Compressive forces induce epigenetic activation of aged human dermal fibroblasts through ERK signaling pathway

mechanical rejuvenation, epigenetic activation , ERK signaling 20 pathway, human dermal fibroblasts. 21 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint This PDF file includes: 22 Main Text 23 Figures 1 to 7 24

25 Age-related changes in human dermal fibroblasts (HDFs) contribute to impaired 26 wound healing and skin aging. While these changes result in altered 27 mechanotransduction, the epigenetic basis of rejuvenating aging cells remains 28 a significant challenge. This study investigates the effects of compressive 29 forces on nuclear mechanotransduction and epigenetic rejuvenation in aged 30 HDFs. Using a compressive force application model, the activation of HDFs 31 through alpha -smooth muscle actin ( ɑ-SMA) is demonstrated. Sustained 32 compressive forces induce significant epigenetic modifications, including 33 chromatin remodeling and altered histone methylation patterns. These 34 epigenetic changes correlate with enhanced cellular migration and 35 rejuvenation. Small -scale drug screening identifies the extracellular signal -36 regulated kinase (ERK) signaling pathway as a key mediator of compression -37 induced epigenetic activation. Furthermore, implanting aged cell spheroids to 38 an aged skin model and subjecting the tissue with compressing forces resulted 39 in increased collagen I protein levels. Collectively, these findings demonstrate 40 that applying compressive force to aged fibroblasts activates global epigenetic 41 changes through the ERK signaling pathway, ultimately rejuvenating cellular 42 functions with potential applications for wound healing and skin tissue 43 regeneration. 44 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Significance Statement 45 46 Partial rejuvenation of aging cells is desirable but is still a major challenge. In 47 this paper, we demonstrate that aged human dermal fibroblasts, embedded in 48 a 3D collagen hydrogel matrix as spheroids, subjected to external static 49 compressive force exhibit partial rejuvenation. Through immunofluorescence, 50 small-scale inhibitor screen and gene expression analysis, we identify some of 51 the critical mechanotransduction pathways in this process. Collectively, our 52

provide compelling evidence that tissue compression results in the 53 activation of potential rejuvenation pathways in aging cells. 54 55

56 Cellular aging is accompanied by various changes in the characteristics of cells, 57 such as genomic instability, telomere attrition, epigenetic changes, loss of 58 proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, cellular 59 senescence, stem cell exhaustion, and altered intercellular communication (1). 60 For example, dermal fibroblasts secrete various extracellular matrix proteins 61 (ECM) into the dermal compartment and contribute to matrix stiffness of the 62 skin (2). Many studies have shown that aging leads to accumulation of 63 senescent fibroblasts resulting in decreased ECM production thereby leading 64 to a loss of skin tissue integrity and of wound healing properties (3). A major 65 challenge in the field is how such aging cells can be activated or rejuvenated. 66 Current strategies to combat aging include induced pluripotent stem cells 67 (iPSC) and mechanical reprogramming (4, 5), metabolic manipulation (daily or 68 intermittent caloric restriction), blood transfusion, small molecule drugs 69 (Rapamycin, Metformin, Ascorbate and Aspirin) and senescent cell ablation 70 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint (Senolytics) (6). Mechanical forces (stretch, shear, compression) have also 71 been shown to activate major signaling pathways, cytoskeleton/chromatin 72 remodeling and gene expression (7 –10). Since cells sense extracellular 73 mechanical cues in tissue microenvironments, we hypothesized that 74 compressive forces could enable the activation/rejuvenation of aging cells. 75 Recent literature has also shown that cancer cells under tissue compression 76 could get activated to a metastatic phenotype (11). Based on previous studies, 77 including our own, on the effects of compressive force on cellular function (12, 78 13), we designed an engineered tissue embedded with aging cells and revealed 79 that tissue compression could provide important avenues for cell 80 activation/rejuvenation. 81 82 In this paper, we develop a force application device, which includes human 83 dermal fibroblasts (derived from a 75 year old healthy male donor) embedded 84 in a 3D collagen hydrogel matrix and subjected to external loading in the form 85 of static compressive force. Using a fibroblast spheroid model, we show that 86 HDFs can be activated by compressive force, as evidenced by increased levels 87 of ɑ Smooth Muscle actin (ɑSMA) and the accompanying cellular memory 88 responses. In particular, we measure the levels of phosphorylated myosin light 89 chain (pMLC) levels, cytoskeletal remodeling, chromatin modifications, reduced 90 DNA damage, global gene expression, and cell migration to demonstrate the 91 activation of HDFs. A key element of the activation process involves the 92 clumping of aged cells into spheroids before the application of force, as single 93 cells embedded in a collagen matrix under compressive force do not undergo 94 activation. We also validated our findings in an artificial aged skin model and 95 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint observed an increase trend in collagen 1protein levels in the spheroid injection 96 group compared to the single -cell injection group. Collectively, our results 97 provide compelling evidence that tissue compression results in the activation of 98 potential rejuvenation of aging cells. 99 100

101 Compressive force induces transient activation of fibroblasts and 102 promotes rejuvenation 103 We established two models for 3D cell culture: one involve s embedding aged 104 fibroblasts as single cells in a collagen hydrogel (hereafter called the single cell 105 model), and the other involves aged fibroblasts as spheroids embedded in a 106 collagen hydrogel (called spheroid model) (Figure 1A, Figure S1E). To achieve 107 the spheroid model, GM08401 fibroblasts (75 year old donor, old group) and 108 GM09503 fibroblasts (10 year old donor, young group) were cultured on 109 fibronectin-coated micropatterns overnight to form spheroids with diameters 110 ranging from 50 μm to 150 μm (Figure S1B), resulting in spheroids with cell 111 numbers ranging from 30 to 100 (Figure S1C). We added collagen hydrogel 112 1 mg/ml concentration on top of the spheroids and applied a metal or glass ring 113 to confine the 3D matrix, preventing collagen hydrogel shrinkage during the cell 114 culture process (see Methods). Finally, a compressive force ~5% strain 115 (referred as 1xload) and ~15% strain (referred as 2xload) was added on top of 116 the collagen hydrogel (Figure S1A, S1D and S1E) followed by culturing for 48 117 hours. 118 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint After two days of culture, we evaluated fibroblast activation using 119 immunofluorescence markers such as ɑSMA and pMLC (Figure 1B and C). We 120 found that cells in the single cell 3D model showed low expression of ɑSMA, 121 when compared to the 3D spheroid model (Figure 1B), although the F -actin 122 expression level is higher in the single cell model as shown in Figure 1B. These 123

