key: cord-0937583-3iyef4qn
authors: Asthana, Abhishek; Gaughan, Christina; Weiss, Susan R.; Silverman, Robert H.
title: Specificity and Mechanism of Coronavirus, Rotavirus and Mammalian Two-Histidine-Phosphoesterases That Antagonize Antiviral Innate Immunity
date: 2021-06-17
journal: bioRxiv
DOI: 10.1101/2021.06.16.448777
sha: 228ebbf300410940cf417869ea332b7a6232f6b7
doc_id: 937583
cord_uid: 3iyef4qn

2’,5’-oligoadenylate(2-5A)-dependent endoribonuclease, RNase L, is a principal mediator of the interferon (IFN) antiviral response. Therefore, regulation of cellular levels of 2-5A is a key point of control in antiviral innate immunity. Cellular 2-5A levels are determined by IFN-inducible 2’,5’-oligoadenylate synthetases (OASs) and by enzymes that degrade 2-5A. Importantly, many coronaviruses and rotaviruses encode 2-5A degrading enzymes thereby antagonizing RNase L and its antiviral effects. A-kinase anchoring protein 7 (AKAP7), a mammalian counterpart, could possibly limit tissue damage from excessive or prolonged RNase L activation during viral infections or from self double-stranded-RNAs that activate OAS. We show these enzymes, members of the two-histidine-phosphoesterase (2H-PE) superfamily, constitute a sub-family referred here as 2’,5’-PEs. 2’,5’-PEs from mouse coronavirus (CoV) MHV (NS2), MERS-CoV (NS4b), group A rotavirus (VP3), and mouse (AKAP7) were investigated for their evolutionary relationships and activities. While there was no activity against 3’,5’-oligoribonucleotides, all cleaved 2’,5’-oligoadenylates efficiently, but with variable activity against other 2’,5’-oligonucleotides. The 2’,5’-PEs are shown to be metal ion-independent enzymes that cleave trimer 2-5A (2’,5’-p3A3) producing mono- or di- adenylates with 2’,3’-cyclic phosphate termini. Our results suggest that elimination of 2-5A might be the sole function of viral 2’,5’-PEs, thereby promoting viral escape from innate immunity by preventing or limiting the activation of RNase L. IMPORTANCE Viruses often encode accessory proteins that antagonize the host antiviral immune response. Here we probed the evolutionary relationships and biochemical activities of two-histidine-phosphoesterases (2H-PEs) that allow some coronaviruses and rotaviruses to counteract antiviral innate immunity. In addition, we investigated the mammalian enzyme, AKAP7, which has homology and shared activities with the viral enzymes and might reduce self-injury. These viral and host enzymes, that we refer to as 2’,5’-PEs, specifically degrade 2’,5’-oligoadenylate activators of the antiviral enzyme RNase L. We show that the host and viral enzymes are metal ion independent and exclusively cleave 2’,5’- and not 3’,5’-phosphodiester bonds, producing cleavage products with cyclic 2’,3’-phosphate termini. Our study defines 2’,5’-PEs as enzymes that share characteristic conserved features with the 2H-PE superfamily but which have specific and distinct biochemical cleavage activities. These findings may eventually lead to pharmacologic strategies for developing antiviral drugs against coronaviruses, rotaviruses, and other viruses.

ABSTRACT 24 25 2',5'-oligoadenylate(2-5A)-dependent endoribonuclease, RNase L, is a principal 26 mediator of the interferon (IFN) antiviral response. Therefore, regulation of cellular 27 levels of 2-5A is a key point of control in antiviral innate immunity. Cellular 2-5A levels 28 are determined by IFN-inducible 2',5'-oligoadenylate synthetases (OASs) and by 29 enzymes that degrade 2-5A. Importantly, many coronaviruses and rotaviruses encode 30 2-5A degrading enzymes thereby antagonizing RNase L and its antiviral effects. A-31 kinase anchoring protein 7 (AKAP7), a mammalian counterpart, could possibly limit 32 tissue damage from excessive or prolonged RNase L activation during viral infections or 33 from self double-stranded-RNAs that activate OAS. We show these enzymes, 34 members of the two-histidine-phosphoesterase (2H-PE) superfamily, constitute a sub-35 family referred here as 2',5'-PEs. 2',5'-PEs from mouse coronavirus (CoV) MHV (NS2), 36 MERS-CoV (NS4b), group A rotavirus (VP3), and mouse (AKAP7) were investigated for 37 their evolutionary relationships and activities. While there was no activity against 3',5'-38 oligoribonucleotides, all cleaved 2',5'-oligoadenylates efficiently, but with variable 39 activity against other 2',5'-oligonucleotides. The 2',5'-PEs are shown to be metal ion-40 independent enzymes that cleave trimer 2-5A (2',5'-p 3 A 3 ) producing mono-or di-41 adenylates with 2',3'-cyclic phosphate termini. Our results suggest that elimination of 2-42 5A might be the sole function of viral 2',5'-PEs, thereby promoting viral escape from 43 innate immunity by preventing or limiting the activation of RNase L. 44

