CD38 endows local antigen-specific Treg cells with stress resilience for control of compartmentalized CNS inflammation

Chen, Hsin-Hsiang; Tyystjärvi, Sofia; Ruiz Navarro, Diego; Kant, Ravi; Groll, Tanja; Wagner, Ingrid; Domínguez Moreno, Helena; Bonafonte-Pardàs, Irene; Öllinger, Rupert; Afzali, Ali Maisam; Heink, Sylvia; Richter, Lisa Charlotte; Sie, Christopher; Lepennetier, Gildas; Seeholzer, Lea; Steiger, Katja; Merkler, Doron; Rad, Roland; Schotta, Gunnar; Schubert, Benjamin; Muschaweckh, Andreas; Korn, Thomas

doi:10.1038/s41590-025-02416-z

Download PDF

Article
Open access
Published: 29 January 2026

CD38 endows local antigen-specific T_reg cells with stress resilience for control of compartmentalized CNS inflammation

Nature Immunology volume 27, pages 516–529 (2026)Cite this article

8372 Accesses
1 Citations
10 Altmetric
Metrics details

Subjects

Abstract

Foxp3-expressing regulatory T (T_reg) cells protect against systemic autoimmunity. However, little is known about the significance of T_reg cells in inflammation-experienced tissues. Here, we use an experimental autoimmune encephalomyelitis model and show that T_reg cells accumulate and persist in the central nervous system (CNS) long after the resolution of the bulk of the inflammatory infiltrate. CNS-specific depletion of postinflammatory T_reg cells, but not systemic depletion of T_reg cells, results in autoimmune inflammatory flares in the CNS by residual local effector T cells. Expression of the NAD-consuming ectoenzyme CD38 is crucial for the functional adaptation of postinflammatory CNS T_reg cells to a stressful microenvironment, in which access to interleukin-2 (IL-2) is limited. CD38 counteracts ADP-ribosylation of the IL-2 receptor and thus maintains its high sensitivity to IL-2. This fully functional high-affinity IL-2 receptor prevents the loss of tissue-resident antigen-specific T_reg cells. These ‘stress-tolerant’ CNS T_reg cells impede the collapse of immune homeostasis in the CNS once acute inflammation is controlled.

Regulatory T cells function in established systemic inflammation and reverse fatal autoimmunity

Article 23 August 2021

Clonal CD8⁺ T cells populate the leptomeninges and coordinate with immune cells in human degenerative brain diseases

Article Open access 27 January 2026

T cell activation and deficits in T regulatory cells are associated with major depressive disorder and severity of depression

Article Open access 16 May 2024

Main

Foxp3⁺ regulatory T (T_reg) cells have regained attention due to their ability to adapt to tissue niches and exert noncanonical functions in tissue homeostasis and regeneration^1,2. Although the developmental trajectory of tissue-resident T_reg cells is consistent with a thymic origin of these cells in most tissues, peripherally induced T_reg cells might be prevalent in some barrier-associated tissues with a high load of commensal microbiota³. A widely accepted model of tissue T_reg cell development suggests a multistep process, involving the imprinting of a tissue T_reg cell precursor subset in secondary lymphoid tissue and the terminal instruction to tissue T_reg cells by the specific milieu in the target tissue^4,5. The dwell time of T_reg cells in nonlymphoid tissues may vary depending on the tissue context⁶, and egress of T_reg cells from these tissues, rejoining the circulation, may occur⁷. However, the description of T_reg cell population dynamics and functional phenotype has mostly been performed in the steady state^8,9. It is unclear how T_reg cells behave in nonlymphoid tissues with low-grade chronic inflammation, as is observed in most autoimmune diseases, including multiple sclerosis.

It is well established that T_reg cells are essential in re-establishing immune homeostasis in any adaptive immune response, where IL-2-driven feedforward loops lead to the expansion of effector T cells¹⁰. Once autoimmune inflammation in the central nervous system (CNS) has been initiated, thymus-derived T_reg cells are recruited to the CNS¹¹. In the target tissue, they adopt an effector phenotype, which is steered by a network of transcription factors, including BLIMP1 and BATF^12,13,14. This transcriptional network also secures the preservation of T_reg cell identity in the face of inflammation. Because highly activated effector T cells are resilient to T_reg cell-mediated control¹¹, it has been suggested that conventional T cell-intrinsic down-modulatory mechanisms must be initiated before efficient T_reg cell-mediated suppression can take effect.

T_reg cells that accumulate in the CNS in autoimmune inflammatory diseases (but also after stroke, in traumatic brain injury and perhaps even neurodegenerative diseases) were suggested to be involved in tissue repair and regulation of organ function by a series of different mechanisms^{15,16,17,18,19,20}. The relevance of Foxp3⁺ T_reg cells for these processes has been demonstrated in loss- and gain-of-function approaches. However, whether T_reg cells that accumulate in the CNS after the peak of experimental autoimmune encephalomyelitis (EAE; a paradigmatic model for autoimmune CNS disease) actually contribute to the re-establishment and subsequent maintenance of immune homeostasis in the CNS is not known. The postinflammatory CNS is a hostile environment for T_reg cells, as it is low in interleukin-2 (IL-2) due to the contraction of the conventional T (T_conv) cell compartment and simultaneously holds high levels of nicotinamide adenine dinucleotide (NAD⁺)²¹. High levels of extracellular NAD⁺ have been shown to induce apoptosis in T_reg cells, in part by facilitating the ADP-ribosylation of the ATP-gated ion channel P2RX₇ (ref. ²²), which induces a gain-of-function of this molecule with a downstream increase in intracellular Ca²⁺ (ref. ²³). Nevertheless, Foxp3⁺ T_reg cells constitute a stable population in the postinflammatory CNS.

Here, we show that the NAD⁺ consuming enzyme CD38 controls stress resilience of CNS-resident T_reg cells by preventing ADP-ribosylation and subsequent loss of function of the IL-2 receptor α-subunit (IL-2Rα; CD25). This function enables CNS T_reg cells to survive despite a limited amount of IL-2 in the microenvironment and to exert compartmentalized immune control in the inflammation-experienced CNS.

Results

T_reg cells persist in the CNS after recovery from an autoimmune response

To investigate the population dynamics and function of T_reg cells in the CNS during recovery from autoimmune inflammation, we used the MOG_35–55-induced EAE model in Foxp3 (green fluorescent protein (GFP)) reporter mice²⁴. After an acute monophasic inflammatory episode, MOG_35–55-induced EAE is characterized by a prolonged period of recovery, during which clinical symptoms improve and stabilize but usually never fully resolve (Fig. 1a). Although the neurological sequelae in this model are due to incomplete recovery from axonal damage caused during the acute phase of the disease, the inflammatory infiltrate in the CNS in the recovery phase is largely reduced. Here, we focused on this ‘re-established’ homeostatic condition in recovery and interrogated the CNS T_reg cell compartment.

Fig. 1: Treg cells persist in the CNS in chronic EAE. — **Fig. 1: T_reg cells persist in the CNS in chronic EAE.**

Although T_conv cell numbers declined significantly between the peak and early recovery phase (Fig. 1b,c), T_reg cell numbers remained largely stable during this time frame in both the brain and spinal cord (Fig. 1b,d). Consequently, the local frequency of T_reg cells within the CNS CD4⁺ T cell compartment increased significantly in the postinflammatory CNS (Fig. 1d). Notably, T_reg cells persisting in the postinflammatory CNS were not evenly distributed but were located in certain niches around the third ventricle and in the meningeal space (Extended Data Fig. 1).

To gauge the contribution of the systemic immune compartment to the CNS T_reg cell pool, we labeled mitoDendra2-expressing T cells in the draining lymph nodes of MOG-immunized mice at the peak of EAE and during recovery (day 30) and tracked photoconverted T cells to the spleen and CNS (Fig. 1e). Although CD4⁺ T cells still trafficked to the CNS during peak disease, the recruitment of T cells from the initial priming site to the CNS significantly dropped during recovery (Fig. 1f) and was negligible for T_reg cells (Fig. 1g). Conversely, the fraction of proliferating CNS T_reg cells was about 10% at the peak of EAE and dropped to 3% in the postinflammatory CNS (Fig. 1h,i). Together, these data suggest that the T_reg cell pool in the postinflammatory CNS was essentially disconnected from its priming site in the systemic immune compartment and, to some extent, had self-renewing capacity.

To investigate the stability of postinflammatory T_reg cells in their CNS environment, we used Foxp3^Cre-ERT2 × Rosa26^LSL-tdTomato mice to perform fate mapping of Foxp3⁺ T_reg cells. Tamoxifen was administered at EAE recovery (Fig. 1j). This allowed us to categorize T_reg cells into stable (Foxp3 (GFP)⁺tdTomato⁺) and unstable (Foxp3 (GFP)⁻tdTomato⁺) populations. We termed those T_reg cells that persisted as stable Foxp3 expressors in inflammation-experienced tissues ‘stress-tolerant’ T_reg cells. Notably, most CNS T_reg cells maintained stable Foxp3 expression in the recovery phase, with only 15 to 20% of CNS T_reg cells losing Foxp3 expression (Fig. 1k). To identify potential drivers of instability in recovery CNS T_reg cells, RNA-sequencing (RNA-seq) analysis was performed in stable and unstable T_reg cell populations isolated during the recovery phase (Extended Data Fig. 2a,b and Supplementary Tables 1 and 2). Stress-tolerant T_reg cells expressed key immunosuppressive molecules, including Il2rb, Ctla4 and Areg. Conversely, unstable T_reg cells exhibited increased expression of Bach2, a transcription factor known to reduce Foxp3 expression²⁵, and Tcf7, which suppresses the transcription of genes co-bound by Foxp3 (ref. ²⁶). These findings suggest that the downregulation of Foxp3 in unstable T_reg cells is linked to the activity of Bach2 and Tcf7. By contrast, the dominant pathway associated with CNS T_reg cell stability in the postinflammatory CNS was the ‘IL-2–STAT5 signaling’ pathway (Fig. 1l and Extended Data Fig. 2c). To confirm our RNA-seq findings in stress-tolerant T_reg cells, we analyzed the expression of IL-2Rα (CD25) in stable and unstable T_reg cells using flow cytometry. Consistent with the transcriptomic data, IL-2Rα expression was significantly lower in unstable than in stable CNS T_reg cells (Fig. 1m), further highlighting the strong association between IL-2 signaling and Foxp3 stability in stress-tolerant T_reg cells.

