Oligomers of mutant Glial Fibrillary Acidic Protein (GFAP)inhibit the proteasome system in Alexander disease astrocytes, andthe small heat shock protein, αB-crystallin, reverses the inhibition

Guomei Tang1, Ming D Perng2 , Sherwin Wilk3, Roy Quinlan2, James E Goldman1,*

1Department of Pathology and Cell Biology, Columbia University, New York, NY10032, 2School ofBiological and Medical Science, University of Durham, Durham, DH1 3LE, UK, 3Department ofPharmacology and Biological Chemistry, Mount Sinai School of Medicine, New York, NY 10029,USA.

Running title: AxD mutant GFAP accumulation inhibits proteasome functionAddress correspondence to: James E Goldman, Department of Pathology, P&S Rm 15-420, ColumbiaUniversity College of P&S, 630 W. 168th St., New York, NY10032. Phone: 212-305-3554; Fax: 212-305-4548; Email: jeg5@columbia.edu

The accumulation of the intermediatefilament protein, GFAP, in astrocytes ofAlexander Disease (AxD), impairsproteasome function in astrocytes. Wehave explored the molecular mechanismthat underlies the proteasomeinhibition. We find that both assembledand unassembled wild type (wt) andR239C mutant GFAP protein interactswith the 20S proteasome complex, andthat the R239C AxD mutation does notinterfere with this interaction. However,the R239C GFAP accumulates to higherlevels and forms more proteinaggregates than wt protein. Theseaggregates bind components of theubiquitin-proteasome system and thusmay deplete the cytosolic stores of theseproteins. We also find that the R239CGFAP has a greater inhibitory effect onproteasome system than wt GFAP.U s i n g a ubiquitin-independentdegradation assay in vitro, we observedthat the proteasome cannot efficientlydegrade unassembled R239C GFAP,and the interaction of R239C GFAPwith proteasomes actually inhibitsproteasomal protease activity. The smallheat shock protein, αB-crystallin, whichaccumulates massively in AxDastrocytes, reverses the inhibitoryeffects of R239C GFAP on proteasomeactivity and promotes degradation of

the mutant GFAP, apparently byshifting the size of the mutant proteinfrom larger oligomers to smalleroligomers and monomers. Theseobservations suggest that oligomericforms of GFAP are particularlyeffective at inhibiting proteasomeactivity.

Alexander disease (AxD) is a rarebut fatal disease of the central nervoussystem, characterized by the presence inastrocytes of Rosenthal fibers (RFs),cytoplasmic protein aggregates thatcontain the intermediate filament protein,glial fibrillary acidic protein (GFAP),ubiquitinated proteins and small heatshock proteins (shsp) (1). Because of theloss of myelin and oligodendrocytes and avariable loss of neurons, AxD has beenhistorically thought of as a leukodystrophyor neurodegenerative disorder. AxD is,however, a disease of astrocytedysfunction, since mutations in GFAP arefound in the large majority of AxDpatients (2). All mutations detected so farare heterozygous coding mutations, andmost cases of AxD disease arise throughde novo, dominant, GFAP mutations (2).Most of the AxD mutations reside the 1A,2A and 2B segments of the conservedcentral rod domain of GFAP, but severalare located in the tail region (2). AlthoughGFAP mutations must underlie the

1

http://www.jbc.org/cgi/doi/10.1074/jbc.M109.067975The latest version is at JBC Papers in Press. Published on January 28, 2010 as Manuscript M109.067975

Copyright 2010 by The American Society for Biochemistry and Molecular Biology, Inc.

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

pathogenesis of AxD, the mechanisms bywhich GFAP mutations are toxic have notbeen fully elucidated.

The R239 residue is a “hot spot”for GFAP mutagenesis, since substitutionsat this arginine are the most frequentmutations found in AxD and appearamong different ethnic groups (2). Ourprevious work has concentrated on themost common substitution, R239C, andhas demonstrated that overexpressingGFAP, either wild type (wt) or the R239Cmutant, leads to the formation ofintracellular protein aggregates (3, 4, 5).GFAP accumulation also inhibitedproteasome protease activity and increasedlevels of ubiquitinated GFAP species (5).Although overexpressing wt GFAP canlead to intracellular protein aggregates, theR239C mutation exacerbates GFAPaccumulation and aggregation, aggravatesthe effects of GFAP accumulation ininducing intracellular stress responses andin inhibiting proteasome activities (5).

In the present study, we haveexplored the mechanisms that underlie theimpaired proteasome function by GFAPaccumulation. We asked if GFAPinteracted with proteasomes, in cells andin vitro , and further asked if thisinteraction resulted in proteasomeinhibition. Finally, we examined if αB -crystallin, a small heat shock protein(shsp) and molecular chaperone, couldreverse the GFAP-mediated proteasomeinhibition. Both assembled (10 nm widepolymers) and unassembled (soluble at80kxg) GFAP protein, both wild type andmutant, interact with the 20S proteasomecomplex. The unassembled GFAPincludes a mix of monomeric andoligomeric forms. Compared to the wtGFAP, the unassembled R239C GFAPshowed a marked shift toward largeroligomers. In addition, the unassembledR239C had a far greater inhibitory effecton proteasome proteolytic activity in vitrothan did the wt GFAP. Addition of αB-crystallin, which accumulates massively inAxD astrocytes, reversed the inhibitoryeffects of the R239C GFAP on proteasome

activity. αB-crystallin also changed theoligomer-monomer distribution of theR239C GFAP to a point at which itappeared identical to that of the wt GFAP.Our observations direct a focus onoligomeric forms of GFAP as effectiveproteasome inhibitors and suggestmechanisms that underlie the proteasomeinhibition in astrocytes of the AxD brain.In addition, since the R239C GFAPaggregates could bind components of theubiquitin-proteasome system, they maydeplete the cytosolic stores of theseproteins.

EXPERIMENTAL PROCEDURES

Reagents Cell culture medium (DMEM),and molecular biological reagents wereobtained from Invitrogen and Qiagen.Primary antibodies included anti-GFAPpolyclonal antibody (pAb, DAKO), antiGFAP mAb, anti-αB-crystallin pAb, anti-ubiquitin mAb (Chemicon), anti-GAPDHmAb (Encor), and anti-20S pAb (Biomol).The fluorescence-conjugated and HRPconjugated secondary antibodies werefrom Chemicon and AmershamBioscience. Imidazol-HCl, EDTA,MG132, Lactacystin, camptothecin, anti-FLAG M2 agarose affinity gel and proteinG Sepharose were from Sigma-Aldrich.

Brain tissue Tissues stored at –80oC fromcerebral hemispheric white matter ofcontrol subjects or AxD cases withdifferent mutations was obtainedpostmortem. For immunohistochemistry,the tissue was fixed in formalin and thenembedded in paraffin. All AxD cases werediagnosed based on histopathologicalexamination and confirmed by themolecular genetic analysis for GFAPmutation. Controls included frozen CNStissue from two children, one with noneurological disease (Control I) and onewith Werdnig-Hoffman disease (non-AxD,non-leukodystrophy neurological diseasewithout RFs) (Control II).

