The adenosine analog prodrug ATV006 is orally bioavailable and has preclinical efficacy against parental SARS-CoV-2 and variants

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus driving the ongoing coronavirus disease 2019 (COVID-19) pandemic, continues to rapidly evolve. Due to the limited efficacy of vaccination in prevention of SARS-CoV-2 transmission and continuous emergence of variants of concern (VOC), orally bioavailable and broadly efficacious antiviral drugs are urgently needed. Previously we showed that the parent nucleoside of remdesivir, GS-441524, possesses potent anti-SARS-CoV-2 activity. Herein, we report that esterification of the 5′-hydroxyl moieties of GS-441524 markedly improved antiviral potency. This 5′-hydroxyl-isobutyryl prodrug, ATV006, demonstrated excellent oral bioavailability in rats and cynomolgus monkeys and exhibited potent antiviral efficacy against different SARS-CoV-2 VOCs in vitro and in three mouse models. Oral administration of ATV006 reduced viral loads and alleviated lung damage when administered prophylactically and therapeutically to K18-hACE2 mice challenged with the Delta variant of SARS-CoV-2. These data indicate that ATV006 represents a promising oral antiviral drug candidate for SARS-CoV-2.


INTRODUCTION
The outbreak of the coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has been ongoing for over 2 years and has resulted in over 440 million confirmed infections and over 6 million reported deaths worldwide as of 4 March 2022 (1). SARS-CoV-2 is a positive-sense, single-stranded RNA virus belonging to the genus Betacoronavirus of the family Coronaviridae (2). Two other members of the same genus, namely, SARS-CoV and Middle East respiratory syndrome coronavirus, have caused outbreaks with substantial fatality rates in 2003 and 2012, respectively (3,4). Given the repeated and accelerating emergence of highly pathogenic coronaviruses, it is increasingly important to develop broadly effective anticoronaviral agents to combat the pandemics of COVID-19 and the future emerging coronaviruses.
Although the coronavirus has a certain proofreading ability (5), it still has a high mutation rate. Progressive mutational change in the virus is therefore inevitable, which leads to the emergence of new variants. At present, the World Health Organization has classified many variants of concern (VOCs) of SARS-CoV-2 (6), such as the Alpha (B.1.1.7), Beta (B.1.351), Delta (B.1.671.2), and Omicron (B.1.1.529) variants. The Omicron and Delta variants are highly contagious and have rapidly become dominant, accounting for more than 99% of recent SARS-CoV-2 infections. The Delta variant may lead to vaccine breakthrough infections associated with higher viral load and longer duration of shedding (7)(8)(9). The Omicron variant has higher transmission capacity and exhibits stronger ability to evade protection conferred by prior infection and currently available vaccines (10)(11)(12)(13); however, the symptoms caused by the Omicron variant are thought to be reduced compared with the Delta variant (14,15). Therefore, with the current limited protection of vaccines against these SARS-CoV-2 variants, there is an urgent need to develop orally available and broadly efficacious anticoronaviral agents.
The RNA-dependent RNA polymerase (RdRp), a key component of the viral replication and transcription machinery, is considered a primary target for development of anti-SARS-CoV-2 drugs (16)(17)(18)(19). Several nucleoside or nucleotide analogs, including remdesivir, molnupiravir (EIDD-2801), favipiravir, and AT-527, originally developed by targeting the RdRp of other RNA viruses, have been repurposed for SARS-CoV-2 (20)(21)(22)(23)(24). Among them, remdesivir is the first approved drug by the U.S. Food and Drug Administration (FDA) for the treatment of patients with COVID-19 (25,26), and it has been shown to reduce the risk of hospitalization or death by 87% when administered early to nonhospitalized, high-risk patients (27). Molnupiravir and a main protease inhibitor, PAXLOVID, were granted emergency use authorization by the FDA for the treatment of nonhospitalized patients with COVID-19 (28,29). However, AT-527 and favipiravir did not meet primary end point in clinical trials for patients with mild to moderate disease (30,31).
In this study, we reported a series of ester prodrugs of GS-441524 with improved antiviral potency. The 5′-isobutyryl derivative ATV006 demonstrated improved oral pharmacokinetic (PK) profiles and potently inhibited the replication of SARS-CoV-2 and its variants. In three different animal models, ATV006 could suppress infection and pathogenesis of SARS-CoV-2. These results indicate that ATV006 is a promising drug candidate for further clinical development for COVID-19.

