{"id":"df735dbb-1a02-4559-a639-e5a4494043ee","arxiv_id":"2508.01046","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Epithelial fingers emerge from correlated active cell motion interacting with a contractile edge, not from leader-cell or signalling instabilities.","lead":"Finger-like protrusions in healing sheets of epithelial cells can form purely from the collective, correlated movement of cells, without needing specially programmed leader cells or chemical signals. The work combines lab movies, simulations, and a physical theory to show that this biological pattern is an expected consequence of dense active matter physics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Height-equation and polymer predictions rely on an un-derived 25x reduction of the AVM boundary stiffness (SI II.D), making finger-scale predictions partly fitted.","rationale":"The reader's weakest assumption is exactly the load-bearing weakness I identify: the ad hoc 25x reduction of line tension and bending stiffness in the mapping from AVM boundary springs to the continuum height equation and polymer model. This is more than a cosmetic parameter: it is the difference between a derived theory and a fitted one. However, the concern does not overturn the central sufficiency claim, because the AVM itself—with only a single fitted boundary stiffness ks—produces the finger-like morphology and matches the experimental tangent-tangent, roughness, and spectral statistics. The 25x factor only affects the continuum theory's quantitative predictions and its claim to have independently reproduced the AVM and experimental results. The paper is transparent about the factor, so this is a correctness risk rather than a hidden flaw. Conditional acceptance is the right verdict, with the proposed relaxation-rate measurement as a decisive check: it would determine whether the 25x reduction is a real emergent property of the dynamic boundary or an unprincipled fit. I therefore keep the reader's CONDITIONAL verdict unchanged.","tokens_in":34185,"tokens_out":7770,"duration_ms":105635,"concrete_test":"In the non-expanding AVM simulations, measure the deterministic relaxation rate Γ(q) of boundary height modes from the long-time autocorrelation of h(q,t) at several q values (e.g., q = 2πn/Lx, n = 1..50), in the absence of active driving (v0 = 0), by initializing a small sinusoidal perturbation and tracking its decay. Fit to the height-equation prediction Γ(q) = (λ_eff q^2 + κ_eff q^4)/(ζ + η q^2). Compare λ_eff and κ_eff with (i) the naive mapping λ = ks/(ζ a^2), κ = kb/(ζ a^4) and (ii) the 25x-reduced values used in SI II.D. If the measured Γ(q) matches the naive mapping, the 25x reduction is an arbitrary fitting parameter and the theory's quantitative match is not independently validated; if it matches the reduced values, the reduction reflects a genuine viscoelastic softening from dynamic boundary point insertion and the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that correlated active cell motion alone suffices to produce fingers is established by the AVM itself, but the paper's quantitative theory—the height equation (Eq. 8) and the 2D polymer model (Eq. 7)—matches experiments only after an unexplained rescaling. SI Appendix II.D states: 'we found that the estimated line tension and bending stiffness on the boundary of the AVM should be taken roughly a factor of 25 smaller, to match with theory', and the main text admits 'we had to fit a very small value of spring constant ks'. This means the continuum theory is not actually derived from the AVM boundary springs; λ and κ are effectively free parameters, absorbed into a factor that is justified only by a hand-waved 'relaxation effects when new points are added'. Since finger width, roughness, and lifetime in the height equation scale with λ and κ, the agreement of Figs. 4c,f,i could be a consequence of tuning this 25x knob rather than a genuine validation. The AVM's own ks is also a fitted parameter, but it is a single calibration to the shape of the tangent-tangent correlation, after which other statistics are predicted; the additional 25x factor breaks the claimed dimensional mapping between AVM and theory. If the real actomyosin cable were 25x stiffer than the value used in the theory, the observed correlated interior motion might be insufficient to generate the observed finger amplitudes, and the sufficiency result would be an artifact of the soft, passive edge used in the model.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper combines MDCK wound-healing experiments, self-propelled Voronoi/active vertex model (AVM) simulations, and a continuum viscoelastic theory with a boundary height equation and active polymer model to argue that fingers at the edge of expanding epithelial monolayers arise as long-lived active fluctuations driven by correlated interior cell motion, without requiring a finite-wavelength instability or leader-cell feedback. The interior velocity statistics are matched by the AVM and by an active viscoelastic continuum model, while the boundary is modelled as a semiflexible polymer driven by interior noise; the height equation reproduces tangent-tangent, roughness, and temporal correlation functions. The paper concludes that leader cells, signalling, and proliferation modulate but do not trigger finger formation.","tokens_in":34482,"tokens_out":4420,"duration_ms":62622,"significance":"If the mechanism is correct, the paper is significant: it provides a quantitative null model for epithelial fingering, connects finger length and time scales to bulk active-matter correlation lengths, and makes falsifiable predictions about how perturbations to adhesion, friction, crawling speed, and cable properties affect fingers. Strengths include the direct experimental measurement of velocity