aberrant glycosylation in the human trabecular meshwork

130 Proteomics Clin. Appl. 2014, 8, 130–142DOI 10.1002/prca.201300031

RESEARCH ARTICLE

Aberrant glycosylation in the human trabecular

meshwork

Adam E. Sienkiewicz1, Brandon N. Rosenberg1, Genea Edwards1,2, Teresia A. Carreon1,2

and Sanjoy K. Bhattacharya1,2,3

1 Bascom Palmer Eye Institute, University of Miami, Miami, FL, USA2 Department of Biochemistry and Molecular Biology, University of Miami, Miami, FL, USA3 Neuroscience Program, University of Miami, Miami, FL, USA

Purpose: To determine the difference in protein glycosylation and glycosylation enzyme levelsbetween glaucomatous and control trabecular meshwork (TM).Experimental design: Glaucomatous and normal donor (n = 12 each) TM tissues, lectinfluorescence, fluorophore-assisted carbohydrate analyses, and quantitative MS were used todetermine the glycosylation levels. Primary TM cells and glycosylation inhibitors were used todetermine their effect on cell shape and motility.Results: In contrast to elevated levels of glycoproteins determined by lectin fluorescence,simultaneous hyper- and hypo-glycosylation in glaucomatous TM was revealed by fluorophore-assisted carbohydrate analyses. Analyses of enzymes showed elevation of beta-glycosidase 1 anddecrease in galactosyltransferase family 6 domain containing protein 1 in the glaucomatousTM. Quantitative MS identified select protein level changes between glaucomatous and normalTM. Primary TM cells were treated with inhibitors to elicit hypo-glycosylation, which affectedcell shape, motility, and fluorescent tracer transport across a layer of TM cells.Conclusions and clinical relevance: Global protein glycosylation is aberrant in glaucomatousTM compared to controls. The results presented here suggest that the alteration in global TMprotein glycosylation encompassing cellular and extracellular matrix proteins contributes toglaucoma pathology likely mediated through changes in properties of TM cells.

Keywords:

Carbohydrate electrophoresis / Glaucoma / Glycosylation / Trabecular meshwork

Received: April 24, 2013Revised: September 19, 2013

Accepted: October 14, 2013

� Additional supporting information may be found in the online version of this article atthe publisher’s web-site

Correspondence: Dr. Sanjoy K. Bhattacharya, Bascom Palmer EyeInstitute (McKnight Building) 1638 NW 10th Avenue, Suite 706A,University of Miami, Miami, FL 33136, USAE-mail: [email protected]: +1-305-326-6547

Abbreviations: DAB, 3,3′-diaminobenzidine; ECM, extracel-lular matrix; FACE, flourophore-assisted carbohydrate elec-trophoresis; GAG, glycosaminoglycan; GEO, Gene ExpressionOmnibus; IOP, intraocular pressure; LT, laser trabeculoplasty;MeQ, 4-(methylsulfanyl)-2-(quinolin-2-ylformamido)-butanoicacid; NDRI, National Disease Research Interchange; PhQ,N-(8-quinolinylsulfonyl)-phenylalanine; POAG, primary openangle glaucoma; RPE, retinal pigment epithelium; SEMG2,semenogelin II; SS, soyasaponin I; TM, trabecular meshwork

1 Introduction

The functional states of a multicellular organism are gov-erned by a constant dynamic interaction between expressionof intrinsic genetic information and environmental factors.Many of life’s secrets reside at the level of modulation of thecellular signals to bring functional changes in response to al-tered environments. Posttranslational modifications (PTMs)are a rapid and energy cost-effective means for cells to alter theinformation on protein structures and thereby the functionin response to changing environments [1]. Their importancein aging and disease states is now being increasingly real-ized [2, 3]. Addition of sugar/glycosyl moieties on proteins istermed as glycosylation. Target proteins can exist as hypo-and hyper-glycosylated at a given state of the cell compared to

Colour Online: See the article online to view Figs. 1–3 in colour.

C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.clinical.proteomics-journal.com

Proteomics Clin. Appl. 2014, 8, 130–142 131

another state (for instance, normal to pathological). Glycosy-lation on two different target proteins may potentially bringtwo functionally opposite responses, for example, activationand deactivation.

The glaucomas are a family of progressive irreversibleblinding optic neuropathies affecting over 60 million peo-ple worldwide [4]. Primary open angle glaucomas (POAGs),which usually account for a large majority, are late onset dis-eases. Angle closure glaucomas are also part of the cohort butunlike POAG, they are anatomically explicable and are betterintervened by opening the blocked angle by surgical means.Current medications or intervention strategies halt glaucomaprogression and delay the disease process (especially POAG)but do not offer a cure [4,5]. Glaucoma presents a large healthburden with regard to quality of life and productivity. Elevatedintraocular pressure (IOP) has been shown to be a major riskfactor for the development and progression of glaucomatousoptic nerve damage. Reduction in IOP (by pharmacologicalor surgical means) is the only proven neuro-protectant in de-laying glaucoma progression [5]. Elevated IOP occurs as aresult of reduced aqueous outflow and increased resistanceat the level of the trabecular meshwork (TM), a very smallregion in the anterior chamber angle between the iris andcornea that serves as a filter for aqueous humor outflow [6].Increased outflow resistance at the level of the TM presum-ably occurs due to changes in the cytoskeleton and the TMextracellular matrix (ECM). Due to passage of aqueous flowthrough this region, TM cells are expected to be responsive tomechanotransduction in response to changing shear stress.The surface proteoglycan layer (glycocalyx) acts as a sensorof fluid shear stress in endothelial and other types of cells,transmitting mechanosensation initiating responsive actionsin the actin network of the cytoskeleton [7, 8].

The changes in glycoproteins and proteoglycans in theTM tissue have been the subject of prior studies [9–12].These studies have encompassed analyses for expression ofmRNA [9,10] and proteins [13] but not the actual glycosylationchanges on proteins. Poor correlation between mRNA andprotein levels has been shown in previous investigations [14].Analyses of glycosylation or any other PTMs need to be per-formed independent of analyses of proteins and mRNAs.Analyses of TM proteins as well as their contributing en-zymes have yet to be done. This study aims to determine gly-cosylation changes between normal and glaucomatous TM.Furthermore, we aimed to capture the alteration of cellularproperties in response to glycosylation changes induced withenzyme inhibitors in vitro.

2 Materials and methods

2.1 Tissue procurement and histochemical

lectin-binding studies

Glaucomatous and normal tissues were procured from Na-tional Disease Research Interchange (NDRI), Philadelphia,

Pennsylvania and from Florida Lions Eye Bank, Miami,Florida, respectively. All eyes used for this study were enucle-ated within 24 h of death, from donors who were not subjectedto laser treatment (laser trabeculoplasty (LT); Supporting In-formation Table 1). Histochemical studies have utilized 17biotinylated lectins (EY Laboratories, San Mateo, CA) at aworking concentration of 9 �g/mL, 3,3′-diaminobenzidine(DAB), HRP and published protocols [15, 16], or Alexa-594conjugated streptavidin (Invitrogen, Carlsbad, CA) for theimmunofluorescent detection of bound lectins (SupportingInformation Table 2). Images were taken on a Leica TSP5 con-focal microscope (Leica Corporation, Mannheim, Germany)and quantified using Image J software (Rasband W.S, USNIH, Bethesda, MD). All experiments and procedures werereviewed and approved by the University of Miami’s institu-tional review boards and are in accordance with the declara-tion of Helsinki. All histochemical data presented here arerepresentative of several analyses (n = 12 each glaucoma andcontrols) performed during this study.

