key: cord-1003069-siquvpty
authors: Sun, Jialei
title: Discovery of five HIV nucleoside analog reverse-transcriptase inhibitors (NRTIs) as potent inhibitors against the RNA-dependent RNA polymerase (RdRp) of SARS-CoV and 2019-nCoV
date: 2020-11-01
journal: bioRxiv
DOI: 10.1101/2020.11.01.363788
sha: 114e9fb8bda9346bd0621de148f455653c5138b9
doc_id: 1003069
cord_uid: siquvpty

The outbreak of SARS in 2002-2003 caused by SARS-CoV, and the pandemic of COVID-19 in 2020 caused by 2019-nCoV (SARS-CoV-2), have threatened human health globally and raised the urgency to develop effective antivirals against the viruses. In this study, we expressed and purified the RNA-dependent RNA polymerase (RdRp) nsp12 of SARS-CoV and developed a primer extension assay for the evaluation of nsp12 activity. We found that nsp12 could efficiently extend single-stranded RNA, while having low activity towards double-stranded RNA. Nsp12 required a catalytic metal (Mg2+ or Mn2+) for polymerase activity and the activity was also K+-dependent, while Na+ promoted pyrophosphorylation, the reverse process of polymerization. To identify antivirals against nsp12, a competitive assay was developed containing 4 natural rNTPs and a nucleotide analog, and the inhibitory effects of 24 FDA-approved nucleotide analogs were evaluated in their corresponding active triphosphate forms. Ten of the analogs, including 2 HIV NRTIs, could inhibit the RNA extension of nsp12 by more than 40%. The 10 hits were verified which showed dose-dependent inhibition. In addition, the 24 nucleotide analogs were screened on SARS-CoV primase nsp8 which revealed stavudine and remdesivir were specific inhibitors to nsp12. Furthermore, the 2 HIV NRTIs were evaluated on 2019-nCoV nsp12 which showed inhibition as well. Then we expanded the evaluation to all 8 FDA-approved HIV NRTIs and discovered 5 of them, tenofovir, stavudine, abacavir, zidovudine and zalcitabine, could inhibit the RNA extension by nsp12 of SARS-CoV and 2019-nCoV. In conclusion, 5 FDA-approved HIV NRTIs inhibited the RNA extension by nsp12 and were promising candidates for the treatment of SARS and COVID-19.

Severe acute respiratory syndrome coronavirus (SARS-CoV) and 2019-nCoV CoV-2) belong to Betacoronavirus genus, Coronaviridae family and they have a single-39 stranded, positive-sense RNA genome which is approximately 30kb [1, 2] . SARS-CoV 40 originates from bat and is transmitted to human through an intermediate host, palm 41 civet [3, 4] . 2019-nCoV is also widely accepted to have a bat origin as it is highly similar 42 to a bat coronavirus RaTG13 throughout the genome with 96.2% identity [5] . The high 43 identity suggests a close relationship between the two viruses and there is no 44 recombination event for both of them after they separated from each other evolutionally 45

[4]. However, it is still unclear whether and when an intermediate host have been 46 involved in the zoonotic transmission to human. Soon after the outbreak, pangolin was 47 proposed to be the intermediate host when a pangolin-CoV was discovered which 48 shared 91.02% identity with 2019-nCoV on whole genome level [6, 7] . The discovery 49 provides evidence for the presence of an intermediate host, but 91.02% identity is still 50 too low to suggest a direct transfer from the pangolin to human. Recently, it is reported 51 that ACE2 from multiple animals can bind the spike protein of 2019-nCoV and support 52 infection [8, 9] , especially the ACE2 from monkeys which is identical to and binds as 53 efficient as human [8, 10] . These studies also provides evidence for the involvement 54 of an intermediate host. In addition, studies have revealed that 2019-nCoV has been 55 under different types of natural selection. Its genome contains uncommon high ratio of 56 neutral mutations which are featured by the predominance of C-U substitution [11] [12] [13] . 57

