key: cord-0302378-78ua6qu5
authors: Rookhuizen, Derek C.; Bonte, Pierre-Emmanuel; Ye, Mengliang; Hoyler, Thomas; Gentili, Matteo; Burgdorf, Nina; Durand, Sylvère; Aprahamian, Fanny; Kroemer, Guido; Manel, Nicolas; Waterfall, Joshua J; Milne, Richard; Goudot, Christel; Towers, Greg J.; Amigorena, Sebastian
title: Induction of transposable element expression is central to innate sensing
date: 2021-09-11
journal: bioRxiv
DOI: 10.1101/2021.09.10.457789
sha: 55531fbce2eeaca8feaf3464ea726995dff93379
doc_id: 302378
cord_uid: 78ua6qu5

Evidence indicates that transposable elements (TEs) stimulate innate sensing pathways in various pathologies but it is not clear whether they are sensed during normal physiological responses. Here we show that, during activation with an exogenous pathogen associated molecular pattern (PAMP), dendritic cells (DCs) epigenetically remodel heterochromatin at TEs by depleting the methyltransferase Suv39h1 and reducing histone-3 lysine-9 trimethylation (H3K9me3). TLR4 signaling activates TE expression to enhance innate responses through the DNA sensor cGAS. Cytosolic cGAS-bound DNA comprised LINE1 TEs as the predominant endogenous ligands. Concordantly, LINE1 inhibition attenuated the type-I IFN response to LPS and rescued influenza virus infection. We propose that in healthy cells, exogenous PAMPs epigenetically activate self-derived PAMPs (LINE1) that engage cGAS to enhance responses. These data explain why pathogens employ redundant and broad innate immune countermeasures, to suppress activation of host PAMPs and illustrate a hitherto unappreciated role for host genome-derived PAMPs in response to pathogens.

SUMMARY 24

Evidence indicates that transposable elements (TEs) stimulate innate sensing pathways in 25 various pathologies but it is not clear whether they are sensed during normal physiological 26 responses. Here we show that, during activation with an exogenous pathogen associated 27 molecular pattern (PAMP), dendritic cells (DCs) epigenetically remodel heterochromatin at TEs 28 by depleting the methyltransferase Suv39h1 and reducing histone-3 lysine-9 trimethylation 29 (H3K9me3). TLR4 signaling activates TE expression to enhance innate responses through the 30 DNA sensor cGAS. Cytosolic cGAS-bound DNA comprised LINE1 TEs as the predominant 31 endogenous ligands. Concordantly, LINE1 inhibition attenuated the type-I IFN response to LPS 32 and rescued influenza virus infection. We propose that in healthy cells, exogenous PAMPs 33 epigenetically activate self-derived PAMPs (LINE1) that engage cGAS to enhance responses. 34

These data explain why pathogens employ redundant and broad innate immune 35 countermeasures, to suppress activation of host PAMPs and illustrate a hitherto unappreciated 36 role for host genome-derived PAMPs in response to pathogens. 37

The ability to distinguish self from non-self is a central principle of immunity. Invading pathogens 40 must be recognized as non-self to trigger an adequate response while self-antigens must be 41 distinguishing different modes of transcriptional regulation across the three retrotransposon 144 classes. These observations reveal a previously unappreciated role for Myd88/Nfkb and Trif/Irf3 145 in driving TE expression in addition to gene induction. 146 147 Our results indicate that autonomous TE expression is part of the normal transcriptional 148 landscape downstream of classical LPS/TLR4 signaling. Concordantly, using LINE1 as an 149 exemplar of TE, we found that RELA, RELB, and IRF3 binding motifs were enriched in the 5' 150 promoter region of intact elements ( Figure 1E ). As a positive control for our observations, YY1, 151 a known LINE1 transcription factor in human (25) and mouse (26) also mapped to the LINE1 152 promoter region ( Figure 1E distribution of total LINE1 expression from mRNA-seq data (21-31%, Figure 1F ). Meanwhile, the 158

Myd88/Trif-dependent group contributed the largest proportion of total DE LINE1 mRNA (55-159 65%), whereas elements dependent on Trif alone contributed the least (12-20%, Figure 1F ). 160 Therefore, co-regulation by Myd88 and Trif induced greater TE RNA expression, compared to 161 either pathway alone. We conclude that both pathways contribute to induction of TE expression 162 downstream of LPS detection by TLR4. 163 164 TEs are transcriptionally silenced by epigenetic modifications (19) such as H3K9me3 which is 165 catalyzed by histone-lysine N-methyltransferases SUV39H1/2 and SETDB1, the latter 166 cooperating with TRIM28. Previous work showed that SUV39H1 selectively trimethylated H3K9 167 at intact LTR and LINE1 loci in a mouse embryonic fibroblast cell line (27) . Two studies have 168 subsequently shown that, in cell lines, the HUSH complex also targets evolutionarily young TEs 169 in collaboration with SETDB1 and TRIM28 (28, 29). Intriguingly, we found that LPS stimulation 170 of BMDC rapidly and selectively depleted Suv39h1 and Setdb1 mRNA ( Figure 1G , S1C). 171