2.2 Fluorophore-assisted carbohydrate

electrophoresis (FACE)

FACE analyses of protein extracts prepared from 20 mg ofnormal and glaucomatous TM were performed followingpublished protocols [17, 18]. Briefly, proteins were digestedwith 10 �L of bovine testicular hyaluronidase (0.5 U) andchondroitinase ABC (10 U), divided into two equal aliquots(control and mercuric acetate treatment), and electrophoresedwith standards in a Dermato-or-Hyaluro-Disaccharide Kit(Seikagaku Corporation, Japan). All samples were deriva-tized using 2-aminoacridone prior to electrophoresis on amonosaccharide polyacrylamide gel (Glyko/Prozyme, San Le-andro, CA). Following electrophoresis, the gel was illumi-nated under UV light (365 nm) and imaged using ChemiDocXRS+ System (Bio-Rad Labs).

2.3 Quantitative proteomics

We utilized 12 glaucomatous and 12 normal samples for8-plex ITRAQ labeling using a kit (catalog number4390811, Applied Biosystems, Carlsbad, CA) containing113–119 and 121 isotopic tags following procedures rou-tinely used in our laboratory [19]. The samples were an-alyzed on an ABI 4800 MALDI TOF-TOF machine (Ap-plied Biosystems). Quantitative proteomics were separatelyperformed for both the TM protein extracts as well asfor lectin immunoprecipitation products. Quantificationof lectin immunoprecipitation products was done as fol-lows: briefly, protein extracts were prepared from glau-comatous and normal donor TM using 100 �L of thebuffer 125 mM Tris-HCl (pH 7.0), 100 mM NaCl, 0.1%Triton X-100, and 0.1% Genapol C-100 per 25 mg of tissue.The homogenized suspension was centrifuged at 10 000 rpm


132 A. E. Sienkiewicz et al. Proteomics Clin. Appl. 2014, 8, 130–142

and precipitated using 10 �L of a 9 �g/�L workingconcentration of a biotinylated lectin, (e.g. Caraganaarborescens lectins; Supporting Information Table 5). The pre-cipitate was recovered with 25 �L of 4 �g/�L streptavidin-coupled magnetic beads (catalog number S1420S, New Eng-land Biolabs, Ipswich, MA). The eluent was subjected toacetone precipitation using a 4:1 ratio of acetone to eluent(v/v) following standard procedures. The IP products weresubjected to 8-plex iTRAQ labeling [19].

2.4 Mass spectrometric analyses

To identify the proteins and perform quantification, pro-tein bands separated on 10% SDS-PAGE were excised, thegel slices were destained with 50% ACN/water and weresuspended in 0.5 M triethylammoniumbicarbonate (number17902, Sigma, St. Louis, MO) pH 8.5 and reduced with 10 mMTris-(2-carboxyethyl) phosphine (Sigma). The proteins weresubsequently alkylated in the dark using 55 mM solution ofiodoacetamide (catalog number RPN6302V; GE Healthcare,Buckinghamshire, UK) and in-gel digested with sequencinggrade-modified trypsin (catalog number V5113; Promega,Madison, WI; 0.1 �g/15 �L in 15 mM N-ethylmorpholin)overnight at 37�C. The peptides were extracted twice with50 �L 0.1% TFA/60% ACN and finally with 30 �L of ACNand dried in a SpeedVac. The extracted peptides were in-cubated with 8-plex iTRAQ reagents in 0.5 M triethylam-moniumbicarbonate containing 60% v/v isopropanol. Pep-tides derived from protein bands of glaucomatous TM tissuewere subjected to incubation with reagents 113, 114, 115,and 116 (4381557, 4381557, 4381557, 4381560; ABI, FosterCity, CA) whereas peptides derived from control TM proteinbands were incubated with reagents 117, 118, 119, and 121(4381561; 4381562; 4381563; 4381564). The incubation mix-tures were dried in a SpeedVac, mixed together, and loadedonto a slurry of 500 �L of cation exchange buffer in 12 mMammonium formate in 25% ACN at pH 2.5–3.0, and sub-jected to chromatographic separation.

2.5 Two dimensional-liquid chromatography (2D-LC)

separations

Strong cation exchange (SCX) separations of peptides wereperformed on a passivated Waters 600E HPLC system, usinga 4.6 × 250 mm polysulfoethyl aspartamide column (PolyLC,Columbia, MD) at a flow rate of 1 mL/min. Buffer A contained10 mM ammonium formate, pH 2.7, in 20% ACN/80% wa-ter. Buffer B contained 666 mM ammonium formate, pH 2.7,in 20% ACN/80% water. The gradient was buffer A at 100%(0–22 min following sample injection), 0–40% buffer B (16–48 min), 40–100% buffer B (48–49 min), then isocratic 100%buffer B (49–56 min), then at 56 min switched back to 100%buffer A to re-equilibrate for the next injection. The first26 mL of eluant (containing all flow-through fractions) was

combined into one fraction, then 14 additional 2 mL frac-tions were collected. All 15 of these SCX fractions were drieddown completely to reduce volume and to remove the volatileammonium formate salts, then resuspended in 9 �L of 2%v/v ACN, 0.1% v/v TFA, and filtered prior to reverse phaseC18 nano-flow-LC separation. For second dimension sep-aration by reverse phase nano-flow LC, each SCX fractionwas autoinjected onto a Chromolith CapRod column (150 ×0.1 mm, Merck) using a 5 �L injector loop on a Tempo LCMALDI spotting system (ABI-MDS/Sciex). Buffer C was 2%ACN, 0.1% TFA, and buffer D was 98% ACN, 0.1% TFA. Theelution gradient was 95% buffer C/5% buffer D (2 �L/minflow rate from 0 to 3 min, then 2.5 �L/min from 3 to8.1 min), 5% buffer D to 38% buffer D (8.1–40 min), 38%buffer D to 80% buffer D (41–44 min), 80% buffer D to5% buffer D (44–49 min; initial conditions). Flow rate was2.5 �L/min during the gradient, and an equal flow of MALDImatrix solution was added postcolumn (7 mg/mL recrystal-lized CHCA, 2 mg/mL ammonium phosphate, 0.1% TFA,80% ACN). The combined eluant was automatically spottedonto a stainless steel MALDI target plate every 6 s (0.6 �L perspot), for a total of 370 spots per original SCX fraction.

After drying the sample spot mentioned above, 13 cali-brant spots (ABI 4700 Mix) are added to each plate manually.MALDI target plates (15 per experiment) were analyzed in adata-dependent manner on an ABI 4800 MALDI TOF-TOF.As each plate is entered into the instrument, a plate calibra-tion/MS default calibration update is performed, and thenthe MS/MS default calibration is updated. MS spectra weretaken from 5500 MALDI spots, averaging 500 laser shotsper spot at Laser Power 3100. A total of 3249 MS/MS spectrawas taken with up to 2500 laser shots per spectrum at LaserPower 3600, with CID gas air at 1.2–1.3 × 10−6 Torr. Afterthe MS and MS/MS spectra from all 15 plates in a sample sethave been acquired, protein identification and quantitationwas performed using the Paragon algorithm as implementedin ProteinPilot 3.0 software (from ABI/MDS-Sciex) andMatrix Sciences MASCOT algorithm version 2.1. Protein-Pilot software was utilized for searches with followingsearch parameters: Cys Alkylation—Iodoacetamide; IDFocus—Biological Modifications; Search Effort—Thorough.For trypsin digestion 0 missed cleavages were specified. Thevariable PTMs, which were searched, were deamidation, ox-idation, carbamidomethylation, and acetylation. The January1, 2010 Human NCBI database sequences contain 512 785protein sequences, plus 156 common lab contaminants. Forestimation of “false discovery rate” (FDR), a simultaneoussearch was performed on a concatenated Decoy database,which is the exact reverse of each protein sequence in thedatabase plus 156 common human and lab contaminants.Total protein sequences searched in Database plus contam-inants plus concatenated reverse Decoy database: 1 025 652.The protein identifications at 95% confidence level wereretained. The preset “Thorough” (iTRAQ or Identification)search settings were used where identifications musthave a ProteinPilot unused score >1.3 (>95% confidence



interval). In addition, the only protein IDs accepted musthave a local “FDR” estimation of no higher than 5%, as calcu-lated from the slope of the accumulated Decoy database hitsby the Proteomics System Performance Evaluation Pipelineprogram [20]. This local FDR estimate is much more strin-gent than p < 0.05 or 95% confidence scores in MASCOT,SEQUEST, ProteinPilot, or the aggregate FDR estimations(2 × number of Decoy database IDs/total IDs at any chosenthreshold score) commonly used in the literature, andcombined with the ProGroup algorithm included inProteinPilot gives a very conservative as well as fully MIAPE-compliant list of proteins identified (i.e. MASCOT andother lists of “proteins ID’d at p < 0.05” will produce morenumerous “significant” IDs from the same data, butthose larger lists are highly likely to contain many morefalse-positive IDs).

