Characterization of cGAS homologs in innate and adaptive mucosal immunities in zebrafish gives evolutionary insights into cGAS-STING pathway

Zhi-fei Liu1 | Jian-fei Ji1 | Xiao-feng Jiang1 | Tong Shao1 | Dong-dong Fan1 |
Xin-hang Jiang1 | Ai-fu Lin1 | Li-xin Xiang1 | Jian-zhong Shao1,2
1Key Laboratory for Cell and Gene Engineering of Zhejiang Province, College of Life Sciences, Zhejiang University, Hangzhou, People’s Republic of China
2Laboratory for Marine Biology and Biotechnology, Qingdao National Laboratory for Marine Science and Technology, Qingdao, People’s Republic of China

Correspondence

Jian-zhong Shao, Key Laboratory for Cell and Gene Engineering of Zhejiang Province, College of Life Sciences, Zhejiang University, 866 YuHangTang Road, Hangzhou 310058, China.
Email: [email protected]

Funding information
National Natural Science Foundation of China, Grant/Award Number: 31572641 and 31630083; the National Key Research and Development Program of China, Grant/Award Number: 2018YFD0900503 and 2018YFD0900505; Stem Cell and Translational Research, the National Key
Research and Development Program of China, Grant/Award Number: 2016YFA0101001;
the Open Fund of the Laboratory for Marine Biology and Biotechnology, Qingdao National Laboratory for Marine Science
and Technology, Qingdao, China, Grant/ Award Number: OF2017NO02; the Open Funding Project of the State Key Laboratory of Bioreactor Engineering and the Zhejiang Major Special Program of Breeding, Grant/ Award Number: 2016C02055-4

Abbreviations: A h, Aeromonas hydrophila; CDNs, cyclic dinucleotides; cGAMP, cyclic GMP-AMP; cGAS, cyclic GMP-AMP synthase; Co-IP,
co-immunoprecipitation; CTT, C-terminal tail; DmcGAS, Drosophila melanogaster cGAS; DncV, dinucleotide cyclase in Vibrio cholera; dpi, days post infection; DrcGASa, Danio rerio cGASa; DrcGASb, Danio rerio cGASb; ESI-LC-MS, electrospray ionization liquid chromatography mass spectrometry; FCM, flow cytometry; GALT, gill-associated lymphoid tissue; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HKLs, head kidney leukocytes; hpf, hours post fertilization; hpi, hours post infection; HscGAS, Homo sapiens cGAS; HsSTING, Homo sapiens STING; IFN-I, type I IFN; IgT, immunoglobulin T; IgZ, immunoglobulin Z; IKK, IκB kinase; IRF-3/7, interferon regulatory factor 3/7; LVs, lentiviruses; Mab21, male abnormal 21; MALT, mucosa- associated lymphoid tissue; MmcGAS, Mus musculus cGAS; MOs, morpholino oligonucleotides; NMP, nucleoside monophosphate; ns, not significant; NTase, nucleotidyltransferase; NTP, nucleoside triphosphate; NvcGAS, Nematostella vectensis cGAS; NvSTING, Nematostella vectensis STING; OAS1,
2′-5′-oligoadenylate synthetase; PBLs, peripheral blood leukocytes; STING, stimulator of IFN genes protein; TBK1, TANK-binding kinase 1; US, unstimulated.
© 2020 Federation of American Societies for Experimental Biology

The FASEB Journal. 2020;00:1–24.

wileyonlinelibrary.com/journal/fsb2 | 1

1 | INTRODUCTION

Cyclic GMP-AMP Synthase (cGAS) is one of the best- characterized cytoplasmic DNA sensors in humans and other mammals due to its critical roles in innate immunity and var- ious diseases.1-4 cGAS is a member of the MAB21 family, which belongs to the nucleotidyltransferase (NTase) fold pro- tein superfamily.5 Mammalian cGAS contains a long unstruc- tured N-terminal region, an NTase domain that overlaps with a C-terminal male abnormal 21 (Mab21) domain, and three DNA-binding surfaces.6,7 Upon direct interaction with DNA, cGAS is activated by forming a 2:2 cGAS-DNA complex, leading to the synthesis of cyclic GMP-AMP (cGAMP) from ATP and GTP.8,9 The mammalian cGAS-derived cGAMP is a noncanonical 2′3′-cGAMP isomer that contains one phosphodiester bond between the 2′-OH of GMP and the 5′-phosphate of AMP and another between the 3′-OH of AMP and the 5′-phosphate of GMP.10,11 This 2′3′-cGAMP serves as a second messenger that binds to the endoplasmic reticu- lum resident receptor protein stimulator of interferon genes (STING) and induces a conformational change in STING; such change is important for the phosphorylation of IRF3 or IKK by TBK1 and the subsequent activation of the down- stream type I IFN (IFN-I) and NF-κB signaling pathways.12,13 cGAS can be activated by a broad range of DNA viruses, certain retroviruses, bacterial pathogens, and self-originated DNA in a DNA sequence-independent but DNA length- dependent manner.14-16
Despite numerous investigations in mammals,15,17-20 the occurrence and functional correlation of cGAS and STING in nonmammalian species remain limited. Several previous studies preliminarily identified cGAS and STING homo- logs in some other vertebrates20,21 and ancient metazoans (Monosiga brevicollis, Nematostella vectensis, Ixodes scap- ularisnematode, and Drosophila melanogaster).7,22 These studies suggested that cGAS and STING originated from early metazoans at least 600 million years ago. However, functional investigations showed that the Nematostella vectensis (N vectensis) cGAS (NvcGAS) does not respond to DNA stimulation; the product of NvcGAS is a 3′3′-cGAMP isomer rather than a 2′3′-cGAMP.22 Drosophila melano- gaster (D melanogaster) cGAS is inactivated in response to Listeria monocytogenes and invertebrate iridescent virus 6 infection.23 Thus, cGAS in ancient metazoans may not act as a DNA sensor as mammalian cGAS does, whose enzy- matic activity is independent of the association with DNA.24 In contrast to the diverse performance of NvcGAS relative to

that of its mammalian counterparts, the STING of N vecten- sis (NvSTING) shares conservative structures with mam- malian STINGs and has the potential for 3′3′-cGAMP and 2′3′-cGAMP signaling.22 These findings imply the potential functional correlation between cGAS and STING that is connected by 3′3′-cGAMP in N vectensis. Hence, an ances- tral cGAS-STING signaling pathway possibly originated in primitive metazoan species. STINGs in many ancient organ- isms preferentially activate NF-κB but not IFN-I signaling, whereas mammalian STINGs predominantly contribute to IFN-I signaling.22-24 This phenomenon might be attributed to the absence of an IFN-I system from ancient organisms and the lack of a C-terminal tail (CTT) domain in ances- tral STINGs; this domain is essential for the recruitment of TBK1 and IRF3 for IFN-I induction.25 Thus, the ances- tral functions of the cGAS-STING pathway likely include the activation of innate antibacterial immunity through the NF-κB pathway but not that of antiviral immunity via the IFN-I pathway. The functional role and mechanism of the cGAS-STING pathway may undergo an evolutionary shift from ancient metazoans to modern vertebrates, which in- cluding the conversion of the second messenger from 3′3′- cGAMP to 2′3′-cGAMP for strong activation upon STING that is advantageous for IFN-I induction, change in the catalytic mechanism of cGAS for cGAMP synthesis from 3′3′-cGAMP to 2′3′-cGAMP, and acquisition of new func- tional domains (eg, CTT) by STING for adaptation to the coevolution of the main components of the IFN-I signaling pathway, such as TBK1 and IRF3/IRF7. These issues should be clarified through systematic characterization of cGAS and STING from low metazoans, including invertebrates, to mammals. To this end, teleost fish, which is an important evolutionary link between invertebrates and vertebrates, be- comes an integral part. In the present study, we provide mo- lecular and functional supports for these hypotheses using zebrafish (Danio rerio) as a model.

2 | MATERIALS AND METHODS
2.1 | Experimental fish and embryo
Wild-type AB zebrafish were maintained in a circulat- ing water bath at 28°C under standard laboratory con- ditions.26 The fish used in adoptive transfer assay were siblings generated after at least two generations of in- breeding. Zebrafish embryos were prepared as previouslydescribed.27 Animal experiments were conducted in accordance with the guiding principles and approved by a local ethics committee.

2.2 | Molecular cloning
The total RNA of Danio rerio cGASa/b (DrcGASa/b) was extracted from zebrafish gill tissues using RNAiso Plus kit (Takara, Dalian, Liaoning, China). DrcGASa/b cDNAs were amplified by RT-PCR using primers shown in Table S1. cDNA products were purified, inserted into pGEM-T vec- tor, transformed into E coli TOP 10 cells and sequenced fol- lowing the previously described protocols.28 Bioinformatics analysis, including gene and protein structures, multiple amino acid sequence alignment, and phylogenetic tree was performed using previously described databases and soft- ware programs.29,30

2.3 | Plasmid constructions
The eukaryotic expression vectors for DrcGASa/b proteins with EGFP, Myc, Flag, and HA tags were constructed using pEGFP-N1, pCMV-N-myc, pcDNA3.1-EGFP-Flag-His, and pcDNA3.1-EGFP-HA-His plasmids.31 The point mutants of DrcGASa/b prepared using a Quick Mutation Site-Directed Mutagenesis Kit (Beyotime, Shanghai, China) were con- structed into pcDNA3.1-EGFP-Flag-His and pcDNA3.1- EGFP-HA-His. DrSTING encoding sequence was amplified from pcDNA1.1-STING donated by Ste´phane Biacchesi and constructed into pCMV-N-Myc for the expression of DrSTING-Myc fusion protein.32 Human and mouse cGAS expression plasmids were donated by Zhijian J. Chen.12 Expression constructs for human wild-type and mutant (R232H) STINGs and Vibrio cholera (V cholera) DncV were donated by Philip J. Kranzusch.22 Zebrafish IFNφ1 expres- sion plasmid was a gift from Victoriano Mulero.33 Human IFN-β and NF-κB luciferase reporter constructs were pur- chased from Clontech. pRL-TK renilla luciferase reporter vector was purchased from Promega. Zebrafish IFNφ1/3 reporter plasmids were produced in our laboratory.34 The primers used for construct generation are listed in Table S1. Plasmids for transfection and microinjection were prepared free of endotoxin using an EZNA Plasmid Mini Kit (Omega Bio-tek, Doraville, GA, USA).

2.4 | Real-time PCR
Real-time PCR was carried out with SYBRR Premix Ex Taq kit (Takara, Dalian, Liaoning, China) using an Eppendorf Realplex Mastercycler instrument (Eppendorf, Mittelsachsen, Saxony, Germany). PCR thermal profile consisted of an ini- tial activation of 2 minutes at 95°C, followed by 40 cycles of amplification at 95°C for 30 seconds and 60°C for 20 sec- onds. Melt-curve analysis (95°C for 15 seconds, 55°C for 1 minute, increase of 0.5°C per 5 seconds until 95°C) was performed to ensure the specificity of the amplification. The 2−ΔΔCt method was used for relative gene expression analy- sis. mRNA abundance was normalized against that of β-actin. The primers used are shown in Table S1. All PCR reactions were performed in triplicate and repeated at least three times.

