The American Journal of Pathology, Vol. 183, No. 6, December 2013

ajp.amjpathol.org

MOLECULAR PATHOGENESIS OF GENETIC AND INHERITED DISEASES

Def-6, a Novel Regulator of Small GTPases in Podocytes,Acts Downstream of Atypical Protein Kinase C (aPKC) l/iKirstin Worthmann,* Michael Leitges,y Beina Teng,* Marcello Sestu,z Irini Tossidou,* Thomas Samson,x Hermann Haller,*Tobias B. Huber,{k and Mario Schiffer*

From the Division of Nephrology,* Department of Medicine, Hannover Medical School, Hannover, Germany; the Biotechnology Centre of Oslo,y University ofOslo, Oslo, Norway; Faculty of Medicine,z Interdisciplinary Centre for Clinical Research (IZKF) Leipzig, University of Leipzig, Leipzig, Germany; Departmentof Cell and Developmental Biology,x Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina; the Renal Division,{

University Hospital Freiburg, Freiburg, Germany; and the BIOSS Centre for Biological Signalling Studies,k Albert-Ludwigs-University, Freiburg, Germany

Accepted for publication

C

P

h

August 26, 2013.

Address correspondence toMario Schiffer, M.D., Divisionof Nephrology, Department ofMedicine, Hannover MedicalSchool, Carl-Neuberg-Str 1,30625 Hannover, Germany.E-mail: schiffer.mario@mh-hannover.de.

opyright ª 2013 American Society for Inve

ublished by Elsevier Inc. All rights reserved

ttp://dx.doi.org/10.1016/j.ajpath.2013.08.026

The atypical protein kinase C (aPKC) isotypes PKCl/i and PKCz are both expressed in podocytes; however,little is known about differences in their function. Previous studies in mice have demonstrated thatpodocyte-specific loss of PKCl/i leads to a severe glomerular phenotype, whereas mice deficient in PKCzdevelop no renal phenotype. We analyzed various effects caused by PKCl/i and PKCz deficiency in culturedmurine podocytes. In contrast to PKCz-deficient podocytes, PKCl/i-deficient podocytes exhibited a severeactin cytoskeletal phenotype, reduced cell size, decreased number of focal adhesions, and increased acti-vation of small GTPases. Comparative microarray analysis revealed that the guanine nucleotide exchangefactor Def-6 was specifically up-regulated in PKCl/i-deficient podocytes. In vivo Def-6 expression issignificantly increased in podocytes of PKCl/i-deficient mice. Cultured PKCl/i-deficient podocytes exhibitedan enhanced membrane association of Def-6, indicating enhanced activation. Overexpression of aPKCl/i inPKCl/i-deficient podocytes could reduce the membrane-associated expression of Def-6 and rescue the actinphenotype. In the present study, PKCl/i was identified as an important factor for actin cytoskeletal regu-lation in podocytes and Def-6 as a specific downstream target of PKCl/i that regulates the activity of smallGTPases and subsequently the actin cytoskeleton of podocytes. (Am J Pathol 2013, 183: 1945e1959; http://dx.doi.org/10.1016/j.ajpath.2013.08.026)

Supported by a grant from the Else Kröner-Fresenius-Stiftung Founda-tion and by grant EXC 294 from the Excellence Initiative of the GermanFederal and State Governments (T.B.H.); grants Schi 587/3, 587/4, and587/6 and grant SFB566 (M.S.); and grant GM029860 to Keith Burridge(principal investigator of T.S.) from the NIH.

T.B.H. and M.S. contributed equally to this work as senior authors.

The protein kinase C (PKC) family consists of 10 serine/threonine kinases, which can be divided into three sub-families on the basis of their different regulatory domains.Conventional PKCs (PKCa, PKCbI, PKCbII, and PKCg)and novel PKCs (PKCd, PKCq, PKCh, and PKCε) can beactivated by calcium or diacylglycerol (DAG) via C1 andC2 domains. The third subfamily, atypical PKCs (aPKCs)[PKCl/i (PKCl in mouse and PKCi in human) and PKCz],lack the calcium-sensitive C2 domain and possess an atyp-ical C1 domain that does not bind DAG but binds phos-phatidylinositol 3,4,5-triphosphate (PIP3) or ceramideinstead.1 The amino acid sequences of PKCl/i and PKCzare 72% identical2; however, the functions of these twoproteins in vivo seem to be distinct. Because completeknockout of PKCl/i in mice is lethal before embryonic day9,3 we created a podocyte-specific PKCl/i knockout(PKCl/i�/�) mouse by mating mice with a floxed PKCl/i

stigative Pathology.

.

gene and Podocin-Cre mice4 with expression of the Crerecombinase dependent on the podocyte-specific podocinpromoter. Podocyte-specific deletion of PKCl/i led to asevere glomerular phenotype with glomerulosclerosis, pro-teinuria, and death at age 4 to 5 weeks.5 In contrast toPKCl/i, complete knockout of PKCz (PKCz�/�) is rathermild. PKCz�/� mice develop no obvious renal phenotypebut exhibit changes in their secondary lymphatic organs as aresult of a disturbed NF-kB pathway.6 Recently, we showedthat double-knockout of aPKCs in podocytes leads to a

Worthmann et al

developmental phenotype with no formation of a normalfoot processes network, defective glomerular maturation,incomplete capillary formation, and mesangiolysis.7 Thesevere podocyte foot process effacement in the podocyte-specific PKCl/i�/� mice and the inability to form foot pro-cesses in the double-knockout mice suggests an importantrole for the actin cytoskeleton of podocytes. The actin cyto-skeleton of podocytes is a highly dynamic and complexstructure8e11 that also has an important role in maintenanceof the slit diaphragm in various renal diseases.12 When theactin cytoskeleton of podocyte foot processes changes fromparallel bundles of actin13 to a dense network of short fila-ments, foot process effacement and, in turn, proteinuria candevelop.14 A group of key regulators for actin cytoskeletalregulation are the small GTPases, which are molecularswitches that change between a GDP-bound inactive formand a GTP-bound active form. Guanine nucleotide exchangefactors (GEFs) control the GDP/GTP exchange by increasingthe exchange rate, which leads to activation of small GTPa-ses.15 Recent studies have demonstrated the critical involve-ment of small GTPases in regulation of podocyte actincytoskeleton in normal podocyte homeostasis and in acquiredand genetic glomerular diseases.16e19 Herein we demonstratethat differentially expressed FDCP-6 (Def-6), a GEF for theRho GTPases Rac1 and Cdc42,20 is differentially affected bythe aPKC isoforms l/i and z in podocytes. Def-6, also knownas swap70-like adaptor of T cells or IRF-4ebinding protein,consists of an EF-hand motif at the N terminus, a pleckstrinhomology domain, and a Dbl homologyelike domain (DHL)at its C terminus. Def-6 contains 631 amino acids, has apredicted molecular weight of 74 kD,21 and exhibits muchhomology to the B cellespecific GEF swap70. To date, mostinvestigations of the function or expression of Def-6 havebeen performed in B or T cells because Def-6 is highlyexpressed in lymphatic organs.21 Herein we report for the firsttime an important role for Def-6 in a renal epithelial cell type.

Materials and Methods

Plasmids, Antibodies, and Reagents

The plasmids Def-6 FL (bp 1e2015 in pEGFP-C2) and Def-6 DHL (bp 1066e1934 in pEGFP-C3) have been used inprevious studies.22 Primary antibodies used were mousemonoclonal anti-vinculin (clone hVIN-1) (Sigma-AldrichCorp., St. Louis, MO), mouse anti-CD29 (integrin b1) (BDTransduction Laboratories, BD Biosciences, San Jose, CA),goat anti-podocalyxin (R&D Systems GmbH, Wiesbaden,Germany), mouse monoclonal anti-RhoA, mouse monoclonalanti-Rac1, mouse monoclonal anti-Cdc42 (Cytoskeleton,Inc., Denver, CO), mouse monoclonal antieDef-6 (AbnovaCorp., Taipei, Taiwan), affinity purified rabbit antieDef-6(according to a previous study23), rabbit antiserum anti-Gia3(Upstate Biotechnology, Lake Placid, NY), rabbit polyclonalantieglyceraldehyde-3-phosphate dehydrogenase (GAPDH)(Santa Cruz Biotechnology, Inc., Santa Cruz, CA), rabbit

1946

polyclonal anti-Sp1 (Santa Cruz Biotechnology), rat mono-clonal anti-entactin/nidogen (Millipore Corp., Billerica, MA),and rabbit monoclonal anti-PKCl/i (Cell Signaling Tech-nology, Inc., Beverly, MA). Polyclonal antibody specific forPKCz was raised in rabbits and has been described previ-ously.24 All secondary antibodies used for immunostainingwere obtained from Invitrogen Corp. (Carlsbad, CA) andincluded rabbiteAlexa Fluor 488, mouseeAlexa Fluor 488,goateAlexa Fluor 488, rabbiteAlexa Fluor 555, mouseeAlexa Fluor 555, and rateAlexa Fluor 488. All secondaryantibodies used for Western blotting were obtained fromSanta Cruz Biotechnology and included goat anti-mouseIgGehorseradish peroxidase and goat anti-rabbit IgG horse-radish peroxidase. Further reagents included Alexa Fluor 546phalloidin (Invitrogen, Darmstadt, Germany), PKCi siRNA,control siRNA (Santa Cruz Biotechnology), aPKC pseudo-substrate and scrambled peptide (BioSource International,Inc., Camarillo, CA), and Def6 (human) recombinant protein(Abnova).