suggest that the activation of fibroblasts was more pronounced in 124 spheroid models, highlighting the importance of compressive forces in 3D 125 spheroid to regulate cellular function. In subsequent studies, we will therefore 126 use the spheroid model to assess its applications to cellular rejuvenation. 127 Under compressive force conditions, both old and young fibroblasts became 128 activated, as indicated by the increased level of ɑSMA compared to the unload 129 group (Figure S3A and Figure S3B). The increased level of pMLC also implies 130 that these cells are in an active state (Figure 1C). Since senescence is a 131 hallmark of aging, we sought to determine if the application of compressive 132 force alters senescence -associated properties. Towards this we performed 133 beta-galactosidase staining and found fewer positively stained cells in the 134 2xload group, suggesting that the applied load reduces senescence of aged 135 cells as shown in Figure 1D. Next, we measured the persistence of the 136 activated state of the fibroblasts in the 2xload group. To do this we removed the 137 load after culturing the cells in 3D under load condition for 48 hours and 138 continued to culture for 5 days as shown in Figure 1E and 1F. We found that 139 the ɑSMA level increased at day 1 and then decreased at day 4 and 5 (Figure 140 1F) suggesting that fibroblasts were transiently activated. Further, many of the 141 cellular and nuclear morphometric parameters that changed with compressive 142 force were also reversed upon the removal of compressive force (Figure 1F and 143 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure S3C and S3D). In summary, the increased cell contractility, as 144 evidenced by elevated ɑSMA and pMLC levels, along with reduced senescence 145 under compressive force conditions and the ability to revert back to a non -146 activated state after removing the load, suggesting that our spheroid model with 147 load application has the potential to induce rejuvenation properties in aged 148 cells. 149 Mechanical force stabilizes microtubules and facilitates chromatin 150 remodeling. 151 Since cellular aging is accompanied by alterations in cytoskeletal and chromatin 152 remodeling, we next evaluate the role of compressive forces on such 153 cytoskeletal and chromatin remodeling. Upon application of compressive 154 forces, microtubule reorganization was more evident as shown in Figure 2A, 155 compared to F -actin (Figure S2E) . α -tubulin intensity, a component of 156 microtubules, is higher after the application of compressive load , and the 157 microtubule network is much more complex compared to the unload group 158 (Figure S9A-C). In addition, Lamin A/C, a nuclear protein playing a key role in 159 force transmission from the cytosol into the nucleus (14), showed not much 160 significant differences between the two groups (Figure 2A). Lamin B interacts 161 with chromatin and contributes to nuclear organization, by anchoring 162 heterochromatin to the nuclear periphery (15). Compared to unload group, 163 Lamin B intensity increased in 2xload group (Figure 2A and Figure S6C). 164 Cells undergo chromatin remodeling in response to mechanical forces as a 165 protective mechanism to maintain genome integrity (16). Our subsequent 166 investigation focused on chromatin organization, highlighting key players such 167 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint as H3K9me3 (Histone 3 Lysine 9 Trimethylation), H3K4me3 (Histone 3 Lysine 168 4 Trimethylation) and HP1a (Heterochromatin Protein 1 alpha). H3K9me3 and 169 H3K4me3 are commonly associated with distinct chromatin regions, with 170 H3K9me3 linked to heterochromatin and H3K4me3 associated with 171 euchromatin (17, 18), while HP1a, binding to H3K9me3, plays a crucial role in 172 forming and maintaining heterochromatin structures (19). In line with these 173 findings, compressive forces on HDFs resulted in increased levels of H3K9me3, 174 and HP1a, and decreased levels of H3K4me3, as indicated in Figure 2B and 175 Figure S6A, S6B. Interestingly, upon removal of the load, HP1a level exhibited 176 an increase trend as shown in Figure S6B. Consistent with Figure 2B, high 177 resolution DAPI -stained images of nuclei showed increased puncta like 178 structure in 2xload group and an analysis of nuclear morphology and chromatin 179 intensity features demonstrated condensed chromatin under mechanical load 180 (Figure 2C) and an increased ratio of heterochromatin to euchromatin (Figure 181 2D, Figure S6D and TableS4). Taken together, these findings suggest that 182 mechanical force not only stabilizes microtubules but also contributes to 183 increased chromatin condensation possibly contributing to genome stability. 184 Mechanical force enhances cell migration primarily through nucleus -185 cytoskeleton axis 186 Upon observing the activation of fibroblasts and increased microtubule 187 organization with compressive forces, we subsequently investigated its role on 188 cell migration. Since cellular aging results in reduced cell migration, the goal of 189 these experiments was to assess if mechanical forces increased cell migration, 190 as possible routes to cellular rejuvenation. In both old and young groups, under 191 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint compressive force conditions, cell migration was significantly higher than in the 192 unloaded groups, as shown in Figure 3A and Figure S2A. In addition, cell 193 migration speed in the aged cells increases with the applied load, as depicted 194 in Figure S2B. Figure S2C and S2D also shows cell migration enhanced with 195 compressive load and after removal of the load. Since cellular perception of 196 compressive forces are transduced via membrane proteins such as G protein-197 coupled receptors (GPCRs), Piezo, integrins, and calcium channels, and 198 transduce these signals into the nucleus via cytoskeletal components including 199 actin, microtubules, and intermediate filaments (20, 21), we carried out a small-200 scale drug screen to identify critical signaling intermediates in our compressive 201 force induced HDF activation and migration. 202 In Figure 3 B-E (large area as shown in Figure S 4A-C), we applied several 203 inhibitors to perturb possible intermediators shown in Figure 3C. We found that 204 Latrunculin A, Nocodazole, and PD98059 are the three most effective inhibitors 205 which inhibit cell migration among these interventions. Apart from the above-206 mentioned inhibitors, the Y27632 (ROCK inhibitor) group reduces the cell 207 migration area but the cell number increases. On the other hand, in the PF -208 573228 (FAK inhibitor (22)) group, cell migration was not significantly affected 209 compared to the 2xload group. Collectively, we identified that inhibition of actin, 210 microtubules and ERK pathway had a critical role in force induced HDF 211 activation and migration. Given the specific roles of ERK pathway in aging and 212 rejuvenation, we next assessed the interplay between compressive forces, 213 chromatin organization, DNA damage and transcription control with a particular 214 focus on ERK inhibition. 215 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Mechanical load orchestrates chromatin reorganization, and diminishes 216 DNA damage through ERK Signaling 217 Based on our findings showing that PD98059 inhibits cell migration to a greater 218 extent, our next step is to investigate the effect of ERK inhibitor on chromatin 219 remodeling and gene expression. As depicted in Figure 4A, 220 immunofluorescence of pERK reveals that ERK undergoes phosphorylation 221 and translocates into the nucleus upon compressive forces. After adding drug 222 PD98059, pERK level decreased in the nucleus as shown in Figure 4B (images 223 are shown in Figure S8). 224 Next, we measured H3K9me3 and H3K4me3 levels and observed an increase 225 of these markers in the PD98059 group as shown in Figure 4C and Figure S6A. 226 Increase in H3K9me3 and H3K4me3 suggests that ERK plays a role in the 227 regulation of chromatin organization through histone modification. Lamin B 228 expression level decreased after adding PD98059 in Figure 4C and Figure 229 S6C. This suggests that the nuclear translocation of pERK may potentially 230 affect chromatin structure and the activity of transcriptional regulators. 231 We then examined whether cells experienced DNA damage by assessing 232 γH2AX, a well -established DNA damage marker (23). Surprisingly, our 233 observations in Figure 4D revealed a notable mitigation of cellular DNA damage 234 under mechanical load as indicated in foci number. In Figure S7A, where the 235 fixed unload group and fixed load group serve as control groups, DRAQ7 236 staining data revealed the appearance of dead cells in the spheroid center 237 under compressive force conditions. In Figure S7B, upon addition of the drug 238 PD98059, we observed an increase in the level of γH2AX in the unload 239 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint condition, indicating an elevation in DNA damage. To further investigate the 240 impact of mechanical load on DNA damage response, we introduced cisplatin, 241 a known inducer of DNA damage. Remarkably, in the 2xload+cisplatin group, 242 cellular damage decreased compared to the unload+cisplatin group (Figure 243 S7C). 