Biochemical and Biophysical Research Communications 432 (2013) 339–344

Contents lists available at SciVerse ScienceDirect

Biochemical and Biophysical Research Communications

journal homepage: www.elsevier .com/locate /ybbrc

High glucose and diabetes modulate cellular proteasome function: Implicationsin the pathogenesis of diabetes complications

Saeed Yadranji Aghdam a, Zafer Gurel a, Alireza Ghaffarieh a, Christine M. Sorenson b, Nader Sheibani a,c,⇑a Department of Ophthalmology & Visual Sciences, University of Wisconsin School of Medicine and Public Health, Madison, WI 53792-4673, USAb Department of Pediatircs, University of Wisconsin School of Medicine and Public Health, Madison, WI 53792-4673, USAc Department of Pharmacology, University of Wisconsin School of Medicine and Public Health, Madison, WI 53792-4673, USA

a r t i c l e i n f o

Article history:Received 24 January 2013Available online 4 February 2013

Keywords:HyperglycemiaDiabetic retinopathyDiabetic nephropathyPericytesPA28 proteasome regulator

0006-291X/$ - see front matter � 2013 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.bbrc.2013.01.101

⇑ Corresponding author. Address: University of Withalmology and Visual Sciences, 600 Highland Avenu53792-4673, USA.

E-mail address: nsheibanikar@wisc.edu (N. Sheiba

a b s t r a c t

The precise link between hyperglycemia and its deleterious effects on retinal and kidney microvascula-ture, and more specifically loss of retinal perivascular supporting cells including smooth muscle cell/peri-cytes (SMC/PC), in diabetes are not completely understood. We hypothesized that differential cellularproteasome activity contributes to sensitivity of PC to high glucose-mediated oxidative stress and vascu-lar rarefaction. Here we show that retinal endothelial cells (EC) have significantly higher proteasomepeptidase activity compared to PC. High glucose treatment (HGT) increased the level of total ubiquitin-conjugated proteins in cultured retinal PC and EC, but not photoreceptor cells. In addition, in vitro protea-some activity assays showed significant impairment of proteasome chymotrypsin-like peptidase activityin PC, but not EC. The PA28-a/-b and PA28-b/-c protein levels were also higher in the retina and kidneyglomeruli of diabetic mice, respectively. Our results demonstrate, for the first time, that high glucose hasdirect biological effects on cellular proteasome function, and this modulation might be protective againstcellular stress or damage induced by high glucose.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Diabetic retinopathy (DR) and diabetic nephropathy (DN) arethe leading causes of blindness and end-stage renal failure in dia-betic patients, respectively [1–3]. Diminished number of retinal PC[4–5] and hypertrophy of the renal mesangial cells, as perivascularcells of the glomerulus with similar functions as PC, are hallmarksof established DR and DN [1–3,6–7]. The HGT of both retinal PC andEC promote oxidative stress, endoplasmic reticulum (ER) stress,and apoptosis in PC, but not in EC [8–9]. However, the molecularbasis for selective vulnerability of PC in diabetic retinal and kidneycomplications, and higher sensitivity of retinal PC to high glucosecompared to EC are not completely understood. We hypothesizedthat differential cellular proteasome activity contributes to selec-tive sensitivity of PC to hyperglycemia.

Ubiquitin Proteasome System (UPS) is responsible for proteinquality control and degradation. The conjugation of ubiquitin totarget proteins primes them for UPS-mediated degradation [10].The proteasomes are multiprotein barrel shaped assemblies withan approximate molecular weight of 2500 kDa and are considered

ll rights reserved.

sconsin, Department of Oph-e, k6/456 CSC, Madison, WI

ni).

the proteolytic machinery, which regulates the turnover of eukary-otic proteins. Proteasomes are composed of 20S core subunit and19S or 11S regulatory subunits capable of binding to both endsof the barrel shaped core subunit and stimulating the proteolyticactivity of the proteasomes. UPS also comprises many other en-zymes involved in the ubiquitin activation (E1s), conjugation(E2s), ligation (E3s) and removal from target proteins by deubiqui-tylating enzymes (DUBs) [10].

The 11S proteasome regulatory subunit comprises the PA28-a,PA28-b and the PA28-c proteins encoded by three different genes.All these proteins form a heptameric ring-shaped complex and,while the PA28-a and PA28-b proteins form a heptameric complextogether (a3b4), the PA28-c only exists in homoheptameric form.Although the important role of PA28 proteins in intracellular anti-gen processing and presentation to immune cells has been demon-strated, the PA28-a and PA28-b proteins can also mediate cellularresponse to oxidative stress [11–12].

