A novel model system to address the relevance of aggregation in animal origins

15 The origin of animals from unicellular ancestors remains a fundamental biological question. Cell 16 aggregation, a widespread eukaryotic behaviour, has been underappreciated as a potential pathway 17 to multicellularity. Here, we establish Capsaspora owczarzaki, a close unicellular relative of animals, 18 as a model system to investigate this process. We demonstrate C. owzarzaki aggregates dynamically 19 deploying key metazoan-related genes such as integrins, tyrosine kinases, and Hippo pathway 20 components. Moreover, we further model mathematically the aggregation process, revealing a 21 threshold-like adhesion response to fetal bovine serum (FBS) and dynamic aggregation kinetics 22 driven by cell-cell adhesion and access to FBS. Our findings suggest that cell aggregation could 23 have played a pivotal role in the evolution of animal multicellularity, providing a context for the 24 origin of genes now crucial for animal development. This work positions Capsaspora as a powerful 25 model system for quantitatively studying the evolutionary transition to multicellularity through cell 26 aggregation. 27 28 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 1 Introduction 29 How animals evolved from their unicellular ancestor remains a major biological question. Theoretically, 30 there are three potential paths in which the unicellular ancestor could have given rise to the first animal: 31 through the formation of colonies formed by clonal division; through the formation of multi-nucleate 32 entities, or through facultative aggregates formed by cell aggregation 1. Given the presence of those three 33 cell behaviours in different close unicellular relatives of animals, the jury is still out. 34 Of those three potential paths to complex multicellularity, aggregation has classically been neglected due 35 to the potential negative consequences of the emergence of cheaters 2 . However, cell aggregation is 36 indeed a powerful way to become multicellular. First, cell aggregation is widespread in eukaryotes 37 (including animals); second, it is a key process for development and homeostasis in many animal tissues; 38 third, it is a common response to stressful conditions; and finally, it presents important ecological 39 advantages (such as fast response to predation avoidance, improved extracellular metabolism) 3 4 5. 40 Notably, in aggregates, cells can have more free movement than in clonal colonies or multinucleate 41 entities 6 71 42 To discern the potential of cell aggregation as a pathway towards complex multicellularity, two factors 43 need to converge. First, we need to better understand how aggregation behaves in an extant relative of 44 animals; in particular to discern whether gene expression is dynamically controlled and if so, which genes 45 in particular are dynamically deployed. For example, if genes that are relevant to animal multicellularity 46 are shown to be dynamically deployed during the aggregation, that will be an indication that cell 47 aggregation could have had a role in the origin of those “multicellular genes” that later on were co-opted 48 to work within a multicellular system. 49 Second, we need to establish a model system in which to experimentally interrogate the potential of cell 50 aggregation to evolve, for example, division of labour, 3D structures, or to become obligate multicellular, 51 as well as to analyse the role of different genes in the formation and dynamics of such aggregates. Such a 52 model system should ideally be established in an organism that shares many “multicellular” genes with 53 animals and in which cell aggregation can be easily induced. Finally, to be able to use this system to 54 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint directly interrogate and quantitatively assess how external or internal cues affect the aggregate formation, 55 a mathematical model should also be established. 56 We here provide the first model system to explore how cell aggregation may have contributed to the 57 origin of animals. We use the filasterean Capsaspora owczarzaki (from here on Capsaspora, Fig. 1), one 58 of the closest known relatives of animals that likely shares the largest set of “multicellular” genes with 59 animals, including, among others, a complete integrin adhesome, a Hippo pathway, several protein 60 tyrosine kinases, and various developmental transcription factorsc(TFs) 8. Notably, Capsaspora 61 aggregates can be easily induced 9. Another key advantage is that Capsaspora cultures are axenic, which 62 ensures that microbial interactions do not confound cell behaviour, thereby allowing researchers to isolate 63 the effects of cell aggregation from other biological factors. 64 65 Figure 1 | a, Phylogenetic relationship of the unicellular closest relatives of animals based on (10,11). The 66 filasterean Capsaspora owczarzaki is among the best studied relatives to animals, and it shares with 67 animals many genes that were thought to be animal-specific 8 b, Capsaspora presents three different life 68 stages under culture conditions12. Filopodial stage in which single amoeboid cells adhere to the substrate. 69 Under maintained stressful conditions cells can enter a cystic stage losing their filopodia and forming 70 resistant structures, this stage is reversible upon recovering optimal conditions. Capsaspora cells can also 71 form a multicellular stage by aggregation, in which cells form larger clusters through filopodial 72 interconnections and formation of an extracellular matrix, exhibiting a high plasticity regarding their size. 73 74 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Our morphological and genetic characterization of cell aggregation and disaggregation in Capsaspora 75 demonstrates how genes involved in multicellular functions are dynamically expressed throughout these 76 processes. This suggests the ancestral origin of these now-crucial animal genes may be tied to cell 77 aggregation behaviour. Moreover, we have constructed a mathematical model that describes the kinetics 78 of the aggregation process and that shows that differential cell-cell adhesion and compactness are driving 79 its dynamics. Overall, our results highlight the central role of cell aggregation in discussions of animal 80 origins and position Capsaspora as a powerful new model system for quantitatively evaluating how 81 aggregation may have paved the way for animal multicellularity. 82 2 Results 83 2.1 The aggregative behaviour of Capsaspora owczarzaki is highly reproducible, with sequential stages 84 that can be identified throughout its temporal dynamics. 85 The formation of Capsaspora aggregates has been previously described, including the chemical cues 86 necessary for inducing the aggregation process 9. There is also an RNA-seq analysis of the different life 87 stages of Capsaspora (amoeboid, cystic, and aggregative), that shows different transcriptomic profiles for 88 each life stage 12 . However, those analyses showed only a specific moment of aggregation. If we want to 89 understand the potential role of aggregation in animal origins, we need a systematic approach to the 90 overall dynamics of cell aggregation to unravel which genes are dynamically employed. We here 91 specifically examined whether Capsaspora’s multicellular aggregative behaviour is deploying 92 “multicellular genes” that later on were key for animal multicellularity. To this end, it was crucial to first 93 identify key intermediate stages of the aggregation process. 94 Microscopical observation allowed us to define eight time points of the aggregation process in which a 95 clear change in aggregation dynamics was observed and reproducible (Movie 1) (Figure 2). 96 We start from a confluent culture of single cells seeded in media without fetal bovine serum (FBS) (T0). 97 At time point 1 (T1, 3min after aggregation induction), cells detach from the surface and begin to get 98 close to one another, forming aggregates in groups of ~5 cells. At time point 2 (T2, 12h), smaller 99 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint aggregates of approximately 20-30 cells are formed. By time point 3 (T3, 24h), growth of the aggregates 100 is mainly generated through cell division reaching hundreds of cells; this increase is represented by a 101 smooth increase in the maximum area of the aggregates shown in the graph. At time point 4 (T4, 36h), 102 merging between aggregates becomes more frequent, leading to the formation of macroscopic aggregates. 103 After time point 5 (T5, 48h), the aggregates are reaching their maximum size, and by time point 6 (T6, 104 60h), dissociation signal are observable from their centre. Finally, at time point 7 (T7, 72h), the 105 aggregates revert to single cells. 