Proteomic Profiling Identifies Kaposi’s Sarcoma-Associated Herpesvirus (KSHV)-Encoded LANASIM-Associated Proteins in Hypoxia

ABSTRACT Hypoxia signaling is a key regulator in the development and progression of many types of human malignancies, including viral cancers. The latency-associated nuclear antigen (LANA), encoded by Kaposi’s sarcoma-associated herpesvirus (KSHV) during latency, is a multifunctional protein that plays an essential role in viral episome maintenance and lytic gene silencing for inducing tumorigenesis. Although our previous studies have shown that LANA contains a SUMO-interacting motif (LANASIM), and hypoxia reduces SUMOylated KAP1 association with LANASIM, the physiological proteomic network of LANASIM-associated cellular proteins in response to hypoxia is still unclear. In this study, we individually established cell lines stably expressing wild-type LANA (LANAWT) and its SIM-deleted mutant (LANAdSIM) and treated them with or without hypoxia, followed by coimmunoprecipitation and mass spectrometry analysis to systemically identify the hypoxia-responsive profile of LANASIM-associated cellular proteins. We found that in hypoxia, the number of cellular proteins associated with LANAWT instead of LANAdSIM was dramatically increased. Functional network analysis revealed that two major pathways, which included cytoskeleton organization and DNA/RNA binding and processing pathways, were significantly enriched for 28 LANASIM-associated proteins in response to hypoxia. HNRNPU was one of the proteins consistently identified that interacted with LANASIM in different proteomic screening systems and responded to hypoxia. This study provides a proteomic profile of LANASIM-associated proteins in hypoxia and facilitates our understanding of the role of the collaboration between viral infection and the hypoxia response in inducing viral persistence and tumorigenesis. IMPORTANCE Kaposi’s sarcoma-associated herpesvirus (KSHV) has been reported to be involved in the regulation of host proteins in response to hypoxic stress. LANA, one of the key latent proteins, contains a SUMO-interacting motif (LANASIM) and reduces the association with SUMOylated KAP1 upon hypoxic treatment. However, the physiological systematic network of LANASIM-associated cellular proteins in hypoxia is still unclear. Here, we revealed two major pathways, which included cytoskeleton organization and DNA/RNA binding and processing pathways, that were significantly enriched for 28 LANASIM-associated proteins in hypoxia. This discovery not only provides a proteomic profile of LANASIM-associated proteins in hypoxia but also facilitates our understanding of the collaboration between viral infection and hypoxic stress in inducing viral persistence and tumorigenesis.

also facilitates our understanding of the collaboration between viral infection and hypoxic stress in inducing viral persistence and tumorigenesis. KEYWORDS KSHV, LANA, SIM, hypoxia, HNRNPU, SUMO-interacting motif H ypoxia (low oxygen) is a common feature of the hostile microenvironment that results from the rapid growth of cancer cells (1). It has been demonstrated that hypoxia occurs in many types of human malignancies, including virus-mediated cancers (2). Kaposi's sarcoma (KS)-associated herpesvirus (KSHV), or human herpesvirus 8 , mainly infects two cell types, human endothelial cells and B lymphocytes, and is tightly associated with KS, primary effusion lymphoma (PEL), and a subset of multicentric Castleman's diseases (MCDs) (3). Like other herpesviruses, KSHV has two stages in its life cycle: latent infection and lytic replication. During latency, the KSHV genome is anchored to the host chromosome as an episome, with only a few viral genes being expressed (4). The latency-associated nuclear antigen (LANA), encoded by open reading frame 73 (ORF73), is one of the viral proteins dominantly expressed in latent infection. It has been demonstrated that LANA is a nuclear protein with multiple functions, particularly tethering the viral episome to host chromatin by acting as an adaptor (through linking the tandem repeat [TR] region of the viral genome with histones H1 and H2A/B of host chromosomes via binding to its C terminus and N terminus, respectively) to ensure the maintenance of the viral genome during host cell segregation (5). In addition, LANA can directly or indirectly interact with a number of host proteins, such as binding to the promoter region of host/viral genes and inhibiting viral lytic replication, host cell immune responses, and apoptosis, thereby promoting the establishment and maintenance of KSHV latent infection and ultimately inducing the formation of tumors (6). We and other groups have also shown that LANA not only interacts with HIF1a (one of the key hypoxia-inducible factors) to block lytic replication under normoxic conditions but also protects the cellular replication machinery from hypoxiainduced degradation (7,8), although hypoxia could reactivate KSHV from latency to lytic replication (7). However, the molecular mechanisms of how KSHV controls its life cycle in response to hypoxia remain largely unclear.