Recent studies have demonstrated that high glucose, the glu-cose-mediated post-translational protein modifications, and diabe-tes are modulators of proteasome activity [13–15]. These reportslink diabetes or hyperglycemia to UPS and raise many questionson how diabetes or hyperglycemia can modulate proteasome tar-geting and activity, and whether this modulation occurs in a cell-specific manner.

340 S.Y. Aghdam et al. / Biochemical and Biophysical Research Communications 432 (2013) 339–344

2. Materials and methods

2.1. Cell culture

The isolation and culture of mouse retinal PC and EC were pre-viously described [16–17]. Choroid EC (ChEC) were isolated fromthe mouse choroid using magnetic beads coated with anti-PECAM-1antibody and grown similar to retinal EC. NIH3T3 cellsand 661W mouse photoreceptor cells were obtained from ATCC(Manassas, VA). Low glucose level was 5 mM and for HGT the cellswere treated for 5 days with 30 mM D-glucose (Sigma, St. Louis,MO). For osmolarity control the cells were treated with 5 mMD-glucose and 25 mM L-glucose (Sigma). For cycloheximide (CHX;Sigma) treatment, cells were treated at final concentration of100 lg/ml for indicated times.

2.2. Western blotting

For immunoblots, the cultured cells and the retina tissue werelysed in RIPA lysis buffer (20 mM Tris–HCl (pH 7.5), 150 mM NaCl,1 mM EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate,5 mM NaF, 1 mM Na3VO4) supplemented with protease inhibitorcocktail (Roche, Nutley, NJ). The lysates were collected in 1.5 mltubes, sonicated, and protein concentration was measured usingBCA Protein Assay Kit (Pierce; Rockford, IL).Total protein (50 lg)was separated by SDS–PAGE (4-20% Tris-Glycine gels; Invitrogen,Carlsbad, CA) and transferred to nitrocellulose membrane. Proteinexpression was analyzed by incubating with specific primary anti-bodies (Supplementary Table 1) at 4 �C overnight. Blots werewashed, incubated with appropriate HRP-conjugated secondaryantibody and developed using ECL. Mouse b-actin was used forloading control.

2.3. In vitro proteasome peptidase assay

Proteasome peptidase assays were performed as described byBech-Otschir et al. [18]. The cells or tissues were lysed in the lysisbuffer containing 50 mM Tris/HCl pH 7.4, 1 mM ATP, 10% glycerol,0.1% NP40, 2 mM MgCl2, 1.5 mM DTT and 0.03% SDS. In total, 50 lg(1 lg/ll) of protein lysates were loaded into a dark 96-well micro-plate and probed with 100 lM of fluorogenic peptide substrates:Suc-Leu-Leu-Val-Tyr-AMC for chymotrypsin-like, Ac-Arg-Leu-Arg-AMC for trypsin-like, and Z-Leu-Leu-Glu-AMC for caspase-like(Enzo Life Sciences, Farmingdale, NY). The plates were then incu-bated at 37 �C for 30 min, and the reaction was stopped by theaddition of 200 ll of 100% ethanol. The fluorescence signal inten-sity of the samples was measured using a luminescence plate read-er (PerkinElmer, Wellesley, MA). The excitation and emissionwavelengths were 355 and 460 nm, respectively. Boiled protein ly-sates were used as blank samples and test values were normalizedto the respective blanks.

2.4. Mice

The wild type and heterozygous Akita/+ (Ins2Akita/+) male miceon C57BL/6j background were purchased from Jackson Laborato-ries (Bar Harbor, ME) and maintained in our facility. Genotypingwas performed by PCR using specific primers (SupplementaryTable 2). All animal procedures were approved by the InstitutionalAnimal Care and Use Committee of the University of Wisconsin,School of Medicine and Public Health. Only male Akita/+ micedevelop diabetes by 4-weeks of age and have an average lifeexpectancy of 10 months. They develop some of the early non-proliferative retinopathies by six-months of age [19].

