Oncogenic Herpesvirus Engages Endothelial Transcription Factors SOX18 and PROX1 to Increase Viral Genome Copies and Virus Production

Abstract

Kaposi sarcoma is a tumor caused by Kaposi sarcoma herpesvirus (KSHV) infection and is thought to originate from lymphatic endothelial cells (LEC). While KSHV establishes latency in virtually all susceptible cell types, LECs support spontaneous expression of oncogenic lytic genes, high viral genome copies, and release of infectious virus. It remains unknown the contribution of spontaneous virus production to the expansion of KSHV-infected tumor cells and the cellular factors that render the lymphatic environment unique to KSHV life cycle. We show here that expansion of the infected cell population, observed in LECs, but not in blood endothelial cells, is dependent on the spontaneous virus production from infected LECs. The drivers of lymphatic endothelium development, SOX18 and PROX1, regulated different steps of the KSHV life cycle. SOX18 enhanced the number of intracellular viral genome copies and bound to the viral origins of replication. Genetic depletion or chemical inhibition of SOX18 caused a decrease of KSHV genome copy numbers. PROX1 interacted with ORF50, the viral initiator of lytic replication, and bound to the KSHV genome in the promoter region of ORF50, increasing its transactivation activity and KSHV spontaneous lytic gene expression and infectious virus release. In Kaposi sarcoma tumors, SOX18 and PROX1 expression correlated with latent and lytic KSHV protein expression. These results demonstrate the importance of two key transcriptional drivers of LEC fate in the regulation of the tumorigenic KSHV life cycle. Moreover, they introduce molecular targeting of SOX18 as a potential novel therapeutic avenue in Kaposi sarcoma. SIGNIFICANCE: SOX18 and PROX1, central regulators of lymphatic development, are key factors for KSHV genome maintenance and lytic cycle in lymphatic endothelial cells, supporting Kaposi sarcoma tumorigenesis and representing attractive therapeutic targets.

Figure 1.

Spontaneous virus production from KLECs supports virus spread in LECs. A, Immunofluorescence staining in the indicated cell types at 14 d.p.i. for K8.1 (red). Nuclei are counterstained with Hoechst 33342. Representative images from two independent experiments are shown. B, Titration of infectious virus released from KBECs and KLECs at 14 d.p.i. Infected U2OS target cells were quantified by their GFP expression. Single values from n = 4 biological replicates are shown. Bars, mean ± SD. P value was calculated using two-tailed paired t test. C, Quantification of intracellular viral genome copies from KBECs and KLECs at 14 d.p.i. Single values from n = 3 biological replicates are shown. Bars, mean ± SD. D, KLECs were mixed with noninfected BECs (left) or LECs (right) at a ratio of 1:50; three days later, PAA was added. At the indicated timepoints the percentage of infected (GFP⁺) cells was quantified. Each dot represents one biological replicate. E, Left, LECs were infected at low multiplicity of infection (MOI) 0.5, and at three d.p.i., PAA was added. At the indicated timepoints, the percentages of GFP⁺ cells were quantified. Each dot represents one biological replicate.

Figure 2.

Endothelial TFs and lytic genes are highly expressed in KLECS. Primary BECs and LECs were infected with rKSHV.219 or left uninfected and analyzed at the indicated timepoints (A and B) and at 14 d.p.i. (C–E). A, qRT-PCR for the indicated cellular targets at the indicated timepoint in BECs (top) and LECs (bottom). Single data from n = 3 independent experiments ± SD are shown for each timepoint. P values were calculated using ordinary one-way ANOVA, followed by Dunnett correction for multiple comparisons. *, P < 0.033; **, P < 0.02; ***, P < 0.001. B, Immunoblotting of BECs (top) and LECs (bottom) treated as in A for the indicated targets, and γ-tubulin (TUBG1) was used as a loading control. The experiment was repeated three independent times. Numbers below each blot indicate relative band intensity normalized to TUBG1. C and D, qRT-PCR and immunoblotting, respectively, for the indicated targets at 14 d.p.i., and TUBG1 was used as a loading control. Numbers below each blot indicate relative band intensity normalized to TUBG1. The experiments were repeated three independent times. E, Immunofluorescence staining for PROX1, SOX18, and COUPTF2 in the indicated cell types, treated as in C and D. Representative images are shown.

