Human Cytomegalovirus UL138 Interaction with USP1 Activates STAT1 in infection

Innate immune responses are crucial for limiting virus infection. However, viruses often hijack our best defenses for viral objectives. Human Cytomegalovirus (HCMV) is a beta herpesvirus which establishes a life-long latent infection. Defining the virus-host interactions controlling latency and reactivation is vital to the control of viral disease risk posed by virus reactivation. We defined an interaction between UL138, a pro-latency HCMV gene, and the host deubiquintase complex, UAF1-USP1. UAF1 is a scaffold protein pivotal for the activity of ubiquitin specific peptidases (USP), including USP1. UAF1-USP1 sustains an innate immune response through the phosphorylation and activation of signal transducer and activator of transcription-1 (pSTAT1), as well as regulates the DNA damage response. After the onset of viral DNA synthesis, pSTAT1 levels are elevated and this depends upon UL138 and USP1. pSTAT1 localizes to viral centers of replication, binds to the viral genome, and influences UL138 expression. Inhibition of USP1 results in a failure to establish latency, marked by increased viral genome replication and production of viral progeny. Inhibition of Jak-STAT signaling also results in increased viral genome synthesis in hematopoietic cells, consistent with a role for USP1-mediated regulation of STAT1 signaling in the establishment of latency. These findings demonstrate the importance of the UL138-UAF1-USP1 virus-host interaction in regulating HCMV latency establishment through the control of innate immune signaling. It will be important going forward to distinguish roles of UAF1-USP1 in regulating pSTAT1 relative to its role in the DNA damage response in HCMV infection.


87
Viral latency is a remarkable coup d'etat that allows for the virus to persist in the face of 88 robust antiviral immunity and is a hallmark of all herpesvirus infections. Latency is defined as a 89 quiescent state of persistence whereby viral genomes are maintained in the absence of progeny 90 virus production. During latency, viral gene expression is quieted, but not completely silenced 91 [1][2][3]. The β-herpesvirus, human cytomegalovirus (HCMV), establishes latency in hematopoietic 92 progenitor cells (HPCs within the CD34+ subpopulation) and cells of the myeloid lineage [4,5].

93
HCMV latency is marked by sporadic and frequent asymptomatic reactivation events, which are 94 controlled by the human immune response. In the absence of adequate cellular immunity, such 95 as in the case of solid organ or hematopoietic cell transplantation, HCMV reactivation poses life 96 threatening disease risk [5][6][7]. HCMV infection, reinfection or reactivation during pregnancy and 97 subsequent transmission to the fetus makes HCMV the primary infectious cause of congenital 98 birth anomalies [8,9]. The benefits or consequences of lifelong persistence of HCMV in 99 otherwise healthy individuals remain poorly defined [10][11][12][13][14][15]. Understanding the molecular basis 100 of latency and reactivation is important in developing strategies to target or control the latent 101 reservoir of HCMV and reveals fundamental mechanisms at the pinnacle of virus-host co-102 evolution.
mass spectrometry (IP-MS/MS). Interacting candidates that precipitated from a control pull 139 down using the FLAG antibody and a lysate from cells infected with TB40/E lacking the FLAG 140 epitope were subtracted from the interacting candidate data set. Top-ranking interacting 141 candidates were based on peptide count and coverage, summarized in Table 1. IP-MS/MS 142 peptides and data are provided for these candidates in Supplementary Information (SI) Table 1.

143
This screen previously identified the lower-ranking interaction with EGFR [19]. UAF1/WDR48 144 (herein referred to as UAF1) was the top ranked UL138 interacting protein, followed by WDR20 145 and USP12. Two previous studies have identified the UL138 interaction with UAF1 (referred to 146 as WRD48), WDR20, and USP12 [34,35], independently validating this work. We verified the 147 UAF1 interaction with UL138 by overexpressing UAF1 fused in frame with a hemagglutinin (HA, 148 UAF1HA) epitope tag and UL138 fused in frame with myc epitope tag (UL138myc).
150 Table 1. Summary of UL138 interactors 151 152 USP1 is the most well defined UAF1 interactor, but likely to escape an interactome 153 screen. To determine if USP1 is associated with UL138-UAF1 complexes, we exogenously 154 expressed UL138myc and USP1 fused in frame with a hemagglutinin (HA, USP1HA) epitope tag.

