PREreview of X-ray Crystallography-Guided Design and Synthesis of Cyclopentyl Heteroaryl Carboxylic Acid based Inhibitors of the SARS-CoV-2 Nsp3 Macrodomain (Mac1)

James S. Fraser, W. Patrick Walters

The authors present a series of compounds following up the original fragment screen of Mac1. They elaborate off a fragment (11) that has the typical carboxylate interaction in the oxyanion hole and explore “north” from there to interact with the Phe (where the adenosine of the natural substrate interacts) and even further to Asp22 engaging hydrogen bonds. Much of the work focuses on the synthetic chemistry details for making these compounds. The major strengths of the manuscript are an excellent recounting of the development of inhibitors from the original fragment screen to our in vivo active molecules in the introduction, the exploration of non-adenosine looking interactions with the Phe explored, and the strong supporting X-ray crystallography data. Ultimately, this effort, which was contemporaneous with other efforts starting from the same inactive fragment hits, developed a series of compounds, eventually arriving at analogues 26a-c with IC50 values in the 6–7 µM range. These molecules are ligand-efficient and could be developed further. While the synthetic methodology (specifically the use of photoredox and Barluenga-Valdés coupling to access trans-cyclopentanes) is of interest, the medicinal chemistry insights are limited, and reads more like a catalog of synthetic outputs than a strategic medicinal chemistry campaign. The paper does not clearly show how these findings will help others design better Mac1 inhibitors or how this scaffold overcomes the hurdles faced by existing, more potent series. The manuscript could be strengthened by 1) “zooming” back out at the end (paralleling the wonderful introduction) and placing these molecules in the chemical space of all that has been done, 2) contrasting the syntheses here with what is accessible with “purchasable” space/enamine real as leveraged on the macrodomain in other work arriving at similar potencies.

Some other points to consider:

• Lack of Potency and Innovation Relative to State-of-the-Art The most potent compounds reported here (6–8 µM) are significantly less potent than previously disclosed Mac1 inhibitors, some of which reach the low nanomolar range (e.g., AVI- and ASAP compounds). Furthermore, the scaffold retains a carboxylic acid moiety, which has been established in Mac1 literature as a metabolic and cellular liability. The paper does not address how this scaffold could eventually circumvent this known issue, nor does it offer a competitive advantage over existing chemical series. This should be addressed in the conclusion head on (paralleling the introduction, but pointing the way forward).

• Use of WaterMap to rationalize SAR. The authors rely heavily on a WaterMap analysis to explain the 10-fold potency difference between benzimidazole 20a and its regioisomers. This analysis appears largely retrospective rather than predictive. Given that the benzimidazole substituent is highly solvent-exposed, the claim that a specific water-mediated network accounts for such a large shift in affinity is tenuous. Furthermore, the proposed noncanonical hydrogen bond between the aromatic C–H group and Asp22 exhibits an unfavorable geometry, casting doubt on its contribution to binding. Similarly, the authors emphasize the "water-bridging" capability of 20a as a key driver of activity, yet they fail to provide a compelling explanation for why other analogues capable of similar interactions (such as the thiadiazole 20f) are significantly less active IC50 > 600 µM). Without a more holistic WaterMap analysis of the inactive or weakly active compounds, the structural hypotheses feel "cherry-picked" to fit the few successful co-crystal structures obtained.

• p2 line 36 left column: “In addition, 54 fragments bound in a newly discovered site, labelled the oxyanion subsite; ADPr does not have any direct interactions with residues in this subsite.” Figure 5 of the 2021 fragments screening paper points out the bridging water and that 47! of the fragments bound there contain a carboxylate! This seems important to mention up top as the theme of the paper is dancing around the carboxylic acid interactions and that was present right from the get go, not just introduced in the later compounds in the right column of p2

• p2 line 42 right column: “Finally, in terms of a fragment starting point, there has been a recent report of a hit-to-lead optimisation.32” 32 stops at in vitro, should probably be mentioned here as it comes after the move advanced in vivo molecules. Kind of confusing where it is placed in the overall flow of the (authoritative!) introduction that felt like reliving the last 5 years of JF’s life

• We are a little confused by highlighting “cis-12 (racemic) (PDB: 5S3X, not shown) (Figure 2).” since it isn’t really mentioned again in the manuscript. Is this just because it also projects similarly? If so, there are many others from the original fragment screen that do too and this gets to the larger point of comparing the Adenosine mimicking elements more broadly to understand the novelty contribution of this article.

• p4 - line 35 left column: “The selection of poses was made based on ligand strain, RMSD of the cyclopentyl core with respect to that in the X-ray structure of the cis-11-Mac1 complex (PDB: 5S3T) and hydrogen bonding interactions with Asp22 and/or Ile23 in the adenine binding subsite (see SI). In the priority 1 set (18 compounds), the heteroaryl groups interacted via at least one hydrogen bond with either Asp22 or Ile23. The priority 2 set (6 compounds) contained a polarised -CH group able to interact with the side chain of Asp22.“ The SI helps us understand the interactions for prioritizing, but we are also interested in the strain, RMSD cutoffs. It’s not clear how to reproduce the workflow in spirit never mind in programmatic detail.

• For Figure 3 and following ones that had modeling, it could be interesting to see vs. docking/modeled poses of adenine mimics to see if there are any surprises in the adenosine mimicking elements given the very locked in carboxylate oxyanion interactions.

• It is notable that the molecules highlighted in Figure 5 growing up towards Asp22 are less Adenosine looking molecules (reminiscent of 8539/ aka RR6 in this paper, FrankenRocs merger in Gahbauer 2023 paper) than others previously reported. The overall novelty at this analogous position could be compared across all known macrodomain inhibitor structures. Could there be a comparison in some chemical space representation of how distinct these compounds (or at least the adenosine analogous element) are? Of course this would recognize the convergence of many ideas over many years and that the molecules proposed here were quite novel at time of original synthesis/ideation.

• Dovetailing off of this point, is there an overall meta point about the synthesis pursued here vs. make on demand chemistry in purchasable space (as demonstrated by our frankenROCS paper, PMID: 40435250) to contextualize “The timelines for the early stages of this project were accelerated as fragment hit cis-11 was specifically designed to be synthetically enabled for follow-up work.“ - is this getting at a distinct chemical space? What are the relative tradeoffs and how do we think about the different approaches as a field? The optimization here relies primarily on "nitrogen walks" and standard heterocycle substitutions. No fragment-linking or merging strategies were employed, and the docking approach is standard practice. Consequently, the paper reads more like a catalog of synthetic outputs than a strategic medicinal chemistry campaign or a wide exploration of the space the fragments can inspire. Can this be placed in the broader context of Mac1 inhibitor chemical space explored, contrasting the strategies that explored that space?

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.