2.6 Bioinformatics, analyses of enzymes, and

glycosylation

A list of deglycosylases and glycosyltransferases was compiledafter extensive search using literature, BRENDA, FRENDA,AMENDA, and Ensembl databases. Next, a Boolean searchwas carried out using keywords “trabecular meshwork” and“eye” in Gene Expression Omnibus (GEO) and in UniGenedatabases (National Center for Biotechnology Information,Bethesda, MD). The presence of mRNA for specific degly-cosylases and glycosyltransferases (Supporting InformationTable 3) identified by microarray analyses in the TM was ver-ified using appropriate primer pairs along with PCR analysisusing standard protocols. For enzymes whose mRNA wasconfirmed to be present in the TM (Supporting InformationTable 3), further ELISA and dot blot analyses were performed.Commercially available primary antibodies for GLT8D1,GLT8D2, GLT6D1, and UGT8 (catalog numbers ab83758,ab105655, ab80904, ab84288, Abcam Inc., Cambridge, MA),and GLA, NAGA, GLB1 (�-galactosidase1), FUS2, GALNS,and FUCA1 (catalog numbers ab70520, ab102668, ab616,ab93660, ab76400, ab69957, Abcam Inc.) were used for ELISAand dot blot analyses following published procedures [19].For dot blot analyses, proteins were spotted as 0.5 mm diam-eter spots. No protein extracts or saturated amount of proteinextracts (>50 �g on 0.5 mm spot using vacuum concentrated(10 mg/mL) protein extract) were used to set 0 and 100%level, respectively, for dot blot. The protein extract (1 �g) of asingle TM extract was used to determine glycerol 3-phosphatedehydrogenase (GAPDH) and enzymatic immunoreactivityfor each individual enzyme by dot blot. The densitometricdetermination of individual enzymatic immunoreactivity foreach enzyme was normalized to GAPDH immunoreactivityand the mean result for 12 glaucomatous and 12 normalsamples was determined. This normalized immunoreactivityis referred to as relative immunoreactivity. A similar analysiswas carried out by ELISA for all 12 glaucomatous and normalsamples determining immunoreactivity of each individual

enzyme. Normalizing to GAPDH immunoreactivity (referredto as relative immunoreactivity) for 1 �g protein extract foreach TM sample, mean values for each group were alsodetermined. For ELISA, no protein and 100 �g of GAPDH orsaturating level of protein extract (∼0.9 absorbance) was usedto set 0 and 100% level. The dot blot densitometric quantifi-cation was performed using Image J software normalizedusing GAPDH immunoreactivity. The total TM proteins(10 �g) subjected to partial digestion for 0.5 and 1 h using20 units of PNGase F (catalog number P0704S, New EnglandBioLabs) were subjected to Western Blot analysis (digestsand controls) using antibodies against specific proteins.Total glycosylated proteins were also detected using GelcodeGlycoprotein kit (catalog number 24562; Pierce Biotechnol-ogy, Rockford, IL) and also a Pro-Q Emerald 300 glycoproteinkit (catalog number P21857; Invitrogen) following previouslydescribed procedures [21].

2.7 Glycosyltransferase/de-glycosylase activity

assay methods

In the first protocol, the wells of a 96-well plate were coatedwith 10 �L (0.5 �g) lactotriaosylceramide (Lc3Cer) methanolsolution and 50 �L of 0.08% polyisobutylmethacrylate in asolvent mixture of chloroform/hexane/chloroform (1:5.25:25v/v/v). Solvent was evaporated using a hair dryer and50 �L of 1% BSA in 1× PBS was added for blocking for 2 h.Wells were washed with 1× PBS and pH was adjusted with25 �L of 50 mM cacodylate buffer (pH 6.8 if UDP-galactose;8.5 when UDP-glucose is used as acceptor) containing 0.005%Triton CF-54, 0.2 M KCl, 20 mM MnCl2, and 1.56 mMUDP-galactose and 10 �g of protein test samples or bovinemilk galactosyltransferase (25 �L; 250/�U) as control andincubated at 37�C for 2 h. Wells were washed with 1×PBS and incubated with 50 �L carbohydrate sequencespecific monoclonal antibody 1 �g/mL; in PBS for 2 hat 25�C. After washing with 1× PBS, 50 �L of alkalinephosphatase conjugated secondary antibody and 100 �Lof phosphatase substrate was added then incubated for30 min followed by several washes with 1× PBS and ab-sorbance measurement at 405 nm. A modified protocol usedbovine milk galactosyltransferase, Clostridium perfringenssialidase with NeuAc�2-3nLc4ceramide (1 �g) as substrate,and monoclonal antibody H-11 for recognizing Gal�1–4GlcNAc� as product. In the second protocol that measuredglucosaminidase activities, test TM protein solutions or pu-rified N-acetyl-�-glucosaminidase enzyme was incubated at37�C for 30 min with 0.264 mM methyl-umbelliferyl N-acetyl-�-glucosaminide (prepared from 4-methylumbelliferoneand acetochloroglucosamine) in 0.05 M sodium citratebuffer pH 5.0. The enzymatic reaction was stopped andmethylumbelliferone was converted into the anionic formby addition of 0.2 M sodium glycinate buffer pH 10.65(3 mL) at the end of incubation. The activity was estimatedby measuring absorbance at 365 nm.



2.8 TM cell culture and inhibitor and tracer transport

studies

Primary human TM cells were cultured from cadavericcorneo-scleral sections using protocols routinely used in ourlaboratory [13]. Time-lapse microscopy was performed usingcells grown on glass bottom chamber slides (catalog num-ber 154534, Lab-Tek, Thermo Fisher Scientific, Rochester,NY). Preincubated cells in serum-free media (DMEM, Medi-atech) were further incubated for 4 h with the three inhibitors:soyasaponin I (SS): 150 �M, N-(8-quinolinylsulfonyl)-phenylalanine (PhQ): 75 �M, 4-(methylsulfanyl)-2-(quinolin-2-ylformamido)-butanoic acid (MeQ): 50 �M (1–3× ofthe IC50 for sulfotransferase or sialyltransferase) priorto time-lapse microscopy using the Zeiss AxioVert 200M microscope with cell culture setup (Carl Zeiss) insidea temperature controlled chamber at 37�C in 5% CO2

atmosphere.To achieve inhibition of glycosyltransferases and degly-

cosylases in vitro, we employed synthetic inhibitors PhQ(catalog number OSSL_288713, Princeton BioMolecular Re-search, Monmouth Junction, NJ) and MeQ (catalog numberEN300–59129, Ryan Scientific, Isle of Palms, SC; specificfor GlcNAc-6-O-sulfotransferase family) and a natural com-pound SS ((3�,4�,22�)-22,23-dihydroxyolean-12-en-3-yl O-6-deoxy-�-L-mannopyranosyl-(1–2)-O-�-D-galactopyranosyl-(1–2)-�-D-glucopyranosiduronic acid; catalog number ABS-00019336–005, Chromadex, Irvine, CA)—an inhibitor of ST3beta-galactoside-alpha-2,3-sialyltransferase 1.

