3D cell sheet structure augments mesenchymal stem cell cytokine production

Mesenchymal stem cells (MSCs) secrete paracrine factors that play crucial roles during tissue regeneration. An increasing body of evidence suggests that this paracrine function is enhanced by MSC cultivation in three-dimensional (3D) tissue-like microenvironments. Toward this end, this study explored scaffold-free cell sheet technology as a new 3D platform. MSCs cultivated on temperature-responsive culture dishes to a confluent 2D monolayer were harvested by temperature reduction from 37 to 20 °C that induces a surface wettability transition from hydrophobic to hydrophilic. Release of culture-adherent tension induced spontaneous cell sheet contraction, reducing the diameter 2.4-fold, and increasing the thickness 8.0-fold to render a 3D tissue-like construct with a 36% increase in tissue volume. This 2D-to-3D transition reorganized MSC actin cytoskeleton from aligned to multidirectional, corresponding to a cell morphological change from elongated in 2D monolayers to rounded in 3D cell sheets. 3D culture increased MSC gene expression of cell interaction proteins, β-catenin, integrin β1, and connexin 43, and of pro-tissue regenerative cytokines, vascular endothelial growth factor (VEGF), hepatocyte growth factor (HGF), and interleukin-10 (IL-10), and increased VEGF secretion per MSC 2.1-fold relative to 2D cultures. Together, these findings demonstrate that MSC therapeutic potency can be enhanced by 3D cell sheet tissue structure.

umbilical cord mesenchymal stem cells (hUC-MSCs) are seeded onto a TRCD and grown to confluence under conventional adherent culture conditions, rendering a 2D monolayer. To generate and release a cell sheet, the 2D monolayer undergoes a 2D-to-3D transition: temperature reduction from 37 to 20 °C changes the TRCD surface from hydrophobic to hydrophilic, releasing the 2D monolayer from adherent culture tension and prompting contraction into a 3D cell sheet (Fig. 1b). The 2D monolayer before contraction was reproduced on a cell cultureinsert membrane with identical culture conditions as the 3D cell sheet group. Immediately prior to cell sheet detachment/contraction, insert membrane-cultured monolayers were taken as 2D monolayer (Fig. 1a) controls. In parallel with the gross change in cell sheet macroscopic structure following detachment contraction, the cell morphology undergoes a transition from 2D aligned adherent cell shape (Fig. 1c,e) in 2D monolayers to 3D unaligned rounded cell shape (Fig. 1d,f) in 3D cell sheets. This result suggests that cell sheet contraction into a 3D structure post-release alters cell shape.
Cell sheet contraction from an adherent monolayer yields a 3D tissue-like structure. Hematoxylin and eosin (H&E) staining of hUC-MSC monolayer and sheet cross-sections shows that the 2D monolayers are, indeed, only single-nuclei thick, while the 3D cell sheets are multi-nuclei thick structures (Fig. 2a,b). Contracted hUC-MSC sheet 3D structure is contributed by a 2.4-fold reduction in sheet diameter ( hUC-MSC actin structure (cytoskeleton) changes in response to cell sheet contraction. hUC-MSC 2D monolayer and 3D cell sheet cytoskeletal arrangement was observed with phalloidin (F-actin) fluorescent staining. Imaged from the top-down, hUC-MSCs in 2D monolayers exhibit unidirectional and elongated cytoskeletal structures, aligned in the direction of cell spreading (Fig. 3a). Conversely, hUC-MSCs that undergo cell sheet contraction present a more 3D cytoskeletal structure with random, multidirectional alignment (Fig. 3b). Although no significant differences in β-actin gene expression per hUC-MSC in each group were evident, hUC-MSCs in 2D monolayers showed greater average β-actin gene expression compared to hUC-MSCs in 3D cell sheets (Fig. 3c). Actin structure and gene expression differences indicate that hUC-MSC cytoskeleton remodeled toward a 3D arrangement in response to cell sheet contraction from the 2D adherent monolayer.
