LSm14A Plays a Critical Role in Antiviral Immune Responses by Regulating MITA Level in a Cell-Specific Manner

The innate antiviral immune response represents the first line of host defense against viral infection. The host antiviral responses are initiated by the recognition of structurally conserved viral components called pathogen-associated molecular patterns by host pathogen recognition receptors, which triggers a series of signaling cascades that lead to the production of type I IFNs, proinflammatory cytokines, and other downstream antiviral effectors. These cytokines and downstream effectors mediate inhibition of viral replication, clearance of virus-infected cells, and facilitation of adaptive immune response (1–4).

The host cells have developed various mechanisms for sensing viral infection. Among pathogen recognition receptors, endosomal TLR3 and cytosolic RIG-I–like receptors play important roles in sensing viral RNA (1). Several other proteins have been reported to recognize viral DNA, including cGAS, LSm14A, DDX41, IFI16, RNA polymerase III, DAI, and TLR9 (5–10). Among them, cGAS is believed to be the most widely used viral DNA sensor. Genetic studies demonstrate that cGAS is essential for innate immune responses to all the DNA or DNA viruses examined in cells and mice (11). Recognition of cytosolic DNA by cGAS leads to the synthesis of the second messenger cGAMP from ATP and GTP (5). Synthesized cGAMP binds to and activates MITA (also known as STING, MPYS, and ERIS), which acts as a critical adaptor in the antiviral signaling pathways (12, 13).

MITA was independently identified by us and others as a critical mediator of innate antiviral responses (14–17). Various studies have established the essential roles of MITA in innate immune responses to DNA viruses and certain RNA viruses (18). Gene knockout studies demonstrate that MITA is indispensable for efficient production of type I IFNs and innate antiviral responses following recognition of intracellular viral DNA in all cell types examined (19). However, MITA may play distinct roles in innate immune responses against RNA viruses in different cell types. It has been demonstrated that knockdown of MITA impairs Sendai virus (SeV)– and vesicular stomatitis virus (VSV)–induced type I IFNs production in human transformed cell lines, such as HEK293, HeLa, and Huh7 cells, as well as in human primary macrophage and dendritic cells (DCs) (14, 16, 20). However, genetic studies suggested that MITA deficiency rendered mouse embryonic fibroblasts (MEFs) highly susceptible to VSV, but less susceptibility was observed in bone marrow–derived DCs or bone marrow–derived macrophages (15). In vivo, MITA-deficient mice were defective in type I IFNs production and highly susceptible to lethal infection with VSV but not another RNA virus encephalomyocarditis virus (19). Compared with the universal requirement for MITA in viral DNA-triggered signaling, MITA seems to be involved in innate immune responses against RNA viruses in a virus- and cell type–specific manner.

Because of its central roles in innate antiviral responses, the mechanisms and regulations of MITA-mediated signaling have been extensively investigated. It has been demonstrated that DNA virus infection induces binding of cGAMP to MITA, whereas RNA virus infection triggers recruitment of MITA to the adaptor protein VISA (14, 21). MITA, in turn, recruits downstream kinase TBK1 and the transcription factor IFN regulatory factor 3 (IRF3), leading to activation of IRF3 and induction of type I IFNs. Various studies have suggested that the activity of MITA is modulated by distinct posttranslational modifications, including ubiquitination and phosphorylation (22–26). Whether MITA is regulated by other mechanisms is unclear.

We previously identified LSm14A as a processing body (P-body)–associated sensor of viral nucleic acids that initiates cellular antiviral responses in the early phase of viral infection in human cell lines (6). In this study, we investigated the functions of LSm14A by gene knockout studies in mice. Interestingly, we found that LSm14A deficiency impaired induction of antiviral cytokines and effectors in response to DNA viruses in mouse DCs but not macrophages and MEFs. Biochemically, we demonstrated that LSm14A was important for maintaining the mature mRNA level of Mita in DCs but not in other examined cells by regulating its nuclear mRNA precursor (pre-mRNA) progressing. Our results suggest that LSm14A plays an important role in antiviral innate and adaptive immune responses by modulating MITA level in a cell type–specific manner.

GM-CSF (PeproTech); polyinosinic-polycytidylic acid [poly(I:C)] (high m.w.) and 2′3′-cGAMP (InvivoGen); LPS, digitonin, Br-UTP, sarkosyl, DNase I, polyvinylpyrrolidone, actinomycin D (Act D), and OVA (Sigma-Aldrich); Lipofectamine 2000 (Invitrogen); polybrene (Millipore); SYBR (Bio-Rad); RNase inhibitor, glycogen, and Freund’s incomplete adjuvant (Thermo); ATP, GTP, CTP, UTP, and random primer (Takara); anti-BrdU Ab agarose conjugate (Santa Cruz Biotechnology); rat anti-mouse IgG1 conjugated to HRP (Southern Biotechnology Associates); ELISA kit for murine IFN-β (PBL); ELISA kits for murine IFN-α and IFN-γ (eBioscience); ELISA kit for murine IL-6 (BioLegend); mouse Abs against β-actin (Sigma-Aldrich), phospho-IκBα (S32/36), and IκBα (Cell Signaling Technology); rabbit Abs against phospho-TBK1 (S172) and TBK1 (Epitomics), phospho-IRF3 (S396) and MITA (Cell Signaling Technology), and human IRF3 (Santa Cruz Biotechnology) were purchased from the indicated manufacturers. Mouse antisera against murine IRF3 and cGAS and rabbit antisera against human LSm14A were raised using the respective full-length recombinant proteins as Ags. HEK293T cells were originally provided by Dr. G. Johnson (National Jewish Health, Denver, CO). Huh7 cells were provided by Dr. D. Guo (Wuhan University, Wuhan, China). THP-1 cells were obtained from the American Type Culture Collection. SeV (Cantell strain) (Charles River Laboratories), HSV-1 (KOS strain) (China Center for Type Culture Collection, Wuhan, China), murid herpesvirus 68 (MHV-68) (Dr. H. Deng, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China), and VSV (China Center for Type Culture Collection) viruses have been obtained from the indicated resources.

