Amplification of the Angiogenic Signal through the Activation of the TSC/mTOR/HIF Axis by the KSHV vGPCR in Kaposi's Sarcoma

Background Kaposi's sarcoma (KS) is a vascular neoplasm characterized by the dysregulated expression of angiogenic and inflammatory cytokines. The driving force of the KS lesion, the KSHV-infected spindle cell, secretes elevated levels of vascular endothelial growth factor (VEGF), essential for KS development. However, the origin of VEGF in this tumor remains unclear. Methodology/Principal Findings Here we report that the KSHV G protein-coupled receptor (vGPCR) upregulates VEGF in KS through an intricate paracrine mechanism. The cytokines secreted by the few vGPCR-expressing tumor cells activate in neighboring cells multiple pathways (including AKT, ERK, p38 and IKKβ) that, in turn, converge on TSC1/2, promoting mTOR activation, HIF upregulation, and VEGF secretion. Conditioned media from vGPCR-expressing cells lead to an mTOR-dependent increase in HIF-1α and HIF-2α protein levels and VEGF upregulation. In a mouse allograft model for KS, specific inhibition of the paracrine activation of mTOR in non-vGPCR-expressing cells was sufficient to inhibit HIF upregulation in these cells, and abolished the ability of the vGPCR-expressing cells to promote tumor formation in vivo. Similarly, pharmacologic inhibition of HIF in this model blocked VEGF secretion and also lead to tumor regression. Conclusions/Significance Our findings provide a compelling explanation for how the few tumor cells expressing vGPCR can contribute to the dramatic amplification of VEGF secretion in KS, and further provide a molecular mechanism for how cytokine dysregulation in KS fuels angiogenesis and tumor development. These data further suggest that activation of HIF by vGPCR may be a vulnerable target for the treatment of patients with KS.


Introduction
Kaposi's sarcoma (KS) is a multifocal vascular neoplasm invariably associated with infection with the KS-associated human herpesvirus (KSHV/HHV8), which is characterized by cytokine dysregulation [1]. The driving force of the KS lesion is the KSHVinfected spindle-shaped tumor (spindle) cell, thought to have a vascular endothelial or endothelial precursor origin. The promotion of the angiogenic phenotype in these lesions is supposed to be mediated by elevated levels of pro-inflammatory and proangiogenic secretions (cytokines, chemokines and growth factors) from the KS spindle cells [1]. Indeed, KS is often thought to result from reactive endothelial hyperproliferation induced by the chronic release of these molecules and has served as a model for tumor-and inflammation-induced angiogenesis [1,2].
Among the angiogenic factors elaborated by the KS spindle cells, VEGF is unique for its profound impact on KS pathogenesis.
VEGF is expressed at elevated levels by KSHV-infected endothelial cells in vitro, and by KS spindle cells in vivo; a strict requirement for VEGF has also been demonstrated for KS spindle cells grown in vitro [3,4,5,6]. These observations are suggestive of an autocrine mechanism for this angiogenic factor in the growth of this vascular tumor. Indeed, like VEGF, its cognate receptor, VEGFR2 (KDR), is also upregulated in AIDS-KS primary lesions as well as in AIDS-KS spindle cell cultures [4,7,8]. However, reasonable disagreement remains as to the precise molecular mechanism whereby KSHV promotes VEGF secretion.
Of interest, KSHV ORF74 encodes for a viral G protein-coupled receptor (vGPCR) with close homology to the mammalian CXCR1 and CXCR2 and ligand-independent (constitutive) activity [9,10]. When expressed upon endothelial-specific retroviral infection or as a transgene, vGPCR is sufficient to recapitulate KS-like lesions in mice [11,12,13]. Indeed, emerging data suggest that vGPCR may play a role in KS initiation, progression and maintenance [12,14,15]. vGPCR has proven to be a powerful oncogene and a potent angiogenic activator [1]. Expression of vGPCR in endothelial cells activates intracellular pathways that induce cell survival and transformation. In addition, vGPCR-expressing cells elaborate pro-angiogenic factors that are thought to promote the recruitment and transformation of neighboring endothelial cells [16]. Both VEGF and KDR have been previously reported to be upregulated by cells expressing vGPCR [7,8,17,18]. However, the limited expression of this viral oncogene in only a few tumor cells in both transgenic KS mouse models and human KS tissues raises the question as to the relative contribution of vGPCR to VEGF secretion, suggesting that the relationship between vGPCR and VEGF is not clearly established. Here we set out to further characterize the contribution of vGPCR to the upregulation of VEGF secretion in KS.

