Preclinical characterization of an intravenous coronavirus 3CL protease inhibitor for the potential treatment of COVID-19

The COVID-19 pandemic created an urgent need for effective antivirals beyond existing options such as remdesivir, whose clinical efficacy has been mixed in hospitalized patients. The coronavirus main protease (3CLpro) is essential for viral polyprotein processing and has no close human homologs, making it an attractive, selective antiviral target. Protease inhibitors have succeeded clinically in HIV and HCV, supporting the strategy. PF-00835231, originally discovered after the 2002–2003 SARS outbreak, is a potent SARS-CoV/SARS-CoV-2 3CLpro inhibitor, aided by the near-identical 3CLpro active sites between the two viruses. However, its low solubility and poor oral bioavailability complicate dosing. This study aims to: (1) characterize PF-00835231’s antiviral breadth, selectivity, and in vivo efficacy; (2) design and evaluate a phosphate prodrug, PF-07304814, to enable intravenous delivery and sustained systemic exposure of PF-00835231; (3) define ADME, pharmacokinetics, and safety profiles; and (4) assess combination effects with remdesivir.

Background work establishes remdesivir as the first approved COVID-19 antiviral, though with variable outcomes in hospitalized cohorts. Monoclonal antibodies provide benefit early in infection. Prior outbreaks and structural studies identify coronavirus 3CLpro as essential for replication with highly conserved active site and broad substrate specificity across coronaviruses, enabling structure-based inhibitor design. The clinical success of protease inhibition in HIV and HCV supports proteases as viable antiviral targets. PF-00835231 was previously designed against SARS-CoV 3CLpro and exhibits potent biochemical inhibition. The high sequence identity (96% overall; 100% in active site) between SARS-CoV and SARS-CoV-2 3CLpro justifies repurposing and optimization. Literature also emphasizes potential drug-efflux issues (e.g., P-glycoprotein) in commonly used Vero cells that may mask true intracellular potency, and highlights the utility of combination therapy targeting distinct replication steps to improve efficacy and suppress resistance.

Biochemical binding and enzyme inhibition: A thermal shift assay assessed direct binding of PF-00835231 to SARS-CoV-2 3CLpro, with protein incubated ± PF-00835231 and SYPRO Orange monitored across 30–90 °C to derive Tm. A FRET-based protease activity panel quantified Ki values of PF-00835231 across alpha-, beta-, and gamma-coronavirus 3CL proteases; selectivity was evaluated against a panel of human proteases and HIV protease. In vitro antiviral assays: Cytopathic effect assays were conducted using VeroE6-enACE2 and VeroE6-EGFP cells infected with SARS-CoV-2 (USA-WA1/2020 or Belgium GHB-03021/2020). Potency shifts were measured with and without the P-gp efflux inhibitor CP-100356 (0.5–2 µM) to assess efflux limitations. Additional assays included A549-ACE2 human lung carcinoma and MRC-5 lung fibroblasts (HCoV-229E) to compare potency and P-gp effects in lung-relevant systems. EC50/EC90 were calculated via 4-parameter logistic models; cytotoxicity (CC50) assessed in parallel. Drug combination studies: HeLa-ACE2 high-content imaging assays quantified PF-00835231 and remdesivir alone and in combination. SARS-CoV-2 infection was detected using convalescent COVID-19 patient sera and fluorescent secondary antibodies. Drug interaction data were analyzed by Loewe, Bliss, HSA reference models and ZIP synergy in SynergyFinder to classify additivity or synergy. Mouse efficacy models: Two models were used: (1) Mouse-adapted SARS-CoV (MA15) in BALB/c mice with PF-00835231 administered subcutaneously BID at 30, 100, or 300 mg/kg starting day 0 or delayed to day +1 or +2; lung titers measured day 4, weight loss and lung histopathology assessed by H&E and anti-N IHC. (2) Ad5-hACE2 transduced BALB/c mice infected with SARS-CoV-2 USA-WA1/2020; PF-00835231 100 mg/kg S.C. BID; lung titers measured day 3 by TCID50. Mouse plasma exposures (unbound Cmin) were quantified by LC-MS. ADME and PK: PF-00835231 metabolic stability and clearance were evaluated in pooled human liver microsomes; reaction phenotyping identified CYP3A4/5 as major enzymes. CYP inhibition (reversible and time-dependent) and transporter inhibition (BCRP, P-gp, OATP1B1/1B3, OCT1/2, OAT1/3, MATE1/2K) were profiled. Plasma protein binding and blood:plasma ratios were determined by equilibrium dialysis. Preclinical PK after IV dosing was characterized in rats, dogs, and monkeys; oral bioavailability assessed in rats and monkeys. Human PK parameters were predicted by in vitro–in vivo scaling and allometry. Prodrug design and characterization: A phosphate prodrug (PF-07304814) was synthesized in two steps from PF-00835231 via phosphoramidite coupling and subsequent deprotection. Physicochemical properties (pKa, predicted logD7.4, aqueous solubility) were measured, demonstrating markedly enhanced solubility (>200 mg/mL). In vitro conversion kinetics using liver S9 fractions across species assessed alkaline phosphatase-mediated conversion to PF-00835231; phosphate buffer inhibition confirmed mechanism. In vivo conversion and exposure to PF-00835231 after IV PF-07304814 dosing were quantified in rats, dogs, and monkeys. Human PK and conversion fraction were predicted to simulate clinical steady-state concentrations under continuous IV infusion. Safety pharmacology and toxicology: Genotoxicity (Ames and micronucleus), hERG inhibition, secondary pharmacology profiling, and hemocompatibility were assessed for both PF-07304814 and PF-00835231. A GLP rat 24-hour continuous IV infusion study (PF-07304814) established a NOAEL and evaluated clinical observations, clinical pathology, and histopathology. A non-GLP 4-day continuous IV infusion study of PF-00835231 in rats assessed tolerability and clinical chemistry.