The neutral mutations suggested there was strong selection pressure on 2019-nCoV 58 and raised the possibility of that the virus stayed in a single type of host, possibly bat, 59 for a very long period. The spike of 2019-nCoV appears to have been optimized for MAL-c5x-nsp12-His was transformed into E. coli BL21 (DE3) (Transgen, CD601-02) 271 and protein expression was induced by an auto-induction system (71757-5, EMD 272 Millipore). Upon overnight induction at 25°C, bacterial cell culture was collected and 273 lysed with lysis buffer (20 mM Tris pH 8.0, 500 mM NaCl, 10% glycerol, 0.5% triton X-274 100, 2 mM MgCl2, 0.1 mM ZnCl2, 10 mM DTT, and 0.5X protease inhibitor cocktail, HY-275 K0010, MCE). Cell lysis was centrifuged at 5000 g for 30 min at 4°C to remove cell 276 debris. Supernant was then applied to amylose resin (E8021L, NEB) for 3-5 h at 4°C 277 to bind MBP-nsp12. Resin was then washed three times with washing buffer (20 mM 278 Tris pH 8.0, 500 mM NaCl, 10% glycerol, 1% triton X-100, 1 mM DTT). To release 279 nsp12 from MBP tag, the resin was treated with factor Xa protease (P8010L, NEB) 280 overnight at 4°C. Supernant was collected and applied to a Ni-NTA resin (88222, 281 Thermo) for 1-2h at 4°C to bind nsp12. The resin was washed three times with washing 282 buffer and nsp12 was eluted with elution buffer (20 mM Tris pH 8.0, 50 mM NaCl, 10% 283 glycerol, 0.1% triton X-100, 1 mM DTT, 300 mM imidazole). The concentration of 284 glycerol was then adjusted to 50% and nsp12 stock was stored at -20°C. Nsp7, nsp8, Proteins in this study were expressed in Escherichia coli with MBP at N-terminus and 292 6X His at C-terminus. MBP was removed by factor Xa protease to produce His-tagged 293 proteins for downstream assays. To simplify, his tag was omitted from the protein 294 names. Figure 1A showed the purified SARS-CoV and 2019-nCoV proteins on SDS-295 PAGE and the targets to purify were indicated by stars (*). The identities of proteins 296 were confirmed by Mass spectrometry (Supplementary Table 1 ). Production of 297 SARS-CoV nsp12 per liter was low compared to nsp7, nsp8 and nsp13. Nsp12 had a 298 large size (100 KD) which must have affected its correct folding during expression. The 299 bulky size must also have restricted the access of factor Xa protease to its cleavage site which was located between MBP and nsp12, leading to the decrease of cleavage 301 efficiency. These could explain the low production observed for nsp12. The production 302 of 2019-nCoV nsp12 was low as well, which was consistent with SARS-CoV. In 303 addition, as shown in the figure, there were 5 extra protein bands for nsp12 which were 304 (a) MBP-nsp12, (b) 60 KD chaperone protein GroEL, (c) MBP, (d) elongation factor Tu, 305

and (e) 30S ribosomal protein S3, as identified by Mass. GroEL is a major chaperone 306 for protein folding in E. coli [86] , it might have bound to nsp12 during expression to 307 promote and maintain the correct folding of nsp12. Production of nsp7, nsp8 and nsp13 308 was relatively abundant and nsp8 showed an alternative product at 10 KD, indicated 309 by (#), which should be due to non-specific cleavage by factor Xa protease. The 310 alternative product was confirmed by Mass as well. 311

It has been reported that the RNA polymerase activity of SARS-CoV nsp12 is primer 314 dependent [59]. Therefore, upon purification, a Cy5.5-labeled RNA primer was 315 annealed to its RNA template to form an RNA primer-template (P/T) complex and the 316 activity of nsp12 was determined using the P/T complex. The condition for activity 317 determination contained 50 mM KCl or NaCl combined with various concentrations of 318 MgCl2 or MnCl2. Active nsp12 would extend the primer and fully extended product 319 would have 9 more nucleotides (nt). As shown by Figure 1B , consistent with a previous 320 study, nsp12 only showed activity with distributive product bands in the presence of 321 Mn 2+ , while very low activity was observed for Mg 2+ [58] . However, as Mg 2+ had a higher 322 concentration and was the predominant catalytic metal for DNA/RNA polymerases 323 intracellularly [87], we did not continue the study with Mn 2+ which did not represent the 324 real physiological condition. For Mg 2+ , the activities of nsp12 were evaluated for both 325 K + and Na + . Both showed very weak activities with faint bands indicated by red (>). As 326 extension lengths of the bands were close to the predicted 9 nt, we initially thought 327 they were real template-dependent primer extension activities. However, when we 328 further performed studies with ATP only instead of all 4 rNTPs, nsp12 showed same 329 extension products ( Figure S1 ). This activity was not detected for UTP, GTP, CTP or their combinations. Therefore, nsp12 could utilize ATP as the sole source for primer 331 extension under the conditions tested and the activities observed for Mg 2+ was poly A 332 activity, not template-dependent primer extension. To identify the real activity, we then 333 determined the activity of nsp12 in the presence of nsp7 and nsp8 ( Figure 1C) , as 334 nsp12 has been reported to form complex with nsp7 and nsp8 to be active. As 335 compared to nsp12 alone, the combination with nsp7 and nsp8 had similar level of 336 products, suggesting that nsp7 and nsp8 did not significantly enhance the activity of 337 nsp12 in the condition tested. Nsp7, nsp8 and combination of the two were not active 338 completely. Taken together, these results suggested nsp12 had low activities towards 339 the double-stranded RNA. Despite of repeatings under various conditions for months, we were not able to improve 343 the activity with the double-stranded RNA P/T complex, possibly due to the low 344 concentration of nsp12 (16 nM) which was much lower than reported studies [59, 88] . 345 Then we noticed that, in the presence of K + , there was another weak product band 346 ( Figure 1B) , which was indicated by red star (*). The product had extended 7 nt as 347 compared to the distributive bands catalyzed by Mn 2+ . As the product was weak and 2 348 nt shorter than fully extended product which was 9 nt, we hypothesized that the activity 349 of nsp12 was hindered by the helix force of double-stranded RNA and it needed to 350 couple with helicase to achieve high activity and long extension. To test , we purified 351 the helicase nsp13 of SARS-CoV ( Figure 1A ) which had been reported to have 352 protein-protein interaction with nsp12 [89-91]. Following the purification, we designed 353 a stem loop-forming P/T complex as well as the P/T complex used above with perfect 354 base-pairings for the determination of nsp12 activity in combination with nsp13 ( Figure  355 2A). The molar ratios of nsp12 and nsp13 in the combination were 1:13 and 1:26. 356