Podocyte-Specific aPKCl/i Knockout Mice

Generation of podocyte-specific aPKCl/i knockout miceusing floxed PKCl/i and Podocin-Cre mice has beendescribed previously.5 Podocyte-specific aPKCl/i hetero-zygous mice were crossed with Immorto transgenic mice.25

Homozygous podocyte-specific aPKCl/i mice harboring theImmorto transgene were used for generation of immortal-ized podocyte cell lines.

Isolation of Glomeruli and Establishment ofImmortalized Mouse Podocyte Cell Lines

Glomeruli were isolated from kidneys of 4-week-old miceusing a sequential sieving technique as described previ-ously.26 Only stable cell lines derived from these glomerulithat were positive for the podocyte-specific markers WT-1and synaptopodin were used in the experiments.

Podocyte Culture

Immortalized murine podocytes were cultured as describedby Mundel et al.27 In brief, cells were cultured on collagenIecoated flasks using RPMI 1640 medium containing 10%fetal calf serum, 1% penicillin/streptomycin, and 10 U/mLg-interferon at 33�C for proliferation. For differentiation,cells were cultured at 37�C without g-interferon for 10 to 14days. For stimulation experiments, cells were serum starvedwith 1% fetal calf serum overnight and then treated with 10mmol/L aPKC pseudosubstrate for 2 hours at 37�C or 10mmol/L scrambled peptide, respectively.

Live Cell Microscopy and Cell Movement

Differentiated deficient and control cells were observedunder a microscope (Axio Observer.Z1; Zeiss Microscopy

ajp.amjpathol.org - The American Journal of Pathology

Def-6 Acts Downstream of aPKCl/i

GmbH, Göttingen, Germany) at �20 magnification for 6hours. Images were obtained automatically every 2 minutes.To quantify the movement of cells, all images were im-ported into ImageJ software version 1.44p. Using theManual Tracking plug-in, the nucleus of every selected cell(10 cells of each genotype) was marked on the imagesequence. The program measures the distance between twoframes, and by summing all distances, the distance coveredfrom the start of recording to the end, at 6 hours.

Transfection

The day before transfection, podocytes were seeded on glasscoverslips and cultured at 33�C overnight. Cells weretransfected using FuGENE HD transfection reagent (RocheDiagnostics GmbH, Mannheim, Germany) according to theprotocol recommended by the manufacturer. Transfectionwith PKCl/i siRNA or control siRNA (Santa CruzBiotechnology) was performed on 3-day differentiatedpodocytes using lipofectamine 2000 reagent (Invitrogen)according to the manufacturer’s protocol. After 6 hours inthe incubator, the medium was changed to normal growthmedium. After transfection, cells were cultured for 72 hoursat 37�C and fixed with 4% paraformaldehyde.

Adenoviral Production and Infection

Plasmids for adenovirus production were generated usingGateway technology (Invitrogen). Donor vectors weregenerated by amplifying Green fluorescent protein (GFP)etagged full-length human PKCi or GFP alone with specificprimers and cloning into the pDONR221 vector by using BPclonase II enzyme for recombination. Adenoviral constructswere generated using LR Clonase II enzyme, whichrecombines pDONR221 with pAd/CMV/V5-DEST. Allreactions were performed as recommended by the manu-facturer (Invitrogen). Adenoviral expression and amplifica-tion were performed in HEK293 cells. Podocytes wereinfected with equal amounts of PKCi-GFP or GFP adeno-viral supernatants. After 7 days, podocytes were eitherharvested for RNA isolation or fixed with 4% para-formaldehyde for immunostaining.

Polymerase Chain Reaction

Genomic DNA from cultured podocytes was prepared usingthe basicdna-OLS Kit (OMNI Life Science GmbH & Co.KG, Hamburg, Germany) following the protocol recom-mended by the manufacturer. As controls, genomic DNAsamples from tail biopsies were used. PCR was performedunder standard conditions in a Primus Thermocycler(MWG-Biotech AG, Ebersberg, Germany) with thefollowing primer pairs: Cre forward, 50-AGGTTCGTTCA-CTCATGGA-30; Cre reverse, 50-TGCACCAGTTTAGTT-ACCC-30; loxP forward, 50-TTGTGAAAGCGACTGGAT-TG-30; loxP reverse, 50-CTTGGGTGGAGAGGCTATTC-30;

The American Journal of Pathology - ajp.amjpathol.org

WT reverse, 50-AATGTTCATGTTCAACACTGCT-30;Highzeta5 forward, 50-GCCATCTCCAACAGCCACAG-30; Highzeta5ex, 50-CCGTTGGCTCGGTACAGCTT-30;and MO13, 50-CTTGGGTGGAGAGGCTATTC-30.

Immunofluorescence Staining

Immortalized cultured mouse podocytes were seeded on glasscoverslips and differentiated for 10 to 14 days. After fixationwith 4% paraformaldehyde, cells were permeabilized with0.1% Triton X-100, blocked in 10% normal donkey serum,and incubated overnight at 4�C with the primary antibodiesas indicated. After rinsing with PBS, the cells were incubatedwith a fluorophore-conjugated secondary antibody for 1 hourand mounted in medium containing DAPI.

For immunofluorescence staining of frozen kidneys, 6-mm sections (Leica CM3050S cryostat; Leica Microsystems,Wetzlar, Germany) were fixed in acetone, blocked with 10%normal donkey serum, and incubated with the primary an-tibodies, as indicated, overnight at 4�C. After rinsing withPBS, sections were incubated with conjugated secondaryantibodies for 1 hour at room temperature and mounted inmedium containing DAPI. Confocal images were obtainedusing a Leica Inverted-2 microscope with a 63� oil im-mersion objective.

Western Blot Analysis

After 10 to 14 days of differentiation, cells were lyzed on icein radioimmunoprecipitation assay buffer [50 mmol/L Tris(pH 7.5), 150 mmol/L NaCl 0.5% sodium deoxycholate, 1%nonidet P-40, and 0.1% SDS] supplemented with proteaseinhibitors (cOmplete, Mini; Roche Diagnostics), 1 mmol/Lsodium orthovanadate, 50 mmol/L NaF, and 200 mg/Lokadaic acid. The lysates were centrifuged for 15 minutes at12,000 rpm and 4�C. Aliquots of the supernatants (10 mgprotein per lane) were separated via 10% or 12% SDS-PAGE and transferred to polyvinylidene difluoride mem-brane (Immobilon-P; Millipore). Protein was quantifiedusing the BCA Protein Assay Kit (Pierce Chemical Co.,Thermo Fisher Scientific, Inc., Rockford, IL). After incu-bation with primary and horseradish peroxidaseelabeledsecondary antibody, the complexes were visualized usingSuperSignal West Pico Chemiluminescent Substrate (PierceChemical Co.) according to the manufacturer’s protocol.

Subcellular Fractionation

For subcellular fractionation, differentiated podocytes wereharvested using trypsin-EDTA, and the various cell frac-tions were isolated using the Subcellular Protein Fraction-ation Kit (Pierce Chemical Co.) according to the protocolrecommended by the manufacturer. The fractions of solublenuclear extract and chromatin-bound nuclear extract werepooled to achieve a generic nuclear fraction.

1947

Worthmann et al

Preparation of Plasma Membranes

Plasma membranes were separated using the MembraneProtein Extraction Kit (PromoCell GmbH, Heidelberg,Germany) according to the protocol recommended by themanufacturer. The plasma membrane fraction was dissolvedin 0.5% Triton X-100 in PBS. Protein was quantified usingPrecision Red Protein Assay Reagent (Cytoskeleton) and 10mg of each preparation separated via 10% SDS-PAGE.

Cell Size Measurements and Quantification of FocalAdhesions

Cell size measurements and quantification of focal adhe-sions were performed using ImageJ software (version1.44p), as described by Marg et al.28 On the basis of anti-vinculin stainings, grayscale 16-bit images were obtainedusing a Leica DMLB microscope (�40 magnification) andprocessed using ImageJ software. With the use of severalcommands, the program defines the cell edge and measuresthe cell area. By using the “thresholding” tool to define thecontacts and with the command “Subtract background,” theprogram creates a binary footprint of the adhesion sites. Tomeasure the number of focal adhesions all particles >0.1mm2 were counted using the “Analyze particles” command.