Stress-tolerant T_reg cells control compartmentalized CNS inflammation

To study the functional relevance of CNS-residing stress-tolerant T_reg cells after resolution of acute inflammation (>day 35), we used Foxp3-DTR mice, which harbor a knock-in of the diphtheria toxin (DTx) receptor (DTR) gene expressed under the control of the endogenous Foxp3 locus, for transient ablation of Foxp3-expressing cells via administration of DTx²⁷. Because a complete depletion of T_reg cells from the CNS at late stages of EAE is challenging to achieve using the conventional systemic (intraperitoneal (i.p.) or intravenous (i.v.)) routes of DTx injection, we chose to administer low doses of DTx (15 ng) directly into the CNS using stereotactic intracerebroventricular (i.c.v.) injection (Fig. 2a). Local DTx delivery resulted in almost complete removal of CNS T_reg cells, while the systemic compartment was only transiently (and partially) depleted of Foxp3⁺ T_reg cells (Fig. 2b,c). Depletion of CNS T_reg cells was accompanied by a rapid worsening of EAE symptoms (Fig. 2d and Extended Data Fig. 3), that is, the return of complete hind limb paralysis. Although some relapses were controlled again after 5 days, most DTx-treated mice succumbed to their relapses (Extended Data Fig. 3b,c). The disease relapses of i.c.v. DTx-injected Foxp3-DTR mice were not due to DTx-mediated CNS toxicity because i.c.v. DTx-injected wild-type control mice did not show any exacerbation of clinical symptoms. By contrast, efficient systemic depletion of T_reg cells in recovered EAE mice by i.p. DTx injection, which largely spared the CNS T_reg cell compartment, failed to trigger an EAE relapse (Fig. 2d).

Fig. 2: Treg cells maintain immune homeostasis in the postinflammatory CNS. — **Fig. 2: T_reg cells maintain immune homeostasis in the postinflammatory CNS.**

Only the near-complete (i.c.v. DTx), but not the partial (i.p. DTx), depletion of CNS T_reg cells led to a marked increase in the amount of activated CD4⁺CD69⁺ T cells in the CNS of Foxp3-DTR mice by day 3 after DTx administration (Fig. 2e,f). In histologic analyses, T cells and myeloid cells primarily expanded at the sites of pre-existing inflammatory infiltrates after i.c.v. injection of DTx into post-EAE Foxp3-DTR mice (Fig. 3a,b). Although some effector cells were recruited from the systemic compartment (Fig. 3c), FTY720 did not prevent the massive activation of those T_conv cells already residing in the CNS (Fig. 3d). Their activation status was similar to the activation of CNS T_conv cells in non-FTY720-treated i.c.v. DTx-injected mice (Fig. 3d). Accordingly, FTY720 treatment failed to prevent EAE relapse following complete local depletion of recovery CNS T_reg cells (Fig. 3e), suggesting that residual CNS T_conv cells are sufficient to drive a relapse following removal of local T_reg cells.

Fig. 3: Loss of Treg cells in the postinflammatory CNS rekindles inflammation. — **Fig. 3: Loss of T_reg cells in the postinflammatory CNS rekindles inflammation.**

In summary, stress-tolerant CNS T_reg cells, but not systemic T_reg cells, are essential for maintaining immune homeostasis in postinflammatory EAE mice.

Stress-tolerant CNS T_reg cells have a tissue-type transcriptome

To determine whether the changing inflammatory milieu within the CNS between the peak and the remission phase of EAE would shape the transcriptional profile of stress-tolerant CNS T_reg cells, we performed RNA-seq of bulk CD4⁺Foxp3 (GFP)⁺ T_reg cells and CD4⁺Foxp3 (GFP)⁻ T_conv cells sorted by fluorescence-activated cell sorting (FACS) from the spleen and CNS on days 15 and 40 after EAE induction.

RNA-seq analysis revealed a total of 311 differentially expressed genes (DEGs; >1.5-fold, adjusted P < 0.05) between CNS T_reg cells derived from EAE recovery and EAE peak, with 159 genes upregulated in recovery CNS T_reg cells and 152 genes upregulated in peak CNS T_reg cells (Extended Data Fig. 4a), consistent with a distinct positioning of CNS T_reg cells along principal component 2 (PC2) according to the disease phase (Fig. 4a). The number of disease-phase conditional DEGs was smaller within splenic T_reg cells, with only 15 upregulated and 6 downregulated genes in recovery splenic T_reg cells versus peak splenic T_reg cells. Compared to their disease-phase-specific splenic counterparts, both recovery and peak CNS T_reg cells showed an upregulation of several genes that have previously been associated with a nonlymphoid tissue T_reg cell phenotype, including Areg, Nfil3, Prdm1, Rora, Itgae, Icos, Il10, Il1rl1 and Il2ra (Extended Data Fig. 4b,c). Although most of the tissue T_reg cell-associated genes did not undergo disease-phase-dependent changes in expression, the ‘TNF signaling via NF-κB’ and ‘IL-2–STAT5 signaling’ pathways were the two top upregulated pathways in recovery CNS T_reg cells compared to peak CNS T_reg cells (Extended Data Fig. 4a–h), suggesting that the optimal exploitation of limited amounts of IL-2 after the contraction of the IL-2-producing effector T cell population (see Fig. 1c) might become the most crucial determinant of their ‘resilience to stress’ in CNS recovery T_reg cells.

Fig. 4: Transcriptome analysis of CNS Treg cells during different phases of EAE. — **Fig. 4: Transcriptome analysis of CNS T_reg cells during different phases of EAE.**

To identify upstream and downstream determinants of the stress-tolerant CNS T_reg cell functional phenotype, we first analyzed genes differentially expressed in recovery CNS versus peak CNS T_reg cells, which encoded cell surface proteins or DNA-binding proteins with transcription factor activity (Fig. 4b,c). We then performed functional enrichment analyses, only selecting those DEGs that were not at the same time significantly regulated in recovery versus peak T_conv cells to define a CNS T_reg cell-exclusive signature of up- and downregulated transcripts in the recovery phase (Fig. 4d and Supplementary Table 3). Functional enrichment analyses with the upregulated genes revealed a few significantly regulated cellular processes, including ‘response to stress’ (Fig. 4e). Notably, the genes that were exclusively upregulated in recovery CNS T_reg cells and at the same time intersected the GO terms ‘response to stress’ and ‘cell surface’ comprised Nrp1, Cxcr6, Cxcr4, Sdc4, H2-Aa, H2-Ab, Lag3, Fasl and Cd38 (Fig. 4f). Those recovery CNS T_reg cell DEGs that intersected the GO terms ‘response to stress’ and ‘transcription factor activity’ comprised Rbpj, Nr4a1, Nr4a2, Nfatc1, Klf10, Zeb2 and Irf4 (Fig. 4g). Together, these data suggest that T cell antigen receptor (TCR) signaling and IL-2 signaling are defining features of the functional phenotype of stress-tolerant T_reg cells residing in the chronically inflamed CNS. Interestingly, CD38 was strongly linked to this T_reg cell phenotype.

Stress-tolerant T_reg cells and tissue T_reg cells share features

To test whether stress-tolerant T_reg cells in the postinflammatory CNS are distinct from bona fide tissue T_reg cells, we compared their transcriptome with the transcriptomes of T_reg cells isolated from the CNS of unmanipulated naive mice and prototypic tissue T_reg cells isolated from the visceral adipose tissue (VAT). Due to the scarcity of Foxp3⁺ T_reg cells in the CNS of naive mice, the cell input had to be adjusted to a lower number in this RNA-seq analysis, resulting in a lower overall sequencing depth across all samples. Nevertheless, PC analysis of the 500 top most variable genes revealed a clear separation along PC1 of the transcriptome of CNS-derived T_reg cell populations from the transcriptome of steady-state VAT T_reg cells, in part driven by the considerably higher expression of tissue T_reg-associated genes such as Areg, Klrg1, Gata3 and Il1rl1 in VAT T_reg cells (Fig. 5a). However, stress-tolerant T_reg cells also showed a clear overlap with VAT T_reg cells for the expression of other tissue T_reg-associated genes (Extended Data Fig. 5a), suggesting that stress-tolerant T_reg cells adopt key features of bona fide tissue T_reg cells. Notably, within the CNS T_reg cell compartment, the transcriptome of stress-tolerant T_reg cells resembled that of naive CNS T_reg cells, and both separated along PC2 from the transcriptome of T_reg cells isolated during the peak of disease (Fig. 5a), indicating that the stress-tolerant CNS T_reg cell transcriptional profile reverts back to a more steady-state-like gene expression. However, despite the similarities to the transcriptome of homeostatic CNS T_reg cells, gene set enrichment analysis (GSEA) again confirmed an enrichment in the IL-2 signaling pathway in stress-tolerant CNS T_reg cells versus naive CNS T_reg cells (Fig. 5b and Extended Data Fig. 5b–d), whereas naive CNS T_reg cells were characterized by more prominent oxidative phosphorylation than stress-tolerant postinflammatory CNS T_reg cells (Fig. 5c and Extended Data Fig. 5b–d). A selection of T_reg cell-associated surface markers, including CD69, CXCR4 and CD38, was markedly expressed in stress-tolerant CNS T_reg cells and to a lesser extent in naive CNS T_reg cells, whereas KLRG1 and ST2 were highly upregulated in VAT T_reg cells (Fig. 5d,e). Together, these data indicate that postinflammatory stress-tolerant CNS T_reg cells adopt a CNS-specific tissue signature²⁸ with a particular dependence on IL-2.

Fig. 5: Recovery CNS Treg cells resemble steady-state CNS Treg cells. — **Fig. 5: Recovery CNS T_reg cells resemble steady-state CNS T_reg cells.**

Lack of CD38 impairs the regulatory capacity of CNS Foxp3⁺ T_reg cells

CD38 was upregulated in T_reg cells in the inflamed CNS, whereas we did not observe disease-phase-dependent upregulation of CD38 in T_conv cells (Fig. 6a). CD38 is the major NAD⁺ hydrolyzing ectoenzyme²⁹. In fact, CD38 is indispensable for maintaining the concentration of NAD⁺ in the microenvironment at low levels. High levels of NAD⁺, in turn, drive the ADP-ribosylation of cell surface molecules by ADP-ribosyltransferase 2b (ARTC2.2), a toxin-related, GPI-anchored ADP-ribosyltransferase expressed in T cells. ARTC2.2 has multiple target molecules that can gain or lose activity through ADP-ribosylation³⁰.