vGPCR activates VEGF expression through a paracrine mechanism dependent on mTOR
The KSHV-encoded vGPCR causes angioproliferative lesions remarkably similar to human KS, when expressed upon endothelial-specific retroviral infection in immunocompetent animals ( Fig. 1A-B) [12]. Interestingly, immunohistochemical staining of these vGPCR tumors using a specific vGPCR antibody reveals the presence of this viral protein in only a small percentage of tumor cells, similar to the expression pattern of vGPCR in human KS, suggestive of a paracrine contribution of vGPCR to Kaposi's sarcomagenesis (Fig. 1B) [12,13]. In this regard, vGPCR has been implicated in the induction of the expression of VEGF, a key angiogenic factor highly upregulated in KS [17,18]. However, staining of vGPCR tumors -and human KS -with a specific antibody against VEGF reveals a robust expression of this factor in most tumor cells (Fig. 1B), indicating that vGPCR may also upregulate VEGF in neighboring cells through an indirect mechanism. To explore this possibility, we used an inducible (Tet-on) expression system for vGPCR in immortalized human microdermal endothelial cells (HMEC1). Using this system, we confirmed that induction of vGPCR expression in endothelial cells leads to the potent upregulation of VEGF, as previously reported (Fig. 1C) [17,18]. Moreover, we observed that exposure of HMEC1 to media conditioned by these vGPCR-expressing cells (vGPCR CM) is similarly able to promote VEGF expression (Fig. 1C).
We therefore set out to determine the mechanism whereby vGPCR angiogenic factors can induce VEGF upregulation. In this regard, we have previously reported that vGPCR paracrine secretions activates TSC/mTOR, a signaling route that has been shown to regulate the expression of VEGF [19,20]. We thus treated HMEC1s with supernatants derived from vGPCRexpressing cells (vGPCR CM) or control cells (Control CM), in the absence or the presence of the mTOR inhibitor, Rapamycin. Figure 1D shows that exposure of endothelial cells to vGPCR secreted factors leads to the upregulation of the transcription and translation of VEGF in a Rapamycin-sensitive manner.
To further evaluate the contribution of the vGPCR-induced paracrine activation of mTOR to VEGF upregulation in vivo, we used an allograft model in which (SV-40) immortalized murine endothelial cells (SVECs) expressing vGPCR (EC-vGPCR) are mixed with SVECs co-expressing two non-tumorigenic KSHV latent genes, vCyclin and vFLIP (EC-vCYC/vFLIP), in a (1:10) ratio that approximates the proportion of expressing cells in human KS (Fig. 1E) [15]. Cells expressing vCYC and vFLIP do not show VEGF upregulation in vitro nor are they tumorigenic in vivo (data not shown; [15]). However, these cells (EC-vCYC/ vFLIP) are able to induce allografts in nude mice when co-injected with vGPCR-expressing cells (EC-vGPCR) (Fig. 1E). Immunohistochemical staining of these lesions also shows upregulation of VEGF in most tumor cells (Fig. 1E). Interestingly, when these mixed-cell allografts are treated with Rapamycin, which is able to block vGPCR tumorigenesis [19], a reduction in VEGF expression in treated tumors is observed. Collectively, these results suggest that the paracrine upregulation of VEGF by vGPCR requires the activation of the mTOR signaling cascade in vitro and in vivo.
We have previously shown that induction of AKT activity by vGPCR promotes mTOR activation in neighboring cells [19]. However, the contribution of other kinases upstream of TSC/ mTOR to vGPCR oncogenesis remains unclear. We thus treated HMEC1 with media conditioned by control or vGPCR-expressing cells and assessed the activation of TSC kinases by these supernatants. In addition to AKT, we found that ERK, p38 and IKKb were also activated by vGPCR paracrine secretions ( Fig. 2A). Activation of these kinases correlated with the phosphorylation of TSC2 in Thr 1462 , Ser 664 and Ser 1254 , and phosphorylation of TSC1 in Ser 511 , respectively, and the upregulation of mTOR activity, assessed by S6K phosphorylation ( Fig. 2A). We also evaluated the activation of these TSC1/2 kinases by individual GPCR angiogenic factors [14,28]. Figure S1 shows that all the cytokines tested were able to induce phosphorylation of TSC1 and/or TSC2. Interestingly, IL-1b, IL-10, TNFa and VEGF are each able to promote phosphorylation of TSC1/2 on at least three separate sites (Fig. S1). Collectively, these results suggest that the secreted factors elaborated by vGPCR-expressing cells act together to upregulate multiple intracellular signaling pathways that converge on TSC/ mTOR.
