DJ-1 depletion slows down immunoaging in T-cell compartments

Decline in immune function during aging increases susceptibility to different aging related diseases. However, the underlying molecular mechanisms, especially the genetic factors contributing to imbalance of naïve/memory T-cell subpopulations, still remain largely elusive. Here we show that loss of DJ-1 encoded by PARK7/DJ-1, causing early-onset familial Parkinson’s disease (PD), unexpectedly delayed immunoaging in both human and mice. Compared with two gender-matched unaffected sibling carriers of similar ages, the index PD patient with DJ-1 deficiency showed a decline in many critical immunoaging features, including almost doubled frequencies of non-senescent T cells. The observation of a ‘younger’ immune system in the index patient was further consolidated by the results in aged DJ-1 knockout mice. Our data from bone marrow chimera models and adoptive transfer experiments demonstrated that DJ-1 regulates several immunoaging features via hematopoietic-intrinsic and naïve-CD8-intrinsic mechanisms. Our finding suggests an unrecognized critical role of DJ-1 in regulating immunoaging, discovering a potent target to interfere with immunoaging- and aging-associated diseases.

receptors (KIRs), killer cells lectin-like receptors (KLRs) and exhaustion markers, e.g., KLRG1, CD85j, LAG3, TIM3 and several cyclin-dependent kinase inhibitors including p21, known to increase during aging, have all been decreased substantially in the patient with a homozygous DJ-1 mutation relative to the heterozygous carriers ( Fig. 1g-j). During aging, the TCR repertoire diversity decreases in naïve T cells (Ahmed et al. 2009, Britanova et al. 2014. In line with other observations, the TCR repertoire diversity, which is positively correlated with iChao1 index (Fig. 1k) and richness (Fig. 1l), but negatively correlated with sample clonality index (Fig. 1m), was increased in sorted naïve CD8 T cells (CD8 Tn) of the index patient. This also held true for the TCR repertoire diversity in naïve CD4 T cells (CD4 Tn, Extended Data Fig. 1h-j). It is well accepted that some chronic infections, especially CMV, markedly accelerate immunoaging (Brunner et al. 2011). We therefore applied serological testing for CMV IgG antibody titers and found that all the three subjects were CMV positive (Fig. 1n). Furthermore, there was no clinical evidence for a specific susceptibility to chronic diseases or repetitive infections of all the three siblings. No increase or decrease trend was observed in systematic levels of relevant pro-inflammatory cytokines (Ferrucci and Fabbri 2018) in the index patient versus the two unaffected siblings (e.g., low levels of plasma IFNg, TNFa, IL6 and undetectable or below fit curve range for IL1b, IL4, IL5, IL17a and IL10 among all the three siblings) (Extended Data Fig. 1k-m).
Thus, these data encouraged us to believe that the reduced immunoaging observed in the index patient was driven by DJ-1 deficiency, but not simply due to CMV infection or other chronic infectious diseases or systematic inflammation from either siblings.
DJ-1 loss-of-function mutations are a rare cause of monogenic PD (Pankratz et al. 2006) and we were unable to identify further patients with PARK7-related PD available for biosampling in our extended networks . Therefore, to obtain more statistic power and mechanistic insights, we analyzed wholebody DJ-1 knockout (KO, for simplicity 'whole-body' will be left out afterwards unless different lines) mice.
Here we examined relevant features of immunoaging in DJ-1 KO mice, which were developed elsewhere to study the impact of DJ-1 on neurodegenerative diseases, i.e., PD . As a hallmark of natural aging, memory T cells increase while their counterpart, Tn, decrease (Nikolich-Zugich 2008). However, we detected a significantly higher frequency of CD8 naïve T cells (Tn, CD44 low CD62L high ) accompanied by a lower frequency of effector memory (CD44 high CD62L low , Tem) in aged (~45 wks) DJ-1 KO mice versus ageand gender-matched WT (Fig. 2a-c). A tendency for reduced frequency of central memory CD8 T cells (CD44 high CD62L high , CD8 Tcm) was also observed in aged DJ-1 KO mice (p=0.07, Extended Data Fig. 2a).