Activity Assays

For activity measurements of the small GTPases, we usedRac1, Rho, and Cdc42 Activation Assay Biochem Kits(Cytoskeleton). For each assay, 700 mg protein lysate wasincubated with PAK-PBD protein beads (Rac1 and Cdc42)or Rhotekin-RBD beads (RhoA) for 1 hour at 4�C. Afterwashing, bound proteins were separated via 12% SDS-PAGE and blotted on polyvinylidine difluoride membranes.After incubation with antibodies specific for Rac1, RhoA,and Cdc42 and the appropriate secondary antibodies,membranes were developed using SuperSignal West PicoChemiluminescent Substrate (Pierce Chemical Co.) ac-cording to the manufacturer’s protocol. For determination oftotal amounts of RhoA, Rac1, and Cdc42, 50 mg total pro-tein lysate was used for Western blotting.

Real-Time PCR

Total RNA was prepared using the RNeasy Mini Kit (Qia-gen GmbH, Hilden, Germany) following the protocol rec-ommended by the manufacturer with an additional step ofDNase digestion (RNase-Free DNase Set; Qiagen). TotalRNA, 1 mg, was reverse transcribed using Oligo(dT)15 andrandom primers and Moloney Murine Leukemia VirusReverse Transcriptase (Promega GmbH, Mannheim, Ger-many). Real-time PCR was performed using a Light Cycler480 (Roche Diagnostics). cDNA was amplified using FastStart Taq Polymerase (Roche Diagnostics), SYBR Green(Invitrogen), gene-specific primers, and the following PCR

1948

conditions: 5 minutes at 95�C for 45 cycles, and 10 secondsat 95�C, 10 seconds at 60�C, and 10 seconds at 72�C.Specificity of the amplification product was verified viamelting curve analysis.The samples were measured as multiplexed reactions and

normalized to the constitutive gene mouse hypoxanthinephosphoribosyltransferase 1.All primers for the listed transcripts were designed using

Primer3 software: PKCi: 50-AGGAACGATTGGGTTGT-CAC-30, 50-GGCAAGCAGAATCAGACACA-30; PKCz:50-GCCTCCCTTCCAGCCCCAGA-30, 50-CACGGACTC-CTCAGCAGACAGCA-30; Def-6: 50-CACCAACGTGA-AACACTGGA-30, 50-TGGTGGTGGGTCGCTTAT-30; andHPRT-1: 50-CAGTCCCAGCGTCGTGATTA-30, 50-AGCA-AGTCTTTCAGTCCTGTC-30

Microarray-Based mRNA Expression Analysis (Single-Color Mode)

The Whole Mouse Genome Oligo Microarray (G4122F, ID014868; Agilent Technologies, Inc., Santa Clara, CA) usedin the present study contains 45,018 oligonucleotide probescovering the entire murine transcriptome. Synthesis of Cy3-labeled cRNA was performed using the Quick Amp La-beling kit, one color (No. 5190-0442; Agilent Technologies)according to the manufacturer’s recommendations. cRNAfragmentation, hybridization, and washing steps were alsoperformed exactly as recommended, using the One-colorMicroarray-Based Gene Expression Analysis Protocolversion 5.7. Slides were scanned using the Agilent MicroArray Scanner G2565CA at two different photomultipliertube settings (100% and 5%) to increase the dynamic rangeof the measurements (extended dynamic range mode). Dataextraction was performed using Feature Extraction Softwareversion 10.7.1.1 by using the recommended default extrac-tion protocol file: GE1_107_sep09.xml. Processed intensityvalues of the green channel (gProcessedSignal, or gPS) werenormalized via global linear scaling: All gPS values of onesample were multiplied by an array-specific factor. Thisscaling factor was calculated by dividing a reference 75thpercentile value (set as 1500 for the entire series) by the75th percentile value of the particular microarray (Array i inthe formula below). Accordingly, normalized gPS values forall samples (microarray data sets) were calculated using thefollowing formula: Normalized gPSArray i Z gPSArray i �(75th PercentileReference Array/75th PercentileArray i).A lower intensity threshold was defined as 1% of the

reference 75th percentile value (threshold Z 15). All ofthose normalized gPS values that fell below this intensityborder were substituted for by the respective surrogate valueof 15. Calculation of ratio values of relative gene expressionwas performed using Excel macros.Heat maps were generated using the MultiExperiment

Viewer 4.5.1 program. The intensity values of all geneswere logarithmized, and the mean intensity value of eachsingle gene was subtracted from all values. These data were

ajp.amjpathol.org - The American Journal of Pathology

Def-6 Acts Downstream of aPKCl/i

loaded into the MultiExperiment Viewer and clustered usingthe command “Hierarchical clustering” and the adjustments“Euclidean distance” and “Average linkage clustering.”

Quantification of Def-6 Expression in Glomeruli

For quantification of Def-6 expression, cryosections of wild-type (WT) and knockout kidneys were double stained withnidogen and Def-6. Def-6 expression in the podocytic areaof glomeruli was rated using a semiquantitative score (0 Znone, 1Z less, 2Z median, 3Z enhanced, 4Z strong) bya scientist blinded to the study (M.S.). From each genotype,40 glomeruli from four animals were scored. The podocyticarea was determined by subtraction of the mesangial areamarked by nidogen and the Def-6epositive area in theglomeruli.

Statistical Analysis

Data are given as means � SD and were compared usingunpaired Student’s t-tests. Data analysis was performedusing Excel statistical software. Significant differences wereaccepted at P < 0.05.

Results

Isolation and Characterization of PKCl/i- and PKCz-Deficient Podocytes

In a previous study, we identified PKCl/i as a critical factorfor podocyte polarity and architecture.5 To generate a PKCl/i-deficient podocyte cell line, we crossed PKCl/i flox/floxPodocin-Cre mice with Immorto transgenic animals ex-pressing the temperature-sensitive SV40 large T antigen(tsA58TAg) under control of the g-interferoneinducible H-2Kb promoter.25 Glomeruli were extracted from deficient andcontrol mice, outgrowing cells were isolated, and cell lineswere cultured from single cells. To compare the two atypicalPKC isoforms PKCl/i and PKCz in cell culture experiments,we also generated monoclonal cell lines from mice deficientin PKCz6 and control mice of the same strain. Three celllines of each genotype were characterized using PCR anal-ysis and various podocyte-specific antibodies (SupplementalFigure S1). After differentiation, podocyte marker proteinssynaptopodin29 and WT-130 were detected in all cell linesvia immunofluorescence staining, and expression of nephrin,podocin, and WT-1 were confirmed in cell lysates viaWestern blot analysis (data not shown).

PKCl/i Deficiency Influences the Actin Cytoskeletonand Motility of Podocytes

Under permissive conditions at 33�C, WT podocytes usuallygrow in a confluent cobblestone-like pattern. After 10 to 14days of differentiation, podocytes spread out and show anarborized shape.27 We observed this expected appearance

The American Journal of Pathology - ajp.amjpathol.org

under permissive (33�C) and nonpermissive (37�C) condi-tions in WT but not in PKCl/i�/� podocytes. PKCl/i�/�

podocytes exhibited a more spindle-like shape and grew inclusters, never reaching confluence at 33�C (data not shown).After differentiation at 37�C, PKCl/i�/� podocytes werenotably smaller and possessed longer cell protrusions than didcontrol cells. PKCz�/� podocytes showed a phenotypesimilar to that of WT cells (Figure 1A). Live cell imaging wasused to record the cell movement of WT and deficient cellsfor 6 hours, with an image obtained every 2 minutes. Wemeasured the distance moved for 10 cells per genotype(Figure 1B) by marking and following the cell nucleus in theimage sequence using ImageJ software. PKCl/i�/� podo-cytes moved a substantially longer distance within 6 hours ofmeasurement in comparison with WT cells. The distancemoved by PKCz�/� podocytes was similar to that of WTcells (Figure 1C). We hypothesized that the irregular cellshape and altered movement of PKCl/i�/� podocytes wasdue to a disturbed actin cytoskeleton. Staining of PKCl/i-deficient and WT podocytes with phalloidin to visualize actinrevealed a substantially altered cytoskeleton with fewer stressfibers and less accumulation of actin fibers at the cellularedges in 74% of the PKCl/i-deficient cells (Figure 1D) (10high-power fields, �20 magnification) containing 10 to 40cells were analyzed from each genotype. Cells were charac-terized as having a “rearranged actin cytoskeleton” when theydid not exhibit regular stress fiber patterns but an actinringlike structure with actin accumulation at cellular edges. Incontrast, the percentage of cells with a rearranged cytoskel-eton in PKCz�/� podocytes (16%) was comparable to that inWT cells (23%). In agreement with these findings, treatmentof WT cells with PKCl/i pseudosubstrate and highly effec-tive PKCl/i siRNA (approximately 95%; Figure 1F) resultedin changes to the actin cytoskeleton similar to those observedin PKCl/i�/� podocytes, indicating a role for PKCl/i but notfor PKCz in actin regulation and stress fiber formationin vitro (Figure 1E).