Emerging evidence indicates that small ubiquitin-like modifier (SUMO), one of the posttranslational modifications of proteins, plays a critical role in the regulation of gene transcription and the response to extracellular stress (9,10). In addition to covalently attaching to substrate proteins, SUMO can also noncovalently bind to target proteins through a SUMO-interacting motif (SIM), which contains a consensus sequence including K-x 3-5 -I/V-I/L-I/L-x 3 -D/E/Q/N-D/E-D/E, h-h-x-S-x-S/T-a-a, or V/I-x-V/I-V/I (11)(12)(13). Our previous work showed that there is a SIM localized within the amino terminus of the LANA protein (LANA SIM ), which can recruit a variety of proteins, including SUMO2-modified KAP1 and Sin3A, to form a transcription-inhibitory complex. This complex in turn silences gene expression and maintains the viral episome along with host cell segregation during latency. In hypoxia, the SUMOylated protein, including KAP1, is dissociated from the LANA SIM -associated complex, which opens the chromatin structure to activate viral lytic replication (14). By expressing and purifying the recombinant protein of the SIM of LANA fused with glutathione S-transferase (GST) as bait in vitro to capture the LANA SIM -interacting proteins from nuclear extracts of PEL cells, combined with mass spectrometry (MS) analysis, we identified a series of proteins that are mainly involved in the regulation of the cell cycle, DNA unwinding and replication, and pre-mRNA/mRNA processing associated with LANA SIM (15). However, a limitation of the previous study was that only amino acids 240 to 300 of LANA were used for capturing interacting proteins; this short amino acid sequence might not be representative of the structure of LANA SIM , and in vitro GST pulldown assays could not fully reflect the endogenous protein-protein interactions that physiologically exist within the cellular complexes in cells.
In this study, to further elucidate the physiological functions of LANA SIM in response to hypoxia, we initially generated human umbilical vein endothelial cell (HUVEC) and HEK293 cell lines stably expressing wild-type LANA (LANA WT ) and its SIM-deleted mutant (LANA dSIM ) individually, followed by treatment with or without hypoxia. We then performed coimmunoprecipitation combined with tandem quantitative mass spectrometry analyses to identify and highlight the interacting proteomic profile of LANA SIM in response to hypoxia. These results provide new insights into the collaborative role of LANA SIM -mediated proteins between viral infection and the hypoxia response for viral pathogenesis and tumorigenesis.

RESULTS
Quantitative proteomic analysis of LANA SIM -associated proteins in hypoxia. To identify proteins specifically associated with the SIM of LANA (LANA SIM ) in a physiological setting (Fig. 1A), we generated two types of cell lines (HUVECs and HEK293 cells) stably expressing recombinant wild-type LANA fused with both yellow fluorescent protein (YFP) and Flag tags (YFP-LANA WT -Flag), its SIM-deleted mutant (YFP-LANA dSIM -Flag), or the vector alone (YFP-Flag). The stable cell lines were then individually subjected to normoxia or hypoxia treatment for 48 h, followed by coimmunoprecipitation assays with anti-Flag antibodies and relative quantitative matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) MS analysis and protein sequence identification (Fig. 1A). As shown in Fig. 1B, along with immunoblot (IB) analysis to verify the expression level of LANA as a control, we observed that the bands representing the interacting cellular proteins were enriched by LANA WT or LANA dSIM in both HUVEC and HEK293 stable cell lines, with both hypoxia and normoxia treatments, compared with the vector-alone group (Fig. 1B, top). Interestingly, the deletion of the SIM of LANA or hypoxia treatment consistently reduced the intensity of the interacting proteins pulled down by LANA in both HUVEC and HEK293 stable cell lines, which supported our previous observation that LANA SIM functions in response to hypoxic stress (14). To confirm whether the LANA SIM -interacting proteins present higher sensitivity in response to hypoxia, the immunoprecipitated complexes of LANA WT and LANA dSIM from both HUVEC and HEK293 stable cell lines in normoxia or hypoxia were individually