{"id":"ae7f77bd-c2f1-452a-86fa-b556800335c5","arxiv_id":"1909.00911","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A coarse-grained simulation of TAZ1 with both HIF-1α and CITED2 present shows the CITED2 complex is thermodynamically and kinetically favored.","lead":"This paper uses computer simulations of simplified molecular models to explain why CITED2 outcompetes HIF-1α for binding to the TAZ1 protein domain. The simulations show CITED2 wins because its complex forms faster and is more stable, matching earlier experiments and clarifying a key step in the cell's low-oxygen response.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ternary CB-over-HB ordering rests on a single hand-tuned (βH, βC) point; no sensitivity or uncertainty analysis, so the 1.91 kT gap may be a calibration artifact.","rationale":"The paper is a careful structure-based simulation study, and the qualitative agreement with the known experimental competition result is real. The central claim, however, is the model's prediction that CITED2 dominates both thermodynamically (CB 1.91 kT below HB) and kinetically. What must be true for this claim to hold is that the calibrated model is not accidentally biased toward CITED2 by the tuning procedure. The weakest point is that βH and βC are set to the values that make the two binary binding free energies equal; the ternary ranking is then a derived quantity with no sensitivity analysis or error bars. This is not an internal contradiction, and it does not make the paper fraudulent; it makes the headline quantitative ranking unproven. The reader identified the same weakness in the weakest_assumption field. My proposed check is concrete and would settle whether the ranking is robust. Because the concern is addressable and the experimental direction already supports the qualitative conclusion, I keep the reader's CONDITIONAL verdict.","tokens_in":18623,"tokens_out":8515,"duration_ms":93377,"concrete_test":"Repeat the ternary REMD with two alternative (βH, βC) pairs that also reproduce the binary Kd ≈ 10 nM (for example, βH=1.05 with βC≈0.93 and βH=1.15 with βC≈0.98, in the neighborhood of Fig. S8), plus an uncalibrated βH=βC=1.0 reference run. For each set, recompute the CB-HB gap from the WHAM free-energy surface and estimate its uncertainty by block bootstrapping the REMD replicas. If the gap stays above 1 kT with the same sign in all calibrated sets, the calibration-sensitivity concern is resolved; if it falls below kT or changes sign, the claimed dominance is an artifact of the chosen tuning.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central thermodynamic claim in Section 2.2 is that the CB basin is 1.91 kT lower than HB on the ternary free-energy surface (Fig. 2). This number is produced by a single set of calibrated interaction strengths, βH=1.1 and βC=0.95, chosen in SI Section 1.2 (Fig. S8) so that each binary complex matches the experimental Kd of about 10 nM. Matching two binary Kd values does not constrain the ternary CB-HB gap, and many (βH, βC) pairs are compatible with those binary data. The paper does not vary βH, βC, the weighted contact map, or the hinge-region assignments and re-measure the ternary ordering; no bootstrap or other error estimate is given for the 1.91 kT gap or the mean FPTon values. If an equally valid neighboring calibration reverses or erases this gap, the thermodynamic and kinetic dominance claims, and the pathway analysis built on them, are not established. The Table 1 caption itself concedes that the simulated Kd values are not the experimental apparent Kd values, so the comparison with experiment cannot by itself validate the parameter choice.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs coarse-grained, structure-based models of the binary TAZ1-HIF-1α and TAZ1-CITED2 complexes and a ternary TAZ1-HIF-1α-CITED2 system, using REMD simulations with weighted NMR contact maps and Debye-Hückel electrostatics. The binary models are calibrated so that both complexes reproduce the experimental 10 nM dissociation constant and low unbound helical content. On the ternary free-energy surface (Fig. 2), the CITED2-bound basin (CB) is reported as the lowest free-energy state, 1.91 kT below the HIF-1α-bound basin (HB), with an intermediate state (IS) and an unbound state (UB) at higher free energy. Kinetic simulations (Tab. 2) report faster direct binding of CITED2 (mean FPTon 0.286 ns vs 1.431 ns) and faster replacement of HIF-1α by CITED2 (60.758 ns vs 163.202 ns), with CITED2 reaching the bound state first in 177 of 200 runs. The paper concludes that CITED2 is thermodynamically and kinetically dominant in the ternary competition, consistent with NMR and fluorescence experiments (Berlow et al., Nature 2017), and analyzes binding-order and φ-value differences between