The immunological profile of RC17 hESC-derived dopaminergic neural progenitor cells in vitro: Implications for the STEM-PD clinical trial

TL;DR

This study shows that RC17 hESC-derived dopaminergic cells are not harmful to the immune system and may be safe for use in Parkinson's disease treatments.

Contribution

The study introduces a framework for preclinical immune evaluation of stem cell therapies using in vitro assays.

Findings

RC17 mesDA progenitors do not activate human T cells in vitro.

mesDA progenitors suppress T cell activation and proliferation in co-culture.

mesDA progenitors show equal or lower immunogenicity than human fetal VM tissue.

Abstract

Parkinson’s disease involves the progressive loss of dopaminergic neurons, prompting clinical trials replacing cell loss with neural grafts. This includes the transplantation of pluripotent stem cell-derived mesencephalic dopaminergic neural progenitors (mesDAp), including the RC17 human embryonic stem cell (hESC)-derived cells currently under investigation in the European STEM-PD trial (NCT05635409). To assess potential immune rejection risk, we characterized RC17-mesDAp immunogenicity in vitro, comparing them to human fetal ventral mesencephalic tissue (hfVM), as successfully used in similar clinical trials such as TRANSEURO. Although RC17-mesDAp expressed MHC class I, upregulated by pro-inflammatory cytokines, no peripheral immune response was detected in vitro. Instead, cells exhibited immunomodulatory effects, reducing T cell CD25 expression and proliferation. Transcriptomic analysis showed that both RC17-mesDAp and hfVM upregulated antigen-processing pathways in response to IFN-γ yet remained non-immunogenic. Findings support the immunological safety of RC17-mesDAp and suggest a set of in vitro assays that may be applicable for the preclinical evaluation of other human stem cell therapies.

•RC17 hESC-derived mesDA progenitors do not activate human T cells in vitro•mesDA progenitors suppress T cell activation and proliferation in co-culture•IFN-γ induces antigen presentation genes without inducing immunogenicity in vitro•mesDA progenitors show equal or lower immunogenicity than human fetal VM tissue

RC17 hESC-derived mesDA progenitors do not activate human T cells in vitro

mesDA progenitors suppress T cell activation and proliferation in co-culture

IFN-γ induces antigen presentation genes without inducing immunogenicity in vitro

mesDA progenitors show equal or lower immunogenicity than human fetal VM tissue

Human embryonic stem cell-derived dopaminergic progenitors (mesDA), prepared as in the STEM-PD trial for Parkinson’s, do not exhibit immunogenicity in vitro and instead actively suppress T cell activation in vitro. These findings support the immunological safety of RC17 mesDA progenitors and establish a functional framework for preclinical immune assessment of stem cell-based neural therapies.

Figures

hESC-derived mesDA progenitors express MHC class I, increased by exposure to IFN-γ(A) Cells were untreated (NT; green) or treated with 5 ng/mL IFN-γ (red), TNF-α (blue), or both IFN-γ + TNF-α (orange) for 24 h.(B) Representative expression of MHC-I vs. cell type marker and summary histogram overlays.(C) Quantification of percent cells expressing MHC-I. *N* = 4 hfVM, hESCs, mesDAp, *n* = 3 DAn (one-way ANOVA). (D) Histogram overlays of expression of MHC-II, CD80, CD86, and CD40 and positive controls: PBMC CD14^+^ or CD19^+^ fraction, FMO control (gray), unstimulated (purple), or LPS stimulated

hfVM, mesDA progenitors, and DAn are transcriptomically similar for immune-relevant genes in non-inflammatory conditionsNanoString nCounter analysis of hfVM, hESCs, mesDAp, and DAn without inflammatory stimulation.(A) Log2(normalized counts) for MHC-I, MHC-II, and co-stimulatory genes (*CD80*, *CD86*, and *CD40*).(B) Clustered heatmap of immune-related pathways.(C and D) Differential expression comparing mesDAp to hfVM (C) and to hESCs/DAn (D).(E and F) Violin plots of genes expressed significantly higher (E) or lower (F) in mesDAp versus hfVM.(G and H) Violin plots of genes differentially exp

hfVM and mesDA progenitors both increased expression of antigen processing and presentation genes in response to IFN-γ exposure(A–D) Differential expression after 24 h IFN-γ (5 ng/mL) versus untreated baseline in mesDAp (A and C) and hfVM (B and D).(C and D) Antigen processing and presentation genes significantly upregulated following IFN-γ stimulation.(E–I) Comparison across cell types following IFN-γ treatment: (E) pathway heatmap, (F) IFN-γ-mesDAp versus IFN-γ-hfVM, (G andH) genes significantly higher (G) or lower (H) in IFN-γ-mesDAp versus IFN-γ-hfVM, and (I) IFN-γ-mesDAp versus IFN-γ-hESC

mesDA progenitors do not induce T cell activation or proliferation in *in vitro* immunogenicity assays; instead, they display largely contact-dependent immunomodulatory properties(A–C) T cell activation measured by proliferation index (ModFit) and CD25 expression in CD4^+^ and CD8^+^ T cells following 5-day co-culture with untreated or IFN-γ-treated mesDAp in (A) T cell co-culture (*n* = 16), (B) PBMC co-culture (*n* = 5), and (C) T cell/moDC co-culture (*n* = 10).(D and E) Supernatants from mesDAp co-cultures (T cell and T cell/moDC) and mesDAp-only wells were analyzed by 8-plex cytokine Lumi

Full text

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized by the loss of A9 dopaminergic neurons (DAn) in the substantia nigra pars compacta (SNpc). Dopamine cell loss is responsible for many of the significant motor deficits seen in PD, including bradykinesia and rigidity, which occur alongside non-motor symptoms. Until now, motor symptom management has involved the use of dopaminergic drugs such as levodopa, dopamine agonists, and inhibitors targeting enzymes that break down dopamine. While usually highly effective in the early stages of disease, over time these drugs can lose efficacy and cause side effects. These include neuropsychiatric and autonomic complications and levodopa-induced dyskinesias (LID), which may be disabling, necessitating more invasive interventions such as deep brain stimulation (DBS). Consequently, there is a pressing need for more targeted, physiological replacement of dopamine.

