Myxoma virus lacking the host range determinant M062 stimulates cGAS-dependent type 1 interferon response and unique transcriptomic changes in human macrophages

The evolutionarily successful poxviruses possess effective and diverse strategies to circumvent or overcome host defense mechanisms. Poxviruses encode many immunoregulatory proteins to evade host immunity to establish a productive infection and have unique means of inhibiting DNA sensing-dependent type 1 interferon (IFN-I) responses, a necessity given their dsDNA genome and exclusively cytoplasmic life cycle. We found that the key DNA sensing inhibition by poxvirus infection was dominant during the early stage of poxvirus infection before DNA replication. In an effort to identify the poxvirus gene products which subdue the antiviral proinflammatory responses (e.g., IFN-I response), we investigated the function of one early gene that is the known host range determinant from the highly conserved poxvirus host range C7L superfamily, myxoma virus (MYXV) M062. Host range factors are unique features of poxviruses that determine the species and cell type tropism. Almost all sequenced mammalian poxviruses retain at least one homologue of the poxvirus host range C7L superfamily. In MYXV, a rabbit-specific poxvirus, the dominant and broad-spectrum host range determinant of the C7L superfamily is the M062R gene. The M062R gene product is essential for MYXV infection in almost all cells tested from different mammalian species and specifically inhibits the function of host Sterile α Motif Domain-containing 9 (SAMD9), as M062R-null (ΔM062R) MYXV causes abortive infection in a SAMD9-dependent manner. In this study we investigated the immunostimulatory property of the ΔM062R. We found that the replication-defective ΔM062R activated host DNA sensing pathway during infection in a cGAS-dependent fashion and that knocking down SAMD9 expression attenuated proinflammatory responses. Moreover, transcriptomic analyses showed a unique feature of the host gene expression landscape that is different from the dsDNA-stimulated inflammatory state. This study establishes a link between the anti-neoplastic function of SAMD9 and the regulation of innate immune responses. Author Summary Poxviruses encode a group of genes called host range determinants to maintain or expand their host tropism. The mechanism by which many viral host range factors function remains elusive. Some host range factors possess immunoregulatory functions responsible for evading or subduing host immune defense mechanisms. Most known immunoregulatory proteins encoded by poxviruses are dispensable for viral replication in vitro. The uniqueness of MYXV M062R is that it is essential for viral infection in vitro and belongs to one of the most conserved poxvirus host range families, the C7L superfamily. There is one known host target of the MYXV M062 protein, SAMD9. SAMD9 is constitutively expressed in mammalian cells and exclusively present in the cytoplasm and has an anti-neoplastic function. Humans with deleterious mutations in SAMD9 present disease that ranges from lethality at a young age to a predisposition to myelodysplastic syndromes (MDS) that often require bone marrow transplantation. More importantly, SAMD9 serves as an important antiviral intrinsic molecule to many viruses. The cellular function of SAMD9 remains unclear mostly due to the difficulty of studying this protein, i.e., its large size, long half-life, and its constitutive expression in most cells. In this study we used M062R-null MYXV as a tool to study SAMD9 function and report a functional link between SAMD9 and the regulation of the proinflammatory responses triggered by cGAS-dependent DNA sensing.


