Abstract
The structure and molecular signature of tumor-associated vasculature are distinct from those of the host tissue, offering an opportunity to selectively target the tumor blood vessels. To identify tumor-specific endothelial markers, we performed a microarray on tumor-associated and nonmalignant endothelium collected from patients with renal cell carcinoma (RCC), colorectal carcinoma, or colorectal liver metastasis. We identified a panel of genes consistently upregulated by tumor blood vessels, of which melanoma cell adhesion molecule (MCAM) and its extracellular matrix interaction partner laminin alpha 4 (LAMA4) emerged as the most consistently expressed genes. This result was subsequently confirmed by immunohistochemical analysis of MCAM and LAMA4 expression in RCC and colorectal carcinoma blood vessels. Strong MCAM and LAMA4 expression was also shown to predict poor survival in RCC, but not in colorectal carcinoma. Notably, MCAM and LAMA4 were enhanced in locally advanced tumors as well as both the primary tumor and secondary metastases. Expression analysis in 18 different cancers and matched healthy tissues revealed vascular MCAM as highly specific in RCC, where it was induced strongly by VEGF, which is highly abundant in this disease. Lastly, MCAM monoclonal antibodies specifically localized to vessels in a murine model of RCC, offering an opportunity for endothelial-specific targeting of anticancer agents. Overall, our findings highlight MCAM and LAMA4 as prime candidates for RCC prognosis and therapeutic targeting. Cancer Res; 76(8); 2314–26. ©2016 AACR.
Introduction
Approximately 271,000 new cases of renal cancer are diagnosed each year worldwide, 3% of all cancers, with the highest incidence rate found in North America (1). By far, the most common form of kidney cancer is renal cell carcinoma (RCC), accounting for 90% to 95% of cases (1). Initial treatment is most commonly surgical, with this approach remaining the primary curative intervention (2). Unfortunately, as RCC is often asymptomatic until the tumor is advanced or metastatic, curative surgical treatment is often not possible. Five-year survival drops from 65% to 90% in operable cases and to less than 10% in metastatic disease (3). It is therefore clear that new prognostic and diagnostic biomarkers for RCC are urgently needed.
Inoperable or recurrent RCC is difficult to treat, with the success rate of traditional chemotherapy or radiotherapy at 4% to 5% (4). Because of this, a multitude of alternative therapies have been trialed and shown efficacy in metastatic RCC (mRCC), including VEGF-targeted antiangiogenic therapies, sunitinib, bevacizumab, sorafenib, and axitinib (5). RCC being a cancer characteristic for and highly dependent on excessive VEGF production, due to the loss of the tumor suppressor Von Hippel–Lindau being a common occurrence in clear cell RCC (ccRCC; ref. 6), anti-VEGF therapies have improved the outlook for mRCC (5). Despite this, 5-year survival remains low, and antiangiogenic treatment relapse is very common (5).
The endothelium is directly accessible to the blood, making it an ideal target for an emerging form of therapy, antibody–drug conjugates (ADC). These operate by delivering antibody-guided therapeutics specifically to the tumor. The first proof of principle that this approach could be effective was provided by Burrows and Thorpe (7), when they engineered a neuroendocrine tumor to express interferon-gamma, inducing MHC-class 2 on the tumor vessels, which they then targeted with a monoclonal antibody against MHC-class 2, conjugated to ricin. This therapy rapidly induced hemorrhagic necrosis in the tumor, leading to dramatic tumor regression. A key requirement for this approach is the identification of highly specific tumor vascular markers, and significant effort has been spent to identify these (reviewed in ref. 8).
Tumor-associated vessels are made distinct by many of the same pressures found in all solid tumors, such as hypoglycemia, severe hypoxia, excessive growth factor receptor activation, infiltration of inflammatory cells, and cytokine activation as well as irregular blood flow and shear stress. We therefore decided to profile the endothelium from RCC, colorectal carcinoma, and colorectal liver metastasis (CRM) to determine what common molecular changes occur in the vessels of these tumors compared with healthy tissue vessels, with the aim of identifying tumor-specific endothelial markers showing robust pan-tumor expression.
Amongst the most consistently expressed vascular markers in the cancers investigated were melanoma cell adhesion molecule (MCAM), a cell surface glycoprotein first identified as a marker on the cancerous cells of malignant melanoma (9) and previously linked to tumor cell mobility (10–12), and laminin alpha 4 (LAMA4), an extracellular matrix glycoprotein and component of the laminin complex, recently identified as a ligand for MCAM (13). Strong expression of both was shown to link negatively to RCC patient survival, suggesting a utility for the markers in RCC prognostication. MCAM expression was shown to be induced by VEGF with expression particularly enhanced in ccRCC over many other tissue types. This distinct expression pattern was shown to facilitate the specific localization of a monoclonal anti-MCAM antibody to the vascular bed of a murine RCC tumor, highlighting its potential as a ligand for specific ADC targeting in RCC.