One approach is to graft new dopamine cells into the site of greatest dopamine loss within the striatum. Since the 1980s, different cell sources have been tested preclinically and in clinical trials. Of these, the greatest success has been seen following transplantation of human fetal ventral mesencephalon (hfVM) tissue (Ahlskog et al., 1990; Backlund et al., 1985; Barker et al., 2013; Brundin et al., 2000; Farag et al., 2009; Freed et al., 1992; Freeman et al., 1995; Kefalopoulou et al., 2014; Lindvall et al., 1989, 1990, 1994; Madrazo et al., 1987; Mendez et al., 2000, 2002; Ming et al., 2009; Mínguez-Castellanos et al., 2007; Olanow et al., 1990, 2003; Piccini et al., 1999; Redmond et al., 1993; Wenning et al., 1997; Widner et al., 1992). This tissue contains developing midbrain dopamine cells of the type lost in PD. However, trial outcomes have been variable, likely due to differences in trial design, patient selection, amount of tissue grafted, and the level and duration of immunosuppression used (Barker, et al., 2015a, 2015b; Barker et al., 2013). Nevertheless, hfVM transplants have provided proof of principle that dopamine cell replacement therapy can be effective in PD (Petrus-Reurer et al., 2021; Qarin et al., 2021), with several patients showing marked clinical improvements with normalization of dopamine levels in the striatum, patients coming off all their PD medication, and with evidence of dopamine cell survival up to 24 years post grafting (Li et al., 2016). Due to the limited availability of hfVM tissue, ethical issues associated with its use, and the failure to standardize such a therapy, the development of human pluripotent stem cell (hPSC)-derived dopamine cell products has been actively pursued (Ganat et al., 2012; Kirkeby et al., 2012, 2017; Kriks et al., 2011a, 2011b; Nolbrant et al., 2017, 2020; Schweitzer et al., 2020). Several induced pluripotent stem cell (iPSC)- and human embryonic stem cell (hESC)-derived mesencephalic dopamine (mesDA) neuron progenitor cells (mesDAp) have now been demonstrated to work reproducibly well in animal models of PD (Barker et al., 2015a; Elabi et al., 2022; Ganat et al., 2012; Grealish et al., 2014; Kikuchi et al., 2017; T. W. Kim et al., 2020; Kirkeby et al., 2023; Kriks et al., 2011a; Morizane et al., 2013; Niclis et al., 2017; S. Park et al., 2024; Parmar et al., 2019; Schweitzer et al., 2020; Tiklová et al., 2020) with a good safety profile, which has now led to first-in-human trials (Chang et al., 2025; Kirkeby et al., 2023, 2025; Sawamoto et al., 2025; Studer, 2017; Tabar et al., 2025; Takahashi, 2020). One of these is the European STEM-PD clinical trial initiated in February 2023 testing human mesDAp derived from the RC17 hESC line using the Nolbrant et al. protocol (Nolbrant et al., 2017) in eight people with Parkinson’s in Lund, Sweden, and Cambridge, UK (NCT05635409; Kirkeby et al., 2023). This protocol enables the generation of mesDAp in 16 days in vitro, at which point cells are cryopreserved until their time of implantation into patients.

In this fast-moving field, a significant concern remains the immunogenic potential of the cells and subsequent possibility of immune rejection. Strategies such as hypoimmunogenic cell engineering or co-transplantation of regulatory cells are being explored to mitigate this risk (Deuse et al., 2019; T.-Y. Park et al., 2023; Pavan et al., 2025; Ye et al., 2020); however, the inherent immunogenicity of most stem cell-derived products is poorly understood. To address this for the RC17 hESC-derived product, we performed detailed in vitro immune characterization, using multiple independent and complementary human immune co-culture assays, comparing the cells to previously successfully engrafted dissociated hfVM, tested in our previous TRANSEURO trial (Barker, 2019; Barker et al., 2026). We studied the RC17 hESC-derived mesDAp at different stages of differentiation in vitro, including the undifferentiated hESC (day 0) to mature, terminally differentiated DAn (day 45+). No mesDAp immunogenicity was detected in vitro, even after exposing them to pro-inflammatory cytokines. Instead, unlike the hfVM tissue, the RC17-mesDAp exhibited in vitro immunomodulatory properties, reducing T cell activation and proliferation. These findings suggest that the intrinsic properties of the RC17-mesDAp do not reflect those of a cell type with a high risk of immune rejection, although one must be cautious given the inherent limitations of in vitro assays.

It is important to emphasize that our data are specific to the RC17-mesDAp and that the immunogenicity of other hPSC-derived products may vary, necessitating individual assessment. We suggest that such assessments include the preclinical, in vitro pipeline outlined here. Furthermore, the findings of this study do not suggest that such therapies should be performed without, or with reduced, immunosuppression. To make such claims, further in vivo analyses would be required, which we outline in the discussion.

Results

Generation of the RC17 hESC-derived mesDA progenitors using the STEM-PD trial protocol

First, we confirmed successful differentiation of RC17 hESC to mesDAp using the Nolbrant et al. protocol, as used in the STEM-PD trial. After 16 days of differentiation, RC17-mesDAp were triple-positive for lineage commitment markers FOXA2, LMX1A, and OTX2 as assessed by immunofluorescence (Figure S1A), and flow cytometric analysis confirmed high expression of FOXA2, integrin-associated protein (IAP), and EN1 and a lack of PAX6 and OCT3/4 (Figures S1B and S1C), confirming correct cell identity. A quantitative PCR gene panel was used to further validate cell identity. This demonstrated expression of genes reflecting DAn progenitor state (EN1, FOXA2, LMX1A, LMX1B, OTX1, and OTX2), non-ventral midbrain patterning (PAX6, FOXG1, HOXA2, and BARHL1), positive predictors of graft outcome in vivo (CNPY1, PAX8, SPRY1, and ETV5), and mature DAn (TH, NURR1, and PITX3) and reduced expression of pluripotency genes (OCT3/4 and NANOG) and the negative predictor of graft outcome FEZF1 (Figure S1D). In all cases, gene and protein expression of desired markers was comparable to, or more favorable than, that of the hfVM tissue. To account for variability of hfVM donor tissues, we used n = 3–6 hfVM tissues, aged within the developmental window used in our TRANSEURO clinical trial for all our profiling and functional assays.

To characterize the product following neuronal maturation, mesDAp were differentiated in vitro to 45+ days. While this provides insight into the post-differentiation phenotype, these cells mature more slowly in vivo where they typically only start to exert any functional effects on the grafted animals 4–6 months post transplantation. After differentiation, a high proportion of cells expressed the appropriate markers of DAn differentiation including tyrosine hydroxylase (TH) and MAP2 as detected by immunofluorescence (Figure S1E) and were positive for TH, MAP2, and NURR1 as measured by flow cytometry (Figures S1F and S1G). Quantification was performed using flow cytometry due to the difficulties associated with reliable immunofluorescence quantification in dense, multi-plane neurosphere-rich cultures. However, TH positivity as measured by flow cytometry is likely to be an overestimate, due to dissociation- and suspension-related technical bias.

Given reports of contaminating cell types in the mature products of both hfVM and mesDAp grafts, particularly in vivo (Tiklová et al., 2020), we also examined the expression of markers associated with non-neuronal cell types. In line with previous reports, we detected expression of COL1A1 (VLMC marker) in all samples (higher in embryonic stem cell [ESC], mesDAp, and DAn vs. hfVM) and PDGFRA (VLMC marker) and TTR (CPEC marker) in the ESC-derived mesDAp and DAn at a higher level compared to the hfVM tissue. In addition, GFAP (astrocyte marker) expression was detected, but only in the mature DAn cultures. These data align with previous reports of mesDA-derived graft products and confirm that our downstream immunological analyses have assessed the immunogenicity of cells equivalent to the clinically administered product, in its full cellular composition, including non-neuronal components that may influence immune responses.

hESC-derived mesDA progenitors express MHC class I, increased by exposure to IFN-γ

Next, we examined the immunogenic potential of the mesDAp by assessing their surface expression of immune-relevant molecules by flow cytometry. To mirror the STEM-PD trial, the mesDAp were assessed following freeze thawing, and their profile was compared to dissociated hfVM tissue as was used in our completed TRANSEURO trial (Barker et al., 2026). To explore whether the cells display different immunogenic profiles at different stages of differentiation/maturation, undifferentiated hESCs and day 45+ DAn were also examined. To mimic a likely, post-surgery, local inflammatory response and the known inflamed PD brain (Isik et al., 2023; Kouli et al., 2023), expression was also assessed after 24 h exposure to the pro-inflammatory cytokines IFN-γ and TNF-α (Figures 1A and S2A) and IL-1β (Figure S2A).Figure 1hESC-derived mesDA progenitors express MHC class I, increased by exposure to IFN-γ(A) Cells were untreated (NT; green) or treated with 5 ng/mL IFN-γ (red), TNF-α (blue), or both IFN-γ + TNF-α (orange) for 24 h.(B) Representative expression of MHC-I vs. cell type marker and summary histogram overlays.(C) Quantification of percent cells expressing MHC-I. N = 4 hfVM, hESCs, mesDAp, n = 3 DAn (one-way ANOVA). (D) Histogram overlays of expression of MHC-II, CD80, CD86, and CD40 and positive controls: PBMC CD14^+^ or CD19^+^ fraction, FMO control (gray), unstimulated (purple), or LPS stimulated (magenta).

