MPYS/STING (stimulator of IFN genes) senses cyclic dinucleotides (CDNs), generates type I IFNs, and plays a critical role in infection, inflammation, and cancer. In this study, analyzing genotype and haplotype data from the 1000 Genomes Project, we found that the R71H-G230A-R293Q (HAQ) MPYS allele frequency increased 57-fold in East Asians compared with sub-Saharan Africans. Meanwhile, the G230A-R293Q (AQ) allele frequency decreased by 98% in East Asians compared with sub-Saharan Africans. We propose that the HAQ and AQ alleles underwent a natural selection during the out-of-Africa migration. We used mouse models of HAQ and AQ to investigate the underlying mechanism. We found that the mice carrying the AQ allele, which disappeared in East Asians, had normal CDN–type I IFN responses. Adult AQ mice, however, had less fat mass than did HAQ or wild-type mice on a chow diet. AQ epididymal adipose tissue had increased regulatory T cells and M2 macrophages with protein expression associated with enhanced fatty acid oxidation. Conditional knockout mice and adoptive cell transfer indicate a macrophage and regulatory T cell–intrinsic role of MPYS in fatty acid metabolism. Mechanistically, AQ/IFNAR1−/− mice had a similar lean phenotype as for the AQ mice. MPYS intrinsic tryptophan fluorescence revealed that the R71H change increased MPYS hydrophilicity. Lastly, we found that the second transmembrane (TM) and the TM2–TM3 linker region of MPYS interact with activated fatty acid, fatty acyl–CoA. In summary, studying the evolution of the human MPYS gene revealed an MPYS function in modulating fatty acid metabolism that may be critical during the out-of-Africa migration.

This article is featured in Top Reads, p.

The cyclic GMP-AMP (cGAMP) synthase (cGAS)–cGAMP–STING (stimulator of IFN genes)/MPYS–type I IFNs axis is critical for host defense against pathogen infections and the development of inflammatory diseases (1, 2). MPYS achieves these pathophysiological functions via sensing the second messenger cyclic dinucleotides (CDNs) derived from pathogens or the danger signal cytosolic DNA (3, 4). Notably, MPYS is constitutively highly expressed in most hematopoietic and nonhematopoietic cells, including endothelial cells, alveolar type 2 pneumocytes, bronchial epithelium, alveolar macrophages, monocytes, T cells, and peripheral sensory neurons (57), suggesting that MPYS may have a function at homeostasis. In contrast, cGAS is an IFN-stimulated gene with low basal expression (8). Studying common human MPYS variants may reveal the physiological function of MPYS at homeostasis.

The human MPYS gene is highly heterogeneous (9, 10). Approximately 50% of Americans carry a non–wild-type (WT) (R232) MPYS allele (9). R71H-G230A-R293Q (HAQ) is the second most common human MPYS allele carried by ∼25% of Americans and ∼63% of East Asians (EAS) (9). Surprisingly, we recently found that HAQ knock-in mice do not respond to CDNs in vivo (11). In corroboration, Mogensen and colleagues (12) reported that homozygous HAQ MPYS/STING was enriched in HIV-infected individuals that were long-term nonprogressors. These HAQ individuals had reduced inhibition of CD4 T cell proliferation and a reduced immune response to DNA and HIV, which could contribute to slower disease progression (12). Optiz and colleagues (13) showed that the HAQ allele impairs cGAS-dependent antibacterial responses and increased susceptibility to Legionnaires’ disease in humans. Levy and colleagues (14) showed that whereas WT STING promoted MHC-matched allogeneic hematopoietic stem cell transplantation–induced graft-versus-host disease, the HAQ mice developed reduced MHC-matched graft-versus-host disease (14). Most recently, in a clinical trial (ClinicalTrials.gov ID NCT02471014), we found that HAQ individuals had poor responses to the Pneumovax 23 vaccine (15), a phenotype that we previously found in the HAQ knock-in mouse (11). Thus, the HAQ allele is defective in CDN sensing.

Analysis of 1000 Genomes Project data found that the AQ allele is exclusively found in individuals of African ancestry (11). In contrast, HAQ is rarely present in these individuals (11). In this study, we developed an AQ knock-in mouse. By comparing the HAQ and AQ mice, we revealed a function of MPYS in regulating fatty acid metabolism at homeostasis.

The AQ knock-in mice were generated by transfecting the linearized targeting vector (Supplemental Fig. 2A) into JM8A3.N1 C57BL/6N embryonic stem cells. A positive embryonic stem clone was subjected to the generation of chimera mice by injection using C57BL/6J blastocysts as the host. Successful germline transmission was confirmed by PCR sequencing (Supplemental Fig. 2B). The heterozygous mice were bred to actin-FLPase mice (The Jackson Laboratory, B6.Cg-Tg(ACTFLPe)9205Dym/J) to remove the neo gene and make the AQ-MPYS knock-in mouse. Animals were generated at the National Jewish Health Mouse Genetics Core Facility.

Age- and sex-matched mice (8–52 wk old, both male and female) were used for indicated experiments. HAQ (11) and MPYS−/− (16) mice were generated as previously described. Mice were housed at 22°C under a 12‐h light/12-h dark cycle with ad libitum access to water and a chow diet (3.1 kcal/g, Teklad 2018, Envigo, Somerset, NJ) and bred under pathogen-free conditions in the Animal Research Facility at the University of Florida. Littermates of the same sex were randomly assigned to experimental groups. All mouse experiments were performed by the regulations and approval of the Institutional Animal Care and Use Committee at the University of Florida (approval no. 201909362).

There are five common human MPYS alleles with a population frequency >1%: WT; H232 (rs1131769); HAQ, R71H (rs11554776)–G230A (rs78233829)–R293Q (rs7380824); AQ, G230A (rs78233829)–R293Q (rs7380824); and Q293 (rs7380824). We extracted STING alleles frequency from 1000 Genomes Project phase III: rs1131769 (single-nucleotide polymorphism [SNP])–population genetics (http://useast.ensembl.org/Homo_sapiens/Variation/Population?db=core;r=5:139477834-139478834;v=rs1131769;vdb=variation;vf=166078604); rs7380824 (SNP)–population genetics (http://useast.ensembl.org/Homo_sapiens/Variation/Population?db=core;r=5:139476897-139477897;v=rs7380824;vdb=variation;vf=167755730); rs11554776 (SNP)–population genetics (http://useast.ensembl.org/Homo_sapiens/Variation/Population?db=core;r=5:139480993-139481993;v=rs11554776;vdb=variation;vf=168808284#373509_tablePanel); rs78233829 (SNP)–population genetics http://useast.ensembl.org/Homo_sapiens/Variation/Population?db=core;r=5:139477840-139478840;v=rs78233829;vdb=variation;vf=191805512. MPYS genotypes for each human subpopulation were downloaded and analyzed for the above human SNPs. WT STING allele is defined as R232 but does not contain any non-synonymous SNPs. HAQ allele is identified to have all three SNPs: R71H (rs11554776)–G230A (rs78233829)–R293Q (rs7380824). AQ allele is identified to have only two SNPs: G230A (rs78233829)–R293Q (rs7380824). H232 and Q293 alleles are identified to have only the rs1131769 and rs7380824 SNPs, respectively. Genotype frequency individuals was calculated based on the allele information. The carriers’ population frequency was calculated as the percentage of people in the population carrying at least one copy of the allele. Information on ethnic populations in the 1000 Genomes Project can be found via the following link: https://www.internationalgenome.org/data-portal/population.

The χ2 test of the goodness-of-fit for Hardy-Weinberg equilibrium (HWE) was calculated as [Σ(Observed numbers − Expected numbers)2/Expected numbers]. The degree of freedom is 2 for three genotypes. The critical values are 5.991 and 9.21 for the p values of 0.05 and 0.01, respectively. The ratio of nonsynonymous-to-synonymous substitution rates (dN/dS) value between chimpanzee (XM_034960610.1) and human MPYS genes (NM_001301738.2, the R232 allele) was calculated by the PAL2NAL (http://www.bork.embl.de/pal2nal/) (embl.de) program. The calculation of the fixation index (FST) is done as (HTHS)/HT, where HT is the expected heterozygosity for total populations and HS is the expected heterozygosity for each subpopulation.

Linkage disequilibrium (LD) r2 scores were download from the 1000 Genomes Project (e.g., rs11554776 (SNP)–Linkage disequilibrium plot–Homo_sapiens–Ensembl genome browser 107) (http://useast.ensembl.org/Homo_sapiens/Variation/LDPlot?db=core;pop1=373516;r=5:139471493-139491492;v=rs11554776;vdb=variation;vf=168808284). HAQ and AQ haplotypes in the Luhya in Webuye, Kenya (LWK; 99 individuals) and Southern Han Chinese (CHS; 105 individuals) populations were characterized by 19 SNPs covering 18.5 kb (from 5:1394866846–5:139468316) of the genomic region and further dephased from their respective genotypes. The youngest possible age of the HAQ-1 haplotype was calculated using a mutation rate of 1 mutation in 30 million bp per generation (1720). The HAQ-1 haplotype age = (2 mutations × 30 × 106)/(18.5 × 103) = 3.24 × 103 generations. We used the average 22 y per generation, that is ∼70,000 y.

Bone marrow–derived dendritic cells (BMDCs) were induced from mouse bone marrow cells cultured in RPMI 1640 (Invitrogen, 11965) with 10% FBS, 2 mM l-glutamine, 1 mM sodium pyruvate, 10 mM HEPES buffer, 1% nonessential amino acids, 50 mM 2-ME, and 1% penicillin/streptomycin, with 20 ng/ml GM-CSF (Kingfisher Biotech, RP0407M) for 7 d (11). BMDMs were induced from mouse bone marrow cells in the above RPMI 1640 medium with 20 ng/ml M-CSF (Kingfisher Biotech, RP0462M) for 7 d (11).

Bodyweight was recorded using a conventional weigh scale, and body composition was measured in the unanesthetized mouse by a quantitative magnetic resonance method using an EchoMRI 700 analyzer (EchoMRI, Houston, TX).

After overnight fasting (∼12 h), an i.p. injection of 25% glucose solution at a dose of 2 g/kg body weight was administered. Glucose concentrations were determined from the tail venous blood using a hand‐held glucometer (Metene blood glucose meter, Metene, Santa Clara, CA) at baseline (0) and 30, 60, and 120 min after glucose injection.

The stromal vascular fraction (SVF) was isolated from the visceral adipose tissue as previously described (21). Briefly, 500 mg of epididymal white adipose tissue (eWAT) explants was digested in 1 ml of DMEM supplemented with 2% fatty acid-free BSA (Sigma-Aldrich, 126575), HEPES (10 mM), Liberase TM (thermolysin medium) (25 μg/ml) (Roche, 05401119001), and DNAse (250 μg/ml) (Roche, 10104159001) for 1 h at 37°C. Digested tissue was filtered through a 150-μm mesh into preheated DMEM containing 2% FBS. Cells were centrifuged at 150 × g for 8 min. The floating adipocytes and SVF cell pellets were collected. SVF pellets were suspended in 0.5 ml of ACK buffer to lyse contaminating erythrocytes and spun at 2.2 kg for 5 min at 4°C. SVF cells were resuspended in the FACS buffer.

White adipocytes were isolated from eWAT as the floating fraction and lysed in RIPA buffer (Cell Signaling Technology [CST], 9806). Protein concentration in whole-cell lysis was quantified using a Bio-Rad protein assay kit (Bio-Rad, 5000201EDU). Three micrograms of total proteins was loaded on 4–20% Mini-PROTEAN gels and probed by the indicated Abs. Following Abs were used: anti-STING rabbit Ab (Proteintech, 19851-1-AP), anti-PGC-1α rabbit Ab (EMD Millipore, ABE868), anti-UCP-1 rabbit Ab (CST, 14670), anti-PPARγ rabbit Ab (CST, 2435), anti-HSL rabbit Ab (CST, 4107), anti-ATGL rabbit Ab (CST, 2138), anti-tubulin rabbit Ab (CST, 2148), anti-hemagglutinin (HA) HRP (BioLegend, 901519), anti-stearoyl-CoA desaturase 1 (SCD-1) rabbit Ab (CST, 2794), and anti-rabbit IgG-HRP (CST, 7040).

eWAT explants were isolated and washed in culture plates with DMEM. After removal of blood vessels and connective tissue by dissection, 25 mg of eWAT was weighed and placed in 250 μl of culture media containing DMEM, 2 mmol/l l-glutamine, 2% fatty acid–free BSA, 50 U/ml penicillin, and 50 mg/ml streptomycin for 2 h at 37°C. Free fatty acid (FFA) released into the culture media was determined for total FFA (Sigma-Aldrich, MAK044) and unsaturated FFA (Cell Biolabs, ST1613) according to the manufacturers’ protocols and quantified by a standard curve provided by the manufacturers.