Reaction with 50% glycerol was used as control. As shown by Figure 2B , upon 357 combination, the activity of nsp12 was enhanced by nsp13 in a dose-dependent 358 manner. The enhancement was especially obvious for the P/T complexes with stem-359 loop structure. This result suggested nsp13 did enhance the activity of nsp12. However, 360 extension length of the product was still 7 nt instead of the predicted 9 nt. This 361 observation was against the coupling model between nsp12 and nsp13, as the helix 362 unwinding by nsp13 should lead to longer extension of nsp12. Then, we realized that 363 nsp12 was active towards the single-stranded RNA primer. The enhancement of nsp12 364 activity observed was due to the release of more single-stranded primer by nsp13 from 365 the double-stranded P/T complex. To confirm, the assay was performed with the single-366 stranded primer without the annealing to a template and almost full activity was 367 observed for nsp12 ( Figure 2B ). This type of activity of nsp12 should be due to the 368 back-priming of RNA primer of which the 3' end bends back and pairs within the primer 369 to form an extendable hairpin structure, which has been reported for nsp12 previously 370 [63, 88] . Taken together, these results showed that nsp12 had high activity towards the 371 single-stranded RNA primer. In addition, nsp8 and nsp7+nsp8 complex also showed 372 high activities which was not surprising as nsp8 was presented to be the primase for 373 SARS-CoVs [60, 92]. Interestingly, nsp13 and nsp7 also showed weak activities which 374 possibly had relationship with their abilities to bind RNA. As shown in Figure 1B , the 375 product had 7 nt extension, we analyzed the primer sequence (Cy5.5-5′-376 AACGUCUGUUCGCAAAAAGC-3′) and found that the 5'-AGC-3' at 3'-terminus could 377 form a double parings with 3'-UUG-5' in the sequence. The full extension for this paring 378 was 7 nt (5'-AGACGUU-3') which had 2 A, 2 G, 2 U and 1 C ( Figure 2C ). Therefore, 379 the sequence of the extended product was predicted to be Cy5.5-5′-380 AACGUCUGUUCGCAAAAAGCAGACGUU -3′, which formed a stem-loop structure. 381 382 Catalytic metals were required for nsp12 activity 383

To characterize nsp12, we first performed time-course study to analyze the activity of 384 nsp12 at different time-points using the single-stranded RNA primer ( Figure 3A) . 385 Extended product could be observed as early as 5 min, suggesting that the RNA 386 extension by nsp12 was very fast. The product increased in a time-dependent manner 387 with the decrease of primer. At 30 min, significant amount of product could be observed. 388

At 120 min, almost full activity was observed with little primer left. This study confirmed 389 the activity of nsp12 observed in Figure 2B and suggested the extension of the single- This result showed that a wide range of concentrations of Mg 2+ or Mn 2+ could support 401 the optimal activity for nsp12. To our surprise, the control without adding Mg 2+ or Mn 2+ 402 also showed activity, though sub-optimal. This should be due to the endogenous 403 catalytic metals introduced by protein purification buffers or other reagents used in the 404 assay. To prove this hypothesis, we repeated the assay in the presence of 0.5 mM 405 EDTA ( Figure 3C ), which could chelate catalytic metals in 1:1 ratio. As expected, 406 activity of the control without Mg 2+ or Mn 2+ was blocked completely. The activities of 407 Mg 2+ at low concentrations (62.5-250 μM) were also blocked while optimal activities 408 were observed from 0.5 to 2 mM. For Mn 2+ , activity was blocked at 62.5 μM and optimal 409 activity was observed at 1 mM. The activity blocking by EDTA suggested that the 410 activity in the control was due to endogenous catalytic metals and nsp12 did require 411 Mg 2+ or Mn 2+ for RNA extension. Furthermore, we performed the assay without adding 412 Mg 2+ or Mn 2+ and titrated the concentration of endogenous catalytic metals using 413 serially diluted EDTA ( Figure 3D ). As compared to control without EDTA, activities 414 remained optimal from 1 nM to 10 μM, suggesting EDTA in the range was insufficient 415 to chelate the metals. However, at 100 μM, the activity of nsp12 was fully blocked. As 416 EDTA blocked metals in 1:1 ratio, this result suggested that the concentration of 417 endogenous catalytic metals was below 100 μM. When Mg 2+ was re-introduced into 418 the assay system, activity was re-observed. Taken together, these results suggested 419 that catalytic metals were required for nsp12 activity and optimal activity could be maintained in a wide range of concentrations of Mg 2+ or Mn 2+ , including physiological 421

The activity of nsp12 was K + -dependent 424

Although monovalent cations were not catalytic to DNA/RNA polymerases, they did 425 play an important role in maintaining ion strength which was also essential for 426 polymerase to achieve optimal activities [96, 97]. In addition, monovalent cations also 427 affect the stability of RNA tertiary structure [98-100]. To characterize how monovalent 428 cations would affect nsp12, we determined the activity of nsp12 at various 429 concentrations of K + and Na + ( Figure 3E ). As compared to the enzymatic control which 430 was performed with 50% glycerol instead of nsp12, RNA extension activities were 431 observed for K + , but not for Na + . The extension activities increased in a dose-432 dependent manner from 10 to 50 mM and maintained optimal from 50 to 150 mM. 433

