INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Vascular endothelial growth factor (VEGF) is a specific mitogen for vascular endothelial cells and a potent inducer of vascular permeability, and it plays a critical role in the formation of new blood vessels during both vasculogenesis and angiogenesis (reviewed in reference 13). The latter includes angiogenesis associated with tumor formation and a number of other pathological conditions (reviewed in reference 1). Several proteins related in sequence and structure to VEGF (also called VPF, for vascular permeability factor) have been reported. The VEGF family now includes placental growth factor (PlGF), VEGF-B, VEGF-C, and VEGF-D, each of which shows between 30 and 45% amino acid sequence identity with VEGF. The family members are all structurally similar and share a central VEGF homology domain, which contains eight cysteine residues that form part of a cysteine knot motif. The VEGF family members are ligands for a set of mammalian tyrosine kinase receptors (37). VEGF binds and activates both VEGF receptor 1 (VEGFR-1) (Flt-1) and VEGFR-2 (Flk-1 or KDR), while PlGF and VEGF-B bind only VEGFR-1. VEGF-C and VEGF-D bind VEGFR-2 and VEGFR-3 (Flt-4). VEGF and PlGF-2 also appear to interact with a recently discovered neuronal cell guidance receptor, neuropilin 1. Although the functional significance of each of these ligand-receptor interactions remains to be precisely defined, gene targeting studies have demonstrated the requirement of VEGFR-1 and VEGFR-2 for embryonic development. In general terms, it appears that VEGFR-1 plays a role in vascular endothelial differentiation and migration, VEGFR-2 is involved in vascular endothelial mitogenesis, and VEGFR-3 is involved in the regulation of angiogenesis of the lymphatic vasculature.

We reported previously that the NZ2 strain of the parapoxvirus orf virus encodes an apparent homolog of VEGF (VEGF-ORFV_NZ2) (26) and recently showed, along with others, that the viral factor shares some of the functional features of mammalian VEGF (31, 43). These include the ability of purified VEGF-ORFV_NZ2 to stimulate the proliferation of vascular endothelial cells, to promote vascular permeability, and to bind VEGFR-2 and neuropilin 1. However, the viral VEGF does not recognize VEGFR-1 or VEGFR-3 and, as such, forms a new member of the VEGF family of molecules with a unique profile of receptor recognition. The in vivo activities of this new growth factor have not yet been examined.

Orf virus is the type species of the genus Parapoxvirus and shares numerous features with other members of the poxvirus family (19, 32). However, the distinctive morphology of parapoxvirus virions and the high G+C content of their genomes (44) suggest that a significant genetic divergence from other genera of this family has occurred. This suggestion has been confirmed by sequence analyses of the 140-kb orf virus genome (14, 30). Orf virus produces pustular dermatitis in humans, sheep, and goats (reviewed in reference 19). Infection is initiated in damaged skin, and lesions progress through stages of erythema, papule, vesicle, pustule, and scab, with infection confined to the epidermis. The pathological features of human and ovine orf virus lesions are the same, and in both hosts infection is confined to the epidermis, with no evidence of systemic spread. The lesions are remarkable for extensive vascular proliferation and dilation as well as marked proliferation of the epidermis, with finger-like projections of the epidermis deep into the dermis (18). The expression of an orf virus-encoded member of the VEGF family of molecules may provide an explanation for some of these observations. No other virus has been reported to encode a VEGF.

In this study, we constructed a recombinant orf virus in which the VEGF-like gene was disrupted and used this construct to examine the activities of VEGF-ORFV_NZ2 during orf virus infection of its natural host. In vivo expression of the viral VEGF promoted a strong angiogenic response including erythema, capillary proliferation, and dilation. In addition, these experiments demonstrated an unexpected role for the viral VEGF in epidermal proliferation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and virus. Orf virus strain NZ2 (33) and vaccinia virus strain Lister (34) were propagated in primary bovine testis (BT) or ovine testis (LT) cells using Eagle's minimal essential medium containing 10% fetal bovine serum and 5% lactalbumen hydrolysate.

