Kaposi Sarcoma Herpesvirus (KSHV) Latency-Associated Nuclear Antigen (LANA) recruits components of the MRN (Mre11-Rad50-NBS1) repair complex to modulate an innate immune signaling pathway and viral latency

Kaposi Sarcoma Herpesvirus (KSHV), a γ2-herpesvirus and class 1 carcinogen, is responsible for at least three human malignancies: Kaposi Sarcoma (KS), Primary Effusion Lymphoma (PEL) and Multicentric Castleman’s Disease (MCD). Its major nuclear latency protein, LANA, is indispensable for the maintenance and replication of latent viral DNA in infected cells. Although LANA is mainly a nuclear protein, cytoplasmic isoforms of LANA exist and can act as antagonists of the cytoplasmic DNA sensor, cGAS. Here, we show that cytosolic LANA also recruits members of the MRN (Mre11-Rad50-NBS1) repair complex in the cytosol and thereby inhibits their recently reported role in the sensing of cytoplasmic DNA and activation of the NF-κB pathway. Inhibition of NF-κB activation by cytoplasmic LANA is accompanied by increased lytic replication in KSHV-infected cells, suggesting that MRN-dependent NF-κB activation contributes to KSHV latency. Cytoplasmic LANA may therefore support the activation of KSHV lytic replication in part by counteracting the activation of NF-κB in response to cytoplasmic DNA. This would complement the recently described role of cytoplasmic LANA in blocking an interferon response triggered by cGAS and thereby promoting lytic reactivation. Our findings highlight a second point at which cytoplasmic LANA interferes with the innate immune response, as well as the importance of the recently discovered role of cytoplasmic MRN complex members as innate sensors of cytoplasmic DNA for the control of KSHV replication.


Introduction
the NF-κB pathway [42]. We show here that cytoplasmic LANA isoforms recruit Rad50 and Mre11 in the cytosol and thereby interfere with the activation of the NF-κB cascade induced by transfected DNA, as well as KSHV reactivation from latency. These observations point to yet another antiviral mechanism inhibited by cytosolic LANA isoforms, and highlight the importance of the sensing of cytosolic DNA by the MRN complex in the context of innate immunity against viral infection.
We confirmed the interaction of LANA with Rad50, Mre11 and NBS1 in the PEL-derived cell lines BC3 and BCBL-1, as well as in BrK.219 (a BJAB cell line stably infected with a recombinant KSHV virus [51]) by co-immunoprecipitation with anti-LANA-antibody-coupled beads and immunoblotting for Rad50, Mre11 and NBS1 (see Fig 1A and 1B for co-immunoprecipitation from BC3 cells, see S1A and S1B Fig for co-immunoprecipitation from BCBL-1 and BrK.219 cells). Before immunoprecipitation cell lysates were incubated with benzonase to digest nucleic acids and avoid DNA-mediated interactions. The interaction between LANA and Rad50 could also be shown by immuno-precipitating LANA with anti-Rad50-antibody-coupled beads and checking for LANA binding by immunoblotting ( Fig 1B and S1 Fig). Both assays show that LANA interacts with the MRN complex in latently KSHV infected B cells. Interestingly, as indicated by an arrowhead in Fig 1B, smaller LANA isoforms were preferentially immunoprecipitated by Rad50 compared to those immunoprecipitated with a LANA antibody ( Fig 1B). Subsequently, we investigated which region of LANA ( Fig 1C) is responsible for the interaction with the MRN complex by performing GST-pull down assays with the N-(aa1-312) and C-(aa931-1162) terminal domains of LANA fused to GST (Fig 1D). HEK293 cell lysates were incubated with GST-fused LANA domains and the interaction with endogenous MRN complex proteins was analyzed by SDS PAGE and immunoblotting. The results suggest that all three MRN complex components bind to the C-terminal domain of LANA ( Fig 1D). We next attempted to map the interaction site in LANA more closely by using GST-fused fragments of the LANA C-terminal domain. The results (S2 Fig) suggested that this interaction may involve multiple contact points in the LANA C-terminal region, in particular within the LANA domain binding to viral DNA (S2 Fig), the structure of which has recently been solved [19,20,22].
The NF-κB cascade is considered to play an essential role in the maintenance of KSHV latency and also in the pathogenesis of KSHV-related diseases [52][53][54][55][56]. Prompted by the observation that Rad50 is associated with smaller isoforms of LANA (Fig 1B), which are known to occur in the cytoplasm [25,27], we investigated whether cytoplasmic forms of LANA interact with Rad50 and Mre11. To that end we performed co-immunoprecipitation assays, in which nuclear and cytosolic fractions from BCBL-1 cells were separated and incubated with anti-LANA-beads or IgG-control beads. We found that LANA recruits Rad50 and Mre11 mainly in the cytoplasm (Fig 2A). As a control, we also probed LANA immuno-precipitates with an antibody to Brd4, a nuclear protein known to be associated with LANA [19,24,57]. As expected, LANA and Brd4 interact in the nucleus, indicating that the buffer conditions in the nuclear extracts used for this experiment did not interfere with the interaction of LANA with its nuclear binding partners (Fig 2A). This experiment also showed that lower molecular weight forms of LANA are found predominantly in the cytoplasm (Fig 2A). To confirm these findings, we performed additional co-immunoprecipitation assays using HEK293 cells transiently transfected with constructs expressing full-length LANA or LANA ΔN mutants (Δ161 and Δ282), which lack the NLS and are therefore mainly located in the cytoplasm [25,27]. Benzonase-treated cell lysates were incubated with anti-LANA-antibody-coupled beads and the interaction with endogenous Rad50 was analyzed by immunoblotting ( Fig 2B). Our results confirm that Rad50 can be recruited by both full-length LANA as well as cytosolic LANA ΔN isoforms ( Fig 2B). Our observation that cytoplasmic LANA variants recruit Rad50/Mre11 in the cytoplasm could suggest that cytoplasmic LANA might modulate the recently described CARD9-dependent activation of NF-κB, triggered as a result of cytoplasmic DNA sensed by Rad50 [42]. Since this pathway was shown to operate in myeloid cells [42], we stably infected a human leukemia monocytic cell line (THP-1) with a recombinant KSHV virus (TrK.219, see Materials and methods) and used it to confirm the interaction between LANA and the Rad50-Mre11-CARD9 DNA sensor complex (Fig 2C). Cells were lysed and incubated first with benzonase, then with anti-LANA-coupled (or IgG-coupled) beads. The interactions were analyzed by immunoblotting. Our results (Fig 2C) show that LANA recruits all the cellular proteins (Rad50, Mre11 and CARD9) recently shown to be involved in the sensing of cytosolic viral DNA and the downstream activation of the canonical NF-κB pathway [42].
A cytoplasmic form of LANA is involved in the modulation of the canonical NF-κB pathway In order to determine whether cytosolic LANA isoforms can modulate the activation of NF-κB triggered by cytosolic DNA, we transiently transfected HeLa cells (

