VEGF and VEGFR-1 are coexpressed by epithelial and stromal cells of renal cell carcinoma

Abstract

BACKGROUND.

Tumor angiogenesis is a dynamic process that plays a major role in cancer progression. Vascular endothelial growth factor (VEGF) and its receptors play a pivotal role in angiogenesis. The expression of VEGF and its receptors VEGFR-1 and VEGFR-2 in renal cell carcinoma (RCC) was investigated in the perspective of anti-VEGF treatments.

METHODS.

Total VEGF protein levels were quantified by enzyme-linked immunosorbent assay (ELISA) in tumor tissue samples from surgical specimens of 65 patients with clear cell RCC. At the cellular level the VEGF isoforms VEGFR-1 and VEGFR-2 mRNA were quantified by real-time quantitative reverse-transcriptase polymerase chain reaction (RT-PCR) in laser-microdissected tumoral epithelial as stromal cells and in corresponding normal tissue compartments. Colocalization of VEGF and VEGFR-1 proteins was studied by triple immunofluorescent labeling.

RESULTS.

Protein VEGF in cytosolic extracts was significantly higher in tumoral than in nontumoral tissue (P < .0001). Event-free survival was significantly longer for patients with cytosolic VEGF lower than the cutoff (75th percentile of VEGF protein levels, P = .02). In laser-microdissected epithelial cells, VEGF₁₂₁ and VEGFR-1 mRNA expressions were higher in RCC than in corresponding nontumoral kidney (P = .007 and P = .002, respectively); they were also higher in stromal cells of RCC compared with nontumoral kidney (P = .02 and P = .003, respectively). There was no differential VEGFR-2 expression in epithelial or in stromal cells of tumoral or nontumoral kidney. By immunofluorescent labeling VEGF and VEGFR-1 colocalized on RCC tumor epithelial and stromal cells.

CONCLUSIONS.

Combined laser microdissection and quantitative RT-PCR, as triple immunofluorescent labeling, underlined the preferential expression of the most soluble VEGF isoform, VEGF₁₂₁, and its receptor VEGFR-1, but not VEGFR-2, in epithelial and stromal cells of RCC. Cancer 2008. © 2007 American Cancer Society.

Tumor angiogenesis is a dynamic process that plays a major role in cancer progression, particularly in renal cell carcinoma (RCC).1–3 Vascular endothelial growth factor (VEGF) and its receptors play a pivotal role in physiologic and pathologic angiogenesis. Therapeutic molecules directed against VEGF protein and VEGF receptor signaling are now available for RCC treatment.3, 4 Therefore, further characterization of VEGFR expression on RCC cells and investigation into VEGF as a growth factor in RCC is needed.

VEGF binds with high affinity to the receptor tyrosine kinase VEGFR-1 (Flt-1) and VEGFR-2 (Flk-1, KDR), which upon ligand binding become tyrosine phosphorylated and activate multiple signaling networks.5 VEGF increases microvascular permeability, induces endothelial cell proliferation, survival, migration, and differentiation, and promotes the degradation of the extracellular matrix around the sprouting endothelium by inducing the expression of proteases. VEGF has 5 main isoforms produced by alternative splicing of a single gene located on 6p21.3, VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉, VEGF₁₄₅, and VEGF₂₀₆.6, 7 These isoforms differ primarily in their bioavailability, which is conferred by heparin and heparan-sulfate binding domains encoded by exons 6 and 7. VEGF₁₈₉, VEGF₁₄₅, and VEGF₂₀₆ contain additional stretches of basic residues, resulting in their nearly complete retention in the extracellular matrix. VEGF₁₆₅ exists both as an extracellular matrix-bound and as a soluble form released by proteolysis.2, 8 VEGF₁₂₁, which lacks both exon 6 and exon 7, is a highly diffusible protein. Several studies point out VEGF₁₂₁ and VEGF₁₆₅ as the most expressed isoforms in human tissues and tumors.9–11 Tumor progression involves complex interactions between neoplastic epithelial cells and the surrounding stroma composed of fibroblasts, endothelial cells, smooth muscle cells, inflammatory cells, and of a macromolecular network termed the extracellular matrix. In organs, tissue homeostasis is dictated by interactions between stromal and epithelial cells. During pathologic processes, particularly cancer, both the epithelial and the stromal compartments undergo crucial changes for tumor progression.12 Supportive functions, such as angiogenesis, are provided by the stromal components and allow tumor growth. Both stromal and epithelial cells are able to produce VEGF, and the expression of VEGF receptors R-1 and R-2 by epithelial cells from normal and neoplastic tissue has been reported.2, 13

In the present study we used laser microdissection coupled to real-time quantitative polymerase chain reaction (QRT-PCR) to assess the respective expression of VEGF and its receptor transcripts in RCC tumor and stromal cells. At the protein level we analyzed VEGF quantitative expression in tissue extracts and VEGF and VEGFR-1 distribution and coexpression in the different types of cells.