Cell cultures and Transfection cDNA

2

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

clones encoding wt GFAP or R239C mtGFAP were inserted into EGFP-C1expression vector for expressing a GFP-GFAP fusion protein. To generatepermanent cell lines, U251 astrocytomacells were transfected with the GFAP-GFPplasmid using calcium phosphate. Thetransfected cells were then cultured inDMEM supplement with 10% FCS,antibiotics, 500ug/ml G418. Clonal linesof stably transfected cells were isolatedand confirmed by FACS sorting orWestern blot analysis. MG132 (10µM),Lactacystin (10µM) were added tocultures at 24 hours post-transfection andincubated for a further 24 hours, withDMSO as the vehicle control.

Immunoprecipitation and Western blottingFor protein extraction and fractionation,brain tissue or cells were lysed in total celllysis buffer (2% w/v SDS, 6.25% mM Tris(pH7.5), 5mM EDTA, supplemented withRoches protease inhibitor cocktail tablet)for total protein, or in fractional lysisbuffer to separate soluble and insolublefraction. Briefly, cells or brain tissueswere recovered in S1 buffer (0.5% v/vTriton-X, 2mM Tris, 2mM EDTA andsucrose implemented with Complete MiniRoches protease inhibitor cocktail). Aftercentrifugation for 20min at 15,000 rpm at4°C, the supernatant was collected as the“soluble” fraction and the pellet(“insoluble” fraction) was recovered in S2buffer (2% w/v SDS, 2mM Tris, 2mMEDTA). Protein concentration wasdetermined using a Bio-Rad protein assaykit according to the manufacturer’sinstructions.

For immunoprecipitation, Cos7cells were transiently transfected withFlag-tagged wt and R239C GFAP orvector only. Immunoprecipitations wereperformed 48h post transfection.. Equalamounts of protein from whole cellextracts, soluble fractions were diluted inIP lysis buffer (6) and were incubated withanti-FLAGM2 agarose gel or anti-20Santibody and Protein G Sepharoseovernight at 4°C. Immunoprecipitates

were then separated by SDS-PAGE andsubjected to Western blotting.

Proteasome Activity Assay Brain tissuesand cells were homogenized in ice-coldproteolysis buffer (10mM Tris-HCl,PH7.2, 0.035% SDS, 5mM MgCl2, 5mMATP and 0.5mM DTT). The homogenateswere centrifuged at 13,000g for 10 min.and the resulting supernate was incubatedwith proteasome substrate II (S-II:ZLeuLeuGluAMC) or III (S-III:SucLeuLeuValTyr-AMC) to detect thepeptidylglutamyl peptide hydrolase(PGPH) and chymotrypsin-like peptidaseactivities, respectively. Fluorescence wasmonitored at 360 nm excitation and 450nm emission in a fluorescence plate reader(Bio-Tek FL600).

Immunocyto- and immunohistochemistryImmunostaining of cultured cells wasperformed at 48 hours after transfection,as described before (4). A cytoskeletonbuffer (PHEM buffer: 100mM Pipes,2mM EGTA, 1mM MgCl2, 0.5% Triton-X100, PH6.8) was utilized to characterizethe cytoskelton. Cells were visualized byZeiss LSM510 confocal microscope.

Immunohis tochemis ty wasperformed according to general protocols.10-µm thick consecutive sections wereprepared. Sections were blocked in TBS-T/ 5% BSA for 1h and subsequentlyincubated with antibodies against 20Sproteasome complex (1:100), αB -crystallin (1:100) ubiquitin (1:100) andGFAP (1:200) overnight at 4°C. Negativecontrol sections were held in PBS duringthe primary incubation. Sections werethen incubated with HRP-labeledsecondary antibody 1h at roomtemperature, and developed with 3,3'-diaminobenzidine (DAB).

In vitro GFAP assembly and co-sedimentation assay Recombinant humanwt and R239C GFAP were purified fromEscherichia coli BL21 (DE3) pLysS andassembled in vitro as before (7). Theassembled wt or R239C GFAP filaments

3

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

were incubated with purified 20Sproteasomes (8) at room temperature for1.5 hours. Samples were then separatedinto pellet and supernatant fractions byultra-centrifugation at 80,000g for 30minat 4°C, and assayed by Western blotting.

Statistical analysis Results are expressedas mean±SD. Statistical analysis wasperformed using ANOVA. P<0.05 wasconsidered statistically significant.

RESULTS

GFAP aggregation decreases proteasomeactivity and depletes cytosolicproteasomes

We first assessed the proteasomeprotease activities in the U251astrocytoma cell line stably transfectedwith control, wt GFAP, and R239C GFAPexpression vectors. In comparing cellstransfected with vector only (used ascontrol and set to 100% activity), thepeptidylglutamyl peptide hydrolase(PGPH) activity was reduced in cellsexpressing wt GFAP (p<0.05) and furtherin those expressing R239C GFAP(p<0.01), while the chymotrypsin-likepeptidase activity showed a non-significant reduction in wt and a modestreduction in R239C cells (p<0.01, Fig1A). The decreased proteasome activitydue to GFAP accumulation was not causedby the loss or reduction of proteasomecomplexes since the total amount of the20S complex remained the same in thevector-transfected cells and the GFAPexpressing cells (Fig 1B). However, weobserved a shift of 20S complex from aTriton X-100 soluble fraction to theinsoluble fraction (Fig 1C).

We detected significantlydecreased PGPH and chymotrypsin-likeprotease activities in white matter fromAxD patients with the R239C, R416Wand R239H mutations (Fig 1D), although20S complex protein levels were at thesame levels or increased compared to that

in white matter without Rosenthal fibersfrom a non-AxD subject (Fig 1E).

The 20S proteasome complex hasbeen reported to be associated withaggresomes, intracellular proteinaggregates formed by misfolded proteins(9). In the U251 cells, the GFAPinclusions were labeled with antibodies tothe proteasomal 20S complex and toubiquitin (Fig 2A). Many more of the cellsexpressing the R239C GFAP showedinclusions, compared to cells expressingwt GFAP (Figure 2D). Treatment of cellswith proteasome inhibitors increasedGFAP levels and increased the proportionof cells that contained inclusions (Figure2C, D). When we examined sections fromthe brains of AxD patients, we found asimilar distribution of GFAP, the 20Sproteasome and Rosenthal fibers (Fig 2B).Because this similar distribution can onlysuggest, but cannot prove physicalinteractions among these components, welooked at cell and in vitro systems (seebelow).

