Prolyl 3-Hydroxylase 2 Supports a Pro-Angiogenic Milieu Promoting Colorectal Cancer Progression and Metastasis

Abstract

Prolyl 3-hydroxylase 2 (P3H2) is a key enzyme involved in the architecture of the extracellular matrix (ECM). While previously shown to be regulated by VEGF-A and to play a role in angiogenesis, its function in cancer remains ambiguous. While characterized as a tumor suppressor, its precise function in colorectal cancer (CRC) progression is poorly defined. Bioinformatic analysis and patient data reveal that P3H2 transcript levels are significantly reduced in colon adenocarcinoma tissues, showing a progressive decline in metastatic lesions. Furthermore, VEGF-A exposure upregulates P3H2 transcripts in the HCT116 CRC cell line. To investigate its impact in CRC, we generated a stable HCT116 clone overexpressing P3H2. In vitro studies demonstrated that while P3H2 overexpression inhibited anchorage-independent growth, it significantly enhanced cellular invasion without altering cell proliferation. In vivo, however, P3H2-overexpressing tumors exhibited accelerated tumor growth and a statistically significant increase in lung metastases. P3H2 overexpression remodeled the tumor microenvironment (TME) by modifying its main substrate, Collagen IV, resulting in the induction of increased vessels density. Our study repositions P3H2 as a dynamic enzymatic switch within the TME. This work identifies P3H2-driven ECM remodeling as a promising therapeutic axis in advanced CRC, with particular relevance for combination strategies targeting angiogenesis.

1. Introduction

Prolyl 3-hydroxylase 2 (P3H2, also known as LEPREL1) [1] belongs to the family of prolyl 3-hydroxylases which includes P3H1, P3H2, P3H3, P3H4, and CRTAP, which catalyzes the post-translational hydroxylation of proline residues in collagens [2]. This modification is critical for the proper folding and stability of the collagen triple helix, thereby influencing the mechanical properties of the extracellular matrix (ECM) and basement membrane (BM) [3,4].

The role of P3H2 in cancer is complex and appears to be context dependent. In many cancers, P3H2 exhibits tumor-suppressive properties. Its expression is frequently downregulated via aberrant methylation in estrogen-receptor-positive breast cancer [5], B-cell lymphomas regardless of subtype [6], and in primary and metastatic melanoma [7]. In these contexts, the ectopic restoration of P3H2 expression inhibits tumor cell proliferation. Similarly, P3H2 is downregulated in the hepatocellular carcinoma tissues, through downregulation of cyclins involved in the cell cycle [8], and in osteosarcoma, where its expression shows a potent inhibitory effect on cell proliferation, migration, invasion, and epithelial–mesenchymal transition primarily mediated through the suppression of the oncogenic AKT/mTOR signaling pathway [9]. Furthermore, P3H2 displays a high frequency of deleterious mutations across various tumors, particularly in cutaneous melanoma, reinforcing its potential role as a tumor suppressor [10].

However, this tumor-suppressive function stands in apparent contrast to its recently identified pro-angiogenic role. Angiogenesis, a hallmark of cancer, is heavily influenced by the tumor microenvironment. Collagen IV, the main structural component of the basement membrane and a primary substrate for P3H2 with hydroxylation of 10–15 Pro residues every 1000 amino acids [11], is intimately linked to blood vessel formation [12,13]. We have recently demonstrated that P3H2 is upregulated by the VEGF-A/VEGFR-2 signaling pathway in endothelial cells, where it promotes migration and capillary formation. This pro-angiogenic activity is associated with P3H2-mediated remodeling of Collagen IV into more condensed bundles. In vivo, P3H2 knockdown prevents pathological angiogenesis in the model of laser-induced choroid neovascularization [14].

Based on these divergent findings, we sought to investigate the net effect of P3H2 modulation in colorectal cancer (CRC), a malignancy where angiogenesis is a critical driver of progression.

Here, we confirmed the downregulation of P3H2 expression in CRC patients and cell lines, consistent with observations previously reported in other tumors. Having established that VEGF-A also regulates P3H2 in CRC cells, we analyzed the functional impact of P3H2 overexpression. Using a stable P3H2-overexpressing HCT116 clone, we demonstrated that while P3H2 exerts opposing effects in vitro, suppressing anchorage-independent growth yet enhancing invasiveness, it potently boosts primary tumor growth, metastasis, and neo-angiogenesis in vivo.

