MDSC primary culture was established as previously described protocols (Gharaibeh et al., 2008). Primary MDSCs were isolated from pooled skeletal muscles of 1-month-old C57BL/6 mice. The MDSC population was expanded up to passages 3–5 and cryopreserved in aliquots to establish a working cell bank, which was subsequently used for all in vitro and in vivo experiments. All animal procedures were approved by the institutional ethics committee (Protocol 003/22; June 6, 2022). Human endothelial HUVEC cells were kindly provided by Prof. Dr. Sang Won Han (UNIFESP, São Paulo, Brazil). Two HUVEC preparations were used in this study: i. a GFP-transduced HUVEC line generated by lentiviral transduction (Lv-VEGF-GFP) that was used exclusively in preliminary mixing controls to confirm the targeted 95:5 MDSC:HUVEC composition by fluorescence microscopy; and ii. the parental, non-transduced HUVEC line, which was used for all experiments reported in the manuscript, including spheroid formation, Live/Dead viability assays (Calcein-AM/7-AAD), and any downstream analyses. No GFP-expressing HUVECs were included in viability quantification, thereby avoiding any potential interference with the green fluorescence channel. Unless otherwise specified, MDSC were cultivated in Proliferation Medium (PM), constituted by Dulbecco’s modified Eagle’s medium (DMEM; Thermo Fisher Scientific, Inc., United States) supplemented with 10% Fetal Bovine Serum (FBS, Thermo Fisher), 10% Horse Serum (HS, Thermo Fisher), 0,5% Chicken Embryo Extract (MP Biomedicals, United States) and 1% Penicillin/Streptomycin (Thermo Fisher). Endothelial differentiation followed the protocol previously described, with modifications (Harding et al., 2017). Cells were seeded on plates covered with type I bovine collagen (Thermo Fisher) and cultured with endothelial cell growth media (EGM, Endothelial Cell Growth Medium-2, Lonza) supplemented with 50 ng/mL of VEGF (Peprotech) for 7 days. All cellular cultures were incubated at 37 °C with an atmosphere of 5% CO₂ (MCO-170AICUV(H)L-PA incubator, Panasonic, Japan) and tested free of mycoplasma contamination. All cellular culture manipulations were conducted in a sterile laminar-flow environment.

The spheroids were synthesized as previously described (Seixas et al., 2023). In summary, 1 × 10⁶ trypsinized cells of designed cellular cultures were added into each micromold, yielding 81 spheroids, with an average of 1.2 × 10⁴ cells/spheroid. Each mold was maintained with 2 mL of appropriate culture media. For MDSC/HUVEC spheroids, a mixture of 95% MDSC and 5% HUVECs was achieved before pipetting the cells into molds. In vivo therapeutic spheroids constituted a mixture of 95% MDSC and 5% MDSC-VEGF (MDSC/MDSC-VEGF), which were pre-differentiated into endothelial lineages through cell culture in EGM-VEGF enriched media for 7 days (50 ng/mL VEGF). All in vitro spheroid assays were followed for 21 days. For in vivo therapeutic use, MDSC/MDSC-VEGF spheroids were synthesized by mixing 95% MDSCs and 5% VEGF-preconditioned MDSCs, and used within 24 h of formation. These spheroids were encapsulated on day 1 of culture to preserve viability and functional phenotype. Spheroid DNA quantification was assessed through Quant-iT™ PicoGreen™ (Thermo Fisher, United States) for such in vivo spheroids. For each timepoint, independent vials containing multiple spheroids were processed in parallel, and total DNA was extracted from each vial as a single sample. The resulting DNA quantity was then normalized by the number of spheroids within that vial to derive the value per spheroid. This approach results in multiple measurements per timepoint that originate from biologically independent vials, rather than technical replicates of the same sample. All spheroids were periodically assessed for size (longitudinal diameters) and morphology under bright-field microscopy with Axio Vert.A1 microscope (ZEISS, Germany). Sphericity was defined as the ratio of the vertical over horizontal longitudinal ratios for a specific spheroid. Microphotographic acquisition was followed with vertical and horizontal radius measurement (“Straight, segmented or freehand lines” tool) utilizing ImageJ (National Institute of Mental Health, United States). The same cellular population and quantity designated to formulate the spheroids was plated, as described in Cell culture, and 2-dimensional microphotographic acquisition of adherent cells was performed, to illustrate, in Figure 2, the difference in the size of spheroids not being related to a significant difference in individual cellular size.