Construction of recombinant orf virus. The gene encoding VEGF-ORFV_NZ2 is located within a 1.45-kb SphI-SmaI subfragment of the KpnI E fragment of orf virus (26). The 1.45-kb subfragment was isolated and cloned into pSP70 to generate pVU486. Digestion of pVU486 with AvrII and BsmI removed a 123-nucleotide sequence from the VEGF-ORFV_NZ2 gene and was followed by incorporation of a linker (CTAGGATCCTTTTTATGAATTCCG) which introduced BamHI and EcoRI restriction sites and a poxvirus early transcription termination signal. The orf virus sequences flanking the linker were a left arm of 981 bp, consisting of 869 bp upstream of the VEGF coding region followed by the first 112 bp of the VEGF gene, and a right arm of 345 bp, consisting of the terminal 167 bp of the VEGF gene and 178 bp downstream of it. This plasmid was designated pVU487. pVU488 contained the Escherichia coli guanosine phosphoribosyltransferase (gpt) gene under the control of the orf virus early promoter, PE1 (15), and the E. coli lacZ gene under the control of the orf virus late promoter, PF1 (14). The promoter-reporter elements were kindly provided by D. Lyttle and will be described in detail elsewhere. An 812-bp PE1-gpt fragment was derived from pVU488 by PCR and ligated into the BamHI site of pVU487. The PF1-lacZ element was removed from pVU488 using EcoRI and ligated into the EcoRI site of pVU487. The final plasmid construct was called pVEGF-.

4

Molecular characterization of recombinant orf virus. Viral cores were recovered from lysates of infected BT cells by centrifugation through a sucrose-dextran cushion, and DNA was purified by standard methods (12, 29). Restriction enzyme analyses, Southern blot hybridizations, PCRs, and DNA sequence determinations were performed according to standard methods (16, 29). The primers used in PCRs were as follows: gf1, AGCGCCCGGGATCCATGAAGTTGCTCGTCG; gf2, ACTCGAGGTACCTAGCGGCGTCTTCTGGG; gpt, CCTGTTCAAACCCCGCTTTA; lac, GACAAACTCGGGCAGCGTT; and out, TCACCGAGGCGGAGCCGTT.

Experimental infection of sheep with orf virus. Orf virus-naive merino-cross lambs were inoculated on the wool-free region of the inner surface of the hind legs. An approximately 2-cm-long scratch was made in the skin using a bifurcated needle, and 10⁶ PFU of virus in 20 µl of phosphate-buffered saline (PBS) was applied to the scratch. Cell-free suspensions were prepared from 3-mm punch biopsies of infected skin (22), and the viral titer was determined by a plaque assay on BT cells.

Immunohistochemical staining and quantitation of vascular endothelial cells. Punch biopsies (5 mm) were fixed in 10% neutral buffered formalin and processed onto paraffin wax. Two 4-µm serial sections were taken from three points, 100 µm apart, in the fixed block. The first of the serial sections was stained with hematoxylin and eosin, and the second was used for immunohistochemical analysis of vascularization. Vascular endothelial cells were identified with a polyclonal rabbit anti-human von Willebrand factor antiserum (Dako). The extent of vascularization was quantified from each of three semiserial sections per lesion by counting the stained cells which fell on any of the 400 intersecting points within a grid (0.5 by 0.5 mm), each side of which was divided by 20 equidistant lines. The areal fraction of dermis vascularized was expressed as the fraction of the intersecting points on which a stained cell fell (42).

Preparation of viral conditioned medium (CM). BT cells were infected at a multiplicity of infection of 0.1 PFU per cell and incubated until all cells were clearly infected. LT cells were infected at a multiplicity of infection of 3 PFU per cell and incubated for 24 h. Cells were sedimented at low speed, and the supernatants were filtered twice (0.1-µm-pore-size filters) to remove infectious orf virus.

