Umbilical-Cord mesenchymal stem cells (UC-MSC) were provided by Chiesi Farmaceutici S.p.A, Italy. Cells were stored in liquid nitrogen. UC-MSC were received in passage number 4, and not passaged further. For every experiment a new vial was used.

UC-MSC were cultured in complete MSC growth medium provided by Chiesi Farmaceutici S.p.A, Italy which contains 2000U/ml heparin (Leo Pharma B.V., Copenhagen, Denmark) and 5% human platelet lysate (provided by Chiesi Farmaceutici S.p.A, Italy). According to the standard protocol, on the day of UC-MSC seeding, cells were thawed using a 37° C water bath for 4-6 minutes. Cells were centrifuged at 500 x g for 10 minutes at room temperature. Supernatant was removed and 15 mL complete MSC growth medium was added to the cells. Flasks were placed in a 37°C, 5% CO₂, 95% humidity incubator for 48 h. After 48 h, UC-MSC culture, medium was aspirated, and cells were washed with sterile PBS (Fresenius Kabi, Bad Homburg, Germany). UC-MSC were harvested by using 3 U/ml trypsin/EDTA (Gibco^®Thermo Fisher Scientific, Waltham, USA). Trypsinization was neutralized by using cold basal MSC culture medium. Cells centrifuged at 300 x g for 10 min at 4°C. Cells were washed after aspirating the medium and counted with trypan blue for further usage.

UC-MSC were seeded in T75 flasks with complete MSC growth medium. 24h later, the cells were stimulated with BPL (Thermo Fisher Scientific) in a 1:200 dilution factor (v/v). BPL-treated cells were incubated at 4° C for 16 h, followed by a 4 h incubation at 37° C to inactivate BPL. After treatment with BPL, cells were washed twice with PBS to remove leftover product. Since BPL fixates the cells, UC-MSC were counted manually using trypan blue staining.

UC-MSC were seeded in T75 flasks with complete MSC growth medium, 24h later, complete MSC growth medium was removed and replaced by complete RPMI 1640 (Life technologies), supplemented with 10% human processed serum (HPS) (manufactured in-house), 1 mM pyruvate, 2 mM glutamax, 100 U/ml penicillin, and 100 μg/ml streptomycin (all Thermo Fisher Scientific) for the following 24 h. After a total of 48 h of culture, cellular debris was removed by centrifugation for 15 minutes at 2000 x g, 4°C.

To evaluate the metabolic activity of BPL-stimulated cells, 10.000 cells were seeded per 96-well flat-bottomed plate (Greiner Bio-One, Frickenhausen, Germany) in 150 µl complete MSC growth medium. Cells were left overnight for attachment and 24 h later, cells were centrifuged at 250 x g for 2.30 minutes, and supernatant was removed. 100 µl fresh culture medium with 10 µl of 12mM 3-(4,5-dimethylthiazol-2-yl)-2,5 ditetrazolium bromide (MTT) (Thermo Fisher) was added to each well. The cells were incubated at 37° C for 4 h. Subsequently, cells were centrifuged again at 250 x g for 2.30 minutes, medium was removed and 50 µl DMSO was added to each well. Plates were incubated at 37° C for 10 minutes. Absorbance was measured at 540 nm at TECAN infinite F50 (TECAN, Switzerland).

Blood samples for peripheral blood mononuclear cell (PBMC) isolation were collected in 10 ml EDTA tubes (BD, Labware, NJ). Blood was obtained from healthy adult individuals and prior to sampling blood, donors signed a written informed consent for scientific use according to Dutch law. PBMC were isolated using density gradient centrifugation (Lymphoprep, Nycomed Pharma AS, Oslo, Norway). CD4⁺ T cells were isolated by negative selection using Miltenyi Biotech CD4⁺ T cell Isolation Kit (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), according to manufacturer’s protocol. The purity of the sorted CD4⁺ T cells was typically above 95%. Viability of PBMC and CD4 was above 80% at the beginning of the co-cultures. Isolated cells were cultured in complete RPMI 1640 medium.

