TMEM173 Alternative Spliced Isoforms Modulate Viral Replication through the STING Pathway

In higher organisms, particularly in chordates, activation of innate immunity leads to the expression of hundreds of proteins, including IFN-β, by a coordinated activation of the transcription factors AP1, NF-κB, and IRF-3/7 (1, 2). The innate immune system activation requires sensing of pathogens by pathogen recognition receptors (PRR) (3), which recognize pathogen-associated molecular patterns (4). In eukaryotes, PRRs can be divided in two groups, membrane-bound PRR and cytoplasmic PRR. Among cytoplasmic PRR are the cytoplasmic DNA receptors (CDRs) (5). The CDR family includes proteins like DAI (DNA-dependent activator of IFN regulatory factors) (6), IFI16 (IFN-γ–inducible protein 16) (7), and DDX41 (a member of the DEAD box proteins family of helicases) (8) that, upon detection of DNA, bind and induce the activation of stimulator of IFN genes (STING) also known as MITA, MPYS, or ERIS, a protein encoded by the TMEM173 gene (9, 10). An additional CDR protein named cGAS (cyclic-GMP-AMP synthase) was identified. cGAS synthesizes cyclic-GMP-AMP (cGAMP) upon cytosolic DNA recognition (11). cGAMP acts as a second messenger that binds and activates STING (12). STING biological relevance is highlighted by the fact that STING-null mice are more susceptible to certain bacterial and viral infections (13, 14). Furthermore, although STING and cGAS protein families’ origins could be traced back to a choanoflagellate named Monosiga brevicollis, only evolutionary closer vertebrate STING proteins present high homology and contain a C-terminal tail implicated in IFN-β induction (15).

Human full-length wild-type (wt) STING is a 379 aa protein that contains four amino terminal transmembrane domains (aa 21–173) followed by a dimerization domain and a C-terminal cytoplasmic domain (aa 174–379). Human STING (hSTING) allocates predominantly in the endoplasmic reticulum. Upon activation, hSTING forms dimers or oligomers (10) and translocates to the Golgi apparatus and subsequently to perinuclear bodies by a mechanism that involves autophagocytic vesicle sorting (16). Active hSTING recruits a protein complex that can contain several proteins that activate NF-κB and IRF-3 and induces the transcription of IFN-β as well as other genes.

hSTING inhibition can be regulated by a number of cellular mechanisms, including phosphorylation by UNC-51–like kinase (ULK1) (17) and K48-like ubiquitination and degradation by E3 ubiquitin ligase 5 (18). hSTING-dependent signaling can also be inhibited during infection by viruses such as human coronavirus (19), yellow fever virus (14), hepatitis C virus (20, 21), dengue virus (12, 22, 23), or HSV type 1 (HSV-1) (24) to evade innate immune response.

Alternative splicing has been identified as an important cellular regulatory mechanism in fine-tuning with host IFN signaling activity. For instance, the alternatively spliced variants of RIG-I (25), MAVS (26), TBK1 (27), NOD2 (28), or MyD88 (29) function as dominant negative inhibitors of their respective pathways. One alternative-spliced form of hSTING has been described so far (30, 31). This truncated form lacks exon 7 and produces an hSTING protein with a C-terminal domain differing in 30 aa. This variant is unable to induce IFN and inhibits full-length wt hSTING from triggering IFN-β production.

In addition to the already reported alternative-spliced hSTING form, in this article, two new hSTING alternative-spliced variants are described. The three of them lack exon 7 (E7-less) (E7-less hSTING isoforms). They play a role in regulating IFN-β production and are involved in controlling viral replication by inhibiting wt hSTING through, at least partially, by inducing proteasome degradation of the protein. In addition, different viral infections decreased the ratio of wt hSTING versus alternative-spliced E7-less hSTING isoforms. A sequence analysis across species of TMEM173 alternative splicing transcripts revealed conservation in the expression of truncated STING isoforms, indicating a conservation of this regulatory mechanism along evolution in vertebrates.