VEGF bioassays. Mitogenic assays were conducted with viral CM and human microvascular endothelial cells (HMVEC) (43). HMVEC were seeded at 10⁴ cells per well and allowed to adhere. Samples of CM diluted to 10% in medium with reduced serum and without growth supplements were added. Proliferation was measured by direct counting of cells 72 h later. The Ba/F3-derived cell line Ba/F3-VEGFR-2-EpoR, expressing a chimeric receptor consisting of the extracellular domain of mouse VEGFR-2 and the transmembrane and cytoplasmic domains of the mouse erythropoietin receptor, was used to determine if CM was capable of binding and activating VEGFR-2 (37). Cells were washed free of interleukin 3 and then resuspended in dilutions of CM from virus-infected cells or from COS-7 cells transiently expressing the mouse VEGF isoform consisting of 164 amino acids, VEGF₁₆₄ (37). Cells were incubated for 48 h, and DNA synthesis was quantified by measuring [³H]thymidine incorporation with a beta counter. The induction of vascular permeability by CM was determined by measuring the extravasation of Evans blue dye in guinea pig skin (Miles assay) (43). Anesthetized guinea pigs were injected intracardially with 500 µl of 0.5% Evans blue dye. CM was diluted 1/5 in PBS (pH 7.2), and 100 µl was injected intradermally to shaved areas on the back of the animal. Five nanograms of purified mouse VEGF₁₆₄ was used as a positive control. The animals were sacrificed after 20 min, the skin was excised, dye was eluted in formamide, and the absorbance at 620 nm was recorded.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Disruption of the orf virus VEGF gene. A recombinant orf virus in which the VEGF-ORFV_NZ2 gene had been inactivated was constructed by removing 116 bp of the gene, including a region encoding five cysteines critical for maintaining the tertiary structure of VEGF, and replacing this segment with a reporter cassette (see Materials and Methods). A schematic representation of the VEGF region of this recombinant virus, ovVEGF-, is shown in Fig. 1, along with DNA analyses which confirmed its identity.

View larger version (35K): [in this window] [in a new window]   FIG. 1.   Characterization of the orf virus recombinant, ovVEGF-. (A) Schematic diagram showing the location of the VEGF gene (hatched bar) within a 1.45-kb SmaI-SphI fragment of the orf virus genome (26). The region between the marked AvrII and BsmI sites was deleted from the WT VEGF gene during the construction of ovVEGF-. This region was replaced with the gpt and lacZ genes under the control of an orf virus early promoter (PE1) and a late promoter (PF1), respectively. The box representing lacZ is interrupted to indicate that it is not drawn to scale. The locations of four BamHI sites are marked with B. The distances between these sites, from left to right, are 812, 111, and 5,414 bp. Arrowheads indicate the locations of PCR primers (see below). Sequences denoted as bars were included in the plasmid (pVEGF-) used in the construction of ovVEGF-. Horizontal lines extending from these boxes represent adjacent orf virus genome sequences not included in pVEGF-. (B) BamHI restriction endonuclease cleavage fragments derived from ovVEGF- and WT genomic DNAs were separated on a 0.7% agarose gel (left panel). The 5.4-kb BamHI fragment derived from ovVEGF- is indicated by an arrowhead. The fragments were Southern blotted and hybridized with the lacZ probe (right panel). (C) HindIII restriction endonuclease cleavage fragments derived from ovVEGF- and WT genomic DNAs were separated on a 0.7% agarose gel (left panel). The HindIII fragment unique to ovVEGF- is indicated by an arrowhead. The fragments were Southern blotted and probed with a plasmid containing the 1.45-kb SmaI-SphI fragment shown in panel A (right panel). (D) PCR products derived from WT and ovVEGF- genomic DNAs were electrophoresed on a 1.3% agarose gel and visualized with ethidium bromide. The primer pairs used for PCR amplifications are indicated by arrowheads in panel A, and the bars below those arrowheads indicate the predicted sizes of the PCR products. Template DNA used in the reactions was WT DNA (even-numbered lanes) or ovVEGF- DNA (odd-numbered lanes). Lane 1, low-DNA-mass ladder (Gibco BRL); bands of 1,200, 800, 400, and 200 bp are shown. Lanes 2 and 3, primers gf1 and gf2. Lanes 4 and 5, primers gf1 and gpt. Lanes 6 and 7, primers lac and gf2. Lanes 8 and 9, primers lac and out.