In our preliminary studies, two UC-MSC : PBMC cell ratios were tested, 1:2.5 and 1:5. CD4⁺/CD8⁺ ratio, CD4⁺Ki67⁺ and CD8⁺Ki67⁺ percentages were tested, whereby the results using the 1:5 UC-MSC : PBMC ratio were more outspoken (data not shown). Therefore, the following experiments were performed with the 1:5 ratio (UC-MSC: PBMC or UC-MSC: CD4⁺).

100.000 CD4⁺ T cells were seeded in 96 well U-bottom plates (Greiner Bio-One) with complete RPMI medium. CD4⁺ T cells were activated with αCD3/CD28 beads (Dynabeads™ Human T-Activator, Gibco®Thermo Fisher Scientific) in 1:5 (bead-to-cell) ratio. After 48 h of culture 20.000 UC-MSC were seeded in 96-well U-bottom plates with complete MSC growth medium. Both CD4⁺ T cells and UC-MSC were kept in culture for 24 h to allow for activation and attachment, respectively. The following day, growth medium of UC-MSC was removed, CD4⁺T cells were resuspended in complete RPMI medium and transferred onto seeded UC-MSC. Cells were cultured together for up to 5 days, and flow cytometry measurement was acquired every other day.

250.000 PBMCs were seeded into 24 well plates (Greiner Bio-One) and activated with αCD3/CD28 beads (Dynabeads™ Human T-Activator, Gibco®Thermo Fisher Scientific) in 1:5 (bead-to-cell) ratio. In addition, 48h after pre-culturing, 50.000 UC-MSC were seeded in transwell inserts (ThinCerts™ Greiner Bio-One) with complete MSC growth medium and allowed to adhere overnight. 24h later, complete MSC growth medium was removed from the transwell inserts, and medium replaced by complete RPMI medium. Inserts were placed onto wells containing αCD3/CD28 activated PBMC. Cells were cultured for up to 5 days.

Samples were transferred to V-bottom plates (Greiner Bio-One) and washed once with flow cytometry buffer (FACS Buffer, containing 0.2% bovine serum albumin, BSA (Sigma-Aldrich, St. louis, USA), in PBS (Fresenius Kabi)). Cells were first stained with viability dye for 30 minutes at 4°C, then washed once with FACS Buffer. For surface staining, samples were stained with monoclonal antibodies of interest for 20 minutes at room temperature in the dark. Cells were washed twice with FACS Buffer. For intracellular staining, cells were permeabilized and fixed according to manufacturer’s instructions (eBioscience, San Diego, USA). Fixed and permeabilized cells were washed once with permeabilization buffer. Samples were incubated with conjugated antibodies for 30 minutes at 4°C in the dark. As a last step, cells were washed twice with permeabilization buffer, and transferred to FACS tubes for acquisition using the Navios™ flow cytometry (Beckman-Coulter, Fullerton, CA, USA).

For intracellular cytokine expression, cells were stimulated with phorbol-12-myristate-13-acetate (PMA), ionomycin and brefeldin A (respectively, 12.5 ng/ml, 500 ng/ml, and 5 µg/ml; all from Sigma-Aldrich, St. Louis, USA) for 4 h, at 37° C in a humidified 5% CO₂ incubator.

Supplementary Table 1 summarizes the antibodies used for flow cytometry and sorting. The average viability at the end of co-culture of unstimulated PBMC was always >60%; stimulated PBMC had a viability of >75%; unstimulated CD4⁺ T cells >70% and stimulated CD4⁺ T cells >80%.

CD4⁺ T cells were sorted on a FACSAria III (BD Bioscience, San Jose, CA, USA) at room temperature into polypropylene round bottom Falcon tubes. Sorted cells were rested for 24h. The following day, samples (1-1.5 x 10⁶ cells) were labelled with 0.1 µM carboxyfluorescein succinimidyl ester (CFSE; Gibco®Thermo Fisher Scientific). After labelling, 50.000 CD4⁺T cells were seeded per condition. Samples were stimulated with either 100 U/mL IL2 or 1:5 (bead-to-cell ratio) αCD3/CD28 beads (Dynabeads™ Human T-Activator, Gibco®Thermo Fisher Scientific). Phenotypical changes were acquired after 24h and 48h with flow cytometry.

