{"id":"7d2b03f7-e5d7-4af7-9bfa-aa6e05a215f3","arxiv_id":"2505.20543","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A localized symmetry-breaking defect in a Voronoi tissue model produces multicellular rosettes whose probability depends on cell deformability, motility, and alignment, and neural rosettes appear in the predicted fluid-like regime.","lead":"The authors used computer simulations of a tissue model to find which cell properties make groups of cells cluster into rosettes around a central point. They found that a balance between cell deformability and mobility, plus cell alignment, controls rosette formation, and microscopy data on stem-cell-derived neurons supports the prediction.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Experimental validation may be comparing the wrong variable: Fig. 7 uses measured Voronoi shape factor, while the model's control parameter p0 is the target perimeter in Eq. 3; the claimed agreement therefore does not yet test the model's fluidity/deformability prediction.","rationale":"The reader's weakest assumption concerned whether external potentials and leader cells faithfully represent natural symmetry breakers. That is a legitimate concern, but the model is explicitly a minimal-ingredient sufficiency study and the leader-cell variant already shows that an imposed size defect can replace the potential. The more decisive gap is in the only quantitative experimental test: the paper's headline experimental support is the observation that rosettes appear where average shape factor exceeds 3.95 (Section II.F, Fig. 7c), claimed to match the fluid-like region of the model's parameter space. But the model's phase diagrams are functions of p0, the target perimeter in Eq. 3, not of the realized shape factor, and the manuscript never reports the realized shape factor in simulations. Since self-propulsion and the symmetry-breaking field both deform cells away from their target shape, and rosette cells are elongated by construction, the measured S_t cannot be equated to p0 without a calibration step. The time-course comparison is also vulnerable to reverse causation: a tissue that is becoming rosette-rich will inevitably show higher mean shape factors if rosette cells are more elongated, so maintaining S_t > 3.95 in rosette-forming regions may be a symptom, not a cause. The proposed test is computational and cheap: measure realized shape index in the existing simulation output. If the realized-shape-index map still separates rosette-forming from non-forming regions at S_t > 3.95 with the same v0 and k used in Fig. 7, then the agreement is real. If not, the experimental claim reduces to a correlation whose mapping to model parameters is unverified. This does not undermine the simulation results themselves, but it means the paper has not 'proved' that rosettes are natural outcomes of curvature; it has shown sufficiency under imposed defects and a correlative, possibly circular, experimental signature. The verdict should remain CONDITIONAL: the theory is plausible, but the validation needs the realized-shape-index calibration and a temporal precedence check before the agreement can be accepted.","tokens_in":16136,"tokens_out":11887,"duration_ms":131436,"concrete_test":"Re-run the simulations underlying Figs. 2, 3, and 5 and record the time-averaged realized shape index (perimeter/sqrt(area) of Voronoi cells) for every (p0, v0, J, k) condition, alongside the existing P_r values. Overlay the experimental threshold 3.95 on the realized-shape-index map; if experimental rosette regions do not fall inside the high-P_r region on that map, the experimental validation is invalid. In parallel, in the day 6-11 time-course, compute S_t in each ROI at the earliest frame before any rosette is visually identified; if S_t is not already above 3.95 at that pre-rosette time, the high shape factor is likely a consequence of rosette formation, and the causal ordering claimed in Fig. 7c is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing weakness is in Section II.F's experimental validation. The model's control parameter p0 in Eq. 3 is the preferred perimeter (target shape index), and the phase diagrams in Figs. 2-5 map rosette probability against this parameter, not against the realized shape factor. The experiments, however, measure S_t = p_i/sqrt(A_i) from Voronoi tessellations of nuclear positions (Methods E) and then identify rosette-forming regions by S_t > 3.95. These are not the same observable: in the active Voronoi model the realized mean shape index can differ from p0 because of self-propulsion, confinement, and the symmetry-breaking field itself, and rosette cells are by construction elongated, so high S_t may be a consequence rather than a cause of rosette formation. Consequently, the observation that rosette-forming regions maintain S_t > 3.95 does not place those cells in the high-P_r region of the (p0, v0) phase diagram, and the claimed 'significant agreement' with model predictions is not established. The text also calls the measured quantity the 'average target shape factor' (Methods E), conflating the realized observable with the model parameter.