subjected to MALDI-TOF MS analysis (Fig. 1C), along with the vector alone as a parallel control to exclude the nonspecifically binding proteins. The results showed that 96 cellular proteins interacted with LANA WT from HUVECs in normoxia, while 169 cellular proteins were associated with hypoxia. Notably, 54 cellular proteins appeared in both normoxia and hypoxia (Fig. 1D, right). In contrast, despite the fact that 70 cellular proteins exclusively interacted with the SIM-deleted mutant of LANA (LANA dSIM ) from HUVECs in normoxia, only 56 cellular proteins were associated exclusively with hypoxia. The number of cellular proteins pulled down by LANA WT from stable HUVECs in hypoxia and normoxia was much higher (fold change for the WT versus dSIM of 2.73-versus 0.8-fold) than that pulled down by LANA dSIM (Fig. 1D, right). Consistently, a similar phenomenon (fold change for the WT versus dSIM of 5.35-versus 2.42-fold) was observed in HEK293 stable cell lines (Fig. 1D, left). This indicates that the loss of the SIM may reduce the ability of LANA to deal with hypoxic stress.
To further explore the role of LANA SIM , we analyzed LANA SIM -associated protein profiles from HUVEC or HEK293 stable cell lines under normoxic or hypoxic conditions, based on the screening criterion of an emPAI ratio of LANA dSIM /LANA WT of $2 (increased with the deletion of SIM) or #0.5 (decreased with the deletion of SIM). As shown in the top panels of Fig. 2A, there were 114 cellular proteins associated with LANA SIM from stable HUVECs in normoxia, while there were 187 proteins in hypoxia. Among these, 71 proteins were increased and 43 were decreased in normoxia, and 51 proteins were increased and 136 were decreased in hypoxia. A similar phenomenon was also observed in the HEK293 stable cell lines ( Fig. 2A, bottom). Functional clustering analysis revealed that LANA SIMassociated cellular proteins from stable HUVECs in hypoxia were enriched mainly in cell cycle and cytoskeleton organization pathways and less so for protein transport or protein localization pathways compared to what was seen in normoxia (Fig. 2B, left). In contrast, LANA SIM -associated cellular proteins from HEK293 stable cells in hypoxia were also predominantly enriched in the cytoskeleton organization pathway, in addition to the cellular component biogenesis pathway, and more limited in the association with proteins in the mRNA splicing/spliceosome and metabolic process pathways (Fig. 2B, right). This provides evidence that the key biological processes associated with the SIM of LANA are different in normoxia and hypoxia, and the cytoskeleton organization pathway may be a common target of LANA SIM in hypoxia.
Comparison between LANA SIM -and LANA dSIM -associated proteomic profiles in hypoxia. To validate whether those proteins enriched by LANA SIM in hypoxia are related to SUMO modification, we chose HUVECs as a model for further analysis and compared the differences between LANA SIM -associated proteins and LANA dSIM -associated proteins. As described previously (15), we predicted the potential SUMO modification sites and SIMs from 101 (out of 187) proteins exclusively associated with LANA SIM in hypoxia and 13 (out of 99) proteins exclusively associated with LANA dSIM as a control (Fig. 3A). As shown in Fig. 3B, 88.2% of the identified proteins have a high potential for being SUMOylated in the LANA SIM -associated profile, while a relatively lower number of proteins (71.3%) was identified in the LANA dSIM -associated profile (Fig. 3B). Unexpectedly, the distributions of LANA SIM -and LANA dSIM -associated proteins with different numbers (1, 2, 3, or .3) of WKxE or noncanonical consensus SUMOylated sites were similar (Fig. 3C). The identified proteins containing only one potential SUMOylation site were the majority in both the LANA SIM -and LANA dSIM -associated profiles (Fig. 3C). Consistent with our previous study, the interaction of SUMOylated proteins with LANA SIM has a higher capacity for SIMs, with about 48.2% of the LANA SIM -associated proteins in hypoxia but only 30.5% of the LANA dSIM -associated proteins containing at least 1 SIM (Fig. 3D). This indicates that the LANA SIM -associated proteins in hypoxia are more easily SUMOylated than the LANA dSIM -associated proteins in hypoxia.
