Interleukin-8 and Growth-Regulated Oncogene Alpha Mediate Angiogenesis in Kaposi s Sarcoma

Transcription

1 JOURNAL OF VIROLOGY, Nov. 2002, p Vol. 76, No X/02/$ DOI: /JVI Copyright 2002, American Society for Microbiology. All Rights Reserved. Interleukin-8 and Growth-Regulated Oncogene Alpha Mediate Angiogenesis in Kaposi s Sarcoma Brian R. Lane, 1,2 Jianguo Liu, 3 Paul J. Bock, 1 Dominique Schols, 4 Michael J. Coffey, 5 Robert M. Strieter, 5 Peter J. Polverini, 3 and David M. Markovitz 1,2 * Divisions of Infectious Diseases 1 and Pulmonary and Critical Care Medicine, 5 Department of Internal Medicine, and Graduate Program in Cellular and Molecular Biology, 2 University of Michigan Medical Center, Ann Arbor, Michigan ; Department of Oral Medicine, Pathology, and Oncology, University of Michigan Dental School, Ann Arbor, Michigan ; and Rega Institute for Medical Research, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium 4 Received 13 March 2002/Accepted 13 August 2002 The development of the complex neoplasm Kaposi s sarcoma is dependent on infection with the Kaposi s sarcoma-associated herpesvirus (KSHV) and appears to be greatly enhanced by cytokines and human immunodeficiency virus type 1 (HIV-1) Tat. Interleukin-8 (IL-8) and growth-regulated oncogene alpha (GRO- ) are chemokines involved in chemoattraction, neovascularization, and stimulation of HIV-1 replication. We have previously demonstrated that production of GRO- is stimulated by exposure of monocyte-derived macrophages (MDM) to HIV-1. Here we show that exposure of MDM to HIV-1, viral Tat, or viral gp120 leads to a substantial increase in IL-8 production. We also demonstrate that IL-8 and GRO- are induced by KSHV infection of endothelial cells and are crucial to the angiogenic phenotype developed by KSHV-infected endothelial cells in cell culture and upon implantation into SCID mice. Thus, the three known etiological factors in Kaposi s sarcoma pathogenesis KSHV, HIV-1 Tat, and cellular growth factors might be linked, in part, through induction of IL-8 and GRO-. Although chemokines were originally characterized by their ability to attract and activate leukocytes, these small polypeptide molecules and their receptors have now been shown to function in a variety of biological processes, including human immunodeficiency virus type 1 (HIV-1) infection and angiogenesis (3). HIV-1 enters cells by binding first to the CD4 receptor, which induces a conformational change in gp120, and then to a seven-transmembrane-domain G-protein-coupled receptor, which exposes the fusogenic epitopes of gp41 and permits fusion with the host cell membrane and subsequent entry into the cell. The principal coreceptor for monocytotropic HIV-1 isolates has been identified as CCR5, a C-C chemokine receptor that has as its ligands RANTES, MIP-1, and MIP-1 (1, 15, 22, 24, 25). The major coreceptor for T-cell-tropic HIV-1 is the C-X-C chemokine receptor CXCR4, the ligand for which is SDF-1 (6, 32). Monocyte-derived macrophages (MDM) express both CCR5 and CXCR4 and are readily infected by CCR5-using isolates of HIV (R5 HIV), which can also infect primary T cells (1, 15, 22, 24, 25). In contrast, while CXCR4-using isolates of HIV-1 (X4 HIV) infect primary T cells as well as T-cell lines, X4 HIV has been shown to infect macrophages only under certain circumstances (81, 85, 89, 96). In the past few years, it has been shown that the chemokines * Corresponding author. Mailing address: 5220 MSRB III, 1150 West Medical Center Dr., Ann Arbor, MI Phone: (734) Fax: (734) dmarkov@umich.edu. Present address: Division of Pulmonary and Critical Care Medicine, Department of Medicine, UCLA School of Medicine, Los Angeles, CA Present address: University of Minnesota School of Dentistry, Minneapolis, MN RANTES, MIP-1, MIP-1, and SDF-1 suppress HIV-1 replication (18, 19), although under some circumstances these chemokines can actually enhance HIV-1 replication (48, 82). Recent studies from our laboratory have shown that other chemokines, namely interleukin-8 (IL-8) and growth-regulated oncogene alpha (GRO- ), can stimulate HIV-1 replication as well (53, 55). IL-8, the prototypical member of the C-X-C chemokine family, has been shown to enhance new blood vessel formation, attract T cells, and stimulate monocyte adherence, in addition to its primary function as a chemotactic factor for neutrophils (3, 41, 49, 56, 97). GRO- was initially identified as an autocrine growth factor for malignant melanoma cells and has subsequently been shown to serve as a neutrophil chemoattractant and angiogenic factor as well (3). GRO- is 43% identical to IL-8 at the amino acid level and has many of the same activities as IL-8 (3). C-X-C chemokines containing the sequence Glu-Leu-Arg, the so-called ELR motif, including IL-8 and GRO-, promote angiogenesis, while ELR-negative C-X-C chemokines, such as gamma interferoninducible protein 10 (IP-10), inhibit angiogenesis (88). IL-8 exerts its actions by binding to two C-X-C chemokine receptors, CXCR1 and CXCR2, and GRO- exerts its actions by binding to CXCR2 (3). Elevated levels of IL-8 and GRO- have been detected in the serum and lungs of HIV-infected individuals (23, 57, 60, 90). The presence of elevated levels of IL-8 in individuals infected with HIV-1 has led several groups to suggest that this angiogenic chemokine plays a role in the pathogenesis of HIV-1 disease and infection with opportunistic infections, but little evidence had previously been presented to support these claims. Kaposi s sarcoma is a highly vascularized neoplasm that is one of the two most common malignancies associated with 11570