Trim28, and HUSH complex members (Morc2a, Mphosph8) were also transiently decreased at 172 the mRNA level. Fam208a (HUSH) and Suv39h2 transcript levels were relatively stable in the 173 early phase of the response but increased at later time points ( Figure S1C ). LPS treatment also 174 induced several histone lysine demethylases belonging to the KDM family that target H3K9me3 175 ( Figure S1C ). Strikingly, Suv39h1 was also lost at the protein level on LPS stimulation of BMDC 176

whereas Setdb1 and Mpp8 were not ( Figure 1H ), indicating that additional mechanisms such as 177 Suv39h1 loss on these elements during LPS treatment. In contrast, although LPS stimulated 279 LTR element expression (Figure 1) , it did not extinguish H3K9me3 peaks at ERV1 and ERVK 280 elements as readily as at LINE1 and SINEB1 ( Figure S2G ), consistent with the reduced effect of 281 Suv39h1 knock-out on steady-state H3K9me3 at these loci ( Figure S2E ). We propose Suv39h1-282 independent regulation of these elements downstream of LPS. Our data support a model in 283 which LPS-induced depletion of SUV39H1 ( Figure 1G , H) facilitates heterochromatin loss to 284 directly regulate de-repression of specific LINE1 and SINEB1 elements. This model is 285 consistent with observations that disruption of epigenetic pathways that silence LINE1 causes 286 their expression and activation of innate immunity (13, 30, 31) . We hypothesize that specific 287

TEs are regulated as part of normal innate immune physiology and might naturally contribute 288

PAMPs to enhance, amplify, or broaden a physiological innate immune response. 289

We next sought evidence for TE contributing endogenous PAMP that might be detected and 292 therefore contribute to LPS-driven inflammatory responses. We first examined the effect of 293 overexpressing Suv39h1 on LPS treatment of RAW264.7 macrophages ( Figure 3A ). Suv39h1-294 overexpression (SUV39H1oe) significantly dampened the LPS-mediated induction of a 295 luciferase construct controlled by the ISG54 promoter construct measured as a surrogate for 296 IFN-I production and ISG induction (ISG54-luciferase). LPS induction of the endogenous ISG 297 VIPERIN was also repressed by SUV39H1oe ( Figure 3B ). This effect was confirmed at the level 298 of signal transduction with reduced levels of phosphor-TBK1 and -IRF3 but unaffected levels of 299 total TBK1/IRF3 protein ( Figure 3D ), consistent with Suv39h1oe reducing PAMP levels and thus 300 downstream PRR signaling. Importantly, SUV39H1oe did not impact ISG54-luciferase 301 production after stimulation with a STING ligand, 2'3'cGAMP ( Figure 3C ), demonstrating that 302 cells overexpressing Suv39h1 were capable of producing WT levels of ISG activation 303 downstream of DNA sensing activation. 304 305 Conversely, complete ablation of Suv39h1 had opposite effects. LPS-stimulated Suv39h1KO 306 cells produced significantly higher levels of IFN-I (IFN) mRNA and protein than their WT 307 counterparts ( Figure S3A , B), exhibiting enhanced phosphor-TBK1, -p65, and IRF3 signal-308 transduction ( Figure S3C ). Concordantly, Suv39h1KO cells exhibited a more vigorous and 309 sustained anti-viral response to LPS than WT cells, evidenced by the upregulation of ISGs 310 ( Figure 3E , S3D). We conclude that Suv39h1 levels, up or down, determine the magnitude of 311 the IFN-I/ISG response following LPS stimulation. 312

In parallel experiments, we found that SUV39H1 depletion with shRNA in primary human 314 monocyte-derived DCs also significantly enhanced LPS induced ISG SIGLEC-1 expression 315 whereas it had less pronounced effects on expression of the NFkB-regulated activation marker 316 CD86 ( Figure S3E ,F). Our data are therefore consistent with SUV39H1 regulation of LPS 317 responses being conserved in human cells. 318

We hypothesized that enhanced steady-state expression of TEs in Suv39h1KO cells should 320 also amplify baseline ISG expression. In fact, gene set enrichment analysis (GSEA) of RNAseq 321 data revealed that Suv39h1KO BMDCs exhibited a significantly augmented type-I IFN/anti-viral 322 gene signature at baseline compared to WT cells (P < 10 -10 ) ( Figure 3F , blue). Further, ablation 323 of the type-I IFN receptor (Ifnar) in WT and Suv39h1KO BMDCs, abolished baseline differences 324 in ISG expression (P > 0.05) ( Figure 3F , gold), confirming that these ISG steady-state 325 differences resulted from enhanced type-I IFN signaling. In a second cell type, in RAW264.7 326 macrophages, depletion of Suv39h1 with shRNA ( Figure S3G , H) also enhanced the baseline 327 ISG expression as measured by ISG54-luciferase production ( Figure S3I ). These data suggest 328 that Suv39h1KO is not directly impacting ISG expression because increased ISG expression 329 levels in Suv39h1KO cells is IFN dependent. 330

To further probe whether ISG and cytokine loci are directly regulated by Suv39h1 and 332

