Kaposi's sarcoma (KS) is caused by Kaposi's sarcoma–associated herpesvirus (KSHV) and consists of proliferating spindle cells, which are related to lymphatic endothelial cells (LEC). Angiopoietin-2 (Ang2) is a secreted proangiogenic and lymphangiogenic molecule. Here, we show the expression of Ang2 protein in KS and confirm that KSHV infection up-regulates Ang2 in LEC. We show that a paracrine mechanism contributes to this up-regulation. A lentiviral library of individual KSHV-encoding genes, comprising the majority of known latent genes and a selection of lytic viral genes, was constructed to investigate the underlying mechanism of this up-regulation. Two lytic genes, viral interleukin-6 (vIL6) and viral G-protein–coupled receptor (vGPCR), up-regulated Ang2 expression in LEC. Both vIL6 and vGPCR are expressed in KSHV-infected LEC and caused up-regulation of Ang2 in a paracrine manner. KSHV, vIL6, and vGPCR up-regulated Ang2 through the mitogen-activated protein kinase (MAPK) pathway. Gene expression microarray analysis identified several other angiogenic molecules affected by KSHV, including the vascular endothelial growth factor (VEGF)/VEGF receptor (VEGFR) axis, which is also affected by vIL6 and vGPCR in LEC, and matrix metalloproteinases, which could act in concert with Ang2 to contribute to KS development. These findings support the paracrine and autocrine roles of the lytic KSHV-encoded proteins, vIL6 and vGPCR, in KS pathogenesis and identify Ang2 as a potential therapeutic target for this neoplasm. [Cancer Res 2007;67(9):4042–50]
Kaposi's sarcoma–associated herpesvirus (KSHV) is etiologically linked to Kaposi's sarcoma (KS) and the lymphoproliferative disorders primary effusion lymphoma and multicentric Castleman's disease (1–4). The KSHV genome consists of more than 80 open reading frames (ORFs; ref. 5). Only a fraction of viral genes are expressed during latency and include genes pirated from the host genome, such as the viral cyclin (vcyclin) and viral FLICE inhibitor protein (vFLIP; refs. 6, 7). Other cellular homologues encoded by KSHV include a G-protein–coupled receptor (vGPCR), viral IL6 (vIL6), IFN regulatory factors (vIRF), and chemokine homologues. The majority of these ORFs are expressed only during lytic viral replication (7).
KS is a vascular tumor that consists of sheets of proliferating spindle cells (the KS tumor cells), an inflammatory infiltrate, and abnormal, leaky, slit-like blood vessels (7, 8). KSHV is present in the vast majority of spindle cells (1) and is thought to contribute directly to their proliferation and immune escape (4, 7). The spindle cells express markers of endothelial cells (EC), and their gene expression microarray (GEM) profile is closest to that of lymphatic EC (LEC) (9).
Angiopoietin-2 (Ang2) mRNA has been found in KS lesions by in situ hybridization, and its levels are up-regulated in KS as determined by GEM analysis (9, 10). Furthermore, Ang2 levels are increased in the plasma of individuals with KS, correlate with number of lesions, and decline during antiretroviral therapy when KS resolves (9). Ang2 and Ang1 are two members of the angiopoietin family, both binding with similar affinities to the receptor tyrosine kinase Tie2, which is expressed on EC (11–13). Ang1 acts as an agonist to the Tie2 receptor, whereas Ang2 is characterized as an antagonist, although recent in vitro studies show that (in certain conditions) Ang2 may activate Tie2 at weaker potency than Ang1 (11, 14, 15). Knockdown of Tie2 expression in EC leads to apoptosis, and Tie2 knock-out mice are embryonically lethal, showing a low vascular complexity (16, 17). Embryonic lethality with an abnormal vascular network is also seen in Ang1 knock-out and in Ang2 overexpressing mice, which supports the in vivo agonistic and antagonistic roles of these molecules, respectively (11, 18).
In its role as an antagonist, Ang2 is thought to destabilize EC and prime their response to other proangiogenic stimuli, such as VEGF (11, 14). Ang2 also plays an important role in lymphangiogenesis as manifested in the defective, leaky lymphatic system of Ang2 knock-out mice (14).
Unlike Ang1, Ang2 is overexpressed in many cancers (19). Ectopic expression of Ang2 by cancer cells increases angiogenesis, thereby promoting tumor growth in experimental models (20), and Ang2-specific inhibitors hinder the growth of tumor xenografts by directly preventing angiogenesis (21). Furthermore, Ang2 regulates the expression of inflammatory proteins in EC, which suggests its contribution to tumor-associated inflammation (22).
