{"id":"2a8fbd4b-bcfa-4f65-9e5b-210f4db9011b","arxiv_id":"2512.08356","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Enzyme-catalyzed molecular interconversion plus a shared enzyme reservoir is enough to produce phase-separated membrane domains with explicit kinetic-rate-controlled coexistence, interface, and nucleation properties.","lead":"This paper builds a minimal mathematical model of how enzymes can drive membrane molecules to separate into stable spatial domains. It derives formulas for when such domains coexist, how sharp their boundaries are, and what size a patch must reach to grow, all depending on enzyme speeds rather than equilibrium forces.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Critical-radius formula Eq. (34) is inconsistent with its own derivation Eq. (E13), so the nucleation claim is not quantitatively supported.","rationale":"The paper has real strengths: it starts from a clear microscopic model, derives a mesoscopic stochastic equation, provides a self-contained mean-field analysis, and supports it with Gillespie simulations and an open-source implementation. The qualitative phase diagram and interface-width scaling are plausible and independently checked to some degree. However, the nucleation radius is a headline closed-form prediction, and the inconsistency between Eq. (34) and Eq. (E13) is not a minor typo: the two formulas differ by a factor of ~2.8 in the saturation limit and have different κ-dependence. Because Fig. 8(b) fits Eq. (34) with a free prefactor, the quantitative agreement claimed for nucleation is not evidence for the derived expression. This is load-bearing for the central claim that the theory yields explicit, quantitatively confirmed predictions for fluctuation-driven transitions. The reader's weakest_assumption pointed to the quasi-steady-state enzyme slaving (Eq. (4)); that is a legitimate concern about applicability to real biological timescales, but it is not internally inconsistent and would affect the physical regime rather than the mathematical derivation. The Eq. (34)/Eq. (E13) mismatch is a direct internal contradiction that can be settled by calculation, so it is the more decisive concern. If the inconsistency is repaired and the fits are redone without free prefactors, the manuscript could be acceptable; in its current form, rejection is warranted.","tokens_in":20983,"tokens_out":21288,"duration_ms":176102,"concrete_test":"Recompute R_c for the Fig. 8 parameter sets: evaluate Eq. (34) and Eq. (E13) using k_e^± from Eq. (11) and solving the implicit ⟨ϕ⟩=2πR^2/s^2−1. If the two closed forms differ by more than ~50%, the main-text formula is not the one derived. Then re-derive R_c numerically from the exact A and B in Eqs. (8)–(10) using Eqs. (E9)–(E11), circumventing the polynomial approximation, and compare all three values against the Gillespie-measured critical radius in Fig. 8(a). This distinguishes a typo in Eq. (34) from a deeper flaw in the nucleation calculation.","verdict_should_be":"REJECT","load_bearing_attack":"The central nucleation prediction is undermined by an internal inconsistency. The closed-form critical radius announced in Eq. (34) and the expression obtained from the large-deviation derivation in App. E, Eq. (E13), are not the same function. Already in the saturation limit Km/C→0, Eq. (34) (with f(0)=1) gives a prefactor 3/20 ≈ 0.15, while Eq. (E13) gives 5/12 ≈ 0.417 for the same combination √(D(k_e^+ + k_e^-))/|k_e^+ − k_e^-| — a factor of ~2.8 discrepancy. More generally, the κ-dependence in Eq. (34) via f(κ) differs from the factor (1 + 2κ + (2/3)/(1+2κ))√g(κ) in Eq. (E13); the two expressions are not equivalent. Since Fig. 8(b) validates Eq. (34) using a fit with a free prefactor, that prefactor can absorb the discrepancy; the simulation data therefore do not confirm the derived nucleation radius. As written, the manuscript does not support its headline claim of a quantitatively confirmed critical radius, even though the underlying large-deviation framework may be sound. A separate sign error in App. D Eq. (D4) — φ0 has the opposite sign to the main-text assertion for k_e^+ > k_e^- — further indicates that the polynomial approximation used in the derivation needs careful re-examination.