hUC-MSC nuclear shape changes in response to cell sheet contraction. DAPI-visualized nuclei in 2D hUC-MSC monolayers appeared more elongated than nuclei in 3D hUC-MSC sheets (Fig. 3d,e). To quantify this finding, nuclei circularity was measured, where a value of 1.0 indicates a perfect circle, and values approaching 0.0 indicate an increasingly elongated shape. The average circularity of nuclei in 3D hUC-MSC sheets (0.69 ± 0.092) was closer to 1.0 than nuclei in 2D hUC-MSC monolayers (0.43 ± 0.12), representing a significant difference in nuclei circularity due to cell sheet contraction (Fig. 3f)  Enhanced pro-regenerative cytokine gene expression is related to 3D cell sheet tissue-like structure. Cell interaction-protein gene expression appears to be upregulated in the 3D, tissue-like environ-   www.nature.com/scientificreports/ ment of contracted cell sheets. β-catenin, an intracellular portion of the adherens junction protein complex that binds extracellular cadherin to mediate cell adhesion to neighboring cells 34 , and integrin β1, a protein motif that extracellularly binds ECM ligands 35 , are both significantly upregulated in 3D cell sheets relative to 2D monolayers (Fig. 4a,c, respectively) (p = 0.0043 and p = 0.0051, respectively). Concomitantly, gene expression for celladhering ECM glycoprotein, laminin, is significantly upregulated in 3D cell sheets ( Fig. 4d) (p = 0.036). Gene expression for connexin 43, a gap junction protein that spans the cell membranes of neighboring cells and allows direct intracellular cytoplasmic molecular signaling exchange, is increased on average per hUC-MSC in 3D cell sheets relative to 2D monolayers (Fig. 4b). Vascular endothelial growth factor (VEGF) and hepatocyte growth factor (HGF) gene expression per hUC-MSC are both upregulated on average in 3D cell sheets relative to 2D monolayers ( Fig. 4e,f, respectively), although not significantly. Gene expression of interleukin-10 (IL-10) was undetectable in 2D monolayers but was measurable per hUC-MSC in 3D cell sheets (Fig. 4g). Fibroblast growth factor (FGF) gene expression was slightly increased on average per hUC-MSCs in 2D monolayers relative to 3D cell sheets (Fig. 4h). These data indicate that genes related to a tissue-like microenvironment are upregulated in 3D cell sheets compared to 2D monolayers, improving the pro-regenerative cytokine secretory capacity in 3D cell sheet hUC-MSCs.

Cell sheet contraction increases real cytokine production by hUC-MSCs. Human VEGF secretion
per hUC-MSC was increased 2.1-fold in 3D cell sheets compared to 2D monolayers, measured over a 24-hour (h) culture span (Fig. 5a). To account for differences in cell numbers due to cell proliferation over 24 h, secreted VEGF was normalized to average cell number per group at 24 h. Despite a significantly higher cell proliferation rate in 2D monolayers (1.7-fold increased ± 0.20, p = 0.013) compared to 3D cell sheets (1.3-fold increased ± 0.18) over the 24-h culture span (Fig. 5b), 3D cell sheet hUC-MSCs secreted twice as much VEGF per cell on average.

Discussion
Clinical MSC applications exploit the unique paracrine signaling function of MSCs [36][37][38] , where the therapeutic potency is dependent upon MSC cytokine secretory capacity as single cell-suspension formulations. Toward improving their clinical significance, we assert that engineering MSCs as tissues functionally enhances individual MSCs beyond their dissociated single cell potency. Specifically, MSC cytokine production is clearly stimulated using highly functional MSC-dense tissue-like constructs. This tissue effect was clearly demonstrated in a previous study that compared single MSCs to 2D MSC monolayers 33 . Single MSCs that had been enzyme-dissociated to cleave cell interactions between neighboring cells, as well as degrade ECM proteins and cleave associated cell-ECM binding proteins, were compared to 2D MSC adherent monolayers that preserved confluent cell-cell www.nature.com/scientificreports/ and cell-matrix interactions. Paracrine factors VEGF, HGF, and IL-10 were significantly upregulated in 2D MSC monolayer cultures relative to single dissociated MSCs 33 . This upregulation was clearly attributed to maintenance of a tissue-like microenvironment 33 . In the present study, we further explored the tissue effect on clinically important MSC cytokine production potency, pushing the tissue model one step further by contracting 2D monolayers into 3D cell sheets. The 3D tissue-like microenvironment supports physical (cell shape and spatial arrangement) and chemical (cell interaction and binding protein expression) effects that augment MSC potency beyond 2D monolayer and single cell formulations. Spontaneous cell sheet contraction creates a 3D microenvironment by contracting the monolayer diameter 40% and increasing the thickness 8.0-fold, representing a 36% increase in tissue volume of the contracted cell  The hUC-MSC proliferation rate during this time course was significantly higher in 2D monolayers compared to 3D cell sheets. Values are means ± SE (n = 3: *p < 0.05 and ***p < 0.001). www.nature.com/scientificreports/ sheet compared to the 2D monolayer ( Fig. 2c-e). While the exact mechanism underlying this volume increase is unclear, cell attachment strength is likely implicated. During 2D adherent culture on tissue culture plastic, the cell-material interaction imposes a high basal adhesive force that compacts the cell cytoskeleton and promotes a tight-packed arrangement of cells and deposited ECM as the cells grow to confluence. The actin cytoskeleton uniformly aligns and compacts in the direction of cell spreading (Fig. 3a). Furthermore, cell attachment and spreading may be attributed to water efflux, resulting in reduced cell volume 39,40 . Once the cell-material interface is disrupted by temperature-responsive release from the material surface, cytoskeletal compaction is released and reorganizes to confer a 3D cell shape, with multidirectional actin arrangement (Fig. 3b). Changing MSC β-actin expression further evidences cytoskeletal rearrangement (Fig. 3c). β-Actin gene expression is relatively higher in 2D monolayer cells compared to those in the 3D cell sheet; this could be due to β-actin's relationship to cytoskeletal tension, which would be much higher under plastic adherent culture than in tissue culture 41 .