H3K9me3, we analyzed H3K9me3 peaks in ISG loci and compared them directly with 333 mRNAseq data from the same samples. H3K9me3 assessment by ChIPseq revealed that 334 ISGs, and Ifnb itself, are not associated with Suv39h1-sensitive H3K9me3 marks ( Figure S3J ). 335 Furthermore, both WT and KO cells broadly maintained H3K9me3 peaks in the same set of 336 ISGs, and in fact, Suv39h1KO cells maintained slightly more peaks ( Figure S3J ). In addition, the 337 presence or absence of H3K9me3 peaks did not correlate with ISG mRNA expression at 338 baseline ( Figure S3J) We first assessed activation of the DNA-sensing cGAS pathway by measuring STING 356 phosphorylation (Konno et al., 2013) in LPS activated WT BMDCs by immunoblot. In this 357 experiment, we increased the number of cells loaded into gel wells (8.5x10 5 ) and transfected a 358 low dose of 2'3'cGAMP to allow comparison of phosphor-STING as a control. Strikingly, LPS 359 induced STING phosphorylation and this was lost in Cgas -/cells, consistent with endogenous 360 PAMP detection by cGAS ( Figure 4A ). LPS-activation of IRF3 detected by WB was only 361 modestly reduced by cGAS knock out, consistent with cGAS-independent LPS-induced 362 canonical TRIF and MYD88 activation of IRF3 ( Figure 4A ). IRF3 activation by phosphorylation is 363 a complex process, and one possibility is that cGAS impacts IRF3 phosphorylation at sites that 364 are important for activation but that are not detected here (Robitaille et al., 2016). As expected, 365 transfection of BMDCs with 2'3'cGAMP (250-500ng/ml) induced STING and IRF3 366 phosphorylation independently of cGAS ( Figure 4A ). Concordantly, LPS stimulation of Cgas -/-367 BMDCs induced significantly less IFNB (type I IFN) protein as compared to WT cells ( Figure  368 S4A). As a control, Cgas ablation blocked HT-DNA-induced IFNB whereas 2'3'cGAMP 369 transfection induced similar IFNB levels in both WT and Cgas -/cells as expected since 370 2'3'cGAMP activates STING directly ( Figure S4B ). Similar results were obtained using freshly 371 isolated splenic DCs where LPS treatment of Cgas -/cells also triggered reduced expression of 372 the ISGs VIPERIN and CCL5 relative to WT cells ( Figure S4B ). Again, transfected 2'3'cGAMP 373 induced normal ISG responses in both WT and Cgas -/cells as expected. Importantly, surface 374 expression of CD86, a NFkB-regulated gene, was not impacted by CgasKO ( Figure S4B Unexpectedly, Mavs deletion had no effect on ISG induction by LPS ( Figure 4B in its zinc-thumb domain that prevent bound DNA from activating cGAS enzymatically (cGAS-518 DBM) (59), or a catalytically-dead cGAS mutant (cGAS-CD, E225A/D227A) that cannot produce 519 2'3'cGAMP (Kranzusch et al., 2013) . Reconstitution levels of the WT and cGAS-mutants are 520 shown in Figure S5G . Because cGAS-FL elevated IFN-I in untreated cells ( Figure S5H ), we 521 subtracted baseline values from stimulation induced IFN-I following treatment with HT-DNA, 522 2'3'cGAMP, or LPS ( Figure S5I ). Critically, LPS-induced ISG54-luciferase expression was 523 selectively restored by cGAS-FL, but not by either cGAS mutant ( Figure S5I ). Accordingly, HT-524 DNA transfection selectively activated cGAS signaling when cGAS-FL but not when cGAS-DBM 525 or cGAS-CD were used ( Figure S5I ). In contrast, 2'3'cGAMP stimulated ISG54-luciferase in all 526 settings, confirming that downstream signaling functioned normally ( Figure S5I ). These data 527 reveal: 1) a specific requirement for cGAS during LPS stimulation and 2) that DNA-induced 528 cGAS enzymatic activity is necessary for cGAS enhancement of LPS-triggered ISG responses. 529 530 Because Suv39h1 regulates the magnitude and kinetics of LINE1 expression during LPS 531 challenge, we predicted that Suv39h1 deletion would enhance LINE1 DNA binding by cGAS. 532

Indeed, LINE1 was detectable on cGAS at low levels in untreated WT cells, but LPS treatment 533 significantly increased LINE1 density ( Figure 5E ). Furthermore, in Suv39h1KO BMDCs, steady-534 state LINE1 densities reached magnitudes similar to that found on cGAS in LPS-treated WT 535 cells. As expected these were further enriched upon LPS stimulation ( Figure 5E ). In contrast, as 536 a control, SINE DNA revealed disparate trends with no statistically supported differences 537 ( Figure 5F ). As a control for our HTS data, cGAS-IP, followed by qPCR, detected LPS-induced 538 enrichment of LINE1 and LTR (ERVK) DNA on cytosolic cGAS whereas genic DNA such as 539 To provide further validation for our model, we sought evidence for the role of LINE1s as 548 endogenous PAMPs that induce IFN-I by targeting them for epigenetic repression using a 549 CRISPR dCas9-KRAB fusion construct. In this system, the mutant Cas9 (dCas9) binds DNA but 550 is incapable of generating dsDNA breaks that would likely occur in excess when targeting 551 numerous LINE1 elements and cause cell death (60, 61). The fused KRAB domain recruits 552 epigenetic factors such as SETDB1, a H3K9me3 methyltransferase, driving 553 heterochromatization and subsequent transcriptional silencing of the target region (61). As a 554 positive control, we targeted the 5' untranslated region of Itgax, the gene encoding the highly 555 expressed integrin and myeloid marker, CD11b. Using two different guide RNAs (gRNAs), we 556 succeeded in selectively repressing CD11b while maintaining MHC-II expression ( Figure S5M ). 557 558 We next designed gRNAs to target 20,000-110,000 LINE1 elements belonging to the L1Md 559 family, and specifically enriched either at the L1Md_F2 sub-family (gRNA_1 and gRNA_2) that 560 bound at high levels to cGAS after LPS treatment or, as a negative control, at L1Md_A elements 561 (gRNA_3) which bound at low levels to cGAS ( Figure S5N ). Consistent with the notion that 562 elements that bind cGAS at higher levels should contribute the largest impact to cGAS-563 dependent IFN-I production, gRNA_1 and gRNA_2 significantly impeded IFN-I production; 564 gRNA_3, however, had no effect ( Figure 5H ). Importantly, differential analysis of mRNA-seq 565 data (Ctrl versus gRNA-expressing RAW cells) confirmed specific repression of L1Md 566 expression by gRNA_1 and gRNA_2 ( Figure 5I In this study, we introduce a new paradigm for the mechanism of innate immune sensing in 626 which endogenous PAMPs are produced within cells exposed to exogenous pathogen-derived 627 PAMP. We propose that these endogenous PAMPs are detected by classical innate sensing 628 pathways, amplifying and broadening responses to pathogens. Specifically, we discovered that 629 LPS induces LINE1 expression, which is detected by cGAS to contribute to LPS-elicited IFN-I 630 responses. We find that LPS activates LINE1 expression by driving SUV39H1 depletion that 631 leads to loss of H3K9me3 at TE loci, combined with MYD88/TRIF-driven transcriptional 632