106 107 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 108 109 Figure 2 | Comparison of Capsaspora aggregation dynamics between experimental results and model simulations under 5% FBS. a, Experimental Capsaspora aggregates growth dynamics comparison with different FBS concentrations. (Left panel) The aggregation dynamics of two replicates at 5% FBS (R1 and R2) shown with aggregation dynamics predicted through simulations at 5% FBS (RMod). (Right panel) The dynamics of aggregation at 5% and 10% FBS are compared. The growth and formation of larger aggregates occurs through two mechanisms, the first involves cell proliferation within aggregates, reflected by a continuous and gradual increase in aggregate area. The second occurs through the clustering of pre-existing aggregates, resulting in a sharp increases in the aggregate area. b, Experimental images(top) vs Model snapshots (bottom) comparison, numerical simulation at 1, 12, 24 and 48 hours after administration of 5% FBS. Both experimental and simulated images cover a 1080 x 1080 μm domain. In the simulations, cell colours represent the local FBS concentration, which is also reflected in the background of the medium. The colour bar on the right indicates FBS levels: initial 5% FBS appears grey, while depletion due to metabolic activity of Capsaspora shifts colours to blue, with dark blue indicating no FBS. Numerical simulations were performed with sigmoidal Hill response of cell adhesion strength to FBS levels (see Supplementary Information) and all parameter values are as in Extended Data Table 1. .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 2.2 Aggregation via FBS-upregulated adhesion 110 Experimental studies have shown that FBS components, such as lipoproteins and calcium ions, induce 111 reversible aggregation in Capsaspora. When FBS is abundant, Capsaspora metabolises its components, 112 forming aggregates that merge into larger clusters over time. Once FBS is depleted, aggregates 113 disassemble, and cells disperse unless additional FBS is supplied 9. Motivated by these observations, we 114 developed a mathematical model to capture this chemically induced aggregation process (see Extended 115 Data Figure 1). 116 Our model integrates individual cell behaviour with local FBS concentration, which decreases 117 over time due to consumption by cellular metabolism. Cells move within the domain, responding to 118 chemical cues and interactions with neighbouring cells. Initially, FBS is uniformly distributed, but as cells 119 consume it at a constant rate, the concentration declines. At high FBS levels, cells aggregate. As FBS is 120 depleted and its concentration decreases, aggregates break apart–mirroring experimental findings. 121 Parameters related to cell size, motility, cell proliferation, and FBS consumption were inferred from 122 experiments. Since calcium ions, key FBS components required for Capsaspora aggregation, are known 123 to stabilise cell-cell adhesion 9 , we modelled FBS-dependent aggregation as an increase in adhesion 124 strength. Specifically, our framework includes short-range repulsion to prevent cell overlap and longer-125 range FBS-dependent adhesion. Cells can interact and form adhesion sites via filopodia at a certain 126 distance from each other, with adhesion becoming more pronounced in the presence of FBS. This 127 assumption is supported by experimental observations showing FBS components within the cell body and 128 along filopodia 9. 129 Our model successfully reproduces all stages of FBS-induced aggregation. Within minutes of exposure to 130 5% FBS, cells form small clusters of 2-10 cells. Over the next 20-25 hours, aggregates grow through cell 131 proliferation and merging during random migration. However, as aggregates enlarge, their movement 132 becomes restricted due to stronger adhesion forces. Larger aggregates (hundreds of cells) primarily grow 133 via proliferation rather than migration and only fuse when in close proximity. A key consequence of this 134 behaviour is localised FBS depletion, which accelerates as aggregates grow (Figure 2b, right panels and 135 Extended Data Figure 2b). The larger the aggregate, the faster it consumes FBS. Once depletion reaches a 136 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint critical threshold, cells begin to disaggregate around 50-60 hours in experiments with 5% initial FBS. By 137 70 hours, cells are fully dispersed, indicating global FBS exhaustion. These processes are visualised in 138 Movies 2-3. 139 Although individual realisations of our model vary slightly due to its stochasticity, and we do not 140 account for medium flux effects on single-cell trajectories (see Supplementary Information), our model 141 effectively captures Capsaspora aggregation dynamics. These results support the hypothesis that FBS 142 stabilises cell-cell adhesion, driving multicellular aggregate formation. A key advantage of our model is 143 its ability to visualise the FBS field in real time and explore different functional forms of FBS-induced 144 adhesion, refining predictions beyond the assumption that FBS enhances adhesion strength. 145 2.3 Threshold-like response to FBS 146 To better understand how FBS influences cell adhesion, we performed numerical simulations across a 147 range of initial FBS concentrations. Experimental findings from a previous study 9 indicate that at 148 concentrations below 1% FBS, Capsaspora cells fail to form multicellular clusters, instead exhibiting 149 random migration without a discernible pattern. However, when the FBS concentration reaches 1% or 150 higher, aggregate size increases sharply, as reflected in the mean aggregate size dynamics previously 151 reported 9. This sharp transition suggests that FBS-induced aggregation in Capsaspora follows a 152 threshold-like behaviour: adhesion remains minimal below 1% FBS, but once this threshold is exceeded, 153 adhesion strength rises rapidly before reaching saturation at higher concentrations. 154 To capture this behaviour, we modelled cell-cell adhesion strength as a Hill function of the FBS 155 concentration, whose sigmoidal shape reflects three key features: weak response at low FBS levels, a 156 sharp increase in adhesion strength once the threshold is surpassed, and an eventual plateau at higher 157 concentrations (Figure 3d). In addition to the Hill function, we tested two alternative adhesion models 158 with linear responses to FBS. The first linear model assumed a gradual, moderate increase in adhesion 159 strength with FBS concentration, while the second featured a steeper slope, producing a more abrupt 160 adhesion increase (Figure 3d). 161 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 162 Figure 3 | Comparison of different cell adhesion responses under varying initial FBS concentrations. a, We tested three functional forms for FBS-dependent adhesion: a, a sigmoidal Hill response, b, a shallow linear increase in adhesion strength, and c, a steep linear increase in adhesion strength whose functional forms are sketched in d. The initial FBS concentration, uniformly distributed at t=0 h, is indicated above each column, ranging from 0.5% (left) to 10% (right). To facilitate comparison, panel e presents experimental images from cultures with an initial 5% FBS concentration (scale bar: 100 μm). f, Figure legend for panels a-c. Each plot is divided diagonally by a white line: the upper half shows a snapshot at 12 h, while the lower half displays the system at 36 h. The colour of cells and the surrounding medium in simulations represents local FBS concentration, with red indicating high levels, grey intermediate levels, and blue low levels. Panels outlined in magenta highlight conditions that deviate from experimental observations and thus fail to reproduce the reported aggregation response to FBS in Capsaspora. Each image represents a single simulation, but we conducted five numerical simulations per condition, all yielding consistent results. The domain size for these simulations is 360 x 360 μm, with parameter values detailed in Extended Data Table 1. Full animations of plots shown in panels a-c are available in Movies 4-6, respectively. .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Interestingly, and as shown in Figure 3a, the threshold-like adhesion response closely replicates 163 experimentally observed Capsaspora aggregation dynamics. Specifically, at 0.5% FBS, no aggregates 164 form; at 1%, small clusters emerge gradually over 36 h; and at 5% and 10%, aggregation occurs rapidly, 165 consistent with experimental results. In contrast, the moderate linear response fails to capture the 166 observed aggregation speed at 5% FBS (compare Figures 3b and e). In the model adhesion remains too 167 weak at 12 h to induce aggregation, whereas experimental observations show aggregate formation within 168 minutes. The steep linear response reproduces aggregation at 5% and 10% FBS, but it also predicts 169 aggregation at 0.5%, which wasn’t previously observed 9 . Overall, these results support the hypothesis 170 that Capsaspora adhesion to FBS follows a threshold-like response, where adhesion remains low below a 171 critical concentration but increases sharply once this threshold is exceeded. To further validate this 172 response, we quantified key metrics (area, cell number, and density) of the largest aggregate throughout 173 our simulations. These measurements reinforce the conclusion that Capsaspora adhesion to FBS follows 174 a threshold-like pattern (see Extended Data Figure 3). 