HEK293T, HeLa, Sirc, Vero, and HuH-7.0 cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS, 1% Na-pyruvate, 1% l-glutamine 200 mM, and 1% penicillin–streptomycin (Invitrogen). THP1 cells were maintained in RPMI 1640 medium containing 10% FBS, 1% penicillin–streptomycin, 1% sodium pyruvate (Invitrogen), and 1% l-glutamine 200 mM (Invitrogen). THP1 monocytes are differentiated to macrophages by addition of 100 nM PMA (Sigma). All cells were grown at 37°C in a 5% CO₂ incubator. PBLs were obtained from 10 ml of whole blood from healthy donors. Cells were purified with a Ficoll gradient (Sigma) following the manufacturer’s instructions.

Liver biopsy specimens from four healthy donors were used to measure expression of wt and E7-less hSTING isoforms. Samples and data from the patients were provided by the Biobank of the University of Navarra and were processed following standard operating procedures approved by the Ethics and Scientific Committees and conform to following the ethical guidelines of the 1975 Declaration of Helsinki.

Detection of endogenous isoform 1 (as well as wt hSTING) was performed by using THP1 cells (3 × 10⁷) lysed in a buffer containing 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 0,1% Triton X-100, 1 mM EDTA, 1 mM 2-ME, and 5% glycerol, supplemented with a protease inhibitor mixture (P8340; SIGMA). The cell lysate was incubated with TMEM173 PrecisionAb Ab clone 4H1 diluted 1:50 (Bio-Rad) and goat anti-mouse IgG Ab conjugated to Sepharose beads at 4°C for 1–3 h. The immunoprecipitates were washed thoroughly and boiled in SDS sample buffer for Western blot analysis, using TMEM173 PrecisionAb Ab at a 1:500 Ab dilution.

For immunoprecipitation of flag M2-fused proteins, HEK293T cells seeded into 100-mm dishes were transiently transfected with the indicated plasmids using Lipofectamine 2000 (Invitrogen) and lysed the following day with lysis buffer. Cell lysates were centrifuged twice at 13,000 rpm for 20 min at 4°C and precleared with protein G agarose beads (Roche) for 4 h at 4°C. The immunoprecipitation was performed with EZview Red ANTI-FLAG-M2 Affinity Gel (Sigma) at 4°C overnight. The following day, samples were washed six times with lysis buffer. Immune-precipitated and total proteins were analyzed by Western blotting using monoclonal anti-hemagglutinin (HA) (Sigma), anti-flag M2 (Sigma), or anti-tubulin (Abcam), followed by incubation with anti-rabbit IgG or anti-mouse HRP-conjugated secondary Ab (GE Healthcare).

Blots were revealed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific).

Cells (5 × 10⁵ cells per well) were seeded in 24-well plates (Nunc, Roskilde, Denmark) and incubated with MG132 (Sigma) at 10 μM for 9 h.

HSV-1 17termAR, expressing plasmid, a kind gift from Dr. R. Thompson (32, 33) (University of Cincinnati College of Medicine, University of Cincinnati), was used to rescue the virus following reported procedures. HSV-1 was grown in rabbit Sirc cells, and virus titers were determined in Vero cells by regular crystal violet plaque assay, adding Beriglobin (CSL Behring) to the media. HSV-1 infections were performed at a multiplicity of infection (MOI) of 1–10.

Vesicular stomatitis virus expressing GFP (VSV-GFP) was a kind gift from Dr. A. García-Sastre (Icahn School of Medicine at Mount Sinai, New York). VSV-GFP virus stock was grown for 2 d in Vero cells and titrated according to the method of Reed–Muënch in Vero cells. Virus growth and infection was done following reported procedures (34).

For transient knockdown of E7-less STING isoforms, cells were transfected with 80 nM E7-less STING-specific small interfering RNA (target sequence: 5′-GCGGAACCUGCAGAUGACA-3′) and a recommended control (Thermo Fisher Scientific), using Lipofectamine 2000 (Invitrogen), according to the manufacturer’s instructions.

PMA-derived THP1 macrophages were activated by poly dA:dT (Invivogen) transfection using Lipofectamine 2000.