At 200 mM, the activity started to decrease. In contrast, for Na + , no extension activity 435 was detected throughout the concentrations evaluated (10 to 200 mM). However, 436 abundant pyrophosphorylation products were observed, indicating that nsp12 was also 437 active in the presence of Na + but the activity only led to pyrophosphorylation. In 438 addition, the control without adding K + or Na + which contained 6 mM NaCl introduced 439 by nsp12 and Cy5.5-RNA stocks, also showed pyrophosphorylation. This result 440 suggested that nsp12 depended on K + to have RNA extension activity and optimal 441 activity required high K + concentration. Assays for the evaluation of chain termination were usually non-competitive, in which 466 a nucleotide analog was present but its corresponding natural rNTP, either ATP, UTP, 467 GTP and CTP was absent, which forced the incorporation of the analog. In our 468 competitive assay, all 4 rNTPs were present. Therefore, the generation of fully 469 extended product was possible, even in the presence of a nucleotide analog. These concentration for the 24 nucleotide analogs, except for emtricitabine which was 479 screened at 2 mM due to its low stock concentration. ATP, UTP, GTP and CTP at 4 mM were used as controls. Their activities were averaged and defined as 100%. Relative 481 activity upon nucleotide analog treatment was normalized to the average and percent 482 of inhibition was calculated. Figure 4B showed the extended products by nsp12 upon 483 treatment with the 24 nucleotide analogs. Figure 4C showed the percent of inhibition 484 of the top 10 hits in the screening. Full list of inhibition by the 24 nucleotide analogs 485 was shown in Table 1 The top 10 hits were verified at 1, 2 and 4 mM, using the same assay as screening 508 ( Figure 5A ). All hits showed significant decrease of extension product in a dose-509 dependent manner, except for ganciclovir which showed similar level of extension product to ATP control at 4 mM. In addition, as an approved treatment for COVID-19, 511 remdesivir was also included in the verification assay which confirmed its inhibition on 512 nsp12 at 4 mM as observed in the screening. Inhibition on pyrophosphorylation for 513 clofarabine and 2'-C-M-GTP at 4 mM was also confirmed. Taken together, this study 514 supported the validity of the screening and confirmed all the hits in the screening, 515 except for ganciclovir, had obvious inhibition on nsp12. 516 517 Discovery of stavudine and remdesivir as nsp12-specific antivirals 518

As the competitive screening assay developed in this study was novel, we tried to 519 prove the validity of the assay from different perspectives. One good way is probably 520 to prove the hits in the screening were specific to nsp12. In Figure 2B , nsp8 was shown 521 to possess comparable RNA extension activity to nsp12, although it was tested at a 522 much higher concentration. Nsp8 had been proposed as a primase and certain 523 mutations in nsp8 region could lead to replication deficiency, even lethality, for SARS-524

CoV [88]. Nsp8 was 3-fold smaller in size than nsp12 and lacked the seven 525 conservative motifs A to G, making nsp8 a reasonable control to identify nsp12-specific 526 drugs. We performed screening against nsp8 using the 24 nucleotide analogs ( Figure  527 5B). The hits and their specificity to nsp12 were listed in Table 1 . To our surprise, most 528 of the analogs in nsp8 screening showed similar patterns of inhibition to nsp12. This 529 result either suggested that the inhibition of RNA extension by these nucleotide 530 analogs was independent of nsp12 or nsp8, or suggested that nsp12 and nsp8 might 531 have evolved similar composition in the catalysis active site to better coordinate the 532 RNA replication for SARS-CoV. Interestingly, we did observe that the inhibition of 533 stavudine decreased by 20% from nsp12 to nsp8 and the inhibition of gemcitabine 534 increased by 30%, suggesting that stavudine had more specificity towards nsp12 while 535 gemcitabine had less. In addition, remdesivir, which had 28.44% inhibition against 536 nsp12, showed minimal inhibition against nsp8, suggesting that remdesivir was a 537 nsp12-specific drug completely. The specificity of stavudine and remdesivir to nsp12 538 did support the hits validity of the screening. To note, nsp12 is the major RdRp for 539 SARS-CoV and a specific drug may be favorable as they could improve efficacy and 540 reduce side effects. However, the efficacy can also be enhanced if a drug had dual 541 inhibition on nsp12 and nsp8. Therefore, the comparison between nsp12 and nsp8 542 screenings was only to prove the validity of the screening and not necessarily to be 543 able to identify a more potent hit. in the screening library due to commercial availability or delivery issues. We managed to purchase all 8 NRTIs and tested them on nsp12 in their corresponding active 571 triphosphates. Clofarabine, an anticancer drug, was included as treatment control. A 572 primer control without nsp12 was also included. As shown by Figure 7A , in addition to 573 stavudine (44.64%) and tenofovir (43.23%), another 3 NRTIs, abacavir, zidovudine and 574 zalcitabine, were identified to be effective inhibitors against SARS-CoV nsp12, with 575 50.09%, 34.62%, and 89.67% inhibition, respectively ( Figure 7C) . The nsp12 of 2019-576 nCoV showed identical results ( Figure 7B) , with 53.05%, 37.56% and 84.69% 577 inhibition for the 3 NRTIs, respectively ( Figure 7C) . The inhibition by zalcitabine was 578 almost complete at 4 mM. In addition, though not regarded as a hit, lamivudine also 579 showed low level of inhibition on both viruses (20.39% and 15.02%, respectively). To 580 verify the inhibition of the 3 newly identified NRTIs, they were evaluated at three 581 concentrations (1, 2 and 4 mM), with ATP and tenofovir used as controls (Figure 8) . 582 The inhibition of all 3 NRTIs were confirmed at 4 mM and zalcitabine also showed 583 inhibition at 2 mM. Taken together, these results suggested that 5 HIV NRTIs, tenofovir, 584 stavudine, abacavir, zidovudine and zalcitabine, were effective inhibitors against the 585 nsp12 of both SARS-CoV and 2019-nCoV. 586 587