No change in the memory/naïve compartments of CD8 T cells was observed in young adult (8-15 wks, simplified as 'young' later on) DJ-1 KO mice ( Fig. 2a-c, Extended Data Fig. 2a). Although immunosenescence does not fully overlap with exhaustion(Akbar and Henson 2011), a considerable intersection exists between these two age-related immunophenotypes and their functional consequences. In contrast to the increased PD-1 expression in T cells during natural aging (Lee et al. 2016), we observed a significantly lower expression of the key exhaustion marker PD-1 among homeostatic total CD8 T cells (Fig.   2d) and different CD8 subsets (Extended Data Fig. 2b) in aged, but not young DJ-1 KO versus WT mice. This observation indicates that the exhaustion process might slow down in aged DJ-1 KO mice. In addition to exhaustion, PD-1 is also regarded as one of the key activation markers of T cells (Ahn et al. 2018). Therefore, to determine whether our observation is based on either reduced activation in general or cell exhaustion, we also assessed proliferation markers, e.g., Ki-67 in different T cell subsets under homeostatic conditions.
Although the frequency of Ki-67 + cells among total CD8, Tn and Tcm was significantly decreased (Fig. 2e, f), the homeostatic proliferation of CD8 Tem was significantly augmented (Fig. 2f). Furthermore, the expression of PD-1 and Ki-67 showed a largely mutually-exclusive pattern among CD8 Tem (Fig. 2g). All these indicate that reduced PD-1 expression in the aged DJ-1 KO mice was not simply attributable to a general reduction in proliferation and activation under homeostatic conditions. Relevant cytokines, e.g., IFNg, increase among CD8 T cells during natural aging (Bandres et al. 2000). Therefore, we also analyzed IFNg production in CD8 T cells. Conforming to the other delayed immunoaging phenotypes, IFNg was decreased in CD8 T cells of aged DJ-KO versus WT mice following in vitro stimulation (Fig. 2h). Consistent with the observations in the index PD patient with DJ-1 deficiency, our results show that DJ-1 KO mice display reduced immunoaging in CD8 T-cell compartments.
T-cell-conditional deletion of the mitochondrial transcription factor A (Tfam), suppressing mitochondria DNA content, accelerates inflammaging (Desdin-Mico et al. 2020) and aging impairs mitochondrial homeostasis (Picca et al. 2018). To check whether our observation in cellular phenotypes can be extended to the organelle level we measured mitochondrial mass and membrane potential in different T cell subsets. As expected, both mitochondrial mass and membrane potential of different CD8 T-cell subsets (CD8 Tn, Tem and Tcm) were declined in aged WT mice versus young WT mice ( Fig. 2i-l, Extended Data Fig. 2c,d). Notably, in both CD8 Tn and Tem from aged mice, the mitochondrial mass and membrane potential were significantly enhanced in DJ-1 KO versus WT mice ( Fig. 2i-l). Still similar to other CD8 subsets, CD8 Tcm mitochondria mass (p=0.07) and membrane potential (p=0.09) showed a tendency to increase in aged DJ-1 KO versus WT mice (Extended Data Fig. 2c,d). Consistent with other cellular data in young mice, significant difference was observed in neither mitochondrial mass nor membrane potential of different CD8 T-cell subsets between young DJ-1 KO and WT mice ( Although CD8 T cells are more susceptible to aging related changes (Czesnikiewicz-Guzik et al. 2008), we have also observed delayed immunoaging features in CD4 T cells of the index patent. Therefore, we also analyzed the murine CD4 T-cell compartments. As expected, similar effects were observed in CD4 T cells of aged DJ-1 KO mice (Extended Data Fig. 3a-h). To gain a more comprehensive picture, we performed transcriptomic analysis of CD4 + CD25conventional T cells (Tconv) sorted from aged mice under homeostatic conditions. In accordance with the enhanced proportions of naïve T cells, our pathway enrichment analysis showed that several pathways involved in T-cell receptor signaling pathway and positive regulation of cell differentiation were significantly affected among the downregulated genes in Tconv of aged DJ-1 KO versus WT mice (Extended Data Fig. 3l). Notably, many memory T-cell or aging-related markers (Mogilenko et al. 2021) and/or memory T-cell development-related genes were also among the downregulated genes (Extended Data Fig. 3l). These results show that DJ-1 depletion also causes the delayed immunoaging in CD4 T-cell compartments.