Biochemical potency and selectivity: PF-00835231 binds tightly to SARS-CoV-2 3CLpro, increasing protein Tm by 14.6 °C (from 55.9 ± 0.11 to 70.5 ± 0.12 °C). It inhibits diverse coronavirus 3CL proteases with Ki values from 0.03 to 4 nM, including SARS-CoV-2 3CLpro Ki 0.27 ± 0.1 nM, and shows strong selectivity over human proteases (cathepsin B activity ~6 µM, ~1000-fold weaker than 3CLpro). Cellular antiviral activity: In VeroE6-enACE2 and VeroE6-EGFP cells, PF-00835231 potency improved markedly with 2 µM P-gp inhibitor, from EC50 39.7–88.9 µM to 0.23 and 0.76 µM, respectively; EC90 values were 0.48 and 1.6 µM. In lung-relevant cells (A549-ACE2, MRC-5), P-gp inhibition had minimal effect, and potencies were similar to P-gp-inhibited Vero assays. HeLa-ACE2 assays showed PF-00835231 EC50 0.13 µM and EC90 0.43 µM; remdesivir EC50 0.074 µM and EC90 0.17 µM. Combination with remdesivir: PF-00835231 displayed additive to synergistic effects with remdesivir without antagonism, supported by Loewe/Bliss/HSA and ZIP synergy analyses. Fixed remdesivir concentrations shifted PF-00835231 potency to lower values (e.g., EC50 from 134 to 59 nM at 48 nM remdesivir; EC90 from 365 to 126 nM), indicating combination benefit. In vivo efficacy: In the SARS-CoV MA15 mouse model, PF-00835231 (S.C. BID) initiated on day 0 reduced lung titers by ≥3 log10 in a dose-dependent manner (30, 100, 300 mg/kg) and mitigated weight loss and lung pathology; day 0 treatment was significant, with delayed start (day +1) still reducing titers, though less robust (day +2 not significant). Estimated unbound Cmin exposures at 100 and 300 mg/kg were ~500 nM (~1× in vitro EC90) and ~1700 nM (~3× EC90), respectively. In the Ad5-hACE2 SARS-CoV-2 mouse model, 100 mg/kg BID reduced lung titers by ~1.5 log10; unbound Cmin ~350 nM (~0.7× EC90). ADME/PK and human projections: PF-00835231 is primarily cleared by CYP3A4/5, has low risk for CYP- or transporter-mediated DDIs at projected exposures (CYP IC50s >200 µM; transporter IC50s >20 µM), moderate protein binding (fu 0.33–0.45), moderate IV clearance, low Vdss, short t1/2 (<2 h), and very low oral bioavailability (<2%) due to low solubility/permeability and efflux. Predicted human CL ~6 mL/min/kg, Vdss ~1 L/kg, t1/2 ~2 h, supporting continuous IV infusion to maintain unbound plasma Ceff ~0.5 µM (≈ EC90), estimated at ~320 mg/day of PF-00835231. Prodrug strategy and conversion: PF-07304814 is highly soluble (>200 mg/mL), enabling feasible infusion volumes and dose escalation. In liver S9, rapid, phosphate-sensitive conversion to PF-00835231 was observed (human CLint,u 428 µL/min/mg), implicating alkaline phosphatase. In vivo IV dosing of PF-07304814 in rats, dogs, and monkeys yielded high conversion to PF-00835231 (≈68–81% of active exposure vs direct dosing). Predicted human PF-07304814 CL ~10 mL/min/kg, Vdss ~0.1 L/kg, short t1/2, and ~75% conversion. A 500 mg/day continuous IV infusion of PF-07304814 is projected to yield unbound steady-state PF-00835231 of ~0.5 µM, reaching 90% steady state by ~6 h. The prodrug itself is >600-fold less potent on 3CLpro (Ki 174 nM) and unlikely to contribute meaningfully to antiviral activity in vivo beyond conversion. Safety: Both PF-07304814 and PF-00835231 were negative in genotoxicity assays, showed no hERG inhibition up to 300 µM, and were hemocompatible. A GLP rat 24-hour continuous IV infusion study established a NOAEL of 1000 mg/kg for PF-07304814, and PF-00835231 was tolerated at 246 mg/kg/day over 4 days with only minimal, non-adverse clinical chemistry changes. At a projected human efficacious dose (500 mg/day PF-07304814), unbound Cmax and AUC24 exposure margins were high (PF-07304814: ~97× and 65×; PF-00835231: ~25× and 21×), supporting clinical dose escalation if needed.