pLOV-LSm14A (murine) was constructed by standard molecular biology technique. The 5′-untranslated region (UTR), coding sequence (CDS), and 3′-UTR of Mita cDNA were amplified from templates derived from DCs and cloned upstream or downstream of the Firefly luciferase gene. They were then constructed into pLOV-CMV-eGFP-2A-EF1a-PuroR plasmid by standard molecular biology technique. pLOV-Renilla luciferase reporter plasmid was similarly constructed for normalization of infection efficiency.

Lsm14a knockout mice were generated by knockout-first strategy as illustrated in Supplemental Fig. 1A. The ES cells harboring Lsm14a^{tm1a(KOMP)Wtsi} allele were purchased from the Knockout Mouse Project Repository. The ES cells were then injected into the blastocysts of C57BL/6. Genotyping by PCR was performed using the following primers: WT (551 bp), 5′-CCAAGGTGACCCCAAACTTA-3′ (forward) and 5′-TCCTGCCAAAAGGTCCATAG-3′ (reverse); and knockout (532 bp), 5′-CTGAATGGTTTCTGCTTGCTT-3′ (forward) and 5′-CACAACGGGTTCTTCTGTTAGTCC-3′ (reverse).

To generate Lsm14a hematopoietic–specific knockout mice, Lsm14a^+/− mice were first bred to Flp mice to remove the FRT site–flanked gene-trap cassette, which reverted the mutation to wild-type, leaving loxP sites on either side of exon 3. The mice were then bred to Vav-Cre mice to generate Lsm14a^{f/f: Vav-Cre} mice. Genotyping by PCR was performed using the following primers: Flp (725 bp), 5′-CACTGATATTGTAAGTAGTTTGC-3′ (forward) and 5′-CTAGTGCGAAGTAGTGATCAGG-3′ (reverse); WT (551 bp)/Floxed (689 bp), 5′-CCAAGGTGACCCCAAACTTA-3′ (forward) and 5′-TCCTGCCAAAAGGTCCATAG-3′ (reverse); and Cre (205 bp), 5′-CGTATAGCCGAAATTGCCAG-3′ (forward) and 5′-CAAAACAGGTAGTTATTCGG-3′ (reverse).

All animal experiments were performed in accordance with the Wuhan University Animal Care and Use Committee guidelines.

Monocytes were isolated from tibia and femur. For preparation of macrophages, monocytes (5 × 10⁶) were cultured in 100-mm dishes in 5 ml 10% M-CSF–containing conditional medium from L929 cells. Three days later, adherent fraction was harvested. For preparation of DCs, monocytes (5 × 10⁶) were cultured in 100-mm dishes in 5 ml complete medium (RPMI 1640 medium supplemented with 10% heat-inactivated FBS, penicillin, and streptomycin) containing murine GM-CSF (50 ng/ml). On days 3 and 5, fresh medium supplemented with murine GM-CSF and warmed at 37°C was added. After a 7-d differentiation, all cells were harvested. EasySep Mouse CD11c Positive Selection Kit was used to magnetically enrich CD11c⁺ fraction from GM-CSF culture. MEFs were prepared from day 12.5 embryos and cultured in DMEM supplemented with 10% FBS.

The following oligonucleotides were used to stimulate cells: HSV60, 5′-TAAGACACGATGCGATAAAATCTGTTTGTAAAATTTATTAAGGGTACAAATTGCCCTAGC-3′; HSV120, 5′-AGACGGTATATTTTTGCGTTATCACTGTCCCGGATTGGACACGGTCTTGTGGGATAGGCATGCCCAGAAGGCATATTGGGTTAACCCCTTTTTATTTGTGGCGGGTTTTTTGGAGGACTT-3′; VACV70, 5′-CCATCAGAAAGAGGTTTAATATTTTTGTGAGACCATCGAAGAGAGAAAGAGATAAAACTTTTTTACGACT-3′; DNA90, 5′-TACAGATCTACTAGTGATCTATGACTGATCTGTACATGATCTACATACAGATCTACTAGTGATCTATGACTGATCTGTACATGATCTACA-3′; and ISD45, 5′-TACAGATCTACTAGTGATCTATGACTGATCTGTACATGATCTACA-3′.