Since DJ-1 is a multi-functional protein ubiquitously expressed in different type of tissues and cells (Wilson 2011) and immunoaging involves many types of immune and non-immune cells (Nikolich-Žugich 2018), we next asked whether the delayed immunoaging in aged DJ-1 KO mice is the result of a hematopoietic-intrinsic or non-hematopoietic regulation. To this end, we generated mixed bone marrow (BM) chimeras by transferring BM cells mixed from young CD45.1 DJ-1 WT and CD45.2 DJ-1 KO donor mice into lethally-irradiated young WT recipients (Extended Data Fig. 4a). In this experiment, both DJ-1 KO and WT CD8 T cells developed under the same host WT environmental conditions. If the difference observed between that of KO and WT origins is consistent with the alteration in aged DJ-1 KO versus WT mice, it is attributable to hematopoietic intrinsic regulations. Interestingly, we indeed observed some hematopoietic-intrinsic phenotypes for the DJ-1-mediated immunoaging role. For instance, the expression of the critical immunoaging marker KLRG1 among blood CD8 T cells of DJ-1-KO origin was already significantly lower than that of DJ-1-WT-derived cells (Extended Data Fig. 4b) and also showed a trend to lessen in spleen (p=0.08, Extended Data Fig. 4c).
This clearer phenotype in blood is in accordance with the notion that Tem are mainly distributed in nonlymphoid tissues (Masopust et al. 2001) and KLRG1 is mainly expressed among CD8 Tem. The exhaustion marker PD-1, which increases during aging, also showed a similar change as did KLRG1 (Extended Data   Fig. 4d). The frequency of CD8 Tem was significantly lower in DJ-1-KO-derived cells (Extended Data Fig.   4e,f). Consistent with the data in aged DJ-1 KO mice, the ratios between CD8 Tn and Tem were already higher in the DJ-1-KO-derived cells than that from WT-origin cells (Extended Data Fig. 4g,h). These data indicate that DJ-1 exhibits a hematopoietic-intrinsic role for regulating not only the expression of immunoaging related markers, but also the proportion of CD8 Tem and the ratio between CD8 Tn and Tem.
However, the percentages of CD8 Tn did not show any significant difference between lymphocytes developed from young CD45.2 and CD45.1 BM cells in both spleen and blood (Extended Data Fig. 4i,j). Moreover, the percentages of CD8 Tcm derived from DJ-1 KO BM cells were higher than that from DJ-1-WT origin within young WT recipients (Extended Data Fig. 4k,l). Considering the observation in aged DJ-1 KO mice (Extended Data Fig. 2a), our results from young-donor-BM transplantation show that DJ-1 deficiency mediates accumulation of CD8 Tn and Tcm via either a hematopoietic-extrinsic or an aging-dependent manner or in combination of both. These mixture phenotypes, especially those non-intrinsic related observations, from the young-donor-BM chimeras urged us to investigate the potential effects of aging and non-hematopoietic aspects on the DJ-1-mediated immunaging.
