Molecular Vision 2014; 20:441-457
Received 04 September 2013 | Accepted 28 March 2014 | Published 30 March 2014
1Department of Anatomy-Histology and Embryology, School of Medicine, University of Patras, Rion, Greece; 2Department of Ophthalmology, School of Medicine, University of Patras, Rion, Greece; 3Department of Medical Physics, School of Medicine, University of Patras, Rion, Greece
Correspondence to: Martha Assimakopoulou, Department of Anatomy-Histology and Embryology, School of Medicine, University of Patras, Gr-26504 Rion, Greece; Phone: 302610969186; FAX: 302610969178; email: email@example.com
Purpose: Εnhanced expression of transcription factor hypoxia inducible factor HIF-1α is known to play a critical role in the modulation of cell metabolism and survival pathways as well as having stem-cell–like properties. Furthermore, accumulated data reveal the existence of cross-regulation between the oxygen-sensing and heat shock pathways contributing to the adaptation of cells under stressful conditions. Pterygium, a stem cell disorder with premalignant features, has been reported to demonstrate hypoxia. The purpose of this study was to investigate the co-expression patterns of transcription factor HIF-1α and von Hippel Lindau protein (pVHL)—which normally acts to keep levels of HIF-1α activity low under normoxic conditions—in pterygium and normal conjunctival human samples. Additionally, expression of HIF-1α compared to the activation of heat shock proteins (Hsp90, Hsp70, and Hsp27) was studied. Emphasis was placed on the detection of HIF-1α and Hsp90, which associates with and stabilizes HIF-1α to promote its transcriptional activity.
Methods: Semi-serial paraffin-embedded sections and tissue extracts from pterygium and normal conjunctival samples were studied by immunohistochemistry and western blot analysis, respectively, with the use of specific antibodies. Double labeling immunofluorescence studies on cryostat sections were also included.
Results: Statistically significant increased expression of HIF-1α and Hsps (Hsp90, Hsp70, and Hsp27) in pterygia compared to normal conjunctiva was demonstrated (p<0.05). In contrast, no significant difference was detected for pVHL expression (p>0.05). Immunohistochemical findings revealed nuclear HIF-1α immunoreactivity in all the epithelial layers of 23/32 (71.8%) pterygium tissues. Furthermore, all epithelial layers of the majority (75%) of pterygium samples showed strong cytoplasmic immunoreactivity for Hsp27 while Hsp27 expression was detected in all pterygia (100%) examined. Hsp27 expression was not observed in the superficial layer of goblet cells. In some samples, focal basal epithelial cells exhibited weak Hsp27 expression or were Hsp27 immunonegative. Ιmmunoreactivity of phopsho-Hsp27 showed the same distribution pattern as Hsp27 did. Epithelium of all pterygia (100%) displayed moderate to strong Hsp90 cytoplasmic immunoreactivity. Furthermore, the majority of pterygia, specifically, 30/32 (93.7%) and 27/32 (84.3%) demonstrated, respectively, Hsp70 and pVHL cytoplasmic immunoreactivity. Hsp90, Hsp70, and pVHL immunoreactivity was mainly detected in basal and suprabasal epithelial layers even though strong immunoreactivity in all epithelial layers was also observed in some pterygia. Stroma vessels were immunopositive for Hsps (Hsp90, Hsp70, and Hsp27) and pVHL. A statistically significant correlation between the expression of HIF-1α and the activation status of Hsps (Hsp90, Hsp70, and Hsp27; p<0.05) was observed whereas HIF-1α expression did not correlate with pVHL expression (p>0.05). Double labeling immunofluorescence studies showed nuclear HIF-1α co-localization with cytoplasmic Hsp90 expression in cells distributed in the entire epithelium of pterygia, in contrast to, normal conjunctiva, which exhibited only a few scattered epithelial cells with cytoplasmic HIF-1α expression and basal epithelial cells with Hsp90 expression.
Conclusions: The upregulation of coordinated activation of HIF-1α and Hsps in pterygium may represent an adaptive process for the survival of cells under stressful conditions. The significance of the association of HIF-1α with Hsp90 with respect to the therapeutic approach of pterygium requires further evaluation.