2.5. Immunofluorescence studies

The cryosections, prepared from eyes and kidneys harvestedfrom age-matched male wild type and Akita/+ mice, were fixedin ice-cold acetone for 30 min, blocked (3% BSA, 0.1% Tween-20)for 1 h and incubated with the desired primary (SupplementaryTable 1) and secondary antibodies (Alexa Fluor-488 or -594 conju-gated; Invitrogen). Control samples were incubated with relevantIgG and nuclei were counterstained with DAPI (Invitrogen).Microscopic slides were photographed using a Zeiss fluorescencemicroscope in digital format (Carl Zeiss, Thornwood, NY).

2.6. Statistical analysis

For Western blot densitometry analysis the wand tool of Image-J, and for statistics the GraphPad Prism software were used, respec-tively. Data are expressed as mean ± sem. Statistical significancewas assessed by unpaired t test. Values were considered statisti-cally significant at p < 0.05.

3. Results

3.1. Retinal EC exhibited enhanced proteasome peptidase activity

To test the effects of HGT on proteasomal degradation capacityof retinal PC, EC and retinas, synthetic fluorogenic proteasome sub-strate peptides were assayed in the presence of ATP [18]. Regard-less of any change in the peptidase activity by HGT, we foundthat retinal EC had substantially higher peptidase activity for allthe substrates compared to other cells (Fig. 1A–C). Retinal EC, com-pared to PC, showed approximately 3.4-, 8.1- and 6.5-fold higherchymotrypsin-like, trypsin-like and caspase-like peptidase activi-ties, respectively (Fig. 1A–C). HGT of cultured cells resulted in amodest reduction of peptidase activity for most of the tested sub-strates, except for EC, which showed a significant rise in peptidaseactivity after HGT. Retinal PC after HGT showed a significant reduc-tion in chymotrypsin-like peptide degradation. Similar to EC, but toa lower extent, the Ins2Akita/+ retinas also showed a significant in-crease in chymotrypsin-like and trypsin-like peptidase activitycompared to age-matched control retinas (Fig. 1A–C). Surprisingly,the ChEC responsible for the nourishment of outer retina did notshow as high of peptidase activity as retinal EC (Fig. 1A–C). Thismay stem from the differences in the physiology of these EC, andsuggest that ChEC are more susceptible to cytotoxic effects ofhyperglycemia.

3.2. HGT increased the total levels of ubiquitinated proteins and PA28-a/-b levels in retinal vascular cells

Using Western blot to detect universal ubiquitination in cul-tured cells, we found that HGT resulted in elevation of total proteinubiquitination in retinal PC, EC, and NIH3T3 cells, but not in 661Wphotoreceptor cells (Fig. 2A). The observed rise in ubiquitinationfollowing HGT was similar to the increased ubiquitination reportedfor cultured myocytes incubated with oxidant reagent (H2O2),which was due to attenuation of proteasome activity [11].

We next determined whether HGT-mediated impairment ofproteasome activity was responsible for increased levels of ubiqui-tinated proteins by analyzing the clearance of ubiquitinated pro-teins after CHX treatment and inhibition of new proteinsynthesis. Western blot for ubiquitin after 12 h of CHX treatmentshowed reduction in total amount of ubiquitinated proteins inHGT similar to normal glucose conditions in all cell types examined(Fig. 2A). Thus, HGT enhanced protein ubiquitination with minimal

Fig. 1. The in vitro proteasome peptidase activity of cultured retinal cells and retinas under low and high glucose conditions. (A) Chymotrypsin-like activity was significantlyreduced by HGT in RPC (p < 0.0066), ChEC (p < 0.0026) and 661W (p < 0.0045) whereas HGT increased the degradation activity in REC (p < 0.0120) and retina (p < 0.0003). RECin LGT showed 3.4-fold and 4.8-fold higher chymotrypsin-like activity compared to RPC and ChEC cells, respectively. (B) Trypsin-like degradation was increased in REC(p < 0.0002) and retina (p < 0.0012), and reduced by HGT in ChEC (p < 0.0002), NIH3T3 cells (p < 0.0071) and 661W cells (p < 0.0015). The REC showed 8.1 and 4.6-fold highertrypsin-like activity than RPC and ChEC under LGT conditions, respectively. (C) Significant increases in caspase-like activity by HGT in EC (p < 0.0001), and reduction in ChEC(p < 0.0056) and NIH3T3 (p < 0.0031) cells. The REC showed approximately 6.5- and 4-fold higher caspase-like activity compared to RPC and ChEC, respectively. Samplenumber for all assays (n = 3). Error bars, Mean ± SEM; ⁄statistically significant; ns, not-significant.