VEGF-like activity is not produced by ovVEGF-. CM prepared from cells infected with WT, ovVEGF-, or the Lister strain of vaccinia virus or mock infected with PBS was subjected to three assays of VEGF-like activity: endothelial cell mitogenesis, vascular permeability, and activation of VEGFR-2. CM from WT-infected LT cells was able to promote a threefold increase in the number of HMVEC (Fig. 2A). No such increase was promoted by CM from cells infected with ovVEGF- or vaccinia virus or mock infected.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   VEGF activity is not produced by ovVEGF-. CM was prepared from WT-, ovVEGF--, vaccinia virus (VV)-, and mock-infected cells. (A) Mitogenic activity of CM for endothelial cells. HMVEC were seeded at 104 cells per well in medium supplemented (10%) with CM from the indicated source (LT cells were used), and cell numbers were determined after 72 h. Values are the mean ± standard deviation (SD) of triplicate experiments. (B) Activation of VEGFR-2 by CM. Ba/F3 cells expressing chimeric VEGFR-2 were resuspended in dilutions of CM from virus-infected BT cells (WT or ovVEGF-) and from COS-7 cells transiently expressing mouse VEGF164 (VEGF). Cells were incubated for 48 h, and DNA synthesis was quantified by measuring 3H-thymidine incorporation and  counting. Values are the mean ± SD of duplicate readings. (C) Induction of vascular permeability by CM (Miles assay). Guinea pigs were injected intracardially with Evans blue dye. CM (from LT cells) and 5 ng of purified mouse VEGF164 (VEGF) were injected intradermally. After 20 min, the skin was excised and photographed (top), the dye was eluted in formamide, and the absorbance at 620 nm (OD/620 nm) was recorded (bottom). Values are the mean ± SD of duplicate injections.

Gross pathology of WT and ovVEGF- lesions. In order to assess the role of VEGF-ORFV_NZ2 in disease, we infected sheep by applying 10⁶ PFU of virus to scarified skin. Each of three animals was given four replicate inoculations with WT, ovVEGF-, or PBS. The development of WT lesions followed the general pattern reported previously (24). This pattern was characterized by the progressive development of erythema, pustules, and a scab until the experiment was terminated 14 days p.i. Representative lesions observed on one animal are shown in Fig. 3A. Erythema was visible along the scratch lines 2 days p.i. This reaction was marked by 4 days, at which time pustules first became apparent (data not shown). By 6 days p.i., the intense erythematous margin seen after 4 days was still evident and the pustules had developed further. Pustule formation was at a maximum 10 days p.i., by which time the erythema was reduced. At the last time point, 14 days p.i., erythema was minimal, the pustules had contracted, and a scab was beginning to form.

View larger version (55K): [in this window] [in a new window]   FIG. 3.   Histopathological analysis of the role of VEGF-ORFVNZ2 in infection of sheep. Scarified skin was inoculated with WT or ovVEGF- (VEGF) or mock infected. (A) Gross pathology. Representative lesions at 2, 6, 10, and 14 days (d) p.i., shown at approximately half life size. All lesions are from the same animal (101). (B) Histological comparison of WT-infected (panels 1 and 2), ovVEGF--infected (panels 3 and 4), and mock-infected (panels 5 and 6) lesions at 10 days p.i. (magnification, ×50). Panels 1, 3, and 5 show representative sections stained with hematoxylin and eosin (H + E), while panels 2, 4, and 6 show the adjacent sections from the same biopsy stained with peroxidase-conjugated anti-von Willebrand factor antibody (anti-vWF) and 3,3'-diaminobenzamine to reveal endothelial cells. The pustule (P), epidermis (E), and dermis (D) are indicated where appropriate, and arrows indicate representative blood vessels highlighted by the antibody staining.

