bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022.

The copyright holder for this preprint

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Matrix viscoelasticity controls spatio-temporal tissue organization

Alberto Elosegui-Artola1,2,3,#, Anupam Gupta1,#, Alexander J. Najibi1,2, Bo Ri Seo1,2, Ryan Garry1,

Max Darnell1,2, Wei Gu4, Qiao Zhou4, David A. Weitz1,5, L. Mahadevan1,5,6*, David J. Mooney1,2*

1.- Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University,
Cambridge, MA, USA
2.- Wyss Institute for Biologically Inspired Engineering, Cambridge, MA, USA
3.- Institute for Bioengineering of Catalonia, Barcelona, Spain.
4.- Weill Cornell Medicine, Cornell University, NY, USA
5.- Department of Physics, Harvard University, Cambridge, United States, MA, USA
6.- Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA,
USA

# Equal contributions
* Corresponding authors

1
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

1 The spatio-temporal patterning of multicellular tissues is driven by the collective dynamics

2 of cell proliferation and active movement. These processes are mediated by the

3 extracellular matrix environment via a combination of biomolecular and physical cues.

4 Here we show that the passive viscoelastic properties of the matrix that encapsulate a

5 proliferating ball of cells (e.g. a developing organoid) play a critical role in guiding tissue

6 organization in space and time. By varying the viscoelasticity of well-defined model

7 matrices, we show how a spheroidal tissue of breast epithelial cells breaks symmetry and

8 forms finger-like protrusions that invade the matrix. A computational model allows us to

9 recapitulate these observations and leads to a phase diagram that demarcates the regions

10 of morphological stability and instability as a function of matrix viscoelasticity, tissue

11 viscosity, cell motility and cell division rate. Experiments that use biomolecular

12 manipulations to independently vary these parameters confirm our predictions. To further

13 test our theory, we also study the self-organization of an in-vitro intestinal organoid and

14 show that the morphological changes of this system also fits within our paradigm.

15 Altogether, our studies demonstrate the role of stress relaxation mechanisms in

16 determining the dynamics of tissue growth and the symmetry breaking instabilities

17 associated with branching, a fundamental process in morphogenesis and oncogenesis,

18 and suggest ways of controlling tissue form using the extracellular matrix.

19

20 Introduction The patterning of tissues in space and time is relevant for many fundamental

21 biological processes (e.g., embryonic development, organogenesis, oncogenesis)1-5, and is

22 driven by cell number, size, shape and position changes and leads to symmetry breaking

23 instabilities such as buckling, folding, tearing, budding or branching6-9. At a molecular level, the

24 spatio-temporal organization of tissues is regulated by intrinsic gene expression10, and a variety

25 of environmental chemical and mechanical cues11. While the importance of chemical morphogen

1
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

26 gradients in development has long been appreciated12,13, it is increasingly clear that diverse

27 mechanical cues14-19 in the tissue and the surrounding 3D extracellular matrix (ECM) also regulate

28 tissue organization and morphogenesis. In particular, the dynamic interaction between cell

29 behavior and the matrix, with its time-varying mechanical properties is increasingly thought to be

30 an important player in morphogenesis20-22. Thus, tissue organization is expected to be impacted

31 by the viscoelastic properties of the matrix23 whose behaviors vary from an elastic solid-like

32 response to a liquid-like viscous response, with stress relaxation time scales that range from a

33 second to a few hundred seconds20,24. Here we report an experimental and computational study

34 of the role of the viscoelasticity of well-defined model matrices in regulating tissue organization in

35 two commonly used in-vitro models of development and pathology, breast epithelial growth25 and

36 intestinal organoid development1. These studies demonstrate the role of stress relaxation in

37 determining the dynamics of tissue growth and the symmetry breaking instabilities associated

38 with branching, a fundamental process in morphogenesis and oncogenesis.

39 Results

40 Matrix viscoelasticity regulates breast epithelial tissue organization

41 We first studied the importance of matrix viscoelasticity in the organization and growth of

42 mammary tissues from spheroids of MCF10A non-malignant breast epithelial cells. Hydrogels

43 formed from the natural polysaccharide alginate were chosen as the model matrix system for

44 these studies, as mammalian cells do not express enzymes to degrade these polymers, allowing

45 effects related to matrix degradation to be eliminated26. The relative viscoelastic properties of

46 these gels can be readily altered independently of the stiffness, pore size and adhesive ligands24.

47 This was achieved here by changing the molecular weight of alginate and the calcium crosslinker

48 density in concert (Fig. 1a) to create gel matrices of constant elastic moduli (G’~5000Pa) (Fig.1b),

49 but varying stress relaxation times (𝜏m 𝜖 [30 − 350𝑠]) to achieve matrices that are more elastic

2
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

50 (𝜏m ~350𝑠), or more viscoelastic (𝜏m ~30𝑠) (Fig. 1c). As alginate does not present intrinsic integrin

51 adhesion ligands, Arg-Gly-Asp (RGD) containing peptides were conjugated to the polymer

52 backbone to provide a constant level of cell binding sites in all gels27. MCF10A breast epithelial

53 cells, widely used to study mammary development and oncogenesis25, were formed into

54 spheroids composed of ~2000 cells and encapsulated in elastic and viscoelastic hydrogels.

55 Over time, tissues in elastic matrices grew slowly and were morphologically stable; they increased

56 in size while maintaining their spherical symmetry. However, tissues in viscoelastic matrices grew

57 much faster. As they increased in size, they exhibited a morphological instability of the nominally

58 smooth tissue-matrix interface; eventually the tissues broke spherical symmetry, formed

59 branches, and invaded the matrix leading to a significant increase in the surface area and a

60 decrease in circularity (Fig.1d-f, Extended Data Fig.1a and Video S1). This is similar to the

61 behavior seen in many biological processes that demonstrate symmetry breaking accompanied

62 by epithelial to mesenchymal transitions (EMT)28. In agreement with that precedent, cells in

63 viscoelastic matrices demonstrated an EMT, as vimentin was expressed in branches (Fig.1g) and

64 cytokeratin 14 expression was low in cells in spheroids in viscoelastic matrices (Fig.1h,i). To

65 determine whether viscoelasticity enhanced tissue growth in-vivo, MDA-MB-231 malignant breast

66 epithelial cells encapsulated either in viscoelastic or elastic matrices were injected in NOD-SCID

67 mice. Tissues grew significantly more rapidly in viscoelastic rather than in elastic matrices (Fig.1j

68 and Extended Data Fig.2). As all observed differences in-vitro and in-vivo resulted from a change

69 in the mechanical properties of the matrix, our studies next focused on two major

70 mechanosensitive hubs in cells, focal adhesion kinase (FAK) and the mechanosensitive

71 transcriptional regulator Yes-Associated protein (YAP)29, both with established roles in MCF10A

72 EMT2,30,31. Viscoelastic, but not elastic matrices promoted the expression of phosphorylated pFAK

73 adhesions (Fig.1k), while YAP remained in the cytoplasm in cells in elastic matrices, but

74 translocated to the nucleus in cells in branches in viscoelastic matrices (Fig.1l,m). When FAK was

3
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

75 inhibited (Extended Data Fig.1b,c), breast epithelium was morphologically stable, confirming the

76 importance of mechanotransduction.