Fig. 6: CD38 signaling controls the functional phenotype of CNS Treg cells. — **Fig. 6: CD38 signaling controls the functional phenotype of CNS T_reg cells.**

To test whether CD38 expression by T_reg cells is relevant in autoimmune CNS inflammation, we developed an experimental model in which we adoptively transferred T_reg cell populations enriched for MOG-reactive T_reg cells that were either sufficient or deficient in CD38 into MOG-immunized host mice whose T_reg cell compartment was acutely depleted immediately before T_reg cell transfer (Fig. 6b). The transfer of MOG-activated wild-type T_reg cells attenuated the EAE severity of recipient mice compared to control EAE mice that were T_reg cell depleted but did not receive a subsequent T_reg cell transfer (Fig. 6c). By contrast, the adoptive transfer of MOG-activated CD38-deficient T_reg cells failed to control EAE symptoms of recipient mice that became as sick as control animals without transfer of T_reg cells (Fig. 6c), suggesting that Cd38^−/− T_reg cells had a T_reg cell-intrinsic deficiency to regulate EAE. Because CD38-deficient T_reg cells were equally recruited to the CNS as wild-type T_reg cells (Fig. 6d–f), we speculated that CD38 was directly or indirectly involved in the formation and maintenance of a productive effector T_reg cell population in the CNS.

To gauge the particular significance of CD38 expression for the functional phenotype of stress-tolerant postinflammatory T_reg cells in nonlymphoid tissues (especially in the CNS), we generated a mixed bone marrow chimera (MBMC) model by transplanting a 1:1 mixture of wild-type Foxp3-DTR⁺ (CD45.1/CD45.2) and ‘nondepletable’ Foxp3-DTR⁻ (CD45.2) donor cells of either wild-type or Cd38^−/− origin into Rag1^−/− recipient mice (Fig. 6g). CNS-specific depletion of DTR-expressing wild-type T_reg cells in these MBMCs by local DTx injection during stable post-peak EAE allowed us to follow disease progression in a scenario in which the remaining CNS-residing T_reg cells were either wild-type or CD38 deficient (Fig. 6g–i). Notably, the T_conv cell compartment was composed of either only wild-type cells (Cd38^+/+ MBMC) or a 1:1 mixture of wild-type and Cd38^−/− T cells (Cd38^−/− MBMCs; Fig. 6h). Consistent with our observation that the severity of EAE was reduced in a gene dose-dependent manner in globally CD38-deficient mice (Extended Data Fig. 6a), Cd38^−/− MBMCs experienced a milder disease course than Cd38^+/+ MBMCs (Extended Data Fig. 6b). Nevertheless, following local depletion of Foxp3-DTR T_reg cells, the remaining nondepletable wild-type T_reg cells in Cd38^+/+ MBMCs were sufficient to maintain a stable EAE score (Fig. 6j and Extended Data Fig. 6b), even when faced with a fully CD38-competent CNS-resident T_conv cell compartment. By contrast, Cd38^−/− MBMCs, with only Cd38^−/− CNS T_reg cells remaining after local depletion of the wild-type T_reg cell fraction, rapidly deteriorated and experienced an EAE relapse with similar kinetics and intensity as we had observed after the complete depletion of CNS T_reg cells (Fig. 6j and Extended Data Fig. 6b; see also Fig. 2d). These data highlight that cell-autonomous expression of CD38 in CNS-resident T_reg cells is indispensable for adequate control of local immune homeostasis in the chronically inflamed CNS in vivo.

CD38 is crucial for maintenance of antigen-specific CNS T_reg cells

Foxp3⁺ T_reg cells need to be activated in an antigen-specific manner to efficiently control effector T cells in EAE³¹. Therefore, the most direct link to the failure of CD38-deficient T_reg cells to control EAE would be the lack of MOG-specific T_reg cells within the population of Cd38^−/− CNS T_reg cells. Indeed, in the CNS, MOG_35–55–I-Ab tetramer-binding T_reg cells were virtually entirely comprised in the CD38⁺ compartment, remarkably during EAE recovery (Fig. 7a–c).

Fig. 7: CD38 is essential for persistence of antigen-specific Treg cells in the inflamed CNS. — **Fig. 7: CD38 is essential for persistence of antigen-specific T_reg cells in the inflamed CNS.**

To test whether CD38 status determined the fraction of MOG-specific T_conv and T_reg cells, we induced EAE in MBMCs, whose hematopoietic compartment consisted of wild-type (CD45.1) and Cd38^−/− (CD45.2) cells, and read out the fraction of MOG_35–55–I-Ab-binding cells in T_conv and T_reg cells in each cellular compartment that were present side-by-side in the same chronically inflamed CNS (day 20 after immunization; Fig. 7d). Notably, although CD38 expression was irrelevant for the fraction of MOG-specific T_conv cells in the CNS, the fraction of MOG-specific CNS T_reg cells was significantly lower in CD38-deficient cells than in CD38-sufficient T cells (Fig. 7e,f). Together, these data indicate that the presence of antigen-specific Foxp3⁺ T_reg cells, but not antigen-specific T_conv cells, in the inflamed CNS is dependent on the cell-intrinsic expression of CD38. This may be particularly relevant for high-affinity antigen-specific T_reg cells because MOG–I-Ab tetramer staining identifies high-affinity MOG-specific T_reg cells³².

Stress-tolerant CNS T_reg cells are dependent on CD38 for sensing IL-2

To dissect the mechanism by which CD38 contributes to the maintenance of antigen-specific T_reg cells in the postinflammatory CNS, we referred to its known function as an NAD⁺-consuming enzyme and speculated that T_reg cells might be more susceptible to NAD⁺-induced cell death than T_conv cells in the chronically inflamed CNS. To directly test this hypothesis, we induced EAE in Foxp3 (GFP) reporter mice and, in the recovery stage, challenged the mice by i.c.v. injection of NAD⁺ in the absence or presence of an inhibitor of ARTC2.2 (Fig. 8a). Although NAD⁺ injection directly resulted in the apoptosis of around 90% of all CNS T_reg cells, CNS T_conv cells were spared from apoptosis (Fig. 8b–d). Interestingly, i.c.v. injection of NAD⁺ did not affect T_reg cells or T_conv cells in the systemic immune compartment (Extended Data Fig. 7a–c). In the CNS of recovery EAE mice, NAD⁺-induced apoptosis was entirely rescued by the co-injection of a blocking nanobody to ARTC2.2 together with NAD⁺ (Fig. 8b–d), suggesting that ADP-ribosylation by ARTC2.2 was the key downstream mediator of NAD⁺-induced cell death in vivo. Also, these data were consistent with an in vivo effect of NAD⁺ and not an artificial release of NAD⁺ from cells during the preparation procedure³³. Although the gain-of-function of P2RX₇ through ADP-ribosylation has previously been shown to drive cell death²³, the expression of P2RX₇ was not different between T_reg cells and T_conv cells in the postinflammatory CNS. Yet, T_conv cells were relatively resistant to NAD⁺-induced apoptosis, whereas T_reg cells were not. Therefore, we explored alternative T_reg cell-specific ADP-ribosylation targets. Because IL-2 signaling had turned out to be a major determinant of stress-tolerant CNS T_reg cells, we first focused on the IL-2Rα subunit, which is highly expressed in antigen-specific T_reg cells (Extended Data Fig. 8a–c). Here, we made use of the prior observation that the R178 ADP-ribosylated hypofunctional form of IL-2Rα is not recognized by the 7D4 clone of anti-CD25, whereas the PC61.5 clone of anti-CD25 detects both the native and the ADP-ribosylated form of IL-2Rα³⁴. To investigate IL-2Rα ADP-ribosylation status, wild-type or CD38-deficient T_reg cells were exposed to NAD⁺ in vitro in the absence or presence of ARTC2.2-inhibiting nanobodies. Although IL-2Rα was mostly not ADP-ribosylated in wild-type T_reg cells, 80% of IL-2Rα was ADP-ribosylated in CD38-deficient T_reg cells. Notably, wild-type T_reg cells were essentially resistant to further ADP-ribosylation induced by NAD⁺ exposure. By contrast, IL-2Rα became fully ADP-ribosylated through NAD⁺ in CD38-deficient T_reg cells, a process that was prevented by a blocking nanobody to ARTC2.2 (Fig. 8e,f). In line with the idea that ADP-ribosylation of IL-2Rα leads to deficient IL-2 signaling through the high-affinity IL-2R expressed in T_reg cells (Fig. 8g), NAD⁺ exposure reduced STAT5 phosphorylation significantly more in CD38-deficient T_reg cells than in wild-type T_reg cells in an ARTC2.2-dependent manner (Fig. 8h,i). Finally, IL-2Rα ADP-ribosylation was significantly more pronounced in CD38-deficient T_reg cells than in their wild-type counterparts, even when T_reg cells of both genotypes were cocultured together in the same dish during NAD⁺ exposure (Fig. 8j), suggesting that cell-intrinsic expression of CD38 was crucial to achieve low levels of NAD⁺ in the proximal vicinity of the cell and, thus, prevent ADP-ribosylation of its surface molecules. Together, these data illustrate that T_reg cell-autonomous expression of CD38 is essential for their perception of restricted amounts of IL-2, as would occur in ‘stressful’ nonlymphoid environments.

Fig. 8: Stress-tolerant Treg cells are dependent on CD38 for efficient IL-2 sensing. — **Fig. 8: Stress-tolerant T_reg cells are dependent on CD38 for efficient IL-2 sensing.**

Discussion

In the present study, we show that Foxp3⁺ T_reg cells persist in the CNS for extended periods of time after the contraction of pathogenic effector T cells. These CNS-resident T_reg cells are indispensable for maintaining the re-established immune homeostasis in the CNS after an autoimmune inflammatory episode, as their site-specific depletion immediately triggers a relapse. The long-term persistence of T_reg cells in the postinflammatory CNS is dependent on the efficient exploitation of limited amounts of IL-2, which selects for antigen-specific T_reg cells with high expression of the high-affinity IL-2R CD25. These stress-tolerant T_reg cells, at the same time, must express CD38 to cope with enhanced levels of extracellular NAD⁺, which otherwise dampen T_reg cell-intrinsic IL-2 signaling due to ADP-ribosylation of IL-2Rα. Therefore, the expression of the NAD⁺ hydrolase CD38 critically determines the resilience of CNS T_reg cells to the hostile environment of the postinflammatory CNS and, thus, is a hallmark feature of stress-tolerant T_reg cells.