Figure 3.

SOX18 and PROX1 regulate KSHV productive lytic replication cycle through different mechanisms. A–D, KLECs were treated at 14 d.p.i. with the indicated siRNAs for 72 hours and analyzed. A, qRT-PCR for the indicated targets. Single values from n = 3 independent experiments are shown. Bars, mean ± SD. B, Immunoblotting for the indicated targets, and actin was used as a loading control. The experiment was repeated three independent times. C, Titration of released infectious virus on naïve U2OS cells. Single values from n = 3 independent replicates are shown. Bars, mean ± SD. D, Relative number of intracellular KSHV genomes. Single values from n = 3 independent experiments are shown. Bars, mean ± SD. E and F, KLECs were treated at 14 d.p.i. with the indicated siRNAs or 0.5 mmol/L PAA for 72 hours. E, Cells were analyzed by qRT-PCR for the indicated viral targets. Single values from n = 3 independent experiments are shown. Bars, mean ± SD. F, Titration of released infectious virus as in C. Single values from n = 2 independent experiments are shown. Bars, mean ± SD. P values were calculated in A and C–F using ordinary one-way ANOVA, followed by Dunnett correction for multiple comparisons. *, P < 0.033; **, P < 0.02; ***, P < 0.001.

Figure 4.

Selective inhibition of SOX18 efficiently decreases the number of KSHV episome copies. A–D, KLECs at 10 d.p.i. were treated with the indicated compounds at the concentrations shown for six days. Every second day, the drug-containing media were refreshed. The relative number of intracellular KSHV genome copies was quantified (A and C), and the released infectious virus in the supernatant was measured by infecting naïve U2OS cells (B and D). Single values from n = 3 independent replicates are shown. Bars, mean ± SD. P values were calculated using ordinary one-way ANOVA, followed by Dunnett correction for multiple comparisons. Exact P values are shown. E, iSLK.219 cells were transduced with lentiviruses expressing either SOX18 or NLS-mCherry (mCherry) vector control. One day later, cells were treated either with DMSO or with the indicated drug for 48 hours in the presence or absence of doxycycline for the induction of the lytic cycle. The relative number of intracellular KSHV genome copies was quantified. Single values from n = 3 biological experiments are shown. Bars, mean ± SD. P values were calculated using two-tailed paired t test.

Figure 5.

SOX18 binds to KSHV genome in the proximity of origins of replication. A and B, Relative number of intracellular KSHV genome copies (left) and immunoblotting for the indicated proteins (right) at 48 hours in KBEC transduced with either SOX18 or NLS-mCherry vector control (mCherry) expressing lentiviruses (A) or iSLK.219 cells transduced with increasing doses of lentiviruses expressing either SOX18 or mCherry control (B). Single values for n = 3 biological replicates are shown. Bars, mean ± SD. C and D, Luciferase reporter assays in HeLa cells transfected as indicated. Single values from C (n = 8) and D (n = 4) biological replicates are shown. Bars, mean ± SEM. E–G, Schematic representation of the promoter regions (top; numbers indicate the nucleotides upstream or downstream of the black regions) amplified by qPCR following ChIP using either anti-SOX18 (middle) or anti-Myc antibodies (bottom) to precipitate endogenous or Myc-tagged SOX18. Anti-HA was used as a negative control. Single values for n = 3 biological replicates are shown. Bars, mean ± SEM. P values in all panels were calculated using ordinary one-way ANOVA, followed by Dunnett correction for multiple comparisons. *, P < 0.033; **, P < 0.02; ***, P < 0.001.

Figure 6.