2.5 | Subcellular localization
HEK293T cells were seeded into multiwell plates and cultured in DMEM supplemented with 10% FBS at 37°C in 5% CO2 to allow growth into 70%-80% confluence. The cells were tran- siently transfected with DrcGASa-EGFP or DrcGASb-EGFP plasmid DNA (1-2 μg) combined with FuGENE HD trans- fection reagent (Promega, Madison, Wis., USA).30,35 After transfection for 48 hours, the cells were fixed in 4% para- formaldehyde for 10 minutes and then stained with 10 μM DiI (Beyotime, Shanghai, China) and 100 ng/mL of DAPI (Sigma-Aldrich, St. Louis., MO, USA) at 37°C for 5 minutes. The cells were also stained with ER-Tracker Red and Golgi- Tracker Red (Beyotime, Shanghai, China) according to the manufacturer’s instructions. Images were captured using a two-photon laser scanning confocal microscope (LSM 710, Carl Zeiss, Oberkochen, Baden-Württemberg, Germany).35

2.6 | DrcGASa/b product identification
DrGASa/b catalytic cGAMP products were prepared from HEK293T cells or zebrafish embryos at 48 hours post transfection or microinjection of the DrGASa/b-expression plasmids, Aeromonas hydrophila (A hydrophila) DNA, or morpholino oligonucleotides. After purification with a phenol-chloroform extraction,10 the samples were collected for electrospray ionization liquid chromatography mass spectrometry (ESI-LC-MS) analysis using an Agilent 1200 LC system (Agilent, Pal Alto, CA, USA) coupled with an IonTrap mass spectrometer (LCQ Deca XP MAX, Thermo Finnigan, San Jose, CA, USA) equipped with an electrospray source operating in positive ionization mode under 350°C capillary temperature, 46 V capillary voltage, 65 arb sheath gas flow rate, 10 arb aux/sweep gas flow rate, and 3.5 kV spray voltage. All samples were chromatographed on Agilent Extend-C18 columns (2.1 × 100 mm with 3.5 μm particle size or 4.6 × 150 mm with 5 μm particle size) at 35°C. The analytes were separated by a gradient elution by mixing sol- vent A (10 mM/20 mM TEAB, pH of 8.5) and solvent B (100% MeOH) at different ratios during the chromatography at flow rate in the range of 0.2-1.0 mL/min. A S1 nuclease digestion assay was performed to identify 2′3′-cGAMP or 3′3′-cGAMP in the analytes, and the digestates were exam- ined by thin-layer chromatography and HPLC according to the previously described protocols.10 The chemically synthe- sized 2′3′-cGAMP and 3′3′-cGAMP (InvivoGen, San Diego, CA, USA) were used as standard controls.

2.7 | Preparation of polyclonal antibodies
Two epitope peptides with amino acids CHPNKHPLDKFLN (for DrcGASa) and KNMGGKYADLETPFPSRC (for IFNφ1) were predicted from the ectodomain of DrcGASa and IFNφ1 proteins as previously described.36 These epitope peptides were chemically synthesized and conjugated with ovalbumin (OVA, Bankpeptide, Hefei, Anhui, China). Six- week-old male BALB/c mice (~25 g) were immunized with the peptides (0.1 mg) in complete Freund’s adjuvant (Sigma- Aldrich, St. Louis., MO, USA) on Day 1 and 3 and in in- complete Freund’s adjuvant (Sigma-Aldrich, St. Louis., MO, USA) on Day 28 and 35. At 1 week after the final immu- nization, serum samples were collected when the Ab titers were above 1:10,000 as determined by ELISA. Abs were af- finity purified from the sera into IgG isotypes using protein A-agarose column (Thermo Fisher Scientific, Waltham, MA, USA). The validity and specificity of the Abs were deter- mined by Western blot. Abs against zebrafish immunoglobu- lin Z (IgZ) and immunoglobulin Z2 (IgZ2) and CD40 were produced in our previous studies.37,38

2.8 | Western blot analysis
Cells or tissues were treated with cell lysis or RIPA buffer (Beyotime, Shanghai, China) containing protease inhibitor mixture (Roche, Basel, Basel-Stadt, Swiss). Proteins were separated by 12% SDS-PAGE under reducing conditions and transfected onto PVDF membranes (Millipore, Boston, MA, USA). After blocking with 5% skimmed milk, the mem- branes were incubated with corresponding Abs followed by a secondary HRP-conjugated anti-rabbit/mouse IgG Ab (Abcam, Cambridge, Cambridgeshire, UK). Immunoreactive proteins were visualized with ECL reagents (GE Healthcare, Pittsburgh, PA, USA) by a digital gel image analysis system (Tanon 4500, Shanghai, China). The grayscale quantization of the protein stripe was performed by ImageJ software.

2.9 | Co-immunoprecipitation (Co-IP) assay
Co-immunoprecipitation was performed to detect the inter- action between DrcGASa and DrcGASb. DrcGASa and/

or DrcGASb expression plasmids were transfected into HEK293T cells. At 24 hours post transfection, cells were lysed with cold cell lysis buffer (Beyotime, Shanghai, China) containing protease inhibitor mixture (Roche, Basel, Basel-Stadt, Swiss). Lysates were centrifuged at 10 000 g for 15 minutes at 4°C and the supernatants were incubated with tag antibody or unrelated rabbit/mouse IgG (negative control) at 4°C overnight and then incubated with 50 μL of protein A-agarose beads (Roche, Basel, Basel-Stadt, Swiss) for 4 hours. The beads were washed five times with lysis buffer, and eluted with loading buffer by boiling for 5 min- utes at 95°C. Samples were analyzed by 12% SDS-PAGE and Western blot assay as described above.

2.10 | Morpholino oligonucleotides (MOs)
Morpholino oligonucleotides against DrcGASa mRNA (DrcGASaMO: 5′-CATGATGGCAGGTGTGTTCACG AGC-3′) and DrcGASb mRNA (DrcGASbMO: 5′-GTTCT TGTTCGACTTTCCATGATGG-3′) and standard control MO (ControlMO: 5′-CTCTTACCTCAGTTACAATTTA
TA-3′) were designed and synthesized using Gene Tools (Corvallis, Oregon, USA). To test the efficiency of these MOs, DrcGASa/b cDNA fragments containing partial 5′–untranslated regions and the beginning parts of ORF were amplified and cloned into pEGFP-N1 to generate DrcGASa-EGFP and DrcGASb-EGFP vectors. One-cell stage embryos of zebrafish were injected with DrcGASa- EGFP or DrcGASb-EGFP vector together with ControlMO, DrcGASaMO or DrcGASbMO (4 ng/embryo). Embryos were collected 24 hours after microinjection and GFP fluores- cence was visualized via fluorescence microscopy (Axiovert
40 CFL; Carl Zeiss, Oberkochen, Baden-Württemberg, Germany). MOs against DrSTING mRNA (DrSTINGMO) was used, and its efficiency was tested in a previous study.34

2.11 | Generation of short hairpin RNA (shRNA) encoding lentivirus (LVs)
Generation of short hairpin RNAs carrying the small inter- fering RNAs (siRNAs) targeting DrcGASa/b and DrSTING mRNAs and shRNA-encoding LVs were designed and pro- duced as previously described.26,27 Briefly, shRNAs were initially constructed into a pSUPER vector for efficiency evaluation. Then the effective shRNAs were reconstructed into a lentiviral pLB vector. The shRNA-encoding LVs were generated by cotransfecting HEK293T cells with pLB and packaging vectors. Viral supernatant was concentrated by ultracentrifugation (25 000 rpm, 90 minutes, 4°C). Viral titers were detected through EGFP signature in HEK293T cells under fluorescent microscope or by flow cytometry

(FCM) analysis. The silencing activity of the resulting LVs was determined in zebrafish in peripheral blood leukocytes/ head kidney leukocytes (PBLs/HKLs), and gill/skin tis- sues by real-time PCR after fish were i.p. injected with the LVs (2 × 105 TU/fish) once every 24 hours for three times. Scrambled shRNA-encoding LVs was administered as a neg- ative control.

2.12 | Involvement of DrcGASa/b in type I IFN and NF-κB activation
HEK293T cells were transfected with DrcGASa/ b/DrSTING expression plasmids and IFN-β or NF-κB luciferase reporter construct (200 ng/mL; Clontech, San Francisco, CA, USA) and pGL-TK Renilla luciferase inter- nal control (20 ng/mL; Clontech, San Francisco, CA, USA). After 24 hours, the relative luciferase activity unit was de- termined using a Dual-Luciferase reporter assay system (Promega, Madison, Wis., USA) as previously described.34 Meanwhile, the mRNA level of CCL20 was examined by real-time PCR. One-cell stage embryo was injected with DrcGASa/b/DrSTING expression plasmids (60 pg), DrSTINGMO (4 ng), IFNφ1/IFNφ3, or NF-κB luciferase reporter constructs (50 pg) and pGL-TK Renilla luciferase internal control (5 pg) in different combinations with PBS (control) or A hydrophila DNA (200pg). After 24 hours, the relative luciferase activity unit was determined. The mRNA levels of IRF3, IFNφ1, IFNφ3, IL-6, IL-1β, and hepcidin from zebrafish embryo and head kidney were also examined by real-time PCR. For this procedure, one-cell stage embryo was injected with MOs for DrcGASa/b and DrSTING (4 ng) and PBS or A hydrophila DNA (200 pg) for 24 hours. Adult zebrafish were i.p. injected with shRNA-encoding LVs for DrcGASa/b/DrSTING knock- down, as described above. After the last LVs administra- tion, fish were stimulated with PBS or A hydrophila DNA (5 μg) for 12 hours.

2.13 | Involvement of cGAMP in type I IFN and NF-κB activation
HEK293T cells were transfected with DrSTING expression plasmid (300 ng/mL), IFN-β or NF-κB luciferase reporter construct (200 ng/mL) and pGL-TK Renilla luciferase inter- nal control (20 ng/mL). On the next day, 2′3′-cGAMP and 3′3′-cGAMP (2 μg/mL) were introduced into cells using Lipofectamine 2000 following the manufacturer’s instruc- tion (Invitrogen, Carlsbad, CA, USA). After treatment for 16 hours, the relative luciferase activity unit was determined as described above. Zebrafish adults and one-cell stage embryos were also stimulated with PBS (control) or 2′3′- cGAMP and 3′3′-cGAMP (5 μg/fish and 100 pg/embryo) for 12-24 hours; and the mRNA levels of IRF3, IFNφ1, IFNφ3, IL-6, IL-1β, and hepcidin from the embryo and head kidney were examined by real-time PCR.