Cytokine levels of the supernatant of cell culture as determined using MILLIPLEX Human Cytokine/Chemokine Magnetic Bead Panel (Merck Darmstadt, Germany). The following analytes were measured: IL-1β, IL-2, IL-4, IL-6, IL-10, IL-12p70, IL-17A, TNFα, TNFβ, CCL4, CCL11 (Eotaxin), CX3CL1, MCP3, IFNγ, and CXCL10 (IP-10). The assay was performed according to manufacturer’s protocol. Data acquisition was performed on the FlexMap-3D System (Luminex) using xPONENT Software (Luminex).

For TGFβ, cell culture supernatants were analyzed using an ELISA Kit (R&D System, Minnesota, USA), performed according to manufacturer’s recommendations.

Neutralizing IL-6 (R&D System) or Pan-TGFβ (R&D System) monoclonal antibodies were added to 24h αCD3/CD28 bead activated CD4⁺ T cells, 30 minutes prior to the start of the co-culture with UC-MSC. Based on a pilot dose response experiment, 1 ug/mL and 10 ug/mL neutralizing IL-6 and Pan-TGFβ were used, respectively in subsequent experiments. Neutralizing antibodies were administered each day of the culture. Phenotypic analysis was performed after 3 days of co-culturing in the presence or absence of neutralizing antibodies.

Flow Cytometry data were analyzed using Kaluza Software (Beckman Coulter, analysis version 2.1). Graphics and statistical tests were performed in R version 3.6.2, version 1.6.0. Differences between the levels of outcome were explored using paired-t-test with mean average. The level of statistical significance was set at *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Previously, it was shown that hMSC were able to suppress proliferation of T cells in a mixed lymphocyte reaction (13) or incubation with TCR-specific activators (αCD3/CD28 beads) (15). While some studies suggest elevated CD25 expression of T cells in hMSC cocultures (16), others show a reduction (17). Considering these inconclusive data, we examined the UC-MSC regulated proliferation and activation of T cells in more detail.

PBMC (PBMC Only) were pre-stimulated with αCD3/CD28 mAb (αCD3/CD28 beads) for 24 hours (PBMC+Beads), after which UC-MSC were added (PBMC+Beads+UC-MSC). Phenotypic changes were measured every other day for 5 days (see Supplementary Figure 1A for the experimental setup). Figure 1A shows the gating strategy for PBMC and UC-MSC co-cultures. Lower percentages of CD4⁺CD25⁺ T cells were found on day 1 and day 3. This effect was diminished on day 5 ( Figure 2A ). UC-MSC were not able to inhibit αCD3/CD28 induced proliferation, as measured by CD4⁺Ki67⁺ expression ( Figure 2B ). Furthermore, T cell maturation was investigated in these same co-cultures. We observed a lower percentage of effector memory T cells (T_EM, CD45RA^-CCR7^-) from day 1 onwards ( Figure 2C ), while an increase in central memory T cells (T_CM, CD45RA^-CCR7⁺) was observed in the PBMC+Beads+UC-MSC condition ( Figure 2D ). The percentage of naïve T cells ( Figure 2E ) was also higher in the presence of UC-MSC, albeit at lower levels than T_EM and T_CM.

