How Scientists Discover New Antiviral Drugs (Medicinal Chemistry)

Watch the video on YouTube or read the written post below!

When people with an IQ of 50 hate on the “evil” scientists of big pharma, they often overlook that there is relentless work and ingenious brainpower behind the discovery and optimization of medicines, including antiviral drugs. If drug discovery would be so easy, pharma companies would not spend up to billions of dollars to get just one drug to the market. Also, chemists would be balling instead of complaining about bleak career opportunities on Reddit.

In this post, we will look at the educational drug discovery journey of an antiviral drug. Just by looking at today’s molecule, you should know this is going to be a nice one – and yes, that’s a boron atom in a pharmaceutical. You will learn why using boron in drugs can be powerful, and why it’s not good if your people in clinical trials turn yellow.

Hepatitis C: Significant Innovation on Major disease burden

Hepatitis C is a severe infectious disease, leading to liver disease and serious complications. The hep C virus chronically infects over 170 million people or over 2% of the world’s population! The dilemma is that disease incidence and drug market are inversely related. The key Western Pacific, Southeast-Asian and African regions have the highest prevalence with 130 million infections and no access to HCV drugs. The US and Europe on the other hand have around 15 million infections but are the target market for drugs. HCV occurs in 6 different “genotypes” or variants across the globe which complicates treatment. High-income countries have primarily the genotype 1, which is only 10% of disease burden in low income countries. Without being a scientist, you can guess that these viral genotypes influence drug sensitivity – that’s also what we’ve seen with C19 vaccines and variants such as Omicron. Oh yeah, and there’s a problem of viral resistance.

HCV spreads via blood-to-blood contact, so injection drug use, poorly sterilized medical equipment and other pathways. In contrast to HIV or Hepatitis B, it is not a STD. An infected individual can show no symptoms for decades while increasing their risk of liver failure and cancer. The good news it that unlike infections with HIV or Hepatitis B, HCV is curable. However, while there are approved vaccines for Hep B, only few are still in development for Hep C. Early physiological studies earned the Nobel Prize in 2020, and the last two decades of saw a true revolution of HCV drug discovery.

As HCV is a viral disease, there are multiple potential inhibition points in the virus lifecycle – entry into the host cell, protein synthesis steps and packaging and release. However, if we contrast HCV drug development with HIV, it becomes evident that there was quite the slow start. While the first antiviral drug was approved 3 years after the virus was identified, it took 24 years from HCV discovery to approval. There are many reasons for this – low perceived market value, low pressure from patient associations but also a poor understanding of the nature of HCV as a disease. Without going through all of the details, advancements in structural biology, pharmacology and also clinical trial design ultimately led to the approval of various therapies.

If you squint hard enough, you might be able to read something!

Most are combinations based on interferon proteins – which work by increasing immune defense – administered by subcutaneous injection, and ribavirin – which is an oral broad-spectrum antiviral that is used against various viral fevers. This “one size fits all” had a decent efficacy of around 55% – but true improvement came only with the development of combination stacks which added a direct-acting antiviral drug. These were more targeted and robust against resistance and also came with a halved treatment duration. As the third evolution, all-oral combinations were also developed – such as the last entry.

You might think, we have more than one blockbuster antiviral drug already on the market that cured millions of patients, why should we look further into HCV? Well, all the remaining, untreated, tens of millions of HCV infected individuals can’t afford outrageous prices of up to over $50K per treatment. Neither can the WHO, who aspires to treat 80% of diagnosed populations by 2030. The development of new direct-acting antivirals thus may serve to increase market competition and lower costs, thereby enabling more equitable access. Also, many of the drugs developed were not equally effective against the six genotypes we covered at the start – so there is more work to be done.

Design of aN antiviral drug against Hepatitis C

We already mentioned that various steps in the viral lifecycle can be potential drug targets. If the virus can’t productively infect host cells, the infection will not be sustained. In 2014, GSK published studies investigating inhibitors of the NS5b polymerase, the viral RNA printer. This is the same mechanism of action of Gilead’s Sovaldi.

Inhibitors can directly block active sites of enzymes – so where the catalysis is occurring – or instead, bind to other, allosteric sites, inducing macro-conformational changes that decrease an enzymes activity. In the case of NS5b, there are four well established allosteric sites – and the palm II site is particularly interesting because it is closest to the active site, and there are many amino acids that are highly conserved HCV genotypes – so a potential inhibitor could combat all variants equally effective.

The GSK team found an inhibitor here tagged with the number (3), which nicely docked into this palm site. This impacts the enzyme in two ways: firstly, the allosteric binding of compound 3 stabilizes a so-called closed state which is much less active. Secondly, the head group of the inhibitor also interacts with catalytic Mg2+ cofactors and thereby disrupts placement of incoming nucleotides, making polymerase initiation and propagation more challenging.