direct and replacement pathways.","tokens_in":18937,"tokens_out":3685,"duration_ms":41398,"significance":"If the central claim is established, the paper provides a mechanistic, structure-based explanation for the experimentally observed dominance of CITED2 over HIF-1α in ternary competition with TAZ1, including a concrete intermediate state and distinct binding orders for direct and replacement pathways. Strengths of the study are its explicit use of the full NMR ensembles to build weighted contact maps, calibration to two independent experimental observables (Kd and helical content), direct comparison of simulated φ values with experimental values for HIF-1α, and the clear mapping of thermodynamic basins and kinetic pathways on a two-dimensional free-energy surface. The qualitative agreement with the experimental competition data is encouraging and the pathway analysis generates falsifiable predictions about which ligand regions are most important in each replacement process. The main weakness is that the ternary ranking is computed at a single hand-tuned point in parameter space, with no sensitivity or uncertainty analysis, so the quantitative robustness of the 1.91 kT gap and the kinetic ratios is not yet established.","major_comments":[{"comment":"The central thermodynamic claim, that CB is 1.91 kT lower than HB in the ternary system, is computed with a single parameter set βH=1.1 and βC=0.95, tuned so that the two binary complexes each match the experimental Kd of about 10 nM. Matching two binary binding free energies does not constrain the ternary free-energy balance between two partially overlapping binding sites, and many (βH, βC) pairs near the fitted values would also reproduce the binary data. The paper does not test whether the CB-over-HB ordering survives variations in βH and βC, in the weighted contact-map construction, or in the hinge-region assignments, and no bootstrap or block-averaging uncertainty is given for the basin free energies. Since the entire conclusion that CITED2 dominates thermodynamically rests on this gap, please provide a sensitivity analysis (for example, re-computing the ternary free-energy surface over the βH ranges shown in Fig. S8 and over the βC ranges) and report the uncertainty of the CB-HB free-energy difference, or state explicitly why the ordering is structurally robust.","section":"Section 2.2, Fig. 2; SI Section 1.2, Eq. S3, Fig. S8"},{"comment":"The simulated ternary Kd values (UB to HB 62.8 nM; UB to CB 1.38 nM) differ from the experimental apparent Kd values (900 nM and 0.2 nM) by factors of roughly 14 and 7 in opposite directions, and the Table 1 caption concedes that the simulated and experimental values are not directly comparable because of different calculation methods. The paper nevertheless states that the results are 'consistent with the tendency of experimental apparent Kd.' This comparison would be more convincing if the authors quantified the expected systematic offset of the model (e.g., from the effective simulation concentration, neglect of explicit solvent, or the coarse-grained energy function) and demonstrated that the sign and order of magnitude of the discrepancies are consistent with that offset, rather than treating the trend alone as validation. At minimum, the paper should avoid implying quantitative agreement that the reported numbers do not support.","section":"Table 1, Section 2.2"},{"comment":"The kinetic dominance claim is based on mean first-passage times (0.286 ns vs 1.431 ns for direct binding; 60.758 ns vs 163.202 ns for replacement) obtained from 200 trajectories per state, but no standard errors, confidence intervals, or distribution statistics are reported. The replacement times in particular are described as including many unsuccessful binding attempts, which suggests broad distributions and possibly large sampling uncertainty; the ratio 60.758/163.202 could be sensitive to a few outlier trajectories. Please report the standard error of the mean (or equivalent uncertainty) for each FPTon value and the 177/23 split, and state whether the observed differences are statistically significant under a suitable resampling test.","section":"Section 2.3, Table 2 and Fig. 4"}],"minor_comments":[{"comment":"The phrase 'the simulations prove the dominant position' is too strong for a coarse-grained model with calibrated parameters; 'suggest' or 'indicate' would be more accurate and would better match the exploratory nature of the study.","section":"Abstract"},{"comment":"In the sentence beginning 'In the experiment, the isolate HIF-1α or CITED2...', 'isolate' should be 'isolated'.","section":"SI Section 1.2"},{"comment":"There is a typo 'The mean FPTon of of each possible pathway'; 'of of' should be 'of'. Also, in Table 2 the header row repeats 'mean FPTon' under