Mutations in the non-catalytic polyproline motif destabilize TREX1 and amplify cGAS-STING signaling

The cGAS-STING pathway detects cytosolic DNA and activates a signaling cascade that results in a type I interferon (IFN) response. The endoplasmic reticulum (ER)-associated exonuclease TREX1 suppresses cGAS-STING by eliminating DNA from the cytosol. Mutations that compromise TREX1 function are linked to autoinflammatory disorders, including systemic lupus erythematosus (SLE) and Aicardi-Goutières syndrome (AGS). Despite key roles in regulating cGAS-STING and suppressing excessive inflammation, the impact of many disease-associated TREX1 mutations - particularly those outside of the core catalytic domains - remains poorly understood. Here, we characterize a recessive AGS-linked TREX1 P61Q mutation occurring within the poorly characterized polyproline helix (PPII) motif. In keeping with its position outside of the catalytic core or ER targeting motifs, neither the P61Q mutation, nor aggregate proline-to-alanine PPII mutation, disrupt TREX1 exonuclease activity, subcellular localization, or cGAS-STING regulation in overexpression systems. Introducing targeted mutations into the endogenous TREX1 locus revealed that PPII mutations destabilize the protein, resulting in impaired exonuclease activity and unrestrained cGAS-STING activation. Overall, these results demonstrate that TREX1 PPII mutations, including P61Q, impair proper immune regulation and lead to autoimmune disease through TREX1 destabilization.


ABSTRACT
The cGAS-STING pathway detects cytosolic DNA and activates a signaling cascade that results in a type I interferon (IFN) response.The endoplasmic reticulum (ER)-associated exonuclease TREX1 suppresses cGAS-STING by eliminating DNA from the cytosol.Mutations that compromise TREX1 function are linked to autoinflammatory disorders, including systemic lupus erythematosus (SLE) and Aicardi-Goutières syndrome (AGS).Despite key roles in regulating cGAS-STING and suppressing excessive inflammation, the impact of many disease-associated TREX1 mutations -particularly those outside of the core catalytic domains -remains poorly understood.Here, we characterize a recessive AGS-linked TREX1 P61Q mutation occurring within the poorly characterized polyproline helix (PPII) motif.In keeping with its position outside of the catalytic core or ER targeting motifs, neither the P61Q mutation, nor aggregate proline-toalanine PPII mutation, disrupt TREX1 exonuclease activity, subcellular localization, or cGAS-STING regulation in overexpression systems.Introducing targeted mutations into the endogenous TREX1 locus revealed that PPII mutations destabilize the protein, resulting in impaired exonuclease activity and unrestrained cGAS-STING activation.Overall, these results demonstrate that TREX1 PPII mutations, including P61Q, impair proper immune regulation and lead to autoimmune disease through TREX1 destabilization.

