Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

In brief

Contractility-dependent mechanical interplay between cells and the extracellular matrix (ECM) generates an ordered cell-ECM layer with fluid instability features, leading to spontaneous follicle patterning in an ex vivo tissue reconstitution model.

Authors

Karl H. Palmquist, Sydney F. Tiemann, Farrah L. Ezzeddine, Anna Erzberger, Alan R. Rodrigues, Amy E. Shyer.

Correspondence

arodrigues@rockefeller.edu (A.R.R.), ashyer@rockefeller.edu (A.E.S.)

Highlights

• An ex vivo assay reconstitutes the process of embryonic skin symmetry breaking
• Mesenchymal mechanics are sufficient to spontaneously generate morphological pattern
• Contractility-dependent cell-ECM interplay creates supracellular structure
• Theory and experiment show that the cell-ECM layer behaves as an active contractile fluid

Summary

Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

During vertebrate embryogenesis, cell collectives engage in coordinated behavior to form tissue structures of increasing complexity. In the avian skin, assembly into follicles depends on intrinsic mechanical forces of the dermis, but how cell mechanics initiate pattern formation is not known. Here, we reconstitute the initiation of follicle patterning ex vivo using only freshly dissociated avian dermal cells and collagen. We find that con- tractile cells physically rearrange the extracellular matrix (ECM) and that ECM rearrangement further aligns cells. This exchange transforms a mechanically unlinked collective of dermal cells into a continuum, with coherent, long-range order. Combining theory with experiment, we show that this ordered cell-ECM layer be- haves as an active contractile fluid that spontaneously forms regular patterns. Our study illustrates a role for mesenchymal dynamics in generating cell-level ordering and tissue-level patterning through a fluid insta- bility—processes that may be at play across morphological symmetry-breaking contexts.

Introduction

The skin is the largest organ in the body, requiring long-range co- ordination to achieve proper functional architecture. The most decisive morphological event in the skin occurs during embryo- genesis, when a flat tissue bilayer of dermis and epithelium tran- sitions into one studded with an array of follicle primordia. Each primordium emerges as a multicellular aggregate with an acti- vated follicle gene expression program. This architecture serves as the morphological template for the rest of skin development and persists into the adult body. Decades of research has been conducted under the presumption that for such a robust pattern to form, patterns of gene expression must emerge prior to the onset of new forms and therefore regulate the initiation of follicle gene expression as well as structure (Jung et al., 1998; Mandler and Neubu ̈ ser, 2004; Painter et al., 2018). Recent work that compared the dynamics of gene expression and morphological changes in the avian skin showed that the expres- sion of localized signaling molecules does not precede morpho- logical changes (Shyer et al., 2017). Instead, mechanical self-or- ganization of the dermis is sufficient to initiate a periodic pattern and also initiate follicle gene expression through b-catenin- dependent mechanotransduction.

While the mechanism of how the dermal layer intrinsically gen- erates tissue scale pattern remains unknown, it has been shown that the determining forces are dependent on the contractile behavior of the mesenchymal cells within the dermis (Shyer et al., 2017). Contractile behavior in mesenchymal cells has long been appreciated in studies of cultured fibroblasts (Bell et al., 1979; Harris et al., 1981; Baker et al., 2015), and the skin presents an opportunity to articulate how such individual cellular forces coalesce in a collective to drive morphogenesis. Such a mechanism, whereby multicellular self-organization acts as the regulatory step for initiating both tissue-scale morphology as well as gene expression, has received far less attention in studies of vertebrate organogenesis than models based on molecular pre-patterning.

Importantly, unlike molecular mechanisms, which exhaus- tively exploit a single scale (Peter and Davidson, 2017; Raspo- povic et al., 2014), mechanisms articulating multicellular self-or- ganization provide a natural opportunity for coverage across scales, connecting events from the molecule to the organ. Namely, collective cell mechanics can be tied to the state of the cellular components and underlying molecular regulators thereof. Further, organ and tissue structure can be understood as an outcome of spontaneous transitions in the cellular collective. The lack of multiscale mechanisms in development and morphogenesis has recently been highlighted as a funda- mental challenge in the field (Lenne et al., 2021). The skin, with its simplicity of geometry (a flat sheet) and pattern (a grid), serves as an ideal model system to confront the complexity of a multi- scale mechanism of morphogenesis.

