INTRODUCTION

Vascular endothelial growth factor (VEGF),¹ also known as vascular permeability factor and vasculotropin, is a potent, multifunctional cytokine that exerts several important actions on vascular endothelium (for review, see Ref. 1). VEGF was originally discovered as a tumor-secreted protein that rendered venules and small veins hyperpermeable to circulating macromolecules (2). Subsequent purification and molecular cloning revealed VEGF to be a 34-42-kDa homodimeric, heparin-binding glycoprotein that is secreted by various cell types including keratinocytes (3, 4, 5, 6, 7, 8, 9, 10, 11). Four different VEGF isoforms have been identified which arise from alternative splicing of a single transcript (12, 13). The two larger variants, VEGF₁₈₉ and VEGF₂₀₆, remain cell-associated whereas the two smaller forms, VEGF₁₂₁ and VEGF₁₆₅, are secreted (13, 14). Two transmembrane type III tyrosine kinases, KDR (15, 16, 17) and flt-1 (18), have been identified as high affinity VEGF receptors. They are expressed predominantly, if not exclusively, by vascular endothelial cells. Overexpression of VEGF and both of its receptors has been documented in a number of animal and human tumors (for review, see Refs. 1 and 19), and VEGF is thought to play an important role in the increased vascular permeability and angiogenesis associated with these malignancies. VEGF is also overexpressed by epidermal keratinocytes in certain non-neoplastic processes of the skin which are characterized by increased microvascular permeability and angiogenesis, e.g. cutaneous wound healing (9) and psoriasis (20). Recently, up-regulation of VEGF was also found in bullous pemphigoid, erythema multiforme, and dermatitis herpetiformis (21), which are all associated with hyperpermeable dermal microvessels.

These findings led us to hypothesize that increased expression of VEGF might also play an important role in the development of sunlight-induced erythema. Exposure to solar ultraviolet B (UVB) radiation is known to induce multiple inflammatory and carcinogenous reactions in the skin (for review, see Ref. 22). Upon UVB irradiation, keratinocytes express several proinflammatory cytokines, including tumor necrosis factor  and interleukin 1 and  (IL-1 and IL-1) (23, 24, 25), which promote reactions of other effector cells in the skin, e.g. leukocytes and endothelial cells. Here we show that low dose UVB irradiation leads to a strong increase in VEGF mRNA and protein levels in human keratinocytes in vitro. Since this increase is not inhibited by antioxidants, reactive oxygen species are unlikely to be involved in the UVB-dependent up-regulation of VEGF. Experiments with cycloheximide and with conditioned medium from UVB-treated cells suggest that the induction of VEGF expression by UVB is indirect and depends on the release of soluble factors that subsequently turn on VEGF expression.

MATERIALS AND METHODS

The human keratinocyte cell line HaCaT (26) was maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin (100 units/ml) and streptomycin (100 µg/ml). For experiments with H₂O₂, keratinocytes were grown in RPMI 1640 medium supplemented with the same components as above. This medium does not contain iron salts, which are known to promote the decomposition of H₂O₂. In both cases, cells were grown to confluence without changing the medium and rendered quiescent by a 16-h incubation in serum-free medium. Cells were then incubated for varying periods in serum-free medium containing H₂O₂, pyrrolidine dithiocarbamate (PDTC), cycloheximide, or conditioned medium from UVB-treated cells. Dulbecco's modified Eagle's medium, RPMI 1640, and fetal bovine serum were purchased from Life Technologies, Inc. PDTC, cycloheximide, and H₂O₂ were from Sigma.

For UVB treatment, confluent and quiescent cells were washed twice with phosphate-buffered saline, irradiated while under a thin film of phosphate-buffered saline, and replenished with their own medium thereafter. The UVB source was a parallel bank of three TL20/12 fluorescent tubes (Philips, Hamburg) emitting a continuous spectrum between 280 and 320 nm with a peak emission at 312 nm. Fluence rate at the site of cell irradiation was 0.86 milliwatts/cm² as measured with a Centra radiometer (Osram, Munich). The UVB doses employed ranged from 5 to 20 mJ/cm² (exposure time: 7-26 s) and were sublethal for the cells (data not shown). These doses are a realistic representation of the irradiation reaching basal keratinocytes in vivo (27). Aliquots of cells were harvested before and at different time points after UVB irradiation and used for RNA isolation.

