Kaposi’s sarcoma-associated herpesvirus vFLIP promotes MEndT to generate hybrid M/E state for tumorigenesis

Kaposi’s sarcoma (KS) is an angioproliferative and invasive tumor caused by Kaposi’s sarcoma-associated herpesvirus (KSHV). The cellular origin of KS tumor cells remains contentious. Recently, evidence has accrued indicating that KS may arise from KSHV-infected mesenchymal stem cells (MSCs) through mesenchymal-to-endothelial transition (MEndT), but the transformation process has been largely unknown. In this study, we investigated the KSHV-mediated MEndT process and found that KSHV infection rendered MSCs incomplete endothelial lineage differentiation and formed hybrid mesenchymal/endothelial (M/E) state cells characterized by simultaneous expression of mesenchymal markers Nestin/PDGFRA/α-SAM and endothelial markers CD31/PDPN/VEGFR2. The hybrid M/E cells have acquired tumorigenic phenotypes in vitro and the potential to form KS-like lesions after being transplanted in mice under renal capsules. These results suggest a homology of KSHV-infected MSCs with Kaposi’s sarcoma where proliferating KS spindle-shaped cells and the cells that line KS-specific aberrant vessels were also found to exhibit the hybrid M/E state. Furthermore, the genetic analysis identified KSHV-encoded FLICE inhibitory protein (vFLIP) as a crucial regulator controlling KSHV-induced MEndT and generating hybrid M/E state cells for tumorigenesis. Overall, KSHV-mediated MEndT that transforms MSCs to tumorigenic hybrid M/E state cells driven by vFLIP is an essential event in Kaposi’s sarcomagenesis.

vWF, which are expressed in KS ( Figure 2C), suggesting that the 2-D cell culture 1 system may not faithfully represent the MEndT in tumors. 2 Three-dimensional (3-D) organotypic cultures allow the mimic function of living 3 tissue and probably provide information encoded in tissue architecture. The 3-D 4 culture was used in mesenchymal stem cells in that MSC spheroids display enhanced 5 differentiation capability compared to 2-D culture [31,32]. We established a 3-D 6 spheroid model by seeding mock-and KSHV-infected PDLSCs in a low attachment 7 condition. As time went by, PDLSCs formed a decentralized network, and then 8 numerous small aggregates progressively assembled into a single central spheroid 9 ( Figure 2D). Once aggregated, the spheroid did not change in size but was generally 10 compacted. The expression spectrum of mesenchymal and endothelial cell markers in 11 mock-and KSHV-infected PDLSC spheroids was examined by IFA. As shown in 12 Figure  Matrigel tubule formation assay was carried out for the acquisition of endothelial and 8 angiogenesis property and showed that KSHV-infected PDLSCs exhibited increased 9 ability to form capillary-like structures in comparison to mock-infected PDLSCs 10 ( Figure 3A). Uptake of acetylated low-density lipoprotein (Ac-LDL) is a hallmark of 11 endothelial cells and macrophages [33]. KSHV-infected PDLSCs were found to 12 possess an Ac-LDL uptake capacity similar to HUVECs ( Figure 3B). KSHV induced 13 a notable outgrowth in KSHV-infected PDLSC spheroids whereas rare sprouting was 14 observed in mock-infected PDLSCs ( Figure 3C). Interestingly, KSHV-infected 15 PDLSCs spontaneously formed many vessel-like structures in the 3D spheroids, but 16 such structures were not seen in mock-infected PDLSC spheroids ( Figure 3D). 17 Immunofluorescence staining of these lumens showed the expression of 18 pan-endothelial marker CD31 in KSHV-infected PDLSC spheroids but not in control 19 spheroids ( Figure 3E). 20 13 To investigate the endothelial differentiation grade of KSHV-infected PDLSCs in vivo, 1 KSHV-infected PDLSCs mixed with Matrigel were injected into C57BL/6 mice. 2 Seven days later, the matrigel plugs were striped. Immunohistochemistry analysis 3 using antibodies against CD34 (mature and immature blood vessel marker), hCD105 4 (immature blood vessel marker), and hCD31 (mature blood vessel marker), 5 respectively, showed enhanced staining for CD34, hCD105, and hCD31 in the 6 Matrigel plug from the mice transplanted with KSHV-infected PDLSCs ( Figure 3F). 7 Meanwhile, an enhanced number of leaky blood vessels and neo-vascular cavity were 8 observed in KSHV-infected PDLSC plugs, suggesting that a fraction of 9 KSHV-infected PDLSCs can differentiate into immature and mature vessels, and may 10 also induce mouse endothelial cells to participate in neovascularization to form leaky 11 vessels. Weibel-Palade bodies (WPBs) are endothelial cell specialized organelles. We It was reported that KSHV infection of MSC increases its migration and invasion 3 capability, which was proposed to be responsible for the tendency of KS occurring in 4 injured or inflamed sites of the body [36]. To assess the migration ability of distinct 5 phenotypic state cells, mock-and KSHV-infected PDLSC spheroids were seeded on 6 adherent culture plates, and observed for spindle-shaped cells migrating from the 7 spheroids. We found that more KSHV-infected PDLSCs moved away from their 8 spheroids than mock-infected spheroids. When they were plated on nonadherent 9 surfaces, no migration was observed with both mock-and KSHV-infected spheroids 10 ( Figure 5B). Then the migrating cells were analyzed for PDPN/PDGFRA expression 11 profile to determine their MEndT status. Results showed that the migrating cells were 12 mainly hybrid M/E state cells ( Figure 5C). The invasion abilities of KSHV-PDLSC 13 and M/E cells were assayed using a transwell apparatus as illustrated in Figure 5D.    suspensions. GFP-positive cells were sorted for the first round, and 20 PDGFRA+/PDPN-or PDGFRA+ /PDPN+ cells were isolated by 20% cutoffs for the 1 second rounds of sorting. Isolated cells were allowed to culture for no more than two 2 weeks. 3