So what do these inhibitors look like? Their quest started from this scaffold – a benzofuran core with a cyclopropyl group swagging around – which was published by researchers from another company. Looking at a terminal alcohol as the “head group” at the sulfonamide, they observed strong activity against wildtype genotype 1 HCV, as well as against a common mutation which occurs at the 316 aminoacid position. The activity was assessed by looking at cell-based replication systems, as well as more “raw” data from a biochemical polymerase assay.

Boron in PharmaceuticaLS?

As mentioned previously, the apolar part of the inhibitor is bound in the grey palm site while the head group looks into the polar active site. By screening through different head groups, the GSK team found that by appending a boronic acid pushed activity into the single-digit nanomolar range, even for the critical polymorph 316N. For you morons out there, the lower the concentration, the stronger the inhibition. Aryl boronic acids were also more effective, particularly for wild type 1a. In contrast to the para isomer, the meta substituted aryl boronic acid showed reduced activity – highlighting the need for proper orientation of the head group.

What about the need for boron? Well, while substituting the boronic acid with other polar groups maintained strong IC50 in the assay in the right-column, it led to significantly lower replicon activity. This is because anionic compounds suffer from lower cell-permeability – due to the low acidity of boronic acids, this problem does not occur for this series. However, when looking at drug metabolism and pharmacokinetics, DMPK, they found that these aryl boronic acids had very low bio-availability in their rat model, and were rapidly cleared and excreted. This challenge was significantly improved by adding a fluorine substituent to the phenyl ring – increasing bio-availability 5-fold, while even further increasing activity, particularly against the 316N variant. They tried to further optimize activity by embedding the boronic acid into a ring – while this resulted in better performance in the polymerase assay, especially the 316N efficacy decreased.

Although less active, this analogue has an easy but instructive synthesis that nicely tests your understanding of fundamental reactions. First, this starting material was carbonylated – of course, this only touches the aryl bromide as fluorides usually don’t react with Palladium. Next, a Sandmeyer reaction exchanged the amine into a new aryl bromide, going through a diazonium as an intermediate. Then, radical bromination installed the benzylic bromide which was coupled with the free sulfonamide of the apolar core. An easy borylation again leveraged the aryl bromide as a functional handle. The final cyclization is triggered by reduction of the ester with Lithium borohydride – which is quite cool.

By the way, if you are wondering about the use of boron in pharmaceuticals. The application of boron in medicine dates back to the early 19th century, when boric acid, so B(OH)3, was used as a mild antiseptic. However, boron derivatives were long neglected thereafter as a result of largely unfounded claims that they are unstable and toxic. After the approval for bortezomib in 2003, there has been quite an upsurge in interest and we will further explore why it can be quite powerful.

Back to the medicinal chemistry: After feeling satisfied with the activity, the team performed crystallizations to reveal binding modes. There are various supramolecular interactions at work – quite basic apolar and polar interactions, but also cation-pi interactions of the electron-rich benzofuran with the positively charged Arginine side chain in the protein. This Arginine is conserved across all HCV genotypes and is the reason why this scaffold is well-suited to occupy this position. This even has a ripple effect as this Arginine anchors a network of hydrogen bon interactions to other parts of the inhibitor. Notably, the boronic acid did not form any covalent bonds or complexes – instead, it forms a hydrogend bonds with a bound water molecule, as well as other polar interactions that are not indicated here. 

Optimization and Structure-activity relationship Of the Antiviral Drug

So hey, we got ourselves the final antiviral drug at hand already – right? Well, the molecule we flashed at the very start actually looks slightly different from the last lead compound we saw. Why aren’t we done yet? You see, when the team moved forward to human studies, they saw that the antiviral drug had a very short half-life in blood plasma of 5 hours, resulting in a higher anticipated daily dose for efficacy. Additionally – back to the importance of DMPK – they found that metabolic breakdown resulted in a major metabolite with long half life. This compound was also observed in Phase 2 of another drug, and associated with strong liver toxicity. Although there was no direct evidence of toxicity, the team wanted to avoid its formation to limit adverse reactions and improve its pharmacokinetic profile. The key step to prevent was oxidation of the benzylic carbon – quite evident given that the team could also detect the carboxylic acid product.

To maintain a similar binding mode and not risk starting from scratch, they hypothesized three major approaches that might reduce the propensity for benzylic oxidation. The first two consisted of cyclization, either onto the phenyl ring or in the direction of the sulfonamide moiety, while the third simply excised the benzylic CH2 group. This shortening of course came with the risk of severely disrupting the binding of the head group due to its positional shift. You might think – what the heck, why are we doing these complicated things – can’t we just throw a methyl group on the benzylic position and hope that steric shielding reduces the rate of oxidation? You would be correct – but unfortunately, the authors found that adding a methyl group reduced potency 50-fold. This was probably because the Methyl substituent induces a unfavorable conformational twist – and that’s the authors envisioned the two cyclization approaches to strive towards a more pre-organized conformation.