Associations between 20S proteasome andnon-assembled GFAPs

Based on the fact that proteasomesassociate with different types ofintermediate filaments, including keratin,desmin and vimentin (10, 11), we inferredthat, as a type III intermediate filament,GFAPs may interact with 20S proteasomecomplexes, and the interaction ofproteasome complexes with theaccumulated GFAP might interrupt thenormal proteasome function. To addressthis possibility, we performed a series ofco-immunoprecipitations in Cos7 cellstransiently transfected with FLAG-taggedwt or R239C GFAP. Since it has beenreported that some proteasomes arepresent in the Triton-X-100 solublemembrane and cytosolic fractions (12), welooked for GFAP-20S complexes in theTriton X-100-soluble fractions of Cos7cells (Fig 3A). Reciprocal co-immunoprecipitations using anti-20Santibodies and anti-FLAG antibodiesrevealed that both wt and R239C GFAPs

4

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

were present in complexes with 20Sproteasomes. Protein A-Sepharose beadswithout anti-20S antibodies were used as abackground control, as were non-transfected Cos7 cells and Cos7 cellstransfected with the empty vector. Notethat the polymerized GFAP intermediatefilaments were excluded from this assay,since they are part of the Triton X-100-insoluble, cytoskeletal network, which isalso insoluble in IP buffer. However,when we treated the cells first with acytoskeleton buffer, (PHEM, containing0.5% Triton-X), which extracts solubleproteins, double immunofluorescenceexamination showed that proteasomescontinued to associate with GFAPinclusions even when soluble, cytoplasmicproteins were removed (Fig 3B),suggesting that some of the 20 subunitsassociated with cytoskeletal GFAP.

20S proteasome bind to assembled GFAPfilaments in vitro

To determine whether GFAPintermediate filaments will associatedirectly with proteasomes, we turned to invitro analyses, first using a co-sedimentation assay (7). Purified 20Sproteasome complexes were incubatedwith in vitro assembled wt or R239CGFAP filaments at a weight ratio of 1:1, orwithout GFAP (as a control), for 1.5 hrs.The samples were centrifuged at 80,000xgthrough 40% sucrose, washed withassembly buffer and the PHEM buffer, andfinally resolved by SDS-PAGE.Nitrocellulose blots were probed with ananti-20S antibody and then re-probed withan anti-GFAP antibody. The 20Sproteasomes incubated without GFAPcould not be recovered in the pelletfraction (Fig 3C). In contrast, whenincubated with assembled wt or R239CGFAP filaments, a fraction of the 20Sproteasomes co-sedimented with theGFAP (Figure 3C). This result revealed adirect association between proteasomesand GFAP filaments. Note that the R239Cmutation apparently does not interferewith this binding.

Both soluble and aggregated forms ofGFAP inhibit proteasome function

Our data above indicated a directinteraction between GFAP and proteasomecomplexes, but we could not discern if theinteraction interrupts proteasome function.To address this possibility, we performedan i n v i t ro ubiquitin-independentproteasome degradation assay byexamining the activity of purified 20Sproteasomes in the presence ofrecombinant wt GFAP, R239C GFAP, orMG132, again using fluorescent substratesto monitor proteasomal protease activity.We first examined how “soluble” wt andR239C GFAPs (separated from pelletable,assembled wt and R239C GFAPs bycentrifugation in 40% sucrose at 80,000xgas described above) affected proteasomefunction. Wt and R239C GFAPs wereisolated from the top layers and wereincubated at identical concentrations withpurified 20S proteasome for 30 min. Wethen added fluorescent substrates and 30min. later analyzed the samples.Compared to those incubated withsubstrates only, 20S proteasomecomplexes exposed to R239C GFAPshowed a significantly decreased activity.In contrast, wt GFAP showed lessinhibitory effects on 20S proteasomeactivity (Fig 3D). In the positive control,MG132-treated proteasomes were almostcompletely inhibited.

We also assessed the effect ofassembled wt and R239C GFAPs on 20Sproteasome activity. In vitro assembled wtor R239C GFAP filaments were separatedfrom “soluble” GFAP by centrifugationand pre-incubated with purified 20Sproteasome complexes. The fluorescentsubstrates were added to examineproteasomal proteolytic activity.Compared to MG132 treated samples andthe negative control, the R239C GFAPfilaments exhibited a significant inhibitoryeffect on proteasomes, while the wt GFAPfilaments only slightly inhibited PGPHprotease activity (Fig 3D). However, theseinhibitory effects were far less than those

5

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

produced by non-assembled GFAP on aweight/weight basis. This result suggestedthat the inhibition of proteasomes byR239C GFAP might be predominantlyattributed to “soluble” GFAPs.

In all of the proteasome inhibitionexperiments we kept the GFAPconcentration between 0.25 mg/ml (50µg/200µ l) and 1mg/ml (200µg / 2 0 0µl),which we believe is within the physiologicalrange given the following estimates. GFAPrepresents about 1.0% of total brain proteinin the adult rat (13). If brain isapproximately 11% protein and 9% proteinby wet weight in adult rat and human,respectively (14), or about 90 mg/ml, theGFAP concentration would be 0.90 mg/ml.If astrocytes take up about 10% of brainvolume (15) and since GFAP is only foundin astrocytes, the GFAP concentration inastrocytes is in the order of 9.0 mg/ml (maybe higher in white matter astrocytes, sincetheir GFAP content is greater than that ofgray matter astrocytes). The CNS inpatients with AxD contains markedlyincreased amounts of GFAP, up by at least10-fold, thus giving a GFAP concentrationof 90 mg/ml. The “soluble” GFAP isoperationally defined as material that doesnot pellet at a specified g-force, and in anormal brain represents a small fraction ofthe total GFAP. Thus, in the adult rat brainafter Triton-X-100 solubilization andcentrifugation at 180kxg about 0.8% of thetotal GFAP appears in the “soluble “ fraction(13). Thus, the “soluble” GFAPconcentration in astrocytes would be around0.72mg/ml.

R239C GFAP inhibits proteasome activityin a non-competitive manner

To understand how non-assembled GFAPs inhibit 20S proteasomeactivity, we varied the concentration offluorogenic substrates at a constantconcentration of wt or R239C GFAP andmeasured proteolytic activities. Purifiedproteasomes were incubated with solubleor polymerized wt and R239C GFAPs,and 20 min later, proteasome substrates II

or III were added. For proteasomesunexposed to GFAPs, as a control, thedegradation of proteasome substratesincreased with increasing substrateconcentration, and reached a maximumactivity at a concentration of 100mM(log100=2). With a further increase insubstrate concentration, the proteasomeactivity decreased slightly (Fig 4A),suggesting that increasing substrateconcentration beyond a threshold mayinhibit proteasome proteolytic activities.In the range of 12.5 (log12.5=1.1) -100mM substrate concentration, theactivity of 20S proteasome exposed tofilamentous or unassembled wt GFAPincreased with increasing substrateconcentration (Fig 4A), similarly to theactivity without GFAP, although thereappeared to be some small inhibition athigher SII substrate levels. These dataindicate that the proteasomal inhibitionthat we observed before was not a falsepositive result from substrate binding andsubstrate sequestration. Thus GFAP seemsto inhibit the proteasome activity via aninteraction with the proteasome, ratherthan by binding substrate peptide. TheR239C GFAP produced a significantinhibition of proteasome activity, with theunassembled (“soluble”) GFAP exhibitingthe greatest effect (Fig 4A). Interestingly,proteasome activity did not continue toincrease with increasing concentrations ofsubstrate, suggesting that the interactionbetween soluble R239C GFAP andproteasomes produced a non-competitiveinhibition of proteasome activity.