244 The observed protective effects against DNA damage prompts the hypothesis 245 that compressive force promotes DNA repair. Next, we sought to investigate 246 whether ATM, a key protein kinase, is involved in the cellular response to DNA 247 damage under force conditions. Surprisingly, in our study (Figure 4D), the 248 administration of KU55933 (an ATM inhibitor) did not significantly affect DNA 249 damage compared to 2xload group, and cell migration remained unaffected 250 (Figure 3D and 3E). Similarly, there were no statistically significant changes 251 observed in the γH2AX levels between the 2xl+PD98059 group and the load 252 groups. Moreover, the γH2AX level in the 2xl+PD98059 group shows a non -253 significant decreasing trend compared to the control group (Figure 4D). This 254 prompts the hypothesis that compressive force prevents DNA damage not via 255 an ATM-dependent ERK signaling pathway, but possibly through a mechanism 256 involving physical force-induced chromatin interactions. Next, we investigated 257 whether DNA damage affects the activation properties under mechanical force 258 using immunofluorescence levels of ɑSMA. As shown In Figure 4E, the 259 immunofluorescence images from all four conditions showed lower levels of 260 ɑSMA in cisplatin groups compared to compressive load conditions. These 261 findings underscore the intricate involvement of ERK in governing cellular 262 processes, ranging from chromatin organization to DNA damage response. 263 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Coupling between ERK signaling and differential gene expression upon 264 compressive forces 265 To further characterize transcription profiles and gene expression changes 266 associated with enhanced migration, rejuvenation, and ERK signaling 267 pathways, we conducted global RNA sequencing analysis under different 268 conditions as shown in Figure S10A -D. From the RNA -Seq analysis, we 269 observed significant upregulation of 278 genes in the 2xl group compared to 270 the unload condition (fold change > 2, adjusted p value < 0.1) (Figure 5A), and 271 612 genes were upregulated in the 2xl group compared to the PD98059 group 272 (Figure 5B). Gene Ontology (GO) biological process analysis of these 273 differentially expressed genes (278 and 612 DEGs) revealed enrichment in cell 274 migration along with other important cellular processes (Figure 5A and 5B). 275 Kyoto Encyclopedia of Genes and Genomes (KEGG ) pathway analysis of the 276 278 DEGs highlighted the involvement of Mitogen -activated protein kinases 277 (MAPKs) during this process (Figure 5C and Figure S12A). GO cellular 278 component analysis of the 278 DEGs revealed that they are associated with 279 extracellular matrix, while GO molecular function analysis indicated enrichment 280 in growth factor activity, extracellular matrix structural constituent, and 281 metallopeptidase activity (Figure S10E). Similarly, KEGG analysis of the 612 282 DEGs validated the importance of MAPKs in this process (Figure 5C and Figure 283 S12B). GO cellular component analysis for the 612 DEGs showed enrichment 284 in the extracellular matrix, while GO molecular function analysis highlighted 285 their association with growth factor receptor activity (Figure S10F). The overlap 286 of 119 genes between the 278 DEGs and 612 DEGs showed consistent results 287 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint in terms of GO biological process analysis, GO cellular component analysis, 288 GO molecular function analysis, and KEGG pathway analysis (Figure S10D and 289 S10G). Additionally, FOXO signaling pathway involvement was noted in KEGG 290 pathway analysis, underscoring its significance in rejuvenation. 291 Subsequent analysis of gene expression using DE -seq normalized results 292 revealed changes in several genes. We divided these into groups based on 293 their enrichment in the functional pathways such as ERK -related genes, 294 microtubule-related genes, migration-related genes, DNA repair-related genes, 295 and rejuvenation-related genes (Figure 5D). Among the ERK -related genes, 296 downregulation was observed in most genes in the PD98059 group, while 297 upregulation of Insulin Receptor (INSR), Son of Sevenless homolog 1 (SOS1), 298 Ribosomal Protein S6 Kinase A1 (RPS6KA1), B-Raf Proto-Oncogene (BRAF), 299 Epidermal Growth Factor ( EGF), and Epidermal Growth Factor Receptor 300 (EGFR) in the 2xload group suggested their involvement in ERK activation, 301 directly or indirectly. Microtubule -related genes exhibited diverse expression 302 patterns, notably characterized by the downregulation of STMN1, which is 303 associated with microtubule destabilization, and the upregulation of MAPRE3 304 and RAC1 in the 2xload group, influencing microtubule dynamics and 305 organization. Migration -related gene analysis, including ECM -related genes, 306 mechanosensor-related genes, and Rho signaling pathway -related genes, 307 using qRT-PCR, revealed consistent trends with part of RNA-seq data analysis 308 (Figure S11). DNA repair -related mechanisms encompassing base excision 309 repair, nucleotide excision repair, homologous recombination, mismatch repair, 310 Fanconi anemia pathway, and non -homologous end -joining were examined 311 (KEGG pathway information reviewed in Figure S13 A-F). Notably, upregulation 312 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint of USP1, RMI2, and FANCE in the 2xload group, associated with the Fanconi 313 anemia pathway, was observed among DNA repair -related genes. 314 Rejuvenation-related gene analysis revealed the upregulation of MBD2, KL, 315 SIRT1, NFE2L2, LBR, and FOXO3, alongside the downregulation of MTOR in 316 2xload group, all of which are associated with longevity (KEGG pathway 317 information reviewed in Figure S14 A-D). 318 In summary, our RNA -seq results align with the observations of enhanced 319 migration, involvement of the ERK signaling pathway, and cellular rejuvenation. 320 Compressive forces on implanted spheroids in an FT AGED skin model 321 show aged fibroblast activation. 322 To explore the potential applications in translational medicine, we utilized an 323 artificial aged skin tissue model to investigate whether aged fibroblasts could 324 be activated (Figure S15). We injected either single cells or spheroids into the 325 skin tissue, followed by the application of compressive force or no force as a 326 control. Cells were localized around the injection site, as shown in Figures 327 S16A and S16B. Compared to the control group (without cell injection), we 328 observed relatively higher levels of αSMA and collagen I protein in the 329 experimental groups with cell injections (Figures S17 and S18). Under 330 compressive force conditions, collagen I expression was higher in the spheroid 331 group compared to the single -cell group (Figure 6A). Similarly, elastin protein 332 levels in the spheroid group under compressive force showed an increasing 333 trend compared to the single -cell group under the same conditions (Figure 334 S18). However, under the current experimental conditions, the levels of 335 fibronectin and elastin were much less upregulated with compressive force 336 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint compared to collagen 1 secretion (Figures S17 and S18). 337 338