CoV (NS4b), rotavirus group A (RVA) (VP3-C-terminal domain, CTD), and mouse 113 AKAP7. We show that NS2 and NS4b are remarkably specific for cleaving 2',5'-linked 114 oligoadenylates, whereas AKAP7 and VP3-CTD will also cleave other 2 ',5'-115 oligonucleotides. In contrast, all of the viral and mammalian 2',5-PEs tested lack the 116 ability to cleave 3',5'-oligoribonucleotides. We further show that these enzymes are 117 metal ion-independent and cleave trimer 2-5A (2',5'-p 3 A 3 ) producing mono-and di-118 adenylates with 2',3'-cyclic phosphoryl termini. Our findings suggest that the sole 119 function of the viral 2',5'-PEs may be to eliminate 2-5A allowing some coronaviruses 120 and rotaviruses to evade the antiviral activity of RNase L. To probe the precise molecular mechanism by which 2',5'-PEs allow some viruses to 126 evade the antiviral effector RNase L, we further investigated MHV NS2, NS4b, rotavirus group A (RVA) VP3-C-terminal domain (CTD), and mouse (mu)AKAP7, 128 (Fig. 1B) . A comparison of the domain organization of these 2',5'-PEs shows a related, 129 catalytic domain. Some of these enzymes have additional domains related to 130 intracellular localization, nucleic acid metabolism or protein binding functions indicative 131 of their cellular compartment-specific or accessory functions (Fig. 1B) . For instance, 132 MERS-CoV NS4b and muAKAP7 contain N-terminal nuclear localization signal (NLS) 133 domains (24, 27, 28) . VP3 is a multifunctional enzyme that contains N-terminal 134 guanylyltransferase (Gtase) and methyltransferase (Mtase) domains involved in capping 135 of the 5' termini of viral mRNAs (25, 29) . muAKAP7 also has a carboxy-terminal binding 136 domain for the regulatory subunit II (RII) of cyclic AMP (cAMP)-dependent protein 137 kinase A (PKA-RII-α-BD) (27) . In addition, MHV NS2 protein has a C-terminal 138 extension of unknown identity or function (Fig. 1B) . 139

To determine the phylogenetic relationships between the different 2',5'-PEs, we 141 constructed a tree for amino acid sequences containing the catalytic domains from 142 coronavirus, rotavirus and mammalian 2',5'-PEs ( Fig. 2A ). 2',5'-PEs were distributed 143 into two distinct branches on the phylogenetic tree. VP3 group of proteins clustered into 144 one branch while the other three groups containing NS2, NS4b and AKAP7 formed a 145 separate branch ( Fig. 2A) . Within the VP3 group, RVA and RVB resolved on distinct 146 sub-branches. Previously, full-length VP3 from RVA and RVB were also shown as 147 separate distinct branches analogous to two separate clades (clade A and clade B) (30, 148 31) . Interestingly, the NS2 proteins were mostly closely related to the mammalian 149 AKAP7 catalytic domains, and then to the bat coronaviruses (HKU5 and SC2013) and 150 MERS-CoV. The rotavirus VP3 proteins were most distally related to the other 2',5 153 Based on the phylogenetic relationship and functional relatedness, we further analyzed 154 the sequence conservation by amino acid alignment. 2H-PE superfamily members are 155 characterized by the presence of two H-ɸ-S/T-ɸ motifs, separated by an average of 80 156 amino acids (where ɸ represents a hydrophobic amino acid) (19) . The alignment shows 157 that both motifs are highly conserved across all 2',5'-PE proteins (Fig. 2B, see boxes) . 158

These motifs form the catalytic core that bind to and cleave the 2-5A substrate. 159

