IFI207, a young and fast‐evolving protein, controls retroviral replication via the STING pathway

ABSTRACT Mammalian AIM-2-like receptor (ALR) proteins bind nucleic acids and initiate production of type I interferons or inflammasome assembly, thereby contributing to host innate immunity. In mice, the Alr locus is highly polymorphic at the sequence and copy number level, and we show here that it is one of the most dynamic regions of the genome. One rapidly evolving gene within this region, Ifi207, was introduced to the Mus genome by gene conversion or an unequal recombination event a few million years ago. Ifi207 has a large, distinctive repeat region that differs in sequence and length among Mus species and even closely related inbred Mus musculus strains. We show that IFI207 controls murine leukemia virus (MLV) infection in vivo and that it plays a role in the STING-mediated response to cGAMP, dsDNA, DMXXA, and MLV. IFI207 binds to STING, and inclusion of its repeat region appears to stabilize STING protein. The Alr locus and Ifi207 provide a clear example of the evolutionary innovation of gene function, possibly as a result of host–pathogen co-evolution. IMPORTANCE The Red Queen hypothesis predicts that the arms race between pathogens and the host may accelerate evolution of both sides, and therefore causes higher diversity in virulence factors and immune-related proteins, respectively . The Alr gene family in mice has undergone rapid evolution in the last few million years and includes the creation of two novel members, MndaL and Ifi207. Ifi207, in particular, became highly divergent, with significant genetic changes between highly related inbred mice. IFI207 protein acts in the STING pathway and contributes to anti-retroviral resistance via a novel mechanism. The data show that under the pressure of host–pathogen coevolution in a dynamic locus, gene conversion and recombination between gene family members creates new genes with novel and essential functions that play diverse roles in biological processes.

Here, we show that the Alr locus in Mus species has undergone rapid evolution for at least the past few million years, and that one gene in the locus, Ifi207 (PyhinA), is highly divergent even among closely related inbred mouse strains.We found that IFI207 is unique among the ALRs due to a large repeat region separating the two domains common to all ALRs, the N-terminal pyrin domain (PYD) that facilitates homotypic and heterotypic interactions with PYD-containing and other proteins, and the C-terminal hematopoietic expression, interferon-inducible nature, and nuclear localization (HIN) 200 domain that binds DNA (4,9,27).We found that the number of repeats is variable among inbred wild and laboratory strains of mice, ranging in number from 11 to 25. Due to the evidence of pathogen-driven positive selection on the Alr locus and the evolutionary diversity of Ifi207, we examined the role of IFI207 in innate immune responses.We demonstrate that IFI207 binds to STING protein and stabilizes it via its repeat region.This binding enhances the STING response and contributes to the control of MLV infection in vivo.

Ifi207 is a young gene exclusively found in the Mus genus
We previously reported that the Alr locus is highly variable, with different numbers and complements of Alr genes mapping to the locus in three inbred mouse strains, C57BL/6, DBA/2J, and 129P2/OlaHsd (11).The locus in the GRCm39 reference genome (strain C57BL/6J) encodes 14 genes (Fig. 1A), while strain 129P2/OlaHsd encodes 16, and DBA/2J, 17 genes (11).Many members of the Alr family are similar to each other, in both the coding and non-coding regions, and are surrounded by transposon elements.To further study the genome structure of Alr locus in house mice, we accessed the long read (whole genome PacBio Sequel/Hifi + Chicago HiC and BioNano scaffolding) de novo assemblies of 13 inbred house mouse strains, Mus spretus and Mus caroli from the Mouse Genomes Project; the genome assemblies will be described in a separate study.As shown in Fig. 1A and B, based on gene structure and phylogeny, the Alr gene family consists of nine variant members in house mice and other Mus species, and the number of genes in the locus ranges from 10 (PWK/PhJ) to 17 (DBA/2J, AKR/J, CBA/J) (Fig. 1A).In inbred mouse strains, there is 1 copy of Aim2 at the 5′ end of the locus followed by 2-3 copies of Ifi206 homologs (Ifi206, Ifi213, and Ifi208 in the GRCm39 reference), 1-2 copies of Ifi214-like genes (Ifi214 and Ifi209), 1 copy of Ifi207, 1-3 copies of Ifi202, 1-4 Ifi203 homologs/alleles, 1-2 Mnda members (Ifi205 or Mnda), 1 copy of Ifi204, and possibly 1 copy of MndaL.Ifi208, Ifi207, Ifi204, and 1 copy of Ifi203 are encoded in a stable island in the middle of Alr locus, while other regions are highly dynamic in the evolutionary history of the house mouse.
In addition, we downloaded the whole genome assembly from similar long-read assemblies carried out with other rodent species.When the Mus locus is compared with other rodent species, only orthologues of Aim2, Ifi206, Ifi214, Mnda, and Ifi203 can clearly be found in the latter (Fig. 1B).All rat assemblies available on NCBI (Rattus norvegicus, strain BN/NhsdMcwi, SHRSP/BbbUtx, WKY/Bbb, SHR/Utx, SS/JrNhsd, and GK/Ox; Rattus rattus CSIRO assembly) contain exactly five Alr members.The Golden spiny mouse (genus Acomys, subfamily Deomyinae), deer mice (genus Peromyscus), and grasshopper mice (Onychomys; both from family Cricetidae) have the same genome structure as rats, except with variable copy numbers of Ifi203 and Mnda paralogs.No Ifi207, MndaL, Ifi204, and Ifi202 orthologs can be found in these species (Fig. 1B).Furthermore, we compared the de novo assembly of 29 species in the glires clade (rodents and rabbits) and investigated the whole genome Illumina reads of key species in the Murinae subfamily.Homologs of Ifi207 can only be found in the Mus genus and arguably in the species Apodemus sylvaticus (Fig. S1 and S2, Table S1; see Discussion).Indeed, the genome structure around Ifi207 (including introns) is probably derived from the 3′ end of Ifi203 and the 5′ end of Mnda, from a recent recombination event in the ancestor of the Mus genus.Additionally, the 5′ end of Ifi203 and the 3′ end from Mnda possibly fused into MndaL in the same event (Fig. S3).In comparison to rodent species, the human genome encodes AIM2 and MNDA, and two other genes with ambiguous relationship to rodent Alrs, namely IFI16 and PYHIN1 (Fig. 1B) (10).

