Expression of VEGF(xxx)b, the inhibitory isoforms of VEGF, in malignant melanoma

Abstract

Malignant melanoma is the most lethal of the skin cancers and the UK incidence is rising faster than that of any other cancer. Angiogenesis - the growth of new vessels from preexisting vasculature - is an absolute requirement for tumour survival and progression beyond a few hundred microns in diameter. We previously described a class of anti-angiogenic isoforms of VEGF, VEGF(xxx)b, that inhibit tumour growth in animal models, and are downregulated in some cancers, but have not been investigated in melanoma. To determine whether VEGF(xxx)b expression was altered in melanoma, PCR and immunohistochemistry of archived human tumour samples were used. In normal epidermis and in a proportion of melanoma samples, VEGF(xxx)b staining was seen. Some melanomas had much weaker staining. Subsequent examination revealed that expression was significantly reduced in primary melanoma samples (both horizontal and vertical growth phases) from patients who subsequently developed tumour metastasis compared with those who did not (analysis of variance (ANOVA) P<0.001 metastatic vs nonmetastatic), irrespective of tumour thickness, while the surrounding epidermis showed no difference in expression. Staining for total VEGF expression showed staining in metastatic and nonmetastatic melanomas, and normal epidermis. An absence of VEGF(xxx)b expression appears to predict metastatic spread in patients with primary melanoma. These results suggest that there is a switch in splicing as part of the metastatic process, from anti-angiogenic to pro-angiogenic VEGF isoforms. This may form part of a wider metastatic splicing phenotype.

Figure 1

Alternative splicing of the VEGF gene in the terminal exon results in two families of isoforms – the angiogenic VEGF_xxx and the anti-angiogenic VEGF_xxxb isoforms.

Figure 2

Biacore analysis of recombinant human VEGF₁₆₅b or VEGF₁₆₅ binding to the anti-VEGF₁₆₅b antibody. (A) VEGF₁₆₅b bound with relatively high affinity and dissociation constants (2.9 × 10⁴ M⁻¹ s⁻¹ and 0.011 s⁻¹, respectively). (B) Even at 180 nM,VEGF₁₆₅ showed no binding to the antibody.

Figure 3

Distal and proximal splice variants of VEGF are found in archival melanoma tissues. (A) Extraction of mRNA from 8- to 10-year-old archival sections of melanoma and surrounding skin followed by RT–PCR using primers in exon 7 and the 3′ UTR of the VEGF gene resulted in two products of 150 and 200 bp. These were consistent with VEGF₁₆₅b and VEGF₁₆₅, respectively. (B) Splice site-specific primers to the proximal splice site (PSS) and the distal splice site (DSS) confirmed expression of these isoforms in archival sections.

Figure 4

Expression of VEGF in normal skin. (A) Expression of VEGF determined by staining with antibodies that detect all isoforms (pro- and anti-angiogenic) of VEGF (pan-VEGF antibody). Staining is seen in epidermis (e) and weak diffuse expression in dermis (d) and around blood vessels (bv). Inset is negative control. (B) Expression of VEGF_xxxb in skin. Expression was seen in epidermis, with very weak staining in dermis. Blood vessels were positive for VEGF_xxxb. Scale bars are already given in the figure.

Figure 5

Expression of VEGF_xxxb in metastatic and nonmetastatic melanoma and surrounding skin. Expression of VEGF_xxxb was lower in horizontal (supratumoral) and vertical (intratumoral) growth phase metastatic melanoma than in nonmetastatic melanoma or normal skin. (A–C) Nonmetastatic, (D–F) metastatic melanoma. (A and D) Histologically normal skin >1 mm horizontally from tumour, (B and E) horizontal growth phase (supratumoral), (C and F) vertical growth phase (intratumoral). Scale bars 50 μm.

Figure 6

Staining of VEGF_xxxb but not that of VEGF is reduced in metastatic but not in nonmetastatic melanomas. (A) Mean±s.e.m. score from three independent observers blinded to metastatic status of staining intensity of images of melanoma samples. (B) Pan-VEGF was not altered in nonmetastatic melanoma, but, in contrast to VEGF_xxxb, appeared to increase in metastatic melanoma from normal through horizontal growth phase (HGP) to vertical growth phase (VGP). ^*=Nonmetastatic vs metastatic, VEGF_xxxb, P<0.0001 one-way analysis of variance (ANOVA), ^**=P<0.01 Bonferroni. ⁺=Compared with normal skin. VEGF_xxxb: metastatic, P<0.0001 one-way ANOVA, pan-VEGF: metastatic, P<0.05, ANOVA.⁺=P<0.05 ⁺⁺⁺=P<0.001, Dunnetts.

Figure 7

In metastatic melanomas, VEGF staining, but not VEGF_xxxb staining is upregulated. Although VEGF staining was clearly seen in peritumoral epidermis, and horizontal growth phase (HGP; supra) and vertical growth phase (VGP; intra), this was in contrast to VEGF_xxxb staining in serial sections. In nonmetastatic melanomas however, VEGF staining was similar to VEGF_xxxb staining in all three tissue types. (A–F) Metastatic melanoma, (G–L) nonmetastatic melanoma, (A–C and G–I) VEGF staining, (D–F and J–L) VEGF_xxxb staining, (A, D, G and J) normal skin, (B, E, H and K) HGP(supratumoral), (C, F, I and L) VGP (intratumoral). Scale bars=50 μm.

Figure 8

Relative levels of panVEGF to VEGF_xxxb staining expression are decreased in metastatic melanoma. The normalised ratio of VEGF_xxxb to VEGF staining was significantly reduced in metastatic compared with nonmetastatic melanoma (^*=P<0.05, ^**=P<0.01 compared with metastatic, n=9 metastatic, n=7 nonmetastatic).