These studies show that Ang2 may play a crucial role in KS pathogenesis by contributing to KSHV-infected EC proliferation, angiogenesis, and the associated inflammatory infiltrate. In the present study, we investigate Ang2 expression in KS lesions and the mechanism by which KSHV up-regulates Ang2 in LEC.
Materials and Methods
Cell culture. LEC were obtained, verified as previously described (23), and cultured on fibronectin (Marathon)-coated flasks or wells in Endothelial Cell Growth Medium MV (PromoCell) supplemented with 10 ng/mL of VEGF-C (R&D Systems). LEC were used for experiments at passages 5 to 8. Conditioned media from LEC or infected LEC were obtained by incubating cells for 24 to 48 h with media [overnight for media with only 0.5% fetal bovine serum (FBS)] and passing the media through a 0.1-μm filter to remove any cell debris and KSHV virions. The filtered conditioned media were either added to LEC or stored at −80°C for later use.
293T cells were cultured in DMEM (Invitrogen) containing 10% FBS (Sigma), 100 units/mL penicillin G, and 100 μg/mL streptomycin (Invitrogen). BCBL-1 cells containing a green fluorescent protein (GFP)–expressing recombinant KSHV (24) were cultured in RPMI 1640 (Invitrogen) containing 10% FBS and 400 ng/mL geneticin (Invitrogen).
KSHV production and infection of LEC. KSHV was produced as previously described (9), and 2 mL of the concentrated virus was used to infect ∼1 × 105 LEC. Infections typically resulted in more than 35% of LEC-expressing GFP 3 to 4 days postinfection (p.i.). GFP-positive KSHV-infected LEC (KLEC) were isolated using the cell sorter Moflo (DAKO).
KSHV lentiviral library. KSHV ORFs from ref. (5) and recent literature were cloned into the lentiviral vector pSIN-MCS (Supplementary Fig. S1A) produced from the pCSGW vector (Supplementary Fig. S1A; ref. 25). Cloned KSHV-encoded genes comprise most known latent genes and a selection of lytic genes (Fig. 2A). ORFs were cloned from PCR products using BC-3 (KSHV-positive, EBV-negative cell line) cDNA or genomic DNA or were cloned from previously produced plasmids. PCR used for cloning was done using the high-fidelity PfuTurbo DNA polymerase (Stratagene), and the primer sequences used are available upon request. Previously unrecorded coding polymorphisms detected were confirmed by sequencing PCR products from BC-3 genomic DNA or cDNA.
Lentivirus production and infection of LEC. Vesicular stomatitis virus-G envelope-pseudotyped lentiviral virions were produced by cotransfecting 2 μg lentiviral (pSIN-MCS or pCSGW) construct, 1.5 μg p8.91, and 1.5 μg pMD.G (25) into a 10-cm dish of ∼70% confluent 293T cells using the FuGENE (Roche) protocol. Five hours after transfection, the medium was changed, and 48 h after transfection, the medium containing the lentiviral virions was collected, passed through a 0.45-μm filter, and either aliquoted directly or concentrated and stored at −80°C. Lentiviral infections were done by incubating the desired amount of virus preparation with LEC in culture, typically 1 mL per 1 × 105 LEC, for 5 h, after which the medium was changed. The expression of constructs was confirmed by reverse transcription-PCR (RT-PCR). GFP expression in pCSGW-infected cells was assessed by fluorescent microscopy or flow cytometry using a FACSCalibur flow cytometer (BD Biosciences).
