Neuropilin 1 and neuropilin 2 co-expression is significantly correlated with increased vascularity and poor prognosis in nonsmall cell lung carcinoma




Cell-retained isoforms of vascular endothelial growth factor A (VEGF-A) have been reported to play an essential role in tumor progression through stromal neovascularization in malignant solid tumors. While more than 95% of nonsmall cell lung carcinoma (NSCLC) expresses cell-retained VEGF-A isoform, the clinicopathologic implications of neuropilin (NRP), considered the specific receptor for limited types of VEGF-A isoform, are not well understood.


The authors examined NRP1 and NRP2 mRNA expression in 68 NSCLCs and 15 extraneoplastic tissues by a densitometry–assisted, semi-quantitative reverse transcription-polymerase chain reaction. The authors determined the distinct expression of NRPs using the expression level of NRPs relative by optical density to β2-microglobulin. The authors also investigated VEGF-A isoforms, their receptors, and the clinical implications. Vascularity of NSCLC was morphologically estimated on sections immunostained with anti-CD34 antibody.


Eleven of 15 extraneoplastic specimens showed NRP1 expression (73.3%) and 8 showed NRP2 expression (53.3%). The expression level of NRP1 or NRP2 of neoplasmic tissue was higher than that of extraneoplastic tissues (P < 0.01, Mann-Whitney U test). Fifty-five and 44 NSCLCs expressed NRP1 and NRP2, respectively. Forty patients co-expressing NRP1 and NRP2 showed significantly poorer prognosis and increased vessel counts as compared to those 28 cases without co-expression (P < 0.05, log-rank test; P < 0.05, Mann-Whitney U test).


The co-expression of NRP1 and NRP2 genes is significantly correlated with tumor progression through neovascularization in NSCLC. These results suggest that both NRP1 and NRP2 are key molecules for stromal vascularization by cell-retained VEGF in NSCLC. Cancer 2002;95:2196–201. © 2002 American Cancer Society.

DOI 10.1002/cncr.10936

Vascularization is an important factor for solid tumor growth and metastasis. Most tumors require neovascularization when they reach 2–3 mm.1 Vascular endothelial growth factor–A (VEGF-A) is a major regulator in the induction of endothelial cell proliferation and in increases in vascular endothelium permeability.2, 3 In humans, six isoforms of VEGF-A (VEGF121, VEGF145, VEGF165, VEGF183, VEGF189, and VEGF206) generated by alternative splicing and ranging in length from 121 to 206 amino acids have been characterized and five receptors have been described. VEGF-A has been shown to play a role in angiogenesis by a paracrine mechanism. The isoforms of VEGF-A are characterized by different properties in tumor progression and angiogenesis.4 We previously reported that the expression profiles of VEGF-A isoforms differ with tumor type and are closely associated with the progression of various malignant tumors, including colon carcinoma, nonsmall cell lung carcinoma (NSCLC), renal cell carcinoma, and osteosarcoma.5–10 Two conventional tyrosine kinase type VEGF-A receptors, two neuropilins (NRPs), and heparan sulfate proteoglycans have been reported to differ in binding to the different VEGF-A isoforms.4, 11–13 Fms-like tyrosine kinase-1 (Flt-1, VEGFR-1) and kinase domain-containing receptor (KDR, VEGFR-2) are VEGF-A receptors expressed in endothelial cells in vivo.

Neuropilin was the first discovered neuronal cell guidance receptor of the collapsin/semaphorin family and is a 130–140 kDa cell-surface glycoprotein. Semaphorin is a repellent factor of growing axon tips.14–17 Two subtypes, NRP1 and NRP2, have been isolated. They also show differences in their interactions with semaphorins (Sema): NRP1 binds Sema 3A/Sema III, Sema 3F/Sema IV, and Sema 3C/Sema E, while NRP2 binds Sema 3F/Sema IV and Sema 3C/Sema E but not Sema 3A/Sema III. There is an intimate interplay between NRP1 and NRP2 as neuron guidance factors.14–16, 18–20 The human NRP1 locus is on chromosome 10, 6061 cR distal from AFM353TB5, and NRP2 is on chromosome 2, 0.7 cR proximal to WI-3694.21 Neuropilin was also shown to be an isoform-specific receptor for VEGF-A.22 These two VEGF-A receptors, NRP1 and NRP2, have highly similar structural features;18 NRP1 has been reported to be associated with the development of the cardiovascular system, while the role of NRP2 in vascularization is unknown.23

We reported previously that NRP2 was associated with vascularization and prognosis in 30 osteosarcoma patients.24 It is unclear how NRP expression is involved in the progression of NSCLC through vascularization. In the current study, we analyzed expression of the NRP1 and NRP2 genes and their clinicopathologic significance in NSCLC.