An 8 �m pore size 1 cm diameter polycarbonate mem-brane well insert (catalog number 140644, NalgeNunc In-ternational, Rochester, NY) was used to measure fluores-cein sodium salt (catalog number 46960, Sigma-Aldrich) flowthrough a multilayer (∼5) matrix of primary TM cells colo-nized specifically for this experiment by repeating cellularsuspension depositions daily. Briefly, the first layer of con-fluent cells (∼0.5 × 106 initial cell load) was made on themembrane using routine culture techniques and incubatedfor 16–24 h. The second cell suspension was (∼1 × 106 ini-tial cell load) carefully distributed on the top of the first layerand allowed to assemble as a cohesive bilayer during an addi-tional 16–24 h incubation time. The latter step was repeatedfour times. A second set of experiments was performed usinga layer of collagen matrix (rat tail collagen; catalog number354249; BD Biosciences, San Jose, CA) formed on the mem-brane to facilitate cell adherence prior to the introduction ofthe first cell monolayer. The cells were similarly allowed toform a confluent monolayer over a period of 16–24 h. Thisprocess was repeated to ultimately achieve a confluent penta-layer of cells. Cells treated with inhibitors were preincubatedin serum-free media (Mediatech) and inhibitors for 4 h as de-scribed above. Following inhibitor treatment, the layered cellmatrix was washed twice with warm 1× PBS (Mediatech).A 10 �L volume of sodium fluorescein dye (1:100 dilutionof 1 mg/mL) was introduced into the inner well of the ma-

trix bearing membrane insert suspended in a 500 �L volumeof 1× (PBS). The sampling was initiated and continued at2 min intervals by aspirating from the outer well of the in-sert; the fluorescence of the samples was measured at 485 and525 nm excitation and emission using a spectroflourometer(Lumistar, BMG Labtech, Offenburg, Germany). A Trypanblue exclusion assay of primary TM cells was performed fordetermination of cell viability following previously publishedprocedures [22].

2.9 Data and analyses

ELISA, dot blot, and tracer transport results are expressedas mean ± SD. ELISA, dot blot, and tracer transport resultswere compared between control and glaucoma donors by thetwo-tailed two-sample t-test. Ratios of glaucoma to normalproteins and glycosylation levels were compared to 1.0 by thetwo-tailed paired or unpaired t-test.

3 Results

3.1 Simultaneous hypo- and hyper-protein

glycosylation in glaucomatous TM

Fluorescent histochemical detection of glaucomatous andnormal human TM specimens (Supporting InformationTable 1) with biotin-bound lectins suggested presence of ele-vated glycoproteins or proteoglycans in the glaucomatous TM(Fig. 1A and Supporting Information Table 2). These resultsare consistent with a previous study employing DAB [15] todetect lectin binding in the human TM. Compared to lectin-fluorescence analyses (Fig. 1A), DAB lectin analyses are rela-tively subjective (Supporting Information Fig. 1A). The 7 outof 17 lectins used in our analyses were common with pre-vious studies and showed general agreement with previousfindings with only a few differences (Supporting Informa-tion Table 2). We next analyzed glycosylation of human TMtissue proteins using enzymatic digestion of protein-boundsugars and FACE [17,18] that revealed (Fig. 1B) simultaneouspresence of hypo- and hyper-protein glycosylation in glauco-matous TM compared to the controls. This is in contrastto the lectin-fluorescence analyses (Fig. 1A). The observeddifference between lectin-bound fluorescence (Fig. 1A) andFACE (Fig. 1B) could be reconciled in the following way,that hyper-glycosylated moieties in the glaucomatous TM areprobably more exposed epitopes compared to that in the nor-mal controls (Fig. 1B), while the hyper-glycosylated moietiesin the controls, as revealed by FACE (Fig. 1B), are eitherburied or not recognized by lectins (Fig. 1A). The FACE anal-yses (Fig. 1B) showed simultaneous occurrence of both hypo-and hyper-glycosylation in the glaucomatous TM. To ensurethat equal amount of protein extract was utilized for these



Figure 1. Glycosylation levels and protein glycosylation enzymatic activities in control and glaucomatous trabecular meshwork (TM).(A) Representative microscopic image of histochemical analyses (n = 12 each glaucoma and control) of bound fluorescent lectins in theTM. The cadaver TM from glaucomatous or control donors as indicated was probed using biotinylated lectins as indicated. The detectionwas based on Alexa R©-594 coupled streptavidin binding to biotinylated lectin. Images were taken on Leica TSP5 confocal microscope(Leica Corporation) at 20× magnification. A tissue section that was not probed with any lectin served as negative control. (B) Represen-tative fluorophore-assisted carbohydrate electrophoresis (FACE R©) analyses of 2-aminoacridone derivatized testicular hyaluronidase andchondroitinase ABC-digested products from normal and glaucomatous TM as indicated. The disaccharide standards were obtained fromHyal-Dermato Disaccharide D-kit (catalog number 400572–1; Seikagaku Corporation). To indicate equal loading, an identical aliquot wasloaded on to a 10% SDS-PAGE and stained with Coomassie blue for protein staining with molecular markers as indicated presented in thebottom panel. (C) Relative quantification of glycosyltransferase immunoreactivities using ELISA and dot blot densitometric analyses asindicated. (D) Relative quantification of deglycosylase immunoreactivities using ELISA and dot blot densitometric analyses as indicated.The enzymatic immunoreactivity ratio (normalized to total protein/total of GAPDH) of indicated enzymes (C, D) with GAPDH was quantifiedusing ELISA and dot blot analysis as indicated and as described in experimental procedures (n = 12 samples each for glaucoma andcontrol). Results are mean ± SD (*significantly different results when compared between control and glaucoma donors by the two-tailedtwo-sample t-test: *p < 0.05). Control and glaucomatous samples are represented by red squares and blue circles for dot blots and ELISAanalyses as indicated.

analyses between control and glaucoma groups, an identicalaliquot of same digest was fractionated on SDS-PAGE andstained with Coomassie blue (Fig. 1B bottom panel). We nextinvestigated whether this is caused by enzymatic differences(glycosyltransferase and de-glycosylases) between normal andglaucomatous TM tissues.