Ifi207 is characterized by two highly polymorphic repeat regions
Most ALRs contain an N-terminal PYD domain and one or two HIN domains, termed HIN-A and HIN-B, depending on their sequence (9,10).Comparison of the amino acid sequences of IFI204, MNDA, and IFI207 demonstrated high homology in the PYRIN domain (Fig. 1C).Moreover, IFI203 and IFI207 share a long linker sequence with 72% identity between the PYRIN and HIN-B domains (Fig. 1C; Tables S2 and S3).
From the de novo assembly of the Alr locus in mice, we found two repeat regions in several genes, which are highly polymorphic between mouse strains.At around residue 120 (PYD-proximal repeat) of IFI204, IFI205, MNDA, and IFI207, there are two to six units of the sequence XXTSTA/DQ.All the repeats are encoded in the same exon, precisely seven residues as a unit, and the first two residues are highly divergent (A, E, G, R, V, S, or Q; Fig. 1C).This repeat was also found in the second exon of Mnda in other rodent species, but not in humans (Fig. S1; Tables S2 and S3).
The other repeat unit is unique to IFI207 and is located close to the HIN-B domain, with 11-25 units of the sequence TXSTXSSXLLAPXA (S/T-rich repeat).In 49 unique repeat sequences extracted from 16 distinct alleles (extracted from 15 Mouse Genome Project [MGP] sequences plus nine wild-derived mouse strains), there was up to 36% diversity (average value 15%) between their nucleotide sequence, and up to 65% difference at the amino acid residue level (29% average); this was confirmed with multiple sequencing methods (Fig. 1C; Tables S2 and S3; see Materials and Methods).Like the PYD-side repeat, all the S/T-rich HIN-side repeats are encoded in a single large exon, with precisely 14 residues (42 base pairs) in a unit.Interestingly, in IFI203, the possible origin and closest paralog of IFI207, the sequence surrounding the repeat region is conserved between IFI203 and IFI207; however, in IFI203, only one or two copies of the S/T-rich HIN-side repeats are found, and they are highly conserved between mouse strains (Fig. S2).Moreover, a single copy of the same S/T-rich sequence can also be found close to the HIN-B domain of human IFI16 (Fig. S2; Table S2).These results indicate the IFI207 long S/T-rich repeats were recently introduced and are evolving in a very short time in mouse evolution, approximately 6.5 million years ago in an ancestor of the Mus genus (28) (see Table S3 for the alignment of Ifi203 and Ifi207).
AlphaFold analysis predicts that the S/T-rich repeat region forms a coiled structure not found in the other proteins encoded in the ALR locus; shown for comparison is the predicted structure of IFI203, the most closely related ALR (Fig. 1D).IFI207 from Mus caroli is also predicted to have a similar coiled structure (Fig. S4).The extremely fast evolution of the S/T-rich repeats in Ifi207 makes it one of the most dynamic regions in the mouse genome, and indeed, in mammalian species in general.

Knockout of Ifi202b, Ifi204, or Ifi207 in mice has a minimal effect on expres sion of other genes in the locus
Mouse Alr genes, including both coding and intervening sequences, are highly homologous, making it difficult to precisely delete single genes in the locus (11).However, we were able to use CRISPR/Cas9 to delete exon 3 of the Ifi207 gene in C57BL/6N mice (Fig. S5A).Additionally, mice with floxed alleles of Ifi202b and Ifi204 on a C57BL/6N background were available through the KOMP repository.Ifi202b encodes a protein that lacks a PYD and has two HIN domains, A and B, while Ifi204 encodes a protein with PYD, HIN-A, and HIN-B domains (Fig. 1C).The replacement allele in IFI202 knockout (KO) mice deleted exon 3, and Cre-mediated deletion resulted in IFI204 KO mice lacking exons 3 and 4 (Fig. S5A; see Materials and Methods).The knockout alleles for both Ifi204 and Ifi207 generate truncated proteins predicted to contain only the PYD, while the Ifi202 knockout allele generates a protein predicted to contain a portion of the HIN-A domain and a full HIN-B domain (Fig. S5B).All three KO strains had normal viability and produced offspring.
We examined the expression of the KO alleles in bone marrow-derived macrophages (BMDMs) and dendritic cells (BMDCs) from wild-type (WT; BL/6) and knockout mice using reverse transcription quantitative PCR (RT-qPCR).Ifi202 RNA was barely detectable in both cell types from WT mice, as we previously reported (11); this was also the case in the KO mice (Fig. S6A).Ifi204 and Ifi207 RNAs were not detected in the respective KO strains (Fig. S6A).
To determine if Ifi202, Ifi204, or Ifi207 knockout affected the expression of the other genes in the locus, we examined RNA levels in BMDMs and BMDCs from WT, IFI202, IFI204, and IFI207 KO mice.BMDMs from STING mut mice were also included in these analyses.These mice are resistant to stimulation with DNA and other PAMPs as they encode a mutant STING that is unable to signal (29).STING RNA levels were equivalent in WT, IFI207 KO, and STING mut BMDMs (Fig. S6B).Because it has previously been shown that the genes in the Alr locus are type 1 IFN inducible, we also tested whether deletion of Ifi202, Ifi204, or Ifi207 affected the induction of other Alrs by IFNβ (10,30,31).Neither basal nor IFNβ-induced levels of expression of the different genes were altered in either BMDMs or BMDCs by knockout of any of the ALRs or the STING mutation, apart from Mnda; Mnda RNA levels were consistently reduced in BMDMs and BMDCs in IFI204 KO mice in the absence and presence of IFNβ (Fig. S6A ).The significance of this observation is unknown.Nonetheless, because loss of Ifi207 did not affect the expression of the other Alrs, phenotypes observed in the KO mouse are likely specific to this gene.

IFI207 is an antiviral restriction factor in mice but does not affect the cytokine response to bacteria
We showed previously that multiple activators of the STING pathway, including cGAS and DDX41, were needed to control in vivo MLV infection (17,20,32).To determine if IFI207 played a role in controlling virus infection, we infected newborn IFI207 and IFI204 KO mice, as well as mice heterozygous for the Ifi207 KO allele with MLV.STING mut mice, which we showed previously are highly susceptible to MLV infection, were infected for comparison (17,20).Splenic viral titers were measured at 16 dpi.MLV titers were significantly higher in both IFI207 KO and heterozygous mice compared to WT mice (Fig. 2A).Similar results were seen in mice infected with a mutant virus, MLV gGag , which has an unstable capsid; overall, titers were lower than in wild-type MLV-infected mice, which we showed previously was due to the increased susceptibility of MLV gGag to nucleic acid sensors and APOBEC3-mediated restriction (32).STING mut mice were more highly infected with both wild-type MLV and MLV gGag compared to IFI207 KOs, like our previous studies with knockout of cGAS and DDX41, upstream effectors of STING (17,20).IFI207 heterozygous mice, which express half as much Ifi207 RNA compared to WT mice, also showed higher levels of infection than WT mice, suggesting haploinsufficiency (Fig. 2A).Finally, although it has been suggested that IFI204 is an anti-retroviral restriction factor, IFI204 KO mice had titers equivalent to WT mice (16).These data demonstrate that IFI207 is one of several factors that control MLV infection in vivo.
A recent publication suggested that Ifi207-depleted cells had a diminished TNFα and IFNβ response to bacterial infection because of the loss of RNAPolII at the promoters of the genes (33).We also tested the response of BMDMs from IFI207 KOs to Staphylococcus aureus and Klebsiella pneumoniae.Tnf and Ifnβ1RNA levels were similar in WT, IFI204, and IFI207 KO BMDMs in response to S. aureus.Ifnβ1 but not Tnf expression was reduced in STING mut cells, as previously reported (Fig. S7A) (34).In contrast, there was no significant difference in the Tnf or Ifnβ1 response to K. pneumoniae in any of the mutant cells compared to WT (Fig. S7B).

IFI207 plays a role in STING pathway signaling
Type I interferons generated by activation of the STING pathway are needed to control in vivo infection by MLV (17,20).Several innate immune pathways can be activated to induce the production of interferon, including the TOLL-like receptor (TLR), RIG-I/MDA5, and STING pathways.To determine which pathway might be affected by loss of Ifi207, we used immortalized NR9456 macrophages in which Ifi207 levels were reduced by siRNAs, as well as primary BMDMs from WT, and IFI207 and TLR4 KO mice.In both NR9456 cells and BMDMs, Ifi207 depletion did not affect lipopolysaccharide (LPS) induction of Tnfα or Ifnβ1 RNA (Fig. S8A and B).In contrast, induction of both cytokine RNAs was ablated in BMDMs isolated from TLR4 knockout mice (Fig. S8B).As expected, because Ifi207 is itself type I IFN inducible, LPS treatment increased Ifi207 levels in the immortal ized macrophages, although to a lesser extent in the siRNA-depleted cells (Fig. S8A).LPS-treated WT, but not IFI207 KO BMDMs, had high levels of Ifi207 RNA (Fig. S6B).Similar results were seen in vivo when LPS was subcutaneously injected; WT and IFI207, but not TLR4 KO mice, had increased Tnf RNA in the draining lymph nodes (Fig. S8C).
We then examined whether IFI207 altered STING-mediated type I IFN induction.BMDMs from WT, IFI207 KO, IFI204 KO, and STING mut mice were transfected with increasing amounts of interferon stimulatory DNA (ISD) or cGAMP, the activating ligand of STING.Three hours after transfection, Ifnβ1 RNA levels were measured.The IFN response to ISD and cGAMP was significantly reduced in IFI207 KO BMDMs, while the response by WT and IFI204 KO cells was the same; STING mut cells had no response (Fig. 2B).Similar results were obtained when fibroblasts from WT, IFI207 KO, and STING mut were transfected with ISD or cGAMP (Fig. 2C).The IFN response to the STING ago nist 5,6-dimethylxanthenone-4-acetic acid (DMXAA) was also decreased in IFI207 KO fibroblasts (Fig. 2C).Thus, the presence of IFI207 in multiple cell types enhanced the STING response.