Here, we gain insight into how multicellular self-organization of the dermis leads to ordered tissue structure and determine the molecular and cellular interactions underlying such a sponta- neous tissue transition. We do so through the development of an ex vivo tissue reconstitution, where we recapitulate the or- dered follicle pattern in a collective of dermal fibroblasts without any potential organizing molecular cues from neighboring tis- sues. Experimental and theoretical analysis of the relationship between individual cell features and tissue-scale properties serve to elucidate fundamental principles underlying active soft matter, a still poorly understood class of materials in which the individual constituents within a material consume energy and dynamically adapt to their extrinsic environment (van der Gucht, 2018). In this context, the long-range structural properties of a supracellular material provide a conceptual framework for understanding how cells robustly generate long-range morphological order across an organ.

Results

Characterizing transitions in skin architecture at the multicellular scale

To address the question of how symmetry is broken in the skin, we investigated which multicellular features set the conditions for initiation of a follicle pattern. We took advantage of a critical feature of chicken skin patterning, which is that it initiates as a single row of follicles that propagates laterally across a field (Figures 1A and 1F). The progressive nature of pattern formation simplifies the consideration of organ symmetry breaking to a sin- gle dimension. Therefore, to address the question of periodic patterning in the skin, we sought to explain the transition of a uni- form band of cells (Figure 1A) into a row of regularly spaced multicellular aggregates (Figure 1F).

For decades, the cell-extracellular matrix (ECM) architecture has been hypothesized to be a nexus for morphological changes (Ettinger and Doljanski, 1992). We therefore characterized cell and ECM architecture at the collective cell scale. At embryonic day 6 (E6.0), the cells and ECM within the pattern-forming band of dermis that runs anterior-posterior (A-P) through the center of the back are isotopically organized. Over the course of the next 12 h (E6.0–E6.5), F-actin and nuclei acquire a clear orientation along the A-P axis of the body (Figures 1C and 1E). Notably, F-actin anatomy and nuclear structure remain similar in the epidermis from E6.0 to E6.5 (Figure 1B). This supports previous studies discovering that mesoscale structural changes within the dermis precede force transmission to the epidermis (Shyer et al., 2017). To complement cellular architecture, we characterized ECM changes through the staining of collagen-I and fibronectin, two major components of the skin ECM. Exam- ination of the fibronectin meshwork reveals a greater degree of orientation along the A-P axis at E6.5 as compared with E6.0 (Figures 1D and 1E).

Following cell-ECM alignment, over the next 12 h (from E6.5 to E7.0) the aligned band distributes into regularly spaced aggre- gates (Figure 1F). We find the presence of continued collective cell alignment radially around the periphery of aggregates but the absence of alignment at the core (Figure 1I). At the same time as this transition in the dermis, the epidermis shows the ex- pected morphological shift, with epithelial cells above a nascent aggregate constricted en face (Figure 1H). At E7.0, we note increased fibronectin and collagen I accumulation in nascent feather follicle primordia (Figure 1G), evidencing that the forma- tion of this periodic pattern involves not just the rearrangement of cells but also the ECM.

This characterization reveals a tissue-wide A-P orientation of cells and ECM that precedes the formation of a row of aggregates (Figure 1J). Although heterogeneity in molecular positional information affords the most obvious pattern pre-condition, our results suggest that a uniform transformation across the pattern-forming band lays the foundation. We tested whether such a transformation in the cells and ECM alters tissue-level physical properties. In line with this hypothesis, micropipette aspiration (MPA) reveals that aligned skin deforms less than unaligned skin in response to constant applied force (Figure S1). This finding compelled consideration of how shifts in cell and ECM organization occur and, further, how these shifts endow the tissue with mechanical properties that, when coupled with appropriate tissue geometry, induce the formation of novel tissue-scale structures.

