Phase separation and DAXX redistribution contribute to LANA nuclear body and KSHV genome dynamics during latency and reactivation

Liquid-liquid phase separation (LLPS) can drive formation of diverse and essential macromolecular structures, including those specified by viruses. Kaposi’s Sarcoma-Associated Herpesvirus (KSHV) genomes associate with the viral encoded Latency-Associated Nuclear Antigen (LANA) to form stable nuclear bodies (NBs) during latent infection. Here, we show that LANA-NB formation and KSHV genome conformation involves LLPS. Using LLPS disrupting solvents, we show that LANA-NBs are partially disrupted, while DAXX and PML foci are highly resistant. LLPS disruption altered the LANA-dependent KSHV chromosome conformation but did not stimulate lytic reactivation. We found that LANA-NBs undergo major morphological transformation during KSHV lytic reactivation to form LANA-associated replication compartments encompassing KSHV DNA. DAXX colocalizes with the LANA-NBs during latency but is evicted from the LANA-associated lytic replication compartments. These findings indicate the LANA-NBs are dynamic super-molecular nuclear structures that partly depend on LLPS and undergo morphological transitions corresponding to the different modes of viral replication.

Introduction chromatin domains through multivalent interactions including recognition of methylated histone H3K9me3 [27]. Transcriptional super enhancers form through the condensate properties of multivalent proteins including members of the BRD family that recognize acetylated histones [28,29]. Post-translational modifications of the heptad repeats in RNA polymerase C-terminal tail nucleate condensates that regulate RNA splicing [30]. In addition, many of these membrane-less structures depend on nucleic acid templates, such RNA dependence of paraspeckles and Cajal bodies [31,32], and repetitive ribosomal DNA for the formation of nucleoli [33]. Viral specific structures are likely to be required to protect and sequester viral genomes away from host genomes, as well as to provide sufficient concentration of components for successful viral replication [34,35] (reviewed in [36]). We have previously investigated LANA NBs and shown that these structures depend, in part, on the oligomerization of the LANA DBD and colocalize with histone H3.3 chaperone DAXX during latency [18]. Here we investigate whether LANA NBs are also dependent on LLPS and how these structures may change during the viral life cycle.

LANA nuclear bodies are sensitive to LLPS disrupting agents
Biological structures that depend on LLPS are sensitive to 1,6-Hexanediol (HD), but less sensitive to 2,5-Hexanediol [33]. To determine whether LANA-NBs were dependent on LLPS dynamics, we first treated KSHV positive pleural effusion lymphoma cell line BCLB1 with 1,6-HD assayed for dispersion of LANA-NBs (Fig 1A and 1B and S1A-S1C Fig). We determined that treatment with 3.0-3.5% 1,6-HD for at least 10 min was sufficient to disrupt coilinassociated Cajal bodies (S1 Fig), consistent with published reports for other cell lines [33]. Cajal bodies function in the assembly of ribonucleoprotein complexes, including spliceosomal snRNPs and telomerase, and their ability to assemble through liquid-liquid phase transition has been correlated with these biological activities [37]. We observed that BCBL1 and iSKL cells typically have~2 Cajal bodies and >4 LANA-NBs in the majority of cells. We found that 1,6-HD reduced the number of cells with 2 coilin NBs to <10%, while cells with >4 LANA-NBs were reduced to~30%. LANA-NBs had similar kinetics of disruption and recovery as did coilin foci in response to 1,6-HD (S1 Fig). To further characterize LANA-NBs and obtain higher quality imaging, we utilized the previously characterized adherent cell line iSLK carrying KSHV Bac16 expressing RFP-LANA [18]. We treated iSLK RFP-LANA cells with 3.5% 1,6-HD or the less disruptive control 2,5-HD for 10 min and assayed by IF for LANA-NB or coilin foci. We found that 2,5-HD treatment resulted in~41% of cells with >2 coilin foci compared to <5% in 1,6-HD treated cells (Fig 1D). Similarly, 2,5-HD treatment resulted iñ 78% of cells with >4 LANA-NBs, compared to~48% in 1,6-HD treated cells. Taken together, these findings indicate that LANA-NBs are partially sensitive to LLPS disrupting 1,6-HD, but less sensitive to 2,5-HD in a time-frame similar to coilin-associated Cajal bodies.

