Interleukin-13 and Interleukin-4 Induce Vascular Endothelial Growth Factor Release from Airway Smooth Muscle Cells

MATERIALS AND METHODS

RESULTS

IL-4 and IL-13 caused a concentration- and time-related increase in VEGF protein release from HASM cells (Figure 1), as reported by others (15). IL-4 induced VEGF release at concentrations as low as 0.3 ng/ml, whereas higher concentrations were required for IL-13 (Figure 1A), which is consistent with the relative efficacy of IL-4 and IL-13 for eotaxin release from HASM cells (20, 28).

To determine whether the effects of Th2 cytokines occurred at the transcriptional level, we measured VEGF mRNA expression by real-time PCR. There was constitutive VEGF mRNA expression in HASM cells, and IL-4 caused a concentration-dependent increase in VEGF mRNA expression. A significant increase in VEGF mRNA was observed with IL-4 doses as low as 3 ng/ml, although a trend could be observed even with 1 ng/ml (Figure 2A). IL-13 also increased VEGF mRNA expression (Figure 2A). IL-4 had no effect on VEGF promoter activity even though cells from the same donors treated at the same time with oncostatin M (OSM) showed a significant increase in VEGF promoter activity (Figure 2B) as previously described (19). To determine whether IL-4 modulated VEGF mRNA stability, we measured VEGF mRNA expression in cells treated or untreated with IL-4 for 6 h. Actinomycin D was added to the cells to inhibit transcription, and VEGF mRNA was measured at 0, 1, 2, and 4 h after the addition of actinomycin D. In cells treated with IL-4, VEGF mRNA had a longer half life (273 ± 22 min) than untreated cells (162 ± 39 min) (P < 0.05) (Figure 3).

Great variability in VEGF release was observed among cells from different donors after IL-13 or IL-4 stimulation. To evaluate the role of IL-4Rα genotype in this variability, we assayed VEGF in supernatants of HASM cells from donors chosen for their IL-4Rα genotype and previously stimulated with IL-4 or IL-13 (22). Each cell donor was studied in duplicate on one or two occasions. Although we have previously reported a marked increase in TARC release induced by TNF-α plus IL-4 or IL-13 in cells from donors with at least one Val50/Pro478/Arg551 allele, we observed no effect of IL-4Rα genotype on IL-4– or IL-13–induced VEGF release (Table 1). Two different conventions have been used for numbering amino acids in IL-4Rα, beginning from the start of the signal peptide or from the start of the mature protein. We use the latter method.

Because numerous SNP have also been described in the VEGF gene, we sought to determine whether the observed donor-related differences in VEGF release were dependent on the VEGF genotype of the cells. HASM cells were genotyped for −460T/C, −160C/T, −152G/A SNP in the promoter region; for +405C/G in the 5′-UTR; and for +936C/T in the 3′-UTR of the VEGF gene. Table 2 lists the VEGF genotypes of the 21 cell donors we examined. The VEGF allele frequencies in our bank of 59 HASM cell donors were as follows: −460C 0.52, −160T 0.0, −152A 0.50, +405G 0.61, and +936T 0.14. The frequencies were consistent with the literature (24, 25, 27). For example, others have reported that −460C and 152A are in strong linkage disequilibrium (26, 27). All cell donors we studied who had at least one −460C allele also had at least one −152A allele; these cells were designated as −460C/−152A. Others have also reported linkage disequilibrium between −460C and +405G polymorphisms (24, 27), which was confirmed by our data (see below).

Cells were first stratified according to their genotype for +405C/G (Figure 4A) without regard for the other SNP. ANOVA indicated no effect of +405C/G genotype on baseline or IL-13–, IL-4–, or TNF-α–induced VEGF release. Cells were stratified according to their genotype at −460T/C and −152G/A (Figure 4B). ANOVA indicated a significant difference among cells from donors with −460TT/−152GG, −460TC/−152GA, and −460CC/−152AA genotypes. Because cells from donors heterozygous or homozygous for the −460C/−152A allele had similar behavior, these two genotypes were grouped together and compared with −460TT/−152GG cells. The presence of at least one −460C/−152A allele significantly decreased VEGF release induced by IL-13 (3 ng/ml) (P = 0.004) or IL-4 (1 ng/ml) (P = 0.012) (Figure 4B). Similar results were obtained with higher doses of IL-4 (10 ng/ml) or IL-13 (30 ng/ml) (data not shown). Presence of the −460C/−152A allele also tended to decrease basal and TNF-α–stimulated expression of VEGF by HASM cells, although the differences were not significant (P = 0.08 and P = 0.229, respectively). The −460/−152/+405 haplotype was then analyzed (Figure 5). In our sample, there were no cells wild-type at +405 that had at least one −460C/−152A allele. Thus, only three −460/−152/+405 haplotypes were observed. ANOVA indicated a significant difference among haplotypes after treatment with IL-13 or IL-4. Cells with at least one +405G allele that were wild type at −460/−152 had similar VEGF release to cells that were wild type at +405 and −460/−152. Therefore, cells with these two haplotypes were grouped together and compared with cells with at least one −460C/−152A and at least one +405G allele. The presence of at least one −460C/−152A and at least one +405G allele led to significantly lower IL-4– and IL-13–induced VEGF release than cells with the other haplotypes (Figure 5). Although the same trend was observed in untreated cells and in cells treated with TNF-α, the difference was not significant. We considered the possibility that the genotype effect might be the result of some unperceived bias in the status of the cells or of some unmeasured common genetic variation in another aspect of the IL-4/IL-13 signaling pathway. We reasoned that such effects would likely affect not just changes in VEGF release induced by IL-4 or IL-13, but also other effects of these cytokines. Therefore, we measured MCP-1 in the same cell supernatants because IL-4 and IL-13 have also been shown to induce MCP-1 release (18). MCP-1 release was measured in untreated cells and in cells treated with IL-4, IL-13, or TNF-α. Results were stratified according to −460/−152/+405 haplotype. IL-13, IL-4, and TNF-α each increased MCP-1 release, but ANOVA did not show any difference in MCP-1 release among −460/−152/+405 haplotypes (data not shown), indicating that the effect of VEGF genotype was specific to VEGF.