Pterygium, derived from the Greek word πτέρυξ, meaning “wing,” refers to a disorder of the ocular surface that occurs in people throughout the world, especially in countries with prolonged sunshine and a warm climate . The lesion appears as a triangular overgrowth of (in most cases) nasal but also (occasionally) the temporal bulbar conjunctiva on the cornea that may eventually obstruct the visual axis . Histopathological findings demonstrate epithelial changes such as squamous metaplasia and goblet cell hyperplasia  and elastotic degenerated connective tissue  with prominent neovascularization [5-7] accompanied by an inflammatory cell infiltration .
Although there is no consensus on the pathogenesis of pterygium [9,10], ultraviolet radiation is considered to be the principal environmental factor that affects epithelial stem cells residing at the nasal limbus [11,12]. According to a recent report, pterygium is likely a stem cell disorder with premalignant features . Although pterygium was traditionally recognized as a benign lesion, studies indicate that it is associated with cancers of the ocular surface such as squamous cell carcinoma and malignant melanoma [14-16]. Furthermore, fibroblasts isolated from human pterygia exhibit transformed cell characteristics showing a possible preneoplastic nature in the pterygium outgrowth . Epithelial-mesenchymal transition (EMT), a phenomenon that occurs during oncogenesis , has been proposed as a mechanism for the origin of pterygial fibroblasts . Additionally, pterygium is characterized by local invasiveness [12,20,21] and a high recurrence rate . Recent clinicopathological analysis implies that neovascularization of pterygia is triggered by ocular hypoxia . Additionally, EpoR, which is upregulated by ischemia or hypoxia, is increased in pterygium .
A hypoxic microenvironment activates critical signaling pathways in order for cells to overcome the limited availability of nutrients and oxygen. Hypoxia-inducible factors (HIFs) are transcription factors of the basic helix-loop-helix (bHLH)/PAS family, which plays a key role in adaptive cellular responses to hypoxia. HIFs include HIF-1α, which is expressed differently in most tissues, and HIF-2α, which shows a more restricted tissue expression pattern [25,26]. HIF-1α—the oxygen sensitive subunit of hypoxia-inducible factor-1 (HIF-1)—regulates the transcription of numerous target genes involved in energy metabolism, angiogenesis and apoptosis . In normoxic cells, this protein is expressed at extremely low levels due to degradation by the von Hippel Lindau protein (pVHL) , a tumor suppressor protein [29,30]. Under hypoxic conditions, the stabilization and activation of HIF-1α is the result of the inability of pVHL to associate with and ubiquinate HIF-1α . However, data show that Hsp90 associates with HIF-1α, resulting in stabilization of HIF-1α . It has been reported that Hsp90 inhibitors, such as geldanamycin and 17-allylaminogeldanamycin (17-AAG), block the binding of Hsp90 to HIF-1α, leading to increased ubiquitination and degradation of HIF-1α [33-35] in hypoxic conditions. Interestingly, the induction of Hsps is HIF dependent. Particularly, heat shock factor (HSF) transcription is upregulated during hypoxia due to direct binding of HIF-1 to HIF-1 response elements in an HSF intron [36,37].
Furthermore, overexpression of Hsps has been observed in response to a wide variety of stress conditions including heat shock , radiation , viral infection , oxidative stress , and inflammation as well as ischemia . Hsps are also induced at specific stages of development, differentiation , and malignant transformation . Hsps display strong cytoprotective effects by playing the role of molecular chaperones for numerous clients and signaling proteins involved in mechanisms promoting cell survival [45,46]. Hsp90, its client protein Hsp70, and Hsp27 are the dominantly expressed heat shock proteins after different kinds of stress [45,47]. The elevated levels of these proteins in a wide range of human cancers provide prognostic and predictive implications .
In this study, we examined the co-expression patterns of transcription factor HIF-1α and pVHL in pterygium and normal conjunctival human samples. In addition, the expression of HIF-1α compared to the expression levels of Ηsps (Hsp90, Hsp70, and Hsp27) was studied.