DAXX structures are resistant to LLPS disrupting agents
We previously showed that LANA-NBs colocalize with DAXX, but not with ATRX or PML [18]. We therefore assayed the effects of LLPS disrupting agents on DAXX and ATRX and their relative colocalization with LANA-NBs (Fig 2 and S2 Fig). Using the same iSLK RFP-LANA cells and treatment conditions that were found to disrupt coilin and partially disrupt LANA foci (Fig 1 and S1 Fig), we found that DAXX NBs that colocalize with LANA-NBs were mostly resistant, while ATRX foci that do not colocalize with LANA were highly sensitive to 1,6-HD (Fig 2A-2D). We observed similar resistance to LLPS disrupting agents by DAXX in BCBL1 cells (S2 Fig). We also found that PML-NBs, as monitored by IF to PML, were highly resistant

PLOS PATHOGENS
Dynamic changes in LANA-NBs to LLPS disrupting agents (S2C and S2D Fig). In contrast, we found that ATRX NBs were reduced from 88% in untreated to~5% in 6-HD treated iSLK RFP-LANA cells (Fig 2B-2D). These findings indicate that ATRX-NB are significantly disrupted, while DAXX-associated LANA-NBs and PML-NBs remain stable after treatment with LLPS disrupting agents. DAXX is known to co-immunoprecipitate (IP) with both LANA and ATRX. We therefore tested whether treatment with 1,6-HD altered the ability of LANA, DAXX or ATRX to coIP (Fig 2E). LANA co-IP with DAXX was detected in extracts from untreated, as well from 1,6-HD treated iSLK RFP-LANA cells. ATRX interaction with DAXX was readily detected in untreated iSLK RFP-LANA cells, but could not be detected in 1,6-HD treated iSLK RFP-LANA cells (Fig 2E). These findings indicate that DAXX interaction with LANA is not sensitive to LLPS, while both ATRX foci and interactions with DAXX are sensitive to LLPS disrupting agents.

PLOS PATHOGENS
Dynamic changes in LANA-NBs KSHV transcription using RT-PCR at various times after drug treatment. We found that 1,6-HD had a minor stabilizing effect on LANA protein and induced modest PARP1 cleavage after 1 hr of treatment as measured by Western blot (Fig 3A). In contrast,1,6-HD had no effects on coilin, RAD21, or DAXX protein levels ( Fig 3A). RT-qPCR showed no significant increase in LANA or other KSHV transcripts at any time point up to 6 hrs post treatment ( Fig 3B). Previous studies found that lytic transcription can be induced as early as 30 min after treatment with ER-stress inducing agents, such as DTT [18]. Consistently, we found that DTT treatment for 3 or 6 hrs led to >10 fold increase in lytic transcription for ORF50 and PAN RNA as demonstrated by RT-qPCR ( Fig 3B). We also assayed whether 1,6-HD affected LANA binding to KSHV TR in vivo by ChIP assay (Fig 3C). Consistent with the increase in LANA protein expression, we found a small (~1.75 fold) increase in LANA binding to TR after 1,6-HD treatment in both iSLK RFP-LANA and BCBL1 cells. These findings indicate that 1,6-HD does not lead to lytic transcript reactivation, nor the loss of LANA binding to KSHV genomes, despite the loss of LANA NB morphology in 50% of the cells.

