|Molecular Vision 2007;
Received 7 May 2007 | Accepted 11 October 2007 | Published 30 October 2007
Proteomic profiling of human retinal and choroidal endothelial cells reveals molecular heterogeneity related to tissue of origin
David O. Zamora,1
Stephen R. Planck,1,4,5
James T. Rosenbaum,1,4,5
Larry L. David,2
Justine R. Smith,1,4
(The last two authors contributed equally to this publication)
1Casey Eye Institute, Departments of 2Biochemistry and Molecular Biology, 3Public Health and Preventive Medicine (Division of Biostatistics), 4Cell and Developmental Biology, and 5Medicine (Division of Arthritis and Rheumatic Diseases), Oregon Health and Science University, Portland, OR
Correspondence to: Justine R. Smith, Biomedical Research Building, L467AD, Oregon Health and Science University, 3181 SW Sam Jackson Park Road, Portland, Oregon 97239; Phone: (503) 494-5023; FAX: (503) 494-6875; email: email@example.com
Purpose: The ocular vascular endothelium plays a key role in the development of several leading retinal causes of blindness in Western nations. Choroidal endothelial cells are integral to the subretinal neovascular lesions that characterize the exudative form of late age-related macular degeneration (AMD), and retinal endothelial cells participate in the initiation of diabetic retinopathy and posterior uveitis. Vascular endothelial cells at different sites exhibit considerable molecular diversity. This diversity has implications for understanding the pathogenesis of tissue-specific diseases and for the development of targeted therapies to treat these conditions. Previous work from our group has identified significant differences in the gene transcript profiles of human retinal and choroidal endothelial cells. Because the proteome ultimately determines the behavior of any given cell, however, it is critical to determine whether molecular differences exist at the level of protein expression.
Methods: Retinal and choroidal endothelial cells were separately isolated from five sets of human eyes by enzymatic digestion with type II collagenase followed by anti-CD31 antibody-conjugated magnetic bead separation. Cells were washed to remove serum peptides in the culture medium, and lysed by sonication in buffer containing 2% sodium dodecyl sulfate. Protein was then precipitated with acetone. Retinal and choroidal endothelial samples from each donor were labeled with Cy3 and Cy5, respectively, mixed with a Cy2-labeled pooled protein sample to facilitate spot matching across gels, and separated by two-dimensional difference gel electrophoresis (2D-DIGE). Following a global normalization, differentially abundant protein spots that were visible in at least four of five donor gels were detected by the significance analysis of microarrays method, with false discovery rate set at 5%. Corresponding spots were excised from additional DIGE-labeled or Coomassie-stained 2D electrophoretic gels. Protein identification was performed by liquid chromatography and tandem mass spectrometry.
Results: Of 123 protein spots detected by 2D-DIGE that qualified for statistical analysis, we found 31 spots that demonstrated a significant difference in abundance between retinal endothelial samples versus choroidal endothelial samples. For 17 proteins, over 50% of the spectral counts could be matched to a single protein in the digested spot. Eleven proteins were more abundant in retinal endothelial cells (i.e., inorganic pyrophosphatase, protein disulfide isomerase A3, calreticulin, peroxiredoxin-4, protein disulfide isomerase, serpin B9, F-actin capping protein subunit β, coactosin-like protein, vimentin, cathepsin B, and a high molecular weight form of annexin A3). Six proteins were more abundant in choroidal endothelial cells (i.e., glutathione peroxidase 1, ubiquitin carboxyl-terminal hydrolase isozyme L1 (UCH-L1), heat-shock protein beta-1, superoxide dismutase (Cu-Zn), nucleoside diphosphate kinase A, and a low molecular weight form of annexin 3).
Conclusions: Our data indicate that the proteomes of retinal and choroidal vascular endothelial cells are different. Several differentially expressed proteins are implicated in the regulation of angiogenesis; these include cathepsin B and UCH-L1, proteins with transcripts that were also differently expressed according to microarray. Our observations further suggest that angiogenesis within the retina, a component of severe diabetic retinopathy and posterior uveitis, may be controlled by different mechanisms to those regulating choroidal neovascularization, as occur in exudative AMD. Future studies to establish the role of these angiogenic proteins in disease may suggest potential new targets for tissue-specific therapies.
Diseases of the posterior eye represent major causes of visual impairment globally , reflecting the fact that the retina contains photoreceptors and other non-regenerating neural cells that are critical to visual processing. Age-related macular degeneration (AMD), diabetic retinopathy, and retinal infection and inflammation, which are frequently grouped together as posterior uveitis, are leading causes of irreversible low vision and blindness for adults in Western nations. In the United States, AMD has a 15-year cumulative incidence of approximately 17%; this figure is substantially higher for those aged 75 years or older, of whom 4% suffer from the exudative form of late AMD . Approximately 3.4% of individuals aged 40 years or older are affected by diabetic retinopathy ; the incidence of visual loss increases with time, but has been estimated to be as high as 37.2% at 10 years for older-onset individuals taking insulin . Posterior uveitis is a less common disease, affecting 115 people per 100,000 in the general population , but the burden of disease is substantial. In a Western European study, visual impairment was recorded in approximately half of 314 patients whose ocular inflammation involved the posterior segment of the eye .
