Small Molecule of the Year – 2021

We asked the global drug discovery community to nominate and vote on their favorite molecule from 2021, and the results are in. The 2021 choice for Drug Hunter’s Small Molecule of the Year is Pfizer’s CoV-2 Mpro Inhibitor, PF-07321332 (nirmatrelvir, Paxlovid). Published in a November 2021 Science article, the molecule was praised for its impressive speed to development and potential clinical impact.

Pfizer received emergency use authorization (EUA) for nirmatrelvir at the end of 2021, with an NDA expected in 2022. The EUA was based on a Phase II/III trial, the final analysis of which showed Paxlovid reduced risk of hospitalization and death by 88% (5.8% lower hospitalizations; all 13 deaths observed in placebo group). The drug also appears to be effective in vitro against the Omicron variant.

The Vote and Runners-Up

Paxlovid received >30% of the vote, with the top five receiving 75% of the vote. Runners up included:

nirmatrelvir PCSK9i KB-0742 MRTX1133 ARV-471 H3B-8800 chemical structure Drug Hunter Molecule of the Year

Every molecule nominated received votes from people not involved on the projects, with significant praise for each coming from scientists.

Why It’s Interesting to Drug Hunters

We asked what you thought was impressive or important about nirmatrelvir, and here is some of what was said:

  • “The discovery and advancement of a small molecule to the market in under two years is unprecedented and considering the impact of COVID-19 is transformational for treatment of this devastating disease. The application of fundamental medicinal chemistry principles to identify this orally available molecule such as conformational restriction and minimizing hydrogen bond donors and molecular weight, as well as focusing on developability by choosing a molecule less prone to epimerization, are all noteworthy.”
    Dr. Christopher J. Helal, VP and Head of Medicinal Chemistry at Biogen
  • “This drug will save MILLIONS of lives and it was developed incredibly rapidly.”
    Dr. John L. LaMattina, senior partner at PureTech Health and former head of R&D of Pfizer
  • “Speed of going from concept to patient. Massively important for addressing the pandemic.”
    Dr. Chris Murray, SVP of Discovery Technology at Astex
  • “How fast it was developed, how effective it appears to be, and the fact that it is orally bioavailable.”
    Dr. Walter Moos, Managing Director of Pandect Bioventures

Scientists and research leaders from across the industry praised the impressive speed of development and potential clinical impact:

What You Liked About Nirmatrelvir (Paxlovid API)

A recap of the fascinating Paxlovid drug discovery and development story appears below.

The Unmet Medical Need for Paxlovid

While vaccines and antibodies were developed and authorized for COVID-19 at record speed, both have been challenged by emerging variants due to the rapidly mutating COVID-19 spike protein.

As physician reviewer Dr. Ron Li says, “The reality is that the vaccines, although extremely helpful, have shown lower effectiveness against Omicron, especially for immunocompromised patients. I recently took care of a triple-vaxxed patient in the hospital with severe COVID pneumonia. He was on two immunosuppressants for a kidney transplant. If [a drug] had a clear benefit for that patient population, it would be a big deal.”

An easy-to-administer oral treatment that doesn’t target the spike protein would help address resistance, allow treatment of immunocompromised or unvaccinated individuals, be more accessible worldwide in terms of cost and time-to-treatment, and be cheaper to manufacture and distribute than antibodies.

Several oral candidates were put forward with different mechanisms of action against COVID, with the initial frontrunner being Merck/Ridgeback’s molnupiravir. Molnupiravir showed sharply reduced efficacy in the final analysis of its trial data relative to the interim analysis, however, with the reduction in hospitalization risk for molnupiravir patients falling from 48% to just 30% in the bigger data set. Molnupiravir’s mechanism of action as a viral polymerase inhibitor also brought along concerns about safety and further resistance.

A drug that can be safely taken for a longer period of time could also help prevent or treat long COVID if it reduces the initial severity of the disease. “There is a clearer correlation now between long COVID and initial severity of disease, so Paxlovid could also potentially exert an effect there. Decreased hospitalizations, improved outcomes for immunocompromised patients, and decreased incidence of long COVID would be a triple win!” said Dr. Li.

