{"id":"fa4e1a35-470e-4a7b-98c0-b789f4b6ae02","arxiv_id":"2504.18559","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Molecular dynamics on 10 dual-pocket PfHT1 inhibitors ranks key residues in the allosteric and orthosteric pockets and shows fructose derivatives gain affinity when orthosteric residues are mutated.","lead":"This paper uses molecular dynamics simulations to rank which amino acids in the malaria transporter PfHT1 control binding of dual-pocket carbohydrate inhibitors. It then uses in-silico mutations to test those rankings, and finds that fructose-linked inhibitors gain potency when orthosteric site residues are mutated.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Allosteric residue rankings rest on a forced pocket closure that never occurs in unbiased simulation; the key state is defined by the authors' SMD bias, not by the system, so the central determinant-ranking claim lacks independent support.","rationale":"The central claim has two components: (1) the computed binding free energies capture the experimental IC50 trend, and (2) per-residue decomposition plus mutation analysis ranks the molecular determinants of dual inhibition. The reader's weakest assumption targets component (2), and the manuscript text supports that concern directly. Unlike the trend claim, which is at least a direct comparison between computed and experimental values, the determinant ranking depends entirely on the allosteric pocket obtained after a post hoc reversal of the simulation direction. The authors are transparent about this reversal, but transparency does not establish physical relevance. The absence of error bars on the reported ΔGbinding values is a separate weakness that affects component (1), but the forced-pocket issue is more load-bearing because it determines which residues are even considered as allosteric determinants. This is not an internal inconsistency or a disagreement with consensus; it is a correctness risk in the central construction. The concern is concrete and testable: an unbiased simulation starting from the holo crystal state or a restraint-free continuation of the SMD-closed state would show whether the pocket is a real conformational state. If the test fails, the residue rankings should be treated as hypotheses from a biased path, not as validated determinants. Because the trend claim may still hold and the allosteric pocket could in principle be legitimate, the appropriate verdict remains CONDITIONAL rather than REJECT; the paper should be revised to include this validation or explicitly reframe the rankings as simulation-bias-dependent predictions.","tokens_in":8266,"tokens_out":5815,"duration_ms":58034,"concrete_test":"Perform 1-µs unbiased MD in three replicates starting from (a) the holo 6M2L/6M20 complex before removal of native ligand C3361 and (b) the final SMD-closed HTI9 structure from §3.1 with all pulling restraints removed. Check whether the same allosteric pocket geometry (Lys51–Asp447 contact, Leu47 and Val443 within contact distance, pocket volume matching Fig. 2) forms spontaneously in (a) and remains stable in (b). If the pocket does not appear in (a) and opens in (b), the SMD-reversed state is an artifact, and the allosteric-site residue rankings in Figs. 5–6, 8–9, and 11 are not representative of PfHT1 dual inhibition.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The manuscript's own Section 3.1 states that in the forward allosteric-communication simulation 'no allosteric pocket is found to form after HTI9 arrives at the orthosteric site'. The authors then switch to a 'reversed allosteric communication' protocol in which the TM1e helix is pulled toward the inhibitor by a steered-MD umbrella potential. Every allosteric-site residue ranking in the paper (Lys51, Leu47, Val443, Asp447 in Figs. 5, 6, 8, 9, and 11) and the PMF comparison in Fig. 2(d) are computed on conformations generated by this imposed bias. The load-bearing condition for the central claim is that this forced pocket corresponds to the binding-competent allosteric state of PfHT1. That condition is not independently established: the native allosteric ligand C3361 is removed from the crystal structures 6M2L/6M20, no comparison is made to the experimentally observed allosteric pocket, and no unbiased control trajectory shows the same pocket geometry being visited spontaneously. If the SMD bias selects a non-native conformational route, the residue rankings and the mutation results are artifacts of the pulling protocol rather than molecular determinants of dual inhibition.