Fig. 2. Effect of HGT on ubiquitination and PA28-a/-b protein levels in cultured retinal cells. (A) Western blot for ubiquitinated and PA28-a/-b proteins in retinal PC and EC,NIH3T3, and 661W cells. (B) The quantification of Western blots for PA28-b protein under LGT and HGT showing the percent change in RPC and NIH3T3.

S.Y. Aghdam et al. / Biochemical and Biophysical Research Communications 432 (2013) 339–344 341

effect on the clearance of the ubiquitinated proteins byproteasomes.

We then asked whether HGT could also influence the composi-tion of the proteasome regulatory subunits to handle the rise insubstrate levels. Thus, we investigated the levels of 11S protea-some regulatory subunits (PA28-a/-b in the retinal vascular cellsafter HGT. Indeed, HGT elevated the levels of PA28-a/-b proteinsin retinal PC and NIH3T3 cells by 36% and 310%, respectively(Fig. 2B), but apparently not in retinal EC and 661W cells (Fig. 2Aand not shown). These results suggested differential requirementfor cells of 11S regulatory proteasome subunits under HGT condi-tions to accommodate the rise in ubiquitination. In addition, the

parallel increase in both PA28-a and PA28-b levels by HGT in ret-inal PC and NIH3T3 cells was also in agreement with the ability ofPA28-a and PA28-b to form a heptameric (a3b4) complex allowingtheir concerted stability and activity [20–21].

3.3. Diabetic mice had higher levels of PA28-a/-b proteins in theirretinas

We next asked whether the retinas from Ins2Akita/+ mice withestablished diabetes demonstrate any changes in expression ofPA28-a/-b proteins. Western blot analysis of the total retinal ly-sates from adult Ins2Akita/+ mice with established diabetes showed

342 S.Y. Aghdam et al. / Biochemical and Biophysical Research Communications 432 (2013) 339–344

significant increase in the PA28-a/-b protein levels compared withnon-diabetic mice (Fig. 3A and B). However, contrary to HGT-in-duced rise in ubiquitination in retinal PC and EC, the retina lysatesfrom adult Ins2Akita/+ mice did not show any difference in totalubiquitination levels compared to control animals (not shown).This suggested that hyperglycemia-mediated rise in the PA28-a/-b proteins was independent of universal ubiquitination in the dia-betic retina.

Following the observed rise in PA28-a/-b levels in diabetic ret-ina, we used immunofluorescence staining to detect PA28-b in theretina of adult wild type and Ins2Akita/+ mice. Similar to adult8-months old wild type mice, the retinal PA28-b expression inIns2Akita/+ mice was only restricted to the superficial nucleatedlayer of the retina in the inner limiting membrane (ILM), whichappeared to be only one layer thick and were covered by GFAPpositive processes of the astrocytes (Fig. 3C and D) [22]. We couldalso detect focal signal for PA28-b on retinal blood vessels in retinalflatmounts (Supplementary Fig. 1). The immunoreactivity site forthe PA28-b in retina suggested that the PA28-b positive cells mightbe retinal ganglion cell, therefore we performed co-stainingwith bIII-tubulin [23] but there was no overlapped staining(Supplementary Fig. 2A). Co-staining with PA28-b and PC markerPDGF-Rb (Supplementary Fig. 2B), and microglial marker Iba1[24](not shown), also did not show overlapped staining.Altogether, the retinal PA28-b immunofluorescence signal labeleda one-layer thick nucleated compartment restricted to the ILMlayer, did not vary between wild type and diabetic mice, and did

Fig. 3. Increased levels of PA28-a/-b proteins in the retinas of diabetic mice. (A) Westernmice. (B) Quantification of the combined mean value of PA28-b levels after normalizing to(C) PA28-b immunoreactivity (red) in the inner limiting membrane (ILM) layer of the wildof retina did not show any immunoreactivity for PA28-b. (D) High magnification of the rpositive cells. Nuclei are counterstained with DAPI. Scale bar 25 lm.

not co-localize with RGC, PC, microglia, and astrocytes. Therewas no detectable fluorescence signal for the PA28-c subunit ofthe 11S proteasome in the mouse retina sections (not shown).