Quantitative PCR for the titration of lentivirus preparations. To determine the number of lentiviral copies per cell, quantitative PCR (qPCR) was done for the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene and the lentiviral packaging signal. Genomic DNA (2–3 days p.i.) was extracted using the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer's protocol. GAPDH primers and probes and their concentrations used for qPCR are previously described (26). Primers and probes for the lentiviral packaging signal and primers used for qRT-PCR were designed using the Primer Express software (Applied Biosystems). The primers for the lentiviral packaging signal and their concentrations used in qPCR were as follows: forward primer, 5′-GCACGGCAAGAGGCGA-3′, 0.3 μmol/l; reverse primer, 5′-CGCACCCATCTCTCTCCTTCTA-3′, 0.3 μmol/l; and the TaqMan probe, 5′-FAM-CGGCGACTGGTGAGTACGCCAAAAAT-3′ TAMRA, 0.15 μmol/l. Real-time PCR was done using a Perkin-Elmer 7700 sequence detector (Perkin-Elmer Applied Biosystems). Reactions were done in 50 μL using the Absolute QPCR ROX and dUTP mix (ABgene) and 0.01 units/μL AmpErase (Applied Biosystems). The qPCR conditions used were as follows: 50°C for 2 min, followed by 95°C for 15 min and 40 cycles of 95°C for 15 s and 60°C for 1 min. DNA mixtures containing linearized pSIN-MCS plasmid and fragments of DNA containing part of the GAPDH gene were used as standards. Copies per cell were determined by adjusting the number of lentiviral constructs present to the number of cells analyzed (using GAPDH). For each sample, two dilutions were run for both the lentiviral packaging signal and GAPDH.
qRT-PCR and RT-PCR. Total RNA from LEC was extracted using TRIzol reagent (Invitrogen) and subjected to DNase I digestion (Invitrogen). The RNA was subsequently purified using the RNEasy Mini kit (Qiagen). About 50 to 1,000 ng of total RNA was used for cDNA synthesis using the SuperScript II reverse transcriptase (Invitrogen). Real-time qRT-PCR was done for GAPDH and Ang2. GAPDH primers and their concentrations were as previously described (25), Ang2 primers and their concentrations used were as follows: forward primer 5′-GTTTGCTACTGGAAAAAGAGGAAAGAG-3′, 0.3 μmol/l; reverse primer 5′-AGGGCTGCTACGCTGCC-3′, 0.9 μmol/l. Reactions were done in 25 μL using the SYBR Green PCR mix (Applied Biosystems). The qRT-PCR conditions used were as follows: 50°C for 2 min, followed by 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 1 min. Relative Ang2 expression was quantitated using the comparative CT method using GAPDH as a housekeeping control gene. qRT-PCR for VEGF-A and VEGF-C was done using TaqMan Gene Expression assays (Applied Biosystems) according to the manufacturer's instructions. GAPDH was used as a housekeeping control gene. All samples were run in duplicate for each gene assayed.
RT-PCR was used to detect the expression of KSHV genes in lentiviral or KLEC and was done using the HotStarTaq DNA polymerase (Qiagen). PCR was done with 25 to 37 cycles, and sequences of the various RT-PCR primers used are available on request.
Immunohistochemistry. Immunohistochemistry was done as previously described (27). Paraffin sections (4 μm) were prepared from formalin-fixed, paraffin-embedded specimens of KS lesions (n = 8) obtained from archival biopsies in the Department of Histopathology, University College London Hospitals. Mouse monoclonal immunoglobulin G2B (IgG2B) anti-Ang2 antibody (R&D Systems) and biotinylated secondary antibodies were used. All immunohistochemistry reactions were followed by final washes and detection by application of the streptavidin-alkaline phosphatase complex using Vectastain ABC-AP Kit (Vector). Nuclei were counterstained with Mayer's hematoxylin, and slides were embedded in Hydromount (BDH). Consecutive sections were used for controls, which included the substitution of primary antibody with isotype-matched mouse IgG (BD Biosciences) used at appropriate concentrations or pre-adsorption of primary antibody with the recombinant human Ang2 (R&D Systems). These served as negative controls to indicate the specificity of the antibodies used. Visualization was done with the use of a Nikon Eclipse E800 microscope and photographs taken with a Nikon CoolPIX900 digital camera.
Ang2 ELISA and Western blotting. Media from uninfected and infected LEC were collected after 48 h in culture and stored at −80°C. Ang2 levels in media were measured using an Ang2 ELISA kit (R&D Systems) according to manufacturer's protocol.
For Western blotting, protein lysates were prepared from LEC or infected LEC using radioimmunoprecipitation assay buffer containing protease inhibitor and phosphatase inhibitor cocktails (Sigma). Typically, 20 μg of protein samples were loaded on 12% SDS-PAGE gels, electrophoresed, and transferred to Immobilon P membranes (Millipore). Membranes were probed with anti-Ang2 antibody (R&D Systems) or anti–phospho-ERK1/2 antibody (Cell Signalling Technology, Inc.) and reprobed with anti-GAPDH antibody (Advanced ImmunoChemical Inc.) or anti-ERK1/2 antibody (Upstate).