PLOS PATHOGENS
Dynamic changes in LANA-NBs

LLPS disruption alters KSHV genome conformation
Previous studies revealed that KSHV forms at least two distinct DNA-DNA loops as measured by chromosome conformation capture (3C) [18]. Loop1 occurs between the latent (LANAp) and lytic (ORF50p) control regions [38]. Loop2 is LANA-dependent loop between TR and LANAp region [18]. We therefore tested whether treatment with LLPS disrupting agents affected either of these 3C loops in both BCLB1 and iSLK cells (Fig 4A). We found that 1,6-HD led to a 30% decrease in Loop 2, but had no effect on Loop 1 in both latency models (Fig 4B and 4C). This suggests that the LANA-dependent 3C loop mediating interaction between TR and LANAp is regulated by LLPS dynamics, while the CTCF-cohesin interactions forming loop1 are not sensitive to LLPS disrupting agents. Red arrowheads indicate previously mapped CTCF binding sites. B. BCBL1 and iSLK RFP-LANA cell lines treated for 1 hr with 3.5% 1,6-Hexanediol were assayed by 3C using anchor primer (Anchor 1) at latency control region (128,264) and acceptor primers at positions indicated on x-axis. 3C-qPCR relative to actin control is reported. C. BCBL1 and iSLK RFP-LANA cell lines treated as in panel B were assayed by 3C using anchor primer (Anchor 2) near TR (position 133872) and acceptor primers at positions indicated on x-axis. 3C-qPCR relative to actin control is reported. � p value < 0.05, two-tailed t-test. https://doi.org/10.1371/journal.ppat.1009231.g004

PLOS PATHOGENS
Dynamic changes in LANA-NBs

LANA-NBs form ring-like structures associated with KSHV DNA during lytic reactivation
LANA-NBs adopt several different nuclear morphologies in latent infection [18]. We next asked whether LANA-NBs undergo any morphological change during lytic reactivation from latency. We induced lytic cycle in iSLK RFP-LANA cells using either doxycycline (Dox) induced RTA, Dox-induced RTA + sodium butyrate (NaB), or JQ1, each of which has been shown to induce lytic cycle to a various extents [39]. We found that each of the lytic inducing agents increased the number, size, and fluorescence intensity of LANA-NBs per cell ( Fig 5A).

Dynamic changes in LANA-NBs
We also observed several distinctive morphologies, including the transformation of LANA-NBs into larger ring-like structures ( Fig 5B). These ring-like structures were most prominent in Dox + NaB suggesting they correspond to more robust lytic reactivation. We next assayed LANA-NBs in combination with KSHV DNA FISH to determine the spatial relationship between LANA NBs and viral genomes during latency and reactivation ( Fig 5B). We found that KSHV DNA colocalized with LANA NBs during latency and after lytic reactivation. In some cases, KSHV DNA colocalized with LANA along the circumference of the ring-like structures during lytic reactivation, while in other instances KSHV DNA appeared to increase within the center of these LANA-rings ( Fig 5B).

Spatial reorganization of KSHV DNA with DAXX and LANA-NBs during lytic reactivation
To investigate the relationship between DAXX and LANA-NB remodeling during lytic reactivation, we next assayed KSHV-DNA FISH in combination with DAXX-IF (Fig 6). In latently