Low molecular weight alginate (MW 50 kDa) used in this study was obtained from Novamatrix (Norway). Alginate peptide residue modification with RGD (GGGGRGDSP-COOH, AminoTech, Brazil) was achieved through carbodiimide chemistry, yielding a degree of substitution of 2. The spheroids were encapsulated in RGD-modified low molecular weight, 0% oxidized, alginate hydrogel, as previously described, with a ratio of 2% weight/volume (w/v) (Whitehead et al., 2021). Spheroids were mixed with unpolymerized alginate. For CaCl₂ gelification assays, the spheroid-alginate mixture was pipetted with a 1,000 µm pipette tip onto a 200 mM CaCl₂ bath, forming macroscopic alginate hydrogel spheres, with a goal of 3 spheroids per hydrogel. The hydrogel spheres were then strained and washed 3 times with sterile normal saline (NaCl 0.9%). A batch of cell-free alginate hydrogels was synthesized, utilizing the same gelation method, for quantifying the cellular impact on hydrogels rheological properties. All hydrogels were cultivated under different culture media at 37 °C and 5% CO₂. Ten spheroids were manually transferred into each hydrogel under sterile conditions using micropipette handling for in vivo experimenting. The number of spheroids per hydrogel was confirmed by direct visual inspection under bright-field microscopy prior to implantation.

Every gel was assessed for its wet weight right after culture media suspension, and posteriorly, for its dry weight, after 24 h of lyophilization. Swelling ratio (Q) was calculated through the formula Q = ((W_s-W_d/P_w) + (W_d/P_p))/(W_d/P_p) where W_s is wet weight, W_d is dry weight, P_w is density of the water and P_p is density of the polymer. Mesh size was measured through interpretation of swelling data. Average mesh size of the network structure was calculated via Canal-Peppas equation, the derived formula of Flory–Rehner theory for anionic polymers: Mesh size = ξ = Q^1/3 * l * (2M_c/M_r)^1/3 * C_n ^1/2, where M_c is molecular weight between crosslinks, l is the length of the repeating unit and M_r is the molecular weight of the repeating unit, and C_n is the characteristic ratio of the polymer. M_c was obtained from the formula derived for anionic alginate polymers: M_c = C_p*R*T/G′, where C_p is the polymer concentration, R is the gas constant, T is temperature and G′ is the storage modulus. For G′ and G″ measurement, alginate hydrogels were submitted to a range of strain of (0.01%–10%) at a constant (10 rad/s), and the tests were performed 1 day after alginate polymerization, in a Anton Paar MCR 702 rheometer, and an average of 12-point analysis was obtained. Temperature was controlled at 25 °C, and rheological analysis was conducted under 8 mm parallel plates. At each timepoint, excess hydrogels were generated, and representative samples were selected based on structural integrity to maximize sample size (n) while maintaining experimental consistency. Further rheological parameters were previously described (Campbell et al., 2018).

Fluorescence microscopy images were acquired to assess cellular viability. Live/Dead assay was performed with 7-AAD (7-Aminoactinomycin D) and Calcein-AM (Thermo Fisher). Fluorescence imaging was obtained utilizing Leica DMi8 microscope (Leica Microsystems, Germany). Spheroids’ cellular viability was assessed through the corrected total cell fluorescence (CTCF) method. After image acquisition, green and red split channels of total cellular fluorescence were quantified with ImageJ (National Institute of Mental Health, Bethesda, Maryland, United States) for integrated density measurement. Three random background areas for each spheroid image were acquired for subtracting background fluorescence, according to the formula: CTCF = Integrated Density – (Area of selected spheroid x Mean fluorescence of background readings). Each spheroid viability (%) was determined following the formula: CTCF (viability, %) = CTCF green/(CTCFgreen+CTCF red).