CD38 is a multifunctional molecule with diverse roles in various cellular processes. It acts as an NAD⁺ consuming enzyme, thereby regulating both intracellular and extracellular NAD⁺ levels^35,36. Additionally, CD38 functions as a receptor of PECAM-1 (CD31), influencing leukocyte migration³⁷. Although PECAM-1/CD38 interactions might affect T_reg cell migration into nonlymphoid tissues, our data indicate that T_reg cells can migrate to the inflamed CNS without CD38. CD38⁺ T_reg cells possess a survival advantage in nonlymphoid tissues. Conversely, CD38-deficient T_reg cells fail to reduce extracellular NAD⁺ concentrations in their micromilieu and, as a consequence, are more susceptible to ARTC2.2/P2X₇-mediated apoptosis, which is dependent on extensive Ca²⁺ influx through the ATP-gated P2X₇ receptor ion channel due to ARTC2.2-mediated ADP-ribosylation of P2RX₇ (refs. ^38,39,40). However, although P2RX₇ is similarly expressed in T_conv and T_reg cells, the exaggerated susceptibility of T_reg (but not T_conv) cells to NAD⁺-induced cell death in the CNS is likely due to further targets of ADP-ribosylation³⁰, such as IL-2Rα, which is orders of magnitude more highly expressed in T_reg cells than in T_conv cells. Here, ARTC2.2-mediated ADP-ribosylation leads to a loss-of-function of high-affinity IL-2Rα with reduced STAT5 activation. CD38 quenches ADP-ribosylation of IL-2Rα through removing membrane-proximal extracellular NAD⁺ (in a cell-autonomous manner) as a substrate of ARTC2.2. Additionally, CD38 also competes for intracellular NAD⁺, thereby influencing intracellular NAD⁺-dependent pathways. Intracellular NAD⁺ is an essential cofactor for sirtuins, which act as deacetylases of histones and, thus, play a role in gene silencing. Intracellular NAD⁺ is also a cofactor for poly(ADP-ribosyl) polymerases, which are involved in DNA damage repair. These functions of NAD⁺-dependent enzymes have been linked to cellular aging⁴¹. In the context of T cells, SIRT1 has been demonstrated to negatively regulate T_reg cell function by decreasing the expression of Foxp3 and CTLA-4, while simultaneously increasing DNMT3a expression⁴². The absence of CD38 could lead to excess NAD⁺ availability, which may enhance SIRT1 activity and dismantle T_reg cell stability. However, here, we did not observe a difference in Foxp3 expression in CD38-deficient T_reg cells compared to wild-type T_reg cells in the same environment. Furthermore, CD38 regulates intracellular Ca²⁺ levels via the cyclic adenosine 5′-diphosphate ribose (cADPR) signaling pathway, which may influence T cell differentiation⁴³. Proper Ca²⁺ influx is critical for T_reg cell functionality, and impaired Ca²⁺ signaling has been associated with the defective immunosuppressive function of T_reg cells in individuals with multiple sclerosis⁴⁴. Finally, T_reg cells require CD38 for adenosine synthesis via the CD38/CD203a/CD73 ectoenzymatic pathway⁴⁵. Yet, we observed a physical loss of antigen-specific T_reg cells in the CNS when they lacked CD38 expression. Therefore, we suggest that CD38 is primarily necessary for the very maintenance of CNS T_reg cells in the chronically inflamed CNS because it enables them to exploit limited sources of IL-2. Prior studies are consistent with the idea that the availability of IL-2 defines the physical and functional niche of T_reg cells in nonlymphoid tissues¹⁶.

The chronically inflamed CNS is characterized by a high load of extracellular NAD⁺²¹. Therefore, we propose that CD38 expression by T_reg cells is a key feature that determines their resilience to the ‘environmental’ stress of this niche. In addition, CNS T_reg cells share much of what has been described as the tissue-resident T_reg cell signature^2,46. It is unresolved whether a universal tissue T_reg cell signature can be defined or whether each tissue has its own unique T_reg cell population, both in terms of TCR repertoire and functional phenotype. Although other investigators have found a comparable T_reg cell TCR repertoire diversity in various tissue sites⁶, these data need to be interpreted with caution as it is virtually impossible to reach the saturation limit in single-cell sequencing of a fully polyclonal repertoire with currently available technologies. Our data would argue in favor of the requirement for a cognate engagement of postinflammatory CNS-resident T_reg cells because the ablation of CD38 induces a more profound loss of MOG-specific T_reg cells than of T_reg cells that do not bind MOG_35–55–I-Ab tetramers. Therefore, the availability of tissue-specific private antigens may require antigen-specific T_reg cells, whose resilience to the local environment then depends on a universal tissue-type transcriptional program, including CD38.

Finally, local depletion of postinflammatory CNS T_reg cells was sufficient to provoke an EAE relapse, whereas systemic T_reg cell depletion was not. Postinflammatory T_reg cells are, thus, not only adapted to their tissue niche, where they may fulfill nonimmune functions^1,15,17,47, but also still absolutely essential to prevent the recurrence of inflammation in a metastable environment. In fact, CNS inflammation is rekindled from niches of persistent immune cells in the inflammation-experienced CNS when T_reg cells are ablated. These findings establish a paradigm that could explain the increasing disconnect of immunopathology from the systemic immune compartment in chronic CNS autoimmunity. Defining their local interaction partners and growth, as well as maintenance, factors will open new avenues for manipulating CNS T_reg cells as auditors of compartmentalized inflammatory processes that may drive immunopathology in progressive CNS autoimmunity.

Methods

Mice

Foxp3^{eGFP-Cre-ERT2} (Foxp3^Cre-ERT2, IMSR_JAX:016961)⁴⁸, Foxp3^DTR (Foxp3-DTR, IMSR_JAX:016958)⁴⁹, CD45.1 congenic (IMSR_JAX:002014)⁵⁰ and B6.129S7-Rag1^tm1Mom/J (Rag1^−/−, IMSR_JAX:002216)⁵¹ mice were obtained from The Jackson Laboratory. C57BL/6J (wild-type, IMSR_JAX:000664) mice were from Charles River Laboratories. Foxp3 (GFP) reporter mice²⁴ were provided by V. Kuchroo (Harvard Medical School and Brigham and Women’s Hospital). Cd38^tm1Lnd (Cd38^−/−, IMSR_JAX:003727) mice⁵² were provided by H.-W. Mittrücker (Medical Center Hamburg Eppendorf). Foxp3^Cre-ERT2 mice and Rosa-CAG-LSL-tdTomato-WPRE::ΔNeo mice (Rosa26^LSL-tdTomato, IMSR_JAX:007914)⁵³ were mated to generate Foxp3^Cre-ERT2 × Rosa26^LSL-tdTomato mice. Pham^T reporter mice (Cd4^Cre × Rosa26^{LSL-mitoDendra2}, that is, IMSR_JAX:022071 mated with IMSR_JAX:018385) were generated in-house⁵⁴. DEREG mice (MMRRC_032049-JAX)²⁷ were provided by T. Sparwasser (University Medical Center Mainz). We used 8- to 20-week-old disease score-matched male and female C57BL/6 mice in all experiments (see Hiltensperger et al.⁵⁴), except for VAT, where we only used 20-week-old male mice⁵⁵. Mice were housed in a specific pathogen-free facility with normal rodent chow at the Technical University of Munich. Protocols were approved by the Bavarian state authorities and conducted according to the relevant guidelines (ROB-55.2-2532.Vet_02-17-69, ROB-55.2-2532.Vet_02-17-234, ROB-55.2-2532.Vet_03-18-53, ROB-55.2-2532.Vet_02-20-1, ROB-55.2-2532.Vet_02-20-23, ROB-55.2-2532.Vet_02-20-187, ROB-55.2-2532.Vet_02-23-71, ROB-55.2-2532.Vet_02-23-118 and ROB-55.2-2532.Vet_02-25-68).

EAE model

Mice were immunized with MOG_35–5₅ (200 μg, Auspep) emulsified in complete Freund’s adjuvant (500 mg Mycobacterium tuberculosis H37Ra, BD Difco). Mice received 200 ng or a suboptimal dose of 50 ng pertussis toxin (Sigma-Aldrich) i.v. on days 0 and 2 after immunization. Clinical scores were recorded as 0 (no disease), 1 (tail tone loss), 2 (hind limb paresis), 3 (hind limb paralysis), 4 (tetraplegia) and 5 (moribund). Peak disease occurred on days 12–15, with recovery starting around days 30–35.

BM chimeras

Six- to 8-week-old Rag1^−/− mice were sublethally irradiated (7 Gy; two doses of 3.5 Gy, 3 h apart) and i.v. injected with 1 × 10⁷ CD90.2-depleted donor BM cells. Mice received enrofloxacin (0.1 mg ml⁻¹) in their drinking water for 2 weeks after BM transplantation. Hematopoietic reconstitution was assessed using peripheral blood at 6 weeks, and the chimeras were immunized for EAE induction 8 weeks after BM grafting.

Tissue and cell isolation

Mice were killed and perfused with 20–30 ml of cold PBS via the left heart ventricle. Spleen tissue and lymph nodes were homogenized and filtered (70-µm strainers, Falcon), and red blood cells were lysed with 2 ml of FACS Lysing Solution (BD Biosciences) for 5 min. Cells were washed and resuspended in FACS buffer (2% fetal calf serum (FCS) in PBS). For CNS cell isolation, brains were dissected, and spinal cords were flushed out with PBS. CNS tissues were digested in 4 ml of digestion buffer (2.5 mg ml⁻¹ collagenase D (Roche Diagnostics), 1 mg ml⁻¹ DNase I (Roche Diagnostics) and 1 µg ml⁻¹ S+16a anti-ARTC2.2 nanobody (BioLegend) in 10% FCS DMEM) at 37 °C for 45 min with shaking at 120 rpm, filtered and separated via Percoll gradient (70–37%, GE Healthcare). Cells were collected, washed and resuspended in cold FACS buffer. VAT from male mice was collected, digested in 3 ml of digestion buffer at 37 °C for 45 min, washed with 30 ml of FACS buffer and treated with 1 ml of FACS lysing solution for 5 min. Cells were washed, pelleted and resuspended in cold FACS buffer. BM was isolated from the femur and tibia. The bones were cut out from each end and flushed out with cold 10% FCS DMEM with a 21-gauge needle. The cells were collected, and red blood cells were lysed with 1 ml of FACS lysing solution for 5 min. The cells were then washed and resuspended in cold FACS buffer.

Cell sorting

Splenocytes and lymph nodes were pooled to isolate T_reg cells. Briefly, CD4⁺ T cells were enriched from single-cell suspensions using a negative selection kit (Miltenyi Biotec). Cells were stained with antibodies for 15 min at 4 °C in the dark, washed and resuspended in FACS buffer. Sorting was conducted using a FACSAria III Cell Sorter (BD Biosciences), achieving >95% purity.