PLOS PATHOGENS
Dynamic changes in LANA-NBs infected cells, we observed strong colocalization of DAXX with KSHV DNA (Fig 6A). This is consistent with previous studies showing strong colocalization of DAXX with LANA-NBs. However, during lytic reactivation, KSHV DNA associated with large ring-like nuclear body morphology, while DAXX was dispersed throughout the nucleus (Fig 6B). These findings suggest that DAXX is evicted from LANA-NBs and dissociated from KSHV DNA during reactivation.
We next used high-resolution confocal microscopy and 3D reconstruction to better analyze the spatial relationship between KSHV DNA and LANA or KSHV DNA and DAXX during latency and lytic reactivation (Fig 6C-6F and S1-S8 Movies). As expected, LANA colocalizes with KSHV DNA in latently infected cells, with most images indicating that LANA sits on compact surfaces of KSHV DNA (Fig 6C). This spatial relationship is altered during reactivation where both larger and more intense bodies of LANA and DNA appear to be interweaving, rather than completely overlapping (Fig 6D). DAXX also colocalizes strongly with KSHV DNA during latency, but appears to be extended over larger portions of the KSHV DNA surface than was observed for LANA ( Fig 6E). In contrast, during lytic reactivation DAXX appears excluded from the larger KSHV DNA nuclear bodies and ring-like structures ( Fig 6F). These findings confirm that KSHV reactivation leads to a change in the spatial re-organization of LANA and KSHV genomic DNA, and the redistribution of DAXX from LANA-NBs during lytic reactivation.
We next tested whether LANA ring-like structure colocalize with KSHV lytic cycle replication proteins, such as K8 or ORF45 (S3 and S4 Figs). While K8 is a b-zip protein implicated in DNA replication, it is not known if it colocalizes with lytic replication compartments. After 24 h induction with Dox + NaB, we found most of the K8 and ORF45 as pan-nuclear, and very few cells with punctate foci. Using K8 or ORF45 as markers for lytic cycle induction, we found that~70% of the K8 or ORF45 lytic cells had LANA ring-like structures (S3 Fig) indicating that most, but not all lytic cells form these LANA superstructures. We observed rare cells with K8 punctate foci that partly colocalized with LANA ring-like structures (S4 Fig), suggesting that colocalization may occur at earlier stages of lytic cycle replication. We also tested whether LANA ring-like structures were sensitive to 1,6-HD using treatment conditions identical to those used for latent cycle LANA-NBs. We analyzed the results in two ways. We first quantified the % of cells with DAXX or LANA-NBs (Fig 7A and 7B). This revealed that 1,6-HD reduced DAXX foci by~5 fold, while having less than a 2-fold reduction of LANA-NBs ( Fig  7B). We also counted more specifically the LANA ring-like structures. These structures were found to be reduced by~3 fold after 1,6-HD treatment (Fig 7C and 7D). Taken together, these findings suggest that both LANA-NBs and lytic induced LANA ring-like superstructures are partially dependent on LLPS.
The function of DAXX in LANA-NB formation and KSHV regulation is not clear. Others have found that DAXX depletion has little effect on KSHV gene expression or lytic replication in BCBL1 cells [40]. We found that partial depletion of DAXX in KSHV iSLK cells led to a decrease in KSHV LANA and ORF45 protein (Fig 8A), and corresponding decreases in viral gene transcripts (Fig 8B). We also observed a reduction in EZH2 protein (Fig 8A). The effect of DAXX depletion on the viral chromatin was assayed by ChIP assay (Fig 8C). DAXX depletion reduced LANA binding and altered histone modifications at several key viral control regions including TR, LANAp and ORF50p (Fig 8C). Consistent with the decrease in viral transcription, we observed a loss of H3K4me3 at the LANAp and an increase of H3K9me3 and H3K27me3 at LANAp and ORF50p, respectively (Fig 8C). ChIP assay also revealed an increase in H3K4me3 at the TR. We next examined the effects of DAXX shRNA depletion by IF ( Fig  8D). DAXX depletion had no obvious effect on LANA-NBs morphology, although average fluorescence intensity of LANA-NBs was reduced. DAXX-depletion also increased the

PLOS PATHOGENS
Dynamic changes in LANA-NBs colocalization of EZH2 and H3K27me3 foci with LANA-NBs (Fig 8D-8F). Taken together, these findings indicate that DAXX associates with LANA-NBs during latency and contributes to epigenetic programming of the latent KSHV genome.