STING protein levels and signaling are decreased in the absence of IFI207
IFI207 could be acting as an upstream sensor to activate STING, or it could be directly interacting with it to enhance responses.First, we determined if reducing endogenous IFI207 levels altered basal levels of STING.Transfection of IFI207 siRNAs into in NIH3T3 cells reduced basal STING levels compared to control siRNA-transfected or untransfected cells (Fig. 3A).Similar results were seen in IFI207 KO macrophages (Fig. 3B and D), fibroblasts (Fig. 3C), and DCs (Fig. S9A) (untreated lanes).Although IFI207 KO cells had lower levels of STING protein, they had similar levels of STING RNA to WT or STING mut cells, demonstrating that IFI207 was not altering STING transcription (Fig. S5B).
We then examined the levels of phospho-STING and TBK, the immediate downstream steps in the STING activation pathway.BMDMs from IFI207 KO and WT mice were transfected with ISD, and the presence of STING, phospho-STING, phospho-TBK1, and TBK1 were examined.The levels of total STING, phospho-STING, and phospho-TBK1 were reduced in the IFI207 KO BMDMs compared to WT cells (Fig. 3B).Similar results were seen in ISD-transfected BMDCs (Fig. S8A).Finally, IFI207 KO fibroblasts and macrophages were treated with DMXAA, and STING and phospho-STING levels were analyzed by Western blot.The KO cells had lower levels of both total and phospho-STING (Fig. 3C and D).
We also used siRNA knockdown to deplete Ifi207 in NIH3T3 cells and infected them with MLV gGag , which, because of its unstable capsid, falls apart in the cytoplasm more rapidly than wild-type virus, and thus induces higher levels of IFNβ (17,32).Knockdown of STING resulted in loss of phospho-TBK and -STING in response to infection (Fig. 3E).
Lower levels of endogenous STING protein in IFI207 KO fibroblasts were also seen by immunofluorescence microscopy (Fig. 3F); fibroblasts were used because of better visualization of the cytoplasm than other cell types.When STING is activated, it first moves to the Golgi and then to an acidic compartment, where it is degraded (35)(36)(37).To determine if this trafficking occurs in the absence of IFI207, we treated WT, IFI207, and STING mut fibroblasts with ISD and stained them with the Golgi marker GM130 or LAMP1 (lysosomal compartment).STING staining was diffuse in both cell types in the absence of ISD (Fig. 3F).In contrast, ISD treatment caused STING to be localized in a perinuclear compartment in both WT and KO cells.No STING was detected in STING mut cells (Fig. S9B).We also found that while there was less co-localization of STING in both compartments of IFI207 KO cells, the levels were particularly low in the LAMP1 positive compartment (Fig. 3F).Taken together, these data suggest, in the absence of IFI207, lower levels of STING are present and may be the cause of the diminished anti-viral response.

IFI207 binds STING
To better understand IFI207's function and the possible reason for the S/T repeats expansion, we cloned the Ifi207 coding regions from RNA isolated from the spleens of C57BL/6N, DBA/2J, and a mouse from the 129P2/OlaHsD background (see Materials and Methods), as representatives of the variants with different numbers of S/T repeats (Ifi207 BL/6 = 12 repeats; Ifi207 DBA = 25 repeats; Ifi207 129 = 24 repeats) (Fig. 1C and see following section).For some experiments, we also used molecular clones of Ifi204 and Ifi203 derived from C57BL/6J mice for comparison (10).
To determine if IFI207 bound to STING, we co-transfected FLAG-tagged STING and all three full-length HA-tagged expression constructs (IFI207 BL/6 , IFI207 DBA , and IFI207 129 ) as well as an HA-tagged IFI204 vector and used anti-HA antibodies for immunoprecipita tion and Western blots.All three IFI207 constructs co-immunoprecipitated with STING (Fig. 4A).In contrast, although it was expressed at much higher levels than any of the IFI207 constructs, IFI204 only weakly co-immunoprecipitated with STING (Fig. 4A).
Interestingly, we observed that even though equal amounts of STING plasmid were transfected in the presence or absence of IFI207, there was more STING protein in extracts when IFI207 was present (input; Fig. 4A).This was similar to our observation that endogenous STING protein levels were higher in IFI207-containing cells (Fig. 3).This suggested that STING stabilization might be dependent on IFI207/STING interac tion.To determine if this was the case, increasing amounts of the IFI207 expression plasmids were co-transfected with an expression vector containing FLAG-tagged STING, and the levels of STING were examined by Western blot.Co-expression of each of the IFI207 plasmids stabilized STING protein expression compared to empty vector (Fig. 4B).Moreover, both the IFI207 DBA and IFI207 129 constructs stabilized STING more effectively than the IFI207 BL/6 construct (Fig. 4B, top).Increasing amounts of STING plasmid were also co-transfected with constant amounts of the IFI207 plasmids, and again, all three IFI207 constructs increased STING protein expression.Although IFI207 stabilized STING, increasing STING had no effect on IFI207 levels (Fig. 4B, bottom).This effect of IF207 was specific to STING; endogenous tubulin and actin expression was not affected by over-expression or depletion of IFI207 (Fig. 3 and 4), and expression of co-transfected FLAG-tagged red fluorescent protein (RFP) construct was not altered by co-transfection of IFI207 (Fig. S10A).