INTRODUCTION
Type I interferonopathies, such as the monogenic disease Aicardi-Goutières syndrome (AGS), often involve chronic systemic and neurological autoinflammation and high levels of type I interferon (IFN) activity in the blood and cerebrospinal fluid (Crow and Stetson, 2022).AGS can result from loss-of-function (or specific dominant-negative) mutations in TREX1, RNASEH2A, RNASEH2B, RNASEH2C, SAMHD1, and ADAR1, gain-of-function mutations in IFIH1 (Lehtinen et al., 2008;Rice et al., 2007a).Mutations in TREX1 are among the most common in AGS, accounting for nearly one-quarter of all AGS-linked mutations (Crow et al., 2015;Rice et al., 2007b).
Mouse models of TREX1 dysfunction recapitulate hallmarks of AGS and related disorders, including familial chilblain lupus.Trex1-deficient mice exhibit multi-organ inflammation and decreased survival (Grieves et al., 2015;Stetson et al., 2008).Replacement of the wild-type Trex1 gene in mice with the nuclease-deficient Trex1 D18N mutant results in a lupus-like disease (Grieves et al., 2015).The health and viability of Trex1-deficient animals are restored by deletion of Cgas, Sting1, Irf3 and Ifnar, indicating that unchecked DNA sensing is responsible for the observed pathologies (Ablasser et al., 2014;Ahn et al., 2014;Gao et al., 2015;Gray et al., 2015;Stetson et al., 2008).Specific dominant-negative mutations in TREX1 include D18N, D200N, and H195Y, which disrupt key catalytic residues, and more frequently observed recessive alleles, e.g.R114H and R97H, that occur in the dimerization surface and hinder requisite homodimerization of TREX1 (Lehtinen et al., 2008;Rice et al., 2015).Other less-common mutations have been proposed to impede TREX1 function by altering phase separation or by destabilizing the protein (Zhou et al., 2022(Zhou et al., , 2021)).The mechanisms associated with many disease-linked TREX1 mutations are poorly understood.
Outside of its catalytic core, TREX1 possesses a single-pass transmembrane helix at its C-terminus that anchors the protein in the ER and positions the nuclease domain in the cytosol (Lee-Kirsch et al., 2007;Mazur and Perrino, 2001;Mohr et al., 2021;Wolf et al., 2016).Deleting this C-terminal extension ablates TREX1 ER localization but does not affect its catalytic activity (De Silva et al., 2007;Lee-Kirsch et al., 2007).TREX1 mutations that truncate the C-terminus disrupt TREX1-ER association while preserving nucleolytic activity, and are associated with a distinct clinical disease referred to as retinal vasculopathy with cerebral leukoencephalopathy (RVCL) (Crow and Manel, 2015;Yan, 2017).RVCL is inherited in an autosomal dominant manner and lacks clear links to excessive type I IFN production (Rodero et al., 2017).
The non-repetitive proline-rich region termed the polyproline II helix (PPII) is another unique motif present in TREX1, but not found in other nucleases within the larger DnaQ family, including the closely related TREX2 homolog (Brucet et al., 2007;De Silva et al., 2007).Like the TREX1 C-terminal extension, the positioning of the PPII helix distal to the TREX1 active site and its absence from the otherwise closely related, catalytically active TREX2 nuclease suggest that it is also unlikely to participate in catalysis or DNA binding.The functional significance of this domain is not known.
Here, we report that TREX1 P61Q mutations located in the PPII motif are linked with AGS and show how these mutations destabilize TREX1 without directly affecting nucleolytic activity or subcellular localization.We demonstrate that TREX1 P61Q instability causes overactive cGAS-STING signaling, ultimately resulting in cGAMP overproduction and excessive levels of type I IFN expression.Thus, these results indicate that the TREX1 P61Q mutations cause AGS through TREX1 protein destabilization and suggest that protein destabilization may account for a subset of AGS patients with TREX1 mutations.
Since R114H renders the allele null by disrupting obligate dimerization of TREX1 (Lehtinen et al., 2008), these findings suggest a recessive, loss-of-function nature of the P61Q mutation.Indeed, calculation of IFN scores, derived by measuring the expression of six IFN stimulated genes (ISGs) using quantitative polymerase chain reaction, revealed a significant upregulation of IFN signaling relative to persons considered to be controls, thus placing both individuals within the type I interferonopathy spectrum (Crow and Manel, 2015;Rice et al., 2017).
Pro-61 lies in a proline-rich tract termed the polyproline II (PPII) helix (Fig. 1A,B) (Brucet et al., 2007;De Silva et al., 2007).PPII positioning distal to the TREX1 active site and its absence from other catalytically proficient enzymes of the DnaQ family, including TREX2, suggest it is likely to be dispensable for nucleolytic activity.To test this directly, we purified the N-terminal enzymatic domain of TREX1 proteins, including human TREX1, a TREX1 P61Q mutant, and a TREX1 PPII>β-hairpin chimera, in which the TREX1 PPII helix is replaced by the β-hairpin found in the corresponding position within TREX2 (Fig. 1B).As expected, in vitro nuclease assays using purified proteins demonstrated that TREX1 P61Q and β-hairpin mutants digested dsDNA with efficiencies comparable to the wild-type enzyme with >50% of substrate degraded within the first 5 minutes of incubation (Fig. 1C,D).
To further investigate the potential impact of TREX1 PPII mutations we assayed TREX1 exonuclease activity in cell lysates.In brief, lysates were incubated with a dsDNA substrate possessing a fluorescent label at one 5′ end closely positioned next to a 3′ quencher (Methods).TREX1 3′→5′ exonuclease activity is predicted to liberate the fluorescent dye from the 3′ quencher and thus result in the acquisition of fluorescence.Cell lysates were prepared from TREX1-deficient MCF10A cells stably transduced with GFP-TREX1-WT, GFP-TREX1-P61Q, and GFP-TREX1-8PA, in which eight prolines in PPII -excluding P61 -are mutated to alanine (Fig. 1B).Lentiviral transduction of these constructs into TREX1-deficient MCF10A cells yielded stable overexpression of GFP-tagged mutant proteins, with no significant differences in protein levels between the three genotypes (Fig. S1A and S1B).As expected, incubation of the dsDNA probe with lysates prepared from MCF10A TREX1 KO cells reconstituted with GFP-TREX1-WT resulted in the rapid acquisition of fluorescence (Fig. 1E,F).In contrast, TREX1 deletion severely diminished the acquisition of fluorescence, confirming the specificity of this assay for TREX1 exonuclease activity (Fig. 1E,F).Similar to results obtained using isolated proteins, measurement of GFP-TREX1-8PA and GFP-TREX1-P61Q activities exhibited no significant differences from GFP-TREX1-WT (Fig. 1E,F).Taken together, these data indicate that targeted mutations within the PPII helix do not directly interfere with TREX1 exonuclease activity and suggest that the PPII helix is dispensable for TREX1 exonuclease activity against dsDNA.
We previously demonstrated that TREX1 association with the ER is critical for processing a subset of cytosolic DNA substrates including nuclear aberrations like micronuclei (Mohr et al., 2021).Positioning of the PPII within the catalytic core and distal to the ER transmembrane domain at the C-terminus of TREX1 suggested that the PPII domain is likely dispensable for ER association.To test this possibility directly, we performed live-cell imaging of cells overexpressing GFP-TREX1 mutants to characterize their subcellular localization.As previously reported (Mohr et al., 2021;Stetson et al., 2008;Wolf et al., 2016), GFP-TREX1-WT was excluded from the nucleus and its localization significantly overlapped with the ER, as indicated by staining with an ER tracker dye (Fig. 1G,H).GFP-TREX1-8PA and GFP-TREX1-P61Q subcellular localizations could not be distinguished from that of the wild-type enzyme, suggesting that the PPII is dispensable for directing TREX1 ER association (Fig. 1G,H; Figure S1C,D).Together, these data indicate that PPII mutations are unlikely to cause TREX1 dysfunction by interfering with its ER localization.