Discussion
Macromolecular self-assembly and membrane-less compartments provide vital functional utility for many subcellular biological processes [41][42][43]. Viruses are known to induce formation of a diverse intracellular structures and compartments required to complete different aspects of their life cycle within the host cell [44]. Compartments for viral DNA replication [45,46], capsid assembly [47], RNA processing [48], and protein quality control [49] have been described for numerous viruses. Here, we further characterize the dynamic properties of the iSLK RFP-LANA cells treated as in panel A were assayed by RT-qPCR at Day 6 for KSHV latency-associated genes LANA, ORF71, ORF73 or lytic-associated genes ORF50, ORF45, PAN. The bar graph represents means ± s.d. from three independent DAXX depletion experiments. � p value <0.05, �� p <0.01, ��� p < 0.005, ns not significant using two-tailed t-test. C. iSLK RFP-LANA cells infected with shCon or shDAXX for 6 days were assayed by ChIP using antibodies against IgG, LANA, histone H3K4me3, H3K9me3, H3K27me3, and H3K27Ac, and normalized to total H3 ChIP signal for each loci, as indicated. ChIPed DNA was examined by qPCR using primers for KSHV TR (top panel), LANA promoter (middle panel), or ORF50 promoter (bottom panel). Bar graph represents the average value of percentage of input relative to H3 for each ChIP from three independent PCR reactions (mean ± s.d.). � p value <0.05, �� p <0.01, ��� p < 0.005, ns not significant using two-tailed t-test. D. iSLK RFP-LANA cells infected with shCon or shDAXX for 6 days were assayed by IF for RFP-LANA (red) and either DAXX (top), EZH2 (middle), or H3K27me3

PLOS PATHOGENS
Dynamic changes in LANA-NBs LANA-NBs that have been implicated in episome maintenance during KSHV latent infection. While we have previously shown that these LANA-NBs remain organized during latent cycle cell division and undergo orchestrated distribution to daughter cells during mitotic cell division [18], we now show that these structures are partly dependent on phase separation dynamics and undergo morphological transition during the lytic reactivation cycle. We show that LLPS disrupting solvents reduce LANA-NB formation as well as LANA-dependent chromosome conformation of the KSHV genome. Finally, we show that DAXX, but not ATRX, associates with LANA-NBs during latency, and is dispersed from LANA-NBs during reactivation. Taken together, these findings indicate the LANA-NBs are dynamic super-molecular nuclear structures that undergo morphological transitions corresponding the functional state of the viral genome.

LLPS contributes to LANA-NB formation
LLPS involves the condensation of a critical mass of multi-valent components [28,50]. In many cases, these structures are dependent or facilitated by nucleic acid templates, including RNA and DNA [31,51]. LANA-NBs are known to form only when LANA is expressed at sufficient concentration in combination with KSHV terminal repeat DNA [11,52]. The cooperative binding of LANA to 3 adjacent sites in each of the 10-40 tandem copies of the TR is likely to provide sufficient local concentration of LANA to drive phase change reactions. Proteins that form LLPS typically have low-complexity (LC) domains that are inherently unstructured but can adapt conformations that enable a variety of different macromolecular shapes to accommodate diverse ligand interactions. LANA binds to the viral template DNA with its highly structured C-terminal DNA binding domain that is inherently prone to form higher-ordered oligomeric structures, including pentamers, decamers and continuous helical spirals [10,18,20]. In addition to its DNA binding domain, LANA also has low complexity (LC) domains involved in various interactions of functional significance. The N-terminal basic domain can bind directly to an acidic patch in the nucleosome formed by histones H2A and H2B [21]. The proline rich domain (aa194-273) contains sites for protein destabilization, SUMO-2 interaction and KAP1 repressor binding [53,54]. The acidic rich domain of LANA can interact directly with DAXX [55]. The acidic domain of LANA has been implicated in many other interactions important for transcription regulation, including binding to CBP [56]. We propose that LANA-NBs form through a combination of rigid and flexible structures, including a KSHV TR DNA template driven oligomerization of the rigid DBD and an LLPS driven multivalency of the N-terminal LC domains. This combination provides structural flexibility required for viral transcription and replication in both latent and lytic stages of the viral life cycle (Fig 9).
One major limitation of these studies, along with many others in LLPS investigation, is the reliance on a limited number of organic molecules like 1,6-HD to evaluate LLPS. These organic molecules can have pleiotropic effects on other cell structures and processes. However, the selective effects of 1,6-HD relative to 2,5-HD supports the model that LLPS contributes to the formation and dynamic properties of LANA-NBs in both latent and lytic cycle structures.