66
Mammalian hosts have sophisticated regulation for the triggering of pro-inflammatory 67 responses, especially after detecting danger signals in the cytoplasm. Many fundamental 68 sensing instruments and their direct downstream signaling axes have been described, such as 69 cGAS/STING/IRF3 axis for DNA sensing (1-3) and RNA sensing pathways (4) e.g., the RIG-70 I/MAVS/IRF3 axis (5). Additional factors may fine-tune the consequences of dangerous stimulus 71 (e.g., DNA substrates) resulting in distinctive overall cellular and immunological responses. The 72 outcome may also be tissue-and cell-type dependent. We are particularly interested in understanding the immunoregulatory mechanism of monocytes/macrophages. These immune 74 cells are among the first responders to viral infection and also important in the maintenance of 75 M063R does not possess a broad-spectrum host range function (17,19). In our study, the 149 control M063R-null MYXV infection did not induce IRF-dependent luciferase expression (not 150 shown), and similar to the wildtype virus M063R-null MYXV did not cause upregulation of 151 CXCL10 in human CD14 + macrophages ( Figure 2C). 152 The ΔM062R stimulated IRF-dependent gene expression is sensed through cGAS 153 We have reported previously that the infection defect by ΔM062R was due to its inability to 154 overcome host SAMD9 function (8). However, the immunological impact of ΔM062R remains 155 unknown. A computational analysis of SAMD9 across all homologues found putative DNA 156 binding domains (20), and DNA pulldown experiments using either the VACV 70mer dsDNA 157 (21) or herring testes dsDNA (not shown) indicated that SAMD9 was co-immunoprecipitated 158 with dsDNA ( Figure 3A). We found previously that the MYXV M062 protein binds to amino 159 acids (aa) 1-385 of human SAMD9 but not to the first 285 aa residues or c-terminal portion of 160 SAMD9 (18), while the region of 285-385 aa in human SAMD9 overlaps with the putative DNA 161 binding domain, the Alba-2 domain (20). We next examined if a human SAMD9 1-385 aa 162 fragment could also be associated with dsDNA. We used the SAMD9-null HeLa cells we 163 previously engineered (18) for the experiment and by transiently transfecting the cells to 164 express SAMD9 1-385 aa we performed dsDNA pulldown experiment similar to what is shown 165 in Figure 3A. We found that the SAMD9 1-385 aa truncated protein also associated with dsDNA. 166 As a control, we transiently expressed a SAMD9 N-terminal fragment of 1-110 aa that contains 167 the SAM domain in SAMD9-null cells, and found that this SAMD9 fragment was not associated 168 with DNA. We then investigated whether the expression of MYXV M062 might interfere with 169 SAMD9's presence in the dsDNA pulldown content. As a control, we infected HeLa cells 170 expressing intact endogenous SAMD9 with either wildtype MYXV or ΔM062R. Wildtype MYXV 171 infection significantly reduced the amount of SAMD9 associated with dsDNA compared with that 172 from ΔM062R infection ( Figure 3C).
Considering that SAMD9 may function through forming a complex with factors binding to DNA,174 we decided to test whether the ability of ΔM062R to induce IFN-I is due to the activation of DNA 175 sensors. We utilized a luciferase expression in human monocytic THP-1 cells for the study. 176 THP-1 cells can be differentiated into macrophages for testing DNA sensing and downstream 177 outcome, and the firefly luciferase (F-Luc) expression is driven by the IRF recognition domain 178 (Invivogen, San Diego, CA) (22). To test whether DNA sensing plays a role in ΔM062R-induced 179 IFN-I induction and pro-inflammatory responses, we used cGAS-null THP-1 cells that were 180 engineered from the F-Luc expressing parental cells mentioned above (22). We found that 181 ΔM062R mutant virus stimulated robust luciferase expression comparable to that induced by 182 interferon-stimulating DNA (ISD) (21) transfection ( Figure 4A). In the absence of cGAS, the 183 luciferase expression caused by both ΔM062R and ISD was eliminated ( Figure 4A). However, 184 transfection of 2'3'-cGAMP, the messenger molecule generated by cGAS upon DNA binding, 185 successfully bypassed the lack of cGAS in cGAS-null THP-1 to restore F-Luc expression 186 ( Figure 4B). We thus conclude that the immunostimulatory effect of ΔM062R is due to the 187 activation of cGAS-dependent DNA sensing pathway. 