To check whether the observed effects were exclusively dependent on the interaction of UC-MSC with T cells and not through additional by-stander cells in the PBMC mixture, we purified CD4⁺ T cells (CD4⁺ Only) by negative selection and then set up co-cultures with UC-MSC. Thus, 24-hour pre-activated purified CD4⁺ T (CD4⁺+Beads) cells were co-cultured with UC-MSC (CD4⁺+Beads+UC-MSC) for 5 days. Flow cytometric measurements were acquired every other day ( Supplementary Figure 1A ). Figure 1B shows the gating strategy for CD4⁺T and UC-MSC co-cultures. Again, we observed a further increase in the percentage of CD4⁺CD25⁺ activated T cells in the presence of UC-MSC ( Figure 3A ). Proliferation of the CD4⁺ cells was reduced by the addition of UC-MSC, mostly in the early stages of culture, but not anymore on day 5, which would point to a delay in proliferation rather than complete suppression ( Figure 3B ). Lastly, addition of UC-MSC led to a higher percentage of CD4⁺CD25^highCD127^low/negFOXP3⁺ activated T cells ( Figure 3C ), but only on day 1; on the following days, this population was no longer present, pointing towards a short-lived activation rather than the induction of a putative regulatory cell type.

These results suggest that expression of the activation marker, CD25, was impaired in PBMC co-cultures for a short-term, and that the effect was not observed when CD4⁺ purified cells were used. Proliferation was delayed only in the co-cultures set up with CD4⁺ purified cells.

Regarding the effect of UC-MSC on the maturation state of T cells, the effect was most pronounced using purified CD4⁺ T cells in the co-cultures. From day 1 onwards, the CD4⁺ T_EM fraction was reduced in the presence of UC-MSC ( Figure 3D ). In contrast, on day 3 and 5, the CD4⁺ T_CM fraction increased in UC-MSC co-cultures ( Figure 3E ). The naïve T cell population showed an increase on day 5, yet the increase was not as high as seen for T_CM ( Figure 3F ). There was no significant change in terminally differentiated effector memory cells (data not shown). Thus, in both PBMC and purified CD4⁺ T cell co-culture settings, T_EM started to decrease from day 1 onwards, whereas an increase in T_CM was most apparent from day 3 onwards.

Having established that UC-MSC interfere with T cell memory formation, we further investigated a possible underlying mechanism. Previously published studies provided contradicting evidence regarding the involvement of direct cell contact versus paracrine factors in the immunomodulatory function of MSC on T cells (10, 13, 14, 18–23). Here, the effect of direct cell-cell contact (DCC) was examined by co-culturing αCD3/CD28 beads pre-stimulated PBMC with UC-MSC over a 5-day period. The effect of UC-MSC derived paracrine factors was investigated by either using a transwell system (TW), with 400 nm pores or the addition of culture medium derived from UC-MSC (to be able to capture factors greater than 400 nm; conditioned medium, CM). 250.000 αCD3/CD28 bead stimulated PBMC were co-cultured with either UC-MSC (DCC or TW) or UC-MSC derived conditioned medium in final concentrations of 25% and 50% (CM50% and CM25%, respectively). Phenotypic changes were measured every other day over a period of 5 days ( Supplementary Figure 1A ). TW and CM conditions were then compared to the DCC condition.

Regarding the expression of the activation marker CD25, the TW, CM50% or CM25% conditions revealed a trend towards higher levels on day 1 compared to DCC, but this effect did not persist. There were more CD25⁺ T cells in DCC compared to other conditions ( Figure 4A ). Although Figure 2 revealed a trend towards reduction in proliferation in the DCC condition (PBMC+Beads+UC-MSC) on day 1, Figure 4B shows that none of the other conditions led to significant suppression of proliferation (Ki67 expression) in αCD3/CD28 beads activated CD4⁺ T cells. As compared to DCC, higher percentages of proliferating (Ki67⁺) cells were found in TW and the CM conditions.

Subsequently, the T cell memory formation was investigated. Figure 2 showed that the direct presence of UC-MSC efficiently inhibits T_EM formation. TW or the addition of CM50% or CM25% resulted in a less reduction of T_EM, with significantly higher percentages of T_EM compared to DCC ( Figure 4C ). In all conditions a similar percentage of T_CM was present on day 1 of culture. However, over time, in none of the conditions T_CM level was maintained to the degree found in DCC ( Figure 4D ). The frequency of the naïve T cell population was also higher in DCC, compared to the other conditions ( Figure 4E ).