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses the Active Voronoi model to study the formation of multicellular rosettes in epithelial layers. The authors simulate three types of symmetry-breaking perturbations: a harmonic potential localized at a point, a Lennard-Jones potential imitating a lumen, and a leader cell with a different target area. They report that the probability of rosette formation depends on the cell target perimeter p0 and self-propulsion velocity v0, that self-alignment interactions enhance rosette stability and size, and that lumen-centered rosettes grow and polarize with increasing p0. To test the predictions, the authors perform time-lapse fluorescence microscopy on iPSC-derived neural rosettes, estimate v0 and tau from single-cell MSD curves, and find that rosettes appear in regions with measured Voronoi shape factor above 3.95. They conclude that rosettes are natural outcomes of geometric and mechanical responses to curvature.","tokens_in":16461,"tokens_out":10766,"duration_ms":105114,"significance":"The simulation study is systematic: parameter sweeps over p0, v0, J, and potential strengths are accompanied by stationary-state checks, and the use of a Voronoi model avoids hand-implemented topological transitions. The alignment-interaction effect is a plausible mechanism for stabilizing rosettes. However, the experimental validation has a variable-matching problem (realized shape factor vs. target p0), and the rosette propensity proxy is not directly linked to high-order vertices. These issues currently prevent the paper from supporting its stronger causal claims. If the authors can re-analyze the experimental comparison using realized shape factors from the simulations and validate the proxy, the conclusions would be substantially strengthened.","major_comments":[{"comment":"The rosette propensity P_r is defined as the frequency of cells with fewer than 5 edges, but the paper's own definition of a rosette is a vertex where five or more cells meet (Introduction). These are distinct observables: a cell's edge count is not the same as the degree of a Voronoi vertex. No validation is shown that P_r tracks the actual number of rosette vertices (e.g., by comparing P_r with direct vertex-degree counts in the same configurations). Because the central phase diagrams and the trade-off claim rely entirely on P_r, the quantitative results would be more convincing if the rosette frequency were measured directly or if the proxy were benchmarked against direct counts. As written, the proxy is an unverified assumption.","section":"Section II.A and Eq. (3); Figures 2-5"},{"comment":"The experimental validation compares the measured Voronoi shape factor S_t with the model's phase diagram parameter p0. However, p0 in Eq. (3) is the target perimeter (a model input), whereas S_t is the realized shape factor computed from the experimental Voronoi tessellation. The realized shape factor is not equal to p0 in the model: activity, the external potential, and the rosette itself change the actual cell shapes, and the authors themselves note that the external field fluidifies the layer even at low p0. Therefore the observation that rosette-forming regions have S_t > 3.95 does not place those cells in the high-P_r region of the (p0, v0) maps, and the claimed 'significant agreement' is not established. The authors should compute the realized shape factor from the simulations and compare it with the experimental S_t, or otherwise provide a mapping between S_t and p0.","section":"Section II.F and Methods E"},{"comment":"The statement 'our work proved that rosettes are natural outcomes of geometric and mechanical responses to curvature' overstates the evidence. In the simulations, symmetry breaking is inserted by hand via a harmonic potential, a Lennard-Jones potential, or a leader cell with a different target area; the experiments are correlative. The model demonstrates that such imposed defects can produce rosette-like structures, but it does not prove that biological rosettes originate from a localized curvature defect. I recommend softening the claim to 'consistent with' and explicitly stating the assumption that the external potentials represent biological symmetry-breaking cues.","section":"Discussion, third paragraph"},{"comment":"The Discussion argues that the harmonic potential 'inserts the simplest positive correction to the curvature' and changes the Euler characteristic to χ>0. However, the simulations are performed on a flat torus with periodic boundary conditions; Eq. (4) adds a potential energy, not a change in the surface metric. The argument that rosettes are stabilized by an excess of positive topological charge on a curved surface is an analogy, not a property of the simulated model. If the authors wish to support the curvature mechanism, they could simulate cells on an intrinsically curved surface (e.g., a sphere or a bump) and compare with the potential-field results.","section":"Discussion, Euler characteristic/curvature argument"}],"minor_comments":[{"comment":"The text refers to