The functional core network of LANA SIM -associated proteins in hypoxia. Since increased and decreased profiles of the LANA SIM -associated proteins from HEVECs or HEK293 cells in hypoxia were observed once the SIM was deleted ( Fig. 2A, right), we attempted to perform an enrichment analysis of the related functional pathways targeted by LANA SIM . The results showed that functional pathways of the decreased proteins targeted by LANA SIM in hypoxia were mainly related to protein localization, translational initiation, and mRNA processing, while the increased ones mainly participated in the regulation of cytoskeleton organization and spindle organization (Fig. 4A). To further evaluate the consistency of the increased and decreased profiles of the LANA SIM -associated proteins in hypoxia, we analyzed and aligned the LANA SIM -associated proteins from HUVECs and HEK293 cells in hypoxia (Fig. 4B) and found that 16 of the increased proteins and 12 of the decreased proteins were consistently associated with LANA SIM (Table 1). Interestingly, the functional core network analysis of these 28 proteins revealed that the LANA SIM -associated proteins in hypoxia are mainly involved in the regulation of cytoskeleton organization and protein localization and DNA/RNA binding and processing ( Fig. 4C and Table 1). Among these, SUMO3, a well-known SUMO molecule that responds to hypoxic stress, was also observed (Fig. 4C). This supports our hypothesis and previous reports from our group and other investigators (15)(16)(17).
To address whether there are some identified proteins that overlap between the LANA SIM -associated proteome profile in hypoxia and the GST-SIM-associated proteome profile reported previously (15), we aligned each protein within the LANA SIM and GST-SIM profiles. Surprisingly, there were only two proteins, namely, SERPINH1 and HNRNPU, that were identified in both the LANA SIM and GST-SIM profiles, while most of the other proteins were exclusive for each of their independent profiles (Fig. 4D). To further validate the potential association of SERPINH1 and HNRNPU with LANA SIM in response to hypoxia, we performed coimmunoprecipitation analyses by overexpressing LANA or RTA in HEK293 cells. The results showed that SERPINH1 and HNRNPU, but not RTA, associated with LANA to some extent (Fig. 5A). In view of the fact that LANA has a higher affinity for HNRNPU than SERPINH1, the interaction of endogenous LANA with HNRNPU was further confirmed in PEL cells by coimmunoprecipitation assays with LANA-specific antibodies (Fig. 5B). The results from immunofluorescence assays showed not only that the expression of LANA in HEK293 cells dramatically enhanced the expression levels of the endogenous HNRNPU protein in normoxia but also that hypoxia treatment increased HNRNPU localization with LANA in the nuclear compartment (Fig. 6A). These data provide strong evidence that HNRNPU associates with LANA SIM and responds to hypoxia.
The SUMOylation of HNRNPU is enhanced by KSHV latent infection and responds to hypoxia. To further determine the role of HNRNPU in KSHV latently infected cells and in response to hypoxia, we performed immunofluorescence assays to examine the localization of HNRNPU in the iSLK cell line and its derived iSLK.219 cells with KSHV latent infection under normoxia or hypoxia conditions. Consistent with the observations from overexpression with exogenous LANA, the results showed that both the expression and nuclear location of HNRNPU are enhanced by KSHV latent infection in iSLK.219 cells under normoxia conditions, and the nuclear localization of HNRNPU is further induced by hypoxic stress (Fig. 6B). In view of the fact that HNRNPU associates with LANA SIM , we attempted to examine whether HNRNPU undergoes SUMOylation in response to KSHV latent infection and hypoxia. The results from denatured immunoprecipitation (IP) assays showed that HNRNPU was dramatically SUMOylated by SUMO1 and SUMO2/3 in iSLK cells with KSHV latent infection (iSLK.219) under normoxic conditions, while the SUMO2/3-but not the SUMO1-modified form of HNRNPU was inhibited by hypoxia (Fig. 7A), indicating that SUMO2/3-modified HNRNPU induced by KSHV infection is involved in the response to hypoxia.