77 Our experiments show that more elastic matrices (𝜏m ~350𝑠), resist tissue invasion, whereas

78 viscoelastic matrices (𝜏m ~30𝑠), are easily invaded by the motile and proliferating cells. Similarly,

79 our observations show that tissues which are highly proliferative lead to an increase in cell influx

80 and likely generate a mechanical pressure that drives the morphological instability of the tissue-

81 matrix interface. These observations of fingering morphologies in active biological systems have

82 physical analogs that have been studied for decades in simple and complex fluids32,33. In physical

83 systems, morphological instabilities emerge when driven by pressure gradients (of the right sign)

84 at an interface between contrasting either elastic or viscous properties. More recently, these

85 physical instabilities have been revisited in active matter systems34-36. Our experimental

86 observations suggest that the combination of biological activity due to cell migration and/or

87 proliferative pressure at the tissue-matrix interface may lead to a similar symmetry breaking

88 instability exemplified by fingering or branching.

89 Computational model coupling cell motility, proliferative dynamics and matrix

90 viscoelasticity recapitulates tissue organization

91 To understand how the conditions for tissue morphological instability emerge, we consider a

92 minimal theoretical model of the system (Fig.2a and Extended Data Fig.3) starting from a two-

93 phase system of active proliferating cells growing inside a confining passive viscoelastic matrix.

94 We model the individual cells in the tissue as overdamped soft elastic spheres of size 𝑎 in a liquid

95 of effective viscosity 𝜇t , which move under the influence of three forces: (i) the interaction

96 between cells, with (a) a short-range repulsion to prevent overlap and (b) mid-range (two cell-

97 length) attraction with the depth in the attractive well 𝜖 (see SI for details) which together lead to

98 an active proliferative pressure driven by cell-division, (ii) the repulsion between the cell and the

4
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

99 surrounding viscoelastic matrix (modeled as a set of similar spheres of size 𝑎 in a liquid of effective

100 viscosity 𝜇m interacting with each other via (a) an attractive potential -equivalent to storage

101 modulus 𝐺′- and (b) a short-range repulsion to prevent overlap), and (iii) the activity of cells that

102 are assumed to move randomly relative to each other in the bulk, characterized by a motility

103 parameter 𝑀 (or an effective temperature)37,38. Additionally, in the model, the cells at the interface

104 are assumed to have the ability to apply forces on the surrounding matrix39,40. The system evolves

105 as cells proliferate and/or migrate actively and the matrix responds passively to the accompanying

106 forces. In particular, the bonds between the spheres in the matrix as well as those between the

107 cells and the matrix can break when strained beyond a prescribed threshold, allowing new bonds

108 to form; this is most likely to happen at the interface between the tissue and the matrix, and allows

109 the boundary between the two phases to evolve dynamically.

110

111 The parameters in the model allow us to define three dimensionless variables to characterize the

112 scaled matrix fluidity, the passive mechanical relaxation time of the matrix and the relative
𝜇
113 proliferative capacity of the tissue: (i) 𝜇 = 𝜇 𝑡 , the ratio of the tissue viscosity 𝜇𝑡 to the matrix
𝑚

𝜏g
114 viscosity 𝜇𝑚 , (ii) 𝑗 = 𝜏t
, the ratio of the constant timescale to add one cell to the tissue in the

115 absence to stress, 𝜏g , and the varying timescale to add one cell to the confined tissue in the

𝜏a 𝜏g
116 presence of stress, 𝜏t and (iii) 𝐴 = the ratio of the cell activity timescale 𝜏a = 𝑀 where 𝑀
𝜏m 𝜖

117 is the effective motility and 𝜖 is the strength of cell-cell adhesion, and the matrix relaxation
𝜇m
118 timescale, 𝜏m = 𝐺′
, where 𝐺′ is the shear (storage) modulus of the matrix. In our experiments,

119 𝜏𝑔 is ~30s for MCF10A, if one starts with the 2000 cells used in our studies,41 which is in the range

120 of our matrices stress relaxation times (~30-350s). It is known that increasing the mechanical

121 stress prevents division,42,43 which will lead to a slower rate of addition of cells to the tissue (𝜏t )

5
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

122 and a smaller scaled cell flux 𝑗. Each of these dimensionless parameters can be large or small

123 (relative to unity) and plays a role in controlling morphological stability of the growing tissue.

124

125 Systems with low 𝜇 correspond to relatively viscous matrices, while those with high 𝜇 correspond

126 to matrices that are relatively fluid. When the scaled cell flux 𝑗 is small, the pressure due to the

127 growing tissue is not large enough to create fingers in the matrix, while when 𝑗 is large, branching

128 likely arises as cells actively intrude into the matrix. Finally, systems with low values of 𝐴

129 correspond to matrices that mechanically relax very slowly, while systems with high values of

130 𝐴 correspond to matrices that relax very quickly. In our experiments and simulations, the ratio
𝜇𝑡 𝜏a
131 𝜇= ∈ [0.001 − 2], 𝜏a ∈ [7 − 54]s, while τm ∈ [1 − 350]s, so that the ratio 𝐴 = ∈ [0.1 − 100],
𝜇𝑚 𝜏m

132 and finally with spheroid sizes R ∼ 100 μm , and proliferative tissue timescale 𝜏t ∼ [4 − 500] 𝑠,
𝜏g
133 the ratio 𝑗 = ∈ [0.002 − 0.25 ].
𝜏t

134

135 We start our simulations within this framework with a spherical ball of cells that is loosely packed

136 within a viscoelastic matrix, and then allow the cells to divide and push each other into the matrix,

137 straining it. To determine when divisions are energetically favorable, we use a Metropolis-

138 Hastings algorithm44. Depending on the rheology of the matrix, this can either cause (i) the matrix

139 to break, flow and be remodeled even as tissue cells form finger-like protrusions, or (ii) the matrix

140 to respond purely elastically by straining, but not breaking, thus preventing the tissue cells from

141 further division and maintaining a spherical boundary with the matrix. Indeed, as we decrease the
𝐺′
142 relaxation time scale making the matrix behave more like a liquid (i.e. making 𝐴 = 𝜇 𝜏a large by
m

143 decreasing 𝜇m ) we see the appearance of an interfacial morphological instability (Fig. 2b-d and

144 Video S2), in accordance with findings of experiments (Fig. 1). Notably, instabilities were found

145 to occur both in simulations and experiments when tissue spheroid diameter was ~10a.