To determine IFN-β, IRF-3, ISG54, and NF-κB induction, HEK293T cells plated on 12-well plates (2–5 × 10⁵ cells per well) were transfected with the corresponding pathway-specific reporter plasmid (driving the expression of the firefly luciferase) and the control plasmid pRL-TK (driving the expression of Renilla luciferase) using Lipofectamine 2000. Cells were lysed 24 h posttransfection, and firefly luciferase and Renilla luminescence were determined using the dual-luciferase assay kit (Promega) according to the manufacturer’s instructions. The relative potency of pathway activation was calculated as firefly luminescence relative to Renilla luminescence. The IRF-3-5D plasmid was a kind gift from Dr. A. García-Sastre (Icahn School of Medicine at Mount Sinai, New York).

Total RNA was isolated with TRIzol reagent (Invitrogen) and reverse transcribed using random primers and Moloney murine leukemia virus reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. To amplify the human coding DNA sequence of hSTING, PCR was performed with a GenAmp PCR system 2400 thermocycler using Expand High Fidelity PCR System (Roche) and the following primers: forward 5′-TGCGTAGCACCATGCCCCACTCCAGCCTGCATC-3′ and reverse 5′-TGATGACTCGAGTCAAGAGAAATCCGTGCGG-3′. For quantitative PCR, cDNA was amplified using iQ SYBR Green Supermix (Bio-Rad) according to the manufacturer’s instructions in a C1000 Thermal Cycler (Bio-Rad). The following primers were used: wt hSTING: forward, 5′-GCTCCAGGCCCGGATTCGAAC-3′, reverse, 5′-CCTATCCTCCCGGCTAAAGCC-3′ and E7-less hSTING: reverse, 5′-CTGCTGTCATCTGCAGGTTCCGC-3′; IFNB1: forward, 5′-GTCAGAGTGGAAATCCTAAG-3′ and reverse, 5′-ACAGCATCTGCTGGTTGAAG-3′; and B-actin: forward, 5′-AGCCTCGCCTTTGCCGA-3′ and reverse, 5′-CTGGTGCCTGGGGCG-3′.

E7-less isoforms 1, 2, and 3 and wt hSTING cDNA were amplified by RT-PCR using RNA purified from THP1 or HeLa cells as a template. PCR products were cloned into pcDNA3.1 V5 His (Invitrogen) using the TOPO-TA cloning system and sequenced. The DNA fragment containing the entire open reading frame of hSTING or isoforms 1, 2, or 3 was excised from pcDNA3.1 V5 His and subcloned into the pCAGGS vector to generate the recombinant proteins fused to a C-terminal HA tag.

To generate the wt hSTING fused to the fluorescent protein mCherry (wt hSTING-Cherry), Cherry fluorescent protein was amplified by PCR and inserted into position after the transmembrane domain and before the cytoplasmic hSTING domain. The generation of MAVS-expressing plasmids were described before (35).

For flow cytometry analysis, an eight-color BD FACS Canto II equipped with a 488-, 633-, and 405-nm laser or a four-color FACSCalibur with two lasers (argon-488 and 635-nm diode) were used. FACS data were analyzed with FlowJo and DIVA software.

Statistical analysis was performed using PRISM version 5.0 (GraphPad). Data are presented as mean ± SD. All experiments were performed at least in triplicate replicas and verified two or more times. Comparisons between two groups were made using a two-tailed unpaired t test. Multiple groups were compared using ANOVA followed by Bonferroni posttest. Statistical significance was assigned to p values <0.01 (***), <0.1 (**), or <0.5 (*).

Transcript sequences related to TMEM173 in human and other species were searched in National Center for Biotechnology Information databases using “tBLASTx” and “BLASTn” (36). A total of 48 sequences belonging to 35 species were selected to create a sequence database subsequently used to create a multiple alignment based on all known exons of TMEM173 using “Muscle” (37). Note that gaps were manually added to sequences where appropriate, to combine all of them in a single multiple alignment, which was subsequently refined using GenDoc (38). A phylogenetic tree analysis was inferred based on this alignment using the maximum likelihood method of phylogenetic reconstruction implemented in PhyML (39).

A sequence logo representation was constructed using Checkalign (40) and an alignment input of 20 nt for each splice site of TMEM173 in eight mammalian species (Cercocebus atys, Equus asinus, Gorilla gorilla, Homo sapiens, Microcebus murinus, Pan troglodytes, Physeter catodon, and Saimiri boliviensis boliviensis).