To further investigate the biological role of HNRNPU in the regulation of KSHV reactivation of lytic replication, we knocked down the expression of HNRNPU in iSLK.219 cells by the introduction of small interfering RNA and looked at the expression level of RTA (the master regulator of lytic replication) and virion production in the culture medium in the presence and absence of hypoxia treatment. The results showed that inhibition of HNRNPU expression did not significantly impair RTA expression or virion production in normoxia (Fig. 7B, left). In contrast, although the expression of RTA did not have a significant effect, inhibition of HNRNPU greatly increased virion production induced by hypoxia (Fig. 7B, right). These data suggest that HNRNPU is involved in the regulation of the KSHV life cycle and responds to hypoxia.

DISCUSSION
To globally address the biological role of the SIM of LANA in KSHV-mediated latent  infection, we report the systematic analysis of cellular proteins associated with the SIM of LANA in hypoxia by generating HUVECs and HEK293 cells stably expressing wildtype LANA (LANA WT ) and its SIM-deleted mutant (LANA dSIM ). Specifically, we found that molecules involved in the regulation of two major cellular pathways, which include cytoskeleton organization and DNA/RNA binding and processing pathways, are significantly targeted by LANA SIM in response to hypoxia. It is worth mentioning that although our previous studies revealed a proteomic profile of 155 LANA SIM -associated proteins when we used a recombinant protein of GST fused with LANA SIM (the region from amino acids 240 to 330 of LANA) as bait to pull down the interacting proteins from nuclear extracts of KSHV-positive PEL cells in vitro (15), it remains uncertain as to whether the LANA SIM -associated proteins play functional roles under physiological conditions in KSHV-infected cells. To further identify the proteomic profile of the LANA SIM -interacting proteins with the intact SIM of LANA, we selected two types of endothelial cells, HUVECs (one of two cell types mainly infected by KSHV) and HEK293 kidney cells (as a parallel control), to stably express wild-type LANA or its SIM-deleted mutant, followed by mass spectrometry analysis after coimmunoprecipitation. To increase the specificity and accuracy of the identified cellular proteins associated with LANA SIM , the vector-alone group along with the SIM-deleted mutant group were used as controls to rule out the nonspecific background of protein-protein interactions. Unexpectedly, we obtained two major proteomic profiles of identified LANA SIM -interacting cellular proteins from HEK293 cells and HUVECs that were significantly different in hypoxia, although there were 28 proteins identified in both types of cells. To address the potential possibility of a difference because of the two cell lines, we analyzed the cell cycle, expression level of LANA, cell viability, and proliferation of stable cell lines in hypoxia. Interestingly, we found that HUVECs stably expressing LANA dSIM were more sensitive to hypoxia than HUVECs stably expressing LANA WT . Cell apoptosis was also increased when cells were treated in hypoxia (data not shown), which could be an explanation for why cellular proteins interacting with LANA WT were increased in hypoxia while they were decreased in the LANA dSIM group. This evidence indicates that the SIM plays an important role in the response to hypoxic stress for inhibition of cell apoptosis. In contrast, among 28 LANA SIM -associated proteins in hypoxia, only 2 molecules, SERPINH1 and HNRNPU, appear in the profile of 155 LANA SIM -associated proteins using the GST-SIM recombinant protein as bait. This suggests that both SERPINH1 and HNRNPU are commonly and significantly targeted by LANA SIM in these two different screening systems. Further investigation revealed that SUMOylation of HNRNPU is greatly enhanced by KSHV latent infection, while only the SUMO2/3-and not the SUMO1-modified form of HNRNPU induced by KSHV responds to hypoxia. Knockdown of HNRNPU increased KSHV virion production induced by hypoxia, indicating that SUMO2/3-modified HNRNPU is involved in the regulation of the hypoxia response and KSHV latent-lytic life cycle control. The fact that SUMO3, as one of 28 LANA SIM -associated proteins in both HUVECs and HEK293 cells, was consistently identified, and there are higher percentages of identified proteins containing at least one SIM in the LANA SIM group than in the LANA dSIM group, further supports the results from the LANA SIM -associated screening system. To analyze the universal biological function of LANA SIM in a variety of cell lines, we chose proteins with similar changing trends and that were identical between HUVECs and HEK293 cells after the SIM was deleted in hypoxia. Among the 28 target proteins, ACTB, ACTN1, ACTN4, CAPZA1 CAPZB, DBN1, TPM1, and VASP are associated with building and maintaining the cytoskeleton. Supporting these studies, a recent report showed that SUMO is covalently attached to components of each of the four major cytoskeletal networks, including microtubule-associated proteins, septins, and intermediate filaments (18,19). For example, ACTN4 is involved in the formation of F-actin and the regulation of cell migration and tumor invasion (20,21), and DBN1 plays a role in cell migration and actin polymerization (22). In addition, some groups discovered that hypoxia treatment also induced actin cytoskeleton remodeling by regulating the binding of CAPZA1 to F-actin (23,24), which further supports our observation that the SIM of LANA participates in the regulation of cell migration and tumor invasion in hypoxia.