2 VOL. 76, 2002 IL-8, GRO-, AND ANGIOGENESIS IN KAPOSI S SARCOMA HIV-1 infection (8). Kaposi s sarcoma is characterized by the presence of complex spindle cell neoplasms, which are composed of endothelial cells, fibroblasts, dermal dendrocytes, and inflammatory cells (36, 42, 61). In the early stages of disease, Kaposi s sarcoma is not a monoclonal tumor but an angioproliferative lesion driven by inflammatory cytokines (33, 77). Although the origin of Kaposi s sarcoma is still controversial, mounting evidence suggests that endothelial cells are the progenitors of the Kaposi s sarcoma spindle cells (7, 8). Since it was first reported that over 90% of AIDS-associated Kaposi s sarcoma tissue samples were positive for a new member of the gammaherpesvirus family, the Kaposi s sarcoma-associated herpesvirus (KSHV), also called human herpesvirus 8, KSHV has been believed to be necessary for the development of Kaposi s sarcoma (14, 38, 63). KSHV infects endothelial cells and is found in spindle cells, and prior KSHV infection clearly predisposes AIDS patients to develop Kaposi s sarcoma (7, 39, 79). The mechanisms by which HIV-1 and human cellular factors influence the development of Kaposi s sarcoma is currently of much interest. While HIV-1 infection alone is not sufficient for the development of Kaposi s sarcoma, AIDS-associated Kaposi s sarcoma is more aggressive, disseminated, and resistant to treatment than the other forms, including posttransplant Kaposi s sarcoma (10, 37, 86). Immunosuppression alone cannot explain the much greater prevalence of Kaposi s sarcoma in AIDS patients than in other immunosuppressed individuals, the frequent presentation of Kaposi s sarcoma prior to immunosuppression in HIV-infected individuals, and the association of Kaposi s sarcoma with HIV-1 but not HIV-2 infection in West Africa (2, 5, 64, 75). Thus, there has been much interest in HIV-related factors that might potentiate the development of Kaposi s sarcoma. In this report, we demonstrate that exposure of MDM to HIV-1 leads to increased IL-8 production, an effect mediated by the HIV-1 transactivator protein Tat and the inflammatory cytokine tumor necrosis factor alpha (TNF- ) as well as by gp120. We also show that KSHV infection of human dermal microvascular endothelial cells (HDMEC) stimulates the production of IL-8 and GRO-. KSHV infection of HDMEC induces IL-8- and GRO- -dependent angiogenic activity in vitro. Furthermore, we show that IL-8 is central to the development of the angiogenic lesions seen when KSHV-infected endothelial cells are implanted into SCID mice. We implicate HIV-1- and KSHV-induced IL-8 and GRO- production in the angiogenic phenotype of Kaposi s sarcoma, offering one potential link between the three major branches of Kaposi s sarcoma pathogenesis KSHV, HIV-1 Tat, and cellular growth factors. MATERIALS AND METHODS Reagents. Recombinant human IL-8, polyclonal antibodies to IL-8, gamma interferon, and IL-1, and monoclonal antibodies to CXCR1, CXCR2, IL-8, and GRO- were obtained from R&D Systems (Minneapolis, Minn.) and added as indicated in the figure legends. Neutralizing antibodies to TNF- were used at a dilution of 1:200 (46). Antibodies capable of blocking HIV-1 binding to CD4 (RPA-T4), CCR5 (2D7), and CXCR4 (12G5) were added as indicated (no azide/low endotoxin format; PharMingen, San Diego, Calif.). AMD3100 was dissolved in phosphate-buffered saline (PBS) at 5 g/ l and used at doses ranging from 5 ng/ml to 5 g/ml (9, 83). HIV-1 BH10 Tat antiserum was obtained from the National Institutes of Health (NIH) AIDS Reagent Program and used at a 1:200 dilution. Lipopolysaccharide (LPS) from Escherichia coli strain O55: B5, polymyxin B, the E-TOXATE reagents kit, and all other chemicals were obtained from Sigma (St. Louis, Mo.). Cell lines. Human dermal microvascular endothelial cells (HDMEC) were purchased from Clonetics (Walkersville, Md.) and Cell Systems Corp. (Kirkland, Wash.). HDMEC were plated in tissue culture flasks (Corning Inc., Corning, N.Y.) coated with 0.01% gelatin and maintained in microvascular endothelial cell-specific growth medium (EGM-MV; Clonetics). BCBL-1, a body cavitybased lymphoma cell line infected with KSHV, was maintained in RPMI 1640 with 10% fetal bovine serum (FBS) and used to propagate KSHV as described previously (78). Isolation and preparation of human PBMC, PBL, and PBM. Peripheral blood mononuclear cells (PBMC) were collected by venipuncture of healthy volunteers as described previously (54). PBMC contained approximately 20% CD14 monocytes as determined by flow cytometry. In order to separate the PBMC into subpopulations composed mainly of monocytes or lymphocytes, PBMC were subjected to a plate adherence step for 2 h. Adherent cells were consistently 90% peripheral blood monocytes (PBM) as determined by Diff-Quik analysis and 85% CD14 by flow cytometric staining with a phycoerythrin-conjugated mouse anti-human CD14 monoclonal antibody (M5E2; PharMingen), as well as 99% viable as determined by trypan blue dye exclusion. PBM were differentiated into monocyte-derived macrophages (MDM) by culture in Dulbecco s modified Eagle s medium supplemented with 10% FBS, 2 mm glutamine, 100 U of penicillin per ml, and 100 g of streptomycin per ml (complete Dulbecco s modified Eagle s medium) for up to 2 weeks (3 days in most experiments) prior to infection. In the experiments in which MDM were cocultured with HDMEC, the adherent MDM were detached by incubation in PBS with 10 mm EDTA for 30 min at 4 C and then added to cultures of adherent HDMEC. The nonadherent cells following the plate adherence step were enriched for lymphocytes and contained less than 2% CD14 monocytes. These monocytedepleted PBMC (peripheral blood lymphocytes [PBL]) were cultured at to /ml in RPMI 1640 supplemented with 10% FBS, 2 mm glutamine, 100 U of penicillin per ml, and 100 g of streptomycin per ml (complete RPMI). PBL were stimulated with 5 g of phytohemagglutinin (Sigma) per ml for 1 to 3 days and then maintained in IL-2 (40 U/ml; Hoffmann-La Roche, Nutley, N.J.). In some experiments, PBL were also depleted of CD8 cells with magnetic Dynabeads (M-450 CD8) according to the manufacturer s instructions (Dynal, Lake Success, N.Y.). Preparation of HIV-1 stocks. All of the HIV-1 isolates used in this report were originally obtained from the NIH AIDS Reagent Program. Stocks of HIV-1 BaL were prepared by infection of HOS-CD4-CCR5 cells, and stocks of HIV-1 BRU were prepared by infection of CEM-SS cells. For some experiments, viral stocks were prepared by infection