PKCl/i Deficiency Leads to Increased Activation ofSmall GTPases, Predominantly Rac1

Formation and motility of the actin cytoskeleton is regu-lated by a group of small GTPases.31 Small GTPases canexist in inactive GDP-bound or active GTP-bound forms.Activation of small GTPases is regulated by GEFs thattrigger the exchange rate of GDP and GTP.15 GTPase-activating proteins lead to inactivation of small GTPa-ses.32 The Rho GTPases (Rho, Rac, and Cdc42) controlsignaling pathways for assembly and disassembly of theactin cystoskeleton and associated integrin adhesion com-plexes.33 Activation levels of small Rho GTPases RhoA,Rac1, and Cdc42 were determined using pull-down assays,and were normalized to total levels of each respectiveprotein (Figure 2A). Furthermore, the values of the defi-cient cell lines were normalized to the appropriate WT cellline. Normalization revealed significantly increased levels

1949

Figure 1 Phenotype of PKCl/i- and PKCz-deficient podocytes in vitro. A: Phenotype ofdeficient or control podocytes after 10 days ofdifferentiation at 37�C. Images of live cells wereobtained using a light microscope (Axio Observ-er.Z1, �20 magnification). PKCl/i�/� podocytesgrow spindle-like and do not reach confluence, incontrast to control and PKCz�/� cells. Scale bar Z100 mm (applies to all panels). B: Migration ofdeficient and control podocytes recorded using alive microscope (Axio Observer.Z1, �20 magnifi-cation). The observed cells are shown in green, andthe covered distance is illustrated as red dots.Scale bar Z 50 mm (applies to all panels). C:Distance covered from startup to 6 hours,measured using ImageJ software (ManualTracking), was significantly increased in PKCl/i�/�

podocytes. n Z 10 cells per genotype. D: Phal-loidin staining reveals increased number of cellswith rearranged cytoskeleton in PKCl/i�/� podo-cytes, podocytes treated with aPKC pseudosub-strate (PS) or with PKCl/i siRNA, in comparisonwith WT podocytes. PKCz�/� podocytes showed nodifference from WT cells. Scale bar = 50 µm (ap-plies to all panels). E: Quantification of cells withrearranged cytoskeleton. Each treated cell line wascompared with an appropriate control. Ten high-power fields (�20 magnification) containing 10to 40 cells were analyzed from each genotypeusing a Leica DMBL microscope. F: Transfectionwith PKCl/i siRNA significantly reduced theamount of PKCl/i in WT podocytes. **P < 0.01.

Worthmann et al

of activated RhoA and Cdc42 and a large increase inactivated levels of Rac1 in PKCl/i�/� podocytes. Rac1activity in PKCz�/� podocytes was also slightly increased,but to a much lesser degree than in PKCl/i�/� podocytes.In addition, in both deficient cell lines, total RhoA wasconsiderably reduced, whereas total Rac1 and total Cdc42were only slightly affected. Equal sample loading wasconfirmed by testing the lysates for the expression levelsof GAPDH. Quantification of signal intensity from fourexperiments in three independent cell clones confirmedthese results (Figure 2B). These data indicate a regulatoryeffect of PKCl/i on all tested small GTPases, with apredominance for Rac1.

PKCl/i Deficiency Influences Cell Size and Number ofFocal Adhesions of Podocytes

The focal adhesion complex can influence the actin cyto-skeleton, and vice versa, via direct connection of stress

1950

fibers to this multiprotein complex. The primary componentsof focal adhesions that link the extracellular matrix to thecytoskeleton are vinculin and integrins.34 Vinculin stainingsof differentiated podocytes revealed differences in the dis-tribution pattern and number of focal adhesions in PKCl/i-deficient cells compared with WT control cells (Figure 3A).Twenty high power fields (×40 magnification) of each ge-notype were analyzed, and the experiments were performedthree times in different cell clones. The total cell area wassignificantly reduced in PKCl/i�/� podocytes (Figure 3B),whereas the cell area of PKCz�/� podocytes was significantlyincreased in comparison with WT cells (Figure 3C). Inaddition, the number of focal adhesions was significantlyreduced in PKCl/i�/� podocytes, whereas the number offocal adhesions in PKCz�/� cells was unaffected. Overall, thesize of the focal adhesions was similar among deficient celllines and WT cells; however, analysis of the number of ad-hesions relative to 100-mm2 cell area revealed a significantincrease in the density of adhesions in PKCl/i�/� podocytes

ajp.amjpathol.org - The American Journal of Pathology

Figure 2 PKCl/i deficiency influences activation of small GTPases. A: Protein lysates of deficient and control podocytes were used in activity assays for thesmall GTPases RhoA, Rac1, and Cdc42 and were analyzed using Western blot analysis. For loading control, the blotted membranes were incubated with GAPDH.B: Quantification of active GTPase versus total GTPase expression levels of RhoA, Rac1, and Cdc42. At least three different experiments of independent cellclones were used for each quantification. *P < 0.05, **P < 0.01.

Def-6 Acts Downstream of aPKCl/i

and decreased density in PKCz�/� cells. Similarly, wedetected an influence on the relative adhesion area bybuilding the ratio of adhesion area to whole cell area(Figure 3, B and C). This analysis revealed that although thenumber of focal adhesions was significantly reduced inPKCl/i�/� cells, the reduced cell area leads to an increaseddensity of adhesions and relative adhesion area. This resultwas confirmed in various clones from each genotype. Thisclearly indicates that the primary effect influencing thephenotype is reduced cell spreading and not the number offocal adhesions. Western blot analysis of integrin b1strengthened evidence for an influence of PKCl/i deficiencyon the focal adhesion complexes. Integrin b1 expression wassignificantly decreased in PKCl/i�/� podocytes but was notsignificantly affected in PKCz�/� podocytes (Figure 3D).Although the small increase in integrin b1 in PKCz�/�

podocytes was not significant in comparison with PKCzþ/þ

or PKCl/iþ/þ podocytes, it is an interesting finding. Thecommon connection of integrin and aPKCs in migration andcell polarity35 could represent counterregulation of the aPKCisoforms z and l/i (Figure 3E). The expression of vinculinwas decreased in both deficiencies; however, densitometricanalysis of three independent experiments in various cellclones indicated no statistical significance (Figure 3E).

Def-6 mRNA Is Increased in PKCl/i�/� Podocytes

To identify downstream targets of aPKCs that potentiallyregulate the actin cytoskeleton in podocytes, we performeda comparative microarray analysis of aPKC�/� and WTpodocytes. For all array experiments, mRNA samples from

The American Journal of Pathology - ajp.amjpathol.org

two independently generated cell lines were used per ge-notype to exclude clonal effects. The lists of differentiallyexpressed genes were sorted, and only the genes more thantwofold up-regulated or down-regulated in both cell lineswere used in the final analysis. The resulting lists of up-regulated or down-regulated genes (Supplemental TablesS1 through S4) in aPKC-deficient cell lines revealed 67common genes that are more than twofold regulated in bothaPKC isoform deficiencies, whereas 297 genes weredifferentially regulated in only PKCz�/� cells, and 593genes were differentially regulated in only PKCl/i�/� cells(Figure 4).

On the basis of the discrepancy between the phenotypes ofPKCl/i and PKCz deficiency, we focused on genes beingchanged in their expression in PKCl/i- but not PKCz-defi-cient cells. The top 30 up-regulated and down-regulatedgenes regulated only in PKCl/i�/� podocytes were clusteredin heat maps using the MultiExperiment Viewer program(Figure 4). This analysis revealed that several genes exclu-sively regulated in PKCl/i�/� podocytes but not in PKCz�/�

cells (Tables 1 and 2) were changed more than 100-fold.Gene ontology analysis performed using Onto-Expressrevealed that a substantial number of genes involved incytoskeletal adaptor activity, G-protein coupled receptor ac-tivity, cell adhesion, and filopodium formation were affected(Supplemental Table S5). Furthermore, Pathway-Expressanalysis detected the cell adhesion molecules as one of thesignificantly changed pathways in PKCl/i�/� but notPKCz�/� podocytes (Supplemental Tables S6 through S8).

Def-6, a 74-kDa GEF,21 was one of the strongest regu-lated genes in PKCl/i�/� podocytes (24-fold up-regulated).

1951

Figure 3 PKCl/i deficiency influences the cellsize and adhesion molecules of podocytes. A:Deficient and control podocytes were stained withan antibody against vinculin, and grayscale imageswere obtained using a Leica DMBL microscope (�40magnification). Binary inverted images were con-structed using ImageJ macros to obtain footprintsof vinculin distribution. Measurement of cell areaand number and size of focal adhesions was per-formed using ImageJ for PKCl/i-deficient cells (B)and PKCz-deficient cells (C) compared with WT cells.Twenty high-power fields (�40 magnification) ofeach genotype were analyzed, and the experimentswere performed three times in different cell clones.Scale bar Z 50 mm (applies to all panels). D:Western blot analysis reveals significantly reducedlevels of integrin b1 in PKCl/i�/� podocytes and anonsignificant reduction in vinculin in PKCl/i�/�

and PKCz�/� podocytes. Lysates were normalized toGAPDH. E: Quantification of at least three inde-pendent experiments of different cell clones. *P <

0.05, **P < 0.01, and ***P < 0.001.