Total RNA was isolated for quantitative real-time PCR (qPCR) analysis to measure mRNA levels of the indicated genes. Data shown are the relative abundance of the indicated mRNA normalized to that of Gapdh. qPCR was performed using the following primers: murine Ifnb1, 5′-TCCTGCTGTGCTTCTCCACCACA-3′ (forward) and 5′-AAGTCCGCCCTGTAGGTGAGGTT-3′ (reverse); murine Ifna4, 5′-CCTGTGTGATGCAGGAACC-3′ (forward) and 5′-TCACCTCCCAGGCACAGA-3′ (reverse); murine Il6, 5′-TCTGCAAGAGACTTCCATCCAGTTGC-3′ (forward) and 5′-AGCCTCCGACTTGTGAAGTGGT-3′ (reverse); murine Lsm14a, 5′-CAGTGAAGGGAATGCTGATGAGG-3′ (forward) and 5′-TCTTCAGCCCAAGTTGGTCTCC-3′ (reverse); murine Isg56, 5′-ACAGCAACCATGGGAGAGAATGCTG-3′ (forward) and 5′-ACGTAGGCCAGGAGGTTGTGCAT-3′ (reverse); murine Ccl5, 5′-TCACCATATGGCTCGGACACCAC-3′ (forward) and 5′-TTGGCACACACTTGGCGGTTC-3′ (reverse); murine Tnfa, 5′-GGTGATCGGTCCCCAAAGGGATGA-3′ (forward) and 5′-TGGTTTGCTACGACGTGGGCT-3′ (reverse); murine cGAS, 5′-ACCGGACAAGCTAAAGAAGGTGCT-3′ (forward) and 5′-CAGCAGGCGTTCCACAACTTTAT-3′ (reverse); murine Mita, 5′-AAATAACTGCCGCCTCATTG-3′ (forward) and 5′-TGGGAGAGGCTGATCCATAC-3′ (reverse); murine Tbk1, 5′-GACATGCCTCTCTCCTGTAGTC-3′ (forward) and 5′-GGTGAAGCACATCACTGGTCTC-3′ (reverse); Murine Irf3, 5′-CGGAAAGAAGTGTTGCGGTTAGC-3′ (forward) and 5′-CAGGCTGCTTTTGCCATTGGTG-3′ (reverse); murine Gapdh, 5′-ACGGCCGCATCTTCTTGTGCA-3′ (forward) and 5′-ACGGCCAAATCCGTTCACACC-3′ (reverse); human LSm14A, 5′-CAGTCCGAGTTCCTTAGTTGGG-3′ (forward) and 5′-AAGGCAGAACCAACCGCACTAC-3′ (reverse); human MITA, 5′-CCTGAGTCTCAGAACAACTGCC-3′ (forward) and 5′-GGTCTTCAAGCTGCCCACAGTA-3′ (reverse); human GAPDH, 5′-GAGTCAACGGATTTGGTCGT-3′ (forward) and 5′-GACAAGCTTCCCGTTCTCAG-3′ (reverse); HSV-1, 5′-GCATCCTTCGTGTTTGTCATT-3′ (forward) and 5′-GCATCTTCTCTCCGACCCCG-3′ (reverse); and SeV, 5′-AAACGCATCACGTCTCTTCC-3′ (forward) and 5′-TTCTCAGCTCTGCTTAGGGG-3′ (reverse).

HEK293T cells were transfected by standard calcium phosphate precipitation method. DCs, macrophages, and MEFs were transfected by Lipofectamine 2000. Briefly, DCs, macrophages, and MEFs (1 × 10⁶) were seeded on 6-well plates and transfected the following day with the indicated nucleic acids (4 μg/ml) by Lipofectamine 2000, according to the manufacturer’s instructions.

Snap-frozen brains were weighed and homogenized three times for 5 s in MEM. After homogenization, the brain suspensions were centrifuged at 1620 × g for 30 min, and the supernatants were used for plaque assays on monolayers of Vero cells seeded in 24-well plates. The cells were infected by incubation for 1 h at 37°C with serial dilutions of brain suspensions. After 1 h infection, 2% methylcellulose was overlaid, and the plates were incubated for ∼48 h. The overlay was removed, and cells were fixed with 4% paraformaldehyde for 15 min and stained with 1% crystal violet for 30 min before plaque counting.

The cells were treated with cGAMP in digitonin permeabilization solution (50 mM HEPES [pH 7], 100 mM KCl, 3 mM MgCl₂, 0.1 mM DTT, 85 mM sucrose, 0.2% BSA, 1 mM ATP, 0.1 mM GTP, and 10 μg/ml digitonin) at 37°C for 30 min. The cells were then incubated in regular medium for 4 or 12 h before qPCR or ELISA experiments were performed, respectively.

Reconstitution of LSm14A into DCs was performed by lentiviral-mediated gene transfer. Briefly, HEK293T cells plated on 100-mm dishes were transfected with pLOV-LSm14A (8 μg) together with the pVSV-G (4 μg), pRSV-REV (4 μg), and pMDL g/p RRE (4 μg) plasmids. Two days after transfection, the viruses were harvested and used to infect DCs in the presence of polybrene (8 μg/ml). Another 2 d later, the infected cells were collected for qPCR analysis.

Cells were stained in ice-cold PBS containing 1% FBS using appropriate Ab-fluorophore conjugates. The following Abs were purchased from BD Biosciences: CD4-PerCP, CD8-PB, CD3-FITC, B220-allophycocyanin, CD11b-PE, CD11b-FITC, CD11c-PE, CD11c-FITC, Gr-1-FITC, CD8-allophycocyanin, IFN-γ-PE, GL7-FITC, and FAS-PE. For immune cell development analysis, lymph nodes and spleens were obtained from the indicated genotypes of mice, and single-cell suspensions were prepared. After depletion of RBCs by ammonium chloride, cells were subject to staining.