155
We detected UL138-USP1 interaction through the co-immunoprecipitation of USP1HA with 156 UL138myc (Fig. 1B), as well as the reciprocal IP (Fig. 1C). To determine if UL138 and USP1 157 interact during HCMV infection, cells were infected with a TB40/E-UL138myc virus. At 24 hours 158 post infection (hpi), UL138myc IP co-precipitated USP1 (Fig. 1D). While not defined here, the 159 interaction between UL138 and USP1 is likely indirect through UAF1. These results validate the 160 interaction between UL138, UAF1, and USP1.   CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

161
The copyright holder for this preprint this version posted February 7, 2023.

258
Late phase pY701-STAT1 binds to the viral genome and regulates UL138 transcription.

259
Considering the presence of pY701-STAT1 at RCs at 72 hpi, we were interested in whether 260 pY701-STAT1 is binding to the viral genome as a transcription factor. Utilizing PhysBinder, we 261 identified 86 pY701-STAT1 consensus sites on the viral genome. Two of which were within the 262 UL133-UL138 locus upstream of UL138, designated site 1 and site 2 (Fig. 6A). To understand if 263 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint pY701-STAT1 was binding to either of these two sites during HCMV infection, we performed 264 CUT&RUN paired with RT-qPCR to quantitatively detect target sequences bound by STAT1 265 (Fig. 6B). CUT&RUN captures DNA bound by a protein using an antibody specific to the protein 266 and micrococcal nuclease conjugated to protein A-protein G to cleave and release the DNA-267 protein complex [45,46]. After interferon stimulation, pY701-STAT1 binds to an interferon-268 sensitive response element (ISRE) consensus sequence upstream of many ISGs to induce 269 gene expression, including IRF1. We analyzed pY701-STAT1 binding to the IRF1 consensus 270 sequence in uninfected fibroblasts with or without universal interferon stimulation for 30 minutes.

271
In unstimulated cells, very low levels of pY701-STAT1 were bound to the IRF1 consensus site.

272
However, upon universal interferon stimulation, there was a significant increase in pY701-273 STAT1 binding to the IRF1 consensus sequence. In WT infection at 48 hpi, pY701-STAT1 274 bound both site 1 and 2 upstream of UL138 relative to the uninfected IRF1 control, whereas we 275 did not detect binding of STAT1 to these sites in ∆UL138STOP infection. Therefore, this data 276 suggests pY701-STAT1 binds to the UL133-UL138 region of the viral genome during HCMV 277 infection in a manner dependent on UL138.

278
Given pSTAT1 binds to the viral genome at two sites upstream of UL138 during HCMV 279 infection, we analyzed UL138 transcripts in infected fibroblasts in the presence of Ruxolitinib, a 280 JAK1/2 inhibitor. Through RT-qPCR, we analyzed UL138 expression relative to viral genomes 281 when phosphorylation of STAT1 was inhibited (Fig. 6C). With inhibition of JAK1/2, we found a 282 25% diminishment in UL138 gene expression but no effect on immediate early 1 (IE1) 283 expression per viral genome in a TB40/E infection relative to a vehicle control. This data 284 suggests that pY701-STAT1 binding to the viral genome regulates UL138 gene expression. 285 286 USP1 and pY701-STAT1 are required for the establishment of viral latency. Given UL138 287 has been defined as important to the establishment of latency, we next wanted to investigate 288 the possible role of USP1 in latency establishment. We chemically inhibited USP1 activity with 289 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint C527 in primary CD34+ HPCs infected with WT or ∆UL138STOP. For these experiments, infected 290 CD34+ HPCs were isolated by fluorescent activated cell sorting and then seeded into long term 291 bone marrow cultures over a stromal cell support to maintain HPC phenotype and function [47].

292
At 10 dpi, HPC cultures were split. Half the cells were seeded by limiting dilution onto fibroblast 293 monolayers in a cytokine-rich media to promote myeloid differentiation and HCMV reactivation.