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents molecular dynamics (MD) simulations of the Plasmodium falciparum hexose transporter 1 (PfHT1) bound to dual orthosteric/allosteric inhibitors. Ten inhibitors are classified into three groups according to linker length, tail-group size, and sugar moiety, and the authors compute MM-GBSA binding free energies, pocket volumes, SMD/WHAM PMF profiles for allosteric pocket formation, and per-residue free energy decompositions. The central claim is that the computed free energies reproduce published experimental IC50 trends, and that per-residue decomposition, combined with alanine mutation analysis, reveals the key molecular determinants, with Lys51, Leu47, Val443, and Asp447 at the allosteric site and several residues at the orthosteric site. The paper also reports an outlier behavior for a fructose derivative, whose binding affinity is enhanced when orthosteric residues are mutated.","tokens_in":8487,"tokens_out":6666,"duration_ms":61304,"significance":"If the results are reliable, the paper offers a computational protocol for ranking residue-level determinants in dual-site inhibition and generates a testable prediction about fructose-derived inhibitors being sensitized by orthosteric mutations. The work has several strengths: it benchmarks against published IC50 data without fitting parameters, uses explicit membrane and solvent environments, performs three replicas per system, and cross-validates the key-residue hypothesis with multiple independent analyses (PMF, pocket volume, decomposition, and mutation studies). The significance is, however, conditional on the allosteric pocket being sampled without a potentially biasing steering protocol and on the statistical reliability of the reported free-energy differences, neither of which is established in the present manuscript.","major_comments":[{"comment":"The allosteric pocket on which all residue rankings rest is generated by an external bias. The authors state that 'no allosteric pocket is found to form after HTI9 arrives at the orthosteric site', and then switch to a 'reversed allosteric communication' protocol in which the TM1e helix is pulled toward the inhibitor (Section 2.1, third paragraph). All subsequent allosteric-site analyses, including pocket-volume comparisons (Figs. 4(b) and 7(b)), per-residue decompositions (Figs. 5, 6, 8, 9), PMF profiles (Fig. 2(d)), and mutation validations (Fig. 11), use conformations produced by this SMD bias. The paper does not show that an unbiased simulation visits a similar closed-pocket geometry, and it does not compare the forced pocket with the experimentally characterized allosteric site of PfHT1, where the native ligand C3361 is bound in the crystal structures 6M2L/6M20. Therefore the key-residue rankings and PMF conclusions are conditional on the unvalidated assumption that the steered pathway corresponds to the physiologically relevant binding-competent state.","section":"3.1 (Figs. 2, 3)"},{"comment":"The claimed agreement with experimental IC50 data is presented qualitatively, with no error bars, correlation coefficients, or significance tests. Although the text states that 'three parallel simulations were performed for all cases' (Section 2.1, first paragraph), the reported ΔGbinding values, pocket volumes, PMFs, and per-residue decomposition energies appear without standard deviations or confidence intervals. Several differences that support the main claims are visually small; for example, the ΔGbinding difference between HTI8 and HTI10 in Fig. 4(c) is modest, as is the difference between Glu-O3 and Glu-O2 in Fig. 10(c). Without per-replica statistics and a rank-correlation measure (e.g., Spearman's ρ with a p-value), the statement in the Introduction of a 'high correlation' with experimental data is not quantitatively supported.","section":"2.1 and 3.2 (Figs. 4, 7, 10)"},{"comment":"The comparison of MM-GBSA binding free energies with quantities labeled ΔGexp=RTlnIC50 is conceptually problematic. IC50 is a functional assay readout, not an equilibrium binding constant, and converting it to an energy via RTlnIC50 introduces an arbitrary offset and assumes a constant relationship between IC50 and binding affinity across chemically diverse compounds. The manuscript should state this approximation explicitly and, for robustness, compare computed ranks with experimental