ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1466

The γ-aminobutyric acid andproton signaling systems in thezebrafish brain

Characterization and effect of stress

ARIANNA COCCO

ISSN 1651-6206ISBN 978-91-513-0344-4urn:nbn:se:uu:diva-348421

Dissertation presented at Uppsala University to be publicly examined in Auditorium Minus,Museum Gustavianum, Akademigatan 3, Uppsala, Saturday, 9 June 2018 at 09:15 for thedegree of Doctor of Philosophy (Faculty of Medicine). The examination will be conductedin English. Faculty examiner: Assistant Professor William H. J. Norton (University ofLeicester).

AbstractCocco, A. 2018. The γ-aminobutyric acid and proton signaling systems in the zebrafishbrain. Characterization and effect of stress. Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Medicine 1466. 88 pp. Uppsala: Acta UniversitatisUpsaliensis. ISBN 978-91-513-0344-4.

The central nervous system of vertebrates is continuously processing sensory informationrelayed from the periphery, integrating it and producing outputs transmitted to efferents. Inthe brain, neurons employ an array of messenger molecules to filter afferent information andfinely regulate synaptic transmission. The γ-aminobutyric acid (GABA) is the major inhibitoryneurotransmitter in the adult vertebrate central nervous system, synthesized from α, L-glutamateby the glutamate decarboxylases (GADs). GABA promotes fast hyperpolarization of targetcells mediated by the ionotropic, chloride-conducting type A GABA (GABAA) receptors. Thosechannels are homo- or heteropentamers and, in the zebrafish, at least twenty-three genes encodefor putative GABAA receptor subunits.

The present PhD thesis presents the expression levels of the almost complete panel of theGABA signaling machinery in the adult zebrafish brain and retinas. The results point towardGABA signaling modalities in zebrafish strikingly similar to those observed in mammals.The most common GABAA receptor subunit combinations in the whole brain were proposedto be α1β2γ2 and α1β2δ, and region-specific GABAA channels were also inferred. Thoseincluded telencephalic α2bβ3γ2, α2bβ3δ, α5β2γ2, α5β3γ2 and cerebellar α4β2γ2 and α4β2δ. A tissuespecific expression was documented for the paralogues α6a and α6b; the former was abundantlytranscribed in the retinas, the latter in the cerebellum. Proposed retinal GABAA receptors wereα1βxγ2, α1βxδ, α6aβxγ2 and α6aβxδ, with either β2 or β3.

Focus was also placed on functional aspects of the GABA signaling system in the adultzebrafish brain, and specifically on the effects of stress on GABAA receptor subunits expression.Treated animals experienced social isolation and repeated confinement, and depicted increasedmRNA levels of several GABAA receptor monomers. It was deduced that a higher number ofextrasynaptic, tonic-current-mediating GABAA channels was synthesized in the brain followingstress. As synaptic transmission promotes extracellular acidification, interest was also placed onthe acid-sensing ion channel (ASIC) subunits. The overall results presented in this PhD thesispoint toward GABA and proton signaling systems in the zebrafish brain that have many commonpoints with those of mammals. Thus, fundamental signaling pathways appear to be conservedacross vertebrates.

Keywords: acid (GABA), receptors, adult zebrafish, central nervoussystem, gene expression profiling, extracellular acidification.

Arianna Cocco, Department of Neuroscience, Physiology, Box 593, Uppsala University,SE-75123 Uppsala, Sweden.

© Arianna Cocco 2018

ISSN 1651-6206ISBN 978-91-513-0344-4urn:nbn:se:uu:diva-348421 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-348421)

γ-aminobutyric GABAA

Alla mia famiglia in Italia

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Cocco, A., Rönnberg, A. M. C., Jin, Z., André, G. I., Vossen, L. E., Bhandage, A. K., Thörnqvist, P.-O., Birnir, B., Winberg, S. (2017) Characterization of the γ-aminobutyric acid signaling system in the zebrafish (Danio rerio Hamilton) central nervous system by reverse-transcription polymerase chain reaction. Neuroscience, 343:300-321.

II Cocco, A., Williams, M. J., Thörnqvist, P.-O., Winberg, S. (2018) Neural expression patterns and protein modeling of the zebrafish (Danio rerio Hamilton) GABAA receptor ζ subunit. Submitted to Zebrafish.

III Cocco, A., Vossen, L. E., Näslund J., Thörnqvist, P.-O., Winberg, S. (2018). Confinement affects the mRNA levels of key elements in the γ-aminobutyric acid and proton signaling systems in the zebrafish brain. Manuscript.

Reprints were made with permission from the respective publishers.

Contents

Introduction ................................................................................................ 11

Background ................................................................................................. 13The γ-aminobutyric acid signaling system ............................................ 13

GABA turnover .................................................................................. 14Anabolism ........................................................................................ 14Catabolism ........................................................................................ 16 GABA and energy metabolism ........................................................ 19

Receptors ............................................................................................. 23 GABAA receptors ............................................................................. 23GABAB receptors ............................................................................. 29

The proton signaling system ................................................................... 30Protons and synaptic transmission ....................................................... 30The proton receptor system .................................................................. 33

Defensive survival circuits ..................................................................... 37Anatomical considerations ................................................................. 37GABAergic transmission in defensive survival circuits .................. 46

GADs ................................................................................................ 46GABAA receptors ............................................................................. 48

Acid sensing in defensive survival circuits .............................................. 53

Aims ............................................................................................................. 56

Experimental procedures ......................................................................... 57Animals .................................................................................................... 57

Zebrafish .............................................................................................. 57Three-spined stickleback .................................................................... 60

Purification of nucleic acids ................................................................... 61RNA extraction ................................................................................... 61Plasmid purification ............................................................................ 63

Gene expression profiling ....................................................................... 64Design and test of primer pairs .......................................................... 64

Considerations on the results and future research directions ........... 68

Acknowledgements ................................................................................... 74

References ................................................................................................... 76

Abbreviations

3α,5α-THDOC 5α-pregnan-3α,21-diol-20-one AB Accessory basal nucleus of the amygdala ABAT/GABA-T 4-Aminobutyrate aminotransferase/GABA

transaminase ACTH Adrenocorticotropic hormone af Ventral amygdalofugal pathway AHA Anterior hypothalamic area ASIC Acid-sensing ion channel ATP Adenosine triphosphate B Basal nucleus of the amygdala BNST Bed nucleus of the stria terminalis CA Carbonic anhydrase CA3 cornu Ammonis subdivision 3 CBG Corticosteroid-binding globulin CeA Central amygdala CoA Anterior cortical nucleus of the amygdala CoP Posterior cortical nucleus of the amygdala CRH Corticotropin-releasing hormone Dd Dorsal part of the dorsal area of the telencephalon DG Dentate gyrus Dl Lateral part of the dorsal area of the telencephalon Dm Medial part of the dorsal area of the telencephalon DMH Dorsomedial hypothalamic nucleus ec External capsule fx Fornix inf Infundibulum GABA γ-aminobutyric acid GABARAP GABAA receptor-associated protein GAD Glutamate decarboxylase GAT GABA transporter GHBA γ-hydroxybutyric acid GR Glucocorticoid receptor HPA axis Hypothalamic-pituitary-adrenocortical axis HPI Hypothalamic-pituitary-interrenal tissue axis LA Lateral nucleus of the amygdala M Medial amygdala

MR Mineralocorticoid receptor nat Anterior tuberal nucleus (nucleus anterior tuberis) nlt Lateral tuberal nucleus (nucleus lateralis tuberis) pLGIC Pentameric ligand-gated ion channels Pir Piriform cortex PLP Pyridoxal 5’-phosphate PMP Pyridoxamine monophosphate POA Preoptic area POMC Proopiomelanocortin psb Pallial-subpallial boundary PVN Paraventricular nucleus RT + Reverse transcriptase positive RT - Reverse transcriptase negative RT-qPCR Reverse transcription-quantitative polymerase chain

reaction SSAL Succinic semialdehyde st Stria terminalis V Ventral area of the telencephalon vGAT/VIAAT Vesicular GABA transporter/vesicular inhibitory ami-

no acid transporter vSUB Ventral subiculum III Third ventricle References within the thesis introduction are “in inverted commas”; refer-ences to the Papers are in bold.

11

Introduction

Life is made possible by the presence of biological membranes, which define an inner and outer space and thus delineate the fundamental living unit, id est the cell. This entity can itself constitute an organism, or associate with other cells to form pluricellular, complex creatures. In both cases, it is part of an environment which undergoes continuous changes of its chemical compo-sition; with its living activity, a cell also contributes in shaping extra- and intracellular cellular chemical perturbations. The fundamental living unit needs to sense those variations and it responds to produce an appropriate output. The biological membrane contouring the cell cytoplasm is a site of active chemical communication, and determines the ability of a cell to detect changes in the intra- and extracellular environments.

The central nervous system of animals is the center that processes inner and outer stimuli, integrates them and produces an output, manifested through animal behavior. To achieve those complex functions, cellular communication in the brain needs to be highly coordinated and finely tuned. Neurons, the elementary conducting units in the nervous system, are a strik-ing example of cells specialized for communication. Those entities continu-ously experience perturbations of the interstitial chemical composition, par-ticularly in the area of indirect contact with other neurons, i. e. the synapse. Here, extensive fluctuations in the concentration of molecules and inorganic ions occur, ensuring a message to spread from the pre- to the postsynaptic cell. Neurons are strictly organized in circuits, which can be anatomically defined, or through the chemical compound, or neurotransmitter, mostly used for synaptic transmission. In some cases, presynaptic neurons employ a cocktail of neurotransmitters; this is the case, exempli gratia, of secreta-gogues synthesized by hypothalamic paraventricular neurons (Gray, 2005a).

The Regnum animale (Linnaeus, 1788a) is ample and diverse, but the principles governing cellular communication are shared across its members. Fundamental signaling circuits, as those generating responses to threats and the striving for survival, are virtually the same in all vertebrates. Specific neurotransmitters are associated to inhibitory or excitatory effects. Those depend on the neurotransmitters themselves, on the receptors present at the cell surface, and on the ionic composition of the intra- and extracellular envi-ronment of target cells. A body of evidence is present on neural communica-tion pathways in the Classis Mammalia (Linnaeus, 1788a); a smaller amount of information is available for vertebrates of other classes.

12

The present PhD thesis profiles the expression patterns of several genes involved in fundamental signaling circuits in the brain and retinas of the zebrafish (Danio rerio Hamilton, 1822). The primary focus was set on the γ-aminobutyric acid (GABA) signaling system, and the results were integrated in light of the vast mammalian literature on the topic (Papers I and II). Sub-sequently, the research interest continued towards the effects of life threaten-ing conditions on the GABA and proton signaling systems in the brain (Pa-per III). From both lines of research, it emerged that in the teleost brain those two fundamental signaling pathways share many aspects with the mammalian ones. The current PhD thesis thus adds a piece of evidence to-wards the biochemical unity of vertebrates.

13

Background

The primary function of the central nervous system is to connect an animal with its outer and inner environments. Sensory information is constantly relayed to central processing units, which elaborate the signals and produce responses. At the cellular level, cell communication is ensured by electrical and chemical signaling. In both cases, a battery of messenger molecules and proteins is necessary for information transduction to happen.

The present doctoral thesis is divided into two parts. The γ-aminobutyric acid (GABA) and the proton signaling systems are discussed first. The se-cond part focuses on defensive survival circuits and integrates the activity of GABA and protons in those fundamental communication pathways.

The γ-aminobutyric acid signaling system The γ-aminobutyric acid (GABA; Figure 1) is the major inhibitory neuro-transmitter in the adult vertebrate brain (Roberts and Kuriyama, 1968). GABA is also an inhibitory neurotransmitter in the retinas in both mammali-an and non-mammalian species (Connaughton et al., 1999; Marc and Cam-eron, 2002; Lukasiewicz et al., 2004; Hildebrand and Fielder, 2011). The aim of this section is to present the biochemical principles governing GABA metabolism, from its biosynthesis to its catabolic pathways. The link be-tween GABA turnover and the energetic state of cells is mentioned at the end of the section.

Figure 1 Lewis structure (A) and CPK space-fill model (B) of the γ-aminobutyric acid (GABA) in its zwitterionic form. The molecule in A was drawn with the software ACD/ChemSketch Freeware 2015.2.5, www.acdlabs.com. The molecule in B was a courtesy by Professor Silvano Geremia (University of Trieste, Italy) and was visual-ized with The PyMOL Molecular Graphics System, Version 1.8.6.2 Schrödinger, LLC. The figure was drawn with Inkscape 0.92, www.inkscape.org.

14

GABA turnover Anabolism GABA is the product of the decarboxylation of the α-amino acid L-glutamate at the C-α level catalyzed by glutamate decarboxylases (GADs; Figure 2A). Vertebrates have at least two isoforms of GADs, GAD67 and GAD65, the protein products of genes GAD1 and GAD2, respectively (Kauf-man et al., 1991; Anglade et al., 1999; Bosma et al., 1999; Sheikh et al., 1999). A third GAD gene, termed GAD3, was first isolated from the brain of a deep-sea fish species (Bosma et al., 1999). More recently, this gene has been identified in several vertebrates by bioinformatics means, including the zebrafish (Grone and Maruska, 2016). This species also has two GAD1 pa-ralogues, gad1a and gad1b whose expression, together with gad2, was ana-lyzed in Paper I. When catalytically active, GADs are dimers with one unit of pyridoxal 5’-phosphate (PLP) covalently bound to each monomer (Porter et al., 1985; Martin, 1987; Fenalti et al., 2007). As mentioned above, the primary reaction catalyzed by GADs is the decarboxylation of Glu to pro-duce GABA. During the reaction, the cofactor and Glu form an external aldimine in the active sites of the enzyme, which are one per monomer (Fig-ure 2B). During this step, CO2 is released from Glu (Figure 2B). Subse-quently, a Tyr residue from the adjacent monomer protonates C-α of the aldimine such that GABA is released (Figure 2B) (Porter et al., 1985; Fenal-ti et al., 2007). This pathway is preferentially catalyzed by the isoform GAD67, which resides in the soma or both in the soma and in neuronal pro-cesses (Kaufman et al., 1991; Wojcik et al., 2006). On the other hand, high amounts of GAD65 are predominantly found in nerve endings of GABAergic neurons (Tappaz et al., 1977; Okamura et al., 1990; Kaufman et al., 1991). Here, the enzyme forms complexes with other proteins and associates with the membrane of mitochondria and synaptic vesicles (Wojcik et al., 2006; Buddhala et al., 2009). GAD65 alternates GABA production to transamina-tion activity, where the amino group of Glu is transferred to PLP. In the lat-ter case, the cofactor leaves GAD as pyridoxamine monophosphate (PMP) and the enzyme is inactive (Figure 2B) (Martin, 1987; Fenalti et al., 2007). This cyclical GAD65 activity results in pulses of GABA synthesis at the axon terminal, likely coupled to synaptic transmission and linked to the general energy state of the cell (Martin, 1987; Buddhala et al., 2009). The preference for protonation of C-α or C-4’ (see Figure 2B) by GAD67 and GAD65, re-spectively, is dictated by the primary structures of the two isoforms, which have about 64% sequence identity (Fenalti et al., 2007; Grone and Maruska, 2016).

15

Figure 2 A Decarboxylation of glutamate by glutamate decarboxylases (GADs) that releases GABA and CO2. B Details of the mechanism of GADs; pyridoxal 5’-phosphate (PLP) is initially bound to the enzyme via a Schiff-bond base with Lys405 or Lys 396 (human GAD67 and GAD65, respectively) of the core protein (Fenalti et al., 2007). When Glu enters the catalytic site, the bond between Lys and PLP is broken and an external aldimine between Glu and PLP is formed. The decar-boxylation is achieved during this step. Alternative protonation of either C-α of the original Glu molecule, or C-4’ of the cofactor, determines whether GABA (GAD67 arrow) or succinic semialdehyde (GAD65 arrow) will be produced. In the first case, the enzyme retains its catalytic activity; in the second one it undergoes autoinactiva-tion (Porter et al., 1985; Martin, 1987; Fenalti et al., 2007). PMP: pyridoxamine monophosphate. The Figure was drawn with ACD/ChemSketch Freeware 2015.2.5 based on Scheme 1 of Porter et al. (1985) and Berg et al. (2012c). The figure was then finalized with Inkscape 0.92.

The active site of GAD65 is open to the cytoplasm and its carboxyl terminal domain is more flexible compared to that of GAD67. Possibly, a looser cata-lytic pocket would confer more mobility to Tyr425 of GAD65 (Tyr434 in GAD67), which would then be free to protonate C-α or C-4’ with the same

16

probability. Tyr434 of GAD67 could also protonate C-4’; however, this iso-form has a loop which acts as a lid closing the active site. In the mechanism of cofactor transamination, the nucleophilic attack of an oxygen atom of a water molecule is essential to produce PMP and succinic semialdehyde (SSAL; see Figure 2B and Scheme 1 of Porter et al., 1985). The catalytic loop of GAD67 could prevent water molecules to enter the active site and render re-protonation of Tyr434 more likely to happen. The activity of GAD65 incubated with PLP has a two-fold increase compared to that of GAD67 in the presence of the cofactor (Kaufman et al., 1991). This evidence strengthens the hypothesis that the catalytic lid plays a substantial role in packing together the external aldimine and the enzyme core, and favors GABA production. Taken together, those considerations point towards the specialization of GAD67 to produce GABA for metabolic purposes, whereas GAD65 synthesizes GABA for synaptic transmission (Kaufman et al., 1991; Fenalti et al., 2007). Nonetheless, the subcellular localization of GAD67 de-pends on the brain area considered, and this isoform can associate with syn-aptic vesicles as GAD65 (Wojcik et al., 2006). The processes described so far point towards a regulation of GADs activity at different levels. Moreover, several vertebrate species have additional GADs genes, such as GAD3 and the paralogues gad1a and gad1b (Bosma et al., 1999; Grone and Maruska, 2016). Whether those genes are all translated into functional protein products has not yet been investigated. In any case, the presence of three or more GAD isoforms may increase the variety of GABA production pathways and augment the complexity of the balance between alternative protonation events.

Catabolism GABA-mediated synaptic transmission terminates when the amino acid is removed from the synaptic or extrasynaptic space (Conti et al., 1998; Jin et al., 2013). Four GABA transporters (GATs) are present in mammals, of which GAT-1 is found in both neuronal and astrocytic processes, whereas GAT-3 is exclusively confined to the latter structures (Minelli et al., 1996; Conti et al., 1998; Chen et al., 2004). The transport of GABA is associated with a neurotransmitter-coupled and a neurotransmitter-gated current, which originates by the symport of Na+ and Cl– with GABA (Radian et al., 1986; Keynan and Kanner, 1988; Krause and Schwarz, 2005; Zomot et al., 2007). GABA uptake causes a depolarization of the plasma membrane; the transport stoichiometry for GAT-1 is mGABA:nNa+:mCl–, with m = 1 and n = 4 (Keynan and Kanner, 1988; Krause and Schwarz, 2005; Nelson et al., 2008). The depolarization concomitant to GABA uptake results from a first influx of two Na+ ions, upon which the transport is dependent (Keynan and Kanner, 1988; Krause and Schwarz, 2005; Zomot et al., 2007). Cl– is re-cruited to partially neutralize the positive charges and it is essential for the structural rearrangements that allow neurotransmitter reallocation (Zomot et

17

al., 2007). The Cl– binding site is close to Na+-binding site Na1; the residues coordinating the anion are Y86, S295, N327, S331 and Y92, S309, N341, S345 in GAT-1 and GAT-3, respectively (Forrest et al., 2007; Zomot et al., 2007). The second, GABA-gated Na+ current follows the neurotransmitter-transport step and further adds electrogenicity to the overall process (Krause and Schwarz, 2005). GATs are detected as monomers, dimers, or oligomers and are glycosylated in vivo; each unit of GAT mediates the transport de-scribed above (Radian et al., 1986; Bennett and Kanner, 1997; Chen et al., 2004; Hu et al., 2017). When GABA is taken up by the presynaptic neuron, it is directed back into the synaptic vesicles by the vesicular GABA trans-porter (vGAT) (McIntire et al., 1997; Wojcik et al., 2006). vGAT is encoded by the gene SLC32A1; it comprises a long, cytoplasmic amino terminal do-main and nine transmembrane α-helices (Juge et al., 2009). vGAT can me-diate the transport of both GABA and glycine at axon terminals of neurons at inhibitory synapses; hence, the alternative name of vesicular inhibitory ami-no acid transporter (VIAAT) (Wojcik et al., 2006; Juge et al., 2009). The secondary active transport of neurotransmitters into synaptic vesicles is fueled by the activity of the vacuolar H+-ATPase (Moczydlowski, 2012). This protein complex sits in the vesicle membrane and concentrates protons into the lumen by hydrolyzing ATP (Finbow and Harrison, 1997). Therefore, it maintains a membrane potential that is more positive on the inside com-pared to the cytoplasm (Finbow and Harrison, 1997; Moczydlowski, 2012). The translocation of GABA or glycine is strictly dependent on this electrical gradient; vGAT mediates a symport of two Cl– ions with one unit of neuro-transmitter per transport cycle (Juge et al., 2009). The proton concentration gradient also fuels vGAT activity, yet to a smaller extent compared to the electrical component (McIntire et al., 1997; Juge et al., 2009). Therefore, in the presynaptic neuron GABA can be packed back into the vesicles for fu-ture needs in terms of synaptic transmission.

18

Figure 3 A The conversion of GABA into succinic semialdehyde (SSAL) by the 4-aminobutyrate aminotransferase (ABAT). The acceptor of the amino group is α-ketoglutarate, which is converted into the correspondent amino acid Glu. B Succinic semialdehyde is toxic for the cell, and it is promptly converted into succinate by the SSAL dehydrogenase (SSAL-DH) (Hearl and Churchich, 1984). At neutral pH, ABAT and SSAL-DH form an enzyme cluster; the interaction between the two en-zymes depends on the ionic strength of the matrix solution (Hearl and Churchich, 1984). C Alternatively, SSAL is reduced to γ-hydroxybutyric acid (GHBA) by the SSAL reductase (Passarella et al., 1984). The picture was prepared with ACD/ChemSketch Freeware 2015.2.5.

Alternatively, GABA can be transported into the matrix of mitochondria both at axon terminals and in glial cells adjacent to the synapse (Berkich et al., 2007; Jin et al., 2013). A GABA transporter is present in the mitochon-drial inner membrane (Passarella et al., 1984; Berkich et al., 2007), even though no direct evidence for this is currently available in vertebrates. In the mitochondrial matrix, GABA is catabolized to succinic semialdehyde by the 4-aminobutyrate aminotransferase (ABAT), also termed GABA transami-

19

nase (GABA-T). This enzyme is a homodimer and requires PLP for catalytic activity; the two monomers are cross-linked via a 2Fe-2S cluster (Storici et al., 2004). As for GADs, each ABAT monomer has one PLP unit bound to a Lys via a Schiff base bond. PLP mediates the transfer of the amino group from GABA to α-ketoglutarate (Figure 3A) (Storici et al., 2004). The prod-ucts of the reaction are succinic semialdehyde (SSAL) and Glu (Figure 3A). When the concentration of intramitochondrial GABA increases, so does the efflux rate of Glu from the matrix into the cytoplasm (Berkich et al., 2007). In the matrix, ABAT can electrostatically interact with the SSAL dehydro-genase (SSAL-DH), which oxidizes SSAL to succinate and produces reduc-ing power (Figure 3B) (Hearl and Churchich, 1984; Cooper, 1985). SSAL can also be reduced to γ-hydroxybutyric acid (GHBA; Figure 3C) (Passarel-la et al., 1984).

The GABA used for synaptic transmission can either be packed back into synaptic vesicles, or degraded in the mitochondria. The carbon atoms of GABA are subsequently employed to extract electrons to fuel the electron transport chain, as discussed in the next section.

GABA and energy metabolism Life is a costly process, and the biological work performed every day must be paid in terms of adenosine triphosphate (ATP) consumption. Therefore, ATP must constantly be regenerated to sustain the pace of life. Every carbon atom that is still partially reduced is used for electron extraction to fuel pro-ton pumping by mitochondrial respiratory complexes. The whole process of GABA turnover is finely tuned to the energy level of the cell. The four-carbon-atom skeleton of GABA is employed to extract energy by conversion of the neurotransmitter into succinate, the latter compound being an interme-diate of the Krebs cycle (Figure 5B). In the mitochondrion, it is likely that enzyme clusters exist such that subsequent reaction steps are linked together (Hearl and Churchich, 1984). Specifically, SSAL-DH may interact with the succinate dehydrogenase, the enzyme catalyzing the oxidation of succinate to fumarate (Figure 5B; Krebs, 1953; Hearl and Churchich, 1984). This en-zyme is a flavoprotein and it is part of the succinate ubiquinone oxidoreduc-tase, or complex II of the mitochondrial respiratory chain (Sun et al., 2005; Moser et al., 2006). Therefore, a direct pathway is present from GABA ca-tabolism to the discharging of succinate electrons onto the electron transport chain (Figure 4).

20

Figure 4 Electron shunt from GABA to the aerobic metabolism. The physical link between ABAT (magenta) and SSAL-DH (blue) ensures fast communication be-tween GABA degradation and the Krebs cycle. Physical interactions between SSAL-DH and complex II of the respiratory electron transport chain (green) further con-nect GABA catabolism to energy production. The enzymes are not to scale; the figure is meant to give a visual overview of the electron flow. In the Krebs cycle, reducing power is in red and the GTP is in green; m. i. m. = mitochondrial inner matrix. The structures were downloaded from the PDB with codes 1OHV (pig ABAT; Storici et al., 2004), 2W8O (reduced form of human SSAL-DH; Kim et al., 2009), 1ZOY (pig complex II; Sun et al., 2005) and prepared with The PyMOL Molecular Graphic Systems, Version 1.8.6.2 Schrödinger, LLC. In ABAT, the 2Fe-2S cluster and the two units of PLP are visualized as spheres, as are FAD, 2Fe-2S, 4Fe-4S, 3Fe-4S, ubiquinone, heme group of Complex II (Sun et al., 2005; Moser et al., 2006). The Krebs cycle was prepared from Fig. 3 of Krebs (1953) and Fig. 17.15 of Berg et al. (2012a) with Inkscape 0.92. The complete figure is partly based on Fig. 1 of Ippolito and Piwnica-Worms (2014). The figure was finalized with Inkscape 0.92.