Multicellular reconstitution of a spontaneous morphological process

To investigate how a regularly spaced row of aggregates emerges from an isotropic band of cells, we engineered a novel ex vivo platform to recapitulate this transformation outside of the body. Such a platform is necessary to determine whether key cell and tissue transitions are autonomous, self-organized events within the dermis, and also to dissect such a mechanism across biological scales. Freshly harvested embryonic dermal cells were dissociated into a single-cell suspension and plated in a drop atop a collagen gel. Through a natural consequence of fluid mechanics termed the ‘‘coffee ring effect’’ (Deegan et al., 1997), the cells settle onto the collagen substrate around the circumfer- ence of the drop (Figure 2A). Following an hour’s incubation al- lowing cells to settle and attach, media was added to cover the cells. Geometrically, given the width of the ring of cells is small compared with its radius, the additional curvature can be neglected, and the ring serves as a line with no end (Gonza ́ lez et al., 2013), thereby removing confounding irregularities at the edge. This ring state then mirrors the band of isotopically orga- nized cells at E6.0 (Figure 1A). The three-dimensional collagen gel substrate with specific density, volume, and gelation condi- tions is a necessary component to recapitulate the properties of the collagen-rich substrate beneath the dermal cells in vivo. Using MPA, we confirmed that the collagen gel used ex vivo was comparable in deformability and deformation recovery to the in vivo back skin dermis (Figure 2B). Combined with a high- throughput, live-imaging platform, our ex vivo platform allows for live-cell and ECM tracking in large numbers of samples.

When tracking cell arrangement over time, we find our ex vivo reconstitution recapitulates the spontaneous formation of or- dered aggregates observed in the skin (Figures 2C and S2; Video S1). These appear as progressive segregations of both the cells and the ECM (Figures 2C and 2D; Video S2), recapitulating the coupling between the cells and the ECM in vivo. We observe reg- ularity of aggregate pattern within a ring as well as across trials (Figures 2E and 2F). These results demonstrate that dermal cells autonomously generate an ordered pattern of aggregates without the requirement of cues from the epidermis, as previ- ously suggested (Ho et al., 2019). Importantly, not all types of fi- broblasts tested in this culture system undergo such transitions in our ex vivo platform. In particular, a chick embryo fibroblast cell-line fails to properly form a pattern (Figure S2), highlighting the specificity of this patterning forming program and the impor- tance of using fresh primary cells to model an in vivo process.

Dermal cells automatically generate supracellular order through ECM alignment.

Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

Our ex vivo system, where the active component is solely dermal cells, provides an ideal context to investigate the origin and role of the in vivo architecture we uncovered. If this system does not recapitulate the alignment of cells and ECM ex vivo, this would suggest that this transition stems from forces or signals external to the dermal layer of the skin, such as the underlying tissue or axial growth of the body. Further, lack of alignment in the context of ordered pattern formation ex vivo would imply that this step is dispensable for pattern formation.

Instead, we find that we recapitulate cell-ECM alignment ex vivo. In both live-cell imaging of GFP-labeled cells (Figures 3A, 3B, and 3D; Video S2) as well as fixed samples stained for cell morphology markers (actin and nuclei) and

Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

ECM (fibronectin) (Figures 3C and 3D), we capture progressive alignment of both cells and ECM linearly along the pattern axis over time. These results demonstrate that the cell-ECM align- ment observed in vivo is self-organized from within the dermis. Further, the fact that the link between this aligned state and ag- gregation is maintained ex vivo, supports the necessity of this conformation in the process of follicle formation. Of note, upon the formation of aggregates ex vivo, cells at the core become disordered and those at the periphery appear radially oriented (Figures 3E–3G). Therefore, we also recapitulate ex vivo the cell arrangement of the in vivo follicle primordia.