Consistent with the phylogenetic analysis, sequence analysis revealed that 2',5'-PEs 160 clustered into four groups corresponding to NS2, NS4b, AKAP7 and VP3. The two 161 histidines within the conserved motifs were 100% conserved among all the sequences 162 ( Fig. 2B , see asterisks). Several residues with intergroup consensus of >50% were 163 identified in the alignment. The amino acid alignment shows several regions of 164 conservation that exist beyond the two conserved catalytic motifs (H-ɸ-S/T-ɸ) (Fig. 2B,  165 shown above sequence alignment). 166 167 Among the sequences in alignment, AKAP7 proteins of human, rat and mouse origin 168 shared the highest amino acid identity ranging between 85 to 97% (88 to 97% similarity) 169 (Table S1) identity of trimer 2-5A (2',5'-p 3 A 3 ) were confirmed by HPLC (Fig. 3A ) and mass 197 spectrometry [described later for ( Fig. 5J) ]. Purified 2',5,-PE proteins were incubated 198 with 2-5A substrate at 30°C for 1 h and the 2-5A cleavage products analyzed by HPLC 199 using a C18 column. All five wild type proteins cleaved 2-5A as observed by loss of 200 intact 2-5A and appearance of peaks for the different cleavage products ( four products upon extended degradation of 2-5A suggesting a difference in either 204 mechanism or rate of cleavage by these proteins. On the other hand, 2-5A cleavage by 205 human PDE12 (Fig. 3F ) results in the formation of two products corresponding to the 206 elution time of the standard ATP and 5'-AMP, as previously described (32). As 207 expected, the 2',5'-PE catalytically inactive mutant proteins containing a His-to-Arg or 208

His-to-Ala mutations in the conserved histidines did not cleave 2-5A ( Fig. 3G-J) . Human 209 PDE12 with Glu-to-Ala mutation at 351 amino acid residue also did not degrade 2-5A, 210 as described previously (34) (Fig. 3K) . Importantly, these findings show a different mode 211 of 2-5A cleavage between 2',5'-PEs, members of the 2H-PE superfamily, and PDE12, a 212 member of the EEP family of phosphodiesterases. 213

To investigate the expanded substrate specificity of 2',5'-PEs, we tested possible 214 cleavage of various 2'-5' and 3'-5' linked oligoribonucleotides by HPLC. Purified 2',5'-PE 215 proteins (1 µM) were incubated with either 2'-5'-or 3'-5'-linked pentaribonucleotide 216 substrates (10 µM) at 30°C for 1 h. Wild type MERS-NS4b, specifically degraded 2'-5' 217 p5'(rA) 5 by >99% while 2'-5' p5'(rU) 5 , p5'(C) 5 or p5'(G) 5 were not degraded (<4%) 218 (Table S2) . Catalytically inactive mutant MERS-NS4b H182R did not degrade any of the 219 tested substrates under similar conditions. Wild type MHV NS2 also specifically 220 degraded 2'-5' p5'(rA) 5 >99% while 2'-5' p5'(rU) 5 , p5'(C) 5 or p5'(G) 5 were not degraded 221 (<7%) ( Table S2 ). Mutant MHV NS2 H126R did not degrade any of the tested substrates. 222

These results suggest MERS-NS4b and MHV NS2 are remarkably specific in degrading 223 2'-5' linked oligoadenylate compared to the other substrates. We further tested RVA 224 VP3-CTD which degraded 2'-5' p5'(rA) 5 >95%, p5'(rU) 5 ~ 40%, p5'(C) 5 ~90%, and 225 p5'(G) 5 ~6% while mutant RVA VP3-CTD H718A did not degrade any of the tested 226 substrates (Table S2 ). Wild type muAKAP7 degraded 2'-5' p5'(rA) 5 >99%, p5'(rU) 5 >95%, p5'(C) 5 >95%, and p5'(G) 5 >90% while mutant muAKAP7 H93A;H185R did not 228 degrade any of the tested substrates with the exception of 2'-5' p5'(G) 5 ~40% (Table  229 S2). To ensure that exclusive cleavage of 2',5'-oligoadenylates by MERS-NS4b was 230 not due to limiting amounts of enzyme, 10 µM of different 2',5'-linked penta-231 ribonucleotides were incubated with three-fold higher concentrations (3 µM) of MERS-232 NS4b at 30°C for 1h. Wild type MERS-NS4b specifically degraded 2'-5' p5'(rA) 5 >99% 233 while 2'-5' p5'(rU) 5 , p5'(C) 5 or p5'(G) 5 were not degraded (<6%) suggesting MERS-234

NS4b enzymatic activity is specific for degradation of 2',5'-oligoadenylates (Table S3) . 235

Because MERS NS4b and MHV NS2 cleaved 2'-5' p5'(rA) 5 but not other 2'-5'-linked 236 substrates, we further determined if this was due to lack of binding to the other 237 substrates. To test this possibility, 10 µM of 2'-5' p5'(rA) 5 was incubated with 0.2 µM of 238 MHV NS2 in the absence or presence of increasing concentrations of 2'-5' p5'(rU) 5 at 239 30°C for 10 min. The amounts of 2'-5' p5'(rA) 5 degraded by MHV NS2 in the presence 240 of 0, 3.1, 10, 12.5, 25, 50 and 100 µM was determined by HPLC analysis (Fig. S1 ). 241