The dependence of nsp12 activity on K + and divalent catalytic metals 589

In this study, we expressed and purified active SARS-CoV nsp12 which could 590 efficiently extend a single-stranded RNA. The activity depended on K + while Na + led to 591 pyrophosphorylation and optimal activity was observed at concentrations close to the 592 physiological K + concentration (140 to 150 mM) [101]. To our knowledge, this is the 593 first time such dependence had been reported for nsp12. Previously, nsp12 alone was 594 reported to be inactive [65, 88] , which possibly could be explained by the conditions in 595 which the activities were determined, either with Na + or low concentration of K + . In our 596 study, the activity of nsp12 was also dependent on Mg 2+ or Mn 2+ , but due to its 597 intracellular predominance [93] , Mg 2+ should be the major catalytic metal for nsp12 598 physiologically. Interestingly, Mn 2+ is mutagenetic and promotes mis-incorporation 599 during RNA synthesis. Viral RdRps in general had low replication fidelity and a relative 600 high mutation rate. Therefore, it is possible that nsp12 may take advantage of Mn 2+ as 601 a cofactor to enhance mutation during replication by occasionally involving it into the 602 catalysis active site. Furthermore, our study proved that nsp12 alone could be fully 603 active at low concentration (16-32 nM) without the requirement of nsp7 and nsp8. The 

To discover nucleotide analogs against nsp12, a competitive assay was developed 613 which provided a rapid screening method to identify drugs with inhibition on nsp12. 614 Assays in previous studies were usually based on chain termination in which the 615 corresponding rNTP of a nucleotide analog, either ATP, UTP, CTP or GTP, was absent 616 to determine the efficiency of the analog to be incorporated into RNA chain and the 617 efficiency to cause chain termination. In contrast, we developed a competitive assay. 618

All the 4 rNTPs were present as well as a nucleotide analog. The nucleotide analog 619 would compete with its corresponding rNTP, which resembled the intracellular 620 condition during nucleotide analog treatment. Instead of showing chain termination 621 with shorter product, the assay showed decrease of the fully extended product. As a 622 catalyst, nsp12 could catalyze both RNA polymerization and the reverse process, 623 pyrophosphorylation. Therefore, it was highly possible that nsp12 could remove chain-624 terminating nucleotide analogs from RNA chain upon incorporation, leading to the 625 absence of a terminated product. To our knowledge, this is the first time that nsp12 has 626 been shown to be associated with such property which provides a new perspective 627 about how nsp12 replicates SARS-CoVs genome intracellularly. 628 629

In this study, we performed drug screening and identified 10 nucleotide analogs 632 with >40% and 14 with >20% inhibition on nsp12, including remdesivir, gemcitabine 633 and tenofovir. The screening concentration was 4 mM which was high. This should be 634 due to the short extension (7 bp) of the RNA product and high concentration of 635 nucleotide analogs had to be involved to show inhibition. SARS-CoVs have a genome 636 of 30 kbp and nucleotide analogs would be able to achieve similar levels of inhibition 637 at much lower concentrations intracellularly. Thus, further evaluation in cell culture is 638 necessary to better determine the effective concentrations of these analogs. But for 639 primary screening, this assay is sufficient for hits identification. In the screening, 640 remdesivir showed 28.44% inhibition on nsp12 and its effects has been proven in cell 