175 Beyond influencing aggregation, the nonlinear upregulation of adhesion in response to FBS is also 176 crucial to reproduce the aggregate disassembly dynamics. Notably, disassembly in large aggregates 177 begins at the core of the aggregate, leading to an inside-out fragmentation pattern. This process occurs 178 because aggregates develop with higher cell density at the centre, where adhesion and crowding effects 179 are strongest. Consequently, FBS depletion occurs more rapidly in the core due to higher local cell 180 consumption, while outer regions retain FBS longer, delaying disassembly. To visualise this process, we 181 quantified local cell density in our simulations, colouring cells accordingly throughout the numerical 182 experiment, and the results show disassembling at around 55 h, with the denser outer annulus persisting 183 until approximately 70 h post-FBS induction (Movie 7). 184 2.4 Temporal gene expression profiles reflect the dynamism of the aggregation process in Capsaspora 185 To explore the expression of Capsaspora genome during aggregation, we first performed a clustering 186 analysis of the expression profiles at different time points (Extended Data Figure 4). Principal component 187 analysis (PCA) distinguishes two clear groups in the X axis, the first one formed by T0 (immediately 188 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint before aggregation) & T1(3 min) clustering together representing adherent samples, and a second group 189 including all the different stages of the aggregate formation after aggregation induction. The Y axis shows 190 the clustering in order from T2 (12h) to T7 (72h) representing the aggregate growth. Based on sample 191 distances, we differentiated four clusters, the first one formed by all the replicates from T2, T3 (24h) & 192 T4 (36h) which cluster together representing the aggregative stage of Capsaspora. t0 represents the cell 193 before inducing aggregation, also called “adherent stage”. T1 forms a cluster on its own which might be 194 indicative of a specific response to the induction of aggregation. T5 (48h), T6 (60h) & T7 cluster as 195 another group in between aggregative and adherent cells, which represents the disaggregation step in 196 which we can find aggregates together with single cells in the culture. These results reveal that for 5% 197 FBS, T5 (48h) should be considered part of the disaggregation step, situating the disaggregation process 198 even earlier than what we previously anticipated by microscopy observations (55-60h from time lapses), 199 and reveals that Capsaspora cells have an underlying regulatory mechanism to detect FBS depletion 200 before the effects are observable in the culture. 201 Additionally the DESeq2 analysis reveals that out of the 8352 genes analysed 922 genes are significantly 202 upregulated at some point of the aggregation process and 990 genes are significantly downregulated, 203 corresponding to an 11% of the genome in both cases. Interestingly, T1 (3min after inducing aggregation) 204 shows the higher number of up and down regulated genes (5% in both cases), congruent with the fast 205 response of Capsaspora to the addition of FBS. 206 To identify and extract gene expression clusters we used DEGpatterns 33 and observed 52 groups of 207 genes, the cumulative distribution function (CDF) of the consensus index indicates that 6 clusters of 208 different expression patterns are the minimum to capture all the variation in gene expression (Extended 209 Data Figure 5). Thus we grouped the 52 patterns in 6 super groups based on hierarchical clustering 210 (Figure 4) and performed a GO term enrichment analysis to determine the main functions associated with 211 each expression profile. 212 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 213 214 Figure 4 | Gene expression dynamics are captured in six super groups. Which are: Group 1, Decreasing pattern in which genes keep decreasing their expression values after 12h. In this group we find genes related to cell growth and division, mainly the regulation of mitosis and the ribosome biogenesis. The Fluctuating Decreasing group (Group 2) shows genes whose expression levels decrease at 12h followed by an increase in expression and a new decrease after 24 h. This group includes GO terms related to the cell cycle process, the regulation of actin polymerization, as well as GO terms related to multicellular functions such as the positive regulation of Notch pathway, system development, and morphogenesis. The middle-peak group (Group 3) reflects GO terms that are especially relevant in the maintenance of the already formed aggregates, such as post translational modifications and splicing. The middle-dip group (Group 4) includes genes whose levels of expressions are the lowest in the aggregates that are fully formed. Here, we find GO terms related to sterol metabolism, gluconeogenesis and ubiquitinization processes. The Increasing group (Group 5) shows a smooth increase in the expression levels across all time points after 12h, and includes GO terms related to Development and signalling, cell fate determination, migration, proliferation, the regulation of stress (oxidative stress HIF1, and osmotic), and the negative regulation of cell population signalling. The Fluctuating-increasing group (Group 6) depicts an increasing trend after 12h of aggregation, and includes GO terms related to development & growth, the positive regulation of multicellular organs formation, cellular localization, actin cytoskeletal organization, and the negative regulation of the hippo pathway. .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 2.5 Main functions underlying the aggregation process in Capsaspora. 215 In addition to the analysis of overall temporal gene expression patterns mentioned above, we also 216 performed differential gene expression analysis between consecutive time points to identify genes 217 responsible for the transitions between each stage. A striking feature is the fast response of Capsaspora to 218 the addition of FBS and induction of aggregate formation. In the first 3 mins we already observed the 219 highest number of genes being up and down regulated, in which the cell rapidly favours cell growth and 220 proliferation, with an increase in GO terms related to ribosome biogenesis and growth, while genes 221 related to transcription are downregulated. This first response is followed by a period of formation of 222 small size aggregates during the first 12h in which we observed an increase in GOs related to 223 development and regulation of cell cycle and cell-cell signalling, with a reduction of genes related to 224 ribosome and amino acid metabolisms (Figure 5 and Extended Data Figure 6). The period that 225 corresponds to the 12 to 24h in which aggregates have reached larger sizes, we find genes related to lipid, 226 glucose and amino acid metabolisms, which suggest functions related to growth and metabolisms within 227 aggregates. Between 24 and 36 hours the aggregates keep merging with each other whenever they enter in 228 contact one to another. In this period, we also observed down regulation of genes related to actin-based 229 processes and cytoskeletal organization, which could be related to the depletion of FBS in the media (as 230 also predicted by our model simulations; see Movies 2-3), and the initial steps of the disaggregation 231 signal. The period between 48-60h we find upregulation of GO terms related to Golgi vesicle transport 232 and the down regulation of GO terms related to ribosome biogenesis (which is highly energy demanding 233 for the cell), amino acid synthesis and growth, probably indicative of cells being under stress within the 234 aggregates, or that they could even be entering into cyst formation. The final period between 60 and 72h 235 corresponds to the complete dissociation of the aggregates in which we find the downregulation of GO 236 terms related to the Nucleosome organization. 237 In conclusion, as expected, the first two steps of the aggregation process are the one with the largest 238 deployment of genes, revealing the importance of the genetic mechanisms regulating the first response to 239 FBS and the initial formation of the aggregates. Once the aggregates are formed, most of the differences 240 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint are related to metabolism and cell cycle. The other peak of DE genes is at t4 vs t3 when genes related to 241 disaggregation start to be expressed which correlates with the consumption of FBS depletion. 242 243 244 Additionally, to discern specifically which genes are being deployed, we looked at the expression 245 dynamics of genes that are in particular relevant for animal multicellularity, including cell adhesion 246 molecules such as integrins, cadherins and the Dystrophin-associated glycoprotein complex (DGC), 247 development-related transcription factors, receptor TKs and organ growth control components (Extended 248 Data Figure 7). 