Sixty-eight NSCLC specimens were obtained at surgical resection from October 1989 to December 1993 with the patients' informed consent. Tissues were frozen and stored at −80 °C until analyses. Total cellular RNA was prepared from the frozen specimens by standard procedures. Surgical specimens were also processed for routine histopathologic analysis to estimate pathologic stage. The 68 NSCLC specimens comprised 44 adenocarcinomas, 21 squamous cell carcinomas, 2 large cell carcinomas, and 1 adenosquamous carcinoma. Corresponding extraneoplastic lung tissues from 15 NSCLC patients (8 adenocarcinomas, 4 squamous cell carcinomas, 2 large cell carcinomas, and 1 adenosquamous carcinoma) were also examined. The pathologic stages of the NSCLCs were as follows: Stage I, 38 patients; Stage II, 6 patients; Stage IIIa, 21 patients; Stage IIIb, 2 patients; and Stage IV, 1 patient.

NRP1 and NRP2 Gene Expression

We evaluated NRP1 and NRP2 mRNA expression by reverse transcription polymerase chain reaction (RT-PCR) according to previously described procedures.25 The primer sets used for NRP1 were as follows: NRP1-S, 5′-GAAAAATGCGAATGGCTGAT-3′ and NRP1-A, 5′-AATGGCCCTGAAGACACAAC-3′; and for NRP2: NRP2-S, 5′-CAAACACTGTGGGAACATCG-3′ and NRP2-A, 5′-TTCTCAGGAAACCCAGGAGA-3′.

Probes (NRP1: 200 bp, NRP2: 198 bp) were prepared by PCR amplification with primers NRP1-S, A and NRP2-S, A, and their sequences were confirmed with an automated sequencer (ABI PRISM 310, Perkin Elmer, Foster City, CA). Reverse transcription was performed at 42 °C for 60 minutes (1 μg total cellular RNA; 100 pM random primers, Boehringer Mannheim, Mannheim, Germany; reverse transcriptase, Gibco-BRL, Life Technologies, Inc., Gaithersburg, MD). Neuropilin-1 and NRP2 cDNA fragments were amplified by 25 rounds of PCR consisting of 1 minute at 94 °C, 1 minute at 58 °C, and 2 minutes at 72 °C with a Gene Amp PCR System 9600 (Perkin Elmer) and Taq DNA polymerase (TOYOBO, Osaka, Japan). Blots of products (Zeta-Probe; Bio-Rad, Hercules, CA) were hybridized with photochemically labeled probes (ECL; Amersham Life Science Ltd., Buckinghamshire, UK), and exposed to Kodak AR film. The quality of the RNA was estimated by RT-PCR for β2microglobulin (β2m). Semi-quantitative RT-PCR was performed according to previously reported procedure.26–28 The specific products of NRP1 and NRP2 using 1 μg of RNA templates were amplified exponentially by 20–30 cycles, and the slope of RT-PCR products using 250 ng–1 μg of RNA templates showed a linear correlation with 25 amplification cycles. The expression level of β2m was considered to be the mRNA level of each specimen. We used β2m as an internal control for NRP expression. The levels of NRP1 and NRP2 gene expression were estimated by densitometry (Interactive Build Analysis System, Carl Zeiss, Oberkochen, Germany). Relative expression level was calculated by the ratio A1/A0, where A1 and A0 are the optical densities of the bands corresponding to NRP1, NRP2, and β2m, respectively, in each specimen to determine cut-off level.

Gene Expression of VEGF-A Isoforms and Their Receptors

We reviewed the gene expression of VEGF-A, expression patterns of VEGF-A isoforms, and Flt-1 and KDR expression in NSCLC patients by RT-PCR analysis according to previously described procedures.5, 7, 9, 10

Vascularization in NSCLC

Estimation of vascularity was performed as described previously.12 Formalin-fixed (10%), paraffin-embedded sections of the tumor tissue were immunohistochemically examined with murine anti-human CD34 monoclonal antibody (NCL-end, Novo Castra Lab, Newcastle upon Tyne, UK). After blockage of endogenous peroxidase activity (methyl alcohol, 3% H2O2) and nonspecific binding (10% normal goat serum), specimens were incubated with the antibody (1:20) at room temperature for 60 minutes. Sections were serially incubated with biotin-labeled anti-murine immunoglobulin G and horseradish peroxidase-conjugated streptavidin (Nichirei, Tokyo, Japan). Reaction products were visualized with 3,3′-diaminobenzidine. Light microscopy was used to identify the three regions within or immediately adjacent to the tumor containing the highest numbers of vessels. The vessel counts were evaluated at ×200 magnification (×20 objective and ×10 ocular, 0.739 mm2 per field) using a computerized image analysis system (Interactive Build Analysis System).