294
The other half of the cells were mechanically lysed and seeded in parallel by limiting dilution to 295 determine virus produced during the latency culture prior to stimulation of reactivation (pre-  As USP1 was important for the induction of STAT1 in fibroblasts ( Fig. 3 A, B), we were 313 interested in a possible role for USP1 specifically in viral reactivation. We infected CD34+ HPCs 314 with a WT virus, as described above; however, we added C527 or DMSO as a vehicle control at 315 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint the time of reactivation. Inhibition of USP1 at the time reactivation did not impact infectious 316 centers production (Fig. 7B). Therefore, USP1 activity is important for latency establishment but 317 not involved in reactivation.

318
Given the role of USP1 activity in the establishment of latency and the role of the type-I 319 IFN response in reversible silencing of the genome [48], we next wanted to investigate the role 320 of STAT1 in the establishment of latency. We infected CD34+ HPCs with a WT or ∆UL138STOP

332
Taken together, these data suggest that USP1 and pY701-STAT1, as modulated by UL138, are 333 important for repression of viral replication and successful establishment of viral latency.

336
Type 1 interferons include IFN and IFN , which bind to IFNAR and stimulates JAK1/2-337 mediated phosphorylation of STAT1 and STAT2 along with recruitment of IRF9 to generate the 338 ISGF3 complex [49]. The ISGF3 complex translocates to the nucleus and binds to a consensus 339 sequence known as interferon stimulated regulatory element (ISRE). Binding ISREs promote 340 the of expression of many ISGs to limit viral infection [50]. However, herpesviruses encode 341 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

374
USP12 also co-precipitated with UL138 (Table 1), suggesting that in addition to USP1, 375 that UL138 may also regulate USP12. The UAF1-USP12 complex has been implicated in 376 sustaining pY701-STAT1 by binding to CREB-binding protein (CBP) and preventing acetylation 377 of pY701-STAT1 and subsequent dephosphorylation by protein tyrosine phosphatase non-378 receptor type 1 (PTPN2) [64]. Although the role of the UL138-UAF1-USP12 complex will be 379 addressed in future studies, the inclusion of USP12 in the IP/MS data set suggests an additional 380 mechanism by which UL138 sustains pY701-STAT1. In support, JAK1 was also found a low-381 ranking hit identified in the IP/MS and bolsters the conclusions that UL138 interacts with host 382 proteins to sustain STAT1 signaling.

383
UAF1-USP1 has been reported to sustain pY70-STAT1 signaling by deubiquitinating 384 and preventing proteasomal degradation of TBK1 [40]. While we show that pY701-STAT1 385 depends on UL138 and USP1 at late times during infection in fibroblasts (Fig. 2-3), we were 386 surprised that neither total or phosphorylated levels of TBK1 were affected by the absence of 387 UL138 or USP1 as a possible mechanism to sustain pY701-STAT1 (Fig. 2C)  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint by IFNAR. It will be important to determine if this is the mechanism by which USP1 and UL138 394 function to sustain STAT1 signaling. Another study found that specifically pY701-STAT1 could 395 be targeted for proteasomal degradation in the nucleus as an additional method to terminate 396 pY701-STAT1 signaling [66]. This highlights another potential mechanism of UL138-USP1 in 397 sustaining pY701-STAT1 during HCMV infection. Future studies will investigate these 398 possibilities, as well as the interaction with USP12, to define the mechanisms by which UL138-399 UAF1 complexes sustain STAT1 activity during HCMV infection.

400
Contrary to our findings, Kalejta and colleagues showed that UL138 inhibits the innate 401 immune response by targeting STING for degradation [41]. UL138 was shown to target STING, 402 but this effect was only evident when both UL138 and STING were transiently overexpressed.

