Identification of unstable regulatory and autoreactive effector T cells that are expanded in patients with FOXP3 mutations

Abstract

Studies of the monogenic autoimmune disease immunodysregulation polyendocrinopathy enteropathy X-linked syndrome (IPEX) have elucidated the essential function of the transcription factor FOXP3 and thymic-derived regulatory T cells (T_regs) in controlling peripheral tolerance. However, the presence and the source of autoreactive T cells in IPEX remain undetermined. Here, we investigated how FOXP3 deficiency affects the T cell receptor (TCR) repertoire and T_reg stability in vivo and compared T cell abnormalities in patients with IPEX to those in patients with autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy syndrome (APECED). To study T_regs independently of their phenotype and to analyze T cell autoreactivity, we combined T_reg-specific demethylation region analyses, single-cell multi-omic profiling, and bulk TCR sequencing. We found that patients with IPEX, unlike patients with APECED, have expanded autoreactive T cells originating from both autoreactive effector T cells (T_effs) and T_regs. In addition, a fraction of the expanded T_regs from patients with IPEX lost their phenotypic and functional markers including CD25 and FOXP3. Functional experiments with CRISPR/Cas9-mediated FOXP3 knock-out T_regs and T_regs from patients with IPEX indicated that the patients’ T_regs gain a Th2 skewed T_eff-like function, which is consistent with immune dysregulation observed in these patients. Analyses of FOXP3 mutation-carrier mothers and a patient with IPEX after hematopoietic stem cell transplantation, indicated that T_regs expressing non-mutated FOXP3 prevent the accumulation of autoreactive T_effs and unstable T_regs. These findings could be directly used for diagnostic and prognostic purposes and for monitoring the effects of immunomodulatory treatments.

One Sentence Summary:

Autoreactive T cells in patients with IPEX syndrome originate from both effector T cells and unstable, Th2-skewed regulatory T cells.

INTRODUCTION

Autoimmunity occurs when immune cells fail to distinguish “self” from “non-self” and initiate immune reactions against their autologous tissues. To limit T cell autoreactivity, developing T cells are selected in the thymus by medullary thymic epithelial cells (mTECs), B cells, and dendritic cells, which express and present self-antigens to T cells (1), a mechanism referred as central T cell tolerance. The expression of tissue-restricted self-antigens (TRAs) in mTECs is regulated by the autoimmune regulator (AIRE). Autoreactive T cells that recognize self-antigens with high affinity are eliminated (negative selection), whereas T cells that recognize self-peptides presented by major histocompatibility complexes (MHCs) with low affinity are positively selected and become effector T cells (T_effs). Alternatively, some T cells with intermediate to high affinity to self-antigens may be positively selected and adopt a unique developmental fate by differentiating into regulatory T cells (T_regs). Hence, the T cell receptor (TCR) repertoire of T_regs is physiologically self-reactive (2–4). However, the process of negative selection is not 100% accurate. Some autoreactive T_effs escape negative selection and enter the periphery (5–7). To prevent autoimmunity, these autoreactive T cells are controlled by various regulatory mechanisms collectively referred to as peripheral tolerance.

Two human monogenic diseases with autoimmunity exemplify the importance of central and peripheral tolerance. Mutations in AIRE result in failed T cell selection and development of autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy syndrome (APECED), manifested by progressive autoimmune targeting of various tissues (8). Analyses of the T_reg and T_eff TCR repertoires of these patients indicated that some of the common T_reg TCR clones were present within the T_eff compartment, indicating an altered distribution of the self-reactive T cells as a consequence of aberrant selection (9). Mutations in FOXP3, a transcription factor essential for T_reg function, lead to the prototypic example of T_reg deficiency, called immune dysregulation polyendocrinopathy enteropathy X-linked syndrome (IPEX), a severe, early-onset, multiorgan, and life-threatening autoimmunity (10). However, the presence of autoreactive T cells and their lineage origin in patients with IPEX has not been characterized, representing an important gap in understanding the role of FOXP3 and T_regs in controlling autoreactive T cells in humans.

Evaluation of the epigenetic landscape of the T_reg-specific demethylated region (TSDR) is the best approach to define thymus-derived T_regs since demethylation of the TSDR is unique to FOXP3-expressing T_regs and distinguishes them from T_effs that can transiently express FOXP3 upon activation (11). Surprisingly, patients with IPEX have an increased frequency of TSDR-demethylated cells relative to CD4⁺ T cells in whole blood samples (12, 13), although the frequency of T_regs measured by their standard phenotypic markers CD3⁺CD4⁺CD25^high and FOXP3⁺, or CD127⁻, or both FOXP3⁺ CD127⁻ ranges from low to normal (14–17). This discrepancy led us to hypothesize that T_regs from patients with IPEX are unstable or plastic and expand beyond the phenotypic T_reg compartment. Indeed, data mainly from murine studies revealed that defects during T_reg development or inflammatory conditions may give rise to unstable or plastic T_regs often referred as “wannabe” or “ex” T_regs, which originate from physiologically self-reactive T_regs but may acquire T_eff-like phenotype (18–24). These data called for human studies of T_reg plasticity in samples from patients with IPEX to unravel whether some phenotypical T_effs have a T_reg origin, what T_eff phenotype T_regs may acquire, and how T_reg plasticity or instability may affect TCR repertoire autoreactivity.

Here, we show that the T_eff compartment of patients with IPEX, unlike that of patients with APECED or healthy donors (HDs), contains TSDR-demethylated cells demonstrating the presence of a population that displays a T_reg lineage marker but does not express CD25, FOXP3, or both CD25 and FOXP3, two molecules normally expressed by T_regs. We named this population “loss-of-identity” T_regs and characterized, at the single cell-level, their transcriptome, TCR repertoire, and protein expression. We show that the presence of loss-of-identity T_regs in vivo is prevented by healthy T_regs expressing unmutated FOXP3. Moreover, Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9-mediated FOXP3 knockout (FOXP3^KO) HD T_regs and IPEX T_regs showed proinflammatory, and Th2-effector functions, recapitulating the immune dysregulation observed in patients with IPEX. Together with the TCR repertoire analyses in phenotypically and epigenetically defined T cell subsets, our data indicate that loss of FOXP3 function in humans leads to increased autoreactivity in T_effs and expansion of T_regs with T_eff-like phenotype, thereby suggesting an expansion of two sources of autoreactive T cells in patients with IPEX.

RESULTS

The T_eff compartment of patients with IPEX contains TSDR-demethylated cells and a fraction of the TSDR-demethylated cells is FOXP3 negative.

This study was performed on samples from patients with IPEX, mutation carrier mothers, and patients with APECED. Mutations, age, and sex of participants included in the study are summarized in table S1. To investigate T_reg plasticity in patients with IPEX, we measured the frequency of TSDR-demethylated cells in sorted CD3⁺CD4⁺ T cell subsets with different expression of CD25 and CD127, the relevant receptors for interleukin (IL)-2 and IL-7 differentially regulating T_reg and T_eff homeostasis (25). To this aim, two subpopulations of T_effs were sorted as CD25⁻CD127^low and CD25^-/dimCD127^high, hereafter named T_eff1 and T_eff2, respectively (Fig. 1A). In parallel, CD25^highCD127^low T_regs were isolated, and shortly expanded in vitro to obtain a sufficient amount of DNA for the analyses. The total frequencies of T_eff1s, T_eff2s, and T_regs were not different between HDs and patients with IPEX (Fig. 1B), and neither were frequencies of CD4⁺ T cells (Fig. 1C). To assess TSDR demethylation, we used a finely optimized quantitative polymerase chain reaction (qPCR) assay with primers and probes specific for the demethylated TSDR sequence, and thus, instead of analyzing the degree of demethylation in bulk population, this assay determines the frequency of the cells with fully or largely demethylated TSDRs (11, 26). Results show that, unlike in HDs, a significant (p<0.0001) fraction of IPEX T_eff1s was TSDR-demethylated (Fig. 1D, fig. S1). The presence of TSDR-demethylated cells in T_eff2 did not reach statistical significance (Fig. 1D, p=0.0582) As expected, the sorted CD25^highCD127^low T_regs of both patients with IPEX and HDs contained almost entirely TSDR-demethylated cells (Fig. 1D). These results indicate that T cells with this T_reg epigenetic marker were found among the CD25-negative cell population in samples from patients with IPEX.