and boundary correlations, the explicit AVM with a dynamic boundary, the detailed analytical derivations in the SI, and the clear statement of the model as a fluctuation-driven null model. The claim that fingers do not require an instability is supported by the absence of peaks in the Fourier spectra and by the AVM's spontaneous finger formation without division or leader cells. However, the quantitative predictive power of the continuum theory is weakened by an unexplained rescaling of the boundary stiffness, and the dominant-length-scale prediction is partly built into the model construction.","major_comments":[{"comment":"The continuum theory and the 2D polymer model are not actually obtained from the AVM boundary parameters by dimensional mapping: SI Appendix II.D states that the estimated line tension and bending stiffness on the boundary of the AVM should be taken roughly a factor of 25 smaller to match theory, and the main text admits that a very small value of the spring constant ks had to be fitted. Because finger width, roughness, and lifetime in the height equation depend directly on lambda and kappa, the quantitative agreement shown in Fig. 4c, f, and i is partly a consequence of this tuning rather than an independent validation. As written, the abstract's claim that the model 'quantitatively predicts' the edge statistics is overstated; the authors should either derive the reduction factor from the boundary point addition/removal relaxation process or explicitly treat lambda and kappa as fitted parameters and state how many free parameters the comparison uses.","section":"SI Appendix II.D and main text Eq. (8)"},{"comment":"The noise-generating chain modulus for the 2D polymer is set to mu_h/zeta_h = 600 um^2/h 'by hand from fitting the real space spatial correlation of the 2d polymer noise generating chain velocity to the real space velocity correlations of the AVM'. This is an additional free parameter that is not listed in Table I and is calibrated against the same AVM velocity field that was already matched to experiments. This further weakens the claim that the polymer and height-equation simulations are parameter-free descendants of the AVM; mu_h should be listed with its uncertainty, and the sensitivity of the boundary correlation functions to its value should be reported.","section":"SI Appendix II.D"},{"comment":"The statement that finger length scales are dominated by the interior shear correlation length q1 = sqrt(zeta/(mu*tau + eta)) is close to a restatement of the model construction: the boundary noise vf in Eq. (S41) is derived from the same viscoelastic interior field with the same mu, eta, zeta, and tau, so the edge inheriting that scale is a consistency check rather than an independent prediction. A stronger, falsifiable test would be a perturbation experiment or simulation, for example varying KP, zeta_pair, or v0 and showing that the finger length and lifetime scales track the predicted q1; the Discussion proposes such dependencies but does not report them. The authors should either perform such a perturbation test or clearly label the dominant-scale result as a self-consistency statement.","section":"Main text, Eq. (S86) and Discussion"}],"minor_comments":[{"comment":"The fitted interval for ks/zeta is reported as 78 +/- 50 h^-1, which is a very large range; the authors should show the sensitivity of the tangent-tangent correlation to ks within this range, for example as a supplemental figure, so that the constraint is visually verifiable.","section":"Table I"},{"comment":"The phrase 'again computed using dimensional scaling only' is misleading because the AVM parameters themselves were fitted to the experimental velocity statistics; the continuum theory is parameterized by mapping those fitted AVM parameters, so it is not an independent calculation.","section":"Main text, Fig. 2 caption and text near Eq. (3)"},{"comment":"The text states that bond lengths can be approximated by a constant a(t) about a(0) = 5 um, but Fig. S10 shows a(t) growing from about 5 um to about 10 um over 15 h; the approximation should be justified more carefully or its effect on the mapping estimated.","section":"SI Appendix II.D, Fig. S10"}],"recommendation":"major_revision","confidential_remarks":"The central sufficiency claim is convincing and the AVM is a strong piece of evidence, but the quantitative theory's match to experiments currently relies on an unexplained 25x stiffness reduction and on an additional hand-fitted noise-chain modulus. I would not reject the paper, but the authors should either derive these factors or substantially moderate the 'quantitatively predicts' language and treat the continuum theory as a consistency check with fitted parameters."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The useful core: they build a minimal null model — self-propelled Voronoi with ABP crawling and a semiflexible contractile edge — and show it produces fingers with the same qualitative and quantitative statistics as MDCK monolayers: interior velocity correlations, tangent-tangent correlations, roughness, lifetime, and no finite-wavelength instability. That is a real result. It shifts the debate: fingers do not need leader cells or biochemical feedback to emerge; those may amplify or modulate a purely physical fluctuation-driven process. The SI derivations are thorough, and the three-way comparison (experiment, AVM, continuum theory) is carefully done.