Materials and Methods
Tissue collection and ethics
Tumor and distant healthy tissue were obtained immediately after surgery. Full patient consent and ethical approval were granted (Queen Elizabeth Hospital Birmingham: Colorectal cancer and CRM, South Birmingham REC, No. 2003/242; RCC, No. 12-090).
Endothelial isolation using Ulex lectin–magnetic beads
The process is summarized in Fig. 1A and based on the protocol published as described (14). Briefly, tissue was minced and digested in DMEM containing 2 mg/mL collagenase type V (Sigma), 7.4 mg/mL actinomycin D (Sigma), and 30 kU/mL DNase I (Qiagen), shaking at 37°C for 2 hours (colorectal carcinoma and colon) or 1.5 hours (RCC, kidney, CRM and liver). Endothelial cells were isolated from the digested single-cell suspension by positive magnetic selection using 1.4 × 107 g−1streptavidin-coated Dynabeads (Invitrogen) conjugated to biotinylated Ulex europaeus lectin (Sigma).
Isolation of endothelial cells using Ulex lectin–coated magnetic beads for microarray analysis. A, the workflow of the main steps involved in the endothelial isolation procedure. B, confirmation of endothelial isolation efficiency by qRT-PCR for markers of leukocytes (CD11b), macrophages (CD68), epithelium (EPCAM), smooth muscle (PDGFRA), and endothelium (PECAM) in the endothelial isolates (EC) from RCC (n = 8), colorectal carcinoma (CRC, n = 8), CRM (n = 7), and associated healthy tissues (n = 8, 8, and 7, respectively), standardized to flotillin 2 (a housekeeping gene) and normalized for marker expression in patient matched bulk tissue. The fold change of marker expression between the endothelial and bulk fraction is shown. Confidence limits are ±SEM. C, Venn diagram showing the commonality of genes at least 2-fold upregulated in four tumors vs. healthy tissue analyses and present in a list of known endothelial-restricted genes.
Microarray and analysis
RNA was isolated from the Ulex lectin–bead isolated endothelium, reverse transcribed to cDNA, transcribed, amplified, and labeled with Cy3 in accordance with the manufacturers' protocols (low input quick amp labelling kit; Agilent Technologies). Labeled cRNA samples were then hybridized to 8 × 66 k whole human genome expression microarrays (Agilent).
The R programming language (Lucent Technologies), marray (15), and the Limma (Bioconductor) plug-in were used to subtract background, quantile normalize probe signal intensities, and perform differential gene expression analyses on the microarray data. Raw and processed data from this analysis are deposited in the Gene Expression Omnibus (GEO) repository (accession number: GSE77199).
Quantitative real-time PCR
RNA isolation was performed using the miRNeasy mini Kit (Qiagen), and complementary DNA was generated using the High Capacity cDNA Reverse Transcription Kit (Invitrogen) in accordance with the manufacturers' protocols. Quantitative real-time PCR (qRT-PCR) was performed using the Exiqon universal probe system (Roche) as previously described (16). Primer sequences are provided in Supplementary Table S1. The Delta-Delta Ct method was used to compare the expression levels between samples, and Flotillin-2 was used to standardize expression.
Immunohistochemistry
Immunohistochemistry of human tissues and arrays was performed using 2 μg/mL mouse polyclonal antisera to CD31 (clone JC70; Dako) and rabbit polyclonal antisera to MCAM (HPA008848; Atlas Antibodies; Sigma; 0.3–0.6 μg/mL) and LAMA4 (HPA015693; Atlas Antibodies; Sigma; 0.5 μg/mL), and stained and visualized using the ImmPRESS universal antibody Kit and ImmPACT NovaRed chromagen (Vector Labs). The sections were then counterstained with Mayer's hematoxylin (Sigma), dehydrated, and mounted in distyrene–plasticizer–xylene resin (Sigma). All images were acquired using a Leica DM6000 light microscope (Leica). The following tissue arrays were used: MA2, MAN2, MB4, MBN4, MC4, MCN4, CD4, CDN4, CDA3, CL2 (Superbiochip), Hkid-CRC180Sur-01, KD951a, BC07001, and KD807 (US biomax). Human tissue cohort scoring was performed independently by three observers (Joseph W. Wragg, Henry J.M. Ferguson and, Jane A. Anderson) blinded to the patient data.
MCAM induction
Human umbilical vein endothelial cells (HUVEC) were isolated from human umbilical cords as previously described (17) and used at passage 2. Human dermal microvascular endothelial cells (HDMEC) were purchased from PromoCell and used within two passages. HUVEC and HDMEC isolates were plated at a density of 14,000 cells/cm2, grown in M199 (Sigma), l-glutamine (Gibco) and 1% FCS (Gibco) for 16 hours, then cultured ±100 ng/mL recombinant human VEGF (Peprotech) for 24 hours.