2.14 | In vivo immunoprotection assay
For antibacterial assay, the DrcGASa/b/DrSTING mor- phornts and wild-type control embryos at 24 hours post ferti- lization (hpf) were exposed to A hydrophila or Edwardsiella tarda (E tarda, 1 × 108 CFU/mL).34,39 After immersion in the bacterial suspension for 5 hours, the mortality of each group was recorded at 12-72 hours post infection (hpi). For antiviral assay, the DrcGASa/b/DrSTING morphornts and control embryos were injected with DNA stimulant (50 pg/ embryo; A hydrophila origin) and EGFP-tagged vesicular stomatitis virus (VSV, 1 × 103 PFU/embryo) at one-cell stage.40 After treatment for 48 hours, the mortality of each group was recorded at 24-72 hpi. For 2′3′-cGAMP antibacte- rial assay, adult zebrafish were stimulated with 2′3′-cGAMP or PBS (control) for 12 hours, followed by i.p challenge with A hydrophila (2.1 × 105 CFU). Mortality in each group was monitored during the 5 day-period at 24 hours of interval and the relative survival rate (RSR) was calculated as previously described.34

2.15 | ELISA for IgZ/IgZ2 immunoglobins
The abundance of the IgZ/IgZ2 Abs in gill/skin mucus against E tarda was examined by ELISA. The mucus was dissolved in PBS (pH 7.2) containing 1× protease inhibitor cocktail (Roche, Basel, Basel-Stadt, Swiss), 1 mM phenyl- methylsulfonyl fluoride (PMSF, Sigma-Aldrich, St. Louis., MO, USA), 0.1 mg/mL of soybean trypsin inhibitor (Sigma- Aldrich, St. Louis., MO, USA) and 0.5% BSA (Sigma- Aldrich, St. Louis., MO, USA) and centrifuged at 400 g for 10 minutes.41 E tarda (1 × 108 CFU/mL) was added into mi- crotiter wells and incubated at 37°C overnight.42 After block- ing with 2% BSA for 2 hours, wells were incubated with diluted mucus samples for 1 hour at 37°C, followed by in- cubation with the rabbit anti-IgZ or anti-IgZ2 Ab for 2 hours at 37°C and HRP-conjugated goat anti-rabbit Ab (Abcam, Cambridge, Cambridgeshire, UK) for 1 hour at 37°C. Colors were developed by a tetramethylbenzidine (TMB) substrate, and the absorbance (A450) value was read by Synergy H1 hy- brid reader (Biotek Instruments, Winooski, VT, USA). Abs titer is defined as the highest mucus dilution at which the A450 ratio (A450 of postimmunization mucus/A450 of preim- munization mucus) is greater than 2.1.36

2.16 | Immunofluorescence staining
Cells were fixed with 2% paraformaldehyde for 10 minutes. After washing with PBS, cells were permeabilized with 0.1% Triton X-100, blocked with 5% normal goat serum, and in- cubated with primary Abs or unrelated control IgG and secondary Abs at 4°C for 1 hour. Gill tissues were fixed in 4% paraformaldehyde overnight, and the paraffin sections were prepared as described.31 After dewaxing the paraf- fin and retrieving Ags, the sections were blocked with 5% normal goat serum and incubated with primary Abs at 4°C overnight. After washing with PBS, the sections were per- meabilized with 0.1% Triton X-100, blocked with 5% normal goat serum, and incubated with primary Abs at 4°C over- night and secondary Abs (PE-conjugated goat anti-mouse Ab and FITC-conjugated goat anti-rabbit Ab) at 4°C for 1 hour. Additional staining with DAPI was performed before pho- tomicrography. Samples were photographed under a laser scanning confocal microscope (FV3000, Olympus, Tokyo, Japan).

2.17 | Magnetic cell sorting (MACS) for γδ T cells
Leukocytes were enriched from the peripheral blood, spleen, and head kidney through Ficoll-Hypaque (1.080 g/mL) cen- trifugation.43 The cell suspension was blocked with 5% nor- mal goat serum for 15 minutes at 10°C, incubated with anti-γ or anti-δ for 15 minutes at 10°C, washed with MACS buffer (PBS containing 2 mM EDTA and 0.5% BSA), and incu- bated for 15 minutes at 10°C with anti-IgG magnetic beads (MiltenyiBiotec, BergischGladbach, Nordrhein-Westfalen, Germany). The cell suspension was applied to a LS separa- tion column in accordance with the manufacturer’s instruc- tions. Positive cells were cultured in L-15 medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% FBS (Thermo Fisher Scientific, Waltham, MA, USA), 100 U/mL of penicillin, and 100 μg/mL of streptomycin at 28°C overnight to detach the magnetic beads. The purity of the sorted γδ T cells was detected through FCM analysis.

2.18 | FCM analysis
The cells were blocked with 5% normal goat serum for 1 hour at 4°C and incubated with the corresponding primary Abs, unrelated control IgG isotypes and secondary Abs (PE- conjugated goat anti-mouse Ab and FITC-conjugated goat anti-rabbit Ab) for 1 hour at 4°C. Fluorescence signals were determined using a FACScan flow cytometer (BD Bio- sci- ences, San Jose, CA, USA) at 488 nm. FCM analysis was based on forward/side scatter (FSC/SSC) characteristics and

PE/FITC-conjugated fluorescence with CellQuest program as previously described.44,45 At least 10,000 events were col- lected from the lymphocyte gate.

2.19 | Function of DrcGASa/b-DrSTING in mucosal immunity
The function of DrcGASa/b-DrSTING axis in mucosal immunity was evaluated by LVs-based DrcGASa/b and DrSTING knockdown in gill/skin tissues, as described above. At 2 days after treatment, fish were infected with E tarda (1 × 108 CFU/mL) by immersion exposure. At 1, 3, or 7 days after infection, gill/skin tissues and their mucus were col- lected for IgZ/IgZ2 examination at mRNA and protein levels by real-time PCR, Western blot or ELISA. Leukocytes were collected from gill tissues for examination of IgZ+CD40+ and IgZ2+CD40+ B cells by FCM, as previously described.44 Fish received scrambled shRNA-encoding LVs served as negative control in the assay.

2.20 | Function of IFNφ1 in mucosal immunity
Gill/skin tissues and their mucus were collected from DrcGASa/b and DrSTING knockdown fish for the examina- tion of IFNφ1 expression, IgZ/IgZ2 production, and IgZ+/ IgZ2+ CD40+ B cell activation by real-time PCR, Western blot/ELISA and FCM, respectively, at 1-7 days after the immersion infection of fish with E tarda (1 × 108 CFU/ mL) as described above. For IFNφ1 rescue assay, recombi- nant IFNφ1 was produced in HEK293T cells and collected from the cultural supernatant, as previously described.46 DrcGASa/DrSTING knockdown fish were i.p. injected with the IFNφ1-containing supernatant (10 μL), and then the IgZ/IgZ2 production and CD40+ B cell activation were examined in their gill/skin tissues and mucus. Fish received scrambled shRNA-encoding LVs and cultural supernatant from HEK293T cells transfected with the empty plasmid (pcDNA6-myc-His B) served as negative controls in the assay.

2.21 | Correlation of DrcGASa-DrSTING- IFNφ1 and γδ T cells in mucosal immunity
The correlation was evaluated by γδ T cell depletion and transfer assays in combination with the knockdown of DrcGASa/b or DrSTING. Fish were injected i.p. three times with anti-γ/anti-δ Abs or unrelated control IgG at a dose of 10 μg/fish.36 At 3 days after the last Ab administration, the percentage of the γδ T cells was examined by FCM. γδ Tcells were also magnetically sorted from the wild-type and DrcGASa/b or DrSTING knockdown fish that prestimulated with mock PBS (control) and E tarda (1 × 108 CFU/mL) by immersion infection for 3 days. After washing three times with L-15 medium, the cells were adjusted to 1 × 108 cell/ mL and injected i.p. into the recipient fish (10 μL). Then, the fish were challenged with E tarda (1 × 108 CFU/mL) by im- mersion infection. At 3 days after infection, gill/skin tissues and their mucus were collected for the examination of IgZ/ IgZ2 production and IgZ+/IgZ2+ CD40+ B cell activation as described above.

2.22 | Functional evaluation for γδ T cells in vitro
γδ T cells and leukocytes were isolated from the gills of fish without or with E tarda (1 × 108 CFU/mL) immersion in- fection by MACS and density gradient centrifugation using Percoll (Sigma-Aldrich, St. Louis., MO, USA) as described previously.30 Then 2 × 104 γδ T cells were added into each Transwell chamber (Corning, Corning, NY, USA), while loading leukocytes (1 × 104) per well in a 12-well plate. Cells were cultured in L-15 medium (Thermo Fisher Scientific, Waltham, MA, USA) containing 10% FBS (Thermo Fisher Scientific, Waltham, MA, USA) and 100 U/mL of penicillin and 100 μg/mL of streptomycin. Meanwhile, 1 μg/mL anti- IFNφ1 Ab or unrelated rabbit IgG was added into the culture to neutralize IFNφ1 activity. After 72 hours of coculture, the percentage changes in IgZ+CD40+ and IgZ2+CD40+ B cells in total leukocytes were analyzed by FCM, as described above. Leukocytes culturing alone without γδ T cells were examined as control.

2.23 | Statistical analysis
All data are presented as the mean ± SD of each group. Statistical differences between the means of the experimental groups were evaluated by a one-way ANOVA and two-tailed Student t test. Statistical significance was considered when P < .05 or P < .01. All experiments were replicated at least three times.

3 | RESULTS
3.1 | Characterization of DrcGASa and
DrcGASb homologs
Two homologous cGAS (ie, DrcGASa and DrcGASb) genes were retrieved from the zebrafish genome database using the Homo sapiens cGAS (HscGAS) gene sequence as a query.The DrcGASa/b gene comprises five/six exons and four/ five introns and is located on chromosome 13/19 within a 4212/5916 bp genomic fragment ( S1A,B). Genes ad- jacent to the HscGAS locus, such as DDX43, EEF1A1, and SLC17A5 are clustered around the DrcGASa gene, although their loci follow a reverse order. However, genes adjacent to the DrcGASb locus are different from those of DrcGASa. This difference suggests that the two genes have diverse ori- gins ( S1B). The cloned DrcGASa/b cDNA consists of 2500/1798 bp, including a 88/322 bp 5′-untranslated region (5′-UTR), a 1779/1374 bp ORF, and a 633/102 bp 3′-UTR
(GenBank accession no. XM_680019.5/XM_003200639.5). The DrcGASa/b protein contains 592/457 amino acids and shares similar domain structures to those of mamma- lian cGASs, including an N-terminus, a NTase domain, a Mab21 domain with a zinc ribbon motif, and a C-terminus. DrcGASa/b exhibits an overall conserved tertiary structure in comparison with mammalian cGASs ( S1C). The structure includes three DNA-binding surfaces, a dimer in- terface, and a cGAMP-catalyzing/binding pocket. The major conservative and critical residues in mammalian cGASs were identified in DrcGASa. These residues include K458, K481, and K485 on the primary DNA-binding surface; R298, K316, H395, K416, and K422 on the second DNA- binding surface; K348, R365, K366, K367, and K503 on the third DNA-binding surface; K416, K468, and E472 on the dimer interface; and G274, S275, E287, D289, D387, T273, K431, R450, S452, and Y510 in the cGAMP-catalyzing/ binding pocket. TheTRAF6-binding motif in HscGAS also existed in DrcGASa ( 1). Compared with DrcGASa, DrcGASb contains conserved critical residues in primary DNA-binding surface (K260, K283, and R287), dimer in- terface (K221, K270, and E274), third DNA-binding surface (K142, R149, R152, R168, and K169), and partially criti- cal residues on the second DNA-binding surface (R101 and K221) and cGAMP-catalyzing/binding pocket (G77, S78, E90, D92, D193, K236, R252, S254, and Y310). DrcGASb
is characterized by a short N-terminus and a long C-terminus with a potential transmembrane domain, but without TRAF6- binding motif ( 1). Phylogenic analysis showed that DrcGASa/b is grouped with their mammalian cGAS coun- terparts (S2).