The vascular endothelial cell plays a critical role in the pathogenesis of these posterior ocular diseases. An imbalance of angiogenic and anti-angiogenic factors promotes the proliferation of choroidal endothelial cells in subretinal neovascular networks that ultimately lead to the exudative detachments of the retinal pigment epithelium as well as the neurosensory retina that define late AMD . In diabetic retinopathy, retinal endothelial cell dysfunction and neoangiogenesis correlate with accumulation of advanced glycation end-products . Infiltrating leukocytes or invading microbes must migrate across the so-called blood-retinal barrier, of which the retinal endothelial cell is a key component, in order for posterior uveitis to be initiated [9,10]. Due to its major participation in the development of the pathology, the vascular endothelial cell is an obvious choice for targeted drug therapies to treat posterior ocular disease. Of particular interest are the molecular distinctions that characterize the retinal and choroidal endothelial cell populations specifically.
Published studies utilizing gene expression microarray have confirmed vascular endothelial diversity between cells from different organs and from different tissues within an organ [11-13]. Our group recently reported a 9% difference in the gene expression of retinal cells versus choroidal endothelial cells . While gene expression profiling has provided much evidence in support of endothelial diversity, mRNA levels do not necessarily correlate with protein levels and certainly do not reflect post-translational modifications. Protein expression ultimately determines the structure and function of a cell and thus arguably is of primary interest in defining cell diversity. In this study, we used two-dimensional difference gel electrophoresis (2D-DIGE) in combination with tandem mass spectrometry to compare the relative abundance of proteins isolated from separately cultured, but donor-matched, human retinal and choroidal endothelial cells. To the best of our knowledge, this is the first time a proteomics approach has been employed to study ocular endothelial cell diversity.
Isolation and culture of retinal and choroidal vascular endothelial cells
Retinal and choroidal vascular endothelial cells were derived from five sets of human eyes obtained from the Oregon Lions Eye Bank. Age at death and gender for the anonymous donors were as follows: 49-year-old male (Donor 1); 40-year-old female (Donor 2); 35-year-old female (Donor 3); 46-year-old male (Donor 4); and 49-year-old male (Donor 5). Donors had no known history of ocular disease. The limited systemic medical history available to us indicated that there were no specific vascular diseases affecting a majority of donors. Time from death to isolation varied from 13 to 19 h. Use of human cadaveric cells for the purposes of these experiments was approved by the Oregon Health and Science University Institutional Review Board.
Retina and choroid were dissected from posterior eyecups. The retinal pigment epithelium was removed from the choroidal tissue with a cotton swab. Endothelial cells were isolated from the two choroids and two retinas from each donor by the following general procedure. Tissues were separately treated with graded solutions (up to 3 mg/ml) of type II collagenase (Sigma-Aldrich, St Louis, MI) in HEPES-buffered MCDB-131 medium (Sigma-Aldrich; catalog number 8537), supplemented with 2% fetal bovine serum (FBS; Hyclone, Logan, UT) and 1 μg/ml amphotericin B (Gibco, Invitrogen, Grand Island, NY), at 37 °C and 5% CO2, until the tissue was visibly digested. In the case of retina, digestion with type II collagenase was preceded by an overnight incubation with 0.3 mg/ml dispase (Gibco, Invitrogen). Endothelial cells were isolated from cell suspensions using magnetic beads (Dynabeads; Dynal, Invitrogen, Brown Deer, WI) coated with mouse monoclonal anti-human CD31 antibody (BD Pharmingen, San Diego, CA). Retinal and choroidal endothelial cells were cultured in separate dishes in MCDB-131 medium supplemented with 10% FBS, and 1 μg/ml amphotericin B and endothelial growth factors (Clonetics, Cambrex Bioscience, Walkerville, MD; EGM-2 SingleQuot catalog number cc-4176, omitting gentamicin, hydrocortisone, and FBS), at 37 °C and 5% CO2, and passed using 0.05% trypsin (Gibco, Invitrogen). If cells were not used immediately in the experiments described below, they were snap frozen in 10% dimethyl sulfoxide and stored in liquid nitrogen. Cells were used at or below passage five in this study.