To study the effects aforementioned, we generated mixed BM chimeras by transferring aged CD45.1 DJ-1 WT BM cells and CD45.2 DJ-1 KO BM cells into irradiated young WT or KO recipients (Fig. 3a). Due to the ethic limitation, we were unable to use aged mice as recipients. Similar to that in young-donor-BM chimeras (Extended Data Fig. 4b-f), the percentages of cells expressing the immunoaging related markers, e.g., KLRG1 ( Fig. 3b,c) and PD-1 (Fig. 3d) as well as the frequency of CD8 Tem (Fig. 3e) were lower among CD8 cells reconstituted from aged DJ-1 KO versus WT BM donors, independent from the types of recipients, again supporting a hematopoietic-intrinsic mechanism. The ratios between CD8 Tn and Tem developed from aged DJ-1 KO versus WT BM donors were significantly higher in young KO recipients or with a tendency to increase also in young WT recipients (p=0.08, Fig. 3f). These consistent data from both young-and ageddonor-BM chimeras together suggests that DJ-1 exhibits an aging-BM-independent, but hematopoieticintrinsic role for regulating the expression of KLRG1 and PD-1 among CD8 T cells, the frequency of CD8 Tem as well as the ratio between CD8 Tn and Tem.
Interestingly, the mixed BM chimeras did not show a significant difference in the frequency of CD8 Tn developed from aged DJ-1 KO versus WT BM donors as long as the recipients were WT mice, no matter from young (Extended Data Fig. 4i,j) or aged donors (Fig. 3g). Notably, consistent with the observation in aged DJ-1 KO versus WT mice, following reconstitution within young DJ-1-KO, but not -WT recipients, the proportion of blood CD8 Tn developed from aged DJ-1 KO versus WT BM cells was already significantly higher ( Fig. 3g). At the same time, no significant change was observed in the frequency of CD8 single positive cells in thymus among aged DJ-1-KO versus -WT origins, within either type of young DJ-1 KO or WT recipients (Extended Data Fig. 5a), ruling out an abnormal thymic development of matured CD8 T cells.
These results suggest that the enhanced frequency of CD8 Tn in aged DJ-1 KO versus WT mice requires the involvement of DJ-1-deficient non-hematopoietic cells. Opposed to the observations in aged DJ-1 KO versus WT mice, the percentages of CD8 Tcm of aged DJ-1-KO versus -WT origin were higher within both types of young recipients (Extended Data Fig. 5b,c). The results about CD8 Tcm percentages were consistent between young-and aged-BM models, regardless of within WT or KO recipients. These together essentially suggest the involvement of DJ-1-deficient aging microenvironmental factors in reduced accumulation of CD8 Tcm within aged DJ-1 KO mice. In short, our data firmly demonstrate that DJ-1 depletion enhances the ratio between CD8 Tn and Tem and inhibits the expression of immunoaging related markers in a hematopoieticintrinsic but aging-BM-independent way, while regulating the CD8 Tn and Tcm accumulation in a more complicated manner.
The immunoaging process in T-cell compartments involves many types of cells (Nikolich-Žugich 2018) and BM-derived cells include far more than T cells. Previous data has shown that the homeostatic proliferation of CD8 Tn is enhanced during natural aging (Rudd et al. 2011) and the homeostatic proliferation of CD8 Tn drives the differentiation into CD8 Tem in Rag1 -/lymphopenic mice (Cho et al. 2000). In aged DJ-1 KO versus WT mice, the homeostatic proliferation of CD8 Tn was reduced (Fig. 2f). Therefore, we hypothesized that this reduced homeostatic proliferation of CD8 Tn might impede or dysregulate the development into Tem cells and consequently the aging process of T cells. To test whether DJ-1 depletion has a T-cell-intrinsic effect on the transition from CD8 Tn into Tem subsets, we performed adoptive transfer of CD8 Tn sorted from young or aged DJ-1 KO or WT mice into young Rag1 -/recipients, where the development of CD8 Tem is radically accelerated under lymphopenia (Cho et al. 2000) (Fig. 3h). Remarkably, the essential murine immunoaging marker KLRG1 was significantly lower in total CD8 T cells developed from aged DJ-1 KO vs WT donor cells (Fig. 3i,j). This phenomenon in total CD8 T cells was mainly due to the fact that from aged donor cells, CD8 T cells mainly consisted of CD8 Tem in this model (Extended Data Fig. 6a-c), where KLRG1 was significantly reduced in KO vs WT-originated cells (Fig. 3k). Furthermore, the frequency of the doublepositive cells expressing both KLRG1 and PD1 among total CD8 T cells and CD8 Tem was significantly lower in aged DJ-1 KO CD8-Tn-originated cells (Fig. 3l,m). In line with the reduced expression of key immunoaging markers, the expression of the T-cell activation marker CD69 was significantly increased in total CD8 T cells and CD8 Tem derived from CD8 Tn of aged DJ-1-KO versus -WT origins (Extended Data Fig. 6d,e).