3.2 | Subcellular localization and tissue/ embryo expression
Subcellular localization analysis showed that DrcGASa/b- EGFP fusion proteins were predominantly distributed in the cytoplasm of the HEK293T cells ( 2A,B). Although DrcGASb has a potential transmembrane domain at its C-terminus, it is not colocalized with the intracellular mem- branes of the endoplasmic reticulum and Golgi apparatus

1 Multiple alignment of DrcGASa and DrcGASb with HscGAS, Mus musculus cGAS (MmcGAS), NvcGAS, and D melanogaster cGAS (DmcGAS). Amino acid residues potentially involved in plasma membrane localization and a long DNA length preference in HscGAS are indicated by blue arrows and black arrows, respectively. The critical residues for cGAS structural switch and in cGAS dimer interface are indicated by black triangles above and under the sequence, respectively. The critical residues on the primary DNA-binding surface, second DNA-binding surface, and third DNA-binding surface are indicated by white, blue, and purple triangles, respectively. The critical residues for cGAMP synthesis and substrate binding are indicated by red and orange triangles. The zinc ribbon domain is encircled by a green box and the critical residues in

it are indicated by green triangles. The predicted traf6-binding site and transmembrane domain are encircled by an orange box and a blue box, respectively

2 Subcellular distribution analysis of DrcGASa and DrcGASb in HEK293T cells and Hela cells. A-D, HEK293T cells were transfected with the expression plasmid of DrcGASa-EGFP or DrcGASb-EGFP, fixed, and stained with the cell nuclei probe DAPI, cell membrane probe DiI, ER-tracker or Golgi tracker. E and F, Hela cells were transfected with the expression plasmid of DrcGASb-HA (E) or DrcGASb-Δ63- HA (F) and stained for DrcGASb by immunofluorescence staining. The nuclei were stained with DAPI. Scale bar, 5 μm( 2C,D). The subcellular localization of a truncated cGASb mutant with the deletion of 63 C-terminal residues (cGASb-Δ63) shifted from the cytoplasm to the nucleus (2E,F). DrcGASa/b mRNA was extensively detected in all examined tissues. The abundance of DrcGASa mRNA was the highest in the skin, gill, brain, and heart, followed by the spleen, intestine, and kidney. This expression pattern of DrcGASa is similar to that of DrcSTING, thereby indi- cating the functional correlation between the two molecules. DrcGASb was highly expressed in the skin and brain; moderately in the heart and muscle; and lowly in other tissues ( 3A). DrcGASa was constitutively expressed in zebrafish embryo from 1 hpf to 72 hpf and showed sig- nificant upregulation at 6-72 hpf ( 3B). The expres- sion pattern of DrcGASa in the embryo is similar to that of DrSTING, as it is in adult tissues. DrcGASb mRNA was moderately expressed at 1 hpf to 3 hpf and decreased signifi- cantly at 6-72 hpf

3 Spatiotemporal expression patterns of DrSTING, DrcGASa, and DrcGASb. A, Gene expression of DrSTING, DrcGASa, and DrcGASb in different tissues of adult zebrafish. Values represent mean ± SD (n = 3). B, Developmental expression analysis of DrSTING,

DrcGASa, and DrcGASb by qPCR at 1, 3, 6, 12, 24, 36, 48 hours, and 72 hours post fertilization (hpf). Data are representative of three independent experiments as mean ± SD

4 Identification of 2′3′-cGAMP produced by DrcGASa or DrcGASb and catalytic activity comparison between DrcGASa and DrcGASb. A, HPLC analysis of the DrcGASa enzymatic product and standard synthetic 2′3′-cGAMP and 3′3′-cGAMP digested by S1 nuclease. B, TLC analysis of the DrcGASa enzymatic product, standard synthetic 2′3′-cGAMP and 3′3′-cGAMP digested by S1 nuclease. 1: synthetic
2′3′-cGAMP, 2: synthetic 3′3′-cGAMP, 3: synthetic 2′3′-cGAMP cleaved by S1 nuclease, 4: synthetic 3′3′-cGAMP cleaved by S1 nuclease, 5: DrcGASa product, 6: DrcGASa product cleaved by S1 nuclease. The asterisk (*) indicates samples not cleaved by S1 nuclease; and the plus (+) indicates products of samples digested by S1 nuclease. C, Western blot analysis of the expressed DrcGASa and DrcGASb proteins in HEK293T cells. D, Comparison of 2′3′-cGAMP abundance between DrcGASa-expressed HEK293T cells and DrcGASb-expressed HEK293T cells under normalization of DrcGASa and DrcGASb protein levels. GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase. Data are representative of three independent experiments as mean ± SD (**P < .01)

5 Examination of the functional roles of DrcGASa and DrcGASb in the activation of IFN-I and NF-κB signaling pathways. A, Functional roles of DrcGASa and DrcGASb in the activation of IFN-I and NF-κB signaling pathways as determined by the IFN-β and NF- κB luciferase reporters in HEK293T cells and mRNA expression level of CCL20 via quantitative real-time PCR. The expression plasmids of DrcGASa, DrcGASb, DrSTING and empty vector (control), together with the IFN-β or NF-κB luciferase reporter vector were transfected into
HEK293T cells in different combinations. Ns indicates not significant. B, Functional roles of DrcGASa and DrcGASb in the activation of IFN-I and NF-κB signaling pathways as determined by the IFN-φ1, IFN-φ3, and NF-κB luciferase reporters in zebrafish embroys. The expression plasmids of DrcGASa, DrcGASb, DrSTING, and empty vector (control), together with the IFN-φ1, IFN-φ3, or NF-κB luciferase reporter vector were microinjected into embryos or morphants at 0 hpf under A h DNA stimulation (200 pg/embryo) and the luciferase activities were assayed at 24 hours after injection. Data are representative of three independent experiments as mean ± SD (*P < .05, **P < .01)

3.3 | Enzymatic activity of
DrcGASa and DrcGASb
By in vitro analysis, four ion peaks with the mass-to-charge ratios (m/z) of 675, 776, 877, and 978 were identified in fractions from HEK293T cells expressing DrcGASa, DrcGASb, HscGAS, and MmcGAS by ESI-LC-MS. The given m/z values of these ion peaks were equivalent to those of cGAMP (675), and cGAMP compounds with one or more triethylamine molecules, including Et3N·cGAMP (776), 2Et3N·cGAMP (877), and 3Et3N·cGAMP (978), which were also detected in the standard control 2′3′-cGAMP and 3′3′-cGAMP samples. The retention time of the DrcGASa/ b-catalyzed products is the same as that of standard 2′3′- cGAMP but differs from that of 3′3′-cGAMP; this result is comparable with that of HscGAS- and MmcGAS-derived products ( S3A,B). For confirmation, the DrcGASa product was purified through HPLC and digested by S1 nu- clease, which can hydrolyze internal 3′-5′ phosphodiester linkages. Thin-layer chromatography and HPLC analyses revealed that the DrcGASa product was cleaved into a linear dinucleotide, the cleavage pattern of the DrcGASa product is the same as that of standard 2′3′-cGAMP but differs from that of 3′3′-cGAMP in control groups, and the latter can be cleaved into mononucleotides ( 4A,B). These find- ings suggest that DrcGASa and DrcGASb have catalytic ac- tivities for producing 2′3′-cGAMP rather than 3′3′-cGAMP. The resulting cGAMP levels were semi quantified by measuring their peak areas of selected ion chromatograms. cGAMP abundance from DrcGASa was significantly higher than that from DrcGASb under the normalization of the two protein levels ( 4C,D). Through in vivo analysis, 2′3′- cGAMP was significantly induced in wild-type embryos by the stimulation of A hydrophila DNA at 48 hpf, wherein DrcGASa was expressed at high levels naturally; this induc- tion significantly decreased in DrcGASa morphants (S4A,B). These observations suggested that 2′3′-cGAMP production depended on DrcGASa in a DNA-regulated manner.

3.4 | Involvement of DrcGASa/b in IFN and NF-κB signaling pathways
In vitro functional assay showed that the exotic expres- sion of DrSTING in HEK293T cells with naturally mini- mal expression of cGAS and STING significantly induced (P < .05) the activation of IFN-β and NF-κB luciferase reporters in a dose-dependent manner. In contrast, IFN-β and NF-κB reporters were minimally activated in cells expressing only DrcGASa/b (5A). This result may be due to the absence of DrSTING in HEK293T cells. Thus, a DrSTING compensation assay was performed bycoexpressing DrcGASa/b and DrSTING in cells. The in- duced activation of IFN-β and NF-κB reporters was signifi- cantly (P < .01) strengthened as predicted. For confirmation, similar alteration of CCL20 expression was detected in DrcGASa/b/or DrSTING introduced cells, thereby reflect- ing the downstream event of the NF-κB signaling pathway ( 5A). In vivo examination showed that DrcGASa/b and/or DrSTING overexpression in embryos significantly enhanced (P < .01) the activation of zebrafish IFN-I (IFNφ1 and IFNφ3) and NF-κB reporters under A hydrophila DNA stimulation. Nevertheless, these enhancements were sig- nificantly impaired along with the MOs-based knockdown of DrSTING. The DrcGASa-induced activation of IFN-I and NF-κB was stronger than that of DrcGASb. This re- sult was consistent with the observation that DrcGASa had high enzymatic activity for 2′3′-cGAMP production (5B). Hence, DrcGASa and DrcGASb participate in the activation of the IFN and NF-κB signaling pathways in a DrSTING-dependent manner with preferential activation by DrcGASa.