both 'Direct binding process' and 'Replacing process'; this is redundant but not confusing.","section":"Section 2.3, Fig. 4 caption and text"},{"comment":"The comparison with experimental φ values is partial because only a few HIF-1α residues are labeled and no CITED2 experimental φ values are shown; the V825A discrepancy is discussed, but a brief statement of how many experimental points were compared and the overall agreement rate would help the reader judge the comparison.","section":"Section 2.4, Fig. S5B"},{"comment":"The notation 'εDHVDebye−H ¨uckel' is typeset awkwardly; please define the symbol for the Debye-Hückel term explicitly and use a consistent notation in Eq. S1 and Eq. S2.","section":"SI Section 1.1, Eq. S1"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a computational biophysics/biomolecular simulation journal. The central qualitative result is plausible and consistent with experiment, but the lack of sensitivity analysis for the ternary ordering is a genuine load-bearing gap. I would not recommend acceptance until the authors demonstrate that the CB-over-HB conclusion is robust to reasonable parameter variation and provide at least a minimal uncertainty estimate for the reported free-energy differences and kinetic times. The abstract's 'prove' wording should also be softened in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a solid, if parameter-sensitive, coarse-grained study of the ternary TAZ1-HIF-1α-CITED2 competition. The new thing is the ternary model itself—previous work and the Berlow experiments only looked at binary complexes or final competition outcomes. Simulating all three molecules together lets the authors identify an intermediate state (IS) and show that CITED2 wins both thermodynamically (CB lower than HB) and kinetically (faster direct binding, faster replacement). That qualitative result is consistent with the experimental apparent Kd values from Berlow et al., and it is not a trivial consequence of the binary fits, because the ternary ordering emerges only when the two ligands compete for an overlapping site.\n\nThe paper also does a few things well. The weighted contact map from all 20 NMR conformers is a reasonable way to handle structural heterogeneity. The calibration strategy is transparent: helical content at low values and binary Kd to set βH and βC. The phi-value comparison with experiment is partial but honest—they flag the V825A mismatch and offer a plausible explanation. The kinetic simulations with 200 runs per process are cheap but give consistent ordering.\n\nThe soft spots are real. The central 1.91 kT gap between CB and HB is computed at a single (βH, βC) point, and while the binary fits constrain those parameters, they do not pin down the ternary difference. The SI shows free-energy curves for βH = 1.0/1.1/1.2 and βC = 0.92/0.95/0.98, but the authors don't report the resulting ternary ordering for those neighboring values. That leaves an unquantified uncertainty on the headline number. The mean FPT values also have no error bars. The abstract says 'prove', which is too strong for a structure-based model; 'support' or 'are consistent with' would be accurate. No code or input files are provided, which limits reproducibility, though the method section is specific enough to rebuild.\n\nNone of this sinks the central argument: the ternary ordering is not wired in by the binary calibration, and it matches the experimental competition data. What is missing is demonstration that the ordering is stable to reasonable parameter variation. That is an addressable revision, not a fatal flaw.\n\nThis paper deserves a serious referee. It is the kind of study a molecular biophysicist in the IDP/TAZ1 niche would want to see, and the mechanistic predictions (IS state, binding order differences) could guide future mutagenesis experiments. I would not cite it in my own next paper, but I might bring it to a reading group.","headline":"A competent ternary SBM study that likely gets the qualitative competition right, but the central free-energy gap rests on a single calibrated point with no sensitivity analysis.","tokens_in":19450,"tokens_out":2525,"would_cite":false,"duration_ms":21006,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A coarse-grained molecular dynamics study of the ternary TAZ1–HIF-1α–CITED2 system claims that CITED2 outcompetes HIF-1α for TAZ1 in both thermodynamics and kinetics, with the CITED2-bound basin lowest by 1.91 kT and a five-fold faster…","keywords":["TAZ1","HIF-1α","CITED2","intrinsically disordered proteins","structure-based model","molecular dynamics","binding competition","free energy surface"],"falsifier":"A competition experiment at 1:1:1 TAZ1:HIF-1α:CITED2 with single-molecule or stopped-flow kinetics could directly measure whether CITED2 