Nuclei were isolated from DCs and the reaction was performed as described previously (27). Briefly, DCs (5 × 10⁶) were incubated for 5 min at 4°C with cold swelling buffer (10 mM Tris-HCl [pH 7.5], 2 mM MgCl₂, and 3 mM CaCl₂). After centrifugation at 1000 rpm for 10 min, the pellet was resuspended in lysis buffer (10 mM Tris-HCl [pH 7.5], 2 mM MgCl₂, 3 mM CaCl₂, 0.5% Nonidet P-40, 10% glycerol, and 2 U/ml RNase inhibitor) and gently pipetted up and down 20 times using a p1000 tip with the end cut off. The nuclei were pelleted at 1000 rpm for 10 min and washed with lysis buffer and then with storage buffer (50 mM Tris-HCl [pH 8.3], 40% glycerol, 5 mM MgCl₂, 0.1 mM EDTA, and 2 U/ml RNase inhibitor). The nuclei were added to reaction buffer (10 mM Tris-HCl [pH 8.3], 5 mM MgCl₂, 1 mM DTT, 300 mM KCl, and 100 U/ml RNase inhibitor) containing ATP, CTP, GTP, and Br-UTP or UTP (500 mM each) in the presence of 0.5% sarkosyl. After transcription at 30°C for 30 min, nuclear run-on (NRO)-RNAs were extracted using the TRIzol reagent, according to the recommendation of the manufacturer, and treated with DNase I. The NRO-RNAs were denatured at 70°C for 10 min and then incubated with preblocked (PBS with 0.1% Tween 20, 0.1% polyvinylpyrrolidone, and 0.1% UltraPure BSA) agarose beads conjugated with anti-BrdU Ab. After 30 min at room temperature on a rotating platform, beads were washed three times with PBSTR (PBS with 0.1% Tween-20 and 8 U/ml RNase Inhibitor). Immunoprecipitated NRO-RNAs were extracted using the TRIzol reagent, and reverse transcription was performed following standard protocol, except that glycogen was added during isopropanol precipitation step, and random primer was used for reverse transcription instead of oligo-dT. qPCR was performed using the following NRO-primers, which were designed for nascent pre-mRNAs: NRO-Mita, 5′-CTTGCCTTTGTCTCCTCTCC-3′ (forward) and 5′-AAAGAGAGACCCTGCCTCAA-3′ (reverse); NRO-Gapdh, 5′-GCCCTATAGGCCAGGATGTA-3′ (forward) and 5′-ATCTGGTTTCTGGAGGATGG-3′ (reverse); and NRO-Actb, 5′-TGCTAAGAAGGCTGTTCCCT-3′ (forward) and 5′-CATGTCGTCCCAGTTGGTAA-3′ (reverse).

The pLOV-Firefly luciferase reporter plasmids containing the full-length, 5′-UTR, CDS, or 3′-UTR of Mita, and pLOV-Renilla luciferase reporter plasmid were expressed in Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} DCs via lentiviral-mediated gene transfer. Two days after lentiviral infection, Firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega).

To measure the half-life of Mita mRNA, cells were incubated in culture medium containing Act D (the RNA polymerase II inhibitor; 5 μg/ml). Total RNA was isolated at the indicated times.

For keyhole limpet hemocyanin (KLH) experiments, Lsm14a^{f/f: Vav-Cre} mice and their wild-type controls (6–8 wk old) were immunized with KLH (0.15 mg/mouse) emulsified in IFA in the absence or presence of cGAMP (10 μg/mouse) s.c. Seven days after immunization, the mice were sacrificed and analyzed individually. Spleen cells (3 × 10⁶) were seeded in 48-well plates and stimulated with KLH for 24 h. The cell culture supernatants were collected and analyzed for IFN-γ concentration by ELISA. CTLs were determined by surface staining with CD8-allophycocyanin, followed by intracellular staining with IFN-γ-PE. Germinal center (GC) B cells were determined by staining with B220-allophycocyanin, GL7-FITC, and FAS-PE.

For OVA experiments, mice were immunized with OVA (10 μg/mouse) in the absence or presence of cGAMP (10 μg/mouse) via the i.m. route. Seven days after immunization, the sera were collected to measure anti-OVA-specific IgG1 by ELISA. In brief, 96-well plates were precoated with OVA (10 μg/ml) overnight and then blocked with PBS containing 10% FBS. Plates were washed and overlaid with serially diluted sera for 1 h at room temperature. After washing, Abs were detected with rat anti-mouse IgG1 conjugated to HRP. After further washing, the plates were stained using the tetramethylbenzidine substrate. The reaction was stopped with 1 N H₂SO₄, and the absorbance was measured.

Data were analyzed using a Student unpaired t test, multiple t test, or two-way ANOVA with Prism 6. The number of asterisks represents the degree of significance with respect to p values, with the latter presented within each figure or figure legend.