Western blot
VEGF-treated HUVEC were harvested by scraping, lysed, and subjected to 8% SDS-polyacrylamide gel electrophoresis. The protein was blotted onto nitrocellulose, stained with 0.3 μg/mL rabbit polyclonal antisera to MCAM (HPA008848; Atlas Antibodies; Sigma), visualized with ECL peroxidase-linked donkey anti-rabbit IgG (NA9340V; GE Healthcare) and ECL detection reagent (GE Healthcare) and used to develop Amersham Hyperfilm ECL (GE Healthcare).
Murine tumor and antibody localization experiments
Mice were handled and treated in accordance with home office requirements (Licence number, PPL. 40/3339). The RENCA cell line was originally obtained from the ATCC, resuscitated from early passage liquid nitrogen stocks, treated as described in ref. 18, and used in this experiment less than 1 month after the re-initiation of culture. Cells were tested negative for mycoplasma contamination. The ATCC uses morphology, karyotyping, and PCR-based approaches to confirm the identity of human cell lines. The RENCA murine RCC cell line was used to develop subcutaneous tumors. RENCA cells (1.25 × 105) were injected into the flank of 8-week-old male Balb/c mice, purchased from Harlan Laboratories UK. Tumors were permitted to grow to 1 cm3. Mice were then intravenously inoculated with 20 μg monoclonal rat anti-mouse MCAM antibodies (clone 733216; R&D Systems) 1 hour prior to cull. The tumors and several other organs were then collected snap frozen in liquid nitrogen and stored at –80°C for immunofluorescent staining.
Immunofluorescent staining
Immunofluorescent staining of frozen mouse tissue was performed using 4 μg/mL goat polyclonal antisera to rat IgGs conjugated to alexafluor 546 (A11081; Invitrogen) to visualize localized anti-MCAM antibodies. The tissue was also stained with 75 ng/mL rabbit polyclonal antisera to PECAM-1 (ab28364; Abcam) and 4 μg/mL donkey polyclonal antisera to rabbit IgGs conjugated to alexafluor 488 (A21206; Invitrogen) and mounted in ProLong Gold mounting media containing DAPI (Invitrogen). Quantification of fluorescence was conducted using the ImageJ software package (19).
Statistical analysis
Mann–Whitney, Kaplan–Meier, log-ranks, χ2, and Cox-regression statistical analyses were performed using the SPSS statistics suite (IBM).
Results
Identification of tumor vascular–associated genes
In order to identify the difference in expression profile between tumor-associated and healthy vasculature, fresh matched healthy and tumor tissues were collected from resections of colorectal carcinoma (colorectal carcinoma, n = 8; Supplementary Table S2), CRM (n = 7; Supplementary Table S3), and RCC (n = 8; Supplementary Table S4), which had not received any neoadjuvant therapy, and were processed within 3 hours of surgery. The endothelium was isolated from these tissues, according to the workflow shown in Fig. 1A, using magnetic beads conjugated to Ulex agglutinin I, a lectin isolated from Ulex europaeus, which binds specifically to the L-fucose residues restricted to glycoproteins on the surface of human endothelial cells. Endothelial-specific isolation was confirmed by qRT-PCR for markers of common cell types, leukocytes (CD11b), macrophages (CD68), epithelium (EPCAM), smooth muscle (PDGFRA), and endothelium (PECAM) in the magnetic bead isolates and normalized to the expression levels in the bulk tissue (Fig. 1B). qRT-PCR analysis determined that endothelium was enriched by between 7- and 17- fold by this procedure.
Microarray analysis was performed on four selected biologic replicates of endothelium isolated from each of RCC, colorectal carcinoma, CRM, as well as that isolated from patient-matched healthy (nonmalignant) tissues. The microarray analysis was performed using the R (64bit) software package with the Limma and marray plugins and separated into four comparison matrices: RCC versus kidney, colorectal carcinoma versus colon, CRM versus Colon, and CRM versus liver. The CRM data were compared with the colon as well as the liver from which its vessels are derived, in order to set up a direct comparison between the markers induced by the colorectal primary tumor and the colorectal metastasis. This analysis identified multiple matrix metallopeptidases consistently upregulated within the three tumor types studied, particularly MMP9, MMP12, and MMP14, whereas MMP7 and 11 were consistently upregulated in the vessels of tumors derived from colorectal malignancy alone (Supplementary Table S5). Many collagens were also modulated by exposure to the tumor environment (Supplementary Table S6). Collagen type IV family members in particular were altered in their expression levels, whereas alphas 1 and 2 were upregulated in the tumors, and alphas 3, 4, and 5 were downregulated (Supplementary Table S6).