of phytohemagglutinin-activated, CD8-depleted PBL; results were identical to those obtained with the cell line-derived viral isolates (data not shown). In these experiments, mock-purified virus from uninfected PBL was used as a control. For each experiment, multiple wells of MDM were treated with equal reverse transcriptase counts of each isolate of HIV-1 ( to cpm of reverse transcriptase used per 10 5 cells). Intracellular cytokine staining. PBMC were collected as described above and cultured in RPMI with 10% FBS. The next day, PBMC were treated for 6 h with monensin (GolgiStop; PharMingen) and incubated with virus or TNF- as indicated in the figure legends. PBMC were then collected and stained for the presence of intracellular IL-8 and various cell surface markers according to the manufacturer s instructions (PharMingen). Adherent cells were collected by incubation in PBS with 10 mm EDTA for 30 min at 4 C. PBMC (both adherent and nonadherent cells) were incubated in staining buffer (Dulbecco s PBS with 1% FBS and 0.09% sodium azide) with mouse immunoglobulin G (IgG) for 20 min at 4 C to block nonspecific binding of IgG to target cells. PBMC were then washed with staining buffer and stained for cell surface antigens for 30 min at 4 C with R-phycoerythrin-conjugated mouse anti-human CD4 (phycoerythrin-cd4) or CD14 (phycoerythrin-cd14) monoclonal antibodies or phycoerythrin-mouse IgG1 isotype control immunoglobulin (PharMingen). After two washes with staining buffer, PBMC were fixed and permeabilized with Cytofix/Cytoperm (PharMingen). PBMC were washed twice in Perm/Wash solution and then stained with either fluorescein isothiocyanate (FITC)-conjugated mouse anti-human IL-8 monoclonal antibody or FITC-mouse IgG2a immunoglobulin isotype standard (PharMingen). Analysis of cell staining was performed with an EPICS Elite cell sorter (Beckman-Coulter, Fullerton, Calif.). The monocyte and lymphocyte subpopulations were gated first according to the pattern of forward scatter and side scatter and then by expression of CD14 (monocytes) or CD4 (lymphocytes). Cytokine ELISAs. Cellular supernatants were collected and stored at 70 C until analysis. Extracellular immunoreactive IL-8 and GRO- were measured

3 11572 LANE ET AL. J. VIROL. with a sandwich-type enzyme-linked immunoassay (ELISA) with capture and biotinylated detection antibodies according to the manufacturer s instructions (R&D Systems). The lower limit of detection for the IL-8 assay is 32 pg/ml and for the GRO- assay is 8 pg/ml. IL-8 and vascular endothelial growth factor (VEGF) were also evaluated with an in-house ELISA protocol as previously described (31, 55). This ELISA method consistently detected cytokine levels of 50 pg/ml. Northern blot analysis of IL-8 mrna. Total cellular RNA was extracted from uninfected and HIV-1 BRU -infected MDM with Trizol (Gibco-BRL) according to the manufacturer s instructions. Equal amounts of RNA were then analyzed for IL-8 gene expression by Northern (RNA) blot analysis. Ethidium bromidestained gels were photographed to demonstrate equivalent amounts of 28S and 18S rrnas in each sample. The IL-8 oligonucleotide probe was end labeled with polynucleotide kinase and [ - 32 P]dATP (87). The probe was annealed for 2.5 h at 68 C in QuikHyb solution (Stratagene, La Jolla, Calif.). Hybridization and low-stringency wash were performed in accordance with the manufacturer s instructions. KSHV infection. BCBL-1 cells (10 7 ) that were untreated or treated with phorbol 12-myristate 13-acetate (PMA) for 48 h were washed twice with RPMI 1640 and then resuspended in 1 ml of EGM-MV. Cell lysates were prepared by freeze-thawing the cells three times in liquid N 2 and filtering them through a m membrane. KSHV-containing lysates were then added to a 25-mm flask containing HDMEC at 80 to 90% confluency. After 48 h, infected cells were split and grown in EGM-MV. HDMEC and KSHV-infected HDMEC were grown to confluence in EGM-MV and passaged at least four times before being used in experiments. Serum-free conditioned medium containing KSHV was then collected from confluent cells after 24 h and concentrated and dialyzed against deionized water in a Centricon filter with a nominal molecular weight exclusion setting of M r 10,000 (Amicon, Beverly, Mass.). PCR. Total DNA present in the cellular lysates or supernatants was purified 12 passages after initial infection with KSHV, according to the manufacturer s instructions (Qiagen, San Clarita, Calif.). PCR was performed with primers that recognize and amplify KSHV-specific viral cyclin D, as previously described (21). PCR mixtures (50- l volume) containing 0.25 g of isolated DNA, 50 pmol of each primer, 200 M each deoxynucleotide triphosphate, 1.5 mm MgCl 2,50mM KCl, 20 mm Tris-HCl (ph 8.4), and 2.5 U of Taq DNA polymerase were set up. DNA was amplified as follows: 94 C for 3 min (1 cycle), 94 C for 30 s, 72 C for 2 min (40 cycles), and 72 C for 10 min (1 cycle). PCR products were electrophoresed through a 1% agarose gel containing ethidium bromide. As a positive control, KSHV DNA was isolated from PMA-treated BCBL-1 cells. In vitro angiogenesis assay. HDMEC were cultured in EGM-MV and used at passage 9, 10, or 11 after infection with KSHV. To measure migration, cells were washed three times and incubated overnight in MCDB-131 medium (Gibco- BRL) with 0.1% bovine serum albumin, harvested, resuspended in MCDB-131 with 0.1% bovine serum albumin, and plated at /well on the lower surface of a gelatinized 0.5- m filter (Nuclepore Corp., Pleasanton, Calif.) in an inverted modified Boyden chamber. After 1 to2hat37 C, during which time the cells adhered to the filter, the chamber was reinverted, test substance was added to the top well, and the chamber was incubated for 3 to 4hat37 C to allow migration. Chambers were then disassembled, the membrane was fixed and stained, and the number of cells that had migrated to the top of the filter in 10 high-power fields was counted. MCDB-131 with 0.1% bovine serum albumin was used as a negative control, and medium with recombinant IL-8 (100 ng/ml) or VEGF (50 