3.5 | Functional characterization of domain structures and residues
A number of DrcGASa/b mutant proteins with mutations of various functional domains/sites or N-terminus and C-terminal tail were generated through the deletion or substitution of the predicted critical residues or sequences in accordance with their counterparts evidenced in the HscGAS homolog (Table S2). The activity of these mu- tants was evaluated based on their stimulatory effects on the activation of IFN-β luciferase reporter in HEK293T cells with the cotransfection of Homo sapiens STING (HsSTING) or its allele HsSTING-R232H, which cannot respond to 3′3′-cGAMP. Most mutations in cGASa (eg, K458E, K485E, K468E, G274A/S275A, E287A/D289A, C470A/C471A, and Δ222) and cGASb (eg, K260E, K283E, R287E, K221E, K270E, G77A/S78A, E90A/D92A,
C272A/C273A, and Δ63) completely or partially lost their abilities to activate IFN-β ( 6A-C). These dysfunc- tions are also the cases for the corresponding mutations of HscGAS as previously described.9,15,47 However, the vari-
ation in the activities of various mutations differed from that observed in HscGAS mutations. For example, cGASa with K416E, K422E and K481E mutations at the DNA- binding surface remain active, and DrcGASb with R101E mutation is completely inactive, whereas those with corre- sponding mutations in HscGAS are inactive or partially ac-
tive.9 These observations suggested that a minor difference
exists between the association of DrcGASa/b and HscGAS with DNA ( 6A,C). Additionally, minimal activity was detected from mutant cGASa (R450I or T273Q/R450I)

6 Analysis of the functions of DrcGASa and DrcGASb mutants. A, Expression vectors of WT DrcGASa and mutant DrcGASa carrying different mutations were transfected into HEK293T cells together with HsSTING. B, Increasing amounts of WT DrcGASa or DrcGASa-Δ222 and low dose of HsSTING expression plasmids were transiently transfected into HEK293T cells, along with the IFN-β-Luc

reporter. C, Expression vectors of WT DrcGASb and mutant DrcGASb carrying different mutations were transfected into HEK293T cells together with HsSTING and the IFN-β luciferase reporter. D, Expression vectors of WT DrcGASa/b and mutant DrcGASa/b carrying different mutations and DncV were transfected into HEK293T cells together with HsSTING or HsSTING-R232H and the IFN-β luciferase reporter. Luciferase assay was performed after 24 hours. Data are representative of three independent experiments as mean ± SD (**P < .01)and cGASb (R252I or S76Q/R252I) with mutations at their substrate-binding sites in the presence of HsSTING-R232H and HsSTING, which were responsive to 2′3′-cGAMP and 2′3′-cGAMP/3′3′-cGAMP ( 6D). This behavior dif- fers from that of HscGAS, wherein corresponding mutations remain active in the presence of HsSTING but not in that of HsSTING-R232H due to the change in catalysates from 2′3′-cGAMP to 3′3′-cGAMP.48 Therefore, reprogramming the enzymatic activity of HscGAS and DrcGASa/b from synthesizing 2′3′-cGAMP to 3′3′-cGAMP had different requirements, which may include multiple parameters that determined linkage specificity, such as the twisting force provided by DNA.
3.6 | DrcGASa/b functions in an oligomerization manner
To investigate whether DrcGASa/b was activated in an oligomerization manner, we performed Co-IP assay in HEK293T cells by coexpressing DrcGASa or/and DrcGASb fusion proteins with flag, hemagglutinin (HA), or myc tags. Clear association signals were found between DrcGASa and DrcGASb proteins . A strong association signal was detected between DrcGASa and DrcGASb when the two proteins were coexpressed in cells, indicating the occurrence of DrcGASa/DrcGASb hetero-oligomerization (7C). Functionally, 2′3′-cGAMP production under this condition

7 Characterization of DrcGASa and DrcGASb oligomerization. A, HEK293T cells were transiently transfected with the Flag- tagged DrcGASa and the HA-tagged DrcGASa, followed by immunoprecipitation with anti-Flag. Western blot analysis was performed with the anti-HA and anti-Flag Ab. The lysates were immunoblotted to monitor the expression of Flag-tagged DrcGASa and the HA-tagged DrcGASa.

B, HEK293T cells were transiently transfected with the HA-tagged DrcGASb and the Myc-tagged DrcGASb, followed by immunoprecipitation with anti-HA. Western blot analysis was performed with the anti-HA and anti-Myc Ab. The lysates were immunoblotted to monitor the expression of HA-tagged DrcGASb and the Myc-tagged DrcGASb. C, HEK293T cells were transiently transfected with the Flag-tagged DrcGASa and the HA-tagged DrcGASb, followed by immunoprecipitation with anti-HA. Western blot analysis was performed with the anti-HA and anti-Flag Ab. The lysates were immunoblotted to monitor the expression of Flag-tagged DrcGASa and the HA-tagged DrcGASb. D, Investigation of the impact of DrcGASb on cGAMP production of DrcGASa. Protein depleted fractions of HEK293T cells transfected with DrcGASa expression plasmids and increasing amounts of DrcGASb expression plasmids from the C18 column were analyzed for the relative abundance of cGAMP. Data are representative of three independent experiments as mean ± SD (*P < .05, **P < .01)significantly (P < .05 or P < .01) decreased relative to that under DrcGASa oligomerization ( 7D). Therefore, the hetero-oligomerization of DrcGASa and DrcGASb inhibits the enzymatic activity of DrcGASa. This finding suggests the existence of a potential mutual regulation between DrcGASa and DrcGASb.
3.7 | Involvement of DrcGASa/b-DrSTING axis in innate immune reactions
The involvement of DrcGASa/b-DrSTING axis in innate immune defenses was examined by the expression of the representative IL-1β, IL-6, hepcidin, IFNφ1, IFNφ3, and IRF3 genes of the NF-κB and IFN-I signaling pathways in zebrafish adults and embryos. Three LVs with the most ef- ficient shRNAs for DrcGASa, DrcGASb, and DrSTING knockdown (by 66%/72%, 75%/69%, and 68%/77% in PBLs/ HKLs) were used in the study ( 8A-L). Administration with A hydrophila DNA as a stimulant significantly (P < .01) induced the expression of IL-1β, IL-6, hepcidin, IFNφ1,
IFNφ3, and IRF3 genes in zebrafish head kidneys and em- bryos. The induction of these genes remarkably (P < .01) decreased with the knockdown of DrcGASa or DrSTING in zebrafish adults or embryos by LV-/MO-based mRNA in- terference ( 9A,B). This suppression was followed by decrease in the survival of zebrafish embryos at each time point (hpi) under challenge with A hydrophila, E tarda, and

8 Silencing of DrcGASa, DrcGASb, or DrSTING gene by the lentivirus-based siRNA delivery system. A, E, and I, Screening of effective siRNAs against DrcGASa (A), DrcGASb (E), and DrSTING (I). Designed siRNAs targeting different regions of DrcGASa, DrcGASb, or DrSTING mRNA were, respectively, inserted into the pSUPER vector. The HEK293T cells were cotransfected with siRNA1-5 for DrcGASa (A), siRNA1-5 DrcGASb (E), or siRNA1-3 DrSTING (I) or the control plasmid (pSUPER vector harboring the scrambled siRNA) and the expression plasmid of DrcGASa (A), DrcGASb (E) and DrSTING (I). The efficacy of siRNAs was measured by real-time PCR. B, F and J, The silencing efficiency of lentivirus-delivered siRNA (siRNA2 for DrcGASa, siRNA5 for DrcGASb, and siRNA1 for DrSTING) against DrcGASa (B), DrcGASb (F) or DrSTING (J) in vivo was analyzed by real-time PCR. C, J, and K, Detection of the infection of LV-DrcGASa (C), LV-DrcGASb (J), or LV-DrSTING (F) efficacy in HEK293T cells determined by GFP fluorescence under a fluorescence microscope (Zeiss Axiovert 40 CFL, original magnification 3400). Most HEK293T cells strongly expressed GFP, demonstrating that the constructed LVs exhibited a highly infectious efficacy. D, H and L, Lentivirus titers of LV-DrcGASa (D), LV-DrcGASb (H), or LV-DrSTING (L) were assessed according to the percentage of
GFP+ HEK293T cells after exposure to different LV dilutions by FCM. The fluorescence of control cells without lentivirus infection is presented as outlines in the diagrams. The number above the bracketed line shows the percentage of GFP+ cells in image. Values in (A), (B), (E), (F), (I), and (J) are representative of three independent experiments as mean ± SD (**P < .01)VSV. The cumulative survival rates of zebrafish embryos de- creased to 16.33% ± 1.15% (A hydrophila), 12.67% ± 1.61% (E tarda), and 22.33% ± 1.44% (VSV) at 72 hpi relative to those of the control groups (38.50% ± 1.80%, 40.17% ± 2.02%, and 31.00% ± 1.50%) without DrcGASa knockdown. Upon the impairment of the gene expression, the survival rates of the DrcGASa-knockdown embryos in response to A hydrophila, E tarda, and VSV infections were lower than those of DrcGASb-knockdown embryos ( 9C-E). This finding suggests that the DrcGASa-DrSTING axis may play an important role in innate defense immunity. The adminis- tration of 2′3′-cGAMP to zebrafish or embryos significantly stimulated innate immune reactions and antibacterial immu- nity, as indicated by the upregulation of IL-1β, IL-6, hep- cidin, IFNφ1, IFNφ3, and IRF3 and the increased survival rates of fish. 3′3′-cGAMP exhibited a similar effect, but its reaction was weaker than that of 2′3′-cGAMP (10A- C). These phenomena were further confirmed in HEK293T cells through the exotic expression of DrSTING under 2′3′- cGAMP and 3′3′-cGAMP administration (10D,E). These results provided insights into the potential application of 2′3′-cGAMP or 3′3′-cGAMP as a new therapeutic adju- vant for fish on the basis of DrcGASa/b-DrSTING pathway.