Mouse LSm14A is consisted of 462-aa residues and shares 94.6% sequence identity with its human ortholog. To further elucidate the physiological roles of LSm14A in vivo, we generated LSm14A-deficient mice by knockout-first strategy (28) (Supplemental Fig. 1A). The successful targeting of Lsm14a gene was verified by genotyping, immunoblot analysis and qPCR (Supplemental Fig. 1B–D). Unexpectedly, of 727 offspring from Lsm14a^+/− parents, only 18 Lsm14a^−/− mice were identified (Supplemental Fig. 1E). Further experiments indicated that the genotypes of day 12.5 embryos matched the Mendelian ratio, suggesting that LSm14A plays a nonessential role for the late phase of embryo development. In addition, the body sizes and weights were markedly different among different genotypes of livable mice. Although the average body weight of Lsm14a^+/+ mice at week 8 was 22.05 ± 4.15 g, the average body weight of Lsm14a^−/− mice was 18.48 ± 2.32 g, which was 15% lower than their wild-type counterparts (Supplemental Fig. 1F). The total cell numbers and compositions of major immune cells in lymph nodes and spleens were similar between Lsm14a^+/+ and Lsm14a^−/− mice (Supplemental Fig. 1G), indicating that LSm14A is dispensable for immune cell development.

Because it was difficult to obtain enough Lsm14a^−/− mice for in vivo experiments, we used the benefit of the knockout-first strategy to obtain Lsm14a hematopoietic–specific knockout strain (Lsm14a^{f/f: Vav-Cre}). The successful targeting of Lsm14a gene in monocytes, macrophages, and DCs were verified by genotyping, immunoblot analysis, and qPCR (Supplemental Fig. 2A–C). Lsm14a^{f/f: Vav-Cre} mice were viable and healthy, and their immune cell development was normal (Supplemental Fig. 2E). To our surprise, Lsm14a^{f/f: Vav-Cre} mice exhibited a 15% decrease at body weight as well (Supplemental Fig. 2D). The reason why Lsm14a^{f/f: Vav-Cre} mice were smaller will need further investigation.

To investigate the functions of LSm14A, we generated MEFs, bone marrow–derived macrophages and DCs from sex- and age-matched wild-type and LSm14A-deficient mice. qPCR experiments indicated that the transcription of Ifnb1, Ifna4, and Il6 genes in responding to the DNA viruses HSV-1 and MHV68 as well as the RNA virus VSV was significantly impaired in Lsm14a^−/− DCs in comparison with their wild-type counterparts (Fig. 1A). In these experiments, transcription of the cytokine genes in response to the RNA virus SeV as well as poly(I:C) (a ligand for TLR3) or LPS (a ligand for TLR4) was not affected by LSm14A deficiency in DCs (Fig. 1A). Consistently, HSV-1–, MHV68-, and VSV-, but not SeV-, poly(I:C)-, and LPS-triggered secretion of cytokines, including IFN-β, IFN-α and IL-6, was significantly decreased in Lsm14a^−/− DCs in comparison with their wild-type counterparts (Fig. 1B). In similar experiments, transcription of Ifnb1 and Il6 genes in response to all the examined stimuli, including HSV-1, MHV68, VSV, SeV, poly(I:C), and LPS, was not affected by LSm14A deficiency in bone marrow–derived macrophages and MEFs (Fig. 1C). In parallel experiments, the mRNA levels of HSV-1 and SeV genes showed no differences in Lsm14a^+/+ and Lsm14a^−/− DCs, which indicated that LSm14A-deficiency did not impair virus infection (Fig. 1D). Biochemically, HSV-1– but not SeV-triggered phosphorylation of TBK1, IRF3, and IκBα, which are hallmarks of the activation of virus-triggered induction of downstream cytokines and antiviral effectors, were markedly decreased in Lsm14a^−/− DCs in comparison with their wild-type counterparts (Fig. 1E). The above results suggest that LSm14A plays an important role in innate immune responses to DNA and certain RNA viruses in DCs but not in other examined cell types.

Because LSm14A regulates innate immune response in a cell type–specific manner, the primary culture of bone marrow–derived macrophages and DCs were further characterized. In the adherent fraction of M-CSF culture, >90% were macrophages (CD11b⁺CD11c⁻ population). At the same time, culturing with GM-CSF without further purification leads to a mixed population of cells in which DCs (CD11c⁺ population) accounted for 71∼77% of total cells. The CD11c^- fraction of GM-CSF culture contained a significant number of Gr-1⁺ granulocytes (Supplemental Fig. 3A). To examine which cell population in the GM-CSF culture accounts for the cell type–specific function, we magnetically separated the GM-CSF culture into CD11c⁻ and CD11c⁺ fraction. We found that LSm14A deficiency resulted in impaired Ifnb1 transcription upon HSV-1 infection only in CD11c⁺ (bone marrow–derived DCs) but not CD11c⁻ fraction (Supplemental Fig. 3B, 3C). This result was consistent with our conclusion that LSm14A plays an important role in innate antiviral immune response in DCs but not in monocytes or macrophages.

Because LSm14A plays an important role in DNA virus–triggered induction of cytokines in DCs, we next determined whether LSm14A was required for the induction of downstream genes triggered by intracellular DNA stimulation in DCs. Previously, it has been demonstrated that transfected dsDNA (60- or 120-mer) representing the genomes of HSV-1 (HSV60 or HSV120), dsDNA (70-mer) representing the genomes of vaccinia virus (VACV70), dsDNA of ∼90 bp (DNA90), and IFN stimulatory DNA of 45 bp (ISD45) are efficient at inducing the expression of type I IFNs (19, 29). We found that the levels of Ifnb1, Ifna4, and Il6 mRNAs (Fig. 2A), as well as the secreted IFN-β, IFN-α, and IL-6 cytokines (Fig. 2B), were significantly lower in Lsm14a^−/− in comparison with Lsm14a^+/+ DCs following transfection of the respective synthetic DNAs. In similar experiments, mRNA levels of the cytokine genes induced by transfected DNA90 and ISD45 were comparable between Lsm14a^+/+ and Lsm14a^−/− macrophages and MEFs (Fig. 2C). These results suggest that LSm14A is required for efficient induction of cytokines in response to intracellular DNA stimulation in DCs but not in macrophages and MEFs.