Statistical Analysis

The Fisher exact test or the χ–square test was applied for comparisons between group frequencies. P < 0.05 was considered significant. Differences in survival between subgroups of patients were compared with the log-rank test, and survival curves were plotted according to the method of Kaplan-Meier. The statistical significance of differences in mean vessel counts between groups was examined by the Mann-Whitney U test. Data are showed as means ± standard error.


Gene Expression of NRP1 and NRP2

The relative expressions of NRP1 and NRP2 genes were shown at various levels in NSCLC (1.04 ± 0.03, 0.92 ± 0.044, respectively). The expression levels of NRP1 and NRP2 in corresponding extraneoplastic lung tissue were 0.33 ± 0.04 and 0.16 ± 0.03 (Fig. 1). Expression levels of both NRP1 and NRP2 genes in neoplastic tissue were higher than extraneoplastic lung tissue (P < 0.01, Mann-Whitney U test). We classified NRP1+ or NRP2+ as relative expression levels of more than 0.5 for cut-off level. Fifty-five of 68 NSCLC specimens showed NRP1+ gene expression (80.9%), and 44 showed NRP2+ gene expression (64.7%).

Figure 1.

Neuropilins (NRPs) and β2-microglobulin (β2m) gene expressions in nonsmall cell lung carcinoma (NSCLC). NSCLC, lanes 1–7; extraneoplastic lung tissue, lanes 8–11.

VEGF-A Gene Expression and VEGF-A Receptors

Sixty-five of 68 NSCLC patients showed VEGF-A expression. As we reported previously, VEGF-A isoform patterns are divided into three types: type 1, VEGF-A121; type 2, VEGF-A121/165; and type 3, VEGF-A121/165/189. Six NSCLCs were classified as type 2, and 59 NSCLCs were type 3. Three NSCLCs did not show any VEGF-A121, 165, 189 gene expression. Fifty-one NSCLCs were positive for Flt-1 gene expression, and 65 were positive for KDR gene expression (Table 1). There was no significant correlation between NRP gene expression and VEGF-A receptors.

Table 1. Univariate Analysis of the Associations Between NRP1 and NRP2 Gene Expression and Patient or Tumor Characteristics in NSCLC
Variablea+/+b−/−P value
  • a

    +/+: Co-expression of NRP1 and NRP2.

  • b

    −/−: No expression of NRP1 and/or NRP2.

  • NRP: neuropilin; NSCLC: non small cell lung carcinoma; Flt: Fms–like tyrosine; KDR: kinase domain–containing receptor; VEGF: vascular endothelial growth factor.

Flt-1 expression    > 0.999
KDR expression    0.564
VEGF-A121 expression    0.065
VEGF-A165 expression    0.065
VEGF-A189 expression    > 0.999
Type 2 VEGF-A121/165    0.389
Type 3 VEGF-A121/165/189    > 0.999
TNM-staging    0.464
 II, IIIa, IIIb, IV1623.5%1420.6% 
T factor    0.595
 T2, T3, T42942.6%1826.5% 
N factor    0.803
 N1, N2, N31623.5%1014.7% 
Metastasis    0.166

NRP Expression and VEGF-A Gene Expression

We divided the 68 NSCLC patients into two groups. One group consisted of 40 patients with co-expression of NRP1 and NRP2 (NRP1+/NRP2+). The other was composed of 28 patients (NRP1− and/or NRP2−), including 15 patients positive for NRP1 gene expression but negative for NRP2 expression (NRP1+/NRP2−), 4 patients negative for NRP1 expression but positive for NRP2 expression (NRP1−/NRP2+), and 9 patients negative for both NRP1 and NRP2 expression (NRP1−/NRP2−). The co-expression of NRP1 and NRP2 did not show apparent correlations with VEGF-A121 and VEGF-A165 gene expression (P = 0.065, χ–square test; Table 1). TNM-stage and TNM-factors showed no correlation with NRP expression.

Clinical Implications of NRP Expression

The post-surgical treatment observation period ranged from 0.8 to 132.9 months (55.4 ± 4.8 months). There were significant differences in prognosis between NRP1+/NRP2+ patients and NRP1− and/or NRP2− patients (P = 0.035, log-rank test). The patients with co-expression of NRP1 and NRP2 showed significantly poorer prognosis than the other groups (Fig. 2).