Total RNA from cultured cells was extracted with Trizol reagent (Life Technologies). The cDNA was obtained from the extracted RNA using the High-Capacity Kit (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems^TM - Thermo Fisher) according to the manufacturer’s instructions. The primers and probes utilized for the RT-qPCR were: Ly6a_mouse_F CCT ACC CTG ATG GAG TCT GTG T, Ly6a_mouse_R CAC GTT GAC CTT AGT ACC CAG G, R18S_F GCC GCT AGA GGT GAA ATT CT, R18S_R CGA ACC TCC GAC TTT CGT TC and Quanti-Nova SYBR-Green (Qiagen). The reaction was performed in Quanti-Studio (Applied Biosystems), and was analyzed by a relative comparison method (2^−ΔΔCt). RNA extraction from 3D alginate-encapsulated spheroids was piloted but yielded insufficient, inhibitor-free RNA at the required implantation timepoints. Despite alginate lyase digestion and additional purification, polymer carryover impaired downstream RT-qPCR. To avoid delaying time-sensitive implantation, stemness assays were therefore performed on early-passage 2D MDSCs; spheroid-level profiling is planned in future work.

All animal procedures were approved by the institutional ethics committee (Protocol 003/22; June 6, 2022) and reported in accordance with ARRIVE 2.0. Three-month-old male C57BL/6 mice (n = 30; 25–28 g) were obtained from CEDEME (Center for the Development of Experimental Models in Medicine and Biology, Brazil). Volumetric muscle loss (VML) was induced in the left gastrocnemius under anesthesia with ketamine (100 mg/kg) and xylazine (10 mg/kg), following a previously described procedure with minor modifications (Hu et al., 2025; Greising et al., 2018; Basten et al., 2023). Briefly, the muscle was surgically exposed and a 4-mm circular defect was created using a sterile biopsy punch (Rhosse Instrumentos e Equipamentos Cirúrgicos EIRELI EPP, Brazil). Animals were identified by ear notching. For treatment, alginate hydrogel–encapsulated spheroids, cell-free alginate hydrogels, or spheroids alone were applied to the defect; a sham surgery group served as control. The muscle fascia, subcutaneous tissue, and skin were then closed with PDS II sutures (Ethicon, Inc.).

Muscle performance was assessed 15 and 30 days postoperatively, utilizing RotaRod testing (RotaRod, EFF 411, Insight, Brazil). The animals were familiarized with a period of training on the device in the 2 days prior to the test days, being submitted to 2 races of 3 min at 24 RPM, being removed within each fall. On the day of the test, the animals were placed in the RotaRod with an initial speed of 10 RPM, progressing to 24 RPM over 5 min (Felipone et al., 2025). The fall time was recorded for each animal individually and displayed graphically as individual plotted data for each group (encapsulated spheroids, cell-free hydrogels, spheroids alone and sham surgery).

Thirty days following the surgical procedure and after RotaRod testing, animals were euthanized for the evaluation of muscle regeneration by histological analysis. Gastrocnemius muscles were harvested, fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned for staining with Hematoxylin and Eosin (H&E) and Picrosirius Red. Image acquisition and quantification were performed using ImageJ and ImagePro Plus (Media Cybernetics, Rockville, MD, United States). To account for inter-sample variability, all histomorphometric measurements were normalized to the injury area of each individual sample, and data were expressed as the percentage of each histological parameter relative to the total injury area. The histological evaluation encompassed the analysis of granulation tissue development, the extent of muscle fiber degeneration and regeneration, and collagen deposition within the injured tissue. Fibrotic areas were defined by collagen accumulation in the endomysial and perimysial spaces between muscle fibers, as visualized by Picrosirius Red staining. The region of interest (ROI) was delineated within the central and regenerative zones of the injury site, deliberately excluding the survival zone and the epimysial cap (Figure 5C), in accordance with anatomical definitions established by Hurme et al. (1991). Quantification of fibrotic tissue was adapted from the methodology described by Martins et al. (2014), while the overall histomorphometric criteria were based on the protocol recently validated by Felipone et al. (2025). For samples with near-complete remodeling at Day 30, the ROI was confined to the central/regeneration zones (Figure 7C), excluding survival zone/epimysial cap; injured area extent was also measured and is reported in Supplementary Table S1 to document differences in residual lesion size. Additional details regarding the histopathological evaluation criteria can be found in the Supplementary Material (Supplementary Figure S2; Supplementary Table S1).

