Equine bone marrow derived MSCs (BM-MSCs) were administered twice to MHC-matched and MHC-mismatched recipient horses, and the systemic cellular immune response was serially analyzed using a modified one-way MLR system. Three horses homozygous for the MHC-haplotype (HapPRE10/HapPRE10, HapPRE11/HapPRE11, and HapMAI04/HapMAI04) were selected as MSC donors (17). The recipient horses were categorized into two groups: MHC-matched and MHC-mismatched with the donors. The MHC-matched group included eight heterozygous horses that shared one haplotype with one of the donors: three HapPRE10 heterozygotes, three HapPRE11 heterozygotes and two HapMAI04 heterozygotes. For the MHC-mismatched group, nine horses were selected, all of them carrying haplotypes other than HapPRE10, HapPRE11 and HapMAI04. Within each group of recipients (MHC-matched and MHC-mismatched), animals were arranged into three subgroups of 2–3 animals. Each subgroup received either basal MSCs (MSC-naïve), proinflammatory-primed MSCs (MSC-primed) or chondrogenically differentiated MSCs (MSC-chondro). In addition, homozygous donors (n = 3) acted as autologous controls receiving each one MSC-naïve, MSC-primed or MSC-chondro (Figure 1).

Each animal received simultaneously three alginate hydrogel scaffolds, each one containing 5 × 10⁶ MSCs of the corresponding type (MSC-naïve, MSC-primed or MSC-chondro). The scaffolds were placed in the neck by creating subcutaneous pockets. This approach was taken to retain the MSCs in a specific anatomic site and minimize interferences of their migration. In addition, this system allowed recovering the scaffolds to be used in a separate study assessing the local response (data not published). In the present study, peripheral blood was collected from each animal to obtain lymphocytes for the one-way MLR assays. This was done prior to scaffold placement (T0) and at weeks 1 (T1), 3 (T2) and 6 (T3) following the implantation. One month later, each horse was subjected to the same procedure to mimic the effect of a second administration: each animal received three new scaffolds in the contralateral neck side, containing the same type and dose of MSCs than in the first administration. Blood was collected at the same time-points (T4: prior to MSC re-exposure; T5: 1 week after the re-exposure; T6: 3 weeks after the re-exposure and T7: 6 weeks after the re-exposure). Schematic timeline and methods are shown in Figure 2A.

A modified one-way MLR assay was used to evaluate the cellular immune response at the systemic level and the possible development of cellular-mediated immune memory. Following blood collection, peripheral blood lymphocytes (PBLs) were obtained from all the recipients and at all the time-points indicated. The PBLs of each recipient were re-exposed in vitro to the same type of MSCs that were previously administered in vivo (MSC-naïve, MSC-primed, MSC-chondro; MHC-matched or MHC-mismatched). In this co-culture system, PBLs were not mitogen-activated in order to assess their responsiveness upon re-encountering with the same MSCs. Such response was assessed in terms of proliferative response of different lymphocyte subsets and production of IFNγ by T cells. To do this, PBLs were stained with carboxyfluorescein succinimidyl ester (CFSE) to evaluate their proliferation after 5 days of co-culture with MSCs. At this point, PBLs were collected and labelled with a panel of antibodies to analyze by flow cytometry the changes in the frequency and proliferation of the main lymphocyte subsets (T cells: cytotoxic, helper and regulatory, and B cells). At the same time, co-culture supernatants were collected to determine the secretion of IFNγ by ELISA, serving as an indicator of T cell activation (Figure 2B).

Twenty healthy horses aged between 2 to 12 years (12 geldings, 7 mares and 1 stallion), with no previous history of receiving MSCs or transfusions, were selected based on their MHC haplotypes as determined by microsatellite typing using a validated panel of 10 highly polymorphic intra-MHC regions, as previously described (12, 35). The methodology for DNA isolation, multiplex polymerase chain reaction (PCR), and fragment analysis was performed exactly as previously reported by our group (17).