The flow cytometric data displayed in Figure 1B show MHC class I expression against the cell-type markers—IAP for hfVM, hESC, and mesDAp (with higher expression of IAP denoting ESC to mesDA progenitor differentiation) and TH for DAn. The summary data (Figure 1C) show the percentage of MHC class I expression of the entire product (gated on live single cells) as would be transplanted into the patient. MHC class I was expressed by ESCs, mesDAp, and DAn, with the percent expression higher than that seen in cells derived from the hfVM tissue (Figures 1B and 1C). A 24-h treatment (time and concentration course shown in Figures S2A–S2C) with IFN-γ induced significant upregulation of MHC-I on hfVM, mesDAp, and DAn and a small, non-statistically significant increase on hESCs. Cytokine concentrations were selected based upon effective upregulation of surface markers without induction of significant cell death or a change in morphology. Treatment with TNF-α alone had a smaller effect (Figures 1B and 1C) and no clear additive effect when given in combination with IFN-γ. Very little MHC-II expression was observed on any cell type. No upregulation was seen following exposure to IFN-γ, TNF-α, or IFN-γ + TNF-α. No cell type displayed expression of the co-stimulatory molecules CD80, CD86, or CD40 (Figure 1D).

These data suggest that the mesDAp may be more immunogenic than hfVM, particularly in an inflammatory environment, based upon their potential to interact directly with CD8^+^ T cells via their higher MHC-I expression. However, at no differentiation stage do they express all the molecules required for direct T cell activation.

hfVM, mesDA progenitors, and DAn are transcriptomically similar for immune-relevant genes in non-inflammatory conditions

Having identified key surface molecule expression, we next performed a targeted transcriptomic analysis on all cell types using the NanoString nCounter Human Immunology V2 panel (which assesses the expression of 579 immunologically relevant genes) plus an additional 20 gene codeset designed to capture additional genes related to T cell activation/inhibition and graft rejection (see pathway annotations and custom codeset in Figure S3A). Cells were examined after 24 h in culture, either with no treatment (NT) or following exposure to 5 ng/mL IFN-γ for 24 h—a dose and time frame selected based on a preliminary experiment that, in line with the surface expression data, revealed stronger immune-related responses to IFN-γ treatment compared to TNF-α and the fact that combined treatments did not appear to be additive (Figure S3B).

In keeping with the surface protein expression, nCounter analysis confirmed expression of HLA-A, -B, and -C, with ESCs and DAn expressing higher levels of HLA-A and HLA-C than the mesDAp and hfVM, and low expression of the MHC-II-encoding genes in all cell types. HLA-DPB1 was the most constitutively expressed class II gene across all cell types, alongside very low levels of HLA-DPA1 and -DMA. Constitutive CD40 expression was only above the limit of detection for hESC. No cell types expressed CD80 or CD86. Expression levels of these key genes were comparable for mesDAp and hfVM tissue (Figure 2A).Figure 2hfVM, mesDA progenitors, and DAn are transcriptomically similar for immune-relevant genes in non-inflammatory conditionsNanoString nCounter analysis of hfVM, hESCs, mesDAp, and DAn without inflammatory stimulation.(A) Log2(normalized counts) for MHC-I, MHC-II, and co-stimulatory genes (CD80, CD86, and CD40).(B) Clustered heatmap of immune-related pathways.(C and D) Differential expression comparing mesDAp to hfVM (C) and to hESCs/DAn (D).(E and F) Violin plots of genes expressed significantly higher (E) or lower (F) in mesDAp versus hfVM.(G and H) Violin plots of genes differentially expressed in mesDAp versus hESCs and/or DAn.(I) IL2RA expression and flow cytometry validation of CD25 in mesDAp. N = 3 (one-way ANOVA).

Gene (Figures S3F and S3G) and pathway (Figure 2B) clustering analyses, performed based on the entire gene set, revealed that the hfVM, and mesDAp and DAn in particular, are transcriptionally similar, while the hESCs appeared distinct. Differential gene expression analysis revealed 45 upregulated and 22 downregulated genes (significance = p < 0.01) in mesDAp compared to cells dissociated from the hfVM (Figure 2C) as well as changes in immune-relevant gene expression through their differentiation from hESC to mesDAp to DAn (Figure 2D).

Genes significantly more highly expressed in mesDAp compared to hfVM were largely anti-inflammatory in nature, including co-inhibitory molecules and anti-inflammatory mediators (CD112, CD276, CSF1, and CTNNB1), TGF-β pathway genes (TGFB1 and TGFBR2), and non-canonical MHC-I genes such as HLA-E (Figure 2E), whereas genes more highly expressed by hfVM were more pro-inflammatory, including chemokines (CCL26 and CXCR4—although notably CXCR4 also has a role in DAn migration in development [Bodea et al., 2014; Yang et al., 2013]), complement and antiviral response genes (C5 and MX1), and genes commonly expressed by antigen presentation cells (CD14 and CTSS) (Figure 2F), together suggesting that at the transcript level, mesDAp appear less pro-inflammatory in nature than hfVM cells.

Performing differential gene expression analysis across cell differentiation stages revealed greater expression of TGF-β pathway-associated genes in mesDAp compared to hESC (TGFB1, TGFBR1, and TGFBR2), with expression of TGFBR1 also being significantly higher in mesDAp than in DAn. mesDAp and DAn also expressed higher levels of the co-inhibitory molecules CSF1, CD276, CD112, and CD113 than the hESC, with CSF1 and CD112 expression being greatest at the day 16 progenitor stage (Figure 2G). Furthermore, several pro-inflammatory genes were found to be expressed at lower levels by the mesDAp compared to the hESC and/or DAn, including genes involved in antigen processing (PSMB8, PSMB9, and TAPBP), which were lower in the mesDAp than hESC and DAn, except TAPBP in which expression levels correlated with differentiation stage. Complement genes showed opposing directionality, with C1QBP decreasing and C4A/B increasing with differentiation, and other pro-inflammatory genes were expressed at a lower level by mesDAp than the hESC or DAn (CCL26 and MX1) (Figure 2H). Several genes differentially expressed between hESC, DAn, and the day 16 mesDAp were related to cell cycle, division, and neuronal cell function, as would be expected, as the cells transition from a replicative to a terminally differentiated state. We also examined expression of molecules relevant to the immunosuppressive regimen used in the STEM-PD clinical trial, particularly CD25 (IL2RA), the target of basiliximab, which is given peri-operatively. CD25 was not detected in mesDAp or DAn, either at baseline or following exposure to inflammatory cytokines at the gene or protein level. Activated T cells, the intended target of basiliximab, were used as a positive control (Figure 2I).