The 3′-UTR of the VEGF gene participates in post-transcriptional regulation of VEGF expression by stabilization of its mRNA (30), and the +936 C/T mutation in the 3′-UTR has been associated with VEGF plasma levels (25). To examine the importance of +936C/T SNP at the 3′-UTR of the VEGF gene on IL-13–, IL-4–, and TNF-α–induced VEGF release, HASM cells were stratified according to +936 C/T genotype. The +936 C/T genotype had no impact on VEGF release (Figure 6), but because we only had heterozygotes in our population, we cannot rule out the possibility that homozygosity for +936T could have an impact.

DISCUSSION

Our data indicate that the ability of Th2 cytokines to induce VEGF from HASM cells is likely mediated at the level of VEGF mRNA stability. To our knowledge, this is the first report of an effect of Th2 cytokines on VEGF mRNA stability in any cell type. Our data also indicate that VEGF genotype modulates the ability of IL-4 and IL-13 to induce VEGF release from HASM cells. This is the first report of a role for VEGF genotype in these events in any cell type.

Our data confirm reports of others that IL-13 and IL-4, Th2 cytokines that play an important role in asthma, induce VEGF release from HASM cells (15, 18). We also report that IL-4 and IL-13 increased IL-4 mRNA expression (Figure 2A), an effect not previously reported in any cell type. Although IL-4 increased VEGF mRNA, IL-4 did not alter VEGF promoter activity (Figure 2). The absence of any effect at the promoter level could be the result of the reporter construct used. Although the construct contained 2.4 kb of the VEGF promoter and was activated by another stimulus (OSM), we cannot rule out the possibility that IL-4 activates elements upstream of this or exerts its effects on intronic elements in the VEGF gene. On the other hand, IL-4 increased VEGF mRNA stability (Figure 3), suggesting that effects of IL-4 on VEGF mRNA are post-transcriptional rather than transcriptional. The VEGF mRNA 3′-UTR, like many labile mRNAs, contains AU-rich elements that are associated with rapid mRNA turnover (30, 31). Consequently, increases in VEGF mRNA stability would be expected to exert profound effects on the expression of this gene. Indeed, hypoxia, one of the most potent stimuli for VEGF expression, increases VEGF release in part through stabilization of VEGF mRNA (32, 33). Previous reports have demonstrated the presence of binding sites for hypoxia- or stress-inducible proteins in the VEGF mRNA 3′-UTR (30), but the molecular mechanisms underlying VEGF mRNA stability remain poorly understood.

Our data indicate that the VEGF genotype of the cells substantially affected IL-4− and IL-13–induced VEGF release. In particular, we found a decrease in IL-13– and IL-4–induced VEGF release in the presence of the −460C/−152A/+405G haplotype (Figure 5), whereas the genotype of the cells at +405 considered independently of −460/−152 did not have an impact (Figure 4A). This effect was not a function of some unperceived bias in the status of the cells or of some unmeasured common genetic variation in some component of the IL-4/IL-13 signaling pathway because when cells were analyzed for another IL-4/IL-13–induced effect (MCP-1 release) and the results stratified according to their VEGF promoter genotype, no effect was observed.