Silencing of Mre11 and inhibition of the NF-κB pathway by cytoplasmic LANA promotes KSHV lytic reactivation
Latently KSHV-infected HeLa.CNX cells (HeLa.CNX.rKSHV) were treated with a recombinant baculovirus expressing the regulator of the lytic replication cycle, RTA [58], and sodium butyrate [27,32,53,59,60] to induce the lytic phase and thereby confirm that KSHV could be reactivated in this cell line (Fig 4Ai and 4Aii). Following the treatment with 20% RTA (vol/vol, see Materials and methods) and 1.5 mM sodium butyrate for 24 hours, the expression of the RFP lytic reporter in the recombinant KSHV.219 virus [51,58] used for these experiments was switched on (Fig 4Ai) and the early KSHV protein K-bZIP and the Orf45-encoded tegument protein were expressed (Fig 4Aii). HeLa.CNX.rKSHV cells have much higher levels of phosphorylated p65/RelA than uninfected HeLa.CNX cells (Figs 4Bi, 4Bii and 5A), in line with the known ability of several latent KSHV proteins such as vFLIP and LANA to activate the NF-κB pathway [52,53,61]. To assess the role of the MRN complex in NF-κB activation and in KSHV lytic reactivation, we inhibited Mre11 expression by siRNA transfection in HeLa.CNX.rKSHV In these cells Mre11 silencing triggers KSHV lytic reactivation, as indicated by an increase in K-bZIP levels and a reduction in the levels of phosphorylated p65/RelA (Fig 4Bii). We next explored if this contribution of Mre11 to the maintenance of KSHV latency also applied to other KSHV-infected cell lineages. Similar to the results obtained in KSHV-infected HeLa cells, we found that silencing of Mre11 in the PEL cell line BCBL-1 as well as in the KSHV-infected THP-1 cell line TrK.219, resulted in KSHV reactivation from latency, as indicated by increased levels of, respectively, K-bZIP or ORF45, along with a decrease in p65/RelA phosphorylation (Fig 4C and 4D). Together, these results indicate that in these KSHV-infected cells Mre11 contributes to the activation of the NF-κB pathway that promotes KSHV latency [14,56]. We could not achieve an efficient silencing of Rad50 in PEL cells (or any other latently KSHV-infected cell lines), and therefore we were not able to assess if Rad50 contributes to the inhibition of the lytic cycle in a way similar to Mre11.
To explore if cytoplasmic LANA could modulate NF-κB via Mre11 and thereby affect lytic reactivation, infected and uninfected HeLa.CNX cells were transfected with LANA Δ161 or the empty vector and p65/RelA phosphorylation was analyzed by immunoblotting ( Fig 5A). Our results show that LANA Δ161 overexpression reduces p65/RelA phosphorylation level in HeLa.CNX.rKSHV cells (Fig 5A). Furthermore, HeLa cells were treated with low amounts of RTA (5% vol/vol, Fig 5B) to induce the lytic reactivation only at a minimal level and were additionally transfected with LANA Δ161 or the empty vector. Our results show that the LANA Δ161 overexpression supports the lytic reactivation in HeLa.CNX.rKSHV cells induced by low levels of RTA, as highlighted by increased levels of K-bZIP expression (Fig 5B). In addition, levels of phosphorylated p65/RelA were reduced following transfection of LANA Δ161 and upon lytic reactivation indicating an antagonistic role of truncated LANA for canonical NF-κB activation (Fig 5A and 5B). In addition, the co-expression of Mre11 together with Δ161 LANA counteracts the Δ161 LANA-mediated downmodulation of p-p65 levels (S5 Fig). To explore the role of NH 2 -terminally truncated cytoplasmic LANA variants further, we compared the ability of full-length LANA, LANA Δ161 and LANA Δ282 to activate an NF-κB dependent reporter vector in HEK293 cells ( Fig 5C). As previously reported [61], full-length LANA was found to activate NF-κB-dependent transcription (Fig 5C). In contrast, LANA Δ161 and LANA Δ282 failed to do so ( Fig 5C). However, when we explored the ability of LANA Δ161 to modulate the activation of the NF-κB pathway by the potent NF-κB activator and IKKγ ligand vFLIP [53,[62][63][64][65], we found that LANA Δ161 could inhibit vFLIP-induced NF-κB activation in a dose-dependent manner, while full-length LANA could not (Fig 5D). Taken together, our results suggest that cytoplasmic forms of LANA may target Rad50 and Mre11, and thereby antagonize the activation of NF-κB and NF-κB-dependent suppression of the KSHV lytic cycle (Fig 6).