PROX1 is a positive regulator of viral gene expression, interacts with ORF50, and binds to its promoter region during KSHV lytic replication cycle. A, KLECs (left) and iSLK.219 cells (right) in n = 3 independent replicates were treated as indicated and subjected to RNA-seq. Fold change ± SEM in the viral transcripts over the appropriate control is shown. B, KLECs (top) and KBECs (bottom) were transduced with the indicated lentiviral vectors for 72 hours. Cells were analyzed by qRT-PCR for the indicated targets. Single values from n = 2 independent experiments are shown. Bars, mean ± SD. C, Luciferase reporter assay in HEK293FT transfected with the indicated plasmids for 36 hours. Single values for n = 4 biological replicates are shown. Bars, mean ± SD. D, Coimmunoprecipitation of PROX1 with ORF50 in KLECs. The experiment was repeated two independent times. E, Immunofluorescence for PROX1 (red) and ORF50 (green) in KLECs infected with WT KSHV at 16 d.p.i. A representative image is shown; the experiment was done two times. F, PLA with antibodies against PROX1 and ORF50, or IgG control. Top and middle, PLA signal is shown by yellow puncta. Nuclei were counterstained with Hoechst 33342. Bottom, quantification of PLA puncta/nucleus; n, number of nuclei quantified. The experiment was repeated two independent times. G, Top, schematic representation of the ORF50 promoter regions (numbers indicate the nucleotides upstream of the ORF50 TSS) amplified by qPCR following ChIP. Middle, ChIP in iSLK.219 cells transduced with lentivirus expressing Myc-tagged PROX1 reactivated (doxycycline) for 24 hours. ChIP was performed with anti-PROX1 and anti-Myc and DNA was amplified in the promoter regions of the ORF50 gene. Bottom, ChIP in KLECs 14 d.p.i. using anti-PROX1 antibody as above. Bars, mean ± SD. P values in A–C, F, and G were calculated using ordinary one-way ANOVA, followed by Dunnett correction for multiple comparisons. *, P < 0.033; **, P < 0.02; ***, P < 0.001.

Figure 7

PROX1, SOX18, COUPTF2, and K8.1 are expressed in Kaposi sarcoma tumors. A, Representative images of consecutive sections from AIDS-Kaposi sarcoma biopsy stained as indicated. The red box indicates the portion of the biopsy magnified in the right inset. B–D, Multiplex IHC staining of 19 Kaposi sarcoma biopsies for PROX1, SOX18, LANA, and K8.1. Correlation was calculated considering the mean intensity of each staining for 8345 tumor cells across the 19 biopsies (mean: 439 cells/biopsy; range: from 102 to 754 cells/biopsy analyzed). K8.1 versus LANA: Pearson correlation coefficient (PCC):0.181, P < 0.0001; K8.1 versus PROX1: PCC:0.407, P < 0.0001; LANA versus PROX1: PCC:0.626, P < 0.0001; LANA versus SOX18: PCC:0.434, P < 0.0001. K8.1 versus SOX18: PCC:0.148, P < 0.0001. P values were corrected for the false discovey rate using the Bonferroni correction. B, Representative image. Nuclei were counterstained with DAPI. Orange arrowheads, cells positive for all the markers; white arrowhead, a cell positive for LANA but not for the other markers; cyan arrowhead, uninfected cells. C and D, Violin plots showing marker intensities with continuous values (y-axis) and categorized values for cells negative (−) or positive (+) for the indicated marker (x-axis). For the categorized LANA and PROX1, the lowest quartile intensities were set as the negativity/positivity thresholds, equaling to 2.7-fold and 2.6-fold background intensities (minimum cell value), respectively. a.u., arbitrary units. ***, P < 0.0001. E, Schematic model of the role of KSHV spontaneous lytic replication in the expansion of viral infection in a LEC culture (left) and of the roles of SOX18 (top right) and PROX1 (bottom right) as positive regulators of viral episome copy numbers, lytic gene expression, and infectious virus release, respectively.