Immunofluorescence assay. LEC or KLEC grown on coverslips were fixed with 4% paraformaldehyde (Sigma) and subsequently permeabilized with 0.2% Triton X-100 (Sigma). Cells were blocked with 5% FBS in PBS and exposed to primary antibodies for 60 min at room temperature, after which cells were washed. The primary antibodies used are anti–LANA-1 (LN53), anti-K8.1A (Advanced Biotechnologies Inc.), anti-Ang2 (R&D Systems) and its isotype-matched mouse IgG (BD Biosciences). Cells were then exposed to the appropriate R-phycoerythrin–conjugated (Molecular Probes) or FITC-conjugated (DAKO) secondary antibodies for 45 min at room temperature. After washing, slides were mounted using VECTASHIELD mounting medium with 4′,6-diamidino-2-phenylindole (DAPI; Vector) or Prolong Gold antifade reagent (Molecular Probes). Visualization was done with the use of a Nikon Eclipse E800 microscope and photographs taken with a Nikon DS-fi1 camera. For the quantification of the percentage of LANA-1 or K8.1A-positive cells in a population, between 5 and 14 fields are counted, resulting in between 300 and 1,100 cells being analyzed.
Pharmacologic inhibition studies. For the VEGF receptor-1 (VEGFR-1) and VEGFR-2 inhibition studies, LEC were seeded onto 24-well plates (1.1 × 104 cells) and, 24 h later, exposed to conditioned media or conditioned media with either 0.25% DMSO or 0.25% DMSO with 50 μmol/L VEGFR-1 and VEGFR-2 inhibitor, 4-[(4′-chloro-2′-fluoro)phenylamino]-6,7-dimethoxyquinazoline (Calbiochem). After 18 h incubation, RNA from LEC was collected, and qRT-PCR was done. Before the addition of the conditioned media, LEC were preincubated for 5 h with either standard media or media containing either 0.25% DMSO or 0.25% DMSO with 50 μmol/L VEGFR-1 and VEGFR-2 inhibitor. Conditioned media from LEC and lentivirus or KLEC were used.
For the mitogen-activated protein kinase (MAPK) inhibition studies, LEC or infected LEC were seeded onto 24-well plates (1.1 × 104 cells) and, 72 h p.i., subjected to standard media or media with 0.1% DMSO, 0.25% DMSO, or different concentrations of the following inhibitors: mitogen-activated protein (MAP)/extracellular signal-regulated kinase (ERK) kinase (MEK) inhibitor, PD98059, and U0126 (Calbiochem), each with 0.1% DMSO or c-jun-NH2-kinase (JNK) inhibitor, SP600125 (Calbiochem) with 0.25% DMSO. After 8.5 h incubation, RNA was collected, and qRT-PCR was done.
GEM analysis. Processed GEM data from six LEC and six KLEC (3 or 4 days p.i. and at least 50% positive for GFP expression) hybridized to Affymetrix hg-u133+2 GeneChips were obtained and analyzed as previously described (23). To determine the effect of KSHV infection upon the LEC angiogenesis transcriptome, a list of 239 angiogenesis-related genes was compiled based on publicly available databases,34
SuperArray Biosciences, http://www.superarray.com.
Ang2 expression in Kaposi's sarcoma and KLEC. Immunohistochemistry was done to investigate whether Ang2 protein is present in KS and, therefore, relevant to the pathogenesis of KSHV (Fig. 1A). Ang2 immunostaining for Ang2 was observed in all KS lesions studied and is in accordance with previous studies demonstrating Ang2 mRNA expression in this tumor (9, 10). Ang2 staining was particularly strong in patch stage (early KS) lesions. We employed an Ang2 ELISA to show that KSHV infection of LEC increases Ang2 secretion (Fig. 1B), which correlates with a previous study (9). Western blot analysis showed that intracellular Ang2 protein was also up-regulated in KLEC compared with LEC (Fig. 1B). The two distinct bands seen in the Ang2 Western blot are most likely due to its alternative splice forms as reported previously (28). A GFP-expressing recombinant KSHV (24) was shown to up-regulate Ang2 transcription through qRT-PCR on GFP-expressing KLEC isolated by cell sorting (Fig. 1C).