To gain insights into the importance of LSm14A in host defense against viral infection in vivo, we investigated innate antiviral immune responses in Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice. Age- and sex-matched Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice were infected with HSV-1, VSV via the i.p. route, or SeV via the intranasal route. HSV-1– and VSV-induced cytokines in the serum, including IFN-β, IFN-α, and IL-6, were severely impaired in Lsm14a^{f/f: Vav-Cre} in comparison with Lsm14a^f/f mice (Fig. 3A). In parallel experiments, the levels of serum cytokines induced by SeV infection were comparable between Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice. Because HSV-1 is a neurotropic virus and the leading cause of sporadic viral encephalitis, we investigated the effects of LSm14A deficiency on HSV-1–induced expression of downstream genes and viral loads in the brain. The results indicated that the levels of Ifnb1, Isg56, Ccl5, Tnfa, and Il6 mRNAs were markedly reduced in the brains of Lsm14a^{f/f: Vav-Cre} mice in comparison with their wild-type counterparts (Fig. 3B). When the brains of Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice were extracted to measure viral titers on day 5 postinfection, significantly higher levels of HSV-1 were detected in Lsm14a^{f/f: Vav-Cre} comparing to Lsm14a^f/f mice (Fig. 3C). The survival rate of Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice upon HSV-1 infection showed no evident difference (Fig. 3D). It is possible that the DC-specific effects of LSm14A is not dramatic enough for affecting the HSV-1–induced survival of the mice. Collectively, these results suggest that LSm14A is essential for host defense against viral infection in mice.

It has been demonstrated that, after recognizing cytosolic DNA, cGAS catalyzes synthesis of the small messenger molecule cGAMP, which binds to MITA to activate the downstream signaling pathways. To investigate the mechanisms on how LSm14A is involved in DNA virus-triggered signaling in DCs, we performed bioassays to measure cGAMP activity in dsDNA-transfected Lsm14a^+/+ and Lsm14a^−/− DCs. To do this, we transfected HSV60 or VACV70 into Lsm14a^+/+ or Lsm14a^−/− DCs for 4 h. The cell extracts were then prepared and heated at 95°C to denature most proteins, which were removed by centrifugation. The supernatants containing cGAMP were delivered to digitonin-permeabilized MEFs and induction of Ifnb1, Ifna4, and Il6 mRNAs in the MEFs was then measured by qPCR. These assays showed that the extracts from dsDNA-transfected Lsm14a^+/+ and Lsm14a^−/− DCs contained comparable cGAMP activity (Fig. 4A), suggesting that LSm14A deficiency does not affect cytosolic DNA-triggered synthesis of cGAMP.

We next determined whether LSm14A is required for cGAMP-triggered downstream signaling. To do this, chemically synthesized cGAMP was delivered to digitonin-permeabilized DCs and induction of downstream cytokine genes was measured. The results indicated that the levels of Ifnb1, Ifna4, and Il6 mRNAs (Fig. 4B), as well as the secreted IFN-β, IFN-α, and IL-6 cytokines (Fig. 4C), were markedly lower in Lsm14a^−/− DCs in comparison with their wild-type counterparts. In similar experiments, cGAMP-triggered induction of the cytokine genes was comparable between Lsm14a^+/+ and Lsm14a^−/− macrophages and MEFs (Fig. 4D, 4E). Consistently, cGAMP-triggered phosphorylation of TBK1 and IRF3 was also inhibited in Lsm14a^−/− DCs in comparison with their wild-type counterparts (Fig. 4F).

These results suggest that LSm14A acts downstream of cGAMP in innate antiviral responses in DCs.

To investigate how LSm14A acts downstream of cGAMP in innate antiviral responses, we examined whether the levels or posttranslational modifications of related signaling components were affected by LSm14A deficiency. Surprisingly, we found that the protein level of MITA, but not other examined components including cGAS, TBK1, and IRF3, was markedly decreased in Lsm14a^−/− DCs in comparison with their wild-type counterparts (Fig. 5A, Supplemental Fig. 4A). Most interestingly, the protein level of MITA was not affected by LSm14A deficiency in mouse monocytes, macrophages, MEFs, and CD11c⁻ fraction of GM-CSF culture (Fig. 5A, 5B, Supplemental Fig. 4B). Knockdown of LSm14A in the human monocytic cell line THP-1 also did not affect the level of MITA (Fig. 5B). These results suggest that LSm14A is specifically required for maintaining MITA level in mouse DCs but not in other examined cell types.