Let’s look at their results. As you can see in this table, cyclization onto the aryl ring actually reduced inhibition significantly, particularly for 316 variants. So, no good.

Within approach B, the team replaced the sulfonamide with carbonyl containing groups because it appeared that only of the oxygens made meaningful contact with the NS5b protein. The results were mixed – for some analogs, even wild-type inhibition decreased significantly. However, the oxazolidinone 25 showed very good potency. Just again demonstrating that the boronic acid is critical, the team found that removing it resulted in an over >100-fold loss of activity.

So, compound 25 was quite encouraging – and the team figured shifting the ring to an aromatic system should be even more metabolically stable, improving the clinical profile of the molecule. They found that the triazoles, such as compound 31, were basically as potent as the previous oxazolidinone. Removing one of the nitrogens decreased activity significantly as it removed a critical hydrogen bond with the protein. Notably, adding substitution to the triazole diminished activity, highlighting the steric constraints in the pocket. In summary, these aromatic designs were not really better – so the team also looked at the approach C – directly eliminating the culprit, the benzylic CH2 group.

Binding Mode of the antiviral drug

This series was right on the money. Recognizing that shortening of the head group would change the position, the team also looked at the meta-boronic acid – but the para-substitution continued to perform better. Here you can also see the dramatic impact of electron-withdrawing groups on activity, particularly on 316N and 316Y genotypes. But the chloro compound 47 was even more potent – this is why they moved away from the fluorine. The final refinement was closure to a benzoxaborole to enhance chemical and metabolic stability – and thankfully, this modification did not lower the activity too much.

Looking at crystal structures revealed an extensive network of interactions, mediated by highly ordered water molecules – the 3 red balls – which the authors did not observe with other series. Here we see the beauty and complexity of supramolecular interactions – the oxaborole moiety interacts directly with two water molecules: One contacts the backbone N–H of a glycine and the second H-bonds to another ordered water molecule bridging an arginine and asparagine. A close look at the boron reveals that its trigonal planar geometry is slightly distorted – it looks like the proximal water is well-positioned to occupy the empty p-orbital of boron and induce a more tetrahedral configuration. The distance between boron and water is 2.5 Angstrom, which is close enough for a strong interaction but not as short as predicted for a covalent bond. This setup is basically an equilibrium between a water-bound and unbound boronate complex. This interconversion of planar to trigonal binding, leading to multiple potential binding modes, is why boron is such a powerful and flexible functionality.

The final question was to check whether this chemical optimization was actually reflected in an improved pharmacokinetic profile of the antiviral drug. Looking at PK in rat, they found that compounds 33 and 49 had much lower in vivo clearance, and were much more bio-available. Looking into drug metabolism, the team also compared cytochrome p450 inhibition. These enzymes are the major route of elimination for multiple drugs and their disruption is one of the most common mechanisms leading to harmful drug-drug interactions and side effects. For example, analog 33 had a micro-molar activity versus major CYPs, raising a potential risk for clinical development. On the other hand, lead compound 49 was less active, posing significantly less risk.

This improved profile was demonstrated when in first-in-human studies, where they saw much longer drug half life and no oxidation to the potentially toxic metabolite which triggered them to re-explore their strategy. GSK then progressed this asset to phase 2, combining it with an RNA-based treatment called RG-101 of Regulus Therapeutics. After they saw two cases of serious jaundice – so patients turning yellow – the FDA put a hold on RG-101 and GSK actually also decided to not further develop this compound. Maybe not the success story we were all expecting.

I hope you learned a thing or two from this story. See you next time!

References on Hepatitis C Antiviral Drug Discovery

  • Design of N-Benzoxaborole Benzofuran GSK8175—Optimization of Human Pharmacokinetics Inspired by Metabolites of a Failed Clinical HCV Inhibitor: J. Med. Chem. 2019, 62, 7, 3254
  • Discovery of a Potent Boronic Acid Derived Inhibitor of the HCV RNA-Dependent RNA Polymerase: J. Med. Chem. 2014, 57, 5, 1902
  • HCV796: A Selective Nonstructural Protein 5B Polymerase Inhibitor with Potent Anti-Hepatitis C Virus Activity In Vitro, in Mice with Chimeric Human Livers,and in Humans Infected with Hepatitis C Virus: Hepatology 2009, 49, 3, 745





Leave a Reply