In this study, we identified 5 HIV NRTIs to be effective inhibitors against SARS-CoV 652 and 2019-nCoV. Among them, tenofovir, abacavir and zidovudine have superior safety 653 profiles than stavudine and zalcitabine [106, 107]. Zalcitabine has been discontinued 654 since 2006. Abacavir and stavudine share the same sugar backbone with a carbon-655 carbon double bond. They probably can be used as an alternative to each other. But 656 as abacavir has a superior safety profile, it would be the first choice. Therefore, 657 tenofovir, abacavir and zidovudine are the top 3 hits we would like to propose as the 658 candidates to be further investigated for COVID-19 treatment. In addition, as proven 659 by antiretroviral therapies, drug combination is a powerful approach to improve therapy efficacy and solve drug resistance issues. Therefore, we would also like to propose the 661 combinations of tenofovir, abacavir and zidovudine to be evaluated for COVID-19 662 treatment. During SARS outbreak, receiving highly active antiretroviral therapy 663 (HAART) was reported to be a protection factor against the virus [108]. In a recent 664 cohort study, taking tenofovir disoproxil fumarate (TDF) combined with emtricitabine In this study, we expressed and purified active SARS-CoV nsp12 which could 681 efficiently extend single-stranded RNA in a K + and Mg 2+ -dependent manner. We 682 developed a competitive assay for antiviral screening of nucleotide analogs against 683 nsp12 and identified 10 hits with more than 40% inhibition. We also discovered that 684 stavudine and remdesivir were specific antiviral to nsp12. In addition, 5 FDA-approved 685 HIV NRTIs, tenofovir, stavudine, abacavir, zidovudine and zalcitabine, were identified was annealed to its template. Reaction mix was incubated at 37°C for 2h and analyzed 706 by denaturing Urea-PAGE. The gel was scanned and extended products were 707 indicated by red (*) or (>). (C) The activity of nsp12 were also evaluated in combination 708 with nsp7 and nsp8. Concentrations of nsp12, nsp7 and nsp8 were 16 nM, 400 nM 709 and 400 nM, respectively. 50% Glycerol was used as control. 210 or 420 nM nsp13, 100 μM rNTPs (25 μM each), 10% Glycerol, 6 mM NaCl and 720 0.02% triton X-100. The activities of nsp12, nsp13, nsp7 and nsp8 on the single-721 stranded primer without a template were also determined. 50% Glycerol was used as 722 controls. Reaction mix was incubated at 37°C for 2h. (C) Back-priming mechanism was 723 proposed. The sequence and secondary strucuture of the extended product was 724

predicted. Primer was shown in black and the extension was shown in red. which showed the activity of nsp12 at various time-points towards the single-stranded 731 RNA primer without a template. was observed at 4 mM which was used for the following screening concentration. 753 Water was used as negative control. Remdesivir at 2 mM was also evaluated which 754 showed no inhibition. (B) The inhibition of 24 nucleotide analogs in their corresponding 755 triphosphate forms were screened at 4 mM using the competitive assay. Emtricitabine 756 was screened at 2 mM due to low stock concentration. Same concentration of primer 757 was loaded as negative control. 100 μM rNTPs without adding extra nucleotide analog 758 was also included as assay control. ATP, UTP, GTP and CTP at 4 mM was used as 759 normalization controls and their activities were averaged and normalized to 100%. 760 Relative percent of inhibition was calculated and hits with >40% inhibition were 761 highlighted in bold. Inhibition on pyrophosphorylation was also observed for with >40% inhibition were verified at 1, 2, and 4 mM using the same assay as screening. 767 Dose-dependent decrease of extended RNA product was observed. Remdesivir which 768 had 28.44% inhibition on nsp12 was included as well. ATP was used as normalization 769 control. Same concentration of primer was used as primer control. 100 μM rNTPs 770 without adding extra nucleotide was used as rNTP control. (B) The inhibition on nsp8 771 of the 24 nucleotide analog triphosphates were also screened to identify nsp12-specific 772

inhibitors and their inhibition percentages on nsp8 were compared to nsp12. Drugs 773 with >20% differential inhibition between nsp12 and nsp8 were highlighted in bold. product of the RNA primer by nsp12 upon NRTI treatment was analyzed by Urea-PAGE. 795 Product was quantified and percent of inhibition was calculated. ATP was used as 796 normalization control. Clofarabine, an anticancer drug, was used as treatment control. 797 Assay with 50% glycerol instead of nsp12 was used as primer control. (C) showed the 798 percent of inhibition of the 8 NRTIs, compared to ATP control. SARS-CoV was 799 presented by white bar with slashes. 2019-nCoV was presented by gray bar with dots. 800 801 802 803 Figure 8 Inhibition verification of abacavir, zidovudine and zalcitabine on nsp12. 804 The three newly identified NRTIs, abacavir (carbovir-TP), zidovudine and zalcitabine 805 triphosphates, were verified at three concentrations (1, 2 and 4 mM) for their inhibition 806 on the RNA extension by nsp12, using SARS-CoV as a model. The Cy5.5-labeled RNA 807 primer was used and its extended product as well as pyrophosphorylation product by 808 SARS-CoV nsp12 were visualized. ATP was used as treatment negative control. 809 Tenofovir was used as treatment positive control. Reaction with 50% glycerol instead 810 of nsp12 was used as primer control. For abacavir and zidovudine, inhibition was 811 confirmed at 4 mM. For zalcitabine, inhibition was observed at both 2 and 4 mM. 812 813 814 815 Figure S1 Poly A activity of nsp12 on double-stranded RNA. An RNA primer 816 annealed to a template was used to determine the template-dependent poly A activity 817 of nsp12, with either single rNTP (UTP, CTP, GTP, ATP) or their combinations (UG, 818 UCG, UCGA). The concentration for each rNTP was 250 μM. Water was used as 819 control and a control with 25 μM each of UCGA was also included. The assay was 820 performed with 50 mM either KCl (A) or NaCl (B). ATP alone showed identical 821 extended products as UCGA while UTP, CTP, and GTP showed no extension, and the 822 extension length (9 nt) was close to the template. This study suggested nsp12 had 823 poly A activity. 824 825 

Genome Composition and Divergence of the Novel Coronavirus (2019-829 nCoV) Originating in China

Coronavirus genome structure and replication

Animal origins of the severe acute respiratory syndrome coronavirus: insight 833 from ACE2-S-protein interactions

Discovery of a rich gene pool of bat SARS-related coronaviruses provides 835 new insights into the origin of SARS coronavirus