ng/ml) was used as a positive control. Each sample was tested in quadruplicate within an experiment, and each experiment was repeated at least twice. Endothelial cells treated in parallel showed no toxicity, as monitored by trypan blue dye exclusion (73). Sponge implantation in SCID mice. Porous poly(l-lactic acid) (PLA) sponges were fabricated as previously described (47). Briefly, PLA (Aldrich Chemical Co., Inc., Milwaukee, Wis.) was dissolved in chloroform to yield a solution of 10% (wt/vol) polymer, and 0.12 ml of this solution was loaded into Teflon cylinders packed with 0.4 g of sodium chloride particles. The solvent was allowed to evaporate, and then the sponges were immersed for 16 h in an aqueous solution containing 10 mg of polyvinyl alcohol (Aldrich Chemical) per ml in PBS. The sponges (measuring approximately 6 mm by 6 mm by 1 mm, with an average pore diameter of 180 m) were dried, lyophilized, and sterilized by exposure to gamma radiation. The sponges were then soaked in 100% ethanol for 2 h, washed for 1 h in PBS, and left overnight in fresh PBS. HDMEC and KSHV-infected HDMEC were grown to confluence in EGM-MV and passaged four or five times. Just before implantation, 10 6 HD- MEC or KSHV-infected HDMEC were resuspended in a 1:1 mixture of EGM-MV and Matrigel (Collaborative Biomedical Products, Cambridge, Mass.) and allowed to adsorb into the sponges. More than 90% of the seeded cells were ultimately delivered in vivo. Male SCID mice (CB0.17.SCID; Taconic, Germantown, N.Y.), 3 to 4 weeks old, were anesthetized with ketamine and xylazine, and two sponges were implanted subcutaneously in the dorsal region of each mouse. At 7 and 14 days after implantation, mice were sacrificed, and the sponges were retrieved, fixed overnight in 10% buffered formalin, dehydrated through graded ethanol, embedded in paraffin, and mounted on Superfrost (Fisher Scientific, Pittsburgh, Pa.) glass slides for histological examination. Three sponges each from four to five mice were evaluated for each treatment condition. CD31 staining and microvessel assessment. Paraffin was removed from the sections with sequential dilutions of xylene and ethanol. The sections were then washed in fluorescent antibody buffer and treated with 0.5% trypsin 0.1% calcium chloride for 45 min at 37 C for antigen retrieval. Seventy-five microliters of rat anti-mouse CD31 (PECAM-1) (MEC13.3; PharMingen) at a concentration of 25 g/ml was applied to sections for 1 h at 37 C, and the anti-rat IgG Vectastain Elite ABC kit (Vector Laboratories, Inc., Burlingame, Calif.) was used according to the standard protocol. The endothelial cells were stained with 3-amino-9-ethyl carbazole developing solution for about 7 min and then counterstained with hematoxylin for 1 min. Microvessels were quantitated by a modification of a previously published method that enumerates vessels in the most vascular portion of the involved granulation tissue (91, 92). Sections were examined under low power (40 to 100 ) to identify the regions of highest vessel density, and vessels were then counted in each of 10 high-power fields (400 ). A vessel lumen was not required for identification of a microvessel; cell clusters were also counted. Counts were then expressed as the total number of microvessels in 10 high-power fields. Statistics. The data shown are representative of at least three experiments and are expressed as the mean of triplicate wells ( standard deviation) throughout unless otherwise noted. Statistical significance was evaluated by t test for normally distributed data and by Wilcoxon signed ranks test for nonparametric data. A P value of 0.05 was regarded as statistically significant. RESULTS IL-8 production by individual PBMC exposed to HIV-1. We have recently observed that elevated levels of IL-8 are present in the lymphoid tissue of individuals with AIDS (53). Previous studies have found increased concentrations of IL-8 in the serum and tissue of HIV-infected individuals as well (12, 23, 51, 60, 70). In order to determine which cells are making IL-8 in response to HIV-1, we analyzed the production of IL-8 by individual PBMC following an encounter with HIV-1 by flow cytometry. Fresh PBMC were cultured overnight without phytohemagglutinin stimulation and then exposed to HIV-1 BaL, HIV- 1 BRU, or TNF- together with monensin to prevent the secretion of IL-8. Staining after 6 h revealed that IL-8 was present in 90% of CD14 monocytes treated with either HIV-1 BRU or TNF-, in 57% of CD14 monocytes treated with HIV- 1 BaL, and in 10% of control monocytes (Fig. 1). IL-8 staining was detected in fewer than 5% of CD4 lymphocytes under all conditions (Fig. 1). Similarly, while phytohemagglutinin-activated PBL constitutively produce appreciable levels of IL-8, HIV-1 only marginally stimulated IL-8 production by PBL above this endogenous level (data not shown). Therefore, the cells responsible for the production of IL-8 in response to HIV-1 appear to be mononuclear phagocytes. Production of IL-8 by MDM exposed to HIV-1. We have previously shown that an encounter with HIV-1 virions or the viral envelope glycoprotein gp120 stimulates the production of GRO- by MDM via interaction with the chemokine receptor CXCR4 (55). Although isolates of HIV-1 that use CXCR4 as an entry cofactor (X4 HIV) did not infect MDM, they stimulated far greater production of GRO- by MDM than CCR5- using HIV-1 (R5 HIV) (55). In contrast, when IL-8 production by MDM was examined following an encounter with several