Together, these data show that at baseline, the mesDAp appear to express higher levels of genes encoding molecules with anti-inflammatory properties and lower levels of pro-inflammatory genes, when compared to hfVM tissue. Furthermore, the mesDA progenitor stage of differentiation also appears to be less immunogenic than the hESC or mature DAn, though the differences between mesDAp and DAn states were minimal.

hfVM and mesDA progenitors both increased expression of antigen processing and presentation genes in response to IFN-γ exposure

In response to IFN-γ exposure, the mesDAp upregulated 43 genes (Figure 3A). Of these, 31 were also upregulated in hfVM (Figure 3B) and 25 in DAn (Figure S3C) (of a total of 37 and 35 genes, respectively), showing similar responses to the inflammatory stimulus. As expected, many of these were IFN-response genes, including IRF1, GBP-1, and GBP-5, showing that mesDAp, DAn, and hfVM tissue have the capacity to respond to IFN-γ. The majority of genes significantly upregulated by mesDAp, hfVM, and DAn were those involved in antigen processing and presentation pathways, many of which were below the limit of detection in the NT condition (Figures 3C, 3D, and S3D), including TAP-1-2, TAPBP, PSMB8-10, and MHC-I antigen presentation genes (HLA-A, -B, -C, B2M). DAn did not upregulate HLA-A, -B, or -C in response to IFN-γ, instead only upregulating the non-canonical MHC-I gene HLA-E (Figures 3A, 3B, and S3C). However, baseline differential expression analysis showed that DAn express significantly higher levels of these genes without exposure to inflammatory signals (Figure 2A). mesDAp and hfVM also increased expression of TLR3 and HVEM following IFN-γ exposure, two common pro-inflammatory receptors involved in the antiviral response.Figure 3hfVM and mesDA progenitors both increased expression of antigen processing and presentation genes in response to IFN-γ exposure(A–D) Differential expression after 24 h IFN-γ (5 ng/mL) versus untreated baseline in mesDAp (A and C) and hfVM (B and D).(C and D) Antigen processing and presentation genes significantly upregulated following IFN-γ stimulation.(E–I) Comparison across cell types following IFN-γ treatment: (E) pathway heatmap, (F) IFN-γ-mesDAp versus IFN-γ-hfVM, (G andH) genes significantly higher (G) or lower (H) in IFN-γ-mesDAp versus IFN-γ-hfVM, and (I) IFN-γ-mesDAp versus IFN-γ-hESC/DAn. N = 3 (one-way ANOVA).

mesDAp, hfVM, and DAn also all upregulated several anti-inflammatory genes in response to IFN-γ, including CD274 (encoding co-inhibitory molecule PD-L2) and IDO-1 (responsible for the catabolism of tryptophan to the anti-inflammatory kynurenine). Only 6 genes were uniquely upregulated by hfVM and not mesDAp, 2 of which were T cell chemoattractants CXCL9 and CXCL10 (Figure 3B). Notably, the two genes with the highest upregulated fold change by DAn were CXCL10 and CXCL11 (Figures S3C and S3D). In line with previous reports (Chen et al., 2020), hESC did not show any transcriptional response to IFN-γ (Figure S3E), despite expression of the IFN receptor IFNGR1, resulting in further separation of the ESC cluster on pathway cluster analyses of IFN-γ-treated cells (Figure 3E).

Following IFN-γ exposure, the number of genes significantly differentially expressed between mesDAp and hfVM increased from 69 (without treatment) to 82 (Figure 3F). 65% of the genes were retained from the untreated comparison, while additional genes, more highly expressed by IFN-γ-treated mesDAp than IFN-γ-treated hfVM, included pro-inflammatory genes associated with TNF pathway activation (TRAF3, TRAF5, and TNFSF11) and caspase activation (CASP1 and CASP8), perhaps suggesting that mesDAp may be more susceptible to inflammation-induced cell death than hfVM tissue (Figure 3G). IFN-γ-treated hfVM was found to express significantly higher levels of complement genes (C3), antigen presentation markers (CD83, HLA-DRA, and HLA-DPA1), and other pro-inflammatory genes, such as CCL3 (Figure 3H), when compared to IFN-γ-treated mesDAp. In contrast, the differentially expressed genes between IFN-γ-treated mesDAp and IFN-γ-treated DAn remained similar to those without IFN-γ treatment. The IFN-γ-treated hESCs were of course strongly separated from both of their lineage committed progeny (Figure 3I).

mesDA progenitors do not induce T cell activation or proliferation in vitro but are immunomodulatory

Next, all cell types (hESCs, mesDAp, DAn, and hfVM), with and without pre-exposure to IFN-γ, were tested in three functional in vitro assays involving co-culture with primary human immune cells labeled with a proliferation-tracking dye. After 5 days, T cell activation was assessed by measuring proliferation, surface marker expression, and secretion of cytokines known to be associated with allorecognition. The three assays were designed to assess different allorecognition pathways: (A) co-culture with magnetic-activated cell sorting (MACS)-enriched T cells to examine direct allorecognition, (B) co-culture with whole peripheral blood mononuclear cells (PBMCs) to examine indirect and semi-direct allorecognition, and (C) co-culture with MACS-enriched T cells plus autologous monocyte-derived dendritic cells (moDCs) to increase sensitivity for detecting indirect and semi-direct responses by enriching for professional antigen-presenting cells (APCs). In assays (A) and (B), cells were also co-cultured with pre-activated T cells or PBMCs to assess potential immunoregulatory effects.

First, assays were optimized (Figure S4); successful generation of moDCs was confirmed by surface expression of key antigen presentation molecules (Figure S4A) and morphology (Figure S4B), as well as by demonstrating their ability to engulf (Figure S4C) and present antigen to T cells (Figure S4D). Keyhole limpet hemocyanin (KLH), a hyperimmunogenic antigen, was used as a positive control in the T cell-moDC assays, while allogeneic and autologous moDCs were also used as positive and negative controls, respectively, in the T cell co-culture assays (Figure S4E). While our primary readouts presented in Figures 4 and 5 focus on T cell activation, we also confirmed that the mesDA progenitor survival was not affected by co-culture with T cells (Figure S4F).Figure 4mesDA progenitors do not induce T cell activation or proliferation in in vitro immunogenicity assays; instead, they display largely contact-dependent immunomodulatory properties(A–C) T cell activation measured by proliferation index (ModFit) and CD25 expression in CD4^+^ and CD8^+^ T cells following 5-day co-culture with untreated or IFN-γ-treated mesDAp in (A) T cell co-culture (n = 16), (B) PBMC co-culture (n = 5), and (C) T cell/moDC co-culture (n = 10).(D and E) Supernatants from mesDAp co-cultures (T cell and T cell/moDC) and mesDAp-only wells were analyzed by 8-plex cytokine Luminex (IFN-γ, TNF-α, IL-2, IL-6, IL-8, IL-4, IL-10, and GM-CSF) (D) and TGF-β ELISA (E); heatmap shows cytokines relative to untreated T cell-only wells.(F and G) Contact versus transwell (0.4 μm) mesDAp-T cell co-culture; proliferation and CD25 measured (F) and quantified (G). N = 3 for (D)–(G) (one-way ANOVA).Figure 5hfVM are neither immunogenic nor immunosuppressive in vitro(A–C) Proliferation index (ModFit) and CD25 expression in CD4^+^ and CD8^+^ T cells after 5-day co-culture with untreated or IFN-γ-treated hfVM in (A) T cell co-culture (n = 8), (B) PBMC co-culture (n = 5), and (C) T cell/moDC co-culture (n = 5).(D) Supernatants from hfVM co-cultures (T cell and T cell/moDC) and hfVM-only wells analyzed by 8-plex cytokine Luminex (IFN-γ, TNF-α, IL-2, IL-6, IL-8, IL-4, IL-10, and GM-CSF); heatmap shows cytokines relative to untreated T cell-only wells. N = 3 (one-way ANOVA).