In order to generate a shortlist of consistently upregulated tumor endothelial markers, comparative Venn analysis was performed on genes 2-fold upregulated in the cancer with a P value < 0.001, from each of the four comparison matrices, together with a list of known endothelial genes identified previously by in silico analysis (Fig. 1C; ref. 20). From this analysis, a shortlist of key genes of interest was generated (Table 1). This list included a number of well-known pan-tumor endothelial markers, such as angiopoietin 2, lysyl oxidase, apelin, and neuropilin, validating the approach as a method of identifying tumor endothelial markers. One of the most consistently upregulated genes in this analysis was LAMA4, a component of the laminin complex, a major noncollagenous constituent of the basal lamina. The recently identified endothelial receptor for LAMA4, MCAM (13), was also identified by this analysis as consistently upregulated in cancer. We therefore decided to investigate these two interacting proteins further.
Consistently upregulated vascular genes across cancer types (gene expression comparison shown as log2 fold change)
Validation of MCAM and LAMA4 as tumor endothelial markers
In order to confirm the findings of the microarray analysis, qRT-PCR was performed to assess mRNA expression levels of MCAM and LAMA4 in endothelial isolates from RCC (n = 8), colorectal carcinoma (n = 8), and CRM (n = 6) and associated healthy tissues (n = 8, 8, and 6, respectively). This analysis showed a pronounced and consistent upregulation of both genes in the endothelium of the cancers when compared with the matched healthy tissues (Fig. 2A). The identity of these genes as markers of tumor endothelium was further confirmed by semiquantitative analysis of their protein expression level by IHC, compared with that of the pan endothelial marker PECAM-1 (Fig. 2B). Vessels of both RCC and colorectal carcinoma strongly stained for MCAM and LAMA4, whereas the vessels of the associated healthy kidney and colon did not stain or stained only weakly at equivalent antibody concentrations. This was particularly apparent in healthy kidney tissue where PECAM was strongly stained within the glomerulus, a highly vascular structure, whereas neither MCAM nor LAMA4 staining was detected in glomeruli. Immunohistochemical staining for both targets was absent in a small cohort of colorectal liver metastases probed (n = 6, data not shown).
MCAM and its extracellular matrix-binding partner LAMA4 are specific markers of endothelium in both renal and colorectal cancers. A, confirmation of cancer-specific enrichment of MCAM and LAMA4 by qRT-PCR on endothelial isolates from RCC (n = 8), colorectal carcinoma (CRC; n = 8), CRM (n = 6), and associated healthy tissues (n = 8, 8, 6). Expression standardized to flotillin 2; confidence limits, ±SEM; statistical analysis: Mann–Whitney U test (*, P < 0.05; **, P < 0.01; ***, P < 0.001). B, confirmation of cancer-specific enrichment of MCAM and LAMA4 by IHC. Representative images of kidney (arrows, glomeruli), RCC, colon, and colorectal carcinoma stained for PECAM (endothelial marker), MCAM, and LAMA4. Scale bar, 50 μm. C, analysis of marker expression in 18 common cancers and associated healthy tissues by IHC. The proportion of tissue of each type scored as exhibiting strong staining is shown. D, VEGF significantly induced MCAM expression in endothelial cells. HUVEC isolates (n = 6) and HDMEC isolates (n = 2 mixed isolates, each in triplicate) were serum and growth factor starved for 12 hours, and then cultured with serum-depleted media ± 100 ng/mL recombinant human VEGF (hVEGF); MCAM expression was determined by Western blot and qRT-PCR. Confidence limits, ±SEM; statistical analysis, Mann–Whitney U test; **, P < 0.01.
In order to identify the tissue-specific expression profile of MCAM and LAMA4, 10 samples each of 18 common cancers and associated healthy tissues were stained and scored for strength of staining (Fig. 2C). Both MCAM and LAMA4 demonstrated markedly specific vascular expression profiles, with vessels the primary source of staining in all tissues other than melanoma, where MCAM expression was primarily found on the tumor cells (data not shown). Of note, 90% of kidney tumors showed strong vascular staining for MCAM, in excess of any healthy tissue and most cancerous tissues examined, highlighting MCAM as a promising vascular target in RCC. LAMA4 on the other hand was shown to be highly expressed in a broad range of both tumor and healthy tissues (Fig. 2C).
MCAM expression is induced by VEGF
MCAM expression on tumor vessels appears to be highly RCC specific. High VEGF expression is common in cases of ccRCC (6), due to the loss of the tumor suppressor Von Hippel–Lindau, as previously discussed. In order to investigate whether enhanced VEGF production within the tumor could be the cause of high MCAM expression, isolated human umbilical cord vein endothelium (HUVEC) and commercially purchased HDMEC were exposed to recombinant VEGF. Six HUVEC isolates and two multisource HDMEC isolates, each used in triplicate, were serum starved for 16 hours and then cultured with or without 100 ng/mL recombinant VEGF for 24 hours before being harvested. qRT-PCR and Western blot analysis for both cell types showed that MCAM was significantly upregulated by VEGF (Mann–Whitney U test, P < 0.01; Fig. 2D).