The full-length human TMEM173 mRNA gene (corresponding to wt hSTING protein) comprises eight exons. The wt hSTING protein coding sequence starts at TMEM173 mRNA exon 3 (mRNA nt 303) and finishes inside exon 8 (mRNA nt 1338). To amplify the wt hSTING coding DNA sequence, we designed specific primers that annealed at the 5′ and 3′ end of the described cDNA sequence (Fig. 1A). Using total RNA obtained from THP1 cells as a template, we obtained the expected full-length wt hSTING amplification product (1140 nt) and three additional transcripts. All products were cloned in pcDNA3.1 TOPO-TA plasmid and were sequenced, revealing that the additional transcripts were the result of hSTING mRNA alternative splicing (Fig. 1B, 1C). Alternative splicing-derived isoforms and wt hSTING share a common amino terminal region (aa 1–76) (Fig. 1D) that contains the first two transmembrane domains. All three spliced transcripts were E7-less (E7-less isoforms 1, 2, and 3). In addition, E7-less isoform 3 mRNA contains an unspliced intron after exon 3, whose first triplet is a stop codon. E7-less isoform 2 lacks exon 4, and the resulting open reading frame contains a C-terminal peptide unique to this isoform. E7-less isoform 1 corresponds to the alternatively spliced isoform described previously as MITA-related protein (30), which lacks only exon7. A graphic representation of the theoretically translated proteins is shown in Supplemental Fig 1A. They share the first two transmembrane domains with full-length wt hSTING. Truncated isoforms have either a sudden interruption by a stop codon or unique peptides at the C-terminal side. Sequencing of splicing sites revealed the nature of the alternative splicing isoforms (Supplemental Fig. 1B).

To compare the relative expression levels of wt STING and E7-less isoforms mRNA in human material, we designed specific quantitative PCR assays with similar coefficient of efficiency, and RNA was isolated from total PBLs and from liver biopsy specimens obtained from healthy volunteers. Our analysis detected a large variability in the expression of full-length compared with E7-less mRNA transcripts among individuals, in both PBLs (Fig. 1E) and liver samples (Fig. 1G). However, both transcripts could be detected in all cases. The same analysis was performed in human cell lines of different origin. Expression of wt hSTING and E7-less hSTING transcripts is variable and depends on the origin of the cell line (Fig. 1F). In general, a higher expression of E7-less hSTING was observed in most of the cell lines of hematopoietic origin such as THP1, HEL, K562, Jurkat, and Karpa299, which also correlated in general with a higher expression of wt TMEM173. Wt and truncated isoforms were also detected in pancreatic adenocarcinoma BxPC3, liver carcinoma Hep3B, and HeLa-derived HEp-2 carcinoma cell lines. We noticed a very low expression of all TMEM mRNA forms in many cell lines, including cells used to characterize hSTING isoforms like HEK293 or derivatives such as HEK293T (Fig. 1F).

Finally, to detect the expression of E7-less hSTING isoforms, we tried to generate Abs in rabbits using peptides covering the conserved region for all STING isoforms. However, despite several attempts, we could not generate any reliable material to detect the different isoforms. Only one of the commercial Abs tested was able to recognize wt STING and isoform 1 by Western blot after immunoprecipitation (Fig. 1H). Unfortunately, isoforms 2 and 3 could not be detected.

To further characterize, wt hSTING and E7-less isoforms were subcloned into a mammalian expression plasmid, and their ability to induce the IFN-β pathway was analyzed using an IFN-β reporter system after cotransfection in HEK293T cells. Only plasmid expressing wt STING, but none of the E7-less isoforms expressing plasmids, was able to induce IFN-β (Fig. 2A). Next, the ability of the different isoforms to induce the transcriptional activity of IRF-3- and ISG54-reporter plasmids was also tested. None of the E7-less isoforms were able to induce IRF-3 (Fig. 2B) or ISG54 (Fig. 1C) as compared with wt hSTING.

cGAS is a key player in the activation of the innate immune system against many pathogens, including RNA and DNA viruses. Activation of cGAS requires STING activity to induce IFN-β (11). cGAS overexpression alone was unable to induce IFN-β-, IRF-3- (data not shown), ISG54-, or NF-κB reporter plasmids (Fig. 2D, 2E) in HEK293T cells. Coexpression of cGAS and wt hSTING in HEK293T cells induced ISG54 reporter activity at higher levels than the one induced by wt hSTING alone (Fig. 2D). In contrast, none of the E7-less isoforms (1, 2, or 3) showed any significant induction when cotransfected with cGAS. In addition, coexpression of cGAS and wt hSTING, but not E7-less isoforms, induced a consistent NF-κB–dependent reporter activity (Fig. 2E).