To determine the relative contribution of these TSC kinases to the paracrine mTOR activation by vGPCR, we used specific inhibitors of PI3K/AKT, MEK/ERK, p38 or IKKb. Surprisingly, we observed that either pharmacological inhibition of AKTmediated phosphorylation of TSC2 or IKKb-mediated phosphorylation of TSC1 lead to a complete inhibition of mTOR, measured by S6K phosphorylation (Fig. 2A). Conversely, pharmacological inhibition of either ERK or p38 leads to only partial inhibition of ERK-mediated or p38-mediated S6K phosphorylation, respectively. We then used siRNA to specifically knock-down expression of AKT, ERK1 and 2, p38, or IKKb. Knock-down of these kinases was only sufficient to partially inhibit mTOR (Fig. 2B). Collectively, these results suggest that there is a redundancy in the pathways leading to the phosphorylation of TSC1 and 2 and the activation of mTOR by the paracrine secretions of vGPCR-expressing cells.
We then stained murine vGPCR tumors and human KS tissues with specific antibodies against the phosphorylated (activated) forms of AKT, ERK, p38 or IKKb or against the corresponding TSC2/TSC1 phosphorylated form, induced by each kinase . VEGF upregulation in HMEC1 exposed for 2 h or 6 h to supernatants collected from vGPCR-expressing cells (vGPCR CM). (D) Supernatants from vGPCR-expressing cells (vGPCR CM) or control cells (Control CM) were used to treat HMEC1 in the presence or absence of Rapamycin (50 nM). RT-PCR and Western blot analysis were used to determine levels of VEGF mRNA and protein, respectively. (E) EC-vGPCR (10%) were mixed with EC-vCYC/vFLIP (90%), and injected into athymic nu/nu mice for allograft formation. Tumor weight curves, and immunohistochemical detection in tumor tissue of (AU5-tagged) vGPCR-expressing cells, phosphorylated ribosomal S6 protein or VEGF, upon treatment with (10 mg/kg) Rapamycin of vehicle (Control), are shown. doi:10.1371/journal.pone.0019103.g001 (Fig. 2C). In support of our in vitro observations, we found phosphorylation of these four kinases and the corresponding targeted aminoacids in TSC2 or TSC1, in both vGPCR tumors and human KS (Fig. 2C). These findings suggest a role for these kinases upstream of TSC/mTOR in vGPCR tumorigenesis and Kaposi's sarcomagenesis in vivo.
Paracrine activation of TSC/mTOR by vGPCR angiogenic factors results in the upregulation of HIF-1a/2a TSC/mTOR has been shown to regulate Hypoxia Inducible Factor (HIF), a family of transcription factors containing an inducible a subunit and a constitutive b subunit [20,29]. HIF promotes neovascularization and vascular remodeling by controlling the expression of key angiogenic proteins, including VEGF [30]. Since vGPCR angiogenic factors activate TSC/mTOR through a variety of intracellular routes, we next investigated whether this activation could lead to the upregulation of VEGF through a HIF-dependent mechanism. To this end, we treated HMEC1 with media conditioned by vGPCR-expressing cells or control cells, in the absence or presence of Rapamycin, and examined the levels of HIF-1a and HIF-2a mRNA and protein.

Inhibition of HIF blocks vGPCR tumorigenesis in vivo
Of interest, several drugs have been recently described that target HIF expression or activity; these drugs have demonstrated anti-angiogenic and anti-cancer effects in vivo [30]. The identification of HIF as a key factor in vGPCR angiogenic amplification prompted us to explore inhibition of HIF as a potential therapeutic approach for KS treatment. We therefore established tumor allografts by injecting mixed-cell populations of EC-vGPCR and EC-vCYC/vFLIP cells in athymic nu/nu mice (Fig. 6). Animals were then treated with either (2 mg/kg) Digoxin, a cardiac glycoside that has been shown to inhibit HIF-1a synthesis and block tumor formation, or vehicle (Control) (Fig. 6) [31]. Drug toxicity, as assessed by weight loss, was minimal in the treated group (reduction ,5%) during the treatment period (results not shown). Inhibition of tumor growth by the treatment with Digoxin was sustained for the duration of the experiment. At the end of the study, we observed that the average estimated weight of vehicle-treated tumors was 702 mg (a 4.8 fold increase) vs. an average estimated weight of 234 mg (a 1.7 fold increase) of Digoxin-treated tumors (Fig. 6). Immunohistochemical analysis of these lesions demonstrated a dramatic reduction in the levels of HIF as well as VEGF in the Digoxintreated animals compared to control mice (Fig. 6). Taken together, our data provide the basis for the early assessment of HIF inhibitors as an anti-KS therapy.