339 340 In summary, this paper presents a new approach for activating/rejuvenating 341 aging dermal fibroblasts using compressive forces. We selected 5% and 15% 342 strain as indicators of compressive force, based on the assumed Young's 343 modulus of collagen hydrogel around 100 Pa (Figure S1D) (24–26). This differs 344 from other studies that use pressure as an index (10, 11, 13, 27, 28). Our 345 previous work demonstrated that compressive force induces actin 346 depolymerization and leads to transcriptional quiescence at 2D single cell level 347 (13). However, in 3D, compressive forces have been shown to activate cancer 348 cell migration (11). In this paper, we hypothesized that by forming spheroids of 349 aged cells which lead to the depolymerization of cytoskeletal filaments, may 350 trigger activation pathways upon the application of compressive forces. 351 In particular, cells and nuclei adopt a relatively rounder morphology, compared 352 to polarized single cells, when clumped in the designed pattern (Figure S1C). 353 In this configuration, they exhibit a soft, highly sensitive state, poised to respond 354 to external cues and initiate cellular processes. When such cells experience 355 external forces at a global level, the highly viscoelastic cytoskeletal organization 356 can activate cytoskeletal remodeling pathways (21, 29). This process entails 357 the dynamic reorganization of actin filaments, microtubules, and intermediate 358 filaments, forming networks that serve as mechanotransducers. These 359 networks could facilitate the direct transmission of mechanical signals from the 360 extracellular environment to the nucleus via the linker of nucleoskeleton and 361 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint cytoskeleton (LINC) complex, ultimately influencing downstream gene 362 transcription (8, 9). Conversely, the nucleus, which exhibits active rheological 363 properties, can function as the primary mechano -sensor when subjected to 364 sudden compressive forces (30). In our model, we observed a clear increase in 365 cell contractility, as indicated by elevated levels of phosphorylated myosin light 366 chain (pMLC). Additionally, we noted a complex reorganization of the 367 microtubule network, accompanied by an increase in Lamin B expression and 368 enhanced chromatin remodeling marked by increased levels of H3K9me3 and 369 HP1a, while H3K4me3 decreased. These findings suggest that mechanical 370 forces trigger cytoskeleton reorganization and chromatin remodeling. 371 Previous study has demonstrated that mechanical forces increase ɑSMA 372 expression and its incorporation into actin filaments by activating two distinct 373 signaling pathways: Rho/Serum Response Factor (SRF) and Mitogen -374 Activated Protein Kinase p38 (MAPK p38)/SRF (31). Consistent with these 375 findings, our results indicate an increase in αSMA expression under 376 compressive loading conditions, as shown in Figure 1B. As cells age, it is 377 commonly observed that H3K9me3 levels decrease while H3K4me3 levels 378 increase, accompanied by decreased HP1a levels and reduced LaminB 379 expression (1). Surprisingly, in our novel model, we observed a converse trend. 380 These findings suggest that the clumped HDFs (spheroids), upon sensing 381 compressive forces, initiate fibroblast activation pathways including their 382 potential rejuvenation, a notion further supported by our beta -galactosidase 383 staining results. 384 Previous research has elucidated the role of mechanical compression in 385 regulating cancer cell migration through the MEK1/ERK1 signaling pathway 386 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint (12). In this study, compressive force enhanced cell migration, but whether this 387 migration is driven by microtubules is unknown. During migration, cells extend 388 filopodial or lamellipodial protrusions, form focal adhesions, and undergo 389 cytoskeletal reorganization to generate force (32, 33). ERK phosphorylates key 390 proteins involved in these processes, including microtubule-associated proteins 391 (MAPs), focal adhesion kinase (FAK), calpain, and myosin light chain kinase 392 (MLCK), thereby providing essential cues for cell migration (34–37). 393 Additionally, phosphorylated ERK functions as a mechanosensory transcription 394 factor capable of shuttling between the nucleus and cytosol to regulate gene 395 expression, thereby exerting influence over cell migration dynamics (38). In our 396 study, we demonstrated the significance of the force transmission pathway 397 between the cytoskeleton and nucleus, as well as ERK signaling, in influencing 398 cell migration. Employing a targeted approach, we conducted small -scale 399 inhibitor screening and utilized a dominant -negative (DN) KASH construct to 400 dissect the specific pathways involved. Our findings revealed that inhibitors 401 targeting the cytoskeleton, including Blebbistatin, Cytochalasin D, Latrunculin 402 A, Nocodazole, and Withaferin A, significantly impeded cell migration (Figure 403 3B-E). This underscores the critical role of the cytoskeleton in mediating force 404 transmission essential for cell motility. Moreover, disruption of the LINC 405 complex by the DN KASH (Figure 3B -E) construct corroborated these 406 observations, further emphasizing the importance of the cytoskeleton -nucleus 407 connections in governing cell migration dynamics. Inhibition of ERK1/ERK2 408 with PD98059 resulted in a notable decrease in cell migration distance. This 409 highlights the pivotal role of ERK signaling in cell migration. In contrast, 410 inhibition of FAK with PF -573228 had minimal effects on cell migration, 411 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint highlighting the differential roles of focal adhesion-related signaling pathways. 412 413 Upon activation, ERK undergoes phosphorylation, resulting in its transformed 414 state, pERK, which then translocates into the nucleus to regulate histone 415 modifications. While a previous study suggested that ERK activation increases 416 H3K4me3 levels in 2D culture conditions (39), interestingly in our 3D model with 417 compressive forces, inhibition of ERK led to an increase in H3K4me3. We 418 propose a possible mechanistic explanation for this observation: under 419 compressive force conditions, perturbation of ERK affects cytoskeletal 420 structure, resulting in reduced cytoskeletal formation and increased softness of 421 the nucleus, as indicated by decreased laminB levels. This could result in higher 422 levels of both H3K4me3 (a euchromatin marker) and H3K9me3 (a 423 heterochromatin marker), reflecting a global change in chromatin architecture 424 in response to ERK perturbation under compressive force conditions. 425 426 Our RNA -seq data revealed that under compressive force conditions, the 427 expression of genes such as INSR, BRAF, EGF, and EGFR increased. These 428 genes are upstream regulators of the ERK pathway, which initiates ERK 429 signaling cascades in response to various extracellular stimuli (36). 430 Additionally, the expression of RPS6KA1, also known as RSK1, increased. 431 RPS6KA1 is a downstream effector of the ERK pathway and is involved in 432 mediating cellular responses to ERK activation. Several studies have 433 highlighted the role of mechanical force in regulating key signaling molecules 434 such as the insulin receptor, BRAF, and EGFR (40–42). Our study presents 435 novel findings demonstrating that mechanical force activates aged dermal 436 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint human fibroblasts through the ERK signaling pathway. As mechanosensory 437 genes, we observed an upregulation of YAP1, TRPV4, and PIEZO2 gene 438 expression levels under 2x load conditions compared to the unloaded state, 439 while PIEZO1 was downregulated under 2x load conditions. Intriguingly, when 440 the ERK inhibitor PD98059 was applied, TRPV4 and PIEZO1 gene expression 441 levels were further inhibited, indicating a potential link between ERK and 442 TRPV4/PIEZO1 in our model. Consistently, previous research has 443 demonstrated that mechanical forces can regulate TRPV4 and PIEZOs (43). 444 Moreover, another study has shown that mechanical force can activate TRPV4, 445 subsequently leading to the induction of the ERK signaling pathway (44). These 446 findings align with our observations and support the notion of a mechanistic 447 relationship between mechanical force, TRPV4/PIEZO1, and activation of ERK 448 signaling pathway. Upon inhibiting ERK, we observed the upregulation of 449 CLASP2, MAP2, MAPT, and HSPA1A genes, which are known to be involved 450 in microtubule dynamics. This suggests the presence of a compensatory 451 response or feedback mechanism triggered by the inhibition of ERK signaling, 452 highlighting the intricate interplay between mechanical force, ERK signaling, 453 and cellular responses related to microtubule dynamics. 