MATERIALS AND METHODS

Patients

Between January 1998 and May 2001, 100 patients with untreated RCC were admitted to the Department of Urology of Hospital St.-Louis (Paris, France). RCC was histologically subclassified according to the current classification of adult renal epithelial neoplasms.14 Sixty-five patients with conventional cell renal carcinoma were included in this study (47 men, 18 women; median age 65 years; range, 31–88) (Table 1). Among the 35 RCC samples excluded, 16 were papillary carcinomas, 5 were chromophobe carcinomas, and 9 were conventional RCC in patients for whom no follow-up was possible. All patients consented to this study. Before surgery, all patients underwent clinical examination and thoracoabdominal scan. Staging was done according to the TNM classification.15 T and N were always defined as pT and pN. The 65 patients underwent nephrectomies (radical, n = 59; partial, n = 2; tumorectomies, n = 4) and were classified as: T1, n = 45; T2, n = 14; T3, n = 5; T4, n = 1. Patients with lymph node invasion were staged as N+. Metastasis was defined according to the clinical findings if no histopathologic data on metastasis were available. In 10 patients distant metastases were discovered at the time of diagnosis (lung, n = 7; bone, n = 1; brain, n = 1; distant lymph node, n = 1). In 9 patients metastasis occurred after surgery (lung, n = 5; bone, n = 1; inferior vena cava, n = 1; brain n = 1; distant lymph node, n = 1). Fuhrman nuclear grade was as follows: grade 1, n = 7; grade 2, n = 35; grade 3, n = 19; grade 4, n = 4.16

Table 1. Characteristics of Patients

Patients	N = 65
Median age, range, y	65, 31–88
Sex
Men	47
Women	18
Nephrectomy
Radical	59
Partial	2
Tumorectomy	4
TNM size (pT)
T1	45
T2	14
T3	5
T4	1
Nodes (pN)
N+	4
N−	61
Metastasis
M+	10
M−	55
Furhman grade
n 1	7
n 2	35
n 3	19
n 4	4

Sample Collection

Surgical pieces were immediately analyzed macroscopically. Samples from tumoral areas and from macroscopically normal areas were isolated. Half of them were immediately snap-frozen in liquid nitrogen, the other half were fixed in formaldehyde and further processed for paraffin embedding. A systematic microscopic control of the samples was performed and the tumor samples containing areas of necrosis were not used.

VEGF Protein Level Quantification in Whole Tissue Sections

Sixty whole-tissue sections (10 μm thick) were taken from frozen tumoral and nonnormal renal tissue. Sections were placed in 10 mM Tris-HCL buffer, pH 7.4, containing molybdat and dithiotreithol (Sigma, St.-Quentin Fallavier, France). Cytosolic extracts from RCC and corresponding normal kidney (n = 52) were obtained by centrifugation at 100,000g for 1 hour at 4°C. VEGF levels were assayed in duplicate by specific sandwich enzyme immunoassay techniques (Quantikine R&D Systems, catalog number DVE00, Minneapolis, Minn). The minimum detectable concentration was estimated to be 9 pg/mL. The enzyme-linked immunosorbent assay (ELISA) test for VEGF recognized the different VEGF isoforms.

Triple Immunofluorescence Study

Triple immunofluorescence labelings were performed on 5-μm-thick frozen sections from tumoral and nontumoral areas. The sections were incubated with primary antibodies directed against VEGF (mouse antihuman VEGF-A antibody, Santa Cruz Biotechnology, Santa Cruz, Calif), VEGFR-1 (goat antihuman VEGFR-1, R&D Systems) at 1:100 dilution or VEGFR-2 (goat antihuman VEGFR-2, Santa Cruz Biotechnology) at 1:100 dilution. The sections were then incubated with FITC-conjugated chicken antimouse IgG (Santa Cruz Biotechnology) at 1:200 dilution and PE-conjugated donkey antigoat IgG (Santa Cruz Biotechnology) at 1:500 dilution. Subsequently, the sections were incubated with mouse cytokeratin antibodies (clone AE1/AE3, at 1:100 dilution, Chemicon, Temecula, Calif) which were prelabeled with Zenon Alexa Fluor 350 mouse IgG labeling kit (Molecular Probes, Eugene, Ore). Using an AX 70 Olympus microscope with SIS software, successive pictures were captured on the same area analyzed with different fluorochrome wavelength filters. For each selected area the successive pictures were overlaid. Cells expressing cytokeratins were considered epithelial cells when cells not expressing cytokeratins were considered stromal cells.