T cell proliferation (measured by a leftward shift in histograms as a result of proliferation dye dilution through subsequent cell divisions), expression of the T cell activation marker CD25, and, in a subgroup of experiments, cytokine production were measured at day 5. No significant T cell proliferation, activation, or cytokine production was seen in any of the assays in response to mesDAp—even after pre-exposure to IFN-γ (Figure 4; representative dotplots displayed in Figures S5A–S5C). Proliferation, activation, and production of GM-CSF, IFN-γ, and TNF-α cytokines was observed following T cell co-cultures with allogeneic moDCs—confirming the ability of this assay to detect T cell allo-responses—and polyclonal anti-CD3/28 bead stimulation (TCA) (Figures 4A–4D and S4E). In fact, rather than inducing an immune response, mesDAp were found to suppress T cell proliferation, activation, and the production of some pro-inflammatory cytokines when added to anti-CD3/28 polyclonally stimulated T cells (Figures 4A, 4D, and S5A). Furthermore, the concentration of the immunosuppressive cytokine TGF-β was significantly higher in co-cultures of activated T cells with mesDAp or IFN-γ-mesDAp (Figure 4E). mesDAp were also observed to suppress T cell responses to alloDCs, although somewhat variably across donors (Figures S5D–S5F).

mesDA progenitor-mediated suppression is largely contact dependent

To understand the importance of direct cell contact in mesDAp-mediated T cell suppression, we repeated the T cell/mesDAp co-culture assay using a transwell. This revealed significantly reduced suppression of T cell activation and proliferation when the cells were physically separated, suggesting that suppression is largely, but not entirely, contact dependent (Figures 4F and 4G).

To examine the in vitro immunomodulatory mechanism, we screened our nCounter dataset for potentially immunosuppressive molecules expressed by untreated and IFN-γ-treated mesDAp, as both had been shown to modulate T cell activation. This revealed high levels of co-inhibitory molecules CD276 (B7-H3), CD112, CD113, and CD155, while CD274 (PD-L1), PDCDLG1 (PD-L2), or VTCN1 (B7-H4) were not expressed (Figure S6A). However, neither inhibition of CD276 (neutralizing antibody; Figure S6B) nor TIGIT—the receptor of CD112/CD113/CD155 (blocking antibody; Figure S6C)—inhibited mesDA progenitor-mediated T cell suppression. CD47 (also referred to as IAP) is a well-established regulator of innate immune clearance through its interaction with SIRPα on macrophages and NK cells and is widely exploited in the generation of hypoimmunogenic cell products. It is also used as a lineage-defining molecule for our mesDAp product. Given its high constitutive expression on mesDAp at the gene and protein level (Figures S6D and S6E), we examined whether CD47 might indirectly contribute to the observed immunomodulatory effects; however, CD47 blockade did not reverse T cell suppression in our assays. Nonetheless, high constitutive expression of this protein may be beneficial for the survival of this product in vivo.

Due to the incomplete inhibition of suppression in the transwell assay, contact-independent mechanisms were also investigated. mesDAp did not express genes encoding the anti-inflammatory cytokines IL-4, IL-10, or IL-13 but did express high levels of TGFB (Figure S6G). We therefore went on to demonstrate that rhTGFβ is able to suppress polyclonally stimulated T cells in a dose-dependent manner (Figure S6H) as this has been shown previously as a suppressive mechanism for other hPSC-derived mesDAp (Liu et al., 2013). However, mesDAp-mediated suppression of T cells was not inhibited by the addition of TGFβR1 nor TGFβR2 inhibitors (Figure S6I).

IDO-1/TDO-mediated catabolism of tryptophan to kynurenine is a well-reported mechanism of suppression in other hPSC-derived products (M. Kim and Tomek, 2021; Romano et al., 2021). mesDAp did not express IDO-1 or TDO at baseline, but upregulated IDO-1 expression was seen following exposure to IFN-γ (Figure S6J), so this mechanism was tested in case IFN-γ-treated mesDAp employ a different mechanism of suppression to the untreated cells. Inhibition of IDO-1, using 1-LMT, however, had no significant effect on suppression (Figure S6K).

Finally, as mesDAp express (at low level) surface CD39 and CD73 (ectonucleotidases known to convert ATP to immunosuppressive adenosine; Figure S6L), we tested the role of adenosine in mesDA progenitor-mediated T cell suppression by blocking the main T cell adenosine receptor A2A. This did not affect mesDAp-mediated suppression (Figure S6M). While we were not able to determine the immunomodulatory mechanism of the mesDAp, our approach ruled out several well-known mechanisms.

hfVM, ESCs, and DAn are neither immunogenic nor immunosuppressive in vitro

Like mesDAp, cells dissociated from hfVM tissue, ESCs, and DAn did not induce T cell proliferation or CD25 expression in vitro (Figures 5 and S7). However, in contrast to mesDAp, these cells did not suppress polyclonally stimulated T cell proliferation (Figures 5A and S7A–S7C), suggesting that T cell suppression is a unique feature of the progenitor cell state. A small reduction in CD25 expression, only in CD8^+^ T cells, was found following addition of hfVM to polyclonally stimulated T cells and PBMCs, but no effect on T cell proliferation was seen (Figures 5A and 5B).

Co-culture of hfVM, IFN-γ-treated-hfVM, DAn, and IFN-γ-treated-DAn led to a slight increase in some supernatant cytokines, including IL-6 and IL-8 production (Figures 5D and S7E), when compared to the mesDAp (Figure S5G), again highlighting the low immunogenicity of the mesDAp in culture.

Discussion

After comprehensive analysis, using multiple independent and complementary human immune co-culture assays, we show that RC17-derived mesencephalic dopaminergic neural progenitors—generated and handled according to the STEM-PD clinical trial protocol—do not elicit detectable immune activation in vitro and, in some settings, appear immunoregulatory. These observations are consistent with previous reports describing neural progenitor cells as immunoregulatory, suppressing T cell activity, particularly in the contexts of spinal cord injury and inflammatory neurological diseases (Fainstein et al., 2008; Pluchino et al., 2003, 2009; Riemann et al., 2018; Willis et al., 2020). Throughout the study, we compared RC17-mesDAp to hfVM tissue, selected because this tissue has previously been shown to survive long-term in the brains of people with PD, beyond the initial period of post-transplant immunosuppression (Barker et al., 2013; Kefalopoulou et al., 2014; Li et al., 2016; Lindvall et al., 1994). Our data indicate that mesDAp are at no greater risk of immune rejection than hfVM tissue, and may, in fact, be at lower risk due to their apparent immunomodulatory qualities and transcriptional profiles. This conclusion is further supported by the fact that all hfVM analyses in our study were performed using tissue from a single donor, whereas clinical hfVM transplants typically involve cells pooled from multiple fetuses, which will increase the likelihood of alloreactivity.

A limitation of this study is that we focused exclusively on RC17-derived mesDAp, raising the question of how generalizable our results are to other stem cell-derived dopaminergic products. This focus was deliberate, as progenitors derived from the RC17 cell line are currently being tested in the STEM-PD clinical trial, and the goal of our work was to characterize their immune properties, particularly as we transition to the next stage of clinical trials using this cell source. Importantly, by studying these cells, we have established a set of optimized, complementary in vitro experimental protocols that provide a practical framework for evaluating the immunogenicity and immunomodulatory potential of other stem cell-derived neuronal products, with relevance for the wider research and clinical community.