The energy stored in C-1 and C-4 of GABA (see neurotransmitter structure in Figure 3A) is released during the joint steps catalyzed by ABAT and

21

SSAL, which produce reducing power (Figure 3B and equation 3 of Figure 5A). The rest of the electrons are rescued from C-2 and C-3 from the succin-ate enter point into the Krebs cycle and onwards (Krebs, 1953; Figure 5B). Of the five carbon atoms of the original Glu molecule needed for GABA synthesis, four are used to extract energy and one is lost in the decarboxyla-tion step catalyzed by GADs (see Figure 2A). The decarboxylation of Glu is essential to the formation of the external aldimine between PLP and the ami-no acid (see Figure 2B). This step occurs irrespectively whether GAD cata-lyzes the formation of GABA, or autoinactivates to release PMP and SSAL (see “The γ-aminobutyric acid system, GABA turnover, Anabolism” and Figure 2B). Both GABA and SSAL can be converted into succinate, there-fore linking neurotransmitter synthesis or enzyme autoinactivation to ATP production. The overall flow of electrons from Glu to the Krebs cycle is termed GABA shunt, and it is schematically represented in Figure 5. The shunt is powered by the reactions catalyzed by GADs, ABAT, SSAL (eqs. 1, 2, 3 of Figure 5A, respectively); the process senses the general energy level of the cell (Passarella et al., 1984; Martin, 1987). Variations of the energy status are relayed by changes in the cytoplasmic [ATP]/[AMP]. When this ratio increases, ATP is being actively produced and it allosterically inhibits GADs (Martin, 1987). Conversely, increases in the cytoplasmic [Pi] promote GABA synthesis by its biosynthetic enzymes (Martin, 1987). At least one isoform of GAD is associated with mitochondrial membranes (GAD65; Mar-tin, 1987; Buddhala et al., 2009); it is licit to infer that a crosstalk between GAD65 and mitochondrial anion antiporters exists (Berg et al., 2012b). Those include nucleoside di- and triphosphate translocase, as well as phos-phate:hydroxide antiporter (Berg et al., 2012c). The transport ratios of those proteins likely reflect the general status of aerobic energy production. There-fore, changes in ATP and Pi fluxes from or to the matrix can promptly be communicated to GADs and modulate their activity. After ABAT catalysis, Glu efflux from the mitochondrion can also be a signal for GADs inhibition, as it promotes GADs inactivation. Rather than being an allosteric inhibitor, increasing [Glu] in the cytoplasm may simply increase the probability of GAD65 to catalyze the transamination of the cofactor (Porter et al., 1985; Martin, 1987). GABA and Asp can also hamper GADs activity, whereas PLP promotes the formation of the holoenzymatic form of the enzyme until saturation (Martin, 1987; Kaufman et al., 1991).

22

Figure 5 A The GABA shunt, or the electron flow from four carbon atoms of Glu into the Krebs cycle. The shunt starts with the activity of GADs (equation 1), con-tinues with GABA catabolism by ABAT (equation 2), and finishes in the aerobic energy production with SSAL-DH (equation 3). Collectively, the GABA shunt al-lows one unit of α-ketoglutarate to jump one Krebs cycle’s intermediate, succinyl-CoA, and directly produce succinate (equation 4; see B). B The Krebs cycle. The intermediates are in bold; reducing power is in red; GTP is in green. A was drawn from Cooper (1985) with ACD/ChemSketch Freeware 2015.2.5. B was drawn from Fig. 3 of Krebs (1953) and Fig. 17.15 of Berg et al. (2012a) with Inkscape 0.92. The figure was finalized with Inkscape 0.92.

Possibly, when there is a higher GABA influx into the matrix compared to a basal situation, a neuron is not synaptically active. This means that all the GABA that could be taken up by synaptic vesicles has been taken up, and the rest is available for transport into the matrix. The reasoning would imply a competition between vGAT and the mitochondrial GABA transporter, and would also suggest that vGAT has a higher affinity for GABA than the mito-chondrial transporter. If GABA concentration in the matrix increases, this would be a signal that the neuron does not need to take up any further GABA in the vesicles and that the rest is available for energy production. Therefore, there would be an increase in the cytoplasmic GABA concentra-tion as soon as synaptic vesicles are filled with GABA and vGAT is saturat-ed. In this way, there are the conditions for GADs inactivation. The inhibi-tion would also come from the excess Glu produced intramitochondrially by ABAT. Berkich et al. (2007) demonstrated that the rate of Glu efflux from the mitochondrion increases when the concentration of GABA in mitochon-dria-bathing medium is increased. All those signals would relay to GADs the information that no further GABA is needed, because there is so much in the cell that it can be burned along the energy metabolism. This view is quite tempting; however, further experimental evidence of mitochondrial GABA transporter is needed for speculations in this direction.

23

Receptors The inhibitory effect of GABA on target cells is mediated by two types of GABA receptors, the type A and type B receptors (GABAA and GABAB, respectively). The first ones are ligand-gated anion channels and promote fast hyperpolarization of the membrane of the target cell (Olsen and Sieghart, 2008). GABAB receptors are G-protein coupled receptors and trig-ger a cytoplasmic signaling cascade upon GABA binding (Bowery et al., 2002). In the next section, details are provided on the GABAA receptor. GABAB receptors are briefly mentioned.

GABAA receptors GABAA receptors are ionophores belonging to the pentameric ligand-gated ion channel (pLGIC) superfamily of membrane receptors, also referred to Cys-loop channels (Schofield et al., 1987). Members of this family are also the ionotropic glycine receptors (GlyRs; Huang et al., 2015), the nicotinic acetylcholine receptors (nAChRs; Unwin, 2005), and the serotonin type 3 receptors (5-HT3Rs; Maricq et al., 1991; Hassaine et al., 2014). GABAA receptors are homo- or heteropentamers (Figure 6A) and mediate the flow of small anions upon binding of the ligand (Olsen and Sieghart, 2008). In a physiological context, the chemical species which are conducted through open GABAA receptors are Cl– and HCO3

– ions (Takeuchi and Takeuchi, 1967; Bormann et al., 1987; Kaila and Voipio, 1987). The human genome comprises nineteen genes encoding for GABAA receptor subunits, divided in the eight subfamilies α (1-6), β (1-3), γ (1-3), δ, ε, π, θ, ρ (1-3) (Simon et al., 2004). The zebrafish has seven additional genes for GABAA receptor mon-omers, for a total of twenty-six (H. J. Haines and D. Larhammar, personal communication). The ε or θ subunits are not found in this species, but they present a ζ which is absent in humans, resulting in the seven subfamilies α (1-6b), β (1-4), γ (1-3), δ, π (1, 2), ζ, ρ (1-3b). In teleosts, the expansion of the panel of GABAA receptor subunits may be partly due to the third round of whole genome duplication, occurred during the evolution of this clade (Vandepoele et al., 2004). All GABAA monomers share a common architec-tural plan (Figure 6B), which frames the multimeric receptors into the pLGIC superfamily. Some two hundred amino acid residues, accounting for more than half of the protein, compose the amino terminal, extracellular domain (ECD) of the monomer (Figure 6C; Miller and Aricescu, 2014; Coc-co et al., 2017). In this region, the polypeptide chain is folded into a β-sandwich made by ten β-strands (Miller and Aricescu, 2014; see also Paper II). N-glycosylation consensus sequences are present in the ECD, and sugar moieties have been solved in the human β3 GABAA receptor subunit (Figure 6A, B) (Miller and Aricescu, 2014; Cocco et al., 2017). The residues down-stream β6 to the middle of β7, including two cysteine residues forming a disulfide bridge, constitute the Cys-loop (Miller and Aricescu, 2014). Of the

24

fifteen residues in the loop, Cys136, Pro144, Asp146, Gln148, and Cys150 are strictly conserved among most of the pLGICs (GABAA receptor β3 subu-nit numbering; Miller and Aricescu, 2014). Pro144 confers a bend to the loop, and it is always preceded by an aromatic residue, either Phe or Tyr (Schofield et al., 1987; Simon et al., 2004; Miller and Aricescu, 2014; Cocco et al., 2017). The polypeptide continues with a transmembrane domain (TMD) formed by four α-helices (M1-4); in fully assembled receptors, each subunit contributes with its M2 α-helix to line the ion pore (Figure 6A, B; Olsen and Sieghart, 2008; Miller and Aricescu, 2014). GlyRs, nAChRs and 5-HT3Rs monomers all share those structural features with GABAA receptor subunits (Figure 6C) (Schofield et al., 1987; Simon et al., 2004; Unwin, 2005; Miller and Aricescu. 2014; Hassaine et al., 2014; Huang et al., 2015). The binding of the ligand to the orthosteric site triggers a series of conforma-tional changes that start in the ECD and propagate to the TMD. The ECD and TMD communicate via surface interactions; the β1-β2 loop in the ECD contacts the carboxyl terminal portion of M2, as well as the N-terminal seg-ment of the loop between the transmembrane M2 and M3 (see Figure 6B) (Miller and Aricescu, 2014; Nemecz et al., 2016). The Cys-loop establishes contacts with the N-terminals of M1 and M3 and with the C-terminals of M2 and M4, therefore touching every transmembrane element (Miller and Aricescu, 2014; Nemecz et al., 2016). The subunits in GABAA receptors are arranged such that it is possible to define a principal and a complementary face for each of them. Those are defined by looking at the receptor from the ECD starting with the principal face on the right, then moving clockwise with respect to the axis of the receptor (see Fig. 3a of Miller and Aricescu, 2014, and Fig. 2a, b, d of Miller et al., 2017). The Cys-loop in the ECD of the principal face of one subunit is contacted by the ECD β8’-β9 loop from the complementary face of the adjacent subunit (Miller and Aricescu, 2014; Nemecz et al., 2016). Additional surface interactions are present between the M2-3 loop of the principal face and the N-terminal of M1 in the complemen-tary face (Miller and Aricescu, 2014; Nemecz et al., 2016). The receptor exists in several conformational states, each with a different affinity for the ligand and with a different probability to be bound to GABA. As recalled above, there are eight families of GABAA receptor subunits in the human genome, and the α and β subunits are always found in heteropentameric GABAA receptors (see Table 3 of Olsen and Sieghart, 2008). Additionally, GABAA receptors exclusively composed by ρ subunits are found in the reti-nas (Lukasiewicz et al., 2004; Olsen and Sieghart, 2008). In α- and β-containing receptors, the binding site for GABA is located at the interface of the ECDs of adjacent subunits (Olsen and Sieghart, 2008; Miller and Arices-cu, 2014). Specifically, it involves the region of the β4 strand, the C-terminal of the β7 strand into the β7-β8 loop, and the β9-β10 loop extending into the β10 strand on the principal face of a β subunit (Boileau et al., 1999; Miller and Aricescu, 2014). The complementary α face completes the agonist bind-

25

ing site with a stretch of twelve amino acid residues in the ECD β2 strand and the N-terminal half of the β6 (Boileau et al., 1999; Miller and Aricescu, 2014).

Figure 6 A Cartoon representation of the human β3 homopentameric GABAA recep-tor (PDB 4COF) as solved by Miller and Aricescu (2014). The authors generated a truncated form of the β3 subunit where the long intracellular loop between M3-4 was replaced by a linker sequence (see Miller and Aricescu, 2014, for details). The sec-ondary structure elements were color-coded following Figure 1a, b of Miller and Aricescu (2014). α-helices are in firebrick, β-sheets in blue; the pore-lining M2s in teal. N-linked glycans are in stick representation with the carbon skeleton in orange and the other elements with standard colors. To be continued on page 26.

26

Figure 6 A (continued from page 25) The β1-β2 loop is in violet, the Cys-loop in green, the M2-3 loop in yellow. Those structural elements constitute part of the ECD (β1-β2 loop, Cys-loop) and TMD (M2-3 loop) interacting surfaces (see main text for more details). Chloride ions are represented by green spheres. EC fluid: extracellular fluid; Cyt: cytoplasm. B A β3 human monomer with the same highlighted structural elements as in A (Miller and Aricescu, 2014). A benzamidine unit is visible in stick representation colored in chartreuse. C Multiple sequence alignment of the chain A of the human β3 GABAA homopentamer with chain A of the α3 subunit of the hu-man homopentameric GlyR (5CFB; Huang et al., 2015), chain A of the heteropen-tameric nAChR from Torpedo marmorata (2BG9; Unwin, 2005), chain A of the murine homopentameric 5-HT3R (4PIR; Hassaine et al., 2014). The N-glycosylation consensus sequences are in bold orange. The Cys residues forming the bridge and lining the loop are in white on black background. The residues in Cys-loop are in green, apart from the conserved P144 (β3 numbering; bold black) and the N residue of the glycosylation consensus sequence. The transmembrane helices M1-4 are dou-ble underlined; the intracellular α-helices between M3-4 of nAChR and 5-HT3R are underlined with dots (Miyazawa et al. 2003; Unwin, 2005; Hassaine et al., 2014). For 4COF, β1-β2 loop is in violet and M2-3 loop in yellow, color-coded as in A. The multiple sequence alignment was made with MUSCLE (MUltiple Sequence Com-parison by Log-Expectation, https://www.ebi.ac.uk/Tools/msa/muscle/). The mole-cules in A and B were prepared with The PyMOL Molecular Graphic System, Ver-sion 1.8.6.2 Schrödinger, LLC, then finalized with Inkscape 0.92.

In the neurotransmitter binding pocket, aromatic residues are essential in coordinating the positively charged N-terminal of GABA (see Figure 1A; Boileau et al., 1999; Miller and Aricescu, 2014 Nemecz et al., 2016). This feature is also observed in the agonist binding cavity of other pLGICs mem-bers (Nemecz et al., 2016). Considering a GABAA receptor with a stoichi-ometry of 2α:2β:1x, and assuming that subunits that belong to different fam-ilies alternate, two GABA-binding sites per receptor can be identified (Olsen and Sieghart, 2008). The opening of the ion pore is a probabilistic shift be-tween different conformational states, with the binding of the first GABA increasing the probability for the binding of the second neurotransmitter unit. Altogether, those changes lead to a preactivated state of the channel, which will thermodynamically fall into the ligand-bound, open state of the receptor (Gielen et al., 2012; Nemecz et al., 2016). The structural reorganization starts at the orthosteric site, propagates to the M2-3 loop, and from here to the ECD (Nemecz et al., 2016) thanks to the surface interactions described above. In pLGICs, those events lead to a constriction of the ECD vestibule; the ultimate rearrangements involved the TMD helix bundle (Nemecz et al., 2016). This moves such that the M2s tilt further apart from each other, com-pared to the closed state (Nemecz et al., 2016). The reduction of the vesti-bule diameter may contribute in breaking the hydrogen bonds between the hydration shell and the ions conducted by the receptors. GABAA receptors also present allosteric binding sites that can accommodate agonists potentiat-ing GABA action. Those compounds include barbiturates, benzodiazepines

27

and neurosteroids; by binding before GABA, they promote structural shifts such that the receptor becomes more affine to the endogenous ligand (Bow-ery et al., 1984; Gielen et al., 2012; Miller et al., 2017). Stress hormone metabolites also potentiate the action of GABA on target cells in a region-specific manner (Belelli and Lambert, 2005; Wang, 2011). The three-dimensional rearrangements of the receptor upon ligand binding are a matter of time and probability. After GABA binding, or in conditions of repetitive agonist stimulation, the receptor spontaneously stops conducting. This de-sensitization phenomenon arises from the M2 helices, which act as rigid bodies and move such that their N-terminal domains come closer together compared to the open state (Gielen et al., 2015). The consequent steric hin-drance at the cytoplasmic end of M2 results in the occlusion of the main pore (Gielen et al., 2015). M2 movements are also transmitted to the other helices in the TMD bundle, and ultimately to the TMD:ECD interface, further stabi-lizing the desensitized, non-conducting state of the receptor (Gielen et al., 2015). In GABAA receptors with the stoichiometry 2α:2β:1x, the fifth subu-nit is usually γ2 or δ and strongly influences the channel conducting proper-ties (Haas and Macdonald, 1999; Bianchi and Macdonald, 2002; Olsen and Sieghart, 2008). GABAA receptors with the γ2 subunit activate faster and mediate currents with higher amplitudes compared to δ-containing channels (Haas and Macdonald, 1999; Bianchi and Macdonald, 2002; Wohlfarth et al., 2002). They also desensitize more promptly and with fast and intermedi-ate components absent with the δ subunit (Haas and Macdonald, 1999; Bian-chi and Macdonald, 2002). Receptors with chimeric δ subunit containing γ2 residues in the M2 segment presented faster desensitization (Bianchi and Macdonald, 2002). Thus, the pore-lining α-helix is a key element in the on-set of this phenomenon (Bianchi and Macdonald, 2002; Gielen et al., 2015).

The anchoring of mature GABAA receptors to the plasma membrane and their correct surface compartmentalization are the result of interactions be-tween the channels and cytoplasmic elements. The GABAA receptor-associated protein (GABARAP) mediates contacts between the receptors and tubulin, and promotes receptor clustering enhancing the total conductivity of the cell (Coyle et al., 2002; Everitt et al., 2004). GABARAP monomers are globular proteins whose N-terminal domain flips by approximately 180° when it associates with another GABARAP unit (Coyle et al., 2002). The polymerization involves an N-terminal β-strand allocated into the preceding monomer, where it interacts with a β-sheet in the C-terminal domain (Coyle et al., 2002). The head-to-tail polymerization fashion of GABARAP is simi-lar to that observed for ubiquitin. The directionality of GABARAP mole-cules define two poles of interactions; the N-oriented is responsible for tubu-lin binding, the C-oriented for establishing contacts with the GABAA recep-tor (Coyle et al., 2002). Several other cytoplasmic elements contact GABAA channels and regulate their location, trafficking and internalization (Chen and Olsen, 2007). Among those is gephyrin, found in indirect association

28

with γ2-subunit containing receptors at postsynaptic sites (Essrich et al., 1998). This protein is essential for targeting the channel at the synapse and it promotes receptor clustering, as does GABARAP (Essrich et al., 1998; Everitt et al., 2004). Interactions between GABAA receptors and GABARAP occur via the γ2 subunit alike (Coyle et al., 2002; Chen et al., 2007b). Specif-ically, they involve an eleven-residue stretch at the C-terminal of the large intracellular loop between M3-4 of the TMD (Coyle et al., 2002). This recognition sequence is highly conserved among vertebrates; in zebrafish, it is GAWRHGRLHIR (Cocco et al., 2017), just one amino acid different (L8I) from the mammalian one (Coyle et al., 2002). The M3-4 loop of γ2 is also involved in the internalization of GABAA receptors by interacting with the clathrin adaptor protein complex via the six-residue stretch YGYECL (Kittler et al., 2008). Those contacts are regulated by phosphorylation events on both Tyr residues, which hinder the interactions and prevent internaliza-tion (Kittler et al., 2008). In the M3-4 loop of γ2, phosphorylation can also occur on a Ser residue part of a protein kinase C consensus sequence (Whit-ing et al., 1990). This sequence is located in an eight-residue stretch, which is alternatively spliced in mammals and zebrafish (Whiting et al., 1990; Cocco et al., 2017).

The variety of GABAA receptor subunits, as well as the possibility of al-ternative splicing, render the array of potential GABAA receptors ample and diverse (Olsen and Sieghart, 2008). Different subunits confer specific physi-cal-chemical properties, including ligand affinity, agonist gating efficacy, conductance features and desensitization rates (see discussion above; Haas and Macdonald, 1999; Bianchi and Macdonald, 2002; Böhme et al., 2004; Farrant and Nusser, 2005; Lindquist and Birnir, 2006; Olsen and Sieghart, 2008). Moreover, the subunit composition of a GABAA receptor predicts its compartmentalization in the plasma membrane. Fast-desensitizing, γ2-containing GABAA receptors are ubiquitously found in the plasmalemma; this subunit often co-localizes with α1, β2, and β3 (Somogyi et al., 1996; Nusser et al., 1998). On the other hand, GABAA channels with the δ subunit are restricted to the peri- and extrasynaptic space (Nusser et al., 1998; Wei et al., 2004; Farrant and Nusser, 2005). α6 can be found with δ in extrasynaptic membrane domains, and both in the synapse and outside it with γ2 (Nusser et al., 1998). Gephyrin interacts with γ2 and targets GABAA receptors to postsynaptic membranes (see discussion above; Essrich et al., 1998). This subunit is also essential for receptor clustering promoted by gephyrin and GABARAP (Essrich et al., 1998; Coyle et al., 2002; Everitt et al., 2004; Farrant and Nusser, 2005). The presence of γ2 or δ in the receptor is mutually exclusive; therefore, the absence of the former in favor of the latter would be an indirect sign to address the GABAA channel to peri- or extrasynaptic membrane domains (Essrich et al., 1998; Farrant and Nusser, 2005). Moreo-ver, δ-containing GABAA receptors might be freer to move in their final membrane location compared to channels with γ2, as interactions with δ and

29

clustering proteins have not been reported. The α4 and α5 subunits are pref-erentially assembled in receptors whose destination is extrasynaptic, whereas α1, β2, β3 are usually situated in the synapse (Nusser et al., 1995; Nusser et al., 1998; Caraiscos et al., 2004; Farrant and Nusser, 2005). The distribution of GABAA channel monomers, as well as their level of expression, depends on the brain region considered. The mammalian α6 is only produced in the cerebellar granule cells, where it can associate with δ giving rise to extrasyn-aptic channels (Kato, 1990; Seeburg et al., 1990; Nusser et al., 1998). In zebrafish, the expression of α6b is restricted to the cerebellum (Cocco et al., 2017). The high affinity for the agonist conferred by α6 and δ is essential for extrasynaptic receptors to intercept GABA molecules that spill over from the synaptic cleft (Rossi and Hamann, 1998; Böhme et al., 2004; Farrant and Nusser, 2005; Lindquist and Birnir, 2006). Such GABAA channels ensure a basal, tonic level of inhibition, which is prolonged in time compared to the phasic inhibition occurring at the synapse (Rossi and Hamann, 1998; Farrant and Nusser, 2005). The δ subunit confers slow and ultraslow phases of de-sensitization, thus making extrasynaptic channels open longer compared to γ2-containing synaptic ones (Bianchi and Macdonald, 2002). Peri- and extra-synaptic GABAA receptors with the α5 subunit also display high affinity for GABA and slower desensitization, compared to α1-containing channels (Caraiscos et al., 2004). Possible GABAA receptor subunit combinations in the zebrafish, as well as comparison with documented mammalian GABAA receptors, are presented in Paper I.

GABAB receptors The G-protein coupled GABAB receptors are heterodimers composed of two seven-transmembrane helices subunits, B1 and B2 in mammals (Bowery et al., 2002). The ligand-binding domain is located in the B1 element; B2 regu-lates the surface expression of the dimer and interacts with the G-protein (Bowery et al., 2002; Burmakina et al., 2014). The conformational shifts triggered by GABA binding in B1 are relayed to B2 via two α-helices, one per subunit, engaged in a coiled-coil structure holding together the receptor (Burmakina et al., 2014). The interactions between the helices consist of a buried hydrogen bond, and polar and hydrophobic interactions (Burmakina et al., 2014). Specifically, residues with hydrophobic side chains are of key importance in ensuring dimerization and address the protein to the cell sur-face (Burmakina et al., 2014). Mammals alternatively splice the B1 subunit, giving rise to different combinations of GABAB receptors (Bowery et al., 2002). The effect of GABAB receptors activation upon ligand binding is slower compared to that mediated by ionotropic GABAA channels. In fact, the formers trigger a cytoplasmic signal cascade with ultimate effects on membrane Ca2+ and K+ conductivity; intracellular production of cyclic aden-osine monophosphate (cAMP) is also regulated (Bowery et al., 1984; Wojcik and Neff, 1984; Bowery et al., 2002). As cAMP is a second cellular

30

messenger acting at allosteric sites of protein kinases, the activation of GABAB receptors may lead to changes in terms of phosphorylation of cyto-plasmic targets. A thorough review on the GABAB receptors goes beyond the scope of this thesis; nonetheless, in Paper I the expression level of the three GABAB receptor subunits in the zebrafish was measured.

The proton signaling system Neurons bathe in the extracellular fluid, whose chemical composition de-pends on transport exchanging activities of the blood-brain barrier, choroid plexuses, glial cells and neurons themselves. Synaptic transmission deter-mines ample fluctuations in the concentration of several ions including H+, therefore affecting the pH of the surrounding medium (Sinning and Hübner, 2013). In the next section, the role of protons as neuromodulatory agents is discussed, as well as membrane proton-sensing mechanisms.

Protons and synaptic transmission The minimal unit for neuronal communication envisages a relaying cell, the presynaptic neuron, and a receiver one, the postsynaptic or target neuron. During signal transmission, both elements undergo profound changes in terms of proton distribution across their axonal and dendritic membranes, respectively. Therefore, the pre- and postsynaptic neurons are involved in shaping pH fluctuations in the extracellular environment, yet in different ways. In the presynaptic cell, the loading mechanism of neurotransmitters into synaptic vesicles envisages the employment of a proton gradient created by the vacuolar H+-ATPase. This enzymatic complex resides in the mem-brane of the vesicle and mediates a primary active transport where ATP hy-drolysis is coupled to the concentration of H+ into the vacuolar space (Fin-bow and Harrison, 1997). As a consequence, a ΔpH is generated, which fuels the packing of neurotransmitters into the cellular compartment (Moczy-dlowski, 2012). The positive membrane potential on the inside of the synap-tic vesicles also contributes to energizing transmitter transport, as happens for vGAT (see discussion above; McIntire et al., 1997; Juge et al., 2009). The H+-ATPase activity results in a vesicular environment that is some two pH units lower compared to the cytoplasmic one, i. e. ~5.7 (Miesenböck et al., 1998). Therefore, fusion of synaptic vesicles to the axon terminal mem-brane following axonal depolarization not only releases specific neurotrans-mitters employed by the neuron, but also the proton content of the vesicle. The result of this process is a transient acidification of the synaptic cleft; a longer lasting extracellular alkaline shift follows (Krishtal et al., 1987). The incorporation of the vacuolar H+-ATPase in the axon terminal membrane may also contribute to the acidification of the synaptic cleft following neu-ronal communication (Krishtal et al., 1987).