Fifty-four patients with primary or recurrent pterygia who underwent routine pterygium excision surgery in the Department of Ophthalmology, University Hospital Patras, Greece, from 2008 through 2010 were included in this study. Normal bulbar conjunctival tissues (n = 18, mean age 72) from patients undergoing glaucoma or cataract surgery were collected from the same department during the same period. Τhe study population was of Greek origin. Use of the human specimens was in accordance with the University Ethics Commission. All research protocols were conducted, and patients were treated in accordance with the tenets of the Declaration of Helsinki. Clinicodemographic features regarding the study population are presented in Table 1.
All tissues for immunohistochemistry were fixed in formalin and embedded in paraffin. Consecutive (semi-serial) 4 μm sections of ocular pterygium and normal conjunctival tissue samples were collected on poly-L-lysine coated slides. Pterygia were oriented such that sections were cut longitudinally through the head and the body of the pterygium. One section for each sample was stained with hematoxylin and eosin (H&E). For immunohistochemical studies, the histological sections were deparaffinized in xylene and rehydrated in graded alcohols up to water. Antigen retrieval was performed either by microwaving the slides in 0.01 M citrate buffer (pH 6) for antibodies against pVHL and Hsps (Hsp90, Hsp70, Hsp27, and phosphorylated Hsp27 [phospho-Hsp27 (Ser 82)]) or incubating them with pepsin at 37 °C for 15min for the antibody against HIF-1α. Endogenous peroxidase activity was quenched by treatment with 1% hydrogen peroxide for 20 min. Incubation with an appropriate protein blocking solution was performed. Sections were subsequently incubated with primary antibodies (Table 2). Detection was performed using the Envision Plus Detection System kit, according to the manufacturer’s instructions (DakoCytomation, DAKO, Carpinteria, CA), with 3,3′-diaminobenzidine (DAB) as a chromogen, which yielded brown reaction products. Sections were counterstained with Mayer’s hematoxylin solution, dehydrated, and mounted. To ensure antibody specificity, negative controls included the omission of primary antibody and substitution with non-immune serum. Control slides were invariably negative for immunostaining. As positive controls, cancer specimens known to express HIF-1α, pVHL, and Hsps were used. Optimal conditions (dilutions, incubation time, and protein blocking) for reproducible results with each primary antibody are shown in Table 2.
Two observers independently evaluated the results of immunohistochemistry in epithelium. Interobserver agreement for evaluation of immunostaining was within 15% (Cohen’s kappa = 0.82) . To determine the labeling index (% labeled cells) for each antibody, two observers independently assessed ten non-overlapping, random fields (X400 total magnification) for each case and manually counted 100 epithelial cells in each field with the aid of an ocular grid. Immunopositive endothelial and stromal cells were excluded from the cell counts. Expression of proteins included in this study was examined in adjacent (semi-serial) sections of each sample . Microphotographs were obtained using a Nikon DXM 1200C digital camera mounted on a Nikon Eclipse 80i microscope and ACT-1C software (Nikon Instruments Inc., Melville, NY).
For the double labeling immunofluorescence studies, pterygium and normal conjunctiva frozen samples were embedded in optimum cutting temperature (OCT) compound (BDH, Poole, UK) and cryosectioned at a thickness of 10μ. The sections were post fixed with 4% PFA for 15 min at RT, washed with PBS and 0.3% Triton-X 100 in PBS. After incubation for at least 1 h with blocking solution (0.1% Tween-20 in PBS, 10% fetal bovine serum, and 3% BSA [BSA, Sigma]), sections were labeled overnight at 4 °C with the following primary antibodies mixed together in blocking solution: HIF-1α and Hsp90. The same dilutions that applied in immunohistochemistry were used (Table 2). After being washed, sections were incubated at room temperature (RT) for 1 h with secondary antibodies Alexa Fluor 488 and 568 goat anti-mouse and anti-rabbit (1/500 and 1/500 respectively; Molecular Probes–Invitrogen, Paisley, UK). The specificity of the staining was determined by omitting the primary antibodies, and 4,6-diamidino-2-phenylindole (DAPI) was used to label cell nuclei. Slides were coverslipped using an aqueous-based mounting medium (Vectashield Hard Set; Vector Laboratories, Burlingame, CA). In addition, double immunofluorescence was performed on deparaffinized sections following the same procedure as described above. Photos were taken with a Nikon Eclipse TE2000-U microscope and collected with the Nikon camera Digital Sight DS-L1 (Nikon, Tokyo, Japan). All images were processed with Adobe Photoshop software (Adobe Systems, San Jose, CA).