A comparison of immune profiles across various stages of differentiation—from hESC to mature DAn—demonstrated that day 16 mesDAp (the stage at which they are transplanted to patients) exhibited particularly low expression of immunogenic molecules and high expression of immunoregulatory molecules compared to the hESC and DAn. This is important as the risk of immune rejection is likely to be highest in the early post-transplant period, when post-surgery-induced inflammation and a disrupted blood-brain barrier are present. To mimic the post-surgical setting and known inflamed environment of the PD brain, we profiled our cells pre- and post-exposure to inflammatory cytokines. As the cytokine environment at the transplant site is largely unknown, we performed our analysis on cells treated with selected cytokines that we experimentally confirmed could upregulate MHC molecules required for direct T cell activation. Of the cytokines tested, IFN-γ induced the most marked effects, even at relatively low concentrations. However, while hfVM, mesDAp, and DAn all upregulated MHC-I antigen processing and presentation genes, they all remained non-immunogenic in our assays, providing reassurance that even in an inflammatory environment, the mesDAp are unlikely to induce a significant direct T cell response.

An additional limitation is the relative insensitivity of in vitro immunogenicity assays, though our assay could detect T cell responses (proliferation, activation, and cytokine production) to other allogeneic cells (allo-moDCs). These assays are also inherently reductionist, lacking critical in vivo elements, including the blood-brain barrier, glial cell populations, tissue architecture, surgical trauma, and prolonged antigen exposure. These factors may explain why hfVM tissue did not elicit a detectable response in our assays, despite reports of immune activation in some clinical studies. However, other factors, such as the use of multiple donors in clinical hfVM grafts, might also explain these differences. Importantly, despite some evidence of immune activation, hfVM grafts have demonstrated long-term survival and functional benefit in patients, after a transient period of immunosuppression, consistent with their observed low immunogenicity. Our assays were designed to assess the likelihood of grafted cells activating peripheral T cells, rather than the full spectrum of immune responses that may arise in the post-transplant brain environment. Our data suggest that RC17-derived mesDAp share this low immunogenicity, providing confidence that similar immunosuppression strategies could be applied in the STEM-PD trial as those used for hfVM tissue (e.g., TRANSEURO).

2D or 3D organoid models containing neurons, glia, and T cells may be able to address some but not all of the limitations of in vitro models—however, immunogenicity testing would require the generation of multiple iPSC-derived cell types to ensure that all “recipient cells” are autologous. Given this, researchers, including ourselves, are increasingly turning to human immune system (HIS) mice to examine immune interactions in vivo. Nevertheless, these models also come with their own set of major challenges. These include (1) variability in the engraftment and functionality of the human immune system, (2) the presence of murine microglia, which may cause a xenogenic response with little relevance to the clinical situation, and (3) the limited time frame for assessing immune responses and graft functionality before graft-versus-host disease (GvHD) develops. This is particularly problematic for our work, as mesDAp require up to 6 months post transplantation to mature and exert functional benefits in a grafted host (Kirkeby et al., 2023).

Given the limitations of all preclinical models, translating data from in vitro or humanized mouse models to clinical decision-making is inherently very challenging. For example, although it is known that hfVM tissue shows low immunogenicity in vitro, grafts in patients with PD who have not received immunosuppression survive poorly (Freed et al., 2001), whereas short-term immunosuppression, using organ-transplant regimens, has supported long-term graft survival (Li et al., 2016). Consequently, although our analyses provide reassurance that the RC17 hESC-derived dopamine cell product exhibits low intrinsic immunogenicity, perhaps lower than that of hfVM, we cannot definitively determine the optimal immunosuppressive regimen for patients undergoing neuronal transplantation.

Ultimately, only clinical data can resolve this question. In the STEM-PD trial, we are systematically collecting samples from patients both before and after transplantation, with and without immunosuppression, to better understand the human immune response to these grafts. These patient-derived data will be critical for refining immunosuppressive strategies and improving the safety and efficacy of future stem cell-based therapies for PD.

Methods

Cell culture conditions

All long-term cultures were performed in the presence of 100 U/mL penicillin-streptomycin. Standard culture conditions involved the use of Corning tissue culture plates and incubation at 37^o^C/5% CO_2_. All reagent suppliers are listed in Table S1.

hESCs and derivatives

RC17 hESCs (Roslin cells, hPSCreg name: RCe021-A) are a clinical grade, healthy female hESC line with normal karyotype derived under GMP and xeno-free conditions; for these experiments, research-grade RC17 hESCs were utilized. Cells were obtained from the inner cell mass of a 3-day-old blastocyst stage embryo, surplus or unsuitable for clinical use. RC17 has normal pluripotency marker expression and can differentiate to all three germ layers in vitro. RC17 hESCs were utilized under study protocol REC No: 21/EE/0051.

Undifferentiated ESCs were cultured in StemMACS iPS-Brew, XF medium on culture plates coated with 0.5 μg/cm^2^ Laminin-521. All single-cell dissociations were done by incubating with StemPro Accutase (5 min, 37°C), and replating was done with 10 μM StemMACS Y-27632 ROCK inhibitor (ROCKi). Media was changed every 24 h.

Day 16 mesDAp were cultured in 2% B-27-supplemented Neurobasal CTS Media (plus 2 μM L-glutamine and 0.2% vol/vol penicillin-streptomycin) plus 20 ng/mL BDNF, 0.2 μM ascorbic acid (AA), and 100 ng/mL FGF8b on culture plates coated with 1 μg/cm^2^ Laminin-111. All single-cell dissociations were done by incubating with Accutase (5 min, 37°C), and replating was done with 10 μM StemMACS Y-27632 ROCKi. Media was changed every 48 h.

Mature DAn were cultured in the same mesDAp media with the addition of 1 μM DAPT and 500 μM cyclic (c)AMP on culture plates coated with 0.01% Geltrex in DMEM. Half media was changed every 48 h.

hfVM collection and maintenance

Fresh hfVM tissues were dissected from fetuses aged 7–10 weeks post conception collected at pregnancy termination from the Rosie Hospital, Cambridge, under full ethical approval (ethics ref: 96/085). hfVM tissues were stored in Hibernate-E medium for up to 4 days before washing in DMEM and incubation with Accutase (30 min, 37°C), with agitation at 10–15 min intervals. Accutase was removed, and the cells were then resuspended in DMEM with 2% B-27-supplemented 1% FBS and 1% PSF (Penicillin-Streptomycin-Fungizone) and plated on PDL (poly-D-lysine)/Laminin-2020-coated culture plates.

Differentiation of hESCs to mesDA progenitors/DAn

hESCs to mesDAp: RC17 hESCs were differentiated to in vitro using the 16-day GMP-compliant protocol derived by Nolbrant et al. (2017). Briefly, on day 0, RC17 hESCs, from a 70%–90% confluent well, were detached with EDTA and plated on Laminin-111-coated plates at a density of 10,000 cells/cm^2^ with 10 μM ROCKi. Cells were cultured in N-2 media supplemented with 10 μM SB431542, 100 ng/mL Noggin, 300 ng/mL Shh-C24II, and 0.85 μM CHIR9902 until day 9. Media was changed to N-2 media plus 100 ng/mL FGF8b until day 11, when they were Accutase-dissociated and replated (0.8 × 10^6^ cells/cm^2^) onto Laminin-111-coated plates with 10 μM ROCKi. Cells were cultured from day 11 to day 16 in 2% B-27-supplemented Neurobasal CTS media plus 20 ng/mL BDNF, 0.2 μM ascorbic acid, and 100 ng/mL FGF8b. At day 16, mesDAp underwent quality testing to verify correct identity using quantitative PCR and immunofluorescence. Cells were either cryopreserved for use in experiments or terminally differentiated to mature DAn in vitro (see the following text).

mesDAp to DAn: terminal differentiation and maturation was performed as per Nolbrant et al. (2017); cells were replated at a density of 250–400,000 cells/cm^2^ on 0.01% Geltrex/DMEM-coated plates (1 h, 37°C) and then cultured until at least day 45 in B-27-supplemented Neurobasal CTS media with 20 ng/mL BDNF, 0.2 μM ascorbic acid, 100 ng/mL FGF8b, 1 μM DAPT, and 500 μM cAMP.