The S/T-rich repeat region is important for STING stabilization
Although there are minor sequence differences in the PYD and HIN domains of the Ifi207 genes found in BL/6, 129, and DBA mice, the major difference we found was in the sequence and number of the HIN-B-proximal S/T-rich repeat region; BL/6 IFI207 had only 12 repeats, while IFI207 in 129 mice lacked a single S/T repeat compared to DBA mice (24 vs 25 repeats, respectively) but was otherwise identical (Fig. 1C; Tables S2 and  S3).To assess whether the repeat region played a role in STING stabilization, we made constructs in the backbone of each allele that lacked all the S/T repeats (R1 BL/6 ; R1 DBA ; R1 129 ; Fig. 4C).These three constructs, as well as the full-length constructs from which they were derived (FL BL/6 ; FL DBA ; FL 129 ), were co-transfected into 293T cells with STING.Expression of all three R1 deletion constructs was much higher than the full-length IFI207 constructs.Nonetheless, deletion of the repeat region abrogated the ability of IFI207 to stabilize STING, because the level of STING after co-transfection with R1 BL/6 , R1 DBA , stripped and probed with anti-actin to control loading.(C and D) Fibroblasts (C) and macrophages (D) from IFI207 KO and wild-type mice were treated with 100 µg/mL DMXAA, and Western blots examining STING and phospho-STING levels were measured at the indicated times.Shown below the graph is the quantification of three independent experiments.Two-way ANOVA was used (*, P ≤ 0.03; **, P ≤ 0.001; ****, P < 0.0001; ns, not significant).(E) NIH3T3 cells were transfected with siRNAs targeting IFI207 or STING.Non-targeting (siC) siRNA was also used.At 48 h post-transfection, the cells were infected with MLV gGag for 2 h (MOI = 5).Protein expression was examined by Western blot with the indicated antibodies.Right panel: Ifi207 KD was verified by RT-qPCR.Average + SD from two independent experiments is shown.(F) Primary fibroblasts were transfected with 8 µg/mL ISD, and at 2 h post-transfection, the cells were fixed and stained with anti-STING and anti-LAMP1 (left) or anti-GM130 (right).Shown to the right of the graphs is the quantification of STING staining from untreated cells (NT) and treated cells.Different exposure times were used to quantify STING levels in the absence (4.5 s) and presence (2 s) of ISD.Shown to the bottom right are the fraction of cells with colocalized STING and LAMP1 or STING and GM130.The panels to the right show quantification of 15 fields for each cell type/condition.or R1 129 was like that seen with empty vector (EV in Fig. 4C).Moreover, STING dimer formation increased with full-length but not the repeat deletion constructs (Fig. S10B).
To further confirm that the repeat region was critical for STING stabilization, we introduced the IFI207 BL/6 HIN-B-proximal repeat region into the equivalent site in isoform 1 of IFI203 (Iso1) (RefSeq ID NP_001289578; see Fig. 1C and 4D).This isoform shares the same structure as the carboxy terminal half of IFI207 and is expressed more abundantly in primary macrophages than the other isoforms (11).The previously reported IFI203 expression vector contained the shortest isoform 3 (IFI203Iso3) (10).We introduced the IFI207 BL/6 repeat region into IFI203Iso1 to make a chimeric IFI203 with 12 HIN-side S/T-rich repeats (IFI203Iso1R; Fig. 1B).We co-transfected IFI203Iso1, IFI203Iso1R, FL-IFI207 BL/6 , and FL-IFI207 DBA with STING.IFI203Iso1R containing the BL/6 repeat region, but not IFI203Iso1 without the repeats, stabilized STING (Fig. 4D).

STING stabilization does not rely on DNA binding or nuclear localization but does require the PYD domain
The PYD domain of IFI16 is required for its ability to interact with ASC, BRCA1, and STING (38,39).We thus tested whether IFI207's PYD domain was also required for STING stabilization.We deleted the PYD domain from both the FL-IFI207 DBA and R1-IFI207 DBA constructs and co-transfected them with STING.Only the full-length construct, but not the ΔPYD constructs, stabilized STING (Fig. 4E).This suggests that although the repeat region was important for STING stabilization, the PYD domain was needed for STING binding.Indeed, the ΔPYD construct did not co-immunoprecipitate with STING (Fig. 4E).
In addition to the recent study suggesting that IFI207 enhances RNApolII binding at the transcription start sites of cytokine genes, it has been suggested that IFI16 restricts HIV-1 infection by interfering with the transcription factor Sp1, thereby suppressing transcription of the integrated provirus (16).To determine where IFI207 was located within the cell, we first performed fractionations of cells transfected with IFI207 BL/6 , IFI207 DBA , and IFI207 129 expression vectors in the presence and absence of STING expression plasmids.Although a large fraction of IFI207 was in the nucleus, it was also found in the cytoplasm, independent of the number of repeats; the BL/6, 129, and DBA constructs all showed both nuclear and cytoplasmic localization (Fig. S11A).STING also appeared in the nuclear fraction, likely due to its association with perinuclear membrane fractions (40,41).Some of the IFI207 found in the nucleus may be due to a similar association.
We next examined three additional mutants for their ability to stabilize STING: ΔNLS, which deletes a putative nuclear localization signal downstream of the PYD; HD, with alanine substitutions at amino acids 978 and 980 in FL-IFI207 DBA , which we showed previously abrogated nucleic acid binding when introduced into IFI203 (17); and HIN, which encodes only the HIN domain and was tagged with V5 (Fig. 4F).The ΔNLS and HD constructs showed decreased nuclear localization (Fig. S11B and C).We co-transfected expression constructs encoding these mutants as well as FL-IFI207 DBA and the R1 mutant with STING, and examined STING stability by Western blots.Constructs bearing IFI204, IFI203Iso3, and the DNA binding mutant IFI203 HD were also co-transfected with STING.We found that both IFI207 ΔNLS and HD mutants, but not the HIN construct, stabilized STING (Fig. 4F).Neither the IFI204 nor the IFI203 constructs stabilized STING.
To determine whether any of these regions also contributed to STING binding, we also carried out co-immunoprecipitations on extracts from cells co-transfected with STING and IFI207 DBA or the mutants on the DBA backbone.Although they were expressed at much lower levels than the R1 or HIN proteins, the FL, ΔNLS, and HD mutants, but not the R1 or HIN domain constructs, efficiently bound and stabilized STING (Fig. 4G).As we showed previously, IFI203Iso3 also bound to STING, albeit less effectively than repeat-containing IFI207 and at levels similar to the R1 mutant (Fig. 4G) (17).
We also tested whether binding of reverse transcripts correlated with IFI207's ability to stabilize STING.We showed previously that DDX41 and IFI203, but not the DNA binding mutant, IFI203 HD, precipitated MLV reverse transcription products from infected cells (17).MCAT cells (293T cells that stably express the MLV entry receptor ATCR1 [42]) were transfected with IFI207 DBA -FL, -R1, and -HD, as well as IFI204, IFI203Iso3, and DDX41, and then infected with MLV gGag , which, because the capsid dissociates more rapidly than wild-type virus, produces high levels of early reverse transcripts in the cytoplasm (17,32).At 4 h post-infection, the ability of the different constructs to precipitate MLV reverse transcripts was determined.IFI207 DBA -FL and -R1 strongly bound MLV reverse transcripts, while IFI207 HD weakly bound viral DNA (Fig. S12).IFI204 and IFI203Iso3 also weakly bound viral DNA, as we previously showed (17).These data show that nucleic acid binding/activation is not required for IFI207 stabilization of STING, at least under conditions of over-expression.
Finally, we studied the co-localization of the different mutants with STING by immunofluorescence.Both the ΔNLS and HD mutants were more frequently localized in the cytoplasm, while like the WT protein, the R1 mutant was seen in both the nucleus and cytoplasm, both in the presence (Fig. 4H) and absence (Fig. S11B and C) of STING.STING immunostaining was brighter in the presence of IFI207, similar to what was seen in WT cells compared to IFI207 KO cells (Fig. 2).STING was also more often found in concentrated perinuclear foci in cells expressing the WT than the R1 constructs (Fig. 4H).