146 Additionally, when cell motility was reduced (by changing 𝑀), the model predicts that tissues

6
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

147 growing in matrices would be unable to grow, break symmetry or form branches (Fig. 2e and

148 Extended Data Fig. 4a,b).

149

150 To test these predictions, we first carried out experiments using matrices without cell adhesion

𝐺′
151 ligands, as cell adhesion and thus motility would be lost in this condition (𝐴 = 𝜇 𝜏a ~0). Tissues
m

152 were found to grow slowly, in a morphologically stable manner (Fig. 2f, Extended Data Fig. 4c

153 and Video S3). Next, potential mechanisms driving tissue motility and proliferation at the cellular

154 scale were explored. Cell motility can be regulated by: 1) cells pulling on the matrix via contractile

155 forces generated by acto-myosin interactions involving ROCK and Non-Muscle Myosin II, 2) cells

156 pushing on the matrix via protrusions created by Rac1 or Arp2/3 activity or 3) ion channel-

157 mediated changes45 (Fig. 2g). Only the inhibition of Rac1 or, the Rac1 pathway downstream

158 molecule, Arp2/3 by pharmacological inhibitors (NSC23766 and CK666, respectively) inhibited

159 tissue growth (Fig.2h and Extended Data Fig. 4d,e), in accordance with our model predictions.

160 This finding indicates that cells generate space for division and migration by pushing on the matrix.
𝜏g
161 Consistent with this, when the rate of cell proliferation in the model was inhibited (𝑗 = ~0)
𝜏t

162 simulations predicted tissue growth and instability would be dramatically diminished (Fig. 2i,

163 Extended Data Fig. 5a,b and Video S4). Experiments in which cell proliferation was inhibited

164 confirmed this prediction (Fig. 2j and Extended Data Fig. 5c). Further, the model predicts that for

165 cells in an elastic matrix, cell division would be spatially confined to the boundary between the

166 growing tissue and the substrate, but for cells in a viscoelastic matrix, the divisions would be more

167 broadly distributed throughout growing tissues (Fig. 2k and Extended Data Fig. 6). Experimental

168 analysis of the spatial distribution of proliferating cells in elastic versus viscoelastic matrices

169 confirmed these predictions as well (Fig. 2k and Extended Data Fig. 6). Altogether, these results

170 show that cell motility and proliferation, both of which are regulated by the viscoelasticity of the

171 matrix, control tissue spatio-temporal organization and morphogenesis.

7
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

172 After having considered the role of matrix viscoelasticity and cell proliferation on tissue

173 organization, we now turn to adapt our computational model to include the experimentally known

174 role that links an increase in matrix stiffness with an increase in cell motility46. We assume a

175 minimal model for this, via the relation 𝑀 ∝ 𝐺′ (Fig.3a). Simulations with this additional

176 assumption in the model predicted that tissue morphological instability would be enhanced with
𝐺′
177 an increase in the modulus of the matrix G’ in viscoelastic matrices (making 𝐴 = 𝜇 𝜏a large) , but
m

178 there would not be a significant impact in more elastic matrices (Fig.3b-d, Extended Data Fig.7

179 and Video S5). To validate these simulations experimentally, the previously developed matrices

180 were modified to change their modulus 𝐺′ (by changing crosslinking to yield

181 𝐺 ′ ~400,1700 and 500 Pa) and independently controlling the relaxation time (by changing the

182 molecular weight of alginate) and thus make the matrix more or less viscoelastic (Fig. 3e and
𝐺′
183 Extended Data Fig.8). In low viscosity matrices, i.e., large 𝐴 = 𝜇 𝜏a , the increase in the modulus
m

184 𝐺 ′ resulted in greater tissue growth and branching, as predicted (Fig.3f,g). To further determine

185 if these differential responses were again mediated by cell motility and proliferation, in silico

186 predictions of this model were compared to in-vitro studies performed under similar conditions.

187 As predicted by the model, inhibition of cell motility by inhibition of Rac1 and Arp2/3 complex led
𝐺′
188 to a greater impact on tissue growth in stiff matrices, i.e. large 𝐴 = 𝜇 𝜏a rather than soft
m

𝐺′
189 viscoelastic matrices, i.e. small 𝐴 = 𝜇 𝜏a (Fig. 3h,I, Extended Data Fig.9 and Video S6). Both
m

190 simulations and experiments revealed that cell division increased with stiffness both in elastic and

191 viscoelastic matrices although significantly more in viscoelastic matrices (Fig.3j,k and Extended

192 Data Fig.10). The significant increase in cell flux 𝑗 with modulus 𝐺′ in the viscoelastic matrices

𝐺′
193 ( 𝐴 = 𝜇 𝜏a is large) emerges from the increase in motility 𝑀46. When cell proliferation is inhibited,
m

194 the simulations show that tissues do not grow (Extended Data Fig.10 and Video S7).

8
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

𝜏g
195 Having studied the emergence of an active scaled cell flux from motility 𝑀 (with 𝑗 = ∼ 𝑂(1)),
𝜏t

𝜏g
196 we turn to passively inject a cell flux to the tissue (making 𝑗 = 𝜏t
≫ 1) to examine the role of

197 passive tissue pressure, known to regulate tissue growth12,41, on morphological stability. To

198 achieve this, we developed a microfluidic system where cells were injected at a constant rate into

199 the tissue, displacing the matrix (Extended Data Fig.11). We find that tissues break symmetry and

200 branch out into elastic matrices but are unable to break symmetry when the matrix is viscoelastic,

201 consistent with our simulations that show a similar response (Extended Data Fig.11, 12, 13c and

202 Video S8). The morphological instability occurring in this cell flux driven situation is similar to the

203 Saffman-Taylor instability in hydrodynamics and its elastic analog32,33, wherein a low viscosity

204 (low stiffness) material forms branches when driven into a high viscosity (high stiffness) material

205 in a confined geometry. Altogether, our simulations and experiments show that the tissue-matrix

206 interface becomes morphologically unstable when the matrix is viscoelastic and can easily relax

207 in response to stresses, or when the tissue proliferative pressure is high in more elastic matrices.

208 We summarize these results in a morphological phase diagram that quantifies the stability of the

209 growing front shown in Fig. 3I and Extended Data Fig.13.