Single-cell suspensions were stained with fluorescent dye–conjugated Abs in FACS buffer (PBS containing 2% FBS and 1 mM EDTA) (22). For intracellular cytokine or transcription factor staining, cells were fixed and permeabilized with the Foxp3 staining buffer set (eBioscience, 00-5523-00) (23). To measure IFN-γ, IL-4, and IL-17a, cells were cultured at 37°C for 4 h with a cell activation mixture (BioLegend, 423301) containing PMA, ionomycin, and brefeldin A. Cells were washed and stained with surface markers. Cells were then fixed and permeabilized for intracellular cytokine staining. Data were acquired on a BD LSRFortessa and analyzed using the FlowJo software package (FlowJo). Cell sorting was performed on the BD FACSAria III flow cytometer and cell sorter. The following flow Abs were used: anti-mouse PPARγ (Thermo Fisher Scientific, PA5-25757), anti-mouse CD4 PE/Cy7 (clone GK1.5) (BioLegend, 100422), anti-carnitine palmitoyltransferase 1 (CPT1)A rabbit Ab (Proteintech, 15184-1-AP), anti-mouse IL-4 allophycocyanin (clone 11B11) (BioLegend, 504106), anti-mouse IL-17a PE (clone TC11-1810.1) (BioLegend, 506903), anti-mouse IL-12p35 PE (clone 27537) (Invitrogen, MA5-23559), anti-mouse IL-4 allophycocyanin (clone 11B11) (BioLegend, 504105), anti-mouse CD45 PerCP/Cy5.5 (clone 30-F11) (BioLegend,103131), anti-mouse GATA3 (BioLegend, 653809), anti-mouse MHC class II (I-A/I-E) Brilliant Violet 421 (clone M5/114.15.2) (BioLegend, 107636), anti-mouse MHC class II (I-A/I-E) Alexa Fluor 700 (clone M5/114.15.2) (BioLegend, 107622), anti-mouse CD11c allophycocyanin/Cy7 (clone N418) (BioLegend, 117323), anti-mouse/human CD11b PE/Cy7 (clone M1/70) (BioLegend, 101216), anti-mouse/human CD11b Brilliant Violet 605 (clone M1/70) (BioLegend, 101237), anti-mouse CD64 PerCP/Cy5.5 (clone X54-5/7.1) (BioLegend, 139307), anti-mouse CD36 (BioLegend, 102616), anti-mouse inducible NO synthase (iNOS) (Invitrogen, 125920), anti-mouse IL-10 allophycocyanin (clone JESS-16E3) (BioLegend, 505016), anti-mouse Foxp3 Pacific Blue (clone MF-14) (BioLegend, 26410), anti-mouse/human Arg1 FITC (R&D Systems, IC5868F), anti-mouse T1ST2 allophycocyanin (clone D1H4) (BioLegend,146605), anti-mouse F4/80 PerCP/Cy5.5 (clone BM8) (BioLegend, 123127), anti-mouse MGL2/CD301B (BioLegend, 146807), anti-mouse CD19 PerCP/Cy5.5 (clone 1D3/CD19) (BioLegend, 152405), and anti-phospho-AMP-activated protein kinase (AMPK; Thr172) (CST, 2535).

To measure fatty acid uptake, cells were incubated in RPMI 1640 medium containing C1-BODIPY 500/510 C12 (Life Technologies, D-382) at a final concentration of 1 μM for 15 min at 37°C (24). Cells were washed with FACS buffer.

Groups of mice (four mice per group for the Ab experiment) were i.m. administered Prevnar 13 (0.125 µg in 50 μl of ultrapure PBS). Blood was collected before and after immunization at the indicated time. Anti-pneumococcal polysaccharide (PPS) type 3 IgG was determined by ELISA. PPS Ag was obtained from American Type Culture Collection (61810463).

Groups of mice (four mice per group) were immunized intranasally with 5 μg of 2′3′-cGAMP (InvivoGen, vac-nacga23) adjuvanted OVA (InvivoGen, vac-pova) (2 µg) or OVA alone. Mice were immunized twice at a 14-d interval. Sera and bronchoalveolar lavage fluid were collected 28 d after the last immunization. The OVA-specific Abs were determined by ELISA. Secondary Abs used were anti-mouse IgG-HRP or anti-mouse IgA-HRP (SouthernBiotech, 1040-05, 7040). To determine the Ag-specific memory Th cell response, splenocytes from OVA or 2′3′-cGAMP + OVA immunized mice were stimulated with 5 μg/ml OVA for 4 d in culture. Th1, Th2, and Th17 cytokines were measured in the supernatant by an ELISA kit from Thermo Fisher Scientific (mouse IFN-γ ELISA kit, 88-7314-22; mouse IL-17a ELISA kit, 88-8711-22; mouse IL-13 ELISA kit, 88-7137-22).

BMDCs were induced with 20 ng/ml GM-CSF as before (11). Cells were activated with 10 µg/ml CDA, CDG, 2′3′-cGAMP or 5 µg/ml Rp-Rp-ssCDA or 5 µg/ml HSV DNA as before (11). Mouse TNF-α and IFN-β were measured in culture supernatant after 5 h by an ELISA kit from Thermo Fisher Scientific (mouse TNF ELISA kit, 88-7324-22; mouse IFN-β ELISA kit, PBL Assay Science, 42410).

BMDMs were induced with 20 ng/ml M-CSF as before (11). Oleic acid (OA)/linoleic acid (LA) (Sigma-Aldrich, L9655) was added on day 3 of the BMDM differentiation. Cells were harvested on day 7 for flow analysis.

Cells were lysed in Fos-choline-13 buffer (n-tridecylphosphocholine) (Anatrace, F310S) containing 1% Fos-choline-13 (w/v), 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, protease inhibitors, and phosphatase inhibitors. Cells were then homogenized with a Precellys 24: 5000 rpm (high) for 2 × 30 s with an intermediate 30-s pause. The homogenate was rotated at 4°C for 1–2 h and spun down. The supernatant was collected for immunoprecipitation with CoA-agarose beads (Sigma-Aldrich, c7013) or palmitoyl-CoA agarose beads (Sigma-Aldrich, p5297) and run on a 4–20% Bio-Rad Mini-PROTEAN gel.

MPYS/STING-knockout (KO) RAW264.7 line was obtained from InvivoGen (rawl-kostg) and transfected with HA-tagged human MPYS constructs using Lipofectamine 3000. After 48 h, transfected cells were cultured in G418 (1 mg/ml) for 2 wk to establish stable cell lines. For protein purification, ∼100 million cells were harvested and lysed in 1% CHAPS buffer (20 mM Tris-Cl [pH 7.4], 150 mM NaCl, 1% w/v CHAPS with protease inhibitors and phosphatase inhibitors). HA-tagged MPYS was pulled down with EZview Red anti-HA affinity gel (Sigma-Aldrich, e6779) in whole-cell lysates. The anti-HA beads were washed four times in 1% CHAPS buffer. HA-tagged protein was eluted with 0.1% CHAPS buffer containing 1 mg/ml HA peptide (Thermo Fisher Scientific, 26184). The eluant was concentrated with an AMICO size exclusion column (3 kDa cutoff) and quantified with a Bio-Rad protein quantification kit (Bio-Rad, 5000201EDU).

CD4+ T cells were isolated from eWAT HAQ or AQ donor mice using the CD4+ T cell purification kit (STEMCELL Technologies, 19852). After purification, CD4+ T cells were CFSE labeled with 2.5 μM (high dose) or 0.5 μM (low dose), according to the protocol from the manufacturer (Invitrogen, C34554). Then, 1.2 × 106 of HAQ and AQ CD4+ T cells were mixed at a 1:1 ratio and transferred into the eWAT of HAQ or AQ recipient mice. CD4+ T cells in the eWAT were analyzed 24 h following the transfer.

Before starting the experiment, all reagents were kept on ice. Fatty acyl–CoA (Avanti Polar Lipids, 870718 and 870719) was dissolved in 20 mM sodium acetate buffer (pH 6.4) as a 2 mM stock and stored at −80°C. MPYS protein was prepared in micelle containing 4.06 mM MPYS protein, 14 mM 1-palmitoyl-2-oleoyl-sn-glycerol-3- phosphocholine (POPC) (Avanti Polar Lipids, 850457), 25 mM HEPES (pH 7.4), and 150 mM NaCl. Fatty acyl–CoA dilutions were prepared in HEPES buffer (25 mM HEPES [pH 7.4], 150 mM NaCl). To begin the experiment, 30 μl of MPYS (in micelle) was added to 75 μl of the fatty acyl–CoA in the HEPES buffer. Thirty microliters of micelle only was added to the 75 μl of HEPES buffer as a control. The reaction was mixed and incubated at 37°C for 20 min. The protein-POPC solution was transferred in a UV-compatible cuvette (BrandTech Scientific) and excited at 295 nm. Fluorescence measurement was measured in an M3 SpectraMax fluorescence spectrophotometer (Molecular Devices). Fluorescence emission was scanned from 310 to 380 nm with a 2-nm interval at 37°C.

All data are expressed as means ± SEM. Statistical significance was evaluated using Prism 9.0 software. Statistical details including sample sizes, the numbers of biological repeats, and statistical tests used (one-way ANOVA was performed with a post hoc Tukey’s multiple comparison test or Student t test) can be found in the figure legends. A p value of <0.05 was considered significant.

All repeats are biological replications that involve the same experimental procedures on different mice. Where possible, treatments were assigned blindly to the experimenter by another individual in the laboratory. When comparing samples from different groups, samples from each group were analyzed in concert, thereby preventing any biases that may arise from analyzing individual treatments on different days.

To understand the physiological function of MPYS, we examined the natural evolution of the human MPYS gene. Based on genetic data from genome-wide (25, 26), mitochondrial DNA (27), and Y chromosomal analyses (28), anatomically modern humans (AMHs) outside of Africa descended from Africa 50,000–70,000 y ago (the out-of-Africa migration) (Fig. 1A). Human MPYS genes consist of four common nonsynonymous SNPs (population frequency > 1%) (Fig. 1B). These SNPs form four major MPYS alleles: H232, HAQ, G230A-R293Q (AQ), and Q293 (9, 11). Analyzing genotyping data from the 1000 Genomes Project (phase III) revealed striking population frequency differences of HAQ-MPYS and AQ-MPYS in human populations (Fig. 1C, 1D) (11). The HAQ allele frequency increased from 0.7% in sub-Saharan Africans (AFR) to 40% in EAS (Fig. 1E), a 57-fold increase. Meanwhile, the AQ allele frequency decreased from 22.2% in Sub-Saharan AFR to 0.5% in EAS, an ∼98% reduction (Fig. 1E). In comparison, the H232 allele frequency is similar between AFR and EAS (Fig. 1E).

FIGURE 1.

Natural selection of HAQ and AQ-MPYS during the out-of-Africa migration. (A) Timeline of the out-of-Africa migration. (B) Information of common human nonsynonymous SNPs. (C and D) Human MPYS genotypes and their frequencies in EAS and sub-Saharan AFR (Nigerians, Gambians, Kenyans, and Sierra Leoneans). (E) Allele frequency of the HAQ, AQ, and H232-MPYS in sub-Saharan AFR and EAS. (F) Carriers’ population frequencies of human MPYS alleles in sub-Saharan AFR, Europeans, South Asians, and EAS. Data are from 1000 Genomes Project phase III (11). (G) A comparison of the synonymous and nonsynonymous SNP changes between chimpanzee and human MPYS gene. dN/dS was calculated as described in Materials and Methods. (H) LD score of the HAQ allele in CHS, GBR, Puerto Rican in Puerto Rico (PUR), and Bengali in Bangladesh (BEB) populations from the 1000 Genomes Project. An r2 score >0.8 is highlighted. (I and K) AQ haplotypes were analyzed for the 19 SNPs in the LWK population described in the Materials and Methods. The minor allele nucleotides (mutations) are highlighted in red. (J and L). HAQ haplotypes were analyzed for the 19 SNPs in the CHS population described in the Materials and Methods. The minor allele nucleotides (mutations) are highlighted in red.

FIGURE 1.

Natural selection of HAQ and AQ-MPYS during the out-of-Africa migration. (A) Timeline of the out-of-Africa migration. (B) Information of common human nonsynonymous SNPs. (C and D) Human MPYS genotypes and their frequencies in EAS and sub-Saharan AFR (Nigerians, Gambians, Kenyans, and Sierra Leoneans). (E) Allele frequency of the HAQ, AQ, and H232-MPYS in sub-Saharan AFR and EAS. (F) Carriers’ population frequencies of human MPYS alleles in sub-Saharan AFR, Europeans, South Asians, and EAS. Data are from 1000 Genomes Project phase III (11). (G) A comparison of the synonymous and nonsynonymous SNP changes between chimpanzee and human MPYS gene. dN/dS was calculated as described in Materials and Methods. (H) LD score of the HAQ allele in CHS, GBR, Puerto Rican in Puerto Rico (PUR), and Bengali in Bangladesh (BEB) populations from the 1000 Genomes Project. An r2 score >0.8 is highlighted. (I and K) AQ haplotypes were analyzed for the 19 SNPs in the LWK population described in the Materials and Methods. The minor allele nucleotides (mutations) are highlighted in red. (J and L). HAQ haplotypes were analyzed for the 19 SNPs in the CHS population described in the Materials and Methods. The minor allele nucleotides (mutations) are highlighted in red.

Close modal

Notably, the most common MPYS genotype (34.3%) in EAS is R232(WT)/HAQ, not R232(WT)/R232(WT) (22.0%) (Fig. 1C). In contrast, the R232/HAQ in sub-Saharan AFR, including Nigerians, Gambians, Kenyans, and Sierra Leoneans, accounts for 0.6% of the population (Fig. 1D). In total, 63.7% of EAS are HAQ carriers whereas only 1.4% of sub-Saharan AFR have the HAQ allele (Fig. 1F). As an internal control, the carriers’ population frequency of the R232H allele of the MPYS gene does not differ between sub-Saharan AFR and EAS populations (Fig. 1F).

For human subpopulations, whereas all AFR subpopulations have an AQ allele frequency of ∼20%, most non-African subpopulations do not have the AQ allele, that is, GBR (British) (0/91), CEU (Northern/Western European) (0/99), FIN (Finnish) (0/99), TSI (Italian) (0/107), PEL (Peruvian) (0/85), MXL (Mexican) (1/64), GIH (Gujarati Indian) (0/103), ITU (Telugu Indian) (0/102), PJL (Pakistani) (0/96), STU (Sri Lankan Tamil) (1/102), KHV (Vietnamese) (0/99), CDX (Chinese Dai) (0/93), and JPT (Japanese) (1/104). In contrast, HAQ frequency is 0 in two sub-Saharan populations, ESN (Nigerian) (0/99) and MSL (Sierra Leonean) (0/85), and 1 in the YRI (Nigerian) population (1/107). In comparison, the HAQ frequency in non-African subpopulations ranges from 9 to 19% in Northern/Western Europeans, Italians; 21 to 33% in Pakistanis, Sri Lankan Tamils; 30 to 39% in Mexicans, Peruvians; and 37 to 44% in Vietnamese, CHS.

Human nonsynonymous SNPs with large allele frequency differences between subpopulations are extremely rare (29, 30). In the entire phase II HapMap, there are only 13 nonsynonymous SNPs with a frequency difference >90% between Nigerian and EAS (30). Random genetic drift is unlikely to cause a 57-fold increase or ∼98% decrease in the allele frequency in 70,000 y. The founder effect or bottleneck effect during the out-of-Africa migration was also less likely because all non-Africans have a similar HAQ increase and AQ disappearance (Fig. 1F). The human out-of-Africa migration is characterized by serial founder effects, not one founder effect (31, 32). The long-distance migrations involve movements followed by periods of settlement. Different non-African subpopulations likely had a series of different founders during the out-of-Africa migration. Finally, despite the general acceptance of the Africa origin of AMHs, the precise location of the exit from Africa is still under intense debate (33). Researchers have proposed two distinct paths. The first involves crossing over from northern Egypt to the Sinai Peninsula, whereas the second involves crossing the Bab al Mandab Strait to Yemen in the southernmost part of the Arabian Peninsula (33, 34). Each route likely had different founders. Yet, all of the non-Africa AMHs had the same increase in the HAQ allele and the loss of the AQ allele (Fig. 1F).