DISCUSSION
Pathogens infect different cell types and tissues, and many replicate in distinct cellular compartments.TLRs, ALRs, cGAS, and other sensors mediate innate immune responses to pathogens.The need for targeted responses to different pathogens has likely led to the expansion of PRRs with different signaling pathways.These specific sensing and signaling pathways are also important targets for pathogens, and as a result, greatly accelerate host-pathogen co-evolution.Indeed, in house mouse strains, Alrs, Ifns, and even Ifn receptor genes are part of "strain-specific diverse regions" (SSDRs) with high sequence diversity (43).Some of the SSDRs are among the most dynamic regions in the genome.Both genome recombination (44) and non-crossover gene conversions (45) between large gene families with two divergent haplotype genomes will create novel gene structures.Unlike housekeeping genes, newly created immune-related genes have a higher chance of being beneficial and thus showing genetic drift in the population because of the "red queen hypothesis" (45,46).On the other hand, the destruction of an existing gene by recombination or gene conversion is more likely to be tolerated when it is part of a gene family because other family members may have (semi-)redundant functions.As a result, SSDRs are possibly also hot spots of new gene generation.
The Alr locus is a typical SSDR.With the exception of some bat species, it is found in all mammalian genomes (47).The ALRs belong to one family of innate immune sensors that may have diversified in response to the selective pressures of pathogens (9)(10)(11)(12)48).We show here that several new Alr family members have recently been introduced into the genomes of the Mus genus, probably by one or a series of unequal genome recombination events.Novel ALR members like MNDAL and IFI207 have acquired new roles in variable biology processes: MNDAL has become a cell growth regulator (49), and as we show here, IFI207 has acquired a key role in the STING-mediated antiviral signaling pathway, using non-redundant and different mechanisms than the other ALRs.These new members may have modified the functional capacity of mouse ALRs, because in rats and humans, fewer ALR members are sufficient for normal immune function.For example, humans have AIM2, IFI16, PYHIN1, and MNDA, while rats have Rn7 (AIM2), Rhin2 (Ifi206), Rhin3 (Ifi214), Rhin4 (Ifi203), and Rhin5 (Mnda) (Fig. 1B).Thus, no Ifi207 orthologs are required in these species.
This kind of functional innovation by gene conversion or recombination may happen independently and repeatedly.In support of this is the structure of the Ifi207-like genes in Apodemus sylvaticus.During our analysis, we found more than five copies of novel genes with a structure similar to Mus Ifi207 in this species.These genes are A. sylvaticus species-specific and lack the characteristic S/T-rich repeats found in Mus Ifi207.However, they have undergone a recent gene duplication and seem to be under selection in other regions of the genes, resulting in highly polymorphic PYD-side repeats (Fig. S1 and S2).In addition, although these genes are possibly formed by recombination between Ifi203 and Mnda as well, they seem to have different recombination sites from the Mus version.
These results suggest that Ifi207 in Mus and A. sylvaticus were introduced into genome independently and evolved to have different functions.
Although the entire Alr locus shows great diversity among different house mouse strains, Ifi207 was strikingly diverse.We found that this gene contains a large HIN-Bproximal S/T-rich repeat region that varies extensively among inbred and wild mice strains.We analyzed the repeat-region sequence in multiple mouse strains and found a large collection of different sequences of repeating units in the IFI207 repeat region.In addition to sequence variation of the repeat region, these mice also had different numbers of repeating units.Apart from this repeat region, the rest of the protein sequence was almost identical in different inbred strains, suggesting that the repeat region specifically is under strong selective pressure.There is a paradox between the young age (a few million years for the Mus genus, which may accumulate <2% mutation under neutral selection) and the extreme high variation (up to 36% at the nucleotide level) of the repeat units.The mechanism by which the sequence evolution occurs is yet to be resolved, but a better understanding of how the Ifi207 repeat region has evolved might offer clues about its role as an innate immune effector as wild mice encounter a wide array of pathogens that may have shaped its evolution.
The data presented here, using knockdown, knockout, and over-expression approaches, show that IFI207 stabilizes STING protein and alters downstream signaling.The repeat region, which is predicted to form a coiled structure similar to a leucine zipper, makes IFI207 unique among the ALRs.The length of the repeat region also affected binding to STING and its stabilization.The IFI207 proteins encoded in DBA/2 and 129 mice, which have 25 and 24 repeats, respectively, stabilized STING more effectively than did the BL/6 protein, which has only 12 repeats.This suggests that these repeats expanded and were subsequently retained in the encoded protein, likely because of exposure to pathogens by the progenitor mice from which different inbred strains were derived.Several ALRs, including IFI16, IFI203, IFI204, and IFI205, bind STING, but these proteins lack the extended repeat region and as we show here, bind STING to the same extent as IFI207 with no repeats.In contrast, IFI203 engineered to contain the repeats increased STING protein levels.Although the PYD domain of IFI207, like other ALRs, confers binding to STING, it is the IFI207 repeat region that confers its ability to stabilize STING and likely enhances the response to different ligands and viruses that activate the STING/IFN response pathway.
A previous study suggested that knockout of the entire Alr locus in mouse macro phages had no effect on the response to ISD (50).However, a more recent study showed that these same ALR KO mice were more highly infected with MLV (16).A number of studies have implicated IFI204 in the response to herpesviruses, polyomavirus, and several bacteria such as Francisella and Staphyloccocus, and IFI16 is important in human cells for the innate immune response to Listeria, herpesviruses, and HIV (13,16,19,(51)(52)(53).We showed previously that IFI203, in conjunction with DDX41, recognized MLV and that DDX41 KO mice were as susceptible to MLV infection as cGAS KO mice (17,20).We show here that IFI207 alters the response to DNA, cGAMP, DMXAA, and MLV.Because MLV has been in Mus species for at least 1 million years, it is not surprising that there is an arsenal of restriction factors in mice that control MLV infection, including multiple ALRS (54).Whether IFI207 KO mice are more susceptible to infection with other pathogens that activate the STING pathway is currently under investigation.
Our results are somewhat at odds with a recent report by Baran and colleagues showing that IFI207 acts to increase RNApol II-mediated transcription of both NFκB and IRF3-stimulated genes (33).Both groups found that IFI207-depleted cells had impaired responses to DNA stimulation.However, unlike Baran et al. (33), we found no significant effect of either LPS (TLR4 ligand) or bacteria treatment in a macrophage cell line depleted for IFI207 or primary BMDMs from IFI207 knockout mice.While the Baran paper demonstrated that IFI207-depleted cells had lower levels of TNFα and IFNβ1 in response to LPS or K. pneumoniae, IFI207 knockout mice in their study sustained lower levels of Klebsiella infection than WT mice.This is in contrast to infection of other mice with defects in the TLR or STING pathways, which are typically more highly infected by bacteria, and by our finding that IFI207 knockout mice are more susceptible to infection with MLV, consistent with the reduction in STING activation in the absence of IFI207.
IFI205 and IFI16 also alter gene expression, by acting as transcriptional repressors of ASC and integrated HIV proviruses, respectively, and these activities require nuclear localization (14,16).Although IFI204 was shown to act as an anti-MLV and -HIV restriction factor by suppressing proviral transcription, we did not see any effect in IFI204 KO mice on infection levels, suggesting that it is not anti-viral in vivo.IFI207 also has an NLS, but we show that its ability to stabilize STING was independent of this sequence.Interestingly, at least in over-expression studies, the mutant lacking repeats still bound MLV reverse transcripts but failed to stabilize STING, while the HD mutant that has minimal nucleic acid binding still showed STING-stabilizing activity.The role of nuclear localization and the nucleic acid-binding domain in IFI207 function thus remains to be determined.
There is incredible diversity in the ALR locus among mammalian species.For example, the horse locus has six ALR genes, dogs and pigs have two, cows have one, rats have five, humans have four genes, and bats have only a truncated AIM2 gene, all of which are encoded in a single locus, which sits between the cell adhesion molecule 3 (CADM3) and spectrin alpha, erythrocytic 1 (SPTA1) genes (9)(10)(11)(12)47).Although we show here that IFI207 enhances the anti-MLV and other STING-dependent innate immune responses, we do not know what pathogens may have shaped the diversity of this locus in different species.In addition to their roles in innate immunity, several ALRs have been implicated in adipogenesis (55,56), autoimmune diseases like lupus erythematosus (57,58) and inflammatory responses (14,59).Identification of the pressures that shaped the locus in different species is thus important for understanding the many biological processes in which these genes play a role.