RESULTS
Design and synthesis of prodrugs of adenosine analogs that target SARS-CoV-2 RNA polymerase Although the adenosine analog GS-441524 was effective against SARS-CoV-2 in vitro and in mouse models, as shown in our previous study, it suffers from poor oral bioavailability, which hampers its further development as an oral drug (34). To overcome this limitation, we devoted our efforts to synthesizing GS-441524 prodrugs by using short-chain fatty acids (SCFAs) or amino acids to mask the polar hydroxyl or amino groups. For this, 21 compounds with different substitutions at the positions of R 1 , R 2 , R 3 , or R 4 of GS-441524 were designed and chemically synthesized (Fig. 1). The antiviral effect of the compounds was initially evaluated using a SARS-CoV-2 replicon system (pBAC-SARS-CoV-2-Replicon-Luciferase), established in our previous work (43), which carries all the genes essential for SARS-CoV-2 The synthetic process of GS-441524 derivatives is shown. R 2 OR 2 , acid anhydrides with carbonyl of R 2 ; rt, room temperature; DCC, dicyclohexylcarbodiimide; THF, tetrahydrofuran. (B) Compounds (cmpd) ATV001 to ATV004 were synthesized from GS-441524 by one-step acylation reactions with related acid anhydride in the presence of 4-dimethylaminopyridine (DMAP) and ethylene dimethacrylate (EDMA). (C) For the synthesis of ATV006-024, 2′,3′-hydroxyl moieties of GS-441524 were protected with acetonide by using 2,2-dimethoxypropane in the presence of sulfuric acid (H 2 SO 4 ). Different aliphatic acids, aromatic acids, or amino acids were reacted with intermediate 1 to product the corresponding esters by DCC/DMAP-mediated condensation, respectively. The subsequent hydrolysis reaction of compound 2 with 6N hydrochloric acid (HCl) yielded desired compounds ATV006 to ATV024.
replication, including RdRp, and the luciferase reporter gene but does not produce infectious virus particles. Promising candidates with improved potency and bioavailability were empirically selected for further testing with live SARS-CoV-2 in cell culture and mouse models after considering their antiviral activity, cytotoxicity, PK properties, and synthetic cost (fig. S1).
We first tested the percentage inhibition in the replicon system ( fig. S2A) and then selected 17 compounds with inhibition rate higher than 90% at 10 M to measure their concentration for 50% of maximal effect (EC 50 ), which ranged from 0.217 to 2.351 M ( fig. S2B and table S1). The compounds ATV001 and ATV002 with isobutyryl amide or acetyl amide at the base moiety showed decreased antiviral activities ( fig. S2, A and B), probably due to the biostable amide group that obstructed hydrogen bond formation between the inhibitor and the RNA template (44,45). Triacetyl esterification of the hydroxyl groups on C5′ (R 1 ), C2′ (R 2 ), and C3′ (R 3 ) positions (ATV003) did not substantially change the activity, whereas the tri-isobutyryl on C5′ (R 1 ), C2′ (R 2 ), and C3′ (R 3 ) (ATV004), which has the same structure to GS-621763 (41,42), and mono-isobutyryl modification of 5′-hydroxyl group (ATV006) improved the inhibitory activities in the replicon system in comparison with its parent GS-441524 ( fig. S2B and table S1). The compounds ATV007-024 contained a R 1 group modified with straight, cyclic, or branched SCFAs; benzyl acyl groups; or amino acid groups. Some of these compounds displayed an improvement in potency relative to GS-441524. Six compounds (ATV019-024) bearing l-or d-amino acid ester were designed to improve drug absorption by targeting the peptide transporter family 1 (PepT1) (46). However, these compounds did not show improved activity against the replicon or improved permeability in Caco-2 cells (table S2). Together, SCFA esterification at the C5′ position could improve the potency relative to GS-441524. Therefore, we selected the top six compounds ranked by low to high EC 50 in replicon system for further analysis of anti-SARS-CoV-2 activity with live virus, including different variants, in cell culture models.

The adenosine analog prodrug ATV006 potently inhibits the replication of authentic SARS-CoV-2 and its VOCs
The antiviral efficacy of the six SCFA prodrugs (ATV006, ATV009-011, ATV013, and ATV017) was evaluated in Vero E6 cells infected with different strains of SARS-CoV-2, including the early strain (B.1) and three SARS-CoV-2 VOC variants (Beta, Delta, and Omicron) (Fig. 2, Table 1, and table S3). The compounds showed improved potency against SARS-CoV-2 as compared with remdesivir and GS-441524, which was consistent with the results of the SARS-CoV-2 replicon system ( Fig. 2 and Table 1). Among them, ATV010 exhibited low micromolar EC 50 value with early strain B.1 and Beta variant. In contrast, ATV006 had an overall >4-fold and >12-fold potency improvement in inhibiting the replication of Delta and Omicron variants, with EC 50 values reaching 0.349 and 0.106 M, respectively. The compounds exhibited similar antiviral activity in Huh7 cells to that in Vero E6 cells ( fig. S3 and table S4). Together, these results indicate that 5′OH-esterified compounds could potently inhibit the replication of SARS-CoV-2 and its variants.
The cytotoxicity of the compounds was evaluated on Vero E6 cells with the CCK-8 assays ( fig. S4). The results showed that most of the compounds had low toxicity with the concentration of 50% cytotoxicity (CC 50 ) values greater than 50 M, except for ATV010, which had a CC 50  ATV006 has favorable PK profiles in rats and cynomolgus monkeys To assess the oral absorption of ATV006, PK studies were conducted in Sprague Dawley rats using GS-441524 as a control. After oral dosing of 25 mg/kg in rats, ATV006 displayed high oral bioavailability (F %) of 81.5%, whereas GS-441524 had poor oral bioavailability with a F % value of 21.7% (Tables 2 and 3), the latter being similar to the data from the National Center for Advancing Translational Sciences (NCATS) of National Institutes of Health (NIH) (47). The maximum blood concentration (C max ) of 8.2 M was achieved 0.5 hours after the oral administration of ATV006, indicating its effective blood exposure to be 3.8-to 30-fold higher than the concentration for 90% of maximal effect (EC 90 ) of ATV006 ( Fig. 3A and Table 2). The C max of 3.4 M was achieved 1 hour after oral administration of GS-441524, indicating that the absorption time of GS-441524 is longer and the final concentration is lower when compared with ATV006 ( Fig. 3B and Table 3). In cynomolgus monkeys, the average C max of 3.71 M was reached at 1.5 hours after the oral administration of ATV006 (10 mg/kg) and the plasma concentration decayed with a half-life (T 1/2 ) of 4 hours ( Fig. 3C and Table 2). The oral bioavailability of ATV006 was about 30% in macaques (Table 2), which had an apparent increase when compared with the 8.3% of GS-441524 previously shown by NCATS of NIH (47). As a compound with an oral bioavailability of greater than 10% has the potential for development as an oral drug (48), ATV006 well met such standard as an oral drug candidate for further testing in animal models.
Next, we explored the tissue distribution of a single oral dose of ATV006 and GS-441524 at 100 mg/kg in rats by measuring its parent nucleoside GS-441524 and its monophosphate and triphosphate forms (Fig. 3, D to F, and Table 4). The results revealed that oral administration of ATV006 achieved broad distribution in the plasma, liver, and kidneys as well as in the lung, the major target organ for SARS-CoV-2 infection. In contrast, after oral administration of GS-441524, the concentrations of GS-441524, remdesivir nucleotide monophosphate (RMP), and remdesivir nucleotide triphosphate (RTP) in various tissues were lower than that of ATV006 (Fig. 3, D to F). Likewise, oral administration of ATV006 showed extensive distribution in the plasma, liver, kidney and lung of C57BL/6 mice ( fig. S5). Together, the PK results showed that ATV006 had improved oral bioavailability and tissue distribution relative to GS-441524.