activation. This discovery is important because it reveals an entirely new level of regulation of 633 innate immune responses and ensuing inflammation that may explain inconsistencies in the 634 field. 635 636

The notion that TE should be permanently switched off derives from the observation that TE 638 expression is typically associated with pathology, particularly autoinflammation and oncogenesis 639 (66-68). We propose that our observation of regulated TE expression, and sensing by cGAS 640 after LPS exposure, evidences an essential role for TE expression in normal innate immune 641 responses. This is supported by our observation that disparate PAMPs also activate TE 642 expression ( Figure 1L ). 643

Indeed, we hypothesize that diverse innate detection pathways converge on distinct epigenetic 644 regulators to induce TE PAMPs that enhance innate immunity such as type-I IFN anti-viral 645 responses. 646 647 Previous studies of TE regulation have suggested that expression of the youngest LINE1 is 648 epigenetically regulated by an evolutionary Red Queen style arms race akin to that between 649 host and pathogen (20). We hypothesize that such studies, characterizing the evolution of 650 LINE1 regulation, demonstrate how host genomes have evolved to specifically regulate their 651 expression when required. Strikingly, our data highlight the importance of LINE1 regulation in 652 enhancing innate immunity and provide a direct example of the evolutionary benefits of retaining 653 the capacity to regulate TE expression, rather than simply suppress it. Epigenetic marks such as 654

H3K9me3 and DNA methylation switch gene expression on and off, and we propose that TE are 655 simply another example of sequences that are epigenetically regulated in normal physiology, 656 rather than comprising a distinct class of elements that must be permanently silenced (19, 69) . The capacity of TEs to elicit innate immunity has also been linked with diseases other than 743 cancer, particularly inflammatory disease. Patients with Aicardies Goutiere Syndrome (AGS) 744 suffer an interferonopathy from early development which has been associated with TE 745 expression, reverse transcription, and activation of cGAS (37, 39, 71) . LINE1 DNA has also 746 been revealed as a source of STING-dependent neuroinflammation in a TREX1-mutant model 747 of AGS (12). To date, no associations have been made between the H3K9me3 pathway and 748 AGS, although a recent paper linked diminished SUV39H1 expression with inflammation in the 749 lungs of patients with chronic obstructive pulmonary disease (COPD) (100). Our data are 750 consistent with these observations and may provide mechanistic explanation. TE expression 751 has even been suggested as a driver of aging-associated inflammation (30, 31) showing z-scores of ISG mRNA expression (FPKM) in untreated Suv39h1WT (n=3) or Quiagen columns using the Qiagen RNeasy mini kit and eluted in RNase/DNase-free water. 1300

RNA preparations performed with this protocol routinely yielded an A260/280 ratio above 2 and 1301 averaging 2.08. HTS libraries were constructed using the TruSeq Stranded mRNA kit from 1302

Illumina and sequenced to a depth of at least 50x10 6 reads per sample using 100bp paired-end 1303 sequencing. 1304 1305

For qPCR, cDNA was generated from 500ng of RNA using random hexamers (Promega, 1306 C1181) and the SuperScript III First-Strand kit (Themro Fisher, 18080051) in a 20µl reaction. 1307

2µl of RNaseH was added and the reaction carried out at 37°C for 20 minutes. Following cDNA 1308 synthesis, 160µl of RNAse/DNase-free H 2 0 was added, and 4µl of this was used per qPCR 1309 reaction containing 1µl of 5µM primers and 5µl of SybrGreen reagent (LifeTechnologies, 1310 4367659 We used the Homer Software as previously described. However, we considered only TEs for 1325 the analysis if the row sum across all samples was FPKM 5. LINE1 and LTR elements greater 1326 than or equal to 4kb in length were considered intact. SINE elements were considered intact 1327 when they diverged less than 10% from their consensus sequences. 1328 1329

Differential Analysis 1330

Using the raw reads matrix, we filtered out the non-expressed genes and TE separately by 1331 requiring more than 5 reads in at least 2 samples for each features and normalized using 1332

DESeq2 R-package v1.18.1 (Love et al., 2014) . Then, the differential expression analysis was 1333 performed using DESeq2, only the Benjamini Hochberg (BH) adjusted p-value below 0.05 were 1334 considered as significant. 1335 1336 PCA 1337