188 Knocking down SAMD9 expression in monocytes/macrophages attenuated their 189 proinflammatory responses 190 MYXV M062 inhibits SAMD9 function, leading to a productive viral infection. We next examined 191 whether SAMD9 played a role in regulating the proinflammatory responses induced by 192 ΔM062R. We generated stable SAMD9 knock-down THP1 cells using lentivirus expressing 193 shRNAs targeting human SAMD9. As control, we engineered THP1 cells stably expressing 194 scrambled shRNAs. We infected differentiated THP1 control or SAMD9 knockdown cells with 195 ΔM062R for 18 hrs before examining pro-inflammatory cytokine production via RT-PCR. We 196 found that reduced SAMD9 expression indeed attenuated ΔM062R-induced pro-inflammatory 197 responses ( Figure 5A). Transfection of ISD in these cells showed a similar attenuation in the IFNβ and ISG expression in SAMD9-knockdown THP1 cells ( Figure 5C) that is similar to the 200 response in the control THP1 cells. We thus concluded that ΔM062R infection stimulated a 201 unique pro-inflammatory state that is cGAS-dependent and also regulated by SAMD9. 202 Next generation sequencing showed a unique gene expression profile during ΔM062R infection 203 We conducted a next generation sequencing study in the macrophage-like THP-1 cells to 204 investigate the global transcriptomic change caused by ΔM062R. We hypothesized that since 205 ΔM062R infection of macrophages stimulated cGAS-dependent IFN-I response, the infection 206 will lead to similar results as dsDNA stimulated changes at the transcriptomic landscape. As 207 control, we included the cells transfected with ISD dsDNA. Using the dual RNAseq bioinformatic 208 analyses, we found that ΔM062R infection in monocytes/macrophages stimulated a very 209 different gene expression profile from that of the ISD group (PCA plot Figure 6A, Venn diagram 210  We next compared the viral gene expression pattern at 8 hours post-infection between ΔM062R 219 mutant virus and wildtype MYXV. Surprisingly, in the ΔM062R infection we detected all the viral 220 genes expressed in the wildtype MYXV infection (Supplemental Figure 2). We found 57 viral 221 genes in ΔM062R infection were differentially expressed compared to wildtype MYXV with 222 statistical significance (p value ranging from 1.03e-03 to 2.72e-88). Most of them (52 viral genes) ( Table 2) showed slightly higher levels in ΔM062R infection than that in wildtype MYXV 224 infection. M136R expression in the ΔM062R infection is 2.6-fold (logFC=1.4) higher than that 225 expressed in wildtype MYXV infection ( Table 2). Only 5 viral genes showed significant reduction 226 at RNA levels during ΔM062R infection compared to that in wildtype MYXV infection. Among 227 these viral genes, only M062R transcript was noticeably reduced (logFC=-3.95) due to the 228 deletion of the central 80% of the M062R coding sequence (7) (Supplemental Figure 3) and 229 the reduction in the levels of the remaining viral genes are in the range of 0.44-0.64 (logFC 230 between -1.16 and -0.652) ( Table 2). 231

232
Poxviruses are large dsDNA viruses with an exclusively cytoplasmic life cycle. These viruses 233 can effectively inhibit IFN-I induction through many strategies. These strategies include blocking 234 IFN-I signaling through decoy receptors (23), inhibiting key signaling effector molecules of the 235 sensing pathways (9, 11), and/or reprograming host gene expression in the nucleus (24-26). 236 One critical ability is for mammalian poxviruses, especially virulent poxviruses, to circumvent 237 host immunosurveillance and the antiviral immune responses induced as a result of DNA 238 sensing (9). Myxoma virus belongs to the genus of Leporipoxvirus and has a narrow host 239 tropism, causing infectious disease only in rabbits. In European rabbits, virulent MYXV can 240 cause lethal infection with 100% lethality and profound immunosuppression (27). Despite its 241 limited host tropism for infectious disease, we found wildtype MYXV inhibits the DNA sensing 242 pathway in human cells comparably to VACV, suggesting the presence of an antagonistic 243 mechanism directed against host DNA sensing in a species-independent manner. 244 Host detection of cytoplasmic DNA by the cyclic GMP-AMP synthase (cGAS) leads to the 245 production of the second messenger molecule 2'3'-cGAMP; binding of cGAMP to the adaptor, a 246 stimulator of interferon genes (STING), triggers the activation of IFN-I production through a 247 series of signaling events involving STING activation, recruitment and activation of TBK1, and phosphorylation of IRF3 (28). Many poxvirus proteins inhibit DNA sensing through different 249 strategies, such as direct degradation of 2', 3'-cGAMP by Poxvins (12,13), inactivation of 250 STING through mTOR by VACV F17 (15), inhibition of IRF3 by VACV C6 (29), and inhibition 251 NF-κB activation by many VACV proteins including B14 (30) and F14 (31) showed reduced DNA replication without late viral proteins being detected through western blot 275 (7). Interestingly, in our RNAseq analyses we not only detected post-replicative viral RNA, 276 especially late viral RNAs during ΔM062R infection, but also found ΔM062R viral RNA synthesis 277 patterns in macrophages closely resembling that of the wildtype MYXV