IFI16 Isoforms with Cytoplasmic and Nuclear Locations Play Differential Roles in Recognizing Invaded DNA Viruses

Recognition of nucleic acids from invaded viruses by pathogen recognition receptors is critical for initiating and amplifying host antiviral immunity (1, 2). Inductions of type I IFN (IFN-I) and IFN-I downstream IFN-stimulated genes (ISGs) are the key events in protecting the host against viral infection (3). Although many candidate sensors of cytosolic DNA have been proposed (4), cyclic GMP-AMP (cGAMP) synthase (cGAS) and IFN-γ–inducible protein 16 (IFI16) are considered as two of the most important DNA sensors in inducing IFN-I–dependent antiviral immunity during DNA viral infection (1, 2). Upon viral DNA recognition, cGAS dimerizes and stimulates the formation of the second messenger cGAMP (5, 6), which binds to the endoplasmic reticulum–bound protein stimulator of IFN genes (STING). The interaction between STING and cGAMP induces conformational changes on the STING C-terminal domain that allow STING dimerization, recruitment, and phosphorylation of TANK-binding kinase 1 (TBK1). TBK1 binding to STING initiates a complex cascade of events, including phosphorylation of STING and IRF3, and drives IFN-I transcription (7). IFI16, a cytosolic and nuclear protein, triggers the STING-TBK1–IRF3–IFN-β signaling axis and activates IFN-α/β receptor downstream antiviral signaling after recognizing both double-stranded viral DNA (dsDNA) and single-stranded viral DNA from invaded viruses such as HSV type 1 (HSV-1) (8), Kaposi’s sarcoma–associated herpesvirus (KSHV) (9), EBV (10), hepatitis B virus (11), vaccinia virus (VACV) (8), and HIV type 1 (12). IFI16 is also required for cGAS-mediated cGAMP production and IFN-I induction in human macrophages (13). Moreover, cooperation of IFI16 and cGAS triggers optimal STING activation and IFN-I production in human keratinocytes, although the structural basis for the role of IFI16 in the cGAS-STING pathway remains unclear (14). In addition to these cytosolic DNA sensors, hnRNPA2B1 has been identified recently as a nuclear DNA sensor to specifically recognize nuclear viral DNA from HSV-1 and then initiate an IFN-I–dependent antiviral immune response (15).

IFI16 is a unique DNA sensor because it not only recognizes both single-stranded DNA from HIV type 1 and dsDNA from HSV-1 (8, 12) but also shuttles between the cytoplasm and the nucleus (16, 17). IFI16 interacts with the adaptor molecular apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) and caspase-1 to form a functional inflammasome in the nucleus during KSHV infection (9). Interestingly, this IFI16-mediated inflammasome induction by KSHV is associated with subcellular redistribution of ASC, caspase-1, and IFI16 from the nucleus to the cytoplasm (9). Genome recognition of herpesvirus, including KSHV, EBV, and HSV-1 by IFI16 in the nucleus leads to IFI16-ASC-caspase1 inflammasome assembly, cytoplasmic translocation, and IL-1β induction (16). Meanwhile, HSV-1 genome recognition results in IFI16–STING interaction in the cytoplasm to induce IFN-β production (16). Acetylation of IFI16 by the acetyltransferase p300 during HSV-1 infection is essential for the IFI16 nucleocytoplasmic redistribution (16), whereas histone deacetylases promote IFI16 nuclear import (17). The DNA-sensing ability of IFI16 is modulated by acetylation of lysine (K) 99 and K128 within its nuclear localization sequence (NLS), which is between the PYD and HINa domains of IFI16 (17).

Human IFI16 is encoded by the IFI16 gene (National Center for Biotechnology Information [NCBI] gene identifier 3428), which is located on 1q23.1. Alternatively spliced transcript variants encoding four isoforms (IFI16-iso1, 2, 3, and 4) have been identified for the IFI16 gene, although only three distinct protein species (IFI16A, B, and C) are usually detected by immunoblotting (18). IFI16-iso2, which corresponds to the studies reported previously (IFI16B), is the most abundantly expressed and extensively studied as a tumor suppressor and as a DNA sensor (8, 9, 19, 20). IFI16-iso2 protein is encoded by the transcript variant 2 (NCBI reference sequence NM_005531.3), resulting from the deletion of exon 7a (168 bp) to encode a protein of 729 aa (18). Compared with the IFI16 transcript variant 2, variant 1 (NCBI reference sequence NM_001206567.2) lacks an in-frame exon near the 5′-coding region that encodes the corresponding NLS motifs 2, 3, and 4 in IFI16-iso2, but it has an additional equal-length in-frame exon in the 3′-coding region (17, 18, 21). As a result, IFI16-iso1, which is encoded by the transcript variant 1, is of the same length (729 aa) as the IFI16-iso2 but lacks the core NLS motifs. However, it is totally unknown whether IFI16-iso1 mainly locates in the cytoplasm and plays distinct roles in triggering host antiviral immunity against different DNA viral infections.