ranks rather than plotting both quantities on the same energy scale, or should clearly indicate that only the trend, not the absolute values, is meaningful.","section":"3.2 (Figs. 4(c), 7(c), 10(c))"}],"minor_comments":[{"comment":"The phrase 'Our binding free energy analysis capture' should be 'Our binding free energy analysis captures'; also 'essential trend' is used twice (abstract and Introduction) and could be replaced with 'overall trend' for clarity.","section":"Abstract"},{"comment":"The sentence describing the SMD protocol is ambiguous: it mentions 'constant-velocity (0.1 nm/μs)' pulling but then says 'by an umbrella potential,' which are two different approaches. Please clarify whether this is a single constant-velocity pull or an umbrella-sampling setup.","section":"2.1, third paragraph"},{"comment":"The blank entries in Table 2 (Ile310 for Glu-O2, Phe403 for Glu-O2 and Fru, Trp436 for Fru) are not explained; please indicate whether these are non-favorable interactions, interactions below a threshold, or missing values, and add a footnote.","section":"Table 2"},{"comment":"The cross-reference '(Figs. 10 (d) and (e))' for the free energy decomposition appears incorrect: Fig. 10(d) is labeled 'Experimental IC50 trend,' while the decomposition is shown in Fig. 10(e). Please fix the reference.","section":"3.2.3"},{"comment":"The sentence 'This is in line with the experimental findings' regarding Lys51 and Asp447 is too broad; the cited experimental work identifies the allosteric ligand and overall pocket, not necessarily a residue-level mechanism, so the specific experimental observation being compared should be stated.","section":"3.1, final paragraph"},{"comment":"No data availability statement or reproducibility details (topology files, input scripts, MM-GBSA parameters such as dielectric constants and entropy treatment) are provided; at least the MM-GBSA setup should be reported in the Methods or Supplementary Information.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere's my take on arXiv:2504.18559. The genuinely new content is the per-residue free-energy ranking of dual-site PfHT1 inhibitors and the in silico result that a fructose derivative gains affinity when orthosteric residues are mutated to alanine. The authors group ten inhibitors by linker length, tail size, and sugar moiety, run MD/MM-GBSA, and show their computed binding free energies track the experimental IC50 trends within each group. That agreement is a real success, not a circular one: no parameters are fitted to the IC50s. The mutation analysis is a sensible way to test the rankings, and the fructose sensitization with a changed binding pose is a concrete, falsifiable prediction that medicinal chemists could check.\n\nThe soft spot is load-bearing. Section 3.1 states plainly that in the forward allosteric-communication simulation, no allosteric pocket forms after the inhibitor reaches the orthosteric site. The authors then switch to a 'reversed' protocol where they pull the TM1e helix toward the inhibitor with an umbrella potential. Every allosteric-site residue ranking and the PMF in Fig. 2 depend on conformations generated by that imposed bias. The manuscript does not show that this forced pocket matches the allosteric pocket observed in the experimental structures, nor does any unbiased trajectory visit the same geometry. So the rankings for Lys51, Leu47, Val443, Asp447 are conditional on the pulling protocol. That is the main thing I would push the authors to fix.\n\nOther issues are milder. No error bars or convergence metrics are reported despite three replicas. No code, data, or parameter files are provided, which makes the computational results harder to audit. The mutation validation is entirely in silico, which is fine for a predictive study but should be framed as such. The claim of consistency with experiment [10,11] is plausible but could use more explicit referencing for the Lys51–Asp447 interaction.