249 In animals ECM-cell adhesion is regulated by the interaction of integrins and cytoplasmatic associated 250 proteins such as vinculin. Our data reveals the upregulation of two out of the four integrin B present in 251 Capsaspora, specifically both integrin B1 together with vinculin have a peak at expression at 24h of 252 aggregation after which they decrease recovering their basal levels after disaggregation. This dynamic fits 253 with our model predictions in which FBS concentration would regulate the aggregation dynamics and 254 Figure 5 | GO terms enrichment analysis of DE genes during Capsaspora aggregation, The aggregation process was divided in 7 key moments that reflect its dynamism. GO enrichment analyses of DE genes is performed between consecutive time points, black GO terms are significantly enriched while grey GOs present higher p values. .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint disaggregation signalling starts earlier than observed. Another interesting integrin is integrin B2, which 255 had been shown to be expressed in Capsaspora filopodia 13 , and that presents a fast, strong response to 256 the addition of FBS and its expression levels keep increasing during the entire process of aggregation. 257 These results are also consistent with the expression pattern of many Capsaspora’s cytosolic tyrosine 258 kinases homologous to the animal's counterparts 14, which peak in their expression values at 12h, once the 259 aggregates are formed. Other metazoan relevant ECM-cell adhesion mechanisms like the Dystrophin-260 associated glycoprotein complex (DGC) display a very fluctuating pattern of expression. With regards to 261 cell-cell adhesion, Cadherins and C-type lectins are downregulated during the formation and growth of 262 the aggregates and recover their regular levels of expression after disaggregation. 263 Regarding development related-TFs, bZIP and TBox transcription factors (TFs) are among the most 264 fluctuating TFs found in Capsaspora. Other interesting TFs are Runx 1 & 2 which both are rapidly 265 expressed upon the addition to FBS. Mef2 and p53 show a fast downregulation during the first moments 266 of aggregation. NfKB is downregulated during the formation and maintenance of the aggregates, 267 recovering its expression levels during disaggregation, which might be indicative of their implication in 268 cell proliferation or stress response programs. 269 Another key developmental program in metazoans is the Hippo pathway. We find many of its core 270 regulatory components to be dynamically expressed during aggregation. For example, Merlin and Kibra 271 are upstream regulators of the pathway involved in mechanical contact sensing. Both of them are highly 272 upregulated during aggregate growth, and down-regulated during the disaggregation phase. The main 273 components like Hippo/Mst and Warts/Last do not seem to be directly regulated as in animals, as each of 274 them presents their own dynamic of expression. But we do find the expected correlation between 275 downstream effectors in Warts/Lasts which negatively regulates the expression of Sd/TEAD, which is 276 reflected in their mirroring expression dynamics, suggesting its implication in growth regulation. 277 We also found Capsaspora homologs of animal G receptors, “Gαv” & “Gαi/t/o”, that display a positive 278 fast response upon the addition of FBS to the media, which could be indicative of the detection of FBS 279 and trigger of aggregation. In addition to the receptors homologous to those of animals, we found 280 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint potential G receptors specific to Capsaspora, (COWC_0686, COWC_090031, COWC_01017) that share 281 the same response dynamic making them putative targets for FBS detection and the initial trigger of 282 aggregation. 283 Overall these results indicate that Capsaspora, a close unicellular relative of animals, already display a 284 dynamic regulatory mechanism involving genes that are relevant to animal multicellularity, which is an 285 indication that cell aggregation could have had a role in the origin of those “multicellular genes” that later 286 on were co-opted to work within a multicellular system 287 288 3 Discussion 289 Here we have shown that Capsaspora owczarzaki, one of the closest unicellular relatives of animals, 290 possesses a high dynamic plasticity regarding the formation of multicellular structures by aggregation, 291 which involves the deployment of many genes that are key for animal multicellularity. 292 Our transcriptomic data reflecting the dynamics of gene expression through the aggregation process 293 shows a very fast response at aggregate induction within minutes, this fast response resembles the 294 immediate-early response process, which is very common in animals stress response, the immune system 295 or in differentiation 15. Our results might indicate that a similar process driven by a rapid gene activation 296 could play a role during the aggregation response, this response would be widespread across the tree of 297 life and would remark the biological importance of aggregation as fast response mechanisms to face rapid 298 environmental changes and as part of the stress response as might be observed in many other organisms 299 such as Dictyostelium. 300 Our modelling study suggests that FBS-upregulated cell adhesion is a possible mechanism 301 underlying Capsaspora aggregation dynamics. By calibrating our model with experimental data, we 302 successfully captured all stages of the process, from initial clustering to aggregate growth and eventual 303 disassembly following FBS depletion. Moreover, our results indicate that Capsaspora adhesion exhibits a 304 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint nonlinear, threshold-like response to FBS: minimal adhesion at low concentrations, a sharp increase past 305 a critical threshold, and saturation at higher levels. 306 Mathematical modelling has been previously used to describe aggregation of cells (or larger 307 species) in various contexts. Our model aligns most closely with non-local adhesion models 16 17 18 19, 308 which describe aggregation driven by cell adhesion. While these models are often used to explain cell 309 sorting through differential adhesion strengths, they have also been applied to multicellular cluster 310 formation 17 1819. Similarly to these models, our approach incorporates random motility and cell-cell 311 adhesion but is specifically calibrated for Capsaspora. In contrast, other commonly used clustering 312 models do not align with Capsaspora behaviour. In particular, Capsaspora cells do not exhibit 313 coordinated or persistent movement (see Extended Data Figure 8 and Supplementary Information), ruling 314 out swarming-type models such as Vicsek and Cucker-Smale 2021 and self-propulsion-based models that 315 generate multicellular clusters through motility-induced phase separation 22 23. Additionally, chemotaxis-316 based models that can reproduce clustering 24 25 26 are also inconsistent with Capsaspora aggregation, as 317 there is no evidence of chemical signalling between cells, nor do they move up an FBS gradient during 318 disaggregation. 319 Our model provides a detailed, agent-based representation, capturing the behaviour of individual 320 cells with parameters specifically calibrated for Capsaspora. Additionally, our model explicitly couples 321 cell aggregation with the dynamics of FBS, a crucial factor for reproducing the reversible nature of 322 Capsaspora aggregation. This explains why larger aggregates dissociate from the centre while smaller 323 aggregates merge more rapidly under higher FBS concentrations. These predictions are supported by the 324 GO enrichment analysis of differentially expressed genes during the aggregation process in which GOs 325 related to growth, development, and cell-cell signalling among others are highly enriched in the first time 326 points after the aggregate induction, reflecting the high diffusion rate of FBS in the media and the fast 327 response to Capsaspora. Furthermore, a significant decrease on GO terms related to actin-based process, 328 cytoskeletal organization and growth is captured between 24 and 36h of aggregation, which indicates that 329 the beginning of the disaggregation is genetically regulated and starts earlier than observed in the 330 microscopy imaging. Additionally, after 72h when aggregates are fully dissociated reaggregation can be 331 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint re-induced by adding again the same concentration of FBS to the media triggering a very rapid 332 aggregation response (extra supplementary video 8), as we expected from the model, and observed in 333 previous work 9 . 