Degradation of 2'-5' p5'(rA) 5 by MHV NS2 decreased as the amount of 2'-5' p5'(rU) 5 in 242 the reaction increased beyond 10 µM (i.e. ratio >1:1) (Fig. S1 ). Our results suggests 243 that 2'-5' p5'(rU) 5 was able to bind MHV NS2 and competitively interfere with MHV NS2 244 ability to cleave 2'-5' p5'(rA) 5 . 245

We next tested degradation activity of 2',5'-PEs against 3'-5' linked p5'(rA) 5 , p5'(rU) 5 246 and p5'(C) 5 . One µM of enzyme was incubated with 10 µM of the substrate at 30°C for 1 247 h. Wild type MERS-NS4b and its mutant MERS-NS4b H182R , wild type MHV NS2 and its 248 mutant NS2 H126R , RVA VP3-CTD and its mutant RVA VP3-CTD H718A , and wild type 249 muAKAP7 and its mutant muAKAP7 H93A; H185R (Table S2 ) did not degrade the 3'-5' linked substrates 3'-5' p5'(A) 5 , 3'-5' p5'(U) 5 , and 3'-5' p5'(C) 5 . [We were unable to 251 obtain 3'-5' p5'(G) 5 because of repeated failures of its chemical synthesis and/or 252 purification, therefore this oligonucleotide could not be tested]. Our results suggest that 253 all of the 2',5'-PEs examined are highly specific for cleaving 2',5' linked 254 oligoribonucleotides. Among 2',5' linked substrates MERS-NS4b and MHV NS2, are 255 specific for cleaving 2'-5' oligoadenylate, whereas RVA VP3-CTD cleaved in order: 2'-5' 256 pA5>pC5>pU5>>pG5 and muAKAP7 cleaved all of 2',5' linked pentanucleotides with 257 similar efficacy. 258 Based on the differential enzymatic activity of these 2',5'-PEs in degrading different 259 types of 2',5'-linked phosphodiester substrates, we tested if they could degrade 2',3'-260 cyclic-GMP-AMP (cGAMP). cGAMP is a cyclic-dinucleotide secondary messenger with 261 mixed phosphodiester linkages between 2'-OH of GMP to 5'-phosphate of AMP and 3'-262 OH of AMP to 5'-phosphate of GMP , synthesized by cyclic GMP-AMP-synthase 263 (cGAS) in response to cytoplasmic dsDNA (35). cGAMP was incubated either with or 264 without wild type and mutant 2',5'-PEs at 30°C for 1h and analyzed by HPLC. Wild type 265 MERS-NS4b, MHV NS2, RVA VP3-CTD, and muAKAP7 did not degrade 2',3'-cGAMP 266 whereas they did degrade 2',5'-p 3 A 3 (served as a positive control) under similar 267 conditions (Table S4 ). Catalytic mutants of 2',5'-PEs tested did not degrade 2',3'-268 cGAMP or 2',5'-p 3 A 3 under similar conditions. The results suggest that 2',5'-PEs are 269 capable of cleaving 2',5'-phosphodiester bonds in linear homo-ribonucleotides but not in 270 the cyclic-mixed phosphodiester linked 2',3'-cGAMP. Our results suggest that the 2',5'-PEs activity of these proteins is independent of Mg 2+ 288 ions and its presence either slightly decreases or has no effect on the activity of these 289 enzymes. 290 291 2',5'-PEs cleave 2',5'-linked oligoadenylate leaving products with cyclic 2',3' 292 phosphoryl termini. 293

Differences in the 2-5A cleavage products as determined by HPLC (Fig. 3) suggested 294 that viral and mammalian 2',5'-PEs cleave 2-5A via a different mechanism than human 295 PDE12, which degrades 2-5A to produce ATP and AMP (32). Among 2',5'-PEs, NS4b and MHV NS2 degraded 2-5A to give two cleavage products whereas RVA VP3-297 CTD and muAKAP7 gave four cleavage products. Therefore we decided to determine 298 the precise cleavage sites in 2-5A by 2',5'-PEs. 2-5A was partially digested with MERS-299 NS4b ( and were found to correspond to ApA and p 3 A>p (where ">p" represents a 2',3' cyclic 316 phosphate), respectively. Intact 2-5A gave an m/z ratio of 584 for the double charged 317 ion (Fig. 5J ). Moreover, the collected peak of p 3 A>p (shown in Fig. 5D ) was subjected to SAP mediated 5' dephosphorylation which results in the peak corresponding to A>p 319 ( Fig. 5E , F). This experiment suggested that MERS-NS4b degrades 2-5A to produce a 320 5'-product with 2',3' cyclic phosphate terminus in the form of p 3 A>p and a 3'-product of 321