Every DNA virus with its distinct life cycle differentially replicates in the cytoplasmic compartment or nucleus. Herpesviruses, such as HSV-1 and KSHV, are representative of nucleus-replicating DNA viruses (22, 23). After infection, the HSV-1 capsids move along microtubules to nuclear pores, where the virions release viral DNA into the nucleus for amplification (23). VACV, a prototype member of the poxvirus family, is unique among most DNA viruses because its replication occurs in the cytoplasm (e.g., cytoplasmic virosomes and endoplasmic reticulum–enclosed cytoplasmic mini-nuclei) of the infected cells (24, 25). Both HSV-1 and VACV infections induce IFI16 protein expression (26). IFI16 recognizes viral DNAs from both HSV-1 and VACV and triggers IFN-I–dependent antiviral immunity (8). Endogenous IFI16 has been shown to colocalize with the transfected dsDNA derived from VACV and HSV-1 in the cytoplasm of the differentiated THP-1 monocytes (8). In contrast, nuclear IFI16 colocalizes with KSHV and HSV-1 genomic DNA during early infection, which is accompanied by IFI16 acetylation and cytoplasmic translocation. Given that cytoplasm-located IFI16-iso1 has more opportunities than IFI16-iso2 to recognize the cytoplasmic viral DNA, such as transfected HSV60mer, VACV70mer, and infected VACV, IFI16-iso1 may play an important role in recognizing cytoplasmic viral DNA as the acetylated IFI16-iso2.

Cytoplasmic viral DNA mimics such as ISD45mer, GHV50mer, HSV60mer, VACV70mer, and G3-YSD (YG3) are usually used to study cytosolic DNA sensors, including IFI16 and cGAS (8, 27, 28). Besides the live nucleus-replicating DNA viruses, nuclear viral DNA mimics will be very useful to study the nuclear DNA sensors. To distinguish the functional difference between cytoplasmic IFI16-iso1 and nuclear IFI16-iso2, we have developed potent viral DNA agonists for nuclear DNA sensors by adding a hexanucleotide motif. Moreover, we have shown that IFI16-iso1 and IFI16-iso2 with different subcellular locations play differential roles in recognizing viral DNA and inducing host innate immunity against DNA viruses.

Primary Ab anti-IFI16 (sc-8023) was purchased from Santa Cruz Biotechnology (Dallas, TX); anti–p-TBK1(S172) (5483S), anti–p-STAT1(Y701) (9167), anti-TBK1 (51872), anti-STAT1 (14994), anti-OAS1 (14498), and anti–β-actin (3700) Abs were obtained from Cell Signaling Technology (Danvers, MA); anti–α-tubulin (T5168) and anti-Flag (F1804) were purchased from Sigma-Aldrich (St. Louis, MO); and anti-BrdU (600-401-C29) was purchased from Rockland (Gilbertsville, PA). Secondary fluorescent Abs anti-mouse IRDye800CW, anti-rabbit IRDye800CW, anti-mouse IRDye680RD, and anti-rabbit IRDye680RD were purchased from LI-COR Biosciences (Lincoln, NE). The plasmids Flag-IFI16-iso1, Flag-IFI16-iso2, and Flag-STING were gifted from Genhong Cheng Laboratory (University of California, Los Angeles). The hnRNPA2B1 plasmid was from the Xuetao Cao Laboratory (Second Military Medical University, Shanghai, China). The Flag-IFI16-iso2-K128Q vector was mutated from Flag-IFI16-iso2 by using a Muta-direct kit (SBS Genetech, Beijing, China), and the sequence was validated by Sanger sequencing.