Preparation of protein samples
After retinal and choroidal endothelial cells reached confluence in separate 10 cm dishes, the medium was replaced with fresh MCDB-131 medium supplemented with 5% FBS and endothelial growth factors, and cells were cultured for a further 4 h. Subsequently the dishes were gently washed four times with phosphate buffered saline (Gibco, Invitrogen, catalog number 14040) at room temperature to remove serum proteins and then snap frozen at -80 °C in preparation for the study. When dishes were thawed, 0.5 ml of lysis buffer containing 40 mM Tris and 2% sodium dodecyl sulfate (pH 7.5) and one tablet of protease inhibitor cocktail per 10 ml solution (Complete, mini EDTA-Free, Roche Applied Science, Indianapolis, IL) was added to each. Cells were removed with a cell scraper. Each resultant suspension was transferred to a separate 1.5 ml centrifuge tube, which was sonicated three times for 10 s at a power setting of 4 W with cooling between bursts (60 Sonic Dismembrator, Fisher Scientific, Hampton, NH). Subsequently, samples were centrifuged at 20,000 g for 15 min at 4 °C. Following removal of the supernatant, proteins were precipitated by incubation in 3.5 times the volume of ice cold acetone with 1 mM HCl at -20 °C for one hour. Finally, precipitates were isolated by centrifugation at 15,000 g for 10 min at 4 °C. Pellets were air-dried for 5 min.
Labeling of protein samples with Cy dyes
Each pellet derived from one dish was dissolved in 0.2 ml of lysis buffer containing 30 mM Tris (pH 8.5), 7 M urea, 2 M thiourea, and 4% (w/v) CHAPS by sonication as described in the previous paragraph. Protein content was determined by the bicinchoninic acid (BCA) assay (Pierce Chemical, Rockville, IL) using bovine serum albumin as a standard, yielding approximately 500 μg of protein per dish. Separate solutions containing 50 μg of one retinal or choroidal sample were labeled with Cy3 or Cy5 minimal dyes, respectively, and a common pool of protein created from each sample was labeled with Cy2 minimal dye. Labeling was performed as recommended by the manufacturer (GE Healthcare, Piscataway, NJ) using 300 pmoles dye per 50 μg protein. Labeled samples were diluted with an equal volume of buffer containing 7 M urea, 2 M thiourea, 100 mM dithiothreitol (DTT), and 4% (v/v) pH 4-7 immobilized pH gradient (IPG) buffer (GE Healthcare). After a 10-minute incubation on ice, samples were mixed such that Cy3- and Cy5-labeled proteins from the retinal or choroidal endothelial cells of the same donor, and 50 μg of Cy2-labeled pooled protein from all donors, were combined. This mixture was diluted to a final 340 μl volume in 7 M urea, 2 M thiourea, 2% (w/v) CHAPS, 50 mM DTT, and 2% (v/v) pH 4-7 IPG buffer, with a trace of bromphenol blue.
Two-dimensional difference gel electrophoresis and image analysis
Protein preparations were applied to 18 cm (pH 4-7) IPG strips (G.E. Healthcare) by overnight re-swelling. Strips were focused using a Protean IEF Cell (Bio-Rad, Hercules, CA) at 20 °C under the following conditions: 0-500V for 6 h (rapid), 500-3500 V for 3 h (linear), 3500 V for 8 h, with a 50 μA limit per gel. Subsequently strips were then reduced and alkylated. A second dimension separation was performed using 12% polyacrylamide gels as previously described , with the exception that prior to polymerization, one glass plate was treated with Bind-Silane, and the other was treated with Repel-Silane as recommended by the manufacturer (GE Healthcare) to facilitate spot excision. Five gels were run, each containing Cy3-labeled retinal samples and Cy3-labeled choroidal samples from a single donor, as well as the Cy2-labeled pooled protein. Gels were scanned at 100 μm resolution using a DIGE Enabled Typhoon 9400 scanner (GE Healthcare) to obtain individual images for paired choroidal and retinal endothelial samples and the pooled protein sample. Intensities (volumes) of Cy3- and Cy5-labeled spots were determined using Phoretix 2D Evolution software (version 2005, Nonlinear Dynamics, Durham, NC). Simultaneous separation of samples within single gels greatly simplified spot matching between retinal and choroidal endothelial proteins, and addition of the pooled protein sample to each gel in the experiment facilitated spot identification between gels.