Next, we assayed the PGPH andchymotrypsin-like protease activities atdifferent GFAP concentrations (50,100,150, 200µg/200µ l total reactionvolume), with the fluorescent peptidesubstrate concentration fixed at 100mM.The R239C GFAP inhibited PGPH andchymotrypsin-like activity more than thewt GFAP at all GFAP concentrations (Fig4B). For example, at the 100µg/200µl, theR239C GFAP inhibited almost 50% ofproteasome activity, while the wt inhibitedonly 10% of total activity; at

6

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

200µg/200µl, the R239C GFAP produceda 90% percent inhibition, while the wtGFAP was far less inhibitory. MG132, as apositive control, blocked the proteaseactivity markedly (Fig 4B).

“Soluble” wt GFAP, but not R239CGFAP, is partially degraded by the 20Sproteasome

We then asked whether the wt andthe R239C GFAPs could be efficientlydegraded by 20S proteasomes via aubiquitin-independent manner in vitro.Soluble wt and R239C GFAPs wereincubated with 20S proteasomes andproducts were analyzed by Westernblotting. After exposure to 20Sproteasomes, a fraction of the wt GFAPwas broken down into smaller fragments(Fig 4C). In contrast, the R239C GFAPwas degraded only minimally (Fig 4C). Inthe presence of MG132, as a positivecontrol, neither wt nor R239C GFAP wasdegraded. Thus, although both wt andR239C GFAP can interact withproteasome 20S complexes, the R239CGFAP cannot be efficiently degraded.

The molecular chaperone, aB-crystallin,restores proteasome proteolytic activitythat was reduced by R239C GFAP

We then asked if αB-crystallincould rescue proteasome function from theR239C GFAP toxicity. We incubatedpurified 20S proteasomes with soluble oraggregated wt or R239C mt GFAPs, in thepresence or absence of αB-crystallin (at aweight/weight ratio of GFAP to α B -crystallin of 1:2), and then analyzed thePGPH and chymotrypsin-like proteasomeproteolytic activities. At the GFAPconcentration used here (100µg/200µl),the soluble R239C GFAP inhibited bothPGPH and chymotrypsin activities and thepelleted R239C GFAP produced a milderinhibition (Fig 5A). The introduction ofrecombinant αB-crystallin attenuated thetoxicity of R239C GFAP on proteasomeactivities (Fig 5A).

To validate the protective effects

of αB-crystallin on proteasome function invivo , we increased intracellular αB -crystallin levels using a recombinantadenovirus in U251 cells stably expressingR239C GFAP. By determining theproteasome activity using fluorescentsubstrates, we found that theoverexpression of exogenous αB -crystallin attenuated the proteasometoxicity induced by R239C GFAP (Fig5 B ) . A s v i s u a l i z e d b yimmunofluorescence, the overexpressedα B-crystallin dispersed the GFAPaggregates, and shifted them intofilamentous structures (Fig 5C).

a B - c r y s t a l l i n c h a n g e s t h eoligomerization state of R239C GFAPand promotes GFAP proteolysis

The R239C mutant GFAP canassemble into 10nm filaments, like the wtGFAP (4). However, a fraction of theGFAP remains in suspension aftercentrifuging the reassembled GFAP at80,000xg, as noted above. This “soluble”GFAP fraction was electrophoresedthrough a non-SDS, non-reducing, 8%PAGE gel and then immunoblotted toreveal the different GFAP species. The“soluble” wt GFAP resolved as a mix ofmonomers (50kD) and higher MW bands,the two main ones representing dimers andtetramers (Fig 5D, left panel). In contrast,the R239C GFAP showed a pronouncedshift to higher MW forms (~200-300 kDa,perhaps reflecting 4-6 GFAP molecules).When we added αB-crystallin to the“soluble” R239C GFAP and thenelectrophoresed the protein, we found apattern of GFAP monomers and oligomersmuch more similar to that of the wt GFAP(Fig 5D, right panel). We then asked ifαB-crystallin would promote moreproteolytic degradation of the R239CGFAP by the 20S proteasome. When the“soluble” R239C GFAP was pre-incubatedwith αB-crystallin, we found that thedegree of proteolytic breakdown of theGFAP was now similar to that of the wtGFAP (Fig 5E, F). Thus, αB-crystallin not

7

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

only shifts the monomer-oligomerequilibrium of the R239C GFAP, but italso allows proteolysis.

DISCUSSION

GFAP accumulation inhibits proteasomeactivityIn this study, we investigated how GFAPaccumulation in the astrocytes of AxDmight affect the ubiquitin-proteasomesystem. Using fluorogenic peptide assays,we demonstrated decreased proteasomeactivities in U251 astrocytoma cells stablyexpressing GFAP and in AxD whitematter. These results corroborated ourprevious findings with a GFP-Uproteasome reporter system, whereoverexpressing GFAP i m p a i r e dproteasome function (5). Both wt andR239C GFAP accumulation inhibitedproteasome activity, but the R239C GFAPexhibited a greater inhibitory effect.R239C GFAP tended to aggregate intoinclusions (5, 16), which were associatedwith components of the proteasomesystem, such as proteasome 20S corecomplexes, ubiquitinated proteins andsmall heat shock proteins, as evidenced byimmunostaining of cells and of Rosenthalfibers in AxD brains. Furthermore, wefound that ubiquitinated GFAP proteinaccumulated in GFAP overexpressingU251 cells and in AxD brain tissues, andthat GFAP further accumulated upontreatment with MG132 and Lactacystin,two proteasome inhibitors. Theseobservations point to a role of theubiquitin-dependent proteasome pathwayin the removal of excess GFAP. Theinhibition of proteasome appear, in turn, toaggravate the GFAP accumulation andaggregation, since we observed anincrease in the GFAP protein level and inthe percentage of cells bearing GFAPaggregates (5).

A recent study (17) showed inastrocytoma cells that proteasome inhibitorsdownregulate GFAP gene transcription andthereby decrease GFAP protein levels.

Furthermore, proteasome inhibition preventsGFAP accumulation in reactive astrocytes ina model in which a microdialysis probethrough which an inhibitor was infused wasplaced in the rat CNS. The trauma of probeplacement itself produced increased GFAPimmunostaining, which was prevented byinhibitor. However, this effect only occurredduring the early increase in GFAP, but notafter 7 days, suggesting that GFAP geneexpression is initially susceptible toproteasome inhibition, but is no longersusceptible at the higher levels of GFAPprotein that had accumulated by the laterstages of astrogliosis. We do not know ifthis model applies to the proteasomeinhibition in AxD. For example, in mousemodels of AxD, the levels of GFAP mRNAand GFAP protein are both elevated (18).