\n\nWho gets value from this: researchers working on PfHT1 or dual-pocket inhibition, and computational medicinal chemists interested in MM-GBSA workflows. The paper deserves a serious referee and, after revisions, could be a useful contribution. I would send it to peer review, but I would insist on a response to the allosteric-pocket validity question—ideally by comparing the SMD-generated pocket to the experimentally bound state and running an unbiased enhanced-sampling control.","headline":"Solid computational follow-up with a useful residue-level map, but the allosteric rankings ride on a forced pocket closure that needs independent validation.","tokens_in":9028,"tokens_out":3094,"would_cite":false,"duration_ms":31152,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Molecular dynamics identifies the residues that control dual-site inhibition of the malaria transporter PfHT1 and reproduces experimental potency trends.","keywords":["PfHT1","antimalarial drug resistance","orthosteric-allosteric dual inhibition","molecular dynamics","binding free energy","per-residue decomposition","mutation analysis","carbohydrate inhibitors"],"falsifier":"Determine the co-crystal structure of PfHT1 with HTI9 and check whether the allosteric pocket is closed with a Lys51–Asp447 salt bridge matching the simulated forced-closure conformation; if the crystal shows an open or differently shaped pocket, the residue rankings built on that trajectory are not representative. A complementary check would measure IC50 values for Lys51Ala and Asp447Ala mutants against HTI8 or HTI9 and compare the predicted affinity losses.","tokens_in":8040,"feed_emoji":"💊","tokens_out":9969,"duration_ms":86281,"temperature":0.7,"pith_summary":"This paper uses molecular dynamics simulations to explain why carbohydrate-based inhibitors of the malaria transporter PfHT1 differ so widely in potency, even when they share the same sugar core and tail group. The authors compute binding free energies for ten dual-site inhibitors divided into three groups by linker length, tail size, and sugar type, and report that the computed energies reproduce the experimental IC50 ordering in each group. Per-residue energy decomposition then ranks the residues that matter most: Lys51, Leu47, and Val443 dominate allosteric-site binding, with Asp447 stabilizing the pocket, while orthosteric-site residues control sugar-specific recognition. Mutation simulations validate the rankings, with one clear exception: a fructose derivative binds more tightly when orthosteric residues are mutated to alanine because the mutation changes its pose. If these rankings are right, they give a residue-level map for designing dual-site antimalarial inhibitors that can circumvent resistance.","feed_headline":"Simulation matches IC50 trends and ranks PfHT1 dual-site key residues","feed_subtitle":"Calculated binding energies track IC50 trends across linker, tail, and sugar variants, pointing to Lys51, Leu47, Val443, and Asp447.","key_machinery":"The rankings come from two linked computational devices. First is the per-residue decomposition of the MM-GBSA binding free energy (molecular mechanics with generalized Born and surface-area solvation, a standard end-state estimate of binding affinity), which sums each amino acid's contribution and produces the residue rankings. Second is the reversed allosteric communication protocol, in which steered molecular dynamics pulls the TM1e helix toward the inhibitor and WHAM (weighted histogram analysis) yields the potential of mean force; because no allosteric pocket formed in ordinary forward simulations, this forced-closure trajectory is the basis for identifying Lys51 as pocket-forming and the Lys51–Asp447 salt bridge as pocket-stabilizing.","core_discovery":"The paper's central claim is that MM-GBSA binding free energies from molecular dynamics reproduce the experimentally observed potency order of the published PfHT1 dual inhibitors, and that per-residue decomposition identifies which residues carry each part of inhibitor binding. For the allosteric pocket, Lys51, Leu47, and Val443 are the dominant determinants, while the Lys51–Asp447 interaction gates pocket closure and stabilization. For the orthosteric pocket, residues including Asn311, Asn435, Gln305, Gln306, Trp390, and Phe403 discriminate the sugar moiety. Since no allosteric pocket formed spontaneously after HTI9 reached the orthosteric site in forward simulations, the authors deliberately pulled helix TM1e toward the inhibitor to close the pocket, treating this as reversed allosteric communication, and used pocket volume and PMF profiles to argue that Lys51 forms the pocket and Asp447 stabilizes it. Mutating the ranked sites to alanine generally lowers calculated affinity, with the outlier that fructose gains affinity on orthosteric mutation because its binding pose shifts and new residues become available.","pith_inferences":["Editorial inference: if the forced-closure pocket is the true cryptic