454 We observed a reduction in DNA damage under compressive force conditions, 455 as indicated by decreased γH2AX immunostaining, suggesting enhanced DNA 456 repair mechanisms. Our RNA sequencing data revealed the involvement of the 457 Fanconi anemia pathway, known for anti-oxidative stress (45). This reduction in 458 DNA damage aligns with the activation or rejuvenation required for aging cells. 459 Previous study demonstrated decreased DNA damage under load conditions 460 attributed to heterochromatin organization and low levels of H3K9me3 (16). 461 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint However, our results showed an upregulation of H3K9me3, potentially due to 462 differences in cell culture models (2D vs. 3D). Notably, we observed no change 463 in nucleus stiffness, as indicated by Lamin A/C immunostaining data. In our 464 RNA sequencing data, rejuvenation-related genes such as MBD2, KL, SIRT1, 465 NFE2L2, LBR, and FOXO3 were upregulated. MBD2 can reduce CpG 466 methylation levels, delaying aging (46). KL has been shown to improve 467 cognitive function, serving as a longevity factor (47). SIRT and FOXO3 are 468 known to slow cellular senescence (48), while NFE2L2 acts as a transcription 469 factor sensitive to reactive oxygen species (ROS) and nitric oxide (NO), induced 470 by exercise, and protects cells against cytotoxic and oxidative damage (49). 471 Upregulation of LBR can increase cell proliferation and suppress genomic 472 instability, supporting the rejuvenating proces s (50). Upon load removal, cells 473 maintained their migration behavior, initially exhibiting sustained high levels of 474 αSMA followed by a decrease in our model, along with sustained increased 475 expression of HP1 a. The persistent expression of HP1 a indicates that 476 chromatin organization remains unchanged even after load removal within 5 477 days. The sustained expression of αSMA initially followed by a decrease in our 478 model represents an intriguing finding, particularly considering that prolonged 479 αSMA expression is associated with fibrosis. In the skin tissue model, after two 480 days of incubation, the spheroid group exhibited higher secretion of collagen 1 481 compared to the single-cell group, highlighting the significance of compressive 482 force induced tissue regeneration properties (Figure 6A). 483 Collectively, the activation of fibroblasts upon compressive force, cytoskeletal 484 and chromatin remodeling, and transition from a mesenchymal to collective 485 migration mode of aged HDFs may signify cellular rejuvenation (Figure 6B), 486 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint crucial for both tissue regeneration and wound healing and could serve as a 487 valuable platform for drug screening. 488

489 490 Fabrication of micropatterned PDMS stamps and microcontact printing 491 Polydimethylsiloxane (PDMS, SYLGRAD TM 184 Silicone Elastomer Kit) 492 elastomer is prepared by blending the base and curing agent at a 10:1 ratio. A 493 typical quantity of 20 -25g proves sufficient to entirely coat the custom wafer 494 surface (1,800 μm2 rectangles, (aspect ratio 1:5), distance between rectangles 495 is 500μm). PDMS is poured onto the wafer, and subjected to degases within a 496 vacuum chamber for 30 minutes until the absence of air bubbles on the surface 497 is achieved. Subsequently, the curing process is initiated at a temperature of 498 60°C for a duration of 3 hours. Following the cooling phase, the PDMS material 499 is carefully detached from the substrate using a pair of tweezers. It is then 500 sectioned into a round 1cm 2 square pieces and stored within clean containers 501 to avert the accumulation of particulate matter. The PDMS surface is activated 502 by a Plasma machine (Henniker Plasma, HPT -200). Briefly, O2 gas was used 503 by exposing the stamps to it for 1.5 min at 75% power, with pressure 0.4mbar. 504 A mixture solution was prepared which includes fibronectin (MERK, F1141) and 505 protein labeling kit (invitrogen, A20170A) in PBS at a concentration of 10% and 506 3%, respectively. For microcontact printing (mCP), a 10ul fibronectin mixture 507 solution is applied onto the PDMS surface and observed under microscope for 508 appropriate drying. After the fibronectin deposited on PDMS was dried, it was 509 then stamped onto the un -coated IBIDI dishes (ididi, µ -Dish 35 mm, high, 510 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint uncoated Cat.No:81151) and pressed gently using tweezers after which the 511 PDMS is then carefully lifted. Finally these imprinted micro patterns were 512 observed under EVOS fluorescence microscope..After careful selection of the 513 dishes, they were passivated using 0.2% pluronic acid (Sigma, P2443) for 10 514 min followed by washing with PBS three times before seeding the cell. 515 Cell culture 516 GM08401 (75 years old) and GM09503 (10 years old) healthy human dermal 517 fibroblast cells (male origin) were obtained from the NIGMS Human Genetic 518 Cell Repository at the Coriell Institute for Medical Research. The HDFs are 519 cultured in MEM (Gibco, 11090 -081) with 1 5% FBS (Thermo Fisher, 520 16141079), 1 % P/S (Penicillin and Streptomycin) (PAN BIOTECH, P06 -521 07300), 1% Glutamax (100x, Gibco, 35050-038) and 1% NEAA (100x, Gibco, 522 11140-035) under 5% CO2 and 37 °C. HEK293T cells (gift from Dr. Deborah 523 Walter) were cultured in high -glucose DMEM supplem ented (BioConcept, 1 -524 26F03-1) with 10% (v/v) fetal bovine serum (Dominigue Dutscher, S1900-500B) 525 and 1% P/S. 526 527 Application of static compressive force in 3D culture model. 528 70,000 old HDF cells were seeded on fibronectin -coated micropatterns in an 529 ibidi dish overnight to form spheroids. 1mg/ml Collagen gel mixture was 530 prepared from Collagen type I from rat tail (Gibco, A1048301) according to the 531 manufacturer protocol. 400ul of 1mg/ml collagen hydrogel was applied on top 532 of the spheroid and allowed for the polymerization for 1 hour in the incubator at 533 37 °C. After 1h incubation metal ring or glass ring was placed on top of collagen 534 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint gel to avoid shrinkage during prolonged culture conditions. Glass coverslip 535 (VWR, 631-1577, 12mm round) was used for compressive force and placed 536 carefully without disrupting the gel inside the ring. For 5% applied compressive 537 force 3 stacked coverslips and for 15% 7 stacked coverslips were used (Figure 538 S1D). 539 Immunostaining 540 Collagen hydrogel samples were fixed with 4% paraformaldehyde (Merk, 541 F8775-25ML) for 1h and the coverslip, used to apply compressive forces, was 542 then removed carefully. Gels are then washed with 100mM glycine (Roth, 543 Nr.3790.3) three times to prevent excess fixation. Permeabiliz ation was done 544 for 20 minutes with 0.8% Triton X -100 for γH2AX and ERK staining and 545 0.5%Triton X-100 was used for the rest of the markers. This was followed by 546 washing with 100mM glycine for three times. Samples were then blocked with 547 10% NGS (Abcam, ab7481) in wash buffer PBS (PanReac Applichem, A0964 548 9050) containing 0.2%Triton and 0.2% Tween20 (Sigma, SLBZ8563) for three 549 hours at room temperature (For γH2AX and ERK staining PBS containing 550 0.3%Triton and 0.2% Tween20 wash buffer used for blocking ). Primary 551 antibody staining was done with 10% goat serum in the wash buffer for two 552 days at 4 °C. Next day, the gels were washed with a buffer ( PBS containing 553 0.2%Triton and 0.2% tween20 ) for 10 -15 minutes each wash three times. 554 Secondary antibody staining 1:300 dilution was done in 5% goat serum in the 555 wash buffer and incubated for three hours at room temperature. Followed by 556 washing with a wash buffer for 15 minutes once and then twice with PBS 15 557 minutes each. D API (Thermo Fisher Scientific, R37605), nucleus stain, and 558 ActinGreen (Thermo Fisher Scientific, R37110), actin stain, was incubated in 559 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint PBS overnight at 4°C and washed three times with PBS. Finally, 100 ul of PBS 560 was added and imaged using Nikon confocal imaging system. Antibodies used 561 in this paper are listed in Supplementary Table 2. For the beta -galactosidase 562 assay, CellEvent™ Senescence Green Flow Cytometry Assay Kit (Invitrogen, 563 C10840) was used as per manufacturer protocol and confocal images were 564 captured. DRAQ7 (Biolegend, 424001) was used to discriminate between 565 live/dead cells. 