Worthmann et al

Def-6 has been analyzed in various cell types; however, todate, no functional studies of renal cells have been pub-lished. Western blot experiments and real-time PCR mea-surements were performed to confirm the results of themicroarrays. Western blot analysis clearly demonstratedthe absence of PKCl/i protein in PKCl/i-deficient podo-cytes and the absence of PKCz protein in PKCz-deficientcells. mRNA levels of PKCl/i and PKCz were signifi-cantly down-regulated in the appropriate deficient celllines. Total Def-6 protein levels were comparable in WTand both deficient cell lines; however, as detected in themicroarray, the mRNA of Def-6 was significantly up-regulated in PKCl/i�/� podocytes. In PKCz�/� cells,

1952

there was no detectable changes in WT cells (SupplementalFigure S2).

Def-6 Expression in Glomeruli and Localization inPodocytes

Def-6 is highly expressed in lymphatic organs such asspleen and thymus.21 Def-6 could also be detected invarious muscle tissues and in renal HEK293 cells.23 Inas-much as currently nothing is known about Def-6 in thekidney, it was of great interest to examine in which renalcompartments Def-6 is expressed. Therefore, we stainedPKCl/iþ/þ and PKCl/i�/� kidney cryosections with an

ajp.amjpathol.org - The American Journal of Pathology

Figure 4 Microarray analysis of PKCl/i- andPKCz-deficient podocytes. Diagram shows the totalnumber of regulated genes in cells deficient forPKCl/i or PKCz and the number of more thantwofold up-regulated [ or more than twofolddown-regulated Y genes (in brackets) and theintersection of the two compared conditions. Thetop 30 genes up-regulated or down-regulated inPKCl/i- but not in PKCz-deficient cells are illus-trated in heat maps. Intensity of regulation isshown as low expression (blue) or high expression(pink). Genes are clustered according to similarintensity using the Software Multi ExperimentViewer at the left side of each heat map. For eachgenotype, two different cell lines were analyzed inthe microarray.

Def-6 Acts Downstream of aPKCl/i

antibody against Def-6. The staining demonstrated that inthe kidney Def-6 is predominantly expressed in theglomeruli and is only weakly expressed in the tubules(Figure 5A). Furthermore, to test the specificity of the Def-6antibody, we preincubated the antibody for 45 minutes withequal concentrations of Def-6 blocking peptide (Def6 re-combinant protein); both secondary antibody and antibodyblocking control were normal (Figure 5A). Co-staining withthe podocyte-specific marker podocalyxin and Def-6demonstrated that Def-6 is expressed in podocytes(Figure 5B). Semiquantitative scoring analysis of Def-6expression in glomeruli of WT and PKCl/i�/� mice (n Z4 mice of each genotype, �10 glomeruli per mouse) showedthat Def-6 protein expression is significantly higher inpodocytes of PKCl/i�/� glomeruli than of WT mice(Figure 5C). The medium score of Def-6 expression in thepodocytic area of PKCl/i�/� glomeruli was significantlyhigher in comparison with WT glomeruli (3.2 versus 2.0)(Figure 5D). The changed score distribution of PKCl/iþ/þ

The American Journal of Pathology - ajp.amjpathol.org

and PKCl/i�/� glomeruli is shown in Figure 5D. Toinvestigate the expression of Def-6 in vitro, we co-staineddifferentiated WT and PKCl/i�/� podocytes with anti-bodies against Def-6 and vinculin to mark focal contacts.We detected that endogenous Def-6 localized to the cellularedges and co-localized with vinculin at focal adhesions inpodocytes. Furthermore, PKCl/i�/� cells had more pro-nounced membrane-associated expression of Def-6(Figure 5E). Subcellular fractionation analysis of proteinlysates revealed that Def-6 is primarily expressed in thepodocyte membrane and only negligibly in the cytosolic ornuclear fraction. The membrane-associated expression issignificantly enhanced in PKCl/i�/� podocytes (140% incomparison with the WT membrane fraction), which wasconfirmed via quantification of fractionations from threeindependent experiments of different cell clones(Figure 5F). Moreover, in specific preparations of isolatedplasma membranes of PKCl/iþ/þ and PKCl/i�/� podo-cytes, we could specify the enhanced membrane localization

1953

Table 2 Highest down-regulated genes in PKCl/i�/� podocytesin comparison with WT, which are not regulated in PKCz�/�

podocytes

Gene Systematic Name Fold Change

Pitx1 NM_011097 262.07Armcx2 NM_026139 250.64Gbp1 NM_010259 234.54Sdk1 NM_177879 168.47Nkx2-3 NM_008699 104.63Napsa NM_008437 96.27Gypc NM_027863 91.91Gcnt1 NM_010265 69.51Cd74 NM_010545 63.38Cd302 NM_025422 53.39Krt8 NM_031170 52.63Unc93b1 NM_019449 52.00Tlr1 NM_030682 45.94Klk1b24 NM_010643 32.84

Table 1 Highest up-regulated genes in PKCl/i�/� podocytes incomparison with WT, which are not regulated in PKCz�/�

podocytes

Gene Systematic Name Fold Change

Aplp1 NM_007467 367.78Pdgfd BC030896 269.44Cxcl12 NM_013655 166.29Emb NM_010330 134.34Htatip2 NM_016865 126.05Six1 NM_009189 86.55Dlgh3 NM_016747 86.51Galntl4 NM_173739 81.09Armcx1 NM_030066 71.40Flrt3 NM_178382 64.97Krt18 NM_010664 60.59Abcg1 NM_009593 59.72G6pc2 NM_021331 58.98Tgfb3 NM_009368 50.17Dpysl4 NM_011993 49.01Has2 NM_008216 48.59Lox NM_010728 43.37Cspg2 D16263 41.96Msx2 NM_013601 39.62Foxc2 NM_013519 34.89Pcsk6 XM_355911 30.30Cbr3 NM_173047 29.14Aytl1 NM_173014 27.49Fbln5 NM_011812 27.00Antxr1 NM_054041 24.53Def6 NM_027185 23.91Tram1l1 NM_146140 23.13Hist3h2ba NM_030082 20.82Zfp125 AJ005350 19.73Smoc2 NM_022315 19.22

Fold change is given as the geometric mean for all genes more than two-fold up-regulated in both tested cell clones.

Worthmann et al

of Def-6 in the plasma membrane of PKCl/i�/� podocytes(three independent experiments performed in different cellclones) (Figure 5G).

Tbx15 NM_009323 32.23Kcnj14 NM_145963 26.07Lama1 NM_008480 23.63Ucp2 NM_011671 22.67Zcchc3 NM_175126 21.17Casp1 NM_009807 20.24Ly75 AK157921 18.66Chst1 NM_023850 18.26Aqp1 NM_007472 17.55Hunk NM_015755 16.81Slc16a13 NM_172371 16.42Lrrc33 NM_146069 16.33Ceecam1 NM_207298 14.29Lbp NM_008489 13.98Gldc NM_138595 13.30C3 NM_009778 13.02

Fold change is given as the geometric mean for all genes more than two-fold down-regulated in both tested cell clones.

Actin Phenotype of PKCl/i�/� Podocytes Is Rescuedby Overexpression of PKCl/i

We examined the effect of PKCl/i overexpression in PKCl/i�/� podocytes on the actin cytoskeletal architecture.Podocytes were transduced with adenoviruses expressingPKCi-GFP or GFP alone as control. At 7 days after infec-tion, RNA was isolated or cells were fixed for immuno-staining. Because there were no detectable differences in thepresence of collagen (data not shown), all experiments wereperformed in the absence of collagen. Real-time PCRanalysis revealed partial rescue of PKCi mRNA expression(60%) in PKCl/i-deficient cells after infection with PKCi-GFP (Figure 6A). As a consequence, Def-6 mRNAexpression decreased significantly (40% in comparison withPKCl/i�/�) (Figure 6A). Western blot analysis confirmed