Flow cytometry

Cell suspensions were stained with Live/Dead Aqua, Live/Dead violet or Near-IR dyes (Thermo Fisher) for 15 min at 4 °C in PBS and washed with FACS buffer. Cells were incubated with an antibody cocktail targeting surface antigen and CD16/CD32 Fc block in FACS buffer for 15 min at 4 °C in the dark, washed twice and resuspended in FACS buffer. For Annexin V staining, cells were stained with an antibody cocktail in Annexin V binding buffer (BioLegend). For intracellular staining, cells were stained with antibodies to surface antigens, fixed with 4% paraformaldehyde (PFA) for 15 min at room temperature (protected from light) and washed with FACS buffer. Fixation and permeabilization were performed using a eBioscience Intracellular Fixation and Permeabilization Buffer Set (Invitrogen). Cells were incubated overnight at 4 °C with an intracellular antibody cocktail and Fc block in permeabilization buffer, washed and resuspended in FACS buffer.

For pSTAT5 Y694 staining, fixation and permeabilization were performed using a Transcription Factor Phospho Buffer Set (TFP, BD Pharmingen), following the manufacturer’s protocol. In short, after fixation and permeabilization, cells were incubated with anti-pSTAT5 Y694, anti-Foxp3 and anti-surface antigen with Fc block in TFP Perm/Wash Buffer for 50 min at 4 °C. Cells were washed twice with TFP Perm/Wash buffer and resuspended in FACS buffer for flow cytometry analysis. A list of the applied antibodies and resources is provided in Supplementary Table 4. Antibody dilutions are detailed in the Reporting Summary. Data were collected using CytoFLEX (Beckman Coulter), FACSAria III (BD Biosciences) or Aurora (Cytek Biosciences) and analyzed using FlowJo software (version 10.8.1, BD Biosciences).

T_reg cell cluster visualizations were based on the data acquired on the Aurora system. T_reg cells were identified as Aqua⁻Foxp3⁺CD4⁺ through primary gating in each sample in FlowJo. To standardize cell numbers across tissues, downsampling was conducted by pooling and selecting 300 T_reg cells, which were then subjected to secondary FlowSOM (v3.0.18) clustering. Further subclustering of T_reg cells was achieved using ClusterExplorer (v1.6.5). The clusters were visualized with uniform manifold approximation and projection through R and the FlowJo plugin packages.

I-Ab tetramer staining

CNS and spleen cells were treated with neuraminidase (0.35 U ml⁻¹, Sigma-Aldrich) and dasatinib (100 nM, Selleckchem) in serum-free DMEM at 37 °C with 5% CO₂ for 25 min. Cells were then washed twice with FACS buffer, blocked with anti-mouse CD16/CD32 Fc block on ice for 15 min and stained with T-Select I-A^b MOG_35–55 Tetramer-PE (Medical and Biological Laboratories, TS-M704-1) or the control T-Select I-A^b OVA_323–339 Tetramer-PE (TS-M710-1) at room temperature for 2 h, with gentle resuspension every 30 min. After tetramer staining, cells were centrifuged (430g, 4 °C, 5 min), washed twice with FACS buffer and incubated with anti-phycoerythrin MicroBeads (Miltenyi Biotec) on a 4 °C shaker for 30 min to stabilize tetramer binding. The cells were washed with FACS buffer and subjected to surface and intracellular staining for flow cytometry analysis. The tetramer⁺ population was identified by gating on NIR⁻, Lin (CD19, CD8a, F4/80, Ly6G)⁻ and MOG_35–55 tetramer–PE⁺ cells, with the OVA_323–339 tetramer as a negative control.

Bulk RNA-seq

Approximately 2,000 or 250 T_reg or T_conv cells (viability > 90%) were sorted directly into TCL buffer (Qiagen) plus 1% 2-mercaptoethanol (Merk) and stored at −80 °C. RNA was extracted and purified using an RNeasy Plus Micro kit (Qiagen). RNA quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). Library preparation followed a previously described protocol⁵⁶. Barcode-specific cDNA was synthesized via reverse transcription with Maxima RT polymerase (Thermo Fisher) using an oligo(dT) primer containing barcodes, unique molecular identifiers (UMIs) and adaptor sequences. The 5′ end of the cDNA was then extended using primers complementary to the template switch oligonucleotide site, followed by amplification of full-length cDNA. The cDNA was fragmented using a Nextera XT kit (Illumina), and the 3′-end fragments were amplified with primers incorporating Illumina P5 and P7 overhangs. The P5 and P7 sites were modified to optimize cluster recognition during sequencing, allowing for sequencing of the cDNA in read 1 and the barcodes/UMIs in read 2.

Sequencing was performed on an Illumina NextSeq 500 platform with 75 cycles for read 1 and 16 for read 2. The sequencing data were processed using the Drop-seq pipeline⁵⁷ to generate UMI tables organized by sample and gene. The GRCm38 reference genome was aligned with transcript and gene annotations derived from ENSEMBL release 75.

Differential gene expression analysis was conducted in R (v4.5.0) using the DESeq2 package (v1.48.1)⁵⁸. Significant DEGs were defined as genes with a Benjamini–Hochberg-adjusted P value of <0.05 (in some cases 0.1 as indicated) and an absolute fold change in expression of >1.5. A log₂(fold shrinkage) estimation was performed using the R package glmGamPoi (v1.20.0)⁵⁹. GO term annotations were derived from the R package org.Mm.eg.db (v3.21.0). For visualization, count data were transformed using the variance stabilizing transformation (‘vst’) function in DESeq2. GSEA was performed on DESeq2 normalized counts using the Broad Institute GSEA Desktop Application (v4.3.2) in conjunction with the Molecular Signatures Database (v2024.1.Mm), with permutation type ‘gene_set’ and an FDR threshold of 0.25 applied. GO enrichment analysis of adjusted P value ranked DEGs was performed using the gprofiler2 R package⁶⁰.

In vivo cell proliferation assay

EAE was induced in C57BL/6 wild-type mice. At various time points during EAE progression, EdU (0.05 μg mg (body weight), Click-iT EdU, Sigma-Aldrich) was administered i.p. 18 h before euthanasia. Circulating cells were labeled with anti-CD45.2 BV650 (3 μg, BioLegend) via i.v. injection 3–5 min before euthanasia. EdU staining was performed according to the manufacturer’s protocol. Cells from the CNS and spleen were stained with surface antigens and fixed with 4% PFA (Thermo Fisher) at room temperature for 15 min in the dark. After two washes with FACS buffer, cells were incubated in a saponin-based fixation-permeabilization buffer at room temperature for 20 min, followed by centrifugation. A fresh EdU detection cocktail was prepared, and cells were incubated at room temperature for 30 min. After two washes with the saponin-based fixation-permeabilization buffer, cells were stained with anti-Foxp3 in the same buffer overnight at 4 °C. Cells were washed twice again, resuspended in FACS buffer and analyzed by flow cytometry.

In vivo T cell trafficking assay

Mice were anesthetized with a cocktail of medetomidine, midazolam and fentanyl, and the fur was removed using depilatory cream (Veet). For photolabeling, the inguinal lymph node on the immunization side was irradiated for 20 min using a 405-nm diode laser (LuxX 405-60, Soliton) as previously described⁵⁴. Two days after photoconversion, the mice were killed, and cells from the CNS and spleen were collected to assess the migration of photolabelled lymphocytes.

In vivo Foxp3 fate mapping

For tamoxifen preparation, tamoxifen (Sigma-Aldrich) was dissolved in 100% ethanol (Merck) at a concentration of 1 g ml⁻¹, vortexed at 37 °C for 15 min and mixed with Miglyol (Caelo) to achieve a final concentration of 10 mg ml⁻¹. The solution was incubated and vortexed at 37 °C for 30 min to ensure complete dissolution. We followed the protocol by Jahn et al.⁶¹ for tamoxifen treatment. Foxp3^Cre-ERT2 × Rosa26^LSL-tdTomato mice were given daily i.p. injections of 1 mg 100 μl⁻¹ tamoxifen for 5 days. Fourteen days after the last injection, the mice were killed, and their spleen and CNS cells were collected for either flow cytometry or RNA-seq analysis.

In vivo T_reg cell depletion

EAE was induced in Foxp3-DTR and control Foxp3 (GFP) mice. During the EAE recovery phase (between 32 and 35 days after immunization, with a clinical score of less than 2), DTx (from Corynebacterium diphtheriae, Sigma-Aldrich) or PBS was administered either i.c.v. (15 ng) or i.p. (500 ng). The i.c.v. injection was performed according to the protocol described by Taylor et al.⁶², modified to administer 3 μl of DTx at a rate of 1 μl min⁻¹. Buprenorphine (0.1 μg per mg (body weight)) was given before, after and 1 day after surgery for pain management. T_reg cell depletion efficiency was confirmed by flow cytometry, and clinical scores were monitored daily from day 0 to day 3 after DTx injection. For FTY720 treatment, FTY720 (1 μg per mg (body weight), Sigma-Aldrich) or vehicle control was administered i.p. starting 1 day before the DTx i.c.v. injection and continuing daily until 1 day after surgery. Clinical scores were recorded from day 0 to day 3.

In vivo T_reg cell suppression

EAE was induced in both donor (either Cd38^+/+ or Cd38^−/−, CD45.2) and recipient DEREG (CD45.1) mice. DTx (1 μg, 50 μg per kg (body weight)) was administered i.p. on days 9 and 10 after induction to deplete T_reg cells in recipients. On day 11, donors were pretreated i.p. with 50 µg of S+16a (anti-ARTC2.2 nanobody, BioLegend) 30 min before euthanasia to prevent cell apoptosis. Donor splenic T_reg cells (LIVE/DEAD Aqua⁻CD4⁺CD25 (clone 7D4)⁺GITR⁺CD45.2⁺) were sorted, and 10⁵ T_reg cells were transferred i.v. into recipients.

In vivo NAD⁺ i.c.v. injection

EAE was induced in Foxp3 (GFP) mice. During the recovery phase, mice received an i.c.v. injection of either freshly prepared NAD⁺ (Thermo Fisher; 100 μg in 3 μl of cold PBS) or vehicle control (3 μl of PBS). Buprenorphine (0.1 μg per mg (body weight)) was administered before and after surgery. Mice were killed 24 h after injection.