4 VOL. 76, 2002 IL-8, GRO-, AND ANGIOGENESIS IN KAPOSI S SARCOMA FIG. 1. Intracellular IL-8 is present in the monocytic subpopulation of PBMC exposed to HIV-1. PBMC were treated with monensin (control) along with HIV-1 BaL, HIV-1 BRU, or TNF- (100 ng/ml). PBMC were harvested after 6 h and analyzed for intracellular IL-8 protein by flow cytometry. (A) Lymphocyte and monocyte subpopulations were gated according to the pattern of forward scatter and side scatter. (B) Background staining in the lymphocytes and monocytes was determined by incubation with phycoerythrin-mouse IgG1 isotype control (mouse IgG-PE) and FITC-mouse IgG2a isotype control (mouse IgG-FITC). The histograms show fluorescence intensity on logarithmic scales along the x axis (FITC) and y axis (phycoerythrin). The percentage of FITC-positive cells is indicated for both the double-positive and single-positive quadrants. (C) Lymphocytes and monocytes were stained with an FITC-conjugated mouse anti-human IL-8 antibody (IL-8-FITC) and either phycoerythrinconjugated mouse anti-human CD4 (CD4-PE) or phycoerythrin-conjugated mouse anti-human CD14 (CD14-PE), respectively. The percentage of CD4 and CD14 cells staining positive for IL-8 in each condition is indicated. Data shown are representative of four independent experiments. isolates of HIV-1, an increase in levels of IL-8 was found with isolates that use either CCR5 or CXCR4 as a coreceptor (Fig. 2A). In experiments with MDM from a large number of donors, the stimulation of IL-8 production was similar after an encounter with either a prototypical R5 HIV isolate, HIV-1 BaL (15- fold), or an X4 HIV isolate, HIV-1 BRU (28-fold) (Fig. 2A). IL-8 production is independent of viral replication, as the X4

5 11574 LANE ET AL. J. VIROL. FIG. 2. HIV-1 and the viral proteins Tat and gp120 stimulate IL-8 production by MDM. (A) MDM were left untreated (control) or treated with either HIV-1 Tat (10 ng/ml; from HIV-1 HXB2 ), R5 gp120 (1 g/ml; from HIV-1 CM235 ), HIV-1 BaL, X4 gp120 (1 g/ml; from HIV-1 MN ), or HIV-1 BRU. Supernatants were collected from cultures in triplicate after 24 to 48 h and tested for IL-8 by ELISA. A box plot of the amounts of IL-8 for each treatment group is shown. Each point represents a single experiment, with the center horizontal line marking the median of the sample. The length of each box shows the range within which the central 50% of the values fell (Hspread), with the box edges (hinges) at the first and third quartiles. The whiskers show the range of values that fell within 1.5 Hspreads of the hinges. Values outside the whiskers are plotted with asterisks. Also indicated for each treatment condition are the mean amount of IL-8 produced, the number of experiments performed (n), and a statistical measure of the difference in the amount of IL-8 relative to the untreated controls with the Wilcoxon signed ranks test for nonparametric data (P value). (B) Total cellular RNA was extracted from control MDM ( ) and MDM exposed to HIV-1 BRU ( ) after 2 days. RNA from two different donors was then analyzed for IL-8 gene expression by Northern (RNA) blot analysis with an IL-8-specific probe. isolate HIV-1 BRU does not replicate in MDM (32, 55), and IL-8 is produced several days before viral replication is detected in MDM infected with HIV-1 BaL (53, 55). In addition, Northern blot analysis of MDM exposed to HIV-1 BRU indicated an increase in the steady-state level of IL-8 mrna (Fig. 2B). This change in mrna is consistent with the increase detected at the protein level by ELISA, indicating that the increase in IL-8 levels is due, at least in part, to an increase in transcript stability and/or an activation of IL-8 transcription. In order to assess the specificity of the response of MDM to HIV-1, we examined the ability of other viral and bacterial antigens to stimulate IL-8 production. Exposure to neither adenovirus nor Epstein-Barr virus stimulated MDM to produce amounts of IL-8 comparable to those in MDM exposed to HIV-1 (data not shown). Lipopolysaccharide (LPS) in the stocks was unlikely to be responsible for the production of IL-8 by MDM, as treatment with polymyxin B (10 g/ml) blunted the effect of LPS but did not significantly reduce the increases in IL-8 seen with either HIV-1 or recombinant proteins (data not shown). More importantly, the effect of HIV-1 was inhibited by antibodies that block the action of gp120 and Tat but not by control antibodies (see below), identifying specific mechanisms by which HIV-1 stimulates IL-8 production. Therefore, IL-8 is not produced because MDM are activated nonspecifically by viral antigens or by LPS in the preparations of virus. Rather, these data, along with the findings described below, indicate that MDM produce IL-8 specifically in response to an encounter with HIV-1 virions and proteins and independently of productive infection by HIV-1. HIV-1 gp120 binding to CXCR4 is partially responsible for induction of IL-8 production by MDM. Previous investigators have shown that the HIV-1 proteins Tat and gp120 are capable of inducing IL-8 production by monocytes (13, 51), although they did not study these effects in the context of the whole virus. We hypothesized that the interaction of HIV-1 gp120 with CXCR4 is the major mechanism by which HIV-1 stimulates MDM to produce IL-8, as we had found with GRO- (55). In order to test this hypothesis, we first incubated MDM with recombinant HIV-1 gp120 from R5 HIV and X4 HIV, as well as with recombinant HIV-1 Tat. MDM produced IL-8 in response to treatment with soluble gp120 from X4 HIV but not from R5 HIV, and gp120 from X4 virus did not stimulate IL-8 production to the same extent as did whole virus (Fig. 2A). In contrast to our findings with GRO- (55), HIV-1 Tat induced a significant increase in the amount of IL-8, comparable to that induced by HIV-1 BaL (Fig. 2A). In order to determine which factor(s) was responsible for the increase in IL-8 production observed in the context of live HIV-1, we next incubated MDM with antibodies that prevent interaction with CD4, CCR5, or CXCR4 prior to exposure to