Treatment of cells with inflammatory stimuli

Cells were seeded at a density of 0.5 × 10^6^ cells/cm^2^ on 24- or 96-well plates with their respective media and coatings as described earlier. For initial optimization experiments, cells were treated for 0–72 h with 0–50 ng/mL IFN-γ, TNF-α, IL-1β, or combinations of two/three. Once the duration and concentration of inflammatory stimuli was selected, all further treatments were done for 24 h with 5 ng/mL IFN-γ and/or TNF-α.

Immunofluorescence staining of ESC, mesDA progenitors, and DAn

Cells were fixed with 4% paraformaldehyde (PFA), blocked/permeabilized for 1 h in 0.25% Triton + 5% donkey serum and then incubated with primary antibodies (1 h at room temperature [RT] then overnight at 4°C) and secondary antibodies (1 h, RT) with PBS-Tween washes between steps. Nuclei were stained with DAPI (1:1,000; 10 min), washed three times, and then imaged on an EVOS FL fluorescent microscope. All scale bars represent 10 μm unless stated otherwise. See Table S2 for antibody details.

Flow cytometry of hESCs, mesDA progenitors, DAn, and hfVM

Cells were treated with/without inflammatory stimuli for 24 h, harvested with Accutase, and washed. Viability staining used Zombie NIR/Aqua (15 min, 4°C) and then cells were blocked with PBS + 7.5% FBS (15 min on ice) and stained with antibody cocktails (45 min, 4°C) with fluorescence-activated cell sorting (FACS) buffer washes between steps (PBS + 2.5% FBS). For extracellular panels, cells were fixed in 0.25% PFA in PBS (10 min, 4°C) and then resuspended in PBS for acquisition. For intracellular staining, cells were washed twice with PBS and then fixed/permeabilized using the eBioscience Transcription Factor Staining kit (45 min fix/perm at RT and two washes in perm/wash buffer) before staining with intracellular antibody cocktails (45 min, 4°C). Cells were washed twice then resuspended in PBS for acquisition. Fluorescence single stains for compensation were performed using single-stained cell samples and/or UltraComp eBeads. Flow cytometry was performed on LSR Fortessa or LSR-II Analyzers, and data were analyzed using FlowJo software. See Table S2 for antibody details.

RNA extractions for qPCR and nCounter

Cells with/without treatments (5 ng/mL IFN-γ, TNF-α, or both) were resuspended in RLT buffer and stored at −80°C until use. RNA was extracted using the RNeasyPlus Mini Kit according to the manufacturer’s instructions. RNA quality and quantity was confirmed using the Agilent 2100 Bioanalyzer and then frozen at −80°C until used for quantitative reverse-transcription PCR (RT-qPCR) or NanoString nCounter.

RT-qPCR

0.5–1 μg of RNA was reverse transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit. Following this, SYBR-green or TaqMan qPCR reactions were performed with gene-specific primers/probes in triplicates on the Quantstudio 12K Flex qPCR machine. GAPDH and ACTB were used as the reference genes for all samples. qPCR data were analyzed using average cycle threshold (CT) values in the Thermo Fisher Connect cloud software, and all data are represented using the double-delta CT method.

NanoString nCounter

RC17 hESCs (three vials; differentiated independently to day 16 mesDAp and day 45 DAn) and hfVM (three donors) were cultured for 24 h ±5 ng/mL IFN-γ, TNF-α, or both and then RNA extracted, quantified, and QC’d. NanoString nCounter was performed according to the manufacturer’s instructions using the Human Immunology V2 Panel with a custom Panel Plus of 20 genes (see Table S2). Briefly, 100 ng of extracted RNA was hybridized with 50 nucleotide capture/reporter probes (20 h, 65°C) and run on the MAX/FLEX system. 555 fields of view (FOV) were imaged per sample. Counts were normalized to the geometric mean of housekeeping genes, and a threshold for limit of detection was determined by 3^∗^mean of negative controls. Data were QC’d, normalized, and analyzed using nSolver, NACHO (NAnostring quality Control dasHbOard), and R packages EnhancedVolcano and Heatmap2.

Immunogenicity assays

Blood collection, PBMC isolation, and cell type enrichment/sorting

Blood was drawn by clinicians from healthy volunteers (aged 21–63) with informed consent in compliance with NHS Health Research Authority ethical approval reference no. 21/EE/0051. Exclusion criteria excluded individuals donating blood within 1 month of vaccination or any individual with diagnosed immune-related or autoimmune disorders. PBMCs were freshly isolated from heparinized blood using Ficoll-Paque Plus density gradient centrifugation.

Desired cell types (pan T cells and monocytes for moDC differentiation) were enriched from total PBMCs using MACS. CD14^+^ monocytes were first positively selected using CD14^+^ MicroBeads, and then T cells were negatively selected from the CD14^−^ subset using the PanT cell isolation kit. Purity of enriched populations was measured using flow cytometry.

T cell staining with eFluor cell proliferation dye

MACS-enriched pan T cells and PBMCs were stained with the eBioscience eFluor 450 cell proliferation dye according to the manufacturer’s instructions. Briefly, cells were resuspended in PBS, and then the cell proliferation dye was added dropwise while agitating the cells to a final concentration of 5 μM and then incubated (10 min, 37°C). Cells were washed with 4× volume of ice-cold RPMI + and 10% FBS then incubated (5 min on ice) and washed before cryopreservation.

moDC differentiation

All monocytes were used fresh due to the known issues of poor viability of cryopreserved monocytes. A full detailed protocol can be found on protocols.io (Curle et al., 2023). Monocytes, MACS-enriched from fresh PBMCs, were seeded at 0.5 × 10^6^ cells/cm^2^ in RPMI plus 5% heat-inactivated human AB serum and cultured with 50 ng/mL rhIL-4 and 50 ng/mL rhGM-CSF for 6 days with a half media change on day 3. On day 6, the cells were activated with 100 ng/mL lipopolysaccharide (LPS) for 24 h and then washed twice with PBS to remove residual LPS before co-culture.

Successful differentiation to a moDC phenotype was confirmed by morphology and by flow cytometry to determine the presence of mature dendritic cell (DC) markers (MHC-I, MHC-II, CD80, and CD86) and the absence of the monocyte marker CD14. Phagocytic function was demonstrated using 1 μm yellow-green FluoSpheres, and antigen presentation function was demonstrated by successful activation of T cells following 24-h moDC pre-exposure to the hyperimmunogenic hemocyanin from Megathura crenulata (KLH).

In vitro co-culture assays

Target cells (hESCs, mesDAp, DAn, or hfVM) were plated 24 h prior to co-culture on coated plates at 0.5 × 10^6^ cells/cm^2^ (96-well: 150,000 cells/well; 24-well: 500,000 cells/well) ±5 ng/mL IFN-γ. On the day of co-culture, cells were washed with PBS, then cryopreserved eFluor 450-stained PBMCs or T cells were thawed and added at a 1:1 ratio in a 1:1 mixture of complete RPMI media with respective cell media (see “Cell culture conditions”; RPMI-mixed media). Co-cultures were maintained for 5 days with half-media changes every 48 h. On day 5, supernatants were collected for enzyme-linked immunosorbent assay (ELISA)/Luminex analysis, PBMCs/T cells were collected and stained, and T cell proliferation and activation were measured using flow cytometry. Several positive controls were utilized. T cell/PBMC assays: (1) polyclonal T cell stimulation with soluble αCD3 (0.5 μg/mL; clone OKT3) and αCD28 (1 μg/mL) or Dynabeads Human T activator CD3/CD28 for T cell expansion and activation (1:30 bead:T cell ratio) or (2) allogeneic moDCs.