Monolayer adherent culture restricts cell adhesion to 2D because of high basal adhesion, limiting cell-cell contacts to the perimeter of adjacent cells. In contrast, cells in 3D culture can make cell interactions in all directions, between the encompassing matrix and between neighboring cells 42 . For this reason, 3D culture provides higher abundance cell interactions relative to 2D conditions 43 . Consistently, our study demonstrated that MSCs in 3D cell sheets also increase cell interactions (Fig. 4a-c). Intracellular catenin, directly binding cytoskeletal F-actin, forms a transmembrane complex with extracellular cadherin to connect adjacent cells 34 . β-Catenin gene expression was significantly upregulated in 3D cell sheet MSCs relative to 2D monolayer MSCs, likely due to increased cell-cell interactions as well as in response to contraction-imposed change in actin cytoskeletal structure. Connexin 43, a major MSC gap junction protein 44 , is similarly upregulated upon increased cell-cell contact in 3D cell sheets. Furthermore, integrin β1, an extracellular adhesion protein connecting cells and ECM 45 , is upregulated in 3D cell sheet MSCs. This observed integrin β1 gene expression increase is consistent with cytoskeletal remodeling as well as an increase in laminin gene expression (Fig. 4d), suggesting that cytoskeleton-bound integrin β1 binds ECM component, laminin, to facilitate greater cell-ECM adhesion in 3D cell sheets relative to 2D monolayers.
MSC paracrine function is arguably one of their most clinically beneficial attributes 46,47 , mediated by MSC secretion of myriad pro-regenerative cytokines and subsequent paracrine activity in host tissue. VEGF, HGF, FGF, and IL-10 directly regulate tissue repair and regeneration, with specific implications in vascularization 48 , fibrosis mitigation 49 , cell regeneration 50 , and inflammation mediation 51 , respectively. MSC production of major therapeutic cytokines involved broadly across tissue regeneration was assessed: 3D cell sheets increased protissue regenerative VEGF, HGF, and IL-10 gene expression (Fig. 4e-g), while IL-10 was undetectable in MSCs in 2D monolayers. VEGF production doubled per MSC in contracted 3D cell sheets relative to 2D monolayers (Fig. 5a), despite significantly higher MSC proliferation rates in 2D monolayers (Fig. 5b). This difference in proliferative activity is to be expected, as it is widely recognized that 2D adherent culture generally promotes stromal cell proliferation rates faster than 3D culture conditions due to excessive basal adherence and cell spreading 39,52 . Normalizing for cell number and proliferation, we attribute greater cytokine-production potency to a 3D tissue effect: in part due to structural changes in cell morphology and cytoskeletal tension, and partly due to chemical cell interactions that are both deficient in 2D culture and absent in single cell suspension 19,53,54 . Particularly, β-catenin plays a specific role in mediating adipose-derived MSC HGF secretion via enhanced cell-cell adhesion 55 . Gap junction proteins that allow direct molecular signal exchange across lipid membranes of neighboring cells have been similarly identified for their key role in boosting individual MSC VEGF secretion, promoting angiogenesis 56 . Also, tissue-like cell-cell and cell-ECM interactions within 3D MSC culture systems significantly improved immune mitigation in an inflammatory arthritis model, due to notable upregulation of MSC-secreted IL-10 57 . FGF is strongly related to MSC proliferative activity 58 ; based on this published evidence combined with FGF expression and MSC proliferative data shown in Figs. 4 and 5 contrasting 2D and 3D MSC properties, 2D adherence mediated MSC proliferation would be expected to yield similar FGF gene expression in both 2D monolayer and 3D cell sheet MSCs (Fig. 4h).