Combined Laser Microdissection and QRT-PCR

Laser microdissection was performed on epithelial cells and stromal cells on 5-μm-thick frozen sections of tumoral (n = 33) and nontumoral renal tissue (n = 22). On an Olympus BX inverted microscope with a PALM laser microdissector, the 2 different cell populations were successively laser-microdissected and catapulted into caps of tubes containing Trizol reagent for RNA extraction. The surface of the different laser-microdissected areas was calculated using PALM robot software v. 3.0. The systematic quantitative assessment allowed us to laser microdissect a minimum of 500,000 tumor cells and 100,000 stromal cells.

RNA extraction and reverse transcription

RNA was extracted using Trizol reagent (Invitrogen, France). Reverse transcription of total RNA was performed using 200 U of Superscript II RNase H-reverse transcriptase with random hexamers (Invitrogen) according to the manufacturer's instructions.

Primer and probe design

Quantification of the transcripts coding for VEGF₁₂₁, VEGFR-1 and VEGFR-2, and control gene, β2 microglobulin (B2M), was performed using LightCycler technology (Roche Diagnostics, France). Primers and probes were chosen using Primer Express Software (Applied Biosystems, France) (Table 2). Hydrolyzation probes were labeled with a reporter dye (6-carboxy-fluorescein phosphoramidite) at the 5′ end and a quencher dye (5-carboxy-tetramethylrhodamine) at the 3′ end.

Table 2. Specific Primer and Probe Sequences Used for Real-Time QRT-PCR

Name	Sequence	Exon junction	Product size (bp)
QRT-PCR indicates quantitative reverse transcriptase-polymerase chain reaction; bp, basepair.
VEGF121 - sense	5′-gAgCTTCCTACAgCACAACAAA-3′	5–8	99
VEGF121 - antisense	5′-CTCggCTTgTCACATTTTTC-3′
VEGF121 - probe	5′-TgCAgACCAAAgAAAgATAgAgCAAgACA
VEGFR-1- sense	5′-CgACgTgTggTCTTACggAgTA-3′	24–25	107
VEGFR-1- antisense	5′-CTTCCCTCAggCgACTgC-3′
VEGFR-1- probe	5′-TgTgggAAATCTTCTCCTTAggTgggTCTC-3′
VEGFR-2- sense	5′-TCTCAATgTggTCAACCTTCTAgg-3′	19–20	79
VEGFR-2- antisense	5′-AAATTTgCAgAATTCCACAATCAC-3′
VEGFR-2- probe	5′-TgTACCAAgCCAggAgggCCACTC-3′
B2M- sense	5′CgCTCCgTggCCTTAgC 3′	1–2	70
B2M- antisense	5′ gAgTACgCTggATAgCCTCCA 3′
B2M- probe	5′TgCTCgCgCTACTCTCTCTTTCTggC 3′

QRT-PCR

PCR optimization and specificity of RT-PCR products were conducted using SYBR Green technology, melting curves, and agarose gel electrophoresis. Transcript quantification was performed using TaqMan technology. Quantitative PCR was performed using either the SYBR Green or LCFastStart DNA Master Mix kit (Roche Diagnostics) according to the manufacturer's instructions. All experiments were performed in duplicate.

To determine the absolute copy number of the target transcripts, the cDNAs for VEGF₁₂₁ mRNA and its receptors were cloned in TOPOII TA cloning Kit (Invitrogen) following the manufacturer's recommendations. Cloned products were digested with EcoR I (Invitrogen, Cergy Pontoise, France), extracted from 2% agarose gel, purified with the PCR purification Kit (Qiagen, Courtaboeuf, France). Finally, the products were measured in a spectrophotometer and molecule concentrations were calculated. A standard curve for each transcript was generated using serial dilutions of cloned products ranging from 1 or 10 to 10⁹ molecules/μL. The copy number of unknown samples was calculated by setting their PCR cycle number (Crossing Point) to the standard curve. To correct for differences in both RNA quality and quantity between samples the expression levels of interest transcripts were normalized to the housekeeping B2M gene transcripts. The results are presented as copies of target gene per 10⁶ copies of B2M.