We further examined the mRNA level of signaling adaptors in Lsm14a^+/+ and Lsm14a^−/− DCs. As shown in Fig. 5C and Supplemental Fig. 4C and 4D, only mRNA of Mita but not other examined genes including cGas, Tbk1, and Irf3 was dramatically reduced in Lsm14a^−/− DCs, suggesting that LSm14A is specifically required for maintaining the mRNA level of Mita in DCs. We also examined the mRNA levels of Mita in various mouse tissues. The results showed that LSm14A deficiency resulted in downregulation of the mRNA level of Mita in the liver, heart, lymph nodes, thymus, brain, and spleen but not the kidney and lung (Supplemental Fig. 4E), suggesting that LSm14A regulates Mita mRNA level in a tissue-specific manner. Because LSm14A plays an important role in maintaining Mita mRNA level in mouse liver, we further examined whether Mita mRNA was regulated by LSm14A in human hepatocellular carcinoma cell line Huh7. As shown in Supplemental Fig. 4F and 4G, knockdown of LSm14A decreased the mRNA and protein level of MITA in Huh7 cells, which is consistent with our previous observations that knockdown of LSm14A impaired virus-triggered induction of type I IFNs in these cells (6). These results suggest that the regulation of Mita level by LSm14A is not restricted in DCs.

To further confirm that LSm14A positively regulates virus-triggered innate immune signaling via maintaining the mRNA level of Mita, we reconstituted Lsm14a^−/− DCs with LSm14A by lentiviral-mediated gene transfer. As shown in Fig. 5D and Supplemental Fig. 4H, reconstitution of LSm14A significantly restored the mRNA and protein level of MITA in Lsm14a^−/− DCs. Consistently, reconstitution of LSm14A also significantly restored HSV-1 induced Ifnb1 and Ifna4 genes transcription in DCs (Fig. 5E).

To investigate how LSm14A regulates the mRNA level of Mita in DCs, we first examined whether the transcription of Mita gene was affected by LSm14A deficiency. We performed NRO assays to measure the transcriptional activity via the quantification of biochemically labeled nascent RNA derived from nuclear isolates (27). As shown in Fig. 6A, nascent transcription of Mita mRNA was comparable between Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} DCs, indicating that LSm14A does not regulate transcription of the Mita gene. In NRO assays, we noticed that the Mita pre-mRNA was comparable between Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} DCs, and only mature Mita mRNA was downregulated in Lsm14a^{f/f: Vav-Cre} DCs (Fig. 6B).

We next examined whether the decrease of mature Mita mRNA was a result of its faster degradation in Lsm14a^{f/f: Vav-Cre} DCs. To do this, the full-length, 5′ untranslated region (UTR), CDS or 3′-UTR of Mita mRNA were cloned upstream or downstream of a luciferase reporter gene as indicated in Fig. 6C. The constructs were then expressed in DCs via lentiviral-mediated gene transfer. Comparable luciferase expression from all the constructs was detected between Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} DCs, indicating that LSm14A does not regulate Mita mRNA stability in a sequence-specific manner. A further quantitative evaluation of the mRNA decay rate revealed a half-life of 2.9 h for Mita mRNA in Lsm14a^f/f DCs and 3.6 h in Lsm14a^{f/f: Vav-Cre} DCs, respectively (Fig. 6D). The results further demonstrated that LSm14A did not increase Mita mRNA stability. Taken together, these results suggest that LSm14A regulates Mita pre-mRNA processing but not its transcription or stability. Because pre-mRNA processing occurs in the nucleus, we examined the cellular localization of LSm14A in DCs. As shown in Fig. 6E, LSm14A localized in the cytoplasm in DCs, which is consistent with previous reports that LSm14A is specifically localized in the P-body. Because we have demonstrated that LSm14A deficiency led to reduced mature Mita mRNA in DCs but not Mita pre-mRNA, we examined the levels of mature Mita mRNA in nucleus and cytoplasm, respectively. As shown in Fig. 6F, both of the mature Mita mRNA levels in nucleus and cytoplasm were reduced in LSm14A-deficient DCs. These results suggest that the regulation of mature Mita mRNA level is most likely happened in the nucleus, and LSm14A may regulate Mita level indirectly through a mediator.

It has been shown that MITA is indispensable in mediating the adjuvant activity of DNA vaccine or cGAMP in vivo (11, 19). To investigate whether LSm14A is involved in these functions, Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice were immunized with KLH emulsified in IFA in the absence or presence of cGAMP s.c. Splenocytes were harvested and restimulated with KLH 7 d after immunization. We found that T cells from Lsm14a^{f/f: Vav-Cre} mice produced lower levels of IFN-γ compared with Lsm14a^f/f mice (Fig. 7A). In addition, KLH-specific CTL frequency in draining lymph nodes (dLNs) was markedly reduced in Lsm14a^{f/f: Vav-Cre} mice (Fig. 7B), which collectively indicated that LSm14A was required for cGAMP-mediated T cell activation. To investigate the role of LSm14A on cGAMP-mediated B cell activation, germinal center (GC) B cells frequency in dLNs was examined. As shown in Fig. 7C, GC B cells in dLNs were markedly decreased in Lsm14a^{f/f: Vav-Cre} mice in comparison with their wild-type counterparts. When Lsm14a^f/f and Lsm14a^{f/f: Vav-Cre} mice were immunized with OVA in the absence or presence of cGAMP via the i.m. route, Lsm14a^{f/f: Vav-Cre} mice produced significantly less serum OVA-specific IgG1 compared with controls (Fig. 7D). Taken together, these results suggest that LSm14A is required for effective cGAMP-mediated adaptive immune responses.