6 VOL. 76, 2002 IL-8, GRO-, AND ANGIOGENESIS IN KAPOSI S SARCOMA FIG. 3. HIV-1 gp120 and Tat both induce IL-8 production by MDM exposed to X4 HIV. (A) MDM were treated with mouse IgG, anti-cd4, anti-ccr5, or anti-cxcr4 (each at 20 g/ml) and then infected with HIV-1 BRU. Supernatants were collected 1 or 2 days after infection and analyzed for IL-8 by ELISA. Data shown are the means standard errors from five independent experiments. The reduction in IL-8 production due to anti-cxcr4 was significant (P 0.04) relative to that of the mouse IgG control by Student s t test. (B) MDM were treated with the CXCR4 inhibitor AMD3100 (5 g/ml), normal rabbit serum (NRS, 1:200), anti-tat (1:200), or anti-tnf- (1:200) as indicated, alone or in combination, just prior to exposure to HIV-1 BRU and every 3 days thereafter. Supernatants were collected 8 days after infection and analyzed for IL-8 by ELISA. Data shown are the means standard errors from three independent experiments. HIV-1. Anti-CXCR4 attenuated the increase in IL-8 production with HIV-1 BRU in part, while anti-cd4 and anti-ccr5 had no effect on IL-8 production induced by either HIV-1 BRU or HIV-1 BaL (Fig. 3A and data not shown). These findings are FIG. 4. Heat inactivation inhibits the stimulation of IL-8 production by HIV-1 in MDM. HIV-1 BaL or HIV-1 BRU was either left untreated, heated at 56 C for 30 min, or heated at 100 C for 10 min and then used to infect MDM from two different donors. Supernatants were collected 40 h after infection and analyzed for IL-8 by ELISA. IL-8 production is presented as a percentage of the amount produced by MDM infected with active HIV-1 by the following formula: [(IL-8 in heat-inactivated HIV-infected MDM IL-8 in uninfected MDM)/ (IL-8 in active HIV-infected MDM IL-8 in uninfected MDM)] 100. In these two experiments, the mean amount of IL-8 induced by HIV-1 BaL was 206 ng/ml, and that induced by HIV-1 BRU was 422 ng/ml. similar to our previous observation, as well as those of others, that signals induced in response to HIV-1 gp120 ligation of CXCR4 can occur independently of interaction with CD4 (55, 58, 98). Incubation of HIV-1 BRU -infected MDM with the CXCR4-specific inhibitor AMD3100 (9, 83) also reduced IL-8 production by about 50% (Fig. 3B), confirming the involvement of CXCR4. These findings suggest that while ligation of CXCR4 by gp120 induces IL-8 production, HIV-1 isolates that bind to either CXCR4 or CCR5 induce IL-8 by an independent mechanism involving Tat. MDM were next exposed to virus that had been subjected to either of two heat treatments, 56 C for 30 min or 100 C for 10 min. The first treatment has been reported to inactivate most proteins (e.g., Tat) but not more heat-stable factors (e.g., gp120), while incubation at the higher temperature destroys both heat-labile and relatively heat-stable factors, including gp120 (17). Heating either HIV-1 BaL or HIV-1 BRU to 100 C for 10 min eliminated their ability to stimulate IL-8 production by MDM (Fig. 4). Interestingly, HIV-1 BaL -induced IL-8 production was also completely sensitive to the 56 C heating, while IL-8 production by HIV-1 BRU -treated MDM was reduced by only 50% (Fig. 4). These data implicate a heat-stable factor, such as gp120, in the production of IL-8 by MDM in response to HIV-1 BRU, as well as a heat-labile factor, such as HIV-1 Tat, in the response to both HIV-1 BRU and HIV-1 BaL. HIV-1 Tat and TNF- together induce IL-8 production by MDM. Previous studies have found that HIV-1 Tat, in concert with costimulatory signals, can stimulate IL-8 production in PBL and endothelial cells (45, 70). We thus investigated whether Tat and inflammatory cytokines that are themselves

7 11576 LANE ET AL. J. VIROL. FIG. 5. HIV-1 Tat and TNF- stimulate IL-8 production by MDM. (A) Dose-response curve for treatment of MDM with recombinant HIV-1 Tat (1 to 1,000 ng/ml). ELISA data are plotted as the mean amount of IL-8 present in 24-h supernatants from multiple wells from two experiments ( standard deviation). Statistical significance (*, P 0.05) was determined with a paired-sample t test. (B) MDM were treated with normal rabbit serum (NRS, 1:200), anti-tat (1:200), or anti-tnf- (1:200) as indicated, alone or in combination, just prior to exposure to HIV-1 BaL and every 3 days thereafter. Supernatants were collected 1, 2, 4, 6, and 8 days after infection and analyzed for IL-8 by ELISA. Data shown are representative of three independent experiments. FIG. 6. The KSHV cyclin D gene is detected in KSHV-infected HDMEC by PCR after multiple passages in culture. After 12 passages, total DNA present in the cellular lysates was purified from uninfected HDMEC (lane 2), HDMEC treated with unstimulated BCBL-1 cell lysate (mock-infected, lane 3), and HDMEC treated with 48-h PMAstimulated BCBL-1 cell lysate (KSHV-infected, lane 4). DNA from PMA-stimulated BCBL-1 cells (lane 5) served as a positive control. PCR analysis demonstrated the presence of the viral cyclin D gene in the BCBL-1 cells and in KSHV-infected HDMEC. HindIII-digested DNA markers (lane 1) are shown for size comparison. regulated by HIV-1 (44, 62, 71) might be involved in the stimulation of IL-8 production by MDM exposed to HIV-1. Recombinant HIV-1 Tat induced IL-8 production by MDM in a dose-dependent manner, with peak activity at 100 ng/ml (Fig. 5A). An even lower dose of Tat (10 ng/ml) caused an increase in IL-8 comparable to that caused by exposure to whole virus (Fig. 2A). The effect of Tat on IL-8 production is likely physiological, as Tat is present in the serum of HIV-infected individuals at levels as high as 1 ng/ml (94) and would presumably be present at a much higher level in the microenvironment of infected tissue. In contrast, GRO- production was not affected by Tat treatment but was significantly stimulated by exposure to HIV-1 BRU and X4 gp120 (55). In order to further test whether Tat was responsible for part of the increase in IL-8 seen in the context of exposure to whole virus, we incubated MDM with antibodies known to neutralize Tat and the inflammatory mediators TNF-, IL-1, and gamma interferon prior to exposure to HIV-1. While neutralization of these cytokines did not significantly affect GRO- production, neutralization of TNF-, especially, reduced the IL-8 production stimulated by HIV-1 (Fig. 3B and 5B and data not shown). In MDM infected with HIV-1 BaL, anti-tat and anti-tnf- each substantially reduced and in combination eliminated IL-8 production throughout the course of infection (Fig. 5B). Anti- Tat and anti-tnf- also reduced IL-8 production by MDM exposed to the X4 isolate HIV-1 BRU and, when added along with the CXCR4 inhibitor AMD3100, prevented nearly all of the increase