Overexpressed TREX1 PPII mutants suppress cGAS-STING signaling
To test whether PPII mutations affect cGAS activation, we quantified intracellular cGAMP via ELISA (Fig. 2A).MCF10A cells lack high levels of cytosolic DNA and do not show strong cGAS activity at baseline, even upon TREX1 deletion (Mohr et al., 2021;Zhou et al., 2021).We therefore stimulated cGAS activation by herring testes (HT-) DNA transfection.
We next sought to determine if TREX1 PPII mutations impacted the downstream cGAS-STING response by using RT-qPCR to measure expression of IFNB1 and interferon-stimulated genes (ISGs) such as OAS2, OAS3, ISG54, and ISG56 (Fig. 2B-F).As expected, RT-qPCR revealed strong increases in IFNB1 and ISG mRNA levels in TREX1 KO MCF10A cells upon HT-DNA stimulation relative to parental controls (Fig. 2B-F).In line with our cGAMP ELISA results, GFP-TREX1-WT, GFP-TREX1-8PA and GFP-TREX1-P61Q suppressed IFNB1 and ISG expression to similar degrees upon overexpression in TREX1 KO cells (Fig. 2B-F).Thus, counter to expectations based on the association between the TREX1 P61Q mutations and AGS (Fig. 1A), these results indicate that TREX1 PPII mutants are functionally proficient to suppress cGAS activation and downstream ISG expression upon overexpression in MCF10A cells.