LSm14A is P-body–associated protein that has been implicated in regulation of cellular antiviral responses in certain human cell lines (6). In this study, we performed gene knockout studies on the roles of LSm14A in innate antiviral immune responses. Surprisingly, we found that LSm14A plays an important role in innate antiviral responses in DCs but not in macrophages and MEFs. Our results also suggest that LSm14A functions by regulating pre-mRNA processing of MITA, a central component in the antiviral signaling pathways.

Several experiments suggest that LSm14A plays a critical role in innate antiviral responses in a cell-specific manner. Induction of IFN-α, IFN-β, and IL-6 following infection with HSV-1, MHV68, and VSV, or transfection with various synthetic DNA, was markedly impaired in Lsm14a^−/− DCs but not in macrophages and MEFs. The functions of LSm14A in DCs are specific because induction of these cytokines by poly(I:C) (a ligand for TLR3) and LPS (a ligand for TLR4) in DCs, as well as in macrophages and MEFs, was not impaired by LSm14A deficiency. These results suggest that LSm14A is critically involved in innate antiviral responses triggered by the cytoplasmic DNA/RNA sensors in DCs but not in macrophages and MEFs. In addition to DCs, LSm14A may be also important for innate antiviral responses in some other cell types. A survey of various tissues indicated that the mRNA levels of Mita in the liver, heart, lymph nodes, thymus, brain, and spleen, but not in the kidney and lung, were downregulated in Lsm14a^−/− mice. In addition, knockdown of LSm14A decreased the mRNA and protein level of MITA in human hepatic Huh7 cells. Because LSm14A functions by regulating MITA pre-mRNA processing and protein level, it is highly possible that LSm14A is important for innate antiviral immune responses in certain cell types of the liver, heart, lymph nodes, thymus, brain, and spleen. The mechanisms on how LSm14A acts in a cell-specific manner are unknown at this point. Apparently, the mechanisms are not related to the molecular sizes and abundance of LSm14A in DCs, macrophages and MEFs and it is believed that LSm14A is regulated by unknown mechanisms rather than protein levels or protein posttranslational modifications in various cell types. It is possible that DCs specifically express one or more unknown components, which are important for LSm14A-mediated processing of Mita mRNA precursor.

There are several lines of evidence suggesting that LSm14A acts in innate antiviral responses by regulating the proper level of MITA, an essential adaptor protein in the antiviral signaling pathways. First, Lsm14a^−/− DCs produce less type I IFNs and proinflammatory cytokines postinfection with the DNA viruses HSV-1 and MHV68, the RNA virus VSV but not SeV. This parallels with the roles of MITA, as it has been demonstrated that cytokine production triggered by HSV-1 and VSV but not SeV was impaired in Mita^−/− DCs. Second, cGAMP synthesis triggered by transfected cytoplasmic DNA was not affected in Lsm14a^−/− DCs, but cGAMP-triggered induction of downstream cytokines was impaired in Lsm14a^−/− DCs. These results further suggest that LSm14A regulates innate antiviral responses at the level of MITA. Third, the protein and mRNA levels of MITA were downregulated in Lsm14a^−/− DCs but not in macrophages and MEFs, whereas reconstitution of LSm14A in Lsm14a^−/− DCs significantly restored MITA expression and virus-triggered induction of antiviral type I IFNs. Previously, it has been shown that LSm14A is involved in SeV-induced signaling in certain human cell lines (6). In the current study, we found that LSm14A was not required for SeV-induced signaling in mouse cells. These results are consistent with previous reports that MITA plays an important role in SeV-induced transcription of downstream genes in human cell lines but is not required for SeV-induced signaling in primary mouse cells (14–16, 19, 20).

We further investigated the mechanism on how LSm14A regulates MITA mRNA level. Reporter and mRNA decay assays indicate that MITA mRNA in LSm14A-deficient DCs is as stable as in the wild-type DCs. NRO experiments indicate that transcription of the Mita gene is not affected in LSm14A-deficient DCs. Interestingly, results from the NRO experiments indicate that the mature MITA mRNA but not its pre-mRNA is downregulated in LSm14A-deficient DCs. The simplest explanation from these results is that LSm14A plays an important role in regulating MITA pre-mRNA processing.

Consistent with a critical role for regulation of MITA level by LSm14A in DCs, hematopoietic-specific Lsm14a knockout mice exhibit impaired innate antiviral responses, such as significantly reduced production of serum IFN-α, IFN-β, and IL-6 following infection with HSV-1 and VSV but not SeV. The levels of Ifnb1, Isg56, Ccl5, Tnfa, and Il6 mRNAs were markedly reduced, whereas the viral titers were significantly increased in the brains of Lsm14a^{f/f: Vav-Cre} mice in comparison with their wild-type counterparts. In addition, our results indicate that LSm14A is essential for effective cGAMP-mediated T and B cell activation. Collectively, our results suggest that LSm14A is essential for host defense against viral infection in mice.

In conclusion, our findings reveal a cell-specific role for the P-body–associated protein LSm14A in antiviral immune responses. Control of MITA pre-mRNA processing and therefore its expression level by LSm14A represents an unexpected and exciting regulatory mechanism of antiviral immune responses.

We thank Dr. Xi Wang at Tianjin Medical University for providing Vav-Cre mice and the Model Animal Research Center of Nanjing University for help with generating Lsm14a knockout mice.

The authors have no financial conflicts of interest.