210 Matrix viscoelasticity also regulates intestinal organoid patterning

211 To explore the generality of these findings as captured by the phase diagram in Fig. 3I, we decided

212 to explore the impact of matrix viscoelasticity in a synthetic context associated with the in-vitro

213 growth and development of self-organizing intestinal organoids. When Lgr5+ stem cells are

214 emplaced in a complex, laminin-rich extracellular matrix termed Matrigel, they develop into

215 complex three-dimensional structures containing all cell-types present in adult intestine, and

216 mimic intestinal tissue organization1,8. To allow for a comparison with the published literature, we

217 modified our alginate matrix system to enable incorporation of Matrigel (Fig. 4a), while still

218 allowing independent control over gel stiffness and viscoelasticity31. The interpenetrating

9
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

219 networks of two different stiffness (𝐺 ′ ~0.5kPa and 1.5kPa) allowed for both elastic and viscoelastic

220 matrices (Fig. 4b). As previously described17,47,48, organoids growing in elastic matrices exhibited

221 slow expansion and were morphologically stable. In contrast, intestinal organoids grew rapidly,

222 broke symmetry and formed branches when within viscoelastic matrices (Fig. 4c-e and Extended

223 Data Fig.14). Apart from demonstrating tissue morphological instability, organoids in viscoelastic

224 substrates exhibited cell patterning and differentiation representative of intestinal development

225 (Fig. 4f,g). Matrix viscoelasticity favored the generation of high curvature tissue regions that

226 concentrated Lgr5+ stem cells, consistent with past reports on the impact of curvature on

227 differentiation18. To determine if organoid spatio-temporal organization was regulated by internal

228 pressure generated inside organoid lumens, as previously reported with other systems14,15,49,

229 organoids were pharmacologically treated to impair the function of Na+/K+ ATPase pumps and

230 block fluid influx14,50. No significant differences in organoid morphology or patterning in

231 viscoelastic substrates were noted (Extended Data Fig. 15). To further test the ability of

232 viscoelasticity to control organ growth, organoid development was monitored in matrices of

233 varying stiffness. The percentage of Lgr5+ organoids and number of colonies were higher in

234 viscoelastic matrices rather than elastic matrices, independent of 𝐺′ (Fig.4h,i). This finding is

235 consistent with previous research as symmetry breaking and organoid development are

236 associated with a higher percentage of Lgr5+ organoids1. Increasing 𝐺 ′ of viscoelastic matrices

237 again led to greater growth of intestinal organoids, symmetry breaking and branch formation, but

238 organoids grew more slowly and maintained their spherical symmetry in elastic matrices (Fig.4j-

239 l).

240 Discussion

241 Our experiments and guiding simulations demonstrate that passive matrix viscoelasticity couples

242 to cell motility and cell proliferation to drive tissue growth, symmetry breaking and branching. The

243 resulting morphology is reminiscent of interfacial instabilities in passively driven physical systems

10
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

244 but modified fundamentally in living systems by the active processes of cell motility and cell

245 proliferation that can destabilize the interface and are relevant to a number of processes

246 including embryogenesis3,14, oncogenesis2,51 , branching morphogenesis6,7, and angiogenesis.

247 Our studies of two different systems: breast epithelia and intestinal organoids, show that the

248 properties of the viscoelastic extracellular matrix relative to that of the tissue, quantified in terms

249 of three experimentally-manipulatable dimensionless parameters, emerge as regulators of spatio-

250 temporal tissue organization.

251

252 More broadly, our results are consistent with prior observations that the increase in ECM fluidity

253 of the mesenchyme drives normal embryonic airway branching52, and an increase in tissue fluidity

254 drives wound healing53, tissue elongation54 or neural crest development55. Furthermore, invasive

255 branches are characterized by either an increase in matrix fluidity, as has recently been observed

256 in glioblastoma56,57, breast58 and liver cancer59 (compared to benign lesions and healthy ECM), or

257 an increase in tissue fluidity, as tumor single cells are less viscous60,61 and tumor tissues acquire

258 more liquid-like properties62-64 (e.g. EMT, unjamming). The increased expression of low molecular

259 weight hyaluronic acid in malignant tumors65 can explain the decrease in tumor ECM viscosity.

260 Our results also suggest that when tumors migrate and grow and push the stroma, this may lead

261 to the passive generation of stroma fingers in the healthy tissue, as the stroma has more liquid-

262 like properties than healthy tissue56-59.We can also rationalize previous apparently contradictory

263 findings that tissues maintained a stable morphology when encapsulated in synthetic materials66

264 of increasing stiffness, while becoming unstable in natural matrices as stiffness was raised2,67

265 (e.g. Matrigel, collagen, fibrin). From our perspective, the explanation is due to the elastic nature

266 of the synthetics that are covalently crosslinked, in contrast to the intrinsic viscoelasticity of

267 physically cross-linked natural matrices. Finally, in addition to providing a framework to

268 understand tissue morphology and organization in normal and pathological states, our study

11
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

269 yields a phase diagram that might help provide a strategy to guide tissue morphology in

270 regenerative medicine and related fields.

271

272

273

274 References:

275 1 Sato, T. et al. Single Lgr5 stem cells build crypt-villus structures in vitro without a
276 mesenchymal niche. Nature 459, 262-265, doi:10.1038/nature07935 (2009).
277 2 Paszek, M. J. et al. Tensional homeostasis and the malignant phenotype. Cancer cell 8,
278 241-254, doi:10.1016/j.ccr.2005.08.010 (2005).
279 3 Deglincerti, A. et al. Self-organization of the in vitro attached human embryo. Nature
280 533, 251-254, doi:10.1038/nature17948 (2016).
281 4 Moris, N. et al. An in vitro model of early anteroposterior organization during human
282 development. Nature 582, 410-415, doi:10.1038/s41586-020-2383-9 (2020).
283 5 Krndija, D. et al. Active cell migration is critical for steady-state epithelial turnover in the
284 gut. Science 365, 705-710, doi:10.1126/science.aau3429 (2019).
285 6 Hannezo, E. & Simons, B. D. Multiscale dynamics of branching morphogenesis. Current
286 opinion in cell biology 60, 99-105, doi:10.1016/j.ceb.2019.04.008 (2019).
287 7 Varner, V. D. & Nelson, C. M. Cellular and physical mechanisms of branching
288 morphogenesis. Development 141, 2750-2759, doi:10.1242/dev.104794 (2014).
289 8 Serra, D. et al. Self-organization and symmetry breaking in intestinal organoid
290 development. Nature 569, 66-72, doi:10.1038/s41586-019-1146-y (2019).
291 9 Hannezo, E. et al. A Unifying Theory of Branching Morphogenesis. Cell 171, 242-255
292 e227, doi:10.1016/j.cell.2017.08.026 (2017).
293 10 Briggs, J. A. et al. The dynamics of gene expression in vertebrate embryogenesis at
294 single-cell resolution. Science 360, doi:10.1126/science.aar5780 (2018).
295 11 Hannezo, E. & Heisenberg, C. P. Mechanochemical Feedback Loops in Development
296 and Disease. Cell 178, 12-25, doi:10.1016/j.cell.2019.05.052 (2019).
297 12 Green, J. B. & Sharpe, J. Positional information and reaction-diffusion: two big ideas in
298 developmental biology combine. Development 142, 1203-1211, doi:10.1242/dev.114991
299 (2015).
300 13 Kondo, S. & Miura, T. Reaction-diffusion model as a framework for understanding
301 biological pattern formation. Science 329, 1616-1620, doi:10.1126/science.1179047
302 (2010).
303 14 Chan, C. J. et al. Hydraulic control of mammalian embryo size and cell fate. Nature 571,
304 112-116, doi:10.1038/s41586-019-1309-x (2019).
305 15 Mosaliganti, K. R. et al. Size control of the inner ear via hydraulic feedback. Elife 8,
306 doi:10.7554/eLife.39596 (2019).
307 16 Nelson, C. M. et al. Microfluidic chest cavities reveal that transmural pressure controls
308 the rate of lung development. Development 144, 4328-4335, doi:10.1242/dev.154823
309 (2017).
310 17 Gjorevski, N. et al. Designer matrices for intestinal stem cell and organoid culture.
311 Nature 539, 560-564, doi:10.1038/nature20168 (2016).