The cyanine 5 (Cy5)-labeled oligonucleotides HSV60mer, HSV60-DNLS, HSV60-RNLS, YG3-DNLS, YG3-RNLS, VACV70mer, and VACV70-DNLS used in confocal microscopy were synthesized by Synbio Technologies (Suzhou, China). The unlabeled dsDNA oligonucleotides HSV60mer, HSV60-DNLS, HSV60-NC, VACV70mer, and VACV70-DNLS were synthesized by GENEWIZ (Suzhou, China). All these viral DNA mimics were hPAGE purified, dissolved in endotoxin-free double-distilled H₂O (Sigma-Aldrich), and annealed gradually (90°C, 1 min; 80°C, 1 min; 70°C, 1 min; 60°C, 1 min; 50°C, 1 min; store at 4°C). The sequences for HSV60mer, VACV70mer, and YG3 were described previously (8, 27).

Human lung carcinoma A549 cells, human embryonic kidney (HEK) 293T cells, human monocyte leukemia THP-1 cells, and Vero monkey kidney cells were purchased from the American Type Culture Collection. The wild-type (WT) and Hnrnpa2b1^−/− L929 cells were provided by the Xuetao Cao Laboratory (Second Military Medical University). WT and IFI16^−/− HaCaT cells were provided by Professor Leonie Unterholzner (University of Edinburgh, Edinburgh, UK). SW620, SW480, HT29, and U251 cells were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). A549, HEK293T, HaCaT, SW620, SW480, HT29, U251, and Vero cells were cultured in DMEM. L929 and THP-1 cells were cultured in RPMI 1640 medium. All the cell lines were cultured in the medium supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C and 10% CO₂. For bone marrow–derived macrophage (BMDM) differentiation, bone marrow cells were harvested from WT C57BL/6 mice (∼8 wk old, male) and differentiated in DMEM containing 10 ng/ml recombinant mouse M-CSF (R&D Systems, Minneapolis, MN). Plasmids were transiently transfected into A549 and HEK293T cells by polyethyleneimine transfection reagents (Polysciences, Warrington, PA) at the ratio of 3:1 to 1:1 (μl reagent/μg DNA) according to the manufacturer’s instructions. Viral DNA mimics were transfected into cells by using Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA) at the ratio of 2.5:1 (μl reagent/μg DNA) according to the manufacturer’s instructions. The final concentration of viral DNA mimics transfected into cells was 1 μg/ml unless noted otherwise in the figure legends.

Total RNA from A549, BMDM, or L929 cells in 12-well plates was extracted using TRIzol (Thermo Fisher Scientific), and the first-strand cDNA was synthesized by using 500 ng total RNA as a template according to the manufacturer’s instructions for PrimeScript RT Master Mix (Takara, Beijing, China). Real-time QPCR amplification was performed to analyze gene mRNA expression levels using TB Green Premix Ex Taq (Takara) on a Roche LightCycler 480 II system. The relative mRNA level of genes was normalized to the internal control ribosomal protein RPL32 gene using the comparative cycle threshold method (29). The primer sequences for QPCR were generated from a primer bank (30). All the primer sequences are available upon request.

Total RNA from virus-infected cell lines was extracted by using TRIzol (Thermo Fisher Scientific), and the first-strand cDNA was synthesized by using 500 ng total RNA as a template according to the manufacturer’s instructions for PrimeScript RT Master Mix (Takara). The cDNA libraries of uninfected lung cancer, colorectal cancer, ovarian cancer, cervical cancer, and glioblastoma cell lines were provided by Professor Sudan He (Suzhou Institute of Systems Medicine, Suzhou, China). The cDNA library of uninfected gastric mucosa and gastric cancer cell lines was a gift from the Xuetao Cao Laboratory (Second Military Medical University). Then the cDNA was used as a template for endogenous IFI16 isoform PCRs using the primers 5′-CCGAGGTGATGCTGGTTTGGG-3′ (forward) and 5′-GGGAGTTACCTGACATTTGGCC-3′ (reverse). The amplicon sizes were 250 bp and 418 bp for IFI16-iso1 and IFI16-iso2, respectively. The PCR products were cloned into a T vector (pMD19-T Vector Cloning Kit; Takara) and examined by Sanger sequencing.

Whole-cell lysates were prepared in cold lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% [v/v] Nonidet P-40, 0.5 mM EDTA) supplemented with cOmplete protease inhibitor mixture (Roche, Basel, Switzerland) and Roche PhosSTOP. The total cell lysates were subjected to SDS-PAGE and immunoblot analysis with the indicated primary Abs and infrared dye–labeled secondary Abs. The fluorescence bands were visualized using the Odyssey CLx Imaging System (LI-COR Biosciences).