Although E7-less hSTING isoforms 1, 2, or 3 could not induce IFN-β, we tested whether they were able to block IFN-β production induced by wt hSTING. For this purpose, E7-less hSTING 1, 2, and 3 were cotransfected with wt hSTING and the IFN-β reporter plasmid. Coexpression of the E7-less isoforms blocked hSTING-mediated IFN-β induction (Fig. 3A). This inhibition was specific for hSTING because they were not able to inhibit the ISG54 activation induced by murine STING (Supplemental Fig. 1C) although human and mouse STING have 69% identity and 81% residue similarity. In addition, this inhibitory effect occurs upstream of IRF-3 activation because none of the E7-less STING isoforms were able to block the ability of constitutively active IRF-3-5D to induce IFN-β expression (Supplemental Fig. 1D). However, E7-less hSTING isoforms inhibit the activation of IRF-3 signaling induced by wt hSTING (Fig. 3B). Similar results were obtained when using the ISG54 reporter assay (Fig. 3C).

To determine the mechanism by which E7-less isoforms mediate inhibition of wt hSTING, we transfected HEK293T cells with plasmids expressing a tagged wt hSTING and the E7-less isoforms. Western blot analysis of wt hSTING protein expression revealed a significant reduction in its amount when coexpressed with E7-less isoforms 1 and 2; E7-less isoform 3 has a weaker effect (Fig. 3E, 3F). These results correlate with a dose-response effect of E7-less isoforms on wt STING ability to trigger IFN-β (Supplemental Fig. 1E). The reduction in wt STING protein levels cannot be explained by an effect on TMEM173 transcription because wt STING mRNA expression levels remained unchanged (Fig. 3G) when coexpressed with the E7-less isoforms in 293T cells (endogenous hSTING cannot be detected because these cells express insignificant levels of full-length TMEM173 transcript; Fig. 1F).

To study the reduction of wt STING upon coexpression of the isoforms in more detail, we generated a plasmid expressing a functional full-length wt hSTING-Cherry protein and developed an assay to quantify the relative fluorescence in vivo over time. Wt hSTING-Cherry shows similar functionality to wt hSTING (Supplemental Fig. 1F). HEK293T cells were cotransfected with wt hSTING-Cherry plasmid and a plasmid expressing each particular E7-less isoform, and in addition, all samples were cotransfected with a plasmid expressing GFP under a β-actin promoter to normalize transfection efficiency. We could observe that in the presence of E7-less hSTING isoforms, the red fluorescence associated to wt hSTING-Cherry decreased slowly over time and almost disappeared at 48 h posttransfection in contrast to cells transfected with a control plasmid (Fig. 3H, Supplemental Fig. 1G, 1H). Similar inhibition and degradation of wt hSTING was observed in the presence of cGAS (Supplemental Fig. 2A–F).

To determine if the mechanism by which E7-less isoforms reduced the amount of wt hSTING was proteasome mediated, HEK293T cells were cotransfected with plasmids expressing limited amounts of HA-tagged full-length wt hSTING together with the three isoforms. Twenty-four hours later, cells were incubated with the proteasome inhibitor MG132 for 9h, and hSTING expression was analyzed by Western blot. MG132 treatment partially restored wt hSTING protein levels in cotransfected cells (Fig. 3I, Supplemental Fig. 1J), indicating that proteasome degradation was partially involved. Next, we tested the potential role of autophagy in the reduction of the protein by treating the cells with the autophagy inhibitor Bafilomycin A. No differences in the reduction of wt STING were observed between cotransfected treated and untreated cells, indicating autophagy is not involved in this process (Supplemental Fig 1I).

