PREreview of Mutational Profiling of SARS-CoV-2 PLpro in human cells reveals requirements for function, structure, and drug escape

Summary:

In this study the authors comprehensively examined the mutational effects on PLpro proteolytic activity and stability. The authors have designed a FRET-based assay composed of N-terminal mClover3 donor and C-terminal m-Ruby3 acceptor fluorophores separated by a linker containing the Nsp2/3 PLpro cleavage motif to measure the proteolytic activity of PLpro. From DMS data the authors infer PLpro active site mutations ablates activity. Their study also revealed residues required for cleavage of the Nsp2/3 site, identified features of substrate binding pocket and the sequence requirements of the blocking loop. The authors have given explanations for their observations in the Discussion section. Overall, the paper is supported with follow-up enzymology and crystallography experiments of key residues. The major limitation of this study is leaky expression of mutations can mask clinically relevant mutations that can arise due to viral evolution and might have the potential to evade inhibitor treatment. Study of such mutations can provide more information about the potential escape routes open to the virus to evade developing therapeutics. Moreover, incorporation of statistical analysis could strengthen the confidence in inferences drawn from the deep sequencing data and improve the quality of the manuscript. 

The following points can improve the quality of the manuscript:

Major points:

1. In Figure 1f, it is unclear that why is the PLpro activity increasing with increase in inhibitor concentration? Perhaps the Y axis is mislabeled as while inhibitor concentration is increasing, PLpro activity should decrease. However, FRET signal would increase (and maybe should be the axis label), since there will be no cleavage. 

2. Line 175 – 178 - What does 0 represent in the normalized dataset? What is the rationale used for selecting minimum 10 reads in the unselected library as the read cut-off. 10 reads is pretty low cut-off. From the data, it seems the distribution tails off before cut-off chosen for the s.d.- by eye. 0.3 s.d and 20 as read cut-off might be a  better option to eliminate sequencing artifacts.

3. Line 187-190 – What is the number of reads for the mutants that showed lower activity scores? There is a possibility that due to low read cutoff, these mutants might be lying in the range with low reads in the unselected library.

4.  Line 223-224 – Authors mention they find a good correlation between activity and abundance score. Although this is noticeable from the scatter plot but supporting high-throughput data with statistical parameters like pearson correlation coefficient, a metric that provides comparison between 2 datasets will make this data reporting more quantitative and informative.

5. Line 1340- Figure 3b- What do authors mean by variants with small enough error? Please be precise.

6.  Line 315-319/ Line 1550-1556- Extended data Fig 15c – It is difficult to interpret the inference reported that is based on the data in Extended Fig 15. There is no data reported for Normalized AMC cleavage for Y268W. Interpretation can be more comprehensible by plotting a scatter plot between the Normalized Activity Fitness Scores obtained from DMS data and Normalized AMC cleavage (%). Through this plot, the reader can easily make out the outlier.

7. Line 364-367 – Authors mention “M208W strikingly increases the protein melting temperature by over 5C, indicating a substantial improvement in thermal stability. Increased stability, and thus reduced turnover in cells, may provide a mechanism to explain leaky expression in our cellular assay and increased yield of recombinant protein for E.coli expression.” Since, leaky expression is a different issue, it is confusing why will leaky expression be a plausible reason for increased stability but less activity? 

8. Since Extended data Fig 11a shows that variants display substantial amount of leaky expression, how have the authors taken this information into account while inferring results from DMS activity scores, especially since they are quantifying at the RNA level and not at the DNA level? Can the activity scores obtained for the mutants be normalized to leaky expression scores in some way, for example by subtracting the scores obtained from the leaky expression dataset in order to measure the true activity of each mutant?

9. Solvents are known to affect an enzyme’s activity, selectivity and stability. In Figure 5, authors should consider and comment about the role of solvent in understanding the mechanism of Michelis-Menten kinetics of M208 variants using substrates Z-RLRGG-AMC, Ubiquitin-Rhodamine and ISG15-Rhodamine.

 Minor points:

1. Figure numbers need to be reformatted. Figure 3 onwards they are incorrectly labelled. For eg. ‘Fig 3’ is labelled as ‘Fig 1’.

2. Line 426-429 – In Figure 3b, L and R domains of papain should be labelled or highlighted in separate colors for the ease of understanding for the reader.

3. Overall, different DMS datasets obtained from different assays in the paper have different read cut-offs such as 10, 13 and 18. A consistent statistical logic for obtaining different read cut-offs across different DMS datasets will be helpful. Also, increasing the read cut-off might improve the data quality and minimize sequencing artefacts.

• Reviewed by Priyanka Bajaj and James Fraser

Competing interests

The authors declare that they have no competing interests.