ApA. To test if p 3 A>p and ApA are end products of the reaction, we subjected 2-5A to 322 extended degradation by MERS-NS4b and monitored the area under the peak 323 corresponding to ApA at 0 h, 1 h, 4 h and 24 h ( Overall mechanisms and differences in degradation of 2-5A by representative EEP 385 (PDE12) and 2',5'-PEs family members are summarized in figure 7. Human PDE12 386 degrades trimer 2-5A into ATP and 2 (5'-AMP)s in the presence of Mg 2+ ions, as has been reported (32). On the other hand, mammalian and viral 2',5'-PEs, act in a metal-388 ion independent way, degrading 2-5A to form 5' products with 2',3' cyclic phosphates. 389 All 2',5'-PEs quickly cleave active anti-viral 2-5A into inactive molecules that is, the 390 products are not capable of activating RNase L because of a requirement for at least 391 three adenylyl residues (37) The 2',5'-PEs studied here exclusively cleaved 2',5'-and not 3',5'-phosphodiester 402 bonds. There was also a strong preference for cleavage of 2',5'-oligoadenylates by NS2 403 and NS4b and, to a lesser extent, by VP3-CTD, whereas AKAP7 had similar activities 404 against the different 2',5'-linked pentamers of A, U, C and G. Therefore, although 405 AKAP7 and VP3-CTD are not the mostly closely related 2',5'-PEs, they can both cleave 406 2',5'-oligoribonucleotides other than 2-5A ( Fig. 2A and Table S2 ). Interestingly, OASs 407 are 2'-nucleotidyl transferases that not only use ATP as substrates but can produce 408 diverse molecules with 2',5' linkages. NAD + , tRNAs, A5'p 4 5'A, and mono-and poly-409 OAS can add other 2'-terminal ribo-and deoxy-nucleotide monophosphates to 2-5A 411 (38) (39) (40) (41) . However, which, if any, of these alternative 2-5A-like molecules can be cleaved 412 by 2',5'-PEs remains to be determined. The cGAMP, a cyclic dinucleotide that activates 413 STING, has one 2',5'-linkage and one 3',5'-linkage, but it is not cleaved by the 2',5'-PEs 414 examined here (Table S4 ). VP3 was phylogenetically distal and has the most distinct 415 mechanism of 2-5A cleavage compared to all of the tested 2',5'-PEs. It is also 416 interesting to note that the two coronavirus 2',5'-PEs (NS4b and NS2) are less closely 417 related than NS2 is to the host enzyme, AKAP7 ( Fig. 2A & Table S1 ). Our results 418 suggest that the main, and perhaps only, function of these activities is to degrade 2-5A 419 thus preventing RNase L activation and viral escape, or in the case of AKAP7 reducing 420 cell and tissue damage from RNase L activity. These 2',5'-PEs are also metal ion 421 independent enzymes, as is RNase L (42) . 422