Cy5-labeled DNA mimics transfected A549 or HaCaT cells plated on glass coverslips were fixed using 4% formaldehyde for 20 min and permeabilized with 0.2% Triton X-100 for 10 min. The cells were washed with PBS three times and then stained with DAPI (0.1 μg/ml) for 10 min. After 30 min washing with PBS, slides were evaluated using a Leica TCS SP8 confocal microscope (Leica Microsystems). Cell images obtained were exported into TIF format using the LCS software package and analyzed by ImageJ software.

Pretreated A549 cells plated on glass coverslips were fixed using 4% formaldehyde for 20 min and permeabilized with 0.2% Triton X-100 for 10 min. The cells were then blocked with 1% goat serum and incubated with mouse anti-Flag primary Abs overnight. After washing three times with PBS, cells were incubated with anti-mouse secondary Abs for 1 h. After that, the cells were washed with PBS three times and then stained with DAPI (0.1 μg/ml). After 30 min washing with PBS, slides were evaluated using a Leica TCS SP8 confocal microscope (Leica Microsystems). Cell images obtained were exported using the LCS software package into TIF format for further analysis.

The Duolink in situ PLA detection kit (Sigma-Aldrich) was used to identify in situ IFI16–VACV or IFI16–HSV-1 interaction. Before staining, A549 cells plated on glass coverslips were transfected with Flag-IFI16-iso1 or IFI16-iso2 plasmids for 24 h and infected with BrdU-labeled VACV or HSV-1 for another 18 h. The pretreated cells were then fixed by using 4% formaldehyde for 20 min and permeabilized with 0.2% Triton X-100 for 10 min. The fixed cells were blocked with blocking buffer for 30 min and incubated with mouse anti-Flag and rabbit anti-BrdU primary Abs overnight. Secondary Ab incubation, ligation, and complex signal amplification were performed according to the manufacturer’s instructions. The amplification step would generate a signal when the two primary Abs against each Ag were in close enough proximity, as described previously (26). After DAPI staining, cells were detected on a Leica TCS SP8 confocal microscope. Images were exported into TIF format using the LCS software package. Cellular interaction signals (red spots) were measured by ImageJ software for statistical analysis.

HEK293T cells were seeded in 24-well plates at a density of 1.5 × 10⁵ cells per well overnight, and the cells were transfected with a mixture of IFN-β promoter firefly luciferase reporter (IFN-β-luc) plasmid (150 ng), Renilla luciferase control (20 ng), and other plasmids or viral DNA mimics using polyethyleneimine transfection reagents, or they were infected by DNA virus. At 24 h or 30 h after transfection, cells were lysed, and the luciferase activity was measured using the Dual-Luciferase Reporter Assay System according to the manufacturer’s instructions (Promega, Madison, WI).

Vero cells were used for HSV-1 and VACV amplification. Cells with a density of ∼50% were infected with HSV-1 or VACV (multiplicity of infection [MOI] 1), and the medium was replaced with fresh medium 1 h after infection. The viruses were collected 48 h later, titrated, and calculated by median tissue culture–infectious dose (TCID₅₀) assay as described previously (31). The viral genome DNA was labeled by BrdU using labeling reagent ordered from Thermo Fisher Scientific as described previously (16, 32). Briefly, BrdU labeling reagent was added to the cell culture medium in a 1:100 (v/v) ratio at 24 h or 72 h, and the labeled viruses were collected on the fifth day. The viral titration was measured by TCID₅₀ assay, and the BrdU labeling efficiency was identified by flow cytometry using an anti-BrdU Ab. The HSV-1 and VACV titration was performed by using Vero cells. The cells were plated in a 96-well plate, and serial dilutions of virus stocks were added to the culture medium after 24 h. At 48 h after infection, the cells in each well were inspected for cytopathic effect or death and classified as infected or not. The results were calculated using the classical Spearman-Kärber cytopathic effect reading method and shown as TCID₅₀/ml (31).

The L929 and A549 cells were plated in 12-well plates overnight, and these cells were transfected with DNA virus mimics (0.5 μg/ml) for 12 h or infected by HSV-1/VACV (MOI, 1) for 18 h. THP-1 cells were differentiated by PMA (100 ng/ml) for 24 h and transfected with DNA virus mimics (2.5 μg/ml) for 4 h or 8 h. The culture supernatants were subjected to ELISA for secreted IFN-β and IL-1β detection by the mouse IFN-β ELISA detection kit (PBL Assay Science, Piscataway, NJ), human IFN-β ELISA detection kit (Cloud-Clone, Wuhan, China), or human IL-1β ELISA detection kit (Dakewe Biotech, Shenzhen, China) according to the manufacturer’s instructions.