the measured quantity as the 'average target shape factor'; this is a misnomer because the quantity is computed from actual cell geometries, whereas the target shape factor is the model parameter p0. Use 'measured shape factor' instead.","section":"Methods E"},{"comment":"The phrase 'i. e.,τ≫τ' should read 't≫τ'.","section":"Section II.F, after Eq. (7)"},{"comment":"The sentence 'Thus, l̃p = f·l_p = 10l_c' is confusing; please clarify the notation (l_c vs. 10 l_c) and the scaling factor.","section":"Section II.F"},{"comment":"The text refers to 'Figure 4l' but the caption lists no panel l; the panel labels j-k appear out of order. Please renumber.","section":"Figure 4"},{"comment":"The phrase 'cells with less than 5 edges' is grammatically ambiguous; if the intended proxy is 'fewer than 5 edges,' state it explicitly, or correct to 'more than 5 edges' if that is what was meant.","section":"Section II.A"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses a timely topic and the modeling framework is appealing. The most serious concern is the mismatch between the experimental observable (realized shape factor) and the model's control parameter (target shape factor); this undermines the claimed quantitative agreement and needs to be resolved before publication. The P_r proxy should also be validated. The paper's language should be moderated from 'proved' to consistency claims. I recommend major revision rather than rejection, as the core simulation results appear sound."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere is my take on arXiv:2505.20543. The paper does something genuinely new: it uses the Active Voronoi model to ask what minimal ingredients produce multicellular rosettes, and answers with a clear trade-off between tissue fluidity (target shape index p0) and single-cell motility (v0), with self-alignment interactions (J) as a strong stabilizer. The three symmetry-breaking scenarios—harmonic potential, Lennard-Jones lumen, and leader cell—are sensible, and the parameter maps in Figures 2–5 are a useful contribution. The experimental part is also more credible than many attempts: v0 and tau are fitted from mean-squared displacement, not from rosette occurrence, so the prediction is not self-fulfilling. The observation that neural rosettes appear in regions with shape factor above 3.95 is suggestive.\n\nThe soft spot is the experimental validation. The model's control parameter p0 is the target perimeter in Eq. 3, but the experiments measure S_t = P/sqrt(A) from Voronoi tessellations of nuclear positions. These are not the same observable: the realized shape index in the model can differ from p0 because of self-propulsion, confinement, and the symmetry-breaking field itself. Moreover, rosette cells are elongated by construction, so high S_t in rosette regions could be a consequence of rosette formation rather than a cause. The paper should re-plot the simulation results against the realized mean shape index and show that rosette probability correlates with that. The time-course in Fig. 7c helps, but it still uses the wrong axis for the comparison. This is addressable, not fatal.\n\nOther minor issues: the rosette proxy P_r (fraction of cells with <5 edges) is indirect; experimental rosette identification is manual; no public code or data; and the Discussion's 'proved' overstates what the evidence supports. None of these change the core simulation findings, which appear internally consistent and well checked for stationarity.\n\nWho is this for? Anyone working on tissue biophysics, developmental biology, or organoid self-organization. The simulation results deserve a serious referee, and the experimental part can be fixed with a re-analysis. I would send this to peer review with a request to address the p0-vs-S_t mismatch.\n\nBest","headline":"Simulations are solid and the parameter sweeps are new, but the experimental comparison uses the realized shape factor where the model predicts in target p0, so the central agreement claim is not yet established.","tokens_in":16977,"tokens_out":4136,"would_cite":true,"duration_ms":39021,"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":"Rosette assemblies in epithelia are symmetry-breaking responses to local curvature, with probability set by a fluidity–deformability trade-off and stabilized by self-alignment.","keywords":["Active Voronoi model","multicellular rosettes","tissue fluidity","cell deformability","curvature defects","self-alignment","neural rosettes","iPSC differentiation"],"falsifier":"Track the shape factor and motility of cells in many rosette-forming and non-forming regions of a living epithelium; the model predicts rosettes appear only in fluid-like regions with average shape factor above about 3.9 and slow cells, and that flat homogeneous monolayers without any localized defect produce essentially no rosettes. Observing rosettes in stiff, fast-moving, defect-free regions would refute the central claim.","tokens_in":15930,"feed_emoji":"🔬","tokens_out":7932,"duration_ms":78914,"temperature":0.7,"pith_summary":"Rosettes—assemblies of five or more cells meeting at a common