3.8 | DrcGASa/b-DrSTING axis was differentially involved in adaptive mucosa immunity
Given the high expression levels of DrcGASa and DrSTING in gill, intestine, and skin, which are the three major mucosa- associated lymphoid tissues of fish, DrcGASa-DrSTING axis was supposed to play a role in adaptive mucosa immunity. This suggestion was initially supported by an induced-expression assay showing that DrcGASa and DrSTING were signifi- cantly upregulated in the gills/skins after E tarda infection at 1 day post infection (dpi), whereas DrcGASb was not in- duced ( 11A,B). With the knockdown of DrcGASa (by 68%/63%) and DrSTING (by 66%/70%) but not of DrcGASb
(by 71%/65%) in the gills/skins via LVs-based mRNA inter- ference (e 8B,F,J), the production of IgZ Abs in the gills/ skins upon E tarda infection was inhibited, as shown by the decline in the abundance of IgZ and IgZ2, which are two sub- class members of the IgZ family, at protein and mRNA levels ( 11C-F). These findings suggested that the DrcGASa- DrSTING axis preferentially participated in adaptive mucosal immunity. Afterward, zebrafish IFNφ1 was upregulated in the gills/skins after 3 and 7 days of E tarda infection. Meanwhile, IFNφ2, IFNφ3, and IFNφ4 did not increase ( 12A,B). DrcGASa/DrSTING knockdown inhibited IFNφ1 expres- sion in the gills/skins accompanied by a decrease in IgZ Abs ( 12C); the latter can be restored through the i.p. administration of IFNφ1 produced by exotic expression in HEK293T cells ( 12D-H). DrcGASa/DrSTING knockdown also decreased the percentages of IgZ+CD40+ and IgZ2+CD40+ B cells in the gills, and this decrease was rescued by administering IFNφ1 to the DrcGASa/DrSTING knockdown fish ( S5). These findings also suggested that CD40-mediated costimulatory signals were essential for the activation of the IgZ+ and IgZ2+ B cells.
Through immunofluorescence staining using Ab against DrcGASa generated in this study, DrcGASa was prevalent in the gills and prominently distributed in γδ T cells ( 13A and S6A). This observation suggested that γδ T cells may participate in DrcGASa-DrSTING-IFNφ1 axis-mediated mucosal immunity. Thus, a γδ T cell deletion assay was per- formed using anti-γ and anti-δ Abs, and the amount of γδ T cells in the gills of fish with E tarda infection decreased by approximately 50% by FCM analysis ( 13B). The γδ T cell-depleted fish displayed low IgZ/IgZ2 mRNA and pro- tein levels in the gill, and their IgZ+CD40+ and IgZ2+CD40+ B cell counts decreased from 28.81% ± 0.71% to 11.98% ± 0.37% and 29.21% ± 0.64% to 12.21% ± 0.57% compared with those of fish under E tarda stimulation alone. The down- regulation of IgZ/IgZ2 Abs and IgZ+CD40+ IgZ2+CD40+ B cells was rescued by transferring γδ T cells from E tar- da-stimulated gills pretreated with LV-DrcGASb for DrcGASb knockdown or PBS (control) without intervention

9 Evaluation of functional roles of DrcGASa/DrcGASb/DrSTING in IFN-I and NF-κB signaling pathway in vivo by knockdown assays. A and B, The expression of the representative effector or regulatory genes of the NF-κB and IFN-I signaling pathways in response to A hydrophila DNA in the embryos (A) and head kidneys from adult fish (B) was measured by real-time PCR. US indicates unstimulated group. C and D, Relative survival rates of A hydrophila (C) and E tarda (D) infected zebrafish embryos in knockdown experiments. Zebrafish embryos were microinjected with ControlMO, DrcGASaMO/DrcGASbMO/DrSTINGMO and raised to 24 hpf. Then, the embryos were exposed to PBS, A hydrophila, or E tarda at 1 × 108 cfu/mL for 5 hours by immersion and monitored for 72 hours. The survived embryos were calculated at 12, 24,
36, 48, and 72 hpi. E, Relative survival rates of VSV-infected zebrafish embryos in knockdown experiments. Zebrafish embryos were microinjected with ControlMO, DrcGASaMO/DrcGASbMO/DrSTINGMO and A h DNA and raised to 48 hpf. Then, the embryos were injected with PBS or VSV and monitored for 72 hours. Values in (A) and (B) are representative of three independent experiments as mean ± SD (**P < .01)for the DrcGASa-DrSTING pathway. In contrast, no signif- icant restoration was observed in γδ T cell-depleted recip- ients that were transferred with γδ T cells pretreated with LV-DrcGASa and/or LV-DrSTING for the knockdown of the DrcGASa-DrSTING pathway ( 13C,D and S7A). γδ T cells were also sorted from E tarda-stimulated gills and cocultured with gill lymphocytes for 72 hours. The results showed that the percentage of the IgZ+CD40+/IgZ2+CD40+ B cells in the total lymphocytes cocultured with the stim- ulated γδ T cells increased from 6.71% ± 0.97%/6.65% ± 0.42% to 21.01% ± 0.82%/22.34% ± 0.67% compared with those in the mock control that was not cocultured with γδ T cells. This enhancement was inhibited by supplementing cocultures with anti-IFNφ1 Ab (1 μg/mL) to neutralize the IFNφ1 cytokine ( S6B and S7B). These findings sug- gested that γδ T cells regulated IgZ/IgZ2 production in gills by upregulating IgZ+/IgZ2+ B cells via the paracrine action of IFNφ1, which depends on the activation of the DrcGAS- DrSTING signaling pathway in these cells.
4 | DISCUSSION
Nucleotidyltransferase fold proteins constitute a superfam- ily with various family members that transfer nucleoside monophosphate (NMP) from nucleoside triphosphate (NTP) to an acceptor hydroxyl group to produce a set of 3′-5′- and/or 2′-5′-linkage oligoadenylates and cyclic dinucleo- tides (CDNs), including c-di-AMP, c-di-GMP, c-di-UMP, 3′3′-cUAMP, 3′3′-cGUMP, 3′3′-cUCMP, and 3′3′-/2′3′- cGAMP.49,50 Early NTase family members, such as V cholera DncV, NvcGAS, and Geodia cydonium 2′-5′-oligoadenylate synthetase (OAS1) synthesize 3′3′-cGAMP, or mixed 3′-5′- and 2′-5′-linkage oligoadenylates that are independent of DNA stimulation.22,48,51 This ability indicates that 3′3′- and 2′3′-cGAMPs once coexist during various stages of metazoan evolution, and 2′3′-cGAMP is ultimately chosen from a cer- tain species. Thus, identifying which species begins to choose 2′3′-cGAMP is an interesting topic. In the present study, DrcGASa/b produced 2′3′-cGAMP but not 3′3′-cGAMP in

10 Evaluation of functional roles of 2′3′-cGAMP and 3′3′-cGAMP in IFN-I and NF-κB signaling pathway. A and B, The expression of the representative effector or regulatory genes of the NF-κB and IFN-I signaling pathways in response to 2′3′-cGAMP or 3′3′-cGAMP in the head kidneys from adult fish (A) and the embryos (B) was measured by real-time PCR. C, Relative survival rates of A
hydrophila-infected zebrafish with or without 2′3′-cGAMP stimulation. Zebrafish were ip-injected with 2′3′-cGAMP (1ug/fish) or PBS. After 12 hours, each fish were ip-injected with A hydrophila (2.1 × 105 cfu/fish) and monitored for 5 d. The survived embryos were calculated at 0, 1, 2, 3, 4, and 5 dpi. D and E, DrSTING expression plasmids or empty plasmid together with the IFN-β (D) or NF-κB (E) luciferase reporter was
transfected into HEK293T cells. The next day, 2′3′-cGAMP or 3′3′-cGAMP (2 μg/mL) was transfected into HEK293T cells. Reporter gene assays were conducted at 16 hours after transfection. Values in (A), (B), (D), and (E) are representative of three independent experiments as mean ± SD (**P < .01)vivo and in vitro. This finding showed that the evolutionary choice of 2′3′-cGAMP occurred as early as in teleost fish. The acquirement of the ability of cGAS to associate with DNA may have enabled cGAS to synthesize 2′3′-cGAMP instead of 3′3′-cGAMP. A reason for this reaction may be that the activity of cGAS to produce 2′3′-cGAMP required a pronounced conformation alteration after the interaction of cGAS with DNA, which triggered the repositioning of active site residues in the catalytic pocket fold. Among the active site residues, GS, E-D, and D (eg, G212S213, E225-D227 and D319 in HscGAS) located in the NTase core domain par- ticipated in the formation of 3′-5′- and/or 2′-5′-linkage phos- phodiester bonds. In contrast, R and Y (eg, R376 and Y436 in HscGAS) located in Mab21 domain and T (eg, T211 in HscGAS) distributed in the NTase core domain participated in substrate (ATP and GTP) docking, orientation, and rota- tion in the catalytic pocket fold, which determines 3′-5′- and/or 2′-5′-linkage specificity.48 Therefore, the repositioning of R, Y, and T in the catalytic pocket induced by the association of cGAS with DNA may be a major event for the produc- tion of 2′3′-cGAMP with 2′-5′- and 3′-5′-linkages. Further study is needed to clarify this hypothesis, which depends on a detailed crystal structure analysis of cGAS in response to DNA stimulation.

The acquisition of the ability of cGAS to interact with cy- tosolic DNA is an important evolutionary event for modern cGASs becoming DNA sensors that trigger STING-mediated NF-κB and IFN signaling by their enzymatic 2′3′-cGAMP products. This event may depend on the acquirement of var- ious functional elements in cGAS for DNA association. In the present study, DrcGASa/b possessed functional elements, including a zinc ribbon domain, a long N-terminal domain, a dimer surface, and three DNA-binding domains that are rich in conserved positively charged amino acid residues. Given
11 Evaluation of functional roles of DrcGASa, DrSTING, and DrcGASb in IgZ and IgZ2 production via knockdown assays. A and B, Real-time PCR analysis of the mRNA levels of DrcGASa, DrcGASb, and DrSTING in the gills (A) and skins (B) of fish at 1, 3, and
7 days after infection with E tarda. C, Real-time PCR analysis of the mRNA levels of IgZ and IgZ2 in each treatment group in response to E tarda infection. D, ELISA analysis of the titers of IgZ and IgZ2 in the mucus of gills or skins in each treatment group in response to E tarda infection. The titer was ascertained based on the highest serum dilution at which the A450 ratio (A450 of postimmunization mucus/A450 of preimmunization mucus) is ≥2.1. E, Western blot analysis of the protein levels of IgZ and IgZ2 in each treatment group in response to E tarda infection. Data are representative of three independent experiments as mean ± SD (**P < .01)that all these elements are absent from primitive metazoan cGASs, our findings suggested that modern cGASs origi- nated from teleost fish by acquiring functional domains for DNA sensing and 2′3′-cGAMP synthesis. This outcome was consistent with the beginning of the evolutionary event of 2′3′-cGAMP selection in fish. Zinc ribbon domain that is included in the Mab21 region triggers the interaction of cGAS with the major groove of the B-form dsDNA through association with a zinc ion.52 This action suggests that thezinc ribbon domain serves as a molecular ruler for scaling the specificity of cGAS toward dsDNA. Thus, the acquisition of the zinc ribbon domain by cGAS is a key event for the devel- opment of cGAS into a DNA sensor. Long N-terminal region, dimer surface, and DNA-binding domains also contribute to the conformational change of the catalytic pocket fold after the insertion of the zinc ribbon domain into dsDNA; this phe- nomenon results in the repositioning of the active residues for 2′3′-cGAMP synthesis. To clarify these hypotheses, we
12 Evaluation of functional roles of IFNφ1 in IgZ and IgZ2 production. A and B, Real-time PCR analysis of the mRNA levels of IFNφ1, IFNφ2, IFNφ3, and IFNφ4 in the gills (A) and skins (B) of zebrafish at 1, 3, and 7 days after infection with E tarda. C, Real-time PCR analysis of the mRNA levels of IFNφ1 in the gills and skins of zebrafish after in vivo knockdown assays and infection with E tarda for 3 days. D and E, Real-time PCR analysis of the mRNA levels of IgZ and IgZ2 in gills and skins (D) and ELISA analysis of the titers of IgZ and IgZ2 in the
mucus of gills or skins (E) in each treatment group in response to E tarda infection. The titer was ascertained based on the highest serum dilution at which the A450 ratio (A450 of post immunization sera/A450 of preimmunization sera) is ≥2.1. F, Western blot analysis of the supernatant liquids of HEK293T cells transfected with the IFNφ1-flag expression construct or an empty control plasmid 32 hours after transfection using the Flag mAb. G and H, Western blot analysis of the protein levels of IgZ and IgZ2 in gills (G) and skins (H) in each treatment group in response to E tarda infection. Data are representative of three independent experiments as mean ± SD (**P < .01)
constructed a series of mutant DrcGASa/b with mutations at each functional position in these domains. As predicted, most of the mutants demonstrated abolished or severely impaired 2′3′-cGAMP synthesis activity and downstream NF-κB and IFN signaling.
The origins of cGAS/STING homologs and their signal- ing components of NF-κB, TBK1, and IKK can be traced