CcRCC alone is characteristic for high VEGF expression, and all the RCC tumors in the multiorgan expression analysis were clear cell. In order to investigate whether MCAM expression is specific for ccRCC, tissues from five different types of renal malignancy were stained by IHC. This analysis revealed that a significantly greater proportion of clear cell tumors stained strongly for MCAM than nonclear cell tumors (61%, n = 64 vs. 30.5%, n = 36; χ2, P < 0.005). A breakdown of results for individual renal malignancy histology types is shown in Supplementary Table S7. This result supports the suggestion that VEGF is playing a part in enhancing MCAM expression in ccRCC. In addition, this result suggested that ccRCC should be the focus of investigation.
Identification of strong MCAM and LAMA4 expression as potent adverse prognostic indicators in ccRCC
In order to investigate whether the strong expression of MCAM and LAMA4 in the vessels of many RCC and colorectal carcinoma tumors could have some prognostic value, excised tissues from cohorts of ccRCC [n = 81 (cohort 1), 48 (cohort 2), 47 (cohort 3)] and colorectal carcinoma (n = 90) were stained by IHC for each marker and semiquantitatively scored. Demographic information is shown for each cohort in Supplementary Tables S8 (RCC cohort 1), S9 (RCC cohort 2), S10 (RCC cohort 3), and S11 (colorectal carcinoma cohort).
For effective investigation of prognostic linkage to marker expression, the analytical tools must be sensitive to the full range of marker expression within the cohort. As the multiorgan tissue array staining for MCAM resulted in 90% of RCC samples staining strongly, the antibody concentration was titrated down from 0.6 to 0.3 μg/mL, to a level where a range of MCAM staining in RCC could be observed. Antibody concentration of 0.6 μg/mL was used for the results listed in Fig. 2, and 0.3 μg/mL was used for all other IHC analyses.
Each cohort was split into tumors exhibiting strong or weak marker staining, as judged by three independent scorers (Joseph W. Wragg, Henry J.M. Ferguson and, Jane A. Anderson; representative images of each group are shown in Fig. 3A, Supplementary Fig. S1). Tumor marker staining was correlated with patient survival by the Kaplan–Meier analysis (Fig. 3B). This analysis identified a significant decrease in survival in patients whose tumors exhibited strong MCAM and LAMA4 staining in both RCC cohort 1 (log-ranks, P = 0.001 and 0.0005, respectively) and cohort 2 (P = 0.08 and 0.001, respectively; Fig. 3B). In patients with tumors exhibiting strong staining for both MCAM and LAMA4 together, this effect was more pronounced, with only 18% surviving to date versus 75% in tumors that were not strongly stained for either marker in RCC cohort 1 and 27% versus 81% in RCC cohort 2 (P < 0.0005, both cohorts; Fig. 3B). In the colorectal carcinoma cohort, there was no significant association between survival and strong MCAM (P = 0.809) or LAMA4 (P = 0.353) expression alone or when coexpressed (P = 0.713; Fig. 3B). This suggests that MCAM and LAMA4 play a unique, tumor-specific role in ccRCC patient survival.
High MCAM and LAMA4 tumor vessel expression has a significant detrimental effect on the survival of RCC patients but not colorectal carcinoma (CRC) patients. A, representative images demonstrative of scoring for staining intensity. Images are of MCAM staining in RCC at a weak (i and ii) or strong (iii and iv) level. Scale bar, 50 μm. B, Kaplan–Meier survival analysis of patients from RCC cohorts 1, 2, and 3 and the colorectal carcinoma cohort, with strong vs. weak staining for MCAM, LAMA4, and marker coexpression. Statistical analysis, log-ranks test, P and n for each test are shown. Crosses, censored cases.
No significant survival effect was observed in RCC cohort 3, with the exception of where the markers were coexpressed (P = 0.013). RCC cohorts 1 and 2 are primarily made up of nonmetastatic tumors, with only five with metastasis across the two cohorts, and survival to last checkup is 64% and 66%, respectively (Supplementary Tables S8 and S9). These types of cohort are ideal for looking at survival effects, as the potential for a separation in survival between two groups is great. RCC cohort 3 is entirely made up of metastatic tumors, and survival to last checkup is 21% (Supplementary Table S10). It is reasonable to suggest therefore that a single-marker survival effect is not seen as too few patients survive. This does however suggest that prognostication using MCAM and LAMA4 marker expression will find its greatest utility in early-stage disease.
Multivariate cox-regression analysis on the largest RCC cohort, cohort 1, identified MCAM (P = 0.006) and LAMA4 (P = 007) individually and in combination (P = 0.002) as independent risk factors for reduced survival in patients with tumors exhibiting strong staining, independent of gender, age, histopathologic tumor grade, or T stage (Table 2). This analysis additionally identified a considerably greater risk of death in patients exhibiting strong MCAM and LAMA4 expression: MCAM, OR 3.4, confidence interval (CI), 1.4–8.1; LAMA4, OR 3.3, CI, 1.4–7.9; coexpression, OR 4.1, CI, 1.7–10 (Table 2).
