Compliant substrates mitigate the senescence associated phenotype of stress induced MSCs

Abstract

Mesenchymal stromal cells (MSCs) are a promising cell population for musculoskeletal cell-based therapies due to their multipotent differentiation capacity and complex secretome. Cells from younger donors are mechanosensitive, evidenced by changes in cell morphology, adhesivity, and differentiation as a function of substrate stiffness in both two- and three-dimensional culture. However, MSCs from older individuals exhibit reduced differentiation potential and increased senescence, limiting their potential for autologous use. While substrate stiffness is known to modulate cell phenotype, the influence of the mechanical environment on senescent MSCs is poorly described. To address this question, we cultured irradiation induced premature senescent MSCs on polyacrylamide hydrogels and assessed expression of senescent markers, cell morphology, and secretion of inflammatory cytokines. Compared to cells on tissue culture plastic, senescent MSCs exhibited decreased markers of the senescence associated secretory phenotype (SASP) when cultured on 50 kPa gels, yet common markers of senescence (e.g., p21, CDKN2A, CDKN1A) were unaffected. These effects were muted in a physiologically relevant heterotypic mix of healthy and senescent MSCs. Conditioned media from senescent MSCs on compliant substrates increased osteoblast mineralization compared to conditioned media from cells on TCP. Mixed populations of senescent and healthy cells induced similar levels of osteoblast mineralization compared to healthy MSCs, further indicating an attenuation of the senescent phenotype in heterotypic populations. These data indicate that senescent MSCs exhibit a decrease in senescent phenotype when cultured on compliant substrates, which may be leveraged to improve autologous cell therapies for older donors.

Introduction

Advanced age is a significant comorbidity for fracture incidence as well as fracture complication requiring inpatient care.^(1,2) Current clinical treatments to address large bone defects rely on autologous bone grafts which suffer from harvest site morbidity, residual pain, and incur additional costs to the patient.⁽³⁾ Cell and tissue engineering approaches are under investigation as an alternative for autologous bone grafts, but few have focused on the unique needs for bone regeneration in aged individuals. Mesenchymal stromal cells (MSCs) are a multipotent population that readily differentiate into the osteogenic, chondrogenic, and adipogenic lineage in vitro and are a promising candidate cell population for musculoskeletal tissue regeneration.⁽⁴⁾ However, MSCs from aged donors are more scarce and exhibit limited proliferation and osteogenic differentiation compared to those from younger donors.^(5–7) The mechanism for this altered phenotype has been attributed to cellular senescence and limits the utility of cell populations from aged donors in tissue engineered strategies.

Cellular senescence is a quiescent cellular state characterized by broadened morphology, increased anti-apoptotic gene expression, and a pro-inflammatory senescence associated secretory phenotype (SASP).^(8,9) Senescent accumulation can be achieved through a combination of proliferative exhaustion and external stressors such as reactive oxygen species or radiation,^(10,11) with 5–20% of cells in the elderly demonstrating the senescent phenotype.^(12,13) The buildup of senescent cells may be a potential mechanism for MSC dysfunction, with MSCs from aged donors exhibiting increased senescent phenotypic markers.^(12,14) While the influence of substrate biophysical properties clearly influences the response of otherwise healthy cells^(15,16), the effect of the underlying substrate on senescent cells has not been thoroughly investigated. Continuous culture of MSCs on substrates of variable stiffnesses did not alter the senescent burden as determined by β-galactosidase expression.⁽¹⁷⁾ However, the effect of substrate stiffness on established senescent cells is not understood, and such an investigation is needed to improve the use of senescent cells in tissue engineered approaches.

Material properties such as ligand density, stress relaxation, and stiffness can alter MSC differentiation.^(16,18,19) The influence of substrate stiffness has been described for its effect on MSC differentiation,⁽¹⁶⁾ with higher stiffnesses promoting osteogenesis and softer substrates supporting adipogenesis. Stiffness within the bone microenvironment is heterogeneous depending on the compartment, ranging from 0.01 – 10 MPa, and can change with age.^(20,21) The mechanism of MSC mechanosensitivity is primarily due to transient receptor potential activating calcium influx, which can be identified by the translocation of Yes-associated protein (YAP) to the nucleus.^(22,23) MSCs sourced from children are more mechanosensitive, evidenced by increased YAP translocation to the nucleus, with profound differences in osteogenic and angiogenic profiles compared to MSCs from the elderly.⁽⁵⁾ This decrease in mechanosensitivity has been postulated as a mechanism for lower osteogenic potential and may facilitate or promote the senescent phenotype.