back to early or early-branching metazoans.7 However, the key component IRF3/IRF7 homologs of the cGAS-STING pathway for IFN-I induction are absent from almost all early metazoans before the divergence of teleost fish. This phenom- enon indicated that the modern cGAS-STING pathway that plays a major role in IFN-mediated innate antiviral immunity may have originated from teleost fish with the occurrence of
13 Functional evaluation of DrcGASa-DrSTING-IFNφ1 signaling pathway in γδ T cells in mucosal immunity. A, Images of immunofluorescence staining of zebrafish gill taken by laser scanning confocal microscopy. The gill was stained for DrcGASa (red) and γ (green). The nuclei are stained with DAPI. Original magnification, ×630. Scale bar, 10 μm. B, Effects of anti-γ and anti-δ Abs on in vivo depletion of γδ T cells were assessed by FCM. Negative controls treated with PBS or normal rabbit IgG were created. The background staining with an irrelevant Ab on the target gated population is presented as outlines in the diagrams. Numbers above the marker bars indicate the percentage of γδ T cells in each treatment group. Each result was obtained from 30 fishes. Means ± SD of three independent experiments are shown. C, Real-time PCR analysis
of the mRNA levels of IgZ and IgZ2 in the gills and ELISA analysis of the titers of IgZ and IgZ2 in the mucus of gills in each treatment group in response to E tarda infection. The titer was ascertained based on the highest serum dilution at which the A450 ratio (A450 of postimmunization mucus/A450 of preimmunization mucus) was ≥2.1. D, Western blot analysis of the protein levels of IgZ and IgZ2 in each treatment group in response to E tarda infection. The fish received various treatments before infection, including administration of anti-γ and anti-δ Abs for γδ T cell deletion (indicated as Anti-γ/δ), in which the non-specific IgG was administrated into the fish as a negative control (indicated as IgG), and transference of γδ T cells from E tarda-stimulated gills pretreated with PBS, LV-DrcGASa, LV-DrSTING, LV-DrcGASb, or γδ T cells from unstimulated gills into the fish with γδ T cell deletion (indicated as Anti-γ/δ+γδ T, Anti-γ/δ+DrcGASa−γδ T, Anti-γ/δ+DrSTING−γδ T, Anti-γ/
δ+DrcGASb−γδ T, and Anti-γ/δ+USγδ T, respectively). Values in (C) and (D) are representative of three independent experiments as mean ± SD
(**P < .01)

IRF3/IRF7. Here, we showed the ability of DrcGASa/b and DrSTING in the activation of IFN-I production and antiviral immunity, thereby providing experimental evidence to support this notion. Structurally, all STING homologs had a central c-di-GMP-binding domain. However, a CTT domain that is essential for IRF3/IRF7 activation and autoinhibitory regula- tion was observed only in the STINGs of fish and other verte- brates and missing from ancient STINGs. Thus, the acquisition of CTT from fish STINGs is an important evolutionary event that provides structural basis for constructing a modern cGAS- STING pathway. This event was coevolved with the switch of cGAS to a DNA sensor for 2′3′-cGAMP synthesis as deter- mined by the coevolutional coefficient (0.908; P < .01) be- tween cGAS/cGASa and STING ( S8). As a support, the 2′3′-cGAMP is more effective for STING activation due to its higher binding affinity to STING than that of 3′3′-cGAMP.22,53 Herein, two cGAS isoforms (ie, DrcGASa and DrcGASb) were identified from zebrafish. Phylogenetic analysis showed that these two isoforms were paralogs that may have origi- nated from a common ancestor gene. Similar paralogs of cGAS could also be predicted from some other lower organ- isms, such as amphioxus, reptiles, and platypus. Through the calculation of the Ka/Ks ratios of cGAS/cGASa/b, STING, and TBK1 within four metazoan lineages (ie, mammals, fish, insects, and cnidarians), the Ka/Ks ratios of the fish cGASa, cGASb, STING, and TBK1 protein pairs are 0.12, 0.30, 0.23, and 0.01, respectively, and are similar to those of other protein pairs.7 These Ka/Ks ratios are less than 1.0, thereby suggesting that all these proteins are subjected to functional constraints by purifying selections. Small Ka/Ks ratios are indicative of strong functional constraints. The higher Ka/Ks ratio of the cGASb than that of cGASa and STING revealed that cGASa and STING undergo strong purifying selections in the IFN- induction pathway, whereas the functional constraint on cGASb has been relaxed. In the present study, DrcGASa had strong functional activities in 2′3′-cGAMP production and IFN-inducing reactions, thereby suggesting the preferential in- volvement of DrcGASa in STING-dependent IFN-production pathway. cGAS performs other biological functions, such as the regulation of DNA repair, tumorigenesis, cellular senes- cence, and autophagy, independent of STING.1,3,4 Hence, whether DrcGASb differentially participates in these newly known biological activities that are fundamental for cell me-
tabolism and survival should be clarified.
The cGAS-STING signaling pathway plays diverse roles in innate immunological activities.54-56 However, the func- tional performance of the cGAS-STING pathway in adaptive mucosal immunity remains largely unknown. In the present study, we explored the involvement of the cGAS-STING pathway in the production of IgZ and IgZ-mediated antibac- terial immunity in zebrafish gill-associated lymphoid tissue (GALT), which is a typical mucosa-associated lymphoid tissue (MALT) specialized in fish.57 IgZ or its equivalent

immunoglobulin T (IgT) is a newly identified immunoglob- ulin (Ig) class from teleost fish. This Ig class is characterized by its involvement in MALTs for mucosal defense against parasitic and bacterial infections.57 Several IgZ/IgT subclass members, including IgZ and IgZ2 in zebrafish and common carp and Igτ1, Igτ2, and Igτ3 in rainbow trout, have been identified.38,58,59 However, the regulatory mechanisms un- derlying IgZ/IgZ2 production and reaction remain limited. In the present study, DrcGASa preferentially contributed to IgZ/ IgZ2 induction in response to E tarda infection by upregu- lating IFNφ1 expression in gill γδ T cells. The 2′3′-cGAMP product of DrcGAS can be used as a strong adjuvant for IgZ/ IgZ2-mediated antibacterial immune reactions. To our knowl- edge, this study is the first to report the function of the cGAS- STING pathway in mucosal immunity and that this process is closely associated with γδ T cells. Given that IgZ/IgT is an 3′,3′-cGAMP  equivalent of mammalian IgA, teleost fish may serve as a favorable animal model for the study of cGAS-STING-medi- ated IgA immunology and diseases.

ACKNOWLEDGMENTS
This work was supported by grants from the National Natural Science Foundation of China (31572641, 31630083), the National Key Research and Development Program of China (2018YFD0900503, 2018YFD0900505), Stem Cell
and Translational Research, the National Key Research and Development Program of China (2016YFA0101001), the Open Fund of the Laboratory for Marine Biology and Biotechnology, Qingdao National Laboratory for Marine Science and Technology, Qingdao, China (OF2017NO02), the Open Funding Project of the State Key Laboratory of Bioreactor Engineering and the Zhejiang Major Special Program of Breeding (2016C02055-4). The authors thank Ms She-long Zhang for technical support for two-photon laser confocal scanning microscope capture and ultracentrifugation.
CONFLICT OF INTEREST
The authors declare that there is no conflict of interest in con- nection with the work submitted.
AUTHOR CONTRIBUTIONS
Z.F. Liu, A.F. Lin, L.X. Xiang, and J.Z. Shao designed the experiments and wrote the article; Z.F. Liu, J.F. Ji, X.F. Jiang, X.H. Jiang, and D.D. Fan performed the experiments;
Z.F. Liu, J.F. Ji, T. Shao, and J.Z. Shao analyzed the data; and all authors approved the final version of the article.
REFERENCES
1. Liu H, Zhang H, Wu X, et al. Nuclear cGAS suppresses DNA re- pair and promotes tumorigenesis. Nature. 2018;563:131-136.
2. Gao D, Li T, Li XD, et al. Activation of cyclic GMP-AMP synthase by self-DNA causes autoimmune diseases. Proc Natl Acad Sci U S A. 2015;112:E5699-E5705.