Intensities of all Cy3 (retinal endothelial protein) and Cy5 (choroidal endothelial protein) spots were transformed into the log2 scale. For each channel, median signal intensity of the gel was subtracted and median signal intensity across all gels was added, to correct for potential bias related to general differences in signal intensity between the gels. This method of normalization is equivalent to performing a global normalization in microarray analysis, i.e., one channel is re-scaled to the other channel by multiplying by a constant factor . Spots detected in at least four of five gels, and therefore for four of the five donors, were subjected to further statistical analysis. Significant differences in the log2 intensities of Cy3 versus Cy5 spots (abundance of retinal endothelial protein spots versus choroidal endothelial protein spots) were determined by the significance analysis of microarrays (SAM) method . Missing intensities were imputed by the default function of the SAM. The SAM allows for a small number of samples, and it uses the concepts of the false discovery rate (FDR)  and the q value . The FDR differs from more conventional corrections for multiple comparisons in that instead of controlling for false positives, it controls for the expected ratio of false positives among significantly expressed genes. The q value is a posterior Bayesian p value, and it refers to the minimum FDR at which a test is deemed to show a statistically significant difference. In this analysis, the FDR was set at 5%, with a significant difference in protein abundance defined as one yielding a q value less than 0.05.
Identification of differentially abundant proteins
Proteins significantly differing in abundance were obtained for identification by either directly excising spots from Cy dye-labeled gels or pooling samples from the five retinal or choroidal endothelial cell samples and running 2-dimensional electrophoretic (2-DE) gels as described in a previous section, with the following exceptions: approximately 400 μg of protein was applied; no silanization was performed; and gels were stained with Coomassie G-250 . Proteins of interest were manually excised using a OneTouch manual spot picker (P2D3.0, The Gel Company, San Francisco, CA). Gel plugs were shaken twice in 500 μl of 50 mM ammonium bicarbonate, 50% (v/v) acetonitrile wash solution for 30 min, dried by vacuum centrifugation, and rehydrated by a 15-min incubation on ice in 30 μl of trypsin digestion solution containing 50 mM ammonium bicarbonate, 4 mM calcium chloride, and 12.5 μg/ml sequencing grade modified trypsin (Promega, Madison, WI). Following rehydration, excess solution was removed, and 60 μl of a solution of 50 mM ammonium bicarbonate and 4 mM calcium chloride was added. Samples were then incubated at 37 °C overnight. Peptides were removed by addition of 3 μl of 88% formic acid, shaking for 15 min, and spinning briefly. Solutions were transferred directly to auto-sampler tubes for analysis.
Peptides were injected onto a 1 mmx8 mm trap column (Michrom BioResources, Inc., Auburn, CA) at 20 μl/min in a mobile phase containing 0.1% formic acid. The trap cartridge was then placed in-line with a 0.5 mmx250 mm column containing 5 mm Zorbax SB-C18 stationary phase (Agilent Technologies, Palo Alto, CA), and peptides were separated by a 2-30% acetonitrile gradient over 60 min at 10 μl/min using a 1100 series capillary HPLC (Agilent). Peptides were analyzed using a LTQ linear ion trap fitted with an Ion Max Source and 34-gauge metal needle kit (ThermoFinnigan, San Jose, CA). Survey mass spectrometry (MS) scans were alternated with three data-dependent MS/MS scans using the dynamic exclusion feature of the software to increase the number of unique peptides analyzed. Peptides were identified by comparing observed MS/MS spectra to theoretical fragmentation spectra of peptides generated from a protein database using Sequest (version 27, rev. 12, ThermoFinnigan). We employed a human-only version of the Swiss-Prot database (Swiss Institute of Bioinformatics, Geneva, Switzerland) containing 13847 entries. Scaffold (version Scaffold-01_06_00, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Protein identifications were accepted if they could be established at greater than 99.0% probability and contained at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm .
When imaged by Phoretix 2D Evolution software, the 2D-DIGE gels of choroidal and retinal endothelial cell protein samples contained approximately 2,514 protein spots. Of these spots, we selected a subset of the most abundant for further study. As we analyzed only those spots that were present in four of five gels, or detected in four of five donors, a total of 123 protein spots were analyzed for differential abundance in retinal endothelial cells versus choroidal endothelial cells. Using the SAM, with the FDR set at 5%, we identified 31 protein spots that were significantly different in abundance between retinal and choroidal endothelial protein preparations, including 20 spots that were more abundant in retinal samples and 11 spots that were more abundant in choroidal samples. These 31 spots are highlighted on the representative gel image shown in Figure 1.