KSHV library screen identified genes that up-regulate Ang2 expression. A lentiviral library of selected KSHV genes was constructed (Fig. 2A) using the pSIN-MCS lentiviral vector (Supplementary Fig. S1A) to investigate the mechanisms by which KSHV induces Ang2. A number of novel coding polymorphisms were observed when cloning and sequencing the viral ORFs (Fig. 2A). These polymorphisms did not affect protein function for the genes tested (data not shown). The ability of lentiviruses to infect LEC was shown via a GFP-encoding lentivirus (pCSGW; Supplementary Fig. S1A). Over 80% of LEC were GFP positive when ∼10 copies per cell were present in LEC (Supplementary Fig. S1B and C). This was typically achieved by adding 1 mL of unconcentrated pCSGW lentivirus preparation to 1 × 105 LEC.
A screen for the up-regulation of Ang2 was done using the selected KSHV lentiviral library (Fig. 2B). Ang2 levels in the culture media were measured 72 h after infection with the different lentiviral preparations. vGPCR and vIL6 increased Ang2 secretion, compared with empty vector (pSIN) or uninfected LEC. The expression of the different lentiviral constructs used in the screen was confirmed by RT-PCR, and the lentiviral copies per cell, as determined by qPCR, in each of the different infections are shown (Fig. 2B).
vGPCR and vIL6 up-regulate Ang2. Increasing amounts of vGPCR and vIL6 were delivered to LEC, and the cell lysates were subjected to Western blot analysis to confirm that vGPCR and vIL6 up-regulate Ang2 (Fig. 3A). Ang2 total cellular protein was up-regulated by both vGPCR and vIL6, with vIL6 strongly up-regulating Ang2 total protein in LEC even at low levels of expression. Similar copies per cell of vGPCR and vIL6 in LEC yielded comparable levels of Ang2 secretion (Fig. 3B). qRT-PCR analysis showed that Ang2 mRNA expression is up-regulated by both vGPCR and vIL6 and displays similar fold changes when compared with the Ang2-secreted protein (Fig. 3B). This indicates that the up-regulation of Ang2 secretion in LEC is at least partly attributed to an up-regulation of Ang2 mRNA by these genes. vGPCR and vIL6 seem to have a weak additive effect in their up-regulation of Ang2: when expressed together in a population of LEC, they up-regulated Ang2 total protein slightly more than either of the genes alone (Fig. 3C). Both vGPCR and vIL6 were expressed in KLEC (Fig. 3D), further suggesting that both contribute to the up-regulation of Ang2 upon KSHV infection (Fig. 1B and C).
Paracrine up-regulation of Ang2. To investigate further how lytic replication and vIL6 and vGPCR in particular are involved in the up-regulation of Ang2, immunofluorescence assays were done (Fig. 4A). Both LEC and KLEC expressed Ang2, with Ang2 immunoreactivity being higher in the latter. All cells in KLEC culture expressed similar levels of Ang2 despite only a small percentage of cells (∼1%) undergoing lytic replication as determined by K8.1A staining (Fig. 4A). This suggests that cells undergoing lytic replication may use paracrine mechanisms to contribute to the increased Ang2 expression in KLEC. To investigate whether KSHV up-regulates Ang2 through a paracrine mechanism, conditioned KLEC media was added to LEC. Ang2 mRNA and intracellular protein were up-regulated by KLEC-conditioned media, and Ang2 mRNA was up-regulated as soon as 8 h after the addition of the conditioned media (Fig. 4B). Similarly, vGPCR- and vIL6-conditioned media up-regulated Ang2, indicating that they are both likely to contribute to the ability of KSHV to up-regulate Ang2 through a paracrine mechanism (Fig. 4B). Interestingly, this mechanism does not involve VEGF-A or other molecules signaling through VEGFR-1 and VEGFR-2. Blocking VEGFR-1 and VEGFR-2 by chemical inhibition does not prevent Ang2 up-regulation by KSHV-infected or vIL6- and vGPCR-expressing LEC (Fig. 4C).
KSHV, vGPCR, and vIL6 up-regulate Ang2 by way of the MEK MAPK pathway. The MAPK pathway has been previously shown to be activated by both vIL6 (29) and vGPCR (30, 31) and has been implicated in Ang2 transcription in bovine EC (32). Western blot analysis showed that vGPCR, vIL6, and KSHV phosphorylate ERK in LEC (Fig. 5A). KSHV infection and expression of vGPCR, but not vIL6, result in a constitutive up-regulation of phosphorylated ERK compared with LEC or pSIN. vIL6 instead promotes a pulse increase in ERK phosphorylation as shown when the conditioned media of vIL6-infected LEC are added to serum-starved LEC (Fig. 5A). vIL6-conditioned media resulted in a stronger and prolonged phosphorylation of ERK compared with pSIN-conditioned media. The increase in ERK phosphorylation caused by pSIN is likely to be mainly due to the background phosphorylation of ERK by FBS in the conditioned media. The increase in ERK phosphorylation by vIL6 compared with pSIN was more pronounced when conditioned media with only 0.5% FBS was used (Supplementary Fig. S2).