In summary, we have identified a proteomic profile of host proteins related to the SIM of LANA in hypoxia through coimmunoprecipitation combined with MS analysis and found that the main cellular pathways regulated by LANA SIM are different when cells are cultured in normoxia or hypoxia. Based on the analysis of the intersecting proteins of both HUVECs and HEK293 cells, we revealed that the SIM of LANA is involved in two major pathways, including cytoskeleton organization and DNA/RNA binding and processing pathways, in hypoxia, for which further experiments for protein validation and functional analysis are required.
Establishment of LANA stable expression cell lines. The pLVX-Puro vectors carrying LANA WT and LANA dSIM cDNAs were cotransfected with lentivirus packaging plasmids into HEK293T cells for 48 h to generate lentiviruses. For transduction, the packaged lentiviruses were individually added to HUVECs and HEK293 cells and centrifuged for 1 h at 1,500 Â g with 6 mg/ml Polybrene (catalog number H9268; Sigma, Shanghai, People's Republic of China). Medium was replaced after the spin, and 2 mg/ml (final concentration) of puromycin was added at 24 h posttransduction. Selection was carried out until colonies were clonal with 100% green fluorescence for transduced cells.
Protein fractionation and mass spectrometry. The sodium dodecyl sulfate (SDS)-PAGE gel lanes of YFP (vector), LANA WT , and LANA dSIM from stable HEK293 cell or HUVEC lines were individually cut into pieces using a sterile in-house cutting device. Protein standard bands served as a guide for the excision of gel slices of various molecular weight ranges. The protocols for gel slice recycling and protein digestion were performed as described previously (15,28). Extracted peptides were resuspended in 10 ml of a solution containing 5% acetonitrile and 0.1% trifluoroacetic acid and run on a NanoAcquity ultra high performance liquid coupled two-dimensional linear ion trap electrostatic Orbitrap XL hybrid mass spectrometer equipped with electron transfer dissociation (UPLC/LTQ Orbitrap XL ETD) tandem quantitative mass spectrometer (Thermo Fisher Scientific, Shanghai, People's Republic of China).
Protein identification by mass spectrometry and data analysis. We chose emPAI values (the emPAI value is a quantitative protein value that can predict protein abundance based on the number of detected peptides) to analyze the relative quantitative profiles of nonlabeled proteins (15). Proteins in the YFP group were used as a blank control, and proteins in the LANA groups with emPAI values $2 times the values of the control group were selected for analysis. The data were analyzed with GPS explorer/Analyst QS software (15) and searched with Mascot software (Matrix Science Ltd.) against the National Center for Biotechnology Information (NCBI) database. Pathway and function annotation information was downloaded from MSigDB database v5.0, and Venn diagrams were created with Venny2.1. Enrichment analysis was fulfilled between the core and extended networks and the currently available pathways and functional categories. The most significant pathways and function categories were picked based on the enrichment P values. The enrichment P value is calculated according to the