Importantly, HAQ is a dominant MPYS allele. We first showed that cotransfecting HAQ with WT MPYS inhibited WT MPYS-induced type I IFN expression in HEK293 cells in 2011 (9). Later, we found that PBMCs from HAQ/WT individuals had decreased type I IFNs to CDNs (11). Most recently, in a clinical trial (ClinicalTrials.gov ID NCT02471014), we found that HAQ heterozygous individuals had reduced responses to the Pneumovax 23 vaccine compared with the WT/WT individuals (15). Mechanistically, MPYS exists as a homodimer (35) and is activated as oligomers (36, 37). Thus, the presence of HAQ likely impairs the WT-MPYS function.

Lastly, the goodness-of-fit test of HWE for HAQ and AQ alleles in the EAS and sub-Saharan AFR revealed a well HWE. Considering the drastic allele frequency differences of HAQ and AQ that exist in all non-African subpopulations versus sub-Saharan AFR (Fig. 1F), we suspected that the natural selection of HAQ and AQ happened in the past during the out-of-Africa migration. Future characterization of large numbers of ancient DNA from non-African AMHs will be very helpful.

We first examined the evolution of MPYS genes by comparing human and chimpanzee MPYS genes. The human and chimpanzee share a common ancestry ∼6 million years ago. The human and chimpanzee MPYS proteins are 99.5% identical and differed by only two mutations: R78W and G230A (Fig. 1G). Thus, the MPYS gene is well-preserved between humans and chimpanzees. We further applied the dN/dS analysis to determine the evolutionary selection between human and chimpanzee MPYS genes. dN/dS measures selection between species by comparing the rate of nonsynonymous substitutions (neutral change) per site with the rate of synonymous substitutions (functional change) per site (38). There are more synonymous than nonsynonymous substitutions between chimpanzee and human MPYS genes (Fig. 1G). The dN/dS value between chimpanzees and humans, calculated by the PAL2NAL program (39), is 0.1571, which is well below 1. The interpretation is that the MPYS gene structure is well conserved to preserve its function possibly by ongoing negative/purifying selection against deleterious mutations in the MPYS gene.

Next, we applied FST to compare the human MPYS gene in two subpopulations of AMHs: the LWK (East Africa; 99 individuals) and CHS (East Asia; 105 individuals). CHS has the highest HAQ allele frequency (44%) in all human subpopulations. Meanwhile, Africa has the highest levels of human genetic variation in the world. We chose LWK because East Africa is the place of the out-of-Africa migration (40). FST measures the level of genetic structure between two populations (41). FST ranges from 0 (no genetic differentiation) to 1 (complete genetic differentiation). The rationale is that the FST at the locus where natural selection occurs would be larger than the FST at other loci where the differences are due to random genetic drift. We calculated the FST scores of 21 SNPs covering 18.5 kb of the genomic region that contains the human MPYS gene (Supplemental Fig. 1A). The highest FST score (0.2474) belongs to rs11554776 (the SNP separating HAQ and AQ alleles). However, the FST score of 0.2474 does not deviate much from the average FST score of 0.19 reported previously between sub-Sahara AFR and EAS populations (42, 43). HAQ is a dominant allele where heterozygosity also has the selective advantage that likely masked the FST signal of rs11554776. Because the analysis from FST is inconclusive, we turned to the haplotype analysis in LWK and CHS populations.

The HAQ allele consists of three SNPs in LD covering an ∼4-kb genomic region (9). The best indicator of recent positive natural selection is a local elevation of allele frequency together with long stretches of LD (34, 41, 4446). The rationale is that under neutral selection, new variants require a long time to reach high frequencies (if it happens) in the population. Because of recombination and mutation, the LD around SNPs will decay substantially during this long period. In contrast, alleles under natural selection will reach high frequency rapidly and maintain long LD. We thus focus on the LD of SNPs covering the human MPYS genes in the LWK and CHS populations.

The human MPYS gene is ∼7.2 kb long. The LD data from the 1000 Genomes Project show that the HAQ allele in non-Africans (e.g., CHS, GBR, Puerto Rican in Puerto Rico [PUR], and Bengali in Bangladesh [BEB]) are in LD with five to six SNPs in the MPYS gene with an r2 near 1 (Fig. 1H). Furthermore, an additional four to five SNPs with r2 > 0.8 were found at the 5′ and 3′ ends of the MPYS gene (Fig. 1H). In total, the HAQ haplotype in non-Africans is in LD with a length of 18 kb (Fig. 1H). As a reference, another common MPYS allele, the R232H allele (rs1131769) in the CHS population, does not exist in LD (Supplemental Fig. 1B). In contrast, the LWK population does not have this long LD in the MPYS gene (Supplemental Fig. 1C, 1D). The intrinsic determinants of LD (rates of recombination, mutation, gene conversion) are the same in the MPYS gene across these populations (LWK versus CHS). The long LD HAQ haplotype found in all non-Africans thus reflects environmental/extrinsic determinants during the out-of-Africa migration.

To understand the evolution of the long LD HAQ haplotypes, we dephased all of the HAQ and AQ genotypes in LWK and CHS populations based on 19 SNPs (minor allele frequency > 1%) in the 18-kb region. We detected 10 AQ and 2 HAQ subhaplotypes in LWK, and 6 HAQ subhaplotypes in the CHS population (Fig. 1I, 1J). Because the nucleotide mutation number is proportional to the time of the evolution (one mutation in 30 million or 100 million bp per generation in humans) (1720), we reconstructed the evolution of AQ and HAQ haplotypes in LWK and CHS based on numbers of mutations in each subhaplotype. First, based on the mutation pattern, HAQ-2 is the founder HAQ haplotype that was derived from the AQ-1 subhaplotype in LWK, whereas HAQ-1 is a direct decedent of HAQ-2 (Fig. 1I). Second, among the six HAQ subtypes, HAQ-1 is dominant with a 34.76% haplotype population frequency in CHS followed by HAQ-2 with a 4.29% population frequency (Fig. 1J, 1L). Notably, besides the rs11554776, the dominant HAQ-1 haplotype contains the additional SNPs (rs7380062 and rs75746446) with high FST scores (Supplemental Fig. 1A), suggesting that the HAQ-1 subhaplotype is distinct between LWK and CHS. In contrast, the allele frequency of AQ subhaplotypes in LWK fits a good stepwise pattern (Fig. 1K). The AQ-1 subtype has the highest frequency (7.58%) in LWK immediately followed by AQ-2 (5.56%), AQ-3 (2.5%), and AQ-4 (1.5%) (Fig. 1I, 1K), reflecting a pattern of random genetic drift.

Third, the HAQ-1 subhaplotype has two fewer mutations than the most recent subtype HAQ-5. Using the one mutation in 30 million bp per generation mutation rate, we calculated the youngest possible age of the HAQ-1 allele as ∼70,000 y, indicating that the dominant HAQ-1 haplotypes exist before the out-of-Africa migration. The age of HAQ-1 could be twice as old as 70,000 y, as some recent studies concluded that the human mutation rate is twice as slow as previously thought (1720). Nevertheless, the HAQ-1 haplotype frequency in the LWK population is 1% (in Africa) (Fig. 1K) compared with ∼35% in CHS (in the Far East) (Fig. 1L), which strongly suggests that the HAQ-1 haplotype underwent a positive natural selection outside of Africa. Overall, the HAQ-1 haplotype has a high population frequency (∼35%) and long LD (18 kb) that signals a positive natural selection during the out-of-Africa migration.

Finally, the most important question for natural selection concerns determining what the advantageous and deleterious biological traits were of the HAQ and AQ alleles, respectively. Most human traits are influenced by multiple genes. The HAQ and AQ alleles differ by one amino acid. Yet, the HAQ allele became dominant whereas the AQ allele disappeared in EAS. Could a single amino acid change in a gene have made such an impact on human fitness during the out-of-Africa migration?

To answer this question, we built mouse models of HAQ and AQ.

We previously generated a mouse model of HAQ. To understand the physiological function of AQ in vivo, we generated an AQ mouse. These mice were made in the same C57BL/6NJ ES line as the HAQ mouse (11) (Supplemental Fig. 2A). Genomic sequencing confirmed the presence of A229 and Q292 mutations (the equivalent of A230 and Q293 in human STING) but not the H71 mutation (Supplemental Fig. 2B). Similar to the HAQ mice (11), AQ mice had lower MPYS protein expression than did wild-type mice (Fig. 2A).

FIGURE 2.

Adult HAQ mice have more energy storage than do AQ mice on a chow diet. (A) Western blot of white adipocytes from indicated mice probed with anti-MPYS/STING Ab. Data are representative of three independent experiments. (B) BMDCs from WT, HAQ, AQ, and MPYS−/− mice were activated by 10 μg/ml CDA, CDG, 2′3′-cGAMP, or 5 μg/ml Rp Rp-ssCDA, HSV-DNA for 5 h. Mouse IFN-β and TNF were measured in cell supernatants. Data are representative of three independent experiments. (C) Body weight was determined weekly in the indicated mice (n = 5–8 mice/group). Data are representative of three independent experiments. (D) Body composition was measured in the unanesthetized mouse by a quantitative magnetic resonance method using an EchoMRI 700 analyzer (EchoMRI, Houston, TX) (n = 5–8 mice/group). Data are representative of three independent experiments. (E) An i.p. glucose tolerance test (IPGTT) was performed as described in Materials and Methods (n = 5–8 mice/group). Data are representative of three independent experiments. (F) The weights of brown adipose tissue (BAT), inguinal white adipose tissue (iWAT), and epididymal white adipose tissue (eWAT) from indicated 6-mo-old male mice were recorded (n = 5–8 mice/group). Data are representative of three independent experiments. (G and H) H&E stain of eWAT from mice in (G) showing the unilocular lipid droplets (LDs). Original magnification ×20. The size of LDs was calculated by the Adiposoft plug-in for Fiji. Data are representative of three independent experiments. (I) Total triglyceride in 100 mg of eWAT was quantified with a BioVision triglyceride kit (catalog no. K622). Data are representative of three independent experiments. (J) Total numbers of adipocytes in eWAT were calculated by dividing the total eWAT weight by the average weight of an adipocyte. The average weight of an adipocyte was calculated by multiplying the size of an adipocyte by the density of an adipocyte (∼0.9 g/cm3). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (D, E, and G) or unpaired student t test (F and H–J). *p < 0.05, **p < 0.01. n.d., not detected; n.s., not significant.

FIGURE 2.

Adult HAQ mice have more energy storage than do AQ mice on a chow diet. (A) Western blot of white adipocytes from indicated mice probed with anti-MPYS/STING Ab. Data are representative of three independent experiments. (B) BMDCs from WT, HAQ, AQ, and MPYS−/− mice were activated by 10 μg/ml CDA, CDG, 2′3′-cGAMP, or 5 μg/ml Rp Rp-ssCDA, HSV-DNA for 5 h. Mouse IFN-β and TNF were measured in cell supernatants. Data are representative of three independent experiments. (C) Body weight was determined weekly in the indicated mice (n = 5–8 mice/group). Data are representative of three independent experiments. (D) Body composition was measured in the unanesthetized mouse by a quantitative magnetic resonance method using an EchoMRI 700 analyzer (EchoMRI, Houston, TX) (n = 5–8 mice/group). Data are representative of three independent experiments. (E) An i.p. glucose tolerance test (IPGTT) was performed as described in Materials and Methods (n = 5–8 mice/group). Data are representative of three independent experiments. (F) The weights of brown adipose tissue (BAT), inguinal white adipose tissue (iWAT), and epididymal white adipose tissue (eWAT) from indicated 6-mo-old male mice were recorded (n = 5–8 mice/group). Data are representative of three independent experiments. (G and H) H&E stain of eWAT from mice in (G) showing the unilocular lipid droplets (LDs). Original magnification ×20. The size of LDs was calculated by the Adiposoft plug-in for Fiji. Data are representative of three independent experiments. (I) Total triglyceride in 100 mg of eWAT was quantified with a BioVision triglyceride kit (catalog no. K622). Data are representative of three independent experiments. (J) Total numbers of adipocytes in eWAT were calculated by dividing the total eWAT weight by the average weight of an adipocyte. The average weight of an adipocyte was calculated by multiplying the size of an adipocyte by the density of an adipocyte (∼0.9 g/cm3). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (D, E, and G) or unpaired student t test (F and H–J). *p < 0.05, **p < 0.01. n.d., not detected; n.s., not significant.

Close modal

Previously, using B cells immobilized from the 1000 Genomes Project, we showed that AQ individuals generate type I IFNs in response to CDN stimulation in vitro (47). We examined CDN responses in BMDCs from AQ mice. Various CDNs and DNA stimulated similar IFN-β and TNF production in AQ BMDCs as WT BMDCs but not the HAQ or MPYS−/− BMDCs (Fig. 2B). We then examined the in vivo immune responses in the AQ mice. We found that the AQ mice responded similarly to WT mice with regard to Pneumovax 23 immunization (Supplemental Fig. 2C). In contrast, HAQ mice have impaired anti-PPS IgM response to Pneumovax 23 (11, 15). Next, we compared Prevnar 13 immunization in HAQ and AQ mice. Whereas HAQ mice were defective in Prevnar 13 responses, AQ mice and WT mice responded similarly to Prevnar 13 immunization (Supplemental Fig. 2D, 2E). Lastly, we examined the responses of AQ mice to 2′3′-cGAMP in vivo. Adjuvant 2′3′-cGAMP-induced Ab responses and memory Th1/2/17 responses were unaltered in AQ mice (Supplemental Fig. 2F–H). We concluded that AQ mice have unaltered CDN responses in vivo and in vitro.