414
We show that the prolonged pY701-STAT1 stimulated during HCMV infection correlates 415 with an enhanced and sustained expression of several type 1 IFN-associated ISGs in a UL138-416 dependent manner. These include MX1, OAS, IFIT1, and IFIT2 (Fig 4B). Type 2 interferons also 417 lead to the phosphorylation of STAT1, which form homodimers which bind to gamma activated 418 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint sequences (GAS) to drive expression of ISGs that are similar and distinct to type 1 IFN driven 419 ISGs [44]. However, we found no induction of a type-2 IFN driven ISG, ICAM1, above 420 uninfected levels. Therefore, elevated pY701-STAT1 levels are most likely part of the type 1 IFN 421 response. In both WT and ∆UL138STOP, the induced type 1 IFN ISGs diminish by 72 hpi despite 422 sustained pY701-STAT1. This is undoubtedly due, in part, to robust inhibition of type-1 IFN 423 signaling by multiple HCMV gene products [53,[67][68][69][70]. However, we also show that pY701-424 STAT1 localizes to nuclear viral RCs (Fig 5A, SI Fig. 2,) and binds to the viral genome to 425 stimulate UL138 expression (Fig. 6). With respect to regulation of viral gene expression, STAT1 426 signaling been shown suppress lytic reactivation of MHV-68, similarly to UL138 but through a 427 different mechanism. The MHV-68 ORF50 promoter, which is essential for the lytic switch, 428 encode STAT1 repressive elements to negatively regulate lytic replication in response to IFNg

443
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

473
We have shown that the HCMV latency protein UL138 forms a unique interaction with

492
assays were performed as previously described [19,26]. In chemical treatments of MRC-5 lung 493 fibroblasts, 0.88uM C527 (ApexBio), 5uM Ruxolitinib (STEMCELL Technologies) every 24 494 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

514
Antibodies were incubated in with a blocking solution of 5% BSA in TBS-T, as per antibody 515 manufacturer specifications. After antibody staining, blots were incubated with fluorescent 516 secondary antibodies and imaged and quantitated using a Li-Cor Odyssey CLx infrared scanner 517 with Image Studio software. Antibodies and sources are defined in Table 3.  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

537
RT-qPCR and qPCR. Cells were infected with 1 MOI of TB40/EGFP and DNA and RNA was 538 isolated using Quick-DNA/RNA miniprep kit (Zymo Research) from 0-72 hpi. RNA was reverse 539 transcribed into cDNA using SuperScript VILO cDNA Synthesis kit (Thermo Fisher Scientific). 540 cDNA and DNA was quantified using LightCycler SYBR Mix kit (Roche) and corresponding 541 primers (Table 2). Assays performed on Light Cycler 480 and corresponding software. Relative 542 expression was determined using the ΔΔCT method normalized to H6PD. Viral genomes were 543 quantified using a standard curve and normalized to RNAseP. For quantification of ISGs, 544 fibroblasts were infected (MOI=1) for 24, 48, and 72 hpi. RNA was isolated using Quick-545 DNA/RNA miniprep kit (Zymo Research) and cDNA was prepared using 1000 ng of total RNA 546 and random hexamer primers (Thermo Fischer Scientific). Real-time qPCR (TaqMan) was used 547 to analyzed cDNA levels with specific TaqMan primer and probe sets (Thermo Fisher Scientifc).

548
Relative expression was determined using the ΔΔCT method using GAPDH as the standard

558
Samples were then normalized to uninfected cells stimulated with universal type 1 interferon 559 (R&D Systems).

560
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint 561 Viral Latency and Reactivation. Infectious centers were quantitated in CD34 + HPCs, as 562 described previously (39). Frequency of infection centers were calculated using extreme limiting 563 dilution analysis (42). For the investigation of USP1 activity during latency establishment, CD34 + 564 HPCs were treated with 1uM/mL C527 (ApexBio) after sorting for CD34 + GFP + populations and 565 when stromals were replaced at 6 dpi. The study for USP1's role in reactivation, 1uM/mL C527

574
Statistics for experiments in this study were calculated using either unpaired student T-test or (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10. 1101/2023 Research Laboratories Division of Biotechnology Cytometry Core Facility for expertise and 587 assistance in flow cytometry. We acknowledge Dean Billheimer for statistics consultation.  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made    . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023       . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint  CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint transcripts encoding UL138 and IE were quantified by RT-qPCR normalized to H6PD. Viral 1006 genomes were analyzed through qPCR using primers for b2.7kb and RNaseP as a loading 1007 control. Relative expression was determined using the ΔCT method and values were 1008 normalized to the DMSO control and graphed for statistical analysis. Statistical significance was 1009 calculated using unpaired student t test and represented by asterisks. Graphs represent the 1010 mean of three replicates and error bars represent SEM. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.07.527452 doi: bioRxiv preprint