3. Liang Q, Seo GJ, Choi YJ, et al. Crosstalk between the cGAS DNA sensor and Beclin-1 autophagy protein shapes innate antimicrobial immune responses. Cell Host Microbe. 2014;15:228-238.
4. Yang H, Wang H, Ren J, Chen Q, Chen ZJ. cGAS is es- sential for cellular senescence. Proc Natl Acad Sci U S A. 2017;114:E4612-E4620.
5. de Oliveira Mann CC, Kiefersauer R, Witte G, Hopfner KP. Structural and biochemical characterization of the cell fate deter- mining nucleotidyltransferase fold protein MAB21L1. Sci Rep. 2016;6:27498.
6. Xie W, Lama L, Adura C, et al. Human cGAS catalytic domain has an additional DNA-binding interface that enhances enzymatic activity and liquid-phase condensation. Proc Natl Acad Sci U S A. 2019;116:11946-11955.
7. Wu X, Wu FH, Wang X, et al. Molecular evolutionary and struc- tural analysis of the cytosolic DNA sensor cGAS and STING. Nucleic Acids Res. 2014;42:8243-8257.
8. Li X, Shu C, Yi G, et al. Cyclic GMP-AMP synthase is activated by double-stranded DNA-induced oligomerization. Immunity. 2013;39:1019-1031.
9. Zhang X, Wu J, Du F, et al. The cytosolic DNA sensor cGAS forms an oligomeric complex with DNA and undergoes switch- like conformational changes in the activation loop. Cell Rep. 2014;6:421-430.
10. Ablasser A, Goldeck M, Cavlar T, et al. cGAS produces a 2′-5′-linked cyclic dinucleotide second messenger that activates STING. Nature. 2013;498:380-384.
11. Diner EJ, Burdette DL, Wilson SC, et al. The innate immune DNA sensor cGAS produces a noncanonical cyclic dinucleotide that ac- tivates human STING. Cell Rep. 2013;3:1355-1361.
12. Sun L, Wu J, Du F, Chen X, Chen ZJ. Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon path- way. Science. 2013;339:786-791.
13. Fang R, Wang C, Jiang Q, et al. NEMO-IKKbeta are essential for IRF3 and NF-kappaB activation in the cGAS-STING pathway. J Immunol. 2017;199:3222-3233.
14. Xia P, Wang S, Gao P, Gao G, Fan Z. DNA sensor cGAS-mediated immune recognition. Protein Cell. 2016;7:777-791.
15. Civril F, Deimling T, de Oliveira Mann CC, et al. Structural mechanism of cytosolic DNA sensing by cGAS. Nature. 2013;498:332-337.
16. Luecke S, Holleufer A, Christensen MH, et al. cGAS is ac- tivated by DNA in a length-dependent manner. EMBO Rep. 2017;18:1707-1715.
17. Gao P, Ascano M, Wu Y, et al. Cyclic [G(2',5′)pA(3′,5′)p] is the metazoan second messenger produced by DNA-activated cyclic GMP-AMP synthase. Cell. 2013;153:1094-1107.
18. Zhang X, Shi H, Wu J, et al. Cyclic GMP-AMP containing mixed phosphodiester linkages is an endogenous high-affinity ligand for STING. Mol Cell. 2013;51:226-235.
19. Wang J, Chu B, Du L, et al. Molecular cloning and functional char- acterization of porcine cyclic GMP-AMP synthase. Mol Immunol. 2015;65:436-445.
20. Rolf J, Siedentop R, Lutz S, Rosenthal K. Screening and identifi- cation of novel cGAS homologues using a combination of in vitro and in vivo protein synthesis. Int J Mol Sci. 2019;21:105.
21. Novotna B, Vanekova L, Zavrel M, et al. Enzymatic preparation of 2′-5′,3′-5′-cyclic dinucleotides, their binding properties to stim- ulator of interferon genes adaptor protein, and structure/activity correlations. J Med Chem. 2019;62:10676-10690.

22. Kranzusch PJ, Wilson SC, Lee AS, Berger JM, Doudna JA, Vance RE. Ancient origin of cGAS-STING reveals mechanism of univer- sal 2′,3′ cGAMP signaling. Mol Cell. 2015;59:891-903.
23. Martin M, Hiroyasu A, Guzman RM, Roberts SA, Goodman AG. Analysis of drosophila STING reveals an evolutionarily conserved antimicrobial function. Cell Rep. 2018;23:3537-3550.e6.
24. Margolis SR, Wilson SC, Vance RE. Evolutionary origins of cGAS-STING signaling. Trends Immunol. 2017;38:733-743.
25. Zhang C, Shang G, Gui X, Zhang X, Bai XC, Chen ZJ. Structural basis of STING binding with and phosphorylation by TBK1. Nature. 2019;567:394-398.
26. Guo DY, Cao C, Zhang XY, Xiang LX, Shao JZ. Scavenger re- ceptor SCARA5 acts as an HMGB1 recognition molecule nega- tively involved in HMGB1-mediated inflammation in fish models. J Immunol. 2016;197:3198-3213.
27. Feng W, Gu YF, Nie L, Guo DY, Xiang LX, Shao JZ. Characterization of SIGIRR/IL-1R8 homolog from zebrafish pro- vides new insights into its inhibitory role in hepatic inflammation. J Immunol. 2016;197:151-167.
28. Shao T, Shi W, Zheng JY, et al. Costimulatory function of Cd58/ Cd2 interaction in adaptive humoral immunity in a zebrafish model. Front Immunol. 2018;9:1204.
29. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: molecular evolutionary genetics analysis using maxi- mum likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol. 2011;28:2731-2739.
30. Gu YF, Fang Y, Jin Y, Dong WR, Xiang LX, Shao JZ. Discovery of the DIGIRR gene from teleost fish: a novel Toll-IL-1 receptor family member serving as a negative regulator of IL-1 signaling. J Immunol. 2011;187:2514-2530.
31. Shi W, Shao T, Li JY, et al. BTLA-HVEM checkpoint axis reg- ulates hepatic homeostasis and inflammation in a ConA-induced hepatitis model in zebrafish. J Immunol. 2019;203:2425-2442.
32. Biacchesi S, Merour E, Lamoureux A, Bernard J, Bremont M. Both STING and MAVS fish orthologs contribute to the induction of interferon mediated by RIG-I. PLoS ONE. 2012;7:e47737.
33. Biacchesi S, LeBerre M, Lamoureux A, et al. Mitochondrial an- tiviral signaling protein plays a major role in induction of the fish innate immune response against RNA and DNA viruses. J Virol. 2009;83:7815-7827.
34. Ma JX, Li JY, Fan DD, et al. Identification of DEAD-box RNA helicase DDX41 as a trafficking protein that involves in multiple innate immune signaling pathways in a zebrafish model. Front Immunol. 2018;9:1327.
35. Meng Z, Zhang XY, Guo J, Xiang LX, Shao JZ. Scavenger recep- tor in fish is a lipopolysaccharide recognition molecule involved in negative regulation of NF-kappaB activation by competing with TNF receptor-associated factor 2 recruitment into the TNF-alpha signaling pathway. J Immunol. 2012;189:4024-4039.
36. Wan F, Hu CB, Ma JX, Gao K, Xiang LX, Shao JZ. Characterization of gammadelta T cells from zebrafish provides insights into their important role in adaptive humoral immunity. Front Immunol. 2016;7:675.
37. Gong YF, Xiang LX, Shao JZ. CD154-CD40 interactions are essential for thymus-dependent antibody production in zebraf- ish: insights into the origin of costimulatory pathway in helper T cell-regulated adaptive immunity in early vertebrates. J Immunol. 2009;182:7749-7762.
38. Hu YL, Xiang LX, Shao JZ. Identification and characterization of a novel immunoglobulin Z isotype in zebrafish: implications

for a distinct B cell receptor in lower vertebrates. Mol Immunol. 2010;47:738-746.
39. Li JY, Gao K, Shao T, et al. Characterization of an NLRP1 in- flammasome from zebrafish reveals a unique sequential activation mechanism underlying inflammatory caspases in ancient verte- brates. J Immunol. 2018;201:1946-1966.
40. Meng F, Zhou R, Wu S, et al. Mst1 shuts off cytosolic an- tiviral defense through IRF3 phosphorylation. Genes Dev. 2016;30:1086-1100.
41. Zhang YA, Salinas I, Li J, et al. IgT, a primitive immunoglob- ulin class specialized in mucosal immunity. Nat Immunol. 2010;11:827-835.
42. Abdillahi H, Poolman JT. Whole-cell ELISA for typing Neisseria meningitidis with monoclonal antibodies. FEMS Microbiol Lett. 1987;48:367-371.
43. Zhu LY, Lin AF, Shao T, et al. B cells in teleost fish act as pivotal initiating APCs in priming adaptive immunity: an evolutionary perspective on the origin of the B-1 cell subset and B7 molecules. J Immunol. 2014;192:2699-2714.
44. Xu XG, Hu JF, Ma JX, et al. Essential roles of TIM-1 and TIM-4 homologs in adaptive humoral immunity in a zebrafish model. J Immunol. 2016;196:1686-1699.
45. Traver D, Paw BH, Poss KD, Penberthy WT, Lin S, Zon LI. Transplantation and in vivo imaging of multilineage engraftment in zebrafish bloodless mutants. Nat Immunol. 2003;4:1238-1246.
46. Xiong R, Nie L, Xiang LX, Shao JZ. Characterization of a PIAS4 homologue from zebrafish: insights into its conserved negative reg- ulatory mechanism in the TRIF, MAVS, and IFN signaling path- ways during vertebrate evolution. J Immunol. 2012;188:2653-2668.
47. Tao J, Zhang XW, Jin J, et al. Nonspecific DNA binding of cGAS N terminus promotes cGAS activation. J Immunol. 2017;198:3627-3636.
48. Kranzusch PJ, Lee ASY, Wilson SC, et al. Structure-guided repro- gramming of human cGAS dinucleotide linkage specificity. Cell. 2014;158:1011-1021.
49. Kuchta K, Knizewski L, Wyrwicz LS, Rychlewski L, Ginalski K. Comprehensive classification of nucleotidyltransferase fold pro- teins: identification of novel families and their representatives in human. Nucleic Acids Res. 2009;37:7701-7714.
50. Whiteley AT, Eaglesham JB, de Oliveira Mann CC, et al. Bacterial cGAS-like enzymes synthesize diverse nucleotide signals. Nature. 2019;567:194-199.
51. Pari M, Kuusksalu A, Lopp A, Reintamm T, Justesen J, Kelve M. Expression and characterization of recombinant

2′,5′-oligoadenylate synthetase from the marine sponge Geodia cydonium. FEBS J. 2007;274:3462-3474.
52. Kato K, Ishii R, Goto E, Ishitani R, Tokunaga F, Nureki O. Structural and functional analyses of DNA-sensing and immune activation by human cGAS. PLoS ONE. 2013;8:e76983.
53. Shu C, Li X, Li P. The mechanism of double-stranded DNA sens- ing through the cGAS-STING pathway. Cytokine Growth Factor Rev. 2014;25:641-648.
54. Gentili M, Lahaye X, Nadalin F, et al. The N-terminal domain of cGAS determines preferential association with centromeric DNA and innate immune activation in the nucleus. Cell Rep. 2019;26:2377-2393.e13.
55. Collins AC, Cai H, Li T, et al. Cyclic GMP-AMP synthase is an innate immune DNA sensor for mycobacterium tuberculosis. Cell Host Microbe. 2015;17:820-828.
56. Gui X, Yang H, Li T, et al. Autophagy induction via STING traf- ficking is a primordial function of the cGAS pathway. Nature. 2019;567:262-266.
57. Xu Z, Takizawa F, Parra D, et al. Mucosal immunoglobulins at respiratory surfaces mark an ancient association that predates the emergence of tetrapods. Nat Commun. 2016;7:10728.
58. Ryo S, Wijdeven RH, Tyagi A, et al. Common carp have two subclasses of bonyfish specific antibody IgZ showing differ- ential expression in response to infection. Dev Comp Immunol. 2010;34:1183-1190.
59. Zhang N, Zhang XJ, Chen DD, Oriol Sunyer J, Zhang YA. Molecular characterization and expression analysis of three sub- classes of IgT in rainbow trout (Oncorhynchus mykiss). Dev Comp Immunol. 2017;70:94-105.

SUPPORTING INFORMATION
Additional Supporting Information may be found online in the Supporting Information section.