1954

PKCi-GFP or GFP overexpression in WT and deficientcells. Total Def-6 protein expression levels remained unaf-fected (Figure 6B). Furthermore, phalloidin stainings ofrescued PKCl/i�/� podocytes revealed normalized actinstress fibers in approximately 50% of tranfected cellscompared with GFP-infected or GFP-uninfected cells(Figure 6C). The experiments were repeated three times intwo different cell clones and confirmed these results. Vin-culin stainings of transduced podocytes were used to mea-sure the cell area (Figure 6D) and number of focal adhesions(Figure 6D) (20 cells of each genotype, repeated in twodifferent cell clones). Both cell area and number of focaladhesions could be partially rescued after infection ofPKCl/i�/� podocytes with PKCi-GFP. The cell area andnumber of adhesions were significantly increased; however,rescue was detectable in only 50% in comparison withWT podocytes, which might be due to only partial rescueof PKCi mRNA. Figure 6E shows representative vinculinstainings after virus transduction. Overexpression of GFP-tagged Def-6 FL and constitutively active Def-6, whichconsists of only the DHL domain,22 in WT podocytes

ajp.amjpathol.org - The American Journal of Pathology

Figure 5 Def-6 expression is enhanced inPKCl/i�/� glomeruli and localizes to cellularedges in podocytes. AeC: Staining of kidney cry-osections from PKCl/iþ/þ and PKCl/i�/� mice. A:Upper panels: Overview staining with an antibodyagainst Def-6 (�10 magnification) reveals strongglomerular expression. Lower panels: Controlstaining with secondary antibody or Def-6 blockingpeptide validated the specificity of the Def-6 anti-body. Scale barsZ 200 mm. B: Co-staining of Def-6(red) and the podocyte marker podocalyxin (green)reveals enhanced podocytic expression of Def-6 inPKCl/i�/� glomeruli (boxed areas in Merge views,and arrowheads in Detail views). Nuclei are visual-ized with DAPI (blue). Scale barsZ 25 mm (appliesto all panels). C: For quantification of Def-6expression, sections were double-stained againstDef-6 (red) and nidogen (green). Nuclei were visu-alized using DAPI (blue). Def-6 expression isenhanced in the podocytic areas of PKCl/i�/�

glomeruli (boxed areas in Merge views, and arrow-heads in Detail views). Scale barsZ 25 mm (appliesto all panels). D: Images of PKCl/iþ/þ and PKCl/i�/�

glomeruli (40 glomeruli from four animals of eachgenotype) were obtained and analyzed using asemiquantitative score ranging from 0 (no expres-sion) to 4 (strong expression). Compared with WTglomeruli, PKCl/i�/� glomeruli show a significantlyhigher score and an obvious change in score distri-bution. nZ 4 mice of each genotype,�10 glomeruliper mouse. ***P < 0.001 E: Endogenous Def-6 lo-calizes to the cellular edges and co-localizes withvinculin, in particular in the PKCl/i�/� podocytes.Immunostaining of differentiated podocytes with anantibody against Def-6 (red) and vinculin (green)(insets in Merge views, and arrows in Detail views).Scale bar Z 50 mm (applies to all panels). F: Def-6membrane localization is significantly enhanced inPKCl/i�/� podocytes. Subcellular fractionation ofdifferentiated PKCl/iþ/þ and PKCl/i�/� podocytes.Lysates were normalized with Gia3 (membrane frac-tion, M), Sp1 (nuclear fraction, N), and GAPDH(cytosolic fraction, C). Quantitative analysis of Def-6expression in three independent experiments of twoindependent cell clones. **P < 0.01. G: Isolation ofplasma membranes from PKCl/iþ/þ and PKCl/i�/�

podocytes reveals increased localization of Def-6 inplasma membranes of PKCl/i�/� podocytes in com-parison with WT cells on Western blot analysis (leftpanel). For quantification, lysates of three indepen-dent experiments from two independent cell cloneswere normalized with Gia3 (right panel). *P< 0.05.

Def-6 Acts Downstream of aPKCl/i

confirmed membrane localization of active Def-6 in podo-cytes. In particular, constitutively active Def-6 DHL domainexhibited enhanced Def-6 localization at the cell membraneof the leading edges in transfected podocytes (Figure 6F).Next we performed Def-6 staining on transduced podocytes.After overexpression of PKCi-GFP, the membrane locali-zation of Def-6 in PKCl/i�/� podocytes could be reversedin >50% of the cells (Figure 6G). Both experiments wereperformed three times.

In summary, our data suggest that PKCl/i functions asan upstream regulator of Def-6 activation in podocytes.

The American Journal of Pathology - ajp.amjpathol.org

Absence of PKCl/i leads to an increase in Rac1 activation.In contrast, RhoA and Cdc42 activation is only mildlyaffected (Figure 7).

Discussion

Podocytes are highly specialized, differentiated cells thatform and maintain a network of arborized primary and sec-ondary foot processes. The formation of this foot processnetwork is maintained via a tightly orchestrated actin cyto-skeleton. The gaps between the secondary foot processes are

1955

Figure 6 Overexpression of PKCl/i rescues thephenotype of PKCl/i�/� podocytes. A: Real-timePCR analysis of PKCl/iþ/þ and PKCl/i�/� podo-cytes infected with adenoviral PKCi-GFP or GFP ascontrol. Transduction with PKCi-GFP rescued themRNA expression of PKCi and reduced the level ofDef-6 mRNA in PKCl/i�/� podocytes. B: Westernblot analysis demonstrated overexpression of PKCi-GFP or GFP control virus. C: Phalloidin stainings ofPKCl/iþ/þ and PKCl/i�/� podocytes transducedwith adenoviral PKCi-GFP or GFP as control.Transduction with PKCi-GFP rescued the actincytoskeletal phenotype of the PKCl/i�/� podo-cytes. Images were obtained using the Leica DMBLmicroscope (�20 magnification). Scale bar Z 50mm (applies to all panels). D: Measurement of cellarea and number of focal adhesions of vinculin-stained transduced podocytes. **P < 0.01, ***P< 0.001. E: Representative vinculin staining aftervirus transduction. Scale bar Z 50 µm (applies toall panels). F: Overexpression of Def-6 in podo-cytes. In transfected podocytes, Def-6 FL GFP lo-calizes to the cellular edges. Overexpression ofconstitutively active Def-6 DHL leads to enhancedmembranous localization (boxed areas in GFPviews and arrows in Detail views). As control,podocytes were transfected with pEGFP-N1. G:Transduced PKCl/iþ/þ and PKCl/i�/� podocytesstained with an antibody against Def-6. Over-expression of PKCi reversed the enhanced mem-brane localization of Def-6 in PKCl/i�/� podocytes(boxed areas in GFP views and arrows in middleDetail view).

Worthmann et al

spanned by a multimolecular complex of transmembranemolecules, the slit diaphragm. Various molecules of the slitdiaphragm complex, such as CD2-associated protein, bind tothe actin cytoskeleton and influence the formation of actinstress fibers,18,36,37 causing the slit diaphragm to function as asignaling hub. This signaling hub consists of importanttyrosine-phosphorylated signaling molecules nephrin andneph1,38 as well as several other molecules important forpodocyte signal transduction. In previous studies of the PKCfamily of serine/threonine kinases, we found that the con-ventional isoform PKCa is involved in TGF-beinducedapoptosis26 and nephrin endocytosis39 in podocytes. We alsodescribe that aPKCl/i is involved in fundamental podocytefunctions such as cell polarity and maintenance of the 3Dpodocyte architecture, and knockout of atypical isoformPKCz leads to no distinct renal phenotype.5,40,41 We recentlyhave found that double knockout of both isoforms inpodocytes leads to a severe developmental defect of the

1956

glomerulus, indicating a partial molecular compensation ofaPKCl/i for aPKCz in development and maintenance of thepodocyte.7 The objective of the present study was to identifyvarious molecular targets of the two aPKC isoforms, PKCl/iand PKCz, in podocytes in vitro that explain their distinctin vivo phenotypes in the single-knockout mouse models.3,5,6

To accomplish this, we performed in vitro studies andanalyzed isolated aPKC single-knockout cells in tissueculture. This revealed clear differences between the twoaPKC deficiencies in podocytes. The initial finding was astriking morphologic difference between PKCl/i�/� podo-cytes and control or PKCz�/� podocytes. The observedphenotypical features included altered cell shape, increasedcell movement, and a rearranged actin cytoskeleton. Thisaltered cell shape along with the actin cytoskeletal pheno-type was similar to the phenotype of podocytes describedin various publications regarding the actin cytoskeletonand small GTPases.19,33 Therefore, we speculated that the

ajp.amjpathol.org - The American Journal of Pathology

Figure 7 Proposed model for influence of PKCl/i via Def-6 on the actincytoskeleton. Deficiency of PKCl/i leads to increased activation of Def-6and thereby to imbalance of the small GTPases, in particular Rac1. ActiveRac1 influences several actin cytoskeletaledependent mechanisms.