Protein spots were excised from either Cy dye-labeled or Coomassie-stained gels during 2D-DIGE, digested with trypsin, and subjected to tandem mass spectrometry with a LTQ linear ion trap. After excluding proteins that were either carry-over or contaminants (keratins and trypsin), we were able to identify specifically 16 proteins for which over 50% of the spectral counts could be matched to a single protein in the spot digest, including 11 proteins, represented by 12 spots, that were more abundant in retinal endothelial cells and six proteins, represented by six spots, that were more abundant in choroidal endothelial cells. Three protein spots were too heterogeneous for specific protein identification. Ten additional spots could not be identified. Proteins that were significantly more abundant in retinal endothelial cells included: inorganic pyrophosphatase, protein disulfide isomerase A3, calreticulin, peroxiredoxin-4, protein disulfide isomerase, serpin B9, F-actin capping protein subunit beta, coactosin-like protein, vimentin, and cathepsin B. Proteins that were significantly more abundant in choroidal endothelial cells included: glutathione peroxidase 1, ubiquitin carboxyl-terminal hydrolase isozyme L1 (UCH-L1), heat-shock protein β-1, superoxide dismutase (Cu-Zn), and nucleoside diphosphate kinase A. Annexin 3 was identified as the major component of one additional spot that contained significantly more retinal protein, and it was also identified in a second additional spot of lower molecular weight that contained significantly more choroidal protein. This was explained on the basis of a selective degradation of annexin 3 in choroidal endothelial cells. Table 1 and Table 2 present lists of differentially abundant protein spots, including fold changes and q values. In Appendix 1, we present comprehensive protein composition data for the spots in which we were able to identify protein species.
In this study, we used 2D-DIGE to compare the protein profiles of primary cultured human donor-matched retinal and choroidal endothelial cells. These endothelial cells have been characterized in a previous publication ; they exhibit typical cobblestone morphology, express CD31 and von Willebrand factor, take up acetylated low-density lipoprotein, and form capillary-like networks on a synthetic basement membrane. A majority of proteins were not differentially expressed between retinal and choroidal cells, consistent with the common microvascular endothelial phenotype. Yet, these two endothelial cell subtypes exhibited reproducible differences in abundance of approximately 25% of the protein spots that qualified for statistical analysis. For gels prepared separately for each of five donors, 123 protein spots were found suitable for comparison, based on the strict requirement of visible presence in four of five gels. Statistical analysis indicated that 31 of the 123 spots were of significantly different abundance in retinal endothelial cells versus choroidal endothelial cells. Using tandem mass spectrometry, we identified 11 retinal endothelial cell-specific and six choroidal endothelial cell-specific proteins.
Our findings are consistent with results of our recently published gene expression microarray analysis , which demonstrated that retinal and choroidal endothelial cells display distinct patterns of gene expression. In showing that molecular differences between these cells also exist at the level of the proteome, the present study provides direct support for the hypothesis that retinal and choroidal endothelial cells have different molecular compositions and activities. On first consideration, vascular endothelial diversity within the posterior eye may appear surprising, given that the retina and choroid are immediately opposed anatomically and that their vascular beds derive from the same source. On the other hand, this is consistent with respective tissue locations on different sides of the blood-retinal barrier, the fact that retina and choroid perform unique functions within the eye, and participation of the two vascular endothelial cell populations in very different ocular diseases.
Ocular neovascularization frequently occurs in blinding forms of posterior eye disease that are common in Western countries. Proliferative diabetic retinopathy is defined by the presence of new retinal vessels on the optic disc or at distant retinal locations. In aggressive forms of posterior uveitis, such as Behcet's disease and sarcoidosis, retinal neovascularization is well recognized. Choroidal neovascular networks are characteristic of exudative AMD. Consequently it is of immediate interest to note that a number of the differentially expressed proteins are involved in the regulation of angiogenesis, including calreticulin and cathepsin B, which are more abundant in retinal endothelium, and glutathione peroxidase 1, UCH-L1 and superoxide dismutase (Cu-Zn), which are more abundant in choroidal endothelium. In a report from the First ARVO/Pfizer Ophthalmics Institute meeting, Campochiaro et al.  highlighted evidence that angiogenesis is likely to be differentially regulated at different tissue sites. Identification of retinal and choroidal specific angiogenic regulators provides support for this phenomenon within the eye.
Of the differentially expressed angiogenic regulators, cathepsin B and UCH-L1 deserve particular attention; we found these two molecules to be differentially expressed by gene expression microarray . Cathepsin B is a lysosomal cysteine protease that participates in the turnover of extracellular matrix, both directly and by activation of other proteolytic enzymes. It promotes the growth of capillary-like tubes by human umbilical venous endothelial cells on Matrigel matrix , and it has been implicated in angiogenesis in brain tumors . Cathepsin B transcript and protein are significantly more abundant in retinal endothelial cells than choroidal endothelial cells, suggesting that the protein might promote retinal neovascularization in diabetic retinopathy as well as posterior uveitis. Interestingly, however, in experiments reported by Im et al. , over-expression of cathepsin B reduced the formation of vascular endothelial growth factor (VEGF)-induced capillary tubes by bovine retinal endothelial cells. Inhibition of cathepsin B permitted tube growth in the absence of angiogenic stimuli, as endothelial cells synthesized VEGF and hypoxia-inducible factor (HIF)-1α, but decreased the production of endostatin. The authors offered the plausible explanation that in retina, cathepsin B acts to preserve the normal non-angiogenic state of endothelial cells.