RNA isolation was performed as described (28). RNase protection assays were carried out as published recently (29). Briefly, a 159-base pair fragment corresponding to nucleotides 339-498 of the human VEGF₁₂₁ cDNA (30) was cloned into the transcription vector pBluescript KSII(+) (Stratagene) and linearized at the 5-end. An antisense transcript was synthesized in vitro using T3 RNA polymerase and [³²P]UTP (800 Ci/mmol, Amersham). 20 µg of total cellular RNA was hybridized at 42 °C overnight with 100,000 cpm of the labeled antisense transcript. Hybrids were digested with RNases A and T1 for 1 h at 30 °C. Protected fragments were separated on 5% acrylamide, 8 M urea gels and analyzed by autoradiography. The increase in VEGF mRNA levels was quantitated by PhosphorImager analysis (FUJI BAS 1000, Fuji).

5 ml of serum-free Dulbecco's modified Eagle's medium/10-cm Petri dish was conditioned by UVB-irradiated quiescent HaCaT cells. 8 and 24 h after irradiation the conditioned medium from three Petri dishes was harvested and centrifuged to remove cell debris. Heparin-binding proteins were precipitated from the supernatant with 100 µl of heparin-Sepharose (1:1 slurry; Pharmacia) overnight at 4 °C. Heparin-Sepharose beads were precipitated by centrifugation and washed three times with 20 mM Tris-HCl, pH 7.4, 300 mM NaCl. Heparin-Sepharose-bound proteins were extracted by a 5-min incubation in Laemmli sample buffer at 95 °C and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. After transfer to nitrocellulose membranes, VEGF proteins were detected using a polyclonal antiserum directed against the amino terminus of VEGF (Santa Cruz Biotechnology) and an alkaline phosphatase detection system (Promega). Conditioned medium from nontreated cells was used as a negative control, whereas a cell lysate from COS cells transiently transfected with a VEGF cDNA was used as a positive control.

RESULTS

To investigate the role of sunlight on VEGF expression in the skin, we studied the effect of UVB on VEGF expression in the human keratinocyte cell line HaCaT. For this purpose, cells were irradiated with physiologically relevant doses of UVB and harvested at different time points after UVB treatment. As shown in Fig. 1A, low levels of VEGF mRNA were detected in quiescent, nontreated keratinocytes. Within 5 h after UVB irradiation (10 and 20 mJ/cm²), a large induction of VEGF mRNA expression was observed with highest levels 8 h after exposure to 20 mJ of UVB/cm² (Fig. 1, A and B). At this time point, VEGF mRNA levels were 11-fold higher compared with the basal level. This effect was long lasting, and 24 h after exposure of the cells to UVB, VEGF mRNA levels were still elevated (2-3-fold).

Four different VEGF proteins have recently been identified in vitro which arise from differential splicing at the 3-end of VEGF primary transcripts (12, 13). The hybridization probe that we used for RNase protection assays corresponds to the 3-end of the shortest form of VEGF (VEGF₁₂₁) and thus enabled us to distinguish among the different isoforms of this growth factor (Fig. 1C). A protected fragment corresponding to the complete coding sequence of the hybridization probe is generated by mRNA encoding VEGF₁₂₁ (upper band in Fig. 1A), whereas two shorter protected fragments are generated by mRNAs encoding longer variants of VEGF (lower bands in Fig. 1A). mRNAs encoding VEGF₁₂₁ and other splice variants were induced to a similar extent (Fig. 1A).