Figure 1. The T_eff compartment of patients with IPEX contains TSDR-demethylated cells and a fraction of them are FOXP3⁻.

(A) Representative image of a gating strategy to sort and analyze the frequencies of T_eff1, T_eff2, and T_reg subpopulations. (B and C) Quantification of the frequencies of T_eff1s, T_eff2s, and T_regs (B) and CD3⁺CD4⁺ T cells (C) from PBMC of 11 patients with IPEX (green) and 11 HDs (blue). (D) Analysis of the frequencies of TSDR-demethylated cells in sorted T_eff1s, T_eff2s, and T_regs of 10 patients with IPEX and 16 HDs. Cells were sorted as shown in (A). (E) Analyses by FACS of FOXP3 expression in T_eff1 and T_eff2 populations in 7 patients with IPEX. The FACS gating strategy is shown in fig. S2A. (F) Representative FACS plots show the gating strategy used to sort FOXP3⁻ T_eff1s and T_eff2s and FOXP3⁺ T_eff1s and T_regs from patients with IPEX and HDs for subsequent TSDR demethylation analysis. The T_eff1s, T_eff2s, and T_regs were gated as shown in (A). (G) Quantification of the frequency of TSDR-demethylated cells in FOXP3⁻ T_eff1s and T_eff2s, and FOXP3⁺ T_eff1s and T_regs in 7 (3 for FOXP3⁺ T_eff1s) patients with IPEX and 13 HDs. Data points shown in gray (IPEX T_eff2s) represent samples with a low cell number or genome copies, where the values may be partially imprecise. (H) FACS analyses (Median Fluorescence Intensity, MFI, and percentages, %) of FOXP3, CTLA4, CD45RA, TIGIT, CD39, and HELIOS expression in FOXP3⁺ T_regs and FOXP3⁺ T_eff1s in 7 patients with IPEX. The data are analyzed in a pairwise fashion comparing the respective populations within the same patients. Data are presented as mean ± standard error of the mean (SEM). Significance was evaluated using the Mann-Whitney test (B to G) or Wilcoxon matched-pairs test (H). n.s., not significant; *p<0.05, ***p<0.001, ****p<0.0001.

The vast majority of IPEX-causing mutations do not abolish FOXP3 expression but diminish FOXP3 function (13, 15). However, FOXP3⁺ cells were detected mainly in T_eff1 compartment of patients with IPEX (Fig. 1E and fig. S2A). The results from this fluorescence-activated cell sorting (FACS) analysis were an important part of the initial rationale for further characterizing the T_eff1 and T_eff2 compartments (fig. S2B), and to test if i) the FOXP3⁺ T_eff1s are TSDR demethylated and ii) the TSDR-demethylated cells are present also among FOXP3⁻ T_effs. To this aim, we analyzed the frequency of TSDR-demethylated cells in the sorted FOXP3⁻ T_eff1, FOXP3⁻ T_eff2, FOXP3⁺ T_eff1, and FOXP3⁺ T_reg subsets (patients with IPEX: n=7 and HDs: n=13) (Fig. 1F and G, fig. S2C). We found that FOXP3⁺ T_eff1s are comprised almost exclusively of TSDR-demethylated cells and that TSDR-demethylated cells are also present among FOXP3⁻ T_eff1s in patients with IPEX, but not in HDs. Compared with phenotypic T_regs, the FOXP3⁺ T_eff1s are predominantly CD45RA⁻, express lower FOXP3 and cytotoxic T-lymphocyte-associated protein 4 (CTLA4), but express more CD39 and TIGIT (Fig. 1H, fig. S2A and D). Both phenotypic T_regs and FOXP3⁺ T_effs expressed Helios. In summary, these analyses suggest that the FOXP3⁺ T_eff1s are antigen-experienced T_regs that downregulated FOXP3 expression. Collectively, our data show that in patients with IPEX, a fraction of the epigenetically-defined T_regs do not express CD25, have reduced or no expression of FOXP3, and some gain CD127 expression.

Characterization of T_reg subpopulations using single-cell multi-omic profiling of CD4⁺ T cells from patients with IPEX revealed a disease-specific population of T_regs that is dominated by CD25⁻ cells.

To gain further insight into T_reg heterogeneity in samples from patients with IPEX, we performed single-cell multi-omic RNA, protein, and TCR profiling using the 10X genomics platform. We assayed CD4⁺ T cells isolated from three HDs and three patients with IPEX. After data filtering, we obtained 17,865 T cells from HDs and 17,566 T cells from patients with IPEX, relatively evenly distributed amongst the samples (Fig. 2A). Cells were clustered and visualized using uniform manifold approximation and projection (UMAP) according to their transcriptomics (Fig. 2B). Clusters were annotated according to their differential expression of key genes (Fig. 2C) and the expression of CD45RA protein (based on sequencing of antibody derived tags [ADT]) (fig. S3A). We identified the following populations: cytotoxic (GZMK⁺, GZMA⁺), Th1/Cytotoxic (TBX21⁺, GZMB⁺), Th2 (GATA3⁺), Th17 (RORC⁺, CCR6⁺), naïve (CD45RA⁺ [ADT]), memory (adt_CD45RA⁻), T_reg naïve (FOXP3⁺, IKZF2⁺, CD45RA⁺ [ADT]), and two populations of memory T_reg (FOXP3⁺, IKZF2⁺, CD45RA⁻) (Fig. 2B and C, fig. S3A). The T_reg memory 1 population included cells from both HDs and patients with IPEX, whereas the T_reg memory 2 population was almost exclusively composed of cells isolated from patients with IPEX (Fig. 2D). The T_reg memory 2 population was also among the most frequent memory T cell populations in these patients. TCR analyses demonstrated that T_reg memory 1 and T_reg memory 2 were clonally related, suggesting a common origin of these populations (Fig. 2E and fig. S3B). Compared with T_reg memory 1, T_reg memory 2 cells demonstrated reduced expression of FOXP3 and IL2RA, but increased expression of IL7R, CD69, and ZPF36L2, which was previously shown to be a negative regulator of IKZF2 expression (27) (Fig. 2F, data file S1). In addition, the T_reg memory 2 population showed differential expression of genes associated with tumor necrosis factor (TNF)-α signaling. Indeed, the enrichment in expression of TNF-α signaling target genes was confirmed by pathway enrichment analysis (Fig. 2G and fig. S3C). These data demonstrate that patients with IPEX have both typical (T_reg memory 1) and atypical (T_reg memory 2) memory T_regs. We observed that the disease-specific T_reg memory 2 are clonally most related to typical T_regs, they display a gene expression signature indicative of TNF-α exposure, and demonstrate reduced mRNA expression of prototypical T_reg markers.

Figure 2. Characterization of T_reg subpopulations using single-cell multi-omic profiling of CD4⁺ T cells revealed a population of T_regs specific for patients with IPEX.