KSHV replication compartments
We found that LANA-NBs undergo morphological change during latent and lytic cycles and remain closely associated with viral DNA. During lytic reactivation, LANA-NBs formed large ring-like structures with LANA enriched along the perimeter and KSHV DNA localized mostly within the rings. Confocal imaging suggests that the lytic replication compartments are two-dimensional rings and not cross sections of globular spheres. KSHV replication

PLOS PATHOGENS
Dynamic changes in LANA-NBs compartments have been described in other studies [57][58][59][60]. Replication compartments similar to what we observed have been described and shown to colocalize with DNA damage repair factors associate with NHEJ, including DNA-PK, ATM, and MRN components [57]. KSHV RNAs, especially the non-coding RNA PAN, has been shown to localize to KSHV replication compartments, suggesting that RNA also contribute to the formation of these super-molecular structures [48]. KSHV lytic replication factors by themselves can self-assemble into subnuclear compartments that colocalized near PML-NBs, but these did not depend on LANA or viral DNA templates [61]. In contrast, we found that KSHV DNA and LANA did not colocalize with PML-NBs. Although DAXX is typically found associated with PML-NBs, it associates with LANA-NBs separate from PML during KSHV latency. This colocalization is disrupted during lytic reactivation, when LANA tended to form large ring-like structures that colocalized with expansion of KSHV DNA. We observed rare punctate colocalization of KSHV lytic replication protein K8 with LANA rings, suggesting that these structures may be distinct from classical replication compartments described for other viruses. LANA has also been reported to accumulated in cytoplasmic foci during lytic replication cycle [60], but our experimental

PLOS PATHOGENS
Dynamic changes in LANA-NBs system using RFP-LANA may only observe full length LANA in the nucleus. While LANA is well known for its role in KSHV latent cycle replication, our new findings suggest that LANA also contributes to the structural organization of KSHV lytic cycle DNA replication. We propose that LANA has a fundamental architectural role in both latent and early phase of the lytic replication cycles of KSHV.

DAXX eviction from LANA-NBs and KSHV DNA during reactivation
KSHV-NBs colocalize with DAXX but not with PML, or other components of the PML-NBs [18]. We show here that ATRX is not associated with LANA-NBs, and that it behaves differently than DAXX in response to LLPS disrupting agents. DAXX is known to function in association with ATRX as a histone H3.3 chaperone [62], but is also known to interact with other proteins, including LANA [55]. Exclusion of DAXX from lytic replication compartments may reflect the block to histone loading on newly replicating viral DNA [36]. The interaction interfaces of LANA and DAXX are both low complexity domains suggesting they may utilize LLPS dynamic interactions. The function of DAXX in LANA-NB formation and KSHV regulation is not clear. While DAXX depletion led to changes in KSHV gene expression and histone modifications in iSLK cells, we and others have also found that DAXX depletion has only a small effect on KSHV gene expression or lytic replication in BCLB1 cells [40]. Furthermore, DAXX depletion has no obvious effect on LANA-NB morphology. The discrepancy of effects of DAXX depletion on KSHV gene expression and chromatin regulation in iSLK and BCBL1 cells remain elusive. DAXX can also interact with SUMO-modified proteins that may regulate protein trafficking and folding [63]. While an essential role of DAXX in LANA-NB formation is not known, it is likely that DAXX contributes to the LLPS dynamics of LANA and morphological change of LANA-NBs during the different stages of KSHV life cycle.