Definitive haplotypes were established for animals homozygous or with direct relatives (parent-offspring pairs or half-sibling groups) and the remaining animals were assigned with provisional haplotypes based on previously known ones reported in bibliography (12, 14, 35, 36). Supplementary material S1 shows the microsatellite alleles of each haplotype identified in the different horses.

All the procedures involving animals were carried out under the Project License PI 15/16 approved by the in-house Advisory Ethics Committee for Animal Research from the University of Zaragoza. The care and use of animals were performed accordingly with the Spanish Policy for Animal Protection RD118/2021, which meets the European Union Directive 2010/63. All animals were kept on paddocks of the facilities of the Animal Research Service of the University of Zaragoza, with free access to water and fed ad libitum with grass hay.

Equine BM-MSCs were obtained and characterized, as previously described (23), as part of a previous study of our group (17). In brief, equine MSCs (n = 3) were isolated and expanded at 5,000 cells/cm² in conventional MSCs medium consisting of low glucose Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 0.1 mg/mL streptomycin and 100 U/mL penicillin (all form Sigma–Aldrich) at 37°C and 5% CO₂.

Mesenchymal stem/stromal cells were used between passage two to four. Cells were detached using trypsin-EDTA 0.25% (Sigma–Aldrich), suspended at a concentration of 5 × 10⁶ MSCs/mL in phosphate buffered saline (PBS, Gibco, Thermo Fisher), and combined in a 1:1 ratio with a 3% alginate solution (ultrapure low-viscosity 67% guluronate, UPLVG; NovaMatrix, FMC Corporation). This resulted in a final concentration of 1.5% alginate in a total volume of 2 mL per scaffold as previously described (32), containing 5 × 10⁶ MSCs each scaffold. Scaffold polymerization was achieved as described earlier (37–39) by incubating them for 30 min with a solution of 102 mM CaCl₂ (Sigma–Aldrich) at 37°C and 5% CO₂. For MSC-naïve and MSC-primed, scaffolds were maintained in conventional MSC culture medium for 48 h. For MSC-primed scaffolds, 12 h before their in vivo placement, 5 ng/mL of TNFα and IFNγ (R&D Systems) were added to the conventional MSC media according to previous reports (22). For MSC-chondro, chondrogenesis was induced according to the methodology previously described (18, 37) using 10 ng/mL of TGFβ-3 (PeproTech) for 21 days.

To remove xenogeneic antigens from the FBS that might affect the immune response, 24 h prior to the in vivo placement of the scaffolds, all of them (MSC-naïve, MSC-primed, and MSC-chondro) were washed twice with PBS and fresh culture media was added containing 10% autologous serum from each recipient to replace the FBS (40). Right immediately before the intervention, scaffolds were washed again with PBS three times.

Scaffolds were placed along the neck tables, from cranial to caudal, following the previously described subcutaneous pocket technique (41). The surgical intervention was carried out on station under appropriate sedation (detomidine 0.01 mg/kg, Sedaquick, Fatro; and butorphanol 0.02 mg/kg, Torbugesic, Pfizer) and local anesthesia (lidocaine 5%, Braun). For the placement of each scaffold, a longitudinal incision of approximately 2 cm in length was performed, leaving about 10 cm between incisions. Subcutaneous tissue was dissected distally to the incision to create a subcutaneous pocket of approximately 2.5 × 2.5 cm, where each scaffold was placed. The incisions were closed in two layers (subcutaneous and skin) with 2/0 USP polyglyconate suture and surgical staples. This method of administration was chosen to facilitate the analysis of implant responses in a separate study. All horses were clinically monitored and received a single dose of flunixin meglumine IV (1.1 mg/kg, Niglumine, Calier) right before surgery and procaine penicillin IM (15 mg/kg once daily for 3 days, Procapen, Livisto) postoperatively.

One-way MLRs were conducted before and serially after each MSC administration for a total of 8 time-points (T0 to T7). The co-cultures were carried out following the methodology previously validated by our group (17).