To investigate whether vIL6-, vGPCR-, and KSHV-induced transcriptional up-regulation of Ang2 in LEC was dependent on the MAPK pathway, pharmacologic inhibition studies were done. In initial studies, vGPCR- and vIL6-expressing LEC were incubated with different concentrations of the MEK-specific inhibitor PD98059 or the JNK-specific inhibitor SP600125, after which RNA was harvested and analyzed by qRT-PCR (Fig. 5B). Expression of vIL6 and vGPCR in infected cells was confirmed by RT-PCR (data not shown). Ang2 mRNA in vGPCR-expressing LEC was significantly down-regulated by PD98059 to <25% of that present in the DMSO control. Ang2 mRNA levels in vIL6-expressing LEC were also decreased by PD98059. In contrast, the JNK inhibitor SP600125 did not inhibit Ang2 expression in either vGPCR- or vIL6-expressing LEC. To investigate further whether vGPCR and vIL6 up-regulate Ang2 through the MEK MAPK pathway, the pharmacologic inhibition studies were repeated using a different and more potent MEK-specific inhibitor, U0126, and were extended to include empty vector–infected cells and LEC infected with both vGPCR- and vIL6-encoding lentiviruses (Fig. 5C). U0126 significantly inhibited the up-regulation of Ang2 mRNA in vGPCR- and vIL6-expressing LEC and reduced the Ang2 mRNA level to that observed in empty vector–infected cells. U0126 was able to inhibit Ang2 transcription to a similar degree in LEC infected with both vGPCR and vIL6, which showed further that both genes up-regulate Ang2 through the MEK MAPK pathway. Finally, U0126 was also able to significantly inhibit the up-regulation of Ang2 in KLEC (Fig. 5C). This further supports that KSHV up-regulates Ang2 at least in part by the effect of vGPCR and vIL6 acting on the MEK MAPK pathway.
A proposed mechanism by which vGPCR and vIL6 contribute to the up-regulation of Ang2 is shown in Fig. 5D. In vGPCR-expressing cells, direct up-regulation of Ang2 by vGPCR is likely to take place through its constitutive activation of the MEK MAPK pathway (30, 33). vGPCR induces Ang2, in addition, indirectly by unknown secreted factor(s) (not VEGF-A). vIL6-induced Ang2 expression is likely to be via gp130—the specific receptor for this viral cytokine (34). However, the contribution of other secreted factors in supernatants from vIL6-expressing cells on Ang2 expression cannot be excluded.
KSHV infection regulates angiogenic factors important for Ang2 function in KS. We analyzed GEM data for 239 genes associated with angiogenesis to determine angiogenic factors affected by KSHV infection of LEC. Of 239 genes associated with angiogenesis, 79 (∼33%) were significantly affected upon KSHV infection of LEC (q < 0.001). Ang2 was the third most up-regulated gene of all angiogenic factors assessed. Supplementary Fig. S3 displays a gene expression heat map that corresponds to angiogenesis-related genes significantly affected upon KSHV infection of LEC. To focus on the genes that are important for Ang2 pro(lymph)angiogenic effects, a diagram of the regulation of such genes by KSHV and their corresponding heat map is shown (Fig. 6A and B). Along with Ang2, other factors related to Ang2 were significantly regulated by KSHV infection. Ang1, which can have an antagonistic effect to Ang2, was not significantly regulated and seemed to be expressed at low levels (Supplementary Table S1). Similarly, both Tie1 and Tie2 were expressed in LEC, but were not significantly affected by KSHV infection. Furthermore, some angiopoietin-like genes were significantly affected by KSHV infection. For example, angiopoietin-like 2 (Angptl2) was one of the most up-regulated genes after infection (Supplementary Fig. S3); however, its actual role and significance in (lymph)angiogenesis remains to be investigated. Importantly, VEGFRs and some of their ligands were significantly up-regulated by KSHV infection (Fig. 6A and B). However, VEGF-A, a factor shown previously to be important for Ang2 to promote angiogenesis, was not directly up-regulated by KSHV. Its receptors VEGFR-1 and VEGFR-2 were significantly up-regulated, thus permitting EC, like LEC, to be more responsive to VEGF-A and the related VEGF-B. VEGF-C and its receptor VEGFR-3 were also both up-regulated by KSHV infection. The VEGF-C/VEGFR-3 axis is essential for LEC, promoting lymphangiogenesis and is required, therefore, along with Ang2, to form a normal lymphatic system (14, 35). Notably, vGPCR and vIL6 up-regulate VEGF-C as well as VEGF-A expression in LEC (Fig. 6C).