With the purpose of clarifying the mechanism of such effect, we analyzed the interaction between wt hSTING and E7-less isoforms 1, 2, and 3 by an immunoprecipitation assay. We found that E7-less hSTING isoforms 1 and 2, but not isoform 3, indirectly interact with wt hSTING (Fig. 3J). Because E7-less hSTING isoform 3 was also able to induce degradation of wt hSTING (Fig. 3F), we cannot rule out a weaker interaction that could not be detected under our experimental conditions. Additionally, we observed that E7-less isoform 1 inhibited dimerization of wt hSTING (marked with an asterisk) when both are coexpressed (Fig. 3I). A weaker interaction of isoforms 2 and 3 with wt STING may explain a weaker dimer signal.

hSTING also can interact and cooperate with MAVS to induce IFN-β (41). We found that E7-less hSTING isoforms 1, 2, and 3 also have a negative effect on MAVS-mediated induction of IFN-β, IRF-3, and NF-κB as shown in Supplemental Fig. 3A–C, respectively. This inhibition is not mediated by MAVS degradation as demonstrated in Supplemental Fig. 3D, although, all isoforms of hSTING, except isoform 3, have a weak interaction with MAVS under our experimental conditions (Supplemental Fig. 3E). Once again, we cannot exclude a weak interaction of isoform 3 not detectable with our technique, correlating with a weaker inhibitory effect on MAVS activation. We have previously demonstrated that the inhibitory effect was not due to the inhibition of downstream signaling because no inhibition of constitutively active IRF-3 5D was observed (Supplemental Fig. 1D).

The biological relevance of endogenous E7-less hSTING isoforms was examined in THP1 cells (because they have high levels of both full-length and truncated TMEM173 mRNA isoforms as shown in Fig. 1F) by knocking down their expression using a specific siRNA targeting a RNA region containing exon 6–exon 8 direct-splicing rearrangements. Several siRNA molecules were tested (data not shown), and siRNA236 was selected based on its specific inhibition of E7-less isoforms but not full-length wt hSTING mRNA (Fig. 4A). To test whether this specific inhibition resulted in differences in hSTING activity, we transfected THP1 cells with an irrelevant scramble siRNA or with siRNA236, and 2 d later, cells were transfected again with poly dA:dT DNA, which is a STING activator. Subsequently, IFN-β mRNA expression was measured. The inhibition of the expression of E7-less hSTING isoforms resulted in higher levels of IFN-β expression after poly dA:dT stimulation (Fig. 4B).

A previous report showed the relevance of THP1 as a model cell line to study IFN-β induction upon HSV-1 infection (42), and wt hSTING has been described as a relevant factor in the detection of HSV-1 (14, 43). To determine whether E7-less hSTING isoforms may influence HSV-1 detection and viral replication, THP1 cells were transfected with siRNA236 or control siRNA and then infected with HSV-1. Significantly higher levels of IFN-β mRNA expression were detected in cells transfected with siRNA236 in comparison with scramble siRNA 9 h after HSV-1 infection (Fig. 4C). In addition, knocking down truncated E7-less STING mRNAs also led to a moderate but consistent reduction in HSV-1 replication (Fig. 4D).

Furthermore, the cGAS/STING-dependent IFN-β induction pathway can become activated as a response against RNA virus infection. Knocking down E7-less isoforms in THP1 cells before infecting them with VSV-GFP at a MOI of 1 for 8 h resulted in replication titers of VSV-GFP that were significantly lower in siRNA236-transfected THP1 cells than in control cells (Fig. 4E). Together these data suggest that a reduction in the expression of the E7-less isoforms is associated with stronger wt STING activation upon viral infection.

We then analyzed whether infection of THP1 cells by different viruses could alter the expression of wt STING and the E7-less hSTING isoforms. Because some viruses inhibit the activity of wt hSTING, we compared whether virus infection could also alter the relative amount of wt versus E7-less hSTING at different times postinfection as a strategy to regulate STING activity. As shown in Fig. 4F, HSV-1 infection of THP1 cells resulted in downregulation of the expression of wt and E7-less hSTING expression between 3 and 6 h postinfection. E7-less isoform expression recovers afterward, whereas wt hSTING expression recovery is delayed. As result, there is a downregulation in the ratio of wt/E7-less mRNA that peaked around 9 h postinfection (Fig. 4G). A fast infection-replicating RNA virus like VSV, however, resulted in a significant downregulation of wt hSTING, whereas no effect could be seen on E7-less isoforms (Fig. 4H). As a result, the ratio of wt/E7-less STING mRNA decreased 3 h postinfection and returned to initial levels by 9 h (Fig. 4I). The effect of virus infection on the expression of E7-less hSTING isoforms was independent of type I or type II IFN signaling because neither IFN-α nor IFN-γ treatments of THP1 cells had a significant effect on altering STING mRNA alternative splicing (Supplemental Figs. 2G, 3H).