31

Figure 7 The reaction catalyzed by CAs (equation 1) and the spontaneous dissocia-tion of H2CO3 at physiological pH (equation 2). The equations were written with ACD/ChemSketch Freeware 2015.2.5 and finalized with Inkscape 0.92.

Mechanisms are present such that the intracellular and interstitial brain pH is maintained within its physiological range. Among those are the carbonic anhydrases (CAs), enzymes located both in the cytoplasm and in the extra-cellular environment catalyzing the reversible hydration of CO2 to carbonic acid, H2CO3 (eq. 1 in Figure 7; Ruusuvuori and Kaila, 2014). The H2CO3 pKa is ~3.5; at physiological pH, this acid promptly dissociates into HCO3

– and H+ (eq. 2 in Figure 7; Mookerjee et al., 2015). Therefore, the catalytic activity of CAs can contribute to increasing the concentration of protons in the solution, but also provides a buffering system via HCO3

– production (Mookerjee et al., 2015). Mammals have at least thirteen catalytically active isoforms of CAs and, in the brain, CA VII locates only in the neuronal cyto-plasm (Ruusuvuori and Kaila, 2014). CAs IV and XIV can be found in the central nervous system alike; they are membrane-bound, with the former associated with both neuronal and glial plasmalemmas, and the latter only in neurons (Ruusuvuori and Kaila, 2014). CA IV is anchored to phosphatidyl-inositol units of cell membranes and accounts for the majority of CO2 hydra-tion in the interstitial space (Tong et al., 2000). Cytoplasmic and extracellu-lar-active CAs engage crosstalks to buffer changes in pH which follow syn-aptic activity (Ruusuvuori and Kaila, 2014). The release of synaptic vesicles depends on Ca2+ influx into the axonal cytoplasm; after transmission, this ion is concentrated back in the extracellular space by the activity of the Ca2+-H+-ATPase. This complex resides in the plasma membrane and couples the hy-drolysis of one ATP unit to the antiport of Ca2+ and H+ with a stoichiometry of 1:1 (Aronson et al., 2012; Ruusuvuori and Kaila, 2014). Therefore, pro-tons are concentrated in the intracellular compartment. The following extra-cellular alkalization is sensed by the interstitial CAs, which promote CO2 hydration and the consequent formation of acid equivalents (see Figure 7; Krishtal et al., 1987; Ruusuvuori and Kaila, 2014). Those result from the spontaneous dissociation of H2CO3, as discussed above (see eq. 2 of Figure 7; Mookerjee et al., 2015). In the axonal cytoplasm, the increase of [H+] stimulates CA VII to produce CO2, which freely diffuses to the interstitial compartment. Thus, the latter acts as CO2 sink, until the physiological pH is restored (Ruusuvuori and Kaila, 2014). Perturbations of the equilibrium of the reaction catalyzed by CAs can originate from postsynaptic activity alike, especially in GABAergic circuits. In fact, GABAA receptors can mediate

32

HCO3– currents; the equilibrium potential for HCO3

– is -10 mV, thus always resulting in an outwardly directed flow of this chemical species (Takeuchi and Takeuchi, 1967; Bormann et al., 1987; Kaila and Voipio, 1987; Chesler and Kaila, 1992). The GABA-induced depolarization of target cells causes an increase in the [HCO3

–]extracellular, which drives the activity of interstitial CAs towards the production of CO2 (see Figure 7; Chesler and Kaila, 1992; Ruuvusuori and Kaila, 2014). On the other hand, the concomitant acidifica-tion of the axonal cytoplasm pressures CA VII to catalyze CO2 hydration for generating buffering capacity in terms of HCO3

– (Chesler and Kaila, 1992; Ruuvusuori and Kaila, 2014). In this case, the CO2 sink is the intracellular environment. The link between GABA signaling and extracellular pH shifts is also explicated through the modulatory action of protons on Cl–-conducting GABAA receptors (Pasternack et al., 1992). Alkalization of the interstitial fluid promotes faster desensitization of GABAA channels, com-pared to a non-shifted situation (Pasternack et al., 1992). Channel conduct-ance is restored as soon as the interstitial pH falls, and it increases together with increases in [H+]extracellular (Pasternack et al., 1992). Therefore, acidifica-tion of interstitial pH promotes inhibitory GABA-mediated Cl– currents in target cells (Pasternack et al., 1992). The same pH shift has a blocking effect on excitatory transmission mediated by Glu (Sinning and Hübner, 2013). Extracellular pH falls in the brain may be associated to a high-alert status of the body in a real or perceived dangerous situation (Wemmie, 2011). None-theless, the pH must be maintained within its physiological range to prevent tissue damage. From a physiological point of view, it is of great sense that inhibitory synaptic transmission is enhanced while the interstitial pH lowers. As recalled above, the acidification of the extracellular environment is a direct consequence of synaptic vesicles fusion with the axonal membrane (Krishtal et al., 1987). Inhibition mediated by augmented activity of pre- and postsynaptic GABAA receptors would ensure fast termination of cellular communication to prevent further increments in the [H+]extracellular. Opposite considerations apply to Glu-mediated excitatory signals, which become hin-dered (Sinning and Hübner, 2013). The present view is quite tantalizing; however, one has to bear in mind that the experiments by Pasternack et al. (1992) were conducted on crayfish muscle fibers, and findings in an inverte-brate species might not be applicable to the vertebrate central nervous sys-tem. Nonetheless, the physiology beyond interstitial pH-driven synaptic transmission makes a lot of sense.

Finally, energy metabolism can induce pH shifts both in the cytoplasm and in the brain interstitial fluid (Mookerjee et al., 2015). To maintain its high metabolic demands, the brain is in continuous need of ATP, and prefer-entially uses glucose to produce the energy for cellular work. That monosac-charide can be either fermented into lactate, or sent into the aerobic metabo-lism in the mitochondria. Astrocytes and neurons engage energetic crosstalks such that the lactate produced in the formers is shunted to the latters. This

33

process, sometimes referred to as the lactate shuttle, is accompanied by a transient decrease in extracellular pH (Erlichman et al., 2008). The rise in [H+]extracellular is determined by the transport mechanism of lactate, made pos-sible by monocarboxylate transporters (Erlichman et al., 2008). Those pro-teins mediate the electroneutral movement of one lactate unit and H+; the anion flows along its concentration gradient (Erlichman et al., 2008). Dec-rements in both intra- and extracellular pH can also be triggered by mito-chondrial CO2 production resulting from energy metabolism (Mookerjee et al., 2015). In fact, complete oxidation of glucose or lactate in the presence of O2 corresponds to six or three CO2 molecules produced, respectively (Mook-erjee et al., 2015; see also Figure 5B). This chemical species is generated by the catalytic activity of the pyruvate dehydrogenase complex, the isocitrate dehydrogenase, and the α-ketoglutarate dehydrogenase complex, all residing in the mitochondrial matrix (Berg et al., 2012a). Thus, the organelle acts as source of CO2, which freely diffuses to the cytoplasm and the extracellular environment (Mookerjee et al., 2015). Here, an increase in [CO2] pushes the reactions depicted in Figure 7 towards the formation of acid equivalents, therefore diminishing the pH of the solution.

The proton receptor system For the considerations presented in the previous paragraphs, it emerges that synaptic transmission causes perturbations of the chemical composition of the interstitial fluid. Moreover, it intrinsically creates the bases for neuro-modulatory effects. In fact, protons are essential for neurotransmitter loading into synaptic vesicles (see discussion above; Finbow and Harrison, 1997; Moczydlowski, 2012). In the case of GABA, Cl– is also strictly required to fuel amino acid concentration into the lumen of the vesicle (see “The γ-aminobutyric acid signaling system, GABA turnover, Catabolism”; Juge et al., 2009). When communication between pre- and postsynaptic neurons occurs, it not only envisages the use of classical neurotransmitters through which circuits are defined. Rather, it releases inorganic ions accompanying the principal messenger molecule, creating several possibilities of neuronal communication tuning. Examples of modulation have been provided in the previous paragraph, where effects on GABA- and Glu-mediated currents dictated by extracellular pH shifts have briefly been presented. However, protons do not solely act on receptors for classical neurotransmitters. They have their own battery of receptors to which they bind and that contributes in shaping postsynaptic responses. The-acid sensing ion channels (ASICs) are the proton receptors and belong to the degenering-epithelial Na+ channel family of cation channels (Waldmann et al., 1997; Wemmie et al., 2006). Mammals have four genes encoding for ASICs, ASIC1-4, which are ex-pressed throughout the peripheral and central nervous system (Waldmann et al., 1997; Wemmie et al., 2006). In the zebrafish genome, there are three

34

paralogues of ASIC1, namely asic1a-c, one copy of asic2, and two copies of the fourth ASIC, asic4a-b (Pauckert et al., 2004). Totally, D. rerio has six genes for ASICs, and ASIC3 seems to be absent in this species (Pauckert et al., 2004). ASICs are channels for small inorganic cations and are involved in noci- and mechanoception (Waldmann et al., 1997; Wemmie et al., 2006). The main species that flows through an open proton receptor is Na+; fluxes of Ca2+, K+, Li+ and H+ themselves are also possible (Waldmann et al., 1997; Wemmie et al., 2006). The fundamental unit of proton receptors is a trimer which can be composed of the same monomer, or of different polypeptides (Wemmie et al., 2006; Jasti et al., 2007). Therefore, homo- or heterotrimers can be found; the ability to detect extracellular pH shifts as well as desensiti-zation and recovery rates are influenced by the subunit composition (Bassi-lana et al., 1997; Askwith et al., 2004; Coryell et al., 2007; Jasti et al., 2007). Moreover, alternative splicing is documented for the mammalian ASIC1 and ASIC2 monomers. Splice variants are synthesized at different rates in the brain and confer specific channel properties to the receptor (Wemmie et al., 2002; Askwith et al., 2004; Wemmie et al., 2006). Cell-specific localization of ASICs are possible; for instance, ASIC1a in only present in neurons, where it localizes on dendrites in the synaptic cleft (Wemmie et al., 2002; Wemmie et al., 2004). The structure of a single mon-omer comprises a large ECD flanked by two membrane-spanning α-helices, M1 and M2 (Jasti et al., 2007). The N- and C-terminals are intracellular, and they interact with cytoplasmic elements in a monomer-specific fashion (Wemmie et al., 2006; Jasti et al., 2007). In the ECD, secondary structure elements are folded into a fist-like conformation and, in the thumb-like por-tion, fourteen cysteine residues are engaged in seven disulfide bridges (Waldmann et al., 1997; Jasti et al., 2007). The thumb-like region extends to the border of the TMD, with which it interacts via Trp288, conserved in all ASICs (numbering of ASIC1; Waldmann et al., 1997; Jasti et al., 2007). Possibly, H+-gating triggers a movement of the ECD subsequently trans-ferred to the TMD. Pivotal to this process may be the rigid, disulfide-rich thumb-like domain; in turn, the latter may communicate the conformational changes to the TMD via Trp288. That residue is positioned such that it might engage stacking interactions with aromatic residues in the loop between M1 and the extracellular domain (Jasti et al., 2007). In other members of the degenering-epithelial Na+ channel family a Tyr is found in place of Trp288 (Waldmann et al., 1997; Jasti et al., 2007). The conservation of an aromatic lateral chain in a key position for putative signal transduction might add strength to the proposed coupling mechanism. It is not clear whether ASIC trimers form a pore; nonetheless, the structure solved by Jasti et al. (2007) presents several extracellular cavities and three V-shaped fenestrations, posi-tioned at the border between the extracellular and lipid environments. Inor-ganic cations might enter the trimer from those openings. Subsequently, they would flow along their electrochemical gradient, facilitated by the negative

35

potential of the buried surface of the TMD (Jasti et al., 2007). Each ASIC monomer presents eight potential sites for protonation; those are four pairs of residues with a carboxylic group on their side chain, namely Asp or Glu (Sawyer and James, 1982; Jasti et al., 2007). Three of the pairs are located at the interface between subunits. One can look at the trimeric proton receptor from the extracellular space and designate the different lateral surfaces of the monomers as principal and complementary in a clockwise fashion. The pro-cedure resembles that employed for defining principal and complementary faces of GABAA receptor subunits (see “The γ-aminobutyric signaling sys-tem, Receptors, GABAA receptors”; Miller and Aricescu, 2014; Miller et al., 2017). By applying this reasoning to proton receptors, each monomer con-tributes to the H+ binding pocket with two acidic pairs on the principal face and one on the complementary one. The result is one acidic hole per inter-subunit interface which is sensitive to protons, and whose pH sensitivity can be modulated by the protein environment and interactions between the car-boxyl groups themselves (Sawyer and James, 1982; Jasti et al., 2007). This line of thought would imply that different ASICs could give a different con-tribution to the acidic pocket in terms of its physical-chemical properties. Thus, the various ASIC monomers would modulate pocket features in differ-ent ways and to different extents. Experimental evidence confirms this view, and heterotrimeric proton receptors have lower H+ sensitivity compared to homotrimeric channels composed by ASIC1 (Bassilana et al., 1997; Askwith et al., 2004). Moreover, mammalian heterotrimers with the splice variants ASIC1a:ASIC2a desensitize faster than homomeric receptors (Askwith et al., 2004). The fourth pair that constitutes the proton antenna system is bur-ied into the lower portion of the ECD, somehow on top of the V-shaped fen-estration (Jasti et al., 2007). The Henderson-Hasselbach equation can help describe the acid-sensing mechanism of ASICs. This equation expresses the pH as a function of both the pKa and the proportion between the protonated and deprotonated forms of a given acid, i. e. pH = pKa + log{[conjugate base]/[acid]} (Chang, 2000). In proteins, it is not uncommon to find pairs of Asp or Glu whose lateral carboxylic groups are close to one another (Sawyer and James, 1982). This proximity creates a microenvironment that usually increases the pKa of those groups compared to the schoolbook value of 4.1 (Sawyer and James, 1982). Asp and Glu pairs of ASICs pH sensors are in-deed very close to one another (see the structure of ASIC, 2QTS, in the PDB; Jasti et al., 2007). In any case, the pKa for those residues remains low-er compared to the typical interstitial pH of 7.3 (Ruusuvuori and Kaila, 2014). Thus, one could expect those channels to respond when the extracel-lular pH decreases below 7.3, and to start becoming saturated as soon as the pH value falls beyond the pKa (Sawyer and James, 1982; Jasti et al., 2007). At resting conditions, the number of protons in the synaptic clef is not enough to bind to all the negative charges of ASICs H+ antennas. In other words, the amount of negative charges exceeds that of the positively charged

36

H+, and the majority of ASICs is in the closed, non-conducting state. How-ever, as soon as synaptic transmission occurs, the synaptic vesicles pour their acidic content into the cleft with the consequent interstitial pH fall. There-fore, an increasing number of Asp and Glu residues in the key pH sensing pairs of ASICs becomes protonated. The acidification of the synaptic cleft may continue until the amount of H+ outnumbers the buffering capacity pro-vided by the negatively charged carboxylate groups. At this point, all the carboxylate lateral chains have turned into carboxylic units and there is no negative charge left to bind the excess H+. The point at which pH = pKa cor-responds to a logarithmic term = 0 in the Henderson-Hasselbach equation, achieved when there are equal amounts of de- and protonated forms of Asp and Glu lateral chains (Chang, 2000). This might be the tiny window during which ASICs are conducting (see the Henderson-Hasselbach equation; Saw-yer and James, 1982; Jasti et al., 2007). In this state, carboxyl-carboxylate interactions may shift the architectural equilibrium of ASICs towards the open state by a shared proton between acidic lateral chains (Sawyer and James, 1982). Therefore, ASICs are being titrated every time synaptic transmission occurs, and the reference internal standard is provided by the known pKa of Asp and Glu lateral chains (Chang, 2000). Somehow in line with those considerations, H+-gated cation currents are measured at pH val-ues that range from 6.5 to 4 (Krishtal and Pidoplichko, 1980; Bassilana et al., 1997; Waldmann et al., 1997; Wemmie et al., 2002; Askwith et al., 2004; Pauckert et al., 2004; Ziemann et al., 2009). Currents with small am-plitude can also be recorded when the pH approaches its interstitial resting physiological value (Kristhal and Pidoplichko, 1982; Bassilana et al., 1997; Pauckert et al., 2004; Ziemann et al., 2009). The subunit composition of ASIC trimers shapes the response to extracellular acidification. The ASIC2 monomer lowers the pH threshold at which channel conductance occurs, therefore diminishing the affinity for protons (Askwith et al., 2004). Moreo-ver, combinations of ASIC1a and ASIC2a subunits lead to receptors which desensitize more promptly compared to ASIC1a or ASIC2a alone (Askwith et al., 2004). As discussed above, the transition between different conforma-tional states might depend on the degree of protonation of key Asp and Glu. A possibility is that ASIC2a creates a different environment in the acidic pocket, if compared to ASIC1a. The differences would regard residues other than the proton antennas Asp and Glu, which are conserved (Jasti et al., 2007). Another possibility is that the subunits in heterotrimeric receptors differently interact compared to those in homotrimers. Perhaps, ASIC2a-containing receptors have weaker subunit-subunit interactions, and thus, the carboxylic groups of the pH-sensing residues are further apart from each other, compared to channels with the sole ASIC1a. This would result in a pKa of lateral chains of key Asp and Glu closer to the typical value of 4.1 (Sawyer and James, 1982). In this case, the Henderson-Hasselbach equation would predict that the point at which the logarithmic term equals 0 is shifted

37

towards a more acidic environment. Therefore, proton receptors containing ASIC2a saturate more slowly compared to other subtypes (Bassilana et al., 1997). Similar conclusions have also been obtained for zebrafish heteromer-ic ASICs in a heterologous expression system (Pauckert et al., 2004). It is licit to infer that the amount of neurotransmitter released during synaptic transmission can vary among circuits. Moreover, there can be stronger or weaker activation of the same circuit, depending on the situation. In those conditions, H+ waves in the synaptic cleft have different amplitudes. The response of postsynaptic neurons to variations in interstitial pH would then be shaped by the proton receptor composition present on its plasma mem-brane. In this way, postsynaptic cells may be sensitive to a range of pH val-ues (Bassilana et al., 1997). As discussed above, an acidic shift in the synap-tic space causes GABA-mediated inhibitory Cl– currents to increase, but dampens Glu-evoked excitatory responses (Sinning and Hübner, 2013). Spe-cifically, it impairs the function of NMDA receptors (Sinning and Hübner, 2013). On the other hand, ASICs are activated when interstitial pH starts to lower, and they mediate cation flows that depolarize the postsynaptic den-dritic membrane. The excitatory effect promoted by ASICs opening might thus restore depolarizing currents when those mediated by Glu start to disap-pear (Wemmie et al., 2002). Nonetheless, ASIC-generated currents are tran-sient, and the channel readily enters a desensitized state (Krishtal and Pidoplichko, 1980; Bassilana et al., 1997; Askwith et al., 2004). The desen-sitization depends on both the pH of the solution and the pKa of the H+ an-tennas.

Finally, mammalian ASIC1 presents a Cl– binding pocket whose lining residues are also found in the zebrafish asic1a-c (Pauckert et al., 2004; Jasti et al., 2007). The binding of this anion to ASIC1a decreases the desensitiza-tion rate of the homomeric proton receptor (Kusama et al., 2010). vGAT mediates the packing of GABA into synaptic vesicles and it is strictly de-pendent on Cl– (see “The γ-aminobutyric acid signaling system, GABA turn-over, Catabolism”; Juge et al., 2009). Therefore, GABA signaling may gen-erate not only fluctuations of interstitial pH, but also changes in the [Cl–

]extracellular. The potentiating effect of Cl– on ASIC currents may counterbal-ance the inhibitory GABAergic transmission. Thus, GABA would be an indirect positive modulator of ASICs function (Kusama et al., 2010).

Defensive survival circuits Anatomical considerations The aim of life is to transmit genes to the next generation and, at least for some animal species, ensure offspring survival. In order to do so, animals face everyday challenges to reach adulthood and reproduce (LeDoux, 2000).

38

The brain defensive survival circuits are those that ensure the capability of animals to recognize a threat and produce a proper behavioral output (Le-Doux, 2014; LeDoux and Brown, 2017). Challenges can come from the out-side world, but they can also be due to perturbations of the internal milieu of an animal. Usually, animals do not choose to respond to a threat. Rather, they cannot avoid undergoing physiological changes that will ultimately produce an output.

A perturbation of the external or internal body environment is peripheral-ly sensed and relayed to hypothalamic centers in the central nervous system (Gray, 2005a). In mammals, dorsomedial parvocellular neurons in the para-ventricular nucleus (PVN) of the hypothalamus produce corticotropin-releasing hormone (CRH) (Gray, 2005a; Herman et al., 2016). Several ver-tebrates possess two CRH paralogues, CRH1 and CRH2, but the latter was lost during the course of evolution in both mammals and teleosts (Grone and Maruska, 2015a; Cardoso et al., 2016). However, teleosts have two copies of the CRH1 gene, namely crha and crhb (Grone and Maruska, 2015b; Cardoso et al., 2016). Those are probably the result of the third round of whole ge-nome duplication that occurred in the teleost lineage (Vandepoele et al., 2004; Grone and Maruska, 2015a, 2015b). Subsequent genome duplication events and gene losses further shaped the pattern of CRH genes in teleosts (Grone and Maruska, 2015b; Cardoso et al., 2016). Upon PVN neurons stimulation, CRH is released in the hypothalamic-pituitary portal vessel sys-tem and reaches target corticotropes in the adenohypophysis (Gray et al., 2005a; Herman et al., 2016). In teleosts, CRH-producing neurons are located in the preoptic area (POA) and directly stimulate target cells in the pituitary (Winberg et al., 2016; Gorissen and Flik, 2016). The regulation of hypo-physeal activity is also mediated by the lateral nucleus of the hypothalamus in the lateral tuberal region (Meek and Nieuwenhuys, 1998). The brain area in close proximity to the third ventricle is therefore of pivotal importance for the stress response (Papez, 1937; Gray, 2005a; Herman et al., 2005). CRH induces the release of adrenocorticotropic hormone (ACTH) by pituitary corticotropes; ACTH is one of the cleavage products of proopiomelanocortin (POMC) (Barrett, 2012). ATCH acts at the level of the mammalian adrenal gland cortex, which produces glucocorticoids then released into the blood stream (Barrett, 2012; Herman et al., 2016). ACTH stimulates cortisol re-lease from cells of the interrenal tissue, which is homologous to the mamma-lian adrenal gland cortex (Clark et al., 2011; Winberg et al., 2016; Gorissen and Flik, 2016). The ultimate activation of glucocorticoid-producing cells upon hypothalamic stimulation is termed hypothalamic-pituitary-adrenocortical (HPA) or hypothalamic-pituitary-interrenal tissue (HPI) axis in mammals and teleosts, respectively (Herman et al., 2005; Clark et al., 2011; Winberg et al., 2016). The neural response to threats also encom-passes the PVN-mediated activation of the autonomic nervous system and the medulla of the adrenal gland (Barrett, 2012; Boulpaep, 2012), or the

39

chromaffin cells in teleosts, homologous to the mammalian medulla (Clark et al., 2011; Winberg et al., 2016).

Several brain areas modulate the action of the hypothalamic PVN. For in-stance, sensory and visceral projections ascend from brain stem nuclei and project in a monosynaptic modality onto the PVN (De Kloet et al., 1998; Gray, 2005a; Herman et al., 2005; Herman et al., 2016). Those inputs are primarily aminergic and cause excitation of the PVN (De Kloet et al., 1998; Herman et al., 2005; Herman et al., 2016). Neurons in this hypothalamic region also receive highly processed afferent information from limbic nuclei (Gray, 2005a; Gray, 2005b; Herman et al., 2005). Among those is the amyg-daloid complex, an almond-shaped telencephalic gray-matter structure locat-ed at the pole of the temporal lobe (Sah et al., 2003; Gray, 2005b; Sobotta, 2011). The amygdala is part of the brain defensive survival circuits, it is activated in case of real or perceived danger and it contributes to producing defense responses (Aggleton and Mishkin, 1986; LeDoux and Brown, 2017). In teleosts, learning and avoidance conditioning experiments revealed that the fish homologue to the amygdala is the medial telencephalic pallium (Portavella et al., 2002; Broglio et al., 2005; Salas et al., 2006) The amygda-loid complex of mammals is composed by several nuclei divided in four groups (Figure 8) (Sah et al., 2003). Those comprise the basolateral, cortical and centromedial groups, which are further divided into subnuclei (LeDoux, 2000; Sah et al., 2003; Gray, 2005b; Usunoff et al., 2009). The fourth divi-sion contains the anterior amygdala area, the amygdalo-hippocampal area and the intercalated nuclei (Sah et al., 2003; Usunoff et al., 2009). The amygdala acts as a funnel to integrate processed cortical sensory information and inputs from the olfactory bulbs, and to relay an output onto diencephalic nuclei (Aggleton and Mishkin, 1986; LeDoux, 2000; Sah et al., 2003; Gray, 2005b). Overall, the flow of information through the amygdalar nuclei is markedly unidirectional; the main entrance points are the lateral division of the basolateral group and nuclei in the centromedial group (LeDoux, 2000; Sah et al., 2003; Gray, 2005b). Massive amygdalar projections arise from the central nulcei, including efferents to the hypothalamus (LeDoux, 2000; Sah et al., 2003). The amygdala influences the activity of the PVN via multi-neuron pathways. On one hand, the ventral amygdalofugal pathway connects the amygdaloid body to the preoptic area and to the anterior hypothalamus (Figure 1) (Gray, 2005a; Herman et al., 2005; Usunoff et al., 2009; Sobotta, 2011; Kamali et al., 2016). This amygdalar efferent is direct and runs paral-lel to the anterior commissure (Kamali et al., 2016). On the other hand, the major efferent of the mammalian amygdaloid body, the stria terminalis, syn-apses onto the bed nucleus of the stria terminalis (BNST) in the diencepha-lon (Figure 1) (Kamali et al., 2015; Kamali et al., 2016). The stria terminalis follows the fornix and reaches the BNST dorsally with respect to the anterior commissure (Kamali et al., 2016). Both amygdalar efferent pathways branch and make synaptic contacts with several other regions of the central nervous

40

system, including limbic system areas and amygdalar afferents (Gray, 2005b; Kim et al., 2011; Kamali et al., 2015; Kamali et al., 2016). The BNST also gives rise to projections that innervate PVN defense effectors and the anterior hypothalamus (Kamali et al., 2016; Gorka et al., 2017). Due to the profound link between the centromedial amygdaloid group and the BNST, the term extended amygdala has been proposed by some authors (Sah et al., 2003; Gorka et al., 2017). The major amygdalar hypothalamic effer-ents originate in the centromedial group from GABAergic neurons (Figure 1) (Sah et al., 2003; Herman et al., 2005; Usunoff et al., 2009). Hypothalam-ic preoptic neurons and the BNST, which project to the PVN, also employ GABA as neurotransmitter (Figure 8) (Herman et al., 2005).