After 5 days in modified one-way MLR co-cultures, the PBLs were collected from the 24-well plates, centrifuged at 310× g for 5 min, resuspended in PBS, and split for the two flow multi-color panels. The supernatants were collected and centrifuged at 500× g for 15 min to remove any contaminating cell and were immediately frozen at −20°C for further ELISA analysis, as will be detailed below. Control PBLs (alone) and their supernatants were processed in the same way.

Two multi-color panels of markers were previously designed by our group to allow assessment of different lymphocyte subpopulations, along with the proliferation (CFSE dilution) (17). In panel 1, anti-CD3 (primary rat anti-human, 1:100, MCA1477, Bio-Rad; secondary mouse anti-rat, 1:100, 12-4812-82, Invitrogen), anti-Pan Ig (primary mouse anti-horse, 1:100, MCA1899, Bio-Rad; secondary goat anti-mouse, 1:200, A-21036, Invitrogen) and anti-CD21 (mouse anti-human, 1:40, 561357, BD Pharmigen) were used to assess the global T and B cell populations. In panel 2, anti-CD8 (mouse anti-horse, 1:5, MCA2385, Bio-Rad), anti-CD4 (primary mouse anti-horse, 1:200, MCA1078, Bio-Rad; secondary rat anti-mouse, 1:200, 406619, BioLegend) and anti-CD25 (primary goat anti-human, 1:50, AF-223, R&D Systems, secondary donkey anti-goat, 1:400, 705-605-003, Jackson Immuno Research) antibodies were used to assess cytotoxic, helper and regulatory T cells, respectively.

All the primary antibodies were selected based on previous reports and previously checked to correctly label equine cells. Primary antibodies were used directly conjugated with fluorochromes or in combination with appropriate secondary antibodies (34, 45, 46). The methodology for staining the equine PBLs with these antibodies has been described in detail in a previous publication of our group that can be openly accessed, and was exactly the same than in this study (17).

All samples were analyzed in a Gallios flow cytometer (Beckman Coulter, Madrid, Spain), acquiring a minimum of 10,000 events per sample. Flow cytometry data was analyzed with FCS Express 7 Flow software (De Novo Software, Pasadena, CA, United States). Unstained PBLs and secondary controls (cells stained with secondary antibodies alone) were used to assess fluorescence background and to establish gates for each marker. Compensation controls were performed using single color stains, while fluorescence minus one (FMO) controls were used to determine the fluorescence spread from other channels. Viability staining was performed with Ghost dye Violet 450 (Tonbo Biosciences, Bio-Rad) for all PBLs.

The gating strategy was the same than in a previous report from our group (17). Briefly, the lymphocyte population was initially gated in the forward and side scatter (FSC × SSC) plot, with doublets being excluded from analysis, followed by exclusion of dead cells that incorporated the viability stain. In the first panel, live cells were further gated as T cells (CD3⁺) or B cells (CD3⁻/Pan Ig⁺/CD21⁺) (47). In panel 2, live cells were gated to differentiate between cytotoxic T cells (CD8⁺/CD4⁻) and helper T cells (CD4⁺/CD8⁻). Furthermore, the subpopulation of regulatory T cells (Treg) was gated from the CD4⁺ cell population based on their high expression of CD25. The subpopulation of CD4⁺ CD25^high is referred as Treg cells in this study for clarity; however, it should be noted that other subpopulations may be present in this gate.

Frequency (%) of each lymphocyte subset over the total lymphocyte population was recorded for all recipients at all time-points. To evaluate the proliferation of PBLs in each specific lymphocyte subset, CFSE dilution was analyzed in cells gated as aforementioned. Autofluorescence (background) was determined using unstimulated and unstained PBLs. Unstimulated and CFSE-labeled PBLs were used to establish the non-proliferating population, considering cells to the left (lower fluorescence intensity) as the proliferating population.

To account for inter-individual variability, the percentage and proliferation of each lymphocyte subpopulation after exposure to MSCs were normalized to the values from classic MLR-matched controls (assigned as value 1). Unspecific proliferation was not observed in the MLR-M controls at any time-point.