point—appear in development, wound healing, and cancer, yet the general conditions that favor them have been unclear. This paper argues that rosettes are the natural response of a cell layer to a localized symmetry break: a defect that introduces local curvature, such as a focus of actin tension, a curved surface, or an oversized cell. Using an Active Voronoi model, the paper shows that such a defect creates topological defects in the tissue, and that the probability of assembling a rosette is governed by a trade-off between tissue fluidity and single-cell deformability. Self-alignment of cell motion with cell elongation strongly stabilizes transient rosettes and enlarges lumen-centered ones. Fluorescence time-lapses of iPSC-derived neural cells confirm the prediction: rosettes appear in regions where the average cell shape factor stays above about 3.9, the fluid-like regime.","feed_headline":"Rosette formation comes down to curvature and tissue fluidity","feed_subtitle":"A Voronoi model and stem-cell videos show rosettes appear where cells are slow, deformable, and aligned.","key_machinery":"The carrying mechanism is the Active Voronoi model: cell centers move under self-propulsion and forces from a vertex energy $$E_v=\\frac{1}{2}\\sum_i\\big[K_p(p_i-p_0)^2+K_A(A_i-A_0)^2\\big],$$ so topological rearrangements happen automatically rather than being inserted by hand. Symmetry breaking is imposed by an external harmonic potential $E_h=\\frac12 k(\\mathbf r_i-\\mathbf r_c)^2$ (transient rosettes), a Lennard-Jones potential centered on one point (lumen-centered rosettes), or a leader cell with different target area (pinwheel-like rosettes). A self-alignment term $J\\sin(\\theta_i-\\phi_i)$ in the orientation dynamics couples self-propulsion direction to cell elongation. The argument connects these ingredients to curvature: a localized positive-curvature defect changes the topological-charge balance, making higher-fold vertices energetically favorable; cell motility and deformability then determine whether the defect is stabilized or washed out.","core_discovery":"The paper's central claim, stated in its own terms, is that rosettes are natural outcomes of geometric and mechanical responses to curvature. In flat space with homogeneous parameters, Euler's relation forces an average of six edges per cell and hexagonal tiling is preferred; higher-fold vertices are unstable and resolve by T1 transitions. Inserting a localized defect—harmonic potential, Lennard-Jones potential, or leader cell of different target area—breaks translational invariance, and the tissue accommodates the induced positive curvature by forming topological defects in which five or more cells meet at a central point. The probability of these transient rosettes increases with cell deformability (target shape index $p_0$) and decreases with self-propulsion speed $v_0$, with an optimal potential strength; self-alignment $J$ boosts the probability and increases the size and polarization of lumen-centered rosettes. The experimental neural-cell data match this picture: rosettes form where the average shape factor is above about 3.9, and single-cell tracks show motion aligned with cell elongation.","pith_inferences":["Beyond the paper: the same curvature-defect logic should apply to other rosette-forming contexts, such as tumor spheroids and organoids, so measurements of shape factor and motility there should predict rosette frequency in the same way.","Beyond the paper: because the external potential is a stand-in for apical actin tension, a direct test is to perturb actomyosin contractility and check whether rosette probability moves along the predicted fluidity–deformability map.","Beyond the paper: rosette probability could be re-expressed as a function of topological-charge density in microscopy images, giving a quantitative imaging biomarker for tissues about to form rosettes."],"forward_implications":["Rosette formation is predictable from tissue-level parameters: deformability (target shape index), self-propulsion speed, persistence time, and alignment strength.","Regions with average shape factor above about 3.9 are the ones that form transient rosettes and can mature into lumen-centered rosettes; regions that stiffen over time do not.","Turning on self-alignment raises the probability of transient rosettes and increases first-layer and second-layer cell numbers by roughly 30 percent in lumen-centered rosettes.","A single oversized leader cell is sufficient to nucleate rosettes without any external potential, showing that a purely mechanical defect can play the role of curvature.","Introducing an external potential fluidizes the layer even for low shape index, shifting the rosette-forming region in parameter space."],"supporting_citations":[{"why":"Supplies the Active Voronoi model, the vertex energy, and the motility-driven solid–fluid transition that the parameter maps are built on.","marker":"[17]"},{"why":"Establishes multicellular rosette formation as a sequence of T1 intercalation events, the topological picture the model reproduces.","marker":"[6]"},{"why":"Provides the