Cell migration/invasion assay 4
Cell migration and invasion assays were performed using 24-well transwell chambers 5 with filter membranes 12 µm pore size (Millipore Corporation, PIXP01250). Cells 6 were detached with trypsin-EDTA, washed once with 1xPBS, and then resuspended 7 in serum-free medium. 3 x10 4 cells were placed in transwell insert, and the medium 8 containing 20% FBS was added to the lower chamber. After 24h incubation, 9 non-migration cells were removed with a cotton swab. The migrated cells were 10 stained and counted under a ZEISS microscope. The cell invasion assay was 11 performed by following the same procedures as cell migration assay except that 12 transwell inserts were precoated with cooled Matrigel (BD Biosciences). 13 For 3D spheroids migration assay, the spheroids were collected, washed with 1xPBS, 14 and resuspended in serum-free medium. 5-10 spheroids were seeded in adherent 15 24-well plates or nonadherent 24-well plates precoated with 0.5% agarose. After 48h 16 incubation, images were captured under a ZEISS microscope. To detect the status of 17 migrated cell away from spheroid, the spheroids were detached by Ophthalmic 18 forceps, and the remaining migrated cell were collected for flow cytometric analysis 19 after staining with PDGFRA-APC and PDPN-PE antibodies. 20 3D spheroids invasion assay was performed using 24-well transwell chambers with 1 filter membranes 12 µm pore size. 5-10 Spheroids resuspended in serum-free medium 2 were seeded in an insert that has been coated with Matrigel (BD Biosciences). Lower 3 compartment was added with medium containing 20% FBS. Plates were then 4 incubated at 37°C for 48h to allow cells to migrate. Non-migration cells were 5 removed with a cotton swab. The migrated cells were stained and counted under a 6 ZEISS microscope. To detect the status of invaded cell away from spheroids, cells on 7 the lower side of the membrane were collected for flow cytometric analysis after 8 staining with PDGFRA-APC and PDPN-PE antibodies. 9

Soft agar colony formation 10
To determine cellular tumor transformation activity, 24-well plates coated with 0.5% 11 agarose medium were prepared. After the agar is solidified, a total of 2,000 cells, 12 suspended in αMEM medium supplemented with 0.3% agarose and 20% FBS, were 13 seeded onto the soft agar coated 24-wells. Cells in soft agar were incubated at 37°C in 14 a 5% CO 2 incubator for 3-4 weeks, and fresh culture medium was added to each well 15 every 3 to 4 days. Colonies larger than the average size of control colonies were 16 counted. 17

Western blotting 18
Cell lysates were prepared as previously described [11]. Whole cell extracts of 30 µg 19 protein were resolved in SDS-PAGE and transferred onto nitrocellulose membranes. 20 The membranes were blocked with 5% non-fat milk/PBS for 30 min and incubated Anti-IR Dye 800 or Dye 680 anti-rabbit or anti-mouse IgG antibodies (LI-COR 7 Biosciences) were used as the secondary antibodies. An Odyssey system (LI-COR 8 Biosciences) was used for detection of proteins of interest. 9

RT-qPCR 10
Total RNA was extracted by Ultrapure RNA Kit (CWBIO, CW0581) in accordance 11 with manufacture's instructions. cDNA was synthesized by reverse transcription. 12 cDNA was diluted 5 times and subjected to real-time PCR using LightCycler 480 13 SYBR Green I Master (Roche) with specific primers for the genes of interest. 14 GAPDH gene was used for calibration. The primer sequences used for RT-qPCR are 15 listed in Table 1. All real-time PCR was done in triplicate. 16