Figure 8 Schematic view of the projections of the centromedial amygdalar group to the hypothalamus. The central amygdala (CeA) sends efferent connections to the bed nucleus of the stria terminalis (BNST) through the stria terminalis (st). The BNST projects to the PVN and also to the anterior hypothalamic nuclei (AHA). The medial amygdala (M) makes efferent synaptic contacts with the preoptic area (POA), which sends neurons to the PVN. The parvocellular neurons in the dorsome-dial PVN release CRH that stimulate corticotropes in the adenohypophysis. PVN magnocellular neurons send axon terminals to the neurohypophysis and release arginine vasopressin (AVP). The amygdala was drawn following the rat amygdala scheme of Figs. 1, 4 from Sah et al. (2003), with basolateral nuclei in teal, cortical-like nuclei in olive, centromedial nuclei in green. The fourth group of nuclei (see text) is not represented in the figure. The hypothalamic nuclei are drawn from Fig. 21.11 of Gray (2005a) and the connections between the amygdala and hypothalamus drawn following Gray (2005a), Herman et al. (2005), Sobotta (2011), Kamali et al. (2015), Kamali et al. (2016). Amygdalar projections traveling through the stria ter-minalis pathway are in black, those through the ventral amygdalofugal (af) pathway in red. CRH-releasing neurons of the PVN are in blue. Neurotransmitters are in bold. AB: accessory basal nucleus; B: basal nucleus; CoA: anterior cortical nucleus; CoP: posterior cortical nucleus; ec: external capsule; LA: lateral nucleus; Pir: piriform cortex; III: third ventricle. The figure was prepared with Inkscape 0.92, www.inkscape.org.

41

Therefore, early amygdalar activation due to threatening stimuli results in an increase of PVN neurons activity, and ultimately in the stimulation of the HPA axis (Herman et al., 2005). The processes described so far are just a simple part of the entire picture. In fact, the amygdala sends projections towards the brain stem and, as recalled before, also to amygdalar afferents. Moreover, the brain displays lateralization; connections of the central amygdala and BNST with afferent and efferent structures are different between the two cerebral hemispheres (Gorka et al., 2017; Hakamata et al., 2017). The amygdaloid body also presents internal pathways of afferent and efferent signals, with different nuclei influencing the activity of their neighbors (Sah et al., 2003). The amygdala is connected to other limbic structures, such as the septal nuclei and the hippocampal formation; the latter is a potent modulator of hypotha-lamic activity (De Kloet et al., 1998; Herman et al., 2005; Kamali et al., 2015; Kamali et al., 2016). The hypothalamus itself projects back to the amygdala (Gray, 2005a). The role of the amygdaloid body as cradle of emotions has been questioned (LeDoux, 2000). LeDoux and Brown (2017) argued that it is not possible to determine whether animals other than humans feel emotions, as they cannot let researchers know how they feel in different situations. None-theless, animals are daily put under pressure and they need to respond to threats (LeDoux, 2014; LeDoux and Brown, 2017). The physiological stress response envisages the coordination of peripheral afferent information, central processing and efferent instructions (Boulpaep, 2012). Those highly coordi-nated events are possible because there are bottom-up and top-down interac-tions between the body and the brain, and in the brain itself. The former ones include hypothalamic afferents from the brain stem and amygdalar efferent projections to different cortexes, e. g. the orbitofrontal and medial prefrontal cortexes (Kim et al., 2011; LeDoux, 2014). Top-down interactions are those between the amygdala and its afferents and the hypothalamus and its targets. Those also include hypothalamic indirect efferent outside the central nervous system. The balance between afferents and efferents to and from the amygdala and the hypothalamus determines the amplitude of behavioral responses to a real or perceived threat (Papez, 1937; Kim et al., 2011). Impaired crosstalks between the aforementioned structures may result in pathological conditions (Kim et al., 2011).

A common architectural plan can be observed in the central nervous sys-tem of vertebrates (Salas et al., 2006; Broglio et al., 2011). Due to their im-portance in ensuring an animal’s survival, it is quite likely that basal defen-sive survival circuits are conserved between mammals and fish. In fact, it has been demonstrated that telencephalic nuclei in the teleosts’ brain are the homologues to mammalian limbic structures (Portavella et al., 2002; Broglio et al., 2005; Salas et al., 2006; Broglio et al., 2011). The telencephalon of actinopterygian fish develops by eversion of the neural tube, rather than the evagination observed in all other vertebrate groups (Meek and Nieuwenhuys, 1998; Nieuwenhuys, 2011).

42

Figure 9 Schematic view of defensive survival circuits in the teleost brain. The me-dial part of the dorsal telencephalic area (Dm), or medial pallium, is considered to be the fish homologue to the mammalian amygdala and projects to various hypothalam-ic nuclei, including the anterior tuberal nucleus (nat). The lateral tuberal nucleus (nlt) sends direct projections to corticotropes in the pituitary, as does the POA. POA efferents to the hypophysis are CRH-releasing neurons, nlt neurons contain GABA (Anglade et al., 1999; Winberg et al., 2016; Gorissen and Flik, 2016). The nlt of the zebrafish may also produce both teleost CRH orthologs as the transcripts crha and crhb have been found in that area (Grone and Maruska, 2015b). The Dm receives highly processed sensory stimuli from all sensory systems apart the olfactory epithe-lium, whose information directly reaches the POA through the olfactory bulbs (Meek and Nieuwenhuys, 1998). The POA is a possible Dm efferent, as is the nlt with respect to both the POA and nat. The telencephalic lobe was drawn from Fig. 7c of Nieuwenhuys (2011) (zebrafish telencephalic lobes) and the hypothalamus following Fig. 15.19 of Meek and Nieuwenhuys (1998). In the telencephalon, the dashed lines divide the three main divisions of the pallium. The dotted line repre-sents the border between the dorsal (pallial) and ventral (subpallial) areas of the fish telencephalon. The asterisk denotes the sulcus externus, whose depth is a sign of the degree of eversion of actinopterygian telencephalon. Sensory projections to the brain are in green; the projection of the Dm to the hypothalamus is in black; innervations of pituitary corticotropes are in blue. Putative connections between the aforemen-tioned structures are in dash-and-dot lines and marked with a question mark. The neurotransmitter is in bold. Dd: dorsal part of the dorsal telencephalic area (dorsal pallium); Dl: lateral part of the dorsal telencephalic area (lateral pallium); inf: infun-dibulum; psb: pallial-subpallial boundary; V: ventral area of the telencephalon (sub-pallium); III: third ventricle. The figure was drawn with Inkscape 0.92, www.inkscape.org.

43

This results in a two-solid-lobe telencephalon with a T-shaped ventricle, which extends over its surface (Nieuwenhuys, 2011). The eversion theory is supported by the presence of ependymal cells lining the surface of the ac-tinopterygian telencephalon and by an external sulcus (see Figure 9) (Broglio et al., 2011; Nieuwenhuys, 2011). The depth of the sulcus is a sign of the eversion degree of the telencephalon (Nieuwenhuys, 2011).

The medial part of the dorsal telencephalic area participates in learning processes when dangerous or aversive stimuli are present in the environment (Salas et al., 2006; Broglio et al., 2011). Afferents to this area are neurons of sensory systems that indirectly project to various thalamic nuclei, and from here to different divisions of the telencephalic pallium, including the medial part (Figure 9). In turn, this area sends efferent connections to the anterior tuberal nucleus of the hypothalamus and to the hypothalamic inferior lobes (Meek and Nieuwenhuys, 1998). The POA is another source of afferents to the tuberal hypothalamus and the periventricular nuclei; moreover, it directly stimulates corticotropes in the pituitary by CRH release (Figure 9) (Meek and Nieuwenhuys, 1998; Winberg et al., 2016; Gorissen and Flik, 2016). In mammals, the periPVN receives inhibitory inputs from the medial amygdala and sends GABAergic projections to the PVN proper, thus modulating its activity (Herman et al., 2016). In teleosts, the hypothalamus itself projects back to the medial pallium, as also observed in mammals between the hypo-thalamus and the amygdala (see above) (Meek and Nieuwenhuys, 1998; Gray, 2005a).

The HPA axis is cyclically activated during the course of a day such that pulses of CRH are released by parvocellular neurons in the PVN (De Kloet et al., 1998; Herman et al., 2016). Therefore, a basal activation of the HPA axis is there even in the absence of a threat. The glucocorticoids secreted by the adrenal gland are essential to maintain the energy levels of an organism and meet its basal metabolic demands. In the short term, they stimulate the mobilization of energy sources by liver and fat tissue cells (Barrett, 2012; Herman et al., 2016). They also promote protein breakdown in the muscles and gluconeogenesis in the liver (Barrett, 2012). The effects of glucocorti-coids on target tissues are mediated by the cytoplasmic mineralocorticoid and glucocorticoid receptor (MR and GR, respectively) (De Kloet et al., 1998). In the brain, there is a differential distribution of MR and GR. The former is mostly expressed in the hippocampus, whereas the latter is ubiqui-tously produced, but it is most abundant in PVN parvocellular neurons and pituitary corticotropes (De Kloet et al., 1998). MR and GR are also produced in the amygdala; interestingly, the former is more abundant in the cytoplasm compared to the latter (Han et al., 2014). Amygdalar GR is abundant in the basolateral group and in CeA and M of the centromedial group (see Figure 8) (Herman et al., 2005; Wang et al., 2014). At rest, monomeric MR and GR are located in the cytoplasm and interact with other proteins to form multi-meric complexes (De Kloet et al., 1998). Glucocorticoid binding promotes

44

the dissociation of the complex, the homo- or heterodimerization of the re-ceptor and its translocation into the nucleus (De Kloet et al., 1998; Han et al., 2014). Here, activated receptors modulate gene expression, thereby af-fecting target cells at the transcriptional level (De Kloet et al., 1998). The majority of glucocorticoids circulate in the blood stream bound to the corti-costeroid-binding globulin (CBG) and, to a smaller extent, to serum albumin (De Kloet et al., 1998; Barrett, 2012). Only a small amount of glucocorti-coids are readily available for receptor binding. MR and GR have different affinities for the ligand; the former is activated at low-glucocorticoid levels, the latter requires higher hormonal concentrations for substrate binding (De Kloet et al., 1998; Herman et al., 2005). Therefore, MRs are active during basal HPA axis activity, whereas GRs come into play when a threat-driven activation of the axis is experienced by an animal (De Kloet et al., 1998). This condition has largely been documented in mammals, but it is also true for fish (De Kloet et al., 1998; Winberg et al., 2016). The hippocampus promotes excitation of the BNST and POA, and thereby stimulates transsyn-aptic inhibition of CRH release via GABAergic projections onto PVN par-vocellular neurons (Herman et al., 2016). When the activity of the HPA axis is higher compared to basal metabolic needs, glucocorticoids level rise such that CBG and serum albumin become saturated (Herman et al., 2016). Therefore, the amount of glucocorticoids available for GRs binding is higher (De Kloet et al., 1998; Herman et al., 2016). At high concentrations, gluco-corticoids directly inhibit PVN neurons and pituitary corticotropes, thus de-creasing the release of CRH and ACTH by those regions (De Kloet et al., 1998). The activity of the limbic system decreases upon GRs activation alike; reduced hippocampal tone onto PVN afferents results in enhanced ACTH release (Figure 10) (De Kloet et al., 1998). On the other hand, re-duced amygdalar output to the BNST and POA inhibits the activity of pitui-tary corticotropes (Figure 10) (De Kloet et al., 1998). Therefore, glucocorti-coids act as modulators of intrinsic circuit neurotransmitters (De Kloet et al., 1998).

Naturally occurring physiological changes cause daily cycles of activation of the stress axis. Interactions between an animal and its environment intro-duce further fluctuations, which result in adaptive responses. Those involve crosstalk between body and brain, and also among brain structures. Defense response circuits are controlled at several different levels on both an anatom-ical and time-scale basis. The presence of defensive survival circuits ensures that animals are able to integrate sensory information and to produce an ap-propriate behavioral output to ultimately strive for surviving.

45

Figure 10 GR-mediated glucocorticoid action on limbic system structures. A De-creased hippocampal excitatory tone results in less stimulation of BNST and POA, determining less inhibition onto the PVN (thin violet projections). PVN neurons thus release more CRH into the hypothalamic-hypophyseal portal vessel system (thick blue projections). B Reduced amygdalar GABAergic output onto the BNST and the POA (thin black and red connections) increases GABAergic tone of those structures (thick black and red connections), inhibiting CRH-releasing neurons in the PVN (thin blue projections). The hippocampus (A) projects to the hypothalamus through the fornix (fx) (Gray, 2005a). Here, the hippocampus and its projections are drawn following Fig. 4 of Herman et al. (2016) and Gray (2005a). The picture in A does not take into account PVN and pituitary corticotropes inhibition by GR present in those structures. CA3: cornu Ammonis 3; DG: dentate gyrus; vSUB: ventral subicu-lum. For the other abbreviations see Figure 1. The figure was drawn with Inkscape v0.92, www.inkscape.org.

46

GABAergic transmission in defensive survival circuits GADs The presence of GADs is a marker for GABAergic neurons, and the activity and distribution of those enzymes in HPA axis-regulating centers has exten-sively been investigated (Tappaz et al., 1977; Okamura et al., 1990). The hypothalamus has higher GAD activity compared to the whole brain, thus witnessing the importance of GABA as a neurotransmitter in this brain re-gion (Okamura et al., 1990; Sheikh et al., 1999). The distribution of GADs, both in terms of transcript and protein, suggests different rates of GABA production across diencephalic nuclei. The medial POA, the AHA and the dorsomedial hypothalamic nucleus (DMH) of the rat have the highest GAD activity among diencephalic structures (Tappaz et al., 1977; Okamura et al., 1990). Further compartmentalization of GADs can be observed in those three regions, which all exert control over the activity of PVN neurons (see Figures 8, 10; Okamura et al., 1990; Herman et al., 2016). The BNST also presents staining for GAD transcripts, especially in its encapsulated part (Okamura et al., 1990). On the other hand, the median eminence of the neu-rohypophysis was lowest regarding enzyme activity and content (Tappaz et al., 1977; Okamura et al., 1990). GAD67 and GAD65 are both produced in the rat brain; hypothalamic nuclei synthesize five-fold more of the latter isoform compared to the former (μg/mg tissue protein; Sheikh et al., 1999). Okamura et al. (1990) reported the difficulty in mapping GADs distribution in the cell bodies of neurons, as they observed a vast predominance of axon terminal labeling for those proteins. GAD65 is exclusively found at nerve endings, whereas GAD67 resides both in the cytoplasm and in association with synap-tic vesicles (see “The γ-aminobutyric acid signaling system, Anabolism”; Kaufman et al., 1991; Wojcik et al., 2006). A higher amount of GAD65 and, in general, a greater proportion of GADs located at axon terminals would confirm that the majority of the GABA produced by the hypothalamus is for synaptic transmission. Tappaz et al. (1977) also mapped the distribution of GABA-containing cells in rat hypothalamic centers, and found no compart-mentalization as that observed for GADs. Moreover, while GADs immuno-reactivity was specific of neurons, glial cells contributed to the GABA-stained cell population. Possibly, their results reflected the activity of the GATs, which are present both at axon terminals and in astrocytes in the proximity of GABAergic synapses (see “The γ-aminobutyric acid signaling system, GABA turnover, Catabolism”). Astrocytic GAT-1 and GAT-3 also intercept GABA in the extrasynaptic space (Conti et al., 1998). The distribu-tion pattern of GADs and GABA in hypothalamic nuclei may reflect the GABA shunt, i. e. the link between GABA production for synaptic release and carbon atom recycling for ATP production (see “The γ-aminobutyric acid signaling system, GABA turnover, GABA and energy metabolism” and Figure 5). This line of thought would imply that not only are there cellular

47

regulation mechanisms of GABA metabolism, but also crosstalk among neurons and astrocytes to sense the general energy level of the tissue. The amygdala is also involved in regulating the pulsatile CRH release by par-vocellular PVN neurons by sending GABAergic projections to hypothalamic PVN afferents (see “Defensive survival circuits, Anatomical considerations” and Figures 8, 10). In this brain area, a greater amount of GAD65 is found compared to GAD67, as observed for the hypothalamus (Sheikh et al., 1999). However, the fold-difference in terms of protein amount was smaller; the amygdala also produces less absolute amounts of GAD (Sheikh et al., 1999). In the mammalian brain, stress differentially regulates the expression of the two GAD isoforms, depending on the stress modality (acute vs. chronic) and the PVN afferent considered (Bowers et al., 1998). Specifically, acute stress experiences promote the transcription of the GAD67 gene in the POA and BNST. On the other hand, time-prolonged stressful conditions cause an in-crease in the GAD65 mRNA level in those regions, as in the periPVN alike (Bowers et al., 1998). The explanation of this difference may rely on the different catalytic activity of the two isoforms, as well as on the connection between GABA turnover and energy metabolism. In the beginning of HPA axis stimulation, it is conceivable that synaptic transmission is increased compared to rest. Therefore, higher amounts of GABA would be needed to sustain cellular communication. GAD67 can be both localized in the cyto-plasm and at the axon terminal, where it synthesizes GABA for metabolic purposes and synaptic transmission, respectively (see “The γ-aminobutyric acid signaling system, GABA turnover, Anabolism”; Wojcik et al., 2006). GAD67 does not catalyze autoinactivation as GAD65 does, and it would thus be more efficient in producing the GABA for short-term usage. On the other hand, GAD65 associates with synaptic membranes and, in this case, all the neurotransmitter produced is packed into the vesicles (Kaufman et al., 1991; Wojcik et al., 2006 Buddhala et al., 2009). During continuous and unpre-dictable HPA axis stimulation, higher amounts of GAD65 could ensure high-er rates of GABA production for synaptic relaying in the POA, BNST, and periPVN (Bowers et al., 1998). Possibly, an increased number of GABA-containing synaptic vesicles is also present in the axonal cytoplasm of PVN afferent in animals experiencing chronic stress.

GADs are present both in the telencephalon and throughout the hypothal-amus of teleosts, with GAD65 and GAD67 isoforms differentially distributed among various areas (Anglade et al., 1990; Martyniuk et al., 2007; Maruska et al., 2017). In Astatotilapia burtoni, GAD67 is almost exclusively produced in the ventral telencephalon (Maruska et al., 2017). GAD65 is abundantly found in the subpallium, but it also has a scattered distribution in the dorsal telencephalon. Martyniuk et al. (2007) reported higher amounts of GAD immunoreactive neurons in the subpallium of the goldfish brain (Carassius auratus), compared to the pallium. However, the difference in the number of GAD-positive cells was not as marked as the one noted for A. burtoni re-

48

garding those two areas (Maruska et al., 2017). Morevoer, the authors did not document differences between the distribution of GAD65 and GAD67 (Martyniuk et al., 2007). Further GAD production is found in the POA and in hypothalamic nuclei of both C. auratus and A. burtoni (Martyniuk et al., 2007; Maruska et al., 2017). In the latter structures, differences in GAD65 and GAD67 levels are documented in both species, with the former isoform more abundant compared to the latter (Martyniuk et al., 2007; Maruska et al., 2017). The hypothalamic nuclei include the nlt and the nucleus of the lateral and posterior recesses (nrl and nrp, respectively). As for C. auratus and A. burtoni, the rainbow trout (Oncorhynchus mykiss) has two cloned GAD isoforms (Anglade et al., 1990). The transcript of GAD65 is distributed in a scattered way in the Dm, but it is abundant in the POA and in the nrl. In this species, the immunoreactivity of GAD65 is exclusively restricted to axon terminals, mirroring the pattern in mammals (Anglade et al., 1990). The pituitary is devoid of GAD65 expression; in mammals, the median eminence in the neurohypophysis has the lowest activity of GADs (Tappaz et al., 1977). In both teleosts and mammals, GABAergic neurons play a pivotal role in the regulation of pituitary corticotropes. Their strategic location around the third ventricle confirms the central role of this area in the modu-lation of the activity of the HPA or HPI axis (Papez, 1937).

GABAA receptors GABAA receptor subunits are widely distributed in defensive survival cir-cuits, where they form both synaptic and extrasynaptic receptors. In the mammalian hippocampus, the transcripts for all α subunits apart from α6 are detected (Wisden et al., 1992; Hörtnagl et al., 2013). α2, α4, and α5 are the monomers expressed to the highest extent; the mRNA production depends on the hippocampal area considered. Specifically, the α2 and α5 subunits are abundanlty transcribed in all CA regions, whereas α4 is found the most in the dentate gyrus (Wisden et al., 1992; Hörtnagl et al., 2013). The α5 subunit is predominantly extrasynaptic and associates with β3 and γ2, generating recep-tors with a high affinity for GABA and slow desensitization rates (Wisden et al., 1992; Hörtnagl et al., 2013; Caraiscos et al., 2004; Farrant and Nusser, 2005). Extrasynaptic receptors intercept low levels of GABA that spills over from the synaptic cleft, and their activity partly depends on the GABA up-take rate by GATs (see “The γ-aminobutyric acid signaling system, GABA turnover, Catabolism”). The ventral subiculum, which indirectly influences hypothalamic PVN neurons firing via projections to the POA and BNST (see Figure 10), abundanlty produces the α1, α2, α3 monomers, all the βs and the γ2 (Hörtnagl et al., 2013). Those receptors subunits usually confer a synaptic location to the pentamer and mediate phasic inhibition (Farrant and Nusser, 2005; Hörtnagl et al., 2013). Synaptic α1β2γ2 or α1β3γ2 GABAA receptors are also found in CA pyramidal cells, CA1 interneurons and granule cells of the dentate gyrus (Seeburg et al., 1990; Wisden et al., 1992; Somogyi et al.,

49

1996; Hörtnagl et al., 2013). In the latter structure, the α4 subunit is abun-danlty transcribed alike, and it exclusively localizes in the perisynaptic space with δ (Wisden et al., 1992; Wei et al., 2003; Farrant and Nusser, 2005; Hörtnagl et al., 2013).

GABAA receptors are produced throughout amygdalar nuclei (Bowery et al., 1987). In the mouse CeA, the most abundanlty produced subunits are α2-

5, β1, β3 and γ1 (see Figure 9; Marowsky et al., 2004, 2012; Hörtnagl et al., 2013). Additionaly, the murine medial amygdala produces α1, β2, γ2 (Hörtnagl et al., 2013). The subunit composition of GABAA receptors dictate their location in the plasma membrane; therefore, it is possible to infer that CeA-GABAA channels with the α4 or α5 subunit be located peri- or extra-synaptically. Moreover, it is possible that α4 forms binary receptors com-posed of only α and β monomers (Olsen and Sieghart, 2008). On the other hand, α1β2γ2 is the most common synaptic GABAA receptor combination, and mediates fast, phasic inhibition (Farrant and Nusser, 2005). Species-specific differences are present in the expression of GABAA monomers in the CeA. In the rat, this structure highly expresses the α2 subunit, all the βs, γ1 and γ2 (Wisden et al., 1992). Moreover, currents mediated by putative ρ-containing GABAA receptors have been documented for this species (Sah et al., 2003). The murine POA, which potently modulates PVN activity, depicts production of all α subunits apart from α6, β1-3, γ1 and γ2; δ is detected at the mRNA level, but not at the protein one (Hörtnagl et al., 2013). CRH-producing PVN neurons abundanlty transcribe α2, β1, β3 and, to a smaller degree, α1 and β2 (Cullinan, 2000). The monomer γ2 is also found throughout PVN neurons (Hörtnagl et al., 2013). Those subunits usually confer a synap-tic location to the GABAA receptor, and a possible subunit combination in parvocellular PVN cells is α2β3γ2 (Farrant and Nusser, 2005; Hörtnagl et al., 2013).