To analyze whether the observed increase in VEGF mRNA levels correlates with the production of immunoreactive VEGF protein, conditioned medium from UVB-irradiated and nonirradiated HaCaT cells was analyzed for the presence of VEGF protein. Because of the kinetics of the VEGF induction at the mRNA level, conditioned medium was harvested 8 and 24 h after irradiation, and VEGF protein was enriched by its capacity to bind to heparin-Sepharose. The presence of VEGF protein in the conditioned medium was detected by immunoblotting using a polyclonal antiserum directed against the amino terminus of VEGF, which is identical in all isoforms. As shown in Fig. 2, two VEGF proteins with estimated molecular masses of 17 and 19 kDa were detected in the conditioned medium of UVB-treated cells. The size of these proteins correlates with the expected sizes of the VEGF gene products. A VEGF specific band of about 36 kDa might represent a dimer of VEGF. The highest levels of VEGF protein were found 24 h after irradiation, whereas no VEGF protein was detectable in conditioned medium of nonirradiated cells. As a positive control, we used lysates from COS cells transiently transfected with a VEGF cDNA. As shown in the right lane of Fig. 2, a VEGF specific protein of about 22 kDa was detected in these cells which was not seen in vector-transfected control cells. Taken together these data demonstrate that the increase in VEGF mRNA after UVB irradiation correlates with accumulation of immunoreactive VEGF protein in the medium.

UV irradiation of cells has been shown to increase the intracellular levels of reactive oxygen species (31). To determine if this effect was responsible for the increase in VEGF expression after UVB exposure, we first analyzed the effect of H₂O₂ on VEGF mRNA expression in HaCaT cells. As shown in Fig. 3A, treatment of the cells with 1 mM H₂O₂ caused a strong increase in VEGF mRNA expression. As expected, pretreatment of the cells with the antioxidant PDTC (20 µM) completely abolished the effect of H₂O₂, whereas PDTC itself did not influence VEGF expression (Fig. 3B). By contrast, no inhibition of the UVB-dependent increase in VEGF mRNA was observed in PDTC-pretreated cells (Fig. 3C). Thus, although H₂O₂ is a potent inducer of VEGF expression, reactive oxygen species are unlikely to be responsible for VEGF up-regulation by UVB.

Recent studies from our laboratory have demonstrated the rapid induction of VEGF expression in HaCaT cells by various growth factors within 10-30 min (32). This increase is independent of de novo protein synthesis. By contrast, up-regulation of this gene by UVB occurred only 5-8 h after irradiation, suggesting that de novo protein synthesis might be required for this effect. Therefore, we investigated the effect of the protein synthesis inhibitor cycloheximide on the UVB- and H₂O₂-induced VEGF expression. As seen with most primary response genes, treatment of quiescent HaCaT cells with this reagent already caused increased VEGF expression (Fig. 4A), and a strong superinduction was seen after addition of both cycloheximide and H₂O₂ (Fig. 4B). These findings suggest a direct effect of H₂O₂ on VEGF expression. In contrast, no superinduction was observed after UVB irradiation of cycloheximide-treated cells. Moreover, the cycloheximide-induced VEGF up-regulation was even reduced by UVB (Fig. 4A). Thus, the effect of UVB is most likely to be indirect. To test this hypothesis further, we analyzed the potency of conditioned medium of UVB-treated cells to induce VEGF expression in quiescent HaCaT cells. As shown in Fig. 5A, medium conditioned for 8 or 24 h by UVB-irradiated cells strongly induced VEGF expression. This induction was comparable to the increase seen after the addition of 10% fetal bovine serum. By contrast, conditioned medium from nonirradiated cells had no effect on VEGF expression. This result suggests that UVB irradiation causes release of soluble mediators that subsequently turn on VEGF expression.

DISCUSSION

Solar radiation causes a variety of biological effects in the skin. Whereas low doses of UV radiation can be useful due to the photosynthesis of vitamin D, larger doses cause sunburn and, when exposure is chronic, can promote tumor formation. Sunburn is characterized by erythema and edema as a result of increased vascular permeability. The mechanisms that underlie these effects have not been elucidated fully. Since erythema is particularly induced by UVB irradiation that only penetrates into the epidermis (33), keratinocyte-derived soluble mediators that diffuse to the dermis are likely to mediate this effect. This diffusion theory is supported by the lag phase between UVB exposure and the appearance of visible erythema and by the increase in the latent period for visible erythema after cooling of the skin (34). The soluble mediators that induce increased vascular permeability have only partially been characterized. Previous studies have suggested an important role of prostaglandins in this process. However, the later stages of erythema cannot be explained fully by the presence of these factors (for review, see Ref. 33). Therefore, additional mediators might contribute to the effect.