Orally administered ATV006 could effectively suppress SARS-CoV-2 replication in mouse models
We next investigated the antiviral activity of orally administered ATV006 in two different mouse models of SARS-CoV-2 (strain B.1), one with knock-in of humanized angiotensin-converting enzyme 2 (KI-hACE2) at the mAce2 gene locus (49) and the other with adenovirus-delivered human ACE2 (Ad5-hACE2) (50). The KI-hACE2 mice were intranasally inoculated with SARS-CoV-2 and treated with ATV006 or vehicle control (Fig. 4A). To better determine the degree of SARS-CoV-2 replication, we detected both the genomic RNA (gRNA) and subgenomic RNA (sgRNA), the latter being produced  (51,52). In the control group, both SARS-CoV-2 gRNA and sgRNA reached high concentrations in the lung, indicating that the mouse infection model was well established. In contrast, SARS-CoV-2 gRNA and sgRNA were hardly detectable at 4 days post-infection (dpi) in the ATV006 treatment groups (Fig. 4, B and C), demonstrating robust inhibition of SARS-CoV-2 replication by ATV006.
We further tested the antiviral potency of ATV006 in the Ad5-hACE2 mouse model, which supports SARS-CoV-2 infection and Table 1   pathogenesis in the mouse lung (50). The mice were treated vehicle, ATV006, or EIDD-2801 as a positive control (Fig. 4D). The results showed that ATV006 (250 mg/kg) and EIDD-2801 (500 mg/kg) could reduce the viral load and pathological damage of the lung (Fig. 4, E and F). Together, ATV006 showed potent anti-SARS-CoV-2 efficacy in different mouse models.

Prophylactic ATV006 treatment reduces lung damage and protects K18-hACE2 mice
We next tested the antiviral potency of ATV006 in the widely used K18-hACE2 transgenic mouse model, which is highly susceptible to SARS-CoV-2 infection and can lead to death of infected mice (53). In a prophylactic model, the mice were challenged with the Delta variant ( Fig. 5A). In this model, the control group gradually lost weight beginning on 4 dpi and died starting on the 6 dpi; all control mice died by 7 dpi, but all mice in the treatment groups survived (Fig. 5, B and C). At 3 dpi, we evaluated the abundance of SARS-CoV-2 virus in mouse lung through quantitative real-time polymerase chain reaction (qRT-PCR) and plaque assay. The amount of viral RNA of the treatment groups was lower than that of the control group, with ATV006 (250 mg/kg) group significantly reducing viral RNA compared with the control (P = 0.002; Fig. 5D). For the ATV006 (250 or 100 mg/kg) and EIDD-2801 (500 mg/kg) groups, virus titers were reduced below the limit of detection (LOD) in plaque assays at 3 dpi (Fig. 5D). Histopathological analyses were performed with the lungs of the mice infected with SARS-CoV-2 at 3 dpi. The vehicle-treated mice showed evidence of pathology, including inflammatory cell infiltration ranging from the trachea and peri-alveolar space to the interstitium; in contrast, ATV006-treated animals had reduced evidence of lung pathology (Fig. 5, E and F). Immunohistochemistry staining showed that the spike (S) protein of SARS-CoV-2 was hardly detectable at 3 dpi in the lungs of mice treated with ATV006 (250 mg/kg) (Fig. 5G). Compared with the ATV006 treatment group, the spleen of the mice in the control group was enlarged, and the white pulp was atrophied to varying degrees (Fig. 5, H and I). Furthermore, ATV006 reduced the production of inflammatory cytokines and chemokines in the lung tissues (table S6). Together, these results demonstrated that ATV006 has robust anti-SARS-CoV-2 efficacy in a mouse model of prophylactic treatment.