Multivariate analysis (Cox regression) of prognostic markers in RCC cohort 1 (n = 81), censored cases (n = 52; 64.2%)
MCAM and LAMA4 expression is enhanced in locally invasive and metastatic disease
Metastatic disease is the area of most therapeutic need for RCC, survival is lowest, and there is the greatest need for systemic therapies. A tumor vascular target would ideally have utility in this setting. In order to investigate the expression of MCAM and LAMA4 in metastatic disease, tumors from ccRCC cohorts 1, 2, and 3 were grouped based on their metastatic status. This revealed that the proportion of metastatic tumors exhibiting strong staining is significantly greater than in tumors with no known metastases (76% vs. 41% MCAM, 68% vs. 36% LAMA4, χ2, P < 0.001; Fig. 4A). It should be borne in mind, however, that the majority of metastatic tumors are from one cohort and nonmetastatic from another, and possible differences in tissue preparation may have affected the result. An additional analysis grouping the tumors based on T stage found that both markers are significantly enriched in tumors exhibiting greater local invasion (χ2, P < 0.001; Fig. 4B). In the case of metastatic disease, it is not just the primary tumor that must be treated, but also the metastases; therefore, MCAM staining in metastases from ccRCC tumors was investigated (Fig. 4C). This analysis revealed that 73% of ccRCC metastases (n = 15) exhibit strong MCAM staining across seven different metastasis locations (Supplementary Table S7). Sample numbers were too low, however, to dissect whether MCAM expression is greater in certain metastasis locations over others. MCAM expression was additionally observed in metastases from papillary and squamous cell RCC; again however, sample numbers (n = 2,2) were too low to make any significant conclusions (Supplementary Table S7). These data collectively identify MCAM and LAMA4 as markers of advanced disease and MCAM as an ideal target for the treatment of metastatic RCC.
MCAM and LAMA4 expression is enhanced in metastatic and locally advanced ccRCC. A, pie-chart proportional representation of strong (black) vs. weak (white) marker expression in metastatic and nonmetastatic RCC, determined by IHC on RCC cohorts 1, 2, and 3. Statistical analysis, χ2, ***, P < 0.001. B, pie-chart proportional representation of strong (black) vs. weak (white) marker expression at different RCC T stages, determined by IHC on RCC cohorts 1, 2, and 3. Statistical analysis, χ2, ***, P < 0.001. C, representative images of MCAM staining in ccRCC metastases to various organs, generated by IHC. Scale bar, 50 μm.
Monoclonal anti-MCAM antibodies specifically localize to murine RCC tumor vessels
The abundance and specificity of MCAM expression in renal tumors open the possibility of using it as a targeting ligand for therapeutic agents. In order to investigate this potential utility, the localization of a monoclonal rat anti-MCAM antibody was determined, following intravenous injection into murine RCC (RENCA) tumor-bearing mice. The RENCA model was chosen as the tumors have previously been shown to express VEGF at a high level (21), closely modeling the majority of human renal cell cancers. In addition, subcutaneous RENCA tumors strongly express MCAM on their vessels (Fig. 5A), identifying it as an ideal model for use in this situation. An hour after antibody infusion, the mice were culled, the organs harvested, and then processed for frozen sectioning. Localized antibody was detected by fluorescently labeled anti-Rat IgGs. Tissues from the RENCA tumor, stomach, heart, liver, colon, kidney, skin, and lung were probed. The average optical density of fluorescence emanating from localized anti-MCAM antibody within the vessels of each tissue in 2 mice and 12 regions of interest was quantified and demonstrated an at least 25-fold greater localization of antibody in the vessels of the RENCA tumors than any other tissue probed (Figs. 5B and C). This finding suggests that RCC therapies could potentially use anti-MCAM antibodies to localize therapeutics to the tumor vasculature specifically, permitting a functional anticancer effect.
A monoclonal anti-MCAM antibody specifically localizes to murine RCC tumor vessels. A, triple immunofluorescent staining of a murine RCC (RENCA) tumor for PECAM-1 (green), MCAM (red), and DAPI (blue). Scale bar, 25 μm. B and C, 20 μg of MCAM monoclonal antibody was intravenously injected into RENCA tumor-bearing mice 1 hour prior to cull. The tumor and selected organs were collected. Frozen sections were stained with anti-rat IgGs (red), PECAM-1 (green), and DAPI (blue). B, the average optical density of fluorescence detected in the anti-rat IgG (red) channel, within regions of vascular (PECAM-1) staining, was quantified. The tissue from two mice was assessed, with six regions of interest selected for each organ and mice (n = 12); confidence limits, ±SEM. C, representative images of MCAM monoclonal antibody localization. Scale bar, 12.5 μm.