2. Results

2.1. Loss of P3H2 Expression Is Associated with CRC Progression

To initially investigate the clinical relevance of P3H2 in CRC, we analyzed P3H2 mRNA levels using the Gene Expression Profiling Interactive Analysis (GEPIA) tool, which integrates data from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases. The analysis revealed that P3H2 expression was significantly lower (p < 0.01) in colon adenocarcinoma (COAD) samples than in normal colon tissue (p < 0.01; Figure 1a), suggesting that P3H2 downregulation is a recurring molecular event in CRC tumorigenesis. We next assessed the prognostic value of P3H2 by performing a Kaplan–Meier survival analysis. Patients with higher P3H2 transcript expression showed a trend toward longer survival compared to those with lower levels; however, this trend did not reach statistical significance (Figure 1b).

In line with the GEPIA data, qRT-PCR analysis of patient-derived CRC samples confirmed that P3H2 transcript levels are progressively reduced during cancer progression. Expression was lowest in metastatic lesions (M) compared to both primary tumors (T) and non-tumor tissues (NT), with a statistically significant difference between M and NT (p = 0.0001) (Figure 1c). Together, these results suggest that loss of P3H2 expression is a frequent and progressive event during CRC advancement, potentially contributing to metastatic evolution.

2.2. P3H2 Expression in Human CRC Cell Lines and Its Regulation by VEGF-A in HCT116

To move from patient data to an in vitro model, we investigated basal P3H2 expression across a panel of human CRC cell lines, using non-tumorigenic colon epithelial cells as a control. P3H2 protein expression was highly heterogeneous: it was significantly downregulated in DLD1 and HCT116 cell lines and nearly absent in GEO cells, relative to the non-tumor control cell line (CRL-1790), which showed the highest expression level (Figure 2a). This pattern is consistent with the trend observed in CRC patient samples. We selected the HCT116 cell line, which retains moderate basal P3H2 expression, to generate stable P3H2-overexpressing cell line.

Given our previous findings identifying P3H2 as one of the most upregulated genes in endothelial cells upon VEGF-A stimulation [14], we investigated this regulatory across CRC cell lines. After VEGF-A stimulation qRT-PCR analysis revealed a significative increase in P3H2 transcript levels in CRC cell lines analyzed (Figure 2b). THBD (Thrombomodulin) transcript levels were also measured and served as an established positive control for VEGF-A signaling activation.

Collectively, these data establish the HCT116 cell line as a suitable in vitro model for dissecting the link between pro-angiogenic stimuli and P3H2 function in CRC.

2.3. Functional Characterization of P3H2-Overexpressing Stable Cell Line

To functionally characterize the role of P3H2 in CRC pathogenesis, we generated stable HCT116 clones over-expressing P3H2. The overexpression of P3H2 was achieved transfecting HCT116 with an expression plasmid coding P3H2 cDNA (pSF-P3H2). Successful overexpression was verified in terms of P3H2 protein abundance by Western blot analysis. As shown in Figure 3a, pSF-P3H2 stable clone showed a 2.1-fold increase in P3H2 protein, as compared to the parental cell and control clone generated by transfection with a control plasmid carrying the cDNA of Firefly Luciferase (pSF-FLuc).

To assess tumorigenic properties of stable clones we performed proliferation, invasion, and anchorage-independent growth assays. P3H2 overexpression did not significantly alter the proliferation rate over 72 h when compared to both parental and control cells (Figure 3b). However, further analysis revealed a complex, dual role for P3H2 in malignancy (Figure 4). While P3H2-overexpressing clones exhibited a marked and significative decrease in anchorage-independent colony formation (Figure 4a), they displayed a significative fourfold increase in invasive capacity relative to both parental and vector controls cells (Figure 4b).

These findings indicate that P3H2 exerts opposing effects on tumor cell behavior: it suppresses growth under non-adherent conditions while potently enhancing invasiveness. This suggests a context-dependent, multifaceted role for P3H2 in colorectal cancer progression.