The effect of GABA on target cells depends on the distribution of Cl– across the plasma membrane; in mature neurons, it is primarily dictated by the K+:Cl– symporter KCC2 (Payne et al., 1996; Rivera et al., 1999). This protein is a twelve-helices membrane spanning polypeptide with intracellular N- and C-termini and four sites of N-glycosylation between helices 5-6 (Payne et al., 1996). The active glycosylated, ion pumping form of the pro-tein is probably a multimer; each monomer has a molecular mass of 123.6 kDa (Payne et al., 1996; Rivera et al., 1999; Hewitt et al., 2009). KCC2 mediates the secondary active transport of Cl– by coupling it to the thermo-dynamically favored K+ extrusion from the cytoplasm to the extracellular fluid (Payne et al., 2003). In the mammalian brain, the production of KCC2 is silenced during the first days of extrauterine life, when GABAergic synap-ses are excitatory (Cherubini et al., 1999; Rivera et al., 1999). Several other factors influence the distribution of chloride across plasma membranes, as the synthesis and activity of HCO3

–:Cl– pumps. Those proteins also exert an effect on GABAA receptor conductance, as the channels are permeable to Cl–

50

as well as to HCO3– (Bormann et al., 1987). The contribution of HCO3

– to the net effect of GABA on target cells may be relevant in some physiological conditions (Bormann et al., 1987; Kaila and Voipio, 1987; Olsen and Sieghart, 2008). Additionaly, the expression of KCC2 in the mammalian neonatal brain has a region-specific pattern, and differences between species are documented (Rivera et al., 1999). In rats, the transcript for KCC2 is de-tected in the thalamus already at birth, but it is absent in other areas such as the hippocampus (Rivera et al., 1999). Possibly, different brain regions have a different pace of development, and the gene expression patterns are a mir-ror of such time diversities. Silencing or activating the KCC2 gene in a re-gion-specific fashion may underlie excitatory or inhibitory functions of GABA in the developing brain. Implicit to this reasoning is that the rest of the protein battery necessary to establish ion gradients is expressed at the same level. The excitatory function of GABA might be inversely correlated to the development of glutamatergic synapses; the production of KCC2 pos-sibly times with the full development of Glu as main excitatory neurotrans-mitter (Cherubini et al., 1991).

For the stress response to initiate, it is necessary that GABA-mediated in-hibition on pituitary corticotropes afferents be removed (see Figures 8-10; Hewitt et al., 2009). This can be achieved by a decrease of GABAergic tone onto target regions by the BNST and POA (Figure 8). The action of cortico-steroids on hippocampal MR receptors partly mediates this phenomenon (see discussion above and Figure 3; De Kloet et al., 1998). Down regulating the activity of KCC2 in parvocellular PVN neurons is an additional strategy to reduce GABA-induced hyperpolarization and therefore remove GABA inhi-bition on corticotropes (Hewitt et al., 2009). Excitatory projections to the PVN arise from the brain stem and employ amines as neurotransmitters (Herman et al., 2005; Herman et al., 2016). Those fast, monosynaptic con-tacts indirectly regulate the activity of PVN KCC2 by stimulation of α1 ad-renergic receptors (Hewitt et al., 2009). Under such conditions, the pumping activity of KCC2 is hindered and a depolarizing shift in the GABA equilib-rium potential (EGABA) is measured in parvocellular PVN neurons (Hewitt et al., 2009). The cells become thus more prone to allow the outflow of Cl–, rendering GABA a depolarizing neurotransmitter (Hewitt et al., 2009). This mechanism may be pivotal in the onset of the stress response, as activation of parvocellular PVN neurons is essential for ACTH release by the pituitary (see Figure 8; Hewitt et al., 2009). Moreover, it may create sort of a stressor memory in the cells, where the memory component is the more positive EGABA resulting from adrenergic stimulation (Herman et al., 2005; Hewitt et al., 2009; Herman et al., 2016). This shift may be the memo stick of a re-cently experienced stress, ensuring a faster activation of parvocellular PVN neurons and therefore a faster response to danger (Hewitt et al., 2009). Sev-eral other mechanisms are involved in the regulation of the HPA or HPI axis and they act on short and long time-scales (see “Anatomical considera-

51

tions”). Moreover, basal HPA axis activity is present every day and ensures coping with circadian rhythms (see “Anatomical considerations”). In condi-tions of chronic stress, higher levels of freely circulating corticosteroids are present in the blood stream. In fact, the HPA axis is more frequently activat-ed compared to a basal situation, and the adrenal glands are hypertrophic (Verkuyl et al., 2004; Herman et al., 2016). Therefore, the lower-affinity GRs are activated, as there is more corticosteroid readily available for recep-tor binding (see “Anatomical considerations”; De Kloet et al., 1998). Corti-costeroid-driven reduction of tonic GABAergic tone, but no of phasic one, is observed in the lateral amygdala (Liu et al., 2014). GABAA receptor subu-nits confer specific plasma membrane locations to mature receptors, and influence their physical-chemical properties (see “The γ-aminobutyric acid signaling system, Receptors, GABAA receptors”). Possibly, GR-mediated loss of tonic inhibition depends on selective silencing of genes encoding for extrasynaptic GABAA receptor monomers (Liu et al., 2014). The activity of corticosteroids on target cells is explicated on the modulation of the intrinsic circuit transmission systems (De Kloet et al., 1998). In parvocellular PVN neurons of rats, chronic stress alters the expression of the extrasynaptic α5 and δ monomers (Verkuyl et al., 2004). Specifically, the transcript of the former subunit is more abundant, whereas there is a reduction in δ mRNA (Verkuyl et al., 2004). A general loss of GABAergic tone is documented in PVN afferents, possibly because chronic activation of the HPA axis causes re-shaping of neurons and reduced number of GABAergic synapses (De Kloet et al., 1998; Verkuyl et al., 2004).

Neurosteroids affect GABAergic transmission by allosterically modulat-ing the activity of GABAA receptors (Belelli and Lambert, 2005). Those compounds can directly be produced in the brain by neurons and glial cells (Belelli and Lambert, 2005). Alternatively, peripherally synthesized steroid hormones can cross the blood-brain barrier and enter brain cells. Here, they may activate specific nuclear receptors, or be metabolized to neurosteroids (Belelli and Lambert, 2005). Those substances act as both agonists and an-tagonists on GABAA receptors, depending on their chemical structure (Wang, 2011). The subunit composition of the channels also determines the neurosteroid-driven degree of allosteric modulation (Wohlfarth et al., 2002; Belelli and Lambert, 2005). The neurosteroid binding pocket is located at the interface between the principal and complementary faces of β3 and one of the α subunits, respectively (see “The γ-aminobutyric acid signaling system, Receptors, GABAA receptors”; Miller et al., 2017). In the latter ones, Gln245 in M1 is essential for engaging a hydrogen bond with the 3α-hydroxyl group of the brain steroid (Miller et al., 2017). Moreover, Trp249 forms stacking interactions with the B and C rings of the hydrocarbon core (Miller et al., 2017). Leu297 in M3 of the β3 principal face also determines the potentiating effect of neurosteroids on GABAA receptor-mediated cur-rents (Miller et al., 2017). The brain-steroid-binding site is placed closer to

52

the cytoplasmic portion of the TMD, than to the ECD region (Miller et al., 2017). Moreover, it is partially exposed to the lipid environment of the membrane (Miller et al., 2017). It is tempting to speculate that the mem-brane, and especially cholesterol, might represent a site of regulation of GABAA receptors. Different brain steroids selectively bind to specific GABAA receptors, depending on their subunit combination (Belelli and Lambert, 2005). In the hippocampal dentate gyrus (see Figure 10A), GABA-mediated tonic currents are enhanced by the 5α-pregnan-3α,21-diol-20-one (3α,5α-THDOC), synthesized from the adrenal-produced deoxycorti-costerone (Belelli and Lambert, 2005; Wang, 2011; Barrett, 2012). In this brain area, GABAA receptors containing the δ subunit are present (see dis-cussion above; Wisden et al., 1992; Wei et al., 2003; Hörtnagl et al., 2013). This monomer does not contribute to the neurosteroid binding site; rather, it potentiates the current when the allosteric modulator is bound between the principal and complementary face of β and α (Wohlfarth et al., 2002; Wang, 2011). Neurosteroids also exert effect on ternary γ2-containing receptors, yet to a smaller degree compared to channels with δ (Wohlfarth et al., 2002). The subtype of α subunit contributes in shaping the brain-steroid-driven modulation of GABAA receptors (Wohlfarth et al., 2002). Receptors with the α6 monomer detect lower amounts of 3α,5α-THDOC compared to α1-containing channels. The combined effect of GABA and 3α,5α-THDOC does not differ between receptors with γ2 or δ, when α6 is present (Wohlfarth et al., 2002). Probably, the neurosteroid-mediated modulation of GABAA receptors with α6 is strong enough that the contribution of δ weighs less, compared to α1-containing channels. In the CA regions of the hippocampus (see Figure 10A), 3α,5α-THDOC fails to exert an effect on extrasynaptic GABAA receptor currents, but other neurosteroids do influence the activity of GABAA channels (Belelli and Lambert, 2005). Possibly, the α5 subunit, abundantly found in those brain areas, dictates the preference for brain ster-oids other than 3α,5α-THDOC (Caraiscos et al., 2004; Belelli and Lambert, 2005). The effect of a brain steroid on GABAA receptors is also determined by the concentration of the neurotransmitter and the steroid itself (Wohlfarth et al., 2002). The synthesis of neurosteroids is affected by several factors, including the levels of GHBA. This compound is produced in the mitochon-drion and can be an indirect sign of increased [GABA] in the matrix (see Figure 3A, C; Passarella et al., 1984). Possibly, a remote link between GABA turnover and neurosteroid modulation is present in neurons, all to ensure the energy balance of the cell. Increased GABA flow into the mito-chondrial matrix may mirror a situation of neuronal resting, or end of synap-tic transmission (see “The γ-aminobutyric acid signaling system, GABA turnover, Catabolism”). In that cellular compartment, such neurotransmitter can be either converted into SSAL, or into GHBA (see Figure 3). In the lat-ter case, and increase in [GHBA] may modulate the activity of neurosteroid-synthesizing enzymes in the mitochondrial matrix (Belelli and Lambert,

53

2005). In fact, GHBA can influence the production of brain steroids (Belelli and Lambert, 2005). In turn, neurosteroid modulation of presynaptic GABAA receptors might contribute to switching off the firing activity of the neuron. All these considerations would point towards a tight integration of GABA production and transmission systems to ultimately save the precious energy needed for life to go on.

Acid sensing in defensive survival circuits In a situation of real or perceived danger, synaptic transmission is initially enhanced compared to a resting situation. This implies that a higher number of synaptic vesicles be fused with the axonal membrane, and the synaptic cleft acidifies faster than during basal neuronal communication. In fact, the content of synaptic vesicles is acidic (see “The proton signaling system, Pro-tons and synaptic transmission”; Miesenböck et al., 1998). If neurons are more synaptically active, it is also possible that their metabolic demands be higher compared to rest. They would inevitably face more frequent changes in ion gradients distribution across membranes, and need more ATP to re-store original ion concentrations. A higher aerobic metabolic rate leads to higher amounts of mitochondrially produced CO2, which originates in the matrix from the pyruvate dehydrogenase complex, isocitrate dehydrogenase, and α-ketoglutarate dehydrogenase complex (see “The proton signaling sys-tem, Protons and synaptic transmission”; Berg et al., 2012a; Mookerjee et al., 2015). CO2 freely diffuses across biological membranes; higher rates of CO2 production likely result in higher amount of this species both in the cytoplasm and in the extracellular environment. Thus, CAs in both com-partments would be pushed to form H2CO3. (see “The proton signaling sys-tem, Protons and synaptic transmission”). As recalled above, the pKa of this acid is 3.5, which implies that it is mostly dissociated in HCO3

– + H+ at the neutral pH of 7.1 (see the Henderson-Hasselbach equation; Ruusuvuori and Kaila, 2014). In this situation the overall buffering system, which is main-tained by CO2-driven crosstalks between cytoplasmic and interstitial CAs, is hindered. The direct consequence of such a scenario is the acidification of both the intra- and extracellular spaces, a quite dangerous circumstance for tissue integrity. Several studies on the arousal of fear-related behaviors link CO2 increase with fear manifestations (Ziemann et al., 2009; see Wemmie, 2011 for review). The possible biochemical reason beyond CO2-induced fear is the aforementioned, consequent brain acidosis (Wemmie, 2011). Increased lactate also induces fear behavior, and the reason is possibly the same that links CO2 to panic. In fact, lactate movement across plasma membrane en-visages the symport of a proton per monocarboxylate unit. This transport causes the acidification of the compartment to which lactate is being moved (Erlichman et al., 2008).

54

The amygdaloid bodies receive highly processed sensory information that is in turn relayed to other integrating structures (see “Anatomical considera-tions”; Aggleton and Mishkin, 1986). The feeling of fear might not be the product of the sole amygdalas, and the topic of feelings and emotions is dif-ficult to assess in animals other than humans (see “Anatomical considera-tions”; LeDoux and Brown, 2017). However, it is quite safe to assume that animals are capable of sensing danger and, even if they might not conscious-ly be aware of it, they usually produce behavioral outputs to maximize the survival odds. As recalled above, a higher alert status of the body, compared to rest, produces higher synaptic transmission, at least in the beginning of such a state (Wemmie, 2011). In turn, this leads to higher metabolic de-mands by neurons (see “The proton signaling system, Protons and synaptic transmission”). Therefore, dangerous situations may be associated with brain acidification, and the sole CO2 may already be a signal of danger (Ziemann et al., 2009). It is likely to infer that every brain region, and not just those constituting the defensive survival circuits, experiences enhanced neuro-transmitter release, higher rates of CO2 production, and the consequent inter-stitial acidosis. This is indeed the case; however, a high amount of ASIC1 is expressed throughout the defensive survival circuits, and especially in the amygdala (Wemmie et al., 2003; Ziemann et al., 2009). This reasoning im-plies that a high density of proton receptor confers the amygdala a higher sensitivity to pH falls. The ASIC1 gene is pivotal for fear conditioning, as demonstrated by experiments with knockout mice, or by enhanced expres-sion of that gene (Wemmie et al., 2003; Wemmie et al., 2004). In this type of fear manifestations, animals are exposed to a conditioned stimulus at the end of which a brief, unconditioned stimulus occurs (LeDoux, 2000; Salas et al., 2006). Animals learn to associate the former, innocuous input to the lat-ter one, which is noxious. Therefore, exposing trained animals to the condi-tioned stimulus is sufficient to trigger the fear manifestation, and ASIC1 plays a major role in this (Wemmie et al., 2003). Interestingly, the sole ex-posure to CO2 is sufficient to trigger fear behavior (Ziemann et al., 2009). Possibly, animals learn to associate that a higher CO2 production corre-sponds to a threatening situation. In this case, CO2 itself would be the condi-tioned stimulus. More generally, experimental evidence points towards a role of ASIC1 in memory establishment; fear conditioning is just a particular type of memory, and animals learn through association. Two gross types of memories can be identified, i. e. that associated with emotional information, and that linked to spatial, relational and temporal cues (Salas et al., 2006). The former is supported by the amygdala, whereas the latter is sustained by the hippocampus. ASIC1 is found in both structures; in the amygdaloid body, this gene is abundantly expressed in the LA and B of the basolateral group (see “Anatomical considerations” and Figure 8; Sah et al., 2003; Wemmie et al., 2003; Coryell et al., 2007). Expression of ASIC1 is also detected in other amygdalar nuclei and in the BNST (Coryell et al., 2007).

55

The flow of information through the amygdala is directional, with the baso-lateral group as point of entrance and the central group as output unit (see “Anatomical considerations”; LeDoux, 2000; Sah et al., 2003). ASIC1 is present on structures that receive both excitatory and inhibitory inputs (Wemmie et al., 2002; Coryell et al., 2007). Knockouts for that gene respond less to fear conditioning and present deficits in spatial learning (Wemmie et al., 2002; Wemmie et al., 2003; Coryell et al., 2009). Moreover, ASIC1a seems to be essential in innate fear behaviors. Those manifestations are not learned by exposure to threats; rather, they are somehow already written in the brain circuits (Coryell et al., 2007). One of the behavioral outputs trig-gered by both conditioned and innate fear is freezing. By doing so, an animal might save energy in preparation for survival striving. Another possible, concomitant explanation is that the brain enters a stand-by state during freez-ing, when synaptic transmission is temporarily dampened. In this case, the brain has some time to restore the ionic concentrations across neuronal membranes. Most importantly, the pH can be brought back to physiological values, preventing acidosis-induced damages. The fast desensitization of ASICs might remotely be linked to this process (Krishtal et al., 1980; Waldmann et al., 1997; Wemmie et al., 2006). In the generation of fear be-havior, mechanisms other than H+-gated ASIC1-mediated currents might be involved. It is true that homotrimeric ASIC1a channels exist and, for the experimental evidence presented above, trigger defensive survival behaviors (Coryell et al., 2007). However, intensive training can restore learning defi-cits in ASIC1 knockouts, which present baseline fear as wild-type animals do (Wemmie et al., 2002; Wemmie et al., 2003). Maybe, proton receptors composed of other ASIC subunits might exist and play some role in learning processes. Indeed, ASIC2 and ASIC4 are expressed in defensive survival circuits; the splice variant ASIC2a generates weak currents when the extra-cellular pH is lowered to about 4 (Askwith et al., 2004; Hoshikawa et al., 2017). Speculations in this sense are quite tentative, as virtually no infor-mation is available on co-localization of ASIC subunits other than ASIC1 with ASIC2. An alternative possibility is that protons shape the response of postsynaptic cells by acting on allosteric sites of other receptors. This is the case of GABAA channels and NMDA Glu receptors, which possibly present proton antennas as the ASICs (see “The proton signaling system, Protons and synaptic transmission”; Sawyer and James, 1982; Pasternack et al., 1992; Sinning and Hübner, 2013). Thus, a structural analysis of closely lo-cated carboxyl-carboxylate residue pairs in GABAA and NMDA receptors may provide insights into their H+-sensing mechanism (Sawyer and James, 1982). Several questions remain open on the physiological relevance of pro-tons as neurotransmitters. Further work is required to elucidate the complete role of extracellular pH shifts in synaptic transmission.

56

Aims

This thesis is based on three works, whose aims are briefly outlined here:

• Paper I: To characterize the GABA signaling system in the adult zebrafish (Danio rerio) brain, grouped brain areas and retinas at the transcriptional level. This work shed light on an ancient and con-served signaling pathway that was previously poorly characterized in the model species.

• Paper II: To characterize the ζ subunit of the GABAA receptor in

terms of transcript abundance and distribution, and protein structure, in the adult zebrafish brain and retinas. This work established a solid methodological pipeline that strengthened the methods developed for Paper I.

• Paper III: To test whether stimulation of the zebrafish HPI axis re-

sulted in change of the expression level of key players in GABA and proton signaling systems. This work highlighted candidate genes for defensive system activation in the zebrafish and pointed towards stimulating future research directions.

57

Experimental procedures

The current section describes the methodological techniques learned, opti-mized and employed in the experiments during the course of the PhD train-ing. For brevity reasons many experimental details were omitted from the Papers; yet they are here presented.

Animals Zebrafish

Figure 11 A Photograph of a small school of zebrafish (D. rerio) in captivity condi-tions at the fish facility at the Uppsala Biomedical Center (BMC), Uppsala. B Adult zebrafish female, conceived on 7th March 2016, anaesthetized to depict the position of the ventral fins (pinnae ventrales) with respect to the pectoral (pinna pectoralis) ones. The zebrafish is a Piscis abdominalis (Linnaeus, 1788). The belly of the indi-vidual in this photograph is possibly full of eggs. The Figure was prepared with Inkscape 0.92.

The zebrafish (Figure 11) is the model organism upon which the present thesis was based. The species was first described by Francis Hamilton in 1822, during his stay in India. In his opera An account of the fishes of the river Ganges and its branches, the author documented the abundance of new ichthyic species in the aforementioned river basins. Among those was the “Cyprinus Rerio, Danio vittis cæruleis argenteisque nonnullis; corpore compressissimo; cirrhis 4, quorum 2 capite paulo longiores” (pg. 390; Ham-ilton, 1822). The author classified Cyprinus Rerio in the Cyprinus Danio

58

division, or VI. division, of the genus Cyprinus. Several traits of the zebrafish were already recognizable in the Cyprinus description by Linnaeus in his Systema Naturae, including the “Caput ore edentulo” and the “Pinnae ventrales saepe novemradiate” (pg. I409; Linnaeus, 1788b). The position of the ventral fins with respect to the pectoral ones was also used as a criterion to classify fish into orders (pg. I129; Linnaeus, 1788b). The Pisces ab-dominales, so termed because their “[pinnae ventrales sunt] pone pinnas pectorales” (pg. I129), comprise the Cyprinus genus. The location of the zebrafish ventral fins is that described by Linnaeus for the Pisces ab-dominales, as can be seen in Figure 2B and Plates 1(a, b) of Laale (1977). The number of rays of those fins, rather than being novem, “[is] only seven” (pg. 324; Hamilton, 1822).

In his opera, Hamilton denoted that locals in India referred to the fish of the VI. division of Cyprinus with the Bengalese word Dhani; hence the name Danio. The beautiful zebrafish, “with several blue and silver stripes on each side”, was found “in the Kosi river” (pg. 323; Hamilton, 1822). This stream originates in the Chinese slope of Himalaya, crosses Nepal, enters India and flows into the Ganges in the proximity of Bhangalpur. After the work of Hamilton, several expeditions collected zebrafish specimens in an area as vast as the drainage basin of the Ganges (McClure et al., 2006; reviewed in Engeszer et al., 2007). The zebrafish has been found in several regions in Northwest India, including low-elevation localities of the state of West Ben-gal, but also mountainous areas of the state of Meghalaya (Engeszer et al., 2007). Spence et al. (2006) documented the presence of D. rerio in Bangla-desh, especially in the central Mymesingh District, in the drainage basin of the Brahmaputra. Generally, this species had a preference for slow-flow water bodies. The natural habitats of the zebrafish spanned a broad range of temperature, pH, transparency of the water and bottom composition, being very heterogeneous in both studies (Spence et al., 2006; Engeszer et al., 2007). Thus, this species depicts a remarkable natural resilience, a feature that makes D. rerio suited for sustaining captivity conditions. Previous re-ports of collection of zebrafish in areas close to Pakistan and Myanmar have been questioned (Engeszer et al., 2007; Spence et al., 2011). Nonetheless, human activities may have reduced the original natural areal of distribution of this “beautiful fish” (pg. 323 of Hamilton, 1822; Engeszer et al., 2007).

The zebrafish became a model organism thanks to pioneering work initi-ated by researchers at the University of Oregon (see e. g. Streisinger et al., 1981; Grunwald and Eisen, 2002). The species had though been employed since the Thirties in fisheries’ research (reviewed by Laale, 1977). Overall, D. rerio is easy to handle, robust, and reproduces well also in laboratory conditions (direct experience of the candidate; Streisinger et al., 1981). Moreover, its genome has completely been sequenced (http://www.sanger.ac.uk/science/data/zebrafish-genome-project; Grunwald and Eisen, 2000; Howe et al., 2013), thus making this species suitable for

59

molecular work. As noted before, the resilience of the zebrafish in the wild may explain the aforementioned traits (Grunwald and Eisen, 2002; Engeszer et al., 2007). Adult zebrafish are small; the average length of all the animals employed in the Papers of this thesis was 3.2 ± 0.39 cm. During the PhD training, the candidate acquired solid skills in breeding zebrafish and obtain-ing vial offspring. This was the result of a patient work of observation of the zebrafish embryos and larvae development, and the behavior of the small creatures during this period (Kimmel et al., 1995). The breeding procedure was carried out by placing a plastic grid covered with glass beads in a 1 L rectangular container, subsequently inserted in the tank with the adults. The grid is necessary to prevent the parents to eat their own fertilized eggs. In fact, zebrafish display cannibalism, both to eggs and, when adults, to their own conspecifics (observation of the candidate; Spence et al., 2006). The method described above was performed late in the afternoon, when the par-ents were also given extra food in prevision of the energy-costly task of mat-ing. The day after, the eggs were collected and carefully analyzed with the help of a stereomicroscope. Fertilized eggs were separated from unfertilized ones and impurities, rinsed and placed in a Petri dish in an incubator at 28 °C. Methylene blue, a compound with antimycotic activity, is recommended in several protocols concerning fish embryos’ handling (see e. g. https://zfin.org/zf_info/zfbook/chapt7/7.52.html; Brady et al., 2016). It was employed in the initial tries of zebrafish breeding by the candidate of this thesis. However, there was no difference in growth and development between embryos and larvae in the presence or absence of that compound. Therefore, it was not used in further attempts. The newly conceived zebrafish were incubated at 28 °C for four to five days in Petri dishes; it was noticed that the density of the embryos influenced their growth rate. In fact, Petri dishes with more creatures displayed an overall faster development of the embryos. Probably, those produced signaling molecules released in the water with mutual effects on each other. After this observation, the water in the Petri dish was filled a bit every day, rather than being completely changed. That procedure saved the precious compounds produced by the organisms. The zebrafish larvae were then placed in 3 L tank with steady water for fifteen days; a third of the volume was replaced two times. After that first developmental period, they were transferred to a water recirculating system. Larval zebrafish were fed with the appropriate food, which was dis-persed over the surface of the water to form a cloud. In that way, the small creatures could have easy access to it. This detailed procedure was used for obtaining the fish used in the reverse transcription-quantitative polymerase chain reaction (RT-qPCR; see “Gene expression profiling, Design and test of primer pairs”) in Papers I and II.

It is widely accepted that zebrafish become adults at around three months of age (see e. g. Streisinger et al., 1981). Laale linked mating capacity with the length of the fish, which he reported being achieved at the second and

60

half month of life (Laale, 1977). In one breeding attempt made during the PhD training, zebrafish actually spawned at two months and twenty-eight days of age. It is almost unquestionable that reaching sexual maturity de-pends on the diet. Therefore, the sooner a researcher wishes to have a new generation of zebrafish, the more the parents-to-be must be provided with food.

Three-spined stickleback The Gasterousteus aculeautus (Linnaeus, 1758) is a Piscis thoracicus, whose ventral fins are “sub pinnis pectoralibus (pg. I129; Linnaeus, 1788b; Figure 12; Fig. 1 in Wotton, 2009). In the words of the eminent Swedish academic, the three-spined stickleback, or storspigg in Swedish, has a “caput compressum, anterius declive; oculi prominuli, iride argentea; mandibulae aequales; branchiarum operculum magnum argenteum; […]; linea lateralis aspera, dorso propior; pinnae flavicantes” (pg. I324; Linnaeus, 1788b). All these features can well be observed in Figure 12; see also Fig. 1 in Wotton (2009). Sometimes, as noted by Linnaeus, the stickleback is a fish whose “gula et pectus [sunt] ruberrima”.