Discussion
The aim of this study was to identify vascular markers with pan-tumor expression and demonstrate their utility as specific ligands against which to target immunological therapies. This study identified MCAM and LAMA4 as promising markers with specific overexpression in endothelial isolates from both colorectal and renal malignancies. A significant link between high expression of these markers and poor patient survival, invasive local disease and metastasis, was demonstrated in ccRCC, but not colorectal carcinoma. MCAM expression was found to be highly enriched in the vessels of ccRCC, in excess of other tumor and healthy tissues, possibly due to VEGF induction demonstrated in this article. These data highlighted MCAM as a potential specific target in RCC, and this utility was demonstrated by specific localization of MCAM monoclonal antibodies to the tumor vessels in a model of RCC.
Comparative analysis of vessels derived from colorectal carcinoma, CRM, and RCC with patient matched healthy tissues identified a small group of endothelial genes consistently upregulated in these tumors. Many of these genes are stimulatory to angiogenesis and tumor invasion, such as LOX (22), MCAM (23), LAMA4 (24), NRP1 (25), MMP1 (26), APLN (27), and SPARC (28), suggesting a signature characterized by active angiogenesis.
One of the most consistent pan-tumor endothelial markers in the analysis was LAMA4, an extracellular matrix glycoprotein and component of the laminin complex. Laminins are made up of three chains, alpha, beta, and gamma, and have been implicated in a wide variety of cellular processes, from cell attachment and differentiation, to influences on cell shape and movement, maintenance of tissue phenotype, and promotion of tissue survival (29). The function of individual laminin chains is poorly understood; however LAMA4, a constituent of laminin-8, 9, and 14 (30), has been shown to have an endothelial-specific expression pattern and also to promote angiogenesis (24). LAMA4 has been shown to codistribute and interact with integrins αvβ3, α3β1, and together with α6β1 mediate endothelial cell–LAMA4 interactions and blood vessel formation (24).
The exact role LAMA4 plays in cancer is unclear. In this study, it is shown to be strongly upregulated on tumor blood vessels in colorectal and renal malignancies, when compared with surrounding nonmalignant tissue. Its tissue distribution appears to be diverse when more broadly investigated; however, appearing in both healthy and tumor tissues. This analysis suggests that using LAMA4 as a target for cancer therapy could be problematic. This study does however demonstrate a highly significant link between LAMA4 expression in RCC and poor patient survival, which is not shared in colorectal carcinoma. A strong association between LAMA4 expression and both metastasis and local invasion is also shown. LAMA4 has previously been associated with increased tumor invasion and metastasis in hepatocellular carcinoma (31), the transition from premalignant to malignant breast carcinomas and reduced relapse-free survival in estrogen receptor–negative breast cancer patients (32), marking LAMA4 as a useful prognostic marker in certain cancers, not least RCC. A recombinant form of the LAMA4 chain containing Laminin-411 (Laminin-8) has been reported to have an antiadhesive effect on RCC cells grown on fibronectin (33). This report, combined with our observations, suggests a potential mechanism in which heightened vascular LAMA4 might impair RCC tumor adhesion and thereby increase metastasis, negatively affecting patient survival. The functional relevance of this mechanism warrants further investigation.
This study also identified MCAM as a potent vascular marker in ccRCC. The role of MCAM in the vascular endothelium is poorly understood, but it is thought to promote angiogenesis (23) and act as a coreceptor for VEGFR-2, thus enhancing endothelial migration and microvessel formation (34). Endothelial conditional knockout of MCAM in mice results in impaired vessel formation in VEGF-dependent angiogenesis assays (34). MCAM is also thought to play a role in cell-to-cell junctions and vascular permeability (35). Additionally, MCAM overexpression has been associated with prosurvival signaling, including protein kinase B (PKB) phosphorylation and downregulation of BCL2-associated agonist of cell death (BAD) expression (36). It is therefore plausible that upregulation of MCAM in tumor vessels could act as a survival mechanism, as well as affect angiogenesis and vascular integrity.
MCAM was first identified as a marker on the carcinoma cells of malignant melanoma, emerging as a potential prognostic indicator of cancer progression (9). MCAM expression has also been reported on the carcinoma cells of prostate (10), breast (11), and ovarian cancers (12), suggesting that MCAM could be a widely expressed tumor antigen. However, in this study, MCAM expression was only occasionally observed on the tumor cells of tissue examined, with MCAM expression being almost exclusively reserved to tumor vessels in all malignancies aside from melanoma, calling into question the relative importance of tumor and endothelial cell MCAM expression in these malignancies.