Statistical Analyses

Values for VEGF protein levels and for the gene expression levels of VEGF₁₂₁, VEGFR-1, and VEGFR-2 in tissue extracts are given as median and range. Differences for VEGF protein and mRNA expression between different renal tissue compartments (tumoral and nontumoral, epithelial, and stromal) were analyzed by Wilcoxon test for paired data. Univariate analysis of the correlation between the different parameters was done using the nonparametric Spearman rank test. For survival analysis the stopping date was January 1, 2006. The overall survival (OS) rates were measured from the date of surgery to the study endpoints, which were the stopping date, the date of death from RCC, or the time of last visit before the patient was lost to follow-up. For the patients free of tumor after surgery, the event-free survival (EFS) was measured from the date of surgery to the study endpoints. Survival rates were estimated by the method of Kaplan and Meier. Univariate analysis was performed using log-rank tests. Results for comparison of major endpoints were regarded as significant if the 2-sided P was < .05. Statistical analysis was performed using the SAS v. 913 software package (SAS Institute, Cary, NC).

RESULTS

VEGF Protein Levels in Whole Tissue Sections

The VEGF protein level was significantly higher (P < .0001) in tumoral tissue than in corresponding nontumoral renal tissue (253 pg/mg, 10–79,959 vs 45 pg/mL, 10–2491; n = 52) (Fig. 1). Tumor VEGF protein level was associated with tumor size (P = .03), tumor grade (P = .02), and metastasis at diagnosis (P = .006) (Table 3). Tumor protein levels were significantly higher (P = .01) in patients with metastasis detected during the follow-up (2061 pg/mg, 78–6179, n = 9) than in patients with no metastasis at all, ie, neither before surgery nor during the follow-up (207 pg/mg, 10–12,812, n = 46).

Figure 1. Individual values of VEGF (pg/mg) in renal cell carcinoma (RCC) and normal tissue. Data points, mean of 2 determinations, bars, median values; P-value was calculated using the nonparametric Wilcoxon test for paired data.

Table 3. VEGF (pg/mg) in RCC Tissue According to Histopathology, Staging, and Grading

Size P = .03		Grade P = .02		Metastasis at diagnosis P = .006
<7 cm (45)*	≥7 cm (20)	Low† (42)	High (23)	M0 (46)	M (10)
Results are given as median (range). Wilcoxon test, P < .05 was considered statistically significant. RCC indicates renal cell carcinoma. * Numbers in parentheses represent the number of patients in each group. † Low grade is grade 1–2 and high grade is grade 3–4.
241 (10–79959)	413 (14–12812)	164 (14–6087)	470 (10–79959)	207 (10–12812)	1516 (86–79959)

For survival analysis the VEGF protein level cutoff value was chosen after quartile analysis and was the value at the 75th percentile of the values found in tumoral tissue for the 65 patients. The median follow-up was 72 months (range, 58–94 months). Univariate analysis of the survival curves demonstrated the benefit of lower levels of VEGF for EFS (P = .02) (Fig. 2). This parameter was not an independent prognostic factor in multivariate analysis.

Figure 2. Event-free-survival after surgery was estimated according to the VEGF levels in tumor extracts. The cutoff value for VEGF is the 75th percentile of VEGF levels in the tumor; P-value was calculated using the log-rank test.

Combined Laser Microdissection and QRT-PCR According to Metastatic and Nonmetastatic Status

For the quantification of each transcript we developed a highly specific and sensitive QRT-PCR test adapted to small quantities of laser-microdissected cells (Fig. 3). For all assays intra- and interrun variability (calculated from triplicate samples and comparing the results of samples in 10 different runs) showed an average standard deviation (SD) for the crossing points of 0.15 and 0.55, respectively. With this combined method the studied transcripts, VEGF₁₂₁ and its receptors VEGFR-1 and -2, were detected and quantified in all specimens of stromal and epithelial compartments from either tumoral or normal kidney.

Figure 3. VEGF₁₂₁, VEGFR-1, and VEGFR-2 mRNA calibration curves and agarose gel electrophoresis TaqMan probe polymerase chain reaction (PCR) products showing specificity of real-time reverse-transcriptase (RT-PCR). V9, V8, V7, V6, V5, V4, V3, V2, V1, and 1 correspond respectively to: 10⁹, 10⁸, 10⁷, 10⁶, 10⁵, 10⁴, 10³, 10², 10 and 1 copies of transcripts. For all the assays, intra- and interrun variability (calculated from triplicate samples and comparing the results of samples in 10 different runs) showed an average standard deviation (SD) for the crossing points of 0.15 and 0.55, respectively. The DNA marker used on agarose gel electrophoresis was pUC19 DNA/Msp1 (HpaII) Marker, 23 (Fermentas, France).