We observed that the adult AQ mice had a decreased body weight compared with the HAQ or WT mice (Fig. 2C, Supplemental Fig. 3A–D) on a chow diet. Body composition scans indicated that decreased body weight in male AQ mice was the result of decreased fat mass (Fig. 2D, Supplemental Fig. 3E, 3F). The decreased body weight at 16 wk in male AQ mice was associated with more efficient glucose metabolization than for their HAQ counterparts (Fig. 2E).

To determine the source of increased fat mass, inguinal WAT, eWAT, and brown adipose tissue (BAT) were extracted and weighed. Male AQ mice had significantly less WAT, but not BAT, than did HAQ mice (Fig. 2F). Consistently, the expression of thermogenesis genes in BAT, including PGC-1α, PPARγ, and UCP-1, were similar in all mice (Supplemental Fig. 3G). H&E staining showed that unilocular white adipocyte size from eWAT was smaller in AQ than HAQ eWAT (Fig. 2G, 2H). Similarly, AQ mice had less eWAT diacylglycerol/triacylglycerol content than did HAQ mice (Fig. 2I). Notably, HAQ and AQ eWAT had similar numbers of total adipocytes (Fig. 2J). Altogether, these data support a previously unknown role for AQ-MPYS in suppressing fat storage in mice.

Male MPYS−/− mice had a similar lean phenotype to that of the AQ mice (Supplemental Fig. 3A). To determine the cellular mechanism by which MPYS mediates fat storage, we generated AdipopcreMPYSfl/fl mice that delete MPYS in adipocytes. AdipopcreMPYSfl/fl mice have similar body weight, fat mass, and eWAT diacylglycerol/triacylglycerol as for the MPYSfl/fl mice (Supplemental Fig. 4). Thus, adipocyte expression of MPYS is dispensable for fat storage.

The anti-inflammatory cells in eWAT, for example, regulatory T cells (Tregs), alternatively activated macrophages (M2 macrophages), eosinophils, and group 2 innate lymphoid cells (ILC2s), actively maintain metabolic homeostasis in adipose tissue (48). These cells are abundant in eWAT in lean mice but are greatly reduced in obese mice (49).

We isolated the SVF from eWAT where immune cells are located. AQ mice had fewer eWAT CD45+ leukocytes than did HAQ mice or WT mice (Fig. 3A). Similarly, AQ mice had fewer neutrophil and Ly6Chi monocytes in eWAT than di the HAQ mice (Fig. 3B, 3C). In contrast, AQ mice had more anti-inflammatory eWAT ILC2s and IL-4+ CD4+ Th2 cells than did HAQ or WT mice (Fig. 3D–F). Taken together, the data indicated that eWAT from the AQ mice have less inflammation than the HAQ mice.

FIGURE 3.

AQ mice have less eWAT inflammation than do HAQ mice. (A and E) CD45+ cells (A) and CD4+ T cells (E) were determined by flow cytometry in the stromal vascular fraction (SVF) of eWAT from WT, HAQ, AQ, and MPYS−/− mice (males, 30–32 wk old) at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (BD and F) Flow cytometry analysis of eWAT neutrophils (B), Ly6Chi monocytes (C), and ILC2s (CD45+CD25+CD127+CD3CD19B220, CD11c, CD11bCD64Nk1.1Ly6GF4/80) (D) and IL-4+ CD4 Th2 cells (F) in WT, HAQ, AQ, and MPYS−/− mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A–E) or an unpaired Student t test (F). *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 3.

AQ mice have less eWAT inflammation than do HAQ mice. (A and E) CD45+ cells (A) and CD4+ T cells (E) were determined by flow cytometry in the stromal vascular fraction (SVF) of eWAT from WT, HAQ, AQ, and MPYS−/− mice (males, 30–32 wk old) at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (BD and F) Flow cytometry analysis of eWAT neutrophils (B), Ly6Chi monocytes (C), and ILC2s (CD45+CD25+CD127+CD3CD19B220, CD11c, CD11bCD64Nk1.1Ly6GF4/80) (D) and IL-4+ CD4 Th2 cells (F) in WT, HAQ, AQ, and MPYS−/− mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A–E) or an unpaired Student t test (F). *p < 0.05, **p < 0.01, ***p < 0.001.

Close modal

Lean mice have increased numbers of eWAT M2 macrophages producing anti-inflammatory cytokines, limiting inflammation, and maintaining adipose tissue homeostasis (50). The total F4/80+ CD11b+ eWAT macrophages did not differ between HAQ and AQ mice (Fig. 4A). Macrophages can be polarized into iNOS-expressing classically activated (M1) or Arg1-expressing alternatively activated (M2) macrophages (51). The M2-like Arg1+CD301b+ eWAT macrophages were increased by 2-fold in the AQ mice compared with the HAQ or WT mice (Fig. 4B). HAQ mice had more iNOS+ macrophages in the eWAT than did the AQ mice (Fig. 4C).

FIGURE 4.

AQ mice have more eWAT M2 macrophages than do HAQ mice. (A) F4/80+CD11B+ macrophages were determined by flow cytometry in the stromal vascular fraction (SVF) of eWAT from WT, HAQ, AQ, and MPYS−/− mice (males, 30–32 wk old) at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (B and C) Flow cytometry analysis of M2 macrophages (Arg1+CD301b+) (B) and iNOS+ M1 macrophages (C) in WT, HAQ, AQ, and MPYS−/− mice at a steady state (n = 4–7 mice/group). Data are representative of three independent experiments. (DH) Flow cytometry analysis of CD36 (D), BODIPY-C12 (E), p-AMPK (F), CPT1A (G), and PPARγ (H) expression in eWAT F4/80+CD11B+ macrophages from HAQ and AQ mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (B and C) or an unpaired Student t test (D–H). *p < 0.05, **p < 0.01.

FIGURE 4.

AQ mice have more eWAT M2 macrophages than do HAQ mice. (A) F4/80+CD11B+ macrophages were determined by flow cytometry in the stromal vascular fraction (SVF) of eWAT from WT, HAQ, AQ, and MPYS−/− mice (males, 30–32 wk old) at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (B and C) Flow cytometry analysis of M2 macrophages (Arg1+CD301b+) (B) and iNOS+ M1 macrophages (C) in WT, HAQ, AQ, and MPYS−/− mice at a steady state (n = 4–7 mice/group). Data are representative of three independent experiments. (DH) Flow cytometry analysis of CD36 (D), BODIPY-C12 (E), p-AMPK (F), CPT1A (G), and PPARγ (H) expression in eWAT F4/80+CD11B+ macrophages from HAQ and AQ mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (B and C) or an unpaired Student t test (D–H). *p < 0.05, **p < 0.01.

Close modal

M2 macrophages have enhanced fatty acid oxidation (FAO) (52). Thus, we hypothesized that AQ eWAT macrophages would have enhanced FAO compared with HAQ. Efficient cellular FAO depends on enhanced fatty acid uptake via CD36, the activation of AMPK (53, 54), transcription factor PPARγ, and mitochondrial outer membrane protein CPT1 (55). CPT1a transports acyl-CoA into mitochondria for oxidation and is the rate-limiting enzyme of FAO. Inhibition of CPT1 with etomoxir increased responder T cell activation (55). In contrast, the activation of AMPK inhibits acetyl-CoA carboxylase, the rate-limiting enzyme in FAS, and enhances FAO (53, 54).

We found that AQ eWAT macrophages had enhanced fatty acid uptake indicated by increased CD36 expression and BODIPY-C12 stain (Fig. 4D, 4E). BODIPY-C12 is a fluorescent fatty acid analog. It forms excimer and emits fluorescence when incorporated into living cells. Furthermore, AMPK activation and CPT1A expression were also increased in AQ eWAT macrophages (Fig. 4F, 4G). Lastly, AQ eWAT macrophages had increased PPARγ expression (Fig. 4H) and promoted FAO (50).

eWAT inflammation has been associated with a dramatic reduction of eWAT Tregs in several animal models of obesity (5658). Adult AQ male mice had similar total eWAT CD4+ T cells as for the HAQ male mice (Fig. 3E). Importantly, the eWAT Tregs were doubled in the AQ mice compared with the HAQ mice (Fig. 5A). Furthermore, there were more IL-10+ eWAT Tregs in AQ mice than in HAQ mice (Fig. 5B).

eWAT-specific Tregs have increased expression of PPARγ, the master regulator of the eWAT Tregs (50). PPARγ promotes fatty acid metabolism and stimulates the suppressive activities of the eWAT Tregs (50). Lastly, eWAT Tregs express T1/ST2 and ICOS, which are important for their suppressive function (59). Thus, consistent with their lean phenotype, the AQ mice had markedly increased numbers of immune-suppressive eWAT Tregs.

FIGURE 5.

AQ mice have more suppressive eWAT Tregs than do HAQ mice. (A) Flow cytometry analysis of eWAT Tregs in WT, HAQ, AQ, and MPYS−/− mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (BE) Flow cytometry analysis of IL-10+ (B), PPARγ+ (C), ICOS+ (D), and T1ST2+ (E) eWAT Tregs in HAQ and AQ mice (n = 3–6 mice/group). Data are representative of three independent experiments. (FI) Flow cytometry analysis of CD36 (F), BODIPY-C12 (G), p-AMPK (H), and CPT1A (I) expression in eWAT Tregs from HAQ and AQ mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values are determined by a one-way ANOVA with Tukey’s multiple comparison test (A and C) or an unpaired Student t test (B and F–I). *p < 0.05, **p < 0.01.

FIGURE 5.

AQ mice have more suppressive eWAT Tregs than do HAQ mice. (A) Flow cytometry analysis of eWAT Tregs in WT, HAQ, AQ, and MPYS−/− mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (BE) Flow cytometry analysis of IL-10+ (B), PPARγ+ (C), ICOS+ (D), and T1ST2+ (E) eWAT Tregs in HAQ and AQ mice (n = 3–6 mice/group). Data are representative of three independent experiments. (FI) Flow cytometry analysis of CD36 (F), BODIPY-C12 (G), p-AMPK (H), and CPT1A (I) expression in eWAT Tregs from HAQ and AQ mice at steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values are determined by a one-way ANOVA with Tukey’s multiple comparison test (A and C) or an unpaired Student t test (B and F–I). *p < 0.05, **p < 0.01.

Close modal

Similar to AQ eWAT macrophages, eWAT Tregs from AQ mice strongly expressed PPARγ, T1/ST2, and ICOS (Fig. 5C–E). AQ eWAT Tregs also had higher CD36 expression and more BODIPY-C12 staining than did the HAQ eWAT Tregs, indicating increased fatty acid uptake (Fig. 5F, 5G). AMPK activation and CPT1a expression were similarly increased in AQ eWAT Tregs (Fig. 5H, 5I). Altogether, AQ mice had stronger anti-inflammatory responses in the eWAT than did the HAQ mice, which likely contribute to the lean phenotype in AQ mice.

To examine the role of MPYS in eWAT macrophage differentiation, we deleted MPYS from eWAT macrophages using CD11CcreMPYSfl/fl mice. eWAT macrophages express CD11C, which was turned on after they were in the adipose tissue. eWAT from CD11CcreMPYSfl/fl mice had increased Arg1+ M2 macrophages and decreased iNOS+ M1 macrophages (Fig. 6A, 6B), suggesting that MPYS suppresses intrinsic M2 macrophage differentiation.

FIGURE 6.

Macrophage intrinsic and extrinsic control of immune tolerance by MPYS. (A and B) Flow analysis of Arg1+, iNOS+ eWAT macrophages from MPYSfl/fl and CD11CcreMPYSfl/fl mice (n = 3 mice/group). Data are representative of two independent experiments. (C and E) Measurement of total free fatty acids (C) and unsaturated free fatty acids (E) in ex vivo eWAT culture of indicated mice. Data are representative of three independent experiments. (D) Western blot analysis of SCD-1 expression in white adipocytes from WT, MPYS−/−, HAQ, and AQ mice. Data are representative of three independent experiments. (F) Flow analysis of HAQ and AQ BMDMs. Data are representative of three independent experiments. (G and H) Flow analysis of HAQ and AQ BMDMs cultured in the presence of OA plus linoleic acid (LA) with indicated dosages for 3 d. Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A and B) or an unpaired Student t test (C and E–G). *p < 0.05, **p < 0.01. n.d., not detected.

FIGURE 6.

Macrophage intrinsic and extrinsic control of immune tolerance by MPYS. (A and B) Flow analysis of Arg1+, iNOS+ eWAT macrophages from MPYSfl/fl and CD11CcreMPYSfl/fl mice (n = 3 mice/group). Data are representative of two independent experiments. (C and E) Measurement of total free fatty acids (C) and unsaturated free fatty acids (E) in ex vivo eWAT culture of indicated mice. Data are representative of three independent experiments. (D) Western blot analysis of SCD-1 expression in white adipocytes from WT, MPYS−/−, HAQ, and AQ mice. Data are representative of three independent experiments. (F) Flow analysis of HAQ and AQ BMDMs. Data are representative of three independent experiments. (G and H) Flow analysis of HAQ and AQ BMDMs cultured in the presence of OA plus linoleic acid (LA) with indicated dosages for 3 d. Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A and B) or an unpaired Student t test (C and E–G). *p < 0.05, **p < 0.01. n.d., not detected.

Close modal

Adipocytes produce and secrete FFAs that influence eWAT immunity (60, 61). Macrophages and Tregs from AQ eWAT had enhanced fatty acid uptakes (Figs. 4, 5). We assessed fatty acid metabolism in eWAT. The total FFA from eWAT was similar in HAQ and AQ mice (∼0.2 mmol/l) (Fig. 6C). However, SCD-1 was increased in white adipocytes of MPYS−/− and AQ mice compared with WT and HAQ mice (Fig. 6D). SCD-1 catalyzes a rate-limiting step in the synthesis of unsaturated fatty acids. The unsaturated FFA from eWAT was ∼4 mg/ml in AQ and MPYS KO but ∼0.5 mg/ml for WT and HAQ (Fig. 6E). The m.w. of palmitic acid (C16) is 256.42 g/mol. Thus, unsaturated FFA is ∼1% of total FFA. We concluded that AQ eWAT secreted more unsaturated fatty acids than did WT or HAQ eWAT (Fig. 6E).