In vitro NAD⁺ incubation and IL-2 stimulation

Splenocytes (5 × 10⁶) from Cd38^+/+ or Cd38^−/− mice were incubated with or without NAD⁺ (concentrations are specified in the figure legends) and anti-ARTC2.2 (1 µg ml⁻¹) in serum-free DMEM at 4 °C for 15 min. Cells were washed with cold FACS buffer, stained with antibodies and analyzed by flow cytometry. For the pSTAT5 Y694 assay, following the initial incubation, cells were stimulated with mouse IL-2 (Miltenyi Biotec, 1.25 ng ml⁻¹) in DMEM containing 2% BSA at 37 °C for 10 min. Cells were immediately fixed and stained for anti-pSTAT5 Y694.

Histological analysis

Mice were killed and perfused with 20–30 ml of cold PBS via the left ventricle. Brains were dissected and fixed in 4% PFA at 4 °C for 48 h, washed with PBS and paraffin embedded. Representative 2-μm sections were cut with a rotary microtome, and immunohistochemistry was conducted using a Leica Bond Rxm autostainer with a Bond Polymer Refine Detection kit (DS9800, Leica Biosystems). Heat-induced epitope retrieval was performed by using citrate-based pH 6 Bond Epitope Retrieval Solution 1 (AR9961, Leica Biosystems). Antibody details are listed in Supplementary Table 4. Slides were then scanned with a Leica Aperio AT2 system and analyzed in QuPath (v0.6.0) using automated positive cell detection.

To generate multiplex immunostainings, 2-µm sections were deparaffinized and subjected to antigen retrieval (citrate buffer pH 6.0). To prevent myelin autofluorescence, slides were treated with 4.5% H₂O₂ and 20 mM NaOH in PBS under a light-emitting diode for 45 min twice at room temperature. Nonspecific binding was blocked with 2.5% goat serum in PBS, and samples were incubated overnight with rabbit anti-CD4 (Cell Signaling). Signals were detected with anti-rabbit horseradish peroxidase (Dako, K4003) with amplification (TSA vivid 520, Tocris, 7534). Following denaturation (Discovery Roche, CC2), slides were blocked with Fab fragment anti-mouse IgG (Jackson Immunoresearch) and 2.5% goat serum in PBS and incubated with rat anti-Foxp3 (eBioscience) overnight. Signals were detected with anti-rat horseradish peroxidase (Vectorlab) with amplification (TSA vivid 570, Tocris, 7535). Nuclei were stained with DAPI (Invitrogen), and slides were mounted in Fluoromount aqueous mounting medium (Sigma-Aldrich, F4680). Immunostained sections were scanned using an Akoya PhenoImager HT 2.0 (Akoya Biosciences).

For cryosections, brains were fixed in 4% PFA at 4 °C for 48 h and subjected to a 30% sucrose gradient for 48 h. Ten-micron cryosections (CM3050 S, Leica) were subjected to immunofluorescence stainings. The antibody details are listed in Supplementary Table 4. Images were acquired on an SP8 confocal microscope (Leica) and analyzed using LAS X software (v.3.5.6.21594, Leica).

Statistical analysis

Statistical analyses were performed using Prism 10 (GraphPad) or R (v4.4.0). No statistical methods were used to predetermine sample sizes, but our sample sizes were similar to those reported in previous publications⁵⁴. Data distribution was assumed to be normal, but this was not always formally tested. In any case, individual data points are provided in the figures. Significance of the differences between two groups was evaluated using two-tailed Student’s t-tests. Differences among three or more groups were analyzed using a one-way or two-way ANOVA with a Tukey’s HSD post hoc test, and results are presented as mean values + s.d. Adjusted P values from DEGs were derived from DESeq2 using a two-tailed Wald’s test with correction for multiple testing using the Benjamini and Hochberg method. Unless stated otherwise in the figure legend, animals were assigned to experimental groups in an age- and sex-matched manner. Scoring of EAE was performed in a manner blinded to the experimental conditions. Unless otherwise stated in the figure legends, all animals were included in the analysis. The significance of differences in clinical scores was determined using two-way ANOVA with a Tukey’s post hoc test, and the results are presented as mean values + s.e.m. The significance of the survival rate was analyzed with a Mantel–Cox test. The individual sample sizes and statistical methods are provided in the figure legends. The individual P or adjusted P values are indicated in the figures.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Sequencing data have been deposited in the Gene Expression Omnibus database under accession code GSE311963. All other data are present in the article and Supplementary Information. Source data are provided with this paper.