in IL-8 production (Fig. 3B). Thus, Tat appears to be the sole viral factor responsible for the stimulation of IL-8 production following exposure to R5 HIV and a major mediator of IL-8 production in response to X4 HIV. KSHV infection stimulates IL-8 and GRO- production by HDMEC. Kaposi s sarcoma is an angioproliferative disorder known to be caused by KSHV in concert with HIV-1 Tat and cellular growth factors (8, 26). As the ELR-positive C-X-C chemokines IL-8 and GRO- are known angiogenic factors, we hypothesized that the increase in the levels of IL-8 and GRO- caused by HIV-1 may create an angiogenic environment which would encourage the development and progression of Kaposi s sarcoma in individuals infected with HIV-1. To test this hypothesis, we developed and used a model of Kaposi s sarcoma in which human dermal microvascular endothelial cells (HDMEC) are infected with KSHV and then studied either in tissue culture or following injection into the flank of SCID mice. KSHV was harvested from the cell-free supernatants of PMA-stimulated BCBL-1 cells (35) and used to infect the HDMEC in vitro. HDMEC treated with PMA-stimulated BCBL-1 cell lysates were passaged multiple times in order to generate a population of HDMEC infected with KSHV. HD- MEC maintained viral genomes after as many as 12 passages, as evidenced by the detection of the KSHV cyclin D gene by PCR analysis (Fig. 6). Uninfected HDMEC and HDMEC

8 VOL. 76, 2002 IL-8, GRO-, AND ANGIOGENESIS IN KAPOSI S SARCOMA FIG. 7. KSHV infection of HDMEC induces the release of the angiogenic factors IL-8, GRO-, and VEGF. HDMEC (80 to 95% confluent) were left uninfected (control) or infected with KSHV and maintained in culture for one to four passages. Fresh medium was added to the KSHV-infected HDMEC and collected 24 h later and analyzed by ELISA for IL-8 (A), GRO- (B), or VEGF (C). Data shown in panels A and B are the means from four experiments ( standard deviation), and those in panel C are from a representative experiment. treated with unstimulated BCBL-1 lysates did not contain detectable amounts of viral DNA (Fig. 6). In addition, HDMEC, which exhibit a typical cobblestone morphology when cultured in EGM-MV, displayed a spindle-shaped morphology after incubation with lysates of BCBL-1 cells (data not shown), similar to that of the spindle cells present in Kaposi s sarcoma lesions in vivo. Previous investigators have seen this spindle cell morphology following infection with KSHV (11, 16), providing further evidence that the HDMEC were in fact infected with KSHV. In our hands, HDMEC that have been infected with KSHV maintain viral genomes for as many as 12 passages in culture and exhibit phenotypic characteristics similar to those of the spindle cells present in Kaposi s sarcoma lesions. Since Kaposi s sarcoma is a highly vascular lesion, we hypothesized that infection with KSHV might shift the balance of cytokines present in favor of angiogenesis (43). To test this hypothesis, we analyzed the supernatants of multiply passaged KSHV-infected HDMEC and uninfected parental HDMEC for the presence of several factors known to play a role in angiogenesis (49, 67). IL-8 and GRO- production were increased threefold in the KSHV-infected HDMEC relative to uninfected HDMEC (Fig. 7A and 7B). While the increase in chemokine was not as pronounced as that following exposure of MDM to HIV-1, the stimulation was statistically significant for both IL-8 and GRO- (P 0.05). As these HDMEC had been passaged multiple times after infection and before being used in the experiments, nonspecific factors in the KSHV stock originally used to infect the HDMEC are extremely unlikely to have caused the increase in IL-8 or GRO- production. The amount of VEGF was also increased in the KSHV-infected HDMEC (Fig. 7C). As HIV-1 stimulates MDM to produce IL-8 and GRO-, as does KSHV infection of HDMEC, HIV-1 infection might contribute to the development of Kaposi s sarcoma by further shifting the chemokine balance to favor angiogenesis. We therefore examined whether KSHV and HIV-1 combine to stimulate even greater IL-8 or GRO- production by infected cells than either virus alone. Indeed, KSHV and HIV-1 together induced a 34-fold increase in IL-8 production and a 15.6-fold increase in GRO- production in cocultures of HD- MEC and MDM (data not shown). Thus, in individuals infected with both viruses, HIV-1 may further augment IL-8 and GRO- production, potentiating the proangiogenic effect of KSHV alone. IL-8 promotes angiogenesis in vitro and in an animal model of Kaposi s sarcoma. To investigate how KSHV infection might stimulate angiogenesis and to determine whether IL-8, GRO-, and VEGF contribute to this angiogenic activity, we analyzed the supernatants of KSHV-infected HDMEC in an endothelial cell migration assay (69). Migration was standardized against that of recombinant IL-8, which could be depleted to less than 5% of its activity by an antibody that neutralizes IL-8 bioactivity (Fig. 8). While supernatants from uninfected HDMEC did not induce the migration of endothelial cells, the conditioned medium from KSHV-infected HDMEC contained substantial migratory activity (Fig. 8). As IL-8, GRO-, and VEGF are induced by KSHV and the latter has been implicated in Kaposi s sarcoma pathogenesis (20, 93), we tested whether they contribute to the induction of endothelial cell migration by using the conditioned medium from KSHV-infected HDMEC. Anti-IL-8 and anti-gro- antibodies each neutralized about 50% of the angiogenic activity, and addition of both antibodies eliminated almost all migratory activity (Fig. 8). In contrast, anti-vegf antibody did not affect the migration of KSHV-infected HDMEC in this assay (Fig. 8). We have shown previously that this anti-vegf antibody prevents both the chemotactic and angiogenic activities of VEGF (67). These studies suggest that IL-8 and GRO- play important roles in the early angiogenic activity induced by KSHV infection, while VEGF, which likely contributes to the pathogenesis of Kaposi s sarcoma at a later step, plays little part in this early activity.