Taken together, our results highlight several key features of cell sheet technology that collectively augment MSC cytokine secretory function: (1) temperature-induced detachment and spontaneous MSC monolayer contraction produces a 3D construct by spontaneous structural and morphological 3D transitions, (2) this 3D transition increases cell-cell and cell-matrix interactions endogenously derived during cell sheet fabrication, and (3) this 3D tissue effect enhances MSC cytokine secretory capacity relative to 2D MSC culture conditions (Fig. 6). For these reasons, cell sheet technology represents a 3D culture platform that enhances MSC paracrine capacities attributed to improved MSC clinical utility.

Conclusions
Spontaneous cell sheet contraction upon release from adherent 2D monolayer culture produces a 3D tissue-like microenvironment that facilitates a 3D MSC shape and cytoskeletal organization. Additionally, this 3D transition upregulates MSC cell-cell, cell-ECM, and gap junction interactions. 3D cell sheet culture increases MSC paracrine activity due to a tissue effect, characterized by cell-experienced structural and chemical changes. As a 3D MSC cultivation system, cell sheets are a promising new platform to boost MSC paracrine effects without exogenous biomaterials and without sacrificing crucial cell-matrix interactions. Collectively, these findings describe a 3D-engineered tissue with enhanced MSC paracrine-relevant secretory function relative to adherent monolayer MSC culture.  www.nature.com/scientificreports/ the monolayer or cell sheet (n = 3), and 5 linear measurements from the apical to basal plane of the monolayer or cell sheet were made per picture using AmScope Software (AmScope) and averaged per group. Tissue volume was calculated using 10 measurements of thickness and diameter per group. Percent change in tissue volume was calculated from the average volumes. Nuclei shape measurement. Nuclei circularity was quantified from DAPI-visualized nuclei in hUC-MSC 2D monolayer and 3D cell sheet cross-sectional images. Images were threshold and the "particle counter" function in ImageJ software was used to measure the circularity of particles with an aspect ratio between 0.0 and 1.0, with 0.0 corresponding to a completely elongated object and 1.0 corresponding to a perfect circle. Ten nuclei were measured per group (n = 3 sheets/group) and reported with mean and standard error (SE). Soluble VEGF secretion normalized to cell number and cell proliferation rate. Immediately following cell sheet detachment and re-plating onto insert membranes, 2D monolayer and 3D cell sheet media were exchanged for fresh growth media with 10% FBS and samples were cultured for 24 h (37 °C, 5.0% CO 2 ). FBS media (10%) alone were cultured for 24 h as a control. The 24-h supernatants (n = 3 per group) were collected and aliquoted, then centrifuged at 1200 RPM for 5 min to pellet cellular debris and preserve soluble proteins in the supernatant. The concentration of soluble VEGF secreted per 2D monolayer and 3D cell sheet was quantified using a human VEGF Quantikine ELISA kit (R&D Systems, MN, USA) and normalized to cell numbers to determine VEGF secreted per cell. Briefly, samples were rinsed twice with PBS and trypsin-EDTA (0.05%, 2 mL) (Gibco) was added directly on the samples to be incubated at 37 °C first for 10 min in a humidified incubator, then for 15 min in a 37 °C water bath. Afterwards, trypsin was removed by centrifugation (1200 RPM, 5 min) and supernatant aspiration. Cell pellets were dispersed with 0.5 mL collagenase P (0.05%, Sigma Aldrich) and incubated for 10 min in a 37 °C water bath. At this point, a single cell suspension had been rendered. Cell suspensions were reconstituted to 1.0 mL with 10% FBS media and exact cell numbers were counted using a trypan blue exclusion assay (Cell Culture Tested Trypan Blue Solution, Sigma Aldrich). Proliferation rate was quantified as the fold change increase in cell number from 0 to 24 h.

Statistical analysis.
All statistical analysis was conducted on data sets of n ≥ 3 biological replicates, with quantitative values expressed as a mean ± SE. D' Agostino-Pearson omnibus K2 test was used to determine a normal distribution for each data set, and therefore a parametric analysis of significance was appropriate. A twotailed, paired, Student's t test was used to measure statistical significance using GraphPad Prism version 9.0.0 for Windows (GraphPad Software, San Diego, California USA, http:// www. graph pad. com). Statistical significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001. No statistical significance was defined as p > 0.05. www.nature.com/scientificreports/