Unless specified, the results obtained are herein expressed as mean ± SD (standard deviation) and submitted to statistical analysis. Statistical analysis was performed by one-way ANOVA with Sidak’s multiple comparisons test or Mann-Whitney test when appropriate. The software used was the GraphPad Prism version 9.0 (GraphPad Software, Inc., La Jolla, CA, United States). The results obtained were considered statistically significant if p < 0.05. Different letters imply statistical significance.

MDSC were cultivated in PM on low confluency (<30%) for stem cell properties maintenance (Supplementary Figure S1A). When MDSC reached higher confluences (60%), clustered differentiation was observed under bright-field microscopy. The visual aspect of round, small cells with greater nuclei proportion was typical of MDSC with preserved stemness. Fusiform, bundled group of cells with multinucleated fibers was typical of terminal differentiation (Supplementary Figure S1B). To quantify stem cell potential, RT-qPCR was performed for key transcription factors. Sca-1 (stem cell antigen-1 - Ly6a) was assessed as a proxy for stemness. Ly6a expression was 3-fold higher in MDSC cultured in EGM media in comparison to PM (Supplementary Figure S1C, p < 0.05).

We initially conducted a morphological and growth characterization of MDSC, HUVEC, or a mixture of MDSC/HUVEC (95:5) spheroids before their encapsulation with alginate. A workflow illustrating the synthesis, culture, and encapsulation process of MDSC spheroids is depicted in Figure 1. All three spheroid groups underwent the same synthesis process to evaluate baseline morphological characteristics over the 21-day experimental period while being cultured in endothelial growth media (EGM). All spheroids exhibited a spherical morphology and homogeneity (Figure 2A). Despite variations in radius over time (Figures 2A,B), all spheroids maintained their spherical shape throughout the 21 days (Figure 2C).

MDSC spheroids showed a bimodal growth pattern, with an initial mean radius of 199.3 ± 2.18 µm. They experienced an initial decrease of 18.86% in mean radius from day 1 to day 7, followed by a subsequent 9.09% increase until day 21. By the end of the experiment, the mean overall radius decreased by 11.49% (Figure 2B).

HUVEC spheroids exhibited distinct behavior, with an average radius double that of MDSC spheroids, measuring 356.9 ± 25.55 µm, which continuously decreased throughout the 21-day experiment period, with a mean total radius decrease of 54.02% (Figure 2B). No differences were detected between the areas occupied by MDSC and HUVEC under conventional cell culture conditions (Figure 2D).

MDSC/HUVEC spheroids initially had a radius averaging between MDSC and HUVEC spheroids, at 233.2 ± 27.25 µm, and were statistically different from both MDSC and HUVEC on day 1 (Figure 2B, p < 0.05). However, as the experiment progressed, MDSC/HUVEC spheroids tended to behave more similarly to MDSC spheroids, with no statistically significant difference observed at the end of the experiment (Figure 2B).

The Live/Dead assay of MDSC/HUVEC spheroids revealed a mean viability of 87.05% ± 7.915% for HUVEC spheroids, 83.938% ± 2.084% for MDSC spheroids, and 82.53% ± 6.978% for MDSC/HUVEC spheroids on day 1, with no initial statistical differences observed (Figures 2E,F). However, MDSC/HUVEC spheroids exhibited superior viability on day 7 compared to MDSC spheroids, with viabilities of 81.113% ± 0.850% and 74.813% ± 4.15%, respectively (Figures 2E,F). The superior viability of MDSC/HUVEC spheroids remained consistent until the end of the experiment, although not statistically significant.