3.2 Enzymatic activities and aberrant glycosylation

in the glaucomatous TM

We prepared a curated short list of all glycosyl metabolizingactivities known to be present in the human TM at themRNA level and a comprehensive search was performed



for glycosyltransferases in the literature, GEO, and Uni-gene databases (Supporting Information Table 3a). Forde-glycosylases, additional searches were performed inBRENDA, Ensembl, and GEO databases (Supporting Infor-mation Table 3b). We further verified their local productionin the TM using appropriate primer pairs and PCR analysis.We next investigated protein expression for confirmedmRNA presence (data not shown; Supporting InformationTable 3a and b) using commercially available antibodies.Measurements of enzymatic activity using available relativenonspecific assays showed agreement and correlation withimmunoreactivity. The immunoreactivity of glycosyl metab-olizing enzymes determined that ELISA analyses correlatedwith that of the dot blot analyses (Supporting InformationFig. 1B) and showed differences between glaucomatous andnormal TM (Fig. 1C and D). We divided these activities in twocategories: glycosyltransferases (the enzymes that transfersugars onto proteins: GLT8D1, GLT8D2, GLT6D1, DDOST,and UGT8; Fig. 1C; Supporting Information Table 3a) andprotein de-glycosylases (the enzymes that remove sugarsfrom modified proteins: GLA, NAGA, GLB1, GALNS, FUS2,and FUCA1; Fig. 1D; Supporting Information Table 3b), re-spectively. We found a small decrease in glycosyltransferaseGLT6D1 and UGT8 levels. UGT8 or UDP-galactose-ceramidegalactosyltransferase is known to catalyze the transfer ofgalactose to ceramide. While this may be relevant for biosyn-thesis of galactocerebrosides or sphingolipids of myelinmembranes, it is irrelevant for the present investigation.GLT6D1 or galactosyltransferase family 6 domain-containing1 is the protein product of an uncharacterized gene andhas been identified as a susceptibility locus for periodon-titis [23]. Significant increased levels of de-glycosylaseGLB1 or beta-galactosidase was found in glaucomatouscompared to normal TM (Fig. 1D). GLB1 cleaves beta-linkedterminal galactosyl residues from glycoproteins and gly-cosaminoglycans (GAGs) [24]. Having determined the levelof glycosyltransferases and de-glycosylases, our next goal wasto determine which proteins are differentially glycosylatedbetween glaucoma and control. However, the determinationof differential glycosylation posed another problem, thatprotein levels themselves could also undergo a differentialchange between glaucoma and control tissues. We thereforeattempted to identify changes in protein levels betweenglaucomatous and normal TM followed by further analyses ofa select subset of proteins for determining their glycosylationstates between control and glaucomatous tissue. Quantitativeanalyses of changes in protein levels enabled us to accountfor changes in glycosylation of select protein in proportion totheir level changes between control and glaucomatous tissue.

3.3 Alteration of glycosylation is independent of

protein level changes

We next subjected 8-plex iTRAQ-labeled quantitative analy-ses of total TM proteins from glaucomatous (tags 113–116)

and normal donors (tags 117–121) to identify the changesin protein levels (Supporting Information Table 4) betweenglaucomatous and control donors. We also attempted 8-plexiTRAQ-labeled quantitative proteomic analyses of precipi-tated lectin-bound proteins from glaucomatous (tags 113–116) and normal TM (tags 117–121) extracts (Supporting In-formation Table 5). The latter attempt was made to captureenriched glycosylated proteins. In both experiments, we iden-tified several known glycosylated proteins and proteoglycans.We compared these quantitative proteins with GEO datasets(GDS359 and GDS 1640). We further compared the resultswith published reports on TM glycoproteins and proteogly-cans [9,10,13,25,26]. We selected a few protein candidates todetermine for their potential hypo-glycosylation in glaucoma-tous tissue (hyper-glycosylated in normal) in comparison tothe controls. The proteins that undergo hyper-glycosylationare also an important component but in the present stud-ies we focused on identification of proteins that are hypo-glycosylated in glaucomatous TM with the goal of mimick-ing hypo-glycosylation via inhibitors in primary TM cells tofurther investigate the effects. We selected semenogelin II(SEMG2), Keratocan and Biglycan, which showed less thantwofold changes in protein levels for further analyses of to-tal glycosylation per protein. Our objective was to find pro-teins whose level either do not change at all (ratio of 1.0with iTRAQ labeling) or change less than twofold (ratio of2.0) for investigation of level changes in their glycosylation(Supporting Information Table 4). Our bioinformatic anal-yses (PSORT, String, and other databases) revealed that anumber of glycoproteins, including Keratocan and Biglycan,have the potential to be secreted into the ECM and have thepotential for multiple protein–protein interactions (Support-ing Information Table 6). We used partial enzymatic PNGaseF digestions for removal of N-linked glycosylation followingpreviously published procedures [27], electrophoretic sepa-ration, and Western blot analyses (Fig. 2) to determine theprotein glycosylation levels. Briefly, the undigested and PN-Gase F digested collapsed protein bands (Fig. 2A, arrows)serve as estimates of total protein and glycosylated fractionsof the protein. The ratio of total protein and total protein nor-malized glycosylated fractions in normal and glaucomatousTM, with or without comparison to total protein load or ahousekeeping protein (GAPDH or beta-actin), provides therelative glycosylation level for the specific protein. The West-ern blot analyses of Keratocan without PNGase F (Fig. 2A)served as total proteins before digestion. After the PNGase Fdigestion, the total amount of lower digested band (indicatedby arrow head) was estimated using densitometry and usedfor determination of the ratio of digested band compared tototal, providing an estimate of relative protein glycosylation.Keratocan in normal TM tissue was more glycosylated com-pared to glaucomatous TM (Fig. 2A). Biglycan in the TMtissue (Fig. 2B), as well as in the primary TM cells with andwithout inhibitor treatment (Fig. 2C), was subjected to similaranalyses. A series of partial PNGase F digestion enables con-firmation of the estimates. We used microtubule-associated



Figure 2. Glycosylation differences of select target proteins in glaucomatous TM compared to controls (n = 12 each) and in responseto glycosylation inhibitors. (A) Representative Western blot analyses of Keratocan in control and glaucomatous TM protein extracts(10 �g) with or without PNGase F treatment as indicated. The ratio of intact to collapsed protein band on digestion provided the ratio ofglycosylation for normal or glaucoma. The ratio of intact band between glaucoma to normal provided the level of protein (normalizedto total protein, glycerol 3-phosphate dehydrogenase (GAPDH) or beta-actin). (B) Representative Western blot analyses of Biglycan incontrol and glaucomatous TM protein extracts with PNGase F treatment. (C) Representative Western blot analyses of Biglycan in controland inhibitor-treated (SS: 150 �M, PhQ: 75 �M, MeQ: 50 �M) cell extracts subjected to PNGase F treatment. GAPDH immunoreactivity asindicator of loading control for panels A through C has been presented as indicated. (D) Densitometric estimation of proteins (hollow bars)and glycosylation (hashed bars) expressed as ratio for each specific protein as indicated. The ratios for proteins and glycosylation areimmunoreactivity in glaucoma/normal, except for Biglycan†, which indicates ratio of inhibitor treated/untreated control. The glycosylationlevel in normal or glaucoma was the normalized ratio of glycosylation estimated by partial digestion of PNGase treatment compared toundigested control band (as shown for Keratocan in A). The overall ratio of glycosylation estimated in glaucoma to control (for Biglycan†,inhibitor treated to control) using this procedure has been depicted with hashed bars. The results are mean ± SD of three independentexperiments. The mean ratio was subjected to a two-tailed paired t-test; *p ≤ 0.05.

protein 2 and SEMG2, whose protein levels have been foundto not undergo any change (ratio of 1.0 between glaucomaand normal; Supporting Information Table 4), as controlsand analyzed the ratio of proteins between glaucomatous andnormal TM. In all these analyses, we have used total proteinload as well as GAPDH and beta-actin as protein loading ornormalizing controls. While the microtubule-associated pro-tein 2 and SEMG2 were found to be unchanged in Westernblot analyses, the Keratocan and Biglycan protein levels werefound to have a ratio of 1.9 and 1.3, respectively, between glau-comatous and normal TM (Fig. 2D; Supporting InformationTable 4). Their glycosylation levels, however, clearly demon-strate lower levels in glaucomatous compared to normal TM.We used in vitro experiments for confirmation of detection ofglycosylation using partial PNGase F treatment. The Biglycanprotein and glycosylation levels were estimated between con-trol untreated and glycosylation inhibitor treated (describedin greater detail in the next section) primary TM cells for

comparison of protein and glycosylation levels in vitro. Thecells showed a higher level of protein in the inhibitor-treatedcells (ratio of 1.4) while the glycosylation level (Fig. 2C)showed a clear decrease in levels in inhibitor-treated cells(Fig. 2D). Performed simultaneously with analyses for tis-sues, these results confirm the consistency of the analysismethod. Previously optimized [21] staining using GelcodeGlycoprotein and Pro-Q Emerald 300 glycoprotein kits esti-mated Keratocan and Biglycan glycosylation levels to be con-sistent with Western blot findings (Fig. 2D).