After the clinical samples were resected, they were placed immediately on dry ice and stored at −80 °C. Before western blotting, tissue sections were cut onto glass slides and stained following the routine staining protocol. The remainder of the sample was directly transferred into a reaction tube followed by solubilization of the tissue in 100 μl 1% SDS w/v with 4 μl Sigma Phosphatase Inhibitor Cocktail I, 4 μl Sigma Phosphatase Inhibitor Cocktail II, and 4 μl Sigma Protease Inhibitor Cocktail, sonicated, and boiled for 15 min. Protein concentration was measured for each sample by with a BSA protein kit (Pierce, Rockford, IL) and spectrophotometry. Duplicate samples (50 μg of total protein) were separated on 7% (for HIF-1α and Hsp90) and 12% (for pVHL, Hsp70, Hsp27, and phospho-Hsp27) polyacrylamide gel and then transferred on nitrocellulose membranes. Non-specific binding sites were blocked for 1 h with 10% nonfat dried milk at RT. The blots were probed with the primary antibodies at appropriate dilutions (Table 2) and incubated overnight at 4°C. The blots were rinsed with Tris buffered saline (TBS)-Tween and then incubated with goat anti-rabbit horseradish peroxidase-linked immunoglobulin G (IgG) or goat anti-mouse horseradish peroxidase-linked IgG for 1 h at RT. After rinsing, immunoreactive bands were visualized with an enhanced chemiluminescence detection system (ECL, Amersham International, Cardiff, UK). Molecular masses were determined by comparison with protein molecular weight marker standards from Biomol. To ensure equal protein loading, membranes were stripped and reprobed with β-tubulin monoclonal antibody (1:2,000 dilution; Sigma, St. Louis, MO). The blots obtained from at least three independent experiments were analyzed. Luminescence from the blots was detected by exposing the membranes to Fuji-Hyperfilm for 1 to 10 min, to ensure operation within the linear range of the film, followed by digital scanning of the developed film in transparency mode. The scanned image of the membranes and band intensities were calibrated and quantified using NIH ImageJ software (version 1.61).
Parametric statistical tests depended on the assumption that data are sampled from a normal (Gaussian) distribution. Testing composite normality was performed using the Shapiro-Wilk W test (along with the Shapiro-Francia correction for leptokurtic samples), when data were not large-sized (<2000). In case of scale data, mean comparisons were performed using the Student t test (for bipartite comparisons) and one-way ANOVA (for multipartite comparisons). Otherwise, median comparisons were performed with non-parametric Wilcoxon’s Rank-Sum test (equivalent to the Mann-Whitney U test) and the Kruskal-Wallis test, respectively. Associations of categorical (nominal or ordinal) data were examined using Pearson’s Chi-Square test of independence (or the z-test for bipartite proportion comparisons). Correlation analysis of parametric data was performed using Pearson’s linear correlation, while the analysis of non-parametric data was performed using Kendall’s τ (or Spearman’s ρ) rank correlation. In either case, a line to fit data (in a least-squares sense) was generated. In all statistical tests, the significance level was defined as p<0.05. Statistical analysis was implemented in MATLAB, release 7.12 (R2011a), using the Statistics Toolbox.