Therapeutic administration of ATV006 reduces virus titers and lung damage caused by Delta variant infection in K18-hACE2 mice
Next, we tested the therapeutic antiviral efficacy of ATV006 in K18-hACE2 mice (Fig. 6A). The body weight of the mice in each group did not change over the first 3 days after inoculation of the virus (Fig. 6B), which resembled the previous results (Fig. 5B). At 3 dpi, we evaluated the abundance of SARS-CoV-2 virus in mouse lung through qRT-PCR and plaque assay (Fig. 6, C and D). The mice treated with ATV006 at different dosages had reduced viral load in the lungs, and the effect was in a dose-dependent manner. Viral titer in the high ATV006 dosage groups (80 and 150 mg/kg) was reduced below the detection limit at 3 dpi (Fig. 6D). In contrast, treatment with GS-441524 (either 80 or 150 mg/kg) did not inhibit virus replication in this model. Histopathological analyses were performed with the lung tissue isolated from infected mice at 3 dpi. ATV006 (150 mg/kg)-treated mice alleviated the symptoms in the lungs, except for evidence of a small amount of hemorrhage, whereas GS-441524 (150 mg/kg)-and vehicle-treated mice showed multiple sites of injury and inflammatory cell infiltration (Fig. 6E). Together, our results showed that oral administration of ATV006 could efficiently inhibit SARS-CoV-2 replication and ameliorate SARS-CoV-2-induced lung disease in mouse models of both prophylactic and therapeutic treatment. Prophylactic dosing of ATV006 could prevent the death of the K18-hACE2 mice challenged with the Delta variant of SARS-CoV-2. These results demonstrated the potential of ATV006 as an orally bioavailable anti-SARS-CoV-2 drug.

DISCUSSION
COVID-19, caused by SARS-CoV-2, is still spreading globally, threatening human health and economic development. Vaccineinduced or naturally acquired protective immunity necessary to disrupt transmission chains had been hampered by the rapid evolution and emergence of VOCs (8)(9)(10)(11)(12)(13)54). Therefore, effective and broad-spectrum anti-SARS-CoV-2 drugs are desperately needed to treat patients with COVID-19 with breakthrough infections. These infections are increasingly common in vaccinated populations and are driven by VOCs that can evade preexisting immunity. As COVID-19 is an acute infectious disease, antiviral treatment can exert its best effect at the early stage of the infection. In contrast, at the late stage of COVID-19 in patients who are hospitalized, antiinflammatory therapy may be more beneficial in lessening the symptoms. Therefore, orally bioavailable anti-SARS-CoV-2 drugs suitable for outpatients are superior to injectable drugs applied to hospitalized patients.
Among the three drugs currently approved by the FDA for the treatment of COVID-19, remdesivir and molnupiravir target the viral RdRp for their antiviral activity. However, the obligatory Table 2. PK profile of ATV006 in Sprague Dawley rats and cynomolgus monkeys. ATV006 was administered by the indicated route, and the parameters were calculated on the basis of the LC-MS/MS analysis of the parent nucleoside, GS-441524 (n = 3 per group). IV, intravenous; AUC, area under the curve.  intravenous administration of remdesivir limits its clinical application (25). Previously, we reported that GS-441524, the major metabolite and the parent nucleotide of remdesivir, has a better inhibitory activity on SARS-CoV-2 and mouse hepatitis virus A59 (34). However, the unfavorable oral PK prevented its further development into an oral drug (34,47). To address this issue, we synthesized a series of SCFA and amino acid prodrugs of GS-441524 aiming at overcoming these limitations. Among the compounds synthesized, the isobutyryl adenosine analog ATV006 improved oral absorption and potently inhibited the replication of SARS-CoV-2, including the Delta and Omicron variants. Furthermore, compared with remdesivir, ATV006 is structurally simpler and easier to synthesize by a threestep transformation with GS-441524 as starting material, which may reduce the cost and accelerate mass production. After administration of ATV006, it is rapidly hydrolyzed by cellular esterases to produce the parent nucleoside GS-441524 (55,56), which then undergoes three steps of phosphorylation and is transformed to the active triphosphate form, the same active component as that of remdesivir and GS-441524. Therefore, ATV006 and remdesivir have the same mechanism of stalling SARS-CoV-2 polymerase by targeting viral RdRp (45,(57)(58)(59). It is expected that ATV006 could be active against different variants of SARS-CoV-2, as the RdRp is the highly conserved among coronaviruses. Our current study demonstrated that ATV006 has a broad antiviral activity against different variants of SARS-CoV-2. The parent nucleoside GS-441524 was previously reported to be broadly active against viruses that belong to the families Paramyxoviridae, Coronaviridae, and Filoviridae (36)(37)(38)(39)60), indicating the potential for more broad antiviral application of ATV006.