These results suggest that both direct UC-MSC interaction with T cells and UC-MSC-derived factors modulated T cell activation as measured by CD25 expression, while only direct contact delayed proliferation. UC-MSC clearly affected memory T cell formation, whereby direct cell contact efficiently suppressed T_EM and increased T_CM percentages, and paracrine factors failed to significantly affect T_EM, and T_CM formation.

Various methods have been proposed to investigate the contribution of surface interactions in cell function, such as heating (24), γ-irradiation (25), and chemical agents (e.g. formaldehyde) to block the secretion and proliferative capabilities of the cells (26). Here, we used beta-propiolactone (BPL), effectively inactivating cells by crosslinking DNA with proteins, thereby blocking secretion from UC-MSC, including cytokine secretion, but still allowing cell-cell interaction (27). BPL inactivation causes relatively low damage to surface antigens as compared to other methods (26). UC-MSC were inactivated with BPL (BPL-UC-MSC) for 24h ( Supplementary Figure 1A ). To confirm that UC-MSC were metabolically inactive a MTT assay was conducted ( Supplementary Figure 1B ).

Figure 5A shows that removing UC-MSC specific paracrine factors from the co-culture (CD4⁺+Beads+BPL-UC-MSC), prevents the increase in the percentage of CD25⁺ T cells compared to the CD4⁺+Beads+UC-MSC condition. Proliferation rates (% of Ki67⁺ cells) were similar for both conditions ( Figure 5B ). In the early activated CD4⁺ T cell population, BPL-treated MSC led to a significantly lower percentage of CD4⁺CD25^highCD127^low/negFOXP3⁺ cells compared to untreated UC-MSC ( Figure 5C ). Notably, BPL-UC-MSCs were not able to reduce the T_EM and induce the T_CM populations to levels observed for untreated UC-MSC ( Figures 5D, E ), with only significant differences on day 5 of co-culture ( Supplementary Figures 2A, B ). There was no significant change in the naive T cell population ( Figure 5F ).

Overall, these findings suggest that surface marker interaction between UC-MSC and T cells alone led to lower level of CD25⁺ T cells. Moreover, there were less CD25^high cells with a possibly regulatory phenotype. Also, surface interaction alone (BPL-treated UC-MSC) was not as effective as the combined effect of surface interaction with paracrine factors (UC-MSC) on the induction of memory formation in CD4⁺ T cells.

Previously, it was shown that hMSCs were able to reduce the pro-inflammatory mediators such as IL-1β, IFNγ, IL-6, TNFα and VEGFα; and increase TGFβ, and prostaglandin E₂ (PGE₂) (23, 28, 29) levels. Abundance of these paracrine factors varies under different experimental and disease conditions [reviewed in (30)]. In order to assess the impact of UC-MSCs in our co-culture system, selected cytokines or chemokines were measured by either ELISA or Luminex.

In the DCC (PBMC+Beads+UC-MSC) condition ( Figure 6A ), IFNγ expression was significantly reduced on day3. Although not significant, the concentration of IFNγ at day 5 was lower in the DCC conditions compared to beads stimulated PBMC. Although IL-2 levels did not show any significant reduction, there was a trend towards a reduction in IL-2 concentration in the DCC condition at day 5. Further, IL-6 levels in the DCC condition were significantly higher over the whole period of co-culture. We found that the TNFα concentration was significantly reduced from day 1 onwards in the DCC condition. The presence of chemokine factors in culture medium was evaluated by measuring IFNγ-induced protein 10 (IP-10 or CXCL-10) and Eotaxin levels. IP-10 was elevated in the DCC condition from day 1 onwards, whereas no changes were observed in Eotaxin levels. Without direct cell contact, in TW conditions, there was a trend towards a reduction in IL-2 concentration at day 5, and IL-6 was significantly elevated only at day 5. TNFβ was only reduced in TW on day 1 and day 3. Lastly, chemokine factor, IP-10 was significantly elevated at day 5. The dynamics and profiles of cytokine secretion were different between DCC and TW.