3.4. Diabetic mice had higher levels of PA28-b/-c, Cul1 and Cul3proteins in their intraglomerular capillaries

Kidney sections from Ins2Akita/+ mice were also analyzed byimmunofluorescence for variation in 11S regulatory subunits ofthe UPS. The immunostaining pattern for 11S proteasome regula-tory subunits in the kidney sections varied between the wild typeand Ins2Akita/+ mice, and among the Ins2Akita/+ mice with longerduration of diabetes and severity of nephropathy. The severity ofthe nephropathy was judged both visually based on the degree ofkidney atrophy upon dissection and microscopically based on theglomerular hypertrophy revealed by immunofluorescence. Thekidneys from 3-months old wild type and Ins2Akita/+ mice did notshow any detectable staining for PA28-b or PA28-c (not shown).The adult 8-months or older non-diabetic mice did not revealany detectable immunoreactivity in their kidney sections for 11Sregulatory subunits (Fig. 4A and C). In contrast, 8-months old In-s2Akita/+ mice demonstrated PA28-b/-c staining in the glomeruli,which co-stained with PDGF-Rb (a mesangial cell marker; Fig. 4Band D) [25–26]. We also observed a similar co-staining patternwith VE-cadherin as EC marker and PA28-b/-c in adult 8-monthsold InsAkita/+ mice (Supplementary Fig. 3A and B). Collectively the11S regulatory proteins were increased in the vascular units of

blot analysis showed increased retinal PA28-a/-b levels in 7-months old Ins2Akita/+

b-actin band intensity showing fold change between wild type and Ins2Akita/+ mice.type mouse retina. Cells in the outer nuclear layer (ONL) or inner nuclear layer (INL)etinal ILM layer showing GFAP-positive astrocytes (green) surrounding the PA28-b

Fig. 4. Increased immunoreactivity for PA28 and Cul1/Cul3 proteins in the glomeruli of Ins2Akita/+ mice. No immunoreactivity for PA28-b (red, A) and PA28-c (red, C) in the 7-months old wild type glomeruli. Ins2Akita/+ mice showed high immunoreactivity for both PA28-b (red, B) and PA28-c (red, D) which overlapped with PDGF-Rb labeling of themesangial cells (green). Immunoreactivity for Cul1 (red, E) and Cul3 (red, G) were determined in 7-months-old wild type (non-diabetic) glomeruli. Ins2Akita/+ mice showedhigh immunostaining for both Cul1 (red, F) and Cul3 (red, H) which overlaped with PDGF-Rb labeling of the mesangial cells (green). Nuclei are counterstained with DAPI. Scalebar 25 lm.

S.Y. Aghdam et al. / Biochemical and Biophysical Research Communications 432 (2013) 339–344 343

the glomerulus. Thus, severe nephropathy was associated withmore robust 11S proteasome regulatory subunit accumulation inthe glomerular capillaries. The increased 11S proteasome regula-tory subunits in the diabetic kidney glomeruli were suggestive ofenhanced proteasome activation.

We then examined other components of the UPS in diabetic kid-neys, namely the Cul1 and Cul3, which are E3 ubiquitin ligasemembers of Cullin family [27,28]. A member of cullin family isshown in complex with O-GlcNAcase, an enzyme which regulatesprotein O-glycosylation events [29]. Thus, it is possible that cullinsmay be involved in specific targeting of glycosylated proteins forproteasome degradation. The antibodies against Cul1 and Cul3 inkidney sections showed similar pattern of immunoreactivity anal-ogous to 11S proteins (Fig. 4E–H). The Cul1 and Cul3 expressionwas almost undetectable in wild type mice (Fig. 4E and G) whereasIns2Akita/+ mice showed strong immunoreactivity and co-localiza-tion for these proteins in the glomeruli with PDGF-Rb (Fig. 4Fand H), which correlated with the severity of nephropathy. Similarto 11S proteins, we did not find any obvious changes in immuno-staining pattern of tubular components of the kidneys for theCul1 and Cul3 (not shown), indicating that glomerular capillariesof the kidney are targets of altered proteasome activity in diabetes.