(A) Number of cells identified per donor. (B) UMAP projection of the T cell populations identified in HDs’ samples (left) and samples from patients with IPEX (right). (C) Expression of selected genes in the different populations identified by the specific color. The dot color intensity indicates the amount of expression and the size of the dot indicates the number of positive cells for a given gene in the population. (D) Distribution of the identified populations in the CD4⁺ T cells from HDs and patients with IPEX. The frequency of each population is defined as a percentage of cells present in the indicated populations from the total number of cells in a given sample. (E) TCR repertoire overlaps between T_reg memory 2 populations and all other populations identified. The color of an arrow indicates the proportion of the overlapping clones between the populations. Arrows always start from a population to which the proportion is indicated. (F) Differential expression analyses between T_reg memory 2 and 1. The relevant genes are indicated by a label. The genes with positive fold change (FC) are upregulated in the T_reg memory 2 compared with the T_reg memory 1. P values were calculated using the Wilcoxon rank sum test. n.s, not significant. (G) TNF-α signaling was identified as the most significantly differentially expressed gene set by fast gene set enrichment analysis (FGSEA) comparing T_reg memory 2 and 1. Genes were ranked based on fold changes; the top panel shows the enrichment score and the bottom panel the fold changes of the genes. (H) Representative FeatureScatter plot of CD25 and CD127 expression using antibody-derived tag data [ADT]. The plot shows one IPEX and HD sample. The FOXP3 positivity was determined based on the mRNA expression. (I and J) Contribution of each T cell population from (B to D) to the total number of FOXP3⁺CD25^highCD127^low (I) and FOXP3⁺CD25^lowCD127^low/high (J) cells. FOXP3⁺CD25^highCD127^low and FOXP3⁺CD25^lowCD127^low/high cells were identified as shown in (H). “Other clusters” are pooled non-T_reg populations from (B to D). Data are presented as mean ± standard deviation (SD).

To validate our cytometric findings in Fig. 1, we assessed CD127 and CD25 expression as determined using ADT in conjunction with FOXP3 transcript expression (Fig. 2H). In both HDs and patients with IPEX, we observed a high frequency of FOXP3⁺ cells among the CD25^highCD127^low cells (fig. S3D). In contrast to HDs, where the FOXP3⁺ cells made up a minor fraction of the CD25^lowCD127^low/high, about 6% of all CD25^lowCD127^low/high cells in patients with IPEX were FOXP3⁺ (fig. S3E), recapitulating our cytometric findings. Next, we assessed how the FOXP3⁺ CD25^lowCD127^low/high and the FOXP3⁺ CD25^highCD127^low cells were distributed across the different transcriptionally-defined T_reg populations. The FOXP3⁺ CD25^highCD127^low cells from both HDs and patients with IPEX associated primarily with the T_reg memory 1 and T_reg naive population (Fig. 2I). However, about a third of the FOXP3⁺ CD25^highCD127^low cells of patients with IPEX were also present in the T_reg memory 2 populations. FOXP3⁺ CD25^lowCD127^low/high cells, although infrequent in HDs (fig. S3E), were spread among the non-T_reg populations in HDs, and in patients with IPEX were primarily associated with the T_reg memory 2 population (Fig. 2J).

These data show that the atypical T_reg memory 2 population is mainly composed of T_regs that have lost CD25 and gained CD127 expression. Consistent with the FOXP3⁺CD25^lowCD127^low/high cells displaying a T_reg memory 2 transcriptional signature, when comparing FOXP3⁺CD25^highCD127^low versus FOXP3⁺CD25^lowCD127^low/high from patients with IPEX we observed a similar set of differentially expressed genes (fig. S3F, data file S2) and enrichment pathways (fig. S3G and H).

TCR repertoire analyses indicate increased autoreactivity in CD4⁺ T cells from patients with IPEX.

We analyzed the TCR repertoires in CD4⁺ T cells from patients with IPEX to potentially illustrate the FOXP3-dependent role of T_regs in controlling autoreactive T cell expansion in humans in vivo. We performed DNA sequencing of TCRB of T_eff1s, T_eff2s, and T_regs from seven patients with IPEX and ten HDs. In total, we identified 5.4×10⁵ and 7.4×10⁵ T_eff1, 5.1 ×10⁵ and 6.9×10⁵ T_eff2, and 1.7 ×10⁵ and 2.3 ×10⁵ T_reg TCR sequences from patients with IPEX and HDs, respectively. The number of sequences identified per sample for each subpopulation was well-balanced between the HDs and patients with IPEX (Fig. 3A). The fraction of productive rearrangement did not show a significant difference in all three cell populations of patients with IPEX as compared with those of HDs (T_regs p=0.2698, T_eff1s p=0.3148, T_eff2s p=0.4747), and the overall Simpson clonality index was largely normal (Fig. 3B and C). We did not find any major bias in TCRB gene usage (fig. S4). Alteration of the TCR repertoires of T_eff1s and especially T_eff2s from patients with IPEX were evidenced by their different distributions of CDR3ß length compared with their counterparts in HDs. Indeed, whereas T_regs from both patients with IPEX and HDs display a similar CDR3ß length distribution, T_eff1s and T_eff2s in patients with IPEX were enriched in clones with longer CDR3ß loops (Fig. 3D). To investigate potential TCR autoreactivity, we analyzed the presence of autoreactivity promoting and limiting amino acid (AA) doublets at positions 6 and 7 of the CDR3ß loop. These AA doublets are TCR features associated with negative and positive T cell selection, respectively, and were suggested to be a mean for distinguishing normal versus autoimmunity-prone T cell repertoires (3, 4, 28, 29). We found increased frequencies of the autoreactivity promoting AA doublets in the TCR repertoires of both T_eff1s and T_eff2s from patients with IPEX (Fig. 3E). Consistently, we found a significantly reduced frequency of autoreactivity limiting AA doublets in T_eff2s (p=0.0431, Fig. 3F). We conclude that the differential length usage of CDR3ß, increased frequency of autoreactivity AA promoting doublets, and reduced frequency of autoreactivity limiting AA doublets in T_effs from patients with FOXP3 mutation indicate increased autoreactivity in the patients’ T_eff compartments. Although T_eff1 is composed of both TSDR-demethylated and bona fide T_eff, T_eff2 contains only a negligible number of TSDR-demethylated cells. Therefore, the abnormal TCR repertoire in T_eff2 is consistent with the idea that increased autoreactivity originates from bona fide T_effs.

Figure 3. TCR analyses indicate increased TCR repertoire autoreactivity in patients with IPEX.

(A) Shown is the number of TCRB sequences that were identified in T_eff1s, T_eff2s, and T_regs from patients with IPEX (green) and HDs (blue). (B) Shown is the fraction of TCRB productive rearrangements for T_eff1s, T_eff2s, and T_regs from patients and HDs. (C) Shown is the Simpson clonality index for T_eff1s, T_eff2s, and T_regs from patients with IPEX and HDs. (D) Analysis of CDR3ß length of T_regs (left), T_eff1s (middle), and T_eff2s (right) from patients with IPEX and HDs. (E and F) Frequency of autoreactivity promoting (E) and autoreactivity limiting (F) AA doublets in CDR3ß sequences in T_eff1, T_eff2, and T_reg populations from patients with IPEX and HDs. TCRB sequencing was performed in all three T cell subsets of 10 HD and 7 IPEX patients (A to F). Analyses in (C to G) were performed from productive rearrangements. Data are presented as mean ± SEM. Statistical significance was evaluated using the Mann-Whitney test; n.s., not significant; *p<0.05, **p<0.01.

In addition, we found a positive correlation between the frequency of TSDR-demethylated cells and the frequency of autoreactivity promoting AA doublets in T_eff1s, suggesting that TSDR-demethylated cells in the T_eff1 compartment contribute to changes in T_eff1 TCR repertoire, indicative of increased autoreactivity (Fig. 4A). There was no significant correlation between the frequency of TSDR demethylated cells in T_eff2 compartment and frequency of autoreactivity promoting AA doublets in T_eff1 compartment (p=0.2357, Fig. 4A), and neither between the frequency of TSDR-demethylated cells and the presence of autoreactivity promoting AA doublets within T_eff2 compartment (p=0.1667, Fig. 4B). However, we found a positive correlation between the presence of TSDR-demethylated cells in T_eff1 and the frequency of autoreactivity promoting AA doublets in T_eff2 (p=0.0341, Fig. 4B). These data suggest that the presence of TSDR-demethylated cells in T_eff1 is predictive of an abnormal TCR repertoire that may be associated with increased autoreactivity of the T_eff2 compartment composed mostly of bona fide T_effs. Hence, the loss-of-identity T_reg population may favor the expansion of autoreactive T_effs.