To evaluate the characteristics of the alginate hydrogel prior to spheroid encapsulation, we monitored its rheometric parameters while immersed in different cell culture media. The clean alginate hydrogels exhibited consistent wet and dry weights throughout the 21-day experiment period, regardless of the cell culture media used. Initially, the mean wet weights of the hydrogels were 30.10 ± 5.216 mg for PM and 25.10 ± 4.078 mg for EGM, with corresponding mean wet weights at day 21 of 19.60 ± 4.491 mg for PM and 20.983 ± 5.071 mg for EGM. Statistical analysis revealed no significant differences between the PM and EGM groups at any time point (Figure 3A). Wet and dry weight (Figures 3A,B) stability led to a stable swelling ratio, with an initial mean of 1.03 ± 0.005 for PM, and 1.024 ± 0.004 for EGM, and a day 21 mean of 1.024 ± 0.013 for PM and 1.020 ± 0.005 for EGM (Figure 3C). Ultimately, our alginate hydrogel sphere had an initial mean mesh size of 52 nm on average, more specifically, 52.281 ± 0.088 nm for PM and 52.191 ± 0.065 nm for EGM (Figure 3D), serving as an initial baseline characterization. Comparing the rheological data obtained from our hydrogel, the hydrogel’s Loss Modulus approximated closely to human gastrocnemius Loss Modulus in vivo rheological assay measurements (Table 1) (Green et al., 2012).

Following encapsulation, MDSC spheroids were cultured within the same alginate hydrogel formulation for 21 days in both PM and EGM (Figure 4A). Initial assessment of wet weight before immersion in specific culture media revealed a mean of 51.714 ± 4.972 mg for both PM and EGM on day 0, already 71%–106% heavier than clean hydrogels. By the end of the 21 days, the mean dry weight for PM was 47.80 ± 3.077 mg, and for EGM, it was 35.92 ± 3.226 mg (Figure 4B). Serial assessments of wet and dry weights demonstrated stability in PM and instability in EGM, with a decreasing trend (Figures 4B,C). This behavior became more pronounced over time (Figures 4B,C, p < 0.05 for EGM day 0 versus 21). Consequently, the swelling ratio of the alginate spheres cultivated in EGM also decreased over time (Figure 4D, p < 0.05 for EGM day 0 versus 21).

Finally, the mesh size of spheroids cultured in EGM significantly decreased by day 21 compared to its initial size, reaching a mean of 52.355 ± 0.051 nm, indicating a reduction of 0.4% (Figure 4E, p < 0.05 for EGM day 0 versus 21). In contrast, the PM group’s mesh size experienced a minimal reduction of 0.13% from its initial mean of 52.623 ± 0.084 nm.

The Live/Dead assay of spheroids within the hydrogel demonstrated initial viabilities of 49.778% ± 13.347% for both groups, assessed before immersion in specific culture media (Figures 4F,G). The viability of PM and EGM groups on day 21 revealed means of 48.905% ± 4.622% and 57.11% ± 10.682%, respectively. Although the 14.72% increase in viability for the EGM group was not statistically significant (Figures 4F,G), EGM viability was significantly superior to PM on day 7 (Figure 4E, p < 0.05 for day 7 p.m. versus EGM). This viability assay thus revealed a stability and growth pattern of EGM-cultured spheroids over time, contrasting with PM-cultured spheroids.

Based on the findings that an endothelial microenvironment (EGM) enhances MDSC stemness (Supplementary Figure S1C) and that direct co-culture with endothelial cells (HUVECs) improves spheroid viability (Figures 2E,F), we hypothesized that inducing an endothelial-like phenotype in the MDSCs themselves could be a superior and more translatable strategy.