References

Feuerer, M. et al. Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters. Nat. Med. 15, 930–939 (2009).
Article CAS PubMed PubMed Central Google Scholar
Panduro, M., Benoist, C. & Mathis, D. Tissue T_regs. Annu. Rev. Immunol. 34, 609–633 (2016).
Article CAS PubMed PubMed Central Google Scholar
Russler-Germain, E. V., Rengarajan, S. & Hsieh, C. S. Antigen-specific regulatory T-cell responses to intestinal microbiota. Mucosal Immunol. 10, 1375–1386 (2017).
Article CAS PubMed PubMed Central Google Scholar
Li, C. et al. TCR transgenic mice reveal stepwise, multi-site acquisition of the distinctive fat-T_reg phenotype. Cell 174, 285–299 (2018).
Article CAS PubMed PubMed Central Google Scholar
Korn, T. & Muschaweckh, A. Stability and maintenance of Foxp3⁺ T_reg cells in non-lymphoid microenvironments. Front. Immunol. 10, 2634 (2019).
Article CAS PubMed PubMed Central Google Scholar
Burton, O. T. et al. The tissue-resident regulatory T cell pool is shaped by transient multi-tissue migration and a conserved residency program. Immunity 57, 1586–1602 (2024).
Article CAS PubMed Google Scholar
Delacher, M. et al. Single-cell chromatin accessibility landscape identifies tissue repair program in human regulatory T cells. Immunity 54, 702–720 (2021).
Article CAS PubMed PubMed Central Google Scholar
DiSpirito, J.R. et al. Molecular diversification of regulatory T cells in nonlymphoid tissues. Sci. Immunol. 3, eaat5861 (2018).
Fan, X. et al. CD49b defines functionally mature T_reg cells that survey skin and vascular tissues. J. Exp. Med. 215, 2796–2814 (2018).
Article CAS PubMed PubMed Central Google Scholar
Abbas, A. K., Trotta, E., Dimitre, R. S., Marson, A. & Bluestone, J. A. Revisiting IL-2: biology and therapeutic prospects. 3, eaat1482 (2018)
Korn, T. et al. Myelin-specific regulatory T cells accumulate in the CNS but fail to control autoimmune inflammation. Nat. Med. 13, 423–431 (2007).
Article CAS PubMed PubMed Central Google Scholar
Cretney, E. et al. The transcription factors Blimp-1 and IRF4 jointly control the differentiation and function of effector regulatory T cells. Nat. Immunol. 12, 304–311 (2011).
Article CAS PubMed Google Scholar
Garg, G. et al. Blimp1 prevents methylation of Foxp3 and loss of regulatory T cell identity at sites of inflammation. Cell Rep. 26, 1854–1868 (2019).
Article CAS PubMed PubMed Central Google Scholar
Delacher, M. et al. Precursors for nonlymphoid-tissue T_reg cells reside in secondary lymphoid organs and are programmed by the transcription factor BATF. Immunity 52, 295–312 (2020).
Article CAS PubMed PubMed Central Google Scholar
Ito, M. et al. Brain regulatory T cells suppress astrogliosis and potentiate neurological recovery. Nature 565, 246–250 (2019).
Article CAS PubMed Google Scholar
Yshii, L. et al. Astrocyte-targeted gene delivery of interleukin 2 specifically increases brain-resident regulatory T cell numbers and protects against pathological neuroinflammation. Nat. Immunol. 23, 878–891 (2022).
Article CAS PubMed PubMed Central Google Scholar
Dombrowski, Y. et al. Regulatory T cells promote myelin regeneration in the central nervous system. Nat. Neurosci. 20, 674–680 (2017).
Article CAS PubMed PubMed Central Google Scholar
Shi, L. et al. T_reg cell-derived osteopontin promotes microglia-mediated white matter repair after ischemic stroke. Immunity 54, 1527–1542 (2021).
Article CAS PubMed PubMed Central Google Scholar
Marin-Rodero, M. et al. The meninges host a distinct compartment of regulatory T cells that preserves brain homeostasis. Sci. Immunol. 10, eadu2910 (2025).
Midavaine, E. et al. Meningeal regulatory T cells inhibit nociception in female mice. Science 388, 96–104 (2025).
Article CAS PubMed Google Scholar
Chiarugi, A. Inhibitors of poly(ADP-ribose) polymerase-1 suppress transcriptional activation in lymphocytes and ameliorate autoimmune encephalomyelitis in rats. Br. J. Pharmacol. 137, 761–770 (2002).
Article CAS PubMed PubMed Central Google Scholar
Aswad, F., Kawamura, H. & Dennert, G. High sensitivity of CD4⁺CD25⁺ regulatory T cells to extracellular metabolites nicotinamide adenine dinucleotide and ATP: a role for P2X₇ receptors. J. Immunol. 175, 3075–3083 (2005).
Article CAS PubMed Google Scholar
Seman, M. et al. NAD-induced T cell death. Immunity 19, 571–582 (2003).
Article CAS PubMed Google Scholar
Bettelli, E. et al. Reciprocal developmental pathways for the generation of pathogenic effector T_H17 and regulatory T cells. Nature 441, 235–238 (2006).
Article CAS PubMed Google Scholar
Kim, E. H. et al. Bach2 regulates homeostasis of Foxp3⁺ regulatory T cells and protects against fatal lung disease in mice. J. Immunol. 192, 985–995 (2014).
Article CAS PubMed Google Scholar
Osman, A. et al. TCF-1 controls T_reg cell functions that regulate inflammation, CD8⁺ T cell cytotoxicity and severity of colon cancer. Nat. Immunol. 22, 1152–1162 (2021).
Article CAS PubMed PubMed Central Google Scholar
Lahl, K. et al. Selective depletion of Foxp3⁺ regulatory T cells induces a scurfy-like disease. J. Exp. Med. 204, 57–63 (2007).
Article CAS PubMed PubMed Central Google Scholar
Wakim, L. M., Woodward-Davis, A. & Bevan, M. J. Memory T cells persisting within the brain after local infection show functional adaptations to their tissue of residence. Proc. Natl. Acad. Sci. USA 107, 17872–17879 (2010).
Article CAS PubMed PubMed Central Google Scholar
Zeidler, J. D. et al. Endogenous metabolism in endothelial and immune cells generates most of the tissue vitamin B3 (nicotinamide). iScience 25, 105431 (2022).
Article CAS PubMed PubMed Central Google Scholar
Leutert, M. et al. Identification of the mouse T cell ADP-ribosylome uncovers ARTC2.2 mediated regulation of CD73 by ADP-ribosylation. Front. Immunol. 12, 703719 (2021).
Article CAS PubMed PubMed Central Google Scholar
Stephens, L. A., Malpass, K. H. & Anderton, S. M. Curing CNS autoimmune disease with myelin-reactive Foxp3⁺ T_reg. Eur. J. Immunol. 39, 1108–1117 (2009).
Article CAS PubMed Google Scholar
Sabatino, J. J. Jr., Huang, J., Zhu, C. & Evavold, B. D. High prevalence of low affinity peptide–MHC II tetramer-negative effectors during polyclonal CD4⁺ T cell responses. J. Exp. Med. 208, 81–90 (2011).
Article CAS PubMed PubMed Central Google Scholar
Georgiev, H. et al. Blocking the ART2.2/P2X₇-system is essential to avoid a detrimental bias in functional CD4 T cell studies. Eur. J. Immunol. 48, 1078–1081 (2018).
Article CAS PubMed Google Scholar
Teege, S. et al. Tuning IL-2 signaling by ADP-ribosylation of CD25. Sci. Rep. 5, 8959 (2015).
Article CAS PubMed PubMed Central Google Scholar
Shubinsky, G. & Schlesinger, M. The CD38 lymphocyte differentiation marker: new insight into its ectoenzymatic activity and its role as a signal transducer. Immunity 7, 315–324 (1997).
Article CAS PubMed Google Scholar
Chini, C. C. S., Zeidler, J. D., Kashyap, S., Warner, G. & Chini, E. N. Evolving concepts in NAD⁺ metabolism. Cell Metab. 33, 1076–1087 (2021).
Article CAS PubMed PubMed Central Google Scholar
Deaglio, S. et al. Human CD38 (ADP-ribosyl cyclase) is a counter-receptor of CD31, an Ig superfamily member. J. Immunol. 160, 395–402 (1998).
Article CAS PubMed Google Scholar
Krebs, C. et al. CD38 controls ADP-ribosyltransferase-2-catalyzed ADP-ribosylation of T cell surface proteins. J. Immunol. 174, 3298–3305 (2005).
Article CAS PubMed Google Scholar
Hubert, S. et al. Extracellular NAD⁺ shapes the Foxp3⁺ regulatory T cell compartment through the ART2–P2X₇ pathway. J. Exp. Med. 207, 2561–2568 (2010).
Article CAS PubMed PubMed Central Google Scholar
Andrejew, R. et al. The P2X₇ receptor: central hub of brain diseases. Front. Mol. Neurosci. 13, 124 (2020).
Article CAS PubMed PubMed Central Google Scholar
Covarrubias, A. J., Perrone, R., Grozio, A. & Verdin, E. NAD⁺ metabolism and its roles in cellular processes during ageing. Nat. Rev. Mol. Cell Biol. 22, 119–141 (2021).
Article CAS PubMed Google Scholar
Beier, U. H., Akimova, T., Liu, Y., Wang, L. & Hancock, W. W. Histone/protein deacetylases control Foxp3 expression and the heat shock response of T-regulatory cells. Curr. Opin. Immunol. 23, 670–678 (2011).
Article CAS PubMed PubMed Central Google Scholar
Huang, X. & Rudensky, A. Y. Regulatory T cells in the context: deciphering the dynamic interplay with the tissue environment. Curr. Opin. Immunol. 89, 102453 (2024).
Article CAS PubMed PubMed Central Google Scholar
Schwarz, A. et al. Fine-tuning of regulatory T cell function: the role of calcium signals and naive regulatory T cells for regulatory T cell deficiency in multiple sclerosis. J. Immunol. 190, 4965–4970 (2013).
Article CAS PubMed Google Scholar
Horenstein, A. L. et al. A CD38/CD203a/CD73 ectoenzymatic pathway independent of CD39 drives a novel adenosinergic loop in human T lymphocytes. Oncoimmunology 2, e26246 (2013).
Article PubMed PubMed Central Google Scholar
Burzyn, D., Benoist, C. & Mathis, D. Regulatory T cells in nonlymphoid tissues. Nat. Immunol. 14, 1007–1013 (2013).
Article CAS PubMed PubMed Central Google Scholar
Vasanthakumar, A. et al. The transcriptional regulators IRF4, BATF and IL-33 orchestrate development and maintenance of adipose tissue-resident regulatory T cells. Nat. Immunol. 16, 276–285 (2015).
Article CAS PubMed Google Scholar
Rubtsov, Y. P. et al. Stability of the regulatory T cell lineage in vivo. Science 329, 1667–1671 (2010).
Article CAS PubMed PubMed Central Google Scholar
Kim, J. M., Rasmussen, J. P. & Rudensky, A. Y. Regulatory T cells prevent catastrophic autoimmunity throughout the lifespan of mice. Nat. Immunol. 8, 191–197 (2007).
Article CAS PubMed Google Scholar
Shen, F. W. et al. Cloning of Ly-5 cDNA. Proc. Natl Acad. Sci. USA 82, 7360–7363 (1985).
Article CAS PubMed PubMed Central Google Scholar
Mombaerts, P. et al. RAG-1-deficient mice have no mature B and T lymphocytes. Cell 68, 869–877 (1992).
Article CAS PubMed Google Scholar
Cockayne, D. A. et al. Mice deficient for the ecto-nicotinamide adenine dinucleotide glycohydrolase CD38 exhibit altered humoral immune responses. Blood 92, 1324–1333 (1998).
Article CAS PubMed Google Scholar
Madisen, L. et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010).
Article CAS PubMed Google Scholar
Hiltensperger, M. et al. Skin and gut imprinted helper T cell subsets exhibit distinct functional phenotypes in central nervous system autoimmunity. Nat. Immunol. 22, 880–892 (2021).
Article CAS PubMed PubMed Central Google Scholar
Vasanthakumar, A. et al. Sex-specific adipose tissue imprinting of regulatory T cells. Nature 579, 581–585 (2020).
Article CAS PubMed PubMed Central Google Scholar
Parekh, S., Ziegenhain, C., Vieth, B., Enard, W. & Hellmann, I. The impact of amplification on differential expression analyses by RNA-seq. Sci. Rep. 6, 25533 (2016).
Article CAS PubMed PubMed Central Google Scholar
Macosko, E. Z. et al. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell 161, 1202–1214 (2015).
Article CAS PubMed PubMed Central Google Scholar
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Article PubMed PubMed Central Google Scholar
Ahlmann-Eltze, C. & Huber, W. glmGamPoi: fitting γ-Poisson generalized linear models on single cell count data. Bioinformatics 36, 5701–5702 (2021).
Article PubMed PubMed Central Google Scholar
Kolberg, L., Raudvere, U., Kuzmin, I., Vilo, J. & Peterson, H. gprofiler2—an R package for gene list functional enrichment analysis and namespace conversion toolset g:Profiler. F1000Res 9, ELIXIR-709 (2020).
Jahn, H. M. et al. Refined protocols of tamoxifen injection for inducible DNA recombination in mouse astroglia. Sci. Rep. 8, 5913 (2018).
Article PubMed PubMed Central Google Scholar
Taylor, Z. V., Khand, B., Porgador, A., Monsonego, A. & Eremenko, E. An optimized intracerebroventricular injection of CD4⁺ T cells into mice. STAR Protoc. 2, 100725 (2021).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We are grateful to H.-W. Mittrücker (Universitätsklinikum Hamburg Eppendorf) for generously providing CD38-deficient mice. We would like to thank K. Schumann (University of Konstanz) and M. Schmidt-Supprian (Technical University of Munich School of Medicine) for fruitful discussions. We appreciate B. Lunk, M. Schmidt and V. Husterer (all Technical University of Munich School of Medicine) for their assistance with mouse colony management. We thank the Technical University of Munich Comparative Experimental Pathology Unit for expert support with histological analyses. This study was supported by the Deutsche Forschungsgemeinschaft (TRR274 (ID 408885537) to T.K., TRR355 (ID 490846870) to T.K. and A.M., GRK2668 (ID 435874434) to T.K. and EXC 2145 (SyNergy, ID 390857198) to T.K.) and by the Hertie Network of Clinical Neuroscience (to T.K.).

Funding

Open access funding provided by Technische Universität München.

Author information

These authors jointly supervised this work: Andreas Muschaweckh, Thomas Korn.

Authors and Affiliations

Institute for Experimental Neuroimmunology, Technical University of Munich School of Medicine, Munich, Germany
Hsin-Hsiang Chen, Sofia Tyystjärvi, Diego Ruiz Navarro, Helena Domínguez Moreno, Ali Maisam Afzali, Sylvia Heink, Lisa Charlotte Richter, Christopher Sie, Gildas Lepennetier, Lea Seeholzer, Andreas Muschaweckh & Thomas Korn
Department of Biotechnology and Bioinformatics, School of Life Sciences, University of Hyderabad, Hyderabad, India
Ravi Kant
Institute of Pathology, Technical University of Munich, Munich, Germany
Tanja Groll & Katja Steiger
Department of Pathology and Immunology, Division of Clinical Pathology, Geneva Faculty of Medicine, Centre Médical Universitaire, Geneva, Switzerland
Ingrid Wagner & Doron Merkler
Institute for Computational Biology, Helmholtz Munich, Campus Neuherberg, Neuherberg, Germany
Irene Bonafonte-Pardàs & Benjamin Schubert
Institute of Molecular Oncology and Functional Genomics, TranslaTUM Cancer Center, Technical University of Munich School of Medicine, Munich, Germany
Rupert Öllinger & Roland Rad
Department of Neurology, Technical University of Munich School of Medicine, Munich, Germany
Ali Maisam Afzali & Thomas Korn
Division of Molecular Biology, Biomedical Center, Faculty of Medicine, LMU Munich, Martinsried, Germany
Gunnar Schotta
Munich Cluster for Systems Neurology, Munich, Germany
Thomas Korn

