Review papers with raw data transparency

Concise critique

The study presents a highquality deep mutational scanning (DMS) map of HIV1 Vif residues 12–115 in two clade B backgrounds (LAI and 1203) to quantify Vif fitness for APOBEC3G antagonism using a twopass SupT1A3G selection and codonlevel libraries (~2000 missense variants per strain). Major findings are (1) pervasive purifying selection with most substitutions deleterious, (2) pockets of unexpected mutational tolerance at some structurally important interfaces (eg H42 and Q83 substitutions), and (3) pronounced straindependent epistasis (56 variants across 38 positions with divergent fitness between LAI and 1203) backed by infectivity validations for Q83 variants DMS mapping of Vif fitness vs A3G

Detailed paper review and critique

1 Summary of aims and approach

The authors aimed to isolate the contribution of APOBEC3G (A3G) selective pressure on HIV1 Vif sequence space by applying codonlevel deep mutational scanning (DMS) to residues 12–115 of Vif in two divergent clade B backgrounds (labadapted LAI and primary isolate 1203). Libraries were designed to avoid introducing A3Gpreferred 5CC motifs, cloned into a replicationcompetent proviral backbone that decouples vif from pol overlap, packaged in HEK293T and selected by two serial lowMOI infections in SupT1 cells expressing physiological A3G (SupT1A3G). Postselection and pooled preselection libraries were deepsequenced and variant enrichment (log enrichment scores) computed; A3Gmutation footprints (GtoA at A3G motifs) were quantified as an orthogonal proxy for loss of Vif function DMS assay design and selection

2 Major strengths

• Appropriate experimental design to isolate A3G pressure — The proviral backbone decoupling vif from pol removes overlappingframe constraints that confound many evolutionary analyses and allows direct interrogation of Vif sequencefunction relationships for A3G antagonism Pol overlap decoupling
• Codonlevel design minimizing A3Gconfounding — Choosing alternative codons during oligo design to avoid creating A3Gtarget 5CC motifs is an important practical control that reduces artifactual depletion from A3Ghypermutation of the encoded variant calls Codon design to avoid A3G motifs
• Orthogonal validation — The authors validated enrichment scores with A3Gmutation footprints (GtoA at predicted motifs) and with independent infectivity and packaging assays for select variants (eg Q83C, Q83S), strengthening claims that enrichment reflects A3Gantagonism rather than generic replication effects Correlation enrichment with A3G footprints and infectivity assays
• Comparative crossstrain analysis — Running DMS in two Vif backgrounds (LAI, 1203) exposed epistasis and evolutionary contingency (56 variants across 38 positions displayed divergent fitness consequences), a major conceptual advance for understanding hostvirus coevolution Crossstrain divergent variants

3 Key results and interpretation

The DMS data show a librarywide median log enrichment of approximately 1.23 depletion for LAI (most single aa substitutions reduce fitness under A3G selection). Sitelevel medians identified 9 constrained positions in LAI and 24 sites with relatively increased tolerance; residue 42 (H42) unexpectedly tolerated many substitutions while still retaining activity, mapped to the RNAbinding interface in the cryoEM VifA3GVCBCRNA structure (PDB 8CX0). Q83 variants were strongly backgrounddependent: Q83C and Q83S were highly enriched in LAI but depleted in 1203; infectivity assays confirmed LAIspecific enhancement of A3G antagonism for Q83C/Q83S, while 1203 showed no advantage for these variants Sitelevel and Q83 results