Discussion
An involvement of some DDR proteins in the innate immune response is increasingly appreciated [42][43][44]. This highlights the similarities between the recognition of host DNA damage for subsequent repair, and of foreign DNA for the purpose of triggering an innate immune response leading to the activation of type I interferon and NF-κB-dependent pathways. In particular, components of the MRN DSBs repair complex, Rad50 and Mre11, have recently been shown to sense cytoplasmic "foreign" DNA and to activate the NF-κB pathway in a CARD9dependent manner [42].
In the present study, we found that LANA recruits Rad50 and Mre11 mostly in the cytosol of naturally KSHV-infected B cells (Fig 2A), that a cytoplasmic form of LANA may antagonize the activation of NF-κB induced by transfected DNA (Fig 3) or vFLIP ( Fig 5C) and that silencing of Mre11 promotes KSHV lytic replication in parallel to reduced NF-κB p65 phosphorylation (Fig 4Bii). These results are in line with the newly described function of DDR proteins in the context of cytosolic DNA sensing and inflammasome response [42]. In contrast to full-length nuclear LANA, which is found in the characteristic nuclear speckles [8,9], cytoplasmic LANA shows a diffuse distribution [25,27]. This absence of any cytosolic Cytoplasmic KSHV LANA isoforms recruit and antagonize cellular DNA sensor proteins cGAS as well as the Rad50-Mre11-CARD9 complex to inhibit innate immunity responses (IFN-β and NF-κB) and support KSHV lytic reactivation from latency. During KSHV lytic reactivation, free viral DNA in the cytosol is detected by host DNA sensors, such as cGAS and the Rad50/Mre11/CARD9 complex. The cGAS-STING cascade leads to IFN-β production, whereas the Rad50/Mre11/CARD9 complex is responsible for NF-κB cascade activation. A cross-talk between these two pathways may also be possible as indicated by the dashed arrows (Fig 6). Triggering of the Rad50/Mre11/CARD9 complex leads to the activation and nuclear accumulation of NF-κB p65 and the subsequent production of chemokines and cytokines. These events would interfere with the efficient KSHV lytic replication and therefore KSHV LANA ΔN isoforms block these signalling cascades by recruiting and inhibiting the upstream activators (cGAS as well as Rad50/Mre11).
https://doi.org/10.1371/journal.ppat.1006335.g006 LANA-containing structure prevented us from showing a LANA-Mre11/Rad50/CARD9 colocalization in the cytoplasm of infected cells and we therefore had to rely on co-immunprecipitation experiments from cytosolic fractions as shown in Fig 2A. We observed the interaction of cytoplasmic LANA with Mre11 and Rad50 in cells without a detectable CARD9 expression (e.g. BCBL-1, BJAB.rKSHV) and therefore believe that it is unlikely that CARD9 is responsible for bridging Mre11 and Rad50 to cytoplasmic LANA. However, we cannot formally exclude this possibility.
We have previously reported that cytoplasmic forms of LANA can promote lytic reactivation by antagonizing another cytoplasmic DNA sensor, cGAS [27]. To discriminate between the effect of cGAS-dependent interferon induction and MRN-dependent NF-κB activation on KSHV latency or reactivation, we took advantage of the fact that the HeLa.CNX subline appears to be deficient for cGAS-dependent IRF3 phosphorylation (S3 Fig). In this somewhat artificial experimental setting, we can therefore demonstrate that cytoplasmic LANA isoforms can promote KSHV reactivation by repressing NF-κB activation.