The PCA was performed using the Ade4 package (S Dray, 2007) of the R software (version 1338 3.3.2). The barycenters were computed from the set of observations in each condition and 1339 projected into the PCA plot. 1340 1341

Chromatin Immunoprecipitation (ChIP)-sequencing 1342

1123 and L. Joannas for assistance with various techniques and/or experiments

1125 for helpful reading of the manuscript

We thank T. Jenuwein for providing the 1127 Suv39h1-deficient mice. S. A. received funding from the Institute Curie; Institut National de la 1128 Santé et de la Recherche Médicale; Centre National de la Recherche Scientifique; ANR 1129 "ChromaTin

received 1130 funding from la Ligue Contre le Cancer

Association de Recherche Contre le Cancer (ARC); grant ERC

Epigenomics of breast cancer". D.C.R was 1135 supported by funding from Institut National du Cancer (INCA) (grant 2017-1-PL BIO-05). High-1136 throughput sequencing has been performed by the ICGex NGS platform of the Institut Curie 1137 supported by the grants

Association 1141 pour la recherche sur le cancer (ARC)

1142 Fondation pour la Recherche Médicale (FRM); a donation by Elior; Equipex Onco-Pheno-1143 Screen

Gustave Roussy Odyssea, the 1144 European Union Horizon 2020 Projects Oncobiome and Crimson; Fondation Carrefour

This study indicated times with HT-DNA, LPS, poly(I:C), or 2'-3'cGAMP, at the indicated concentrations in 1212 6-well or 96-well tissue culture treated plates (Costar) following the same protocol as for 1213 BMDCs. 1214 1215 Dendritic cell infection wit IAV 1216 IAV/PR8/34 H1N1

D-G) was used for infection experiments. Briefly, 1219 BMDCs or JAWS II cells were plated in serum-free IMDM media supplemented with non-1220 essential amino acids and infected with IAV/PR8 for 1hour at 37°C, before addition of complete 1221 media with a final concentration of 10%FBS. Infections were allowed to proceed for the times 1222 specified in figure legends. For RTI, treatment, emtricitabine and disoproxil fumarate were 1223 administered two hours prior to innoculation and maintained at 10µM for the duration infection 1224 (120hrs). DMSO vehicle control was administered alongside RTI treatment as a control. 1225 1226 Generation of Ifnar -/-and Mavs -/-RAW264.7 macrophage cell lines with CRISPR 1227 WT RAW264.7 macrophages were deleted using recombinant Cas9 protein (IDT, Alt-R S

Mm.Cas9.IFNAR2.1AA, Mm.Cas9.IFNAR2.1AB), or Mavs

Briefly, gRNAs and Tracr RNA duplexes were 1232 generated by melting at 95°C for 5 minutes followed by downramping to 20°C at 5°C/minute

WT RAW264.7 cells were subjected 1235 to two rounds of nucleofection with Cas9-RNPs using the SF Cell Line 4D-Nucleofector X Kit S 1236 (Lonza) and the 4D Nucleofector apparatus precisely as specified by the manufacturer. Ifnar1 1237 and Ifnar2 gRNAs were delivered in the first and second rounds of nucleofection, respectively; 1238 both Mavs gRNAs were delivered in first and second rounds. Ifnar deletion was confirmed by 1239 luciferase assay as described under "Luciferase assay

Mavs deletion was confirmed by western 1241 blot. 1242 1243 Suppression of LINE1 using CRISPR-dCas9-KRAB 1244 In brief, two lentiviral vectros were used to target endogenous LINE1 elements in RAW264.7 1245 macrophages

2014) for 1248 epigenetic suppression of gRNA targets. gRNAs against LINE1 5' regions were generated using 1249 CRISPOR and selected based on their predicted in silico enrichment in full length LINE1 1250 elements and paucity of genic targets

Buffy coats were prepared from peripheral adult human blood and CD14 + monocytes were 1254 isolated by magnetic separation (Miltenyi 130-050-201). CD14 + monocytes were differentiated 1255 into DCs (MDDCs) in RPMI containing Glutamax, supplemented with 10%FBS (GIBCO), 1256 50µg/ml Gentamicin (GIBCO), 0.01M HEPES (GIBCO), 10ng/ml GM-CSF (Miltenyi premium 1257 grade), and 50ng/ml IL-4 (Miltenyi premium grade) at a density of 10 6 cells/ml

Briefly, 293FT in 1262 one well of a 6-well were transfected with 1µg psPAX2, 0.4µg pCMV-VSV-G and 1.6µg of a 1263 lentiviral vector (for human shSUV39H1

TransIT 293 (Mirus, Biomedex) in Optimem (GIBCO). 2.6µg of 1265 pSIV3 and 0.4µg of pCMV-VSV-G were transfected into 293FT cells to generate SIV-VLPs

Twelve to fourteen hours later, the medium was changed to 3 mls of MDDC culture medium 1267 without cytokines. 30-32 hours later, supernatants were harvested, passaged over a 0.45µm 1268 filter, and used immediately for transduction of freshly isolated CD14 + monocytes at 1.5x10 6 1269 cells/well of a 6-well plate in 1ml of MDDC culture medium. 1 ml of each lentiviral particles and 1270 SIV-VLPS were added to each well for 3mls final