566 567 Drug treatment 568 All the drugs with specific concentrations used in our assays are mentioned in 569 Figure S1E and supplementary table 3. 1 ml medium with the respective drug 570 concentrations were added to the overnight spheroids and incubated for 1h, 571 before covering it with the collagen gel. After 1h of gel polymerisation, 2ml of 572 new complete medium with respective concentration of drug was added and 573 incubated for 2 days. Finally, these gels were processed for immunostaining 574 and imaging. 575 576 Real-time PCR assay. 577 For RNA purification, at least 10 gels from aged fibroblasts (with and without 578 load) were used. Single cells were isolated from these gels using collagenase 579 at a concentration of 2 mg/ml (Merk, C0130) and incubated at 37 0C for 30 580 minutes. After centrifugation at 1000 rpm for 4 minutes, the supernatant was 581 removed and pellet was collected to be processed for RNA isolation using 582 RNeasy Plus Micro Kit (QIAGEN, 74034). cDNA was prepared using iScript ™ 583 cDNA Synthesis Kit (BIO-RAD, 1708890). Real-time PCR was done using Sso 584 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint advanced SYBER mix (BIO -RAD, 1725274). Relative fold change was 585 calculated with the 2^ -ΔΔCT method using GAPDH as a housekeeping gene 586 for normalization. All primers used in this study are shown in Supplementary 587 Table 1. 588 589 RNA seq analysis 590 RNA library preparation and sequencing was performed at Genomics Facility, 591 ETH Zurich in Basel. Unload group, 1XL group, and 2XL group in triplicates 592 were performed by NovaSeq S4 PE 2x101bp. The PD98059 group in triplicates 593 was performed by NextSeq PE 2x38bp. Standard pipelines such as DEseq 594 were used for RNA seq analysis (51, 52). To summarize, the paired-end reads 595 were aligned to the human genome GRCh38.84 from UCSC. Reference 596 genomic indexes using the HISAT2 sequence -alignment tool (version 2.2.1) 597 was used. The cloud indexes (grch38_trans) for HISAT2, was accessed on 598 June 25th, 2020 from https://registry.opendata.aws/jhu -indexes. Combining 599 reads from four technical replicates for each biological sample served as the 600 input for HISAT2, utilizing default parameters. Subsequently, single aligned 601 reads were enumerated using htseq -count (version 1.99.2). The counts for all 602 expressed genes were then employed for the differential expression analysis 603 and analyzed using DESeq2 (Version 1.36.0). Batch information was 604 incorporated into the DESeq2 design formula. Differentially expressed genes 605 were identified based on adjusted P values (Benjamini –Hochberg) below 0.1 606 false discovery rate (FDR) and fold change above or below 2. Enrichment 607 analysis was performed using ShinyGO (version 0.80). Python script was 608 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint utilized for generating heat maps that compare gene expression across 609 different biological conditions, based on DESeq2 normalized counts. 610 In vitro 3D reconstructed skin models under compressive force 611 In this study, we used the Phenion FT AGED skin model as a substitute for 612 aged human skin. This model includes senescent fibroblasts, reduced ECM 613 proteins (such as collagen and elastin), and elevated MMP secretion due to 614 treatment with mitomycin C. We divided the samples into six groups, as shown 615 in Figure S15A. Single-cell injection refers to the collection of cells from 2D cell 616 cultures. Spheroid injection involves collecting cells following the method 617 described above. The elastic modulus of the reconstructed skin model was 618 approximately 7kPa (53), and deformation of up to 12.6% was achieved under 619 compressive force, as indicated in Figure S15B. Cells were injected at three 620 different points, with a concentration of ~70,000 cells per point and an injection 621 volume of 50 µL. The injection sites were located approximately 2 mm from the 622 center, and a wound was made at one site to indicate the direction (Figure 623 S15C). The reconstructed skin models were cultured in a specific Air -Liquid 624 Interface Culture Medium provided by the supplier. 625 626 Cryo-sectioning and immunofluorescence of tissue sections 627 After two days of culture, the tissues were placed in cryomolds and embedded 628 in OCT medium (Leica Biosystem, 14020108926). Samples were cryo -629 sectioned at a thickness of 20 µm at -15 °C using a cryo-microtome and stored 630 at -80 °C until staining. For immunostaining, the tissue sections were fixed in 631 pre-cooled acetone (VWR, 20063.296) for 15 minutes at -20 °C. After air-drying 632 for 5 minutes, a PAP pen (Sigma -Aldrich) was used to encircle the tissue. 633 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Sections were then blocked with 10% goat serum for 1 hour. Subsequently, the 634 samples were incubated with primary antibodies diluted in 1% BSA and 0.3% 635 Triton-X 100 in PBS overnight at 4 °C. After three washes in PBS (5 minutes 636 each), the sections were incubated with secondary antibodies diluted in 1% 637 BSA and 0.3% Triton-X 100 in PBS overnight at 4 °C. Following another three 638 washes in PBS, the samples were stained with Hoechst 33342 in PBS (one 639 drop per 1ml) for 1 hour at room temperature. Finally, the sections were 640 mounted with ProLong Gold Antifade Reagent (Thermo Fisher Scientific), 641 covered with a coverslip, and sealed at the edges with a thin layer of nail polish. 642 The slides were stored at 4 °C until imaging. 643 Image acquisition and analysis 644 All confocal images were obtained using the Nikon confocal ti2 imaging system. 645 Briefly, collagen hydrogel was imaged using a 40X oil immersion objective NA 646 1.25 or 60 X oil immersion objective NA 1.4. All bright-field images in this study 647 were acquired using EVOS M5000 (Thermo Fisher Scientific) and slide scanner 648 (sysmex) for skin tissue. For the analysis of mean intensity and the γH2AX foci 649 number, Fiji image tool was used. For nuclear marker analysis, DAPI channel 650 was used to generate the mask, whereas for cytosolic protein, either the actin 651 or protein channel was used for mask generation. Nuclear and chromatin 652 features analysis was done using the code from previously published paper 653 (54). The importance of each attribute was measured by Relief F/Gini/Gain ratio 654

via orange software. Internuclear pairwise distance (IPD) analysis was 655 performed using R package dist. Labeled images were processed using the 656 StarDist2D plugin in Fiji. Microtubule meshwork generation and directionality 657 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint histograms analysis were conducted by SOAX software and Fiji software with 658 plugin directionality. For skin tissue samples, Fiji software was used to measure 659 the mean fluorescence intensity and to count cell numbers using the StarDist2D 660 plugin. Normalization was calculated as the ratio of the mean intensity of the 661 immunofluorescence-positive injected area to the cell number. 662 Statistical analysis 663 All plots and statistical analysis were performed with Origin 2024. For box-and-664 whisker plots: The box represents the interquartile range (IQR), encompassing 665 the middle 50% of the data. The bottom of the box marks the first quartile (25th 666 percentile), and the top marks the third quartile (75th percentile). The line inside 667 the box indicates the median (50th percentile). The whiskers extend to the 668 smallest and largest values within 1.5 times the IQR, while outliers are 669 represented by asterisks. Unpaired, two -tailed student -t test was used to 670 compare two groups. One-way ANOVA (Tukey test) was employed to compare 671 groups comprising more than two. 672 Data and materials availability: 673 All codes used in this paper are available from the corresponding author upon 674 request. All illustration graphs shown in this study were created by 675 Biorender.com. All data are available in the main text or the supplementary 676 materials. 677 Acknowledgments 678 We thank GVS group members for their comments on the manuscript and in 679 particular Drs. Nicholas Lawler and Yagyik Goswami for critical reading of the 680 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint manuscript. This work was supported by the Swiss National Science 681 Foundation grant 310030_208046; China Scholarship Council (Grant Number: 682 202008440471). 683 684