Comparing epithelial and stromal cells in tumoral and corresponding nontumoral areas (n = 22), VEGF₁₂₁ mRNA was significantly higher in tumoral epithelial (P = .007) and stromal cells (P = .016) than in corresponding nontumoral cellular populations. Similarly, VEGFR-1 mRNA was significantly higher in tumoral epithelial (P = .002) and stromal cells (P = .003) than in corresponding nontumoral cellular populations (Table 4). No difference in VEGFR-2 mRNA expression was found in laser-microdissected epithelial and stromal compartments from tumoral and nontumoral areas. Comparing epithelial and stromal compartments in tumors (n = 33), no significant difference was found either for expression of VEGF₁₂₁, VEGFR-1 or -2. Considering the patients with or without metastasis at diagnosis, no difference was observed regarding VEGF₁₂₁, VEGFR-1 and -2 expressions. In this group of 33 patients with RCC analyzed with combined laser microdissection and QRT-PCR, no association of VEGF₁₂₁, R1, and R2 mRNA values were found with the stage of the disease or survival.

Table 4. VEGF₁₂₁, VEGFR-1 and -2 Transcript Expression in Laser-Microdissected Compartments (Epithelium and Stroma) of Normal and Tumoral Renal Tissue According to Metastatic Status

		Whole population (n = 22)	M0 (n = 12)*	M+ (n = 10)
* M0 indicates nonmetastatic; M+, metastatic; TE, tumoral epithelial cells; NTE, nontumoral epithelial cells; TS, tumoral stromal cells; NTS, nontumoral stromal cells. † Numbers in parentheses represent the number of patients in each group. ^c Results are given as median (range). ‡ Wilcoxon test for paired data, P < .05 was considered statistically significant.
TE vs NTE	VEGF121	4.8 (0–125) vs 2.6 (0–8.6), P = .007	4.3 (0–125.0) vs 2.5 (0.1–8.6),†P = .03	6.0 (1.6–42.0) vs 3.5 (0–5.6), P = .11‡
	VEGFR-1	2.9 (0.1–197.5) vs 0.5 (0–10.2), P = .002	2.0 (0.05–197.5) vs 0.6 (0–3.4), P = .02	4.3 (0.4–68.6) vs 0.5 (0–10.3), P = .03
	VEGFR-2	9.5 (0–104) vs 8.3 (0–41.4), P = .51	11.4 (0.2–104) vs 10.2 (0–41.4), P > 0.99	9.5 (0–54.3) vs 2.8 (0.3–27.3), P = .14
TS vs NTS	VEGF121	6.9 (0.6–165.0) vs 1.0 (0–23.2), P = .02	8.0 (0.9–165.0) vs 0.5 (0–15.2), P = 0.02	6.9 (0.6–31.1) vs 2.6 (0–23.2), P = .40
	VEGFR-1	5.3 (0–435.0) vs 0.5 (0–54.7), P = .003	4.5 (0–435) vs 0.3 (0–54.7), P = .12	7.8 (0–108.7) vs 0.6 (0–4.7), P = .007
	VEGFR-2	6.5 (0–132.3) vs 1.3 (0–76.9), P = .33	6.5 (0–54) vs 1.3 (0–76.9), P = .87	9.8 (0–132.3) vs 1.3 (0–65.7), P = .14

Immunofluorescent Labeling of VEGF, VEGFR-1, VEGFR-2, and Cytokeratins

In tumoral areas (Fig. 4), VEGF was strongly expressed by tumor cells and by some stromal cells. Coexpression of VEGF and VEGFR-1 was observed in tumor cells, and interestingly, in stromal cells of tumoral areas. No cell expressed VEGFR-1 alone. Around 1 of 10 cytokeratin-positive cells coexpressed VEGF and VEGFR-1. In corresponding nontumoral tissue VEGF was seldom expressed. It was not found in glomerular or vascular areas, but in rare tubular sections. In nontumoral kidney, no coexpression of VEGF and VEGFR-1 was observed and no stromal cell was labeled. In tumoral and nontumoral areas (Fig. 5), no cell expressed VEGFR-2.