Experimental repeats are indicated in the figure legends. All bar graphs are shown as the mean ± SD. Statistical analysis was performed with the Student t test in GraphPad Prism 8 software. A p value less than 0.05 was considered statistically significant.

IFI16-iso2 mainly locates in the nucleus because it has an evolutionarily conserved multipartite NLS in its N-terminus (17). Viral infection induces acetylation of IFI16-iso2 NLS on K99 and K128 and thus drives IFI16 translocation into the cytoplasm (16, 17). Because mutation of K to glutamine (Q) mimics the acetylation state, we made the IFI16-iso2 NLS-acetylated mutants and confirmed that the WT IFI16-iso2 located in the nucleus and the IFI16-iso2-K128Q mutant mainly accumulated in the cytoplasm (Fig. 1A). Next, we analyzed the amino acid sequences of IFI16 isoforms and found that both IFI16-iso1 (NP_001193496.1) and IFI16-iso2 (NP_005522.2) had 729 aa with the exact same length. However, compared with IFI16-iso2, IFI16-iso1 lacked a fragment containing NLS motifs 2, 3, and 4. The key acetylation site K128 was located in NLS motif 2 of this fragment (Fig. 1B). Consistent with the analysis of amino acid composition, IFI16-iso2 mainly located in the nucleus, whereas IFI16-iso1 located in the cytoplasm, which was similar to the IFI16-iso2-K128Q mutant (Fig. 1C; Supplemental Fig. 1A).

The difference of subcellular locations suggests that IFI16-iso1 and IFI16-iso2 may have unique functions during DNA viral infection. To prove this hypothesis, we tested two DNA viruses, HSV-1 and VACV, which replicate in the nucleus and cytoplasm, respectively. Interestingly, cytoplasmic IFI16-iso1 formed typical speckles (white arrows, Supplemental Fig. 1A) in the VACV-infected cells, which were similar to the filamentous structure or oligomerization complexes formed in restriction of herpesvirus (33). Multiple cell lines were used to discover the endogenous IFI16-iso1. PCR results showed that IFI16-iso1 was ubiquitously expressed in colorectal cancer cells (e.g., HT-29, SW480, and SW620) and gastric cancer cells (e.g., MKN45 and HGC27) (Fig. 1D). Infection with HSV-1 or VACV greatly induced IFI16-iso1 expression in SW620 or SW480 cells (Fig. 1E). A549 cells, in which the basal levels of IFI16 and STING signals were found to be low but detectible (Supplemental Fig. 1B), were used in the exogenous expression experiment to identify the functions of IFI16-iso1 and IFI16-iso2 in sensing different viral DNA during host antiviral immunity. IFI16-iso1 formed complexes with the downstream adaptor STING in the cytoplasmic compartment, whereas IFI16-iso2 formed complexes with STING mainly in the cytoplasmic compartment and less in the nuclear region (Supplemental Fig. 1C). Together, these results describe the different subcellular locations of IFI16-iso1 and IFI16-iso2 and indicate the differential roles of IFI16 isoforms in sensing viral DNA during host antiviral immunity.

To distinguish the roles of IFI16-iso1 and IFI16-iso2 in recognizing viral DNA and triggering IFN-I–dependent antiviral immunity, potent agonists are required for stimulating both cytoplasmic and nuclear DNA sensors. However, there was no established agonist to activate the nuclear DNA sensors such as IFI16-iso2 and hnRNPA2B1. Therefore, we tried to develop novel nucleus-localized viral DNA mimics by adding a hexanucleotide motif 5′-AGUGUU-3′, which was reported to direct miR-29b nucleus import (34). We synthesized and annealed several viral dsDNA mimics with the 3′-terminal hexanucleotide element 5′-AGUGUU-3′ (RNA form of nuclear localization sequence [RNLS]) or 5′-AGTGTT-3′ (DNA form of nuclear localization sequence [DNLS]) (Supplemental Fig. 2A). These viral DNA mimics were labeled with Cy5 fluorescent dye at the 5′ or 3′ terminus (Supplemental Fig. 2A). Both of the Cy5-labeled HSV60mer and HSV60-RNLS were detectable in the cytoplasm of A549 cells. However, HSV60-RNLS rather than HSV60mer was gradually enriched in the nucleus 2 h after transfection, and much more HSV60-RNLS than HSV60mer was detected in the nucleus after 4-h or 8-h transfection (Fig. 2A, 2B). In addition to A549 cells, we checked the subcellular location of the transfected HSV60mer, HSV60-RNLS, and HSV60-DNLS in the human keratinocyte cell line HaCaT. Much more HSV60-RNLS and HSV60-DNLS were observed in the nucleus than HSV60mer (Fig. 2C, 2D), which suggests that both hexanucleotide RNLS and DNLS can direct HSV60mer nuclear import. We tested another two viral DNA mimics: YG3 (26 bp) and VACV70mer (70 bp). More YG3-RNLS, YG3-DNLS, and VACV70-DNLS were detected in the nucleus than their respective controls (Fig. 2E, 2F; Supplemental Fig. 2B, 2C). These results indicate that RNLS and DNLS are able to effectively direct the nucleus translocation of viral DNA mimics with lengths ranging from 26 bp to 70 bp.