We previously demonstrated that acute irradiation generates a stress induced premature senescent state in MSCs that is similar to replicative induced senescence with regard to proliferation arrest, spread morphology, impaired differentiation (Supplementary Fig. 1), and pro-inflammatory protein secretion.⁽¹¹⁾ We hypothesize that stress induced senescent MSCs (siMSCs) possess a muted osteogenic response on substrates of increased stiffnesses compared to healthy MSCs (MSCs). We used collagen-coated polyacrylamide (PA) hydrogels to interrogate the effect of stiffness on senescent phenotype. We further hypothesized that siMSCs would exhibit an increased senescent phenotype when cultured on softer substrate stiffnesses, coordinating the physiological environment with cellular phenotype.

Materials and Methods

Cell culture

Human bone marrow-derived MSCs (RoosterBio, Frederick, MD, Donor 238) were expanded on tissue culture plastic (TCP) in growth media (GM) composed of alpha minimum essential media (alpha-MEM) supplemented with 10% fetal bovine serum (GenClone, Genesee Scientific, San Diego, CA) and 1% penicillin-streptomycin (Gemini Bio Products, West Sacramento, CA) in standard cell culture conditions. GM supplemented with 0.01 μM dexamethasone, 50 μg/mL L-ascorbic acid 2-phosphate, and 10 mM sodium β-glycerophosphate was used as osteogenic differentiation media (OM). GM or OM were refreshed every 48 hr for the duration of the experiments. MSCs were maintained in culture until ~80% confluent and serially passaged until use at passage 5. Generation of stress induced senescence via irradiation was performed as we described.⁽¹¹⁾ Briefly, MSCs were lifted with trypsin, resuspended in GM at a concentration of 1×10⁶ cells/mL, and subjected to 10 Gy of irradiation using a MultiRad225 X-ray Irradiator (Precision X-ray, North Branford, CT). Populations interrogated included healthy MSCs (MSC), irradiated stress induced senescent MSCs (siMSC), and a 3:1 mix of control and stress induced senescent MSCs (MSC+siMSC), respectively. We used a mixed population including 25% stress induced senescent cells, as it was close to physiological levels but sufficiently distinct to isolate changes due to substrate stiffness or surrounding cell populations.^(12,13)

Polyacrylamide culture surface preparation

Polyacrylamide (PA) “easy coat” substrates with Young’s modulus of 50 kPa (Matrigen LLC, Irvine, CA) were covalently conjugated with 10 μg/cm² rat tail collagen 1 (MilliporeSigma, St. Louis, MO) via an amide-quinone reaction for 1 hr at room temperature per the manufacturer’s instructions. Tissue culture or glass substrates were coated with an equivalent amount of rat tail collagen 1 and incubated at room temperature for 1 hr. All culture substrates were washed with PBS prior to seeding with MSCs at 20,000 cells/cm² of respective cell populations.

Determination of MSC viability and metabolism

Total DNA content was quantified by scraping MSCs in the presence of 1X passive lysis buffer (Promega, Minneapolis, MN) followed by a 15 s sonication and centrifugation at 5000xg for 5 min to pellet cell debris. DNA from the supernatant was quantified using the Quant-It PicoGreen dsDNA kit (Invitrogen, Waltham, MA) following the manufacturer’s instructions. MSC metabolic activity was assessed using a 1:10 solution of alamarBlue (Invitrogen) in growth media for 1.5 hrs in standard cell culture conditions. Supernatant was dispensed into a black 96 well plate, and fluorescence (530/590 nm) reported as relative fluorescent unit (RFU). Caspase 3/7 activity was quantified using a Caspase-Glo 3/7 assay (Promega) per the manufacturer’s instructions and reported as a relative luminescent unit (RLU). alamarBlue and caspase 3/7 activity was normalized to DNA content.