How Paxlovid Was Discovered: Target Rationale

The whole genome sequence of SARS-CoV-2 was completed and deposited to GenBank by January 5, 2020. As soon as the SARS-CoV-2 virus was isolated and characterized, antiviral experts worldwide started thinking about how to tackle the disease. Though the genome revealed 29 different viral proteins, antiviral drug hunters had a good idea of where to start based on decades of experience.

Many viruses, including HIV, HCV, Dengue, and COVID, translate their genome into a large polyprotein that is subsequently cleaved into components by cellular or viral proteases. Although there are numerous possible proteases to target, viral polyprotein proteases have been successfully drugged for HIV (e.g., amprenavir) and HCV (e.g., boceprevir, simeprevir) in the past, making the main COVID-19 polyprotein protease, SARS-CoV-2 Mpro, particularly attractive. Furthermore, because the recognition sequence of Mpro (L-N*S-A-G) is dissimilar to any human protease cleavage sites, toxicity related to human protease inhibition seemed unlikely.

It turned out that SARS-CoV-2 was highly similar to SARS-CoV (79.6% sequence identity), the virus which caused SARS nearly two decades ago. Better yet, the COVID-19 main protease (Mpro/3CLpro) was 96% sequence identical between the two viruses, and 100% conserved in the active site. Crystallization of the original SARS Mpro had already been accomplished during research on SARS, allowing rapid crystallization of the new virus’s Mpro enzyme, confirming similarity between the two enzymes. Accordingly, activity of Mpro inhibitors against SARS-CoV-1 and SARS-CoV-2 were found to track together.

Fortunately, Pfizer had a research program on the original SARS in the early 2000s that led to an IV preclinical candidate for SARS-CoV-1 (PF-00835231, WO2005113580). Though the development of PF-00835231 was halted due to the end of the SARS outbreak, it turned out to have similar potency against SARS-CoV-2, and its novel phosphate prodrug (PF-07304814) was taken into development as an IV drug for Covid-19.

An IV SARS-CoV-2 Drug from a SARS-COV-1-Inhibitor PF-00835231

How Paxlovid Was Discovered: Chemistry Strategy

But the Pfizer team recognized how important an oral option would be and conducted a parallel campaign to identify such a drug. PK optimization, however, is anything but trivial for protease active-site inhibitors, as the long history of HCV protease inhibitors shows. The starting points against proteases, the substrates of which are peptides, tend to be highly peptidic and, hence, easily metabolized and poorly absorbed.

If anyone knows how to make a drug bioavailable, it’s a Pfizer scientist. The team recognized that hydrogen bond donors were likely a problem and systemically evaluated H-bond donor replacements, separating absorption from metabolism issues by tracking fraction absorbed by the gut prior to first-pass liver metabolism (Fa x Fg). The replacement of the hydroxyketone warhead with a nitrile led to a >10x increase in (Fa x Fg) and 5x increase in permeability, though potency suffered by ~100x.

Improved Absorption form Gastrointestinal Tract

With the general directions of fewer hydrogen bond donors and fewer rotatable bonds set, and with preferred warhead fragments in hand, the remaining fragments were quickly evaluated using peptide couplings with pieces derived from what are now readily available building blocks, thanks to decades of HIV and HCV research. Though the Pfizer team had data suggesting that the nitrile could ultimately be more bioavailable, initial analogs were generated with a benzothiazole ketone (e.g., compound 5, Figure 5) possibly because it is more amenable to parallel chemistry (the right-hand amine precursor is isolable, readily protonated for MS detection, and UV-active, whereas the nitrile is introduced in an additional step after peptide coupling, and the amide precursor is not readily protonated or UV-active, a pain for detection during parallel synthesis).

A Benzothiazole Ketone Lead vs. PF-073211332 "compound 5" PF-07321332/nirmatrelvir

PF-07321332 (nirmatrelvir) ended up being one of 20 compounds prepared in a week in July 2020, within five months of the program’s initiation. Interestingly, PF-07321332, a nitrile, was selected over a benzothiazole ketone (“compound 5”) as a clinical candidate primarily based on ease of synthetic scale-up, isomeric stability and formulation considerations for preclinical toxicology. This choice makes PF-07321332 one of the rare examples of a drug with a reversible covalent nitrile warhead (other previously approved reversible covalent and irreversible covalent nitriles being saxagliptin and vildagliptin, both vs. DPP4).