Taken together, our observations therefore suggest that cytoplasmic forms of LANA antagonize not only cGAS-dependent type I interferon responses but also the Rad50-Mre11-CARD9-dependent activation of NF-κB pathway in response to cytoplasmic DNA (Fig 6), which is present during herpesviral lytic replication [27,42,66,67]. The fact that cytoplasmic LANA appears able to neutralize both these pathways testifies to their importance in restricting "lytic", productive, KSHV replication. The NF-κB pathway has previously been shown to be required for maintaining the latency of γ2-herpesviruses [14,56], and the KSHV vFLIP protein, known to activate both NF-κB and the expression of interferon-dependent cellular genes, also contributes to the maintenance of KSHV latency [53,55,[68][69][70][71]. This is supported by the observation shown in Figs 4Bi and 5A that KSHV-infected HeLa cells display higher levels of NF-κB p65 phosphorylation than uninfected controls. It is thus conceivable that KSHV needs to counteract both these restrictive pathways to successfully reactivate from latency. This may also be necessary as a cross-talk between these two pathways (cGAS-STING activating NF-κB and vice versa) may be possible [44]. Cytoplasmic isoforms of LANA, which lack the NLS-containing N-terminal region, have been shown to be more strongly expressed during lytic reactivation [27] and may result from the use of alternative in-frame translational start codons [25] or the cleavage of an N-terminal LANA fragment by Caspase 3 [72]. Together with our previous report [27], our recent findings may therefore indicate a role for cytoplasmic LANA isoforms as viral antagonists of the innate immune response. Furthermore, our observations ( Fig  5C and 5D) indicate that cytoplasmic LANA isoforms may act as antagonists of full-length, nuclear LANA, at least with regard to antagonizing the activation of the NF-κB pathway, which is thought to contribute to the establishment and/or maintenance of latency [52,54,61]. Cytoplasmic LANA would thus support the action of the lytic switch protein, RTA, encoded by ORF50, which has been shown to counteract vFLIP-dependent NF-κB activation and its contribution to the maintenance of latency by aiding the degradation of vFLIP by the proteasome [53,71]. Taken together, our results suggest a role for cytoplasmic LANA variants in modulating NF-κB activity by recruiting components of the MRN DNA repair complex and thereby regulating KSHV latency.  [51,73] and the human leukemia monocytic cell line THP-1 (ACC-16 from the German Collection of Microorganisms and Cell Cultures-DMSZ) stably infected with rKSHV.219 (TrK.219) were grown in RPMI medium 1640 (containing L-glutamine) supplemented with 20% FCS, and in case of BrK.219 and TrK.219 with 4 μg/mL puromycin (Sigma, P8833). Cells were grown at 37˚C in a 5% CO 2 incubator. Adherent cells were plated in 6-well plates 24 hours before transfection (5x10 5 cells per well), or were microporated (1x10 6 cells per well, in 12-well plates). The suspension cells were split at a ratio 1:2 one day before microporation (1x10 6 cells per condition) or lysed for binding assays (1x10 7 cells per condition). HeLa.CNX cells were latently infected with a recombinant KSHV virus containing a puromycin-resistance cassette, which had been produced using BrK.219 cells. Briefly, BrK.219 cells were stimulated with α-IgM (2.5 μL/mL) for 48 hours. After centrifugation, supernatant, containing infectious virions, was collected and filtered using a 0.45μm pore-size filter to remove cell debris and stored at +4˚C. HeLa.CNX cells were seeded in a 12-well plate and one day later infected at an MOI of 10 with BrK.219-derived virus. After 48 hours, puromycin (1 μg/mL) was added to the medium for selection of the KSHV-infected (+) HeLa cells. Three weeks later, the stably KSHV infected cell line was tested for viral proteins expression (by immunoblotting). The TrK.219 cell line was established by infecting THP-1 cells with rKSHV.219 at an MOI of 10. Puromycin was added to the medium for selection at final concentration of 4 μg/mL. After four weeks, KSHV stably infected THP-1 cells (TrK.219) were tested by immunoblotting and PCR.