occupies TAZ1 first and more often than HIF-1α; the paper predicts CITED2 should win the first-binding event in the large majority of runs, and that mutating CITED2 residues Phe222/Ile223 should slow CITED2 binding more than mutating HIF-1α residues Gly791/Leu792/Gln824 slows HIF-1α binding.","tokens_in":18427,"feed_emoji":"🧬","tokens_out":5027,"duration_ms":50357,"temperature":0.7,"pith_summary":"This paper tries to establish that in the three-molecule system of TAZ1 with HIF-1α and CITED2, CITED2 wins the competition not just thermodynamically but also kinetically. Using coarse-grained structure-based molecular dynamics, it finds the TAZ1–CITED2 bound state is the global free-energy basin, about 1.91 kT below the TAZ1–HIF-1α bound state, and that CITED2 binds TAZ1 roughly five times faster in direct binding. The simulations also map an intermediate state through which one ligand displaces the other, giving a concrete mechanism for why CITED2 can efficiently replace HIF-1α. If correct, this provides a mechanistic explanation for the experimentally observed dominance of the TAZ1–CITED2 complex and identifies the residues that matter most at the early step of binding.","feed_headline":"CITED2 dominates TAZ1 binding in thermodynamics and kinetics","feed_subtitle":"In a 1:1:1 simulation, CITED2's bound state sits lowest and forms five times faster, matching experiments.","key_machinery":"The central object is a weighted structure-based coarse-grained model: a Cα model with native contacts weighted by their frequency across the 20 NMR conformers of each complex, and with Debye–Hückel electrostatics added. The binary complexes are parameterized so that each reproduces the experimental dissociation constant of about 10 nM, and the same interaction strengths are then used in the ternary simulation. The argument is carried by the ternary free-energy surface projected on the fractions of native inter-molecular contacts, Qinter(TAZ1–HIF-1α) and Qinter(TAZ1–CITED2), and by mean first-passage times computed from kinetic simulations. The key mechanistic state is the intermediate IS, where one ligand is partly bound while the conserved LPQL or LPEL motif of the other ligand occupies the shared binding site; this state links the direct-binding and replacement pathways.","core_discovery":"The central claim is that TAZ1–CITED2 is the dominant complex in a ternary TAZ1:HIF-1α:CITED2 system, both at equilibrium and in kinetics. On the two-dimensional free-energy surface projected onto the binding reaction coordinates of the two ligands, the CITED2-bound basin is lowest at 0.00 kT, the HIF-1α-bound basin is 1.91 kT higher, and an intermediate state where both ligands partially engage TAZ1 is 3.07 kT higher. Direct binding of CITED2 has a mean first-passage time of 0.286 ns versus 1.431 ns for HIF-1α, and in 200 unbiased binding runs the system reaches the CITED2-bound state first in 177 runs. Replacement of HIF-1α by CITED2 is also faster than the reverse (60.8 ns versus 163.2 ns mean first-passage time), mostly through the intermediate state. The paper concludes that CITED2 dominates both thermodynamically and kinetically, in line with the experimental findings.","pith_inferences":["A testable extension the paper does not make: if CITED2's kinetic advantage comes from its simpler single-helix fold, truncating the N-terminal L1 region of CITED2 should slow direct binding more than truncating the C-terminus of HIF-1α slows its binding; the paper's transition-state contact maps predict this asymmetry.","The ternary ordering depends on the calibrated interaction strengths βH=1.1 and βC=0.95; re-running the ternary simulation with those two values swapped would isolate whether CITED2 dominance is a structural feature of the complexes or an artifact of the calibration.","A potential practical consequence is that a molecule mimicking the N-terminal region of CITED2 could act as a competitive inhibitor of HIF-1α–TAZ1 binding, since the direct-binding pathway of CITED2 begins at that region.","The paper's emphasis on binding order suggests that in vivo, the timing of expression of CITED2 versus HIF-1α could determine which complex forms, even if both ligands are present at similar concentrations; this is an implicit biological consequence the paper does not spell out."],"forward_implications":["If CITED2 is truly kinetically favored, then at equal concentrations CITED2 will occupy TAZ1 before HIF-1α has time to establish its complex, making the competition a race rather than a simple equilibrium selection.","The free-energy gap of 1.91 kT in the ternary system means that even though the two binary affinities are nearly equal, the presence of both ligands amplifies the small difference and shifts the population strongly toward CITED2.","The intermediate state