\n\nThe soft spots are in the continuum theory. The height equation and 2D polymer model require an ad hoc 25x reduction of the boundary line tension and bending stiffness relative to the dimensional mapping from the AVM (SI II.D). That's a fitted knob, not a derived parameter, and it directly enters finger width, roughness, and lifetime. The stress-test note is on target: the quantitative agreement of Figs. 4c,f,i is partly a result of tuning this factor. It doesn't kill the central sufficiency claim, because the AVM itself — where ks is also fitted but to the tangent-tangent shape — reproduces the velocity correlations and finger statistics in one model. But it does mean the continuum theory is not a fully first-principles derivation; it is a motivated fitting exercise with a clean analytical form.\n\nSecond soft spot: the abstract says leader cells, signalling, and proliferation 'modulate, but do not trigger' fingers. A null model only shows they aren't necessary for the basic pattern. It doesn't show they never trigger it. The discussion is more careful, but the abstract overstates.\n\nThird, several interior parameters (tau, v0, zeta_pair, KP) are calibrated to the same experimental correlation functions the paper then says it matches. That's standard practice, but it weakens the claim of predictive power. The paper would be stronger if it predicted a holdout statistic or a perturbation (e.g., changing substrate stiffness or adhesion) rather than matching the same data used for fitting.\n\nWho's this for? Groups working on tissue mechanics, active matter, and wound healing models. It deserves serious peer review — the core sufficiency result is important and the data/theory package is rich. I'd accept with major revision: address the 25x factor or at least test its robustness, soften the abstract, and consider showing at least one genuinely predictive comparison. I'd engage with it if it came to me.","headline":"Strong sufficiency result: fingers emerge from correlated cell noise plus a soft contractile edge, but the continuum theory's 25x stiffness rescaling and the abstract's causal claims need fixing.","tokens_in":35059,"tokens_out":2579,"would_cite":true,"duration_ms":31708,"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":"Correlated active cell motion alone, without leader cells or signalling, is sufficient to produce the finger-like protrusions at the front of spreading epithelial sheets.","keywords":["epithelial monolayers","finger formation","collective cell migration","active matter","correlated velocity fluctuations","active vertex model","actomyosin cable","wound healing"],"falsifier":"Measure whether the advancing edge is actually passive: if inhibiting actomyosin contractility at the leading edge abolishes fingers while the interior velocity correlations (length scale of about 100 micrometres, persistence of about 1 hour) remain unchanged, the passive-cable picture fails; if fingers persist, the claim that interior correlations alone drive them survives.","tokens_in":33973,"feed_emoji":"🧫","tokens_out":8578,"duration_ms":103696,"temperature":0.7,"pith_summary":"This paper tries to establish a minimal physical explanation for finger-like protrusions at the advancing edge of healing epithelial sheets. It claims that uncorrelated crawling of individual cells, once filtered through elastic and viscous interactions between cells, produces a correlated velocity field in the monolayer interior, and that this field alone drives the contractile actomyosin cable at the edge into long-lived fingers. The authors support the claim by showing that an agent-based model and a continuum theory, with parameters calibrated to measured interior velocity correlations, quantitatively reproduce the tangent-tangent, roughness, and Fourier-spectrum statistics of experimental boundaries. If correct, the result would move finger formation out of the category of specialised biological control and into the generic physics of dense active matter, with leader cells and signalling as modulators rather than triggers.","feed_headline":"Finger formation needs no leader cells; interior motion suffices","feed_subtitle":"Fingers at a healing edge can be long-lived fluctuations, driven by interior cell swirls rather than leader cells.","key_machinery":"The load-bearing object is the advancing edge as a stretched, contractile semiflexible polymer, an effective worm-like chain driven by a correlated active noise field that represents interior cell motion. In the continuum description this becomes a linear height equation for the boundary height $h(x,t)$, whose relaxation is controlled by line tension $\\lambda$, bending stiffness $\\kappa$, and an internal pair-dissipation coefficient $\\eta_h$, and whose driving is the velocity field of a viscoelastic model of the monolayer interior. The analysis shows that the dominant finger length scale is the shear length scale of the interior velocity correlations, $\\xi_{\\perp,p}=\\sqrt{(\\mu\\tau+\\eta)/\\zeta_c}$, and that finger lifetimes are set by the slow $q^{-2}$ relaxation of the height modes at that scale. Alongside it, the active vertex model supplies the agent-based testbed: cells crawl as active Brownian particles with persistence time $\\tau$, interactions include substrate friction and pair friction, and the boundary is implemented as coupled linear and angular springs that can grow and shrink.","core_discovery":"On its own terms, the paper's central claim is: correlated active cell motion alone suffices to produce fingers; leader cells, signalling, and proliferation modulate, but do not trigger, this pattern. The authors present fingers as long-lived active fluctuations of the boundary rather than the product of a finite-wavelength instability. They demonstrate this by combining in vitro imaging of spreading epithelial monolayers with simulations of an active vertex model whose edge is a contractile semiflexible polymer, and with a linear viscoelastic height-equation theory for the edge driven by correlated noise from the interior. The quantitative match between theory, simulation, and experiment on tangent-tangent correlations, roughness