Role of LLPS in viral assemblies
LLPS and biomolecular condensate forces are likely to drive formation of many biological structures, especially rapidly transitioning viral compartments [36]. Replication compartments for several viruses, including measles and VSV, are known to be LLPS dependent [64,65]. Super-enhancers induced by viral transcription factors, such as EBV EBNALP were found to be LLPS dependent [66]. Numerous viral-specific structures are also suspected of being LLPS dependent [26,36]. CMV assembly compartments [47], viral induced chaperoneenriched (VICE) domains that serve as nuclear protein quality control centers [49], AdV replication compartments undergo morphological changes during replication and viral maturation [44], and HSV replication compartments that can fuse to form larger superstructures [45,46], are all likely to involve LLPS. However, a recent study found that HSV replication compartments were mostly insensitive to LLPS disrupting agents [67]. We found that only a small percentage of LANA NBs are dispersed by LLPS disrupting agents, indicating that physical forces other than LLPS contribute to the formation and stability of LANA-NBs. LANA-NBs are highly stable during latency, and rather than disperse, are remodeled and repurposed into replication compartments during lytic reactivation. It is likely that LANA-NBs utilize combinations of interaction forces, including LLPS-dependent, to form malleable gel-like polymers [68]. Since LANA-NB are "templated" by repetitive DNA [33], they may have properties that distinguish them from lower complexity structures. In conclusion, LANA-NBs have dynamic structural properties involving phase separation condensates of low-complexity domain interactions in addition to stable solid-state components providing the necessary plasticity for genome conformational changes associated with diverse transcriptional responses and

PLOS PATHOGENS
Dynamic changes in LANA-NBs different modes of viral DNA replication. Such modular structures provide protection and adaptability necessary for viral persistence and transmission.

Indirect immunofluorescence (IF) assay
For adherent cells (iSLK), 1x10 5 (80% confluence) cells were plated on 12 mm diameter glass coverslips in 24-well plate. For suspension cells (BCBL1) cells were collected, washed with 1xPBS and plated on the 12 mm glass coverslips in amount 1x10 5 using the Cytospin (SHAN-DON) at 1000 rpm for 5 minutes, on high acceleration. Cells mounted on slides were washed with 1XPBS and fixed for 15 min with 2% paraformaldehyde (Electron Microscopy Sciences) in 1XPBS, washed twice with 1XPBS and permeabilized with 0.3% TritonX-100 (Sigma) in PBS. All procedures were performed at room temperature. After washing with PBS, slides were incubated in blocking solution (0.2% fish gelatin, 0.5% BSA in 1XPBS) for 30 min. Primary antibodies was diluted in blocking solution and applied on cells for 1h followed with 1xPBS washing. Cells were further incubated with fluorescence-conjugated secondary antibodies in blocking solution for 1h, counterstained with DAPI and mounted in SlowFade Gold mounting solution (LifeTechnologies). Images were acquired with a Nikon 80i Upright Microscope (Nikon Instruments) at 60X or 100X lens using Nikon ImagePro Plus software (Media Cybernetics) and Adobe Photoshop CS6 for image capture and processing.

PLOS PATHOGENS
Dynamic changes in LANA-NBs then stained with DAPI. Slides were dehydrated with ethanol wash series, air dried and mounted in SlowFade Gold mounting solution. Images were taken at 60X or 100X zoom using Nikon 80i Upright Microscope and processed with Adobe Photoshop CS6.

Confocal microscopy and image processing
High resolution, confocal images of fixed cells were captured using a Leica TCS SP8 WLL scanning laser confocal microscope and Leica LAS-X software (Leica Microsystems, Inc., Buffalo Grove, IL). Image post-processing included importing into Huygens software for deconvolution (Scientific Volume Imaging, Laapersveld, Hilversum, The Netherlands) followed by 2D maximum projection or 3D reconstruction, iso-surface application and video rendition in LAS-X.
Fixed cell preparations were acquired using a 63X/1.40 oil objective, 6X zoom and a pinhole of 1 AU, with 11-15 z-steps through 3-3.5 um stacks, resulting in a voxel size of 60 x 60 x 299nm. Cells were labeled with DAPI (nuclei), Alexa488 (KSHV) and RFP (LANA or DAXX) and acquired with HyD detectors in sequence to maximize signal and minimize cross-talk. Foci and colocalizations were determined and quantified using Nikon NIS Elements AR software, version 5.02, using the Spot Detection Tool. Using the Nikon NIS-Elements AR software, version 5.02 with the Bright Spot Detection Tool, the size of the smaller single LANA dot = 0.79μm, the bigger LANA structure = 2.87μm. Images acquired with the Nikon Upright Microscope utilized the Image-Pro Plus 7.1 software with saturation and exposure time tools to limit signal saturation. ImageJ 1.49 processing program was used to count only NBs that were not overexposed.
Colocalization was determined and quantified with Nikon NIS-Elements AR software, version 5.02 using the Bright Spot Detection Tool. For each detected nuclear body, a 1μm circular binary outline is stored. For the detected objects, the growing feature adjusted to match the object's shape. The same procedure was used for either LANA and DAXX, or LANA and EZH2 NBs colocalizations.
For 3D reconstructions, original confocal files were captured using a Leica TCS SP8 WLL scanning laser confocal microscope (Leica Microsystems, Inc., Buffalo Grove, Ill.). Extended focus Z-stacks of each cell were acquired according to optimized settings based on Nyquist criteria and the resulting files were processed using Huygens deconvolution software (Scientific Volume Imaging, B.V., Hilversum, Netherlands). 3D reconstructions and video animations were then created using the Leica LAS-X 3D software module.