Proteasomes bind to filamentous GFAP Our data demonstrated a direct

association between filamentous GFAPand the 20S proteasome complex. Theassociation of proteasomes withintermediate filament network of thekeratin type has been reported in PtK1 andHela cells (10). In addition, intracellularlocalization of proteasomes overlappedwith the vimentin network in proliferatinghuman fibroblasts, and with desminintermediate filament in myoblasts of themouse myogenic cell line C2.7 (11). Alater study also revealed a directinteraction between proteasomes and theactin network (12). In the present study,we identified an association betweenproteasomes and the GFAP intermediatefilament network in U251 astrocytomacells. This co-localization wasparticularly visible after treatment withTri ton-X-100, which faci l i ta tesobservation of GFAP, since it eliminatesmost of the soluble proteins. In vitro co-sedimentation experiments showed furthera direct interaction between 20Sproteasome and assembled GFAPfilaments. This was consistent withprevious reports on the interaction ofpro teasomes and recons t i tu tedintermediate filaments of the vimentintype (12).

8

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Whether this interaction ofproteasomes with filamentous GFAPinterferes with proteasome activity is notfully resolved. For example, we show thata larger proportion of the 20S subunitassociates with the “pellet” fraction ofU251 cells than in control cells (Figure1B). This association would depleteproteasomes from the “soluble” fractionand thus might account for the loss ofproteasome activity measured in a solublefraction in the cells (Figure 1A), simplybecause there are fewer proteasomes inthis soluble fraction.

The in vitro activity assaysshowed that the filamentous forms ofGFAP did not inhibit proteasomeproteolytic activity as much as the“soluble” forms, on a weight/weight basis,although the filamentous R239C GFAPwas more inhibitory than the wt GFAP.

The interaction between proteasomes and“soluble” GFAP inhibits proteasomeactivity

We demonstrated a directassociation between Triton X-100-solubleGFAP and 20S proteasome complexes byreciprocal immunoprecipitation from celllysates. With this result in mind, we testedthe effects of the “soluble” fraction of thereassembled GFAP on proteasome activityin vitro. The “soluble” fraction consistedof a mix of monomeric and oligomericforms,-. We use the term “oligomer” todenote a molecule larger than themonomeric GFAP but smaller than thepolymerized intermediate filament.However, we know neither the exactstructure of these larger “soluble” forms,nor whether they are normal intermediatesin the polymerization of R239C GFAP,which will assemble into normalappearing 10 nm filaments (4), orabnormal intermediates separate from thepathway of filament assembly. There wasa marked shift to larger oligomeric formswith the R239C mutation, suggesting thatthese larger intermediates played a criticalrole in the inhibition. Indeed, the greaterability of soluble, unassembled R239C

GFAP, compared with assembled R239CGFAP, to inhibit proteasome activityimplies that AxD mutant GFAP inhibitsproteasome function primarily through theaccumulation of monomeric and/oroligomeric forms. We would consider theoligomer forms to be inhibitory, given thatdecreasing oligomer levels with αB -crystallin restored proteasome proteolyticactivity. These observations add to therecent view in protein misfoldingdisorders that oligomeric protein speciesmay be biologically more toxic thanlarger, aggregated fibrils (19, 20).

However, we cannot conclude that wtGFAP is innocuous and can never causeproteasome inhibition. Wt GFAP willinhibit proteasome activity, but requiredhigher concentrations than R239C GFAP.One of the effects of increasing wt GFAPconcentrations would be to increase theconcentration of wt oligomers, but wewould suggest that, because of thedifferences in equilibrium, one needs a farhigher concentration of wt GFAP thanR239C to reach the same oligomerconcentration. Alternatively, largeramounts of wt GFAP monomers maysimply provide a competitive inhibition,since proteasome inhibition might occur athigher rates of substrate degradation, anda competition between the excess GFAPand other cellular proteins for degradationmight occur.

The kinetic analysis showed that theproteolysis of fluorogenic substrates wassaturable at a constant proteasomeconcentration (Fig 4). The soluble R239CGFAP produced a marked inhibition, witha relatively flat activity curve, suggestinga non-competitive component to theinhibition. This could reflect, for example,R239C oligomer binding to theproteasome. The oligomer would not bedegraded, but its binding would inhibit theproteolysis of other substrates. Themechanistic model of proteasomedegradation proposes that substrates aredirected into proteasomes and degraded inthe inner channel of 20S proteasomes,where the active sites of the catalytic

9

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

subunits are located. However, onlyunfolded monomeric peptides are able toslide into the peptide channel, whichfunctions by size exclusion (21). Oneimplication of this model is that proteinsthat are in oligomeric assemblies will notbe efficiently degraded, but will only bedegraded after depolymerization into amonomeric form. Since monomeric andoligomeric forms of GFAP must exist inequilibrium, any changes to the proteinthat shift the equilibrium towardoligomeric forms will slow proteasomaldegradation. One of the dramatic effectsof the R239C mutation is the shift in MWsize in the “soluble” GFAP pool towards alarger class of oligomers. We do not yetknow why this equilibrium is so shifted.Presumably the larger oligomers of theR239C GFAP are more stable than thoseof the wt GFAP, but the structural basis forthis stability is unclear. Theseobservations suggest that further studies ofmonomer-oligomer equilibrium with thevarious GFAP mutations of Alexanderdisease are worth pursuing.

Ubiquitin-mediated and ubiquitin-independent degradation of GFAP

The 20S proteasome particlecatalyzes proteasomal hydrolysis whichcan take two forms: ubiquitin-dependentprotein degradation, representing themajor fraction, relies on the 26Sproteasomes, and ubiquitin-independentproteolysis of certain unfolded ormodified proteins can be carried out bythe single 20S proteasome (22, 23). Ourdata suggests that GFAP can be degradedvia both ubiquitin-dependent andubiquitin-independent processes. Theubiquitin-independent GFAP proteindegradation has been demonstrated above.We also found that in cell cultures andAxD brains, GFAPs were ubiquitinated.Upon proteasome inhibition, there was afurther accumulat ion of theseubiquitinated species (5). This result wasconsistent with our previous observationsthat defects in proteasome degradationoccurred when GFAP were overexpressed

in the GFP-U proteasome reporter system,suggesting that overloaded GFAP proteinsalso damaged the ubiquitin-dependentproteasomal degradation. We also do notexclude the possibilities that the R239CGFAP may interact with 26S proteasomeat other sites, such as the proteasomesubunit alpha4 (24) and the 19Sproteasomal component S6 (25). Thesesites exert important roles in mediating thetoxicity of alpha-synuclein and parkin (25,26). However, whether these sites wereoccupied by R239C GFAPs still awaitfurther study.

The R239C GFAP probablyexerted its inhibition by direct binding tothe proteasome. Thus, we found that theprotein degradation of proteasomespreincubated with the R239C GFAP wasinhibited. The binding of R239C GFAPpresumably prevented the fluorogenicpeptide from entering the catalyticcompartment of 20S proteasomecomplexes, either by binding at theentrance of the peptide channel or byinteracting directly with the inner catalyticcompartment of the proteasome, in amanner suggested for the toxicity ofamyloid beta peptide in Alzheimer disease(27). Either external or internal bindingcould block the entrance of othersubstrates into the proteasome chamber oralter the conformation of the activeproteolytic site.