IS, with both ligands simultaneously touching TAZ1, gives a concrete structural target for experiments: mutations that perturb the N-terminus of CITED2 or the C-terminus of HIF-1α should alter the displacement rates in predictable ways.","The contact-map and φ-value analysis predicts different binding orders for direct versus replacement pathways, so experiments that probe specific residues can distinguish whether a mutation affects initial binding or the displacement step.","The simulated φ values agree with most experimental φ values for HIF-1α, supporting the view that the native hydrophobic binding interactions form after the transition state during TAZ1–HIF-1α binding."],"supporting_citations":[{"why":"Supplies the experimental Kd values, the apparent Kd values for displacement, and the NMR observation that TAZ1-CITED2 dominates the ternary mixture; the central claim is calibrated to and compared against these data.","marker":"[15]"},{"why":"Provides the NMR structure 1L8C of the TAZ1-HIF-1α complex, used as the starting model for the binary and ternary simulations.","marker":"[10]"},{"why":"Provides the NMR structure 1R8U of the TAZ1-CITED2 complex, also used as the starting model for the simulations.","marker":"[13]"},{"why":"Supplies the structure-based model potential form and the method for building native-contact models, which is the foundation of the simulation approach.","marker":"[26]"},{"why":"Provides the experimental φ values for HIF-1α that the simulated φ values are compared with at the transition state.","marker":"[25]"},{"why":"Establishes the weighted-contact-map protocol and the simulation framework that the binary and ternary models are built on.","marker":"[16]"}],"fun_headline_variants":["CITED2 beats HIF-1α to TAZ1: 5× faster and 1.9 kT lower","TAZ1 binding: CITED2 dominates, 1.9 kT lower and 5× faster","Simulations: CITED2 binds TAZ1 fastest and most stably","CITED2 wins TAZ1: energy basin lowest, binding speed 5× faster"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central result assumes that the two tuning knobs that set how strongly TAZ1 attracts HIF-1α versus CITED2, calibrated so each complex alone matches the measured 10 nM affinity, also hold in the three-molecule system; if the true ternary strengths differ, the predicted CITED2 dominance could vanish.","fun_headline_variants_meta":{"raw":{"variants":["CITED2 beats HIF-1α to TAZ1: 5× faster and 1.9 kT lower","TAZ1 binding: CITED2 dominates, 1.9 kT lower and 5× faster","Simulations: CITED2 binds TAZ1 fastest and most stably","CITED2 wins TAZ1: energy basin lowest, binding speed 5× faster"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000715,"raw_usage":{"total_tokens":3228,"prompt_tokens":973,"completion_tokens":2255,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":589,"completion_tokens_details":{"reasoning_tokens":2152}},"tokens_in":589,"tokens_out":2255,"duration_ms":19570,"temperature":1.0,"reasoning_tokens":2152,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:32:32.404946+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A competition experiment at 1:1:1 TAZ1:HIF-1α:CITED2 with single-molecule or stopped-flow kinetics could directly measure whether CITED2 occupies TAZ1 first and more often than HIF-1α; the paper predicts CITED2 should win the first-binding event in the large majority of runs, and that mutating CITED2 residues Phe222/Ile223 should slow CITED2 binding more than mutating HIF-1α residues Gly791/Leu792/Gln824 slows HIF-1α binding.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the experimental Kd values, the apparent Kd values for displacement, and the NMR observation that TAZ1-CITED2 dominates the ternary mixture; the central claim is calibrated to and compared against these data."},{"cited_title":"Azia and Y","cited_arxiv_id":null,"evidence_quote":"Provides the NMR structure 1L8C of the TAZ1-HIF-1α complex, used as the starting model for the binary and ternary simulations."},{"cited_title":"The secondary structures as well as the LPQL/LPEL motif are labeled","cited_arxiv_id":null,"evidence_quote":"Provides the NMR structure 1R8U of the TAZ1-CITED2 complex, also used as the starting model for the simulations."},{"cited_title":"Lindstr ¨om, E","cited_arxiv_id":null,"evidence_quote":"Supplies the structure-based model potential form and the method for building native-contact models, which is the foundation of the simulation approach."},{"cited_title":"Kumar, J","cited_arxiv_id":null,"evidence_quote":"Provides the experimental φ values for HIF-1α that the simulated φ values are compared with at the transition state."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the weighted-contact-map protocol and the simulation framework that the binary and ternary models are built on."}],"review_version":1}