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript develops a minimal stochastic model of enzyme-driven interconversion of two membrane-bound molecular states, with enzymes shuttling from a reservoir and product-feedback recruitment. Under assumptions of fast enzyme equilibration and a well-mixed reservoir, the six-field reaction–diffusion system is coarse-grained into a single non-conserved order parameter governed by an active Model A equation with a global constraint. Mean-field analysis yields an explicit phase diagram, coexistence conditions, interfacial tension (Eq. 28), interface width (Eq. 30), and power consumption (Eq. 31). Large-deviation theory is used to derive a critical nucleation radius (Eq. 34). Stochastic lattice-Gillespie simulations are used to test the predictions. The paper argues that the resulting phenomenology is consistent with experiments on phosphoinositide and Rab5 membrane systems.","tokens_in":21390,"tokens_out":11789,"duration_ms":104408,"significance":"If the derivations are correct, this is a valuable theoretical contribution: it offers an analytically tractable microscopic-to-mesoscopic route to active phase separation in a biologically relevant enzyme–substrate module, with closed-form expressions for several observables. The explicit stochastic simulation implementation and detailed appendices are strengths. However, the quantitative validation is weakened by the use of free-prefactor fits, and the critical-radius formula in the main text is inconsistent with the derivation in App. E. The conceptual framework—a global enzymatic reservoir providing the stabilizing constraint—is plausible and connects to the mass-conserved reaction–diffusion literature. The central idea is worth publishing after a thorough revision.","major_comments":[{"comment":"The closed-form critical radius in the main text is not the one derived in the appendix. In the saturation limit K_m/C→0, Eq. (34) gives R_c = (3/20)√(D(k_e^+ + k_e^-))/|k_e^+ - k_e^-|, while Eq. (E13) gives (5/12) of the same combination. The κ-dependent factors also differ: f(κ) in Eq. (35) is not proportional to (1+2κ+(2/3)/(1+2κ))√g(κ) in Eq. (E13). Because Fig. 8(b) fits Eq. (34) with a free prefactor, the simulation cannot resolve this factor-of-~2.8 discrepancy; the quantitative confirmation of R_c is therefore not established. The authors should identify which expression is the actual prediction and test it without a floating prefactor.","section":"Section III, Eq. (34) and App. E, Eq. (E13)"},{"comment":"All three quantitative validations of the closed-form expressions use a free overall prefactor in the fit (explicitly stated for Fig. 6 and Fig. 8(b); implicitly for the B(φ) fit in Fig. 7). This tests only the shape and scaling of the predictions, not the advertised closed-form prefactors. The abstract and conclusions assert that 'analytical results are quantitatively confirmed'; this overstates the evidence. I recommend refitting with all parameters fixed to the simulation parameters, or at minimum reporting the fitted prefactor values and comparing them with unity/the derived values, and softening the quantitative claim.","section":"Figs. 6, 7, 8(b) and Eqs. (9), (30), (34)"},{"comment":"Eq. (D4) gives φ0 = (1+2K_m/C)(k_e^+ - k_e^-)/(k_e^+ + k_e^-)c, which has the opposite sign from the main-text result (Section II: φ0 ∝ k_e^- - k_e^+). With k_e^+ > k_e^-, Eq. (D4) predicts a barrier at positive φ, i.e. a stable minus phase, contradicting the text. Since the polynomial approximation is used in the derivation of Eqs. (28) and (E13), this sign error must be corrected and the derivation re-checked.","section":"App. D, Eq. (D4)"}],"minor_comments":[{"comment":"The timescale separation leading to Eq. (4) is not quantified. An estimate for the cited kinase–phosphatase and Rab5 systems would help connect the theoretical regime to the experiments.","section":"Section I"},{"comment":"The phrase 'centered at ρ=ρ0' is misleading; the physical region is an interval with endpoints ρ0/ρ_+ and ρ0/ρ_-.","section":"Fig. 2 caption"},{"comment":"The notation ξ_R for the reaction noise is not defined in the main text; please define it consistently with Eq. (7).","section":"App. B, Eq. (B8)"}],"recommendation":"major_revision","confidential_remarks":"This is a serious theoretical paper with a well-structured derivation and a reproducible simulation code. The main concern is the internal inconsistency in the nucleation section: Eq. (34) and Eq. (E13) disagree, and the simulation validation uses a free prefactor. These issues are fixable but require careful reworking of the central quantitative claim. I recommend major revision rather than rejection, provided the authors can reconcile the critical-radius formula and either confirm the prefactor without fitting or explicitly restrict their claims to scaling behavior."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a worthwhile paper with a real blemish: the main-text critical radius formula is not the same function as the one derived in the appendix. Eq. (34) gives R_c = (3/20) f(Km/C) sqrt(D(k_e^+ + k_e^-))/|k_e^+ - k_e^-|, and App. E Eq. (E13) gives a different prefactor (5/12 in the saturation limit) and a different dependence on Km/C. Since Fig. 8(b) validates Eq. (34) using a fit with a free prefactor, that validation cannot confirm the derived expression. There is also a sign error in App. D: Eq. (D4) has phi0 with opposite sign to the main-text definition, indicating the polynomial approximation needs another pass.