We investigated the mechanism responsible for this critical self-organization into an aligned cell-ECM band. When consid- ering what features of the dermis might allow the tissue to align itself, we hypothesized that cellular contractility serves as the force generator. Indeed, we find that when contractility is decreased through myosin II inhibition, nuclei and fibronectin fail to orient (Figures 4A and 4B). Further we inhibited contrac- tility in whole skin explants and find we recapitulate these re- sults in the intact skin (Figure S3). Given the coupling between the ECM and cell rearrangement, we explored the role of the ECM’s material properties. We find that when we increase cross-linking of the collagen gel substrate through ribose-medi- ated glycation (Francis-Sedlak et al., 2009; Mason and Rein- hart-King, 2013; Vicens-Zygmunt et al., 2015), alignment is significantly delayed (Figures 4C and 4D). Therefore, this align- ment is dependent on particular ECM physical properties. Together, these results support a mechanism whereby contrac- tile cells rearrange the ECM. Such a realignment of the extracel- lular environment would serve to further align the cells, gener- ating a reciprocal cell-ECM exchange responsible for collective alignment. This tissue-wide alignment represents the coordination of cells and ECM beyond the length scale of an individual cell (i.e., supracellular order).

From a biophysical perspective, we noted that the supracellu- lar alignment of cells and ECM is similar to nematic ordering in liquid crystals. Although initially studied in passive chemicals, more recently, concepts from liquid crystal physics have been applied to multicellular contexts, including the supracellular or- ganization of adult liver and cultured epithelial cells (Morales- Navarrete et al., 2019; Saw et al., 2017). In the context of tissue morphogenesis, we reason that nematic ordering both indicates and confers supracellular coordination, allowing cellular forces to have tissue-level effects.

Ca2+ signaling mediates a cell-ECM co-evolution critical for generating supracellular order

We investigated how the establishment of supracellular order sets the stage for transformation to aggregates. In particular, given that ECM alignment and increasing amounts of fibronectin would predict a supracellular structure that would be less likely to separate (Wang et al., 2014; Ahmadzadeh et al., 2017), why does cell-ECM alignment precede separation into discrete units rather than stabilize the field into a linear strip of oriented cells? The spatial and temporal access inherent to the ex vivo platform provides the opportunity to dissect how critical cellular and intra- cellular mechanisms set the conditions for such a tissue-level transformation.

We hypothesized that increases in cell-ECM alignment might stimulate a concomitant increase in cellular contractility due to increasing tension on the cell, as has been demonstrated in other contexts (Ahmadzadeh et al., 2017; Polte et al., 2004). Thus, cell- ECM alignment would trigger a more contractile dermal layer that could overcome the increasing integrity of the ECM to separate into aggregates. To test whether cells shift contractility over time ex vivo, we developed two methods to measure traction forces in cells collected from ring cultures at early (6 h) and late (25 h) time points (Figure 4E). In both methods, we extract cells from the ring and replate to assure equivalent cell arrangements (Figure S3) needed for accurate comparison of traction forces at different times. First, we compared the rate of ex vivo rearrange- ment across cells collected at these time points when dissociated and replated in a new ex vivo ring culture. We find a difference in contractile behavior between cells collected at 6 h and those collected at 25 h. Specifically, those reseeded from the late time point initiated aggregation almost immediately, whereas those from the early time point took significantly longer, indicative of cells from later ex vivo time points being more contractile (Fig- ure 4F). To test this point directly, we also reseeded cells from these two time points in a collective traction force microscopy (TFM) assay, where contractility could be assayed by deformation of a collagen matrix embedded with GFP beads. In line with the above observations, cells isolated from a later time point con- tracted the collagen gel more strongly as compared with those cells isolated from an earlier time point (Figures 4G, 4H, and S3).