infection. We observed 278 an occasional reduction in RNA levels among a few late genes during ΔM062R infection, and 279 most of the ΔM062R viral transcripts were present at slightly higher levels than that in the 280 wildtype MYXV infection at the same time point. The presence of significant levels of late RNA 281 during ΔM062R infection is unexpected, as in order to synthesize late viral RNA comparable to 282 the wild type virus level, many intermediate proteins must be produced de novo. This phenotype 283 suggests a unique antiviral state stimulated by ΔM062R MYXV infection. In this state, viral 284 protein synthesis, especially late protein production, is inhibited, which effect is coupled with 285 inhibition of viral DNA replication. We speculate that the unique antiviral effect of inhibiting viral 286 protein synthesis may be connected to the DNA sensing event. An alternative possibility is that 287 the absence of M062 during ΔM062R infection may lead to translation deceleration of viral 288 proteins until a complete stop, when a large quantity of late viral proteins are needed to 289 complete the life cycle. In this alternative scenario, the DNA-trigger immune response reported 290 may be caused by suboptimal levels of viral immunoregulatory proteins. This is, however, less 291 probable because of the observed robust IRF-dependent gene expression triggered by 292 ΔM062R, comparable to what is directly induced by dsDNA as shown in the luciferase assay. 293 Cytoplasmic sensing of DNA to trigger protective inflammation plays a key role in host antiviral 294 defense (34, 35). The causes of cytoplasmic DNA may vary during the lifetime of a mammalian 295 cell, such as improperly processed cellular DNA due to DNA repair or replication defect (36), 296 and foreign DNA such as during viral infection (9, 37). Once triggered, DNA sensing induced 297 IFN-I production and inflammation will lead to dramatic changes in the immunological milieu that 298 may ultimately alter the immune responses profoundly. Thus, there must be additional control 299 mechanisms to monitor and then regulate DNA-dependent IFN-I induction. There are known 300 downstream host control mechanisms to fine tune IFN-I responses. Other than negative 301 feedback cascade to restrict the duration and extent of IFN-I responses, e.g., SOCS, regulation 302 of IFN-α receptor (IFNAR), and USP18 (38-40), intracellular signaling events and miRNAs can 303 also perform such function (41). Upstream of IFN-I production, there are also regulatory 304 measures to pattern recognition receptors (PRRs) and their adaptors, e.g., AKT to cGAS (42), 305 TMEM120A to STING (43), and RNF138 to TBK1 (44). Our work provides evidence of a novel 306 mechanism to fine tune the IFN-I response, which may operate through cellular translation 307 regulation to PRR activation (Figure 7), and ultimately alters the global transcriptional 308 importantly, SAMD9 may serve as a signaling hub to fine tune the immunological consequences 322 of the cell. SAMD9 may be targeted by other viruses (46) and has broad-spectrum antiviral effect (47-49). In our pursuit of understanding the mechanism of the SAMD9 antiviral effect, we 324 revealed a key role of SAMD9 in regulating innate sensing for transcriptional remodeling in 325 monocytes/macrophages. This work leads us to investigate its function in connecting innate 326 immune sensing, translation control, and transcriptional refinement in host immune responses in 327 the next step. 328 Inc). Sybr green RT-PCR primers used in this study is listed in Table 1. 392

Generation of SAMD9 knockdown and control THP1 cells 393
Lentiviral particles for stable SAMD9 shRNA expression (Cat# sc-89746-V, Santa Cruz 394 Biotechnology, Dallas, TX) and Lentiviral particles for stable expression of scramble shRNAs 395 (Santa Cruz Biotechnology, Dallas, TX) were used for generating the cell lines. We followed the manufacturer's standard protocol similar to that reported before (7)  Additionally, options were provided to output "Ballgown-ready" files. 435 For gene-level exploratory data analysis (EDA) and differential expression analysis, StringTie 436 output was imported into DESeQ2 v1.32.0 (57), where p-value and adjusted p-value thresholds 437 were set to 0.05 and 0.1, respectively. 438 Differential gene expression analysis (e.g., M062R-null MYXV infection vs. mock and wildtype 439 MYXV infection vs. mock) was uploaded for IPA to host pathway core analyses, and graphic 440 summary was exported for data presentation. 441

Statistical analyses 442
Graphpad Prism 9.1 was used for statistical analyses. Multiple-group comparison with single 443 variable was performed using One-way ANOVA followed by secondary comparisons (e.g., 444 Tuckey multiple comparisons test). Statistical significance is defined by a p<0.05.