Immunohistochemical analysis showed expression of HIF-1α, pVHL, and Hsps with distinct patterns of cellular localization in the epithelium of pterygium and normal conjunctival samples included in this study. Furthermore, in the pterygium stroma, vascular endothelial cells showed cytoplasmic immunoreactivity for Hsps (Hsp90, Hsp70, and Hsp27) and pVHL. The immunohistochemical data are shown in Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, and Figure 6 and are summarized in Table 3. Nuclear immunoreactivity for HIF-1α was detectable in all the epithelial layers of 23/32 (71.8%) pterygium tissues (Figure 1A). In some cases, cytoplasmic immunoreactivity was also noted (Figure 1B,C,E). Only the nuclear immunoreactivity was taken into consideration during the evaluation of staining. Hsp27 expression was detected in all the pterygia (100%) examined. Furthermore, the majority (75%) of pterygium samples exhibited strong cytoplasmic immunoreactivity for Hsp27 in all epithelial layers (Figure 2A-C). In some samples (about 1%), focal basal cells exhibited weak Hsp27 expression or were Hsp27-negative (Figure 2C). Hsp27 expression was not observed in the superficial layer of goblet cells (Figure 3A). Ιmmunoreactivity of phospho-Hsp27 showed the same distribution pattern as Hsp27 did. Additionally, all pterygia (100%) showed moderate to strong cytoplasmic immunoreactivity for Hsp90, mainly in basal and suprabasal cells (Figure 3B) although Hsp90 immunoreactivity was detected occasionally in almost all epithelial layers (Figure 4A-C). Furthermore, basal and suprabasal cells displayed moderate cytoplasmic immunoreactivity for Hsp70 in 30/32 (93.7%) pterygia (Figure 3C, Figure 5A,B). In some samples, however, Hsp70 immunoreactivity was observed in the entire epithelium (Figure 5C,D). Occasionally, strong immunoexpression of Hsp70 in basal cells was detected (Figure 5D). Weak pVHL immunoreactivity was detected in 27/32 (84.3%) of pterygia whereas moderate to strong pVHL immunoreactivity, particularly in basal and suprabasal cells, was also noted (Figure 6A,B). In some samples, pVHL-expressing cells with strong immunoreactivity were distributed in all epithelial layers (Figure 3D, Figure 6C). In the pterygia included in this study, there was considerable staining heterogeneity among different samples concerning the HIF-1α, Hsps, and pVHL immunostaining. Furthermore, variability of positive cells was also common within individual tissues.
Quantitative analyses of the immunohistochemical labeling indices (LIs) showed no significant differences based on age, gender, or grade of pterygium (p>0.05). However, there were significant correlations between the LIs for HIF-1α and all three Hsps (Figure 7A-C) but not between HIF-1α and pVHL (Figure 7F). Furthermore, there were significant correlations between the LIs for pVHL and Hsp90, as well as, between pVHL and Hsp70 (Figure 7D,E) but not between pVHL and Hsp27 (p = 0.06). The expression of each Hsp was significantly correlated with the other Hsps (Figure 8).
In normal conjunctival epithelium, a small number of epithelial conjunctival cells showed nuclear HIF-1α immunoreactivity although in some samples both nuclear and cytoplasmic staining were observed. Intense cytoplasmic immunoreactivity was observed for Hsp27 mainly in basal and suprabasal layers of all samples. Furthermore, Hsp90, Hsp70, and pVHL expression was observed in basal and suprabasal epithelial cells that showed weaker immunoreactivity compared to the pterygia (Figure 1, Figure 2, Figure 4, Figure 5, and Figure 6). Considerable staining heterogeneity of the Hsps and pVHL was observed among and within conjunctival samples. Comparison of mean labeling indices (MLIs) between primary and recurrent pterygia showed a significantly higher expression of HIF-1α, Hsp27, and Hsp90 in primary pterygia (Table 4).