Supernatants collected from the one-way MLR assays and corresponding controls were used to evaluate IFNγ production by using a commercially available ELISA kit (Equine IFN-gamma DuoSet ELISA, R&D Systems, REF: DY1586), as previously reported (17, 44, 48).

The supernatants from the classic MLRs with MHC-matched or mismatched PBLs as stimulators were used as the negative and positive control, respectively. All supernatants were diluted 1:1 in reagent diluent. All the procedures were performed as per manufacturer’s instructions and concentrations determined using a standard curve, including a blank.

The standard curve was set from 62.5 pg/mL to 8,000 pg/mL of IFNγ. All the samples and points of the standard curve were run in triplicate. All the colorimetric assays were analyzed on a microplate reader (Biotek Synergy HT, Winooski, VA, United States) and read immediately at 450 nm with wavelength correction set to 540 nm. The duplicate readings for each standard, control, and sample were averaged, and the average zero standard optical density was extracted. The standard curve was created generating a four-parameter logistic curve-fit and the concentrations extrapolated were multiplied by the dilution factor.

Statistical analysis was performed with IBM SPSS Statistics version 26 statistical package. Analytical statistical tests were used to test for differences in each lymphocyte subpopulation and in IFNγ secretion as the dependent variables of the study.

The independent variables were “MHC group” (two categories: MHC-matched and MHC-mismatched), “MSC type” (three categories: MSC-naïve, MSC-primed and MSC-chondro) and “time” (eight categories: T0, T1, T2, T3, T4, T5, T6, T7). Each dependent variable was analyzed individually and each of the independent variables was analyzed as a factor, to study the differences as follows: differences between MSC types at each time within each MHC group of recipients, differences between MHC groups of recipients for each MSC type at each time, differences over time within each MHC group of recipients for each MSCs type.

The existence of outlier samples was evaluated with the Grubbs test (alpha = 0.05) and the results were analyzed by Shapiro–Wilk test to assess normality of data. Levene’s test was used to test the equality of variances. When data followed a normal distribution and had homogeneous variances, the parametric test ANOVA was used, followed by Bonferroni comparisons test as a post hoc. In normally distributed data with unequal variances, Welch’s t-test was used. In non-normal data and variables with more than three groups, Kruskal–Wallis or Friedman tests followed by Dunn’s test were used as post hoc, for independent (differences between MSC types) or related (differences over time) samples, respectively. The Mann–Whitney U test was used for comparisons between two groups (differences between MHC groups). The significance level was set at p < 0.05 for all analyses. GraphPad Prism 9.2 was used for graphical representation (San Diego, CA, United States).

This study evaluated the effect of two allogeneic administrations of MSCs under different conditions (naïve, primed and chondro) and under different MHC combinations (matched and mismatched) on the response of circulating PBLs. To this end, the proliferation of different lymphocyte populations from recipient horses was measured at different time-points using an in vitro co-culture system with the same MSCs administered in vivo.

The concentration of IFNγ was measured in the co-culture supernatants from all the modified one-way MLR assays as a reflection of T lymphocyte activation. Overall, MSC-primed, either MHC-matched or MHC-mismatched, induced more secretion of IFNγ than the other MSC types and than the positive control (classic MLR-mismatched) (Figure 8). Specifically, the IFNγ secretion induced by MSC-primed was significantly higher compared to MSC-chondro in both MHC-matched and MHC-mismatched co-cultures before and after the first administration (T0, p < 0.05 in both cases; T1, p < 0.01 in both cases; T2, MHC-matched, p < 0.01 and MHC-mismatched, p < 0.05; T3, p < 0.05 in both cases), and right before the second administration (T4, p < 0.05 in both cases) (Figures 8A,B). Moreover, after the second administration, MSC-primed MHC-mismatched also induced higher secretion of IFNγ over MSC-chondro (T6, p < 0.05; T7, p < 0.01) (Figure 8B).