Measurement of gene expression by Quantitative Polymerase Chain Reaction (qPCR)

MSCs were collected in Trizol Reagent (Invitrogen), and mRNA was isolated per the manufacturer’s instructions. 1000 ng of mRNA was reverse transcribed to complimentary DNA using the Quantitect Reverse Transcription Kit (Qiagen, Hilden, Germany) and reconstituted in water to a concentration of 13 ng/μL. MSCs from suspension were collected prior to seeding on culture surfaces as a Day 0 control. Single tube primer sets were purchased from ThermoFisher (Waltham, MA), and expression of CDKN1A (Hs00355782_m1), CDKN2A (Hs00923894_m1), and EGFR (Hs01076090_m1) were determined using a QuantStudio 6 Pro (Applied Biosystems, Waltham, MA). Relative expression was quantified using GAPDH (Hs02786624_g1) as the endogenous control gene. Expression of each gene was calculated as 2^−Δct and then normalized to healthy MSCs on TCP.

Quantification of SASP proteins

All treatment conditions were cultured in 24 well plates with 0.5 mL of media. Wells were refreshed with media 48 hrs prior to collection and conditioned media was collected on Day 14. Pro-inflammatory proteins in conditioned media were quantified using a ProcartaPlex Human Inflammation Panel 20plex kit (ThermoFisher) and read on a Luminex 200 System (Luminex, Austin, TX) per the manufacturer’s instructions. Proteins that were non-detectable or outside the standard curve were not included in the analysis, resulting in 9 proteins being characterized. Protein concentrations were normalized to total DNA content from the same cultures and further normalized to proteins secreted by healthy MSCs for each stiffness.

Immunocytochemical detection of senescence and mechanotransduction factors

After 14 days, MSCs were fixed with 4% paraformaldehyde for 15 min at room temperature. MSCs were permeabilized with 0.5% Triton X, washed with 0.1% Tween 20, and blocked with a 10% goat serum/1% bovine serum albumin (BSA) solution. Permeabilized MSCs were stained with either a p21^cip1 antibody (Novus Biologicals, Centennial, CO, NBP3–15661, 1:150 dilution), vinculin monoclonal antibody (Invitrogen, 50-112-3655, 1:20 dilution), or a YAP antibody (Santa Cruz Biotechnology, Dallas, TX, SC-376830, 1:150 dilution) and incubated overnight at 4^°C. Samples treated with p21^cip1 antibody were then stained with goat anti-rabbit secondary antibody (Abcam, Cambridge, UK, AB150083, 1:300 dilution), AlexaFluor 488 phalloidin (Invitrogen, A12379, 1:400), and DAPI (Invitrogen, 1:250) and incubated for 1 hr at room temperature. γH2AX was detected using a γH2AX Staining Kit (Abcam, ab242296) following the manufacturer’s instructions and counterstained with DAPI (1:250). All stained samples were washed with PBS and imaged on a Stellaris 5 confocal microscope (Leica Microsystems, Deerfield, IL). The integrated density of p21 fluorescent staining was quantified using ImageJ and normalized to cell number. Total cell number within a field of view was determined by applying an auto threshold to the DAPI channel and counting the resulting particles. Similarly, this thresholding method was used to segregate nuclear YAP expression. Raw integrated density of nuclear YAP was compared to that of YAP within the cytosol. In imaging samples on 50 kPa hydrogels, we observed an amplification in fluorescent signal despite the use of consistent image settings across samples. To account for the imbalance of intensity between substrate stiffnesses that would otherwise be reflected in the quantification of p21 expression, we used a scaling factor based on the ratio of intensity of DAPI staining in healthy MSCs on 50 kPa hydrogels compared to TCP. Quantification of γH2AX was performed using a triangle threshold (ImageJ, Bethesda, MD) on individual channels and normalized to cell number from each field of view. Three separate fields of view were quantified from each sample and averaged.