Figure 4. Frequencies of TSDR-demethylated cells correlate with increased frequencies of autoreactivity promoting AA doublets in T_eff compartments of patients with IPEX.

(A) Correlation between % of TSDR-demethylated cells in T_eff1s (left) and T_eff2s (right), and frequency of autoreactivity promoting AA doublets in T_eff1s. (B) Correlation between % of TSDR-demethylated cells in T_eff1s (left) and T_eff2s (right), and the frequency of autoreactivity promoting AA doublets in T_eff2s. Data were analyzed using Spearman correlation; n.s., not significant; *p<0.05. The p-values, Spearman r, and a simple linear regression are shown (n=7).

Mothers of patients with IPEX carrying FOXP3 mutations have a normal distribution of TSDR-demethylated cells and a normal TCR repertoire.

The FOXP3 gene is located in the X chromosome. As a consequence, in carrier mothers of FOXP3 mutations, random chromosome X inactivation results in the expression of mutated FOXP3 in 50% of T_effs (30, 31). In contrast to T_effs, the majority of carrier mothers’ CD25^high T_regs express the non-mutated FOXP3 allele (30, 32). Therefore, unlike in patients with IPEX, the T_regs in the healthy mothers are functional and able to control the partially mutated T_effs. However, it is unknown if i) some FOXP3-mutated TSDR-demethylated cells are present in T_eff compartment, and ii) carrier mothers have abnormal TCR repertoires indicative of increased autoreactivity in the T_effs. We sorted T_eff1s, T_eff2s, and T_regs from five carrier mothers and from five sex-matched HDs. We did not find TSDR-demethylated cells in any of the T_eff cell subsets tested (Fig. 5A). These results demonstrate that in carrier mothers, as well as in HDs, in which functional T_regs are present, TSDR-demethylated cells are restricted to T_reg compartment.

Figure 5. IPEX carrier mothers have a normal distribution of TSDR-demethylated cells and TCR repertoires of T_eff2, and the presence of donor-derived cells prevents expansion and normalizes the distribution of TSDR-demethylated cells in a patient with IPEX post HSCT.

(A) Shown are the frequencies of TSDR-demethylated cells from sorted T_eff1s, T_eff2s, and T_regs from 5 FOXP3 mutation carrier mothers (green) and 5 sex-matched HDs (blue). The sorting strategy was the same as in Fig. 1A. (B) Frequency of autoreactivity promoting and limiting AA doublets in CDR3ß repertoire of T_eff2 cells from 6 FOXP3 carrier mothers and 6 sex-matched HDs. Analysis in (B) was performed from productive rearrangements. (C) FACS analysis of chimerism 2.5 years post-transplantation in a patient with IPEX shows a competitive advantage of healthy donor-derived T_regs over patient-derived T_regs. Donor-derived cells are HLA-A2⁺ and host-derived cells are HLA-A2⁻. (D) TSDR demethylation analysis of sorted host- and donor-derived T_effs from the patient with IPEX before and 2.5 years post-transplantation. In addition, HD T_effs were sorted in parallel as a control (light blue). (Left panel) Gating strategy. (Right panel) Quantification of the TSDR-demethylated cell frequencies in the sorted populations. (E) TSDR demethylation analysis from whole blood at various time points pre- and post-transplantation. The frequencies of TSDR-demethylated cells were normalized to % of CD4⁺ T cells also determined by epigenetic analyses. The mean and SD of HD (previously published data (13); for each value see data file S4) are depicted as full and dashed lines, respectively. (F) Quantification of TSDR:CD4 ratio before transplantation, less than one year post- transplantation, and more than one year post-transplantation. Data are presented as mean ± SEM. Data in (F) were analyzed using Dunn’s multiple comparisons test and in (A and B) using Mann-Whitney test; n.s., not significant; *p<0.05.

We sequenced TCRB repertoire of T_eff2s from six carrier mothers and six sex-matched HDs (Fig. 5B). We did not find any differences in the frequencies of autoreactivity promoting and limiting AA doublets in carrier mothers as compared with HDs. These data further support that the aberrant TCR repertoire in T_effs of patients with IPEX mainly originates from the expansion of autoreactive TCR clones owing to a loss of T_reg function, rather than resulting from a FOXP3 cell-intrinsic effect in T_effs.

Decreased frequency of TSDR-demethylated cells in T_eff were observed in a patient with IPEX after hematopoietic stem cell transplantation, despite low donor chimerism.

We analyzed the frequency of TSDR-demethylated cells in the T_effs of a patient with IPEX prior to and up to 2.5 years post-hematopoietic stem cell transplantation (HSCT), which resulted in mixed donor chimerism. Patient- and the donor-derived cells could be distinguished using FACS by the differential expression of human leukocyte antigen (HLA)-A2. The patient donor chimerism 2.5 years post-transplantation was only about 15% in T_eff compartment and about 38% in T_reg compartment (Fig. 5C). Despite the relatively low chimerism at this time post-HSCT in both T_regs and T_effs, we observed a three-fold reduction in the frequency of TSDR-demethylated cells in the host-derived T_effs as compared with the pre-HSCT sample (Fig. 5D). In addition, the kinetics of the TSDR demethylation in whole blood showed that the proportion of TSDR-demethylated cells relative to the frequencies of CD4⁺ T cells significantly decreased over time (p=0.0498), reaching normal values one-year post-HSCT (Fig. 5E and F). These data suggest that the presence of even a small proportion (about 40%) of the donor-derived T_regs prevents the expansion of FOXP3 mutated T_regs and the accumulation of TSDR-demethylated cells in T_eff compartment.

In patients with APECED, TSDR-demethylated cells are restricted to the T_reg cell subset but TCR repertoire analyses indicate increased autoreactivity.

To get insight into the potential thymic origin of TSDR-demethylated cells in T_effs, we analyzed TSDR demethylation in T_regs and T_effs from patients with APECED, in which a T cell-extrinsic defect in thymic selection leads to impaired T_reg commitment and presence of clones that were meant to be T_regs within the T_eff pool (9). We sorted T_eff1s, T_eff2s, and T_regs from three patients with APECED and five HD or carrier relatives, if available. In contrast to patients with IPEX, TSDR-demethylated cells were not detected in the T_effs of patients with APECED, whereas APECED T_regs were exclusively comprised of TSDR-demethylated cells (Fig. 6A). In line with the aberrant T_eff and T_reg selection, analyses of previously reported TCR sequencing data (9) together with newly generated TCRB sequencing data (total six patients with APECED and six HD) showed an increased frequency of autoreactivity promoting AA doublets in both T_effs and T_regs of patients with APECED as compared with the HDs (Fig. 6B). The frequency of autoreactivity limiting AA doublets in T_effs was also reduced (Fig. 6B). Hence, although patients with APECED present with abnormal TCR repertoire consistent with increased T_eff autoreactivity, the T_reg lineage commitment in AIRE deficiency does not result in the aberrant presence of TSDR-demethylated cells in the T_eff compartment. These results, therefore, suggest that the presence of TSDR-demethylated cells in T_eff compartment of patients with IPEX specifically results from defective FOXP3 function.

Figure 6. Patients with APECED have a normal distribution of TSDR-demethylated cells, but an increased and reduced frequency of autoreactivity promoting and limiting AA doublets, respectively, in T_reg and T_eff compartments.

(A) Frequencies of TSDR-demethylated cells in sorted T_eff1, T_eff2, and T_reg populations from 3 patients with APECED (purple) and 5 HDs (green), including 4 parents and 1 HD. (B) Analysis of TCR receptor autoreactivity from previously reported TCR sequencing data of four patients with APECED and four HDs (9) together with newly obtain CDR3ß sequences from two patients with APECED syndrome and two HDs. (Left graph) Frequency of autoreactivity promoting AA doublets in CDR3ß of T_regs and T_effs. (Right graph) Frequency of autoreactivity limiting AA doublets in CDR3ß of T_regs and T_effs. Analyses were done from productive rearrangements. Data are presented as mean ± SEM. Data were analyzed using the Mann-Whitney test. n.s., not significant; *p<0.05, **p<0.01.