Members of the angiopoietin family play a critical role in tumor angiogenesis, by way of their interaction with the Tie2 receptor (11–13). Increased Ang2 expression in tumors results in increased angiogenesis and tumor growth, whereas Ang2 inhibition has the opposite effect (20, 21). Ang1-mediated activation of Tie-2 acts as a regulator of vessel maturation and EC quiescence, whereas the antagonistic ligand Ang2 causes vessel destabilization and promotes EC to respond to other proangiogenic stimuli, such as VEGF (11, 14).
KS is a vascular tumor that consists primarily of sheets of proliferating spindle cells. KSHV is latently expressed in the majority of these tumor cells, but lytic viral replication is also present in a fraction of cells during all stages of tumor development (7, 36). The spindle cells are surrounded by an inflammatory infiltrate and abnormal leaky blood vessels (7, 8). Ang2 is up-regulated in the sera of individuals with KS, in the media of KSHV-infected LEC, and at the mRNA level in KS lesions (9, 10).
Here, we showed that Ang2 protein is highly expressed in early (patch stage) and late (nodular) KS lesions. It has been shown previously that Ang2 is not expressed in normal adult tissues, except at sites of vessel remodeling (11), and in certain human malignancies, such as breast cancer (19). We show areas of intense staining in patch stage (early) lesions, which could be sites of lytic KSHV replication (expression of vGPCR and vIL6). This correlates with observations that suggest the association of early KS with increased lytic viral replication (36).
Next, we investigated the mechanisms by which KSHV causes the up-regulation of Ang2 using LEC infected with individual KSHV-encoded latent and lytic genes. Lentiviruses were employed to construct the genetic screen because they infect primary cells with a relatively high efficiency and are able to infect both dividing and nondividing cells (37). The efficient infection of LEC was shown by the ability of unconcentrated viral preparations of GFP-encoding lentiviruses to cause GFP expression in more than 80% of LEC. The genes selected for the screen include most of the genes associated with KSHV latency: vIRF1, KapA, vFLIP, vcyclin, LANA-1, and K15/LAMP, and a selection of lytic genes, three of which were previously shown to promote angiogenesis (vIL6, vMIP1, and vGPCR; refs. 7, 38). The Ang2 lentiviral screen and our further studies revealed that both vIL6 and vGPCR up-regulate Ang2. The fold regulation of Ang2 was comparable to that seen in KLEC.
vIL6 is the viral homologue of human IL6 (hIL6) and is expressed in up to 5% of spindle cells in KS lesions (7, 39). Although hIL6 receptor consists of both the gp80 binding subunit and the signal transducing subunit gp130, vIL6 mediates its effects solely by gp130 (34). vIL6 activates both Janus-activated kinase/signal transducers and activators of transcription signaling and the MAPK pathway (29, 34). In vivo, vIL6 promotes tumor angiogenesis and increases VEGF expression (40). Interestingly, EC seem to only express gp130 and not the gp80 signaling subunit necessary for hIL6 signaling (41).
vGPCR is a constitutively active GPCR, which is related to the IL8 receptors CXCR1 and CXCR2, and like vIL6, it is a lytic gene (7, 33). vGPCR activates several signaling pathways, some of which seem to be cell specific, and include the MAPK, ERK, JNK, p38, and the antiapoptotic serine-threonine kinase AKT pathways (30, 31). vGPCR immortalizes EC, and vGPCR-expressing cells lead to the induction of vascularized tumors in vivo (42, 43). The ability of vGPCR to stimulate angiogenesis is at least partly attributed to its ability to up-regulate angiogenic factors such as VEGF and Groα (31). Ang2 up-regulation by the lytic viral proteins vGPCR and vIL6 further supports a role of lytic viral replication in KS pathogenesis and provides additional insight into the mechanism by which these genes promote angiogenesis (40, 43).