For moDC/T cell assays, target cells were collected using Accutase/EDTA and replated onto moDC wells in the presence of 10 μM ROCKi, with eFluor 450-stained T cells added concurrently; KLH antigen presentation served as a positive control.

Transwell assay: mesDAp were plated on Laminin-111-coated 24-well plates (500,000 cells/well) for 24 h ± 5 ng/mL IFN-γ and then co-cultured with PBMCs or T cells either in direct contact or separated by a 0.4 μm transwell insert. Contact wells received 1 mL total RPMI-mixed media, whereas transwell conditions used 500 μL in the plate and 500 μL in the insert; half media was replenished every 48 h in both compartments.

Inhibitors and blocking antibody treatments

To test for mechanisms of mesDAp-mediated suppression, T cell co-culture assays were performed in the presence of inhibitors and blocking antibodies. In all cases, target cells (mesDAp and/or T cells) were exposed to inhibitors or blocking antibodies for 2 h prior to the start of the assay and then refreshed throughout the duration of the 5-day co-culture. The following functional grade blocking/neutralizing antibodies were used, all at 10 μg/mL: monoclonal anti-CD276 (clone MIH35), Ultra-LEAF-purified monoclonal anti-TIGIT antibody (clone A15153A), and monoclonal anti-CD47 (clone B6H12). TGFβR1 inhibition was performed using the SB 431542 inhibitor at 10 μg/mL, and TGFβR1/2 blocking was done using LY2109761 inhibitor at 10 μg/mL. Cells were treated with 200 μM of 1-methyl-L-tryptophan (1-LMT; IDO-1 inhibitor) and 1-methyl-D-tryptophan (1-DMT; 1-LMT enantiomer) for exploration of a possible tryptophan mechanism of immunosuppression. For adenosine pathway investigation, cells were treated with 2.5 μM concentration of ZM241385 (adenosine 2A receptor inhibitor).

ELISA

The Human TGF-beta 1 Duoset ELISA was utilized with the Ancillary reagent kit 1 to measure secreted TGFβ1 in the supernatants of the T cell and moDC assays. 150 μL supernatants were collected at the end of the 5-day assay, centrifuged at 1,500g for 10 min, and then liquid was collected and stored at −80°C. Supernatants were thawed on ice and then pre-activated using the sample activation kit of 1.2N NaOH for activation and 0.5 M HEPES for neutralization (specific to the TGF-β kit to activate latent TGF-β). Sample titrations were performed to determine the optimal dilution ensuring that absorbance fell within the linear region of the standard curve.

Three wash steps were performed with 0.05% Tween 20 between steps. ELISA plates were coated with 2 μg/mL capture antibody overnight and then blocked in 1% BSA in PBS (1 h). Samples or standards (100 μL) were incubated (2 h) followed by 50 ng/mL detection antibody (2 h) and streptavidin-horseradish peroxidase (20 min). TMB substrate (1:1 H_2_O_2_/tetramethylbenzidine) was developed for 20 min and then stopped with 2 N H_2_SO_4_. Absorbance was read at 450 nm with 570 nm correction; concentrations were interpolated from standards (concentration range: 31.2–2,000 pg/mL) in GraphPad Prism.

Luminex

Human cytokine Luminex 8-plex assay (IFN-γ, TNF-α, IL-2, IL-4, IL-6, IL-8, IL-10, and GM-CSF) was run on assay supernatants (with magnetic washes) on a Bio-Plex 200. Beads were incubated with diluted samples/standards, followed by biotinylated detection antibody and streptavidin-PE, and then resuspended for acquisition per manufacturer instructions. Samples were run in technical duplicates; wells with CV >25% were excluded, and inter-plate control CVs were <25% so no inter-plate correction was applied. Inter-plate CVs for each cytokine were as follows: GM-CSF 1%, IFN-γ 4.9%, IL-2 13.5%, IL-4 6.7%, IL-6 10.5%, IL-8 0.3%, IL-10 1.1%, and TNF-α 10.5%.

Analysis, statistics, and visualization

Biological replicates (“n”) are reported in the figure legends and reflect independent biological starting material (not technical repeats). For phenotyping and transcriptomic analyses (Figures 1, 2, 3, and S1–S4), n for hESCs, mesDAp, and DAn represents independent differentiations: e.g., n = 3 means three independent hESC thaw/differentiation runs (separate starting stocks/passage batches) carried through to mesDAp and DAn. For hfVM, n represents the number of independent fetal donors from which ventral midbrain tissue was obtained. For immunogenicity co-culture assays (Figures 4, 5, and S4–S6), n represents independent target-cell preparations and independent immune donors. Specifically, n = 3 indicates three independently generated target-cell preparations (e.g., three separate mesDAp differentiations and/or hfVM donor preparations) tested in co-culture against immune cells from three separate human blood donors.

All data are presented as mean ±1 standard deviation (1SD). All statistical analyses and visualization were performed using GraphPad Prism or R packages. Statistical tests are stated in all figure legends, n.s. = not significant; ^∗^p < 0.05, ^∗∗^p < 0.01, ^∗∗∗^p < 0.005, and ^∗∗∗∗^p < 0.001. All figures were generated using BioRender.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Professor Joanne L. Jones (jls53@medschl.cam.ac.uk).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data reported in this study (NanoString nCounter count matrices and metadata, flow cytometry outputs, microscopy images, Luminex/ELISA data, and qPCR data) will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We would like to thank Dr. Agnete Kirkeby for her support throughout this project. This work was supported by 10.13039/501100019326UK-Regenerative Medicine Platform (UKRMP; MR/S020934/1), 10.13039/100010661Horizon 2020-funded Neural Stem Cell-Reconstruct (NSC-R; G101558), 10.13039/501100000833Rosetrees Trust and 10.13039/501100020400John Black Charitable Fund (G118482), and 10.13039/100014461NIHR Biomedical Research Centre funding to J.L.J. (BRC-1215-20014). A.J.C.’s studentship was supported by 10.13039/501100003343Cambridge Trust and 10.13039/100014000Masonic Charitable Trust Foundation, and A.S.’s studentship was supported by the 10.13039/100010269Wellcome Trust. We performed flow cytometric analysis in the Cambridge 10.13039/100014461NIHR BRC Cell Phenotyping Hub (Department of Medicine, University of Cambridge). The views expressed here are those of the authors and not necessarily those of the NIHR.

Author contributions

A.J.C., S.V.F., S.K.H., S.Q., X.H., and A.S. performed the experiments. A.J.C. and S.Q. analyzed the data. A.J.C. wrote the draft manuscript, with input from J.L.J. and R.A.B. All authors contributed to the design of experiments and data interpretation and approved the final version of the manuscript.

Declaration of interests

R.A.B. has provided consultancy advice around dopamine cell-based therapies to Aspen Neuroscience, BlueRock Therapeutics, FCDI, and Novo Nordisk. J.L.J. reports consultancy work for Enhanc3DGenomics, Sanofi, and Roche. S.Q. is a current employee at ISOgenix Ltd.