Enhanced unsaturated fatty acid uptake leads to FAO that contributed to immune tolerance phenotypes in macrophages and Tregs (60, 61). We developed BMDMs from HAQ and AQ mice. We found that HAQ BMDMs had elevated Arg1+iNOS+ cells compared with AQ (Fig. 6F). Culturing HAQ and AQ BMDMs with OA (18:1), the principal product of SCD-1, and another unsaturated fatty acid linoleic acid (LA, 18:2) reduced iNOS expression in HAQ and AQ BMDMs (Fig. 6G). Furthermore, OA/LA culture enhanced Arg1 expression in AQ BMDMs (Fig. 6H). Thus, M2 macrophage differentiation in AQ eWAT might be enhanced by increased unsaturated FFA from AQ white adipocytes.

We next asked whether HAQ or AQ controls eWAT Tregs intrinsically or extrinsically. We performed adoptive cell transfer of CD4+ T cells into eWAT of adult HAQ and AQ male mice. A mixture (1:1) of CFSE-labeled eWAT CD4+ T cells from adult HAQ or AQ mice was injected into the eWAT of HAQ or AQ mice, respectively (Fig. 7A). After 24 h, eWAT was harvested from the recipient mice and CFSE+ cells were examined. CFSE+ donor HAQ and AQ cells were recovered at the input (1:1) ratio in the HAQ and AQ recipient mice (Fig. 7B). Recovered CFSE+ cells contained similar CD4+CD25+ Tregs in both HAQ and AQ recipient mice (Fig. 7C). Thus, adoptive cell transfer confers no proliferation or survival advantage for CD4+ T cells in either HAQ or AQ mice. Next, we examined fatty acid metabolism in transferred CFSE+ donor HAQ or AQ cells in HAQ and AQ recipient mice by CD36 expression.

FIGURE 7.

Treg-intrinsic controls of fatty acid metabolism by AQ and HAQ. (A) Cartoon illustrating the adoptive cell transfer (intra-eWAT) of mixed (1:1 ratio, 1.2 × 106 each) HAQ and AQ CD4+ T cells into HAQ and AQ recipient mice, respectively. (B) Flow analysis of CFSEhi (HAQ) and CFSElo (AQ) cells in eWAT in the recipient mice (n = 3 mice/group). Data are representative of two independent experiments. (C) Flow analysis of CFSE+ CD4 T cells in eWAT from the recipient mice from (B). Data are representative of two independent experiments. (DI) Flow analysis of eWAT CD4+CD25+ (D–F) and CD4+CD25+ (G–I) cells in the HAQ and AQ recipient mice (n = 3 mice/group). Data are representative of two independent experiments. (J and K). Flow analysis of steady-state Tregs at mesenteric lymph nodes (J) and lung (K) in adult HAQ and AQ mice (n = 3 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (F and I) or an unpaired student t test (J and K). **p < 0.01. n.s., not significant.

FIGURE 7.

Treg-intrinsic controls of fatty acid metabolism by AQ and HAQ. (A) Cartoon illustrating the adoptive cell transfer (intra-eWAT) of mixed (1:1 ratio, 1.2 × 106 each) HAQ and AQ CD4+ T cells into HAQ and AQ recipient mice, respectively. (B) Flow analysis of CFSEhi (HAQ) and CFSElo (AQ) cells in eWAT in the recipient mice (n = 3 mice/group). Data are representative of two independent experiments. (C) Flow analysis of CFSE+ CD4 T cells in eWAT from the recipient mice from (B). Data are representative of two independent experiments. (DI) Flow analysis of eWAT CD4+CD25+ (D–F) and CD4+CD25+ (G–I) cells in the HAQ and AQ recipient mice (n = 3 mice/group). Data are representative of two independent experiments. (J and K). Flow analysis of steady-state Tregs at mesenteric lymph nodes (J) and lung (K) in adult HAQ and AQ mice (n = 3 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (F and I) or an unpaired student t test (J and K). **p < 0.01. n.s., not significant.

Close modal

Compared to endogenous CFSECD4+CD25+ T cells, CFSE+CD4+CD25+ cells transferred from AQ mice retained high CD36 expression regardless of the genotype of the recipient mice (Fig. 7D–F). Similarly, CFSE+CD4+CD25+ cells transferred from HAQ mice had low CD36 expression regardless of recipient mice (Fig. 7D–F). Thus, HAQ and AQ control of CD36 in Tregs is likely cell intrinsic. As an internal control, CD36 expression levels on transferred CFSE+CD4+CD25 HAQ and AQ non-Tregs were similar to the endogenous CD4+CD25 HAQ or AQ eWAT T cells (Fig. 7G–I), indicating that the CFSE+CD4+CD25 non-Tregs adapted to their new environment.

Taken together, these data suggested that HAQ and AQ controlled Treg fatty acid metabolism intrinsically. We thus hypothesized that AQ mice would have increased Tregs in tissues other than eWAT. Indeed, we found that, at steady state, mesenteric lymph nodes and lungs from AQ mice had increased Tregs compared with HAQ mice (Fig. 7J, 7K). We propose that by controlling tissue Treg production, AQ-MPYS promotes, while HAQ-MPYS suppresses, immune tolerance at a steady state.

Type I IFN induction is the hallmark function of MPYS/STING (62). AQ can induce type I IFNs whereas HAQ cannot (Fig. 2). IFN-β can induce immune tolerance (24, 63). To exclude the role of type I IFNs in the lean phenotypes in the AQ mice, we generated AQ/IFNAR1−/− mice to ablate type I IFNs signals in the AQ mice. The AQ/IFNAR1−/− mice had decreased eWAT weight and lipid droplet size similar to the AQ mice (Fig. 8A–C). Furthermore, the AQ/IFNAR1−/− mice had similar eWAT Tregs and M2-like macrophages as the AQ mice (Fig. 8D, 8G). Lastly, PPARγ, CPT1a, and AMPK activation, key players in FAO, were comparable in eWAT Tregs (Fig. 8E, 8F) and macrophages (Fig. 8H–J) from AQ and AQ/IFNAR1−/− mice. Thus, type I IFNs are dispensable for the enhanced eWAT immune tolerance and lean phenotype in AQ mice.

FIGURE 8.

AQ/IFNAR1−/− mice have similar BW and eWAT Tregs and M2 macrophages as for the AQ mice. (A) Weight of eWAT from indicated 6-mo-old male mice was recorded (n = 5–8 mice/group). Data are representative of three independent experiments. (B and C). H&E stain of eWAT from mice in (A) showing the unilocular lipid droplets (LDs). Original magnification ×40. The size of LDs was calculated by the Adiposoft plug-in for Fiji. Data are representative of three independent experiments. (DF) Flow cytometry analysis of eWAT Tregs in AQ mice and their AQ/IFNAR1−/− littermates at a steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (GJ) Flow cytometry analysis of eWAT macrophages in AQ mice and their AQ/IFNAR1−/− littermates at a steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A and B) or an unpaired student t test (D and G). *p < 0.05. n.s., not significant.

FIGURE 8.

AQ/IFNAR1−/− mice have similar BW and eWAT Tregs and M2 macrophages as for the AQ mice. (A) Weight of eWAT from indicated 6-mo-old male mice was recorded (n = 5–8 mice/group). Data are representative of three independent experiments. (B and C). H&E stain of eWAT from mice in (A) showing the unilocular lipid droplets (LDs). Original magnification ×40. The size of LDs was calculated by the Adiposoft plug-in for Fiji. Data are representative of three independent experiments. (DF) Flow cytometry analysis of eWAT Tregs in AQ mice and their AQ/IFNAR1−/− littermates at a steady state (n = 3–6 mice/group). Data are representative of three independent experiments. (GJ) Flow cytometry analysis of eWAT macrophages in AQ mice and their AQ/IFNAR1−/− littermates at a steady state (n = 3–6 mice/group). Data are representative of three independent experiments. Graphs represent the mean with error bars indicating the SEM. The p values were determined by a one-way ANOVA with Tukey’s multiple comparison test (A and B) or an unpaired student t test (D and G). *p < 0.05. n.s., not significant.

Close modal

AQ eWAT Tregs and macrophages have altered fatty acid metabolism. Central to fatty acid metabolism is fatty acyl–CoA, the activated form of fatty acid. Acyl-CoA has the same 3′-phospho-adenosine group found in 2′3′-cGAMP, the ligand for MPYS/STING (Fig. 9A). We hypothesized that MPYS regulates fatty acid metabolism by directly binding acyl-CoA. Furthermore, we hypothesized that the interaction of MPYS–acyl-CoA happens at homeostasis, that is, with a physiological concentration of normal cellular acyl-CoA.

FIGURE 9.

Human MPYS interacts with a physiological concentration of acyl-CoA via its N-terminal region. (A) Structure of acyl-CoA. (B) BMDMs from WT and MPYS-KO mice were lysed in 1% Fos-choline 13 buffer. CoA beads and palmitoyl-CoA beads were used for immunoprecipitation. The blot was probed for MPYS with anti-STING Ab. Data are representative of three independent experiments. (C and D). MPYS/STING-KO RAW264.7 cells stably expressing the indicated human MPYS mutants were lysed in Fos-choline 13 buffer as in (B). Palmitoyl-CoA beads were used for immunoprecipitation. The blot was probed with HA-HRP. Data are representative of three independent experiments. (E) Cartoon illustrating the formation of a POPC-based lipid bilayer with full-length recombinant human MPYS. The four tryptophan residues in MPYS are highlighted in green. (F) WT human MPYS was solubilized in POPC buffer as in (E). cGAMP, stearoyl-CoA, or oleoyl-CoA at the indicated concentrations was added in the MPYS-POPC solution. After 20 min, the MPYS-POPC solution was mixed with a loading buffer (0.5% SDS) and run in a running buffer with 0.1% SDS on a 4–20% Mini-PROTEAN TGX gel (no SDS) for 1.5 h. The blot was probed with anti-STING Ab as in (B). Data are representative of three independent experiments. (GJ) WT, C88xxC91, HAQ, and AQ human MPYS in POPC buffer were mixed with stearoyl-CoA as in (E). The mixed solution was excited at 295 nm and read at emission from 312 to 380 nm with a 2-nm interval. Data are representative of three independent experiments. (K) A summary of the λmax for WT, CxxC, HAQ, and AQ with 1–40 nm acyl-CoA. (L) Cartoon illustrating the interaction of acyl-CoA with the TM2 and TM2–TM3 linker of MPYS.

FIGURE 9.

Human MPYS interacts with a physiological concentration of acyl-CoA via its N-terminal region. (A) Structure of acyl-CoA. (B) BMDMs from WT and MPYS-KO mice were lysed in 1% Fos-choline 13 buffer. CoA beads and palmitoyl-CoA beads were used for immunoprecipitation. The blot was probed for MPYS with anti-STING Ab. Data are representative of three independent experiments. (C and D). MPYS/STING-KO RAW264.7 cells stably expressing the indicated human MPYS mutants were lysed in Fos-choline 13 buffer as in (B). Palmitoyl-CoA beads were used for immunoprecipitation. The blot was probed with HA-HRP. Data are representative of three independent experiments. (E) Cartoon illustrating the formation of a POPC-based lipid bilayer with full-length recombinant human MPYS. The four tryptophan residues in MPYS are highlighted in green. (F) WT human MPYS was solubilized in POPC buffer as in (E). cGAMP, stearoyl-CoA, or oleoyl-CoA at the indicated concentrations was added in the MPYS-POPC solution. After 20 min, the MPYS-POPC solution was mixed with a loading buffer (0.5% SDS) and run in a running buffer with 0.1% SDS on a 4–20% Mini-PROTEAN TGX gel (no SDS) for 1.5 h. The blot was probed with anti-STING Ab as in (B). Data are representative of three independent experiments. (GJ) WT, C88xxC91, HAQ, and AQ human MPYS in POPC buffer were mixed with stearoyl-CoA as in (E). The mixed solution was excited at 295 nm and read at emission from 312 to 380 nm with a 2-nm interval. Data are representative of three independent experiments. (K) A summary of the λmax for WT, CxxC, HAQ, and AQ with 1–40 nm acyl-CoA. (L) Cartoon illustrating the interaction of acyl-CoA with the TM2 and TM2–TM3 linker of MPYS.

Close modal

To detect protein–lipid interaction, we lysed cells in Fos-choline 13 detergent (64). Fos-choline detergents are structurally similar to phospholipids containing an acyl chain linked to choline through a phosphodiester bond. We used palmitoyl-CoA–conjugated beads to pull down palmitoyl-CoA interacting proteins. CoA-conjugated beads were used as a control for palmitoyl-CoA–specific interacting protein. We found that only palmitoyl-CoA, not CoA, beads pulled down MPYS in BMDMs (Fig. 9B), suggesting that MPYS specifically interacted with fatty acyl–CoA via its acyl chain.

MPYS has five hydrophobic domains at its N-terminal (Fig. 9E). We hypothesized that MPYS binds to acyl-CoA via its N-terminal region. To test this, we used an MPYS/STING-KO RAW264.7 line and generated RAW264.7 lines stably expressing HA-tagged human MPYS (1–160), MPYS (1–240), MPYS (41–379), MPYS (81–379), and MPYS (151–379). To study MPYS function at the steady state and avoid MPYS/STING activation during the DNA transfection, we used stable cell lines. MPYS/STING-KO RAW264.7 cells were transfected with the human MPYS constructs and held for 2 wk to select stable lines. We then lysed cells in Fos-choline 13 buffer and pulled down MPYS with palmitoyl-CoA–conjugated beads. MPYS (1–160) and MPYS (1–240), but not MPYS (151–379), precipitated with palmitoyl-CoA (Fig. 9C). Thus, the first 160-aa region of MPYS is sufficient for the interaction with acyl-CoA. Furthermore, MPYS (41–379), but not MPYS (81–379), was pulled down by palmitoyl-CoA beads (Fig. 9D), suggesting that MPYS (aa 41–81) is required for acyl-CoA binding. The aa 41–81 region of MPYS encompasses the second transmembrane (TM) (aa 45–69) and the TM2–TM3 linker (aa 69–91) (Fig. 9E).

The acyl-CoA binding region encompasses aa 71, which accounts for the difference between HAQ and AQ. We hypothesized that HAQ might have an altered acyl-CoA binding from AQ. To investigate the acyl-CoA binding by HAQ and AQ proteins at a physiological concentration of cellular acyl-CoA, we measured MPYS intrinsic tryptophan fluorescence (ITF).