Figure 4. Expression of VEGF, VEGFR-1, and cytokeratin (AE1/AE3) in RCC (A and B) and in nontumoral areas (C). Triple immunofluorescence labeling of VEGFR-1, VEGF, and cytokeratins showed that cytokeratin-positive RCC cells coexpressed VEGF and VEGFR-1 (epithelial cells surrounded by broken lines in A and B); cytokeratin negative round (A) or fusiform (B) cells coexpressed also VEGF and VEGFR-1 (arrows). In the normal kidney, no cell whether tubular (T), glomerular (G) or interstitial expressed VEGFR-1. Few tubular cells expressed slightly VEGF. Scale bars = 10 μm.

Figure 5. Expression of VEGF, VEGFR-2 and cytokeratin (AE1/AE3) in (A) renal cell carcinoma (RCC) and in (B) nontumoral areas. Identical methods of triple immunofluorescence labeling for VEGFR-2, VEGF, and cytokeratins did not show any coexpression of VEGF and VEGFR-2 whether in (A) RCC or in (B) nontumoral renal tissue. T, tubular area; G, glomerular area. Scale bars = 10 μm.

DISCUSSION

In this study of 65 patients with RCC, with an 8-year follow-up, we have shown that VEGF protein levels in tumor cytosolic extracts were associated with tumor progression. VEGF protein level measure in RCC cytosolic extracts is an easy-to-perform and reliable test. Several groups have analyzed VEGF expression in RCC at the mRNA and/or at the protein level using immunohistochemistry or Western blot,17–21 but assessment of VEGF protein level in RCC tumor extracts has not been reported. The 2 quantitative techniques that could be used in clinical situations are QRT-PCR at the mRNA level and ELISA in cytosolic extracts at the protein level. In other types of cancer, particularly in breast cancer,22, 23 VEGF tumor content measured in cytosolic extracts has been shown to have a prognostic significance.

Considering mRNA expression, different groups have reported an increase in VEGF mRNA expression in RCC compared with normal renal tissue.24, 25 A higher tumor VEGF mRNA expression has been associated with a worse prognosis in different series of RCC.20, 21 As for VEGF isoforms a predominant expression of VEGF₁₂₁ and VEGF₁₆₅ transcripts has been reported in different series of RCC,25–27 together with a higher VEGFR-1 and R2 mRNA level in RCC compared with normal kidney in 1 series.27 We have extended these results using laser microdissection coupled with QRT-PCR with the analyses of VEGF and its receptor expression in microdissected RCC epithelial and stromal compartments. When compared with corresponding nontumoral renal cells, VEGF₁₂₁ and VEGFR-1, but not VEGFR-2, mRNA levels were higher in RCC epithelial and stromal cells. This suggests that epithelial as well as stromal cells contribute to the increased VEGF and VEGFR-1 expression in RCC. Along the same line, experimental data in a xenograft model of breast cancer demonstrate that the tumor growth inhibition is the strongest when human VEGFR-1, expressed by the tumor cells, and murine VEGFR-1, expressed by the stromal cells, are concomitantly inhibited.28

To further characterize the type of cell expression of VEGF and VEGFR-1, we used a triple immunofluorescence labeling on whole tissue sections. Interestingly, we observed a colocalization of VEGF and VEGFR-1 on some RCC tumor epithelial and stromal cells. Such a coexpression was not observed in any type of cells in normal kidney.

Taken together, these results are in accordance with recent studies demonstrating a major role for VEGFR-1 signaling in tumor progression.28–32 VEGFR-1 participates in the migration of neoplastic epithelial cells,11, 32 as do several others types of nonendothelial cells, such as hematopoietic stem cells and monocytes. For leukemic cells it has been shown that VEGFR-1 is essential for tumor cell growth via a VEGF/VEGFR-1 autocrine loop.33, 34 During progression of solid tumors hematopoietic progenitor cells expressing VEGFR-1 might initiate the premetastatic niche.31 The demonstration in a murine model of RCC of the inhibition of lung metastasis through administration of a soluble form of VEGFR-1 also underlines the role of VEGFR-1 in RCC progression.30

We have demonstrated that in RCC both epithelial and stromal cells contribute to VEGF and VEGFR-1 overexpression, and that VEGF overexpression is linked to tumor progression. These results contribute to a better knowledge of cell targets of anti-VEGF therapies. Along this line, further studies are needed to establish the value of VEGF protein assay in RCC cytosolic extracts for the patients who could benefit from anti-VEGF therapies.

Acknowledgements

The authors thank Marie-Pierre Podgorniak for technical support.