12
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

312 18 Shyer, A. E., Huycke, T. R., Lee, C., Mahadevan, L. & Tabin, C. J. Bending gradients:
313 how the intestinal stem cell gets its home. Cell 161, 569-580,
314 doi:10.1016/j.cell.2015.03.041 (2015).
315 19 Harris, A. R. et al. Characterizing the mechanics of cultured cell monolayers.
316 Proceedings of the National Academy of Sciences of the United States of America 109,
317 16449-16454, doi:10.1073/pnas.1213301109 (2012).
318 20 Vining, K. H. & Mooney, D. J. Mechanical forces direct stem cell behaviour in
319 development and regeneration. Nature reviews. Molecular cell biology 18, 728-742,
320 doi:10.1038/nrm.2017.108 (2017).
321 21 Khalilgharibi, N., Fouchard, J., Recho, P., Charras, G. & Kabla, A. The dynamic
322 mechanical properties of cellularised aggregates. Current opinion in cell biology 42, 113-
323 120, doi:10.1016/j.ceb.2016.06.003 (2016).
324 22 Ricca, B. L., Venugopalan, G. & Fletcher, D. A. To pull or be pulled: parsing the multiple
325 modes of mechanotransduction. Current opinion in cell biology 25, 558-564,
326 doi:10.1016/j.ceb.2013.06.002 (2013).
327 23 Heisenberg, C. P. & Bellaiche, Y. Forces in tissue morphogenesis and patterning. Cell
328 153, 948-962, doi:10.1016/j.cell.2013.05.008 (2013).
329 24 Chaudhuri, O. et al. Hydrogels with tunable stress relaxation regulate stem cell fate and
330 activity. Nature materials 15, 326-334, doi:10.1038/nmat4489 (2016).
331 25 Debnath, J., Muthuswamy, S. K. & Brugge, J. S. Morphogenesis and oncogenesis of
332 MCF-10A mammary epithelial acini grown in three-dimensional basement membrane
333 cultures. Methods 30, 256-268 (2003).
334 26 Lee, K. Y. & Mooney, D. J. Alginate: properties and biomedical applications. Progress in
335 polymer science 37, 106-126, doi:10.1016/j.progpolymsci.2011.06.003 (2012).
336 27 Rowley, J. A., Madlambayan, G. & Mooney, D. J. Alginate hydrogels as synthetic
337 extracellular matrix materials. Biomaterials 20, 45-53 (1999).
338 28 Yang, J. et al. Guidelines and definitions for research on epithelial-mesenchymal
339 transition. Nature reviews. Molecular cell biology, doi:10.1038/s41580-020-0237-9
340 (2020).
341 29 Elosegui-Artola, A. et al. Mechanical regulation of a molecular clutch defines force
342 transmission and transduction in response to matrix rigidity. Nature cell biology 18, 540-
343 548, doi:10.1038/ncb3336 (2016).
344 30 Zanconato, F., Cordenonsi, M. & Piccolo, S. YAP/TAZ at the Roots of Cancer. Cancer
345 cell 29, 783-803, doi:10.1016/j.ccell.2016.05.005 (2016).
346 31 Chaudhuri, O. et al. Extracellular matrix stiffness and composition jointly regulate the
347 induction of malignant phenotypes in mammary epithelium. Nature materials 13, 970-
348 978, doi:10.1038/nmat4009 (2014).
349 32 Saffman, P. G. & Taylor, G. I. The penetration of a fluid into a porous medium or Hele-
350 Shaw cell containing a more viscous liquid. Proceedings of the Royal Society of London.
351 Series A. Mathematical and Physical Sciences 245, 312-329,
352 doi:doi:10.1098/rspa.1958.0085 (1958).
353 33 Biggins, J. S., Saintyves, B., Wei, Z., Bouchaud, E. & Mahadevan, L. Digital instability of
354 a confined elastic meniscus. Proceedings of the National Academy of Sciences of the
355 United States of America 110, 12545-12548, doi:10.1073/pnas.1302269110 (2013).
356 34 Montel, F. et al. Stress clamp experiments on multicellular tumor spheroids. Phys Rev
357 Lett 107, 188102, doi:10.1103/PhysRevLett.107.188102 (2011).
358 35 Bogdan, M. J. & Savin, T. Fingering instabilities in tissue invasion: an active fluid model.
359 Royal Society Open Science 5, 181579, doi:doi:10.1098/rsos.181579 (2018).
360 36 Risler, T. & Basan, M. Morphological instabilities of stratified epithelia: a mechanical
361 instability in tumour formation. New Journal of Physics 15, 065011, doi:10.1088/1367-
362 2630/15/6/065011 (2013).