The −460 T/C and +405 C/G SNP have been implicated in a number of diseases with an angiogenic basis, although the results have not been consistent (24, 26, 27). For example, +405C has been associated with diabetic retinopathy, psoriasis, and higher serum VEGF levels (26). In contrast, others have reported that +405G allele is associated with higher VEGF protein production by peripheral blood mononuclear cells stimulated with lipopolysaccharide (24). One possible explanation for this apparent discrepancy in the impact of +405C/G is that there are cell-type– or stimulus-dependent effects of VEGF SNP. Alternatively, linkage disequilibrium may account for these apparently disparate observations. For example, our data indicate that the presence of the −460C/−152A allele decreases the ability of HASM cells to release VEGF in response to IL-4 or IL-13. −460C always occurred with +405G, and cells with at least one –460C/−152A/+405G allele also had reduced VEGF expression (Figure 5), whereas we could not observe any effect of +405C/G SNP alone.

Our results indicate no effect of IL-4 on VEGF promoter activity (Figure 2). Rather, IL-4 seemed to act by increasing VEGF mRNA stability. Therefore, we do not think that the impact of the −460C/−152A allele (Figure 5) on IL-4– and IL-13–induced VEGF release was related to effects on VEGF promoter activity. It is more likely that the reason for the decrease in IL-4– and IL-13–induced VEGF release from cells bearing the −460C/−152A/+405G haplotype is linkage disequilibrium with some other SNP in the 3′-UTR, a region important for stabilization of VEGF mRNA (30, 32). Strong linkage disequilibrium has been reported across the VEGF gene (24, 26, 27) and was confirmed in this group of HASM cell donors. Because the +936T allele in the 3′UTR has been associated with reduced VEGF plasma levels (25, 34), we examined its impact on IL-4– and IL-13–induced release in these HASM cells. Our results indicated no impact of the presence of the +936T allele in these cells, although because we had only heterozygous cells in our population, we cannot rule out the possibility that this SNP has the capacity to affect VEGF release. It is more likely that the reduction in VEGF in cells bearing at least one −460C/−152A/+405G allele is the result of linkage disequilibrium between this haplotype and another undescribed mutation elsewhere in the VEGF gene.

Although the presence of the −460C/−152A/+405G haplotype accounted for some of the variability in IL-13– and IL-4–induced VEGF release (Figures 4 and 5), substantial variability in VEGF release remained even in cells that were −460TT/−152GG (Figure 4B, open symbols). In HASM cells, IL-4 and IL-13 act through the type II IL-4 receptor, a dimer consisting of the IL-4Rα and the IL-13RαI (20, 21). We considered the possibility that SNP in the IL-4Rα gene accounted for some of this variability but found no evidence to that effect. We were somewhat surprised by these results because we have previously demonstrated that the Val50/Pro478/Arg551 haplotype of the IL-4Rα is associated with increased IL-13– and IL-4–induced TARC expression (22) and increased IL-13–induced β-adrenergic desensitization in HASM cells (23). This difference is likely related to differences in the regulation of VEGF, TARC, and β-adrenergic responsiveness by Th2 cytokines. It is possible that genetic variation in STAT-6 or in other components of the IL-4/IL-13 signaling pathway account for some this remaining variability or that there are other unidentified SNP in the VEGF gene that contribute to this variability. There may also be other sources of variability, including heterogeneity in VEGF release among cells within a single culture.

The importance of VEGF for asthma is underscored by the observations that inhibiting VEGF reduces lung inflammation and airway hyper-responsiveness in mice and that overexpression of VEGF enhances Th2 inflammation and induces airway hyper-responsiveness (6, 8). Given the ability of IL-4 and IL-13 to induce VEGF and the important role played by Th2 cytokines in the asthmatic airway, it is likely that IL-4 and IL-13 contribute to the increased VEGF expression observed in asthma. Although the magnitude of the change in VEGF release induced by Th2 cytokines may seem small (2- to 3-fold; see Figure 1), it is similar to that reported by Wen and colleagues (15) and similar to that induced by the potent proinflammatory cytokine, TNF-α (Figure 4). Data with another potent proinflammatory cytokine, IL-1β, indicate comparable levels of induction of VEGF (15, 19) in HASM cells. Even hypoxia, one of the most potent stimuli for VEGF expression, induces only a 5-fold increase in VEGF release by lung epithelial cells (11).

Although it is likely that IL-14 and IL-13 contribute to VEGF release in the airways of patients with asthma, we do not know to what extent VEGF released from airway smooth muscle is important. However, the basal VEGF release from HASM cells found in this study is comparable to the basal release of VEGF from lung epithelial cells (11). Moreover, VEGF release from HASM cells has the capacity for autocrine effects because VEGF alters the expression of extracellular matrix proteins by HASM cells (16).

In summary, this study indicates that IL-4 and IL-13 induce VEGF release in airway smooth muscle cells likely at the level of VEGF mRNA stability. In addition, the ability of HASM cells to release VEGF when stimulated with IL-4 or IL-13 is strongly dependent on VEGF haplotype, but it is not affected by IL-4Rα genotype. VEGF is a potent angiogenesis factor released in the asthmatic airway. Our data suggest that VEGF haplotype can have an impact on airway remodeling in asthma.