correlations, height spectra, and long finger lifetimes is presented as evidence that no feedback instability is needed.","pith_inferences":["Inference: the same null model could be tested in other epithelial systems whose interior correlation lengths differ; the theory predicts their finger length scales should track the interior shear length scale rather than any cell-autonomous fingerprint.","Inference: if the edge is passive, then transiently suppressing the actomyosin cable should make fingers wider and shorter-lived, while suppressing interior correlations, for example by raising density toward a jammed state, should suppress fingers even with a normal cable; both experiments are directly doable.","Inference: because the height correlator contains a leading $1/q^2$ term that dominates at long times and large system sizes, very long experiments should show slowly coarsening roughness even though tangent-tangent correlations saturate; this coarsening is an untested corollary."],"forward_implications":["Leader cells appear at the tips of fingers as a response to the local mechanical environment of a forming protrusion, so they should amplify rather than initiate the pattern.","Proliferation and cell flattening contribute to the average border speed but leave the finger statistics unchanged, so division is not the trigger.","Finger length and lifetime are predicted to be set by the interior correlation length and cable parameters: changing cell-cell adhesion, friction, crawling speed, or cable line tension should shift both interior and boundary scales in a predictable way.","The absence of a peak in the boundary height spectrum over time implies that no finite-wavelength instability is at work; the low-$q$ fluctuations saturate to a steady state.","Initial roughness of the border explains the early-time differences between experiment and simulation, and the nonlinear short-scale behaviour shifts the roughness exponent from about 2 to 1."],"supporting_citations":[{"why":"Establishes the experimental wound-healing assay and the observation that finger-like structures appear at the moving front.","marker":"[3]"},{"why":"Provides the PIV-based velocity statistics and the earlier observation of spatially correlated motion that the model is calibrated against.","marker":"[4]"},{"why":"Introduces the relative roughness function used to compare boundary shapes and supplies prior cell-front roughness data.","marker":"[26]"},{"why":"Supplies the dense active matter continuum model of monolayer motion patterns whose velocity correlations the paper extends with viscosity.","marker":"[36]"},{"why":"Provides the self-propelled Voronoi model and solid ground-state selection used in the active vertex simulations.","marker":"[41]"},{"why":"Introduces the Active Vertex Model with dynamic boundary that the paper extends with pair friction and a semiflexible contractile edge.","marker":"[42]"},{"why":"The soft-matter treatment of the worm-like chain that underlies the height-equation derivation of the edge dynamics.","marker":"[58]"}],"fun_headline_variants":["Correlated cell motion alone drives fingers","No leader cells needed: interior motion shapes fingers","Finger formation is a fluctuation, not a leader-driven process","Active matter physics explains finger formation","Correlated swirls, not leaders, produce epithelial fingers"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the actomyosin cable at the edge can be treated as a passive, very soft elastic string whose line tension and bending stiffness are about 25 times smaller than the values obtained by directly mapping the boundary springs of the simulation.","fun_headline_variants_meta":{"raw":{"variants":["Correlated cell motion alone drives fingers","No leader cells needed: interior motion shapes fingers","Finger formation is a fluctuation, not a leader-driven process","Active matter physics explains finger formation","Correlated swirls, not leaders, produce epithelial fingers"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00017,"raw_usage":{"total_tokens":1195,"prompt_tokens":797,"completion_tokens":398,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":413,"completion_tokens_details":{"reasoning_tokens":327}},"tokens_in":413,"tokens_out":398,"duration_ms":5007,"temperature":1.0,"reasoning_tokens":327,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T05:53:31.608624+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure whether the advancing edge is actually passive: if inhibiting actomyosin contractility at the leading edge abolishes fingers while the interior velocity correlations (length scale of about 100 micrometres, persistence of about 1 hour) remain unchanged, the passive-cable picture fails; if fingers persist, the claim that interior correlations alone drive them survives.","supporting_citations":[{"cited_title":"Poujade, E","cited_arxiv_id":null,"evidence_quote":"Establishes the experimental wound-healing assay and the observation that finger-like structures appear at the moving front."},{"cited_title":"Petitjean, M","cited_arxiv_id":null,"evidence_quote":"Provides the PIV-based velocity statistics and the earlier observation of spatially correlated motion that the model is calibrated against."},{"cited_title":"Rapin, N","cited_arxiv_id":null,"evidence_quote":"Introduces the relative roughness function used to compare boundary shapes and supplies prior cell-front roughness data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the self-propelled Voronoi model and solid ground-state selection used in the active vertex simulations."},{"cited_title":"van Saarloos, V","cited_arxiv_id":null,"evidence_quote":"The soft-matter treatment of the worm-like chain that underlies the height-equation derivation of the edge dynamics."}],"review_version":1}