685 1. C. López -Otín, M. A. Blasco, L. Partridge, M. Serrano, G. Kroemer, 686 Hallmarks of aging: An expanding universe. Cell 186, 243–278 (2023). 687 2. M. V. Plikus, et al., Fibroblasts: Origins, definitions, and functions in health 688 and disease. Cell 184, 3852–3872 (2021). 689 3. S. Knoedler, et al. , Fibroblasts – the cellular choreographers of wound 690 healing. Front. Immunol. 14, 1233800 (2023). 691 4. K. Takahashi, S. Yamanaka, Induction of Pluripotent Stem Cells from 692 Mouse Embryonic and Adult Fibroblast Cultures by Defined Factors. Cell 693 126, 663–676 (2006). 694 5. B. Roy, et al. , Laterally confined growth of cells induces nuclear 695 reprogramming in the absence of exogenous biochemical factors. Proc. 696 Natl. Acad. Sci. 115 (2018). 697 6. W. Zhang, J. Qu, G.-H. Liu, J. C. I. Belmonte, The ageing epigenome and 698 its rejuvenation. Nat. Rev. Mol. Cell Biol. 21, 137–150 (2020). 699 7. C. C. DuFort, M. J. Paszek, V. M. Weaver, Balancing forces: architectural 700 control of mechanotransduction. Nat. Rev. Mol. Cell Biol. 12, 308 –319 701 (2011). 702 8. C. Uhler, G. V. Shivashankar, Regulation of genome organization and 703 gene expression by nuclear mechanotransduction. Nat. Rev. Mol. Cell 704 Biol. 18, 717–727 (2017). 705 9. G. V. Shivashankar, Mechanical forces and the 3D genome. Curr. Opin. 706 Struct. Biol. 83, 102728 (2023). 707 10. Y. Cui, et al., Cyclic stretching of soft substrates induces spreading and 708 growth. Nat. Commun. 6, 6333 (2015). 709 11. J. M. Tse, et al. , Mechanical compression drives cancer cells toward 710 invasive phenotype. Proc. Natl. Acad. Sci. 109, 911–916 (2012). 711 12. M. Kalli, et al. , Mechanical Compression Regulates Brain Cancer Cell 712 Migration Through MEK1/Erk1 Pathway Activation and GDF15 713 Expression. Front. Oncol. 9, 992 (2019). 714 13. K. Damodaran, et al. , Compressive force induces reversible chromatin 715 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint condensation and cell geometry–dependent transcriptional response. Mol. 716 Biol. Cell 29, 3039–3051 (2018). 717 14. N. Naetar, S. Ferraioli, R. Foisner, Lamins in the nuclear interior − life 718 outside the lamina. J. Cell Sci. 130, 2087–2096 (2017). 719 15. F. Napoletano, et al., The prolyl -isomerase PIN1 is essential for nuclear 720 Lamin-B structure and function and protects heterochromatin under 721 mechanical stress. Cell Rep. 36, 109694 (2021). 722 16. M. M. Nava, et al., Heterochromatin-Driven Nuclear Softening Protects the 723 Genome against Mechanical Stress -Induced Damage. Cell 181, 800 -724 817.e22 (2020). 725 17. T. Montavon, et al., Complete loss of H3K9 methylation dissolves mouse 726 heterochromatin organization. Nat. Commun. 12, 4359 (2021). 727 18. X. Liu, et al. , Distinct features of H3K4me3 and H3K27me3 chromatin 728 domains in pre-implantation embryos. Nature 537, 558–562 (2016). 729 19. W. Zeng, A. R. Ball Jr, K. Yokomori, HP1: Heterochromatin binding 730 proteins working the genome. Epigenetics 5, 287–292 (2010). 731 20. D. M. Rosenbaum, S. G. F. Rasmussen, B. K. Kobilka, The structure and 732 function of G-protein-coupled receptors. Nature 459, 356–363 (2009). 733 21. N. Wang, J. D. Tytell, D. E. Ingber, Mechanotransduction at a distance: 734 mechanically coupling the extracellular matrix with the nucleus. Nat. Rev. 735 Mol. Cell Biol. 10, 75–82 (2009). 736 22. Y. Pan, et al. , Mechanosensor Piezo1 mediates bimodal patterns of 737 intracellular calcium and FAK signaling. EMBO J. 41, e111799 (2022). 738 23. F. K. Noubissi, et al. , Detection and quantification of γ -H2AX using a 739 dissociation enhanced lanthanide fluorescence immunoassay. Sci. Rep. 740 11, 8945 (2021). 741 24. T. Jin, L. Li, R. C. Siow, K. -K. Liu, Collagen matrix stiffness influences 742 fibroblast contraction force. Biomed. Phys. Eng. Express 2, 047002 (2016). 743 25. C. S. Vidmar, M. Bazzi, V. K. Lai, Computational and experimental 744 comparison on the effects of flow-induced compression on the permeability 745 of collagen gels. J. Mech. Behav. Biomed. Mater. 128, 105107 (2022). 746 26. C. R. I. Lam, et al., A 3D Biomimetic Model of Tissue Stiffness Interface 747 for Cancer Drug Testing. Mol. Pharm. 11, 2016–2021 (2014). 748 27. Y. Tang, Y. Fan, Q. Luo, G. Song, Pressure Loading Induces DNA 749 Damage in Human Hepatocyte Line L02 Cells via the ERK1/2 –Dicer 750 Signaling Pathway. Int. J. Mol. Sci. 23, 5342 (2022). 751 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 28. T. Kanazawa, et al., Biological Responses of Three-Dimensional Cultured 752 Fibroblasts by Sustained Compressive Loading Include Apoptosis and 753 Survival Activity. PLoS ONE 9, e104676 (2014). 754 29. I. Andreu, et al. , The force loading rate drives cell mechanosensing 755 through both reinforcement and cytoskeletal softening. Nat. Commun. 12, 756 4229 (2021). 757 30. Y. Song, J. Soto, B. Chen, L. Yang, S. Li, Cell engineering: Biophysical 758 regulation of the nucleus. Biomaterials 234, 119743 (2020). 759 31. J. Wang, R. Zohar, C. A. McCulloch, Multiple roles of α -smooth muscle 760 actin in mechanotransduction. Exp. Cell Res. 312, 205–214 (2006). 761 32. C. Le Clainche, M. -F. Carlier, Regulation of Actin Assembly Associated 762 With Protrusion and Adhesion in Cell Migration. Physiol. Rev. 88, 489–513 763 (2008). 764 33. K. M. Yamada, M. Sixt, Mechanisms of 3D cell migration. Nat. Rev. Mol. 765 Cell Biol. 20, 738–752 (2019). 766 34. V. J. Fincham, Active ERK/MAP kinase is targeted to newly forming cell -767 matrix adhesions by integrin engagement and v -Src. EMBO J. 19, 2911–768 2923 (2000). 769 35. R. E. Harrison, E. A. Turley, Active Erk Regulates Microtubule Stability in 770 H-ras-Transformed Cells. Neoplasia 3, 385–394 (2001). 771 36. H. Lavoie, J. Gagnon, M. Therrien, ERK signalling: a master regulator of 772 cell behaviour, life and fate. Nat. Rev. Mol. Cell Biol. 21, 607–632 (2020). 773 37. C. Huang, K. Jacobson, M. D. Schaller, MAP kinases and cell migration. 774 J. Cell Sci. 117, 4619–4628 (2004). 775 38. D. A. Berti, R. Seger, “The Nuclear Translocation of ERK” in ERK 776 Signaling, Methods in Molecular Biology., G. Jimenez, Ed. (Springer New 777 York, 2017), pp. 175–194. 778 39. C. Esnault, et al., ERK-Induced Activation of TCF Family of SRF Cofactors 779 Initiates a Chromatin Modification Cascade Associated with Transcription. 780 Mol. Cell 65, 1081-1095.e5 (2017). 781 40. J. Kim, D. Bilder, T. P. Neufeld, Mechanical stress regulates insulin 782 sensitivity through integrin -dependent control of insulin receptor 783 localization. Genes Dev. 32, 156–164 (2018). 784 41. A. Hollósi, K. Pászty, M. Kellermayer, G. Charras , A. Varga, BRAF 785 Modulates Stretch-Induced Intercellular Gap Formation through Localized 786 Actin Reorganization. Int. J. Mol. Sci. 22, 8989 (2021). 787 42. B. Sullivan, et al. , Mechanical disruption of E -cadherin complexes with 788 epidermal growth factor receptor actuates growth factor –dependent 789 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint signaling. Proc. Natl. Acad. Sci. 119, e2100679119 (2022). 790 43. M. Zhang, N. Meng, X. Wang, W. Chen, Q. Zhang, TRPV4 and PIEZO 791 Channels Mediate the Mechanosensing of Chondrocytes to the 792 Biomechanical Microenvironment. Membranes 12, 237 (2022). 793 44. P. S. Nayak, et al. , Mechanotransduction via TRPV4 regulates 794 inflammation and differentiation in fetal mouse distal lung epithelial cells. 795 Respir. Res. 16, 60 (2015). 796 45. W. Du, Z. Adam, R. Rani, X. Zhang, Q. Pang, Oxidative Stress in Fanconi 797 Anemia Hematopoiesis and Disease Progression. Antioxid. Redox Signal. 798 10, 1909–1921 (2008). 799 46. A. A. Johnson, et al., The Role of DNA Methylation in Aging, Rejuvenation, 800 and Age-Related Disease. Rejuvenation Res. 15, 483–494 (2012). 801 47. C. Park, et al., Platelet factors are induced by longevity factor klotho and 802 enhance cognition in young and aging mice. Nat. Aging 3, 1067 –1078 803 (2023). 