Determination of osteogenic differentiation

Intracellular alkaline phosphatase (ALP) activity was determined via the reduction of p-nitrophenyl phosphatase as previously described.⁽¹¹⁾ On Day 14, MSC monolayers were fully dissolved in 1 M HCl for 72 hr, and calcium deposition was quantified using a Calcium Liquid Reagent Diagnostics Kit (Stanbio Labs, Boerne, TX) as described.⁽¹¹⁾

Osteoblast mineralization in conditioned media

Human long bone osteoblasts (PromoCell, Heidelberg, Germany, P3) were seeded on a collagen coated 48-well plate at 4,100 cells/cm² in Osteoblast Growth Medium (PromoCell). One day after seeding, osteoblasts were switched to a 50:50 mix of Osteoblast Mineralization Medium (PromoCell) and conditioned media from MSCs, siMSCs, or MSC+siMSCs cultured on either TCP or 50 kPa substrates and collected on D14. Conditioned media mix was refreshed every 3 d for 21 d. On Day 21, osteoblast monolayers were collected and dissolved in 1 M HCl for 48 hr. Calcium deposition was quantified using a Calcium Liquid Reagent Diagnostics Kit (Stanbio Labs).

Statistical analysis

Data are presented as mean ± standard deviation. Statistical analysis was performed using Prism 9.5.1 (GraphPad, San Diego, CA) software utilizing one-way analysis of variance (ANOVA) with post hoc Tukey’s test or two-way ANOVA with post hoc Šídák’s multiple comparison test with p values less than 0.05 considered statistically significant. Groups with different letters denote significance (p<0.05) while groups that share a common letter are not statistically significant.

Results

Substrate stiffness dictates proliferation and cell metabolism

We generated a population of largely senescent stress induced senescent MSCs (siMSC) via irradiation to interrogate the effect of substrate stiffness on senescent phenotype. We first coated tissue culture plastic (TCP) with an approximate stiffness of 1 GPa or polyacrylamide (PA) with a stiffness of 50 kPa with type 1 collagen to facilitate cellular adhesion (Fig. 1A). These substrates were then seeded with either healthy MSCs (MSC), siMSCs, or a 3:1 population of MSC and siMSC (MSC+siMSC) (Fig. 1B) and cultured for 14 days in either GM or OM prior to analysis (Fig. 1C).

Figure 1: Overview of experimental conditions and experimental outline.

(A) Correlation of modeled substrate stiffness with different bone compartments. (B) Schematic of generation of irradiation induced senescent MSCs and interrogated MSC populations. (C) Experimental setup and timeline. Figure created with BioRender.

We initially cultured cells in both GM and OM to determine whether media composition would alter the response of siMSCs. Total DNA content was lowest when MSCs were cultured on 50 kPa substrates, regardless of media treatment, and siMSCs exhibited the lowest DNA content across both media formulations (Fig. 2A,D). The siMSCs had no differences in proliferation regardless of substrate stiffness, demonstrating that substrate stiffness does not rescue the quiescent phenotype. Quantification of alamarBlue staining, an indicator of cellular metabolic activity, of cells in GM exhibited a higher trend when MSCs were cultured on TCP, regardless of inclusion of siMSC (Fig. 2B). alamarBlue quantification of healthy MSCs increased when cultured on 50 kPa gels, while siMSCs had decreased alamarBlue on 50 kPa gels in OM. Mixed MSC populations in OM exhibited no significant change in alamarBlue activity between stiffnesses but trended toward increasing on 50 kPa gels (Fig. 2E). This result may be due to the normalization of the activity by DNA content and the low DNA content of the siMSC 50 kPa group. Generally, alamarBlue activity was higher in GM, likely due to the differentiation of MSCs in OM. No differences in caspase 3/7 activity emerged between MSC groups on either substrate stiffness in both media formulations with the exception of siMSCs on 50 kPa that achieved a significant reduction in apoptotic activity in OM (Fig. 2C,F). These data indicate media formulation does not alter the proliferation or metabolic activity of siMSCs on these substrates. Therefore, further studies were performed with OM alone to interrogate differences in senescent and osteogenic phenotype when cultured on TCP and 50 kPa substrates.

Figure 2: MSC viability is not influenced by media formulation or substrate stiffness.