Figure 12 Photograph of an adult individual of three-spined stickleback (G. acule-atus) at the fish facility at BMC, Uppsala. The position of the ventral fin with re-spect to that of the pectoral fin is noted. The Figure was prepared with Inkscape 0.92.

This characteristic, together with a blue iris, is typical of dominant males in the reproductive season. The author of this thesis could directly observe those colors during a one-month fieldwork at the Sven Lovén Centre for Marine Infrastructure. This marine station is part of the University of Gothenburg and is located at the mouth of the Gullmarsfjord, on the West coast of Sweden. Three-spine sticklebacks caught in the wild during their reproductive period, “Aprili et Junio ad plantas aquaticas ova pariens pauca” (pg. I324; Linnaeus, 1788b), were housed at the Centre’s facility. The candidate could observe that one or two males among the whole school of fish displayed the mating colors. Moreover, when only one male had the characteristic coloration, the shades were stronger. When two males had red throat and chest and blue eyes, their tinges were fainter. It was highly likely that a strong competition for mating

61

established in the artificial environment. It was also inferred that hierarchies in the schools were not definitive, but could change every day. The natural habi-tats of the three-spined stickleback are freshwater basins of Europe, as de-scribed by Linnaeus (1788b). However, storspigg populations can also be anadromous or completely marine animals, found far from the coast and in a wide range of salinity and temperature conditions (direct observation of the candidate; Wootton, 1984; Wootton, 2009).

The three-spined stickleback was not used in the current thesis as model species. However, the RNA extraction procedure (see “Purification of nucle-ic acids, RNA extraction”) was optimized also with the use of samples of brain and gills collected during the fieldwork month at the Sven Lovén Cen-tre for Marine Infrastructure. The genome of G. aculeatus has been fully sequenced (https://www.broadinstitute.org/stickleback/stickleback-genome-project), as has been that of the zebrafish. During the course of the PhD training, primer pairs based on the three-spined stickleback cDNA sequences were developed and tested with the samples from the fieldwork collection. Those were then used for an experiment of gene expression quantification with the nine-spined stickleback (Pungitus pungitus Linnaeus, 1758), whose genome is not sequenced. This data, however, are not presented in this doc-toral thesis.

Purification of nucleic acids Sample quality is a key factor in obtaining good results. During the course of the PhD training all laboratory techniques aimed to purify RNA and DNA were optimized until reproducible, high-yield and high-quality-sample methods were reached.

RNA extraction The first attempts in RNA extraction were made by using silica-gel mem-brane columns and following the kit’s manufacturer’s instructions. The pro-cedure was tested with different samples, namely brains, brain areas and eyes of zebrafish, and brains of three-spined stickleback; kits from different suppliers were tried out. However, the overall results did not meet the de-sired expectations. The RNA yield was quite low and very different among samples; the procedure felt weak and good results seemed to be more a mat-ter of luck than a real mastery of the technique. There were suspects of co-purification of DNA with RNA by analyzing the results of primer pairs’ testing (Figure 13A, C; see also Figure 15 and its discussion in “Gene ex-pression profiling, Design and test of primer pairs”). Taken together, all the-se considerations lead to a thorough literature search to find a more robust way of extracting the RNA.

62

The first new method employed by the candidate was a standard phe-nol:chloroform extraction based on organic chemistry principles. In the ap-propriate medium, nucleic acids can be separated from each other and from proteins in a pH-dependent fashion (Chomczynski and Sacchi, 1987). Spe-cifically, a denaturing acidic medium promotes the partitioning of RNA in an upper, aqueous phase, whereas DNA will be located in a white interphase and proteins will partition in the lower, organic phase (technical bulletin of Sigma-Aldrich for product 77619; Chomczynski and Sacchi, 1987). The protocol by Rodrigo et al. (2002), which envisages the phase separation de-scribed above, was employed to purify RNA from samples of brain and gills of the three-spined stickleback. Following the guidelines of the authors, a treatment with DNase I was included, and a second phemol:chloroform ex-traction afterwards performed. The integrity of the RNA was checked on a gel and can be seen in Figure 13B. The standard ratios A260/A280 and A260/A230, employed to control for protein on one side, and guanidinium and ethanol presence in the RNA sample on the other, were above 2.0. This indi-cates a high purity of the template (Thermo Scientific T009-Technical Bulle-tin, NanoDrop 1000 & 8000), that could also be noted when those samples were used for primer test for three-spined stickleback (see “Animals, Three-spined stickleback”).

However, the phenol:chloroform extraction presented several critical points. First, the pipetting had to be very precise; if only a small amount of phenol was carried to the final steps, i. e. RNA precipitation and re-suspension, the whole preparation could not be further used. In fact, phenol impairs the measurement of RNA concentration, generating an overestimate of the total amount of nucleic acid. This was observed when the procedure by Rodrigo et al. (2002) was tested with adult zebrafish brains. Phenol is also a very toxic compound; therefore, the author of this thesis looked for another way of extracting the RNA. The solution to this laboratory issue came with the protocol by Eyester and Brannian (2009), which is a combina-tion of the phase separation method with TRIzol® Reagent and silica-gel membrane column purification. As for the standard phenol:chloroform ex-traction, the RNA was first divided from the DNA and proteins by pH-dependent phase separation. The aqueous phase was then transferred to RNA binding columns; by using TRIzol® Reagent, the white DNA interphase was very visible. The RNA was then purified following the kit’s manufacturer’s instructions, and the sample checked for integrity and purity. An example of RNA samples extracted with this latter protocol is provided by Figure 13C. The standard A260/A280 ratio was good for all samples, A260/A230 was more variable. The latter ratio apparently varied in a random way among samples and probably depended on the column itself. A trace of what was probably DNA was still visible as a very faint upper band (compare Figure 13C with 13A). The details of this procedure, as well as small modifications with re-spect to Eyster and Brannian (2009), are described in Paper I, Experi-

63

mental procedures, RNA extraction and cDNA synthesis. This method was robust, yielded good amounts of RNA and was employed with brain and eyes samples in Paper I and individual brain areas in Papers II and III.

Figure 13 A Example of RNA samples from zebrafish extracted and purified with silica-gel membrane columns following manufacturer’s instructions. Clear 28S and 18S ribosomal RNA bands were visible. However a bigger, upper band, possibly representing DNA, was detected in all samples. TM: tecta mesencephali; DC: dien-cephalon; OB+TC: olfactory bulbs + telencephalon; WB: whole brain. B Example of one RNA sample from the brain of three-spined stickleback purified with a standard phenol:chloroform extraction, then treated with DNase I (Rodrigo et al., 2002). Clear 28S and 18S rRNA bands were visible. The bigger band visible in A was ab-sent, therefore strengthening the evidence that silica-gel membrane columns would co-purify DNA together with RNA. C Example of RNA purification with a combi-nation of TRIzol® Reagent and silica-gel membrane columns. Clear rRNA bands were visible. A very thin upper band was also detectable, but it was negligible com-pared to that in A. Procedure from Eyster and Brannain (2009), with the modifica-tions explained in the text.

Plasmid purification The PCR products obtained after primer test for in situ hybridization (see “Gene expression profiling, Design and test of primer pairs”) were inserted

64

into pCRTMII-TOPO vectors (see invitrogen User Guide for the kit 45-0640 and details in Paper II, Experimental procedures, Gene expression anal-ysis by in situ hybridization). The bacteria were transformed and cultured overnight according to standard procedures. The plasmids were first purified according to the lithium miniprep protocol, then the alkaline lysis miniprep procedure was preferred (Engebrecht et al., 2001). The protocol was robust and yielded very concentrated and pure samples.

Gene expression profiling Gene expression profiling was carried out in all three studies presented in this thesis. The analysis of the mRNA presence, in terms of quantity by RT-qPCR (Bustin et al., 2009; Papers I, II, III) and anatomical distribution by in situ hybridization (Schaeren-Wiemers and Gerfin-Moser, 1993; Paper II), started with the design of primer pairs for the genes of interest. The next paragraph describes the procedure and methodological pipeline developed to obtain the primer pairs. The cDNA synthesis method is described in Paper I, Experimental procedures, RNA extraction and cDNA synthesis.

Design and test of primer pairs The procedure established to obtain primer pairs is depicted in Figure 14, and details of the method are provided in Papers I and II. The procedure was adapted case to case, according to the results. Briefly, primer pairs were de-signed with the help of a software or manually and blasted against the Standard Nucleotide BLAST of NCBI. For each primer, its position along the cDNA transcript was checked and noted, as well as its melting temperature. For each gene at least three primer pairs were tested; effort was spent to position the dif-ferent pairs in different regions of the transcript. To place a primer in an exon-exon junction region would theoretically ensure preference of the primer for the mRNA transcript and not DNA possibly present in the sample. However, it was direct experience of the candidate that primer pairs with one or both primers spanning exon-exon junctions were not always specific (Figure 15). Therefore, it was ensured that for each gene at least one primer pair spanned a junction and one did not, to compare the results. The pairs were then tested with RT+ and RT- qPCRs and accepted when the specific product, and only that one, was present in the wells. The melting temperature of the qPCR product was noted down; melting temperatures for the primer pairs used in the experimental work are in Papers I and III. Initially, no product was expected in the RT- reactions, and a primer pair discarded if any signal was detected in those wells. However, being the genes of interest more than forty, the criteria for the negative controls were smoothened. If a primer pair gave a product in the negative reactions, and that product was of different size compared to the specific one, the pair was accepted (example in Figure 16).

65

Yes (to both)

Primer pair design: GETPrime*, Primer-BLAST NCBI, manual

Check primers in each pair with Nucletide BLAST NCBI

Do the primers align to all splice vari-ants of a given gene? Are primers in one pair located in dif-ferent exons, or is one (or both) placed on an exon-exon junction?

No (to one or both)

Order primer pair

Test primer pair with RT+ and RT-** qPCR

Does the pair give one product in the RT+ wells, and is that of the predicted size? Does the pair give no product in the RT- wells***?

No (to first or both)

Accept primer pair

Yes (to first or both)

66

Figure 14 (pg. 66) Flux diagram depicting the gross workflow to develop the primer pairs presented in the Papers of this thesis. The scheme was flexible and adapted to the different primer pairs tested. * Reference for the software GETPrime is David et al. (2017). ** RT+ and -: reverse transcriptase positive and negative, respectively. See Paper I, Experimental procedures, Primer specificity test and reverse tran-scription-quantitative polymerase chain reaction (RT-qPCR) for technical de-tails regarding primer test. *** See text, Paper I and Paper III for discussion on this point.

Figure 15 Results of the primer test, in this example for gene slc12a1 (zfnkcc2). The first primer pair (1) does not span an exon-exon junction, is specific in the RT+ reactions and gives no product in the RT- ones. The first primer of the second pair (2) spans a junction and is unspecific. The third primer pair (3) is not exon-exon junction and it is unspecific. The primer pair selected from this test and used in Pa-per I and Paper III is the first one. The numbers after the sequential identification of the primer pair refers to the expcted size of the amplicon. Gene ruler: 50 bp.

In this case, gels were run; see Paper I and III for more details on this point. Paper I also presents the procedure for designing primer pairs for single splice variants (see Paper I, Experimental procedures, Splice variants of GABAA receptor subunit γγ2). For the synthesis of in situ hybridization probes, the primer pairs were tested with RT+ and RT- PCRs. The pipeline resembles that above described and is presented in detail in Paper II, Ex-perimental procedures, Gene expression analysis by in situ hybridiza-tion. As noted before, it was possible to infer that the standard RNA purifi-cation by silica-gel membrane columns does not discriminate between RNA and DNA. Therefore, this method leads to RNA samples that contain DNA (see “Purification of nucleic acids, RNA extraction” and Figure 13A). The primer pairs published in Paper I were all tested with this kind of samples, including those for slc12a1 presented in Figure 15. Here, it is possible to add an element pointing towards DNA presence in RNA samples. The forward and reverse primers of the third pair for slc12a1 are located on exons thirteen and fourteen, respectively (ENSDART00000166365.2 in Ensembl Genome Browser; last four lanes of Figure 15, marked with “3”). The intron between them is 380 base pairs; this, summed with the amplicon length of 97 base

67

pairs, gives the predicted size of 477 bp. The unspecific product observed in both the RT+ and RT- wells this predicted size (see last four lanes of Figure 15, marked with “3”). All these considerations were applied to the optimiza-tion of the RNA extraction procedure.

Figure 16 Example of accepted primer pair for gad1a that gave unspecific products in the negative reactions, yet was used in the experiment of Paper I. The unspecific products were of different size compared to the specific one.

68

Considerations on the results and future research directions

In the current PhD thesis, the principal focus was the GABA signaling sys-tem in the adult zebrafish brain. The interest for this field of research arose due to the lack of information on this topic. That situation was somewhat surprising, as the zebrafish is a routinely used laboratory model organism and the GABA signaling system is ancient and conserved. In Paper I, the expression level of forty genes encoding for GABA signaling players was measured and integrated in the light of mammalian literature. Twenty-three GABAA receptor subunits were analyzed, and their mRNA levels allowed speculating on possible subunit composition in the mature zebrafish brain. It was proposed that α1β2γ2 be the most common GABAA channel in the whole brain, as documented for mammals (Olsen and Sieghart, 2008). Regional brain differences could be observed in terms of GABAA receptor subunits, corroborated by the results of Paper III. In the zebrafish telencephalon and olfactory bulbs α1, α2b, α5, β2, β3, γ2, δ were expressed to the highest level. While the α1 subunit is in general abundantly transcribed, it is stimulating to speculate on α2b, α5 and β3. In mammals, the α2 subunit is abundanlty tran-scribed in the amygdaloid body and hippocampal formation (Wisden et al., 1992; Hörtnagl et al., 2013), structures of the limbic system pivotal for sen-sory processing and memory formation. It is interesting to note that in the murine CeA, α1 is almost not transcribed, whereas α2 is abundant throughout all its nuclei (Marowksy et al., 2004). The α5 mRNA is detected to the high-est extent in the hippocampus, and specifically in the CA subdivisions (Wis-den et al., 1992; Hörtangl et al., 2013). A certain degree of α5 expression is also documented for the olfactory bulbs (Hörtnagl et al., 2013). In the mammalian CA, this subunit is predominantly found in extrasynaptic α5β3γ2 GABAA channels (Wisden et al., 1992; Caraiscos et al., 2004; Hörtnagl et al., 2013). Based on the results of Paper I, it is possible that such a GABAA receptor is present in the zebrafish olfactory bulbs and telencephalon (Fig-ure 7 of Paper I). In teleosts, areas of the dorsal telencephalon, or telence-phalic pallium, are considered to be homologous to mammalian limbic struc-tures. Specifically, the medial pallium is associated to aversive stimuli con-ditioning, whereas the lateral pallium is required for spatial and temporal learning (Salas et al., 2006). The former is therefore depicting amygdalar functions, whereas the latter is associated with hippocampal ones (Portavella

69

et al., 2002; Broglio et al., 2005). It is possible that α5β3γ2 GABAA receptors are concentrated in the zebrafish lateral pallium, as they are in the mammali-an hippocampal formation (Paper I, Paper III; Wisden et al., 1992; Carais-cos et al., 2004; Hörtnagl et al., 2013). Gene expression profiles in terms of RT-qPCR and in situ hybridization, restricted to the sole telencephalon, could confirm if the speculations reported above go in the right direction. It is possible to further discuss the α3 and α4 GABAA receptor monomers in the olfactory bulbs and telencephalon of the adult zebrafish. Those two sub-units are not abundantly transcribed in the aforementioned regions; in the mouse brain, the amygdala is one site of production of α3 (Marowsky et al., 2004, 2012; Hörtnagl et al., 2013). Here, it is located in the extrasynaptic space, possibly with β3 and γ2 (Marowsky et al., 2012; Hörtnagl et al., 2013). Quaternary amygdalar GABAA receptors are also possible, with two differ-ent α subunits; according to expression levels and protein production rates, α2α3β3γ2 GABAA channels could exist (Marowsky et al., 2004, 2012; Hörtnagl et al., 2013). Based on the expression profiles obtained with Paper I and Paper III, the zebrafish medial pallium could produce α3βxγ2 GABAA receptors, with the β subunit being either β2 or β3. The α4 subunit usually confers extrasynaptic location to a GABAA pentamer and it is found in com-bination with δ in the mammalian hippomcapus. A possibility that ternary α4βxδ receptors are synthesized in the telencephalon of the zebrafish, and that they might locate in the lateral pallium, is present. The contribution of the olfactory bulbs to the overall GABAA receptor population may be in-ferred by taking into account the results of Paper I and data emerging from mammalian literatutre (Wisden et al., 1992; Hörtnagl et al., 2013). In that brain area, synaptic α1βxγ2 and extrasynaptic α1βxδ receptors may be present. As considered for the α5β3γ2 subunit combination, combined RT-qPCR with in situ hybridization experiments could help elucidating if the proposed channels are likely to exist. It was quite difficult to speculate on possible GABAA receptors in the adult zebrafish diencephalon, as the expression profiles obtained in Paper I were not well defined (Figure 6). The fact that the telencephalon and diencephalon were sampled together in Paper III somehow hindered the speculation power of the study in that direction. Nonetheless, expression profiles in mammals can help conceiving GABAA receptor subunits combination in the diencephalon. The majority of thalamic GABAA channels may well be extrasynaptic, as α4 and δ are the predomi-nant monomers found in this brain area (Wisden et al., 1992; Hörtnagl et al., 2013). In fact α4, but above all δ, are exclusively found in extrasynaptic re-ceptors (Farrant and Nusser, 2005). Therefore, GABA may predominantly mediate tonic inhibition in the thalamus. The subunits α1 and β2 are also present in this brain area in mammals, both in terms of mRNA and protein (Wisden et al., 1992; Hörtnagl et al., 2013). Those two monomers can be equally found in synaptic and extrasynaptic GABAA receptors; proposed thalamic channels are thus α1β2δ and α4β2δ (Wisden et al., 1992; Hörtnagl et

70

al., 2013). It is possible that the zebrafish thalamus presents the same GABAA receptors, as emerged from the expression profile of the GABAA channel monomers in the telencephalon and diencephalon of the adult zebrafish (Figure 3A of Paper III). In fact, both α4 and δ were detected to a certain level in the aforementioned brain areas. Paper I confirmed that the zebrafish GABA system shares several common points with that in mam-mals. Thus, α4 and δ synthesis may interest the lateral pallium and thalamus of zebrafish, as it does in the hippocampus of mammals (Wisden et al., 1992; Hörtnagl et al., 2013). In the teleost diencephalon, the extrasynaptic combinations α1β2δ and α4β2δ might be likely to exist.

The mammalian α6 subunit is exclusively produced in the cerebellum (Kato, 1990; Wisden et al., 1992; Hörtnagl et al., 2013). In zebrafish, there are two paralogoues of that subunit, namely α6a and α6b (Klee et al., 2012): They probably originated after the third round of whole genome duplication occurred in the teleost lineage (Vandepoele et al., 2004). From Paper I, it emerged that the two monomers have a tissue-specific expression when con-sidering the retinas and the whole brain. The mRNA of α6a was almost ex-clusively detected in the retinas, whereas α6b was transcribed in both ana-lyzed tissues, yet its transcript was more abundant in the brain. This finding opens an array of tantalizing questions. Mammalian GABAA receptors with α6 are activated by the lowest ligand concentration when compared to chan-nels with other α monomers, maintaining a fixed β and γ stoichiometry (Böhme et al., 2004). Moreover, the presence of α6 in a GABAA channel is sufficient to promote 3α,5α-THDOC potentiation, without needing aid from the δ subunit (Wohlfarth et al., 2002). The role of the two different pa-ralogues in the zebrafish could be investigated in several ways. The first point to be elucidated is whether the α6a and α6b proteins are actually pro-duced in the zebrafish. In fact, the detection of an mRNA transcript does not necessarily mean that the actual polypeptide is synthesized. Such an observa-tion also applies for all the other analyzed genes in this thesis. The second question is whether α6a and α6b confer different affinities to the receptor for GABA. Similarly, the two isoforms might bind 3α,5α-THDOC in distinct ways, and differently potentiate the conducting activity of GABAA channels. Posssible insights into this topic might be provided by structural studies. It is possible to construct three-dimensional protein models based on structural homology, as done in Paper II for the zebrafish ζ subunit (Šali and Blun-dell, 1993; Webb and Sali, 2014). In GABAA receptors, the neurosteroid binding pocket is located between the principal and complementary faces of one β and one α subunit, respectively (Miller et al., 2017). Modeling the three-dimensional structure of both α6a and α6b could provide insights into the amino acid residues whose lateral chains line the steroid binding site. A quick multiple sequence alignment could also help determining the neuro-steroid binding pocket provided by the zebrafish α6 subunits. This would also allow speculating on the interactions between the hormone and the pro-

71

tein, based on the work by Miller et al. (2017). The actual production of neurosteroids and of interrenal-tissue derived steroids may also be investi-gated, both in the brain and in the retinas. The results may integrate the pic-ture on α6a- or α6b-containing GABAA receptors, and possibly elucidate the different functional roles of the two paralogues.

The in situ hybridization experiment presented in Paper II can be viewed as a pilot study to optimize that technique with receptor subunits. In the zebrafish genome, there are at least two orthologues to the mammalian π, namely π and ζ. In mammals, π mRNA is detected to a very low level in the brain, where it localizes in the hippocampus and temporal cortex (Hedblom and Kirkness, 1997). In Paper I, the transcript of π was also detected to a low level in the zebrafish brain and retinas, but that of ζ was indeed pro-duced to a certain degree (Figure 1 C, E and Figure 12 C, E). Moreover, regional brain differences were documented in terms of the abudance of the ζ mRNA (Paper I). Therefore, focus was placed on that subunit regarding its anatomical distribution in the zebrafish central nervous system and retinas (Paper II). In mammals, the αβπ GABAA receptor subunit combination has been listed as a tentative one (Olsen and Siegahrt, 2008). Nonetheless, in-ducing the production of π-containing GABAA receptors in heterologous expression systems resulted in functional channels with specific chemical-physical properties (Hedblom and Kirkness, 1997; Neelands and Macdonald, 1999). Ternary α5β3γ3 and α5β3π GABAA channels had lower affinity for GABA, compared to binary α5β3 (Neelands and Macdonald, 1999). Howev-er, neurosteroids potentiated α5β3π GABAA receptors in a manner similar to that reported for α5β3 and, in both cases, the potentiation was higher than that measured for α5β3γ3 channels (Neelands and Macdonald, 1999). As re-ported above, the π mRNA has been detected in the hippocampus (Hedblom and Kirkness, 1997). The α5 subunit is abundantly produced in that brain area, where it localizes in the CA subdivisions, assembles with β3 and gives rise to extrasynaptic GABAA receptors (Wisden et al., 1992; Caraiscos et al., 2004; Farrant and Nusser, 2005; Hörtangl et al., 2013). Perhaps a small, extrasynaptic subpopulation of α5β3π GABAA channels exists in the mam-malian hippocampus. Crosstalks between peripherally produced steroid hormones and the brain might be partly mediated by π-containing GABAA receptors. Binary β3π channels have lower sensitivity to pregnanolone com-pared to homomeric β3 receptors, and therefore respond to that steroid only when there is a certain concentration available (Hedblom and Kirkness, 1997). Pregnanolone is the 5β enantiomer of allopregnanolone, and those two compounds are both synthesized from progesterone (from Sigma Al-drich, https://www.sigmaaldrich.com/catalog/product/sigma/p8887?lang=en &region=SE and https://www.sigmaaldrich.com/catalog/product/sigma/p 8129?lang=en&region=SE, respectively; Wang, 2011; Gong and Smith, 2014). It is licit to infer that females experience fluctuations of the aforemen-tioned steroid hormones during the ovarian cycle, and those variations in

72

hormone concentration might be centrally sensed by π-containing GABAA receptors. To note, the π subunit has its highest level of expression in the uterus (Hedblom and Kirkness, 1997). Effects of steroid hormones related to the ovarian cycle do influence the GABA signaling system in the mammali-an brain, as demonstred for steroid-driven production of the δ subunit (Maguire and Mody, 2007). As proposed above, investigation of the actual production of neurosteroids, as well as of brain metabolites of peripherally produced steroid hormones, could be a valuable help to get a more complete picture of the GABA signaling system in the adult zebrafish brain. The ques-tion is still open whether the ζ subunit is synthesized and incorporated into functional GABAA channels. Possible in situ hybridization experiments could focus on the co-localization of α5, β3 and ζ transcript in the region of the lateral telencephalic pallium in zebrafish.