3.4 Enzymatic inhibitors mediated global TM

glycosylation reduction alters primary TM cell

properties and tracer transport across cell layers

We next attempted to determine whether inhibition ofglycosylation results in changes in TM cells properties.



Figure 3. Effects of inhibitor-mediated re-duction in glycosylation in vitro (repre-sentative data). (A) Representative micro-scopic images depicting cell morphologyin response to inhibitor-mediated reduc-tion in protein glycosylation in comparisonto controls as indicated in selected hourlytime intervals. (B) The transport of fluo-rophore across multilayered ensemble ofprimary TM cells with or without glycosy-lation inhibitor treatment on a porous 8 �mmembrane, without any collagen embed-ding (solid lines) and with collagen matrixembedding at each cell layer (dashed lines).(C) The cell survival for primary TM cellswas determined for control (filled bars) andinhibitor-treated groups at indicated timeinterval (hollow bars) in hours using Trypanblue exclusion assay. The average ±SD ofpercent of cell survival for indicated groupshas been shown.

Primary TM cells were treated with inhibitors of sulfo-transferases, glycosyltransferases, and sialyltransferases tocause global glycosylation changes. SS, an inhibitor of ST3beta-galactoside-alpha-2,3-sialyltransferase 1 and two syn-thetic inhibitors, PhQ and MeQ, for members of GlcNAc-6-O-sulfotransferase family which are specifically known toinhibit N-acetylglucosamine-6-O-sulfotransferase-2 and N-acetylglucosamine-6-O-sulfotransferase-3, were used for re-duction of glycosylation. PNGase F analyses (Fig. 2) andfurther FACE analyses (data not shown) confirmed reducedglobal TM protein glycosylation as a consequence of treat-ment of cells with these inhibitors. Treatment of primary TMcells with these inhibitors (concentrations of SS: 150 �M,PhQ: 75 �M, MeQ: 50 �M) showed distinct and significantalteration in the shape of the TM cells, that is, a spindle-like elongation of the cell body (Fig. 3A), reduced motil-ity, and increased division time (Fig. 3A, Supporting Infor-mation Videos 1 and 2). These changes were observed at

∼4 h after addition of inhibitors in the cell culture. TM cellscultured to a five-layer cell matrix grown on a porous polycar-bonate membrane subjected to the three inhibitor treatmentshowed an increased level of fluorescein transport comparedto untreated layer setup (Fig. 3B, solid lines). The extendedincubation (∼3 days) of inhibitor-treated cells without colla-gen matrix also showed increased flow of tracer comparedto controls as noted for cells without extended incubation(Fig. 3B, solid lines). This is consistent with elongated cellsrendering larger areas for fluid transport. In a comparisonafter extended incubation (∼3 days), an identical setup of TMcells, cultured to a five-layer cell matrix grown on the sameporous polycarbonate membrane but coated with cross-linkedcollagen at each layer level, displayed a decreased tracer trans-port following the combined inhibitor treatment as comparedto untreated cells (Fig. 3B, dashed lines). This is consistentwith compaction of layers and loss of flexibility in cellularmodeling upon inhibitor treatment enabling tortuous flow



across cell layers thus increasing residence time. PrimaryTM cells suspended in media-enriched polymerized collagengel inside glass microcapillaries [28] displayed a lack of gel ex-pansion/contraction in the combined inhibitor-treated groupand was comparable to untreated cell containing gels, consis-tent with the lack of flexibility for the inhibitor-treated cells.We did not find any significant cell death due to prolongedinhibitor treatment (Fig. 3C). There was not a statisticallysignificant difference in cell survival between control andinhibitor-treated groups at different time intervals (Fig. 3C).

4 Discussion

How PTMs, and in particular, global TM glycosylation affectthe aqueous outflow dynamics is not fully understood. Alter-ations in global TM glycosylation levels were found in POAGas well as juvenile glaucoma [29, 30]. Unlike angle closureglaucoma, the iridocorneal angle remains open in the POAGbut yet the aqueous outflow is decreased which is thought tobe due to increased resistance at the TM. Surface glycopro-teins are important determinants of cellular structural andfunctional properties [15]. Aberrant functioning of glycan-processing enzymes (glycosidases and glycosyltransferases)involved in the biosynthesis and transfer of the carbohydratemoieties results in aberrant glycosylation [31]. Aberrant gly-cosylation alters properties of cells such as adhesion, inva-sion, and capacity to withstand different regimes of fluidshear [7, 32].

Glycocalyx, or the surface protein glycosylation, has beenshown to act as a sensor of fluid shear stress in a variety ofcells that include endothelial [7], osteoblast [33], and smoothmuscle [34] as well as several other cell types [35]. Fluid flowand migration of cells in biological systems is often associ-ated with engagement of specialized adhesion molecules [35].The floating endothelial cells [7] are known to undergo glyco-calyx changes in response to different regimes of changingfluid shear. However, very little is known about the glycoca-lyx of cells from neural crest origin such as TM cells [36].The TM cells function in an environment that is continu-ously subjected to varying levels of fluid shear stress or shear-induced mechanical forces [37]. Actin cytoskeletal remodel-ing and changes in cell shape and mobility occur in responseto mechanotransduction due to a changing regime of fluidshear [7, 8]. Cytoskeletal changes are often commensuratewith dynamic changes in surface protein glycosylation [7,35].

Our understanding of adhesive interactions under dif-ferent flow conditions is also incomplete and whether gly-cosylation is an effector of adhesive interactions has notbeen well studied. Neoplastic transformation is usually as-sociated with altered binding to lectins [38] due to alteredsurface glycosylation and is also commensurate with al-tered mobility on soft agar [39]. Metastatic cancer cells un-dergo migration commensurate with continuous altered ex-pression of adhesion molecules as they undergo differentlymphatic fluid shear stress during their migration [40]. Dif-

ferent degrees of L-selectin binding are shown by differentcancer cell types when they migrate through different shearthreshold regimes. Sialylation, sulfation, and N-glycosylationare effectors of L-selectin binding. Squamous colon andleukemic cancer cells, as well as endothelial cells, show differ-ent glycosylation levels of L-selectin as they migrate throughdifferent shear thresholds [40].

Our analysis with biotinylated lectins revealed modula-tion of glycoprotein and proteoglycans in the glaucomatousTM as compared to control specimens (Fig. 1A) and islikely to capture the cellular surface glycoprotein changes.The PSORT analyses of proteins identified in our quantita-tive proteomic analyses suggest cell surface and ECM as wellas residence of identified glycoproteins (Supporting Informa-tion Table 6).

A previous study of lectin binding on human TM [15] andanother on bovine ocular tissues [16] utilized several commonlectins used here (Supporting Information Table 2). Previousstudies [15, 16] utilized DAB methods that are rather subjec-tive compared to clear contrast obtained using fluorescentdetection (Fig. 1A). We have utilized several previously un-available lectins [15] and despite their overlapping recognitionof the sugar modifications, a number of these lectins presentimportant differences in terms of peptide context for theirsugar modification recognition motif. Our overall findingsare consistent and in general agreement with previous re-sults (Fig. 1A; Supporting Information Table 2) that showedelevated glycoproteins in glaucomatous TM. In contrast, ourFACE analyses (Fig. 1B) supports co-occurrence of hypo- andhyper-glycosylation in the glaucomatous TM with respect toseveral disaccharides.