Histological analysis of WT and ovVEGF- lesions. The reduced pathology seen in superficial examinations of ovVEGF- lesions compared with WT lesions was also apparent in histological examinations of these lesions (Fig. 3B). Histological differences were seen as early as 2 days p.i., when in the WT lesions there was an accumulation of fluid in the dermis that was not seen in the ovVEGF- lesions or PBS controls. After 6 days, there was an intense influx of inflammatory cells in the dermis of WT lesions. Cellular ballooning and degeneration of the upper layers of the epidermis were also evident in WT lesions, and the epidermis had become markedly hyperplastic. In contrast, the influx of inflammatory cells into the dermis of ovVEGF- lesions was less intense, and there was little evidence of the cytopathic effects observed in the epidermis of WT lesions.

Epidermal hyperplasia in orf virus lesions. Epidermal hyperplasia is a feature of the response to orf virus infection. At 14 days p.i., the epidermis of WT lesions was more than sevenfold thicker than that of mock-infected lesions and approximately twofold thicker than that of ovVEGF- lesions (data not shown). A major component of the enlarged epidermis seen in WT lesions was the number of epidermal downgrowths, known as rete ridges. These were apparent 6 days p.i., when WT lesions had developed 9.3-fold more rete ridges than had ovVEGF- lesions (Table 1). The rete ridges seen in WT lesions extended further into the dermis and were, on average, 1.7-fold longer than those few seen in ovVEGF- lesions. This pattern was also apparent 14 days p.i., at which time the rete ridges had increased in length by two- to threefold over those seen at 6 days p.i. (Table 1). Rete ridges were not seen in mock-infected lesions.

View this table: [in this window] [in a new window]   TABLE 1.   Induction of rete ridges in skin infected by WT and ovVEGF- (Rec)a

WT lesions are more vascularized than ovVEGF- lesions. Vascularization at the sites of infection was assessed using anti-human von Willebrand factor antibody. Representative sections from WT-, ovVEGF--, and mock-infected lesions sampled 10 days p.i. are shown in Fig. 3B. The extent of dermal vascularization associated with sites of infection was analyzed by determining the areal fraction of dermis reacting with anti-von Willebrand factor antibody. This quantitative analysis of vascularization showed a time-dependent increase in the dermal vascularization of WT lesions that greatly exceeded any such response in ovVEGF- lesions (Fig. 4). At 2 days p.i., there appeared to be no difference in the vascularization of WT-, ovVEGF--, or mock-infected lesions. By 6 days p.i., however, the dermis of WT lesions was intensely vascularized and the lumen of blood vessels appeared to be enlarged. WT lesions were from 1.8-fold (sheep 102) to 3.8-fold (sheep 103) more vascularized than corresponding mock-infected lesions (Fig. 4). Markedly enhanced dermal vascularization and dilation of blood vessels were also apparent in WT lesions sampled 10 (Fig. 3) and 14 days p.i., with the area vascularized exceeding that seen in mock-infected lesions by an average of 5-fold and by as much as 9.8-fold in one case (sheep 101; day 14). It was clear that orf virus infection of sheep skin induced a substantial proliferation of the vasculature of the underlying dermis. This dramatic response was not seen in ovVEGF- lesions. In all samples taken at days 6, 10, and 14 p.i., ovVEGF- lesions were found to be less vascularized than WT lesions taken at the same time from the same animal (Fig. 4). At these times p.i., the extent of dermal vascularization of the WT lesions exceeded that seen in the corresponding ovVEGF- lesions by 1.9- to 6.2-fold. Analysis of these data using paired t tests for each day confirmed that the vascularization seen in WT lesions was significantly greater than that seen in ovVEGF- lesions at 6, 10, and 14 days p.i. (P values of 0.048, 0.021, and 0.039, respectively) but was not significantly different at 2 days p.i. (P value of 0.485). Figure 4 shows that in all samples taken 6, 10, and 14 days p.i., the extent of dermal vascularization recorded for ovVEGF- lesions exceeded that recorded for mock-infected lesions. This stimulation of vascularization was markedly lower than that seen for WT lesions and, on average, the values for ovVEGF- infection exceeded those for mock infection by 1.7-fold. Only vascularization at 6 days p.i. was significantly different (P value of 0.039). These results demonstrate that the viral VEGF is expressed during natural infection and plays a key role in the development of the vascularized nature of orf virus lesions.