Mice
C57BL/6N, DBA2/J, BALB/cJ, STING mut (C57BL/6J-Sting1gt/J), AIM2 KO [B6.129P2-Aim2Gt(CSG445)Byg/J], and CMV-Cre [B6.C-Tg(CMV-cre)1Cgn/J] mice were originally purchased from The Jackson Laboratory and were bred at the University of Illinois at Chicago (UIC).IFI207 KO mice [C57BL/6N-IFI207em1(IMPC)Wtsi] were generated by targeted deletion by zygote injection of four gRNAs, two flanking each side of exon 3, to induce a frameshift null allele of Ifi207.The embryos used for these experiments were C56BL/6NJ, and after germline transmission, the colony was maintained on this background.Full details of the gRNAs used for gene disruption are provided in Table S3A.IFI202 [Ifi202btm1(KOMP)Mbp/MbpMmucd] and IFI204 KO mice [Ifi204tm1a(KOMP)Wtsi/ MbpMmucd] were derived from sperm purchased from the Mouse Biology Program and UC Davis.The IFI204 mice were initially crossed with CMV-Cre mice to generate knockout strains and then backcrossed to remove the CMV Cre transgene (Fig. S4).Genotyping primers for the Ifi207, Ifi202, and Ifi204 knockout alleles are in Table S3B.cGAS KO mice [Mb21d1tm1d(EUCOMM)Hmgu] were originally provided by Michael Diamond and Skip Virgin (60).129P2/OlaHsd DNA was isolated from mice originally purchased from Harlan (11).Mus musculus castaneus mice were purchased from The Jackson Laboratory.All mice were housed according to the policies of the Animal Care Committee (ACC) of the UIC; all studies were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.The experiments performed with mice in this study were approved by the UIC ACC (protocol no.21-125).

Sequence analysis
The assemblies of non-mouse species are downloaded from NCBI (https:// www.ncbi.nlm.nih.gov), with access names listed in Tables S1 and S2.Whole genome de novo assembly, as well as the raw PacBio/PacBio Hifi reads of 13 mouse strains, Mus spretus and Mus caroli were provided by the Phase 3 Mouse Genomes Project (MGP) from EMBL-EBI (unpublished data; available from NCBI/EBI under project accession: PRJEB47108).The assemblies were produced using third-generation long reads (PacBio Sequel/PacBio Hifi, BioNano) and Hi-C.For the additional mouse strains, the alleles of Ifi207 were assembled from whole genome Illumina data or the transcriptomes of IFNα/γ-induced diaphragm fibroblast cells.Samples were provided by the Institute of Vertebrate Biology CAS.The sequences of HIN-B-proximal S/T repeat regions were double-confirmed by Sanger sequencing of genome amplicons, with the primers described in Table S7.Genomic DNA samples from mouse strain SIN, SIT, BULS, PERC, BULS, STUF, DJO, MOLF, and CIM were acquired from a variety of sources and were stored in Gulbenkian Institute of Science, Portugal.The details of the mouse strains can be found in the website https://housemice.cz/en/strains/.Full-length sequences of DBA and 129 IFI207 were obtained from the cloned cDNAs (see "Plasmids").(GenBank acquisition numbers for the re-sequenced repeat regions and full-length clones are listed in Table S5.For more details, see "Plasmids" .)

BMDM and BMDC cultures
Bone marrow was harvested from mouse limbs.To differentiate BMDMs, cells were cultured for 6-7 d on Petri dishes in DMEM containing 10% FBS, 2 mM L-glutamine, 100 U/mL penicillin, 100 µg/mL streptomycin, 0.1 mM sodium pyruvate, 25 mM HEPES (Gibco), and 10 ng/mL murine M-CSF (Gibco, PMC2044).Fresh media containing 10 ng/mL M-CSF were added on day 3. On day 6 or 7, BMDMs were collected for experiments.To differentiate BMDCs, bone marrow cells were treated with ammoniumchloride-potassium (ACK), and 2 × 10 6 cells were seeded in the center of Petri dishes in 10 mL Roswell Park Memorial Institute medium (RPMI) containing 10% FBS, 2 mM L-glutamine, 100 U/mL penicillin, 100 µg/mL streptomycin, and 50 µM β-mercaptoetha nol plus 20 ng/mL murine granulocyte-macrophage colony-stimulating factor (GM-CSF) (Peprotech, 315-03).On day 4, 10 mL fresh media containing 20 ng/mL GM-CSF was added to plates.On day 8, half of the culture volume was carefully removed to avoid disrupting the cells clustered in the centers of dishes, and 10 mL fresh media with 20 ng/mL GM-CSF was added to plates.Non-adherent cells clustered in centers of wells were collected by gentle pipetting for experiments on day 10.At time of collection, BMDMs and BMDCs were stained for F4/80, CD11b, or CD11c and were analyzed by flow cytometry (LSR-Fortessa flow cytometer; BD Biosciences).Macrophage and DC cell populations were >80% pure.

Expression analysis
BMDMs and BMDCs from BL/6, STING mut , IFI202, IFI204, and IFI207 KO mice were treated with or without 100 U mouse IFNβ (PBL Assay Science, 12405-1).Four hours after treatment, RNA was harvested, and Alr gene expression was analyzed by RT-qPCR.RNA was isolated by RNeasy Mini Kit (Qiagen 74106) with on-column DNase diges tion using RNase-Free DNase Set (Qiagen 79254).cDNA was synthesized with random hexamers from the Superscript III First-Strand Synthesis System (Invitrogen 18080051).RT-qPCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems 4367659) with appropriate primer sets on a QuantStudio 5 Real-Time PCR System (Applied Biosystems).Primers for RT-qPCR analysis of Alrs, Ifnb, Actin, and Gapdh have been previously described (Table S5) (11,17).MLV strong-stop primers (SSS) have been previously described (61).

LPS stimulation of BMDMs and macrophages
NR9456 were reverse transfected with 30 pmol of IFI207 or control siRNAs.Forty-eight hours post-transfection, the cells were treated with 10 ng/mL lipopolysaccharide (LPS; from Escherichia coli O55:B5; Sigma).After 3 h, RNA was collected, and Tnf and Ifnβ1 RNA levels were analyzed by RT-qPCR.Primary BMDMs were treated with 10 ng/mL LPS, and RNA for RT-qPCR analysis was isolated at 3 and 6 h post-treatment.

Bacterial infection of BMDMs
For S. aureus (USA300 LAC), overnight bacterial cultures were added into a new tube containing fresh tryptic soy broth (TSB) and were incubated for 3 h at 37°C.Bacteria were pelleted, washed twice with PBS, and adjusted to an OD 600 of 1.0 (1 × 10 9 CFU/mL).BMDMs from WT, IFI207 KO, STING mut , and IFI204 KO mice were infected by S. aureus at a multiplicity of infection (MOI) of 5 and 100.Gentamicin was added 1 h after infection, the media with bacteria were removed, the cells were washed twice with PBS, and fresh media with 200 µg/mL of gentamicin were added.Four hours after infection, RNA was collected, and Tnf/Ifnβ1 mRNA expression was analyzed by RT-qPCR.For K. pneumoniae (KPPR1), overnight bacterial cultures were grown in lysogeny broth (LB broth), and a 1:10 dilution was inoculated into fresh LB and was incubated for 2.5 h at 37°C.Bacteria were pelleted and adjusted to an OD 600 of 1.0 (1 × 10 10 CFU/mL).BMDMs and BMDCs from WT, IFI207 KO, STING mut , and IFI204 KO mice were infected with K. pneumoniae at an MOI of 10 and 100.At 1 h after infection, the media with bacteria were removed, the cells were washed twice with PBS, and fresh media with 100 µg/mL of gentamicin were added.At 6 h after infection, cells were harvested for RNA isolation.