αB-crystallin shifts monomer-oligomerequilibrium of R239C GFAP andreinstates proteasome degradativefunction

αB-crystallin normalized theequilibrium between oligomers andmonomers to that which resembled wtGFAP. αB-crystallin also mitigated theR239C GFAP-dependent proteasomeinhibition, as measured both byfluorogenic substrate assays and by thedegree of proteolysis of GFAP itself.Thus, a restoration of the normal GFAPsize range correlated with a restoration ofproteolysis. We presume these twophenomena are causally related.

10

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

The interactions between αB -crystallin and GFAP are complex. Shsps,α B-crystallin and hsp27, do bindintermediate filaments, both the solubleand the filamentous forms (7). αB-c rys t a l l i n can inh ib i t GFAPpolymerization (28), and can preventGFAP assembly in vitro , as well asfilament-filament interactions in vitro (7).In cultured astrocytes, αB-crystall inoverexpression can reorganize abnormalintermediate filament aggregates into anormal filamentous network (3) and canspread filament bundles. In the presentstudy, αB-crystallin overexpressionmitigated the proteasome inhibitionproduced by the R239C GFAP.Furthermore, αB-crystallin associates withGFAP aggregates in AxD brains, and withaggregates in cell models produced byoverexpression of R239C GFAP (thisstudy) and R416W (29).

How does αB-crystallin exert itseffects on proteasome dysfunction? Wethink the most likely effect is todepolymerize/disaggregate the largerGFAP oligomers, presumably by bindingto GFAP and interfering with GFAPassembly. It seems less likely that αB -crystallin relieves the proteasomeinhibition by preventing the R239C GFAPfrom binding to proteasomes. Theinteraction between αB-crystallin andGFAP appears to be more critical than anypossible direct interactions between αB-crystallin and the 20s proteasome for tworeasons. First, we pre-incubated αB -crystallin with GFAP and found a shift inthe oligomer profile in the absence ofproteasomes. Second, while αB-crystallinis reported to interact with the proteasomesubunit, C8/α7, its interaction with the full20S proteasome appears much weaker(30). However, we cannot completely ruleout the possibility that a binding of αB-crystallin to proteasomes could relieve theGFAP-induced inhibition, although it isdifficult to formulate such a mechanism.

In cells with accumulated R239CGFAP, the overexpression of αB-crystallin

restored proteasomal protein degradationto a normal level, suggesting that in theliving cell αB-crystallin alters thedistribution of “soluble” unassembledGFAP proteins between monomeric andoligomeric states and helps prevent anaccumulation of toxic oligomers. Onemight ask, however, why there is suchstrong proteasomal inhibition in AxD,given the markedly increased levels ofαB-crystallin in astrocytes. The degree ofinhibition may be a matter of relativelevels of GFAP and αB-crystallin. It isimportant to consider that although thelarge majority of cellular GFAP iscytoskeletal-associated (Triton X-100insoluble), there is a “soluble” pool andcells expressing the R239C GFAP containa relatively larger soluble pool than cellsexpressing wt GFAP (31). A furthercomplication is that αB-crystallin shiftsfrom a largely cytosolic form to acytoskeletal-associated form in thepresence of GFAP accumulation, probablybeing bound up by the excess of filaments.Thus, there is less soluble αB-crystallinavailable in a soluble pool to bind toGFAP oligomers.

Can proteasome inhibition explainmetabolic changes seen in AxD?

Proteasome inhibition would have anumber of consequences for astrocytefunction (Fig 6). It could activate theSAPK/JNK stress pathway, andcompromise the astrocytes’ ability towithstand apoptotic stress stimuli (5).Meanwhile, cel ls wil l developcompensatory mechanisms to compensatefor the dysfunction of proteasomes. Thesemechanisms include the induction of shspgene transcription, the accumulation ofshsps, and the induction of autophagy(13). Inhibiting proteasomes in severalcell systems leads to an induction ofoxidative response genes, via the Nrf2-regulated transcriptional pathway (32).Indeed, as predicted, analysis of braintissues in knock-in and transgenic mousemodels of AxD showed a markedoxidative stress response and the induction

11

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

of several oxidative stress response genes(18, 33).

REFERENCES

1. Brenner, M., A.B. Johnson, O. Boespflug-Tanguy, D. Rodriguez, J.E. Goldman, and A.Messing. (2001) Nat Genet. 27, 117-20.

2. Li, R., A.B. Johnson, G. Salomons, J.E. Goldman, S. Naidu, R. Quinlan, B. Cree, S.Z. Ruyle,B. Banwell, M. D'Hooghe, J.R. Siebert, C.M. Rolf, H. Cox, A. Reddy, L.G. Gutierrez-Solana, A. Collins, R.O. Weller, A. Messing, M.S. van der Knaap, and M. Brenner. (2005)Ann Neurol. 57, 310-26.

3. Koyama, Y., and J.E. Goldman. (1999) Am J Pathol 154, 1563-72.4. Hsiao, V.C., R. Tian, H. Long, M. Der Perng, M. Brenner, R.A. Quinlan, and J.E. Goldman.

(2005) J Cell Sci. 118, 2057-2065.5. Tang G, Xu, Z, and Goldman, JE (2006) J. Biol. Chem., 50:38634-38643.6. Xu Z, Maroney AC, Dobrzanski P, Kukekov NV, Greene LA.(2001) Mol Cell Biol. 21, 4713-

24.7. Perng, M.D., P.J. Muchowski, P. van den IJssel, G.J. Wu, A.M. Hutcheson, J.L. Clark and

R.A.Quinlan. (1999) J. Biol Chem. 274, 33253-43.8. Wilk S, Chen WE. (2001) Curr Protoc Protein Sci. Chapter 21:Unit 21.6.9. Johnston, J.A., C.L. Ward, and R.R. Kopito. (1998) J Cell Biol. 143, 1883-98.10. Grossi, D.S.M., D.S.C. Martins, F. Harper, M. Olink-Coux, M. Huesca, K. Scherrer. (1988) J

Cell Biol. 107, 1517-30.11. Olink-Coux M, Arcangeletti C, Pinardi F, Minisini R, Huesca M, Chezzi C, Scherrer K.

(1994) J Cell Sci. 107, 353-66.12. Arcangeletti C, Sütterlin R, Aebi U, De Conto F, Missorini S, Chezzi C, Scherrer K. (1997)

J Struct Biol. 119, 35-58.13. Malloch GDA, Clark JB, Burnet FR (1987) J Neurochem 48, 299-306.14. M. Banay-Schwartz M, Kenessey A, DeGuzman T, Lajtha A, Palkovits M (1992) Age 15, 51-

54.15. Reichenbach A, Wolburg H (2005). in Neuroglia, Kettenmann H and Ransom B, eds. Oxford

Univ. Press, Oxford. pp.19-3516. Tang G, Yue Z, Talloczy Z, Hagemann T, Cho W, Messing A, Sulzer DL, Goldman JE.