TREX1 PPII mutations destabilize the protein
We reasoned that strong TREX1 overexpression resulting from lentiviral delivery (Fig. S1A and S1B) may obscure defects associated with PPII mutation.We therefore used CRISPR-Cas9 gene editing to endogenously introduce an N-terminal HaloTag concurrently with a PPII edit-proline-to-alanine mutation of all nine prolines in PPII (9PA) or P61Q -into the diploid MCF10A cell line (Fig. S2A).The remaining, unedited allele was deleted, yielding Halo-TREX1/Δ genotypes for all subsequent experiments (Fig. S2A).All gene edits were validated by Sanger sequencing and PCR screening (Fig. S2B-D).Immunoblotting with anti-TREX1 antibodies further confirmed successful insertion of the HaloTag into the endogenous TREX1 locus.(Fig. 3A).
Similar to our prior results from GFP-TREX1 overexpression (Fig. 1E and 1F), all Halo-TREX1 lysates retained the ability to digest dsDNA (Fig. 3E and 3F).However, fluorescence increased at a much slower rate in Halo-TREX1-9PA/∆ and Halo-TREX1-P61Q/∆ lysates than Halo-TREX1-wild-type/∆ lysates, with the area under curve values decreased about two-fold.Thus, TREX1-P61Q and TREX1-9PA mutations lead to significant reductions in protein levels that are associated with corresponding decreases in nucleolytic activity.
Observed reductions in TREX1-9PA and TREX1-P61Q protein levels and activity could not be explained by reduced TREX1 mRNA expression (Fig. S3C).Instead, Thermofluor analysis of purified TREX1 and TREX1-P61Q proteins demonstrated a significant 13.5ºC difference in protein stability with TREX1 exhibiting a melting temperature (Tm) of 51ºC and TREX1-P61Q exhibiting a Tm of 37.5ºC (Fig. 3G).These results indicate that TREX1 PPII mutations destabilize the protein, and thus lead to reduced overall protein levels with 244 corresponding decreases in nucleolytic activity.245   Following cGAS-STING activation, TBK1 phosphorylates the transcription factor IRF3 at multiple residues including S386 and S396, inducing IRF3 dimerization and transcription of type I IFN (Liu et al., 2015).Increased type I IFN signaling results in the phosphorylation and activation of STAT1 (pY701)/STAT2 heterodimers, ultimately culminating in the transactivation of a wide-ranging pro-inflammatory response (Galluzzi et al., 2018).We therefore immunoblotted for phospho-IRF3 (pS386) and phospho-STAT1 (pY701) to assess cGAS-STING signaling downstream of cGAMP production (Fig. 4B-D).Consistent with prior work (Mohr et al., 2021), TREX1 KO cells exhibited significant increases in the phosphorylated forms of IRF3 and STAT1 following HT-DNA stimulation relative to wild-type Halo-TREX1 controls (Fig. 4B-D).Congruent with the observed increase in cGAMP levels, Halo-TREX1-9PA and Halo-TREX1-P61Q mutant cells exhibited increased levels of pIRF3 and pSTAT1 compared to wild-type controls, albeit to a lesser extent than TREX1 KO cells (Fig. 4B-D).
We next performed RT-qPCR to measure IFNB1 and associated ISG mRNA levels to test if increases in cGAMP, and IRF3, and STAT1 phosphorylation are associated with elevated pro-inflammatory gene expression.Indeed, IFNB1, OAS2, OAS3, ISG54, and ISG56 transcripts were elevated across multiple Halo-TREX1-9PA and Halo-TREX1-P61Q mutant subclones relative to wild-type controls to levels that were often indistinguishable from TREX1 KO cells (Fig. 4E-I).Taken together, these results indicate that TREX1 PPII mutations result in defective cGAS regulation and an increased pro-inflammatory transcriptional response, defects most likely stemming from TREX1 protein instability and associated reductions in overall TREX1 protein levels and corresponding decreases in nucleolytic activity.