A pneumonia outbreak associated with a new coronavirus of probable bat 837 origin

Probable Pangolin Origin of SARS-CoV-2 Associated with 839 the COVID-19 Outbreak

Are pangolins the intermediate host of the 2019 novel coronavirus

CoV-2)? PLoS Pathog

Broad and Differential Animal Angiotensin-Converting Enzyme 2 Receptor 843 Usage by SARS-CoV-2

SARS-CoV-2 spike protein predicted to form complexes with host 845 receptor protein orthologues from a broad range of mammals

Rhesus angiotensin converting enzyme 2 supports entry of severe acute 848 respiratory syndrome coronavirus in Chinese macaques

Mutation Patterns of Human SARS-CoV-2 and Bat RaTG13

Coronavirus Genomes Are Strongly Biased Towards C>U Transitions, Indicating Rapid 851 Evolution in Their Hosts

The divergence between SARS-853

CoV-2 and RaTG13 might be overestimated due to the extensive RNA modification

Evidence for host-dependent RNA editing in the transcriptome of 856 SARS-CoV-2. Sci Adv

The proximal origin of SARS-CoV-2

SARS-CoV-2 and bat RaTG13 spike glycoprotein structures inform on 860 virus evolution and furin-cleavage effects

A Multibasic Cleavage Site in the Spike 862 Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike 865 Glycoprotein

Emergence of SARS-CoV-2 through recombination and strong purifying 867 selection

Comparative genomic analysis revealed specific 869 mutation pattern between human coronavirus SARS-CoV-2 and Bat-SARSr-CoV RaTG13

Structural basis of receptor recognition by SARS-CoV-2

Cryo-EM structure of the 2019-nCoV spike in the prefusion 874 conformation

Structure of SARS coronavirus spike receptor-binding domain complexed with 876 receptor

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the 878 ACE2 receptor

The coronavirus spike protein is a class I virus fusion protein: structural 880 and functional characterization of the fusion core complex

An efficient method to make human monoclonal antibodies from 883 memory B cells: potent neutralization of SARS coronavirus

Cross-neutralization of SARS-CoV-2 by a human monoclonal SARS-CoV 886 antibody

A human monoclonal antibody blocking SARS-CoV-2 infection

Human neutralizing antibodies elicited by SARS-CoV-2 infection

Antibody cocktail to SARS-CoV-2 spike protein prevents rapid mutational 892 escape seen with individual antibodies

Potently neutralizing and protective human antibodies against SARS-CoV-894 2

A 193-amino acid fragment of the SARS coronavirus S protein efficiently 896 binds angiotensin-converting enzyme 2

The spike protein of SARS-CoV--a target for vaccine and therapeutic 898 development

SARS-CoV-2 SPIKE PROTEIN: an optimal immunological target for 900 vaccines

Structures of Human Antibodies Bound to SARS-CoV-2 Spike Reveal 902 Common Epitopes and Recurrent Features of Antibodies

Progress and Prospects on Vaccine Development against SARS-CoV-2

Vaccines (Basel)

Receptor-binding domain of SARS-CoV spike protein induces highly potent 907 neutralizing antibodies: implication for developing subunit vaccine

COVID-19 VACCINE TRACKER

Polyprotein cleavage mechanism of SARS CoV Mpro and chemical 911 modification of the octapeptide

The SARS-CoV-2 main protease as drug target

Structures of two coronavirus main proteases: implications for substrate 915 binding and antiviral drug design

Crystal structure of SARS-CoV-2 main protease provides a basis for 917 design of improved alpha-ketoamide inhibitors

Crystallographic and electrophilic fragment screening of the 919 SARS-CoV-2 main protease

The crystal structures of severe acute respiratory syndrome virus main 921 protease and its complex with an inhibitor

Conservation of substrate specificities among coronavirus main 924 proteases

Fast Identification of Possible Drug Treatment of Coronavirus Disease-19 926 (COVID-19) through Computational Drug Repurposing Study

Virtual screening and repurposing of FDA approved drugs 929 against COVID-19 main protease

Potential inhibitors against papain-like protease of 931 novel coronavirus (SARS-CoV-2) from FDA approved drugs

An investigation into the identification of potential inhibitors of SARS-CoV-933 2 main protease using molecular docking study

Targeting SARS-CoV-2: a systematic drug repurposing approach to 935 identify promising inhibitors against 3C-like proteinase and 2'-O-ribose 936 methyltransferase

Computational studies of drug repurposing and synergism of 938 lopinavir, oseltamivir and ritonavir binding with SARS-CoV-2 protease against COVID-19

SARS-CoV) 3CL Protease Inhibitors: Peptidomimetics and Small Molecule Chemotherapy

A Trial of Lopinavir-Ritonavir in Adults Hospitalized with Severe Covid-19

Lopinavir-ritonavir in patients admitted to hospital with COVID-19 946 (RECOVERY): a randomised, controlled, open-label, platform trial. The Lancet

Lopinavir-Ritonavir in the Treatment of COVID-19: A Dynamic 948 Systematic Benefit-Risk Assessment

Arbidol monotherapy is superior to lopinavir/ritonavir in treating COVID-950 19

Triple combination of interferon beta-1b, lopinavir-ritonavir, and ribavirin 952 in the treatment of patients admitted to hospital with COVID-19: an open-label, 953 randomised, phase 2 trial