Cells were harvested on day 4 of culture, plated at 50,000 cells per well of a 1273 96-well round bottom plate, and stimulated overnight with the indicated PAMPs. 1274 1275 Luciferase assay 1276 Production of type-I IFN production in RAW264.7 macrophages was measured by production of 1277 secreted luciferase under the control of the interferon-inducible ISG54 promoter. Briefly, 1278 conditioned supernatants were collected after stimulation, and 10µl was quantified for secreted 1279 luciferase (Renilla) activity in the presence of luciferase substrate

Luminescence was recorded on a FLUOstar OPTIMA microplate reader (BMG 1281 Labtech). 1282 1283 IFN ELISA 1284 IFNb in conditioned supernatants was measured with the VeriKine Mouse Interferon Beta ELISA 1285

RNA was separated from DNA and protein by chloroform extraction. Briefly, 100µl 1291 chloroform was added to 1 ml of Trizol, mixed by shaking for 1 min and incubated for 5 minutes 1292 at RT. After centrifugation (20,000 x g for 25 minutes at 4°C), 400µl of the RNA-containing 1293 aqueous phase was removed and combined with 1µg of RNAse-free glycogen (ThermoFischer) 1294 and 500µg ice-cold isopropanol. The mixture was frozen overnight at -80°C and thawed on ice 1295 before centrifugation (20,000 x g for 25 minutes at 4°C). The RNA pellet was air-dried for 30-1296 60minutes in a laminar hood, resuspended in 10µl RNAse/DNAse-free water, and the DNA 1297 removed by in-solution DNAse digestion (Qiagen's RNase-Free DNase Set, #Cat 79254) for 30 1298 minutes at room-temperature

All Bigwig files were generated with merged biological replicates to improve the sensitivity by 1433 increasing the depth of read coverage and normalized over own input ChIP-seq files using

For heatmap 1436 visualization were generated with --PlotHeatmap after to compute the matrix of scores per 1437 genome regions generated by the tool --ComputeMatrix from deepTools Software

H3K9me3 Suv39h1-dependent elements in WT vs KO (Figure 1C) were selected using bedtools 1439 intersect (default parameters) with WT peak and subtract (-A) with KO peak

They were incubated on ice for 30 minutes, vortexed at high speed for 15 seconds 1449 and then centrifuged for 10 minutes at 13,000xg, aliquoted, snap frozen on liquid nitrogen and 1450 stored at -80°C until further processing. Samples were thawed on ice and denatured in Laemmli 1451 sample buffer supplemented with fresh beta-mercaptoethanol for 15 minutes at 95°C. For 1452 phosphorylated and corresponding total protein levels including IRF3 and STING, cells were 1453 lysed in ice cold RIPA buffer

Lysates were incubated for 30 minutes on 1456 ice, vortexed for 15 seconds at high speed, and centrifuged for 8 minutes at 8,000 x g. Clarified 1457 lystates were aliquoted, snap frozen on liquid nitrogen and stored at -80°C. Samples were 1458 thawed on ice and denatured in Laemmli sample buffer supplemented with fresh beta-1459 mercaptoethanol for 15 minutes at 95°C

Free gels (BioRad) and dry transferred to PVDF membranes (Bio-Rad) with 1461 the Trans-Blot Turbo Transfer System (Bio-Rad). In experiments where phosphor-STING was 1462 measured, lysate from 8.5x10 5 cells was loaded per lane

Tween20 (TBST) and 5% non-fat milk (Carnation, total protein) or 10% Roche blocking reagent 1464 (phosphor-proteins) for 1hr at RT, rinsed in TBST, and then rocked overnight at 4°C in TBST 1465 5% BSA (Fraction V, 04-100-812-C, Euromedex) with primary antibodies: cGAS (1:1000; 1466 31659, CST), gp96 (1:1000; adi-spa-850-D, Enzo), Phosphor-IRF3 (Ser396)(1:1000; 29047, 1467 CST), Phosphor-STING (1:1000; CST), Phosphor-NF-KappaB p65 (Ser536)(1:1000

After 1526 AgeI digestion, PacI and AbsI sites were introduced and one of the AgeI sites preserved 1527 downstream of the SFFV promoter by polylinker ligation using the following oligos: 1528 ACCGGTTTGGGATTAATTAAAATCACCTCGAGGCAGTCCGGT and 1529 TGGCCAAACCCTAATTAATTTTAGTGGAGCTCCGTCAGGCCA. In brief, oligos were 1530 phosphorylated with T4 PNK (NEB) for 30 minutes at 37°C, denatured at 95°C for 5 minutes 1531 and then ramped down to 25°C at 5°C per minute. Ligation proceeded overnight at 16°C using a 1532 thermocycler, and bacterial transformation of NEB 10-beta competent E. coli (NEB) 1533 accomplished by heat shock at 42°C. Minipreps were confirmed by Sanger sequencing. 1534 Sequential digestion of modified pL-SFFV-RFP with AgeI and BspEI were carried out at 37°C 1535 with column purification between reactions (QIAquick, Quiagen). A mouse SUV39H1 gBlock 1536 was synthesized (IDT) and cloned into AgeI/BSpEI-digested pL-SFFV-RFP using Gibson 1537 assembly (NEB) for 1hour at 50°C. NEB10 (NEB) bacteria were transformed and selected on 1538 ampicillin plates after 14hrs at 37°C. Minipreps were grown at 30°C over night with shaking at 1539 250rpm. Plasmids constructs were confirmed by Sanger sequencing. The final SUV39H1 1540 plasmid contained a P2A signal followed by the RFP reporter gene, and an internal ribosome 1541 entry site (IRES), followed by the puromycin resistance gene