Because the tagged nucleus-located viral DNA mimics had additional hexanucleotide elements at the 3′ terminus, it was unknown whether these tails affected the function of the viral DNA mimics in activating IFN-I signaling. In A549 cells, both HSV60-RNLS and HSV60-DNLS triggered the transcription of IFNB1, ISG56, and OAS2 very well, even more robustly than the untagged HSV60mer did (Fig. 3A). Similarly, VACV70-DNLS robustly triggered the transcription of IFNB1, ISG56, and OAS2 as the untagged VACV70mer did (Supplemental Fig. 2D).

To exclude that extra hexanucleotide made the HSV60mer more potent in induction of IFN-I, HSV60-NC, a control viral mimic tagged with mutated DNLS hexanucleotide elements (5′-TCTCAT-3′) were included in the following experiments (34). Compared with the HSV60mer, HSV60-DNLS triggered more IFNB1 as well as the expression of IFN-β downstream ISGs such as ISG56 and OAS2 in A549 cells. However, the HSV60-NC did not exhibit any additional effect as HSV60-DNLS did in inducing IFN-β and ISGs (Fig. 3B). Similar phenomena were observed in the BMDMs (Fig. 3C). Moreover, transfection with HSV60-DNLS drove more IFN-β transcription activity than HSV60mer and HSV60-NC did in the IFN-β promoter reporter assay (Fig. 3D).

hnRNPA2B1 has been identified as a nuclear DNA sensor that initiates and amplifies innate antiviral immunity against DNA viral infection (15). We tested the function of HSV60-DNLS in activating hnRNPA2B1-dependent antiviral immune responses. Overexpression of hnRNA2B1 induced IFNB1, ISG56, and OAS2 genes robustly in A549 cells. HSV60-DNLS, but not HSV60mer or HSV60-NC, facilitated hnRNA2B1-induced IFNB1 and ISGs in these cells (Fig. 3E). HSV60-DNLS drove more Ifnb1, Isg15, and Oas2 transcription than HSV60mer or HSV60-NC in the WT L929 cells but not in the Hnrnpa2b1^−/− cells (Fig. 3F, 3G). IFN-β ELISA results further confirmed that HSV60-DNLS was a potent agonist for the nuclear DNA sensor hnRNPA2B1 (Fig. 3H, 3I).

Taking these results together, we have developed nucleus-located viral DNA mimics that could be used as potent agonists for nuclear DNA sensors such as hnRNPA2B1 and IFI16-iso2.

We next tested the ability of IFI16-iso1 and IFI16-iso2 in recognizing cytoplasmic and nuclear viral DNA mimics. Colocalization analysis by confocal technology showed that IFI16-iso1 recognized HSV60mer and the cytoplasmic portion of HSV60-DNLS (Fig. 4A). Almost no colocalization of IFI16-iso2 with HSV60mer could be detected in the nucleus, whereas IFI16-iso2 and HSV60-DNLS colocalized very well in the nucleus (Fig. 4A). Nucleus-replicating HSV-1 and cytoplasm-replicating VACV were used to further identify the ability of IFI16-iso1 and IFI16-iso2 in viral DNA recognition. In order to detect the complexes of viral DNA with its sensor in situ, the viral genomic DNA was labeled by BrdU for the in situ PLA (Supplemental Fig. 3A). IFI16-iso1 preferably recognized the cytoplasmic VACV and the cytoplasmic portion of HSV-1, whereas the IFI16-iso2 preferably interacted with the nuclear HSV-1 DNA (Fig. 4B–4E; Supplemental Fig. 3B). These results suggest differential roles of IFI16-iso1 and IFI16-iso2 in recognizing DNAs from cytoplasmic and nuclear viruses.