334 This scenario also couples with previously established hypotheses in which morphological variation in 335 response to the environment is an ancient physically-based property already present in the last unicellular 336 common ancestor. This would be of especial relevance in earlier multicellular forms where the underlying 337 regulatory networks wouldn’t be fully established, and the interplay of intrinsic physical properties and 338 external conditions would be even more prevalent. These properties are found to be shared with other 339 aggregative lineages phylogenetically unrelated such as dictyostelids, fungi or even myxobacteria 6 27 , 340 and would have served as templates for the accumulation of stabilizing and reinforcing genetic circuits 341 leading to the development of obligated animal multicellularity. 342 The rapid and dynamic nature of Capsaspora aggregation might be indicative of or reflect their biological 343 i mportance for unicellular organisms, which need to adapt very fast to changing environments, and could 344 have played a crucial role in the early evolution of multicellularity by conferring an adaptive advantage in 345 fluctuating environments. Capsaspora has only been found inside the freshwater snail Biomphalaria 346 glabrata 28 which could lead to think that the aggregation response is a mechanisms to protect against the 347 host response, but Biomphalaria does not have the classical vertebrates lipoproteins used in FBS, which 348 could imply that Capsaspora’s response is an evolved response to chemical cues from the snail that can 349 indicate the identity and physiological state of its host 29 . Alternatively, Capsaspora could also live 350 outside the snail and the signal would come from other neighbouring organisms. This could be of 351 biological relevance since passive aggregation or autoaggregation, can be influenced by the density of 352 surrounding single cells. When competition between aggregates and single cells is low, aggregates 353 experience a growth disadvantage due to restricted access to resources in their interior. However, under 354 high competition, aggregates exhibit increased fitness, as vertical extension above the surface provides 355 cells at the top with improved access to nutrients 30 . Additionally, aggregation may enhance 356 Capsaspora's dispersal capability, facilitating colonization of new locations 31 5 . 357 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Our transcriptome analyses reveal the dynamic expression of relevant genes such as integrin and several 358 tyrosine kinases and TFs associated with multicellular functions in metazoans during the aggregation 359 process, as it happens with Ministeria vibrans ("A close unicellular relative reveals aggregative 360 multicellularity was key to the evolution of animals" by Li et al, submitted). The fact that those genes are 361 being dynamically deployed during aggregation and that some of them originated at the Filozoa clade 362 (Metazoa, Choanoflagellata, and Filasterea), suggest that those genes could have originally originated to 363 function for these facultative aggregation process, providing an important raw genetic material that 364 animals cold later co-opt to work within obligated multicellular entities. If so, this situates aggregation as 365 a relevant mechanism involved, directly or indirectly to the origin of animals. This is in contrast to the 366 classical views that cell aggregation never gives rise to complex multicellularity. It should be mentioned 367 that those are ad-hoc explanations, since they are based on the observation that extant complex 368 multicellular taxa (i.e., animals, plants, fungi, red and brown algae) develop by clonal division, while 369 some (not all) extant simple multicellular taxa (i.e. Dyctilostelium, Acrasis) develop through cell 370 aggregation. However, whatever the development of extant organisms nowadays (clonal or aggregative) 371 cannot be taken as a proxy of how the unicellular-to-multicellular transitions took place. Our data suggest 372 that chemically induced cell aggregation, highly influenced by cell adhesion, was most likely present in 373 the unicellular ancestor of animals and suggests an ancient, environmentally responsive property that may 374 have served as a template for more complex, genetically stabilized multicellularity. This aggregation in 375 the unicellular ancestor used many novel genes that later on were pivotal for animal multicellularity and 376 development. Thus, that the first animal emerged by cell aggregation to then evolve into an obligate 377 multicellular entity with embryonic development cannot be fully discarded 1 . 378 Our work, notably the development of a robust mathematical model, firmly establishes Capsaspora as a 379 powerful model system to further analyse the evolutionary potential of cell aggregation to either evolve 380 genes that later on were crucial to animals or even to create the first animal. In particular, having this 381 mathematical model, together with the recent establishment of transfection and genome editing tools in 382 this organism, will allow researchers to interrogate, in Capsaspora, the potential ancestral function of 383 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint many genes key for animals, as well as to perform experimental evolution studies that can test different 384 aggregation dynamics. 385

386 1. Ruiz-Trillo, I., Kin, K. & Casacuberta , E. The Origin of Metazoan Multicellularity: A Potential 387 Microbial Black Swan Event. 136, 26 (2023). 388 2. Pentz, J. T. et al. Evolutionary consequences of nascent multicellular life cycles. Elife 12, (2023). 389 3. Herron, M. D. et al. De novo origins of multicellularity in response to predation. Sci Rep 9, (2019). 390 4. Pentz, J. T. et al. Ecological Advantages and Evolutionary Limitations of Aggregative 391 Multicellular Development. Current Biology 30, 4155-4164.e6 (2020). 392 5. Tong, K., Bozdag, G. O. & Ratcliff, W. C. Selective drivers of simple multicellularity. Current 393 Opinion in Microbiology vol. 67 Preprint at https://doi.org/10.1016/j.mib.2022.102141 (2022). 394 6. Newman, S. A., Forgacs, G. & Müller, G. B. Before programs: The physical origination of 395 multicellular forms. International Journal of Developmental Biology 50, 289–299 (2006). 396 7. Newman, S. A. Cell differentiation: What have we learned in 50 years? J Theor Biol 485, (2020). 397 8. Suga, H. et al. The Capsaspora genome reveals a complex unicellular prehistory of animals. Nat 398 Commun 4, (2013). 399 9. Ros-Rocher, N. et al. Chemical factors induce aggregative multicellularity in a close unicellular 400 relative of animals. Proc Natl Acad Sci U S A 120, (2023). 401 10. Ros-Rocher, N., Pérez-Posada, A., Leger, M. M. & Ruiz-Trillo, I. The origin of animals: An 402 ancestral reconstruction of the unicellular-to-multicellular transition. Open Biology vol. 11 Preprint 403 at https://doi.org/10.1098/rsob.200359 (2021). 404 11. Ocaña-Pallarès, E. et al. Divergent genomic trajectories predate the origin of animals and fungi. 405 Nature 609, 747–753 (2022). 406 12. Sebé-Pedrós, A. et al. Regulated aggregative multicellularity in a close unicellular relative of 407 metazoa. Elife 2013, (2013). 408 13. Parra-Acero, H. et al. Integrin-Mediated Adhesion in the Unicellular Holozoan Capsaspora 409 owczarzaki. Current Biology 30, 4270-4275.e4 (2020). 410 14. Suga, H. et al. Genomic Survey of Premetazoans Shows Deep Conservation of Cytoplasmic 411 Tyrosine Kinases and Multiple Radiations of Receptor Tyrosine Kinases. 412 www.SCIENCESIGNALING.org (2012). 413 15. Bahrami, S. & Drabløs, F. Gene regulation in the immediate-early response process. Advances in 414 Biological Regulation vol. 62 37–49 Preprint at https://doi.org/10.1016/j.jbior.2016.05.001 (2016). 415 16. Buttenschön, A. , & H. T. Non-Local Cell Adhesion Models. (Springer, 2021). 416 17. Armstrong, N. J., Painter, K. J. & Sherratt, J. A. A continuum approach to modelling cell- cell 417 adhesion. J Theor Biol 243, 98–113 (2006). 418 18. Juliette, G., Peters, R. & Owen, D. M. An agent- based model of molecular aggregation at the cell 419 membrane. PLoS One 15, (2020). 420 19. Noureen, S. R., Mort, R. L. & Yates, C. A. Modeling adhesion in stochastic and mean- field models 421 of cell migration. Phys Rev E 111, 014419 (2025). 422 20. Vicsek, T., Czirok, A., Ben-Jacob, E., Cohen, I. & Shochet, O. PH YS ICAL REVIEW LETTERS 7 423 AUGUs~ 1995 Novel Type of Phase Transition in a System of Self-Driven Particles. vol. 75 (1995). 424 21. Cucker, F. & Smale, S. Emergent behavior in flocks. IEEE Trans Automat Contr 52, 852–862 425 (2007). 426 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 22. Gonnella, G., Marenduzzo, D., Suma, A. & Tiribocchi, A. Motility-induced phase separation and 427 coarsening in active matter. C R Phys 16, 316–331 (2015). 428 23. Caprini, L., Marini Bettolo Marconi, U. & Puglisi, A. Spontaneous Velocity Alignment in 429 Motility-Induced Phase Separation. Phys Rev Lett 124, (2020). 430 24. Keller, E. F., Segelf, L. A. & Lxvision, B. Initiation of Slime Mold Aggregation Viewed as an 431 Instability. J. theor. Biol vol. 26 (1970). 432 25. Palsson, E. & Othmer, H. G. A Model for Individual and Collective Cell Movement in 433 Dictyostelium Discoideum. www.pnas.org (2000). 434 26. Avery, L., Ingalls, B., Dumur, C. & Artyukhin, A. A Keller- Segel model for C elegans L1 435 aggregation. PLoS Comput Biol 17, (2021). 436 27. Arias Del Angel, J. A., Nanjundiah, V., Benítez, M. & Newman, S. A. Interplay of mesoscale 437 physics and agent-like behaviors in the parallel evolution of aggregative multicellularity. EvoDevo 438 vol. 11 Preprint at https://doi.org/10.1186/s13227-020-00165-8 (2020). 439 28. Stibbs, H. H., Owczarzak, A., Bayne, C. J. & Dewan, P. Schistosome Sporocyst-Killing Amoebae 440 Isolated from Biomphalaria Glabra Ta. PATHOLOGY vol. 33 (1979). 441 29. Kidner, R. Q. et al. Lipids from a snail host regulate the multicellular behavior of a predator of 442 parasitic schistosomes. iScience 27, 110724 (2024). 443 30. Trunk, T., Khalil, H. S. & Leo, J. C. Bacterial autoaggregation. AIMS Microbiol 4, 140–164 444 (2018). 445 31. Kragh, K. N. et al. Role of multicellular aggregates in biofilm formation. mBio 7, (2016). 446 447