The data demonstrate that PF-00835231 is a potent, selective, broad-spectrum coronavirus 3CLpro inhibitor with consistent antiviral activity across relevant cell systems and significant in vivo efficacy in mouse models when systemic exposures near the in vitro EC90 are achieved. The phosphate prodrug PF-07304814 overcomes solubility and dosing limitations, enabling continuous IV infusion to maintain target unbound plasma concentrations likely reflective of lung exposures. The observed additivity to synergy with remdesivir supports combination strategies targeting distinct replication steps (protease and polymerase), analogous to combination protease-based regimens used successfully in HIV and HCV, potentially improving efficacy and resistance suppression. The safety and DDI profiles appear favorable, and exposure margins suggest flexibility to achieve higher multiples over EC90 in clinical studies. Collectively, the findings directly address the need for non-polymerase antivirals and provide a viable IV therapeutic candidate for COVID-19, with the potential to serve as a backbone for combination therapy.

PF-07304814 is a soluble phosphate prodrug that rapidly converts in vivo to the active 3CLpro inhibitor PF-00835231, enabling continuous IV dosing to sustain efficacious exposures. PF-00835231 shows tight binding and broad 3CLpro inhibition across coronaviruses, potent cellular antiviral activity, meaningful in vivo efficacy in SARS-CoV and SARS-CoV-2 mouse models, favorable ADME/PK, low DDI liability, and supportive safety margins. Combination studies with remdesivir show additive to synergistic effects without antagonism. These preclinical data justify clinical evaluation of PF-07304814 as a single-agent antiviral for COVID-19 and as a combination partner. Future work should include clinical pharmacokinetic/pharmacodynamic validation, resistance profiling of SARS-CoV-2 under 3CLpro pressure, exploration of dosing duration/optimization, and expanded combination studies with other antivirals.

Resistance selection and mapping for SARS-CoV-2 under 3CLpro inhibition were not performed and remain to be elucidated. In vitro potency assessments in Vero cells are confounded by high P-gp expression; while mitigated experimentally, this highlights assay-dependent variability. Translational assumptions rely on predicted human PK and prodrug conversion fractions; actual human clearance and conversion could differ. The active agent has low oral bioavailability, necessitating continuous IV infusion, which may limit outpatient use. Duration of effective clinical dosing is not established and requires clinical validation.