Discussion
KS is an angioproliferative tumor characterized by the presence of angiogenic and inflammatory mediators [1]. The observation that KS tumors tend to localize to sites of inflammation suggests that these lesions thrive in a cytokine-rich environment [1]. Indeed, KS spindle cells do not appear to be truly transformed; rather, the KS spindle cell elaborates a variety of cytokines, chemokines and growth factors that are essential for their growth and survival. Indeed, isolated KS spindle cells remain strictly dependent on cytokines and growth factors to proliferate in vitro and are not tumorigenic when tested in animal models. Among the numerous angiogenic mediators on which the KS spindle cell is dependent, VEGF has been shown to be essential for KS spindle cell survival in vitro and KS pathogenesis in vivo.
We previously reported a mechanism whereby the KSHV vGPCR promotes the upregulation of VEGF transcription in vGPCR-expressing cells [17]. We report here that vGPCR also upregulates VEGF through a complex indirect (paracrine) mechanism. Upregulation of VEGF in neighboring (non-vGPCR-expressing) tumor cells results in a dramatic amplification of the angiogenic signal promoted by vGPCR and helps provide an explanation for how this unusual viral oncogene can play a role in KS despite the observation that its expression is restricted to only a few tumor cells. vGPCR angiogenic amplification involves the secretion of angiogenic and inflammatory cytokines by vGPCR-expressing cells which then activate in neighboring cells  multiple signaling pathways that ultimately converge on TSC1 and 2, resulting in de-repression of mTOR and the promotion of HIF upregulation and VEGF transcription and secretion (Fig. 7).
TSC kinases activated by vGPCR paracrine secretions include AKT, ERK, p38 and IKKb, inducing phosphorylation of TSC2/ 1 at specific regulatory sites [22,23,24,25,26,27]. Although pharmacological inhibition of AKT-mediated phosphorylation of TSC2 or IKKb-mediated phosphorylation of TSC1 lead to a complete inhibition of mTOR activity, siRNA knock down expression of AKT, ERK1 and 2, p38, or IKKb was only sufficient to partially inhibit S6K phosphorylation. This suggests either that the PI3K/AKT inhibitor, LY294002, and the IKKb inhibitor, BAY 11-7082, have non-specific inhibitory effects, or that the RNAi knockdown of AKT and IKKb were less efficient than their pharmacological inhibition. Although we suspect the former to be true, regulation of mTOR through these pathways has proven to be quite complex and the answer may not prove to be straightforward. Nonetheless, our results collectively suggest that there is a redundancy in these pathways and in the phosphorylation sites for inhibiting TSC activity and further provide insight into the complexity of mTOR regulation by different exogenous stimuli.
A number of vGPCR factors promote the phosphorylation of TSC1/2 through several of these signaling pathways. Surprisingly, we demonstrate here that phosphorylation of TSC1 in Ser 511 by IKKb is activated by most of the cytokines tested (Fig. S1), suggesting that regulation of mTOR by IKKb may be quite promiscuous. Moreover, as IKKb activation leads to the activation of NFkB, this, in turn, may promote a positive feedback loop, further enhancing cytokine -and therefore VEGF -secretion in KS. Of note, paracrine secretions from vGPCR-expressing cells promote a gene expression profile with an NFkB signature in endothelial cells [32]. In light of our results here, this suggests that the NFkB signature may be mediated by the specific secreted cytokines that promote the activation of IKKb. Ultimately, further investigation into the relative contribution of each of these cytokines to the angiogenic phenotype in KS as well as in other tumors may be warranted.
Several additional KSHV genes have also been shown to upregulate cytokine levels. These viral proteins, including vFLIP, kaposin A, kaposin B, K1 and K15, undoubtedly contribute to the inflammatory milieu observed in KS [33,34,35,36,37,38]. As KSHV infection itself has been demonstrated to induce cytokine release [1], the relative contribution of each of the KSHV genes to cytokine dysregulation and the angiogenic phenotype in KS remains to be determined. Moreover, various inflammatory cytokines, including IL-1, TNF-a and interferon-c (IFN-c), are increased upon HIV infection, an important cofactor in KS development [39]. Collectively, these findings suggest that cytokine dysregulation and TSC/mTOR/HIF/VEGF activation may be a general mechanism linking KS co-factors, inflammation, and dysregulated angiogenesis in KS.