allosteric site, a compound that pre-stabilizes the Lys51–Asp447 salt bridge before its sugar reaches the orthosteric site should bind more potently, since the PMF barrier to pocket closure would be lowered.","Editorial inference: the mutation results imply a testable resistance scenario in which parasites carrying orthosteric-site mutations that impair glucose transport become selectively vulnerable to fructose-based inhibitors, because those compounds bind better in the mutant background.","Editorial inference: the reversed-communication protocol is a transferable way to search for cryptic allosteric pockets in other transporters, but its residue rankings would need structural validation before being used for drug design."],"forward_implications":["Allosteric-site design should aim at hydrophobic contacts with Lys51, Leu47, and Val443, and should preserve the Lys51–Asp447 interaction to keep the pocket closed.","The nonmonotonic linker-length dependence (best near n=9) follows the summed contribution of the three allosteric residues rather than pocket volume, so linker design should be tuned against those contacts.","Mutations at orthosteric residues will not weaken all sugar-based inhibitors equally: fructose derivatives are predicted to gain affinity through pose change, suggesting a scaffold for resistant backgrounds.","The per-residue maps provide concrete residue-level targets for designing one-molecule dualsteric compounds that occupy both the orthosteric and allosteric sites."],"supporting_citations":[{"why":"supplies the PfHT1 crystal structures, the ten carbohydrate dual-inhibitors, and the experimental IC50 values against which the computed binding free energies are judged","marker":"[10,11]"},{"why":"provides the MM-GBSA protocol used to calculate ΔGbinding and per-residue energy contributions","marker":"[25]"},{"why":"provides the WHAM procedure used to compute the PMF profiles for allosteric-pocket formation and stabilization","marker":"[23]"},{"why":"identifies the conserved glucose/fructose orthosteric uptake residues used as the experimental baseline for orthosteric mutation effects","marker":"[30]"},{"why":"supplies the docking procedure used to construct the PfHT1–inhibitor starting complexes","marker":"[15,16]"}],"fun_headline_variants":["MD pinpoints PfHT1 dual-site anchor residues","Key residues for PfHT1 dual inhibition revealed","Fructose outlier breaks alanine mutation trend in PfHT1","Allosteric pocket gate: Lys51 and Asp447 in PfHT1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The allosteric-pocket rankings assume that the pocket shape made by pulling one protein helix toward the inhibitor, done only after no allosteric pocket formed on its own, is the same shape a real inhibitor would create.","fun_headline_variants_meta":{"raw":{"variants":["MD pinpoints PfHT1 dual-site anchor residues","Key residues for PfHT1 dual inhibition revealed","Fructose outlier breaks alanine mutation trend in PfHT1","Allosteric pocket gate: Lys51 and Asp447 in PfHT1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000731,"raw_usage":{"total_tokens":3283,"prompt_tokens":970,"completion_tokens":2313,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":586,"completion_tokens_details":{"reasoning_tokens":2243}},"tokens_in":586,"tokens_out":2313,"duration_ms":16145,"temperature":1.0,"reasoning_tokens":2243,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:00:38.046021+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Determine the co-crystal structure of PfHT1 with HTI9 and check whether the allosteric pocket is closed with a Lys51–Asp447 salt bridge matching the simulated forced-closure conformation; if the crystal shows an open or differently shaped pocket, the residue rankings built on that trajectory are not representative. A complementary check would measure IC50 values for Lys51Ala and Asp447Ala mutants against HTI8 or HTI9 and compare the predicted affinity losses.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the MM-GBSA protocol used to calculate ΔGbinding and per-residue energy contributions"},{"cited_title":"Kumar, J","cited_arxiv_id":null,"evidence_quote":"provides the WHAM procedure used to compute the PMF profiles for allosteric-pocket formation and stabilization"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"identifies the conserved glucose/fructose orthosteric uptake residues used as the experimental baseline for orthosteric mutation effects"}],"review_version":1}