9 11578 LANE ET AL. J. VIROL. FIG. 8. IL-8 and GRO- are major components of the angiogenic activity present in the conditioned medium of KHSV-infected HDMEC. Conditioned medium derived from uninfected or KSHV-infected HDMEC, with or without neutralizing antibodies to IL-8, GRO-, or VEGF, was assayed for the ability to stimulate the chemotaxis of endothelial cells. Specific migration was determined after subtracting background migration (unstimulated control). Values are reported as a percentage ( standard error of the mean) of the migration induced by recombinant IL-8. In order to examine the role of IL-8 in the angiogenic phenotype of KSHV-infected lesions in vivo, we developed an animal model of Kaposi s sarcoma by modifying our protocol for implanting SCID mice with endothelial cells seeded into porous PLA sponges (69). The HDMEC which were either left uninfected or infected with KSHV were cultured for at least five passages in vitro (to eliminate the nonspecific influence of any factors in the supernatants originally used to infect the HDMEC) and then seeded into a PLA sponge and implanted into the flank of SCID mice. PLA sponges implanted with KSHV-infected HDMEC stimulated the migration of newly formed vessels into the transplanted tissue, whereas uninfected HDMEC did not (Fig. 9A). Murine blood vessels respond to angiogenic factors produced by HDMEC in the implanted tissue by sprouting and growing into the site of the lesion (69). These vessels can be visualized by immunostaining of the KSHV-infected transplanted tissue for the endothelial cell-specific marker CD31 (91), as shown in Fig. 9B. A significant increase in the number of new vessels was detected in KSHV-infected transplanted tissue compared with tissue transplanted with uninfected HDMEC (Fig. 9C). While implants containing KSHV-infected HDMEC stimulate angiogenesis, they do not induce tumor formation in this model unless the HDMEC overexpress the oncogene Bcl-2 (68), suggesting that angiogenesis precedes tumorigenesis in this model of Kaposi s sarcoma, in agreement with the Hanahan and Folkman model of tumor angiogenesis (43). In order to examine the role of IL-8 and VEGF in the new vessel formation in this in vivo model of Kaposi s sarcoma, we added antibodies that neutralize IL-8 or VEGF to the HDMEC-containing sponges prior to implantation. Administration of anti-il-8 antibody reduced the number of vessels staining positive for CD31 by greater than 50%, while the anti-vegf antibody had little effect on the angiogenic response in the SCID mouse model system (Fig. 9C). Taken together, our data from KSHV-infected HDMEC cultured in vitro and from the animal model demonstrate the involvement of IL-8, and likely GRO-, in the development of the angiogenic phenotype of Kaposi s sarcoma. DISCUSSION Previous investigators have found elevated levels of IL-8 in the serum and tissues of HIV-infected individuals and have suggested but not demonstrated experimentally that this chemokine may impact HIV-1 pathogenesis (12, 23, 51, 60, 70). We have recently found elevated levels of IL-8 in the lymphoid tissue of patients with AIDS and have also discovered that IL-8 stimulates HIV-1 replication (53). We have also previously shown that a related C-X-C chemokine, GRO-, is induced by exposure of MDM to HIV-1 and feeds back to stimulate HIV-1 replication in MDM and T cells (55). In this report, we characterized the mechanisms by which HIV-1 induces IL-8 production and linked IL-8 and GRO- to the angiogenic activity of endothelial cells infected with KSHV. We found that the mechanism responsible for the production of GRO- by MDM exposed to HIV-1, namely, ligation of CXCR4 by viral gp120 (55), is not the sole mechanism responsible for IL-8 production. Our data confirm earlier findings indicating that HIV-1 triggers IL-8 release by PBM and MDM (29, 30) and extend these findings by demonstrating that different factors are responsible for the effects of R5 HIV and X4 HIV. Specifically, R5 HIV stimulates IL-8 production via a mechanism involving Tat and TNF-, while the effect of X4 HIV is mediated by Tat and TNF- as well as by gp120. Indeed, treatment of MDM with recombinant gp120 from X4 HIV but not gp120 from R5 HIV

10 VOL. 76, 2002 IL-8, GRO-, AND ANGIOGENESIS IN KAPOSI S SARCOMA enhanced IL-8 production. However, while gp120 induced IL-8 production, it cannot be the sole mechanism by which HIV-1 acts, since an antibody that prevents gp120 interaction with the HIV-1 coreceptor CXCR4 and an inhibitor of CXCR4 signaling only partially inhibited the stimulation of IL-8 seen with whole virus. In addition, in contrast to GRO- production, induction of IL-8 was much greater following exposure to virus than following treatment with recombinant gp120 (55). While gp120 has been shown previously to induce IL-8 production by monocytes (13), we now confirmed this in the context of the whole virus. Furthermore, we add the finding that only gp120 from X4 HIV is able to have this effect and that even in this case, X4 gp120 is responsible for only part of the effect of HIV-1 on IL-8 production. HIV-1 Tat has been demonstrated to stimulate the release of IL-8 by PBL, PBM, astrocytes, and brain-derived endothelial cells (45, 50, 51, 70). We now confirm the production of IL-8 by PBM and MDM in response to Tat in the context of the whole virus. We observed that antibodies which neutralize Tat and TNF- substantially reduced IL-8 production induced by X4 HIV and completely eliminated IL-8 production in response to R5 HIV. The relevance of these findings is further FIG. 9. KSHV-infected HDMEC stimulate IL-8-dependent angiogenesis in SCID mice. Cells (10 6 ) were passaged five times (to avoid confounding effects from contaminants in the original inoculum of KSHV) before being seeded into a PLA sponge. (A) HDMEC (left panel) and KSHV-infected HDMEC (right panel) were seeded into a PLA sponge and implanted in duplicate into the flanks of SCID mice (n 4). After 14 days, implanted sponges were removed from the mice, and photographs were taken with a Nikon anatomy scope (magnification, 1.5 to 2). (B) The sponges were stained with a rat anti-mouse CD31 antibody at high power (400 ). (C) Cells were seeded in a PLA sponge with or without neutralizing antibodies to VEGF or IL-8 and implanted into the flanks of SCID mice in duplicate. After 14 days, the PLA sponges were removed, and CD31-positive blood vessels were counted in 10 random fields from three independent sponges for each condition. strengthened by our finding of elevated levels of IL-8 in the lymphoid tissue of AIDS patients (53). Tat is the HIV-1 protein with the strongest link to the pathogenesis of Kaposi s sarcoma. Although Kaposi s sarcoma can develop in the absence of HIV-1 infection, as is seen in classic Kaposi s sarcoma and iatrogenic Kaposi s sarcoma, Tat has been implicated in the growth of Kaposi s sarcoma cells and, in combination with KSHV and inflammatory cytokines, is believed to contribute to the formation of Kaposi s sarcoma lesions (26, 28). Tat can act in a paracrine fashion, as extracellular Tat is able to bind to integrin receptors (4, 27). Tat has recently been shown to stimulate intracellular signaling cascades, activating NF- B and leading to the production of several cytokines, including IL-8 (76). Tat can also enhance the activity of cellular mediators of angiogenesis, such as basic fibroblast growth factor (bfgf), but the mechanism by which Tat functions in this capacity has remained unknown (28). Several in vitro models involving either primary Kaposi s sarcoma specimens or KSHV-infected primary effusion lymphoma (PEL) cell lines have been used to study Kaposi s sarcoma pathogenesis and transformation (11, 34, 35, 65, 78). For our tissue culture and mouse models, we infected endothelial cells with KSHV by treating HDMEC with lysates from PMAstimulated BCBL-1 cells. In our hands, this establishes an infection lasting for at least 12 passages in culture, as determined by PCR of the KSHV cyclin D gene. We have also found that KSHV infection of HDMEC stimulates proliferation of