UCH-L1 may also modulate ocular angiogenesis, but in the choroid, since endothelial cells of this tissue express relatively high levels of transcript and protein. Although generally considered specific to neurons and neuroendocrine cells , expression by endothelial cells has been recognized in another proteomic study . This protein is a member of a large family of deubiquitinating enzymes and the ubiquitin carboxy-terminal hydrolase subclass . The ubiquitin-proteasome system is a major mechanism for degradation of cellular proteins, which are conjugated to ubiquitin as the first step. While critical for normal cell function, this system has also been implicated in disease and pathological angiogenesis . Bargagna-Mohan et al.  used a three-dimensional endothelial sprouting assay to show that choroidal angiogenesis might be curbed by inhibiting the ubiquitin-proteasome pathway with natural inhibitors derived from the plant, Withania somnifera; this result implies that UCH-L1 might exhibit anti-angiogenic activity. Yet the situation is not simple. Deubiquitination serves several functions, including protein rescue, but the specific cellular roles of the 90 identified deubiquitinating enzymes, including UCH-L1, remain a subject for research . In one recent report, Selinger et al. suggested that in certain settings, UCH-L1 may counteract VHL gene-driven HIF-1α ubiquitination, stabilizing HIF-1α, which would be expected to promote angiogenesis . Additional studies are indicated to clarify how UCH-L1 might influence the development of choroidal new vessels in AMD.
Results from this work can be contrasted with those of our gene expression microarray study  in two ways. First, in our microarray study, we identified 779 gene transcripts to be significantly differentially expressed between retinal and choroidal endothelial cells, while in this study we identified less than 5% of that number of differentially expressed proteins. Despite the much greater number of cellular proteins over transcripts, present proteomics technology simply does not achieve the number of molecular identifications as gene expression microarray. Second, we found a disparity between the lists of differentially expressed molecules obtained using the two different methodologies. Only three of the 17 differentially abundant protein spots that were specifically identified corresponded with transcripts that were differentially expressed in retinal endothelial cells versus choroidal endothelial cells (cathepsin B, serpin B9, and UCH-L1). We confirmed the differential expression of these three proteins in an analysis of cultured retinal and choroidal endothelial cell protein samples that were extracted from an independent human donor, by 2D liquid chromatography and tandem mass spectrometry, using the spectral count data to establish relative protein abundance  (data not shown). Interestingly, fold differences in expression of cathepsin B, serpin B9, and UCH-L1 transcript by microarray  were comparable to fold differences in abundance measured at the protein level. While it is possible that additional "common molecules" were present in the 15 spots that lacked specific identification, poor correlation between the transcriptome, as profiled by microarray, and the proteome, as studied by proteomics methodology, is a problem recognized by other groups [35-37] and a field of statistical investigation . Certain proteins are difficult to detect by 2D-DIGE, such as those with extreme isoelectric points or molecular weights, low abundance proteins, and membrane-bound proteins. Because much endothelial specialization occurs at the cell surface, this point has relevance to our study of endothelial diversity. Indeed, membrane-bound molecules were one group of transcripts that we noted to be differentially expressed by retinal endothelial cells versus choroidal endothelial cells on microarray. The advent of gel-free "shotgun proteomics" promises to advance the field of endothelial proteomics in terms of identifying membrane proteins and increasing the number of protein identifications in general .
In conclusion, we have shown that the proteomes of retinal and choroidal vascular endothelial cells are different, and that these differences may have implications for local angiogenesis. While experiments using cultured cells have the advantage of studying a specific cell population and providing large amounts of material for analysis, the process of cell culture may alter cell phenotype. To address this concern, we used primary isolates that were of early passage and at 99% purity. Furthermore, phenotypic drift in culture usually decreases cell-specific protein expression. Consequently it is unlikely that the significant differences in protein expression identified between retinal and choroidal endothelial cells were created by the process of cell culture. It is critical to study any putative angiogenic regulators at a functional level in both in vitro and in vivo models of ocular angiogenesis. An obvious approach would be to examine the effect of specific neutralizing antibodies or anti-sense oligonucleotides on capillary-like tube formation by cultured human ocular endothelial cells . Experimental models of ocular vascular disease provide opportunities for additional future studies. Mice and non-human primates expressing high levels of vascular endothelial growth factor in the posterior segment as a result of transgenic technology and gene transfer via virus, respectively, develop proliferative retinal vascular disease [40,41]. Senescent mice deficient in CCL2 or CCR2 present a useful model of neovascular AMD . Apart from the obvious implications of differentially expressed proteins for understanding the pathogenesis of disease, constitutively expressed proteins implicated in the initiation of a disease may provide the target for a specific therapy. In addition, irrelevant proteins expressed at relatively high levels can provide targets for tissue-specific treatments. By elucidating distinctions between retinal and choroidal vascular endothelial cells, we envision future therapies for AMD, diabetic retinopathy and posterior uveitis that can be delivered to specific cells of the posterior eye.