2.4. P3H2 Drives Tumor Growth and Metastasis In Vivo

To reconcile these contrasting findings, we assessed the tumorigenic potential of the pSF-P3H2 cells using a subcutaneous xenograft mouse model. Contrary to the anti-growth effect observed in vitro, pSF-P3H2 cells grow more rapidly generating tumors that reached significantly larger volumes over 16 days compared to both parental HCT116 and pSF-FLuc cells (p = 0.0005 versus HCT116 and p = 0.001 versus pSF-FLuc) (Figure 5a). This suggests that factors within the in vivo microenvironment override its growth-suppressive effects observed in the anchorage-independent assay.

Given the significant in vitro increase in invasion (Figure 4b), we next examined metastatic potential using an experimental lung metastasis model. Following tail-vein delivery, mice injected with pSF-P3H2 cells developed a statistically significant increase in the number of visible lung surface metastases after nine weeks compared to control groups (Figure 5c). Histological analysis of lung sections with hematoxylin and eosin staining confirmed a greater burden of metastatic nodules in the pSF-P3H2 group when compared to the control groups (Figure 5b).

Collectively, these in vivo results demonstrate that the pro-invasive and pro-angiogenic activities of P3H2 ultimately dominate, enhancing both primary tumor growth and metastasis dissemination.

2.5. P3H2 Drives Enhanced Vascularization

To investigate the mechanism underlying the increased tumor growth and, based on our previous demonstration that P3H2 is a molecular player in angiogenesis [14], we assessed tumor vascularization and the surrounding matrix.

We analyzed vessel density by immunohistochemical staining for the endothelial marker CD31. This revealed a significative increase in vascular density in pSF-P3H2 tumors compared to HCT116 and pSF-Fluc tumors (Figure 6), providing a potential explanation for the increased tumor volume, as a richer blood supply would support accelerated nutrient and oxygen delivery to the tumor mass.

Being Collagen IV the main substrates of P3H2 and a major structural component of vascular basement membranes, we examined its deposition around the vessels. Immunofluorescence co-staining for CD31 and Collagen IV showed a pronounced accumulation of Collagen IV surrounding CD31-positive vessels in pSF-P3H2 tumors, which was less marked in the control groups (Figure 7). This suggests that P3H2 overexpression not only promotes angiogenesis but may also influence stabilization of newly formed vessels. Indeed, the Collagen IV-enriched vascular bed indicates that P3H2 activity contributes to establishing a more robust structural framework for endothelial cells, thereby supporting angiogenesis.

Collectively, these findings suggest that P3H2 overexpression drives a pro-angiogenic phenotype characterized by both increased vessel density and enhanced matrix deposition. The elevated perivascular Collagen IV could contribute to improved vascular stability, which may explain the growth advantage observed in vivo by potentially overriding intrinsic anti-growth signals.

3. Discussion

CRC remains a leading cause of cancer-related mortality worldwide. Its progression from a localized adenoma to invasive carcinoma and, ultimately, to metastatic disease is a multistep process orchestrated by accumulating genetic alterations and a dynamic crosstalk with the surrounding TME. A pivotal hallmark of this progression is the induction of angiogenesis—the formation of new blood vessels from pre-existing vasculature. Tumors cannot grow beyond a minimal size or successfully metastasize without developing an adequate blood supply to deliver oxygen and nutrients. This angiogenic switch is driven by an imbalance between pro- and anti-angiogenic factors, with VEGF-A being a central regulator. Consequently, anti-angiogenic drugs have established role in the management of metastatic CRC, underscoring the critical need to understand the molecular regulators of angiogenesis within the CRC TME is crucial for developing novel therapeutic strategies [15].

Within this context, our study identifies P3H2 as a gene with a complex, dual role in colorectal cancer. It functions as a context-dependent tumor suppressor while simultaneously acting as a potent, previously unrecognized modulator of tumor angiogenesis. Consistent with reports in other malignancies, we observed a significant downregulation of P3H2 expression that correlates with advancing disease stages in CRC patient samples. This pattern of frequent loss was also reflected in a panel of established CRC cell lines, suggesting a selective pressure against P3H2 expression during tumor evolution, akin to a classical tumor suppressor. Interestingly, P3H2 expression increased in HCT116 CRC cells following VEGF-A exposure—a regulatory response we also observed in endothelial cells [14]—indicating that this mechanism is conserved across different cellular contexts.