4. Discussion

The exceptionally higher peptidase activity together with thelack of increase in PA28-a/-b levels following HGT in EC comparedwith PC indicated that retinal EC may possess higher proteasomalefficiency for clearance of damaged proteins compared with PC,even without a requirement for rise in 11S proteasome subunits.These observations are consistent with recognition of retinal PC,but not EC, as the early target of hyperglycemia damage and endo-plasmic reticulum stress in retinal vasculature during diabetes[9,30]. Elevated ubiquitination in retinal vascular cells inducedby HGT provides higher amounts of proteasome substrate for deg-radation. Efficient clearance of the ubiquitin-conjugated proteinsafter 12 h of CHX treatment indicated that HGT did not impair pro-

teasome activity in PC and EC. However, the modest but statisti-cally significant decrease in in vitro proteasome activity in PC,ChEC, NIH3T3, and 661W cells pointed to the impairment of thecatalytic activities of the proteasome by HGT. The interpretationof the differential peptidase activity among these cells requiresmore functional assessment of the proteasome activity and charac-terization of the proteasome quantity, subunit composition, andadjunct proteasome-associated proteins.

Photoreceptor cells did not show elevated HGT-mediated ubiq-uitination compared to PC and EC, which was suggestive of differ-ences in the metabolic responsiveness to HGT among retinalneuronal and non-neuronal cells. These results also suggest thatretinal neuronal cells may not be the primary target of hyperglyce-mia and pathogenesis of DR. The link between HGT and ubiquitina-tion is unclear but protein modifications such as protein oxidation[31] and O-glycosylation [15,32] target proteins for ubiquitination.The rise in PA28-a/-b levels in diabetic retinas provided an in vivolink between diabetes and modulation of proteasome activity.

The observed increase in immunoreactivity for PA28 and cullinproteins in the intraglomerular capillaries of Ins2Akita/+ mice and itscorrelation with the severity of the DN is the first report correlatingthe proteasome activation to DN. The primary site of DN damage inpatients is within the glomeruli with the net outcome of reducedlevels of glomerular filtration rate and proteinuria [33]. Thus, over-lapping rise in immunofluorescence signal for UPS components inglomerular capillaries implies a significant role for proteasome inpathogenesis of DN, and changes in expression/levels of these pro-teins may provide a suitable marker for detection of nephropathy.However, it is not clear whether increased proteasome activity isprotecting the glomeruli from diabetes or contributes to the exac-erbation of nephropathy. A recent study showed that MG132 med-iated blockage of proteasome in diabetic rats was protectiveagainst renal oxidative stress and DN [34]. Based on the role of11S subunits in protection against oxidative stress, and the re-ported improvement of DN symptoms with proteasome inhibition[34], it is possible that primary activation of the proteasome underdiabetic conditions in vulnerable organs provides temporary pro-tection against oxidative stress. However, its chronic activation

344 S.Y. Aghdam et al. / Biochemical and Biophysical Research Communications 432 (2013) 339–344

may ultimately result in more devastating tissue damage in vul-nerable organs including kidney.

In summary, the observed HGT-mediated modulation of UPScomponents or UPS-related substrate levels in vitro, and also in tis-sues vulnerable to diabetes damage including retina and kidney,identify hyperglycemia as key modulator of proteasome activity.Thus, differential activation of the 11S proteasome activator in cer-tain subset of vulnerable cells may provide protection againstdamage caused by glucose-mediated protein modifications includ-ing O-glycosylation [15] and protein oxidation [31]. However, thedirect contribution of these protein modifications and their clear-ance by proteasome to pathogenesis of DR and DN awaits futureinvestigation.

Acknowledgments

This work was supported by Grants EY016995, EY021357, andP30-EY016665, from the National Institutes of Health and an unre-stricted departmental award from Research to Prevent Blindness.NS is a recipient of a Research Award from American DiabetesAssociation, 1-10-BS-160 and Retina Research Foundation.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.bbrc.2013.01.101.

References

[1] M. Brownlee, Biochemistry and molecular cell biology of diabeticcomplications, Nature 414 (2001) 813–820.

[2] D. Aronson, Hyperglycemia and the pathobiology of diabetic complications,Adv. Cardiol. 45 (2008) 1–16.

[3] T.M. Curtis, T.A. Gardiner, A.W. Stitt, Microvascular lesions of diabeticretinopathy: clues towards understanding pathogenesis?, Eye (London) 23(2009) 1496–1508

[4] U. Denis, M. Lecomte, C. Paget, D. Ruggiero, N. Wiernsperger, M. Lagarde,Advanced glycation end-products induce apoptosis of bovine retinal pericytesin culture: involvement of diacylglycerol/ceramide production and oxidativestress induction, Free Radical Biol. Med. 33 (2002) 236–247.