View larger version (10K): [in this window] [in a new window]   FIG. 4.   Vascularization of orf virus lesions. Scarified skin was inoculated with WT (black), ovVEGF- (hatched bars), and PBS (grey bars). Five-millimeter punch biopsies were taken from separate lesions of each animal at days 2, 6, 10, and 14 p.i. The dermal areal fraction reacting with anti-von Willebrand factor antibody was calculated from the analysis of three equidistant sections as outlined in Materials and Methods. Values are the mean ± standard deviation. The asterisk indicates that no data were obtained from this biopsy, but regression methods were used to derive an estimate of 0.24; this value was used in subsequent analyses.

Effect of inactivation of VEGF-ORFV_NZ2 on viral yield in vivo. In order to provide an estimate of the production of infectious virus during the course of infection in sheep skin, 3-mm punch biopsies were taken and, after processing, the titers of PFU were determined (Fig. 5). At 2 days p.i., virus was detected in only two of the six samples, but by 6 days p.i., all viral lesions contained significant amounts of virus, with similar titers of virus being detected in WT and ovVEGF- lesions. This result can be seen in the average ratio of WT to ovVEGF- titers, which at day 6 was 0.9. 

View larger version (13K): [in this window] [in a new window]   FIG. 5.   Effect of inactivation of VEGF-ORFVNZ2 on virus yield in infected sheep. Scarified skin was inoculated with WT (black bars) and ovVEGF- (hatched bars). Three-millimeter punch biopsies were taken from separate lesions of each animal at days 2, 6, 10, and 14 p.i. Samples were processed and viral titers were determined as described in Materials and Methods. Values are the mean ± standard deviation of duplicate titrations. Asterisks indicate that the viral titer was less than 5 PFU per sample.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Descriptions of orf virus lesions frequently refer to prominent vascular changes, and as early as 1890, Walley referred to lesions that "readily bleed" and observed "distension of the blood vessels" (41). More recently, Groves described "massive capillary proliferation and dilation [that] gave the impression of an angiomatous lesion" (18). Construction of a recombinant orf virus in which the viral VEGF was inactivated allowed us to examine for the first time the role played by the viral VEGF in the vascular response seen in orf virus lesions.

We demonstrate here that VEGF-like activities expressed by WT-infected cells were not seen in cells infected with the recombinant virus, in which the VEGF-like gene was disrupted. Disruption of the viral VEGF gene and the loss of VEGF activity had no detectable effect on the growth of orf virus in cultured cells; however, infection of its natural host was significantly affected. Analysis of vascularization associated with the viral lesions allowed us to confirm previous reports that the dermis underlying an orf virus lesion is intensely vascularized (18). As early as 6 days p.i., the density of endothelial cells was, on average, fivefold greater than that seen in the dermis of a mock-infected lesion. Furthermore, at this time and up to at least 14 days p.i., there was significantly greater vascularization of WT lesions than of lesions induced by ovVEGF-. These observations demonstrate that the viral VEGF is able to promote angiogenesis during infection of skin.

Although our measure of the vascular response was based on the density of endothelial cells rather than on vascular profiles, we did observe a progressive development of recognizable microvessels with clearly visible lumina (Fig. 3). Confirmation that the vast majority of the vascular response is directed by the viral VEGF rather than by host factors induced by orf virus infection is provided by the virus yield data, which show that at 6 days p.i., the virus yields are similar for both WT and ovVEGF-. If dermal vascularization were a host response to viral infection, it would be expected that the response would be proportional to the extent of viral growth. This is clearly not so. Orf virus replication is known to be confined to the epidermis (23, 24), but our data suggest that the viral VEGF affects endothelial cells of the dermis. This notion is consistent with the results of studies of transgenic mice, in which overexpression of VEGF in keratinocytes stimulated an increase in the density of blood vessels in the dermis (10). Enhanced expression of VEGF has also been reported for a range of malignant and benign tumors of the skin and has been shown for advanced melanoma lesions to directly correlate with a 2.5-fold increase in the vascular density of the associated dermis (11). Studies with purified VEGF-ORFV_NZ2 have shown that it recognizes VEGFR-2 but not VEGFR-1 or VEGFR-3 (31, 43). Our demonstration that in situ expression of this factor promotes angiogenesis is consistent with the view that VEGFR-2 is predominantly responsible for signaling vascular endothelial cell mitogenesis.