This work was supported in part by grants from the National Institutes of Health (EY014909 and EY010572) and Research to Prevent Blindness (Career Development Award to J.R.S., Scholar Award to S.R.P., Senior Scholar Award to J.T.R., and an unrestricted grant to Casey Eye Institute).
1. Congdon NG, Friedman DS, Lietman T. Important causes of visual impairment in the world today. JAMA 2003; 290:2057-60.
2. Klein R, Klein BE, Knudtson MD, Meuer SM, Swift M, Gangnon RE. Fifteen-year cumulative incidence of age-related macular degeneration: the Beaver Dam Eye Study. Ophthalmology 2007; 114:253-62.
3. Kempen JH, O'Colmain BJ, Leske MC, Haffner SM, Klein R, Moss SE, Taylor HR, Hamman RF, Eye Diseases Prevalence Research Group. The prevalence of diabetic retinopathy among adults in the United States. Arch Ophthalmol 2004; 122:552-63.
4. Moss SE, Klein R, Klein BE. Ten-year incidence of visual loss in a diabetic population. Ophthalmology 1994; 101:1061-70.
5. Gritz DC, Wong IG. Incidence and prevalence of uveitis in Northern California; the Northern California Epidemiology of Uveitis Study. Ophthalmology 2004; 111:491-500; discussion 500.
6. Rothova A, Suttorp-van Schulten MS, Frits Treffers W, Kijlstra A. Causes and frequency of blindness in patients with intraocular inflammatory disease. Br J Ophthalmol 1996; 80:332-6.
7. de Jong PT. Age-related macular degeneration. N Engl J Med 2006; 355:1474-85.
8. Wautier JL, Schmidt AM. Protein glycation: a firm link to endothelial cell dysfunction. Circ Res 2004; 95:233-8.
9. Parnaby-Price A, Stanford MR, Biggerstaff J, Howe L, Whiston RA, Marshall J, Wallace GR. Leukocyte trafficking in experimental autoimmune uveitis in vivo. J Leukoc Biol 1998; 64:434-40.
10. Xu H, Forrester JV, Liversidge J, Crane IJ. Leukocyte trafficking in experimental autoimmune uveitis: breakdown of blood-retinal barrier and upregulation of cellular adhesion molecules. Invest Ophthalmol Vis Sci 2003; 44:226-34.
11. Chi JT, Chang HY, Haraldsen G, Jahnsen FL, Troyanskaya OG, Chang DS, Wang Z, Rockson SG, van de Rijn M, Botstein D, Brown PO. Endothelial cell diversity revealed by global expression profiling. Proc Natl Acad Sci U S A 2003; 100:10623-8.
12. Holmen C, Stjarne P, Sumitran-Holgersson S. Heterogeneity of human nasal vascular and sinusoidal endothelial cells from the inferior turbinate. Am J Respir Cell Mol Biol 2005; 32:18-27.
13. Hendrickx J, Doggen K, Weinberg EO, Van Tongelen P, Fransen P, De Keulenaer GW. Molecular diversity of cardiac endothelial cells in vitro and in vivo. Physiol Genomics 2004; 19:198-206.
14. Smith JR, Choi D, Chipps TJ, Pan Y, Zamora DO, Davies MH, Babra B, Powers MR, Planck SR, Rosenbaum JT. Unique gene expression profiles of donor-matched human retinal and choroidal vascular endothelial cells. Invest Ophthalmol Vis Sci 2007; 48:2676-84.
15. Lampi KJ, Shih M, Ueda Y, Shearer TR, David LL. Lens proteomics: analysis of rat crystallin sequences and two-dimensional electrophoresis map. Invest Ophthalmol Vis Sci 2002; 43:216-24.
16. Kreil DP, Karp NA, Lilley KS. DNA microarray normalization methods can remove bias from differential protein expression analysis of 2D difference gel electrophoresis results. Bioinformatics 2004; 20:2026-34.
17. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001; 98:5116-21. Erratum in: Proc Natl Acad Sci U S A 2001 Aug 28; 98(18):10515.
18. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc, B 1995; 57:289-300.
19. Storey J. A direct approach to false discovery rates. J R Stat Soc, B 2002; 64:479-98.
20. Nivinskas H, Cole KD. Environmentally benign staining procedure for electrophoresis gels using Coomassie Brilliant Blue. Biotechniques 1996; 20:380-2,384-5.
21. Nesvizhskii AI, Keller A, Kolker E, Aebersold R. A statistical model for identifying proteins by tandem mass spectrometry. Anal Chem 2003; 75:4646-58.