To identify the mechanism underlying progressively increasing contractility, we investigated the role of Ca2+-mediated mecha- notransduction, which has been shown in other systems to tune contractility in response to extracellular mechanics (Munevar et al., 2004; Matthews et al., 2010; Kim et al., 2014). In line with such a role for Ca2+ in our system, we find that cells are positive for fluorescent imaging plate reader (FLIPR) Calcium 6 staining, a marker of Ca2+ signaling (Figure 5A; Video S3; Sirenko et al., 2013). Further, from 12 to 32 h, cells on average double the frequency of their calcium oscillations, which is consistent with a role for calcium in increasing contractility over time, ex vivo (Figures 5A and 5B; Doyle et al., 2004). To investigate whether dermal cells respond to changes in ECM deformability by chang- ing Ca2+ signaling frequency, we measured Ca2+ signaling on increasingly cross-linked collagen gels. Indeed, we find that dermal cells on more cross-linked collagen gels have signifi- cantly increased Ca2+ signaling frequency as compared with controls (Figure 5C). In line with a role for Ca2+ signaling in increasing contractility necessary for the transition to ordered aggregates, we found that inhibiting Ca2+ signaling using the

Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

L-type channel blocker, nifedipine (Atsuta et al., 2019), abolishes pattern formation ex vivo (Figure 5D). Additionally, inhibiting Ca2+ signaling interferes with cell alignment, consistent with a role for Ca2+ in increasing cellular contractility (Figure 5E). In turn, acti- vating Ca2+ signaling through the addition of capsaicin appears to activate contractility (Figure S4).

To determine whether Ca2+ signaling facilitates pattern forma- tion by mediating the necessary increase in contractility, we investigated whether ectopically increasing contractility in the absence of proper Ca2+ signaling restores pattern formation. Indeed, we can rescue the effect of nifedipine through addition- ally increasing contractility with calyculin A treatment (Figure 5D). In addition, we found that these same results can be achieved using the inhibitor of stretch-activated channels, GdCl3 (Naka- mura et al., 2001; Guilak et al., 1999), further implicating mecha- nosensitive channels in this process (Figure S4). Lastly, to test whether these observations ex vivo are indicative of an unde- scribed role for Ca2+ in vivo, we applied nifedipine to skin ex- plants. Compared with the control, nifedipine abolishes the supracellular alignment observed at E6.5 and also prevents the formation of feather follicles (Figure 5F). Together, these results support a model where Ca2+ signaling is necessary for cells to tune contractility in response to changing extracellular me- chanics. In doing so, Ca2+ signaling mediates an emergent cell-ECM co-evolution critical for generating an aligned supra- cellular structure and establishing the proper contractility and ECM conditions to transition to a pattern of aggregates.

Mesoscopic liquid phase properties enable spontaneous aggregate formation

We then investigated whether a regular pattern of aggregates is the predicted outcome of an aligned, contractile cell-ECM continuum. We therefore performed a theoretical analysis of how an aggregate pattern initiates. In identifying the general mecha- nism by which regularly spaced aggregates form, we may reveal how patterning is tuned by the biological and physical properties of the system. The simple geometry and the extent of experimental control render our ex vivo platform particularly suited for a theoretical analysis of the most salient features contributing to the initiation of an aggregate pattern. In particular, the orientation of the cells and ECM along the perimeter of the ring stipulates that the problem can be considered along a single dimension.

We therefore formalized our understanding of the effective material properties of the aligned cell-ECM layer into a contin- uum theory, relying on the following four assumptions.

(1) Our findings compel a description of the cell-ECM layer as a viscous material. In particular, we find that in our ex vivo assay, cells are enmeshed in ECM (Figure S5) and this cell-ECM layer undergoes irreversible rearrangements on the timescale of 1–10 h (Video S2), and our MPA results suggest that embryonic chicken skin possesses viscous fluid-like characteristics; namely, it continuously deforms, or flows, over time under constant applied force, and it retains that deformation after force is released (Figure S5). Studies of viscous properties in tissues are underexplored, and viscosity is often excluded in favor of studies of elastic behavior (Cantini et al., 2020; Muller et al., 2013; Murray et al., 1983). (2) Because adherent cells exert contractile stresses on their environment (Schwarz et al., 2002; Schwarz and Safran, 2013; Tanimoto and Sano, 2014), we expect that dermal cells act as attractive elements when embedded in an ECM layer, drawing in adjacent cells. (3) We as- sume that the number and spacing of aggregates is determined at the early stages of the patterning process, i.e., the first 10–20 h. In vivo, the density patterns are read out at E7.5 ($12–24 h af- ter patterning onset) by b-catenin signaling (Shyer et al., 2017), while ex vivo, patterns are stabilized on a similar timescale by the depletion of material in between aggregates. Thus, we aim to describe patterning onset and early aggregation and neglect effects such as spatial variations in rheological properties or dynamic parameter changes that may eventually arise from feed- back between cells and the ECM upon strong aggregation. (4) Consistent with our observation that the timescale of patterning occurs on a much faster timescale than cell proliferation (Fig- ure S5), we do not consider fluctuations in the total cell number.