448 Cell culture and aggregate induction 449 Capsaspora owczarzaki cells (strain ATCC 30864) were maintained axenically in ATCC medium 1034 450 (modified PYNFH medium) with the addition of 10% (v/v) fetal bovine serum (Sigma-Aldrich, F9665) at 451 23°C. 452 Aggregate formation was induced chemically following a protocol adapted from 9. Cells were washed 453 twice in FBS-free growth medium by centrifugation at 5000g for 5mins. Subsequently, 1x106 Capsaspora 454 cells were seeded in a 12 well plate (Costar® 12-well plates REF3513) coated with collagen to generate a 455 low adherent surface for Capsaspora 13 456 Collagen coating was performed by adding 600µl of 20µg/mL collagen into each well for a minimum of 457 30mins, after which it is removed and the wells are left to dry completely before seeding the cells in FBS-458 free medium and incubated overnight. Aggregation was then induced by adding FBS at the desired 459 concentrations. 460 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Bright Field imaging was performed using a Zeiss Axio Observer Z.1 epifluorescence inverted 461 microscope equipped with LED illumination and an Axiocam 503 mono camera at 5x magnification for 462 (Supp Videos 1,2/ 5&10%FBS aggregation), or an Eclipse TS100 Nikon epifluorescence inverted 463 microscope equipped with an Intensilight C-HGFI Illuminator and a DS-Fi2 Camera Head at 10× 464 magnification (Supp Video3/0,05%FBS aggregation). Average aggregate areas were measured by batch 465 processing with a macro script in Fiji imaging software version 2.14.0/1.54f 466 Computational model of Capsaspora aggregation 467 W e developed a computational model to investigate the mechanisms of Capsaspora aggregation. Our 468 model consists of two key components: (i) an agent-based cell model to account for Capsaspora 469 proliferation and migration (Extended Data Figure 1a) and (ii) a continuum description of the dynamics of 470 Fetal Bovine Serum (FBS), a chemical stimulant in the culture medium which can affect Capsaspora 471 behaviour (Extended Data Figure 1b). These components are coupled to simulate cell behaviour under 472 varying FBS concentrations (Extended Data Figure 1c). The model has been calibrated using 473 experimental data where available, ensuring that key processes, such as cell proliferation, FBS-dependent 474 motility, and FBS dynamics, are informed by observed Capsaspora behaviour. Here, we summarise the 475 main processes incorporated in our Capsaspora model, while Supplementary Information provides a 476 detailed description of the model. 477 Cell model Th e centre-based approach employed in this work treats each cell as a circular entity in 478 two dimensions (Extended Data Figure 1a). Adult (fully grown after cell division) cells are assumed to 479 have fixed cell radius and thus, we only need to track the position of their centres to know the distribution 480 of cells across the domain. The position of the centre of 𝑖𝑖th cell, 𝑥𝑥𝑖𝑖, evolves according to the following 481 ordinary differential equation: 482 𝑑𝑑𝑥𝑥𝑖𝑖 𝑑𝑑𝑑𝑑 = 1 𝜈𝜈𝐹𝐹𝑖𝑖 𝑚𝑚 + 𝐹𝐹𝑖𝑖 𝑑𝑑. (1) 483 In this equation, 𝜈𝜈 r epresents the drag coefficient, and 𝐹𝐹𝑖𝑖 𝑚𝑚 denotes the mechanical forces arising from 484 interactions with neighbouring cells. These forces include short-range repulsion when cell centres are in 485 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint close proximity and long-range adhesion mediated by Capsaspora filopodia (Extended Data Figure 1a). 486 As shown in 9 , the metabolic activity of Capsaspora in the presence of externally supplied FBS increases 487 the ‘stickiness’ of their filopodia, enhancing long-range interactions between cells. Accordingly, we 488 assume that the strength of these adhesion forces increases with local FBS concentration. 489 In addition to mechanical forces, cell movement is influenced by random motility, represented by 490 the term 𝐹𝐹𝑖𝑖 𝑑𝑑 in Eq (1). We calibrated the random motility component using experimental data on 491 individual cell trajectories at various FBS concentrations. Our analysis, detailed in Supplementary 492 Information, demonstrates that Capsaspora motility is significantly enhanced at higher FBS levels. To 493 capture this behaviour, we model cell motility as an increasing function of FBS concentration. 494 Thus, Capsaspora aggregation observed in experiments can arise in our model simulations due to a 495 balance between FBS-mediated adhesion and random motility. In the absence of FBS, Capsaspora cells 496 fail to form aggregates 9 , which suggests that the adhesion forces are weaker than the random motility 497 that disperses the cells. However, when stimulated with FBS, aggregation occurs, indicating that the FBS-498 dependent cell interactions dominate, even though cell motility also increases in high-FBS environments. 499 In our simulations, we also incorporate cell proliferation, modelled stochastically based on a 500 distribution fitted to experimental data on Capsaspora division times, as reported in 32. Cell division is 501 assumed to be symmetric, consistent with the findings in 32, where each division produces two daughter 502 cells of approximately equal size. After division, the daughter cells gradually grow to reach the size of an 503 adult Capsaspora over a specified period. 504 FBS model The FBS dynamics are modelled by a partial differential equation (PDE) (Extended Data 505 Figure 1b), which describes its diffusion across the medium and consumption by the cells. The diffusion 506 coefficient and uptake rates were fitted based on experimental data on FBS depletion, which leads to 507 Capsaspora disaggregation behaviour. 508 Coupling between cell and FBS models The model couples cell behaviour with FBS concentration 509 (Extended Data Figure 1c), allowing motility and intercellular adhesion to vary with local FBS levels 510 while cell metabolic activity depletes the available FBS. This coupling enables the simulation of cell 511 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint aggregation patterns as a response to varying FBS concentrations, providing insight into the mechanisms 512 driving Capsaspora aggregation. 513 RNA extraction and Temporal transcriptomics analysis 514 Whole RNA was extracted from the cells in triplicate for each condition after 0 min (i.e., immediately 515 before aggregate induction, T(0), 3min (T1), 12h (T2), 24h (T3), 36h (T4), 48h (T5), 60h (T6) & 72h (T7) 516 of incubation, using Trizol (Invitrogen/Thermo Fisher Scientific, 15596026) as described in 33 . RNA was 517 purified with RNeasy minikit QIAGEN (ref 74104). And measured using Quibit 2.0 fluorometer from 518 Invitrogen. Samples were sent to NOVOGENE for sequencing on Illumina Novaseq 6000 platform with 519 paired-end 150bp sequencing strategy, library preparation was performed by polyA capture and cDNA 520 synthesis, purification was checked by PCR. 521 522 The quality of the sequence reads was checked with FastQC 523 ( https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The reads were aligned to the recently 524 assembled version of the Capsaspora owczarzaki genome (DDBJ BioProject ID: PRJDB19057), counted 525 and quantified with RSEM (–bowtie2 option). DESeq2 34 was used to identify differentially expressed 526 genes at false discovery rate (FDR) < 0.05. Identification and visualization of gene expression clusters 527 was performed with DEGpatterns 33, and out of the 52 clusters obtained we determined the best number 528 of consensus clusters based on the cumulative distribution function (CDF) of the consensus index vs the 529 number of clusters resulting in the 6 consensus clusters represented in figure 4 35. The GO term 530 enrichment analysis and annotation was performed with topGO36. A detailed script of each of the steps 531 has been uploaded in our GitHub repository. 532 Data availability statement 533 T he simulation results presented in this paper have been uploaded to FigShare at 534 https://doi.org/10.6084/m9.figshare.28761560. Using the scripts provided in our GitHub repository 535 (https://github.com/daria-stepanova/Capsaspora.git), all figures and movies from this study can be fully 536 reproduced. 537 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint The Movies 1 and 8 presented in this paper have been uploaded to FigShare at 538 10.6084/m9.figshare.29044817. 539 The transcriptomic raw data used for the analyses during the current study is available in the EMBL-EBI 540 repository with the accession code PRJEB89428. 541 The Capsaspora annotated genome used for the analyses is available at the DDBJ repository under the 542 BioProject: PRJDB19057 (PSUB024242) 543 Code availability statement 544 The code for our mathematical model is available on GitHub at https://github.com/daria-545 stepanova/Capsaspora.git. The repository includes a detailed README file with instructions on running 546 the model and reproducing the simulation results presented in this study. 547 548