In an attempt to find evidence of reported sequences of hSTING splicing and trace the evolutionary history of hSTING mRNA isoforms, we screened National Center for Biotechnology Information databases for hSTING transcripts in humans and other vertebrates. Our searches suggest that the wt STING occurs in almost all vertebrates, from fishes to primates (Fig. 5A). A recent publication expands homologous sequences even more (41), whereas all isoform 1 homologs were mainly found in annotations from human cells as well in various other primates. It is worth highlighting that we found isoform 1 to be present in Ochotona princeps, a species belonging to the lagomorph order, which also includes the Leporidae (rabbits and hares). This piece of evidence suggests that E7-less isoform 1 may probably be more widely distributed in mammals.

In Fig. 5B, we provide graphical representation of the exon–intron architecture of several of the isoform transcripts detected in humans, using experimental truncated isoforms 1, 2, and 3 and the wt hSTING as references (reference: 198282.3). We found several references of annotated transcripts for isoform 1. As for truncated isoform 2, we only found two human transcripts supporting the expression of this isoform. Both are consistent with isoform 2 in the absence of exon E4, which results in a premature stop codon that will lead to the isoform 2 protein. However, although the one found in THP1 lacks exon E7, the two candidate transcripts include it (one of them is truncated at 3′, showing only partly in this exon). As for truncated isoform 3, we detected a human transcript that, albeit, is also truncated at 3′ and is therefore incomplete, it provides convincing support for isoform 3 as it carries the two exons (E3′ and E5′) that characterize this isoform. This transcript contains the early stop codon described in truncated isoform 3, which will lead to the isoform 3 protein.

It is worth mentioning that we also detected a transcript from Equus africanus asinus that presents exon E3′, although it lacks exon E5′. In addition, this transcript also presents exon E7. Hence, despite that we cannot consider or classify this last transcript of E. asinus as E7-less isoform 3, this transcript and those found related to isoform 2 in humans and isoform 1 in other species, suggest that not only humans but also other mammals probably transcribe different hSTING alternative-spliced isoforms and that the number of viable isoforms may be greater than the three described in this study. In this regard, Supplemental Fig. 4 presents a sequence logos representation created based on the alignment of the splicing sites of hSTING exons (from E3 to E8) in several species (C. atys, E. asinus, G. gorilla, H. sapiens, M. murinus, P. troglodytes, P. catodon, and S. boliviensis boliviensis). On this representation we can observe that the canonical GT/AG motifs sites are highly preserved in almost all exons, thus suggesting that both canonical and alternative-splice events are viable in the aligned species. The less-preserved exon is E5′, which instead of the canonical splice GT/AG motifs, shows predominance (but not well preserved) of a noncanonical GC/AG motif, indicating that exon, albeit viable isoforms, carrying exon E5′ might, probably, be more rare than other isoforms.

Two reports have described the expression of what we named “E7-less” hSTING isoform 1 in HEK293T and HEK293 cells (30, 31). In those reports, the inhibitory activity of the truncated E7-less hSTING protein over the wt protein was described. In our hands, endogenous expression of both wt hSTING and E7-less hSTING isoforms in HEK293T cells is undetectable. Functional assays (Figs. 2D, 3E) in 293T cells reinforce the absence of a function of endogenous Supplemental Fig. 1D STING in this cell line. Our data comparing the expression of wt hSTING and E7-less hSTING isoforms in cell lines from different origins, together with the data of relative expression levels in hepatic samples and in peripheral blood cells obtained from different donors, indicate that expression of wt hSTING as well as the alternative-splicing forms has a large variability that depends on the origin of the cell, suggesting a cell type–dependent function for this protein. After our initial analysis, a more complete analysis of STING protein and gene expression has been reported (44) in the Human Protein Atlas (www.proteinatlas.org). Hematopoietic origin cells as well as respiratory organs and female reproductive organs seem to have a higher wt hSTING expression compared with other organs. Taking into account the differences observed in hSTING expression, it is very likely that the activity of the E7-less isoforms described in the current study will depend on the origin of the experimental cell. In this sense, our data demonstrate differences in the activity of E7-less hSTING alternative-splicing variants in THP1 in which STING expression levels is well documented, as compared with HEK293 cells that poorly express STING (12, 45).