The viral and mammalian 2',5'-PEs produce cleavage products from trimer 2-5A (2',5'-424 p 3 A 3 ) with cyclic 2',3'-phosphoryl groups, and not 2',3'-OH termini. These conclusions 425 are based on analysis of 2-5A cleavage products by two types of HPLC columns 426 (Dionex and C18) and, importantly, by mass spectrometry. In contrast, our prior studies 427 based on more limited analysis of the 2-5A cleavage products by one type of HPLC 428 column (Dionex) misidentified these cleavage products of NS2, VP3-CTD, and AKAP7 429 as AMP and ATP (22, 25, 27) . 430 431 Interestingly, mammalian and viral 2',5'-PEs have activities highly similar to an 432 invertebrate 2H-PE present in the oyster, Crassostrea gigas (43). The oyster enzyme has sequence similarity to AKAP7, is metal ion independent, cleaves 2',5'-but not 3',5'-434 linked oligonucleotides, and leaves cyclic 2',3'-phosphate and 5'-OH termini on its 435 products. It also degraded tri-phosphorylated 2-5A oligomers with multi-fold efficiency 436 compared to the corresponding non-phosphorylated core 2-5A oligomers. Similarly, we 437 observed RVA VP3-CTD degrades 5'-triphosphorylated-di-adenylate with 2',3' cyclic 438 phosphoryl termini (p3ApA>p) The viral enzymes NS2, NS4b and VP3-CTD are antagonists of innate immunity that 472 support virus replication by eliminating 2-5A and preventing, or reducing, activation of 473 RNase L by 2-5A (20, 22, 24-26). In contrast, mammalian AKAP7 is a nuclear 2',5'-PE 474 that does not affect viral replication, unless its nuclear localization signal peptide is 475 deleted leading to cytoplasmic accumulation (27). A mutant AKAP7 deleted for its N-476 terminal nuclear localization signal peptide accumulates in the cytoplasm was able to 477 rescue an NS2 mutant of MHV (22). While the function of the 2',5'-oligonucleotide cleaving activity of AKAP7 is still unresolved, the phylogenetic tree suggests that the 479 NS2 coronavirus proteins may have evolved from the AKAP7 catalytic domain ( Fig. 2A) . 480 481 Enzymes that degrade 2-5A have significance beyond antiviral innate immunity. Self-482 dsRNA activates the OAS-RNase L pathway leading in some circumstances to 483 apoptosis (12,13). In one example, mutation or inhibition of the dsRNA editing enzyme, 484 ADAR1, leads to accumulation of self dsRNA activating OAS-RNase L leading to cell 485 death, and PKR, inhibiting protein synthesis initiation (16, 48). In another instance, 486 DNA methyltransferase inhibitors, e.g. 5-aza-cytidine, cause self-dsRNA accumulation 487 from repetitive DNA elements leading to OAS-RNase L activation and apoptosis (17, 488 49). Thus, 2-5A is a secondary messenger for cytotoxic and antiviral activities of either 489 non-self (viral) or self-dsRNA (host) whose levels must be tightly controlled to limit Protease inhibitor, Thermo Scientific, USA) and 10% glycerol] and lysed with lysozyme 552 followed by sonication. Supernatants were incubated with Amylose resin (NEB, USA), 553 washed three times with buffer followed by elution with 100 mM maltose. Proteins were 554 concentrated using Centriprep centrifugal filter devices (Millipore; molecular weight 555 cutoff, 10 kDa) and further purified using size exclusion chromatography (SEC) on an 556 AKTA pure 25L protein purification system (GE Healthcare, USA) in buffer C (20 mM 557 HEPES pH 7.5, 100 mM NaCl and 1 mM DTT). Wild type and catalytic mutants of 558 NS4b, and MHV NS2 were purified as described previously (22, 24) . In addition to 559 inactive mutants, purified MBP protein was used as control in experiments with MBP 560 fusion proteins. Protein concentrations were determined using Bio-Rad protein assay 561 reagent (Bio-Rad, USA). All proteins were stored in Buffer C supplemented with 10% 562 glycerol in -80°C. 563 564

substrates 566 2-5A or p 3 5'A(2'p5'A) 2 (2',5'-p 3 A 3 ) was synthesized from ATP by using histidine-tagged RNA oligoribonucleotides (other than 2',5'-p 3 A 3 ) with 2'-5' or 3'-5' phosphodiester 587 linkages were commercially purchased. Oligonucleotide substrates 5'-588

pU3'p5'U3'p5'U3'p5'U3'p5'U -3' were purchased from Integrated DNA Technologies 591 (IDT, USA) while 5'-pC2'p5'C2'p5'C2'p5'C2'p5'C -3', 5'-pC3'p5'C3'p5'C3'p5'C3'p5'C -3' 592 and 5'-pA2'p5'A -3' were purchased from ChemGenes Corporation (Wilmington, USA). 593 Penta-ribonucleotides substrates are shown as p5'(rN) 5 , where N represents A, U, G, or 594 C nucleotide. A2'p5'A standard was prepared by incubating 5'pA2'p5'A with shrimp 595 alkaline phosphatase (ThermoFisher, USA) as per manufacturer's protocol. 2',3'-cyclic- 

The substrates and cleavage products were analyzed on a 1260 Infinity II Agilent 612 technologies HPLC equipped with an Infinitylab Poroshell 120 C18 analytical column 613 (Agilent technologies, 4.6 x 150 mm, 4µm). Eluent A was 50 mM ammonium phosphate 614 buffer pH 6.8 and eluent B was 50% methanol in water. Five µl of processed samples 615 were injected on the C18 column, at a flow rate of 1 ml/min and eluted with a linear gradient (0-40%) of eluent B over a period of 20 min, then 3 min 40% Eluent B, followed 617 by equilibration to initial condition (100% Eluent A). The HPLC column was maintained 618 at 40°C. Spectra were recorded at 256 nm. The products were identified either by 619 comparing the elution time of known standards or by mass-spectrometry analysis. 620

Alternatively, to test expanded substrate specificity, 10 µl of processed samples were 621 injected on a Dionex DNAPac R PA-100 analytical column at a flow rate of 1 ml/min and 622 eluted with a linear gradient of 10-800 mM of NH 4 HCO 3 buffer (pH 7.8) over a period of 623 90 min, followed by 30 min equilibration to initial condition. Open Lab CDS software was 624 used to analyze and calculate area under the peaks in HPLC spectra. 625 626