\n\nThat said, the paper has a solid core. The derivation from microscopic reactions (1)-(3) to a mesoscopic active Model A with a global constraint is careful, and the mean-field phase diagram, interfacial tension, and interface width are internally consistent and supported by Gillespie simulations. The expressions in terms of kinetic rates are genuinely new and potentially useful for experimentalists studying phosphoinositide or Rab5 domains. The power-consumption scaling is a nice touch.\n\nThe main soft spot, aside from the nucleation inconsistency, is the quasi-steady-state slaving of enzyme binding, Eq. (4). The paper never gives an experimental estimate for how fast enzyme on/off rates are relative to phase ordering. If that separation fails, the global feedback loop and every closed-form result lose their justification. This is a standard modeling assumption, but it should be argued quantitatively, at least for the cited systems.\n\nOverall, the paper is for people working on non-equilibrium phase separation on membranes. The mean-field part is a genuine contribution and the nucleation calculation is a promising framework, but the current manuscript does not support its headline claim of a quantitatively confirmed critical radius. I would send it to peer review—the framework deserves referee time—but the authors need to reconcile Eqs. (34) and (E13), fix the sign error, and refit the nucleation data without an arbitrary prefactor. If those repairs hold, it could be a good paper.","headline":"Worthwhile mean-field theory of enzyme-driven phase separation, undercut by an internal inconsistency between the main-text and appendix critical-radius formulas.","tokens_in":21826,"tokens_out":3688,"would_cite":false,"duration_ms":31612,"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":"Stable membrane signaling domains can arise purely from enzymatic cycling, with phase coexistence, interface sharpness, and nucleation sizes fixed by catalytic rates.","keywords":["active phase separation","enzyme kinetics","membrane domains","Model A with global constraint","interfacial tension","nucleation","Michaelis-Menten kinetics","signaling molecules"],"falsifier":"Measure the time it takes enzymes to bind to and leave the membrane in a reconstituted system and show it is not much shorter than the time needed for domains to coarsen; that would break the quasi-steady-state reduction. Alternatively, a FRAP experiment showing domain material recovery times comparable to dense passive condensates would falsify the rapid gas-like exchange prediction.","tokens_in":20896,"feed_emoji":"🧬","tokens_out":10119,"duration_ms":87864,"temperature":0.7,"pith_summary":"The paper tries to establish that the formation of stable signaling domains on membranes does not require equilibrium attractions between molecules: the repeated, energy-consuming interconversion of two molecular states by antagonistic enzymes is enough. Starting from the microscopic reactions of substrate conversion, product-recruiting enzyme binding, and exchange with a cytosolic reservoir, it derives a single stochastic equation for the local state difference, an active Model A with a global constraint. From that equation it obtains explicit mean-field conditions for phase coexistence and closed-form expressions for interfacial tension, domain fractions, maintenance power, and critical nucleation radius, all in terms of kinetic rates. If this is right, the same phenomena we call phase separation can be controlled by biochemical parameters that experiments can tune — catalytic rates, enzyme asymmetry, scaffold affinity — rather than by saturation concentrations. The authors support the derivation with lattice-gas simulations and point to agreement with reconstituted kinase-phosphatase and Rab5 systems.","feed_headline":"Enzyme cycles alone can stabilize membrane domains","feed_subtitle":"Microscopic reaction kinetics yields explicit predictions for coexistence, interface width, and nucleation size from catalytic rates.","key_machinery":"The object doing the work is a reduced Langevin equation for the non-conserved order parameter phi, in the class of Model A with a global constraint: phi evolves by diffusion, a drift from enzymatic interconversion, and multiplicative intrinsic noise, while the effective catalytic rates k_e^pm are functionals of the spatial average <phi>. This global coupling originates from the quasi-steady-state enzyme-slaving relation K_d^pm E^pm = phi^pm E_f^pm together with conservation of the shared enzyme pool, which makes the finite enzyme reservoir feel the whole membrane state. Polynomial approximations of the effective potential and noise amplitude (Appendix D) turn the theory into explicit formul","core_discovery":"The paper's central claim: a two-state membrane molecule, switched by antagonistic enzymes that are recruited by their own product and exchange rapidly