Kaposi’s sarcoma-associated herpesvirus infection of blood endothelial cells induces lymphatic differentiation

Abstract

Kaposi’s sarcoma-associated herpesvirus (KSHV) is necessary for KS, a highly vascularized tumor predominated by endothelial-derived spindle cells that express markers of lymphatic endothelium. Following KSHV infection of TIME cells, an immortalized human dermal microvascular endothelial cell (DMVEC) line, expression of many genes specific to lymphatic endothelium, including VEGFR3, podoplanin, LYVE-1, and Prox-1, is significantly increased. Increases in VEGFR3 and podoplanin protein are also demonstrated following latent infection. Examination of cytokine secretion showed that KSHV infection significantly induces hIL-6 while strongly inhibiting secretion of IL-8, a gene product that is decreased by differentiation of blood to lymphatic endothelial cells. These studies support the hypotheses that latent KSHV infection of blood endothelial cells drives their differentiation to lymphatic endothelial cells.

Introduction

Kaposi’s sarcoma (KS), the most frequent malignancy in AIDS patients, is currently the most commonly reported tumor in parts of Africa (Wabinga et al., 1993). KS tumors are highly vascularized and exhibit extensive neoangiogenesis. KS tumors also contain increased levels of several pro-inflammatory cytokines, including hIL-6 (Oxholm et al., 1989). The most prominent cell in the KS tumor is the spindle cell, a cell of endothelial origins, though there are other cell types present in the tumor including infiltrating lymphocytes. KS spindle cells express elevated levels of vascular endothelial growth factor receptor 3 (VEGFR3) and podoplanin, two specific markers of the lymphatic endothelium (Jussila et al., 1998; Skobe et al., 1999), which led to the hypothesis that KS spindle cells are lymphatic in origin (Jussila et al., 1998; Skobe et al., 1999; Weninger et al., 1999). The lymphatic endothelium controls the fluid and lymphocyte uptake into lymph nodes and is architecturally distinct from the blood vascular endothelium. Lymphangiogenesis is a separate process from blood or hemangiogenesis, utilizing different VEGF family members, primarily VEGF-C and D and their receptor VEGFR3 (Makinen et al., 2001).

Kaposi’s sarcoma-associated herpesvirus (KSHV also known as human herpesvirus 8 or HHV-8) is present in spindle cells of all KS tumors and is necessary for KS progression (Aluigi et al., 1996). KSHV is also associated with two B-cell lymphoproliferative diseases, primary effusion lymphoma (PEL) (Cesarman et al., 1995) and AIDS-associated plasmablastic multicentric Castleman’s disease (MCD) (Soulier et al., 1995). KSHV is a gamma herpesvirus, further subclassified as a rhadinovirus. At least 80 genes are encoded on its 165-kilobase genome (Neipel et al., 1997; Russo et al., 1996). Like all herpesviruses, KSHV has both a lytic and a latent phase. During latency, a small subset of the coding capacity of KSHV is expressed. In KS tumors and primary effusion lymphomas, KSHV is found predominantly in the latent state, although there is always a low background of lytic replication (Staskus et al., 1997).

TIME cells are human dermal microvascular endothelial cells (DMVECs) immortalized with h-tert, the active subunit of telomerase, and are efficiently infected with KSHV. By 24 h postinfection, nearly all of the TIME cells support latent infection while around 1–3% undergo lytic replication, similar to the percentages in KS tumors and PELs (Lagunoff et al., 2002). The studies presented here demonstrate that KSHV infection of TIME cells induces many changes that are consistent with differentiation from blood to lymphatic endothelial cells (BEC to LEC). This supports the hypothesis that KSHV infection of BECs in vivo causes them to differentiate into LECs as they transform into spindle cells.

Results

Markers of lymphatic endothelial cells are induced by KSHV infection

To examine major markers of lymphatic endothelial cells, we used quantitative real-time RT-PCR to examine expression of VEGFR3, podoplanin, and LYVE-1 in TIME cells infected with KSHV or with KSHV that was UV irradiated. UV-irradiated virus can bind and enter cells but there is no viral gene expression. Infection rates were determined by seeding a subset of infected cells onto slides at the time of harvest and staining with antibodies to a latent marker (LANA) and a lytic marker (ORF 59). Only experiments where greater than 90% of the cells expressed latent markers and less than 5% of the cells expressed lytic markers were analyzed. No staining was seen in uninfected cells or cells infected with UV-irradiated virus. Expression of VEGFR3, podoplanin, and LYVE-1 is increased approximately 10-, 13-, and 7-fold, respectively, in KSHV-infected cells (Fig. 1A). Prox-1 is required for the developmental expression of VEGFR3 and podoplanin as BECs differentiate into LECs in vivo (Wigle and Oliver, 1999), thus making the Prox-1 pathway a likely cellular target of KSHV. We could not amplify an RT-PCR product using primers specific for Prox-1 when using mRNA from uninfected or UV-irradiated KSHV-infected TIME cells, but we could amplify product from mRNA isolated from KSHV-infected TIME cells (Fig. 1B). The 2.2-kb product was subcloned, sequenced, and found to be full-length Prox-1 mRNA (data not shown). This indicates that uninfected TIME cells express little, if any, Prox-1, and the gene is upregulated by KSHV infection. Interestingly, we could easily amplify Prox-1 using mRNA from uninfected primary DMVECs (data not shown). It was previously shown that commercially available DMVECs are a mixture of blood and lymphatic endothelial cells (Makinen et al., 2001). Thus, it is likely that the contaminating LECs express the Prox-1 detected. The complete lack of detectable Prox-1 in TIME cells provides evidence that TIME cells are purely BEC in origin.