Immunoblotting experiments on tissue extracts from pterygium and normal conjunctival samples were performed to confirm the expression of HIF-1α, pVHL, and Hsps as detected with immunohistochemistry. These experiments clearly demonstrated the presence of the expected 27 kDa band of the Hsp27 protein and also the expected 27-KDa size for its phosphorylated isoform phospho-Hsp27 in pterygia and normal conjunctiva. Similarly, expected bands of 70 kDa, and 90 kDa were found for the expression of Hsp70 and Hsp90, respectively. Immunoblots for HIF-1α exhibited a specific band at 120 kDa and multiple bands of weaker intensity at 100–120 kDa, which represent post-translational modifications of HIF-1α. No signal for HIF-1α was detected in normal conjunctival samples. Finally, for the pVHL protein, the main band was detected around 30-KDa and one additional band of weaker intensity at 14-kDa was also noted (not shown). The pVHL immunoreactive band was hardly detectable in pterygia, and similar pVHL immunoreactive bands were observed in normal conjunctival samples. A representative western blot is depicted in Figure 9A. Blot quantification showed a significant increase in expression for HIF-1α and Hsps but not for pVHL in pterygia (Figure 9B).
Double immunofluorescence staining in pterygia revealed that nuclear HIF-1α expression was co-localized with cytoplasmic Hsp90 expression in cells distributed in the entire width of the epithelium (Figure 10A). In contrast, in samples where HIF-1α was detected in the cytoplasm, this expression was not co-localized with Hsp90 staining (Figure 10B); specifically, HIF-1α cytoplasmic staining was mainly observed in superficial epithelial cells whereas Hsp90 expression was limited to the basal and suprabasal epithelial cells (Figure 10C). In normal conjunctiva, double immunofluorescence showed Hsp90 expression mainly in the basal epithelial cells (Figures 10D-F) whereas HIF-1α immunoreactivity was detected in the cytoplasm of a few epithelial cells (Figure 10E,F).
In this study, we investigated the co-expression patterns of HIF-1α and pVHL as well as Hsps (Hsp90, Hsp70, and Hsp27) in pterygium and normal conjunctival human samples. Accumulating experimental data reveal that HIFs (HIF-1α and HIF-2α) and their target genes are master regulators of selective adaptation of cells to oxygen and nutrient deprivation [25-27]. Furthermore, the activated HIFs regulate the transcription of genes that may play a critical role in the modulation of the stemness properties of cells . In this study we used immunohistochemistry to identify the enhanced expression of HIF-1α in a high percentage of pterygial epithelial cells compared to normal conjunctiva (Table 3). The localization of HIF-1α was mainly nuclear even though cytoplasmic immunoreactivity was also detected in a proportion of epithelial cells (Figure 1). Western blot analysis demonstrated HIF-1α expression in pterygia samples but not in normal conjunctiva (Figure 9B). These findings suggest that HIF-1α plays a role in the pathogenesis and growth of the pterygium lesion.
Previous studies in pterygium suggest that ocular hypoxia may exist [5,23]. As is known, hypoxia may activate the nuclear translocation of HIF-1α, resulting in its upregulation and activation , but overexpressed HIF-1α is also constitutively localized to the nucleus under normoxic conditions . According to the Groulx and Lee model , nuclear translocation itself seems to be constitutive, independent of oxygen availability or pVHL status in cells infected with an adenovirus expressing GFP fused to the HIF-1α subunit. It has been proposed that the upregulation of HIF activity, even in normoxia, may be differently regulated through the sustained stimulation of growth factor and cytokine pathways [55-60]. In pterygium, ultraviolet irradiation causes the production of a variety of growth factors (e.g., EGF, TGF-beta, bFGF, HB-EGF, and IGFBP-2) . These data indicate that the activation of HIF-1α in pterygium could not only be the result of hypoxia but also of hypoxia-independent mechanisms, such as oncogene activation and growth factor signaling pathways.
Additionally, HIF-1α can be activated under normoxia by loss of the von Hippel-Lindau tumor suppressor protein (which normally acts to keep levels of HIF-1α activity low)  or by the lack of interaction with proteins that promote its ubiquination—even if these are overexpressed—and, therefore, the nuclear translocation of HIF-1α can protect it from degradation . Indeed, recent data in pterygium indicate that SAT1— a highly regulated rate-enzyme in polyamine metabolism—which is involved in the oxygen-independent degradation of HIF-1α, is abundantly expressed . In the current study, we investigated the expression patterns of HIF-1α in relation to pVHL.