TCR-stimulated FOXP3^KO T_regs from HDs and T_regs from patients with IPEX showed greater production of proinflammatory cytokines compared to controls.

Since patients with IPEX have TSDR-demethylated cells present in T_eff compartment and a fraction of these cells do not express FOXP3, we investigated the T_eff function of T_regs following the loss of FOXP3 expression. These experiments serve as a model to study FOXP3-deficient T_regs and to identify cytokines that are the most affected by the loss of FOXP3 expression. We generated FOXP3^KO T_regs from HD T_regs as described previously (33). Briefly, we used CRISPR/Cas9 technology combined with Adeno-Associated Virus Type 6 (AAV6) delivery of FOXP3 homologous template containing truncated nerve grow factor receptor (NGFR) reporter gene under the control of PGK promoter. FOXP3^KO T_regs were purified using NGFR (Fig. 7A). Control T_regs were treated as the FOXP3^KO but electroporated in the absence of GuideRNA/Cas9 and not treated with AAV6. Upon TCR-mediated activation, we observed an enhanced expansion of FOXP3^KO T_regs (Fig. 7B) and a slight but significant increase in the proportions of TSDR-demethylated cells in FOXP3^KO over control T_regs (p=0.0244, Fig. 7C), which is in line with the increased proliferative capacity in the absence of FOXP3 and confirms high purity of the FOXP3^KO T_regs. Of note, analysis of CD25 and CD127 expression is not reliable in T_reg cultures due to IL-2 media supplementation. To assess T_eff-like function in FOXP3^KO T_regs, we collected the supernatant after the 3-day expansion and analyzed the production of cytokines using a multiplex Luminex assay. Data were normalized to the control:FOXP3^KO cell ratio at the end of the experiment, taking into account the higher expansion of the FOXP3^KO T_regs. This normalization may favor the control T_regs. We observed increased production of IL-13, IL-17, TNF-α and granulocyte-macrophage colony-stimulating factor (GM-CSF) by the FOXP3^KO T_regs (Fig. 7D and fig. S5), suggesting that FOXP3-deficient T_regs can confer T_eff-like function, better expand in response to TCR stimulation, and produce Th2 and Th17 cytokines. To exclude non-specific effects of AAV6 or GuideRNA/Cas9, we repeated the experiments with control T_regs and T_regs treated with AAV6 and scramble GuideRNA/Cas9. The data indicate that the above-described differences between FOXP3^KO T_regs and control T_regs were specifically caused by loss of FOXP3 expression (fig. S6). Next, we examined ex vivo stimulated T_regs and T_effs from patients with IPEX and HDs for the production of IL-13, IL-17, TNF-α, and GM-CSF, which we found to be controlled by FOXP3. Unlike HDs, phenotypic FOXP3⁺CD25⁺CD127⁻ T_regs from patients with IPEX produced GM-CSF and IL-13 (Fig. 7E and F and fig. S7). These data indicate that, in contrast to HDs, even the phenotypic T_regs from patients with IPEX can acquire a T_eff phenotype to some extent and participate in the inflammation and polarization of immune response towards a Th2 phenotype.

Figure 7. FOXP3^KO T_regs exhibit greater expansion and increased production of proinflammatory cytokines in response to TCR stimulation in vitro, and TCR stimulated T_regs from patients with IPEX exhibit increased proinflammatory cytokine production ex vivo.

(A) Representative FACS plot (left), quantification of FOXP3 expression (middle), and quantification of NGFR expression (right) in FOXP3^KO and control T_regs from 11 HDs. (B) T_reg expansion upon CD3/CD28 stimulation. Cells were counted 3 days post anti-CD3 and anti-CD28 stimulation. The data are plotted as a fold increase of the cell number from day 0. Lines connect control and FOXP3^KO T_regs from the same donor. (C) Frequency of TSDR-demethylated cells in control and FOXP3^KO T_regs. (D) Analysis of cytokine production from supernatants of control and FOXP3^KO T_regs using a Luminex-based multiplexed assay. Only significantly upregulated or downregulated cytokines are shown. The cytokine concentrations were normalized to the ratio of control:FOXP3^KO T_reg numbers at the time of supernatant collection. (E) FACS analysis of proinflammatory cytokine production by T_regs from patients with IPEX and HDs. The representative gating strategy shows Brefeldin A- and Monensin-treated PBMCs from patients with IPEX and HDs unstimulated or stimulated with PMA and ionomycin for six hours. (F) Quantification of IL-13 and GM-CSF production by stimulated or unstimulated T_regs and T_effs from five HDs and five patients with IPEX. Representative gating strategy is shown in (E). Data are presented as mean ± SEM. Data in (A to D) were analyzed using the Wilcoxon matched-pairs test. Statistical significance in (F) was evaluated using the Mann-Whitney test. *p<0.05, **p<0.01, ***p<0.001.

T_reg subpopulations isolated from patients with IPEX were characterized after TCR stimulation using single-cell multi-omic profiling.

We performed the RNA, protein, and TCR profiling in CD4⁺ T cells from the same samples as in Fig. 2, but we first stimulated the cells for 6 hours with anti-CD3 and anti-CD28 antibody-coated beads. We obtained data for 16,458 T cells from HD and 18,018 T cells from patients with IPEX (Fig. 8A). Once again, cells were clustered and visualized using UMAP according to their transcriptomics (Fig. 8B). Clusters were annotated based on mRNA expression of key genes (Fig. 8C), and CD45RA protein expression (based on ADT) (fig. S8A). We identified the following populations: Th1/Cytotoxic (TBX21⁺, GZMB⁺, GZMK⁺, GZMA⁺), Th2 (GATA3⁺), Th1/Th17 (RORC⁺, TBX21⁺), naïve (CD45RA⁺ [ADT]), memory (CD45RA⁻ [ADT]), T_reg naïve (FOXP3⁺, IKZF2⁺, CD45RA⁺ [ADT]), T_reg memory 1 (FOXP3⁺, IKZF2⁺), T_reg memory 2 (FOXP3⁺, IKZF2⁺), and T_reg/T_eff population (FOXP3^low, IKZF2^low) (Fig. 8B and C, fig. S8A). Similarly, to ex vivo sequenced cells, the TCR-stimulated T_reg memory 2 cells were specific to patients with IPEX (Fig. 8D). To confirm that this stimulated population was analogous to the T_reg memory 2 population in our ex vivo experiment, we analyzed the TCR overlap between ex vivo and stimulated populations (Fig. 8E). Indeed, we detected the highest TCR overlap between TCR-stimulated T_reg memory 1 and ex vivo T_reg memory 1 as well as between TCR-stimulated T_reg memory 2 and ex vivo T_reg memory 2 populations, thereby demonstrating the correspondence between the ex vivo and TCR stimulated populations. (Fig. 8E). Consistent with shared clonality between the T_reg memory 1 and T_reg memory 2 in the ex vivo analysis, we observed shared clonality between TCR-stimulated and ex vivo T_reg memory 1 and T_reg memory 2, and vice versa (Fig. 8E). For the TCR overlap between non-T_reg populations, see fig. S8B. Differential expression analyses between T_reg memory 1 and 2 showed downregulation of IL2RA, FOXP3 and CTLA4, and upregulation of GATA3, suggesting preferential polarization of the atypical T_regs to Th2-like phenotype (Fig. 8F, data file S3). Pathway analyses showed an increased expression of cMYC target genes in T_reg memory 2 (Fig. 8G and fig. S8C). cMYC is known to be upregulated upon TCR stimulation in T cells(34, 35), thus the relative increase in expression of cMYC target genes in T_reg memory 2 over T_reg memory 1 may suggest higher responsiveness to TCR stimulation, which may potentially explain the superior in vitro expansion of FOXP3^KO T_regs compared with FOXP3-competent T_regs (Fig. 7B). Consistent with the in vitro FOXP3^KO T_reg and ex vivo T_reg data from patients with IPEX, we observed IL13 expressing T_regs in both T_reg memory 1 and 2 compartments of patients with IPEX, and only a very limited number of IL13 expressing T_regs in HDs (Fig. 8H and I, fig. S8D). Importantly, T_regs from patients with IPEX accounted for 7 to 20% of all IL13 expressing CD4⁺ T cells, suggesting a substantial contribution of patients’ T_reg to the IL-13 producing cell pool (Fig. 8J). We did not observe an increase in the production of other proinflammatory cytokines from our single-cell multi-omics experiments (fig. S8D and E); however, these data may be partially ascribed to differences in cytokine kinetics or limitations to sequencing approaches.