references 549 32. Pérez-Posada, A., Dudin, O., Ocaña-Pallarès, E., Ruiz-Trillo, I. & Ondracka, A. Cell cycle 550 transcriptomics of Capsaspora provides insights into the evolution of cyclin-CDK machinery. PLoS 551 Genet 16, (2020). 552 33. Ocaña-Pallarès, E., Najle, S. R., Scazzocchio, C. & Ruiz-Trillo, I. Reticulate evolution in 553 eukaryotes: Origin and evolution of the nitrate assimilation pathway. PLoS Genet 15, (2019). 554 34. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for 555 RNA-seq data with DESeq2. Genome Biol 15, (2014). 556 35. Pantano, L. DEGreport: Report of DEG analysis. Preprint at (2025). 557 36. Alexa, A. R. J. topGO: Enrichment Analysis for Gene Ontology. Preprint at (2004). 558 37. Ferrer-Bonet, M. & Ruiz-Trillo, I. Capsaspora owczarzaki. Current Biology vol. 27 R829–R830 559 Preprint at https://doi.org/10.1016/j.cub.2017.05.074 (2017). 560 561

562 Work in the IRT lab is supported by grants PID2020-120609GB-I00 funded by MICIU/ AEI 563 /10.13039/501100011033/ and by “ERDF A way of making Europe”, and by PID2023-153273NB-I00 564 funded by MICIU/AEI /10.13039/501100011033 and FEDER, UE. We also acknowledge support to 565 Departament de Recerca i Universitats de la Generalitat de Catalunya (exp. 2021 SGR 00751) and 566 support by PIE-202120E047- Conexiones-Life. D.S. and T.A. thank the CERCA Program/Generalitat de 567 Catalunya for institutional support. D.S. and T.A. have been funded by grant PID2021-127896OB-I00 568 funded by MCIN/AEI/10.13039/501100011033 ‘ERDF A way of making Europe’. The work of D.S. and 569 T.A. has been supported by the Spanish Research Agency (AEI), through the Severo Ochoa and Maria de 570 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Maeztu Program for Centers and Units of Excellence in R&D (CEX2020-001084-M). K.K. received the 571 support of a fellowship (LCF/BQ/PI20/11760009) from ”la Caixa” Foundation (ID 100010434) and from 572 the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-573 Curie grant agreement No 847648. 574 575 576 577 Author contributions 578 All authors contributed to the conception of the project. D.S. and T.A. developed the computational 579 models. D.S. performed all the simulations. G.B.S. captured videos of Capsaspora aggregation. K.K. and 580 G.B.S. performed the RNA-seq experiments. G.B.S. analysed all the RNA-seq data. G.B.S., D.S., and 581 I.R.T. prepared the first draft of the manuscript. K.K., T.A. and I.R.T were involved in supervision and 582 funding acquisition. All authors contributed to the revision of the manuscript. 583 Competing interest declaration 584 The authors declare no competing interests. 585

and Correspondence 586 Correspondence and requests for materials should be addressed to Tomàs Alarcón ([email protected]) or 587 Koryu Kin ([email protected]). 588 589 Additional information 590 Supplementary information and videos accompanying the manuscript are available at FigShare 591 (https://doi.org/10.6084/m9.figshare.28761560.v1). 592 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Extended Data Figure and Table legends 593 594 595 596 Extended Data Figure 1 | A schematic representation of our modelling approach. a, A centre-based 597 model is used to describe the behaviour of proliferating and migrating Capsaspora cells. In two 598 dimensions, each cell is modelled as a circle of radius, r, and the positions of cell centres are tracked 599 throughout the simulations. Cell movement is influenced by interactions with neighbouring cells and 600 random motility. Short-range repulsion prevents collisions, while filopodia-mediated long-range adhesion 601 can draw cells closer. We note that filopodia are not drawn to scale as they can reach several cell 602 diameters 37. We assume that both filopodia-mediated adhesion and random motility increase with local 603 FBS concentration. b, A partial differential equation (PDE) is used to describe the dynamics of FBS in the 604 culture medium, accounting for its diffusion and uptake by cells. FBS decay is considered negligible 605 within the timescale of interest. c, The full model couples cell dynamics with the continuum description 606 of FBS concentration in the medium. 607 608 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Extended Data Figure 2 | Comparison of Capsaspora aggregation between experimental results 609 and model simulations under 5% FBS in a smaller domain. Experimental images (top row) and 610 snapshots of a numerical simulation (bottom row) of chemically induced Capsaspora aggregation at a, 0-611 24 and b, 36-70 hours after administration of 5% FBS. Both experimental and simulated images cover a 612 360 x 360 μm domain. Scale bar in experimental images corresponds to 100 μm. In the simulation 613 snapshots, cell colours represent the local FBS concentration, which is also reflected in the background of 614 the medium. The colour bar on the right indicates FBS levels: initial 5% FBS appears grey, while 615 depletion due to metabolic activity of Capsaspora shifts colours to blue, with dark blue indicating no 616 FBS. At early time moments of Capsaspora aggregation (0-24 h in a), no noticeable FBS depletion can 617 be seen. However, lower concentrations of FBS in regions surrounding cell aggregates after 30 can be 618 observed, which leads to aggregate disassembly at 60 h and complete disaggregation at 70 h. Numerical 619 simulations were performed with sigmoidal Hill response of cell adhesion strength to FBS levels (see 620 Section I of Supplementary Information). All parameter values are as in Extended Data Table 1. For the 621 complete animation of this simulation, see Movie 2. 622 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 623 624 Extended Data Figure 3 | Quantification of the largest multicellular aggregate (by area) in 625 numerical simulations for different initial FBS concentrations. Panels show the temporal evolution of 626 a, the mean area of the largest aggregate, b, the number of cells within this aggregate, and c, the 627 corresponding aggregate density. Solid lines represent the mean values across five numerical simulations 628 for each condition, with shaded regions indicating standard deviation. Colours distinguish simulations 629 with initial FBS levels of 0.5% (purple), 1% (orange), 5% (blue), and 10% (green). The first column 630 corresponds to a threshold-like Hill response of cell adhesion to FBS, the middle column to a moderate 631 linear increase (shallow slope), and the right column to a steep linear response. Animations of individual 632 simulations are available in Movies 3-5 for the Hill, shallow linear, and steep linear responses, 633 respectively. Notably, variance in the metrics increases significantly as aggregate disassembly begins 634 around 55-60 h. Additionally, panel c reveals that aggregates are more compact at higher FBS 635 concentrations, explaining why they appear smaller at these levels, as also observed experimentally in 9. 636 For these simulations, we set the domain size to 360 x 360 μm, and all parameter values are as in 637 Extended Data Table 1. 638 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 639 Extended Data Figure 4: Most variable during aggregation process a, PCA of the top 100 most 640 variables genes at eight different time points of the aggregation process b, Heatmap of the top 100 most 641 variable genes forms four clusters based on pairwise correlation representing disaggregation t5,t6t7; 642 adherent stage t0; aggregate growth t2,t3,t4; & Aggregate induction t0. 