Rat
We tested the anti-SARS-CoV-2 efficacy of ATV006 with cell culture infection and three different mouse models by four independent research groups at three different biosafety level 3 (BSL-3) Table 3. PK profile of GS-441524 in Sprague Dawley rats. GS-441524 was administered by the indicated route, and the parameters were calculated on the basis of the LC-MS/MS analysis of GS-441524 (n = 3 per group).  facilities, and it was noted that the anti-SARS-CoV-2 activity of ATV006 was robust and consistent. Among the three mouse models (KI-hACE2, Ad5-hACE2, and K18-hACE2) (49,50,53), the highly sensitive K18-hACE2 mice were used both in prophylactic and therapeutic models. Although K18-hACE2 mice can suffer fatal viral encephalitis from the neuroinvasion of SARS-CoV-2 (61) and the metabolites of ATV006 were not detectable in the mouse brain, ATV006 could effectively prevent the death of K18-hACE2 mice infected by SARS-CoV-2 Delta variant in the prophylactic model. One possible explanation for such potent efficacy is that prophylactic administration of ATV006 may effectively block SARS-CoV-2 replication early after infection and consequently avoid the neuroinvasion of SARS-CoV-2. However, our study has several limitations. First, SARS-CoV-2 mouse models cannot fully recapitulate human infection and COVID-19

Fig. 5. Prophylactic ATV006 treatment is efficacious against SARS-CoV-2 in K18-hACE2 mice.
(A) The experimental timeline is shown. K18-hACE2 mice were intranasally (i.n.) inoculated with the SARS-CoV-2 Delta variant (1 × 10 3 PFU per mouse) and treated with vehicle (control, n = 11), ATV006 (250 mg/kg, orally, once daily, n = 11), ATV006 (100 mg/kg, orally, once daily, n = 8), or EIDD-2801 (500 mg/kg, orally, once daily, n = 8). (B) The survival curve is shown. (C) Change in body weight was measured over time after infection. (D) Viral load from lung tissue at 3 dpi was analyzed by qRT-PCR and plaque assay. The detection limit of qRT-PCR was 0.5 copies/l. The red dashed line in (D, right) indicates the LOD for the plaque assay. (E to G) Histopathology (E), gross pathology (F), and immunohistochemistry detection of SARS-CoV-2 S protein (G) are shown for lungs isolated from the indicated treatment groups. S protein is denoted by brown staining in (G). (H and I) Gross pathology (H) and histopathology (I) are shown for spleens isolated from the indicated treatment groups. Data in (C and D) are presented as means ± SD. Statistical analysis was conducted using a log-rank test (survival), a one-way ANOVA with Dunnett's correction for multiple comparisons (lung titer), or a Kruskal-Wallis test with Dunn' s correction for multiple comparisons (lung viral RNA). *P ≤ 0.05, **P ≤ 0.005, and ****P ≤ 0.0001. pathogenesis. For example, K18-hACE2 mice can suffer fatal viral encephalitis from neuroinvasion of SARS-CoV-2 in addition to pneumonia (61). Therefore, the antiviral effect of ATV006 shown in mouse models needs to be tested in nonrodent preclinical models and ultimately in human clinical trials. Second, the oral bioavailability of ATV006 was tested in both rat and macaque models, but because of limitations of nonhuman primate experiments, the oral availability of ATV006 was not compared head to head with its parent nucleoside, GS-441524, in macaques. Third, in the therapeutic model of ATV006, we only tested the dosing scheme with first oral administration at 6 hours after infection and only with treatment twice a day for 3 days. Future studies are needed to explore the latest dosing time and lowest dosages of ATV006 that are efficacious. Last, although we have evaluated the antiviral activity of ATV006 against different SARS-CoV-2 variants (including an early strain, B.1, as well as Beta, Delta, and Omicron), the broad-spectrum antiviral effect of ATV006 on other coronaviruses (including other human and animal coronaviruses) was not tested.
Collectively, our results demonstrated that ATV006 is orally bioavailable and has potent efficacy against SARS-CoV-2. This efficacy extended to variants of SARS-CoV-2. These data suggest that ATV006 represents a promising drug candidate for the treatment for COVID-19 and potentially for emerging coronavirus diseases in the future.

Study design
The aim of this study was to find new antiviral compounds that have potent anti-SARS-CoV-2 activity and can be administered orally. For this purpose, we designed and synthesized a series of derivatives of GS-441524, the parent nucleoside analog of remdesivir, by using SCFAs or amino acid modifications to mask the polar hydroxyl or amino groups. We first measured the antiviral activity of these compounds using a SARS-CoV-2 replicon system and further evaluated their anti-SARS-CoV-2 activity with live virus, including the early strain, B.1, and the Beta, Delta, and Omicron variants in Vero E6 and Huh7 cell cultures. Cytotoxicity was assessed in the Vero E6 cell lines. In vitro experiments were repeated at least twice. By considering the anti-SARS-CoV-2 activity, selectivity index, LogP value, and water solubility of the compounds, ATV006 was selected for further studies. PK, tissue distribution, and efficacy studies in mice, rats, and macaques were intended to gain the data for the oral bioavailability, druggability, and in vivo anti-SARS-CoV-2 activity. Three SARS-CoV-2 mouse models were used in this study, including the mice with adenovirus-delivered hACE2 (Ad5-hACE2), KI-hACE2 at the mAce2 gene locus (KI-hACE2), and transgenic K18-hACE2. Mice were age-, sex-, and genetic background-matched and randomized to different groups of at least three animals per group before the start of each experiment. The preventive and therapeutic effects of ATV006 were studied by weight monitoring, virus detection (qRT-PCR, focus forming assay, plaque assay, or immunostaining analysis), histopathology, and survival analysis. Sample size calculation and blinding were not performed.