Acyl-CoA binds to aa 41–81 in MPYS where aa 82 is a tryptophan (W82). Human and mouse MPYS has a total of four tryptophan residues, all in the N-terminal region, that is, W34, W82, W119, and W161 (Fig. 9E). Tryptophan fluorescence is highly sensitive to the local microenvironment, which is determined by protein conformation. W34, W119, and 161 are in the hydrophobic domains (TM1, TM4, and dimerization region) whereas W82 is exposed to the cytosol (Fig. 9E). We hypothesized that acyl-CoA binding induces a conformational change, affects the local microenvironment of W82, and alters MPYS ITF.

MPYS is a multitransmembrane protein. To mimic the physiological interaction of MPYS with acyl-CoA, we carried out the reaction in 4 mM POPC, 25 mM HEPES buffer (pH 7.4), and 150 mM NaCl buffer (65). POPC forms a lipid bilayer and mimics mammalian phospholipid composition in the cell membrane where MPYS localizes (Fig. 9E).

Full-length human HA-MPYS protein was purified from the MPYS/STING-deficient mouse RAW264.7 cells stably expressing human MPYS and solubilized in the POPC buffer. Cellular MPYS exists as a dimer (35) and can form tetramer or other higher-order multimers on cell membranes (36). Using a seminative gel (with 0.5% SDS loading buffer and 0.1% SDS running buffer, and no SDS in the PAGE gel), we confirmed that MPYS in POPC existed as a dimer and multimer with very few monomer MPYS (Fig. 9F). Furthermore, examining MPYS (1–160) and MPYS (151–379), we confirmed that MPYS multimer formation depended on the N-terminal region (Supplemental Fig. 5A, 5B). The addition of stearoyl-CoA or oleoyl-CoA did not dramatically alter MPYS multimer or dimer formation in the POPC buffer (Fig. 9F).

To determine ITF, WT human MPYS in POPC buffer was excited at 295-nm wavelength to exclusively detect tryptophan fluorescence, not phenylalanine or tyrosine fluorescence. ITF was recorded from emission wavelength 310–380 nm with a 2-nm interval to determine the emission maximum (λmax). Tryptophan is considered buried and in a “non-polar” environment when its λmax is less than ∼330 nm. If the λmax is longer than ∼330 nm, tryptophan is considered in a “polar” environment, which indicates solvent exposure. The WT MPYS in the POPC buffer had a λmax of ∼324 nm (Fig. 9G), suggesting that the four tryptophans in the N-terminal of MPYS are buried.

Next, we examined ITF upon acyl-CoA binding. Given that the physiological concentration of cellular free acyl-CoA in normal conditions does not exceed 200 nM and is most likely <5 nM (66), we examined physiological-relevant acyl-CoA concentrations (1–40 nm) in inducing MPYS ITF. Adding 5 nM acyl-CoA to MPYS POPC solution lowered the λmax from 324 to 322 nm (Fig. 9G). The λmax was further decreased to 320 nm upon 30 nM acyl-CoA (Fig. 9G). Thus, a WT human MPYS-bound physiological concentration (5 nM) of acyl-CoA leads to a decrease of λmax by 4 nm. W82 is the only exposed tryptophan in MPYS (Fig. 9E). The decreased λmax indicated that acyl-CoA binding increased the hydrophobic environment of W82.

Near W82, in the same cytosolic TM2–TM3 linker, are Cys88 and Cys91 (Fig. 9E), which are essential for MPYS/STING function in type I IFN production (67, 68). Furthermore, Cys91 is the target of STING antagonist H151, C167 (69). We hypothesized that the C88S–C91S (CxxC) mutant had an altered ITF. Indeed, the ITF λmax for human CxxC was 330 nm, 6 nm higher than the WT human MPYS (Fig. 9H). CxxC still forms multimers in the POPC buffer (Supplemental Fig. 5C). Thus, mutating Cys88 and Cys91 to serines decreased the hydrophobicity of W82, likely exposing the TM2–TM3 linker.

We next examined acyl-CoA–induced ITF in the C88xxC91 protein. Similar to the WT, acyl-CoA binding decreased λmax to 328 and 326 nm (Fig. 9H). Different from the WT, a higher concentration of acyl-CoA (10 and 30nM) was required for the shift of λmax (Fig. 9H).

HAQ and AQ differ by the R71H change, which is also in the TM2–TM3 linker. We hypothesized that the R71H change will alter W82 ITF, leading to the different λmax for HAQ and AQ. First, we confirmed that HAQ and AQ formed dimers and multimers in the POPC buffer (Supplemental Fig. 5D, 5E). There was no difference in multimer formation between HAQ and AQ proteins (Supplemental Fig. 5D, 5E). Next, we examined the ITF λmax of HAQ and AQ in the POPC buffer. HAQ had a λmax of 326 nm compared with the 322 nm λmax of AQ (Fig. 9I, 9J). Thus, the R71H change in HAQ increased λmax by 4 nm and likely exposed the TM2–TM3 linker.

We next examined acyl-CoA–induced λmax changes in HAQ and AQ proteins. Similar to the WT MPYS, acyl-CoA decreased the ITF λmax of HAQ by 4 nm (Fig. 9I), indicating that acyl-CoA binding increased hydrophobicity of W82 in HAQ. In contrast, acyl-CoA increased the ITF λmax of AQ by 2 nm (Fig. 9J), reducing the hydrophobicity of W82. Furthermore, whereas both HAQ and AQ bound physiological concentrations of acyl-CoA (1–5 nM), the maximum λmax shift was 2 nm in AQ compared with 4 nm in HAQ and WT (Fig. 9J, 9K). The acyl-CoA–induced λmax changes for HAQ, AQ, WT, and CxxC are summarized in (Fig. 9K.

In this study, we described a natural selection of the HAQ and AQ during the out-of-Africa migration and the function of MPYS in regulating fatty acid metabolism at homeostasis. Based on the population frequency difference, extended haplotype LD, low dN/dS score, high FST scores, and the fact that HAQ and AQ predated the out-of-Africa event, we concluded that HAQ and AQ are positively and negatively selected, respectively, during the out-of-Africa migration. Additional genetic tests for natural selection such as the calculation of extended haplotype homozygosity and analysis of ancient DNA from African and non-African AMHs will be helpful to determine the time and magnitude (e.g., selection coefficient) of the event (70, 71). The identification of a precise time for the population frequency changes in HAQ and AQ will help determine the causative environmental factor(s). Why did AQ vanish while HAQ became dominant during the out-of-Africa migration? The answers will reveal the essential role of MPYS in humans.

HAQ promotes fat storage and suppresses immune tolerance, whereas AQ has opposing effects promoting immune tolerance. This previously unsuspected role of MPYS in fatty acid metabolism and fat storage is uncoupled from the CDN-sensing and type I IFN–inducing function because whereas HAQ and WT promote fat storage, HAQ lost CDN responses. In reverse, whereas AQ and MPYS KO lost fat storage function, AQ retained CDN responses. Lastly, ablating type I IFN signaling in the AQ mice did not change its lean phenotype separating the fatty acid metabolism function from the type I IFN induction. AQ MPYS has a normal CDN–type I IFN function. The disappearance of AQ-MPYS from non-Africans suggests that the CDN–type I IFN function of MPYS is dispensable in humans. Instead, the type I IFN–independent energy metabolism function of MPYS may be critical. Supporting this notion, Akhmetova et al. (72) reported that Drosophila with STING deletion has reduced lipid storage and is sensitive to starvation. Drosophila STING lacks the conserved sequence for type I IFN induction and cannot induce type I IFNs by CDNs.

In a high-fat diet–induced obese mouse model, STING promotes insulin resistance, glucose intolerance, and metabolic disorders (7377). These discoveries, in the overnutrition mice, are in line with the proenergy storage role of MPYS/HAQ found in the present study. Nevertheless, two notable differences exist. First, our HAQ and AQ mice study used a chow diet and examined a possible homeostatic function of MPYS. Second, mechanistically, the prevailing hypothesis for the STING-mediated overnutrition-induced metabolic disorder is that obesity promotes the release of mitochondria DNA into the cytosol that activates the STING pathway in adipose tissue (74, 75, 78). In contrast, HAQ mice were defective in CDN sensing but retained fat-storage function as for the WT mice. In addition, Vila et al. (79) recently showed that the metabolic advantage observed in STING−/− mice at homeostasis was absent in cGAS−/− mice, indicating a DNA sensing–independent pathway controlling homeostatic energy metabolism. They further showed that MPYS/STING can inhibit FADS2 (fatty acid desaturase 2). Ablation of STING enhances FADS2 and increases polyunsaturated fatty acid production that inhibits inflammation (79). The emerging link between MPYS/STING and fatty acid metabolism at homeostasis is worth further investigation.

It is a new concept that MPYS directly interacts with a lipid, acyl-CoA, which is a function of the poorly characterized N-terminal region of MPYS. Acyl-CoA binds the second TM and the TM2–TM3 linker of MPYS, which contains multiple positive charged residues (Fig. 9L). The acyl chain likely interacts with the second TM domain whereas the negatively charged phosphate group in CoA interacts with the positive residues in the TM2–TM3 linker (Fig. 9L). Notably, such a model will suggest that MPYS may bind to other similar phospholipids besides acyl-CoA. Further studies on MPYS–lipid interaction may reveal a new mode of function of MPYS in health and disease.

MPYS may have a function in suppressing immune tolerance. The role of MPYS/STING in promoting inflammation is well established. The proinflammatory function of MPYS is mainly attributed to type I IFNs and TNF production after sensing CDNs or DNA. However, CDN or DNA sensing also lead to STING-independent inflammasome activation (80, 81), which promotes inflammation via the production of IL-1β and pyroptosis. Thus, DNA sensing should promote inflammation independent of STING/MPYS. Even for the autoinflammatory disease termed STING-associated vasculopathy with onset in infancy, recent studies showed that type I IFNs were dispensable for its development (82, 83). The underlying mechanisms for type I IFN–independent inflammation caused by MPYS/STING are unclear. In this study, we propose an alternative hypothesis that MPYS/STING promotes inflammation by suppressing immune tolerance. It is the loss of immune tolerance in MPYS mutants that exacerbates inflammation. Understanding the molecular mechanism by which AQ-MPYS induced immune tolerance could lead to new immunotherapies for inflammatory diseases by restoring immune tolerance.

In summary, we uncovered a function of MPYS at homeostasis promoting energy storage and suppressing immune tolerance. Further studies are needed to evaluate how this new function of MPYS/STING influences infections, inflammatory diseases, and cancer immunotherapy.

The study was based on data from HAQ and AQ mice. Further studies are needed to determine whether HAQ individuals have more energy storage than AQ individuals. Second, whether the energy storage difference is the cause for the natural selection of AQ and HAQ in non-Africans has not been established. Third, the genetic analysis relies on the available data in the 1000 Genomes Project, which lacks the Northern and Southern African subpopulations (84). African subpopulations have more genetic variation than non-Africans (84). For example, Batswana clustered with Southern African populations, but distinctly from 1000 Genomes Project African populations (85). As such, our analysis is based only on the African subpopulations in the 1000 Genomes Project.

We thank the Center for Immunology and Transplantation at the University of Florida for assistance. We thank Dr. Hong-bing Shu for the HA-MPYS/MITA mutant plasmids.

This work was supported by National Institute of Allergy and Infectious Diseases Grants AI110606, AI125999, and AI132865 and National Heart, Lung, and Blood Institute Grant HL152163 (to L.J.), as well as by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK116004 and DK125890 (to G.d.L.). S.M. was supported through The American Association of Immunologists Careers in Immunology Fellowship Program.

The online version of this article contains supplemental material.

Abbreviations used in this article:

     
  • AMH

    anatomically modern human

  •  
  • AMPK

    AMP-activated protein kinase

  •  
  • AQ

    G230A-R293Q

  •  
  • BAT

    brown adipose tissue

  •  
  • BMDC

    bone marrow–derived dendritic cell

  •  
  • CDN

    cyclic dinucleotide

  •  
  • cGAMP

    cyclic GMP-AMP

  •  
  • cGAS

    cGAMP synthase

  •  
  • CHS

    Southern Han Chinese

  •  
  • CPT1

    carnitine palmitoyltransferase 1

  •  
  • CST

    Cell Signaling Technology

  •  
  • dN/dS

    ratio of nonsynonymous-to-synonymous substitution rates

  •  
  • EAS

    East Asians

  •  
  • eWAT

    epididymal white adipose tissue

  •  
  • FAO

    fatty acid oxidation

  •  
  • FFA

    free fatty acid

  •  
  • FST

    fixation index

  •  
  • GBR

    British

  •  
  • HA

    hemagglutinin

  •  
  • HAQ

    R71H-G230A-R293Q

  •  
  • HWE

    Hardy–Weinberg equilibrium

  •  
  • ILC2

    group 2 innate lymphoid cell

  •  
  • iNOS

    inducible NO synthase

  •  
  • ITF

    intrinsic tryptophan fluorescence

  •  
  • KO

    knockout

  •  
  • LA

    linoleic acid

  •  
  • LD

    linkage disequilibrium

  •  
  • LWK

    Luhya in Webuye, Kenya

  •  
  • λmax

    emission maximum

  •  
  • OA

    oleic acid

  •  
  • POPC

    1-palmitoyl-2-oleoyl-sn-glycerol-3-phosphocholine

  •  
  • PPS

    pneumococcal polysaccharide

  •  
  • SCD-1

    stearoyl-CoA desaturase 1

  •  
  • SNP

    single-nucleotide polymorphism

  •  
  • STING

    stimulator of IFN genes

  •  
  • sub-Saharan AFR

    sub-Saharan Africans

  •  
  • SVF

    stromal vascular fraction

  •  
  • TM

    transmembrane

  •  
  • Treg

    regulatory T cell

  •  
  • WT

    wild-type

1.
Li
T.
,
Z. J.
Chen
.
2018
.
The cGAS-cGAMP-STING pathway connects DNA damage to inflammation, senescence, and cancer.
J. Exp. Med.
215
:
1287
1299
.
2.
Motwani
M.
,
S.
Pesiridis
,
K. A.
Fitzgerald
.
2019
.
DNA sensing by the cGAS-STING pathway in health and disease.
Nat. Rev. Genet.
20
:
657
674
.
3.
Li
X. D.
,
J.
Wu
,
D.
Gao
,
H.
Wang
,
L.
Sun
,
Z. J.
Chen
.
2013
.
Pivotal roles of cGAS-cGAMP signaling in antiviral defense and immune adjuvant effects.
Science
341
:
1390
1394
.
4.
Burdette
D. L.
,
K. M.
Monroe
,
K.
Sotelo-Troha
,
J. S.
Iwig
,
B.
Eckert
,
M.
Hyodo
,
Y.
Hayakawa
,
R. E.
Vance
.
2011
.
STING is a direct innate immune sensor of cyclic di-GMP.
Nature
478
:
515
518
.
5.
Liu
Y.
,
A. A.
Jesus
,
B.
Marrero
,
D.
Yang
,
S. E.
Ramsey
,
G. A. M.
Sanchez
,
K.
Tenbrock
,
H.
Wittkowski
,
O. Y.
Jones
,
H. S.
Kuehn
, et al
2014
.
Activated STING in a vascular and pulmonary syndrome.
N. Engl. J. Med.
371
:
507
518
.
6.
Jin
L.
,
A.
Getahun
,
H. M.
Knowles
,
J.
Mogan
,
L. J.
Akerlund
,
T. A.
Packard
,
A. L.
Perraud
,
J. C.
Cambier
.
2013
.
STING/MPYS mediates host defense against Listeria monocytogenes infection by regulating Ly6Chi monocyte migration.
J. Immunol.
190
:
2835
2843
.
7.
Donnelly
C. R.
,
C.
Jiang
,
A. S.
Andriessen
,
K.
Wang
,
Z.
Wang
,
H.
Ding
,
J.
Zhao
,
X.
Luo
,
M. S.
Lee
,
Y. L.
Lei
, et al
2021
.
STING controls nociception via type I interferon signalling in sensory neurons.
Nature
591
:
275
280
.
8.
Ma
F.
,
B.
Li
,
S. Y.
Liu
,
S. S.
Iyer
,
Y.
Yu
,
A.
Wu
,
G.
Cheng
.
2015
.
Positive feedback regulation of type I IFN production by the IFN-inducible DNA sensor cGAS.
J. Immunol.
194
:
1545
1554
.
9.
Jin
L.
,
L. G.
Xu
,
I. V.
Yang
,
E. J.
Davidson
,
D. A.
Schwartz
,
M. M.
Wurfel
,
J. C.
Cambier
.
2011
.
Identification and characterization of a loss-of-function human MPYS variant.
Genes Immun.
12
:
263
269
.
10.
Patel
S.
,
L.
Jin
.
2019
.
TMEM173 variants and potential importance to human biology and disease.
Genes Immun.
20
:
82
89
.
11.
Patel
S.
,
S. M.
Blaauboer
,
H. R.
Tucker
,
S.
Mansouri
,
J. S.
Ruiz-Moreno
,
L.
Hamann
,
R. R.
Schumann
,
B.
Opitz
,
L.
Jin
.
2017
.
The common R71H-G230A-R293Q human TMEM173 is a null allele.
J. Immunol.
198
:
776
787
.
12.
Nissen
S. K.
,
J. G.
Pedersen
,
M.
Helleberg
,
K.
Kjær
,
K.
Thavachelvam
,
N.
Obel
,
M.
Tolstrup
,
M. R.
Jakobsen
,
T. H.
Mogensen
.
2018
.
Multiple homozygous variants in the STING-encoding TMEM173 gene in HIV long-term nonprogressors.
J. Immunol.
200
:
3372
3382
.
13.
Ruiz-Moreno
J. S.
,
L.
Hamann
,
J. A.
Shah
,
A.
Verbon
,
F. P.
Mockenhaupt
,
M.
Puzianowska-Kuznicka
,
J.
Naujoks
,
L. E.
Sander
,
M.
Witzenrath
,
J. C.
Cambier
, et al
CAPNETZ Study Group
.
2018
.
The common HAQ STING variant impairs cGAS-dependent antibacterial responses and is associated with susceptibility to Legionnaires’ disease in humans.
PLoS Pathog.
14
:
e1006829
.
14.
Bader
C. S.
,
H.
Barreras
,
C. O.
Lightbourn
,
S. N.
Copsel
,
D.
Wolf
,
J.
Meng
,
J.
Ahn
,
K. V.
Komanduri
,
B. R.
Blazar
,
L.
Jin
, et al
2020
.
STING differentially regulates experimental GVHD mediated by CD8 versus CD4 T cell subsets.
Sci. Transl. Med.
12
:
eaay5006
.
15.
Sebastian
M.
,
C. J.
Hsiao
,
H. S.
Futch
,
R. S.
Eisinger
,
L.
Dumeny
,
S.
Patel
,
M.
Gobena
,
D. S.
Katikaneni
,
J.
Cohen
,
A. M.
Carpenter
, et al
2020
.
Obesity and STING1 genotype associate with 23-valent pneumococcal vaccination efficacy.
JCI Insight
5
:
e136141
.
16.
Jin
L.
,
K. K.
Hill
,
H.
Filak
,
J.
Mogan
,
H.
Knowles
,
B.
Zhang
,
A. L.
Perraud
,
J. C.
Cambier
,
L. L.
Lenz
.
2011
.
MPYS is required for IFN response factor 3 activation and type I IFN production in the response of cultured phagocytes to bacterial second messengers cyclic-di-AMP and cyclic-di-GMP.
J. Immunol.
187
:
2595
2601
.
17.
Xue
Y.
,
Q.
Wang
,
Q.
Long
,
B. L.
Ng
,
H.
Swerdlow
,
J.
Burton
,
C.
Skuce
,
R.
Taylor
,
Z.
Abdellah
,
Y.
Zhao
, et al
2009
.
Human Y chromosome base-substitution mutation rate measured by direct sequencing in a deep-rooting pedigree.
Curr. Biol.
19
:
1453
1457
.
18.
Campbell
C. D.
,
J. X.
Chong
,
M.
Malig
,
A.
Ko
,
B. L.
Dumont
,
L.
Han
,
L.
Vives
,
B. J.
O’Roak
,
P. H.
Sudmant
,
J.
Shendure
, et al
2012
.
Estimating the human mutation rate using autozygosity in a founder population.
Nat. Genet.
44
:
1277
1281
.
19.
Kong
A.
,
M. L.
Frigge
,
G.
Masson
,
S.
Besenbacher
,
P.
Sulem
,
G.
Magnusson
,
S. A.
Gudjonsson
,
A.
Sigurdsson
,
A.
Jonasdottir
,
A.
Jonasdottir
, et al
2012
.
Rate of de novo mutations and the importance of father’s age to disease risk.
Nature
488
:
471
475
.
20.
Ségurel
L.
,
M. J.
Wyman
,
M.
Przeworski
.
2014
.
Determinants of mutation rate variation in the human germline.
Annu. Rev. Genomics Hum. Genet.
15
:
47
70
.
21.
Brestoff
J. R.
2017
.
Isolation of immune cells from adipose tissue for flow cytometry.
Methods Mol. Biol.
1566
:
49
59
.
22.
Blaauboer
S. M.
,
S.
Mansouri
,
H. R.
Tucker
,
H. L.
Wang
,
V. D.
Gabrielle
,
L.
Jin
.
2015
.
The mucosal adjuvant cyclic di-GMP enhances antigen uptake and selectively activates pinocytosis-efficient cells in vivo.
eLife
4
:
e06670
.
23.
Mansouri
S.
,
D. S.
Katikaneni
,
H.
Gogoi
,
L.
Jin
.
2021
.
Monocyte-derived dendritic cells (moDCs) differentiate into Bcl6+ mature moDCs to promote cyclic di-GMP vaccine adjuvant-induced memory TH cells in the lung.
J. Immunol.
206
:
2233
2245
.
24.
Mansouri
S.
,
H.
Gogoi
,
M.
Pipkin
,
T. N.
Machuca
,
A. M.
Emtiazjoo
,
A. K.
Sharma
,
L.
Jin
.
2021
.
In vivo reprogramming of pathogenic lung TNFR2+ cDC2s by IFNβ inhibits HDM-induced asthma.
Sci. Immunol.
6
:
eabi8472
.
25.
Mallick
S.
,
H.
Li
,
M.
Lipson
,
I.
Mathieson
,
M.
Gymrek
,
F.
Racimo
,
M.
Zhao
,
N.
Chennagiri
,
S.
Nordenfelt
,
A.
Tandon
, et al
2016
.
The Simons Genome Diversity Project: 300 genomes from 142 diverse populations.
Nature
538
:
201
206
.
26.
Pagani
L.
,
D. J.
Lawson
,
E.
Jagoda
,
A.
Mörseburg
,
A.
Eriksson
,
M.
Mitt
,
F.
Clemente
,
G.
Hudjashov
,
M.
DeGiorgio
,
L.
Saag
, et al
2016
.
Genomic analyses inform on migration events during the peopling of Eurasia.
Nature
538
:
238
242
.
27.
van Oven
M.
,
M.
Kayser
.
2009
.
Updated comprehensive phylogenetic tree of global human mitochondrial DNA variation.
Hum. Mutat.
30
:
E386
E394
.
28.
Poznik
G. D.
,
Y.
Xue
,
F. L.
Mendez
,
T. F.
Willems
,
A.
Massaia
,
M. A.
Wilson Sayres
,
Q.
Ayub
,
S. A.
McCarthy
,
A.
Narechania
,
S.
Kashin
, et al
1000 Genomes Project Consortium
.
2016
.
Punctuated bursts in human male demography inferred from 1,244 worldwide Y-chromosome sequences.
Nat. Genet.
48
:
593
599
.
29.
Casto
A. M.
,
B. M.
Henn
,
J. M.
Kidd
,
C. D.
Bustamante
,
M. W.
Feldman
.
2012
.
A tale of two haplotypes: the EDA2R/AR Intergenic region is the most divergent genomic segment between Africans and East Asians in the human genome.
Hum. Biol.
84
:
641
694
.
30.
Coop
G.
,
J. K.
Pickrell
,
J.
Novembre
,
S.
Kudaravalli
,
J.
Li
,
D.
Absher
,
R. M.
Myers
,
L. L.
Cavalli-Sforza
,
M. W.
Feldman
,
J. K.
Pritchard
.
2009
.
The role of geography in human adaptation.
PLoS Genet.
5
:
e1000500
.
31.
DeGiorgio
M.
,
M.
Jakobsson
,
N. A.
Rosenberg
.
2009
.
Explaining worldwide patterns of human genetic variation using a coalescent-based serial founder model of migration outward from Africa.
Proc. Natl. Acad. Sci. USA
106
:
16057
16062
.
32.
Ramachandran
S.
,
O.
Deshpande
,
C. C.
Roseman
,
N. A.
Rosenberg
,
M. W.
Feldman
,
L. L.
Cavalli-Sforza
.
2005
.
Support from the relationship of genetic and geographic distance in human populations for a serial founder effect originating in Africa.
Proc. Natl. Acad. Sci. USA
102
:
15942
15947
.
33.
Bae
C. J.
,
K.
Douka
,
M. D.
Petraglia
.
2017
.
On the origin of modern humans: Asian perspectives.
Science
358
:
eaai9067
.
34.
Henn
B. M.
,
L. L.
Cavalli-Sforza
,
M. W.
Feldman
.
2012
.
The great human expansion.
Proc. Natl. Acad. Sci. USA
109
:
17758
17764
.
35.
Jin
L.
,
P. M.
Waterman
,
K. R.
Jonscher
,
C. M.
Short
,
N. A.
Reisdorph
,
J. C.
Cambier
.
2008
.
MPYS, a novel membrane tetraspanner, is associated with major histocompatibility complex class II and mediates transduction of apoptotic signals.
Mol. Cell. Biol.
28
:
5014
5026
.
36.
Shang
G.
,
C.
Zhang
,
Z. J.
Chen
,
X. C.
Bai
,
X.
Zhang
.
2019
.
Cryo-EM structures of STING reveal its mechanism of activation by cyclic GMP-AMP.
Nature
567
:
389
393
.
37.
Ergun
S. L.
,
D.
Fernandez
,
T. M.
Weiss
,
L.
Li
.
2019
.
STING polymer structure reveals mechanisms for activation, hyperactivation, and inhibition.
Cell
178
:
290
301.e10
.
38.
Hurst
L. D.
2002
.
The Ka/Ks ratio: diagnosing the form of sequence evolution.
Trends Genet.
18
:
486
487
.
39.
Suyama
M.
,
D.
Torrents
,
P.
Bork
.
2006
.
PAL2NAL: robust conversion of protein sequence alignments into the corresponding codon alignments.
Nucleic Acids Res.
34
(
Web Server
):
W609
W612
.
40.
Choudhury
A.
,
S.
Aron
,
L. R.
Botigué
,
D.
Sengupta
,
G.
Botha
,
T.
Bensellak
,
G.
Wells
,