22. Silverman MD, Zamora DO, Pan Y, Texeira PV, Planck SR, Rosenbaum JT. Cell adhesion molecule expression in cultured human iris endothelial cells. Invest Ophthalmol Vis Sci 2001; 42:2861-6.
23. Campochiaro PA. Ocular versus extraocular neovascularization: mirror images or vague resemblances. Invest Ophthalmol Vis Sci 2006; 47:462-74.
24. Premzl A, Turk V, Kos J. Intracellular proteolytic activity of cathepsin B is associated with capillary-like tube formation by endothelial cells in vitro. J Cell Biochem 2006; 97:1230-40.
25. Lakka SS, Gondi CS, Rao JS. Proteases and glioma angiogenesis. Brain Pathol 2005; 15:327-41.
26. Im E, Venkatakrishnan A, Kazlauskas A. Cathepsin B regulates the intrinsic angiogenic threshold of endothelial cells. Mol Biol Cell 2005; 16:3488-500.
27. Day IN. Molecular analysis of two general markers of neurons and the diffuse neuroendocrine system: neuron-specific enolase and PGP9.5. Biochem Soc Trans 1988; 16:454-7.
28. Bruneel A, Labas V, Mailloux A, Sharma S, Vinh J, Vaubourdolle M, Baudin B. Proteomic study of human umbilical vein endothelial cells in culture. Proteomics 2003; 3:714-23.
29. Luchansky SJ, Lansbury PT Jr, Stein RL. Substrate recognition and catalysis by UCH-L1. Biochemistry 2006; 45:14717-25.
30. Wang J, Maldonado MA. The ubiquitin-proteasome system and its role in inflammatory and autoimmune diseases. Cell Mol Immunol 2006; 3:255-61.
31. Bargagna-Mohan P, Ravindranath PP, Mohan R. Small molecule anti-angiogenic probes of the ubiquitin proteasome pathway: potential application to choroidal neovascularization. Invest Ophthalmol Vis Sci 2006; 47:4138-45.
32. Seliger B, Fedorushchenko A, Brenner W, Ackermann A, Atkins D, Hanash S, Lichtenfels R. Ubiquitin COOH-terminal hydrolase 1: a biomarker of renal cell carcinoma associated with enhanced tumor cell proliferation and migration. Clin Cancer Res 2007; 13:27-37.
33. Ke Q, Costa M. Hypoxia-inducible factor-1 (HIF-1). Mol Pharmacol 2006; 70:1469-80.
34. Liu H, Sadygov RG, Yates JR 3rd. A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal Chem 2004; 76:4193-201.
35. Chen YR, Juan HF, Huang HC, Huang HH, Lee YJ, Liao MY, Tseng CW, Lin LL, Chen JY, Wang MJ, Chen JH, Chen YJ. Quantitative proteomic and genomic profiling reveals metastasis-related protein expression patterns in gastric cancer cells. J Proteome Res 2006; 5:2727-42.
36. Li R, Wang H, Bekele BN, Yin Z, Caraway NP, Katz RL, Stass SA, Jiang F. Identification of putative oncogenes in lung adenocarcinoma by a comprehensive functional genomic approach. Oncogene 2006; 25:2628-35.
37. Wang D, Jensen R, Gendeh G, Williams K, Pallavicini MG. Proteome and transcriptome analysis of retinoic acid-induced differentiation of human acute promyelocytic leukemia cells, NB4. J Proteome Res 2004 May-Jun; 3:627-35.
38. Cox B, Kislinger T, Emili A. Integrating gene and protein expression data: pattern analysis and profile mining. Methods 2005; 35:303-14.
39. Wu CC, MacCoss MJ. Shotgun proteomics: tools for the analysis of complex biological systems. Curr Opin Mol Ther 2002; 4:242-50.
40. Ohno-Matsui K, Hirose A, Yamamoto S, Saikia J, Okamoto N, Gehlbach P, Duh EJ, Hackett S, Chang M, Bok D, Zack DJ, Campochiaro PA. Inducible expression of vascular endothelial growth factor in adult mice causes severe proliferative retinopathy and retinal detachment. Am J Pathol 2002; 160:711-9.
41. Lebherz C, Maguire AM, Auricchio A, Tang W, Aleman TS, Wei Z, Grant R, Cideciyan AV, Jacobson SG, Wilson JM, Bennett J. Nonhuman primate models for diabetic ocular neovascularization using AAV2-mediated overexpression of vascular endothelial growth factor. Diabetes 2005; 54:1141-9.
42. Ambati J, Anand A, Fernandez S, Sakurai E, Lynn BC, Kuziel WA, Rollins BJ, Ambati BK. An animal model of age-related macular degeneration in senescent Ccl-2- or Ccr-2-deficient mice. Nat Med 2003; 9:1390-7.