rosette definition (five or more cells around one center) and the measured rosette frequencies used as the baseline for the model.","marker":"[9]"},{"why":"Shows apical actin networks exert coordinated mechanical tension, the biological rationale for modeling curvature as an external potential.","marker":"[21]"},{"why":"Supplies the interplay of curvature and rigidity in shape-based tissue models, supporting the fluidification-by-curvature effect.","marker":"[22]"},{"why":"Reports curvature-induced unjamming and fluidization of epithelial layers, the effect the external potential reproduces.","marker":"[23]"},{"why":"Provides the alignment-interaction formalism and its structural effects, the basis for the self-alignment term in the model.","marker":"[24]"},{"why":"Gives the iPSC neural differentiation protocol under which rosettes form after 8–10 days, the experimental system used for validation.","marker":"[25]"},{"why":"Supplies the active-particle mean-squared-displacement expression used to extract self-propulsion velocity and persistence time from single-cell tracks.","marker":"[26]"}],"fun_headline_variants":["Rosettes form when curvature breaks tissue symmetry","Curvature and cell deformability drive rosette assembly","Symmetry-breaking defects seed rosette formation","Rosette odds hinge on tissue fluidity and cell shape","Curvature defects, not chance, create rosette clusters"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on the assumption that the localized pull (or barrier) and the oversized leader cell used in the simulations faithfully represent the real biological triggers of rosettes, such as apical actin tension, organoid curvature, or an atypically large cell; if natural rosettes are initiated by something else, the central causal claim loses its basis.","fun_headline_variants_meta":{"raw":{"variants":["Rosettes form when curvature breaks tissue symmetry","Curvature and cell deformability drive rosette assembly","Symmetry-breaking defects seed rosette formation","Rosette odds hinge on tissue fluidity and cell shape","Curvature defects, not chance, create rosette clusters"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000267,"raw_usage":{"total_tokens":1631,"prompt_tokens":979,"completion_tokens":652,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":595,"completion_tokens_details":{"reasoning_tokens":573}},"tokens_in":595,"tokens_out":652,"duration_ms":13092,"temperature":1.0,"reasoning_tokens":573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:53:02.971273+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Track the shape factor and motility of cells in many rosette-forming and non-forming regions of a living epithelium; the model predicts rosettes appear only in fluid-like regions with average shape factor above about 3.9 and slow cells, and that flat homogeneous monolayers without any localized defect produce essentially no rosettes. Observing rosettes in stiff, fast-moving, defect-free regions would refute the central claim.","supporting_citations":[{"cited_title":"Cristina Marchetti, and M","cited_arxiv_id":null,"evidence_quote":"Supplies the Active Voronoi model, the vertex energy, and the motility-driven solid–fluid transition that the parameter maps are built on."},{"cited_title":"Todd Blankenship, Stephanie T","cited_arxiv_id":null,"evidence_quote":"Establishes multicellular rosette formation as a sequence of T1 intercalation events, the topological picture the model reproduces."},{"cited_title":"Smith, Natalia White, Vivienne Wilkins, Tomoko Watanabe, Abigail Moore, Bradley Joyce, Jacintha Sugnaseelan, Tristan A","cited_arxiv_id":null,"evidence_quote":"Provides the rosette definition (five or more cells around one center) and the measured rosette frequencies used as the baseline for the model."},{"cited_title":"Os- cillatory behaviors and hierarchical assembly of contrac- tile structures in intercalating cells.Physical Biology, (4):045005, July 2011","cited_arxiv_id":null,"evidence_quote":"Shows apical actin networks exert coordinated mechanical tension, the biological rationale for modeling curvature as an external potential."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the interplay of curvature and rigidity in shape-based tissue models, supporting the fluidification-by-curvature effect."},{"cited_title":"Fredberg, and Dapeng Bi","cited_arxiv_id":null,"evidence_quote":"Reports curvature-induced unjamming and fluidization of epithelial layers, the effect the external potential reproduces."},{"cited_title":"Cristina Marchetti, Ignacio Pagonabarraga, and Gi- ancarlo Ruocco","cited_arxiv_id":null,"evidence_quote":"Provides the alignment-interaction formalism and its structural effects, the basis for the self-alignment term in the model."},{"cited_title":"Self- organization of neural tissue architectures from pluripo- tent stem cells.Journal of Comparative Neurology, 522 (12):2831–2844, May 2014","cited_arxiv_id":null,"evidence_quote":"Gives the iPSC neural differentiation protocol under which rosettes form after 8–10 days, the experimental system used for validation."}],"review_version":1}