(2008) Hum Mol Genet. 17, 1540-55.17. Middeldorp J, Kamphuis W, Sluijs JA, Achoui D, Leenaars CHC, Feenstra MGP, van Tijn P,

Fischer DF, Berkers C, Ovaa H, Quinlan RA, Hol EM (2009) FASEB J 23,2710-2726.18. Hagemann TL, Gaeta SA, Smith MA, Johnson DA, Johnson JA, Messing A. (2005) Hum

Mol Genet. 15, 2443-58.19. Kristiansen M, Deriziotis P, Dimcheff DE, Jackson GS, Ovaa H, Naumann H, Clarke AR, van

Leeuwen FW, Menéndez-Benito V, Dantuma NP, Portis JL, Collinge J, Tabrizi SJ. Mol Cell.26, 175-188.

20. Caughey B, Lansbury PT. (2003) Ann Rev Neurosci. 26, 267-298.21. Wenzel T, Baumeister W. (1995) Nat Struct Biol. 2, 199-204.22. Orlowski M, Wilk S. (2003) Arch Biochem Biophys. 415, 1-5.23. Alvarez-Castelao B, Castaño JG. (2005) FEBS Lett. 579, 4797-802.24. Dächsel JC, Lücking CB, Deeg S, Schultz E, Lalowski M, Casademunt E, Corti O, Hampe C,

Patenge N, Vaupel K, Yamamoto A, Dichgans M, Brice A, Wanker EE, Kahle PJ, Gasser T.(2005) FEBS Lett. 579, 3913-9.

25. Snyder H, Mensah K, Theisler C, Lee J, Matouschek A, Wolozin B. (2003) J Biol Chem.

12

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

278, 11753-9.26. Ghee M, Melki R, Michot N, Mallet J. (2005) FEBS J. 272, 4023-33.27. Gregori L, Hainfeld JF, Simon MN, Goldgaber D. (1997) J Biol Chem. 272, 58-62.28. Nicholl ID, Quinlan RA. (1994) EMBO J. 13, 945-53.29. Perng MD, Su M, Wen SF, Li R, Gibbon T, Prescott A, Brenner M, Quinlan RA. (2006) Am J

Hum Gen 79, 197-213.30. Boelens WC, Croes Y, and deJong WW. (2001) Biochim Biophys Acta 1544, 311-319.31. Tian R, Gregor M, Wiche G, Goldman JE. (2006) Am J Pathol. 168, 888-97.32. Kwak MK, Itoh K, Yamamoto M, Kensler TW. (2002) Mol Cell Biol. 22, 2883-92.

32.33. Hagemann TL, Connor JX, Messing A. (2006) J Neurosci. 26, 11162-73.

FOOTNOTES

We Thank Prof Albee Messing and Prof Michael Brenner for helpful discussions, and Prof. TomK Hei and Dr. Michael Partridge for help with our proteasome activity assay. This work wassupported by National Institutes of Health Grant PO1NS42803 and P30-HD03352.

The abbreviations used are: GFAP, Glial fibrillary acidic protein; AxD, Alexanderdisease; shsp, small heat shock protein.

13

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

FIGURE LEGENDS

Figure 1 GFAP accumulation decreases proteasome activity(A) Proteasome proteolytic activity in U251 cells stably expressing EGFP vector (V), wt GFAP (wt)and R239C GFAP (mt). 48h after plating, cells were lysed and the proteasome PGPH andChymotrypsin like protease activities were analyzed by using the fluorescent substrates S-II and S-III,respectively. Protease activities in cells expressing EGFP-C1 vector were set to 100%. The results arethe average from 5 independent experiments. * p<0.05, ** p<0.001, by two-tailed ANOVA.(B) Shift of 20S proteasome from soluble pool to insoluble pool in U251 cells stably expressing wt ormt GFAP. The U251 cells were harvested 48 h after plating at 80% confluency. Triton X-100 soluble(30µg), insoluble (10µg), or total protein (15 µg) was analyzed by Western blotting with an anti-20Santibody and, after stripping, with an anti-GFAP and finally with an anti- GAPDH antibody tonormalize for loading.(C) Quantitation of levels of 20S in total, soluble and insoluble fractions of cell lysates from U251cells expressing EGFP-C1 vector (V), wt (WT) and mt GFAPs (mt). In comparing cells transfectedwith vector onlyas control, protein levels of 20S in wt or mt GFAP expressing cells were reduced markedly in thesoluble fraction but enhanced significantly in the insoluble fraction. * p<0.05, ** p<0.001, by two-tailed ANOVA.(D) Decreased proteasome proteolytic activity in AxD brain with R239C, R239H and R416Wmutation. Protease activities in control subject I was set to 100%. Each result represents a mean±SD of5 independent experiments. * p<0.05, ** p<0.001, by two-tailed ANOVA.(E) Western blotting analysis of 20S proteasome in brains of AxD patients (control subject II notincluded).

Figure 2 R239C GFAP aggregates recruit ubiquitin proteasome system components andproteasome inhibitors regulate GFAP accumulation(A) GFAP inclusions in R239C mutant GFAP cells were labeled with antibodies to ubiquitin (upper)and the proteasomal 20S complex (lower). U251 cells stably expressing mutant GFAP were treatedwith Triton-X containing cytoskeletal buffer and fixed with 4% PFA, then were immunostained forGFAP, 20S or ubiquitin. Merged images on right show 20S or ubiquitin in red, GFAP-GFP in greenand the overlap in yellow. Aggregates are indicated by arrows and nuclei by asterisks. Note that ineach panel cells without aggregates were included for control. After removal of soluble cytosolicprotein, most of the proteasome components remain in cell nuclei. Scale bar, 10µm.(B) Immunohistochemistry of brain tissue of an AxD patient and a control, using antibodies againstGFAP, 20S and ubiquitin. Hematoxylin and eosin staining of the AxD brain showed extensive,brightly eosinophilic Rosenthal fibers (arrows), some of which surround vessels (H&E, 40X). NoRosenthal fibers are present in control white matter. GFAP immunostaining of control white mattershows cells with typical, process-bearing, astrocyte morphology and many thin astrocyte processes,whereas in AxD, GFAP immunostains highlights the Rosenthal fibers. Correlated with the GFAPimmunostaining, the signal for both 20S and ubiquitin was localized to Rosenthal fibers . V, bloodvessel; arrows , Rosenthal fibers. Scale bar, 100 µm.(C) Effects of proteasome inhibition on GFAP accumulation. U251 cells stably expressing vectorcontrol (V), wt GFAP (WT) and R239C GFAP (mt) were incubated with medium with or withoutMG132 or Lactacystin for 12 hrs. Total cell lysates were subjected to SDS–PAGE and Western-blotting analysis with antibodies against GFAP and GAPDH.(D) Percentage of cells with aggregates afterexposure to proteasome inhibitors. Results are average +SD from three independent experiments, with at least 400 cells counted in each. A larger percentage ofthe R239C GFAP expressing cells, compared to wt expressing cells, contained aggregates under alltreatment conditions. Different treatments were then compared within the wt or R239C GFAPexpressing group, using DMSO as control. ** Compared to DMSO control, by two-tailed ANOVA, P<0.001.