Fig. 1.

KSHV infection of TIME cells induces specific markers of lymphatic endothelial cells. (A) Depicted is the fold change of LEC-specific transcripts in KSHV-infected TIME cells compared to UV-irradiated virus-treated cells at 48 h postinfection as determined by real-time RT-PCR (N = 3–5). All values were normalized to GAPDH expression and error bars represent the standard error of the mean. (B) Shown is a representational gel of Prox-1 and GAPDH mRNA amplification by semi-quantitative RT-PCR from mock- and KSHV-infected TIME cells. (C) Western blot analysis of VEGFR3 podoplanin in mock- and KSHV-infected TIME cells. Cox1 is included as a loading control.

We also analyzed the protein levels of lymphatic endothelial-specific markers. Cell lysates from uninfected and KSHV-infected TIME cells were analyzed by Western blot using antibodies to VEGFR3 and podoplanin. Both were initially below the threshold of detection by Western blot, but easily detected following infection (Fig. 1C). We also examined expression of podoplanin in cells using immunofluorescence. There was only light background staining with an anti-podoplanin antibody in uninfected cells while there is strong cytoplasmic membrane staining in KSHV-infected TIME cells (Fig. 2). Importantly, around 90% of these cells express LANA and less than 1% express markers of lytic replication, indicating expression of lymphatic markers is due to latent gene expression. Also of note, the high level of podoplanin staining is only seen in LANA-positive cells in the infected cell panel (Fig. 2). Taken together, these data indicate that latent KSHV infection of BECs leads to differentiation to LECs.

Fig. 2.

Podoplanin expression is increased in latently infected TIME cells. TIME cells that are >90% latently infected and <1% lytically infected with KSHV and matched mock-infected cells were fixed 72 h after infection and co-stained with anti-podoplanin antibody (green), anti-LANA (red), and DAPI (blue) to stain the nuclei. Only those infected cells that express LANA exhibit cytoplasmic membrane staining for podoplanin.

Global changes in mRNA expression after KSHV infection

To identify global changes in gene expression caused by KSHV infection of endothelial cells, we used cDNA microarrays to examine transcripts from uninfected versus KSHV-infected TIME cells during a 4-day time course of infection and also UV-irradiated virus versus KSHV-infected TIME cells to control for contaminants in the inoculum that might effect host cell gene expression. Analysis was done on glass slides spotted with approximately 15000 or 13000 features in duplicate. As above, at the time of harvest, subsets of cells were seeded onto slides and stained with antibodies to LANA and ORF59. The cells infected with UV-irradiated virus did not express LANA or ORF59, while in the wild-type-infected cells, greater than 90% of the cells stained for LANA and less than 5% of the cells expressed ORF59. A total of eight true biological replicates were done in duplicate at three different time points, 24, 48, and 96 h after infection with another set of duplicates performed at 48 h with UV-irradiated virus-treated TIME cells as the control. The effects of changes in gene expression in cells undergoing lytic replication are minimal because changes in expression in less than 5% of the cells will not significantly impact the overall gene expression patterns. In addition, it was recently demonstrated that there is host gene shut off in lytically infected cells (Glaunsinger and Ganem, 2004).

To analyze genes altered by KSHV infection, we chose to only examine genes that were altered in all eight experiments, indicating that the alterations were sustained over the course of 4 days and, because we used UV-irradiated virus in two experiments, alterations also required viral gene expression. Cellular genes (147), about 1% of the total genes on the array, are altered at all time points more than 1.8-fold in intensity with a P value of less than 0.001. Expression of 86 genes is significantly increased at all time points (Table 1) while 61 genes are significantly repressed by infection with KSHV (Table 2). More than 10% of the genes that were identified as altered by KSHV infection are similarly differentially expressed in BECs versus LECs or in BECs differentiated into LECs by forced expression of Prox-1 (Hirakawa et al., 2003; Hong et al., 2002; Petrova et al., 2002). These changes are indicated in bold in Tables 1 and 2. We have confirmed many of these results by real-time RT-PCR looking at transcripts altered by KSHV infection as well as some that were unchanged on the arrays. Over 30 transcripts have been analyzed by real-time RT-PCR and to date all the genes tested are consistent with the results from the microarrays (see Figs. 1A, 2, 3A and data not shown). Expression of podoplanin is significantly increased on the array (listed as T1A-2), confirming results from Fig. 1. VEGFR3 is not spotted on the array.

Table 1.