Compounds, cells, and viruses
The compounds presented here (ATV001-024) were synthesized at our in-house facility, analyzed by 1 H nuclear magnetic resonance (NMR) spectroscopy, 13 C NMR spectroscopy, and high-resolution mass spectrometry, and purified by high-performance liquid chromatography (LC) with a purity of >95%. Human embryonic kidney (HEK) 293T, Caco-2, and Huh7 cells were obtained from American Type Culture Collection. The African green monkey kidney Vero E6 cell line (Vero E6) was provided by H. Zhang (Sun Yat-sen University). HEK 293T, Vero E6, Caco-2, and Huh7 cells were cultured in Dulbecco's modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin at 37°C in a humidified atmosphere of 5% CO 2 .

SARS-CoV-2 replicon assays
The replicon system was reported in our previous work (43). The luciferase assays were performed following the manufacturer's instructions (Promega Corporation). In brief, cells in 24-well plates were transfected with 500 ng of pBAC-SARS-CoV-2-Replicon-Luciferase plasmid and 10 ng of RL-TK plasmid. After 6 to 8 hours of transfection, the supernatant was discarded and replaced with fresh DMEM, followed by addition each compound (described in table S1) to the medium with the final concentration of 50, 10, 5, 2, 1, 0.1, or 0.01 M. After 60 hours, cells were lysed in 200 l of passive lysis buffer. Each lysate (20 l) was transferred into 96-well white plate and then mixed with 20 l of Luciferase Assay Reagent II, followed by 20 l of Stop & Glo solution. The luminescence values of the two-step reaction were recorded using a luminescence detector in Synergy H1 Hybrid Multi-Mode Reader.

Anti-SARS-CoV-2 activity assays in cell culture
Vero E6 and Huh7 cells were seeded at 2 × 10 4 cells per well in 48-well plates. Cells were allowed to adhere for 16 to 24 hours and then infected at multiplicity of infection of 0.05 with SARS-CoV-2 for 1 hour at 37°C. Then, viral inoculum was removed, and cells were washed two times with prewarmed phosphate-buffered saline (PBS). Medium containing dilutions of compounds or dimethyl sulfoxide (DMSO) was added. At 48 hours after infection, supernatants or cells were harvested for qRT-PCR analysis. The dose-response curves were plotted from viral RNA copies versus the drug concentrations using GraphPad Prism 6 software.

qRT-PCR analysis
For SARS-CoV-2 RNA quantification, RNA was isolated by the Magbead Viral DNA/RNA Kit (CWBIO). A SARS-CoV-2 nucleic acid detection kit (Da'an Company) was used to detect the virus.
The LOD of this kit reached 0.5 copies/l. For the detection of the viruses and cytokines in mouse tissues, total RNA was isolated from tissue samples with TRIzol per the manufacturer's instructions. mRNA was reverse-transcribed into cDNA by a PrimeScript RT reagent Kit (Takara). The cDNA was amplified by a fast two-step amplification program using ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co. Ltd) or Taq Pro HS Universal Probe Master Mix (Vazyme Biotech Co. Ltd.). Gapdh was used to normalize the input samples using the Ct method. The relative mRNA expression of each gene was normalized to Gapdh housekeeping gene expression in the untreated condition, and fold change was calculated by the Ct method relative to those in untreated samples. The qRT-PCR primers are listed in table S7.

CCK-8 cell viability assay
To investigate the effect of drugs on cell viability, Vero E6 cells were seeded in 96-well plates at a density of 20,000 cells per well and were treated with drugs at indicated concentrations (0, 0.01, 0.1, 1, 5, 10, 50, 75, 100, and 200 M) for 48 hours. Cell viability was tested by using Cell Counting Kit-8 (CCK-8, Bimake, B34302). The figures were plotted from cell viability percentage versus the drug concentrations using GraphPad Prism 6 software.

pK a determination
The pK a value of test compounds was detected by using a fast ultraviolet (UV) method with Sirius T3 Titrator. First, the blank calibration and fast UV buffer calibration were performed where an empty sample vial was placed in the sample position. The buffer was fully titrated, and reference spectra were collected at all pH values. The sample was then detected with the following process: Methanol was used as a cosolvent to measure the cosolvent dissociation constant (p s K a ). Five microliters of the compound stock solution of 10 mM was added to a sample vial. Three weight/volume ratios of methanol cosolvent were tested, and the total starting volume for titration was set to 1.5 ml. A spectrometer detected the turbidity, and the pH range of titration by a fast UV method was 2 to 12. The titration data were obtained from a proper titration, and the p s K a test result was obtained by the Yasuda-Shedlovsky extrapolation method for the fast UV p s K a method.