Def-6 Acts Downstream of aPKCl/i

deficiency of aPKCl/i podocytes might also influence theactivity of small GTPases. Consistent with this hypothesis,we detected significantly more active Rac1, RhoA, andCdc42 in the PKCl/i-deficient podocytes, with Rac1exhibiting the most hyperactivation (Figure 2). Furthermore,we found an obvious reduction in total RhoA. It is supposedthat Rho and Rac have opposing effects on the actin cyto-skeleton; Rho regulates the formation of actin stress fibers,and Rac induces actin polymerization at the cellular edges tointroduce lamellipodia.33 In the present study, we detectedmore activated RhoA and Rac1 in PKCl/i�/� podocytesand observed fewer stress fibers and more peripheral actin.This could be explained by the strong Rac1 activation thatmight overrule RhoA activation. The second phenotypicalfeature that we analyzed in detail was the overall cell size.aPKCl/i-deficient podocytes exhibited a substantiallyreduced cell size and a significantly reduced number of focaladhesions. This could be explained by the changed activa-tion pattern of the small GTPases because changed signalingof small GTPases to the actin cytoskeleton directly in-fluences formation and turnover of focal adhesions.42 Toidentify distinct molecular targets, we performed acomparative microarray analysis of WT and the two aPKC-deficient podocyte cell lines. Comparison of up-regulatedand down-regulated genes that were specifically changedin PKCl/i-deficient but not PKCz-deficient podocytesrevealed genes involved in cytoskeletal adaptor activity andfilopodium formation. We identified GEF Def-6 as

The American Journal of Pathology - ajp.amjpathol.org

specifically up-regulated in aPKCl/i�/� podocytes. Def-6 ishighly expressed in B and T lymphocytes, in which stim-ulation of the T-cell receptor leads to activation of Def-6 andconsequently to binding of PIP3 and recruitment to theimmunologic synapse.20 Def-6 functions as a GEF for thesmall GTPases Rac1 and Cdc4210 and has the ability tochange the cellular architecture by binding to activated Rac1in other cell types.43

Characterization of Def-6 expression in the kidneydemonstrated for the first time a strong glomerular expres-sion of Def-6 in WT and PKCl/i-deficient mouse kidneys.In glomeruli of PKCl/i-deficient mice, we detected signif-icantly increased Def-6 expression, in particular in thepodocytes. In cultured podocytes, Def-6 was found at thecellular edges and focal adhesions, where it co-localizeswith actin and vinculin (Figure 5). This is similar to earlierdescriptions by Samson et al23 in NIH-3T3 fibroblasts,HeLa epithelial cells, and C2C12 myoblasts. After frac-tionation of podocyte protein lysates, we detected anenhanced amount of endogenous Def-6 protein in themembrane fractions, in particular in the plasma membraneof aPKCl/i-deficient podocytes. When we artificiallyoverexpressed the full-length protein or the constitutiveactive form, we could confirm the observed membranelocalization, in particular the active form in podocytes,much like NIH-3T3 cells in a study by Mavrakis et al.22

To demonstrate that aPKCl/i is the critical factor for theactin cytoskeletal phenotype, we re-expressed aPKCl/i inthe deficient podocytes and observed a phenotype reversalafter overexpression of PKCl/i (Figure 6). Further evidencefor involvement of Def-6 in PKCl/i signaling was thesimilar phenotype achieved in aPKCl/i-deficient and WTpodocytes after artificial overexpression of Def-6.

Our experiments point to the balance of small GTPasesand the upstream acting GEF Def-6 as the causative factorsfor the detected cytoskeletal phenotype. The balance ofsmall GTPases influenced by various GEFs and GTPase-activating proteins is a research area of increasing impor-tance in recent years for podocyte homeostasis. Forexample, Zhu et al19 have reported that activation of RhoAin podocytes induced albuminuria and focal segmentalglomerulosclerosis in a transgenic mouse model. Further-more, Akilesh et al16 have identified a Rac1 GAP, Arh-gap24, that contributes to the balance of Rho and Racsignaling in podocytes and a mutant form of Arhgap24 thatis associated with focal segmental glomerulosclerosis. In ratintestinal epithelial cells, expression of kinase-deficientPKCi blocks the oncogenic Ras-mediated activation ofRac1.44 In partial contrast to these findings, we measuredsimultaneously increased activation of Rac1, RhoA, andCdc42 in PKCl/i�/� podocytes, with the most prominenteffect on Rac1 activation. To our knowledge, Def-6 is thesecond example of molecules relevant in the immunologicsynapse that demonstrate a podocyte-specific function.Similarly, CD2-associated protein was initially described asan important component of the immunologic synapse;

1957

Worthmann et al

knockout of CD2-associated protein did not lead to animmunologic phenotype but to a podocyte-specific pheno-type with a changed actin cytoskeleton and severeproteinuria.45e47 Direct comparison with other epithelialcell types could be misleading because podocytes are ahighly specialized, differentiated cell type with a uniquefunction highly dependent on actin cytoskeleton arrange-ment. It is intriguing to speculate that a distinct cell-specificpathway to regulate activation of the various small GTPasesin podocytes might exist. Indirect interaction of Cdc42 andPKCl has already been shown by Coghlan et al,48 whofound that activated Cdc42 and PKCl induce loss of stressfibers in NIH3T3 cells. In contrast, we showed that inpodocytes the absence of PKCl/i leads to fewer stress fibersand less actin accumulation at the cellular edges. In ourmodel, we not only detected increased Cdc42 activity butalso increased amounts of activated RhoA and in particularRac1. This overall activation of small GTPases might in-fluence the actin structures in several ways.

An interesting finding is that Def-6 mRNA expressionwas increased in microarray and real-time PCR analysis butthat total Def-6 protein levels were comparable in WT andPKCl/i�/� podocytes. However, the enhanced membranelocalization of Def-6 that we detected via immunostainingand fractionation may most likely be the activated form ofDef-6, which is recruited to the membrane and leads to themeasured increase of activated Rac1. In this case, the pro-tein level of Def-6 is not so important for activation of smallGTPases as is recruitment of Def-6 to the plasma mem-brane. When Def-6 is not activated, it is kept in a closedconformation through an intramolecular interaction that in-hibits its GEF function. After phosphorylation and bindingto PIP3, the intramolecular interaction is released and GEFfunction is activated.20 There is no commercially availableantibody against the activated form of Def-6. Thus, onlymembrane association can be used as an indirect indicator ofDef-6 activation. Inasmuch as we found that the membraneassociation of Def-6 vanishes after rescue of aPKCl/i indeficient cells, it is highly probable that Def-6 is animportant downstream effector of aPKCl/i signaling inpodocytes

In conclusion, our data describe an actin cytoskeletalphenotype in PKCl/i�/� podocytes as a consequence ofdisturbed balance of small GTPases, most likely caused byactivation of the GEF Def-6. Differential expression of thenewly identified PKCl/i target Def-6 in glomeruli and inparticular in podocytes is a finding that opens novel researchavenues and describes a previously unknown pathwayrelevant for actin cytoskeletal dynamics in podocytes.

Acknowledgments

We thank Sandy Zachura and Germaine Puncha for tech-nical assistance, Dr. Nathan Susnik for proofreading themanuscript, Dr. Wolfgang Ziegler for help with the

1958

AxioObserver.Z1 microscope, and Dr. Fred Sablitzky(University of Nottingham) for providing the Def-6 con-structs. Microarray experiments were performed by staffmembers of the service project Z02 of the Sonderfor-schungsbereich SFB566. All confocal images were obtainedat the laser scanning facility of the Medical School ofHannover.

Supplemental Data

Supplemental material for this article can be found athttp://dx.doi.org/10.1016/j.ajpath.2013.08.026.

References

1. Steinberg SF: Structural basis of protein kinase C isoform function.Physiol Rev 2008, 88:1341e1378

2. Selbie LA, Schmitz-Peiffer C, Sheng Y, Biden TJ: Molecular cloningand characterization of PKC iota, an atypical isoform of protein kinaseC derived from insulin-secreting cells. J Biol Chem 1993, 268:24296e24302

3. Soloff RS, Katayama C, Lin MY, Feramisco JR, Hedrick SM: Tar-geted deletion of protein kinase C lambda reveals a distribution offunctions between the two atypical protein kinase C isoforms. JImmunol 2004, 173:3250e3260

4. Moeller MJ, Sanden SK, Soofi A, Wiggins RC, Holzman LB: Podo-cyte-specific expression of cre recombinase in transgenic mice. Gen-esis 2003, 35:39e42

5. Huber TB, Hartleben B, Winkelmann K, Schneider L, Becker JU,Leitges M, Walz G, Haller H, Schiffer M: Loss of podocyte aPK-Clambda/iota causes polarity defects and nephrotic syndrome. J AmSoc Nephrol 2009, 20:798e806

6. Leitges M, Sanz L, Martin P, Duran A, Braun U, Garcia JF,Camacho F, Diaz-Meco MT, Rennert PD, Moscat J: Targeteddisruption of the zetaPKC gene results in the impairment of the NF-kappaB pathway. Mol Cell 2001, 8:771e780

7. Hartleben B, Widmeier E, Suhm M, Worthmann K, Schell C,Helmstadter M, Wiech T, Walz G, Leitges M, Schiffer M, Huber TB:aPKCl/i and aPKCz contribute to podocyte differentiation andglomerular maturation. J Am Soc Nephrol 2013, 24:253e267

8. Andrews PM, Bates SB: Filamentous actin bundles in the kidney. AnatRec 1984, 210:1e9

9. Vasmant D, Maurice M, Feldmann G: Cytoskeleton ultrastructure ofpodocytes and glomerular endothelial cells in man and in the rat. AnatRec 1984, 210:17e24

10. Ichimura K, Kurihara H, Sakai T: Actin filament organization of footprocesses in rat podocytes. J Histochem Cytochem 2003, 51:1589e1600