Functional analysis revealed that, in contrast to wt hSTING, the E7-less isoforms failed to induce activation of the IFN-β pathway and inhibited the ability of wt hSTING to induce the IFN type I pathway when coexpressed. Furthermore, the inhibitory effect was maintained when wt hSTING was activated in the presence of its natural upstream inducer cGAS (Supplemental Fig. 2). In a previous study, it was proposed that this inhibitory effect was due to the ability of truncated isoform 1 to prevent wt STING interaction with TBK1 (30). In this article, we propose an additional possibility. We found that the E7-less hSTING isoforms directly interact with wt hSTING and that the overexpression of E7-less isoforms altered the stability of wt hSTING in a dose-dependent manner. The analysis of wt hSTING protein expression by Western blot or FACS analysis demonstrated that the coexpression of the E7-less isoforms affects the stability of wt hSTING protein, resulting in a rapid reduction of the protein level. MG132 used as proteasome inhibitor revealed an additional mechanism involved in the isoform-mediated degradation of wt hSTING through the proteasome. Thus, it is most likely that the direct interaction of E7-less hSTING isoforms with wt STING facilitates such degradation. And therefore, expression of E7-less hSTING isoforms could represent a cell-inherent mechanism to avoid overactivation after hSTING stimulation.

Viruses employ a wide variety of mechanisms to counteract the host machinery involved in the antiviral response (46). Among them, activation of virus-induced alternative splicing is a mechanism that is used to maximize the coding potential of their genomes (47–49). Viruses can also regulate host splicing machinery, like adenovirus L4-33K protein, which modulates alternative splicing in the host cell (50). Another strategy used by viruses is to specifically manipulate the splicing of mRNAs involved in the response to the virus to subvert host defenses. EBV, for example, can specifically modulate alternative splicing of STAT-1 mRNA (51), HSV-2 can evade the antiviral effect of promyelocytic leukemia protein by specifically inducing alternative splicing (52), and HSV-1 enhances the expression of an alternative-splice variant of the MxA that supports viral replication (53).

In addition to the alternative splicing induced by viruses, host cells can also use alternative splicing to increase their antiviral defense fitness. Specific knocking down of truncated isoforms in THP1 cells improved the ability of these cells to respond to transfected DNA or virus infection. Moreover, we observed that infection by two different viruses, such as HSV-1 and VSV, induced an imbalance and a reduction of the relative amounts of wt hSTING as compared with the expression of E7-less isoforms. We speculate that the downregulation of the ratio of wt STING/E7-less isoforms could be exploited by certain viruses to create a cellular state that transiently favors viral replication to allow its spreading to other cells. Another possibility could be that by initially downregulating STING-dependent innate immune activation, macrophages can obtain a higher viral load that enhances their role as APCs.

Despite the divergence of the TMEM173 gene among different species, our screenings (Fig. 5) revealed the presence of identical, or very similar to the ones described in this study, splicing isoforms in different human cells. Some of these isoforms are also present in other mammals, thus supporting the idea that several hSTING isoforms may have a redundant way to control STING activity in humans and other animals. This idea is also supported by the similar intron–exon structure of the gene in mammals and the good degree of preservation of their splicing sites. The functional context favoring the assembly of one isoform or another is yet unclear for us and merits further attention. We think that changes in the splicing machinery, which may result as a consequence of different insults received by cells (such as viral infections), may increase the probability of alternative-splicing variant synthesis and result in effects such as the ones observed in our experimental results.

Overall, E7-less hSTING isoforms seem to play a role in controlling hSTING activity in response to different stimuli by downregulating wt hSTING signal transduction, IFN-β production, and its antiviral activity. This mechanism adds an additional level of complexity to understanding hSTING biology, which may have special relevance during macrophage infection, activation, and Ag presentation, determining the level and kinetics of the antiviral response.

The authors have no financial conflicts of interest.