with a reservoir, obeys a single stochastic reaction-diffusion equation for the difference field phi = phi+ - phi- with multiplicative noise and a global constraint. The constraint — the effective catalytic rates depend on the spatial average <phi> — permits stable coexistence. Mean-field analysis gives explicit coexistence conditions in kinetic-rate space and closed-form expressions for interfacial tension, interface width, maintenance power, and critical nucleation radius; lattice simulations of the full particle process match the formulas.","pith_inferences":["Editorial extension: any pair of antagonistic, product-recruiting enzymes on a two-state membrane substrate should show the same phase diagram, so testing other GTPase systems (for instance Ras or Arf) would reveal whether the predictions extend beyond the phosphoinositide and Rab5 examples the paper cites.","Editorial extension: measuring steady-state domain area fraction versus enzyme concentration in a reconstituted system and comparing with Eq. (23) would extract K_d/C directly; Eq. (34) would then predict the nucleation radius without additional fitting parameters.","Editorial extension: the gas-like rapid-exchange picture implies that FRAP recovery in these active domains should be fast and largely independent of domain size, unlike dense passive condensates where recovery is limited by diffusion through the condensed phase.","Editorial extension: the multi-species extension in the appendix suggests enzymatic kinetics can program spatial ordering of interfaces — disabling one reaction makes a third species wet the interface between the other two — pointing to a biochemical code for multiphase membrane organization."],"forward_implications":["Phase coexistence is selected by nonequilibrium kinetics: it occurs only while the catalytic ratio rho_0 lies between rho_- and rho_+, so changing a single catalytic rate or enzyme number can switch a membrane between homogeneous and domain states.","Interfaces are sharper when enzymes operate near saturation (small K_m) and when catalytic turnover is fast; the width is set by 1/sqrt(k_e^+ + k_e^-) and by the Michaelis constant through g(K_m/C).","Sustaining an interface costs energy: the power per unit length is proportional to the interface width and scales as the square root of the catalytic rate, meaning the domains continuously dissipate ATP or GTP at their boundaries.","Uniform states can be metastable: escape proceeds by nucleating a critical droplet of the favored phase, with R_c decreasing for saturation, larger kinetic asymmetry, and stronger scaffold affinity; without basal catalysis the uniform states become absorbing.","Material exchange across domain boundaries is rapid because particles behave as a gas: stability comes from interconversion, not from reduced mobility, a testable difference from passive condensates."],"fun_headline_variants":["Enzyme-driven switching alone can stabilize membrane domains","Active Model A with global constraint gives explicit phase diagram","Catalytic rates replace saturation curves in phase coexistence","From microscopic kinetics to membrane phase separation in one equation","Testable predictions for active domains from enzyme kinetics"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole reduction stands on the assumption that enzyme binding and unbinding are much faster than phase ordering, so the membrane-bound enzyme distribution is slaved to the substrate distribution; if that separation of timescales fails in a real system, the effective rates, the global feedback, and every closed-form formula lose their justification.","fun_headline_variants_meta":{"raw":{"variants":["Enzyme-driven switching alone can stabilize membrane domains","Active Model A with global constraint gives explicit phase diagram","Catalytic rates replace saturation curves in phase coexistence","From microscopic kinetics to membrane phase separation in one equation","Testable predictions for active domains from enzyme kinetics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000267,"raw_usage":{"total_tokens":1464,"prompt_tokens":767,"completion_tokens":697,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":511,"completion_tokens_details":{"reasoning_tokens":624}},"tokens_in":511,"tokens_out":697,"duration_ms":7208,"temperature":1.0,"reasoning_tokens":624,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T17:42:35.945813+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the time it takes enzymes to bind to and leave the membrane in a reconstituted system and show it is not much shorter than the time needed for domains to coarsen; that would break the quasi-steady-state reduction. Alternatively, a FRAP experiment showing domain material recovery times comparable to dense passive condensates would falsify the rapid gas-like exchange prediction.","supporting_citations":[],"review_version":1}