Figure 2.

Survival curves for the nonsmall cell lung carcinoma patients. Neuropilin (NRP1)+/NRP2+ patients (solid line, n = 40) showed significantly poorer prognosis than NRP1− and/or NRP2− patients (broken line, n = 28; P = 0.0353, Kaplan-Meier, log-rank test).

Vascularization and NRP Expression

The mean vessel count of patients with NRP1 and NRP2 co-expression was 110.8 ± 17.0; in patients without co-expression the mean vessel count was 66.5 ± 10.3. The mean vessel area of patients with NRP1 and NRP2 co-expression was 4.1 ± 0.6; in those without co-expression the mean vessel area was 4.24 ± 0.5. The patients with NRP1 and NRP2 co-expression showed significantly increased vessel counts as compared to the other groups (P = 0.0348, Mann-Whitney U test, Fig. 3).

Figure 3.

Vascularization in nonsmall cell lung carcinoma shown by immunostaining for CD34. (A) Neuropilin (NRP1)+/NRP2+ showed significantly increased vascular density. (B) NRP1− and/or NRP2− showed moderate vascular density. Original magnification ×126.


As a receptor of the VEGF-A family, NRP1 serves as a receptor for VEGF165 and placenta growth factor-2 (PlGF-2), while NRP2 acts as receptor for VEGF165, VEGF145, and PlGF-2. NRP1 and NRP2 show differences in binding to isoforms of VEGF-A as compared with other receptors, including flt-1, KDR, and flt-4. Neither NRP1 nor NRP2 bind VEGF121.11 NRP1 acts as a co-receptor to enhance the binding of VEGF165 to KDR and VEGF165 chemotactic and mitogenic activity in a human melanoma cell line. Blocking VEGF165's access to NRP1 substantially inhibits this binding and the mitogenic activity of VEGF165 for human umbilical vascular endothelial cells.22 The domain of VEGF-A encoded by exon 7 is necessary for binding to NRP1. NRP2 has been suggested to have similar biologic features to NRP1,18 but its interaction with NRP1 in regulation of VEGF-A binding and operation has not been well characterized.

Neuropilin–1 and NRP2 are widely expressed in tissues, including lung.18, 22 In the current study, we evaluated the various expression levels of NRP genes with densitometry and their ratios. The expression level of NRPs in NSCLC was higher than the level of NRPs in corresponding extraneoplastic lung tissue. Brambilla et al. suggested that VEGF acts through NRP1 and NRP2 by showing the different localizations of Sema 3F in different stages of NSCLC. Lower levels of Sema 3F correlated with higher stage in NSCLC. VEGF might compete with Sema 3F for binding to NRP1 and NRP2.29 In vitro Sema 3F especially binds NRP2 with higher affinity than NRP1.18 The NSCLCs with co-expression of NRP1 and NRP2 showed significantly increased vascularity and poorer prognosis than those without co-expression of these molecules. Although the mechanism of the interaction of NRP1 and NRP2 is unclear, these results suggest that there is intimate cooperation between these molecules. To our knowledge, this is the first study to show that NRP1 and NRP2 gene expression are correlated with vascularity and prognosis in NSCLC.

Vascular endothelial growth factor 165 and VEGF121 were co-expressed in 95.6% of the NSCLCs examined. NRP1 and NRP2 were characterized as receptors for VEGF165, but not for VEGF121.11, 22 NRP1 is also known to be a co-receptor for VEGF-A receptor (KDR, VEGFR-2). A neuropilin-binding VEGF splice variant was suggested by Bachelder et al. as enhancing the metastasis of breast carcinoma cells without KDR expression.30 That study agreed with the current results regarding the lack of a significant correlation between KDR and NRP expression. While NRP1 and NRP2 were not significantly correlated with KDR in NSCLC, the current results support the possibility that NRP1 and NRP2 act as VEGF165 receptors and/or co-receptors for KDR.22 Vascular endothelial growth factor 189 expression was noted in NSCLCs (86.8%), but no significant correlation was apparent with NRP1 or NRP2 gene expression.

In conclusion, NRP1 and NRP2 expression cooperatively promotes angiogenesis and results in poorer prognosis in NSCLC with VEGF-A expression. NRP1 and NRP2 gene expression should be helpful for predicting the prognosis of patients with NSCLC.


The authors thank Mr. Yuichi Toda and Mr. Johhu Itoh for technical assistance.