Virus
Wild-type Moloney MLV and MLV gGag were harvested from supernatants of stably infected NIH3T3 cells.Supernatants were centrifuged for 10 min (3,000 rpm) to remove cellular debris, and then filtered through a 0.45-µm filter (Millex Millipore) and treated for 30 min at 37°C with 20 U/mL DNase I (Roche).The virions were pelleted through a 25% sucrose cushion, resuspended in 2% FBS-supplemented DMEM, snap frozen, and stored at −80°C.Virus titers were determined by an infectious center (IC) assay on NIH3T3s.One day before tittering, 8 × 10 4 cells were plated in 6-well tissue culture plates and were incubated overnight at 37°C.Tenfold serial dilutions of virus stocks were prepared with polybrene (8 µg/mL) in supplemented DMEM and were added to cells.Infected cells were incubated at 37°C for 2 h with rocking every 15 min.After 2 h, viral dilutions were aspirated, and 2.5 mL media was added to each well.Plates were incubated for 3 d at 37°C.After 3 d, the plates were washed in PBS and were stained with monoclonal 538 Env antibody (1:300) for 1 h at 4°C with rocking.The plates were washed with 2% FBS in PBS and were stained with goat anti-mouse IgG/IgM Alexa Fluor 488-conjugated secondary antibody (Invitrogen A-10680) diluted in 2% FBS in PBS (1:300) for 45 min at 4°C with rocking.The plates were washed, and ICs were counted by microscopy (Keyence BZ-X710 and BZ-X analyzer) to determine viral titers.For virus stocks derived from cells transfected with molecular clones, viral mRNA was quantified by RT-qPCR with primers for env, and viral proteins were confirmed by Western blotting with MLV antisera, as previously described (30).

In vitro MLV infection
Forty-eight hours post-siRNA transfection, cells were infected with MLV gGag and were harvested at 2 h post-infection.The cells were collected, and lysates were analyzed by Western blot.

In vivo MLV infections
IFI207 heterozygous mice were crossed, and their newborn litters consisting of knockouts and heterozygotes, as well as WT mice, were used for infection.Genotyping was carried out after virus loads were measured.Additional BL/6N pups were also infected and included in the WT group.STING mut and IFI204 KO mice were generated by homozygote matings.Mice were intraperitoneally injected with 2 × 10 3 PFU MLV at 0-48 h post-birth.At 16 days post-infection, splenocytes were harvested and treated with ACK buffer.Tenfold dilutions of the cells were co-cultured with NIH3T3 cells for 3 days to assess viral titers by IC assays.

In vivo LPS treatment
WT, IFI207 KO, TLR4 KO, and IFI204 KO mice were injected into the right hind footpad with 1 µg of LPS.At 2 h after LPS injection, cells were isolated from the popliteal draining (right) and non-draining (left) lymph nodes.RNA was isolated, and Tnf and Ifnβ1 RNA levels were analyzed by RT-qPCR.

Plasmids
Expression plasmids for DBA and 129 IFI207 variants were generated by PCR amplification of cDNA synthesized from splenic RNA of DBA2/J and AIM2 KO mice.AIM2 KO were used because 129/P2 mice were no longer available for purchase, and the Alr locus and Ifi207 in AIM2 KO mice were derived from the parental 129P2/OlaHsd embryonic stem cells in which the knockouts were made (11).An HA tag was added to the C-termini immediately before the stop codon of both IFI207 variants.The PCR products were cloned into pcDNA3.1/Myc-His(+)(Invitrogen).Full-length sequences of DBA and 129 IFI207 were obtained from the cloned cDNAs.IFI207 R1 mutants, ΔNLS, ΔPYD, and HD mutants were generated from full-length IFI207-HA plasmids and the R1-ΔPYD mutant from R1-IFI207 using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs).IFI207-HIN plasmids were generated by PCR amplification of the HIN domain from full-length IFI207-V5 expression plasmids.BL/6-derived IFI203Iso2-HA, IFI207 BL/6 -HA, IFI204-HA, and STING-FLAG plasmids were a gift from Dan Stetson (10).The IFI203Iso2 HD plasmid was previously described (17).V5 tags were introduced to the IFI207 BL/6 , IFI207 DBA , and IFI207 129 constructs by PCR amplification of the IFI207-HA expression constructs with primers designed to replace the HA tags with V5 tags.The V5-tagged PCR products were cloned into pcDNA 3.1/Myc-His(+).
The IFI203Iso1-HA plasmid was generated by in-fusion cloning.In brief, cDNA from C57BL/6N primary fibroblasts was used as a template to amplify full-length IFI203 isoform 1, using primers 203F and 203R.PCR was performed to linearize IFI203Iso2-HA plasmid using primers B6 203 1F and B6 203 2R.The insertion DNA fragment from the isoform 1 cDNA was PCR amplified using primers B6 203 3F and B6 203 4R.In-fusion assembly with the two fragments was performed using the In-fusion snap assembly kit by Takara Bio.
To generate IFI203Iso1R containing the IFI207 repeat region, PCR was performed to linearize IFI203Iso1-HA plasmid using primers 203 VF and 203 VB.The repeat region of IFI207 BL/6 was amplified by PCR using primers 207 21bpF and b6 21bpR.After purification, the PCR products were used for in-fusion assembly.
All cloning primers are described in Table S6 and S7.The myc-tagged DDX41 construct was previously described (17).All plasmids were sequenced to verify the inserts.

Transfections for protein expression analysis
HEK293T cells were transiently transfected using Lipofectamine 3000 for 24 h.For fractionation assays, HEK293T cells were transfected with expression plasmids, collected in ice cold PBS, and fractionated into nuclear, cytoplasmic, and whole cell extracts by the rapid, efficient, and practical (REAP) method (62).The REAP method was modified to lyse nuclear extracts in RIPA buffer and to treat nuclear lysates with 3 U micrococcal nuclease (Thermo Fisher Scientific) per μL lysate for 15 min at room temperature (RT).For IFI207 and STING titration assays, HEK293T cells were transfected either with increasing amounts of IFI207-HA and a constant amount of STING-FLAG or with increasing amounts of STING-FLAG and a constant amount of IFI207-HA.When different amounts of plasmids were transfected within experiments, empty vector (EV) plasmid [pcDNA 3.1/Myc-His(+)] was added to keep the total amount of transfected DNA equal for all samples.

Immunoprecipitations
For immunoprecipitation (IP) assays, HEK293T cells were transfected with expression plasmids for 24 h using Lipofectamine 3000.The cells were harvested in PBS, pelleted, lysed in RIPA buffer, sonicated, and quantified by Bradford Assay (Bio-Rad).IP samples were prepared with Protein A/G PLUS-Agarose (Santa Cruz sc-2003) or HA-tag and V5-tag Sepharose bead conjugate (3956S and 67476S, respectively).Antibodies and agarose or Sepharose beads were incubated with cell lysates at 4°C overnight on a rotating rack.Samples were washed three to five times in RIPA buffer and were centrifuged for 3 min (2,500 rpm).Samples were eluted in 4X Laemmli buffer plus 5% β-mercaptoethanol.