Cytoplasm-located HSV60mer and nucleus-located HSV60-DNLS were used to compare the functions of IFI16-iso1 and IFI16-iso2 in triggering IFN-β production and activating IFN-α/β receptor downstream signaling. IFI16-iso2 triggered more transcripts of IFNB1 and IFN-β downstream ISG56 and OAS2 in the HSV60-DNLS–transfected A549 cells than IFI16-iso1 did (Fig. 5A). By using the IFN-β promoter reporter, we found that HSV60mer induced more IFN-β transcription than HSV60-DNLS in the IFI16-iso1– and STING-overexpressed HEK293T cells, whereas HSV60-DNLS induced more IFN-β transcription than HSV60mer in the IFI16-iso2– and STING-overexpressed cells (Fig. 5B). Consistently, HSV60mer triggered more phosphorylation of TBK1 and STAT1 and induced more OAS1 protein expression in the IFI16-iso1-overexpressed A549 cells than in the IFI16-iso2–overexpressed cells (Fig. 5C). However, HSV60-DNLS triggered more phosphorylation of TBK1 and STAT1 and induced more OAS1 protein expression in the IFI16-iso2–overexpressed A549 cells than in the IFI16-iso1–overexpressed cells (Fig. 5C). The functional difference between IFI16-iso1 and IFI16-iso2 in the regulation of poly(I:C)-triggered IFN-I signals was not so obvious (Supplemental Fig. 3C). These results show the differential effects of IFI16-iso1 and IFI16-iso2 in viral DNA sensing and IFN-I signaling activation due to their different subcellular locations. Besides, both of IFI16-iso1 and IFI16-iso2 sustained the activation of the DNA-responsive inflammasome (Supplemental Fig. 3D, 3E).

VACV and HSV-1 were taken to distinguish the roles of IFI16-iso1 and IFI16-iso2 in initiating and amplifying host antiviral immunity. In the VACV-infected A549 cells, overexpression of IFI16-iso1 induced more IFNB1, ISG56, and OAS2 than IFI16-iso2 did (Fig. 6A). However, in the HSV-1–infected cells, IFI16-iso2 exhibited stronger ability in the induction of IFN-β and downstream ISGs (Fig. 6A). An IFN-β promoter reporter assay and an IFN-β ELISA further demonstrated the differential functions of IFI16-iso1 and IFI16-iso2 in activating host antiviral immunity during cytoplasmic VACV and nuclear HSV-1 infection (Fig. 6B, 6C). In addition, VACV infection triggered more activation of STING and IFN-I signaling in the IFI16-iso1–overexpressed cells than in the IFI16-iso2–overexpressed cells (left panel of (Fig. 6D; Supplemental Fig. 4A). However, HSV-1 infection triggered more activation of STING and IFN-I signaling in the IFI16-iso2–overexpressed cells than in the IFI16-iso1–overexpressed cells (right panel of (Fig. 6D; Supplemental Fig. 4A). However, infection with Sendai virus, which replicates in the cytoplasm, did not trigger additional improvement of IFN-I signaling in the IFI16-iso1–expressed cells (Supplemental Fig. 4B). These results collectively indicate that IFI16-iso1 is more important in sensing cytoplasmic DNA virus, whereas IFI16-iso2 is more critical in sensing nuclear DNA virus. Consistently, the VACV viral particles in cell culture supernatant were much less frequent in IFI16-iso1–overexpressed cells than in the IFI16-iso2–overexpressed cells. However, HSV-1 viral particles in cell culture supernatant were restrained without statistical significance in the IFI16-iso2–overexpressed and the IFI16-iso1–overexpressed cells (Fig. 6E).