The secretion of IFNγ induced by MSC-primed was also higher compared to MSC-naïve in both MHC-matched and mismatched settings, and both before (T0, p < 0.05) and after the first administration (MHC-matched: T2, T3 and T4, p < 0.05; MHC-mismatched: T1, p < 0.05) (Figures 8A,B). After the second administration, IFNγ induction by MSC-primed over MSC-naïve was overall higher in the MHC-mismatched group (T7, p < 0.01) (Figure 8B) but not in the MHC-matched group (Figure 8A). Although the IFNγ concentration detected in the MSC-naïve co-cultures was consistently higher than that observed for MSC-chondro, the difference between them was only significant in the MHC-matched co-cultures at 1 week after each administration (T1, p < 0.01; T5, p < 0.05) (Figure 8A).

When assessing the response to each cell type, MSC-chondro only induced mild secretion of IFNγ, but there were some interesting findings (Figure 8C). Levels of IFNγ were generally similar to the negative control (unstimulated PBLs alone) in the MHC-matched recipients along the time. However, in the MHC-mismatched group, IFNγ levels were higher than the negative control up to T5 and subsequently decreased (T6 and T7 over T1, p < 0.05; T6 over T2, p < 0.01; T6 over T5, p < 0.05; T7 over T3, p < 0.05). Moreover, IFNγ concentration in MHC-mismatched co-cultures was overall higher than in the MHC-matched ones (T1 and T5, p < 0.05).

Regarding MSC-primed (Figure 8D), IFNγ values were very similar between autologous and both allogenic MHC-matched/mismatched co-cultures after the first administration. However, after the second administration, IFNγ values significantly increased in the MHC-mismatched group (T7 over: T0, T1, T4 and T5, p < 0.05) while in the MHC-matched group IFNγ values decreased compared to the first administration (T5 over T1; T6 over T2, p < 0.05). Thus, IFNγ production after the second administration was consistently lower in the MHC-matched co-cultures, but significant differences over the MHC-mismatched co-cultures were only found at the end of the study (T7, p < 0.05).

For MSC-naïve (Figure 8E), even though MHC-mismatched co-cultures tended to induce more IFNγ, the highest IFNγ secretion was observed in MHC-matched co-cultures at 1 week after the first administration (T1 over: T0, T3, T4 and T5, p < 0.05; over MHC-mismatched group, p < 0.01). Aside from this observation, IFNγ secretion was higher in the MHC-mismatched co-cultures after the first administration (T3 over T0, p < 0.05; over MHC-matched group, p < 0.05) and it further increased after the second administration (T7 over T0, T1 and T3, p < 0.05; over T4, T5 and T7 MHC-matched group, p < 0.05).

We found a number of important observations: first, that equine MSCs under different conditions (priming, chondrogenic differentiation, basal) interact differently with the immune system in vivo. We hypothesized that MSCs primed with pro-inflammatory cytokines would be more efficient evading immune recognition while chondrogenically differentiated MSCs would be more easily recognized. In contrast, we observed that MSC-primed exhibited higher immunogenicity compared to MSC-chondro and MSC-naïve in terms of T cell response; however, MSC-primed also induced Treg and barely provoked a response by B cells, while the contrary was observed for MSC-chondro.

Secondly, this study shows that the compatibility for the MHC between donor and recipient is key in the systemic cellular immune response to equine MSCs in vivo. According to our initial hypothesis, MHC-mismatched MSCs of any type tended to induce higher T lymphocyte proliferation compared to MHC-matched MSCs; however, this response was not so marked for B cells. Of note, the response of T cell subsets was barely observed when analyzing their relative frequencies (Supplementary material S2), but it was clear when studying their proliferation rates. Specifically, MHC-mismatched recipients showed the greatest response for most of the T cell populations and at almost all time-points when receiving MSC-primed, and after the second administration of MSC-naïve. Additionally, although MSC-chondro induced the lowest effect on T cells, MHC-mismatched recipients of MSC-chondro showed greater changes in the cytotoxic response compared to MHC-matched ones. Previous studies in vitro also observed higher proliferation of T cells when exposed to MHC-mismatched equine MSCs that expressed high levels of MHC-II, in comparison to either MHC-matched MSCs (13) or MHC-mismatched MSCs but with low MHC-II expression (34). However, other works found that allogeneic MSCs (unknown MHC) were not inherently more immunogenic than autologous MSCs, as both types induced a similarly mild lymphocyte proliferation in vitro (33). This reinforces the notion that the results observed in vitro do not always accurately reflect the immune response that occurs in vivo, which leads to our third main finding.