Nevertheless, the frequency of CD8 Tn, Tem and Tcm in the Rag1-null recipients was not different in both comparisons of CD8 T cells from young DJ-1 KO versus -WT origins as well as that from aged DJ-1-KO versus -WT origins (Extended Data Fig. 6a-c), indicating the existence of CD8-Tn-extrinsic factors and/or requiring aging-microenvironmental factors. Nevertheless, the observed CD8-Tn-intrinsic role for DJ-1 in regulating key immunoaging related hallmarks in CD8 T cells, at least already partially, contribute to the delayed immunoaging phenotype in DJ-1 KO mice.
Our findings reveal an unexpected causal link between deficiency in a key monogenic PD gene PARK7/DJ-1 and delayed immunoaging in the T cell compartments. The data consistency between both human and mice with DJ-1 deficiency support a highly potent strategy to interfere with immunoaging for various complex diseases and infectious diseases, including COVID-19, where immunoaging is believed to be crucial (Koff and Williams 2020). Understanding the detailed molecular mechanisms through which DJ-1 regulates immunoaging still requires further investigation using cell-type-specific conditional DJ-1 KO mice. Our work also offers a unique animal model with delayed immunoaging phenotypes to allow researchers to explore the potential roles of a relatively juvenile immune system in the immune and aging associated diseases.

Materials and Methods
Methods, list of materials used in this work and any related references are provided in Supplementary Information.

Supplementary Information
In the initial submission, it is attached in the end of the pdf file. It will be available in the online version of the paper. . The half black square/circle indicates male/female individuals heterozygous for the DJ-1 mutation, the black square indicates the patient carrying the homozygous DJ-1 mutation. *Obligate heterozygous mutation carrier although dead at the time of the initial study. b, Expression of CD27 and CD28 on peripheral blood CD8 T cells of three participants. c, d, e, f, Expression of CD57 (c), PD-1 (d), EOMES (e) and T-bet (f) on peripheral blood CD8 T cells of three participants. The enlarged number in the corresponding gate represents the corresponding percentage out of the parent population. g-j, mRNA expression of senescence related genes (g), exhaustion related genes (h), KIR or KLR genes (i) and cyclin-dependent kinase inhibitor genes (j) in sorted CD8 T cells of the peripheral blood from three participants. k, l, Comparison of the lower bound of TCR-beta repertoire (k) and richness (l) of naïve CD8 T cells among three participants. m, The sample clonality index of TCR repertoire of naïve CD8 T cells among three participants. n, Serology detection of CMV infection for three participants. The threshold of 11 was decided according to the manufacture instruction. a, Representative flow-cytometry plots of CD44 and CD62L expression on total CD8 T cells of aged DJ-1 KO and age and gender-matched WT mice. b, c, Percentages of CD44 low CD62L high (Tn) (b) and CD44 high CD62L low (Tem) (c) cells among total CD8 T cells of spleen and pLNs from young and aged DJ-1 KO and WT littermates. (young KO, n=5; young WT, n=5; aged KO, n=8; aged WT, n=6; for aged mice, data pooled from 2 independent experiments). d, Representative histogram overlay of PD-1 expression among total CD8 T cells in spleen of aged mice (left panel) and percentages of PD-1 + cells among total CD8 T cells (right panel). e, Percentages of Ki-67 + cells among total