Purified substrates and cleavage product mixtures were dephosphorylated by 628 incubating with SAP (ThermoFisher, USA) at 37°C for 1 h according to the 629 manufacturer's protocol. Samples were prepared for subsequent analysis as described 630 above. 631 632 Sample preparation for mass spectrometry 633 Desired peak fractions (including cleavage products of 2',5'-p 3 A 3 ) were collected by 634 running samples on a Dionex DNAPac R PA-100 analytical column as described above. 635

Collected peaks were subjected to acetone precipitation, supernatants containing 636 cleavage products (from HPLC peak) were collected and lyophilized. Lyophilized 637 samples were suspended in 1 mM NH 4 HCO 3 buffer (pH 7.8) and used for mass 638 spectrometry analysis. 639 640

Prepared samples were subjected to mass spectrometry analysis. The LC/MS/MS 642 analysis was carried out using a triple quadrupole tandem mass spectrometer (TSQ-643 Quantiva, Thermo Scientific, USA) equipped with an electrospray ionization (ESI) 644 interface. The mass spectrometer was coupled to the outlet of HPLC system that 645 consisted of an UHPLC system (Vanquish, Thermos Fisher Scientific, USA), including 646 an auto sampler with refrigerated sample compartment and inline vacuum degasser. 647

The Xcalibur software was used for data processing. The ESI mass spectrometric 648 detection was performed in both the negative and positive ionizations, with ions spray 649 voltage at 2.5kV, sheath gas at 35 Arb and Aux gas at 20 Arb. The ion transfer tube 650 and vaporizer temperatures were set at 350°C and 250°C, respectively. The qualitative 651 analysis was performed using full scan at the range from 200 to 1250 (m/z). Five µl 652 extracted samples were injected on the C18 column (Gemini, 3 µm, 2 x 150 mm, 653

Phenomenex, CA) with the flow rate of 0.3 ml/min at 45°C. Mobile phases were A 654 (water containing 10 mM ammonium acetate and 20 mM ammonium hydroxide) and B 655 (methanol containing 10 mM ammonium acetate and 20 mM ammonium hydroxide). 656

Mobile phase B at 0% was used at 0-2 min, a linear gradient was used starting from 0% 657 B to 100% B at 2-12 min, kept at 100% at 12-26 min, then from 100% B to 0% B at 26-658 27 min and kept at 0% B for 8 min. The peaks shown in full scans were processed to 659 locate and identify the cleavage products of the 2',5'-p 3 A 3 substrate using the Xcalibur 660 software v4.1. Standard adenosine, AMP, ATP and adenosine-2',3'-cyclic 661 monophosphate sodium salt were run for reference. 662

The PDE domain sequences from different 2',5'-PEs were used for creating a multiple 665 sequence alignment using MAFFT version 7 (54) Experiments were performed three times (n=3) and bars represent the standard error of 819 mean. Statistical significance was calculated using unpaired t test (n=3; *, P value < 820 0.05; **, P < 0.005;***, P < 0.001; ns, not significant) in GraphPad Prism (9.0.0) 821 software. 822 

Interferons at age 50: past, current and future impact on biomedicine

Antiviral actions of interferons

Forty years with coronaviruses

SARS-CoV-2 induces double-stranded RNA-mediated innate immune 901 responses in respiratory epithelial derived cells and cardiomyocytes

Identification of genes 904 differentially regulated by interferon alpha, beta, or gamma using oligonucleotide 905 arrays

Synthesis of low molecular weight 907 inhibitor of protein synthesis with enzyme from interferon-treated cells

1978. pppA2'p5'A2'p5'A: an inhibitor of protein synthesis 910 synthesized with an enzyme fraction from interferon-treated cells

The 913 oligoadenylate synthetase family: an ancient protein family with multiple antiviral 914 activities

Genomic structure of the mouse 2',5'-916 oligoadenylate synthetase gene family

Activation of RNase L in 918

Egyptian Rousette Bat-Derived RoNi/7 Cells Is Dependent Primarily on OAS3 919 and Independent of MAVS Signaling. mBio 10

Inhibition of cell-free protein synthesis by 921 pppA2'p5'A2'p5'A: a novel oligonucleotide synthesized by interferon-treated L cell 922 extracts

Expression cloning of 2-5A-dependent 924

RNAase: a uniquely regulated mediator of interferon action

A study of the interferon antiviral mechanism: apoptosis 927 activation by the 2-5A system