Murine cGAS cDNA , Human cGAS 161-522 E225A/D227A, and Human 1552 cGAS 161-522 C395A/C396A were cloned in frame to EGFP in pTRIP-SFFV-EGFP to obtain 1553 pTRIP-SFFV-EGFP-ms cGAS, pTRIP-SFFV-EGFP-cGAS 161-522 C395A/C396A, and pTRIP-1554 SFFV-EGFP-FLAG-cGAS 161-522 E225A/D227A. Leniviral particles were generated as 1555 described and used to transduce WT RAW264.7 macrophages. 1556 1557 Software 1558 Sequencing data was analyzed with R packages and R-Studio

Flow cytometry data was analyzed with FloJo (Tree Star) v9.9.5. Graphs and statistical analysis 1561 were generated with Prism/Graphad (Version 7). Figure layouts were constructed with 1562 Prism/Graphpad, Adobe Illustrator, and Adobe Photoshop

Toll-like Receptors and the Control of Immunity

Immune Sensing Mechanisms that Discriminate Self from 1569 Altered Self and Foreign Nucleic Acids

Interferome v2.0: an updated database of annotated interferon-1571 regulated genes

5'-Triphosphate RNA is the ligand for RIG-I

Structural basis of double-stranded RNA recognition by the RIG-I like 1575 receptor MDA5

Regulation and function of the cGAS-STING pathway of 1577 cytosolic DNA sensing

Cyclic [G(2',5')pA(3',5')p] is the metazoan second messenger produced by 1579 DNA-activated cyclic GMP-AMP synthase

Cyclic GMP-AMP synthase is a cytosolic DNA 1581 sensor that activates the type I interferon pathway

Inhibiting DNA Methylation Causes an Interferon Response in 1583 Cancer via dsRNA Including Endogenous Retroviruses

Silencing of retrotransposons by SETDB1 inhibits the interferon 1585 response in acute myeloid leukemia

Repression of Stress-Induced LINE-1 Expression Protects Cancer 1587 Cell Subpopulations from Lethal Drug Exposure

Modeling of TREX1-Dependent Autoimmune Disease using Human 1589 Stem Cells Highlights L1 Accumulation as a Source of Neuroinflammation

The HUSH complex is a gatekeeper of type I interferon through 1592 epigenetic regulation of LINE-1s

p53 cooperates with DNA methylation and a suicidal interferon 1594 response to maintain epigenetic silencing of repeats and noncoding RNAs

LINE-1 elements in structural 1598 variation and disease

Restricting retrotransposons: a review

LINE-mediated retrotransposition of marked 1602 Alu sequences

An active murine transposon family pair: 1604 retrotransposition of "master" MusD copies and ETn trans-mobilization

Systematic identification of factors for provirus silencing in embryonic 1607 stem cells

An evolutionary arms race between KRAB zinc-finger genes 1609 ZNF91/93 and SVA/L1 retrotransposons

Toll-like receptor 4 in acute viral infection: Too 1611 much of a good thing

Control of RelB during dendritic cell activation integrates canonical and non-1615 canonical NF-κB pathways

Role of adaptor TRIF in the MyD88-independent toll-like receptor 1617 signaling pathway

Cutting edge: a novel Toll/IL-1 receptor domain-containing adapter 1619 that preferentially activates the IFN-beta promoter in the Toll-like receptor signaling

Binding of the ubiquitous nuclear 1622 transcription factor YY1 to a cis regulatory sequence in the human LINE-1 transposable 1623 element

A novel active L1 1625 retrotransposon subfamily in the mouse

Suv39h-dependent H3K9me3 marks intact retrotransposons 1627 and silences LINE elements in mouse embryonic stem cells

Selective silencing of euchromatic L1s revealed by genome-wide screens 1630 for L1 regulators

The HUSH complex cooperates with TRIM28 to repress young 1632 retrotransposons and new genes

L1 drives IFN in senescent cells and promotes age-associated 1634 inflammation

LINE1 Derepression in Aged Wild-Type and SIRT6-Deficient Mice 1636 Drives Inflammation

A comprehensive approach to expression of L1 loci

Activation of individual L1 retrotransposon instances is restricted to 1640 cell-type dependent permissive loci

Single-cell RNA-seq reveals dynamic paracrine control of cellular 1642 variation

hnRNP K coordinates transcriptional silencing by SETDB1 in 1644 embryonic stem cells

DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing 1646 Viral Mimicry by Endogenous Transcripts

Cutting Edge: cGAS Is 1648 Required for Lethal Autoimmune Disease in the Trex1-Deficient Mouse Model of Aicardi-1649 Goutieres Syndrome

Inflammation-driven carcinogenesis is mediated through STING

Trex1 prevents cell-intrinsic 1653 initiation of autoimmunity

Pan-viral specificity of IFN-induced genes reveals new roles for 1655 cGAS in innate immunity

Targeting STING with covalent small-molecule inhibitors

Crude subcellular fractionation of cultured mammalian cell 1659 lines

The N-Terminal Domain of cGAS Determines Preferential Association 1661 with Centromeric DNA and Innate Immune Activation in the Nucleus

cGAS-mediated stabilization of IFI16 promotes innate signaling 1664 during herpes simplex virus infection