We observed that cellular immune memory is generated in vivo by equine MSCs. This is based on the observation that equine lymphocytes responded more markedly to any type of MSCs in the one-way MLRs after the animals had received the MSCs, compared to the baseline (T0, prior to any MSC administration). This response varies along the time and, importantly, it is exacerbated by repeated administration of the same MSCs. However, interestingly, while MSC-primed generally tended to be more immunogenic, the cellular response after their second administration was similar to the first one. In contrast, the response elicited by MSC-naïve and MSC-chondro was further increased after a second administration. These findings also highlight the relevance of conducting in vivo studies to evaluate the immunogenicity of equine MSCs, and not to rely only on in vitro assessments using lymphocytes from animals that have not been exposed to MSCs. Moreover, evaluating only the frequencies of lymphocytes after MSC administration may not be enough to reveal the immune response, as we and others (17, 46) have found limited changes in the percentage of each lymphocyte subset while their proliferative response was marked and thus can provide further information.

Prior to engaging into more detailed discussion of our findings, it is important to acknowledge the limitations of this study. First, the sample size of animals is limited due to the challenges inherent to working with large animals like horses (51). Furthermore, finding animals that were matched for the MHC added additional complexity provided the high diversity of MHC haplotypes (35). The reduced sample size of the study could prevent extrapolating definitive recommendations to the general equine population, but considering the consistent trends observed for several of the analyses performed, our results can provide valuable insight into the cellular systemic immune response when administering equine MSCs. Second, the autologous group consisted of only three horses (one per MSC type), as they were involved solely for control purposes, similarly to previous works (40, 52). Third, the route of administration of MSCs (subcutaneous placement of hydrogels) does not accurately reflect clinical administration. Subcutaneous administration was chosen to make MSCs easily reached by the immune system as other locations (e.g., joints, eyes) may be less accessible, and the scaffold system was selected to immobilize the cells in a definite anatomic location and prevent their loss. In addition, this approach allowed recovering the scaffolds to be used in a separate study assessing the local response induced by the MSCs (unpublished data). However, the hydrogel could have also constituted a physical barrier for the interaction of MSCs and immune cells (41). Fourth, as aforementioned, the Treg subset could not be analyzed at some time-points due to supply issues of the anti-CD25 antibody, for reasons out of our control. Fifth, the possible interference of FBS antigens in the study cannot be completely disregarded, in spite of using the FBS removal strategy explained above that included exchanging FBS by autologous serum in the MSC culture for 24 h prior to administration, and several PBS washes. Moreover, to account for the possible effect of FBS antigens, or other unspecific antigens, we run several controls. It is worth noting that none of the negative controls (PBLs cultured alone and MLR-M) showed unspecific PBL proliferation, neither before administration nor at any time point of the study. Sixth, different methods for chondrogenic differentiation were used for the in vivo administrations and for the in vitro assays, thus potentially adding certain variability between the immune response produced in vivo and the proliferative response studied in vitro. The use of different protocols was needed to meet the requirements of this study for in vivo administration (MSCs embedded in an alginate hydrogel) and for in vitro cocultures (direct contact between MSC-chondro and PBLs, and control of the ratio between cell types). Moreover, to limit the possible interference of using different differentiation methods, we used MSC-chondro cells from the same donor and same passage in both assays and we previously compared the chondrogeneic outcome of both protocols (data not shown).

According to our results, the systemic response of T cells against equine MSCs in vivo is influenced by the MHC compatibility and this response is dynamic, varying along the time after administration. The response of CD3⁺ lymphocytes also reflects how the immune recognition of MSCs in vivo depends on the pre-treatment done to the cells: MSC-naïve may be more quickly recognized after a second administration, while MSC-primed may initially evade this response, which might be attribute to a potentially higher regulatory capacity. While studying the general CD3⁺ T population provides important insight into the immune response (33, 53), dissecting T cell subsets reveals further differences.