Figure 8. Characterization of T_reg subpopulations post-TCR stimulation using single-cell multi-omic profiling revealed increased IL13 expression in T_reg populations from patients with IPEX.

(A) Number of cells identified per donor. (B) UMAP projection of the T cell populations identified in HDs (left) and patients with IPEX (right). (C) Expression of selected genes in the populations. Color intensity indicates the amount of expression and the size of the dot indicates the percentage of positive cells. (D) Distribution of CD4⁺ T cells from HD and patients with IPEX in the identified populations. Frequency of each population is defined as a percentage of cells present in the indicated population from the total number of cells in a given sample. The same color code as in (B) and (C) indicates the population. (E) Morisita overlaps of TCR repertoires between TCR-stimulated populations: T_reg naïve, T_reg memory 1, T_reg memory 2, and T_reg/T_eff populations, as well as all ex vivo populations identified in Fig. 2. (F) Differential expression analyses between T_reg memory 2 and 1. The relevant genes are indicated by a label. The genes with positive fold change (FC) are upregulated in T_reg memory 2 compared with T_reg memory 1. HLA-DRB1 is not shown (p-value = 3.9E⁻¹⁶², log₂ fold change = −2.6). n.s., not significant. (G) cMYC target genes were identified as the most significantly differentially expressed gene set by FGSEA comparing T_reg memory 2 and T_reg memory 1 populations. Genes were ranked based on fold changes; the top panel shows the enrichment score and the bottom panel the fold change of the genes. (H) Feature dot plot shows expression of IL13 in T_reg populations in patients with IPEX and HDs. (I) Quantification of IL13-expressing cells in patients with IPEX and HDs. The frequency on x axis is defined as in (D) and all other populations are presented as “Non-T_reg cells”. (J) % of IL13-expressing cells from the T_reg populations from all IL13.

To provide further evidence of TCR autoreactivity in T_reg and T_eff populations from patients with IPEX, we analyzed the presence of autoreactivity promoting and limiting AA doublets from the scRNA and TCR sequencing data. To obtain enough CDR3ß sequences for the analyses, we pooled the data from ex vivo and TCR-stimulated cells, and from HDs and patients with IPEX in T_reg populations. We observed that the T_reg memory 2 compartment was enriched in the frequency of autoreactivity promoting AA doublets and depleted of autoreactivity limiting AA doublets (fig. S8F), again supporting that the T_reg memory 2 population, composted in majority by FOXP3⁺ TSDR-demethylated CD25^low/high cells, is enriched for autoreactive cells.

DISCUSSION

In the present work, we elucidated two concomitant and potentially related phenomena contributing to the etiology of IPEX, a distinctive example of monogenic autoimmunity and loss of immune regulation. First, CD4⁺ T cells from patients with IPEX with TSDR demethylation, the epigenetic lineage marker of T_regs, were present in the T_eff compartment. Second, the TCR repertoire of these patients was enriched in autoreactive sequences, resembling those of T_regs.

The TSDR demethylated cells in the CD4⁺ T_eff compartment of patients with IPEX comprised both FOXP3⁺ and FOXP3⁻ cells. Since FOXP3⁺ T_effs from patients with IPEX were exclusively TSDR demethylated, we were able to identify and study them. FOXP3⁺ T_effs displayed a memory phenotype and had reduced expression of molecules linked to suppressive function, such as FOXP3 and CTLA4, compared with phenotypic T_regs. Consistent with their memory phenotype, they expressed high abundance of CD39 and TIGIT (36). Using a combination of mRNA, protein, and TCR profiling, we provided evidence that these FOXP3⁺ T_effs correspond to an IPEX-specific, transcriptionally distinct, T_reg population that we called T_reg memory 2. The mRNA analyses of the T_reg memory 2 population revealed an upregulation of IL7R and ZPF36L2, a negative regulator of IKZF2 expression (27), as well as an enrichment of TNF-α signaling target genes. Yet, the TCR repertoire of the T_reg memory 2 population was the most similar to that of classical memory T_regs (T_reg memory 1). Hence, the FOXP3⁺ TSDR demethylated cells present in the T_eff compartment of patients with IPEX are memory T_regs with altered expression of key T_reg molecules, with a gene expression signature indicative of exposure to an inflammatory environment, and a TCR repertoire clonally related with prototypical T_regs that are potentially their precursors. We, therefore, propose that these TSDR-demethylated cells, which expanded beyond the CD25^highCD127^low phenotypical T_reg compartment, are unstable T_regs, which we consider loss-of-identity T_regs. Although we cannot exclude the possibility that some T_regs exit the thymus with loss-of-identity T_reg phenotype, the evidence presented above suggest otherwise. These data provide important insights into the heterogeneity of FOXP3-deficient T_regs observed by Zemmour et al. (37) They reported two populations of T_regs in FOXP3-deficient patients, one of which was only present in the disease. Based on our TSDR data, TCR data, and the simultaneous profiling of T_reg protein and transcriptional markers, we demonstrated that FOXP3 deficiency gives rise to two different subsets of memory T_regs, one of which loses its phenotypical characteristics.

The presence of loss-of-identity T_regs can well explain the overall increased proportion of TSDR-demethylated cells in whole blood of patients with IPEX, recently deemed diagnostically relevant and associated with disease activity (12, 13), as well as its discrepancy with the frequency of T_regs as determined by FACS (13–17). The absence of the loss-of-identity T_regs not only in carrier mothers and healthy donors, but also in patients with APECED, supports dysfunctional mutated FOXP3 as a primary cause of T_reg instability. Therefore, the loss-of-identity T_regs might represent a human counterpart of the “ex” and unstable T_regs described in mice (18–24). However, although loss-of-identity T_regs in patients with IPEX correlated with an increased T cell autoreactivity, the latter occurred in patients with APECED independently from the T_reg expansion, likely caused by direct impaired thymic deletion of autoreactive T cells (13). Importantly, we provide evidence that the presence of functional T_regs in vivo can prevent or control the accumulation or the expansion of the loss-of-identity T_regs, as observed in carrier mothers or in the transplanted patient with IPEX in which functional donor-derived T_regs had a selective advantage.

The phenotype analyses of the loss-of-identity T_regs showed reduced expression or even complete absence of crucial suppressive molecules such as FOXP3, CD25, and CTLA4. As a consequence, these cells likely lose regulatory functions. However, they are still derived from the self-antigen specific selected T_regs, as indicated by the TCR analysis, and are therefore likely to play a pathogenic role in patients with IPEX. In agreement with this scenario, the T_eff2 compartment of patients with IPEX, which was composed mainly of bona fide T_effs, had an increased frequency of autoreactivity promoting AA doublets, as well as a reduced frequency of autoreactivity limiting AA doublets, thus indicating expansion of autoreactive bona fide T_effs in the absence of functional T_regs. Considering that the increase in the frequency of autoreactivity promoting AA doublets in T_eff2 correlated with the frequency of TSDR-demethylated cells in T_eff1, we speculate that the autoreactive T_effs expand freely because of the lack of functional T_regs in patients with FOXP3 mutations (15). Consistent with this hypothesis, IPEX carrier mothers had a normal TCR repertoire, indicating that the expansion of autoreactive T_effs in patients with IPEX is largely caused by a lack of FOXP3-dependent T_reg function.