Genes upregulated by KSHV infection of TIME cells

Sequence name	Sequence description	Accession no.	Fold change
SOCS3	suppressor of cytokine signaling 3	T72915	18.01
FLJ13220	hypothetical protein FLJ13220	AA083577	7.37
CEACAM1	carcinoembryonic antigen-related cell adhesion molecule 1 (biliary glycoprotein)	AA406571	6.02
IL6	interleukin 6 (interferon, beta 2)	W31016	5.82
LIPG	lipase, endothelial	T49529	5.22
GBP5	guanylate binding protein 5	AA487252	4.49
SLC21A2	solute carrier family 21 (prostaglandin transporter), member 2	AA037014	4.33
CALCRL	calcitonin receptor-like	N45301	3.96
T1A-2	lung type-I cell membrane-associated glycoprotein (podoplanin)	AA046430	3.79
PMP22	peripheral myelin protein 22	R26732	3.47
DEPP	decidual protein induced by progesterone	N55269	3.34
SLC2A14	solute carrier family 2 (facilitated glucose transporter), member 14	T97782	3.27
TMEM2	transmembrane protein 2	H75703	3.15
T1A-2	lung type-I cell membrane-associated glycoprotein (podoplanin)	AA149827	3.06
SSB	Sjogren syndrome antigen B (autoantigen La)	H29484	2.96
STOM	stomatin	R62817	2.85
FHIT	Homo sapiens cDNA FLJ46040 fis, clone SPLEN2036475	AA256123	2.79
CA2	carbonic anhydrase II	H23187	2.76
DEPP	decidual protein induced by progesterone	R33362	2.74
LTBP4	latent transforming growth factor beta binding protein 4	R87406	2.69
PRSS2	protease, serine, 2 (trypsin 2)	AA284528	2.68
EST_293336			2.58
ADAMTS1	a disintegrin-like and metalloprotease (reprolysin type) with thrombospondin type 1 motif, 1	R76276	2.56
DAF	decay accelerating factor for complement (CD55, Cromer blood group system)	R09561	2.56
ROM1	retinal outer segment membrane protein 1	H84113	2.45
FLJ11795	hypothetical protein FLJ11795	AA459693	2.43
HIF1A	hypoxia-inducible factor 1, alpha subunit (basic helix-loop-helix transcription factor)	AA598526	2.38
IFIT2	interferon-induced protein with tetratricopeptide repeats 2	N63988	2.35
ATF3	activating transcription factor 3	H21041	2.33
KIAA1450	KIAA1450 protein	W31540	2.31
PBEF	pre-B-cell colony-enhancing factor		2.3
KIAA0963	KIAA0963 protein	AA476273	2.26
CD47	CD47 antigen (Rh-related antigen, integrin-associated signal transducer)	R92801	2.25
PALM2	paralemmin 2	AA488418	2.23
IL6ST	interleukin 6 signal transducer (gp130, oncostatin M receptor)	T60500	2.2
SOD2	superoxide dismutase 2, mitochondrial	AA487750	2.2
FLT1	fms-related tyrosine kinase 1 (vascular endothelial growth factor/vascular permeability factor receptor)	AA149508	2.18
ARRB1	Homo sapiens mRNA; cDNA DKFZp762M127 (from clone DKFZp762M127)	H20911	2.17
GARP	glycoprotein A repetitions predominant	AA122287	2.13
E2-EPF	ubiquitin carrier protein	AA464019	2.12
NRCAM	neuronal cell adhesion molecule	R25521	2.11
LGALS9	lectin, galactoside-binding, soluble, 9 (galectin 9)	AA434102	2.08
C6orf4	chromosome 6 open reading frame 4	R78558	2.08
FLJ31265	hypothetical protein FLJ31265	H80258	2.07
SERPINB1	serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 1	R54664	2.07
UBE2L6	ubiquitin-conjugating enzyme E2L 6	AA292074	2.06
CTSL	cathepsin L	W73874	2.06
RPL7A	ribosomal protein L7a	H23421	2.04
NP	nucleoside phosphorylase	AA430382	2.03
KIAA0960	KIAA0960 protein	N92895	2.02
KIAA1554	KIAA1554 protein	H17860	2.02
KIAA1450	KIAA1450 protein	N52875	2
IL1RL1	interleukin 1 receptor-like 1	AA128153	2
EST_139250			1.99
P2RY5	purinergic receptor P2Y, G-protein coupled, 5	N90783	1.98
PROS1	protein S (alpha)	T74191	1.97
PPAP2B	phosphatidic acid phosphatase type 2B	T72119	1.96
FLJ22269	hypothetical protein FLJ22269	AA463986	1.95
MMRN	multimerin	AA423867	1.95
SLC38A2	solute carrier family 38, member 2	R27255	1.94
ARHJ	ras homolog gene family, member J	W92399	1.93
LAPTM4B	lysosomal associated protein transmembrane 4 beta	AA600214	1.93
PRO1073	PRO1073 protein	AA065090	1.92
NPC1	Niemann-Pick disease, type C1	AA634267	1.92
SP110	SP110 nuclear body protein	R54613	1.91
EST_357298	Homo sapiens mRNA; cDNA DKFZp762M127 (from clone DKFZp762M127)	W93688	1.89
CD53	CD53 antigen	AA132090	1.88
MGP	matrix Gla protein	AA155913	1.87
MTSG1	mitochondrial tumor suppressor gene 1	AA621202	1.87
M160	scavenger receptor cysteine-rich type 1 protein M160	N69291	1.87
MG61	Homo sapiens cDNA clone IMAGE:5092935, partial cds	AA486843	1.85
EPAS1	endothelial PAS domain protein 1	AA680300	1.85
RALA	v-ral simian leukemia viral oncogene homolog A (ras related)	H97948	1.85
WARS	tryptophanyl-tRNA synthetase	AA664040	1.84
NNMT	nicotinamide N-methyltransferase	T72089	1.84
ETS2	v-ets erythroblastosis virus E26 oncogene homolog 2 (avian)	T46844	1.84
HS3ST1	heparan sulfate (glucosamine) 3-O-sulfotransferase 1	T55714	1.84
KIAA0564	KIAA0564 protein	R24860	1.84
PTGS2	prostaglandin-endoperoxide synthase 2 (prostaglandin G/H synthase and cyclooxygenase)	AA644211	1.84
KIAA0963	KIAA0963 protein	AA485433	1.84
FLJ25084	hypothetical protein FLJ25084	R62633	1.83
GDAP1	ganglioside-induced differentiation-associated protein 1	H15302	1.83
RANBP1	RAN binding protein 1		1.83
SPRY4	Homo sapiens cDNA FLJ26120 fis, clone SYN00419	R70506	1.82
CXCL16	chemokine (C-X-C motif) ligand 16	AA416552	1.82
POPDC3	popeye domain containing 3	N20482	1.82