(A) DNA content as an indicator of cell number, (B) quantification of alamarBlue staining as an indicator of metabolic activity, and (C) caspase 3/7 activity in cells cultured in GM. Comparative quantification of (D) DNA content, (E) alamarBlue quantification, and (F) caspase 3/7 activity in cells cultured in OM. Data are mean ± SD (n=4); significance determined by two-way ANOVA with p<0.05 and different letters denote significant differences. Dotted line represents DNA content of cells seeded at Day 0.

MSCs exhibit reduced markers of DNA damage on compliant substrates

Extensive DNA damage facilitates the development of the senescent phenotype and is used as a marker of senescent cells. MSCs across treatments were stained for γH2AX, an epitope for double stranded DNA breaks, to investigate the effect of substrate stiffness on the senescent phenotype. Confocal images illustrate that γH2AX is present in all groups. We observed the greatest amount present in siMSCs cultured on TCP, with a demonstrable decrease in stain prevalence when MSCs were cultured on softer substrates (Fig. 3A). Quantification of the images revealed a significant decrease in positive nuclear staining between TCP and the 50 kPa gel across all cell populations, with the highest γH2AX staining in siMSCs at both stiffnesses (Fig. 3B).

Figure 3: MSCs exhibited fewer DNA strand breaks on softer substrates.

(A) Immunocytochemistry of γH2AX (green) and DAPI (blue) indicating the presence of DNA double strand breaks. (B) Quantification of stain area normalized to cell number as determined by nuclei counts. Data are mean ± SD (n=3), and significance is determined by two-way ANOVA with p<0.05. Groups with different letters denote significant differences.

We further interrogated the influence of cell-substrate interactions by assessing the actin cytoskeleton and vinculin presentation due to its importance in regulating cell mechanical forces.⁽²⁴⁾ Vinculin was heavily expressed in MSCs at the periphery of the cell body across both substrate stiffnesses (Fig. 4). Vinculin was differentially expressed in siMSCs compared to MSCs, but presentation was similar between substrate stiffnesses, indicating that modulation of adhesion complexes is not implicated in reduction of the DNA damage phenotype. MSCs cultured on TCP exhibited similar actin fiber alignment and thickness, while siMSCs had dense actin fiber distribution compared to MSCs on 50 kPa gels. However, actin fiber differences may be due to the low confluency of the siMSCs on the softer substrate as opposed to differential responses to substrate stiffnesses. Collectively, these data indicate that culture on softer substrates decreases the senescent phenotype as indicated by γH2AX staining, and this reduction is not due to differences in cellular adhesion to the underlying substrates.

Figure 4: Vinculin is differentially expressed in senescent MSCs but is independent of substrate stiffness.

MSCs were stained for nuclei (DAPI, blue), vinculin (red), and actin (green). Cells were imaged at 400x, scale bar is 100 μm.

MSCs on compliant substrates exhibit reduced SASP markers

The detrimental effect of senescent cells on tissue homeostasis is commonly associated with the proinflammatory cytokine profile known as the senescent associated secretory phenotype (SASP). To assess the influence of substrate stiffness on the SASP, we collected conditioned media from each group after 14 days of culture in OM and quantified its composition using a multiplexed analyte bead system. The siMSC population secreted the highest quantities of proinflammatory proteins when cultured on TCP, and these quantities were significantly reduced when siMSCs were cultured on the 50 kPa substrates (Fig. 5A). However, the SASP profile of the MSC+siMSC population was similar to the MSC profile with no differences observed across different substrate stiffnesses (Fig 5B). These results establish that softer substrates reduce the production of proinflammatory factors found in the SASP in homotypic populations, while the presence of healthy cells with physiologically relevant numbers of senescent cells can compensate for the senescent burden.

Figure 5: Stress induced senescent MSCs exhibit suppressed secretion of proinflammatory cytokines of the SASP when cultured on softer substrates.

(A) Protein quantification of secreted inflammatory factors from siMSCs and (B) MSC+siMSCs at 14 days of culture. Readings are normalized to proteins secreted by healthy MSCs for each stiffness. Data are mean (n=4); significance between stiffnesses determined by two-way ANOVA with *** p<0.001. Analytes without *** indicate no significance.