A clinically relevant question is to what extent the unstable T_regs are capable of effector functions. We showed that, upon TCR-mediated activation, FOXP3^KO T_regs from HDs upregulated the production of the proinflammatory cytokine TNF-α and the Th2 polarizing cytokine IL-13, both of which we found among the most upregulated cytokines in plasma from patients with IPEX(13). Indeed, and in contrast to HDs, even the CD25^highCD127^low T_regs from patients with IPEX produced IL-13. Similarly, we detected IL-13 production by TCR-stimulated T_reg memory 2 cells using scRNA-seq. These findings are in line with data from a murine model where IPEX-like disease was caused by attenuated FOXP3 expression; in this model, the T_regs with low FOXP3 expression acquired a Th2 T_eff-like phenotype (23). As expected, we also observed increased production of IL-17 by FOXP3^KO T_regs, which reflects the T_reg/Th17 imbalance in the absence of functional FOXP3 (38–40). However, we did not observe IL-17 production in ex vivo experiments, possibly because T_regs isolated from peripheral blood of patients with IPEX and of HDs are preferentially Th2 polarized as indicated by GATA3 expression in the scRNA-seq. Unlike in Lam et al., we did not observe a broad upregulation of cytokine secretion in FOXP3^KO T_regs but rather an increase in specific cytokines, including GM-CSF (41). Interestingly, we found GM-CSF to be produced also by phenotypic CD25^highCD127^low T_regs from patients with IPEX. GM-CSF produced by T cells can act in a paracrine fashion to induce phagocyte activation and production of proinflammatory cytokines and chemokines such as IL-1β, TNF-α, CCL17, and CCL22, a mechanism associated with the pathogenesis of T cell mediated autoimmune diseases, such as rheumatoid arthritis (RA) and multiple sclerosis (MS) (42–45). Moreover, we observed increased plasma concentrations of cytokines and chemokines associated with dysregulation of the phagocytic compartment in patients with IPEX (13), including TNF-α, CCL17, and CCL22. Although further work is necessary to fully unravel the role of GM-CSF in the pathogenesis of IPEX disease, GM-CSF neutralizing antibodies might be an attractive treatment option, as they are currently in clinical trials for RA and MS (42). Overall, our in vitro and ex vivo data indicated that, upon activation, FOXP3 fully- or partially- deficient T_regs upregulate the expression of effector cytokines, specifically those that we described being increased in patients with IPEX. These data might provide at least a partial explanation for the observed Th2 polarization in patients with IPEX and their Th2- and Th17-associated pathologies (such as eczema and enteropathy, respectively).

As IPEX is a rare disease, experimental studies are limited by the amount of material per patient and the number of patients available. In addition, shipment of the collected samples made some of them less suitable for certain types of experiments, such as ex vivo scRNA-seq. Our study missed assessments of multiple time points during the course of the disease prior and after immunosuppressive treatment. Such data could provide a more definitive picture of the temporal dynamic of the unstable T_regs and autoreactive T_effs, which could be associated to the clinical manifestations. Similarly, tissue samples, including from the thymus and target organs, could provide additional information about the development, homing, and function of the unstable T_regs and autoreactive T_effs in patients with IPEX. Finally, T_reg instability analyzed in a larger number of patients would likely allow us to dissect the mutation-specific effects.

Collectively, these data contribute to the basic knowledge about the role of FOXP3 as a gatekeeper of T_reg identity and quiescence as well as in preventing autoreactive T_eff expansion in humans. In addition, these data are of interest from a translational perspective, such as in the context of T_reg-based therapies, not only in patients with IPEX, but also in patients with other autoimmune diseases in which the loss-of-identity T_regs and expansion of autoreactive T cells are still poorly defined or where similar T_reg instability has been suggested (46, 47).

MATERIALS AND METHODS

Study Design

The research objective was to determine the presence and the distribution of autoreactive T cells in patients with IPEX. More specifically, we aimed to unravel how FOXP3 deficiency affects the T cell repertoire autoreactivity and T_reg stability in vivo. The data obtained in IPEX patients were compared with those in APECED patients and HD. The inclusion criteria for patients with IPEX consisted of detected FOXP3 mutation by gene sequencing and the presence of clinical symptoms or signs of autoimmunity. The inclusion criteria for patients with APECED consisted of detected homozygous AIRE mutation by gene sequencing. In patient 63, the AIRE mutation was confirmed with certainty only in one allele, but the patient clinically presented with APECED manifestations including autoimmune adrenal insufficiency, autoimmune hypoparathyroidism, and hypothyroidism, and thus was included in the study. Patient and healthy donor details are summarized in table S1. Patients’ and HDs’ samples were collected upon informed consent on an institutional review board approved protocol (IRB #34131) of the Center for Genetic Immune Diseases, Stanford, CA. In addition, the samples from patients with APECED were collected as described previously (9) and under an institutional review board approved protocol (#0906005336). HDs were sex-matched siblings, parents, and pediatric healthy individuals. In addition, samples from adult HDs were purchased from Stanford Blood Center. For TSDR-demethylation analysis in patients with APECED, parents of patients with APECED were used as a control if available (4 out of 5 samples, non-sex matched), and for TCR analyses, sex-matched HDs were used as controls.

FACS staining, sorting, and analysis

Ficoll-Paque PLUS media (Fisher Scientific)-isolated peripheral blood mononuclear cells (PBMCs) from whole blood obtained from a standard blood draw, buffy coat, or Leukoreduction System Chambers were stained for 30 minutes on ice, washed with phosphate-buffered saline (PBS), analyzed, and sorted using BD FACSAria II Cell Sorter (BD Biosciences). The high cell purities after the cell sorting were confirmed by reanalysis of the sorted samples (when enough cells were available) and by TSDR demethylation analysis. Staining of intracellular antigens was done using Foxp3 / Transcription Factor Staining Buffer Set (eBioscience) according to manufacturer recommendations. FACS data were analysed using FlowJo software (FlowJo LLC). If multiple brilliant violet or ultraviolet dyes were used, cells were stained in Brilliant stain buffer or brilliant staining buffer plus (BD Biosciences). Antibodies used in this study are summarized in table S2.

TSDR demethylation analyses

Sorted CD14⁻CD3⁺CD4⁺CD25⁻CD127^low T_eff1 and CD14⁻CD3⁺CD4⁺CD25^-/dimCD127^high T_eff2 pellets were frozen in PBS and CD14⁻CD3⁺CD4⁺CD25^highCD127^low T_regs were expanded using Dynabeads Human T-Activator CD3/CD28 (1:1 beads:T_reg ratio, Gibco), and cultured in X-VIVO 15 with Gentamicin, L-Gln, Phenol Red (Lonza), supplemented with 5% human serum (Millipore Sigma) and 300 IU/ml IL-2 (Peprotech) for 5 to 10 days to obtain a sufficient number of cells for TSDR demethylation. TSDR demethylation was analyzed using qPCR as described previously (26). Briefly, cell pellets were thawed and DNA was isolated, bisulfite converted, and analyzed using qPCR. Data were analyzed by Epimune GmbH. TSDR-demethylation data obtained from the whole blood samples of transplanted patients were normalized to the number of CD4⁺ T cells, also determined by epigenetic analysis (13). The % of cells with demethylated TSDR is defined as follows: (TSDR demethylated copies)/((GAPDH copies)/2/100), and thus was independent of sex.