Our data show a correlation between the presence of an intact viral VEGF gene and the induction of not only dermal vascularization but also epidermal hyperplasia. Cutaneous angiogenesis in conjunction with epidermal hyperplasia has also been reported for cutaneous infantile hemangiomas and UV-irradiated skin and was connected in both studies to altered levels of angiogenic factors, including VEGF (3, 4). The pathway by which the viral VEGF might induce epidermal proliferation is unclear, but the induction of growth factors, such as fibroblast growth factor 2 (FGF-2), keratinocyte growth factor (KGF), or heparin-binding epidermal growth factor (HB-EGF), is possible. For example, VEGF has been shown to induce HB-EGF in vascular endothelial cells, and HB-EGF is a known mitogen of keratinocytes (2).

In addition to a general thickening of the epidermis, WT lesions showed extensive rete ridge formation that was not apparent in ovVEGF- lesions. These extensions of the epidermis into the dermis are also seen in other pathologies. For example, psoriatic skin shows both enhanced angiogenesis and an increase in rete ridge length (9). A link between rete ridge formation and KGF was suggested by a study of wound healing (38). A role for the orf virus-encoded VEGF in promoting the expression of KGF could be postulated to occur via increased vascular permeability, leading to alterations in the dermal extracellular matrix, including the release of sequestered FGF-2. Such FGF-2 could then signal dermal fibroblasts to produce KGF and might thereby induce epidermal proliferation, with the formation of rete ridges.

The considerable epidermal proliferation seen in ovVEGF- lesions indicates that factors other than VEGF-ORFV_NZ2 also play a role in stimulating this cellular response to orf virus infection. Lesions of other poxviruses typically demonstrate localized cellular proliferation but do not show the vascularization seen in orf virus lesions. A probable explanation for this effect has been provided by the demonstration that several poxviruses each encode a growth factor related to epidermal growth factor (EGF) (5, 8, 40). We have not been able to detect any evidence of an EGF homolog encoded by orf virus. Nor is there any evidence of VEGF-like activity encoded by any virus other than parapoxviruses. However, virus-encoded EGF and VEGF are both examples of a growing family of factors which are expressed by poxviruses and other large DNA viruses, which are often nonessential for viral replication, but which play important roles in modulating the host environment during infection (25, 28, 36).

The epidermal and vascular responses seen in lesions of orf virus are reminiscent of a sustained wound healing response. Orf virus applied to wounded skin establishes replication in the regenerating epidermal layer (23, 24), and extravagantly proliferative and persistent orf virus lesions have been reported for immunocompromised individuals (21, 35, 39). The expression of a viral VEGF might assist in maintaining this regenerative response and thereby might support extended viral growth. Our data are suggestive of extended replication of WT relative to orf virus with a deletion of VEGF, but more detailed examinations will be required to clarify this possibility. Atypically severe orf virus lesions have been reported for thermally injured skin (20), and erythema multiforme reactions are a relatively common complication of orf virus infection. Intriguingly, both of these pathologies, in the absence of orf virus, have been linked to elevated levels of VEGF (6, 17).

Another possible role for the viral VEGF, again with a link to wound healing, is the extensive scab formation seen on healing orf virus lesions. The scab shed from orf virus lesions contains substantial amounts of infectious virus. The envelopment of the virus in the scab provides it with protection from environmental inactivation. In this way, the virus remains available to infect naive animals as much as 1 year after being shed (19). Vascular hyperpermeability, with associated leakage of plasma proteins, is a feature of wound healing and has been linked to the overexpression of VEGF (7). The viral VEGF is able to induce vascular permeability and would seem to contribute to scab formation, since lesions induced by orf virus with a deletion of VEGF have essentially no scab. Our data conclusively demonstrate an in vivo role for the virus-encoded VEGF and provide a natural disease model to study the activities of a new member of the VEGF family that activates VEGFR-2 but not VEGFR-1 or VEGFR-3.