14

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Figure 3 20S proteasome associates with soluble and aggregated GFAPs(A) Co-immunoprecipitation of 20S proteasome and GFAP revealed that both wt and R239C mutantGFAPs were present in complexes with 20S proteasomes. Cos7 cells were transiently transfected withpcDNA3-Flag vector, vectors expressing wt and R239C GFAP. Forty-eight hours after transfection,cells were harvested and lysed with 1X RIPA buffer. GFAP was immunoprecipitated with anti-flag M2agarose affinity gel, and 20S proteasome complex was precipitated by protein A sepharose beadsconjugated with an anti 20S antibody. C, untransfected cells; V, pcDNA3-Flag vector transfected cells;wt and mt, cells expressing flagtagged wt or mt GFAP; NC, Protein A-Sepharose beads without GFAPand 20S antibody.(B) 20S proteasome associates with Triton-insoluble GFAP aggregates in U251 cells. U251 cellsstably expressing wt or mt GFAPs were treated with Triton-X containing cytoskeletal buffer and fixedwith 4% PFA and then examined for GFAP-GFP (green) and 20S(red) fluorescence. GFAP aggregatesare indicated by arrows and nuclei by asterisks. Scale bar, 10µm.(C) 20S proteasome associates with insoluble, filamentous GFAPs in vitro. Both wt and R239C GFAPwere assembled in vitro as described (7), with or without purified 20S proteasomes. Without GFAP(left panel), the 20S proteasomes remain in the soluble “S” fraction and none appeared in the pelletfraction “P”. “C” is a control lane loaded with 20S proteasome subunits. With reassembled wt andR239C GFAP (middle and right panels) the 20S subunits were recovered in the pellet fraction “P”.Western blots were performed with both anti-GFAP and anti-20S antibodies.(D) Both soluble and aggregated forms of GFAP inhibit proteasome function. Purified proteasomeswere incubated without (PS only) or with soluble non-assembled (S) or assembled (P) wt and R239C(RC) GFAPs. One hundred µg of soluble(S) or assembled (P) GFAPs were incubated with proteasome(100µg) for 30 min. in a total volume of 200 µ l. Proteasome activity was then analyzed usingfluorescent substrates (100µM) with a fluorescence microplate reader at an excitation wavelength of360nm and an emission wavelength of 440nm. Compared to those exposed to substrates only (PSonly), both soluble mt GFAP (RC/S) and pellet R239C GFAP (RC/P) significantly inhibitedproteasome PGPH and chymotrysin like proteolytic activities. ** Compared to PS only, by two-tailedANOVA, P <0.001.

Figure 4 R239C GFAP inhibited proteasome activity in a non-competitive manner(A) R239C mt GFAPs inhibited proteasome activity in a non-competitive manner. Purifiedproteasomes (100µg) were incubated without (PS only) or with soluble (S) or polymerized (P) wt andR239C GFAPs (100µg), and 20 min later, proteasome substrates II or III was added at differentconcentration (25, 50, 75, 100, 125µM) and proteasome proteolytic activity was analysed.(B) PGPH and chymotrypsin-like protease activities at different GFAP concentrations (50, 100, 150,200µg/200µl), with the fluorescent peptide substrate concentration fixed at 100µM. Compared with20S only, * p<0.05; ** p<0.001, by two-tailed ANOVA.(C) Proteasome degradation of wt and R239C GFAP. “Soluble” wt and R239C GFAPs were incubatedwith or without 20S proteasomal subunits and with or without MG132, electrophoresed on SDS gels,and immunoblotted with anti-GFAP antibody. This experiment was repeated 3 times with similarresults.

Figure 5 αB-crystallin restores proteasome proteolytic activity reduced by R239C GFAPs(A) In vitro preincubation of R239C GFAP with αB-crystallin (aBC) reversed the inhibitory activityof the R239C GFAP. Unexposed proteasomes (20S only), MG 132 control treatment (MG132). *p<0.05; ** p<0.001.(B) Adenoviral introduction of αB-crystallin increased proteasome activity in U251 cells expressingR239C GFAP. Protease activities in cells expressing EGFP-C1 vector are set to 100%. The results arethe average from 5 independent experiments. Adenoviral infection did not affect the proteasomeactivity. αB-crystallin expression increased PGPH proteolytic activity in both wt and R239C GFAP

15

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

expressing U251 cells and chymotrypsin-like proteolytic activity in R239C GFAP expressing cells. *p<0.05; ** p<0.001.(C) Morphological changes of GFAP expressing U251 cells. U251 cells stably expressing wt orR239C GFAP were infected with αB-crystallin expressing adenovirus, then fixed and processed forGFP fluorescence. αB-crystallin dispersed the GFAP aggregates in wt expressing cells, and shiftedthem into filamentous structures. GFAP aggregates (arrow heads); while αB-crystallin. expressionreversed the formation of small GFAP inclusions, shifting the GFAP to large perinuclear aggreagatesand filaments. Scale bar: 20µm.(D) Oligomerization profile analysis of soluble wt and R239C GFAP preincubated with (right panel)or without αB-crystallin (left panel). The same amounts of soluble wt and R239C GFAP, in theabsence or the presence of αB-crystallin, were subjected to non-SDS, non-reducing, 8% PAGE gelelectrophoresis and then immunoblotted for GFAP.(E) αB-crystallin promotes degradation of the R239C GFAP by the 20S proteasome. The sameamounts of soluble wt and R239C GFAP (100µg), in the absence or the presence of αB-crystallin(200µg), were incubated with 20S proteasome (100µg). The incubation products were then subjectedto 4-12% SDS gradient gel electrophoresis and Western blotting. After preincubation with αB-crystallin, the degree of proteolytic breakdown of the R239C GFAP was similar to that of the wtGFAP.(F) The optical density of degraded GFAP protein bands in (E) was quantitated and plotted to analyzethe effects of αB-crystallin on GFAP protein degradation. Compared to wt GFAP, R239C GFAP wasless susceptible to proteasome degradation (P<0.05). Preincubation with αB-crystallin promotedproteasome degradation of both wt (P<0.05) and R239C GFAP(P<0.05).

Figure 6. Proposed consequences of proteasome inhibition by the AxD GFAP R239C mutationAccumulation of GFAP oligomers and aggregates inhibits proteasome proteolytic activity. Thisinhibition, in turn, activates MLK/JNK/p38 stress kinase pathways, which upregulate αB-crystallintranscription and activate autophagy. Both αB-crystallin and autophagy act to decrease GFAP proteinlevels, the former by inhibiting aggregate formation and reducing inhibitory GFAP oligomer levels,and the latter by degrading GFAP aggregates. Proteasome inhibition also activates ARE-regulatedgene expression, via increased levels of Nrf2.

16

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

17

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

18

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

19

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

20

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

21

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

22

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

23

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Guomei Tang, Ming D. Perng, Sherwin Wilk, Roy Quinlan and James E. GoldmanB-crystallin, reverses the inhibition

system in Alexander disease astrocytes, and the small heat shock protein, alpha Oligomers of mutant Glial Fibrillary Acidic Protein (GFAP) inhibit the proteasome

published online January 28, 2010J. Biol. Chem. 

  10.1074/jbc.M109.067975Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

by guest on March 29, 2018

http://ww

w.jbc.org/

Dow

nloaded from