We further pursued the hypo-glycosylation in glaucoma-tous TM (Fig. 1B) but not hyper-glycosylation (in glaucoma-tous TM). It would be important and interesting to deter-mine the consequences of observed hyper-glycosylation inglaucomatous TM (Fig. 1B) and their reversal as well. How-ever, in the current study we focused on the observed hypo-glycosylation changes observed in glaucomatous TM in con-trast to controls. Primary TM cells were subjected to specificglycosylation inhibitors to achieve reduced global TM proteinglycosylation (Fig. 2D) and to determine consequent changesin the behavior and properties of the TM cells (Fig. 3A). Thesefindings are consistent with enzymatic reduction or removalof N-glycosylation affecting cell surface microenvironment,cell adhesion, and cell migration [41]. The fluid transportstudies in a multilayer cell arrangement demonstrate modu-lation of the fluid flow ostensibly due to altered behavior ofglycosylation-inhibited cells (Fig. 3B). Notably, the collagenmatrix embedded cells showed decreased tracer transport forinhibitor-treated cells compared to controls (Fig. 3B, dashedlines). It is possible that used glycosylation inhibitors are toxicto the cells. However, we did not see any discernable cell deathof primary TM cells with the concentration ranges used. Wehave also used primary retinal pigment epithelium (RPE) aswell as HEK293 cells as controls. RPE or HEK293 cells didnot show changes in cell shape or behavior similar to that



Clinical Relevance

Unlike high-throughput mRNA and protein analyses,those of PTMs, objective of current work, are rela-tively scant. Often it is the posttranslationally mod-ified protein that is active (or inactive) compared tothe nonmodified version of the protein. Understand-ing the PTM differences between normal and glau-comatous TM tissue will give mechanistic insight

into the pathology of glaucoma. Understanding theglycosylated state of proteins will also enable find-ing innovative ways to restore the normal glycosyla-tion condition and determine whether reinstating thenormal state may help restore conditions associatedwith pathology.

demonstrated by TM cells nor did layered RPE primary cellsdemonstrate any changes in tracer transport under identicalsetting (data not shown).

Fluid shear is a form of mechanical stimulus, however,mechanical stimuli in the form of biophysical cues also havebeen suggested to play a role in the TM cell expansion andremodeling [37]. The trabecular cell surfaces as well as the in-tertrabecular spaces are lined with mucinous GAGs [42, 43].Whether GAGs are associated with biophysical cues remainto be determined. Biophysical cues have been shown to resultin changes in shape and motility of TM cells [37]. Notably,LT is one of the effective intervention strategies in glaucoma.Low-threshold LT has been proposed to provide some sortof local mechanical stimulus to TM cells without causingdamage or burn to them. Specific changes due to LT havenot been clearly discernable at an ultrastructural level [44],but global TM glycosylation changes have been confirmedand detected both at cellular [45] and at tissue level in cateyes [21] subjected to LT. The LT-associated glycosylationchanges remain to be revisited using different interdisci-plinary approaches. Changes have been found in cell shapeand behavior after LT treatment [45,46]. It is important to notethat absence of ophthalmic and other history of donor eyes re-mains a problem for overall interpretation and analyses. Usu-ally donor eyes arrive and information follows. Of the 32 totalglaucoma eyes procured for these studies, we could get thestatic perimetry results confirming glaucomatous optic neu-ropathy (mean defect in Supporting Information Table 1)and other pertinent ophthalmic information for only about20 of them (62.5%). Two glaucoma eyes subjected to LT wereexcluded from final analyses therefore we could use only56.3% of the procured glaucomatous donor eyes for our finalanalyses. Another problem with glaucomatous eyes is thefrequent incomplete information about treatment medica-tions, which poses some problems in stratifying results.With recognition of these challenges, we have now initi-ated efforts in procuring glaucomatous donor eyes with morecomplete information and also for a greater medical historyperiod.

In summary, the results presented here suggest that al-tered glycosylation is associated with gross changes in theTM cell shape, motility, and their interaction with matrix con-tributing to the opening and closing of spaces between cells

and thus the general filter behavior of the TM in glaucoma.Modulation in transport of sodium fluorescein across a layerof TM cells embedded in collagen matrix suggests glycosyla-tion to play a role in control of fluid flow across TM tissue.The nonglaucomatous DBA/2J-Gpnmb+/SjJ mouse will be agood model system to study the glycosylation inhibitors stud-ies for determination of their effect on the aqueous outflowpathway. Any affect in a model animal system will suggestan underlying dynamic glycosylation sensitive mechanism inthe regulation of the aqueous outflow.

We thank Manik Goel, Gabriel Gaidosh, Avinash Saraon,Qin Yang, Praseeda Venugopalan, Hayden Sandler, AlejandroLamas, and Julee Sunny for their assistance with experiments andWilliam Feuer for assistance with statistical analyses. This studywas supported by NIH grants EY016112, EY016112S1, P30-EY14801, the Computational Ocular Genomics Training GrantT32EY023194-01 (to GE), and a career award and unrestrictedfunds from Research to Prevent Blindness.

The authors have declared no conflict of interest.

5 References

[1] Stadtman, E. R., Protein modification in aging. J. Gerontol.1988, 43, B112–B120.

[2] Bhattacharya, S. K., Retinal deimination in aging and dis-ease. IUBMB Life 2009, 61, 504–509.

[3] Jaisson, S., Gillery, P., Evaluation of nonenzymatic post-translational modification-derived products as biomarkersof molecular aging of proteins. Clin. Chem. 2010, 56,1401–1412.

[4] Quigley, H. A., Broman, A. T., The number of people withglaucoma worldwide in 2010 and 2020. Br. J. Ophthalmol.2006, 90, 262–267.

[5] Morrison, J. C., Johnson, E. C., Cepurna, W., Jia, L., Un-derstanding mechanisms of pressure-induced optic nervedamage. Prog. Retin. Eye. Res. 2005, 24, 217–240.

[6] Goel, M., Picciani, R. G., Lee, R. K., Bhattacharya, S. K., Aque-ous humor dynamics: a review. Open. Ophthalmol. J. 2010,4, 52–59.

[7] Tarbell, J. M., Weinbaum, S., Kamm, R. D., Cellular fluid me-chanics and mechanotransduction. Ann. Biomed. Eng. 2005,33, 1719–1723.



[8] Yao, Y., Rabodzey, A., Dewey, C. F., Jr., Glycocalyx modulatesthe motility and proliferative response of vascular endothe-lium to fluid shear stress. Am. J. Physiol. Heart. Circ. Physiol.2007, 293, H1023–H1030.

[9] Diskin, S., Kumar, J., Cao, Z., Schuman, J. S. et al., Detectionof differentially expressed glycogenes in trabecular mesh-work of eyes with primary open-angle glaucoma. Invest.Ophthalmol. Vis. Sci. 2006, 47, 1491–1499.

[10] Acott, T. S., Kelley, M. J., Extracellular matrix in the trabecularmeshwork. Exp. Eye. Res. 2008, 86, 543–561.

[11] Haddadin, R. I., Oh, D. J., Kang, M. H., Filippopoulos, T. et al.,SPARC-null mice exhibit lower intraocular pressures. Invest.Ophthalmol. Vis. Sci. 2009, 50, 3771–3777.

[12] Wirtz, M. K., Bradley, J. M., Xu, H., Domreis, J. et al., Proteo-glycan expression by human trabecular meshworks. Curr.Eye. Res. 1997, 16, 412–421.

[13] Bhattacharya, S. K., Rockwood, E. J., Smith, S. D., Bonilha, V.L. et al., Proteomics reveals cochlin deposits associated withglaucomatous trabecular meshwork. J. Biol. Chem. 2005,280, 6080–6084.