Lopinavir-ritonavir in severe COVID-19

Biochemical characterization of a recombinant SARS coronavirus nsp12 956 RNA-dependent RNA polymerase capable of copying viral RNA templates

The RNA polymerase activity of SARS-coronavirus nsp12 is primer 959 dependent

The SARS-coronavirus nsp7+nsp8 961 complex is a unique multimeric RNA polymerase capable of both de novo initiation and 962 primer extension

A second, non-canonical RNA-dependent RNA polymerase in SARS 964 coronavirus

Structure of the SARS-CoV nsp12 polymerase bound to 966 nsp7 and nsp8 co-factors

Structure of replicating SARS-CoV-2 polymerase

A structural and primary sequence comparison of the viral RNA-dependent 970 RNA polymerases

Structural Basis for RNA Replication by the SARS-CoV-2 Polymerase

Discovery of an essential nucleotidylating activity associated with a 974 newly delineated conserved domain in the RNA polymerase-containing protein of all 975 nidoviruses

Analysis of therapeutic targets for SARS-CoV-2 and discovery of potential 977 drugs by computational methods

RNA-dependent RNA polymerase of SARS-CoV-2 as a therapeutic target

SARS-CoV-2 and SARS-CoV: Virtual screening of potential inhibitors 981 targeting RNA-dependent RNA polymerase activity (NSP12)

Targeting SARS-CoV-2 Nsp12/Nsp8 interaction interface with approved 983 and investigational drugs: an in silico structure-based approach

CoV-2 RNA dependent RNA polymerase (RdRp): A molecular docking study

Nucleotide Analogues as Inhibitors of Viral Polymerases. bioRxiv

Sofosbuvir as a potential alternative to treat the SARS-CoV-2 epidemic

Nucleotide Analogues as Inhibitors of SARS-CoV-2 Polymerase. bioRxiv

Nucleotide Analogues as Inhibitors of SARS-CoV-2 Polymerase, a Key 994 Drug Target for COVID-19

Triphosphates of the Two Components in DESCOVY and TRUVADA are 996 Inhibitors of the SARS-CoV-2 Polymerase. bioRxiv

A library of nucleotide analogues terminate RNA synthesis catalyzed 998 by polymerases of coronaviruses that cause SARS and COVID-19

Repurposing of clinically developed drugs for treatment of Middle East 1001 respiratory syndrome coronavirus infection

Gemcitabine, lycorine and oxysophoridine inhibit novel coronavirus 1004 (SARS-CoV-2) in cell culture

Tenofovir Disoproxil Fumarate: New Chemical Developments and 1006 Encouraging in vitro Biological Results for SARS-CoV-2

Antiviral Efficacies of FDA-Approved Drugs against SARS-CoV-2 Infection 1009 in Ferrets. mBio

Ribavirin and interferon-beta synergistically inhibit SARS-1011 associated coronavirus replication in animal and human cell lines

Novel coronavirus treatment with ribavirin: Groundwork for an evaluation 1014 concerning COVID-19

Remdesivir for the Treatment of Covid-19 -Final Report

Analysis of Ribonucleotide 5'-Triphosphate Analogs as Potential Inhibitors of 1018

Zika Virus RNA-Dependent RNA Polymerase by Using Nonradioactive Polymerase Assays

GroEL-mediated protein folding: making the impossible, possible

Basic mechanism of transcription by RNA polymerase II

One severe acute respiratory syndrome coronavirus protein complex 1025 integrates processive RNA polymerase and exonuclease activities

Structural Basis for Helicase-Polymerase Coupling in the SARS-CoV

Mechanism of nucleic acid unwinding by SARS-CoV helicase

Analysis of intraviral protein-protein interactions of the SARS 1032 coronavirus ORFeome

Structural analysis of the putative SARS-CoV-2 primase complex

Cellular magnesium homeostasis

Cellular manganese content is developmentally regulated in human 1038 dopaminergic neurons

Concentrations of physiologically important metal ions in glial cells 1040 cultured from chick cerebral cortex

Effect of monovalent cations on the activity of the DNA 1042 polymerase of Escherichia coli B

Stabilization of RNA tertiary structure by monovalent cations

Measurement of the effect of monovalent cations on RNA hairpin 1048 stability

The influence of monovalent cation size on the stability of RNA tertiary 1050 structures

Potassium physiology

Ribavirin-induced mutagenesis across the complete open reading frame 1053 of hepatitis C virus genotypes 1a and 3a

In vitro and in vivo evaluation of ribavirin and pleconaril antiviral activity 1055 against enterovirus 71 infection

Remdesivir Inhibits SARS-CoV-2 in Human Lung Cells and Chimeric 1057 SARS-CoV Expressing the SARS-CoV-2 RNA Polymerase in Mice

Incidence and Severity of COVID-19 in HIV-Positive Persons Receiving 1060 Antiretroviral Therapy : A Cohort Study

A review of the toxicity of HIV medications

Nucleotide Reverse Transcriptase Inhibitors: A Thorough Review

Present Status and Future Perspective as HIV Therapeutics

Consideration of highly active antiretroviral therapy in the 1067 prevention and treatment of severe acute respiratory syndrome

Antiviral activities of type I interferons to SARS-CoV-2 infection

Type 1 interferons as a potential treatment against COVID-19