In addition to the SASP, the expression and presentation of p21 is a widely used marker of the senescent phenotype. To identify if p21 expression was dependent on substrate stiffness, we stained MSCs for p21 protein at Day 14. The presentation of p21 was greatest in siMSCs on TCP, while p21 was visibly decreased in siMSCs on 50 kPa gels (Fig. 6A), in agreement with reduced SASP expression by siMSCs on 50 kPa gels. We quantified p21 staining across several samples to confirm this observation, yet quantification did not reveal statistically significant differences (Fig. 6B). The amount of p21 staining was similar within MSC populations on each substrate. We subsequently characterized expression of two commonly studied senescence-related genes after 2 weeks. The expression of CDKN2A, the gene that encodes for p16, was similar across all cell populations and stiffnesses (Fig. 6C). Expression of CDKN1A, which encodes for the p21 protein, was highest in the siMSC 50 kPa treatment group, and trends suggest that softer substrates increase the expression of this gene in all cell populations (Fig. 6D). These data are in contrast to p21 staining, in which we did not observe quantitative differences in p21 protein. Gene expression of EGFR, encoding for the epidermal growth factor receptor, was highest in siMSCs on 50 kPa gels, with nearly triple the expression of siMSCs on TCP (Fig. 6E). These data reveal increased p21 protein expression in siMSCs compared to healthy MSCs, as expected. However, we did not appreciate significant differences in the expression of p21 protein or genes commonly used as markers of senescence when cultured on stiff versus compliant substrates.

Figure 6: Substrate stiffness does not influence expression of common markers of senescence.

(A) Representative images of groups stained with DAPI (blue), p21cip (red), and phalloidin (green) after 14 days in OM. (B) Quantification of p21 staining. (C) CDKN2A, (D) CDKN1A, and (E) EGFR expression was normalized to gene expression in healthy MSCs on TCP. Data are mean ± SD (n=3); significance determined by two-way ANOVA with p<0.05. Scale bar is 100 μm.

Mechanosignaling is preserved while osteogenic outputs are influenced by softer substrates

We next investigated if the reduction in senescent phenotype on softer substrates could be leveraged to improve MSC osteogenic differentiation. Cellular mechanosensation is regulated through the Yes-associated protein (YAP), and increased mechanical responsiveness in two-dimensions is evident by translocation of the YAP protein to the nucleus.^(25,26) To understand if the reduction of SASP factors was due to a reduced or hyper-sensitivity to the mechanical environment, we studied the presence and location of YAP in all groups (Fig. 7A). Quantification of YAP staining indicated no difference in YAP translocation within MSC populations or between stiffnesses (Fig. 7B). These data suggest that mechanosignaling is not significantly altered for MSCs between the two substrates, and reductions in senescence markers are due to another mechanism.

Figure 7: The expression of osteogenic markers in stress induced premature senescent MSCs is influenced by substrate stiffness.

(A) MSCs were stained for DAPI (blue), YAP (red), and actin (green). (B) Image quantification of YAP localization at day 14. Quantification of (C) intracellular alkaline phosphatase (ALP) activity and (D) calcium deposition after 14 days in OM. (E) Quantification of calcium deposition by osteoblasts at Day 21 when treated with conditioned media (CM) obtained from MSC, siMSC, and mixed groups. Data are mean ± SD (n=3–4); significance determined by two-way ANOVA with different letters indicating significance. Scale bar is 100 μm.

Intracellular alkaline phosphatase (ALP) was significantly higher in healthy MSCs when cultured on TCP, while siMSC and MSC+siMSC populations exhibited similar, yet lower, ALP activity (Fig. 7C). In agreement with ALP activity, MSCs on 50 kPa gels secreted less calcium than on TCP. This reduction was no longer observed with siMSCs, either in homotypic or heterotypic populations, and the addition of healthy MSCs to siMSCs increased calcium deposition over 14 days (Fig. 7D). To further probe stiffness-driven differences between MSC populations, we analyzed the potential of conditioned media to induce calcium deposition in osteoblasts. Interestingly, osteoblasts cultured with conditioned media from siMSCs on compliant substrates deposited more calcium after 21 days compared to conditioned media from siMSCs on TCP (Fig. 7E). These data suggest the complex secretome of stress induced premature senescent MSCs is affected by substrate stiffness and may influence the osteogenic potential of surrounding cells.