Up-regulation of Ang2 by KSHV, as well as vIL6 and vGPCR, was found to involve a paracrine mechanism. This reveals how the small percentage of cells undergoing lytic replication and expressing vGPCR and vIL6 (7) can act on nearby cells to contribute to the overall large increase in Ang2 expression seen in KLEC or KS (9). Therefore, a relatively large population of cells are up-regulating Ang2 despite the small percentage of lytic cells. vGPCR and vIL6 result in the expression of several secreted factors, such as VEGF-C and vIL6 itself, which may contribute to the paracrine up-regulation of Ang2.
The regulation of Ang2 is affected by a wide range of factors such as hypoxia, Ang1, and basic fibroblast growth factor (44). The MAPK pathway regulates Ang2 expression in EC (32). From our studies, we showed how vIL6, vGPCR, and KSHV activated the MEK MAPK pathway in LEC by increasing phosphorylated ERK and, from our pharmacologic studies, showed that inhibiting the MEK MAPK pathway inhibited their up-regulation of Ang2 transcription. This finding concurs with recent work showing that KSHV induces Ang2 expression via AP-1 and Ets1 (45). vGPCR caused a constitutive increase in phosphorylated ERK levels, whereas vIL6 caused a pulse up-regulation of phosphorylated ERK. This difference in the way the MAPK pathway is activated by vGPCR and vIL6 is probably due to vGPCR being a constitutively active receptor unlike the gp130 receptor of vIL6 (33, 34). It is interesting to note that other genes in the KSHV genome, such as vFLIP and LAMP (K15), are known to activate MAPK pathways (46, 47), yet did not up-regulate Ang2 in our study. This could be because these viral genes activate different pathways in primary LEC, compared with the 293 or Cos7 cells used in previous studies. Whether vFLIP or LAMP activates MAPK pathways in LEC requires further investigation.
Because Ang2 requires other factors for its pro(lymph)angiogenic activity (11, 14, 48), we used GEM analysis to investigate whether KSHV infection of LEC affected such factors. The observation that, unlike Ang2, Ang1 is expressed at relatively low levels and is unaffected by KSHV infection is consistent with the expression profile observed in the majority of other tumors where Ang2 is up-regulated (19). Tie2 was found to be expressed in LEC and not significantly affected by KSHV infection, which indicates how the relative increase of Ang2 expression compared with Ang1 could act to prevent the maturation and quiescence signaled by Tie2 (14). The GEM analysis also showed that VEGFRs and some of their ligands are up-regulated directly by KSHV. This could be important because in the absence of active VEGF-A, high levels of Ang2 can induce capillary regression and EC death (48). Two of the most highly up-regulated angiogenic genes by KSHV infection are members of the matrix metalloproteinase family (including matrix metalloproteinase 9), which degrade the extracellular matrix and release trapped growth factors such as VEGF-A (Supplementary Fig. S3; ref. 49). In addition, vIL6 and vGPCR up-regulated the VEGFR ligands VEGF-A and VEGF-C in LEC. The VEGF-C/VEGFR-3 axis is also essential for lymphangiogenesis (35) and was up-regulated by KSHV infection.
In summary, we show that Ang2 protein is expressed in KS lesions and is up-regulated in LEC after KSHV infection. We found that two KSHV lytic genes, vIL6 and vGPCR, up-regulate Ang2 and, similar to KSHV, cause a paracrine up-regulation of Ang2. KSHV as well as vIL6 and vGPCR up-regulate Ang2 by activating the MEK MAPK pathway. GEM analysis revealed that other molecules that are important for the ability of Ang2 to promote (lymph)angiogenesis are also regulated by KSHV, confirming the notion that Ang2 up-regulation acts in concert with other factors to allow KS development. We conclude that Ang2 up-regulation by KSHV-encoded genes contributes to its high levels in individuals with KS, and that molecular mechanisms regulating its expression might present a target for future anti-KS therapeutics.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
L.L. Nikitenko and D. Lagos contributed equally to this work.
Grant support: Cancer Research U.K., the Medical Research Council, and the Wellcome Trust. R.J. Vart was supported by the Biotechnology and Biological Sciences Research Council and an educational grant from Sanofi-Aventis.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Andrew Godfrey for producing the pSIN-MCS vector, Mali Weisz for helping with cloning into the pSIN-MCS vector, and Nadege Presneau and Marie-Helene Malcles for their technical expertise. We also thank Prof. Peter Isaacson for the KS biopsies.