Determinations of thermodynamic solubility
Test compounds were weighed out for measuring the solubility (about 1.5 mg each in three separate 1.5-ml glass vials). For each compound, one vial was used to do the standard, and the other two were used to measure the solubility in duplicate. The proper volume of ultrapure water was added to each vial of the solubility sample plate using pipettes. One stir stick was added to each vial, and the vials were sealed using a molded polytetrafluoroethylene (PTFE)/ silicone plug. The solubility sample plate was transferred to an Eppendorf Thermomixer Comfort plate shaker and shaken at 25°C at 1100 rpm for 24 hours. Then, the stir sticks were removed using a magnet, and the samples were transferred from the solubility sample plate into a filter plate with pipettes. All compounds were filtered under a vacuum condition. The filtrate was diluted 1000-fold with a mixture of H 2 O and acetonitrile (1:1 in v/v). The plates were vortexed for 5 min. For standards (STDs), the necessary volume was calculated to dissolve each standard sample to the concentration of 1.5 mg/ml. DMSO was added to each standard sample. The standard vials were covered. The standard plate was placed on the Eppendorf ThermoMixer Comfort plate shaker for 5 min at 25°C and 1100 rpm. After 5 min, each compound was completely dissolved. From the DMSO (1.5 mg/ml) standard plate, 5 l of DMSO STD was transferred into the remaining empty plate, and then, 5 l of ultrapure water and 490 l of a mixture of H 2 O and acetonitrile (1:1 in v/v) were added to that plate to have a concentration of 15 g/ml. From the standard plate (15 g/ml), 50 l was transferred into a new plate, and 450 l of a mixture of H 2 O and acetonitrile (1:1 in v/v) was added to that plate to have a final standard concentration of 1.5 g/ml. The solubility sample plate and standard plate were placed into the well plate autosampler at 4°C and evaluated by LC-tandem mass spectrometry (LC-MS/MS) analysis (LC system, Shimadzu; MS system, Triple Quad 5500 instrument from AB Inc. with an ESI interface). All calculations were carried out using Microsoft Excel.

LogP determination
The stock solutions of test compounds were prepared in DMSO at 10 or 30 mM. The 5 l of stock solution (30 mM) or 15 l of stock solution (10 mM) of each sample was placed into their proper 96-well rack. A volume of 500 l of PBS pH 7.4 was added into each vial of the cap-less LogP plate, followed by the addition of 500 l of 1-octanol. The assay was performed in duplicate. One stir stick was added to each vial and sealed with a molded PTFE/silicone plug. The LogP plate was transferred to the Eppendorf Thermomixer Comfort plate shaker and shook at 25°C at 1100 rpm for 1 hour. After incubation, the stir sticks were removed by a magnet. To separate the phases, the samples were transferred and centrifuged at 25°C at 20,000g for 20 min. The upper (1-octanol) and lower (buffer) phases were separated and transferred to the empty tubes. Aliquots of 5 l were taken from the upper phase, followed by 495 l of methanol. After 1 min of a vortex, aliquots of 50 l were taken from the diluent, followed by 450 l of methanol. Aliquots of 50 l were taken from the lower phase, followed by 450 l of methanol. The samples were evaluated by LC-MS/MS analysis, and all calculations were carried out using Microsoft Excel.

PK study in rat
Male Sprague Dawley rats (180 to 220 g, n = 3) were fasted for 12 hours before drug administration. ATV006 was administered intravenously at 5 mg/kg or orally at 25 mg/kg. Blood samples were collected from the jugular vein into anticoagulant EDTA-K2 tubes at 0.083, 0.25, 0.5, 1, 2, 3, 4, 6, 8, and 24 hours for the intravenous group and at 0.25, 1, 0.5, 2, 3, 4, 6, 8, and 24 hours for the oral group, respectively. All samples were centrifuged under 4000 rpm for 10 min at 4°C, and the plasma samples were collected and stored at −65°C for future analysis. An aliquot of 50 l each plasma sample was treated with 250 l of acetonitrile. The samples were centrifuged under 4000 rpm for 10 min and filtered through 0.2-m membrane filters. The concentrations of analytes in each sample were analyzed by LC-MS/MS.

PK study in cynomolgus monkeys
Three male cynomolgus monkeys (2 to 5 years of age, weighing 3 to 5 kg) were orally administrated AVT006 of 10 mg/kg on day 1. Blood samples for plasma were collected from a jugular vein into anticoagulant EDTA-K 2 tubes at 0.083, 0.25, 0.5, 1, 2, 4, 8, 24, and 48 hours after administration. After a 3-day washout period, each animal was administered intravenously with ATV006 at a dose of 5 mg/kg, followed by blood collection from the jugular vein at the specified time points. All samples were centrifuged under 4000 rpm for 10 min at 4°C, and the plasma was collected and stored at −65°C for future analysis. The concentration of analytes in each sample was analyzed by LC-MS/MS.