Alternative splicing (AS) is an essential mechanism for generating functional diversity, which allows individual genes to express multiple mRNAs and encode numerous proteins (35). It has been estimated that 95% of mammalian genes undergo AS, with strong impact on multiple regulatory processes, including chromatin modification and signal transduction (35). The AS of the IFI16 gene was identified three decades ago (18). Initially, three isoforms were cloned and verified by Sanger sequencing, and three protein bands are usually detected by immunoblotting. However, at least four variants of the IFI16 transcripts (NCBI reference sequences NM_001206567.2, NM_005531.3, NM_001376587.1, and NM_001376592.1; https://www.ncbi.nlm.nih.gov/nuccore/) have been deposited in the NCBI GenBank database. Moreover, IFI16-β, a novel IFI16 isoform that contains two HIN domains but lacks the PYD domain, has recently been discovered in various human tissues and cells (36). IFI16-β colocalizes with AIM2 in the cytoplasm and inhibits AIM2 inflammasome activation (36), whereas IFI16-iso2 plays important roles in inducing host antiviral immunity and p53-dependent apoptosis (8, 19). Here, we have shown that IFI16-iso1, with the same amino acid length as IFI16-iso2, mainly localizes in the cytoplasm and preferentially recognizes the cytoplasmic viral DNA. Given the distinct subcellular localization by comparison with the cytoplasm-nucleus shuttling of IFI16-iso2, IFI16-iso1 may have unique functions in sensing viral DNA and triggering antiviral immunity against cytoplasmic DNA viruses.

Our in situ viral DNA recognition experiment has shown that IFI16-iso1 recognizes cytoplasmic VACV more effectively, whereas IFI16-iso2 recognizes nuclear HSV-1 more effectively. IFI16-iso1 is more potent than IFI16-iso2 to induce IFN-I and ISG expression in the cytoplasmic viral DNA-stimulated cells. It also triggers greater antiviral immunity against cytoplasm-replicating VACV than IFI16-iso2 does. Although acetylation of IFI16-iso2 can translocate from the nucleus to the cytoplasm (16, 17), IFI16-iso2 plays a more dominant role in response to the nuclear HSV60-DNLS than the cytoplasmic HSV60mer. IFI16-iso2 and IFI16-β are ubiquitous in human tissues and cells (13, 14, 36, 37). Similarly, IFI16-iso1 is ubiquitously expressed in human cancer cells from different tissues and can be effectively induced by HSV-1 or VACV infection. It will be interesting to know how to control the AS of the IFI16 gene to produce differential IFI16 transcript variants during host antiviral immunity. According to the source sequences of IFI16-iso1 that have been recorded in the NCBI databases (accession numbers AK296228, BM806453, DA945454, DB273519, and M63838), this isoform is expressed in human thalamus, melanotic melanoma, spleen, uterus, and CTL/NK cells (21, 38). Further studies are required to identify the human tissues that express IFI16-iso1 and to investigate the unique function of IFI16-iso1 by RNA interference.

Viral DNA mimics such as B-DNA, Z-DNA, ISD45mer, and HSV60mer have been widely used to discover DNA sensors or study cell signaling transduction. In addition to potently inducing IFN-I production and activating antiviral immunity, viral DNA mimics have their advantages over the DNA viruses in investigating cell signaling and host immune response. For example, a single component of the viral DNA mimics excludes the effects of other viral components such as proteins and metabolites in the regulation of host antiviral immunity. Viral DNA mimics are much easier than viral genomes to edit, modify, and label. Hence, numerous viral DNA motifs that can induce IFN-I production have been identified, and the relative viral DNA mimics, including G3-YSD, GHV50mer, HSV60mer, and VACV70mer, have been developed (8, 27, 28). However, transfection of these viral DNA mimics via regular techniques such as liposome or cationic lipid transfection leads to viral DNA mimic cytoplasmic accumulation and cytosolic DNA sensor activation. Numerous DNA viruses, such as HSV-1, KSHV, and hepatitis B virus, enter the nucleus after infecting host cells (9, 11, 16). Nuclear DNA sensors, including IFI16-iso2 and hnRNPA2B1, are discovered to recognize nuclear viral DNA. To efficiently activate nuclear DNA sensors through the viral recognition motifs to elicit IFN-I, we have developed nuclear viral DNA mimics by adding a hexanucleotide motif RNLS or DNLS into the well-known cytoplasm-located viral DNA mimics. These novel dsDNA agonists preferentially activate nuclear DNA sensors and induce IFI16-iso2/hnRNPA2B1-dependent IFN-I production.

In summary, our study has provided a series of dsDNA agonists to mimic the nucleus-replicating viruses, which will be useful to study the emerging nuclear DNA sensors. By comparing the sequence and function of the transcript variants of IFI16, we also have demonstrated that IFI16-iso1, an isoform of IFI16 without any functional study, mainly locates in the cytoplasm and preferentially detects cytoplasmic DNA viruses.

We appreciate the excellent technical support provided by the RNA technology platform of Suzhou Institute of Systems Medicine.

The authors have no financial conflicts of interest.