A significant association between high vascular MCAM expression, poor RCC patient survival, and increased metastasis and local invasion was demonstrated in this study. Heightened MCAM mRNA expression has previously been reported in bulk tumor tissue from patients with RCC, with the highest levels observed in metastatic disease, indicating a direct correlation between increasing MCAM expression and disease progression (37), partially corroborating our observations. We further this finding by highlighting vascular MCAM as key to this process and demonstrating a direct survival impact. High MCAM expression has additionally been associated with poor survival in patients with non–small cell lung adenocarcinoma (but not squamous cell carcinoma; ref. 38). These data highlight MCAM as an important prognostic marker in cancer, in particular RCC, where coexpression with its recently identified extracellular ligand LAMA4 (13) was shown to be highly predictive of very poor patient survival. This observation suggests that the expression and interaction of MCAM and LAMA4 could be highly significant in RCC progression and should be further investigated. MCAM interaction with the LAMA4-containing laminin-9 complex has been shown to promote migration of tumor cells, when associated with α6β1 integrin (39), but the importance of this interaction for endothelial cell or cancer biology is unknown.
Multi-tissue analysis identified MCAM as highly specific to RCC in its expression. Although it is present in other tissues, its expression in ccRCC vessels is greatly in excess. This study identified VEGF-mediated induction of MCAM in endothelial cells as a potential explanation for this. This is the first study to demonstrate that VEGF, a growth factor highly expressed in RCC (6), will induce MCAM expression in endothelial cells. The regulation of MCAM on vessels is poorly understood. The expression of MCAM in HUVEC was found to be upregulated by culture with media conditioned with a hepatoma cell line (23); however, the exact mechanism was not determined. Tumor necrosis factor has been reported to induce the formation of a soluble form of MCAM in various cell types, including endothelium (40,41). Of note, insulin-like growth factor-binding protein 4 (IGFBP-4) has been reported to induce MCAM in RCC. IGFBP-4–transfected renal tumor cells were found to exhibit enhanced cell growth, invasion, and motility, as well as enhancing MCAM expression (42). However as has been discussed, in this study, MCAM expression was primarily observed on tumor vasculature, with no discernable RCC tumor cell expression, so the significance of IGFBP-4–induced MCAM expression on RCC tumor cells in human cancer is unclear.
The finding that MCAM, a cell surface glycoprotein, was highly specific to RCC vessels marked it out as an ideal target for anticancer immunological therapies. This study demonstrates this utility by showing a monoclonal anti-MCAM antibody specifically localizing to the vessels of a murine model of RCC, accumulating in the tumor at least 25-fold greater density. This constitutes a striking level of antibody specificity to cancer, in excess of other successful tumor localization studies (43,44). A number of successes in anticancer cell targeting using antibodies have been achieved, including FDA-approved ADCs brentuximab vedotin in Hodgkin's lymphoma (45) and transtuzumab emtansine in breast cancer (46). However as mentioned in the introduction, the targeting of tumor blood vessels does offer a number of advantages: the blood vessels are easily accessible for ligand targeting; up to 100 tumor cells are dependent on a single endothelial cell (47) for survival, making the vessels an extremely efficient target; the vasculature is thought to be more genetically stable and so more homogeneous in terms of marker expression (48); and finally therapeutics disrupting the vessels have been shown to cause preferential lysis in the core of large tumors (49), a region poorly treated by antiangiogenics and traditional chemotherapeutics (50). This study therefore identifies and validates MCAM as a new ligand with which to specifically target therapeutics against RCC vessels, potentially improving treatment of this malignancy.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: J.W. Wragg, V.L. Heath, R. Bicknell
Development of methodology: J.W. Wragg, R. Bicknell
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.W. Wragg, J.P. Finnity, J.A. Anderson, H.J.M. Ferguson, E. Porfiri, R.I. Bhatt, P.G. Murray
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.W. Wragg, J.P. Finnity, P.G. Murray, R. Bicknell
Writing, review, and/or revision of the manuscript: J.W. Wragg, J.A. Anderson, H.J.M. Ferguson, E. Porfiri, R.I. Bhatt, P.G. Murray, V.L. Heath, R. Bicknell
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.W. Wragg, H.J.M. Ferguson, R.I. Bhatt
Study supervision: R. Bicknell
Other (specimen identification and consent and collection): R.I. Bhatt
Grant Support
J.W. Wragg, J.P. Finnity, V.L. Heath, and R. Bicknell received funds from the University of Birmingham CRUK Cancer Center. J.A. Anderson, H.J.M Ferguson, E. Porfiri, R.I. Bhatt, and P.G. Murray received funds from the UHB NHS Foundation Trust.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Acknowledgments
The authors thank Dr. Andy Reynolds for his provision of the RENCA cell line and Professor Nick James for funding the RCC tissue collection.
Footnotes
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received May 26, 2015.
- Revision received December 17, 2015.
- Accepted January 1, 2016.
- ©2016 American Association for Cancer Research.
References
Volume 76, Issue 8