To investigate its function, we generated a stable P3H2-overexpressing clone. Subsequent in vitro analyses revealed a striking functional dichotomy. Unlike previous reports in other cancers [7], P3H2 overexpression did not alter proliferation in HCT116 cells. However, it strongly suppressed anchorage-independent growth in soft agar assays (Figure 4a), consistent with a tumor suppressor role. Paradoxically, P3H2 also acted as a potent promoter of cellular invasiveness (Figure 4b).

This contradiction was resolved through in vivo investigations. Xenograft models demonstrated that P3H2 expression significantly enhanced primary tumor growth and, critically, promoted metastatic colonization (Figure 5). This shift in phenotype from in vitro suppression to in vivo promotion suggests that P3H2’s primary oncogenic role is not mediated through cell-intrinsic effects on proliferation. Instead, its function is dictated by its enzymatic activity within the TME. We propose that P3H2 expression is induced by TME-derived signals such as VEGF-A and, in turn, directly contributes to a pro-tumorigenic remodeling of the ECM.

Collagen IV, the main substrate of P3H2 catalytic activity, is the most abundant component of the BM, a multifunctional support that mediates several biological processes including angiogenesis [16,17]. The pro-angiogenic activity of Collagen IV is due to a direct modulation of endothelial cells functions, such as proliferation [18] adhesion and migration [12,19]. Concurrently, the endothelial cell capability to synthetize and deposit Collagen IV is indispensable for vascular survival and maturation in vivo [13,20,21]. Indeed, in the model of capillary sprouting from aortic ring cultured in presence of different concentrations of Collagen IV, a dose-dependent effect on neovessels elongation and survival was observed. At high concentration, Collagen IV was able to stabilize the neovascular outgrowths preventing vascular regression [13,21].

The increased Collagen IV deposition we observed in P3H2-overexpressing tumors (Figure 6) likely facilitates angiogenesis (Figure 7) and stabilizes nascent vessels. This remodeling of the extracellular matrix (ECM) would consequently promote tumor growth and enhance metastatic potential. While our data strongly support a model in which P3H2 drives angiogenesis through Collagen IV remodeling, a key future direction will be to definitively link this function to its enzymatic activity. This could be achieved by comparing the effects of wild-type P3H2 to a catalytically inactive mutant in rescue experiments and by employing pharmacological inhibitors of collagen prolyl hydroxylase activity in our in vivo models. Such studies would establish a direct causal relationship between P3H2-mediated prolyl hydroxylation, Collagen IV matrix assembly, and the resultant pro-tumorigenic vasculature.

In conclusion, our study redefines the role of P3H2 in colorectal cancer. While its frequent downregulation in advanced tumors aligns with a tumor suppressor profile, our findings reveal a more dynamic, context-dependent function. We propose that residual or reactivated P3H2 expression, particularly in response to VEGF-A-rich TME signals, can drive critical pro-tumoral steps. By catalyzing the hydroxylation and facilitating the proper deposition/stability of Collagen IV, P3H2 acts as a master modifier of the ECM landscape, fostering an angiogenic niche that supports tumor growth and metastatic dissemination.

Thus, P3H2 transcends the simple classification of a classical tumor suppressor. It emerges as a dynamic enzymatic switch within the TME, whose function pivots on contextual signals to shape the ECM and regulate angiogenic processes. This opens a new view on the role of collagen-modifying enzymes in malignancy and positions P3H2, and the ECM remodeling pathway it governs, as a potential therapeutic target worthy of further exploration, particularly in combination with anti-angiogenic strategies.

4. Materials and Methods

4.1. Bioinformatic Analysis

P3H2 transcript expression in colon tumor (T, n = 275) and normal (N, n = 359) was analyzed using http://gepia.cancer-pku.cn/index.html accessed in 7 April 2022. Kaplan–Meier survival analysis was also performed using GEPIA, dividing 270 patients into high and low P3H2 expression groups (n_high = 135; n_low = 135).

4.2. Human Tissues

Human primary CRC, CRC liver metastasis and their corresponding adjacent normal snap-frozen tissue samples were obtained from the Biological Resources Center (CRB, Centre de Resource Biologique) of Centre Léon Bérard (protocol number: BB-0033-00050). This study was approved by the ethical review board of Centre Léon Bérard (N° CMT 2015-012). All the patients provided informed consent according to French laws. RNA was extracted from human tissues with a QIACUBE kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. The RNA concentration was measured with a Nanodrop system (Thermo Scientific, Waltham, MA, USA), and the RNA quality was evaluated with a TapeStation system (Life Technologies, Carlsbad, CA, USA).