804 48. K. A. Nath, The role of Sirt1 in renal rejuvenation and resistance to stress. 805 J. Clin. Invest. 120, 1026–1028 (2010). 806 49. T. L. Merry, M. Ristow, Nuclear factor erythroid‐derived 2‐like 2 (NFE2L2, 807 Nrf2) mediates exercise ‐induced mitochondrial biogenesis and the anti ‐808 oxidant response in mice. J. Physiol. 594, 5195–5207 (2016). 809 50. A. En, K. Takemoto, Y. Yamakami, K. Nakabayashi, M. Fujii, Upregulated 810 expression of lamin B receptor increases cell proliferation and suppresses 811 genomic instability: implications for cellular immortalization. FEBS J. 812 febs.17113 (2024). https://doi.org/10.1111/febs.17113. 813 51. B. Roy, T. Pekec, L. Yuan, G. V. Shivashankar, Implanting mechanically 814 reprogrammed fibroblasts for aged tissue regeneration and wound 815 healing. Aging Cell 23, e14032 (2024). 816 52. M. I. Love, W. Huber, S. Anders, Moderated estimation of fold change and 817 dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 818 53. D. Malhotra, et al., Linear viscoelastic and microstructural properties of native male 819 human skin and in vitro 3D reconstructed skin models. J. Mech. Behav. Biomed. 820 Mater. 90, 644–654 (2019). 821 54. S. Venkatachalapathy, D. S. Jokhun, M. Andhari, G. V. Shivashankar, 822 Single cell imaging -based chromatin biomarkers for tumor progression. 823 Sci. Rep. 11, 23041 (2021). 824 825 826 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 827 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure 1. Establishment of 3D collagen hydrogel in -vitro model upon 828 compressive forces and characterization activation, rejuvenation and memory 829 properties of the regenerated phenotypes. (A) Schematic of 3D in-vitro collagen 830 model (single cell embedding and spheroid embedding) and illustration of the 831 cell culture process. (B) Representative ɑSMA immunofluorescence confocal 832 images and quantification data per image of mean intensity in single cell model 833 and spheroid. Nucleus is labeled in blue. (Scale bar, 100 μm). (C) 834 Representative pMLC immunofluorescence confocal images and quantification 835 data per image of mean intensity. Nucleus is labeled in blue. (Scale bar, 100 836 μm). (D) Representative Beta -galactosidase staining confocal images and 837 quantification data per image of mean intensity. (Scale bar, 100 μm). (E) 3D 838 nucleus construction and representative ɑSMA immunofluorescence confocal 839 images under load and load removal condition. (Scale bar, 100 μm). (Unit in 840 green box is μm). (F) Quantification data of ɑSMA mean intensity per image, 841 nucleus volume, Z project area of nucleus and roundness of nucleus. All the 842 experiments were repeated at least three times independently with similar 843 results. P values in Figure (B -D) were calculated by unpaired, two -tailed 844 Student’s t test. P values in Figure (F) were calculated by the one-way ANOVA 845

with Tukey’s post hoc test. *P<0.05; **<0.01; ***P<0.001; No asterisks 846 means not significant. Source data are provided as a Source Data file. 847 848 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 849 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure 2. Mechanical force stabilizes microtubule and fa cilitates chromatin 850 remodeling. (A) Representative α -tubulin, Lamin A/C and Lamin B 851 immunofluorescence confocal images and quantification data per cell of mean 852 intensity. Nucleus is labeled in blue. (Scale bar, 50 or 100 μm). (B) 853 Representative H3K9me3, H3K4me3 and HP1a immunofluorescence confocal 854 images and quantification data per nucleus of mean intensity. (Scale bar, 855 100μm). (C) Representative gray images from DAPI in unload condition and 856 load condition. (Scale bar, 50 m). (D) Heatmap of chromatin and nucleus 857 morphology analysis. All the experiments were repeated at least three times 858 independently with similar results. P values in Figure (A and B) were calculated 859 by unpaired, two -tailed Student’s t test. *P<0.05; **<0.01; ***P<0.001; No 860 asterisks means not significant. Source data are provided as a Source Data file. 861 862 863 864 865 866 867 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 868 869 870 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure 3. Mechanical load enhances cell migration via nucleus -cytoskeleton 871 axis. (A) Windrose plots displaying the distance of the migrated cell nucleus to 872 the center of one spheroid. (B) Schematic illustration of inhibitor targets. (C) 873 Table for inhibitors' description. (D) Representative immunofluorescence 874 confocal images to check spheroid spreading. Nucleus is labeled in blue. F -875 actin is labeled in green. (Scale bar, 300 mm). (E) Quantification data of the 876 spread area of the spheroid. All the experiments were repeated at least three 877 times independently with similar results. P values in Figure (E) were calculated 878 by unpaired, two -tailed Student’s t test. Other groups are compared to the 879 2xload group. *P<0.05; **<0.01; ***P<0.001; No asterisks means not 880 significant. Source data is provided as a Source Data file. 881 882 883 884 885 886 887 888 889 890 891 892 893 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 894 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure 4. ERK role in chromatin remodeling, DNA damage response and gene 895 expression regulation. (A) Representative ERK phosphorylation 896 immunofluorescence confocal images and quantification data of mean intensity 897 per cell (some images shown in Figure S8). Nucleus is labeled in blue. α-tubulin 898 is labeled in red. pERK is labeled in magenta (Scale bar, 50 μm). (B) 899 Representative H3K9me3, H3K4me3 and Lamin B immunofluorescence 900 confocal images and quantification data of mean intensity per nucleus. Nucleus 901 is labeled in blue. (Scale bar, 100 μm). (D) Representative γH2AX 902 immunofluorescence confocal images and quantification data of foci number 903 per nucleus. (Scale bar, 100 μm). (E) Representative ɑSMA 904 immunofluorescence confocal images and quantification data of mean intensity 905 per image. Nucleus is labeled in blue. (Scale bar, 100μm). All the experiments 906 were repeated at least three times independently with similar results. P values 907 in Figure (B, D, E) were calculated by the one-way ANOVA method with Tukey’s 908 post hoc test. P values in Figure (C) were calculated by unpaired, two -tailed 909 Student’s t test. *P<0.05; **<0.01; ***P<0.001; No asterisks means not 910 significant. Source data is provided as a Source Data file. "2xL" is an 911 abbreviated notation for "2x load." 912 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint 913 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint Figure 5. RNAseq analysis. (A) Volcano plot of significant genes in 2xload 914 group compared to unload condition and GO Biological process analysis. (fold 915 change > 2, adjusted p value < 0.1). 278 genes upregulated in 2xload compared 916 to unload. (B) Volcano plot of significant genes in 2xload group compared to 917 PD condition and GO Biological process analysis. PD condition means 918 PD98059 inhibitors plus load condition. (fold change >1, adjusted p value 919 < 0.1). 612 genes upregulated in 2xload compared to PD. (C) KEGG pathway 920 analysis in above two DEG lists (278 DEG list and 612 DEG list). (D) The 921 heatmaps show gene expression level in different groups such as ERK-related 922 genes, microtubule-related genes, migration-related genes, DNA repair-related 923 genes and rejuvenation-related genes. 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint B 939 940 Figure 6. Implanted reprogrammed single cell and spheroids show activation 941 properties under compressive force in an FT AGED skin model . (A) 942 Representative images (20× magnification, Nikon) of collagen I. (Scale bar, 943 100μm). Normalized intensity plots of collagen I at the cell -implanted regions 944 from at least 3 replicates. S+L group: inject single cell under compressive force; 945 O+L group: inject spheroids under compressive force. (B) Illustration of the 946 mechanism of compressive force in cellular rejuvenation. Figure created with 947 BioRender.com. MT: microtubule. 948 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 5, 2024. ; https://doi.org/10.1101/2024.11.04.621794doi: bioRxiv preprint