Discussion

The biophysical properties of materials used as cell delivery vehicles or culture for expansion and preconditioning are instrumental in instructing MSC function and therapeutic efficacy.^(16,19) The importance of these properties has been established using healthy cells from young donors, while little attention has been paid to the response of cells from aged donors. MSCs from aged donors suffer from reduced prevalence, increased senescence, and impaired osteogenic potential.⁽¹⁴⁾ The senescent load within aged mice can be more than double that of young mice, ranging from 5–20% depending on the tissue compartment.⁽¹³⁾ Furthermore, a meta-analysis of senescence and chronological age indicates an increase in senescent markers with increasing age across a broad spectrum of tissues.⁽²⁷⁾ Thus, there is a significant need to study the role of substrate mechanical properties on senescent populations to develop effective cell-based therapies for the elderly.

Numerous materials have been employed to interrogate the effect of mechanical properties on cell behavior. Common materials that offer easily tunable mechanical properties include alginate ⁽¹⁸⁾, collagen ⁽²⁸⁾, polyethylene glycol ⁽²⁹⁾, and polyacrylamide (PA).⁽³⁰⁾ Herein, we used the nondegradable PA platform due to its reproducible mechanical properties, elastic nature, and elimination of the confounding effect that degradation may have on cell behavior. Consistency in collagen coating, as well as the relatively thin collagen adhesion complex compared to the thickness of the PA gel, ensures that cells are responding to the biophysical properties of the underlying substrate.⁽³¹⁾ Others reported that extended culture of adipose-derived MSCs on substrates with variable stiffnesses did not alter the induction of the senescent phenotype or DNA methylation.⁽¹⁷⁾ However, bone marrow-derived MSCs cultured on softer poly(ethylene glycol) hydrogels exhibited reduced MSC proliferation and senescence-associated β-galactosidase staining, in agreement with our findings.⁽³²⁾ To our knowledge, this is the first study that interrogates the effect of varying substrate stiffness on stress induced premature senescent MSCs from human primary cells. We utilized a single PA gel stiffness, 50 kPa, but a library of stiffnesses could prove useful as the marrow space is known to accumulate adipose tissue with age.^(33,34) Moreover, this is the first study to interrogate the response of mixtures of senescent and healthy cells, as are found in vivo, which is key to understand how neighboring cells may compensate for population level responses to a stimulus.

Senescence is defined by arrested proliferation. Herein, stress induced premature senescent MSCs exhibited this quiescence and a lack of metabolic maintenance over a 14 day culture period. Proliferation was improved across all cell populations on TCP, in agreement with previous reports of MSCs exhibiting increased proliferation on stiffer substrates.^(32,35) Senescent MSCs demonstrated a dysregulated metabolic response to softer substrates, with an inverse response to groups that included healthy MSCs. Caspase 3/7 activity increased in all MSC populations cultured on TCP substrates, contrary to literature investigating stiffness effects on apoptosis.⁽³⁶⁾ However, osteogenic differentiation and stem cell maintenance are reliant on non-apoptotic increases in caspase activity and may be responsible for the increase in caspase 3/7 on the stiffer substrates.⁽³⁷⁾

DNA damage is a reliable mechanism to induce premature senescence in vitro and establish the senescent phenotype from donor tissue.^(14,38) We previously reported that senescence induction of MSCs using gamma irradiation increases γH2AX expression, an epitope for DNA strand breaks.⁽¹¹⁾ While γH2AX increased over time from Day 0 in all groups (Supplementary Fig. 2), these data reveal that MSCs cultured on softer substrates possess reduced γH2AX expression, with the largest reduction observed within the homotypic siMSC population. DNA strand repair is dependent on ECM stiffness via activation of MAP4K4/6/7, with lower stiffnesses inhibiting the repair process within cancer cell lines.⁽³⁹⁾ Alternatively, cells from aged donors demonstrated a dysregulated accumulation of lamin A, a purported mechanism of aging, and stiffer substrates increase lamin A expression.^(40,41) The reduction of DNA damage markers suggest that the softer substrates reduce accumulation of DNA damage and warrants further investigation.