Previous data have shown that HIF-1α degradation can occur in both compartments in some specific cell types, such as primary endothelial cells (e.g., from mouse brain) , and that pVHL may be found in both the cytoplasm and nucleus [54,64,65]. In this study, while distinct cytoplasmic localization of pVHL was detected in pterygial epithelial cells (Figure 6), there was no significant correlation between HIF-1α and pVHL immunopositive cells (Figure 7F). Similarly, there was no significant difference in pVHL expression between pterygium and conjunctival samples as detected by immunohistochemistry (Table 3) and western blot quantification (Figure 9B). These results show that pVHL does not play a role in the degradation of HIF-1α in pterygium, indicating another possible mechanism for HIF-1α upregulation in pterygial epithelial cells.
As was mentioned, pVHL was consistently detected in both pterygium and conjunctival epithelial cells. This finding is in accordance with previous results that demonstrate a wide expression of pVHL in normal human tissues [66,67], especially in the cytoplasm of epithelial cells covering the body surface and the alimentary, respiratory, and genitourinary tracts . Remarkably, pVHL strong cytoplasmic immunoreactivity was demonstrated in vessels of pterygia but not in normal conjunctiva (Figure 6). Studies of tumors have also shown expression of pVHL in endothelial cells, fibroblasts, and pericytes . Thus, pVHL may be implicated in mechanisms of angiogenesis in pterygium.
Previous studies have reported the existence of cross-regulation between the oxygen-sensing and heat shock pathways [36,37]. Overexpression of Hsps has been shown to demonstrate their dual roles as regulators of protein conformation as well as anti-apoptotic mediators [46,47]. Immunohistochemical data have demonstrated expression of Hsp90, Hsp70, and Hsp27 in normal conjunctival epithelium . In the current study, western blot quantification revealed significantly increased protein levels of Hsp90, Hsp70, and Hsp27 (including phospho-Hsp27) in pterygia compared to normal conjunctival samples (Figure 9B). Immunohistochemically, a significantly higher percentage of immunopositive epithelial pterygial cells for Hsp90 and Hsp27 was detected in pterygia whereas the expression levels of Hsp70 did not differentiate between normal conjunctival and pterygium epithelium (Table 3). A recent study by Sebastiá et al.  also reported highly statistically significant differences in Hsp90 positive cells between normal conjunctiva and pterygium epithelium.
Hsp90 and Hsp70 were mainly detected in basal and suprabasal layers of epithelium (Figure 4, Figure 5) whereas Hsp27 was distributed in all epithelial layers of pterygia (Figure 2). Interestingly, in some pterygium samples basal epithelial cells were Hsp27-immunonegative (Figure 2C), indicating association of this protein with cell differentiation. In addition, Hsp27 has been reported to be a predifferentiation marker [71,72]. However, differences between various epithelial cell layers of pterygium have been also previously detected, with p53 expression being higher in basal cells compared to more superficial layers .
Not surprisingly, analysis of Hsp expression patterns revealed that each Hsp was significantly correlated with the others (Figure 8). Furthermore, Hsp90 and Hsp70 were significantly correlated with the expression of pVHL in pterygial epithelial cells (Figure 7D,E). Interestingly, expression of Hsps (Hsp90, Hsp70, and Hsp27) was also detected in endothelial cells in pterygia but not in normal conjunctiva, indicating a possible role of these proteins in angiogenesis of pterygium. These findings suggest that, in pterygium, finely tuned interactions between Hsps and their co-chaperones and client proteins mediate the cytoprotective actions of Hsps against stress. However, as it has been proposed for tumors, it is possible that the accumulation of Hsps may contribute to many of the traits that permit the malignant phenotype . Notably, it has recently been shown that Hsp90 plays a critical role in the pathogenesis of conjunctival melanoma .
In this study, an analysis of HIF-1α and Hsp expression patterns was performed. It is known that HIF-1α can bind to the promoters of all Hsp genes in response to stress . Α strong relationship between HIF-1α expression and Ηsps (Hsp90, Hsp70, and Hsp27) expression levels was found (Figure 7A-C), indicating that there may be cross-regulation between HIF-1α and HSF in pterygia as a result of the effort of the cells to protect themselves under stressful conditions. Functional studies will clarify whether the upregulation of the Hsps in pterygia is derived exclusively from the HIF-1α activation.