One of the most potent vascular permeability factors is VEGF, which is overexpressed in a wide variety of physiological and pathological conditions associated with increased vascular permeability, such as cancer and wound healing and also in erythema seen in several blistering diseases (1, 9, 21). The results described in the present study suggest an additional role of VEGF in UVB-induced erythema. UVB irradiation of cultured keratinocytes resulted in a strong increase in VEGF mRNA and protein levels. These elevated levels might either be due to transcriptional activation of the VEGF gene and/or to increased mRNA stability since both mechanisms have been shown to contribute to the regulation of VEGF mRNA levels (35).

The UVB dose and spectrum used in this study were well comparable to the UVB light reaching the human skin upon sun exposure, suggesting that a similar increase in VEGF expression might occur in vivo. Thus, in addition to prostaglandins, VEGF might contribute further to the induction of erythema by UVB irradiation. This hypothesis is supported by a recent study that demonstrated a synergistic effect of prostaglandin E₂ and VEGF in a quantitative model of local plasma leakage in rabbit skin (36), and a similar synergistic effect might occur in human skin after sun exposure.

The effects of UV irradiation are frequently mediated by reactive oxygen species, and increased levels of hydrogen peroxide, hydroxyl radicals, superoxide, and organic hydroperoxides are found in many different cell types following UV exposure (31, 33). These reactive oxygen species are potent modulators of transcription factor activity (37, 38), resulting in significant changes in gene expression. Therefore we speculated about a role of these molecules in the induction of VEGF expression seen after UVB exposure. Indeed, treatment of keratinocytes with H₂O₂ caused a strong increase in VEGF expression which was blocked by antioxidants. By contrast, antioxidants had no effect on the UVB-induced VEGF expression, suggesting that UVB irradiation and H₂O₂ induce expression of this gene via different mechanisms. The induction of VEGF expression by UVB in the presence of antioxidants is in agreement with recent results that demonstrated the involvement of reactive oxygen intermediates in the UVA but not in the UVB response (39).

A series of previous studies suggested an important role of tyrosine phosphorylation but not of protein kinase C activation in the induction of gene expression by UVC and UVB (40, 41, 42, 43, 44), and preliminary studies of our laboratory suggest that similar mechanisms might be responsible for the increased expression of VEGF after UVB irradiation. Thus, pretreatment of the cells with protein tyrosine kinase inhibitors blocked the UVB response, whereas protein kinase C inhibitors had no effect.

Interestingly, induction of VEGF expression in keratinocytes by H₂O₂ and various growth factors and cytokines occurred much more rapidly compared with the induction seen after UVB exposure (32 and data not shown). This finding suggested that indirect mechanisms might be responsible for the UV effect. This hypothesis was supported by the results seen with cycloheximide. Whereas a combination of H₂O₂ and the protein synthesis inhibitor caused a strong superinduction of VEGF expression, induction of this gene by UVB was reduced slightly in the presence of cycloheximide. The indirect nature of the UVB effect was proven further by the induction of VEGF expression by conditioned medium from UVB-treated keratinocytes. This finding demonstrates that UVB irradiation of these cells causes release of soluble factors that subsequently turn on VEGF expression. Proinflammatory cytokines are the most likely candidates for this effect. Thus, expression of IL-1, IL-1, and tumor necrosis factor  have been shown to be induced in keratinocytes by UVB irradiation (23, 24, 25), and expression of these cytokines precedes the increased expression of VEGF in these cells (data not shown). Since IL-1 and tumor necrosis factor  are potent inducers of VEGF expression in HaCaT cells (32), a combined action of these factors might be responsible for induction of VEGF expression by UVB. A similar cytokine-mediated effect has recently been demonstrated for the UVB-induced prostaglandin E₂ expression in keratinocytes (45), suggesting that various UV effects could be mediated by these factors.

In summary, our data demonstrate a strong induction of VEGF expression by UVB which is mediated by the release of soluble factors. These findings suggest a novel and important role of VEGF in the induction of erythema after prolonged sun exposure. Future studies using animal models will help to clarify further the role of this multipotent factor in the UVB response in vivo.