KSHV lytic cycle induction
KSHV lytic reactivation was induced as followed: HeLa.rKSHV cells were treated with a combination of RTA, ectopically expressed from a baculoviral vector (calculated as volume of medium containing baculovirus / volume of total medium in one well, vol/vol), and Sodium Butyrate (see figure legends for further details). Cell pictures to check for GFP and RFP expression were taken using a Nikon Intensilight C-HGFI microscope.

Reagents and plasmids
Full-length LANA was expressed from a vector with the pcDNA3.1 backbone. Human Mre11 was expressed by transfecting a plasmid purchased from Addgene (plasmid # 82033) and the corresponding empty vector (plasmid # 46960) was used as a control. Adherent cells were transfected using Fugene6 (Promega, E269A) according to the manufacturer's instructions. Cells were stimulated with naked DNA (ISD Naked, InvivoGen, tlrl-isdn) by transfection with Lipofectamine2000 (Invitrogen by Life Technologies, 11668-027), using the conditions indicated in the figure legends. siRNAs were purchased from Dharmacon: human Mre11 custom siRNA pool (#1: ccugccucgaguuauuaaguu; #2: cugcgaguggacuauaguguu; #3 gaugccauugaggaauuaguu), siGENOME Non-Targeting Pool#2 (D-001206-14-50). siRNAs were prepared according to the manufacturer's instructions and transfected at the concentrations indicated in the figure legends using the Neon transfection system (Thermo-Fischer Scientific) under the following microporation conditions: 1150V, 30ms, 2 pulses.

Cytosolic/nuclear fractionation
Cytosolic/nuclear fractions were prepared from whole cell lysates using NE-PER nuclear and cytoplasmic extraction reagents (ThermoFischer, 78835) according to the manufacturer's instructions. All extracts were incubated immediately with LANA-beads or stored at -80˚C.

Binding assays
Production of GST fusion proteins and GST-pulldown assays were performed as previously described [15,16]. Endogenous co-immunoprecipitation assays were performed using PEL cell lines (1x10^7 cells/condition) harvested with TBS-T buffer (20 mM TRIS-HCl pH 7.4, 150 mM NaCl, 50 mM MgCl 2 , 1% TritonX-100). Benzonase nuclease (Merck Millipore, 71205-3) was added to whole cell lysates (50U each 2x10 6 cells) for 30 minutes at RT to digest nucleic acids. Subsequently, the samples were centrifuged at 20800 g for 10 minutes at +4˚C and the supernatants used for immunoprecipitation. The input control corresponds to 4% of the lysate used for an individual immunoprecipitation sample. Protein G sepharose beads (GE Healthcare) were washed with TBS-T buffer and incubated for 5 hours at +4˚C with α-LANA (rat, from ABI, 18-210-100) or α-Rad50 (mouse, from GeneTex, GTX70228) or negative control (α-IgG rat or α-IgG mouse) antibody. Finally, antibody-coupled-beads were washed with TBS-T buffer, resuspended in PBS and used immediately or stored shortly at +4˚C.

Luciferase reporter assay
For luciferase reporter assays, HEK293 cells were transiently co-transfected in duplicates with NF-κB reporter plasmid and expression constructs as reported in the figures legends. At the indicated time points, cells were washed once with PBS and lysed using 125μl per well of Reporter Lysis Buffer (Promega). Luciferase activity was immediately measured at the luminometer (DIGENE DIAGNOSTICS, inc.) using 30μl per condition and 100μl Luciferase Buffer (40mM Tricine, pH 7.8, 10mM MgSO 4 , 0.5mM ATP, 10mM DTT, 0.5mM Coenzyme A, 0.5mM D-Luciferine). To test for statistical significance a two-tailed T-test was used.