11. Kriz W, Mundel P, Elger M: The contractile apparatus of podocytes isarranged to counteract GBM expansion. Contrib Nephrol 1994, 107:1e9

12. Pavenstädt H, Kriz W, Kretzler M: Cell biology of the glomerularpodocyte. Physiol Rev 2003, 83:253e307

13. Drenckhahn D, Franke RP: Ultrastructural organization of contractileand cytoskeletal proteins in glomerular podocytes of chicken, rat, andman. Lab Invest 1988, 59:673e682

14. Kerjaschki D: Caught flat-footed: podocyte damage and the molecularbases of focal glomerulosclerosis. J Clin Invest 2001, 108:1583e1587

15. Hall A: Rho GTPases and the actin cytoskeleton. Science 1998, 279:509e514

16. Akilesh S, Suleiman H, Yu H, Stander MC, Lavin P, Gbadegesin R,Antignac C, Pollak M, Kopp JB, Winn MP, Shaw AS: Arhgap24 in-activates Rac1 in mouse podocytes, and a mutant form is associated

ajp.amjpathol.org - The American Journal of Pathology

Def-6 Acts Downstream of aPKCl/i

with familial focal segmental glomerulosclerosis. J Clin Invest 2011,121:4127e4137

17. Wang L, Ellis MJ, Gomez JA, Eisner W, Fennell W, Howell DN,Ruiz P, Fields TA, Spurney RF: Mechanisms of the proteinuriainduced by Rho GTPases. Kidney Int 2012, 81:1075e1085

18. van Duijn TJ, Anthony EC, Hensbergen PJ, Deelder AM, Hordijk PL:Rac1 recruits the adapter protein CMS/CD2AP to cell-cell contacts.J Biol Chem 2010, 285:20137e20146

19. Zhu L, Jiang R, Aoudjit L, Jones N, Takano T: Activation of RhoA inpodocytes induces focal segmental glomerulosclerosis. J Am SocNephrol 2011, 22:1621e1630

20. Gupta S, Fanzo JC, Hu C, Cox D, Jang SY, Lee AE, Greenberg S,Pernis AB: T cell receptor engagement leads to the recruitment of IBP,a novel guanine nucleotide exchange factor, to the immunologicalsynapse. J Biol Chem 2003, 278:43541e43549

21. Gupta S, Lee A, Hu C, Fanzo J, Goldberg I, Cattoretti G, Pernis AB:Molecular cloning of IBP, a SWAP-70 homologous GEF, which ishighly expressed in the immune system. Hum Immunol 2003, 64:389e401

22. Mavrakis KJ, McKinlay KJ, Jones P, Sablitzky F: DEF6, a novel PH-DH-like domain protein, is an upstream activator of the Rho GTPasesRac1, Cdc42, and RhoA. Exp Cell Res 2004, 294:335e344

23. Samson T, Will C, Knoblauch A, Sharek L, von der Mark K,Burridge K, Wixler V: Def-6, a guanine nucleotide exchange factor forRac1, interacts with the skeletal muscle integrin chain alpha7A andinfluences myoblast differentiation. J Biol Chem 2007, 282:15730e15742

24. Helfrich I, Schmitz A, Zigrino P, Michels C, Haase I, le Bivic A,Leitges M, Niessen CM: Role of aPKC isoforms and their bindingpartners Par3 and Par6 in epidermal barrier formation. J Invest Der-matol 2007, 127:782e791

25. Jat PS, Noble MD, Ataliotis P, Tanaka Y, Yannoutsos N, Larsen L,Kioussis D: Direct derivation of conditionally immortal cell lines froman H-2Kb-tsA58 transgenic mouse. Proc Natl Acad Sci U S A 1991,88:5096e5100

26. Tossidou I, Starker G, Kruger J, Meier M, Leitges M, Haller H,Schiffer M: PKC-alpha modulates TGF-beta signaling and impairspodocyte survival. Cell Physiol Biochem 2009, 24:627e634

27. Mundel P, Reiser J, Zúñiga Mejía BA, Pavenstädt H, Davidson GR,Kriz W, Zeller R: Rearrangements of the cytoskeleton and cell contactsinduce process formation during differentiation of conditionallyimmortalized mouse podocyte cell lines. Exp Cell Res 1997, 236:248e258

28. Marg S, Winkler U, Sestu M, Himmel M, Schönherr M, Bär J,Mann A, Moser M, Mierke CT, Rottner K, Blessing M, Hirrlinger J,Ziegler WH: The vinculin-DeltaIn20/21 mouse: characteristics of aconstitutive, actin-binding deficient splice variant of vinculin. PLoSOne 2010, 5:e11530

29. Mundel P, Gilbert P, Kriz W: Podocytes in glomerulus of rat kidneyexpress a characteristic 44 KD protein. J Histochem Cytochem 1991,39:1047e1056

30. Mundlos S, Pelletier J, Darveau A, Bachmann M, Winterpacht A,Zabel B: Nuclear localization of the protein encoded by the Wilms’tumor gene WT1 in embryonic and adult tissues. Development 1993,119:1329e1341

The American Journal of Pathology - ajp.amjpathol.org

31. Sander EE, ten Klooster JP, van Delft S, van der Kammen RA,Collard JG: Rac downregulates Rho activity: reciprocal balance be-tween both GTPases determines cellular morphology and migratorybehavior. J Cell Biol 1999, 147:1009e1022

32. Bernards A: GAPs galore! a survey of putative Ras superfamilyGTPase activating proteins in man and Drosophila. Biochim BiophysActa 2003, 1603:47e82

33. Hall A: Rho GTPases and the control of cell behaviour. Biochem SocTrans 2005, 33(Pt 5):891e895

34. Ziegler WH, Liddington RC, Critchley DR: The structure and regu-lation of vinculin. Trends Cell Biol 2006, 16:453e460

35. Etienne-Manneville S, Hall A: Integrin-mediated activation of Cdc42controls cell polarity in migrating astrocytes through PKCzeta. Cell2001, 106:489e498

36. Asanuma K, Yanagida-Asanuma E, Faul C, Tomino Y, Kim K,Mundel P: Synaptopodin orchestrates actin organization and cellmotility via regulation of RhoA signalling. Nat Cell Biol 2006, 8:485e491

37. Yuan H, Takeuchi E, Salant DJ: Podocyte slit-diaphragm proteinnephrin is linked to the actin cytoskeleton. Am J Physiol Renal Physiol2002, 282:F585eF591

38. Benzing T: Signaling at the slit diaphragm. J Am Soc Nephrol 2004,15:1382e1391

39. Tossidou I, Teng B, Menne J, Shushakova N, Park JK, Becker JU,Modde F, Leitges M, Haller H, Schiffer M: Podocytic PKC-alpha isregulated in murine and human diabetes and mediates nephrin endo-cytosis. PLoS One 2010, 5:e10185

40. Hartleben B, Schweizer H, Lübben P, Bartram MP, Möller CC, Herr R,Wei C, Neumann-Haefelin E, Schermer B, Zentgraf H, Kerjaschki D,Reiser J, Walz G, Benzing T, Huber TB: Neph-Nephrin proteins bindthe Par3-Par6-atypical protein kinase C (aPKC) complex to regulatepodocyte cell polarity. J Biol Chem 2008, 283:23033e23038

41. Simons M, Hartleben B, Huber TB: Podocyte polarity signalling. CurrOpin Nephrol Hypertens 2009, 18:324e330

42. Rottner K, Stradal TE: Actin dynamics and turnover in cell motility.Curr Opin Cell Biol 2011, 23:569e578

43. Oka T, Ihara S, Fukui Y: Cooperation of DEF6 with activated Rac inregulating cell morphology. J Biol Chem 2007, 282:2011e2018

44. Murray NR, Jamieson L, Yu W, Zhang J, Gokmen-Polar Y, Sier D,Anastasiadis P, Gatalica Z, Thompson EA, Fields AP: Protein kinaseCiota is required for Ras transformation and colon carcinogenesisin vivo. J Cell Biol 2004, 164:797e802

45. Lehtonen S, Zhao F, Lehtonen E: CD2-associated protein directly in-teracts with the actin cytoskeleton. Am J Physiol Renal Physiol 2002,283:F734eF743

46. Shih NY, Li J, Karpitskii V, Nguyen A, Dustin ML, Kanagawa O,Miner JH, Shaw AS: Congenital nephrotic syndrome in mice lackingCD2-associated protein. Science 1999, 286:312e315

47. Dustin ML, Olszowy MW, Holdorf AD, Li J, Bromley S, Desai N,Widder P, Rosenberger F, van der Merwe PA, Allen PM, Shaw AS: Anovel adaptor protein orchestrates receptor patterning and cytoskeletalpolarity in T-cell contacts. Cell 1998, 94:667e677

48. Coghlan MP, Chou MM, Carpenter CL: Atypical protein kinasesClambda and -zeta associate with the GTP-binding protein Cdc42 andmediate stress fiber loss. Mol Cell Biol 2000, 20:2880e2889

1959