Similar to PBMC co-cultures, IL-6 was elevated in CD4⁺+Beads+UC-MSC. IL-17 was significantly higher in the CD4⁺+Beads+UC-MSC condition on day 1 and 3. Regarding IP-10, there was no significant increase in UC-MSC conditions compared to αCD3/CD28 beads activated cells alone, yet the concentration of IP-10 was elevated on all days in CD4⁺+Beads+UC-MSC. Although, we clearly observed reductions in intracellular IFNγ and IL-2 expression ( Supplementary Figures 2C, D ) in T_EM from the UC-MSC co-cultures, no significant changes were measured in the culture supernatant, this also held true for IL-12 and Eotaxin. Interestingly, there was no concrete change in IL-6 levels in the BPL-treated UC-MSC and αCD3/CD28 beads activated conditions ( Figure 6B ). This suggests that - as expected- IL-6 was mainly produced by UC-MSCs.

In PBMC co-cultures, on day 3, there was increased TGFβ expression in DCC, but not in TW. On day 5, there was a trend towards elevated TGFβ in DCC ( Figure 6C ). In addition to this, in CD4⁺ co-cultures, TGFβ was significantly induced on day 1 and day 5 in CD4⁺+Beads+UC-MSC. Although, there was no significance on day 3, there was a higher TGFβ concentration in the CD4⁺+Beads+UC-MSC condition ( Figure 6D ).

Overall, a difference in cytokine pattern was observed in PBMC and purified CD4⁺ T cell cultures, with more pronounced differences in the PBMC co-cultures. The most striking changes were found for IL-6, IP-10, TGFβ and TNFα. Since, IL-6 and TGFβ were shown to be explicitly secreted by UC-MSC, we next examined whether these two factors played a role in the phenotypic changes observed in the UC-MSC CD4⁺ T cell co-cultures. Therefore, IL-6 and TGFβ were captured in the co-cultures by using neutralizing antibodies. Even though there was no statistically significant difference, neutralization of IL-6 showed a trend toward higher T_EM percentages ( Figure 7A ) and conversely lower levels of T_CM ( Figure 7B ). TGFβ showed a similar pattern as IL-6, but to a lower intensity. These results are suggestive that IL-6 and TGFβ may have a role in the immunomodulating effect of UC-MSC on T cell memory formation, but that other paracrine factors and/or surface interactions are also required.

Having established that the T_EM population was reduced, T_CM numbers increased, and cytokine production was partially impaired upon co-culture with UC-MSC, it remained to be determined to what extent the UC-MSC induced central memory cells were responsive upon re-stimulation. To this end, after 3 days culture of αCD3/CD28 beads activated CD4⁺ T cells with or without UC-MSC, viable CD45⁺CD4⁺CD62L⁺ (T_CM) cells were sorted and rested for 1 day. After 24 hours, cells were stained with CFSE for subsequent measurement of proliferation and re-stimulated with αCD3/CD28 beads. 24 and 48 hours after stimulation, CFSE levels, as well as CD45RA and CCR7 expression were measured ( Figure 8A ).

The results showed that the sorted T_CM cells were able to proliferate upon restimulation with αCD3/CD28 beads after 48 hours ( Figure 8B ). Furthermore, there was no difference in T_CM proliferation between UC-MSC primed and not primed.

When examining the differentiation potential of these cells, we observed that 24 hours later, there was no significant change. However, after 48 hours, there were less T_CM and more T_EM in pre-UC-MSC-beads exposed and re-stimulated cells, compared to non-re-stimulated cells ( Figures 8C, D , respectively). There was no difference in T_EM and T_CM after 48 hours, upon re-stimulation of pre-UC-MSC-beads exposed and pre-beads exposed cells. These findings highlight a phenotypical skewing from T_EM to T_CM when UC-MSC are present in the co-culture and this effect is reversible after re-stimulation in the absence of UC-MSC.