643 644 645 Extended Data Figure 5: a, Cumulative Distribution Function plot (CDF) calculated for different 646 number of gene expression pattern clusters, after 6 clusters (light blue) the curves start to flatten, meaning 647 there is no addition to the total explained variability indicating that 6 clusters are sufficient to capture 648 most of the variability b, DEGpattern Dendrogram of the 54 clusters obtained by hierarchical 649 clustering, the blue line represents the 6 supergroups used for the gene expression pattern analysis (Figure 650 4) 651 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 652 Extended Data Figure 6: Dotplots of enriched GO terms across the aggregation process. The 653 dotplots represent the top 20 significant GO terms for each of the timepoint pairwise comparisons 654 generated with clusterprofiler, the color of the dots indicates the p-value and the size the number of genes 655 in the GO term, the position indicates the enrichment score, genes with high enrichment score (right 656 position), high p-values (warm colours), and high number of genes associated to the GO (bigger size) are 657 the most significant ones. 658 659 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 660 Extended Data Figure 7: Dynamic expression patterns of metazoan homolog genes related to 661 multicellularity a, Integrins b, Cytosolic tyrosin kinases c, Transmembrane G receptors d, Cadherin e, 662 Dystrophin associated glycoprotein complex f, Hippo pathway g, General transcription factors h, TBox 663 transcription factors i, bZip transcription factors. The error bars represent one standard error of the mean 664 (SEM). 665 666 667 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint Extended Data Figure 8 | Modelling cell geometry, division dynamics, and adhesion response in 668 Capsaspora. a, Capsaspora cells have spherical bodies with long filopodia (not drawn to scale), which 669 facilitate long-range interactions. In our simulations, we simplify their representation to circular cell 670 bodies. b, c, Experimental data from 32 support this circular approximation. b, The histogram shows the 671 distribution of cell radii (𝑟𝑟) inferred from area measurements, with a mean of 3.10 ± 0. 17 𝜇𝜇𝜇𝜇. We use 672 𝑟𝑟 = 3 𝜇𝜇𝜇𝜇 in simulations. c, A comparison of experimental and estimated perimeters (𝑃𝑃∗ = 2𝜋𝜋 𝑟𝑟) shows 673 a low mean relative error ( = 0.0263), validating the circular assumption. d, e, The cell cycle 674 model was calibrated using Capsaspora division time distributions from 32. d, The histogram (purple) 675 shows the distribution of (12 h – division time), fitted to Gamma (magenta) and Inverse Gaussian (blue) 676 distributions. The Gamma shape and scale parameters are 𝑘𝑘� = 6. 30 and 𝜃𝜃� = 0.34, while the Inverse 677 Gaussian mean and shape parameters are 𝜇𝜇 �= 2.15 and 𝜆𝜆̃ = 12.23. e, The cumulative density functions 678 (cdfs) confirm the fit of both distributions to experimental data. f, The rescaled Morse force magnitude, 679 | 𝐹𝐹𝑖𝑖𝑖𝑖 𝑚𝑚 |/ 𝜈𝜈 ,, is plotted against cell separation distance 𝛿𝛿𝑖𝑖𝑖𝑖. The force is repulsive for 𝛿𝛿𝑖𝑖𝑖𝑖 𝛿𝛿𝑒𝑒, with zero force at equilibrium. g, The FBS-dependent adhesion strength, 𝜇𝜇𝑠𝑠(𝐹𝐹𝐹𝐹𝐹𝐹), 681 follows three functional forms: a Hill response (blue) and two linear responses with different slopes 682 (magenta and black). Parameter values are listed in Extended Data Table 1. 683 684 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 685 Extended Data Figure 9 | Single-cell motility of Capsaspora in different FBS conditions. a, b, 686 Trajectories of individual cells under a 0% FBS and b 5\% FBS. The images show the initial cell 687 positions, with the full 1-hour trajectories overlaid (colouring is arbitrary). As such, cells are positioned at 688 the start of their respective tracks. Tracks of clustered cells or those present only briefly were excluded 689 from the analysis. c, d, Velocity distributions of individual cells from c Fpop and d Cl1 cell lines. The 690 distributions are calculated for 33 tracks in c and 14 tracks in d. e, The ensemble-averaged MSD, 691 extracted from individual cell trajectories, is used to estimate the diffusion coefficient for cells in 0% FBS 692 (magenta) and 5% FBS (blue). Solid lines represent the mean MSD values, as a function of 693 the delay time 𝜏𝜏 (mins), while the shaded regions indicate the weighted standard deviation. The diffusion 694 coefficient 𝐷𝐷 is estimated by fitting a linear function of the form = 2 𝑑𝑑 𝐷𝐷 𝜏𝜏, to the first 20 695 minutes of the MSD curves. Here, 𝑑𝑑= 2 indicates the spatial dimension. Using this approach, we obtain 696 𝐷𝐷 = 2. 53 𝜇𝜇𝜇𝜇2𝜇𝜇𝑖𝑖𝑚𝑚−1 for 0% FBS, with a 95%-confidence interval of [2.47, 2.59] and a goodness-of-fit, 697 𝑅𝑅2 = 0.995. For 5% FBS, the diffusion coefficient is 𝐷𝐷 = 15 .98 𝜇𝜇𝜇𝜇2𝜇𝜇𝑖𝑖𝑚𝑚−1, with a 95%-confidence 698 interval of [15.73, 16.24] and a goodness-of-fit, 𝑅𝑅2 = 0.998. f, We also calculated the normalised 699 velocity autocorrelation function, 𝑍𝑍(𝜏𝜏) =/ , f or individual cell 700 trajectories in 0% FBS (magenta) and 5% FBS (blue). Solid lines show the mean autocorrelation, while 701 the shaded regions represent the weighted standard deviation. The statistics in e and f are calculated for 702 78 tracks in 0% FBS and 69 tracks in 5% FBS. 703 704 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint 705 706 Par. Value Description r 3 μm Radius of the Capsaspora spherical core. 𝛿𝛿𝑒𝑒 6 𝜇𝜇𝜇𝜇 Equilibrium distance for the Morse force for adult Capsaspora. 𝜇𝜇𝑠𝑠 900 𝜇𝜇𝜇𝜇⋅ ℎ−1 Rescaled strength of the Morse force (𝜇𝜇𝑠𝑠= 𝜇𝜇𝑠𝑠� /𝜈𝜈, where 𝜈𝜈 denotes the drag coefficient) from Eq (7) in SI. 𝑎𝑎𝑠𝑠 0.67 𝜇𝜇𝜇𝜇−1 Parameter controlling the range of interaction for short-range repulsion in Eq (7) in SI. 𝑎𝑎𝑙𝑙 0.17 𝜇𝜇𝜇𝜇−1 Parameter controlling the range of interaction for long-range adhesion mediated by filopodia in Eq (7) in SI. 𝜇𝜇0 1412.1 𝜇𝜇𝜇𝜇⋅ ℎ−1 Maximum value of the rescaled adhesion strength of the Morse force for the Hill response to FBS (see blue curve in Extended Data Figure 9g). 𝑓𝑓0 3.8494 % FBS concentration producing half of the maximum adhesion strength in the Hill response to FBS (see blue curve in Extended Data Figure 9g). 𝑚𝑚0 1.8153 Hill coefficient in the Hill response to FBS of Capsaspora adhesion strength (see blue curve in Extended Data Figure 9g). 𝑘𝑘𝜇𝜇 1 66 𝜇𝜇𝜇𝜇⋅ ℎ−1 Slope parameter in the linear response (with shallow slope) to FBS of Capsaspora adhesion strength (see magenta curve in Extended Data Figure 9g). 𝑘𝑘𝜇𝜇 2 120 𝜇𝜇𝜇𝜇⋅ ℎ−1 Slope parameter in the linear response (with steep slope) to FBS of Capsaspora adhesion strength (see black curve in Extended Data Figure 9g). 𝑘𝑘 2.6906 Slope parameter in the linear dependency of the Capsaspora diffusion coefficient on FBS concentration (Eq (8) in SI). 𝑏𝑏 2.53 Intercept parameter in the linear dependency of the Capsaspora diffusion coefficient on FBS concentration (Eq (8) in SI). 𝐷𝐷𝑓𝑓𝑓𝑓𝑠𝑠 675 𝜇𝜇𝜇𝜇2 ⋅ ℎ−1 FBS diffusion coefficient. 𝑘𝑘𝑢𝑢 20.7 𝜇𝜇𝜇𝜇2 ⋅ 𝑐𝑐𝑒𝑒 𝑐𝑐𝑐𝑐−1 ℎ−1 Rate of FBS consumption per cell per area of the finite element model (FEM) element (in our simulations, a right triangle with sides equal to 14.4 𝜇𝜇𝜇𝜇). 𝐿𝐿 360 or 1080 𝜇𝜇𝜇𝜇 Size of the square domain used in the model simulations. 𝑁𝑁0 24 or 216 cells Initial cell number scattered across the simulation domain at 𝑑𝑑= 0 for domain with side of 360 𝜇𝜇𝜇𝜇 and 1080 𝜇𝜇𝜇𝜇, respectively. 𝜇𝜇� 2.1530 The mean parameter of the Inverse Gaussian distribution described by the probability density function from Eq (3) in SI which provided the best fit for the experimental data of the quantity [12 h - division time (h)]. 𝜆𝜆̃ 12.2344 The shape parameter of the Inverse Gaussian distribution described by the probability density function from Eq (3) in SI which provided the best fit for the experimental data of the quantity [12 h - division time (h)]. Parameter values after rescaling the space with the cell radius. r 1 Radius of the Capsaspora spherical core. 𝜇𝜇𝑠𝑠 300 ℎ−1 Rescaled strength of the Morse force. 𝑎𝑎𝑠𝑠 2.0 Parameter controlling the range of interaction for short-range repulsion. 𝑎𝑎𝑙𝑙 0.5 Parameter controlling the range of interaction for long-range adhesion via filopodia. 𝜇𝜇0 470.7 ℎ−1 Maximum value of the rescaled adhesion strength of the Morse force for the Hill response to FBS (see blue curve in Extended Data Figure 9g). 𝑘𝑘𝜇𝜇 1 22 ℎ−1 Slope parameter in the linear response (with shallow slope) to FBS of Capsaspora adhesion strength (see magenta curve in Extended Data Figure 9g). 𝑘𝑘𝜇𝜇 2 40 ℎ−1 Slope parameter in the linear response (with steep slope) to FBS of Capsaspora adhesion strength (see black curve in Extended Data Figure 9g). 𝑘𝑘 0.2990 Slope parameter in the linear dependency of the Capsaspora diffusion coefficient on FBS concentration. 𝑏𝑏 0.2811 Intercept parameter in the linear dependency of the Capsaspora diffusion coefficient on FBS concentration. 𝐷𝐷𝑓𝑓𝑓𝑓𝑠𝑠 75 ℎ−1 FBS diffusion coefficient. 𝑘𝑘𝑢𝑢 2.3 𝑐𝑐𝑒𝑒 𝑐𝑐𝑐𝑐−1 ℎ−1 Rate of FBS consumption per cell per area of the FEM element (in our simulations, a right triangle with sides equal to 4.8, dimensionless). 𝐿𝐿 120 or 360 Size of the square domain used in the model simulations. 707 Extended Data Table 1 | Descriptions of parameters included in our model and their baseline 708 values used in this work. For simplicity, we omit bars in the notation of the parameters rescaled with cell 709 radius. SI: Supplementary Information. 710 711 .CC-BY-NC-ND 4.0 International licensemade available under a (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 The copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.14.653760doi: bioRxiv preprint