DISCUSSION
Genetic associations of type I interferonopathies like AGS have been well-characterized, particularly in cases involving TREX1 mutations linked with compromised catalytic activity (Crow and Manel, 2015).Yet, how missense mutations outside of the catalytic site can lead to inflammatory disease has often remained unclear.Here, we identify an AGS-linked P61Q point mutation within the non-catalytic PPII motif of TREX1.Using in vitro biochemical measures of protein stability and endogenous gene editing, we show that TREX1 PPII mutations, including P61Q, destabilize the protein, resulting in significantly decreased TREX1 protein levels, diminished TREX1 exonucleolytic activity, and impaired cGAS-STING regulation.These defects were obscured in lentiviral delivery models where massive overexpression of TREX1 PPII mutants masks reductions in protein stability to maintain effective cGAS inhibition.The distal position of PPII to the catalytic site, along with the lack of differences in the GFP-TREX1 lysatebased nuclease assay, suggests that the nucleolytic defect observed in the endogenous system is due to decreased protein levels, rather than a direct effect of the mutations on catalysis.Thus, our results indicate diminished protein stability and an associated reduction in overall nucleolytic power of TREX1 as a plausible molecular explanation for why TREX1 P61Q mutations lead to severe AGS phenotypes in patients.
Autoinflammatory disease-linked TREX1 missense mutations often affect residues that play direct roles in DNA binding (i.e.R128H, K160R), catalytic activity (D18N/H, H195Y/Q, D200H/N) or dimerization (R97H, R114H).We recently reported that TREX1 mutations may also cause dysfunction by interfering with TREX1 interactions with cGAS-DNA condensates (E198K) (Zhou et al., 2021).Here, the identification of AGS-linked TREX1 P61Q mutations suggests that another class of mutations may compromise TREX1 function by diminishing overall protein stability.Indeed, structural analyses predict that the disease-linked TREX1 T13N, T32R, R185C, and D220G substitutions are likely to diminish protein stability (Zhou et al., 2022).Biochemical experimentation supports this premise as TREX1 T13N, T32R, R185C, and D220G substituted proteins exhibit Tm reductions of 4-8 ºC in vitro (Zhou et al., 2022).Thus, TREX1 protein destabilization may be a common defect occurring across multiple AGS-linked TREX1 mutations.TREX1 P61 is located with the PPII polyproline helix, a proline-rich region containing 9 prolines within a 15 amino acid stretch (Brucet et al., 2007;De Silva et al., 2007).This type of proline-rich segment is a conserved feature of TREX1, as it occurs in all organisms harboring TREX1, including placental mammals and marsupials.The paralog TREX2, as well as the ancient TREX nuclease occurring in non-mammals such as Anopheles and Drosophila, lack a proline-rich motif (Brucet et al., 2007), indicating that PPII likely evolved during the gene duplication event.Interestingly, the emergence of PPII in evolution seems to have coincided with the addition of a long C-terminal intrinsically disordered region.In keeping with structurebased predictions based on the PPII positioning outside of the catalytic core and ER transmembrane domains, our data confirm that the PPII motif is dispensable for TREX1 nucleolytic activity and subcellular localization.The precise function of the PPII motif therefore remains unknown.
The close positioning of the two PPII helices along the same side of the TREX1 dimer interface has been proposed to create a surface that allows for protein-protein interactions without occluding the active sites (Brucet et al., 2007;De Silva et al., 2007).Indeed, their high potential for presenting exposed hydrogen bond donors and acceptors, cause proline-rich motifs to be considered likely protein interaction domains (Adzhubei et al., 2013).The amino acid sequence of PPII matches the binding motif for the WW domain (Brucet et al., 2007), a peptide module characterized by two tryptophan residues (Sudol et al., 1995).Co-immunoprecipitation experiments have previously confirmed that murine TREX1 PPII interacts with the WW domain protein CA150 in vitro (Brucet et al., 2007).Whether human TREX1 also interacts with WW domain proteins and endogenous interactors remains unknown.Outside of a proposed interaction with the nucleosome assembly SET protein (Chowdhury et al., 2006), TREX1 protein partners are largely uncharacterized.Further work is therefore necessary to investigate this exciting hypothesis.
Our study relies heavily on the N-terminal HaloTag for studying the behavior of endogenous TREX1 PPII mutations.We observed an apparent stabilizing effect of the HaloTag on TREX1, as Halo-TREX1(WT)/Δ yielded a stronger immunoblot signal than parental cells (data not shown).This observation is consistent with a prior report, which demonstrated that HaloTags can elicit a significant impact on the detection of proteins by Western blot (Broadbent et al., 2023).Apparent increases of HaloTag protein levels were attributed to enhanced western blot transfer efficiency (Broadbent et al., 2023).Therefore, western blotting analysis may underestimate the full extent of TREX1 P61Q protein instability.A further potential limitation of our study is the use of the non-malignant MCF10A breast epithelial cell line to model AGSlinked TREX1 mutations.MCF10A cells were selected for this study because they possess an intact cGAS-STING-TREX1 pathway (Mohr et al., 2021) and are suitable for facile gene editing.However, it is not clear how well this cell model recapitulates aspects of AGS, a disease that primarily affects the central nervous system.Nevertheless, orthogonal measurements of TREX1 P61Q stability via Thermofluor analysis provide assurance that the P61Q mutation is likely to exert a destabilizing effect across multiple cell types and thus reinforce our proposed mechanism of pathogenesis in patients harboring the TREX1 P61Q mutation.