The mTOR signaling pathway is a key modulator of protein translation and has previously been identified as a positive regulator of HIF and HIF-dependent responses [20,30]. Oncogenes activating this pathway have been implicated in tumor-induced angiogenesis in other tumors. vGPCR promotion of the paracrine activation of mTOR may play a similar role in the regulation of HIF in KS. Indeed, upregulation of HIF activity has been observed upon KSHV infection of endothelial cells in culture and HIF stabilization has been previously reported in AIDS-KS lesions [40,41]. Of note, other KSHV genes (e.g. LANA-1 and IRF-3) have also been suggested to play a role in upregulating HIF activity [42,43,44]. Given the central role of HIF in VEGF regulation, and the importance of VEGF in KS, it is certainly reasonable to argue that KSHV may encode a redundancy of mechanisms to ensure that this critical endothelial cell growth factor is available in growing KS tumors.
Here, we show that pharmacological inhibition of HIF upregulation by vGPCR is sufficient to inhibit vGPCR oncogenesis. As the master regulator of the hypoxic vascular response, it should not be surprising that HIF plays a central role in Kaposi's sarcomagenesis. HIF drives transcriptional activation of hundreds of genes involved in vascular reactivity, angiogenesis, arteriogenesis, and the recruitment of endothelial precursor cells, all key steps toward the development of KS [45]. Indeed, it is tempting to speculate that recent publications describing KS regression in patients with iatrogenic KS following a switch in their immunosuppressive treatment to the mTOR inhibitor, Sirolimus, may -in part -be due to its effect on decreasing HIF activation [46,47]. Collectively, our data ultimately provide the basis for the early assessment of drugs inhibiting HIF in those patients with cutaneous and/or systemic KS.

Tumorigenesis assays
All procedures involving animals were approved by the Institutional Animal Care and Use Committee. Murine vGPCR tumors were obtained as described in ref 12. Briefly, TIE2-tva transgenic mice expressing in vascular endothelium the avian leukosis virus (ALV) receptor, TVA, were injected i.p. with ALVderived retrovirus encoding for KSHV vGPCR. Macroscopic vGPCR tumors developed in 4 months predominantly in ear, tail, paw and GI tract. For tumor allograft formation, cells expressing vGPCR (EC-vGPCR or EC-vGPCR/RR-mTOR) were mixed with cells expressing vCyclin and vFLIP (EC-vCYC/vFLIP or EC-vCYC/vFLIP/RR-mTOR), in a (1:10) ratio (10 5 :10 6 cells). These mixed cell populations were then injected into the right flank of 8wk-old athymic (nu/nu) nude female mice as previously described [15]. For these in vivo studies, Digoxin stock solution (Sigma-Aldrich) was dissolved in DMSO and Rapamycin (LC Laboratories) was dissolved as previously described [19]. Drug treatment was initiated when estimated tumor weight reached around 150-200 mg. For this procedure, tumor-bearing animals were randomly grouped (control, n = 5; drug-treated group, n = 5) and treated with Rapamycin (10 mg/kg) or Digoxin (2 mg/kg) or an equal volume of vehicle. For Rapamycin, treatment schedule was a single injection per animal given i.p. for 5 consecutive days [19]. For Digoxin, treatment schedule was a single injection per animal given i.p. for 10 consecutive days [31]. The animals were monitored twice weekly for tumor formation. The longest length (L) and shortest width (W) of the tumor were measured using a caliper at different time points throughout the experiment. Tumor volume was then converted into tumor weight using the formula LW 2 /2, as described previously [19]. Results of animal experiments were expressed as mean estimated tumor weight 6 SD. When appropriate, animals were euthanized, and tissue was fixed in 4% paraformaldehyde and embedded in paraffin for further analysis.

Human tissues
Four patients diagnosed with KS (multiple lesions) at University of Maryland Dental School or School of Medicine, between 1990 and 2010, were identified. Formalin-fixed paraffin-embedded tissue samples were obtained from the pathology archives for immunohistochemical studies. Representative hematoxylin and eosin sections of each tumor were reviewed and the diagnosis was confirmed by immunohistochemistry using a specific antibody recognizing KSHV Latency-associated Nuclear Antigen-1 (LANA-1). The Institutional Review Board (IRB) protocol was exempt.