[5] B. Liu, M. Bhat, A.K. Padival, D.G. Smith, R.H. Nagaraj, Effect of dicarbonylmodification of fibronectin on retinal capillary pericytes, Invest. Ophthalmol.Vis. Sci. 45 (2004) 1983–1995.

[6] R.N. Frank, Diabetic retinopathy, N. Engl. J. Med. 350 (2004) 48–58.[7] Y.S. Kanwar, L. Sun, P. Xie, F.Y. Liu, S. Chen, A glimpse of various pathogenetic

mechanisms of diabetic nephropathy, Annu. Rev. Pathol. 6 (2011) 395–423.[8] Y. Zhong, J.J. Wang, S.X. Zhang, Intermittent but not constant high glucose

induces ER stress and inflammation in human retinal pericytes, Adv. Exp. Med.Biol. 723 (2012) 285–292.

[9] Q. Huang, N. Sheibani, High glucose promotes retinal endothelial cellmigration through activation of Src, PI3K/Akt1/eNOS, and ERKs, Am. J.Physiol. Cell Physiol. 295 (2008) C1647–1657.

[10] M.B. Metzger, V.A. Hristova, A.M. Weissman, HECT and RING finger families ofE3 ubiquitin ligases at a glance, J. Cell Sci. 125 (2012) 531–537.

[11] X. Dong, J. Liu, H. Zheng, J.W. Glasford, W. Huang, Q.H. Chen, N.R. Harden, F. Li,A.M. Gerdes, X. Wang, In situ dynamically monitoring the proteolytic functionof the ubiquitin–proteasome system in cultured cardiac myocytes, Am. J.Physiol. Heart Circ. Physiol. 287 (2004) H1417–1425.

[12] A.M. Pickering, R.A. Linder, H. Zhang, H.J. Forman, K.J. Davies, Nrf2-dependentinduction of proteasome and Pa28alphabeta regulator are required foradaptation to oxidative stress, J. Biol. Chem. 287 (2012) 10021–10031.

[13] T. Murakami, T. Frey, C. Lin, D.A. Antonetti, Protein kinase cbetaphosphorylates occludin regulating tight junction trafficking in vascularendothelial growth factor-induced permeability in vivo, Diabetes 61 (2012)1573–1583.

[14] H. Liu, S. Yu, W. Xu, J. Xu, Enhancement of 26S proteasome functionalityconnects oxidative stress and vascular endothelial inflammatory response indiabetes mellitus, Arterioscler. Thromb. Vasc. Biol. (2012).

[15] R. Fujiki, W. Hashiba, H. Sekine, A. Yokoyama, T. Chikanishi, S. Ito, Y. Imai, J.Kim, H.H. He, K. Igarashi, J. Kanno, F. Ohtake, H. Kitagawa, R.G. Roeder, M.Brown, S. Kato, GlcNAcylation of histone H2B facilitates itsmonoubiquitination, Nature 480 (2011) 557–560.

[16] E.A. Scheef, C.M. Sorenson, N. Sheibani, Attenuation of proliferation andmigration of retinal pericytes in the absence of thrombospondin-1, Am. J.Physiol. Cell Physiol. 296 (2009) C724–734.

[17] X. Su, C.M. Sorenson, N. Sheibani, Isolation and characterization of murineretinal endothelial cells, Mol. Vis. 9 (2003) 171–178.

[18] D. Bech-Otschir, A. Helfrich, C. Enenkel, G. Consiglieri, M. Seeger, H.G.Holzhutter, B. Dahlmann, P.M. Kloetzel, Polyubiquitin substratesallosterically activate their own degradation by the 26S proteasome, Nat.Struct. Mol. Biol. 16 (2009) 219–225.

[19] A.J. Barber, D.A. Antonetti, T.S. Kern, C.E.N. Reiter, R.S. Soans, J.K. Krady, S.W.Levison, T.W. Gardner, S.K. Bronson, The Ins2Akita mouse as a model of earlyretinal complications in diabetes, Invest. Ophthalmol. Vis. Sci. 46 (2005) 2210–2218.

[20] J.R. Knowlton, S.C. Johnston, F.G. Whitby, C. Realini, Z. Zhang, M. Rechsteiner,C.P. Hill, Structure of the proteasome activator REGalpha (PA28alpha), Nature390 (1997) 639–643.

[21] P. Cascio, A.L. Goldberg, Preparation of hybrid (19S–20S-PA28) proteasomecomplexes and analysis of peptides generated during protein degradation,Methods Enzymol. 398 (2005) 336–352.