Based on these points, we modeled the aligned cell-ECM layer as an active attractive fluid (Bois et al., 2011), obtained the hy- drodynamic equations governing the dynamics of the cell den- sity and material flow along the ring, and investigated the spon- taneous appearance of patterns using linear stability analysis and simulations (Figures 6 and S6; Methods S1).

Our theory recovers that the self-attracting activity of contrac- tile cells in the ECM is capable of generating the spontaneous formation of regular aggregate patterns (Figure 6A). Further- more, our calculations reveal how two key parameters govern the capacity of the cell-ECM layer to form patterns: the cellular contractility and the hydrodynamic length. This length parameter characterizes the spatial range of flows in the material (Mayer et al., 2010) and depends on the mechanical properties of the ECM, such as the cross-linking between its components.

In particular, linear stability analysis and simulations yield a distinct relationship between these parameters and (1) the size and spacing of multicellular aggregates and (2) the timescale of pattern formation (Figures 6B, 6C, and S6; Methods S1). Spe- cifically, we find that the number of aggregates that form should increase as contractile stresses generated by the cells increase, and patterning should take place within a shorter period of time. In turn, the model predicts that an experimental decrease of active cellular contractility would lead to fewer aggregates, which should form on a longer timescale. Moreover, an increase in the hydrodynamic length should lead to fewer aggregates and slower patterning.

In order to test the extent to which our theoretical calculations capture the biophysical dynamics of the cell-ECM layer, we compared our predictions with the experimental results. Our ex vivo platform, where we reconstitute multicellular dynamics in the most minimal context, allows for the perturbation of indi- vidual physical features and observations of the resulting pattern dynamics. It therefore serves as a multicellular diagnostic assay, providing a mechanical fingerprint for each population or pertur- bation. When we inhibit cellular contractility with the myosin inhibitor blebbistatin, the resulting pattern matches the predicted decrease in the number of aggregates. Conversely, treatment

Reciprocal cell-ECM dynamics generate supracellular fluidity underlying spontaneous follicle patterning

with calyculin A, an amplifier of cellular contractility, increased the number of aggregates as expected (Figures 6B–6E). Our cal- culations also predict that the mechanical properties of the ECM influence cellular aggregation by setting the spatial range of cellular flows such that increasing viscosity, for example, leads to fewer aggregates and eventually stalls patterning (Figures 6A–6D). We pretreated collagen with different concen- trations of ribose to modulate the cross-linking of collagen fibers and thereby alter the viscosity of the network (Stojkov et al., 2021; Tian et al., 2016). Indeed, at a concentration of 250 mM ribose, fewer aggregates emerge, and treatment with 500 mM ribose abolishes observable aggregation (Figures 6B–6E). More- over, our experimental observations confirm the predicted changes in the patterning timescales for both manipulations (Fig- ure S6). Further details of our analyses, the simulations, and the parameters used are provided in the Methods S1, the materials and methods, Table S2, and Figure S6. This theoretical study, with its experimental agreement, supports a scenario in which the emergence of an aligned supracellular structure sets the con- ditions for a contractile instability that resolves as the sponta- neous emergence of follicles.