J.
Kumuthini
,
D.
Shriner
,
Y. J.
Fakim
, et al
H3Africa Consortium
.
2020
.
High-depth African genomes inform human migration and health. [Published erratum appears in 2021 Nature 592: E26.]
Nature
586
:
741
748
.
41.
Sabeti
P. C.
,
S. F.
Schaffner
,
B.
Fry
,
J.
Lohmueller
,
P.
Varilly
,
O.
Shamovsky
,
A.
Palma
,
T. S.
Mikkelsen
,
D.
Altshuler
,
E. S.
Lander
.
2006
.
Positive natural selection in the human lineage.
Science
312
:
1614
1620
.
42.
Nelis
M.
,
T.
Esko
,
R.
Mägi
,
F.
Zimprich
,
A.
Zimprich
,
D.
Toncheva
,
S.
Karachanak
,
T.
Piskácková
,
I.
Balascák
,
L.
Peltonen
, et al
2009
.
Genetic structure of Europeans: a view from the North-East. [Published erratum appears in 2010 PLoS One 5: 10.1371/annotation/2849e182-aef5-4e2b-a5ac-0b74b30e5f48.]
PLoS One
4
:
e5472
.
43.
Elhaik
E.
2012
.
Empirical distributions of FST from large-scale human polymorphism data.
PLoS One
7
:
e49837
.
44.
Vitti
J. J.
,
S. R.
Grossman
,
P. C.
Sabeti
.
2013
.
Detecting natural selection in genomic data.
Annu. Rev. Genet.
47
:
97
120
.
45.
Sabeti
P. C.
,
D. E.
Reich
,
J. M.
Higgins
,
H. Z.
Levine
,
D. J.
Richter
,
S. F.
Schaffner
,
S. B.
Gabriel
,
J. V.
Platko
,
N. J.
Patterson
,
G. J.
McDonald
, et al
2002
.
Detecting recent positive selection in the human genome from haplotype structure.
Nature
419
:
832
837
.
46.
Voight
B. F.
,
S.
Kudaravalli
,
X.
Wen
,
J. K.
Pritchard
.
2006
.
A map of recent positive selection in the human genome. [Published errata appear in 2006 PLoS Biol. 4: e154 and 2007 PLoS Biol. 5: e147.]
PLoS Biol.
4
:
e72
.
47.
Patel
S.
,
S. M.
Blaauboer
,
H. R.
Tucker
,
S.
Mansouri
,
J. S.
Ruiz-Moreno
,
L.
Hamann
,
R. R.
Schumann
,
B.
Opitz
,
L.
Jin
.
2017
.
Response to comment on “The common R71H-G230A-R293Q human TMEM173 is a null allele”.
J. Immunol.
198
:
4185
4188
.
48.
Man
K.
,
V. I.
Kutyavin
,
A.
Chawla
.
2017
.
Tissue immunometabolism: development, physiology, and pathobiology.
Cell Metab.
25
:
11
26
.
49.
Zeng
Q.
,
X.
Sun
,
L.
Xiao
,
Z.
Xie
,
M.
Bettini
,
T.
Deng
.
2018
.
A unique population: adipose-resident regulatory T cells.
Front. Immunol.
9
:
2075
.
50.
Cipolletta
D.
,
M.
Feuerer
,
A.
Li
,
N.
Kamei
,
J.
Lee
,
S. E.
Shoelson
,
C.
Benoist
,
D.
Mathis
.
2012
.
PPAR-γ is a major driver of the accumulation and phenotype of adipose tissue Treg cells.
Nature
486
:
549
553
.
51.
Galván-Peña
S.
,
L. A.
O’Neill
.
2014
.
Metabolic reprograming in macrophage polarization.
Front. Immunol.
5
:
420
.
52.
Pearce
E. J.
,
B.
Everts
.
2015
.
Dendritic cell metabolism.
Nat. Rev. Immunol.
15
:
18
29
.
53.
Everts
B.
,
E. J.
Pearce
.
2014
.
Metabolic control of dendritic cell activation and function: recent advances and clinical implications.
Front. Immunol.
5
:
203
.
54.
Munday
M. R.
,
D. G.
Campbell
,
D.
Carling
,
D. G.
Hardie
.
1988
.
Identification by amino acid sequencing of three major regulatory phosphorylation sites on rat acetyl-CoA carboxylase.
Eur. J. Biochem.
175
:
331
338
.
55.
Malinarich
F.
,
K.
Duan
,
R. A.
Hamid
,
A.
Bijin
,
W. X.
Lin
,
M.
Poidinger
,
A. M.
Fairhurst
,
J. E.
Connolly
.
2015
.
High mitochondrial respiration and glycolytic capacity represent a metabolic phenotype of human tolerogenic dendritic cells.
J. Immunol.
194
:
5174
5186
.
56.
Feuerer
M.
,
L.
Herrero
,
D.
Cipolletta
,
A.
Naaz
,
J.
Wong
,
A.
Nayer
,
J.
Lee
,
A. B.
Goldfine
,
C.
Benoist
,
S.
Shoelson
,
D.
Mathis
.
2009
.
Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters.
Nat. Med.
15
:
930
939
.
57.
Ilan
Y.
,
R.
Maron
,
A. M.
Tukpah
,
T. U.
Maioli
,
G.
Murugaiyan
,
K.
Yang
,
H. Y.
Wu
,
H. L.
Weiner
.
2010
.
Induction of regulatory T cells decreases adipose inflammation and alleviates insulin resistance in ob/ob mice.
Proc. Natl. Acad. Sci. USA
107
:
9765
9770
.
58.
Cipolletta
D.
,
D.
Kolodin
,
C.
Benoist
,
D.
Mathis
.
2011
.
Tissular Tregs: a unique population of adipose-tissue-resident Foxp3+CD4+ T cells that impacts organismal metabolism.
Semin. Immunol.
23
:
431
437
.
59.
Shi
H.
,
H.
Chi
.
2019
.
Metabolic control of Treg cell stability, plasticity, and tissue-specific heterogeneity.
Front. Immunol.
10
:
2716
.
60.
Buck
M. D.
,
R. T.
Sowell
,
S. M.
Kaech
,
E. L.
Pearce
.
2017
.
Metabolic instruction of immunity.
Cell
169
:
570
586
.
61.
Han
S. J.
,
A.
Glatman Zaretsky
,
V.
Andrade-Oliveira
,
N.
Collins
,
A.
Dzutsev
,
J.
Shaik
,
D.
Morais da Fonseca
,
O. J.
Harrison
,
S.
Tamoutounour
,
A. L.
Byrd
, et al
2017
.
White adipose tissue is a reservoir for memory t cells and promotes protective memory responses to infection.
Immunity
47
:
1154
1168.e6
.
62.
Ishikawa
H.
,
G. N.
Barber
.
2008
.
STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. [Published erratum appears in 2008 Nature 456: 274.]
Nature
455
:
674
678
.
63.
Mansouri
S.
,
D. S.
Katikaneni
,
H.
Gogoi
,
M.
Pipkin
,
T. N.
Machuca
,
A. M.
Emtiazjoo
,
L.
Jin
.
2020
.
Lung IFNAR1hi TNFR2+ cDC2 promotes lung regulatory T cells induction and maintains lung mucosal tolerance at steady state.
Mucosal Immunol.
13
:
595
608
.
64.
Radhakrishnan
A.
,
L. P.
Sun
,
H. J.
Kwon
,
M. S.
Brown
,
J. L.
Goldstein
.
2004
.
Direct binding of cholesterol to the purified membrane region of SCAP: mechanism for a sterol-sensing domain.
Mol. Cell
15
:
259
268
.
65.
Qian
H.
,
X.
Zhao
,
R.
Yan
,
X.
Yao
,
S.
Gao
,
X.
Sun
,
X.
Du
,
H.
Yang
,
C. C. L.
Wong
,
N.
Yan
.
2020
.
Structural basis for catalysis and substrate specificity of human ACAT1.
Nature
581
:
333
338
.
66.
Faergeman
N. J.
,
J.
Knudsen
.
1997
.
Role of long-chain fatty acyl-CoA esters in the regulation of metabolism and in cell signalling.
Biochem. J.
323
:
1
12
.
67.
Jin
L.
,
L. L.
Lenz
,
J. C.
Cambier
.
2010
.
Cellular reactive oxygen species inhibit MPYS induction of IFNβ.
PLoS One
5
:
e15142
.
68.
Mukai
K.
,
H.
Konno
,
T.
Akiba
,
T.
Uemura
,
S.
Waguri
,
T.
Kobayashi
,
G. N.
Barber
,
H.
Arai
,
T.
Taguchi
.
2016
.
Activation of STING requires palmitoylation at the Golgi.
Nat. Commun.
7
:
11932
.
69.
Haag
S. M.
,
M. F.
Gulen
,
L.
Reymond
,
A.
Gibelin
,
L.
Abrami
,
A.
Decout
,
M.
Heymann
,
F. G.
van der Goot
,
G.
Turcatti
,
R.
Behrendt
,
A.
Ablasser
.
2018
.
Targeting STING with covalent small-molecule inhibitors.
Nature
559
:
269
273
.
70.
Slatkin
M.
,
B.
Rannala
.
2000
.
Estimating allele age.
Annu. Rev. Genomics Hum. Genet.
1
:
225
249
.
71.
McManus
K. F.
,
A. M.
Taravella
,
B. M.
Henn
,
C. D.
Bustamante
,
M.
Sikora
,
O. E.
Cornejo
.
2017
.
Population genetic analysis of the DARC locus (Duffy) reveals adaptation from standing variation associated with malaria resistance in humans.
PLoS Genet.
13
:
e1006560
.
72.
Akhmetova
K.
,
M.
Balasov
,
I.
Chesnokov
.
2021
.
Drosophila STING protein has a role in lipid metabolism.
eLife
10
:
e67358
.
73.
Bai
J.
,
C.
Cervantes
,
J.
Liu
,
S.
He
,
H.
Zhou
,
B.
Zhang
,
H.
Cai
,
D.
Yin
,
D.
Hu
,
Z.
Li
, et al
2017
.
DsbA-L prevents obesity-induced inflammation and insulin resistance by suppressing the mtDNA release-activated cGAS-cGAMP-STING pathway.
Proc. Natl. Acad. Sci. USA
114
:
12196
12201
.
74.
Mao
Y.
,
W.
Luo
,
L.
Zhang
,
W.
Wu
,
L.
Yuan
,
H.
Xu
,
J.
Song
,
K.
Fujiwara
,
J. I.
Abe
,
S. A.
LeMaire
, et al
2017
.
STING-IRF3 triggers endothelial inflammation in response to free fatty acid-induced mitochondrial damage in diet-induced obesity. [Published erratum appears in 2018 Arterioscler. Thromb. Vasc. Biol. 38: e60]
Arterioscler. Thromb. Vasc. Biol.
37
:
920
929
.
75.
Bai
J.
,
C.
Cervantes
,
S.
He
,
J.
He
,
G. R.
Plasko
,
J.
Wen
,
Z.
Li
,
D.
Yin
,
C.
Zhang
,
M.
Liu
, et al
2020
.
Mitochondrial stress-activated cGAS-STING pathway inhibits thermogenic program and contributes to overnutrition-induced obesity in mice.
Commun. Biol.
3
:
257
.
76.
Qiao
J. T.
,
C.
Cui
,
L.
Qing
,
L. S.
Wang
,
T. Y.
He
,
F.
Yan
,
F. Q.
Liu
,
Y. H.
Shen
,
X. G.
Hou
,
L.
Chen
.
2018
.
Activation of the STING-IRF3 pathway promotes hepatocyte inflammation, apoptosis and induces metabolic disorders in nonalcoholic fatty liver disease.
Metabolism
81
:
13
24
.
77.
Yu
Y.
,
Y.
Liu
,
W.
An
,
J.
Song
,
Y.
Zhang
,
X.
Zhao
.
2019
.
STING-mediated inflammation in Kupffer cells contributes to progression of nonalcoholic steatohepatitis.
J. Clin. Invest.
129
:
546
555
.
78.
Oduro
P. K.
,
X.
Zheng
,
J.
Wei
,
Y.
Yang
,
Y.
Wang
,
H.
Zhang
,
E.
Liu
,
X.
Gao
,
M.
Du
,
Q.
Wang
.
2022
.
The cGAS-STING signaling in cardiovascular and metabolic diseases: future novel target option for pharmacotherapy.
Acta Pharm. Sin. B
12
:
50
75
.
79.
Vila
I. K.
,
H.
Chamma
,
A.
Steer
,
M.
Saccas
,
C.
Taffoni
,
E.
Turtoi
,
L. S.
Reinert
,
S.
Hussain
,
J.
Marines
,
L.
Jin
, et al
2022
.
STING orchestrates the crosstalk between polyunsaturated fatty acid metabolism and inflammatory responses.
Cell Metab.
34
:
125
139.e8
.
80.
Abdul-Sater
A. A.
,
I.
Tattoli
,
L.
Jin
,
A.
Grajkowski
,
A.
Levi
,
B. H.
Koller
,
I. C.
Allen
,
S. L.
Beaucage
,
K. A.
Fitzgerald
,
J. P.
Ting
, et al
2013
.
Cyclic-di-GMP and cyclic-di-AMP activate the NLRP3 inflammasome.
EMBO Rep.
14
:
900
906
.
81.
Hornung
V.
,
A.
Ablasser
,
M.
Charrel-Dennis
,
F.
Bauernfeind
,
G.
Horvath
,
D. R.
Caffrey
,
E.
Latz
,
K. A.
Fitzgerald
.
2009
.
AIM2 recognizes cytosolic dsDNA and forms a caspase-1-activating inflammasome with ASC.
Nature
458
:
514
518
.
82.
Warner
J. D.
,
R. A.
Irizarry-Caro
,
B. G.
Bennion
,
T. L.
Ai
,
A. M.
Smith
,
C. A.
Miner
,
T.
Sakai
,
V. K.
Gonugunta
,
J.
Wu
,
D. J.
Platt
, et al
2017
.
STING-associated vasculopathy develops independently of IRF3 in mice.
J. Exp. Med.
214
:
3279
3292
.
83.
Luksch
H.
,
W. A.
Stinson
,
D. J.
Platt
,
W.
Qian
,
G.
Kalugotla
,
C. A.
Miner
,
B. G.
Bennion
,
A.
Gerbaulet
,
A.
Rösen-Wolff
,
J. J.
Miner
.
2019
.
STING-associated lung disease in mice relies on T cells but not type I interferon.
J. Allergy Clin. Immunol.
144
:
254
266.e8
.
84.
Choudhury
A.
,
S.
Aron
,
L. R.
Botigué
,
D.
Sengupta
,
G.
Botha
,
T.
Bensellak
,
G.
Wells
,
J.
Kumuthini
,
D.
Shriner
,
Y. J.
Fakim
, et al
2020
.
High-depth African genomes inform human migration and health. [Published erratum appears in 2021 Nature 592: E26.]
Nature
586
:
741
748
.
85.
Retshabile
G.
,
B. C.
Mlotshwa
,
L.
Williams
,
S.
Mwesigwa
,
G.
Mboowa
,
Z.
Huang
,
N.
Rustagi
,
S.
Swaminathan
,
E.
Katagirya
,
S.
Kyobe
, et al
Collaborative African Genomics Network (CAfGEN) of the H3Africa Consortium
.
2018
.
Whole-exome sequencing reveals uncaptured variation and distinct ancestry in the Southern African population of Botswana.
Am. J. Hum. Genet.
102
:
731
743
.

The authors have no financial conflicts of interest.

This article is distributed under The American Association of Immunologists, Inc., Reuse Terms and Conditions for Author Choice articles.

Supplementary data