[14] Gygi, S. P., Rochon, Y., Franza, B. R., Aebersold, R., Correla-tion between protein and mRNA abundance in yeast. Mol.Cell. Biol. 1999, 19, 1720–1730.

[15] Chapman, S. A., Bonshek, R. E., Stoddart, R. W.,O’Donoghue, E. et al., Glycans of the trabecular meshworkin primary open angle glaucoma. Br. J. Ophthalmol. 1996,80, 435–444.

[16] Tuori, A., Virtanen, I., Uusitalo, H., Lectin binding in the an-terior segment of the bovine eye. Histochem. J. 1994, 26,787–798.

[17] Calabro, A., Benavides, M., Tammi, M., Hascall, V. C. et al.,Microanalysis of enzyme digests of hyaluronan and chon-droitin/dermatan sulfate by fluorophore-assisted carbohy-drate electrophoresis (FACE). Glycobiology 2000, 10, 273–281.

[18] Calabro, A., Hascall, V. C., Midura, R. J., Adaptation ofFACE methodology for microanalysis of total hyaluronanand chondroitin sulfate composition from cartilage. Glyco-biology 2000, 10, 283–293.

[19] Parikh, T., Eisner, N., Venugopalan, P., Yang, Q. et al., Pro-teomic analyses of corneal tissue subjected to alkali expo-sure. Invest. Ophthalmol. Vis. Sci. 2010, 52, 1819–1831.

[20] Tang, W. H., Shilov, I. V., Seymour, S. L., Nonlinear fittingmethod for determining local false discovery rates from de-coy database searches. J. Proteome. Res. 2008, 7, 3661–3667.

[21] Amelinckx, A., Castello, M., Arrieta-Quintero, E., Lee, T. et al.,Laser trabeculoplasty induces changes in the trabecularmeshwork glycoproteome: a pilot study. J. Proteome. Res.2009, 8, 3727–3736.

[22] Aljohani, A. J., Munguba, G. C., Guerra, Y., Lee, R. K. et al.,Sphingolipids and Ceramides of Human Aqueous Humor.Mol. Vis. 2013, 19, 1966–1984.

[23] Schaefer, A. S., Richter, G. M., Nothnagel, M., Manke, T. et al.,A genome-wide association study identifies GLT6D1 as asusceptibility locus for periodontitis. Hum. Mol. Genet. 2010,19, 553–562.

[24] Hinek, A., Biological roles of the non-integrin elastin/lamininreceptor. Biol. Chem. 1996, 377, 471–480.

[25] Keller, K. E., Aga, M., Bradley, J. M., Kelley, M. J. et al.,Extracellular matrix turnover and outflow resistance. Exp.Eye. Res. 2009, 88, 676–682.

[26] Rhee, D. J., Haddadin, R. I., Kang, M. H., Oh, D. J., Matricellu-lar proteins in the trabecular meshwork. Exp. Eye. Res. 2009,88, 694–703.

[27] Alcazar, O., Hawkridge, A. M., Collier, T. S., Cousins, S. W.et al., Proteomics characterization of cell membrane blebsin human retinal pigment epithelium cells. Mol. Cell. Pro-teomics 2009, 8, 2201–2211.

[28] Ilagan, R., Guthrie, K., Quinlan, S., Rapoport, H. S. et al., Lin-ear measurement of cell contraction in a capillary collagengel system. Biotechniques 2010, 48, 153–155.

[29] Kuleshova, O. N., Zaidman, A. M., Korel, A. V., Glycosamino-glycans of the trabecular meshwork of the eye in primaryjuvenile glaucoma. Bull. Exp. Biol. Med. 2007, 143, 381–384.

[30] Kuleshova, O. N., Nepomnyashchikh, G. I., Aidagulova, S. V.,Ultrastructure of connective tissue of eye drainage systemin ophthalmic hypertension associated with primary juvenileglaucoma. Bull. Exp. Biol. Med. 2008, 145, 374–376.

[31] Spicer, S. S., Schulte, B. A., Diversity of cell glycoconju-gates shown histochemically: a perspective. J. Histochem.Cytochem. 1992, 40, 1–38.

[32] Hsu, C. C., Lin, T. W., Chang, W. W., Wu, C. Y. et al.,Soyasaponin-I-modified invasive behavior of cancer bychanging cell surface sialic acids. Gynecol. Oncol. 2005, 96,415–422.

[33] Wang, H., Young, S. R., Gerard-O’Riley, R., Hum, J. M.et al., Blockade of TNFR1 signaling: A role of oscillatoryfluid shear stress in osteoblasts. J. Cell. Physiol. 2011, 226,1044–1051.

[34] Shi, Z. D., Abraham, G., Tarbell, J. M., Shear stress modu-lation of smooth muscle cell marker genes in 2-D and 3-Ddepends on mechanotransduction by heparan sulfate pro-teoglycans and ERK1/2. PLoS ONE 2010, 5, e12196.

[35] Toh, Y. C., Voldman, J., Fluid shear stress primes mouse em-bryonic stem cells for differentiation in a self-renewing en-vironment via heparan sulfate proteoglycans transduction.FASEB J. 2011, 25, 1208–1217.

[36] Tripathi, B. J., Tripathi, R. C., Neural crest origin of humantrabecular meshwork and its implications for the pathogen-esis of glaucoma. Am. J. Ophthalmol. 1989, 107, 583–590.

[37] Gasiorowski, J. Z., Russell, P., Biological properties of trabec-ular meshwork cells. Exp. Eye. Res. 2009, 88, 671–675.

[38] Rambaruth, N. D., Dwek, M. V., Cell surface glycan-lectininteractions in tumor metastasis. Acta Histochem. 2011, 113,591–600.

[39] Agre, P., Williams, T. E., The human tumor cloning assay incancer drug development. A review. Invest. New Drugs 1983,1, 33–45.

[40] Resto, V. A., Burdick, M. M., Dagia, N. M., McCammon, S. D.et al., L-selectin-mediated lymphocyte-cancer cell interac-tions under low fluid shear conditions. J. Biol. Chem. 2008,283, 15816–15824.



[41] Krahling, H., Mally, S., Eble, J. A., Noel, J. et al., The glyco-calyx maintains a cell surface pH nanoenvironment crucialfor integrin-mediated migration of human melanoma cells.Pflugers. Arch. 2009, 458, 1069–1083.

[42] Zimmerman, C. C., Lingappa, V. R., Richards, J. E., Rozsa, F.W. et al., A trabecular meshwork glucocorticoid response(TIGR) gene mutation affects translocational processing.Mol. Vis. 1999, 5, 19–24.

[43] Knepper, P. A., Goossens, W., Hvizd, M., Palmberg, P. F.,Glycosaminoglycans of the human trabecular meshwork inprimary open-angle glaucoma. Invest. Ophthalmol. Vis. Sci.1996, 37, 1360–1367.

[44] Stoiber, J., Fernandez, V., Lamar, P. D., Decker, S. J.et al., Trabecular meshwork alteration and intraocular pres-sure change following pulsed near-infrared laser trabeculo-plasty in cats. Ophthalmic. Surg. Lasers. Imaging. 2005, 36,471–481.

[45] Van Buskirk, E. M., Pond, V., Rosenquist, R. C., Acott, T. S.,Argon laser trabeculoplasty. Studies of mechanism of action.Ophthalmology 1984, 91, 1005–1010.

[46] Acott, T. S., Samples, J. R., Bradley, J. M., Bacon, D. R. et al.,Trabecular repopulation by anterior trabecular meshworkcells after laser trabeculoplasty. Am. J. Ophthalmol. 1989,107, 1–6.


aberrant glycosylation in the human trabecular meshwork

Documents