13
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

363 37 Shaebani, M. R., Wysocki, A., Winkler, R. G., Gompper, G. & Rieger, H. Computational
364 models for active matter. Nature Reviews Physics 2, 181-199, doi:10.1038/s42254-020-
365 0152-1 (2020).
366 38 Marchetti, M. C. et al. Hydrodynamics of soft active matter. Reviews of Modern Physics
367 85, 1143-1189, doi:10.1103/RevModPhys.85.1143 (2013).
368 39 Mallory, S. A., Šarić, A., Valeriani, C. & Cacciuto, A. Anomalous thermomechanical
369 properties of a self-propelled colloidal fluid. Physical Review E 89, 052303,
370 doi:10.1103/PhysRevE.89.052303 (2014).
371 40 Berthier, L. & Kurchan, J. Non-equilibrium glass transitions in driven and active matter.
372 Nature Physics 9, 310-314, doi:10.1038/nphys2592 (2013).
373 41 Araujo, A. R., Gelens, L., Sheriff, R. S. & Santos, S. D. Positive Feedback Keeps
374 Duration of Mitosis Temporally Insulated from Upstream Cell-Cycle Events. Mol Cell 64,
375 362-375, doi:10.1016/j.molcel.2016.09.018 (2016).
376 42 Uroz, M. et al. Regulation of cell cycle progression by cell-cell and cell-matrix forces.
377 Nature cell biology 20, 646-654, doi:10.1038/s41556-018-0107-2 (2018).
378 43 Delarue, M. et al. Compressive Stress Inhibits Proliferation in Tumor Spheroids through
379 a Volume Limitation. Biophysical journal 107, 1821-1828,
380 doi:https://doi.org/10.1016/j.bpj.2014.08.031 (2014).
381 44 Hastings, W. K. Monte Carlo Sampling Methods Using Markov Chains and Their
382 Applications. Biometrika 57, 97-109, doi:10.2307/2334940 (1970).
383 45 Hung, W. C. et al. Confinement Sensing and Signal Optimization via Piezo1/PKA and
384 Myosin II Pathways. Cell Rep 15, 1430-1441, doi:10.1016/j.celrep.2016.04.035 (2016).
385 46 Ulrich, T. A., de Juan Pardo, E. M. & Kumar, S. The mechanical rigidity of the
386 extracellular matrix regulates the structure, motility, and proliferation of glioma cells.
387 Cancer research 69, 4167-4174, doi:10.1158/0008-5472.CAN-08-4859 (2009).
388 47 Hushka, E. A., Yavitt, F. M., Brown, T. E., Dempsey, P. J. & Anseth, K. S. Relaxation of
389 Extracellular Matrix Forces Directs Crypt Formation and Architecture in Intestinal
390 Organoids. Advanced Healthcare Materials 9, 1901214, doi:10.1002/adhm.201901214
391 (2020).
392 48 Cruz-Acuna, R. et al. Synthetic hydrogels for human intestinal organoid generation and
393 colonic wound repair. Nature cell biology 19, 1326-1335, doi:10.1038/ncb3632 (2017).
394 49 Dasgupta, S., Gupta, K., Zhang, Y., Viasnoff, V. & Prost, J. Physics of lumen growth.
395 Proceedings of the National Academy of Sciences of the United States of America 115,
396 E4751-E4757, doi:10.1073/pnas.1722154115 (2018).
397 50 Violette, M. I., Madan, P. & Watson, A. J. Na+/K+ -ATPase regulates tight junction
398 formation and function during mouse preimplantation development. Developmental
399 biology 289, 406-419, doi:10.1016/j.ydbio.2005.11.004 (2006).
400 51 Friedl, P., Locker, J., Sahai, E. & Segall, J. E. Classifying collective cancer cell invasion.
401 Nature cell biology 14, 777-783, doi:10.1038/ncb2548 (2012).
402 52 Spurlin, J. W. et al. Mesenchymal proteases and tissue fluidity remodel the extracellular
403 matrix during airway epithelial branching in the embryonic avian lung. Development 146,
404 doi:10.1242/dev.175257 (2019).
405 53 Tetley, R. J. et al. Tissue fluidity promotes epithelial wound healing. Nature Physics 15,
406 1195-1203, doi:10.1038/s41567-019-0618-1 (2019).
407 54 Mongera, A. et al. A fluid-to-solid jamming transition underlies vertebrate body axis
408 elongation. Nature 561, 401-405, doi:10.1038/s41586-018-0479-2 (2018).
409 55 Kuriyama, S. et al. In vivo collective cell migration requires an LPAR2-dependent
410 increase in tissue fluidity. The Journal of cell biology 206, 113-127,
411 doi:10.1083/jcb.201402093 (2014).

14
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

412 56 Streitberger, K. J. et al. High-resolution mechanical imaging of glioblastoma by

413 multifrequency magnetic resonance elastography. PloS one 9, e110588,
414 doi:10.1371/journal.pone.0110588 (2014).
415 57 Streitberger, K. J. et al. How tissue fluidity influences brain tumor progression.
416 Proceedings of the National Academy of Sciences of the United States of America 117,
417 128-134, doi:10.1073/pnas.1913511116 (2020).
418 58 Sinkus, R. et al. Viscoelastic shear properties of in vivo breast lesions measured by MR
419 elastography. Magnetic resonance imaging 23, 159-165, doi:10.1016/j.mri.2004.11.060
420 (2005).
421 59 Shahryari, M. et al. Tomoelastography distinguishes non-invasively between benign and
422 malignant liver lesions. Cancer research, canres.2150.2019, doi:10.1158/0008-
423 5472.Can-19-2150 (2019).
424 60 Nematbakhsh, Y., Pang, K. T. & Lim, C. T. Correlating the viscoelasticity of breast
425 cancer cells with their malignancy. Convergent Science Physical Oncology 3, 034003,
426 doi:10.1088/2057-1739/aa7ffb (2017).
427 61 Ketene, A. N., Schmelz, E. M., Roberts, P. C. & Agah, M. The effects of cancer
428 progression on the viscoelasticity of ovarian cell cytoskeleton structures. Nanomedicine:
429 Nanotechnology, Biology and Medicine 8, 93-102,
430 doi:https://doi.org/10.1016/j.nano.2011.05.012 (2012).
431 62 Palamidessi, A. et al. Unjamming overcomes kinetic and proliferation arrest in terminally
432 differentiated cells and promotes collective motility of carcinoma. Nature materials 18,
433 1252-1263, doi:10.1038/s41563-019-0425-1 (2019).
434 63 Han, Y. L. et al. Cell swelling, softening and invasion in a three-dimensional breast
435 cancer model. Nature Physics 16, 101-108, doi:10.1038/s41567-019-0680-8 (2020).
436 64 Staneva, R. et al. Cancer cells in the tumor core exhibit spatially coordinated migration
437 patterns. Journal of cell science 132, doi:10.1242/jcs.220277 (2019).
438 65 Monslow, J., Govindaraju, P. & Puré, E. Hyaluronan – A Functional and Structural Sweet
439 Spot in the Tissue Microenvironment. Frontiers in immunology 6,
440 doi:10.3389/fimmu.2015.00231 (2015).
441 66 Taubenberger, A. V. et al. 3D Microenvironment Stiffness Regulates Tumor Spheroid
442 Growth and Mechanics via p21 and ROCK. Advanced Biosystems 3, 1900128,
443 doi:10.1002/adbi.201900128 (2019).
444 67 Kumar, S. & Weaver, V. M. Mechanics, malignancy, and metastasis: the force journey of
445 a tumor cell. Cancer metastasis reviews 28, 113-127, doi:10.1007/s10555-008-9173-4
446 (2009).
447