Discussion

Mesenchymal mechanics can autonomously initiate periodic morphological pattern

Broadly speaking, morphological changes can involve an expan- sion of an already existing morphology (e.g., elongation [Mon- gera et al., 2018] and spreading [Morita et al., 2017]) or an in- crease in complexity, as is seen in the case of the emergence of follicles in the skin. In vivo, it has remained unresolved whether the initiation of follicle primordia occurs through the action of cellular components with different molecular profiles (Painter et al., 2018) or through a spontaneous self-organization of a ho- mogeneous field of cells (Shyer et al., 2017). Although previous studies have presented in vivo and explant data consistent with both models, here we use a reconstitution approach to demonstrate a process of spontaneous cellular self-organization. Going forward, we anticipate that ex vivo reconstitution of morphological processes will be a critical step in de-coupling the effects of multicellular mechanics from molecular signaling.

This ex vivo system fills longstanding gaps in studies of collective cell behavior. In particular, we gain access to mesenchymal dynamics, whereas organoid and collective cell assays predom- inantly focus on epithelial cells and substitute mesenchyme with a commercial ECM. We also augment classic fibroblast culture on collagen (Stopak and Harris, 1982; Harris et al., 1984; da Rocha-Azevedo and Grinnell, 2013) by showing that through incorporating the proper cell density, ECM properties, and geo- metric constraints seen in vivo, mesenchymal self-organization is responsible for tissue-specific, regular patterns. A further advance is that we incorporate time-lapse imaging, providing a window into cell-ECM dynamics.

By reconstituting each of the critical steps observed in vivo, we provide evidence that these events are not the consequence of well-known signals from the epithelium or from forces provided from neighboring tissue layers. Rather, these events are autonomously generated by the dermis. Given the prevalence of mesenchymal tissue layers across the embryo during morphological maturation, the sufficiency of cellular contractility as an intrinsic force generator may be conserved across tissues con- texts. However, our results indicate that the mechanics of mesenchyme needs focused attention.

A consideration of the ECM has been underemphasized in studies of multicellular mechanics due to its diminished role in epithelial tissue models, in addition to the challenges that visual- izing and manipulating the ECM poses. However, we find that the ECM is vital for setting the stage for a tissue-scale spontaneous transformation. Moreover, although mesenchymal collectives are intrinsically more disordered than epithelial collectives, we find that the alignment of the ECM serves to generate supracel- lular order.

Soft matter physical models as a conceptual framework for morphological transformation

In order to fully explain our data, we found it necessary to invoke concepts developed in the field of soft matter physics. First, we noted that supracellular order was actively created by nematic ordering of cells and the ECM. In the case of a developing tissue, such ordering can be advantageous for transitioning a disor- dered layer of mesenchyme into one that has long-range struc- tural properties. Indeed, by probing the mechanical properties of the tissue, we find that nematic ordering accompanies a shift in material properties of the tissue. Such a transition is likely necessary as tissues generate stable morphological structures on the scale of hundreds to thousands of cells. Furthermore, this long-range order can serve to orient the individual behavior of cells—in this case, contractile forces—along a desired axis. Importantly, we note that such nematic ordering differs from ordering observed in passive systems because, in our biological case, the interactions between the individual units driving the alignment depend on active cytoskeletal processes within the units themselves.

While previous studies of morphological emergence have focused on elastic properties of tissues (Savin et al., 2011), here we found evidence of fluid-like properties as a fundamental driver of spontaneous multicellular self-organization. Rheolog- ical phenomena in development and disease have been previ- ously noted but have yet to be applied to the study of non-trivial morphological changes in vertebrates driven by mesenchyme (Wang et al., 2020; Petridou et al., 2021). The instances where developing vertebrate tissues are modeled as fluids focus on flow in driving folding or spreading of a tissue or tissue region (Saadaoui et al., 2020). In contrast, our findings support a model whereby a contractile instability in an active attractive fluid spon- taneously generates a morphological, periodic pattern. Analo- gous mechanisms for pattern formation occur in active fluids at other scales, including the clustering of myosin motors in the actin cortex (Hawkins et al., 2011; Bois et al., 2011; Alvarado et al., 2013; Hannezo et al., 2015; Nishikawa et al., 2017) and on cytokinetic rings (Wollrab et al., 2016; Thiyagarajan et al., 2021).