Interferon action and apoptosis 930 are defective in mice devoid of 2',5'-oligoadenylate-dependent RNase L

RNase L activates the NLRP3 inflammasome during viral infections

Ribonuclease L mediates the cell-937 lethal phenotype of double-stranded RNA editing enzyme ADAR1 deficiency in a 938 human cell line. Elife 6

941 OAS-RNase L innate immune pathway mediates the cytotoxicity of a DNA-942 demethylating drug

945 Epigenetic therapy induces transcription of inverted SINEs and ADAR1 946 dependency

Detection of novel 948 members, structure-function analysis and evolutionary classification of the 2H 949 phosphoesterase superfamily

Viral phosphodiesterases that antagonize 951 double-stranded RNA signaling to RNase L by degrading 2-5A

Structural and 954 mechanistic basis for preferential deadenylation of U6 snRNA by Usb1

Antagonism of the interferon-induced OAS-RNase L pathway by 958 murine coronavirus ns2 protein is required for virus replication and liver 959 pathology

Lineage A Betacoronavirus NS2 962 Proteins and the Homologous Torovirus Berne pp1a Carboxy-Terminal Domain 963

Are Phosphodiesterases That Antagonize Activation of RNase L

Middle East Respiratory Syndrome 966

Homologous 2',5'-phosphodiesterases from disparate RNA 969 viruses antagonize antiviral innate immunity

Phosphodiesterase Activity in Inhibiting RNase L Signaling and Contributing to 974 Intestinal Viral Replication In Vivo

Murine AKAP7 has a 2',5'-phosphodiesterase 977 domain that can complement an inactive murine coronavirus ns2 gene

MERS-CoV 981 4b protein interferes with the NF-kappaB-dependent innate immune response 982 during infection

Rotavirus open cores catalyze 984 5'-capping and methylation of exogenous RNA: evidence that VP3 is a 985 methyltransferase

Predicted structure and 987 domain organization of rotavirus capping enzyme and innate immune antagonist 988 VP3

Analysis of rotavirus 990 species diversity and evolution including the newly determined full-length 991 genome sequences of rotavirus F and G

Identification of 2'-phosphodiesterase, which plays a 995 role in the 2-5A system regulated by interferon

Characterization of human phosphodiesterase 12 and identification of 998 a novel 2'-5' oligoadenylate nuclease -The ectonucleotide 999 pyrophosphatase/phosphodiesterase 1

Molecular mechanisms and cellular functions of 1004 cGAS-STING signalling

Cellular magnesium homeostasis

Intrinsic molecular activities of the interferon-induced 2-5A-dependent RNase

Suppressing PARylation by 1012 2',5'-oligoadenylate synthetase 1 inhibits DNA damage-induced cell death

The 2'5' oligoadenylate 1015 synthetase has a multifunctional 2'5' nucleotidyl-transferase activity

Gene structure and function of 1018 the 2'-5'-oligoadenylate synthetase family

Synthesis, characterisation and biological significance 1020 of (2'-5')oligoadenylate derivatives of NAD+, ADP-ribose and 1021 adenosine(5')tetraphospho(5')adenosine

Ribonuclease L and metal-ion-independent endoribonuclease cleavage sites in 1024 host and viral RNAs

Identification of a 1026 novel member of 2H phosphoesterases, 2',5'-oligoadenylate degrading 1027 ribonuclease from the oyster Crassostrea gigas

Small self-RNA generated 1029 by RNase L amplifies antiviral innate immunity

TLR8 Is a Sensor of RNase T2 Degradation Products

Ten Strategies of Interferon Evasion by Viruses

The Role of Phosphodiesterase 12 (PDE12) as a Negative 1039 Regulator of the Innate Immune Response and the Discovery of Antiviral 1040 Inhibitors

Human ADAR1 Prevents 1043

p53 cooperates with DNA methylation and 1047 a suicidal interferon response to maintain epigenetic silencing of repeats and 1048 noncoding RNAs

Crystal structure of 1055 the 2'-specific and double-stranded RNA-activated interferon-induced antiviral 1056 protein 2'-5'-oligoadenylate synthetase

Monitoring activation of ribonuclease L 1058 by 2',5'-oligoadenylates using purified recombinant enzyme and intact malignant 1059 glioma cells

MAFFT multiple sequence alignment software 1061 version 7: improvements in performance and usability

Human 2'-1064 phosphodiesterase localizes to the mitochondrial matrix with a putative function 1065 in mitochondrial RNA turnover

We thank Dr. Renliang Zhang of Mass Spectrometry Core, Lerner Research Institute, 707Cleveland Clinic for performing LC/MS/MS and to Kristin Ogden (Vanderbilt University)