Hsp90, co-operating with Hsp70, appears to be more specialized to capture client proteins involved in cell signaling, such as transcription factors and protein kinases . Previous data show that Hsp90, acting as a molecular chaperon, stabilizes the HIF-1α in renal carcinoma cell lines while disruption of HIF-1α/Hsp90 promotes the ubiquitination and proteasome-mediated degradation of HIF-1α in an oxygen- and pVHL-independent manner and diminishes HIF-1α transcriptional activity .
In this study, we used double immunofluorescence staining to show co-localization of HIF-1α and Hsp90 in pterygial epithelial cells. Specifically, nuclear HIF-1α was co-localized with Hsp90 cytoplasmic expression. On the contrary, cytoplasmic HIF-1α expression was not co-localized with Hsp90 expression (Figure 10). These findings lend support to the hypothesis that Hsp90 plays a critical role in the stabilization of HIF-1α and its subsequently nuclear translocation in pterygium. As a result, the transcriptional activity of HIF-1α is strongly promoted.
Studies have shown that primary and recurrent pterygia are quite different. In particular, the morphology of recurrent pterygium is different, and its prognosis is worse. Tong et al.  suggested that recurrence is a distinct biologic phenomenon from the formation of primary pterygium. According to their findings, the recurrence of pterygium may be related to an imbalance of growth signals rather than due to mere prolongation of the stimulus that initiated the primary pterygia formation. Indeed, elevated expression of HIF-1α, pVHL, and Hsps was observed in recurrent pterygia compared to normal conjunctiva (Table 4), confirming the view of Tong et al.  regarding the formation of recurrent pterygium. Moreover, Tong et al.  hypothesized that a low level of inflammation plays a critical role in recurrence of a pterygium. Considering the accumulating data implicating HIF in the regulation of immunity and inflammation  as well as data reporting the induction of Hsps because of inflammation , it is logical to speculate that HIF-1α and Hsps may be involved in the pathogenesis of recurrent pterygium.
Notably, Tong et al.  reported that genes with increased expression are not the same in both primary and recurrent pterygium. In contrast, we immunohistochemically demonstrated that the significantly increased genes in recurrent pterygium are the same as those that increased in primary pterygium epithelial cells compared to normal conjunctiva. Specifically, significantly increased expression for HIF-1α, Hsp90, and Hsp27 was detected in primary compared to recurrent pterygium epithelia. These findings enhance the importance of the role of cross-regulation of HIF-1α and Hsps in the development of pterygium and highlight the hypothesis of Tong et al.  that the differentiation in protein expression is likely associated with differences in pathogenesis between the two pterygium types. Thus, HIF-1α and Hsps appear to be implicated in the formation not only of primary but also of recurrent pterygium, probably through different signaling pathways.
In summary, the results of the current study suggest that, in pterygium, HIF-1α upregulation is independent from pVHL expression but dependent on hypoxia and/or hypoxia-independent mechanisms. Furthermore, since the most prevalent hypothesis for the pathogenesis of pterygium is that focal limbal irradiation of corneal epithelial stem cells causes alteration of these cells and a breakdown of the limbal barrier [9,11], the upregulation of activated HIF-1α in pterygium may represent an adaptive process for the survival of stem/progenitor as well as non-stem cells under stressful conditions. The complete eradication of these cells, by targeting the HIF signaling network, might be beneficial in the therapeutic approach to pterygia. Further trials will evaluate the role of HIF-1α and Hsps coordinated activation in pterygium pathogenesis and treatment.
The authors kindly thank Professor F. Angelatou, PhD, Department of Physiology, School of Medicine, University of Patras, for her support in this study, and Assoc. Professor S. Taraviras, PhD, Department of Physiology, School of Medicine, University of Patras, for providing the immunofluorescence microscope. The technical contributions by K. Botsakis, PhD, and M. Spella, PhD, Department of Physiology, School of Medicine, University of Patras, in the Western and Immunofluorescence analysis, respectively, are gratefully acknowledged.