Bold denotes similar regulation in lymphatic endothelial cells.

Table 2.

Genes downregulated by KSHV infection of TIME cells

Sequence name	Sequence description	Accession no.	Fold change
EST_261253	Homo sapiens cDNA FLJ43100 fis, clone CTONG2003100	H98248	−1.82
EST_23588	Homo sapiens, Similar to hypothetical protein LOC216438, clone IMAGE:4444559, mRNA, partial cds	R38369	−1.83
EST_322537	Homo sapiens MSTP157 (MST157) mRNA, complete cds	W15263	−1.83
TPM1	tropomyosin 1 (alpha)	W58009	−1.83
VASP	Homo sapiens transcribed sequence with strong similarity to protein pir:S51797 (H. sapiens) S51797 vasodilator-stimulated phosphoprotein—human	AA410429	−1.84
CDC6	CDC6 cell division cycle 6 homolog (S. cerevisiae)	H59203	−1.85
BIRC5	baculoviral IAP repeat-containing 5 (survivin)	AA460685	−1.86
ITGA10	integrin, alpha 10	H43656	−1.87
MN1	meningioma (disrupted in balanced translocation) 1	R59212	−1.87
FOSL1	Homo sapiens mRNA; cDNA DKFZp666J045 (from clone DKFZp666J045)	T82817	−1.88
CTCF	CCCTC-binding factor (zinc finger protein)	H89996	−1.88
UBE2C	ubiquitin-conjugating enzyme E2C	AA430504	−1.89
AKR1C1	aldo-keto reductase family 1, member C1 (dihydrodiol dehydrogenase 1; 20-alpha (3-alpha)-hydroxysteroid dehydrogenase)	R93124	−1.89
MYBL2	v-myb myeloblastosis viral oncogene homolog (avian)-like 2	AA456878	−1.9
DOC1	downregulated in ovarian cancer 1	W69790	−1.9
EST_782847	Homo sapiens transcribed sequences	AA448283	−1.9
EST_415584	Homo sapiens transcribed sequence with moderate similarity to protein ref:NP_060265.1 (H. sapiens) hypothetical protein FLJ20378 [Homo sapiens]	W78816	−1.91
CARP	cardiac ankyrin repeat protein	AA486364	−1.92
EST_501430	Homo sapiens cDNA FLJ31353 fis, clone MESAN2000264.	AA115248	−1.92
PVRL3	poliovirus receptor-related 3	R62780	−1.92
SEC15L1	SEC15-like 1 (S. cerevisiae)	AA187002	−1.93
UAP1	UDP-N-acteylglucosamine pyrophosphorylase 1	N68465	−1.93
CXCL1	chemokine (C-X-C motif) ligand 1 (melanoma growth stimulating activity, alpha)	W42723	−1.95
COL1A2	collagen, type I, alpha 2	W93067	−1.96
IL8	interleukin 8	AA102526	−1.97
PTTG1	pituitary tumor-transforming 1	AA430032	−1.97
ALCAM	activated leukocyte cell adhesion molecule	R39862	−1.98
EGR3	early growth response 3	R39111	−1.98
FJX1	four jointed box 1 (Drosophila)	H72460	−2
HMGA1	high mobility group AT-hook 1	AA448261	−2.02
KRT7	keratin 7	AA489569	−2.03
HRB2	HIV-1 rev binding protein 2	W52272	−2.05
EST_201562	Homo sapiens transcribed sequences	R97050	−2.05
EST_171753	Homo sapiens surfactant associated protein F mRNA, partial sequence, mRNA sequence	H18335	−2.06
RGS4	regulator of G-protein signaling 4	AA007419	−2.07
EPLIN	epithelial protein lost in neoplasm beta	AA491501	−2.09
MICAL2	flavoprotein oxidoreductase MICAL2	AA412582	−2.1
EST_130826	Homo sapiens transcribed sequences	R22189	−2.11
FN14	IMAGE EST: 135791; Unigene: Hs.10086; type I transmembrane protein Fn14		−2.12
CDCA1	cell division cycle associated 1	AA421171	−2.19
EST_140301	Homo sapiens cDNA FLJ11041 fis, clone PLACE1004405.	R66924	−2.19
ESDN	endothelial and smooth muscle cell-derived neuropilin-like protein	AA431438	−2.24
IRS1	insulin receptor substrate 1	AA460841	−2.27
PDCD1L1	programmed cell death 1 ligand 1	R78465	−2.29
RRM2	ribonucleotide reductase M2 polypeptide	AA187351	−2.29
EST_140966	Homo sapiens LOC158525 (LOC158525), mRNA	R66502	−2.31
HRB2	HIV-1 rev binding protein 2	AA251800	−2.32
TRB@	Homo sapiens TCR BV3 mRNA for T cell receptor beta chain (CDR3 region), partial cds, isolate:HTLV-1 myopathy case3, clone:Tax tetramer-5.	N91921	−2.34
DJ667H12.2	hypothetical protein DJ667H12.2	AA608531	−2.37
CCNA2	cyclin A2	AA608568	−2.37
ESDN	endothelial and smooth muscle cell-derived RG_TX_tbl2neuropilin-like protein	N21309	−2.39
ESDN	endothelial and smooth muscle cell-derived neuropilin-like protein	H99543	−2.44
COL1A2	collagen, type I, alpha 2	AA490172	−2.56
EST_80186	Homo sapiens T cell receptor beta chain BV20S1 BJ1-5 BC1 mRNA, complete cds	T64380	−2.84
SEMA3C	sema domain, immunoglobulin domain (Ig), short basic domain, secreted, (semaphorin) 3C	AA042990	−3.12
LRRC17	leucine rich repeat containing 17	AA423870	−3.19
XRCC1	X-ray repair complementing defective	AA425139	−3.25
CDKN2B	repair in Chinese hamster cells 1 cyclin-dependent kinase inhibitor 2B (p15, inhibits CDK4)	N34320	−3.45