4.3. Quantitative Reverse Transcription PCR

Total RNA was extracted from cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s recommendations. Total RNA was reverse transcribed using a QuantiTec Reverse Transcription Kit (Qiagen, Hilden, Germany). The RT products (cDNA) were amplified via real-time quantitative PCR (Applied Biosystems 7900 HT Fast Real-Time PCR system, Thermo Scientific, Waltham, MA, USA) with Power SYBR Green Master Mix. Each point was done in triplicate. Oligonucleotide primers specific for human P3H2 (forward 5′-GTGCAACTGTCCTGAAAGCA-3′ and reverse 5′-TCGGCAGACCATGTGTGTAT-3′) human THBD (forward 5′-CAGAGAGGCCTTTTGGAATGTG-3′ and reverse 5′-TTCTAACCAGCTCCCATGGG-3′) and gene expression was normalized using the human GAPDH (forward 5′-TGCACCACCAACTGCTTAGC-3′ and reverse 5′-TCTTCTGGGTGGCAGTGATG-3′). The qPCR cycling conditions were 50 °C for 2 min, 95 °C for 10 min, followed by 40 cycles of a two-step amplification program (95 °C for 15 s and 58 °C for 1 min). At the end of the amplification, melting curve analysis was performed using the dissociation protocol from the Sequence Detection system to exclude contamination with nonspecific PCR products. The PCR products were also confirmed by agarose gel electrophoresis, which revealed only one specific band of the predicted size. For negative controls, no RT products were used as templates in the qPCR, and the results were verified by the absence of bands in the gel. The relative expression of the target genes was determined by the 2^–ΔΔCt method.

4.4. Cell Culture and Tumor Stable Clone Generation

All the cell lines were purchased from ATCC and cultured at 37 °C in 5% CO₂. Human colon epithelial cells CRL-1790 were grown in Eagle’s Minimum Essential Medium, while human colon cancer cell line DLD1 in RPMI, HCT116 were grown in McCoy’s 5 A medium, GEO, SW480 and SW620 were grown in Dulbecco’s modified Eagle’s medium, supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM glutamine and a standard concentration of antibiotics (EuroClone).

HCT116 were stably transfected with the pSF-P3H2 vector, the pSF-FLuc vector was used as a control, both previously described [14]. Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) was used for transfection according to the manufacturer’s instructions. 72 h post-transfection culture medium was supplemented with 800 µg/mL geneticin (Euroclone, Pero, MI, Italy). The selective medium was refreshed every two days. Two weeks later, G418-resistant clones were picked, amplified and screened for successful integration of pSF-P3H2 or pSF-FLuc by Western blot analysis.

4.5. Western Blotting Analysis

The cell lines were lysed with lysis buffer (20 mM Tris-HCl, pH 8; 150 mM NaCl; 1% Triton X-100; 10 mM EDTA; 10% glycerol; and 1 mM ZnAc). The proteins were separated via SDS-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membranes (Amersham Biosciences, Little Chalfont, United Kingdom) and probed with the antibodies against the following proteins: P3H2 (Merck, Sigma-Aldrich, St Louis, MO, USA 1:1000), β-Tubulin (Elabioscience, Houston, TX, USA 1:2000). The secondary antibodies, namely, goat anti-rabbit (GeneTex, Irvine, CA, USA) or anti-mouse HRP (ImmunoReagents, Raleigh, USA), were diluted 1:10,000. The signals were visualized by chemiluminescence using an ECL substrate (Advansta, Menlo Park, CA, USA) or by additional sensitive chemiluminescence using a LiteAblot Turbo (EuroClone, Pero, MI, Italy) following the manufacturer’s instructions.

4.6. Cell Proliferation

Proliferation of stable clones and parental cells was evaluated from 24 h up to 72 h using the CellTiter Aqueous One Cell Proliferation Assay (Promega, Madison, WI, USA) following the manufacturer’s procedure.

4.7. Soft Agar Assay

Six-well plates were precoated with a 0.75% basal agar layer containing culture media. HCT116 and pSF-FLuc and pSF-P3H2 stable clones were resuspended in a 0.32% upper agar layer and seeded at a density of 0.5  ×  10³ cells per well. The medium was changed every 3 days. After 15 days cell colonies were fixed with 4% PFA and visualized by 0.005% crystal violet (Merck, Sigma-Aldrich) staining. Images were captured, and the number of colonies was counted by ImageJ software 2.16.0/1.54p (NIH, Bethesda, MD, USA).