Single cell RNA, protein, and TCR profiling sample preparation and analysis

PBMCs were thawed and CD4⁺ T cells were isolated using a CD4+ T Cell Isolation Kit, human (Miltenyi Biotec). High CD4+ T cell purity was confirmed by FACS (fig. S8G). For TCR stimulation, 5e⁵ T cells (except one donor where only 3.3e⁵ where available) were plated in 200 μl of X-VIVO15 with Gentamicin, L-Gln, Phenol Red (Lonza), supplemented with 5% human serum (Millipore Sigma), and stimulated for 6 hours with Dynabeads Human T-Activator CD3/CD28 (1:1 beads:cell ratio, Gibco). Cells and beads were centrifuged 300g for 2 min at the start of the activation. Both stimulated and ex vivo cells were stained with oligo-nucleotide tag-conjugated CD127, CD25, and CD45RA antibodies; and cells from patients with IPEX were stained with hashtag antibody 1 and HD cells with hashtag antibody 2 (for details about the antibodies, see table S2). The antibody premix was centrifuged at 14e³ g for 8 min and staining was performed in PBS with 1% bovine serum albumin (BSA; Miltenyi Biotec) for 30 min on ice. Samples were washed three times, cell concentration was adjusted to 1e6 per ml, and samples from HDs and patients with IPEX were mixed in pairs (always HD and IPEX sample) right before the mixed samples were loaded onto a chromium controller (10x Genomics). Super loading was performed to target 20,000 cells for each pair of samples (48).

Preparation of transcriptome, repertoire (TCR), and protein (ADT and hashtag) cDNA libraries was performed according to the 10x Genomics protocol: “Chromium Next GEM Single Cell 5’ Reagent Kits v2” (CG000330 Rev D). cDNA libraries were sequenced at high-depth using the NovaSeq X Plus Platform. The 10x Genomics Cloud Analysis platform was used to process and align raw reads to the human genome (GRCh38) and to generate count matrices and TCR annotations for single cells.

The analyses were performed in R using standard Seurat workflow (49). Briefly, low-quality cells were removed based on low read counts and high mitochondrial DNA content, RNA and ADT data were normalized and de-hashtagged, and doublets were removed based on hashtag staining. Contaminating CD79A and CD14 double positive and CD3E and CD247 double negative clusters and cells were removed from analyses. Data were scaled, TCR genes were removed from variable genes used for principal component analysis using Trex package (50), and data were integrated using the harmony package (51). Ex vivo data were analyzed separately, and TCR-stimulated data were analyzed together with ex vivo data and separated post-scaling. TCR repertoires were analyzed using scRepertoire package and, for the TCR overlap analyses, the AA sequence of TCRα and β chains were used (52). CD25 and CD127 gating was performed in pairs in which the libraries were prepared, where each pair always contained HD and IPEX samples. Differential gene expression analyses between the populations were performed using the FindMarkers function (Seurat) and the results were plotted using EnhanceVulcano package (53). Pathway analyses were performed using fgsea package; all genes were used for the analyses and their differential expression was ranked based on fold changes (54). In addition, scCustomize package and ColorBrewer were used for visualization (55, 56).

TCR repertoire analyses

Sorted cell pellets were frozen in PBS. Upon thawing, DNA was isolated using DNeasy Blood and Tissue Kit (Qiagen). The TCRB was sequenced by Adaptive biotechnologies. Data were processed by Adaptive and initial analysis was done using immunoSEQ analyzer online tool (Adaptive biotechnologies). The analysis of the frequency of autoreactivity promoting and limiting doublets was performed as described previously (29). Briefly, CDR3ß sequences shorter than 8 amino acids were removed because of the conserved amino acid residues at the end of the CDR3ß loop. The frequency of autoreactivity promoting and limiting doublets defined by B. D. Stadinski et al., were calculated as a sum of frequencies of productive rearrangements of a given set of doublets (3, 29). TCRB sequences are available online at (https://clients.adaptivebiotech.com/). The previously published TCR data from patients with APECED (9) and respective HDs are available under immuneACCESS DOI https://doi.org/10.21417/B7V926 (https://clients.adaptivebiotech.com/pub/sng-2019-sciimmunol).

FOXP3 knock out in primary T_regs

To isolate CD3⁺CD4⁺CD25^highCD127⁻ T_regs, we used PBMCs from Leukoreduction System (LRS) chambers and enriched T_regs using the CD4⁺CD25⁺CD127^dim/− Regulatory T Cell Isolation Kit II, human (Miltenyi Biotec) followed by FACS sorting of the CD4⁺CD25⁺CD127^dim/− cells to obtain a pure population. T_regs were stimulated for 7 to 8 days with Dynabeads Human T-Activator CD3/CD28 (1:1 beads:T_reg cell ratio, Gibco), and cultured in X-VIVO 15 with Gentamicin, L-Gln, and Phenol Red (Lonza), supplemented with 5% human serum (Millipore Sigma) and 300 IU/mL IL-2 (Peprotech). We subsequently edited the cells using Cas9 (Integrated DNA Technologies), chemically modified guide RNA (Synthego corporation), and AAV6 virus (Signagen laboratories, and in house produced)-mediated delivery of homologous template containing NGFR reporter gene under control of PGK promoter as described previously (33). Control T_regs were treated the same as FOXP3^KO T_regs but were electroporated in the absence of CRISPR/Cas9 and not treated with AAV6. Scramble T_regs were nucleofected in the presence of ScrambleGuide/Cas9 (Negative Control Scramble #1, mod, Synthego) and treated with the same batch of the AAV6 virus used to generate the FOXP3^KO T_regs. Cells were expanded for 7 to 10 days and NGFR⁺ cells were purified using MACSelect LNGFR MicroBeads and autoMACS Pro Separator (Miltenyi Biotec). Two donors were expanded for an additional 9 days to obtain sufficient cell numbers for analyses. Subsequently, 1×10⁵ cells were plated in technical duplicates or triplicates, stimulated with Dynabeads Human T-Activator CD3/CD28 (1:1 beads:T_reg cell ratio, Gibco), and analyzed for cell count, FOXP3 expression (in Fig. 7, FOXP3⁺ cells were determined as double positive for two antibody clones 259D and 150D; in fig. S7 only clone 259D was used), and cytokine production 3 days post stimulation. Samples with less than 30% of TSDR-demethylated cells were removed from analyses. For the cytokine production, the technical duplicates were averaged and for each donor, the values were normalized to control:FOXP3 KO cell number ratio at the time of supernatant collection with the following formula: (Technical duplicate average)/((FOXP3 KO)/control [T_reg number])).

In the case of control and scramble T_regs, the averaged cytokine concentrations were not normalized to cell number ratio because scramble and control did not show any difference in proliferation.

Cytokine production analyses

For cytokine production by FOXP3^KO T_regs, supernatants were collected 3 days post stimulation as described above. Cytokines were measured using a Luminex-based multiplexed assay (Th1/2/9/17 18-plex Human ProcartaPlex Panel, Thermo Fisher) per manufacturer recommendation. For cytokine production by ex vivo stimulated PBMCs from patients with IPEX and HDs, cells were thawed and 2 ×10⁶ cells per ml were incubated in X-VIVO 15 with Gentamicin, L-Gln, and Phenol Red (Lonza), supplemented with 5% human serum (Millipore Sigma), Brefeldin A (BioLegend) and Monensin (BioLegend) for 6 hours. Cells were left untreated or treated with PMA and ionomycin (Cell Activation Cocktail (without Brefeldin A), BioLegend). Next, cells were washed and stained as indicated in the FACS staining, sorting, and analysis section above.

Statistical analysis

All raw, individual-level data for experiments where n<20 are presented in data file S4. Statistical analyses were performed using GraphPad Prism software version 9 (GraphPad Software, Inc.) or R studio software packages as indicated in methods. Normal distribution was not assumed, and the number of observations did not allow for proper testing of normality, therefore, non-parametric two-tailed test Mann-Whitney for non-paired values and non-parametric two-tailed Wilcoxon test for paired values were used. In addition, two-tailed Spearman correlation and Dunn’s multiple comparisons test were used to evaluate significance. Adjustments to alpha levels for multiple comparisons were made using Bonferroni correction for differential gene expression analyses between the clusters (adjusted p values are reported in data file S1 to S3) and using Benjamini-Hochberg method for the FGSEA. Statistical tests are also indicated in the figure legend. Significance is indicated as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. N represents the number of donors or values per group.

Data availability:

All data associated with this study are in the paper or supplementary materials. In addition, TCR sequencing data are deposited at https://clients.adaptivebiotech.com/, and single-cell multiomics data are available at GEO under record GSE247274.