Bold denotes similar regulation in lymphatic endothelial cells.

Fig. 3.

KSHV infection of TIME cells alters transcripts involved in angiogenesis. Depicted is the fold change of angiogenesis-related transcripts in KSHV-infected TIME cells compared to UV-irradiated virus treated cells at 48 h postinfection as determined by real-time RT-PCR (N = 3–5). All values were normalized to GAPDH expression and error bars represent the standard error of the mean.

Angiogenic genes

KSHV infection alters the expression of several genes involved in angiogenesis. Included in the significantly altered genes is FLT1 (VEGFR1), which is increased about 2-fold on average (Table 1). Interestingly, FLT1 was previously shown to be decreased in LECs versus BECs (Hirakawa et al., 2003), indicating that KSHV is also altering gene regulation unrelated to lymphatic differentiation. A set of experiments was done to further quantify the alteration in expression of genes involved in angiogenesis and to probe transcripts that were not present on the array. Quantitative real-time RT-PCR was done using commercially available TaqMan probes for each transcript. As before, replicates (N = 3–5) were done on different days with different viral stocks. As the microarray experiments had shown, expression of VEGFR1 (FLT1) in TIME cells is significantly increased, approximately 5-fold, following KSHV infection, while VEGFR2 (KDR) is relatively unchanged (Fig. 3). None of the VEGF family members were greatly increased in expression. In fact, most (VEGF-A, VEGF-B, and placental growth factor (PGF)) appeared to be slightly decreased, while VEGF-C was slightly and consistently increased (Fig. 3). Interestingly, expression of VEGF-B and PGF is decreased in LECs versus BECs (Hirakawa et al., 2003; Petrova et al., 2002), again consistent with KSHV-induced differentiation. Cultured LECs, but not BECs, respond to VEGF-C, both as a mitogen and chemoattractant (Makinen et al., 2001), and VEGF-C is present in KS lesions (Skobe et al., 1999). VEGF-C was previously shown to be decreased significantly in LECs versus BECs. However, it was not repressed by forced Prox-1 expression in BECs (Petrova et al., 2002), only in mixed culture primary DMVECs (Hong et al., 2002). Thus, KSHV infection of BECs appears to drive the differentiation to LECs, while the increase in VEGFR3 expression and the concomitant induction of VEGF-C raise the possibility of an autocrine loop for cell activation.

We also found that carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) was increased about 18-fold (Fig. 3), which was consistently one of the most induced genes upon BEC to LEC differentiation (Hirakawa et al., 2003; Petrova et al., 2002).