[22] R. Kinouchi, M. Takeda, L. Yang, U. Wilhelmsson, A. Lundkvist, M. Pekny, D.F.Chen, Robust neural integration from retinal transplants in mice deficient inGFAP and vimentin, Nat. Neurosci. 6 (2003) 863–868.

[23] M. Hellstrom, M.J. Ruitenberg, M.A. Pollett, E.M. Ehlert, J. Twisk, J. Verhaagen,A.R. Harvey, Cellular tropism and transduction properties of seven adeno-associated viral vector serotypes in adult retina after intravitreal injection,Gene Ther. 16 (2009) 521–532.

[24] A.S. Ibrahim, A.B. El-Remessy, S. Matragoon, W. Zhang, Y. Patel, S. Khan, M.M.Al-Gayyar, M.M. El-Shishtawy, G.I. Liou, Retinal microglial activation andinflammation induced by amadori-glycated albumin in a rat model of diabetes,Diabetes 60 (2011) 1122–1133.

[25] C.E. Alpers, R.A. Seifert, K.L. Hudkins, R.J. Johnson, D.F. Bowen-Pope, PDGF-receptor localizes to mesangial, parietal epithelial, and interstitial cells inhuman and primate kidneys, Kidney Int. 43 (1993) 286–294.

[26] P. Lindahl, M. Hellstrom, M. Kalen, L. Karlsson, M. Pekny, M. Pekna, P. Soriano,C. Betsholtz, Paracrine PDGF-B/PDGF-Rbeta signaling controls mesangial celldevelopment in kidney glomeruli, Development 125 (1998) 3313–3322.

[27] M.J. Emanuele, A.E. Elia, Q. Xu, C.R. Thoma, L. Izhar, Y. Leng, A. Guo, Y.N. Chen, J.Rush, P.W. Hsu, H.C. Yen, S.J. Elledge, Global identification of modular cullin-RING ligase substrates, Cell 147 (2011) 459–474.

[28] M.D. Petroski, R.J. Deshaies, Function and regulation of cullin-RING ubiquitinligases, Nat. Rev. Mol. Cell Biol. 6 (2005) 9–20.

[29] Y. Gao, L. Wells, F.I. Comer, G.J. Parker, G.W. Hart, Dynamic O-Glycosylation ofnuclear and cytosolic proteins: cloning and characterization of a neutral,cytosolic b-N-acetylglucosaminidase from human brain, J. Biol. Chem. 276(2001) 9838–9845.

[30] P. Geraldes, J. Hiraoka-Yamamoto, M. Matsumoto, A. Clermont, M. Leitges, A.Marette, L.P. Aiello, T.S. Kern, G.L. King, Activation of PKC-delta and SHP-1 byhyperglycemia causes vascular cell apoptosis and diabetic retinopathy, Nat.Med. 15 (2009) 1298–1306.

[31] T. Grune, K. Merker, G. Sandig, K.J. Davies, Selective degradation of oxidativelymodified protein substrates by the proteasome, Biochem. Biophys. Res.Commun. 305 (2003) 709–718.

[32] G.V. Shrikhande, S.T. Scali, C.G. da Silva, S.M. Damrauer, E. Csizmadia, P.Putheti, M. Matthey, R. Arjoon, R. Patel, J.J. Siracuse, E.R. Maccariello, N.D.Andersen, T. Monahan, C. Peterson, S. Essayagh, P. Studer, R.P. Guedes, O.Kocher, A. Usheva, A. Veves, E. Kaczmarek, C. Ferran, O-glycosylation regulatesubiquitination and degradation of the anti-inflammatory protein A20 toaccelerate atherosclerosis in diabetic ApoE-null mice, PLoS ONE 5 (2010)e14240.

[33] M.L. Caramori, P. Fioretto, M. Mauer, Low glomerular filtration rate innormoalbuminuric type 1 diabetic patients: an indicator of more advancedglomerular lesions, Diabetes 52 (2003) 1036–1040.

[34] Z.F. Luo, W. Qi, B. Feng, J. Mu, W. Zeng, Y.H. Guo, Q. Pang, Z.L. Ye, L. Liu, F.H.Yuan, Prevention of diabetic nephropathy in rats through enhanced renalantioxidative capacity by inhibition of the proteasome, Life Sci. 88 (2011) 512–520.