During synaptic transmission, the classical neurotrasnmitters thorugh which circuits are defined are not the sole substances acting upon target cells. In fact, the whole content of synaptic vesicles is released in the intersti-tial space; for GABAergic circuits, that encompasses GABA, protons and Cl– (see “The γ-aminobutyric acid signaling system, GABA turnover, Catab-olism” and “The proton signaling system”; Juge et al., 2009). Therefore, an array of compounds acts on target neurons to finely regulate synaptic trans-mission. Extracellular protons not only exert effects on receptors for classi-cal neurotransmitters, but do have their own receptor suite (Kristhal and Pidoplichko, 1980; Wemmie et al., 2006). This discovery opened up flam-boyant possibilities to investigate the role of fluctuations of interstistial pH on synaptic plasticity. In mammals, it is well characterized that the ASIC1 subunit is widely expressed in the brain and that it can be found in both ho-mo- and heterotrimeric channels (Bassilana et al., 1997; Wemmie et al., 2002, 2003; Askwith et al., 2004). ASIC1 mRNA is alternatively spliced in two variants, namely ASIC1a and ASIC1b, with the former much more abundant compared to the latter in the brain (Wemmie et al., 2002). Alterna-tive splicing is also documented for the ASIC2 transcript, producing ASIC2a and ASIC2b (Wemmie et al., 2002; Askwith et al., 2004). The population of functional ASICs in the mammalian brain may be well composed of ASIC1a:ASIC2 channels with different subunit stoichiometry (Askwith et al., 2004). In the zebrafish genome, there are three paralogues of the ASIC1 subunit, asic1a-c (Pauckert et al., 2004). It is possible that the different tele-ost asic1 paralogues are found in place of the mammalian splice variants. It is also possible that the paralogues are themselves alternatively spliced, pre-senting the scenario that zebrafish proton receptors display a potentially wider array of subunit combinations compared to mammals. Homo- and heterometic zebrafish ASICs studied in a heterologous expression system present different current amplitudes, pH sensitivities and desensitization rates (Pauckert et al., 2004). In mammals, the ASIC2 subunit confers lower affinity for H+ compared to ASIC1, and channels with the former monomer

73

require steeper pH falls to be activated (Bassilana et al., 1997; Askwith et al., 2004). In zebrafish, such a condition has been reported for asic1a:asic1c heterotrimers (Pauckert et al., 2004). Future research directions may focus on expression patterns of the asic genes in the adult zebrafish brain, which would allow speculating on possible subunit compositions of H+ receptors. Research interest may also be placed on the ASIC4 gene; in the mammalian brain, its mRNA is detected in several brain areas (Hoshikawa et al., 2017). The ASIC4 monomer, as the ASIC2b splice variant of ASIC2, does not dis-play proton sensitivity, and it may thus not contribute to the proton antenna pairs (see “The proton signaling system, The proton receptor system”; Saw-yer and James, 1982; Wemmie et al., 2006). In zebrafish the two paralogues asic4a and asic4b are found; the former mediates H+-gated cation currents, the latter does not (Pauckert et al., 2004; Chen et al., 2007a). When charac-terized in a heterologous expression system, asic4b is more abundantly found at the cell surface compared to asic4a (Chen et al., 2007a). It is possi-ble that asic4b forms homotrimers gated by chemical spceis other than H+ (Chen et al., 2007a). On the other hand, asic4a may exclusively be found in proton-sensitive, heterotrimeric ASICs and contribute to the establishment of the H+ sensing antennas (Sawyer and James, 1982; Chen et al., 2007a). The aforementioned studies were performed with ASICs produced by heterolo-gous expression systems, and might therefore not mirror the actual in vivo situation. A characterization of ASIC subunit synthesis and their anatomical distribution patterns may further elucidate the proton signaling system in the adult zebrafish brain.

74

Acknowledgements

First and foremost, my deepest gratitude goes to my supervisor Svante Win-berg. I am so grateful for the opportunity to pursue my PhD education in your lab, and that you believed in my ideas, supported and encouraged them with all your means. I also appreciated all the dinners and social events to integrate with Swedish life and culture. My co-supervisor Fredrik Jutfelt believed in me as well, and gave me the chance, together with Svante, to start my PhD education in Sweden. For this I would like to express a big thank you. I am deeply indebted to B. Þrándur Björnsson, who first accepted me as a Master student in the Fish Endocrinology Laboratory group at the University of Gothenburg in 2012. Your passion and enthusiasm for life are a true ex-ample to follow. A big hug also goes to Ingibjörg E. Einarsdóttir, who has always found the right words for me whenever we met. I wish to thank my co-supervisor Per-Ove Thörnqvist for several insightful discussions on protocols and techniques. Bryndis Birnir, co-supervisor of mine alike, has an impressive knowledge on the GABA system. I am thank-ful for sharing that with me, and for always providing fruitful comments to my work. Michael Williams, Sergiy Korol and Zhe Jin dedicated significant amounts of their time to my questions on methods and papers, and I am grateful for that. My colleagues Laura Vossen and Johanna Axling shaped a great working environment, together with all the members of the Unit of Physiology at the Department of Neuroscience. I also would like to express my gratitude to my co-authors for collaborations and help during the experimental work. A thank you is for the colleagues at the Department alike, of whom I would like to mention Misty Attwood and Olga Dyakova. To the first one I am indebted for always finding time to check my written English. To the second one, for celebrating birthday parties and for the nice gift of Кин-дза-дза!. I am also thankful to Mariana Hooli for listening to me on virtually every matter I needed to talk about. I was also very lucky in meeting a true friend in Ningping Gong, who shared life experiences, thoughts and opinions with me. Help or suggestions always come when I need them, and I would like to say thank you for that.

75

Att flytta till Göteborg var en stor utmaning, men också en tur. Jag är så tacksam att jag fick bo i Sverige, lära mig svenska och utveckla mig i den person som jag är nu. Därför skulle jag också vilja tacka Sverige. Infine, il ringraziamento più grande va ai miei genitori, che sono stati sempre con me nonostante sei anni di distanza. Grazie per essere la mia fonte princi-pale di forza e motivazione.

76

References

Aggleton, J. P., Mishkin, M. (1986) Chapter 12: The Amygdala: Sensory Gateway to the Emotions. In: Emotion: Theory, Research, and Experience, Volume 3: Bi-ological Foundations of Emotion, R. Plutchik and H. Kellerman (eds.), pgg. 281-299, © Academic Press, Inc., Orlando

Anglade, I., Mazurais, D., Douard, V., Le Jossic-Corcos, C., Mañanos, E. L., Michel, D., Kah, O. (1999) Distribution of Glutamic Acid Decarboxylase mRNA in the Forebrain of the Rainbow Trout as Studied by In Situ Hybridization. J. Comp. Neurol. 410(2):277-289

Aronson, P. S., Boron, W. F., Boulpaep, E. L. (2012) Chapter 5: Transport of So-lutes and Water. In: Medical Physiology, 2nd edition, W. F. Boron and E. L. Boulpaep (eds.), pgg. 106-146, © Saunders Elsevier, Philadelphia

Askwith, C. C., Wemmie, J. A., Price, M. P., Rokhina, T., Welsh, M. J. (2004) Ac-id-sensing Ion Channel 2 (ASIC2) Modulates ASIC1 H+-activated Currents in Hippocampal Neurons. J. Biol. Chem. 279(18):18296-18305

Barrett, E. J. (2012) Chapter 50: The Adrenal Gland. In: Medical Physiology, 2nd edition, W. F. Boron and E. L. Boulpaep (eds.), pgg. 1057-1073, © Saunders Elsevier, Philadelphia

Bassilana, F., Champigny, G., Waldmann, R., de Weille, J. R., Heurteaux, C., Lazdunski, M. (1997) The Acid-sensitive Ionic Channel Subunit ASIC and the Mammalian Degenerin MDEG Form a Heteromultimeric H+-gated Na+ Channel with Novel Properties. J. Biol. Chem. 272(46):28819-28822

Belelli, D., Lambert, J. J. (2005) Neurosteroids: endogenous regulators of the GABAA receptor. Nat. Rev. Neurosci. 6(7):565-575

Bennett, E. R., Kanner, B. I. (1997) The Membrane Topology of GAT-1, a (Na+ + Cl–)-coupled γ-Aminobutyric Acid Transporter from Rat Brain. J. Biol. Chem. 272(2):1203-1210

Berg, J. M., Tymoczko, J. L., Stryer, L., Gatto, G. J. (2012a) Chapter 17: The Citric Acid Cycle. In Biochemistry, 7th edition, pgg. 515-542, © W. H. Freeman and Company, New York

Berg, J. M., Tymoczko, J. L., Stryer, L., Gatto, G. J. (2012b) Chapter 18: Oxidative Phosphorylation. In Biochemistry, 7th edition, pgg. 543-584, © W. H. Freeman and Company, New York

Berg, J. M., Tymoczko, J. L., Stryer, L., Gatto, G. J. (2012c) Chapter 23: Protein Turnover and Amino Acid Catabolism. In: Biochemistry, 7th edition, pgg. 697-727, © W. H. Freeman and Company, New York

Berkich, D. A., Ola, M. S., Sweatt, A. J., Hutson, S. M., LaNoue, K. F. (2007) Mito-chondrial Transport Proteins of the Brain. J. Neurosci. Res. 85(15):3367-3377

Bianchi, M. T., Macdonald, R. L. (2002) Slow phases of GABAA receptor desensiti-zation: structural determinants and possible relevance for synaptic function. J. Physiol. 544(Pt 1):3-18

77

Böhme, I., Rabe, H., Lüddens, H. (2004) Four Amino Acids in the α Subunits De-termine the γ-Aminobutyric Acid Sensitivities of GABAA Receptor Subtypes. J. Biol. Chem. 279(34):35193-35299

Boileau, A. J., Evers, A. R., Davis, A. F., Czajkowski, C. (1999) Mapping the Ago-nist Binding Site of the GABAA Receptor: Evidence for a β-Strand. J. Neurosci. 19(12):4847-4854

Bormann, J., Hamill, O. P., Sakmann, B. (1987) Mechanism of anion permeation through channels gated by glycine and γ-aminobutyric acid in mouse cultured spinal neurones. J. Physiol. 385:243-286

Bosma, P. T., Blázquez, M., Collins, M. A., Bishop, J. D. D., Drouin, G., Priede, I. G., Docherty, K., Trudeau, V. L. (1999) Multiplicity of Glutamic Acid Decar-boxylases (GAD) in Vertebrates: Molecular Phylogeny and Evidence for a New GAD Paralog. Mol. Biol. Evol. 16(3):397-404

Boulpaep, E. L. (2012) Chapter 25: Integrated Control of the Cardiovascular Sys-tem. In: Medical Physiology, 2nd edition, W. F. Boron and E. L. Boulpaep (eds.), pgg. 593-609, © Saunders Elsevier, Philadelphia

Bowers, G., Cullinan, W. E., Herman, J. P. (1998) Region-Specific Regulation of Glutamic Acid Decarboxylase (GAD) mRNA Expression in Central Stress Cir-cuits. J. Neurosci. 18(15):5938-5947

Bowery, N. G., Bettler, B., Froestl, W., Gallagher, J. P., Marshall, F., Raiteri, M., et al. (2002) International Union of Pharmacology. XXXIII. Mammalian γ-Aminobutyric AcidB Receptors: Structure and Function. Pharmacol. Rev. 54(2):247-264

Bowery, N. G., Price, G. W., Hudson, A. L., Hill, D. R., Wilkin, G. P., Turnbull, M. J. (1984) GABA receptor multiplicity. Visualization of different receptor types in the mammalian CNS. Neuropharmacology 23(2B):219-231

Brady, C. A., Rennekamp, A. J., Peterson, R. T. (2016) Chapter1: Chemical Screen-ing in Zebrafish. In: Methods in Molecular Biology, Zebrafish: Methods and Protocols, 2nd edition, K. Kawakami, E. E. Patton and M. Orger (eds.), pgg. 3-16, © Springer Science + Business Media, New York

Broglio, C., Gómez, A., Durán, E., Ocaña, F. M., Jiménez-Moya, F., Rodríguez, F., Salas, C. (2005) Hallmarks of a common forebrain vertebrate plan: Specialized pallial areas for spatial, temporal and emotional memory in actinopterygian fish. Brain Res. Bull. 66(4-6):277-281

Broglio, C., Gómez, A., Durán, E., Salas, C., Rodríguez, F. (2011) Chapter 15: Brain and Cognition in Teleost Fish. In: Fish Cognition and Behavior, 2nd edi-tion, C. Brown, K. Laland and J. Krause (eds.), pgg. 325-358, © Blackwell Pub-lishing Ltd., John Wiley and Sons, Chichester

Buddhala, C., Hsu, C.-C., Wu, J.-Y. (2009) A novel mechanism for GABA synthesis and packaging into synaptic vesicles. Neurochem. Int. 55(1-3):9-12

Burmakina, S., Geng, Y., Chen, Y., Fan, Q. R. (2014) Heterodimeric coiled-coil interactions of human GABAB receptors. Proc. Natl. Acad. Sci. U. S. A. 111(19)6958-6963

Bustin, S. A., Benes, V., Garson, J. A., Hellemans, J., Huggett, J., Kubista, M. et al. (2009) The MIQE Guidelines: Minimum Information for Publication of Quantia-tive Real-Time PCR Experiments. Clin. Chem. 55(4):611-622.

Caraiscos, V. B., Elliott, E. M., You-Ten, K. E., Cheng, V. Y., Belelli, D. Newell, J. G., et al. (2004) Tonic inhibition in mouse hippocampal CA1 pyramidal neurons

78

is mediated by α5 subunit-containing γ-aminobutyric acid type A receptors. Proc. Natl. Acad. Sci. U. S. A. 101(10):3662-3667

Cardoso, J. C. R., Bergqvist, C. A., Félix, R. C., Larhammar, D. (2016) Corticotro-pin-releasing hormone family evolution: five ancestral genes remain in some lin-eages. J. Mol. Endocrinol. 57(1):73-86

Chang, R. (2000) Chapter 11: Acid and Bases. In: Physical Chemistry for the Chem-ical and Biological Sciences, 3rd edition, pgg. 397-444, © University Science Books, Sausalito

Chen, N.-H., Reith, M. E. A., Quick, M. W. (2004) Synaptic uptake and beyond: the sodium- and chloride-dependent neurotransmitter transporter family SLC6. Pflu-gers Arch. 447(5):519-531

Chen, X., Polleichtner, G., Kadurin, I., Gründer, S. (2007a) Zebrafish Acid-sensing Ion Channel (ASIC) 4, Characterization of Homo- and Heteromeric Channels, and Identification of Regions Important for Activation by H+. J. Biol. Chem. 282(42):30406-30413

Chen, Z.-W., Olsen, R. W. (2007b) GABAA receptor associated proteins: a key factor regulating GABAA receptor function. J. Neurochem. 100(2):279-294

Cherubini, E., Gaiarsa, J. L., Ben-Ari, Y. (1991) GABA: an excitatory transmitter in early postnatal life. Trends Neurosci. 14(12):515-519

Chesler, M., Kaila, K. (1992) Modulation of pH by neuronal activity. Trends Neuro-sci. 15(10)396-402

Chomczynski, P., Sacchi, N. (1987) Single-Step Method of RNA Isolation by Acid Guanidinium Thiocyanate-Phenol-Chloroform Extraction. Anal. Biochem. 162(1):156-159

Clark, K. J., Boczek, N. J., Ekker, S. C. (2011) Stressing zebrafish for behavioral genetics. Rev. Neurosci. 22(1):49-62

Cocco, A., Rönnberg, A. M. C., Jin, Z., Gonçalo, A. I., Vossen, L. E., Bhandage, A. K., Thörnqvist, P.-O., Birnir, B., Winberg, S. (2017) Characterization of the γ-aminobutyric acid signaling system in the zebrafish (Danio rerio Hamilton) cen-tral nervous system by reverse transcription-quantitative polymerase chain reac-tion. Neuroscience 343:300-321

Connaughton, V. P., Behar, T. N., Liu, W.-L. S., Massey, S. C. (1999) Immunocyto-chemical localization of excitatory and inhibitory neurotransmitters in the zebrafish retina. Vis. Neurosci. 16(3):483-490

Conti, F., Melone, M., De Biasi, S., Minelli, A., Brecha, N. C., Ducati, A. (1998) Neuronal and Glial Localization of GAT-1, a High-Affinity γ- Methods Enzymol. 113:80-82

Coryell, M. W., Wunsch, A. M., Haenfler, J. M., Allen, J. E., Schnizler, M., Ziemann, A. E., et al. (2009) Acid-Sensing Ion Channel-1a in the Amygdala, a Novel Therapeutic Target in Depression-Related Behavior. J. Neurosci. 29(17):5381-5388

Coryell, M. W., Ziemann, A. E., Westmoreland, P. J., Haenfler, J. M., Kurjakovic, Z., Zha, X.-m., et al. (2007) Targeting ASIC1a Reduces Innate Fear and Alters Neuronal Activity in the Fear Circuit. Biol. Psychiatry 62(10):1140-1148

Coyle, J. E., Qamar, S., Rajashankar, K. R., Nikolov, D. B. (2002) Structure of GABARAP in Two Conformations: Implications for GABAA Receptor Localiza-tion and Tubulin Binding. Neuron 33(1):63-74

79

Cullinan, W. E. (2000) GABAA Receptor Subunit Expression Within Hypophysio-tropic CRH Neurons: A Dual Hybridization Histochemical Study. J. Comp. Neu-rol. 419(3):344-351

David, F. P.A, Rougemont, J., Deplancke, B. (2017) GETPrime 2.0: gene- and tran-script-specific qPCR primers for 13 species including polymorphisms. Nucleid Acids Res. 45(D1):D56-D60

De Kloet, E. R., Vreugdenhil, E., Oitzl, M. S., Joël, M. (1998) Brain Corticosteroid Receptor Balance in Health and Disease. Endocr. Rev., 19(3):269-301

Engebrecht, J., Brent, R., Kaderbhai, M. A. (2001) Minipreps of Plasmid DNA. Curr. Protoc. Mol. Biol. 15(II)1.6:1.6.1-1.6.10.

Engeszer, R. E., Patterson, L. B., Rao, A. A., Parichy, D. M. (2007) Zebrafish in the Wild: Review of Natural History and New Notes from the Field. Zebrafish 4(1):21-40

Erlichman, J. S., Hewitt, A., Damon, T. L., Hart, M., Kurascz, J., Li, A., et al. (2008) Inhibition of Monocarboxylate Transporter 2 in the Retrotrapezoid Nu-cleus in Rats: A Test of the Astrocyte-Neuron Lactate-Shuttle Hypothesis. J. Neurosci. 28(19):4888-4896

Essrich, C., Lorez, M., Benson, J. A., Fritschy, J.-M., Lüscher, B. (1998) Poststyn-aptic clustering of major GABAA receptor subtypes requires the γ2 subunit and gephyrin. Nat. Neurosci. 1(7):563-571

Everitt, A. B., Luu, T., Cromer, B., Tierney, M. L., Birnir, B., Olsen, R. W., Gage, P. W. (2004) Conductance of Recombinant GABAA Channels Is Increased in Cells Co-expressing GABAA Receptor-associated Protein. J. Biol. Chem. 279(21):21701-21706

Eyster, K. M., Brannian, J. D. (2009) Chapter 5: Gene Expression Profiling in the Aging Ovary. In: Methods in Molecular Biology, Molecular Endocrinology, O.K. Park-Sarge and T. E. Curry, Jr. (eds.), pgg. 71-89, © Humana Press, a part of Springer Science + Business Media, New York

Farrant, M., Nusser, Z. (2005) Variations on an inhibitory theme: phasic and tonic activation of GABAA receptors. Nat. Rev. Neurosci. 6(3):215-229

Fenalti, G., Law, R. H. P., Buckle, A. M., Langendorf, C., Tuck, K., Rosado, C. J., et al. (2007) GABA production by glutamic acid decarboxylase is regulated by a dynamic catalytic loop. Nat. Struct. Mol. Biol. 14(4):280-286

Finbow, M. E., Harrison, M. A. (1997) The vacuolar H+-ATPase: a universal proton pump of eukaryotes. Biochem. J. 324(Pt 3):697-712

Forrest, L. R., Tavoulari, S., Zhang, Y.-W., Rudnick, G., Honig, B. (2007) Identifi-cation of a chloride ion binding site in Na+/Cl–-dependent transporters. Proc. Natl. Acad. Sci. U. S. A. 104(31):12761-12766

Gielen, M. C., Lumb, M. J., Smart, T. G. (2012) Benzodiazepines Modulate GABAA Receptors by Regulating the Preactivation Step after GABA Binding. J. Neuro-sci. 32(17):5707-5715

Gielen, M., Thomas, P., Smart, T. G. (2015) The desensitization gate of inhibitory Cys-loop receptors. Nat. Comm. 6:6829. doi: 10.1038/ncomms7829

Gong, Q. H., Smith, S. S. (2014) Characterization of neurosteroid effects on hy-perpolarizing currents at α4β2δ GABAA receptors. Psychopharmacology 231(17):3525-3535

Gorissen, M., Flik, G. (2016) Chapter 3: The Endocrinology of the Stress Response in Fish. In: Biology of Stress in Fish, C. B. Schreck, L. Tort, A. P. Farrell, C. J. Brauner (eds.), pgg. 76-111, © Elsevier Inc.

80

Gorka, A. X., Torrisi, S., Shackman, A. J., Grillon, C., Ernst, M. (2017) Intrinsic functional connectivity of the central nucleus of the amygdala and the bed nucle-us of the stria terminalis. Neuroimage http://dx.doi.org/10.1016/j.neuroimage.2017.03.007

Gray, H. (2005a) Chapter 21: Diencephalon. In: Gray’s Anatomy: The Anatomical Basis of Clinical Practice, 39th edition, A. R. Crossman and S. Standring (eds.), pgg. 369-385, Elsevier Churchill Livingstone, © Elsevier Ltd

Gray, H. (2005b) Chapter 22: Cerebral hemisphere. In: Gray’s Anatomy: The Ana-tomical Basis of Clinical Practice, A. R. Crossman and S. Standring (eds.), pgg. 387-417, Elsevier Churchill Livingstone, © Elsevier Ltd

Grone, B. P., Maruska, K. P. (2015a) A Second Corticotropin-Releasing Hormone Gene (CRHS2) Is Conserved Across Vertebrate Classes and Expressed in the Hindbrain of a Basal Neopterygian Fish, the Spotted Gar (Lepisosteus oculatus). J. Comp. Neurol. 523(7):1125-1143

Grone, B. P., Maruska, K. P: (2015b) Divergent evolution of two corticotropin-releasing hormone (CRH) genes in teleost fish. Front. Neurosci. 9:365

Grone, B. P., Maruska, K. P. (2016) Three Distinct Glutamate Decarboxylase Genes in Vertebrates. Sci. Rep. 6:30507

Grunwald, D. J., Eisen, J. S. (2002) Headwater of the zebrafish – emergence of a new model vertebrate. Nat. Rev. Genet. 3(9):717-724

Haas, K. F., Macdonald, R. L. (1999) GABAA receptor subunit γ2 and δ subtypes confer unique kinetic properties on recombinant GABAA receptor currents in mouse fibroblasts. J. Physiol. 514(Pt 1):27-45

Hakamata, Y., Komi, S., Moriguchi, Y., Izawa, S., Motomura, Y., Sato, E., Mizu-kami, S., Kim, Y., Hanakawa, T., Inoue, Y., Tagaya, H. (2017) Amygdala-centred functional connectivity affects daily cortisol concentrations: a putative link with anxiety. Sci. Rep. 7(1):8313

Hamilton (formerly Buchanan), F. (1822) An account of the fishes found in the river Ganges and its branches, Archibald Constable and Company, Edinburgh, and Hurst, Robinson, and Co., London

Han, F., Ding, J., Shi, Y. (2014) Expression of amygdala mineralocorticoid receptor and glucocorticoid receptor in the single-prolonged stress rats. BMC Neurosci. 15:77

Hassaine, G., Deluz, C., Grasso, L., Wyss, R., Tol, M. B., Hovius, R., et al. (2014) X-ray stucture of the mouse serotonin 5-HT3 receptor. Nature 512(7514):276-281

Hearl, W. G., Churchich, J. E. (1984) Interactions between 4-Aminobutyrate Ami-notransferase and Succinic Semialdehyde Dehydrogenase, Two Mitochondrial Enzymes. J. Biol. Chem. 259(18)11459-11463

Hedblom, E., Kirkness, E. F. (1997) A Novel Class of GABAA Receptor Subunit in Tissues of the Reproductive System. J. Biol. Chem. 272(24):15346-15350

Herman, J. P., McKlveen, J. M., Ghosal, S., Kopp, B., Wulsin, A., Makinson, R., et al. (2016) Regulation of the hypothalamic-pituitary-adrenocortical stress re-sponse. Compr. Physiol. 6(2):603-621

Herman, J. P., Ostrander, M. M., Mueller, N. K., Figueiredo, H. (2005) Limbic sys-tem mechanisms of stress regulation: Hypothalamo-pituitary-adrenocortical axis. Prog. Neuropsycopharmacol. Biol. Psychiatry 29(8):1201-1213

81

Hewitt, S. A., Wamsteeker, J. I., Kurz, E. U., Bains, J. S. (2009) Altered chloride homeostasis removes synaptic inhibitory constraint of the stress axis. Nat. Neu-rosci. 12(4):438-443

Hildebrand, G. D., Fielder, A. R. (2011) Chapter 2: Anatomy and Physiology of the Retina. In: Pediatric Retina, J. Reynolds and S. Olitsky (eds.), pgg. 39-65, © Springer-Verlag Berlin Heidelberg

Hörtnagl, H., Tasan, R. O., Wieselthaler, A., Kirchmair, E., Sieghart, W., Sperk, G. (2013) Patterns of mRNA and protein expression for 12 GABAA receptor subu-nits in the mouse brain. Neuroscience 236:345-372

Hoshikawa, M., Kato, A., Hojo, H., Shibata, Y., Kumamoto, N., Watanabe, M., et al. (2017) Distribution of ASIC4 transcripts in the adult wild-type mouse brain. Neurosci. Lett. 651:57-64

Howe, K., Clark, M. D., Torroja, C. F., Torrance, J., Berthelot, C., Muffato, M., et al. (2013) The zebrafish reference genome sequence and its relationship to the human genome. Nature 496(7446):498-503

Hu, J., Weise, C., Böttcher, C., Fan, H., Yin, J. (2017) Expression, purification and structural analysis of functional GABA transporter 1 using the baculovirus ex-pression system. Beilstein J. Org. Chem. 13:874-882

Huang, X., Chen, H., Michelsen, K., Schneider, S., Shaffer, P. L. (2015) Crystal structure of human glucine receptor-α3 bound to agonist strychnine. Nature 526(7572):277-280

Ippolito, J. E., Piwnica-Worms, D. (2014) A Fluorescence-Coupled Assay for Gamma Aminobutyric Acid (GABA) Reveals Metabolic Stress-Induced Modula-tion of GABA Content in Neuroendocrine Cancer. PLoS One 9(2):e88667

Jasti, J., Furukawa, H., Gonzales, E. B., Gouaux, E. (2007) Structure of acid-sensing ion channel 1 at 1.9 Å resolution and low pH. Nature 449(7160):316-323

Jin, Z., Mendu, S. K., Birnir, B. (2013) GABA is an effective immunomodulatory molecule. Amino Acids 45(1):87-94

Juge, N., Muroyama, A., Hiasa, M., Omote, H., Moriyama, Y. (2009) Vesicular Inhibitory Amino Acid Transporter Is a Cl– /γ-Aminobutyrate Co-transporter. J. Biol. Chem. 284(50):35073-35078