Senescent cells secrete a pro-inflammatory SASP that is the primary mechanism by which these cells dysregulate tissue homeostasis and regeneration.^(8,42) Due to the detrimental effects of the SASP, we quantified the secretion of nine proteins implicated in inflammation and chemotaxis of immune cells. siMSCs exhibited a significant reduction in eight of these factors when cultured on 50 kPa gels. However, we did not detect differences in pro-inflammatory protein secretion between MSC and MSC+siMSC across the different stiffnesses, suggesting that low amounts of senescent MSCs in vitro do not generate a pro-inflammatory secretome. These findings indicate that senescent MSCs secrete higher concentrations of SASP-related proteins when cultured on stiff substrates, and the utilization of softer substrates for MSC delivery may be beneficial in the elderly. ⁽⁴³⁾ Furthermore, healthy MSCs buffer the inflammatory effect of siMSCs, even when the senescent population is enriched compared to physiological conditions. This finding is in agreement with murine heterochronic parabiosis studies, where the circulation of young mice decreased inflammatory markers within the aged animals.^(44,45) When exploring the influence of substrate stiffness on commonly studied genetic markers of senescence (i.e., CDKN1A, CDKN2A), we did not detect changes as a function of substrate stiffness. We also did not observe dramatic differences in mechanosignaling by siMSCs, regardless of substrate stiffness. Moreover, the osteogenic response of stress induced senescent MSCs was impaired compared to healthy MSCs. As expected, the osteogenic response of healthy MSCs was reduced when cultured on compliant substrates compared to stiff TCP. However, when examining the influence of the secretome of siMSCs on osteoblast differentiation, we observed increased calcium deposition using conditioned media from siMSCs on softer substrates. These data suggest the complex secretome of stress induced premature senescent MSCs is influenced by substrate stiffness and may influence the osteogenic potential of surrounding cells, which is in keeping with differences in inflammatory cytokines within the secretome of siMSCs.

The reliable and consistent determination of markers of the senescent phenotype is a significant challenge. The majority of studies on senescence have used aged or transgenic mice,⁽⁴⁶⁾ but these findings have not translated well with human primary cells.^(47,48) The generation of stress-induced senescent cells is an effective model to overcome the limitations and diversity of tissue sourced from older donors.⁽⁴⁹⁾ Furthermore, the use of stress-induced senescent cells reduces the burdens associated with replicative induced senescence in vitro such as time, cell culture consumables, and phenotypic drift. Stress induced premature senescence (SIPS), induced by irradiation, relies on inducing DNA damage and the subsequent DNA damage response.⁽⁵⁰⁾ However, the generation of senescence in the aged population is primarily driven by replicative exhaustion leading to telomere attrition.⁽⁵¹⁾ The mechanistic difference of senescence induction alters the genetic expression between SIPS and physiological senescence,⁽⁵²⁾ but there is large proteomic overlap within the SASP between SIPS and physiological aging.⁽⁴²⁾ As such, irradiation induction of senescence in human cells possesses value as a model system, but the translation of these findings will require validation using cells from aged donors and tissues. Finally, these studies were performed in two-dimensions to better observe changes in cytoskeletal proteins and senescent markers. Future studies must examine the influence of substrate stiffness in three-dimensional culture to identify effective biophysical properties in cell carriers for use in the elderly.

Conclusion

We demonstrated that substrate stiffness influences senescent phenotype in stress induced senescent MSCs. The pro-inflammatory secretions of senescent cells were reduced on softer substrates. Markers of DNA damage were highest in homotypic populations of stress induced senescent MSCs, regardless of substrate stiffness, while mixed populations of senescent and healthy MSCs exhibited levels of DNA damage and osteogenic potential that more closely mimicked healthy MSCs. These data indicate that the pro-inflammatory secretome of a physiological level of senescent cells can be attenuated by neighboring healthy cells. The differential response of senescent MSCs to substrate stiffness and the impact of this response on neighboring cell function are important characteristics to consider when developing cell-based therapies that may contain a high population of senescent cells.

Data availability statement

Data will be made available upon reasonable request to the authors.