CD8 T cells. f, Percentage of Ki-67 + cells among splenic CD8 Tn, Tem and Tcm in aged DJ-1 KO and age-and gender-matched WT mice. g, Representative flow-cytometry plots of Ki-67 and PD-1 among splenic CD8 Tem of aged mice. h, IFNγ production in CD8 T cells of spleen and pLNs after in vitro stimulation using 50 ng/ml of PMA and 750 ng/ml of ionomycin for 5 h. i, j, k, l, Comparison of splenic CD8 Tn mitochondrial mass (mito, Mitotracker green MF, i) and mitochondrial potential (MitoTracker Deep Red, j) of young and aged DJ-1 KO and WT mice. Comparison of CD8 Tem mitochondrial mass (k) and mitochondrial potential (l) of young and aged DJ-1 KO and WT mice. SP and LN represents spleen and lymph nodes, respectively. Results represent at least four (b-h) or three (i-l) independent experiments. Data are mean± s.d. The P-values are determined by a two-tailed non-paired Student's t-test. ns or unlabeled, not significant, *P<=0.05, **P<=0.01 and ***P<=0.001. b,d, Percentages of KLRG1 + (b) and PD-1 + (d) CD8 T cells derived from aged DJ-1 KO and WT BM cells within young DJ-1 KO or WT recipients. c, Percentages of KLRG1 + among CD8 Tem derived from aged DJ-1 KO and WT BM cells within young DJ-1 KO or WT recipients. e,g, Percentages of CD8+ CD44 hi CD62L low (Tem) cells (e) and CD8+ CD44 low CD62L hi (Tn) cells (g) derived from aged DJ-1 KO and WT BM cells within young DJ-1 KO or WT recipients. f, Ratios between CD8 Tn and Tem cells developed from two types of BM cells within young DJ-1 KO or WT recipients. h, Schematic of the experimental setup of CD8 Tn adoptive transfer. 2.5E5 naïve CD8 T cells isolated from young or aged DJ-1 KO and WT littermates were injected into Rag-1 deficient mice by i.v. injection. 6 weeks later, mice were sacrificed for FACS analysis. i, Representative flow cytometry data of KLRG1 + population among total CD8 T (upper) and CD8 Tem cells (lower) following adoptive transfer of CD8 Tn from aged mice. j, l, Percentage of KLRG1 + (j) and KLRG1 + PD-1 + (l) population among total CD8 T cells. k, m, Percentage of KLRG1 + (k) and KLRG1 + PD-1 + (m) population among CD8 Tem cells. Results from BM transfer and adoptive transfer of CD8 Tn represent two independent experiments. Data are mean± s.d. The P-values are determined by a two-tailed paired (b-g) or non-paired (j-m) Student's t-test. ns or unlabeled, not significant, *P<=0.05, **P<=0.01 and ***P<=0.001.

Extended Data Figures:
Extended Data Figure 1. The DJ-1-devoid index patient showed delayed immunoaging features also in CD4 T cells. a, Percentages of total CD4 and CD8 T cells among the living lymphocyte singlets of peripheral blood of three participates [P1 (heterozygous mutation), P2 (homozygous mutation) and P3 (heterozygous)]. b, c, Coexpression of CCR7 and CD45RO (b), CD27 and CD45RO (c) on peripheral blood CD4 T cells of three participants. d, e, f, Expression of CD57 (d), PD-1 (e), and T-bet (f) on peripheral blood CD4 T cells of three participants. g, Frequency of FOXP3+CD4+ Tregs among total CD4 T cells in the peripheral blood from three participants. h, i, Comparison of the lower bound of TCRbeta repertoire (h) and richness (i) of sorted naïve CD4 T cells of three participants. j, The sample clonality index of TCR repertoire of naïve CD4 T cells of three participants. k, l, m, Cytokine measurement of IFNγ (k), TNF-α (l) and IL-6 (m) in the plasma of three participants. The other five tested cytokines were either undetectable or below fit curve. For cytokine measurement, data are mean± s.d.