Antibodies
The following antibodies were used for immunofluorescence, Co-IP, and Western blotting studies: rat polyclonal anti-HHV8 or anti-LANA

RT-qPCR
Total RNA was isolated from BCBL1 cells treated with either 3.5% 1, 6-HD or DTT at indicated times using Trizol reagents (Life Technologies) as manufacturer's instructions. For RT in shDAXX depleted cells, total RNA was isolated from KSHV iSLK cells infected with lentiviruses encoding shDAXX or shControl at day 6 post-infection and puromycin selection. RT reaction was performed by using Superscript IV Reverse Transcriptase (Invitrogen) as specified by the manufacturer. qPCR was performed using a SYBR green probe with ABI Prism 7900 Sequence Detection System (Applied Biosystems). Relative RT-PCR was determined using ΔΔCT methods relative to control gene Actin or Gapdh, as indicated. Primers used in RT-qPCR are listed in [18].

Co-Immunoprecipitation (Co-IP)
Co-IP assays were performed essentially as described previously [72]. About 5 x 10 6 KSHV iSLK or BCBL1 cells mock-treated or treated with 3.5% 1, 6-HD for 1 hr were lysed in 1ml cold lysis buffer (20 mM Tris, pH 7.5, 137 mM KCl, 1 mM EDTA, 1.5 mM MgCl 2 , 10% glycerol, 1% Triton X-100) fresh supplemented 1 mM DTT and protease inhibitor cocktails (Sigma) for 30 mins. Cell lysates were collected by centrifugation at 14000 rpm for 10 mins at 4 o C, and used for IP by antibodies to LANA, DAXX, and control IgG at 4 o C overnight with rotation. The immunoprecipitates were collected by sepharose A + G beads and washed three times 10 mins each with BC200 (20 mM Tris, pH 7.5, 0.2 mM EDTA, 10% glycerol, 200 mM KCl, 10 mM β-mercaptoethanol, 1 x protease inhibitor cocktails) and once for 20 mins with BC100 at 4 o C. The immune-complex was eluted from beads with 2x Laemmli buffer and subject to Western blot analysis.

PLOS PATHOGENS
Dynamic changes in LANA-NBs Supporting information S1 Fig. Dose titration and time course for treatment and recovery with 1,6-Hexanediol. A. BCBL1 cells were untreated or treated with 1.0, 2.0, or 3.0% 1,6-HD for 15 min and imaged by IF for Coilin(green) or LANA (red) or merged with DAPI (blue). Scale bar = 10 μm. B. BCBL1 cells were treated with 3.5% 1,6-Hexanediol for either 10, 20, or 30 min and imaged by IF as in panel A. Scale bar = 10 μm. C. BCBL1 cells were treated for 30 min with 3.5% 1,6 Hexanediol (top panel) followed by recovery in fresh media for 10, 20, or 30 min and imaged by IF as in panel A. Scale bar = 10 μm. D. iSLK cells were treated with 3.5% 1,6-HD for 10, 20, or 30 min and imaged by IF as in panel A. Scale bar = 10 μm.