Kaila, K., Voipio, J. (1987) Postsynaptic fall in intracellular pH induced by GABA-activated bicarbonate conductance. Nature 330(6144):163-165

Kamali, A., Sair, H. I., Blitz, A. M., Riascos, R. F., Mirbagheri, S., Keser, Z., et al. (2016) Revealing the ventral amygdalofugal pathway of the human limbic sys-tem using high spatial resolution diffusion tensor tractography. Brain Struct. Funct. 221(7):3561-3569

Kamali, A., Yousem, D. M., Lin, D. D., Sair, H. I., Jasti, S. P., Keser, Z., et al. (2015) Mapping the trajectory of the stria terminalis of the human limbic system using high spatial resolution diffusion sensor tractography. Neurosci. Lett. 608:45-50

Kato, K. (1990) Novel GABAA Receptor α Subunit is Expressed only in Cerebellar Granule Cells. J. Mol. Biol. 214(3):619-624

Kaufman, D. L., Houser, C. R., Tobin, A. J. (1991) Two Forms of the γ-aminobutyric Acid Synthetic Enzyme Glutamate Decarboxylase Have Distinct Intraneuronal Distributions and Cofactor Interactions. J. Neurochem. 56(2):720-723

82

Keynan, S., Kanner, B. I. (1988) γ-Aminobutyric Acid Transport in Reconstituted Preparations from Rat Brain: Coupled Sodium and Chloride Fluxes. Biochemis-try 27(1)12-17

Kim, Y.-G., Lee, S., Kwon, O.-S., Park, S.-Y., Lee, S.-J., Park, B.-J., et al. (2009) Redox-switch modulation of human SSADH by dynamic catalytic loop. EMBO J. 28(7):959-968

Kim, M. J., Loucks, R. A., Palmer, A. L., Brown, A. C., Solomon, K. M., Marchante, A. N., et al. (2011) The structural and functional connectivity of the amygdala: From normal emotion to pathological anxiety. Behav. Brain Res. 223(2):403-410

Kimmel, C. B., Ballard, W. W., Kimmel, S. R., Ullmann, B., Schilling, T. F. (1995) Stages of Embryonic Development of the Zebrafish. Dev. Dyn. 203(3):253-310

Kittler, J. T., Chen, G., Kukhtina, V., Vahedi-Faridi, A., Gu, Z., Tretter, V., et al. (2008) Regulation of synaptic inhibition by phosphor-dependent binding of the AP2 complex to a YECL motif in the GABAA receptor γ2 subunit. Proc. Natl. Acad. Sci. U. S. A. 105(9):3616-3621

Klee, E. W., Schneider, H., Clark, K. J., Cousin, M. A., Ebbert, J. O., Hooten, W. M., et al. (2012) Zebrafish: a model for the study of addiction genetics. Hum. Genet. 131(6):977-1008

Krause, S., Schwarz, W. (2005) Identification and Selective Inhibition of the Chan-nel Mode of the Neuronal GABA Transporter 1. Mol. Pharmacol. 68(6):1728-1735

Krebs, H. A. (1953) The citric acid cycle. Nobel Lecture, December 11, 1953. https://www.nobelprize.org/nobel_prizes/medicine/laureates/1953/krebs-lecture.pdf

Krishtal, O. A., Osipchuk, Yu. V., Shelest, T. N., Smirnoff, S. V. (1987) Rapid ex-tracellular pH transients related to synaptic transmission in rat hippocampal slic-es. Brain Res. 436(2):352-356

Krishtal, O. A., Pidoplichko, V. I. (1980) A receptor for protons in the nerve cell membrane. Neuroscience 5(12):2325-2327

Kusama, N., Harding, A. M. S., Benson, C. J. (2010) Extracellular Chloride Modu-lates the Desensitization Kinetics of Acid-sensing Ion Channel 1a (ASIC1a). J. Biol. Chem. 285(23):17425-17431

Laale, H. W. (1977) The biology and use of zebrafish, Brachydanio rerio in fisheries research. A literature review. J. Fish. Biol. 10(2):121-173

LeDoux, J. E. (2000) Emotion circuits in the brain. Annu. Rev. Neurosci. 23(1):155-184

LeDoux, J. E. (2014) Coming to terms with fear. Proc. Natl. Acad. Sci. U. S. A., 111(8):2871-2878

LeDoux, J. E., Brown, R. (2017) A higher order theory of emotional consciousness. Proc. Natl. Acad. Sci. U. S. A. 114(10):E2016-E2025

Lindquist, C. E. L., Birnir, B. (2006) Graded response to GABA by native extrasyn-aptic GABAA receptors. J. Neurochem. 97(5):1349-1356

Linné, C. A. (1788a) Systema Naturae per Regna Tria Naturae, secundum Classes, Ordines, Genera, Species, cum Characteribus, Differentiis, Synonymis, Locis. Tomus I. Digitalized version: https://play.google.com/books/reader?id=n_1cAAAAcAAJ&hl=sv&lr=&printsec=frontcover&pg=GBS.PP7

83

Linné, C. A. (1788b) Systema Naturae. Tom. I Pars III. Digitalized version: https://play.google.com/books/reader?id=IbfwXWnc0uAC&hl=sv&printsec=frontcover&pg=GBS.PA1031

Liu, Z.-P., Song, C., Wang, M., He, Y., Xu, X.-B., Pan, H.-Q., et al. (2014) Chronic stress impairs GABAergic control of amygdala through suppressing the tonic GABAA receptor currents. Mol. Brain 7:32

Lukasiewicz, P. D., Eggers, E. D., Sagdullaev, B. T., McCall, M. A: (2004) GABAC receceptor-mediated inhibition in the retina. Vision Res. 44(28):3289-3296

Maguire, J., Mody, I. (2007) Neurosteroid Synthesis-Mediated Regulation of GABAA Receptors: Relevance to the Ovarian Cycle and Stress. J. Neurosci. 27(9):2155-2162

Maricq, A. V., Peterson, A. S., Brake, A. J., Myers, R. M., Julius, D. (1991) Primary Structure and Functional Expression of the 5HT3 Receptor, a Serotonin-Gated Ion Channel. Science 254(5030):432-437

Marc, R. E., Cameron, D. (2002) A molecular phenotype atlas of the zebrafish reti-na. J. Neurocytol. 30(7):593-654

Marowsky, A., Fritschy, J.-M., Vogt, K. E. (2004) Functional mapping of GABAA receptor subtypes in the amygdala. Eur. J. Neurosci. 20(5):1281-1289

Marowksy, A., Rudolph, U., Fritschy, J.-M., Arand, M. (2012) Tonic Inhibition in Principal Cells of the Amygdala: A Central Role for α3 Subunits-Containing GABAA Receptors. J. Neurosci. 32(25):8611-8619

Martin, D. L. (1987) Regulatory Properties of Brain Glutamate Decarboxylase. Cell. Mol. Neurobiol. 7(3):237-253

Martyniuk, C. J., Awad, R. Hurley, R. Finger, T. E., Trudeau, V. L. (2007) Glutamic acid decarboxylase 65, 67, and GABA-transaminase mRNA expression and total enzyme activity in the goldfish (Carassius auratus) brain. Brain Res. 1147:154-166

Maruska, K. P., Butler, J. M., Field, K. E., Porter, D. T. (2017) Localization of Glu-tamatergic, GABAergic, and Cholinergic Neurons in the Brain of the African Cichlid Fish, Astatotilapia burtoni. J. Comp. Neurol. 525(3):610-638

McClure, M. M., McIntyre, P. B., McCune, A. R. (2006) Notes on the natural diet and habitat of eight danionin fishes, including the zebrafish Danio rerio. J. Fish. Biol. 69(2):553-570

McIntire, S. L., Reimer, R. J., Schuske, K., Edwards, R. H., Jorgensen, E. M. (1997) Identification and characterization of the vesicular GABA transporter. Nature 389(6653):870-876

Meek, J., Nieuwenhuys, R. (1998) Chapter 15: Holosteans and Telesots. In: The Central Nervous System of Vertebrates, vol. II, pgg. 759-937, © Springer-Verlag Berlin Heidelberg

Miesenböck, G., De Angelis, D. A., Rothman, J. E. (1998) Visualizing secretion and synaptic transmission with pH-sensitive green fluorescent proteins. Nature 394(6689):192-195

Miller, P. S., Aricescu, A. R. (2014) Crystal structure of a human GABAA receptor. Nature 512(7514):270-275

Miller, P. S., Scott, S., Masiulus, S., De Colibus, L., Pardon, E., Steyaert, J., et al. (2017) Structural basis for GABAA receptor potentiation by neurosteroids. Nat. Struct. Mol. Biol. 24(11):986-992

Minelli, A., DeBiasi, S., Brecha, N. C., Zuccarello, L. V., Conti, F. (1996) GAT-3, a High-Affinity GABA Plasma Membrane Transporter, Is Localized to Astrocytic

84

Processes, and It Is Not Confined to the Vicinity of GABAergic Synapses in the Cerebral Cortex. J. Neurosci. 16(19):6255-6264

Miyazawa, A., Fujiyoshi, Y., Unwin, N. (2003) Structure and gating mechanism of the acetylcholine receptor pore. Nature 423(6943):949-955

Moczydlowski, E. G. (2012) Chapter: Synaptic Transmission and the Neuromuscu-lar Junction. In: Medical Physiology, 2nd edition, W. F. Boron and E. L. Boul-paep (eds.), pgg. 212-236, © Saunders Elsevier, Philadelphia

Mookerjee, S. A., Goncalves, R. L. S., Gerencser, A. A., Nicholls, D. G., Brand, M. D. (2015) The contributions of respiration and glycolysis to extracellular acid production. Biochim. Biophys. Acta 1847(2):171-181

Moser, C. C., Farid, T. A., Chobot, S. E., Dutton, P. L. (2006) Electron tunneling chain of mitochondria. Biochim. Biophys. Acta 1757(9-10):1096-1109

Neelands, T. R., Macdonald, R. L. (1999) Incorporation of the π Subunit into Func-tional γ-Aminobutyric AcidA Receptors. Mol. Pharmacol. 56(3):598-610

Nelson, R., Bender, A. M., Connaughton, V. P. (2008) Transporter-mediate GABA responses in horizontal and bipolar cells of zebrafish retina. Vis. Neurosci. 25(2):155-165

Nemecz, Á., Prevost, M. S., Menny, A., Corringer, P.-J. (2016) Emerging Molecular Mechanisms of Signal Transduction in Pentameric Ligand-Gated Ion Channels. Neuron 90(3):452-470

Nieuwenhuys, R. (2011) The development and general morphology of the telen-cephalon of actinopterygian fishes: synopsis, documentation and commentary. Brain Struct. Funct. 215(3-4):141-157

Nusser, Z., Roberts, J. D. B., Baude, A., Richards, J. G., Somogyi, P. (1995) Rela-tive Densities of Synaptic and Extrasynaptic GABAA Receptors on Cerebellar Granule Cells As Determined by a Quantitative Immunogold Method. J. Neuro-sci. 15(4):2948-2960

Nusser, Z., Sieghart, W., Somogiy, P. (1998) Segregation of Different GABAA Re-ceptors to Synaptic and Extrasynaptic Membranes of Cerebellar Granule Cells. J. Neurosci. 18(5):1693-1703

Okamura, H., Abitbol, M., Julien, J.-F., Dumas, S., Bérod, A., Geffard, M., et al. (1990) Neurons containing messenger RNA encoding glutamate decarboxylase in rat hypothalamus demonstrated by in situ hybridization, with special emphasis on cell groups in medial preoptic area, anterior hyoothalamic area and dorsome-dial hypothalamic nucleus. Neuroscience 39(3):675-699

Olsen, R. W., Sieghart, W. (2008) International Union of Pharmacology. LXX. Subtypes of γ-Aminobutyric AcidA Receptors: Classification on the Basis of Subunit Composition, Pharmacology, and Function. Update. Pharmacol. Rev. 60(3):243-260

Papez, J. W. (1937) A proposed mechanism of emotion. Arch. Neurol. Psychiat. 38:725-743

Passarella, S., Atlante, A., Barile, M., Quagliariello, E. (1984) Carrier mediated GABA translocation into rat brain mitochondria. Biochem. Biophys. Res. Com-mun. 121(3):770-778

Pasternack, M., Bountra, C., Voipio, J., Kaila, K. (1992) Influence of extracellular and intracellular pH on GABA-gated chloride conductance in crayfish muscle fi-bers. Neuroscience 47(4):921-929

85

Pauckert, M., Sidi, S., Russell, C., Siba, M., Wilson, S. W., Nicolson, T., et al. (2004) A Family of Acid-Sensing Ion Channels from the Zebrafish. J. Biol. Chem. 279(18):18783-18791

Payne, J. A., Rivera, C., Voipio, J., Kaila, K. (2003) Cation-chloride co-transporters in neuronal communication, development and trauma. Trends Neurosci. 26(4):199-206

Payne, J. A., Stevenson, T. J., Donaldson, L. F. (1996) Molecular Characterization of a Putative K-Cl Cotransporter in Rat Brain. J. Biol. Chem. 271(27):16245-16252

Portavella, M., Vargas, J. P., Torres, B., Salas, C. (2002) The effects of telencephalic pallial lesions on spatial, temporal, and emotional learning in goldfish. Brain Res. Bull. 57(3-4):397-399

Porter, T. G., Spink, D. C., Martin, S. B., Martin, D. L: (1985) Transaminations catalyzed by brain glutamate decarboxylase. Biochem. J. 231(3):705-712

Radian, R., Bendahan, A., Kanner, B. I. (1986) Purification and Identification of the Functional Sodium- and Chloride-coupled γ-Aminobutyric Acid Transport Gly-coprotein from Rat Brain. J. Biol. Chem. 261(33):15437-15441

Rivera, C., Voipio, J., Payne, J. A., Ruusuvuori, E., Lahtinen, H., Lamsa, K., et al. (1999) The K+/Cl– co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature 397(6716):251-255

Roberts, E., Kuriyama, K. (1968) Biochemical-physiological correlations in studies of the γ-aminobutyric acid system. Brain Res. 8(1):1-35

Rodrigo, M. C., Martin, D. S., Redetzke, R. A., Eyster, K. M. (2002) A method for the extraction of high-quality RNA and protein from single small samples of ar-teries and veins preserved in RNAlater. J. Pharmacol. Toxicol. Methods 47(2):87-92

Rossi, D. J., Hamann, M. (1998) Spillover-Mediated Transmission at Inhibitory Synapses Promoted by High Affinity α6 Subunit GABAA Receptors and Glo-merular Geometry. Neuron 20(4)783-795

Ruusuvuori, E., Kaila, K. (2014) Chapter 14: Carbonic Anhydrases and Brain pH in the Control of Neuronal Excitability. In: Carbonic Anhydrase: Mechanism, Reg-ulation, Links to Disease, and Industrial Applications, 1st edition, S. C. Frost and R. McKenna (eds.), pgs. 271-290, © Springer Science + Business Media, Dor-drecht

Sah, P., Faber, S. L., Lopez de Armentia, M. Power, J. (2003) The Amygdaloid Complex: Anatomy and Physiology. Physiol. Rev. 83(3):803-834

Salas, C:, Broglio, C., Durán, E., Gómez, A:, Ocaña, F. M., Jímenez-Moya, F., et al. (2006) Neuropsychology of Learning and Memory in Teleost Fish. Zebrafish 3(2):157-171

Šali, A., Blundell, T. L. (1993) Comparative Protein Modelling by Satisfaction of Spatial Restraints. J. Mol. Biol. 234(3):779-815

Sawyer, L., James, M. N. G. (1982) Carboxyl-carboxylate interactions in proteins. Nature 295(5844):79-80

Scharen-Wiemers, N., Gerfin-Moser, A. (1993) A single protocol to detect tran-scripts of various types and expression levels in neural tissue and cultured cells: in situ hybridization using digoxigenin-labelled cRNA probes. Histochemistry 100(6):431-440.

86

Schofield, P. R., Darlison, M. G., Fujita, N., Burt, D. R., Stephenson, F. A., Rodri-guez, H., et al. (1987) Sequence and functional expression of the GABAA recep-tor shows a ligand-gated receptor super-family. Nature 328(6127):221-227

Seeburg, P. H., Wisden, W., Verdoorn, T. A., Pritchett, D. B., Werner, P., Herb, A., et al. (1990) The GABAA Receptor Family: Molecular and Functional Diversity. Cold Spring Harb. Symp. Quant. Biol. LV:29-40

Sheikh, S. N., Martin, S. B., Martin, D. L. (1999) Regional distribution and relative amounts of glutamate decarboxylase isoforms in rat and mouse brain. Neuro-chem. Int. 35(1):73-80

Simon, J., Wakimoto, H., Fujita, N., Lalande, M., Barnard, E. A. (2004) Analysis of the Set of GABAA Receptor Genes in the Human Genome. J. Biol. Chem. 279(40):41422-41435

Sinning, A., Hübner, C. A. (2013) Minireview: pH and synaptic transmission. FEBS Lett. 587(13):1923-1928

Sobotta, J. (2011) Content 12: Brain and Spinal Cord. In: Sobotta, Atlas of Human Anatomy: Head, Neck, and Neuroanatomy, 15th edition, F. Paulsen and J. Waschke (eds.), pgg. 212-342, © Elsevier GmbH, Munich

Somogyi, P., Fritschy, J.-M., Benke, D., Roberts, J. D. B., Sieghart, W. (1996) The γ2 Subunit of the GABAA Receptor is Concentrated in Synaptic Junctions Con-taining the α1 and β2/3 Subunits in Hippocampus, Cerebellum and Globus Palli-dus. Neuropharmacology 35(9-10):1425-1444

Spence, R., Fatema, M. K., Reichard, M., Huq, K. A., Wahab, M. A., Ahmed, Z. F., Smith, C. (2006) The distribution and habitat preferences of the zebrafish in Bangladesh. J. Fish. Biol. 69(5):1435-1448

Spence, R., Gerlach, G., Lawrence, C., Smith, C. (2011) Content III.: (1) Distribu-tion and habitat. In: The behaviour and ecology of the zebrafish, Danio rerio, pgg. 16-18. Biol. Rev. Camb. Philos. Soc. 83(1):13-34

Storici, P., De Biase, D., Bossa, F., Bruno, S., Mozzarelli, A., Peneff, C., et al. (2004) Structures of γ-Aminobutyric Acid (GABA) Aminotransferase, a Pyri-doxal 5’-Phosphate, and [2Fe-2S] Cluster-containing Enzyme, Complexed with γ-Ethynyl-GABA and with the Antiepilepsy Drug Vigabatrin. J. Biol. Chem. 279(1):363-373

Streisinger, G., Walker, C., Dower, N., Knauber, D., Singer, F. (1981) Production of clones of homozygous diploid zebra fish (Brachydanio rerio). Nature 291(5813):293-296

Sun, F., Huo, X., Zhai, Y., Wang, A., Xu, J., Su, D., et al. (2005) Crystal Structure of Mitochondrial Respiratory Membrane Protein Complex II. Cell 121(7):1043-1057

Takeuchi, A., Takeuchi, N. (1967) Anion permeability of the inhibitory post-synaptic membrane of the crayfish neuromuscular junction. J. Physiol. 191(3):575-590

Tappaz, M. L., Brownstein, M. J., Kopin, I. J. (1977) Glutamate decarboxylase (GAD) and g-aminobutyric acid (GABA) in discrete nuclei of hypothalamus and substantia nigra. Brain Res. 125(1):109-121

Tong, C.-K., Brion, L. P., Suarez, C., Chesler, M. (2000) Interstitial Carbonic An-hydrase (CA) Activity in Brain Is Attributable to a Membrane-Bound CA Type IV. J. Neurosci. 20(22):8247-8253

Unwin, N. (2005) Refined Structure of the Nicotinic Acetylcholine Receptor at 4 Å Resolution. J. Mol. Biol. 346(4):967-989

87

Usunoff, K. G., Schmitt, O., Itzev, D. E., Haas, S. J.-P., Lazarov, N. E., Rolfs, A., Wree, A. (2009) Efferent Projections of the Anterior and Posterodorsal Regions of the Medial Nucleus of the Amygdala in the Mouse. Cell Tissues Organs 190(5):256-285

Vandepoele, K., De Vos, W., Taylor, J. S., Meyer, A., Van de Peer, Y. (2004) Major events in the genome evolution of vertebrates: Paranome age and size differ con-siderably between ray-finned fishes and land vertebrates. Proc. Natl. Acad. Sci. U. S. A. 101(6):1638-1643

Verkuyl, J. M., Hemby, S. E., Joëls, M. (2004) Chronic stress attenuates GABAergic inhibition and alters gene expression of parvocellular neurons in rat hypothala-mus. Eur. J. Neurosci. 20(6):1665-1673

Waldmann, R., Champigny, G., Bassilana, F., Heurteaux, C., Lazdunski, M. (1997) A proton-gated cation channel involved in acid sensing. Nature 386(6621):173-177

Wang, M. (2011) Neurosteroids and GABA-A receptor function. Front. Endocrinol (Lausanne) 2:44

Wang, Q., Verweij, E. W. E., Krugers, H. J., Joels, M., Swaab, D. F., Lucassen, P. J. (2014) Distribution of the glucocorticoid receptor in the human amygdala; changes in mood disorder patients. Brain Struct. Funct. 219(5):1615-1626

Webb, B., Sali, A. (2014) Comparative Protein Structure Modeling Using MODEL-LER. Curr. Protoc. Bioinformatics 47:5.6.1-5.6.32

Wei, W., Zhang, N., Peng, Z., Houser, C. R., Mody, I. (2003) Perisynaptic Localiza-tion of δ Subunit-Containing GABAA Receptors and Their Activation by GABA Spillover in the Mouse Dentate Gyrus. J. Neurosci. 23(33):10650-10661

Wemmie, J. A. (2011) Neurobiology of panic and pH chemosensation in the brain. Dialogues Clin. Neurosci. 13(4):475-483

Wemmie, J. A., Askwith, C. C., Lamani, E., Cassell, M. D., Freeman Jr., J. H., Welsh, M. J. (2003) Acid-Sensing Ion Channel 1 Is Localized in Brain Regions with High Synaptic Density and Contributes to Fear Conditioning. J. Neurosci. 23(13):5496-5502

Wemmie, J. A., Chen, J., Askwith, C. C., Hruska-Hageman, A. M., Price, M. P., Nolan, B. C., et al. (2002). The Acid-Sensing Ion Channel ASIC Contributes to Synaptic Plasticity, Learning and Memory. Neuron 34(3):463-477

Wemmie, J. A., Coryell, M. W., Askwith, C. C., Lamani, E., Leonard, A. S., Sig-mund, C. D., et al. (2004) Overexpression of acid-sensing ion channel 1a in transgenic mice increases acquired fear-related behavior. Proc. Natl. Acad. Sci. U. S. A. 101(10):3621-3626

Wemmie, J. A., Price, M. P., Welsh, M. J. (2006) Acid-sensing ion channels: ad-vances, questions and therapeutic opportunities. Trends Neurosci. 29(10):578-586

Whiting, P., McKernan, R. M., Iversen, L. (1990) Another mechanism for creating diversity in γ-aminobuyrate type A reeptors: RNA splicing directs expression of two forms of γ2 subunit, one of which contains a protein kinase C phosphoryla-tion site. Proc. Natl. Acad. Sci. U. S. A. 87(24):9966-9970

Winberg, S., Höglund, E., Øverli, Ø. (2016) Chapter 2: Variation in the Neuroendo-crine Stress Response. In: Biology of Stress in Fish, C. B. Schreck, L. Tort, A. P. Farrell, C. J. Brauner (eds.), pgg. 35-74, © Elsevier Inc.

88

Wisden, W., Laurie, D. J., Monyer, H., Seeburg, P. H. (1992) The Distribution of 13 GABAA Receptor Subunit mRNA in the Rat Brain. I. Telencephalon, Dien-cephalon, Mesencephalon. J. Neurosci. 12(3):1040-1062

Wohlfarth, K. M., Bianchi, M. T., Macdonald, R. L. (2002) Enhanced Neurosteroid Potentiation of Ternary GABAA Receptors Containing the δ Subunit. J. Neuro-sci. 22(5):1541-1549

Wojcik, S. M., Katsurabayashi, S., Guillemin, I., Friauf, E., Rosenmund, C., Brose, N., Rhee, J.-S. (2006) A Shared Vesicular Carrier Allows Synaptic Correlease of GABA and Glycine. Neuron 50(4):575-587

Wojcik, W. J., Neff, N. H. (1984) γ-Aminobutyric Acid B Receptors Are Negatively Coupled to Adenylate Cyclase in Brain, and in the Cerebellum Those Receptors May Be Associated with Granule Cells. Mol. Pharmacol. 25(1):24-28

Wootton, R. J. (1984) Chapter 2: Spatial distribution. In: A functional biology of sticklebacks, 1st edition, P. Calow (ed.), pgg. 4-19, Springer, New York

Wootton, R. J. (2009) The Darwinian stickleback Gasterosteus aculeatus: a history of evolutionary studies. J. Fish. Biol. 75(8):1919-1942

Ziemann, A. E., Allen, J. E., Dahdaleh, N. S., Drebot, I. I., Coryell, M. W., Wunsch, A. M., et al. (2009) The Amygdala Is a Chemosensor that Detects Carbon Diox-ide and Acidosis to Elicit Fear Behavior. Cell 139(5):1012-1021

Zomot, E., Bendahan, A., Quick, M., Zhao, Y., Javitch, J. A., Kanner, B. I. (2007) Mechanism of chloride interaction with neurotransmitter:sodium symporters. Nature 449(7163):726-730

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1466

Editor: The Dean of the Faculty of Medicine

A doctoral dissertation from the Faculty of Medicine, UppsalaUniversity, is usually a summary of a number of papers. A fewcopies of the complete dissertation are kept at major Swedishresearch libraries, while the summary alone is distributedinternationally through the series Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty ofMedicine. (Prior to January, 2005, the series was publishedunder the title “Comprehensive Summaries of UppsalaDissertations from the Faculty of Medicine”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-348421

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2018