Cytokines and KSHV infection

Because many cytokines are expressed in KS tumors and cytokine dysregulation has been proposed to play an important role in KS genesis, the alteration of cytokines was further analyzed. One of the most highly induced genes is hIL-6. hIL-6 is a cytokine involved in B-cell activation but may also play many other roles including a role in tumor formation. Importantly, hIL-6 is present at high levels in the KS tumor (Oxholm et al., 1989). The expression of hIL-6 after infection was further analyzed by real-time RT-PCR, where its transcript was increased approximately 44-fold on average after infection (Fig. 4A). Inhibition of cytokines may play a role in KS formation by immune evasion or more directly through changes in the activation of different cell types. Expression of IL-8, a chemokine that is capable of attracting neutrophils and T-cells (Hoffmann et al., 2002), is significantly inhibited at all three time points after KSHV infection. As described above, the inhibition of IL-8 is also consistent with the differentiation of TIME cells from blood to lymphatic endothelium. The expression of IL-8 was further analyzed by real-time RT-PCR, where a decrease of approximately 8-fold was seen at 48 h postinfection with KSHV (Fig. 4A).

Fig. 4.

KSHV infection of endothelial cells alters cytokine production and secretion. (A) Depicted is the fold change of IL6 and IL8 transcripts in KSHV-infected TIME cells compared to UV-irradiated virus treated cells at 48 h postinfection as determined by real-time RT-PCR (N = 3). All values were normalized to GAPDH expression and error bars represent the standard error of the mean. (B) A cytokine antibody array was used to monitor secreted cytokines from KSHV-infected and mock-infected primary human DMVECs by probing the antibody-spotted membrane with diluted supernatants from 48 to 96 h postinfection.

To analyze the protein levels of IL-6, IL-8, and a large number of other cytokines, a cytokine antibody array was used to monitor secreted proteins, as described in Materials and methods. Media from uninfected and infected cells were used to probe a membrane spotted with antibodies to 79 different cytokines and the relative levels of cytokines were determined. A representative experiment is shown from uninfected and KSHV-infected primary DMVECs at both 48 and 96 h postinfection (Fig. 4B). Similar results were obtained with TIME cells at 48 and 96 h postinfection (data not shown). Matching the mRNA expression data, hIL-6 protein levels in the media was significantly increased while IL-8 protein levels in the media were strongly decreased by KSHV infection. The only other cytokine that was altered was monocyte chemoattractant protein-1 (MCP-1), which was decreased at both 48 and 96 h postinfection. MCP-1 expression is also decreased in LECs versus BECs (Hirakawa et al., 2003; Petrova et al., 2002), again consistent with KSHV-infection reprogramming BECs to behave like LECs. These data also demonstrate that the cytokine changes are not unique to TIME cells but are also seen after infection of primary endothelial cells.

Discussion

In vivo, KS spindle cells express markers of the lymphatic endothelium and it was proposed that KS was lymphatic endothelium in origin (Weninger et al., 1999). The data presented here indicate that, rather than infecting lymphatic endothelial cells, KSHV could infect blood or circulating endothelial precursor cells and drive them to differentiate into lymphatic endothelial cells as they become spindle cells. Developmental work has shown that the lymphatic vasculature buds from the existing blood vasculature as a result of the initiation of a distinct differentiation pathway in the endothelial cells (Oliver and Harvey, 2002). Expression of Prox-1 is specific to lymphatic endothelial cells and mice that were deficient in Prox-1 had a normal blood vasculature but had no lymphatic system. While we could detect Prox-1 in primary microvascular endothelial cells (Carroll and Lagunoff, data not shown), which are known to contain significant percentages of LECs, we could not detect expression in TIME cells, indicating that TIME cells do not contain LECs and are BEC derived. Furthermore, only after infection with KSHV could we detect Prox-1 mRNA in TIME cells (Fig. 1B). We obtained quantitative data for three other major markers of lymphatic endothelium, VEGFR3, podoplanin, and LYVE-1 (Fig. 1A), and data demonstrating an increase in protein levels for two of the markers, VEGFR3 and podoplanin (Fig. 1C). Global analysis of TIME cell changes after infection yielded many more changes consistent with differentiation into the lymphatic endothelium. Ten percent of the most significant alterations after KSHV latent infection was differentially expressed between BECs and LECs and many of these changes were also altered when Prox-1 was used to drive differentiation of BECs to LECs (bold in Tables 1 and 2). Taken together, these data demonstrate that KSHV recapitulates many of the aspects of differentiation from BEC to LEC via Prox-1, but the fact that some of the changes are similar to other array data comparing BECs to LECs (Hirakawa et al., 2003; Petrova et al., 2002), but not with forced expression of Prox-1 alone in BECs (Hong et al., 2002; Petrova et al., 2002), suggests that KSHV may induce differentiation effects distinct from Prox-1 in addition to the changes programmed by Prox-1.