To test this, we pre-treated MDSCs with VEGF (50 ng/mL) to promote endothelization prior to spheroid formation as previously described (Verma et al., 2024). We then generated therapeutic spheroids using a 95:5 mixture of naïve MDSCs and VEGF-pre-treated MDSCs (hereafter referred to as MDSC-VEGF), mirroring the optimal ratio used in the MDSC/HUVEC co-cultures.

MDSC-VEGF spheroids were encapsulated as the therapeutic vehicle in the characterized alginate hydrogel, with 10 spheroids per gel, cultivated in EGM for 15 days, and serially assessed for viability, through two different methods. Live/Dead viability revealed that all spheroids were able to preserve viability (Figure 5A, day 2%–82.95% ± 6.551%, day 15%–80.21% ± 7.831%), and exhibited the standard morphological measurements behavior (Figures 5B,C, radius and sphericity) as previously characterized (Figures 2B,C). Quantification of DNA per spheroid revealed an increase in genetic material per spheroid in EGM-cultured spheroids (Figure 5D, day 2–157.057 ± 14.632 pg, day 15–212.273 ± 62.466 pg, p < 0.05, for EGM day 2 versus day 15). Bright-field microscopic aspect of hydrogels is shown (Figure 5E).

To evaluate the therapeutic efficacy of alginate-encapsulated spheroids in a muscle injury model of VML, we performed a surgical protocol involving partial removal of the gastrocnemius muscle (Figure 6). Subsequently, mice were categorized into different groups based on their therapeutic treatment. These groups included mice receiving therapeutic spheroids, which were characterized in the previous section, included spheroids alone (VML+Sph), hydrogel alone (VML+Alg), spheroids encapsulated by alginate (VML+AlgSph) and VML. Histomorphometric investigations were conducted on the injured gastrocnemius muscle, focusing on regions of interest (ROI) under bright-field microscopy following Hematoxylin and Eosin (H&E) staining (Figures 6C, 7; Supplementary Figure S2).

Mice receiving therapeutic spheroids (VML+Sph and VML+AlgSph) demonstrated the lowest quantifications of granulation tissue per injury area, with therapeutic encapsulated spheroids (VML+AlgSph) showing significantly less granulation tissue than spheroids alone (Figures 7A–C; 48.53% ± 6.79% versus 34.08% ± 5.637%, VML+Sph versus VML+AlgSph, respectively, p < 0.05). Additionally, mice receiving therapeutic encapsulated spheroids presented with the fewest number of degenerating myofibers per injury area, compared to the VML group and VML+Alg groups (Figures 7A–D; 34.97% ± 6.173% for VML, 33.32% ± 8.270% for VML+Alg, 9.317% ± 5.277% for VML+AlgSph, p < 0.05).

Conversely, mice receiving spheroids exhibited the highest quantities of regenerating myofibers per injury area (Figure 7E; 15.78% ± 5.885% for VML, 1997% ± 8.376% for VML+Alg, 50.85% ± 9.452% for VML+Sph, 5988% ± 8.332% for VML+AlgSph; p < 0.05). Bright-field microscopy of fibrotic regions of interest following Picrosirius red (PSR) staining is displayed in Figures 7B–F. Quantification of fibrotic area per ROI revealed that mice receiving spheroids (VML+Sph and VML+AlgSph) had the lowest percentages of fibrosis, though they were not statistically different from each other (Figures 7B–F; 21.73% ± 4.622% for VML, 19.39% ± 3.959% for VML+Alg, 12.77% ± 1.970% for VML+Sph, 10.84% ± 1.566% for VML+AlgSph, p < 0.05).

No significant differences in the gastrocnemius muscle to mice body weight ratio were detected (Figure 7G). Additionally, the Rota-Rod test showed that mice receiving spheroids alone or encapsulated by alginate performed significantly better than the VML group 15 days after the injury (Figure 7H, p < 0.05). Curiously, the group injected only with alginate also showed an improved performance at the Rota-Rod, compared to VML group (Figure 7H, p < 0.05). After 30 days there was no difference among the groups.