Indirect fluorescent antibody (IFA) assay
NIH3T3 cells were seeded on glass coverslips in 12-well plates and were incubated at 37°C overnight.The cells were transfected with IFI207-HA and STING-FLAG expression plasmids for 24 h with Lipofectamine 3000.To stain, the culture media were aspirated, the cells were fixed in 4% paraformaldehyde (PFA) for 10 min, washed in PBS for 3 × 2 min, permeabilized in 0.25% Triton X-100-PBS for 5 min, washed, blocked for 30 min in 10% goat serum/PBS, incubated in primary antibody for 2 h, washed, incubated in secondary antibody for 2 h, and washed a final time.Coverslips were mounted on slides with VECTASHIELD Antifade Mounting Medium with DAPI (Vector Laboratories, H-1200).Primary and secondary antibodies were diluted in blocking buffer.Primary antibodies used were rabbit anti-HA (Abcam 9110) and mouse anti-FLAG M2 (Sigma, F1804).Goat anti-rabbit IgG Alexa Flour 568-conjugated (Invitrogen, A-11011) and goat anti-mouse IgG Alexa Fluor 488-conjugated (Invitrogen, A-11001) secondary antibodies were used.Images were captured on a Keyence BZ-X710 at 60× magnification under oil immersion.Image brightness was manually adjusted to better visualize STING.Scoring of images was performed in a blinded fashion.Images were scored for IFI207 and STING localiza tion (cytoplasmic, nuclear, or both).To quantify localization, cells that were IFI207-posi tive or STING-positive (single transfections) or IFI207-and STING-double positive (double transfections) were counted.Cells scored as cytoplasmic, nuclear, or both are represented as the fraction of the total cells counted within each transfection group (Fig. S11B).
Primary fibroblasts were seeded on glass coverslips in 24-well plates and were incubated at 37°C overnight.ISD was transfected into fibroblasts for 2 h with Lipofect amine 3000 reagent (Invitrogen).Fibroblasts were prepared for staining as described in the preceding paragraph.The fibroblasts were incubated with primary antibody overnight at 4°C and then stained with the secondary antibody (all 1:1,000, Alexa Fluor; Invitrogen) for 1 h.Coverslips were mounted on slides with VECTASHIELD Antifade Mounting Medium with DAPI (Vector Laboratories, H-1200).The antibod ies and concentrations used were as follows: sheep anti-STING (1:100, R and D Systems AF6516, RRID:AB_10718401), rabbit polyclonal anti-GM130 (2 μg/mL, Novus NBP2-53420G, RRID:AB_2890912), and rabbit anti-LAMP1 (1:100, Sigma-Aldrich L1418, RRID:AB_477157).Images were acquired using a Keyence BZ-X710 at 40× and 60× magnification.Data analysis was performed with BZ-X Analyzer software.STING intensity analysis was calculated by averaging the STING fluorescence intensity per total number of cells.

Statistical analysis
All gene expression RT-qPCR analysis experiments were conducted in duplicate or triplicate with two to three technical replicates for each experiment.Each technical replicate was additionally split into triplicates for RT-qPCR analysis.All other experiments were performed at least in triplicate with technical replicates or as indicated in the figure legends.Fluorescence quantifications were carried out in a blinded fashion.GraphPad Prism was used for statistical analysis.Tests used to determine significance are shown in the figure legends.

FIG 1
FIG 1 Ifi207 structure.(A) Genome structure of the Alr locus in inbred mouse strains.Based on gene structure and phylogeny, the nine family members with variant copy numbers are indicated by the color blocks.Ifi208, Ifi207, Ifi204, and one copy of Ifi203 are encoded in a stable contig, and other regions in the Alr locus showed significant numbers of inversions, recombinations, and duplications.(B) Genome structure of the Alr locus in other rodent species and humans.Ifi207, MndaL, and Ifi202 are only found in the Mus genus.Some of the genes in Onychomys and Peromyscus were annotated tandemly, probably caused by low annotation quality.(C) IFI207 protein structure in different Mus genus members.Shown for comparison is the structure of IFI202, IFI204, MNDA, and IFI203 (see Tables S2 and S3 for sequences).Similar PYRIN domains are indicated by color, as are the HIN-A and HIN-B domains.The long linker sequence between these domains in IFI203 and IFI207 is indicated by a blue bar.(D) Comparison of C57BL/6 IFI207 (E9Q3L4) and IFI203 (E9QAN9) predicted protein structures.Structures were downloaded from AlphaFold (https://alphafold.ebi.ac.uk/).Domains are colored as follows: pyrin domain, pink; PYD-proximal repeat, brown; HIN-proximal repeat, teal; and HIN-B, blue.

FIG 2
FIG 2 IFI207 KO mice are more susceptible to MLV infection.(A) Pups of the indicated genotype were infected with 2,000 PFU of MLV (MLV or MLV gGag ).At 16 dpi, splenic viral titers were determined on NIH3T3 cells by infectious center (IC) assays.Significance was determined by two-tail Mann-Whitney test (**, P < 0.003; ****, P < 0.0001).Right panel: RT-qPCR analysis to verify Ifi207 RNA levels in mice.Symbols represent individual mice.Significance was determined by one-way ANOVA (***, P < 0.002; ****, P < 0.0001).(B) BMDMs from the indicated mice were transfected with the indicated concentrations of ISD or cGAMP for 3 h.Ifnβ RNA levels were quantified by RT-qPCR.Shown are the averages of three independent experiments.(C) Primary fibroblasts from IFI207 KO and STING mut mice were transfected with 8 µg/mL of ISD or cGAMP or 100 µg/mL DMXAA, and at 3 h post-transfection, Ifnβ RNA levels were measured.Shown is the average of three experiments (ISD and cGAMP) or two experiments (DMXAA).Significance was determined by two-way ANOVA (****, P < 0.0001).

FIG 4
FIG 4 IFI207 stabilizes STING.(A) HEK293T cells were transfected with full-length HA-tagged IFI207 129 , IFI207 BL/6 , IFI207 DBA , IFI204, and FLAG-tagged STING expression constructs.Anti-HA was used to immunoprecipitate IFI207 in the transfected cell extracts, and Western blots with the indicated anti-HA and -FLAG (STING) as well as anti-tubulin antibodies were performed.(B) HEK293T cells were transfected with increasing amounts of the indicated full-length IFI207 (HA) (top) or STING (FLAG) (bottom) expression constructs.Protein expression was examined by Western blot with antibodies to the tags and to STING.Experiments were performed in triplicate and quantified by densitometry analysis in ImageJ (right panel).(C) HEK293T cells were co-transfected with HA-tagged full-length (FL) or repeat deletion (R1) IFI207 129 , IFI207 B6 , IFI207 DBA (diagram), and FLAG-tagged STING expression plasmids.Protein expression was examined by Western blot with the antibodies to the tags and to STING.Experiments were performed in triplicate and were quantified by densitometry analysis in ImageJ (right panel).Green arrow, IFI207 129 or IFI207 DBA ; purple arrow, IFI207 B6 ; cyan arrow, R1 deletion constructs.(D) The same experiment was carried out with HA-tagged IFI203Iso1, IFI203Iso1R, IFI207 B6 , and IFI207 DBA .(E) IFI207 DBA FL or R1 lacking the PYD domain was examined for their ability to increase STING levels (left panel) and co-immunoprecipitate with STING (right panel).(F) IFI207, IFI203 HD, and IFI207 ΔNLS mutants and HIN constructs were tested for their ability to increase STING levels.All IFI207 plasmids are HA tagged, except the HIN construct, which contains a V5 tag.Protein expression was examined by Western blot with the indicated antibodies; STING was detected with anti-STING antibody.Right panel: triplicate experiments were quantified by densitometry analysis in ImageLab (BioRad).NS, non-specific; EV, empty vector.(G) HEK293T cells were co-transfected with the indicated IFI207, IFI203, IFI204, and FLAG-tagged STING expression plasmids.Immunoprecipitations were carried out with anti-HA (left panel) or V5 (right panel), and Western blots were examined for co-immunoprecipitation with STING.(H) NIH3T3 cells were transfected with the full-length IFI207 BL/6 (HA) with or without STING (FLAG) expression constructs.Cells were fixed and stained with antibodies directed against expression construct tags.Images were acquired by fluorescence microscopy.Image brightness was manually adjusted to better visualize STING in the STING-alone panels because of the lower levels of STING.Images of cells transfected with IFI207 constructs without STING are in Figure S11B, and quantification of subcellular localization is in Figure S11C.