In total, about 600 compounds and 80 cocrystal structures were generated until the candidate was selected in October 2020.

Cocrystal Structures of CoV-2 Mpro Complexes Mpro, PDB: 7KHP, 7JOY. (Top right) CoV-2 Mpro dimer bound to nirmatrelvir, PDB: 7RFS. (Middle left) PF-00835231 bound to CoV-2 Mpro, PDB: 6XHM. (Middle right) PF-07321332/nirmatrelvir bound to CoV-2 Mpro, PDB: 7RFS. (Bottom left) chemical structure of PF-00835231. (Bottom right) chemical structure of nirmatrelvir.

Caption: (Top left) Cocrystal structure of the C-terminal autopeptide substrate bound to CoV-2 Mpro, PDB: 7KHP, 7JOY. (Top right) CoV-2 Mpro dimer bound to nirmatrelvir, PDB: 7RFS. (Middle left) PF-00835231 bound to CoV-2 Mpro, PDB: 6XHM. (Middle right) PF-07321332/nirmatrelvir bound to CoV-2 Mpro, PDB: 7RFS. (Bottom left) chemical structure of PF-00835231. (Bottom right) chemical structure of nirmatrelvir.

The Development of Paxlovid: An Unbeatable Speed Record?

By Sep. 1, 2020, initial rodent PK studies of PF-07321332 had been completed, and by late October, 100 grams of the molecule was produced. Only a month later, in November 2020, production had been scaled up to 1.4 kilos to support toxicology. Fortunately, nirmatrelvir passed a battery of preclinical experiments including two-week regulatory toxicity studies in rats and monkeys, in which the no observed adverse effect levels (NOAELs) were 1000 mg/kg/day and 600 mg/kg/day, respectively (unbound Cmax margins of 273x and 510x).

Recruitment for Phase I started three months later, in February 2021. Nirmatrelvir was studied both as a single agent and in combination with ritonavir (RTV) as a PK booster (RTV is a time-dependent CYP3A4 inhibitor). Significantly greater target coverage based on SARS-CoV-2 EC90 was observed with PF-07321332 250 mg+RTV than with PF-07321332 150mg alone (~18 h vs. <3 h). Hence, Paxlovid is authorized as a regimen of 300 mg nirmatrelvir + 100 mg ritonavir, twice daily for five days.

“Paxlovid sets a speed record in development that may never be broken,” said Drug Hunter reviewer Mike Koehler. The Science paper describing the development came out in the same month that their clinical trial was ended early due to strong efficacy. “I looked up some other recent molecules with rapid development times,” he adds, “and the standouts are the CFTR modulators, which were in trials for only about three years (but took a very long time to go from the phenotypic screens to development) and remdesivir, which received rapid approval, but was developed years earlier as a general antiviral therapy and repurposed.”

Paxlovid Drug Discovery and Drug Development Timeline

To recap, here is a timeline of key milestones in the discovery and development of Pfizer’s COVID-19 drug, Paxlovid:

A decision on the target to pursue was made almost immediately after the virus was identified. The team was resourced within three days of the program being proposed at Pfizer. The candidate was made in four months from project start. The candidate was nominated within six months of project commencement, and Phase I recruitment began less than a year after the program was initiated.

The drug received authorization less than a year after the first patient was dosed, only two years after the pandemic started. Overall, from project start to approval, the program took about 20 months.

In an industry where we are accustomed to 12-year development timeframes, the pace of this molecule’s discovery and development by the Pfizer team is truly breathtaking. Indeed, it’s a testament to how far science and industry have advanced in less than two decades, when the original SARS broke out.

Hope you enjoyed this story, and thanks to all who participated. Explore for more drug discovery content.

Researched and illustrated with Dr. Jennifer Huen and Dr. Vinicius Texeira. Thanks to Dr. Dennis Koester and Dr. Michael Koehler for review of this article. A special thank you to Ukrainian designers Mike Dodukalo and Aleksey Sendetskiy who insisted(!) on continuing work on this article.

Join the Drug Hunter mailing list

to get free content and resources weekly. Trusted by > 5,500 drug hunters worldwide. Unsubscribe anytime.


Join Subscribers from

Reader Interactions