15
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

448

449

450 Figure 1. Matrix viscoelasticity determines symmetry breaking, tissue branching, and
451 epithelial to mesenchymal transition

16
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

452 a, Schematic demonstrating how simultaneously changing the polymer molecular weight and
453 extent of crosslinking allows for constant gel stiffness but altered viscoelastic properties. b,
454 Quantification of the storage modulus of resulting alginate hydrogels (n=5,9 gels per condition).
455 Statistical analysis was performed using two-sided U Mann-Whitney test. c, Quantification of the
456 timescale at which an initially applied stress is relaxed to half its original value (n=19 gels per
457 condition). Statistical analysis was performed using Mann-Whitney U-test. d, Examples of growth
458 of MCF10A spheroids in elastic versus viscoelastic hydrogels over 5 days. Phalloidin in cyan,
459 Hoechst in magenta. e-f, Quantification of the spheroids area (e) and circularity (f), respectively
460 (error bars, s.e.m). n=19-43 spheroids/condition/day. Statistical analysis was performed using
461 Kruskal–Wallis test followed by post hoc Dunn’s test. g. Examples of vimentin, phalloidin and
462 hoechst stainings in spheroids growing in viscoelastic gels. Insets shows a spheroid branch. h.
463 Examples of phalloidin, Hoechst (left) and cytokeratin 14 (right) stainings in spheroids in
464 viscoelastic and elastic hydrogels. Phalloidin in cyan, Hoechst in magenta and cytokeratin 14 in
465 yellow. i, Quantification of average cytokeratin 14 intensity of the outer ring of spheroids. Elastic
466 spheroids average intensity is normalized to 1. n=12,15 spheroids per condition. Statistical
467 analysis was performed using Mann-Whitney U-test. j. Quantification of the tumor volume in mice
468 injected in day 0 with viscoelastic and elastic hydrogels containing MDA-MB231 breast epithelial
469 cells (error bars, s.e.m). k, Representative examples of phosphorylated FAK, phalloidin and
470 Hoechst stainings in MCF10A celll spheroids growing in elastic and viscoelastic gels. pFAK in
471 yellow, phalloidin in cyan and hoechst in magenta. h. Representative examples of phalloidin,
472 Hoechst (upper row) and YAP (lower row) stainings of spheroids in elastic and viscoelastic gels
473 (spheroids core cells and branch leader cells). i. Quantification from stainings of the percentage
474 of cells with nuclear YAP per image for the indicated regions (n=8,11,17 images per condition).
475 All data are mean ± s.d. except where indicated, all scale bars are 75 µm.
476

17
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

477

478

479 Figure 2: 3D theoretical model predicts that spheroids-material physical interaction

480 regulates tissue geometrical evolution

481 a, Schematic depicting the theoretical physical model of tissue growth in a passive viscoelastic
482 matrix. The viscosity of the tissue, viscosity of the matrix, and the elasticity of the matrix can be

18
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

483 tuned independently. b, Examples of simulated tissue growth in elastic matrices (top row) as
484 versus viscoelastic matrices (lower row). c-d, Quantification from the simulations of the projected
485 area and circularity of the spheroids, respectively, over time. e, Model prediction with inhibition of
486 cell motility. f, Representative experimental examples (upper row) and quantification of spheroid’s
487 area (lower row) in hydrogels after 5 days in gels with and without cell adhesive ligand RGD.
488 n=52,52,51,54 spheroids per condition. Statistical analysis was performed using Kruskal–Wallis
489 test followed by post hoc Dunn’s test. g, Schematic showing the inhibitors used to affect cell
490 motility:1) Blebbistatin and Y27632 affect actomyosin cytoskeleton by affecting non-muscle
491 myosin II and ROCK, respectively; 2) Cell protrusion is affected by NSC23766 and CK666 that
492 affect Rac1 and Arp2/3 complex, respectively; and 3) gadolinium affects ion channels. h,
493 Quantification of spheroid area in hydrogels after 5 days in the presence of the indicated inhibitors.
494 n=52,50,51,51,51,50,51,50,51,46,41,51,21,21,24,20,21,25 spheroids per condition. Statistical
495 analysis was performed using Kruskal–Wallis test followed by post hoc Dunn’s test. i, Model
496 predictions with tissue growth inhibition. j, Representative experimental examples and
497 quantification of the spheroid’s area without or with the presence of thymidine to inhibit cell
498 proliferation. n=52,53,51,53 spheroids per condition. Statistical analysis was performed using
499 Kruskal–Wallis test followed by post hoc Dunn’s test. k, Model predictions and experimental
500 results for the numbers and distributions of proliferating cells across spheroids in elastic (upper
501 row) and viscoelastic gels (lower row): left, model predictions of localization of cell division (cyan)
502 from a section of a spheroid; center, representative examples of experimental spheroids showing
503 EdU positive cells (cyan) and cell nuclei (Hoechst, magenta) for spheroids in elastic and
504 viscoelastic gels; right, colormaps of experimental image (center) showing the local percentage
505 of EdU positive cells across the spheroid. n=3,4 spheroids per condition. All data are mean ± s.d.,
506 all scale bars are 200 µm.