Our results suggest that fluid instabilities, such as those observed when thin films of water on glass retract into droplets, offer a powerful organizing framework for periodic pattern formation in tissue biology.

Supracellular regulation bridges molecular events to tissue-level phenotype

The perspective that the fundamental mechanisms responsible for generating form are best framed on scales larger than the mo- lecular scale was posited decades ago (Murray et al., 1983; Os- ter and Alberch, 1982). In contrast, in recent decades, supracel- lular mechanics have been seen as an elaboration of gene and molecular regulation leading to a focused search for gene regu- latory events. Along these lines, studies addressing how a peri- odic pattern of follicles forms, focused predominantly on molec- ular events. The finding that tissue rearrangement and local activation of signals are dependent on forces generated by mesenchymal mechanics broadened the scope of investigation to include larger length-scales (Shyer et al., 2017). Here, we un- cover how such force generation is initiated, uncovering regula- tory events across biological scales.

Regulatory mechanisms respond to a change in system condi- tions and modulate it toward a particular outcome. In the embryo, regulation enables morphological robustness by channeling form toward a specific result rather than veering off course as a result of small fluctuations. Here, pivotal regulatory mechanisms were identified through an analysis of cell and ECM dynamics at the multicellular scale. Although the cell-ECM relationship is often depicted as static or one-directional (Levental et al., 2009; Fraley et al., 2015; Piotrowski-Daspit et al., 2017), we find that the link- age between cells and the ECM evolves dynamically. Such a co-evolution regulates the relative levels of the two key biophys- ical parameters, contractility and ECM viscosity, assuring the formation of a particular pattern out of the entire phase space of pattern possibilities. The presence of stochasticity at the mo- lecular and cellular scale during embryogenesis is commonly observed across organisms. Our results suggest that rather than combating stochasticity through additional molecular regu- lation, mechanisms at the multicellular scale exploit it to generate instabilities that resolve into new patterns of form.

The supracellular regulation observed here undeniably rests on the actions of molecular components, specifically in that the behavior of each cell within a dermal collective is determined by the genes expressed within it. Nevertheless, critical proper- ties emerge at the supracellular scale that could not be discov- ered or understood through persistent dissection of molecular features alone. These results highlight a growing awareness that regulatory events occur at multiple scales and that these events can be simultaneous and linked. A major challenge in studies of morphogenesis will be to strategically select scales of study to find the proper foothold into multiscale processes. We propose that the multicellular scale can bridge the vast cat- alog of molecular states to higher order tissue structure in two ways. First, the multicellular scale can act as a phenomenolog- ical filter, allowing for more focused screening of molecular drivers given known cell properties of interest. Second, multicel- lular properties set the template for organ and tissue structure through spontaneous transitions that emerge from a supracellu- lar continuum.

Limitations of the study

The mechanism we present focuses on the critical step of pattern initiation. However, this is just one key aspect of the complex dynamic process of follicle formation. Once the folli- cle initiates in the skin, signaling molecules are activated through mechanotransduction. Future work will investigate how such signals modify the individual properties of skin pro- genitor cells, thereby altering the macroscopic properties of the dermal tissue to capture the initiating structure and then perpetuate its formation. Such studies of how signals shift dermal tissue properties will inform our theoretical account of follicle formation, as our current model does not account for how the pattern is cemented once initiated. Further, our study exploits the simplicity of symmetry breaking in just one dimension. Therefore, in future extensions of this mecha- nism, it will be necessary to consider how the emergence of the initial midline row mechanically primes the emergence of adjacent rows. Moreover, investigations of pattern emergence in two dimensions across a tissue field will be needed to address whether this mechanism explains follicle patterning during mammalian skin development.

For supplemental information, acknowledgements, references and more, please click this link to view the pdf version of this article.