Nuclease Assay with Recombinant TREX1
In vitro DNA degradation assay was performed as previously described with minor modifications (Zhou et al., 2022).Briefly, 1 μM 100-bp dsDNA (see below for sequence) was incubated with 0.1 μM human TREX1 or TREX1 variants in a 20 μL reaction system (20 mM Tris-HCl pH 7.5, 15 mM NaCl, 135 mM KCl, 5 mM MgCl2, and 1 mg/ml BSA) at 25°C with a time gradient of 5-30 min.DNA degradation was quenched by adding SDS (final concentration at 0.0167% (w/v)) and EDTA (final concentration at 10 mM) and incubating at 75°C for 15 min.The remaining DNA was separated on a 4% agarose gel using 0.5 × TB buffer (45 mM Tris, 45 mM boric acid) as a running buffer.After DNA electrophoresis, the agarose gel was stained with 0.5x TB buffer (containing 10 μg/mL ethidium bromide) at 25°C for 15 min, followed by de-staining with milli-Q water for an additional 45 min.DNA was visualized by ImageQuant 800 Imaging System and quantified using FIJI (Schindelin et al., 2012).Whole cell lysates were generated by resuspending 3 million cells in 80 μL of assay buffer containing 25 mM HEPES 7.5, 20 mM KCl, 1 mM DTT, 1% Triton X-100, 0.25 mM EDTA, and 10 mM MgCl2 supplemented with Complete Mini Protease Inhibitor Cocktail (Invitrogen #11836153001).Cells were lysed by passing the cell resuspension through a 28 G syringe (BD #329461) ten times, incubated on ice for 15 minutes, and then were spun down at 14,000 ✕ g, 4ºC for 15 minutes to remove pellets.1:10 dilution of whole cell lysates in assay buffer were used to quantify protein content using Reducing Agent-compatible Pierce BCA Assay Kit (Thermo Fisher Scientific #23250).