There is a large amount of neo-angiogenesis in KS tumors. Whether KSHV infection directly induces angiogenesis or if the effects are indirect is unknown. If KSHV directly induces angiogenesis, this study indicates that factors involved in lymphangiogenesis not hemangiogenesis should be analyzed. Lymphangiogenesis does not revolve around VEGF-A, but rather VEGF-C or VEGF-D is the main growth factor. The receptor for VEGF-C or D is VEGFR3, which is highly upregulated in infected cells. Interestingly, VEGF-C was shown to induce growth of KS spindle cells in culture (Marchio et al., 1999). While these cells are abnormal in chromosome number and no longer maintain KSHV, they were most likely infected with KSHV in vivo and subsequently lost the virus upon growth in culture. Thus, it is possible that the differentiation to lymphatic endothelium is irreversible in these cells and they now respond to VEGF-C because they are lymphatic endothelial cells. There are other changes in genes involved in angiogenesis that are not completely consistent with lymphatic differentiation. The two most prominent are VEGFR1, which is increased slightly on the array and by real-time RT-PCR, and the slight increase in VEGF-C. An increase in VEGF-C is very interesting because that could induce an autocrine loop in infected cells. Further work is underway to determine the role of VEGF-C in infected cells. VEGFR1 has been described as both a negative and a positive regulator of hemangiogenesis. If it is acting positively, it would be interesting to have an infected cell that can respond gratuitously to both lymphangiogenic and hemeangiogenic stimuli simultaneously.

Changes in the cytokine milieu have been proposed to play a major role in KS pathogenesis. The studies presented here identified three cytokines that are altered following KSHV infection. hIL-6 is one of the most highly upregulated genes following KSHV infection while IL-8 is one of the most significantly downregulated genes. MCP-1 secretion is also decreased upon infection with KSHV. IL-8 and MCP-1 decreases are consistent with differentiation from blood to lymphatic endothelial cells. In KS tumors, hIL-6 is found in high levels. As with VEGFR3 and podoplanin, changes in infected TIME cells in culture mimic gene expression seen in KS tumors in vivo. The role of hIL-6 in KS tumors is unknown but may be part of the interplay of infiltrating lymphocytes and spindle cells.

Three other studies have analyzed host gene expression following KSHV infection of endothelial cells. Two studies used commercially available primary DMVECs (Naranatt et al., 2004; Poole et al., 2002). The third study used a DMVEC line that was immortalized with E6 and E7 from human papilloma virus (Moses et al., 2002), two potent oncogenes. None of these studies noted the large number of lymphatic endothelial cell changes that we found here. Both studies done at later times postinfection contained a few changes consistent with lymphatic differentiation, but neither noted the major markers of lymphatic endothelium. It is possible that the arrays used did not have many of the lymphatic-specific genes. For example, the cDNA array we used did not contain VEGFR3, the changes were determined solely by real-time RT-PCR and at the protein level. However, it is important to note again that commercially available primary cells contain a significant percentage of lymphatic endothelial cells and thus many of the changes due to differentiation could be dampened. We have analyzed many of the genes described in primary DMVECs as well, and all of the genes are altered in a similar, yet in some cases diminished, fashion (Carroll and Lagunoff, unpublished). The most recent study examined cells at 2 and 4 h (Naranatt et al., 2004), before latency has been established and there is much more complex viral gene expression not directly relevant to latent gene expression (Krishnan et al., 2004). In the E6/E7 immortalized DMVECs, the authors found that VEGFR2 is upregulated (Moses et al., 2002). This hallmark of blood endothelial cells is not seen after latent infection of primary cells in the other study or our experiments (Poole et al., 2002). Of note, E7 can destabilize chromosomes leading to aneuploidy (Duensing and Munger, 2002), and E6 targets p53 for degradation (Scheffner et al., 1990), which can have multiple effects on angiogenesis and tumorigenesis (Ravi et al., 2000).

Overall, this study demonstrates the potential for KSHV to alter the differentiation state of endothelial cells. Because KS spindle cells express markers of LECs and KSHV infection of BECs in culture leads to upregulation of lymphatic markers, we believe that KSHV may infect blood endothelial cells or other endothelial cell types in vivo and drive their differentiation to lymphatic endothelial cells. However, the question remains, what is the role of lymphatic differentiation in KSHV biology? It is possible that differentiation induces the cells to a state where latency is better established or where the infected cells are better suited to expand and spread though the body via the lymphatic system and VEGFR3-mediated mitogenic stimulation. Further studies are ongoing to determine the functional effects of the alterations of these genes on endothelial cells and how they might be advantageous for KSHV.

Materials and methods

Cells

BCBL-1 cells are a KSHV-positive, EBV-negative B-cell line derived from a primary effusion lymphoma maintained as described elsewhere (Renne et al., 1996). TIME cells are a tert-immortalized dermal microvascular endothelial cell line (DMVEC) maintained as described elsewhere (Lagunoff et al., 2002). Primary human DMVECs (Clonetics) were cultured under the same conditions as the TIME cells.