We observed that equine MSCs can induce in vivo cytotoxic and helper T cell populations that are able to elicit a proliferative response when they encounter with the same MSCs again. In other words, cellular memory is generated against allogeneic equine MSCs. After the first administration, MSC-primed tended to induce a stronger CD8⁺ and CD4⁺ T cell response regardless of the MHC compatibility, while MSC-chondro and MSC-naïve did not seem to induce memory in the CD4⁺ population. Similarly, previous research (54) neither observed changes in the frequency of CD4⁺ population after in vitro re-exposure to MSCs at 120 days post single intra-articular injection of autologous or allogeneic equine MSC-naïve. However, lymphocyte proliferation was not assessed in that study, which could have revealed further changes (17). In contrast to the first administration, the second administration of MSC-chondro and MSC-naïve induced the response of both helper and cytotoxic T cells, but only in the MHC-mismatched group. However, interestingly, a second administration of MSC-primed did not increase the response of cytotoxic and helper T cells over the first administration. In spite of this, MSC-primed would still be the most immunogenic in our study based on T cell response. This increased immunogenicity could be attributed to the overexpression of MHC-I and MHC-II following MSC priming (13, 55). On the other hand, the enhanced immunomodulatory capacity of MSC-primed (19) could explain the slower and more moderate response of CD4⁺ and CD8⁺ T cells after their second administration.

To investigate deeper the response of T cells to equine MSCs, the supernatants from the in vitro co-cultures were analyzed to determine the presence of IFNγ, a pro-inflammatory cytokine produced by T lymphocytes that is associated with cell-mediated immunity. The secretion of IFNγ indicates the expansion of CD8⁺ or CD4⁺ effector and memory cells against donor MSCs (9). In agreement with the CD8⁺ and CD4⁺ T cell proliferative response, our results for IFNγ secretion also suggest that priming MSCs with pro-inflammatory cytokines can increase their immunogenicity, while MSC-chondro would produce the lowest T cell response. However, it cannot be completely disregarded the possible presence of residual exogenous IFNγ from the priming process, which might have interfered with the results.

Considering overall the T cell response, MHC-matched MSCs of any type prevented or moderated the proliferative and IFNγ production responses of helper and cytotoxic cells, especially after the second administration. The lower immunogenicity exhibited by MSC-chondro when administered in MHC-mismatched horses is consistent with previous reports using repeated administrations of allogeneic MSCs chondrogenically pre-differentiated in the horse (27). However, that study did not account for MHC compatibility, and our results show that MHC-mismatched MSC-chondro induced a significantly higher response than MHC-matched cells. Therefore, it remains to be determined whether the therapeutic effects of chondrogenically differentiated MSCs would be facilitated by MHC-matching. When administering MSC-naïve, the stronger cellular immune response after a second administration in MHC-mismatched recipients could have clinical implications, such as in terms of an adverse event (e.g., inflammatory response) and of potentially diminishing the therapeutic effects of MSCs because of their targeting for elimination. Thus, it would be advisable to consider using MHC-matched MSCs for repeated treatments. Selecting MHC-matched donors would be particularly important if cytokine priming strategies are used (1, 56) or if MSCs are to be injected in a highly inflammatory site, since we observed that MSCs exposed to pro-inflammatory cytokines may lead to an increased T cell response in vivo in MHC-mismatched recipients, even though the second administration of these cells might be better tolerated in terms of T cellular response.