4.8. Transwell Invasion Assay

HCT116 and pSF-FLuc and pSF-P3H2 stable clones (1  ×  10⁵) were seeded into the upper chamber of a 24-well Transwell insert system with a polycarbonate filter with 8 μm pores (Corning) that was coated with 200 μg/mL Matrigel. The lower chamber was filled with complete medium as the positive control and with serum-free medium as the negative control. After 24 hours, the cells on the top of the filter were removed, and those on the bottom side were stained with 4′,6-diamidino-2-phenylindole (DAPI). Images were captured on a Nikon Eclipse Ni-E (Nikon, Corporation Tokyo, Japan) fluorescence microscope. For each group, seven random Transwell chamber fields were counted. Single cells were counted using ImageJ (NIH, Bethesda, MD, USA).

4.9. Animals

CD1 mice were purchased from Charles River. For all the procedures, 7- to 8-week-old male mice were used, and anesthesia was achieved by intraperitoneal injection of 100 mg/kg ketamine hydrochloride and 10 mg/kg xylazine.

4.10. Xenograft Tumor Model

A total of 3 × 10⁶ HCT116 cells and stable clones (pSF-FLuc and pSF-P3H2) were subcutaneously injected into the right flanks of CD1 (n  =  7 mice per group) nude mice. Tumor volume (mm³) was quantified three times a week by measuring the shortest (d) and longest (D) tumor diameters with an electronic caliper, the tumor volume was calculated using the formula D × d²/2. Sixteen days after injection, the mice were sacrificed and the tumor masses were harvested. For the tail vein metastasis assay, 1 × 10⁶ HCT116 cells and stable clones (pSF-FLuc and pSF-P3H2) were injected into CD1 nude mice (n  =  7 mice per group) via the tail vein. Nine weeks after the injection, the lungs were explanted.

4.11. Immunohistochemical Staining

The 10 μm thick cryopreserved tumors sections were fixed in 4% PFA and antigen retrieval was performed for 15 min at 37 °C in 0.2% trypsin and 0.001% CaCl₂ solution. Endogenous peroxidase activity was blocked with 0.3% H₂O₂ for 30 min. After blocking with 10% goat serum and 1% BSA in PBS/0.1% Triton-X-100, the slides were stained with the primary CD31 antibody (BD Pharmigen, San Diego, CA, USA, 1:1000) for 16 h at 4 °C. Then, the slides were stained with rat anti-mouse biotinylated secondary antibodies (DAKO, Santa Clara, CA, USA), and signal amplification was performed using a Vectastain elite ABC kit (Vector Laboratories, Burlingame, CA, USA). The signal was visualized using an ultraView Universal DAB Detection Kit (Merck, Sigma-Aldrich, St. Louis, MO, USA). The slides were counterstained with hematoxylin. All counts were performed on five different random fields from at least four random sections for each animal. All the images were recorded using a Nikon microscope (Nikon, Tokyo, Japan). Mouse lungs were fixed in 4% PFA and embedded in paraffin. Tissue samples were dewaxed, rehydrated and subjected to hematoxylin staining. All the images were recorded using a Nikon microscope (Nikon, Tokyo, Japan).

4.12. Immunofluorescence Analyses

The 10 μm thick cryopreserved tumors sections were fixed in 4% PFA and blocked one hour in 1% BSA, 10% Goat Serum in PBS and 0.1% Tween20 (PBS-T). After blocking, cells were exposed to primary antibodies against Collagen IV (Abcam, 1:100) or CD31 (BD Pharmigen, 1:200) in blocking solution and incubated 16 h at 4 °C in a humid chamber. After three washes in PBS-T, Alexa-Fluor conjugated secondary antibodies diluted 1:500 were incubated for 30 min. Slides were mounted with Vectashield with DAPI (Vector Laboratories, Burlingame, CA, USA) to counterstained nuclei. Images were recorded on Nikon fluorescence microscope (Nikon, Tokyo, Japan).

4.13. Statistical Analysis

The data are expressed as the mean  ±  SEM, with p  <  0.05 considered to indicate statistical significance. Differences among groups were tested by one-way ANOVA.