Acknowledgements 446
The study was supported by NIH K22AI099184 and R01AI139106 to J.L., a fund from the River 447 were performed with ordinary one-way ANOVA followed by Tukey's multiple comparison test 463 and p<0.05 is defined as being statically significant (****p<0.0001). Shown is a representative 464 result of 2 biological replicates and each data point is an average of triplicate measurements 465 (technical replicates). B. In the absence of M062R gene, the resulting ΔM062R MYXV loses the 466 ability to inhibit HT-DNA stimulated luciferase expression. The same luciferase-expressing THP-467 1 cells were differentiated into macrophages as in "A". After viral infection at a high moi of 10 468 with VACV or MYXV, cells were transfected with HT-DNA for 18 hrs. Supernatant was then 469 collected to measure the luciferase activities. M062R-knockout MYXV could no longer inhibit dsDNA-stimulated IRF-dependent luciferase expression. Statistical analyses were performed 471 using ordinary one-way ANOVA followed by Dunnett multiple comparison test and p<0.05 is 472 defined as being statistically significant (****p<0.0001). Shown is a representative of 3 biological 473 replicates and each data point is the average of 2 replicating measurements. 474  of SAMD9 with dsDNA. With the same dsDNA probe described above, we used HeLa cells 506 expressing endogenous SAMD9 and infected them with either wildtype MYXV expressing V5 507 tagged M062 protein or ΔM062R MYXV for dsDNA pull-down assay. Proteins associated with 508 DNA were separated on SDS-PAGE for Western Blot probing for SAMD9 and V5 tagged M062. 509 The input lysates were examined for total SAMD9 expression. 510 host genes identified as being differentially regulated in ISD, ΔM062R, and wildtype MYXV 538 groups compared to mock treated cells. Gene lists were generated by performing differential 539 gene expression analysis using the R library DESeQ2, and only those genes whose adjusted-p-540 value was less than 0.1 were included in the analysis. C. Heatmap showing distinct host gene 541 expression profile stimulated by ΔM062R from that by dsDNA. All significant differentially 542 expressed genes between 2 groups under study were visualized using the R library heatmap.2. 543 Gene counts were normalized using DESeQ2's default normalization. Each row's values were 544 scaled using a z-score method before plotting and ward. D hierarchical clustering was 545 performed using the Euclidean distance measure. D. Distinct host transcription profiles between 546 ΔM062R and wildtype MYXV infection. All significant differentially expressed genes, between 547 the groups under study, were visualized using the R library heatmap.2. Gene counts were 548 normalized using DESeQ2's default normalization. Each row's values were scaled using a z-549 score method before plotting and ward. D hierarchical clustering was performed using the 550 Euclidean distance measure. E. Ingenuity pathway analysis (IPA) revealed signaling pathway 551 stimulated by ΔM062R infection. Host genes differentially expressed during ΔM062R infection at 552 8h post-infection were analyzed using IPA (Qiagen, version 01-20-04) for pathways affected. 553 The graphic summary is shown. F. ISD transfection leads to activation of classic antiviral 554 responses consistent to dsDNA-stimulated signaling events. Shown is the graphic summary 555 generated by IPA using differentially expressed genes from ISD treatment group at 8-hour time 556 point. 557