Live-cell Imaging
Cells were plated onto 4-well glass-bottom μ-slide dishes (Ibidi #80427) 24 h before imaging.Five minutes before imaging, media in each well was replaced with FluoroBrite DMEM Imaging Media (Thermo Scientific #A1896701) containing 1 μM ER Tracker Red (Thermo Scientific #E34250) or ER Tracker Green (Thermo Scientific #E34251).Live-cell imaging was performed at room temperature using Nikon Eclipse Ti2-E equipped with CSU-W1 SoRa
Primary antibodies were diluted (1:4000 for β-actin, 1:1000 for all others) in Intercept T20 (TBS) Antibody Diluent (LI-COR #927-65001) and incubated with membranes overnight at 4 ºC on a nutator.Membranes were washed three times in TBST.Secondary antibodies were diluted 1:10,000 in Intercept T20 (TBS) Antibody Diluent and incubated for 1 hour at room temperature on a shaker.After three rounds of washing with TBST and one round of washing with TBS, membranes were scanned using the Odyssey XL infrared imaging scanner (LI-COR).

RT-qPCR
Total RNA was isolated from 1 million cells using Quick RNA Miniprep Kit (Zymo Research #R1055) according to the manufacturer's instructions.A DNase I digestion step was included prior to eluting the RNA.cDNA was generated from 1000 ng total RNA using the SuperScript IV First-strand Synthesis System (Invitrogen #18091200) with random hexamer and oligo-(dT) priming.Reverse-transcribed samples were treated with RNase H to remove RNA.qPCR was performed with gene-specific primers (see Key Resources Table ) and SYBR Green qPCR Master Mix (Applied Biosystems #A25742).qPCR was performed on QuantStudio 6 (Applied Biosystems), using 10 ng of cDNA and 250 nM of each primer on a MicroAmp 384-well reaction plate (Applied Biosystems #4309849).Relative transcription levels were calculated by normalizing to the geometric mean of ACTB and GAPDH cycle threshold values.

Thermal Denaturation Assay
10 µM of purified TREX1 mutant protein and 3✕ SYPRO Orange Protein dye (Life Technologies) were loaded into a 96-well reaction plate, in a 20 µL reaction containing 20 mM Tris-HCl pH 7.5, 75 mM KCl, and 1 mM TCEP.Reactions were incubated with an increasing temperature from 20 to 95º C in a Bio-Rad CFX thermocycler with HEX channel fluorescence measurements taken every 0.5º C, and melting temperature (Tm) was defined as the temperature at which the half of the maximum fluorescence change occurs.

Statistical Analysis
Information regarding biological replicates, sample size, and statistical testing is provided in the figure legends.

Figure 1 .
Figure 1.Mutations in PPII are linked to AGS but do not compromise intrinsic functions of overexpressed TREX1. A. Location of PPII and P61 (orange) within TREX1.Genotypes of

Figure 2 .
Figure 2. Overexpressed TREX1 mutants can suppress cGAS-STING signaling.A. ELISA analysis of cGAMP production in the indicated cells following the transfection of 4 μg HT-DNA;

Figure S2 .
Figure S2.Generation of TREX1 knock-in mutations.A. Representative schematic of TREX1 gene editing protocol.Briefly, an N-terminal sgRNA and a HaloTag donor plasmid harboring a

Figure S3 .
Figure S3.Mutations in PPII do not interfere with TREX1 transcription or localization.A. Pearson correlation coefficients of the indicated cells as in Fig. 3C; mean ± s.d., n = 5