4 Critical limitations, caveats, and alternative explanations

1. In vitro to invivo extrapolation — SupT1A3G cells are a controlled, useful model but do not recapitulate the multicellular, immune, and populationlevel dynamics shaping Vif evolution in patients; the authors acknowledge low correlation between DMS mutational tolerance and natural conservation, because conservation integrates many pressures (other A3s, immune epitopes, overlapping frames) and phylogenetic history Limits comparing DMS to conservation
2. Assay noise and incomplete depletion — The authors point out that some highly enriched substitutions may be assay artefacts due to incomplete depletion of nonfunctional variants after a single round; they therefore emphasise site medians and multi replicate reproducibility. For rare, largepositive enrichment events, independent lowthroughput validations are essential (and some were provided for Q83) Assay noise caveat
3. Library scope and Cterminus exclusion — The study deliberately excludes Vif residues 116–192 (Cterminal domain) because of VCBC/Cul5 interactions; this narrows the conclusions to the Nterminal functional landscape but leaves gaps about Cterminal contributions to antagonism and multifunctionality (eg PP2A, other host interactions) Cterminus exclusion
4. Hostfactor complexity — 1203 evolved in a host homozygous for A3H haplotype II; this history likely imposed different selective constraints (balance between antagonizing A3G, A3H, and potentially PP2A), so differences between strains reflect both historical selection and current assay context; the authors discuss this and map divergent sites onto known CBFβ/A3G interfaces Hosthistory shaping 1203 Vif

5 Relationship to structural literature and mechanistic plausibility

The DMS findings dovetail with recent highresolution cryoEM structures of Vif bound to human A3G, CBFβ and components of the Cullin5 E3 complex showing RNAbridged interfaces (PDB 8CX0). The structural data provide a plausible mechanistic basis for why substitutions at RNAcontacting residues (eg H42) can be tolerated while maintaining A3G antagonism, because RNA acts as a molecular glue positioning contact residues; conversely, substitutions that perturb CBFβ or Cullin5 interfaces are strongly deleterious CryoEM structure VifA3GVCBC

6 Specific suggestions to strengthen the claims / further experiments

1. Perform DMS selections in an alternative cell line expressing physiologic A3G plus other APOBECs (eg A3F or A3H) singly and combinatorially to quantify pleiotropic constraints from multiple restriction factors and to partition shared vs specific constraints.
2. Publish raw sequence count matrices and processing code (UMI consensus pipeline, codon count tables) in a public repo and deposit reads to SRA/ENA with accession numbers to maximize reproducibility; the paper references supplementary data but accession numbers were not present in the provided text.
3. Targeted doublemutant validation (double mutant cycles) at key epistatic loci (eg background residues that differ between LAI and 1203 near position 83) to quantify pairwise epistasis magnitudes and test whether Q83C benefit in LAI is reversed by specific background residues in 1203 (this would directly probe mechanistic epistasis rather than correlative differences) Double mutant cycles and epistasis caution
4. Measure Vif protein levels and stability for enriched variants to rule out spurious enrichment via altered expression rather than altered A3Gbinding per se (CBFβ affects Vif biosynthesis; differences in steady state levels can confound selection) CBFbeta effects on Vif levels

7 Practical implications and future directions

Mapping Vif fitness constrained by A3G highlights potential targets for smallmolecule or biologic disruption of VifA3G interactions. For example, residues strongly constrained in DMS and contacting CBFβ or Cullin5 represent conserved functional hotspots unlikely to tolerate resistancemutations; structural data (PDB 8CX0) can inform rational inhibitor design. Conversely, pockets of tolerance (eg H42) suggest that targeting those surfaces alone may permit viral escape via alternative substitutions and are less promising therapeutic targets Therapeutic targeting implications

8 Reproducibility assessment

Methods are described in sufficient technical detail (codon pool sourcing, cloning sites MluI XbaI, UMI consensus, two passage schedule, MOI, Spearman reproducibility ranges for replicates provided LAI ρ = 0.46 0.65). The main reproducibility gaps are lack of explicit public read accession numbers and code for UMI processing in the provided text; providing these would increase reproducibility from 8 to 9+ on an objective scale Reproducibility details and gaps

9 Conclusion and balanced judgement

This paper is a rigorous, well controlled application of DMS to a central hostvirus conflict, generating a highresolution map of Vif residues important for A3G antagonism and revealing strain dependent epistasis with orthogonal validation. The principal limitations are (a) in vitro selection context versus invivo complexity and (b) incomplete deposition of raw processing code/read accessions in the provided text. Overall the study meaningfully advances understanding of molecular arms race dynamics between Vif and A3G and provides actionable structural targets for further validation and therapeutic exploration Overall study conclusion

Run an iterative bioinformatics agent to reanalyse raw counts, compute sitelevel epistasis, or model escape pathways with structural mapping