Tight nuclear tethering of cGAS is 1667 essential for preventing autoreactivity

Structural basis of nucleosome-dependent cGAS inhibition

Structural basis for the inhibition of cGAS by nucleosomes

Structural basis for sequestration and autoinhibition of cGAS by 1673 chromatin

Structural mechanism of cGAS inhibition by the nucleosome

The molecular basis of tight nuclear tethering and inactivation of cGAS

Breaching Self-Tolerance to Alu Duplex RNA Underlies MDA5-1679 Mediated Inflammation

Sequence-specific activation of the DNA sensor cGAS by Y-form 1681 DNA structures as found in primary HIV-1 cDNA

Structure of human cGAS 1683 reveals a conserved family of second-messenger enzymes in innate immunity

Cytosolic RNA:DNA hybrids activate the cGAS-STING axis

Phosphoinositide Interactions Position cGAS at the Plasma 1688 Membrane to Ensure Efficient Distinction between Self-and Viral DNA

The N terminus of cGAS de-1690 oligomerizes the cGAS:DNA complex and lifts the DNA size restriction of core-cGAS 1691 activity

Nonspecific DNA Binding of cGAS N Terminus Promotes cGAS Activation

Parkin and PINK1 mitigate STING-induced inflammation

Structural mechanism of cytosolic DNA sensing by cGAS

Enabling large-scale genome editing by reducing DNA nicking

Editing the epigenome: 1703 technologies for programmable transcription and epigenetic modulation

SARS-CoV-2 infection induces a pro-inflammatory cytokine 1706 response through cGAS-STING and NF-κB. bioRxiv

cGAS and STING: At the intersection of DNA and RNA virus-1708 sensing networks

An influenza virus-triggered SUMO switch orchestrates co-opted 1710 endogenous retroviruses to stimulate host antiviral immunity

Transposable elements as sensors of SARS-1714

CoV-2 infection. bioRxiv

Roles for retrotransposon insertions in human 1716 disease

Transposable elements drive widespread expression of oncogenes in 1718 human cancers

Site-specific human 1722 histone H3 methylation stability: fast K4me3 turnover

An autoimmune disease prevented by anti-1724 retroviral drugs

Reverse-Transcriptase Inhibitors in the Aicardi-Goutieres Syndrome. N 1726

Human LINE retrotransposons generate 1728 processed pseudogenes

Distinct 1730 mechanisms for trans-mediated mobilization of cellular RNAs by the LINE-1 reverse 1731 transcriptase

Multiple fates of L1 retrotransposition 1733 intermediates in cultured human cells

Retrotransposition of marked SVA elements by human L1s in cultured cells

Human L1 retrotransposition: cis preference versus trans 1738 complementation

Structural basis for nucleosome-mediated 1740 inhibition of cGAS activity

Mitochondrial DNA stress primes the antiviral innate immune response

Viral unmasking of cellular 5S rRNA pseudogene transcripts induces 1744 RIG-I-mediated immunity

Endogenous retroviruses promote homeostatic and 1746 inflammatory responses to the microbiota

The cGAS-STING Defense Pathway and Its Counteraction by 1748 Viruses

Activation of cGAS/STING pathway upon paramyxovirus infection

Unique and complementary suppression of cGAS-STING and RNA 1762 sensing-triggered innate immune responses by SARS-CoV-2 proteins. Signal 1763 transduction and targeted therapy

The papain-like protease of porcine epidemic diarrhea virus 1766 negatively regulates type I interferon pathway by acting as a viral deubiquitinase

Dengue virus NS2B protein targets cGAS for degradation and prevents 1769 mitochondrial DNA sensing during infection

STING regulates intracellular DNA-mediated, type I 1771 interferon-dependent innate immunity

Zika virus elicits inflammation to evade antiviral response by cleaving 1773 cGAS via NS1-caspase-1 axis

Hepatitis C virus NS4B blocks the interaction of STING and TBK1 to 1775 evade host innate immunity

Molecular and genetic 1777 properties of tumors associated with local immune cytolytic activity

Epigenetic therapy induces transcription of inverted SINEs and 1780 ADAR1 dependency

FBXO44 promotes DNA replication-coupled repetitive element silencing 1782 in cancer cells

LSD1 Ablation Stimulates Anti-tumor Immunity and Enables Checkpoint 1784 Blockade

Epigenetic silencing by SETDB1 suppresses tumour intrinsic 1786 immunogenicity

Suppression of STING signaling through epigenetic silencing and 1788 missense mutation impedes DNA damage mediated cytokine production

cGAS-STING cytosolic DNA sensing pathway is suppressed by JAK2-1791 STAT3 in tumor cells

SUV39H1 Reduction Is Implicated in Abnormal Inflammation in COPD

Characterization of murine 1795 dendritic cell line JAWS II and primary bone marrow-derived dendritic cells in Chlamydia 1796 muridarum antigen presentation and induction of protective immunity

Suv39h1KO (n=4) BMDCs, and right, the corresponding H3K9me3 peak status from H3K9me3 1033ChIPseq data (WT, n=3; KO, n=3): red, H3K9me3 peak positive, grey, no H3K9me3 peak 1034 detected. Several genes along with Ifnb1 are designated in bold to the right as exemplars of 1035 ISGs with a grey or red box indicating H3K9me3 peak status. Statistical differences in panels A, 1036 B, and E were determined by two-way ANOVA, followed by Bonferroni's ad hoc analysis for 1037 individual comparison between groups:*P<0.05, ** P <0.001, ***P<0.001, ****P<0.0001. In I, a 1038 two-tailed student's t-test was performed to determine statistical significance (P < 0.05).