Tissue distribution study in rats
Eighteen Sprague Dawley rats were randomly divided into two groups (n = 9), and each group received a single oral administration of either ATV006 (100 mg/kg) or GS-441524 (100 mg/kg). Three rats in each group were anesthetized at 1-, 4-, and 24-hours after dosing. Blood samples were collected into the heparinized tubes that contain PhosSTOP and DTNB [5′-dithiobis (2-nitrobenzoic acid)] and were centrifuged. Tissue samples, including the liver, kidney, and lung, were harvested after perfusion with saline and immediately homogenized for further analysis as previously described (64). Tissue samples were individually homogenized with four equivalent volumes of H 2 O containing PhosSTOP EASYpack (PhosSTOP) (Roche) and DTNB (≥98%) (Sigma-Aldrich). For GS-441524 determination, 40 l of plasma or homogenate was immediately added to 200 l of acetonitrile-methanol (50/50, v/v) that contains 50 nM nadolol, vortexed thoroughly, and centrifuged at 4°C. For RMP and RTP determination, 50 l of plasma or homogenate was immediately added to 150 l of 2% formic acid and 5 l of 10 M 13 C, 15 N-adenosine triphosphate aqueous solution, then vortexed thoroughly, and centrifuged at 4°C. Supernatant in a volume of 180 l was mixed with an equivalent volume of water for solidphase extraction with a weak anion exchange plate (Waters Oasis WAX Elution plate). Last, the eluate was gently evaporated to dryness under a gentle stream of air, and the residue was reconstituted with 100 l of H 2 O for analysis on a Waters Positive Pressure-96 processor.

Tissue distribution study in mice
Five C57BL/6 mice were fasted for 12 to 16 hours before oral administration with a single dose of ATV006 (100 mg/kg). After 1 hour of administration, 0.5 ml of the blood sample was taken from heart. Liver, kidney, lung, and brain tissues were harvested. The blood samples were processed to the same method as in the rat PK studies. Tissue samples were homogenized and extracted with 70% methanol. After centrifugation at 4000 rpm for 10 min, the supernatants were transferred to a clean tube, and the concentrations of analytes in each sample were analyzed by LC-MS/MS.
The mouse SARS-CoV-2 infection experiments were performed at the BSL-3 facilities of Sun Yat-sen University and Guangzhou Customs District Technology Center, which obtained a license for small animal study of SARS-CoV-2 approved by the China National Health Commission. All animal study protocols were approved by the Institutional Animal Welfare Committee, and all procedures used in animal studies complied with the guidelines and policies of the Animal Care and Use Committee of the respective research units.

SARS-CoV-2 plaque assay
Vero E6 cells were grown in 12-well plates to 70 to 80% confluence and infected with 500 l of medium containing viruses at indicated dilutions. After 1 hour at 37°C, the inoculate were removed. One milliliter of 0.8% agar (Amresco) in DMEM with 2% FBS was overlaid onto cells. Plaques were analyzed at 72 hours. For plaque staining, 1 ml of 70% ethanol containing 0.5% purple crystal (Sigma-Aldrich) was overlaid onto cells. Twenty-four hours later, the stained plaques were counted.

Focus-forming assay
Vero E6 cells were seeded in 96-well plates 1 day before infection. Tissue homogenates were serially diluted and used to inoculate Vero E6 cells at 37°C for 1 hour. Inoculate were then removed before adding 125 l of 1.6% carboxymethylcellulose warmed to 37°C per well. After 24 hours, cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100. Cells were then incubated with a rabbit anti-SARS-CoV-2 nucleocapsid protein polyclonal antibody (catalog no. 40143-T62, Sino Biological), followed by a horseradish peroxidase (HRP)-labeled goat anti-rabbit secondary antibody (catalog no. 109-035-088, Jackson ImmunoResearch Laboratories). The foci were visualized by TrueBlue Peroxidase Substrate (KPL) and counted with an ELISPOT reader (Cellular Technology). Viral titers were calculated per gram of tissue.

H&E staining and immunohistochemistry
For hematoxylin and eosin (H&E) staining, the mouse lung and spleen sections were fixed in zinc formalin and embedded with paraffin. Tissue was sectioned to slices of about 4 m and were stained with H&E. For immunohistochemistry, slides were blocked in 1% bovine serum albumin for 1 hour and stained with a primary antibody to SARS-CoV-2 S protein (GeneTex, catalog no. GTX632604) in blocking buffer at 4°C overnight. The slides were incubated with the biotinylated-secondary antibodies (Jackson ImmunoResearch Laboratories Inc.,715-065-151) for 30 min at room temperature. The slides were then incubated with HRP-conjugated streptavidin (Jackson ImmunoResearch Laboratories Inc., 016-030-084) for 30 min at room temperature in a dark humid box and subsequently stained with 3,3′-diaminobenzidine (Vector Laboratories, SK-4100). Counterstaining with hematoxylin (Beijing Leagene Biotech) and mounting with neutral resin (Solarbio) was performed. Last, the images were scanned by NanoZoomer S360 (Hamamatsu Photonics).

Statistical analysis
All values are presented as means ± SD of individual samples. Data analysis was performed with GraphPad Prism Software (GraphPad Software Inc., version 6.01). A Shapiro-Wilk test was used for analyzing normality distribution. A log-rank test, one-way analysis of variance (ANOVA) with Dunnett's correction for multiple comparisons, or Kruskal-Wallis test with Dunn's correction for multiple comparisons was used for statistical analysis, and respective details are indicated in the figure legends (Figs. 4 to 6). P values < 0.05 were considered statistically significant.