An increase in Treg would be related to enhanced immunosuppressive ability of MSCs, as this subset of lymphocytes would help in dampening the adaptive immune response and preventing rejection of foreign cells by the host (57). As for cytotoxic and helper responses, MSC-primed also produced the highest induction of regulatory T cells. Actually, the response along the time of Treg to MSC-primed in MHC-mismatched recipients suggests that the induction of this population may be a later event, which might be related with the more limited immunogenicity of MSC-primed after the second administration. Similarly, MHC-mismatched recipients of MSC-naïve showed the strongest Treg response steadily over time after the second administration, when the cytotoxic and helper responses were also higher. On the other hand, MSC-chondro, which had the mildest effect on CD8⁺ and CD4⁺ T cells, also presented a limited effect on the Treg subset. Therefore, equine MSCs may be eliciting an in vivo Treg response proportional to their immunogenicity, thus possibly suggesting a compensatory regulatory mechanism for the situations in which MSCs are more immunogenic (58). However, in the case of MSC-naïve, Treg induction would not limit the cytotoxic and helper response, and recipients of MSC-chondro did not show Treg induction corresponding to the time-points with an increased helper and cytotoxic response. Therefore, it might also be suggested that MSCs under different conditions would employ distinct mechanisms of immune evasion.

The principal function of B lymphocytes is producing antibodies against foreign antigens (59). In contrast to Treg, MSC-primed produced minimal activation of B cells while MSC-chondro led to the most marked induction of this population. In agreement with our results, a previous study in rats (18) reported a significant increase in B cells when these were re-exposed in vitro to the chondrogenically differentiated MSCs that were firstly administered in vivo. On the contrary, the only in vivo study in the equine species that has investigated the capacity of chondrogenically induced allogeneic MSCs (ciMSCs) to activate B cells (CD138⁺ plasma cells), showed no response when re-exposing lymphocytes to ciMSCs in vitro (27). Of note, such co-cultures were carried out at variable times after administration (5 days to 1.5 years) and the MHC-haplotype was not examined.

While the type of MSCs influenced the response of B cells as also observed for T cells, the MHC compatibility did not seem to significantly affect their response, particularly in the case of MSC-chondro. The second administration of MSC-naïve and MSC-chondro produced an increase in B cell response, compared to the first exposure, in both MHC-matched and mismatched recipients. However, in MSC-chondro, this B cell induction was similar between MHC-matched and mismatched recipients. This could possibly indicate that chondrogenically differentiated equine MSCs might induce immune memory even in the absence of foreign MHC molecules. While we do not have a clear explanation for this finding, specific immunomodulatory mechanisms might be operating in MSC-chondro and/or the response of B cells might be mainly raised by antigens other than the MHC in differentiated cells. In addition, certain technical aspects of the one-way MLR system might have influenced the proliferation of B cells, such as the ratio MSC-chondro:PBLs, the potential presence of undifferentiated MSCs in the co-cultures, or the impact of TGF-β3 on MSC immunogenic profile. These considerations were also made in our earlier in vitro work (17), and their potential contribution to the present outcomes cannot be totally disregarded.

On the contrary, MSC-naïve produced a higher response in MHC-mismatched over MHC-matched recipients, but only after the second administration. First administration of MSC-naïve produced low proliferation rates for B cells, comparable to the baseline, in both MHC-matched and mismatched recipients, in agreement with previous studies (34). However, after repeated administration of MSC-naïve into MHC-mismatched animals, B cell proliferation progressively increased. Along with the T cell response observed, it might be suggested that MHC incompatibility is more critical for repeated administrations of MSC-naïve than for MSC-chondro and MSC-primed. This may suggest that the regulatory mechanisms of basal MSCs (MSC-naïve) may not be enough to evade the immune system, and thus manipulation prior to administration (differentiation, priming) may enhance their effects in vivo, as it has been suggested for treating equine joint pathologies (22, 27).

In contrast to MSC-naïve and MSC-chondro, MSC-primed showed the lowest B cell response consistently over the time and even after the second administration, suggesting that repeated administrations of MSC-primed would not induce B cell memory. Moreover, MSC-primed neither produced differences between MHC-matched and mismatched recipients. In agreement with our results, a previous in vitro study with human MSCs (60) observed that MSC-primed inhibited the proliferation of B cells over MSCs-naïve. These findings might be related with immune tolerance mediated by Treg, which were strongly induced by MSC-primed as discussed above. Therefore, it seems that MSC-primed may have unique immunomodulatory properties, in spite of their immunogenicity, that might prevent them from triggering a B cell immune response.