Virus isolation and infections

Virus was isolated as before (Lagunoff et al., 2002) and was titered on TIME cells to determine the optimal amount of virus to achieve greater than 90% latent infection and less than 5% lytic infection, as determined by immunofluorescence assay (Lagunoff et al., 2002) using anti-LANA (a kind gift from A. Polson and D. Ganem) and anti-ORF59 (a kind gift from B. Forghani or purchased from ABI). Alexafluor 488 or 594 nm labeled secondary antibodies were used (Molecular Probes, Oregon). Slides were mounted in DAPI containing mounting media, Vectashield (Vector Laboratories, Inc.), then scored on a fluorescent microscope. Immunofluorescence staining with anti-podplanin antibody, purchased from AngioBio Co. (Del Mar, CA), was done similarly as above except methanol was used to fix the cells and fetal bovine serum was used as a block. The pictures in Fig. 2 were taken for each fluor under identical conditions between mock- and KSHV-infected cells.

RNA isolation and cDNA generation

Total RNA was isolated from TIME cells using RNAbee as recommended by the manufacturer (Tel-Test, Texas). Enrichment of polyadenylated mRNA was done using the Oligotex Kit as recommended by the manufacturer (Qiagen). For the first set of experiments, 8 μg of polyA-enriched RNA was labeled for microarray as described elsewhere (Geiss et al., 2001). For the second round of experiments, mRNA was amplified using the RiboAmp RNA Amplification Kit, as recommended by the manufacturer (Arcturus). Labeled cDNA was then made as described above. cDNA for real-time RT-PCR was generated from 100 ng of isolated mRNA using the SuperScript First-Strand Synthesis System according to the manufacturers directions (Invitrogen).

Microarray analysis

Slides for analysis were spotted at the University of Washington’s Center for Expression Array as described elsewhere (Geiss et al., 2001). Labeled cDNAs from the first set of experiments were hybridized to approximately 15000 gene array spotted in duplicate on two slides. Each experiment was done with the control and experimental cDNA labeled with cy3 and cy5 and then repeated with the labels reversed. The second round of experiments was done on approximately 13000 gene array identical to the original array except for the absence of approximately 2000 spots. RNA for each time point was isolated from different infections, performed on different days, using different viral stocks. Microarray data were analyzed using the University of Washington’s Center for Expression Array and the Resolver Biosoftware Package (Rosetta Inpharmatics).

Real-time and semi-quantitative RT-PCR

cDNA was generated as stated above, and commercially available TaqMan probes (Assays-on-Demand) were used, following the manufacturers directions (Applied Biosystems). iTaq DNA Polymerase (Biorad) was used for the amplification of 2 μl of cDNA with the TaqMan probes, and the reactions were monitored on an Icycler thermocycler (Biorad). Changes in relative abundance of transcript were normalized to GAPDH expression (Threshold cycle, C_T) compared to control expression in UV-irradiated KSHV-treated TIME cells using the following equation (C_T of experimental gene in control cells minus C_T of GAPDH) minus (C_T of experimental gene in KSHV-infected cells minus C_T of GAPDH). For semi-quantitative RT-PCR, 5–160 ng of mRNA was amplified using the OneStep RT-PCR kit (Qiagen) as recommended using the following 5′ to 3′ primers; Prox-1 F: CCCGGATCCGTGATGCCTGACCATGAC, Prox-1 R: AAACCCGTC-GACTTTCTACTCATGAAGCAGCTC and GAPDH F: AAGGTGAAGGTCGGAGTCAACG, GAPDH R: TGGAAGATGGTGATGGGATTTC. The Prox-1 primers are tailed with restriction sites. The product resulting from RT-PCR with these primers was sequenced and found to perfectly match the human Prox-1 sequence.

Western blot analysis

Forty-eight hours postinfection, KSHV-infected and mock-infected TIME cells were lysed in RIPA lysis buffer (50 mM Tris–HCL, pH 7.6, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% sodium dodecyl sulfate supplemented with 0.5 mM dithiothreitol, 0.4 mM phenylmethylsulfonyl fluoride, 2 μg/ml each of aprotinin, pepstatin, and leupeptin, and 1 mM sodium ortho-vanadate). Protein in the lysis supernatant was quantified by BCA assay (Pierce), and 20 or 40 μg of total protein was subjected to SDS-PAGE. After transfer to PVDF (Immobilon-P, Millipore), the blot was probed with the following primary antibodies: rabbit anti-podoplanin (Research Diagnostic Inc, RDI-102PA40), rabbit anti-VEGFR3 (Santa Cruz, SC-321), and mouse anti-COX1 (Cayman Chemical, 160110), then washed and probed with either goat anti-mouse or goat anti-rabbit (Jackson ImmunoResearch Laboratories). Chemiluminescence was then performed using ECL+ Plus (Amersham) and exposing to Blue X-Ray Film (Phenix Research Products).

Cytokine antibody array

DMVECs were infected with KSHV or mock infected for 42 and 90 h. Media were replaced with serum-free media and left on the cells at 37 °C for 6 h. The serum-free media were then harvested, cellular debris was centrifuged out, and the supernatant was diluted 1:5 and used as the probe on a Human Cytokine Array V spotted with antibodies to 79 different cytokines (RayBiotech, Inc.). The membrane was treated as recommended by the manufacturer. After the final wash the membrane was developed using ECL+ Plus Western Blotting Detection System (Amersham Biosciences).