Authors

Hsin-Hsiang Chen
View author publications
Search author on:PubMed Google Scholar
Sofia Tyystjärvi
View author publications
Search author on:PubMed Google Scholar
Diego Ruiz Navarro
View author publications
Search author on:PubMed Google Scholar
Ravi Kant
View author publications
Search author on:PubMed Google Scholar
Tanja Groll
View author publications
Search author on:PubMed Google Scholar
Ingrid Wagner
View author publications
Search author on:PubMed Google Scholar
Helena Domínguez Moreno
View author publications
Search author on:PubMed Google Scholar
Irene Bonafonte-Pardàs
View author publications
Search author on:PubMed Google Scholar
Rupert Öllinger
View author publications
Search author on:PubMed Google Scholar
Ali Maisam Afzali
View author publications
Search author on:PubMed Google Scholar
Sylvia Heink
View author publications
Search author on:PubMed Google Scholar
Lisa Charlotte Richter
View author publications
Search author on:PubMed Google Scholar
Christopher Sie
View author publications
Search author on:PubMed Google Scholar
Gildas Lepennetier
View author publications
Search author on:PubMed Google Scholar
Lea Seeholzer
View author publications
Search author on:PubMed Google Scholar
Katja Steiger
View author publications
Search author on:PubMed Google Scholar
Doron Merkler
View author publications
Search author on:PubMed Google Scholar
Roland Rad
View author publications
Search author on:PubMed Google Scholar
Gunnar Schotta
View author publications
Search author on:PubMed Google Scholar
Benjamin Schubert
View author publications
Search author on:PubMed Google Scholar
Andreas Muschaweckh
View author publications
Search author on:PubMed Google Scholar
Thomas Korn
View author publications
Search author on:PubMed Google Scholar

Contributions

Conceptualization: H.-H.C., A.M. and T.K. Methodology: H.-H.C., S.T., D.R.N., R.K., T.G., I.W., H.D.M., I.B.-P., R.Ö., A.M.A., S.H., L.C.R., C.S., G.L., L.S., K.S., D.M., R.R., G.S. and B.S. Investigation: H.-H.C., S.T., D.R.N., R.K., T.G., I.W., H.D.M., A.M.A., S.H., L.C.R. and L.S. Visualization: H.-H.C., A.M. and T.K. Funding acquisition: A.M. and T.K. Supervision: A.M. and T.K. Writing, original draft: H.-H.C., A.M. and T.K. Writing, review and editing: A.M. and T.K.

Corresponding author

Correspondence to Thomas Korn.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Immunology thanks Adrian Liston, Scott Zamvil and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available. Primary Handling Editor: Nick Bernard, in collaboration with the Nature Immunology team.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Stress-tolerant Treg cells populate distinct niches in the chronically inflamed CNS.

EAE was induced in Foxp3 (GFP) mice, and on day 45 after MOG_35-55 immunization, CNS tissue was collected, and serial sections of the brain were analyzed for the presence of CD4⁺Foxp3⁺ T cells (a). Representative coronal section, scale bar, overview 500 μm; scale bars, high power magnifications 50 μm. (b, c) Absolute number of Treg cells per coronal section in relation to the distance from the bregma, around the third ventricle (b) or the meningeal space (c). (d) Visualization of the Treg cell hotspot in the third ventricle area in coronal mouse brain atlas schemes (according to Paxinos and Franklin, The mouse brain in stereotaxic coordinates). (e) Quantification of Foxp3⁺ Treg cells as a fraction of all CD4⁺ T cells in the third ventricle area and the meningeal space. (b, c, e) Sample size: n = 3 biological replicates.

Source data

Extended Data Fig. 2 Transcriptomic analysis of stable and unstable CNS Treg cells in chronic EAE.

In vivo Foxp3 fate mapping was conducted in Foxp3^Cre-ERT2 x Rosa26^LSL-tdTomato mice with MOG-induced EAE. Mice were injected with tamoxifen (i.p.) for 5 days at the EAE recovery stage. 14 days after the last tamoxifen injection, CNS cells were harvested and analyzed by flow cytometry. (a) Stable Treg cells (Foxp3 (GFP)⁺tdTomato⁺) and unstable Treg cells (Foxp3 (GFP)⁻tdTomato⁺) were sorted and analyzed by bulk RNA sequencing (n = 5 biological replicates). Volcano plots showing only genes with annotations to the GO terms ‘GO:0009986 cell surface’ and ‘GO:0003700 DNA binding transcription factor activity’ that exhibit differential expression (≥1.5-fold, adjusted P value < 0.05) between stable and unstable Treg cells. DEGs are highlighted in orange. (b) Heatmap showing the top 50 DEGs (≥1.5-fold, adjusted P value < 0.05) between stable and unstable Treg cells. Adjusted P values are derived from DESeq2 with a two-tailed Wald’s test and correction for multiple testing using the Benjamini and Hochberg method. (c) Gene set enrichment analysis of MSigDB Hallmark gene sets comparing stable and unstable CNS Treg cells.

Extended Data Fig. 3 Treg cells maintain immune homeostasis in the chronically inflamed CNS.

(a) EAE was induced in Foxp3 (GFP) and Foxp3-DTR mice. Diphtheria toxin (DTx) was administered intracerebroventricularly (i.c., 15 ng) or intraperitoneally (i.p., 500 ng) during the recovery phase of EAE. Mean EAE scores + SEM (DTx i.c. ctrl n = 3, DTx i.p. n = 5, DTx i.c. n = 6 biological replicates). (b) Survival of mice over 10 days following DTx injection. Survival curves were analyzed using the Mantel–Cox test (DTx i.c. ctrl n = 6, DTx i.p. n = 3, DTx i.c. n = 10 biological replicates), and corresponding individual EAE scores of mice are plotted in (c).

Source data

Extended Data Fig. 4 Transcriptomic analysis of spleen and CNS Treg cells in EAE peak and recovery stages.

Volcano plots displaying differentially expressed genes (≥1.5-fold, adjusted P value < 0.05) between recovery (Rec) and peak CNS Treg cells (a), peak CNS and peak spleen (SPL) Treg cells (b), and recovery CNS and SPL Treg cells (c). Adjusted P values in (a-c) are derived from DESeq2 using a two-tailed Wald’s test with correction for multiple testing using the Benjamini and Hochberg method (peak n = 4, recovery n = 5 biological replicates). Labeled genes comprise a selection of genes commonly associated with non-nonlymphoid tissue Treg cells (black: not significantly regulated, orange: DEG, light grey: not significant and not part of the selection of tissue Treg cell genes). (d) Summary of gene set enrichment analyses comparing recovery and peak CNS Treg cells using the MSigDb Hallmark gene set collection. (e, f) Enrichment plot of genes part of the ‘TNFα-signaling via NFκB’ (e) and the ‘IL-2 STAT-5 signaling’ signature (FDR, false discovery rate; NES, normalized enrichment score) (f). (g-h) Heatmap and hierarchical clustering of significantly upregulated genes (≥1.5-fold, adjusted P value < 0.05) that contribute to the leading-edge genes for the gene sets ‘TNFα-signaling via NFκB’ (g) and ‘IL-2 STAT-5 signaling’ (h) in recovery (Rec) vs. peak CNS Treg cells.

Extended Data Fig. 5 Transcriptomic analysis of VAT and naive CNS Treg cells.

(a) Volcano plot of differentially expressed genes (≥1.5-fold, adjusted P value < 0.1) of recovery (Rec) CNS Treg cells vs. VAT Treg cells. Labeled genes comprise a selection of genes commonly associated with a non-nonlymphoid tissue Treg cell phenotype (black: not significantly regulated, orange: selected DEG. Adjusted P values are derived from DESeq2 using a two-tailed Wald’s test with correction for multiple testing using the Benjamini and Hochberg method (VAT Treg cells n = 3, recovery CNS Treg cells n = 4 biological replicates) (b) Summary of gene set enrichment analyses comparing recovery (Rec) CNS Treg cells from EAE mice and naive CNS Treg cells using the MSigDb Hallmark gene set collection (n = 4 biological replicates). (c, d) Heatmap and hierarchical clustering of genes that contribute to the leading-edge genes for the gene sets ‘IL-2 STAT5 signaling’ (c) and ‘Oxidative phosphorylation’ (d) in Rec vs. naive CNS Treg cells.

Extended Data Fig. 6 EAE in CD38-deficient mice and MBMC mice.

(a) EAE was induced in wild-type (Cd38^+/+), Cd38^+/−, and Cd38^−/− mice. Mean EAE scores + SEM. P values were calculated using a two-way ANOVA with Tukey’s post-test (Cd38^+/+ n = 6, Cd38^+/− n = 4, Cd38^−/− n = 5 biological replicates). (b) Bone marrow from Foxp3-DTR mice (CD45.1/2) mixed 1:1 with either Cd38^+/+ (CD45.2) or Cd38^−/− bone marrow (CD45.2) was transferred into sublethally irradiated Rag1^−/− recipients. EAE was induced in bone marrow-reconstituted mice. Diphtheria toxin (DTx) was administered intracerebroventricularly (i.c., 15 ng) in stable EAE mice on 28 days post MOG_35-55-immunization. Mean EAE scores + SEM (Cd38^+/+ MBMC n = 4, Cd38^−/− MBMC n = 5 biological replicates).

Source data

Extended Data Fig. 7 Systemic Tconv and Treg cells are not affected by the injection of NAD⁺ into the lateral ventricle.

EAE was induced in Foxp3 (GFP) mice. NAD⁺ or PBS was administered intracerebroventricularly (i.c., 100 µg) with or without anti-ARTC2.2 injection (i.p., 50 µg) during the recovery stage of EAE. Mice were euthanized 1 day after treatment. Spleen (SPL) cells were isolated, and Annexin V and LD violet staining were performed on Tconv (CD4⁺Foxp3⁻) and Treg cells (CD4⁺Foxp3⁺), followed by flow cytometric analysis (a). (b) Percentage of Annexin V⁺ and (c) total counts of viable (Annexin V⁻LD violet⁻) splenic Tconv and Treg cells. Symbols in (b, c) represent individual mice (n = 4 biological replicates).

Source data

Extended Data Fig. 8 Antigen-specific Tconv and Treg cells express higher levels of CD25.

EAE was induced in Foxp3 (GFP) mice, and CNS cells were collected during the EAE recovery stage. (a) The percentage of CD25 expression was determined in MOG-I-Ab tetramer-binding Tconv and Treg cells. An OVA_323-339-I-Ab tetramer was used as a control tetramer. CD25 expression (MFI) in MOG-I-Ab⁻ vs. MOG-I-Ab⁺ Tconv (b) and Treg cells (c). Symbols in (b, c) represent individual mice. Data are means + s.d. P values were calculated using a paired two-tailed Student’s t-test (n = 8 biological replicates).

Source data

Supplementary information

Reporting Summary (download PDF )

Peer Review File (download PDF )

Supplementary Table 1-4 (download XLSX )

Supplementary Tables 1–4.

Source data

Source Data Fig. 1 (download XLSX )

Numerical data (cell numbers, cell fractions and EAE scores).

Source Data Fig. 2 (download XLSX )

Numerical data (cell numbers, cell fractions and EAE scores).

Source Data Fig. 3 (download XLSX )

Numerical data (cell numbers, cell fractions and EAE scores).