19
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

507

20
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

508 Figure 3. Stiffness intersects with matrix viscoelasticity to regulate growth and branching.

509 a, To incorporate the matrix stiffness dependence on the tissue property, now the active motility
510 of the tissue is an increasing function of the matrix stiffness. Which makes the active motility a
511 dependent parameter and in turn it also affects the tissue growth. b-d, 3D final timepoint
512 simulation images (b), projected area (c) and circularity (d) evolution over time of spheroids in
513 increasingly stiff elastic and viscoelastic gels. e, Stiffness of experimental matrices was modified
514 by further altering the extent of crosslinking in both elastic and viscoelastic gels. f, Representative
515 experimental examples (upper row) and quantification of spheroid area (lower row) after 5 days
516 in elastic and viscoelastic matrices of increasing stiffness. n=63,55,84,50,55,50 spheroids per
517 condition. Statistical analysis was performed using Kruskal–Wallis test followed by post hoc
518 Dunn’s test. g, Quantification of spheroid circularity after 5 days in elastic and viscoelastic
519 matrices of increasing stiffness. n=63,55,84,50,55,50 spheroids per condition. Statistical analysis
520 was performed using Kruskal–Wallis test followed by post hoc Dunn’s test. h, Representative
521 model simulation results when cell motility is eliminated in stiff viscoelastic matrices compared to
522 soft viscoelastic matrices. i, Representative experimental examples (upper row) and
523 quantification of spheroid’s area (lower row) after 5 days in soft and stiff viscoelastic matrices with
524 Rac1 (NSC23766) and Arp2/3 (CK666) inhibitors. n=25,22,27,21,24,21,21,24 spheroids per
525 condition. Statistical analysis was performed using Kruskal–Wallis test followed by post hoc
526 Dunn’s test. j, Model predictions for cell proliferation in spheroids of increasing stiffness for both
527 elastic and viscoelastic gels. k, Representative experimental examples (upper row) and
528 quantification of the percentage of EdU positive cells in a spheroid (lower row) after 5 days in
529 elastic and viscoelastic gels of increasing stiffness. n=32,30,28,33,31,33 spheroids per condition.
530 Statistical analysis was performed using Kruskal–Wallis test followed by post hoc Dunn’s test. All
531 data are mean ± s.d., all scale bars are 200 µm. l, phase diagram. Simulations predict, and
532 experiments confirm that regions of tissue growth stability and instability can be predicted based
𝜏g
533 on the values of three dimensionless variables. When the scaled proliferation pressure 𝑗 = 𝜏t
≪
534 1, the tissue grows as a stable spheroid (Fig. 2i,j and Extended Data Fig. 9, 10, 13b). Additionally,
𝜏
535 when the scaled matrix relaxation time 𝐴 = 𝜏 a ≪ 1, the tissue remains spheroidal and is
m
𝜏g
536 morphologically stable as long as the scaled proliferation pressure 𝑗 = ∼ 𝑂(1) (top panel of
𝜏t
𝜏
537 Fig.1d and Fig 2b). When the scaled matrix relaxation time 𝐴 = 𝜏 a ≫ 1: if the scaled proliferation
m
𝜏g
538 pressure 𝑗 = 𝜏t
≪ 1, the tissue grows as a stable spheroid (bottom right of Fig. 2i and bottom
𝜏g
539 panel of Extended Data Fig. 11b); if the scaled proliferation pressure 𝑗 = 𝜏t
∼ 𝑂(1), the growth is
540 unstable and the tissue breaks symmetry and develops branches (bottom panel of Fig.1d and
𝜏g
541 bottom panel of Fig. 2b and 3b); if the scaled proliferation pressure 𝑗 = 𝜏t
≫ 1, the morphological
𝜇 𝜇
542 stability of the tissue depends on µ = 𝜇 t (see Extended Data Fig11d,e and 13c); for 𝜇 = 𝜇 t ≪ 1,
m m
𝜇
543 the tissue remains spheroidal (Extended Data Fig.11d,e, 13c); for µ = 𝜇 𝑡 ≫ 1, growth is unstable
𝑚
544 and the tissue breaks symmetry and develops branches (Extended Data Fig.11d,e, 13c). We
545 have shown representative images from the experiments and the simulations in different regimes
546 of the Phase diagram; one set of images from stable tissues in the blue region

21
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

𝜏a 𝜇t 𝜏g
547 (𝐴 = = 0.4, 𝜇 = = 0.002, 𝑗 = = 0.05); top left is first set of unstable images from a specific
𝜏m 𝜇m 𝜏t
𝜏 𝜇 𝜏g
548 point (𝐴 = 𝜏 a = 400, 𝜇 = 𝜇 t = 2, 𝑗 = 𝜏t
= 0.22); and top right is second set of images of another
m m
𝜏 𝜇 𝜏g
549 unstable point (𝐴 = 𝜏 a = 3.3, 𝜇 = 𝜇 t = 2, 𝑗 = 𝜏t
= 0.14). Scale bars are 200 µm. p<0.05 *; p<0.01
m m
550 **; p<0.001 ***.

22
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

551

552

23
 bioRxiv preprint doi: https://doi.org/10.1101/2022.01.19.476771; this version posted January 20, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

553 Figure 4: Matrix viscoelasticity controls intestinal organoid growth, symmetry breaking,
554 budding and cell patterning.

555 a, Schematic depicting of interpenetrating networks (IPNs) of alginate and Matrigel used in
556 organoid studies. Viscoelasticity is controlled by polymer molecular weight and crosslinker
557 concentration, while the concentration of Matrigel is maintained constant. b, Storage moduli of
558 the elastic and viscoelastic alginate-matrigel IPNs. n=6 gels per condition. Statistical analysis
559 was performed using Mann-Whitney U-test. c, Representative examples of phalloidin and
560 hoechst stainings of intestinal organoids in elastic and viscoelastic hydrogels over 7 days of
561 culture. Phalloidin in cyan, Hoechst in magenta. d-e, Quantification of the organoids area (d)
562 and circularity (e), respectively, over 7 days in elastic and viscoelastic matrices (error bars,
563 s.e.m). n= 24/26,2/24,27/22,31/21,19/23,22/29,21/26 organoids in Elastic/Viscoelastic gels per
564 day. Statistical analysis was performed using Kruskal–Wallis test followed by post hoc Dunn’s
565 test. f, Example of Lgr5+, phalloidin and hoechst staining of intestinal organoids in a stiff
566 viscoelastic gel after 7 days. Left, Lgr5+ (magenta) and hoechst (cyan); right, phalloidin (cyan).
567 g, Example of Lysozyme, phalloidin and hoechst staining of intestinal organoids in a stiff
568 viscoelastic gel after 7 days. Left, lysozyme (magenta) and hoechst (cyan); right, phalloidin
569 (cyan). h, Representative examples of phase contrast and Lgr5+ GFP images (upper row) and
570 quantification of GFP positive Lgr5+ intestinal organoids in the viscoelastic and elastic matrices
571 of different stiffness. n=5,5,5,6 samples per condition. Statistical analysis was performed using
572 Kruskal–Wallis test followed by post hoc Dunn’s test. i, Quantification of the percentage of
573 colony formation per condition. n=20,20,18,24 images per condition. Statistical analysis was
574 performed using Kruskal–Wallis test followed by post hoc Dunn’s test. j, Examples of phalloidin
575 and Hoechst stainings of intestinal organoids in different stiffness elastic and viscoelastic
576 matrices after 7 days. k, Quantification of the organoids area in different stiffness elastic and
577 viscoelastic matrices. n=32,32,38,37 organoids per condition. Statistical analysis was performed
578 using Kruskal–Wallis test followed by post hoc Dunn’s test. m, Example of EdU (cyan) and
579 Hoechst (nuclei) (upper row) and the percentage of EdU positive cells (lower row) of intestinal
580 organoids in different stiffness elastic and viscoelastic matrices. n=10,9,8,8 organoids per
581 condition. Statistical analysis was performed using Kruskal–Wallis test followed by post hoc
582 Dunn’s test. All data are mean ± s.d. except where indicated, all scale bars are 100 µm.

24