Extended Data Figure 2. Extended characterization of delayed immunoaging in CD8 Tcell compartments. a, Percentages of CD44 high CD62L high cells (Tcm) among total CD8 T cells of spleen and pLNs from young and aged DJ-1 KO and WT littermates. b, Percentage of PD-1 + cells among splenic CD8 Tn, Tem and Tcm subsets in the aged DJ-1 KO and age-and gender-matched WT mice. c, d, Comparison of CD8 Tcm mitochondrial (mito) mass (c) and mitochondrial potential (d) of young and aged DJ-1 KO and WT mice. Results represent at least four (a, b) and three (c, d) independent experiments. Data are mean± s.d.. The P-values are determined by a two-tailed non-paired Student's ttest. n.s. or unlabeled, not significant, *P<=0.05, **P<=0.01 and ***P<=0.001.

among total CD4 T cells of spleen and pLNs from young and aged DJ-1 KO and WT littermates. d, Representative histogram overlay of PD-1 expression among total CD4 T cells in spleen of aged mice (left panel) and percentages of PD-1 + cells among total CD4 T cells (right panel). e, Representative histogram overlay of CTLA-4 expression among total CD4 T cells in spleen of aged mice (left panel) and percentages of CTLA-4 + cells among total CD4 T cells (right panel). f, Percentages of Ki-67 + cells among total CD4 T cells. g, IFNγ production in CD4 T cells of spleen and pLNs
after in vitro stimulation using 50 ng/ml of PMA and 750 ng/ml of ionomycin for 5 h. h, i, Comparison of naive CD4 (Tn) mitochondrial (mito) mass (h) and mitochondrial (mito) potential (i) of young and aged DJ-1 KO and WT mice. j, k, Comparison of CD4 Tem mitochondrial mass (j) and mitochondrial potential (k) of young and aged DJ-1 KO and WT mice. l, The selected significantly enriched GO-terms and pathways among the downregulated genes in CD4 Tconv cells from aged DJ-1 KO mice versus the age-matched WT littermates from microarray analysis. m, Volcano plot shows both downregulated and upregulated differentially expressed genes in splenic CD4 T cells from three aged DJ-1 KO mice versus three age-matched WT littermates. SP and LN represents spleen and lymph nodes, respectively. Results represent at least four (b-g) and three (h-k) independent experiments. Data are mean± s.d. The P-values are determined by a twotailed un-paired Student's t-test. n.s. or unlabeled, not significant, *P<=0.05, **P<=0.01 and ***P<=0.001.
Extended Data Figure 4. DJ-1 ablation regulates KLRG1 and PD-1 expression as well as the ratios between CD8 Tn and Tem in a hematopoietic-intrinsic manner but the accumulation of CD8 Tn and Tcm in a complicated manner. a, Schematic of the experimental setup of bone marrow transplantation. A total of 10E6 of bone marrow cells from young DJ-1 KO mice (CD45.2 + ) and WT mice (CD45.1 + ) (1:1 mix) were transferred into lethally-irradiated young WT recipients (CD45.2 + ) by i.v. injection. Mice stably engrafted with donor cells were sacrificed for FACS analysis later. b, c, Percentage of KLRG1 + CD8 T cells derived from young DJ-1 KO and WT donor BM cells in blood (b) and spleen (c) within young WT recipients. d, Percentage of PD-1 + cells among total CD8 T cells derived from young DJ-1 KO and WT BM cells in spleen within young WT recipients. e, f, Percentages of CD8 Tem derived from young DJ-1 KO and WT BM cells in blood (e) and spleen (f) within young WT recipients. g, h, Ratios between CD8 Tn and Tem cells developed from CD45.1 (WT) or CD45.2 (KO) BM cells in blood (g) and spleen (h) within young WT recipients. i, j, Percentages of CD8 Tn in blood (i) and spleen (j) derived from young DJ-1 KO and WT BM cells within young WT recipients. k, l, Percentage of CD8 Tcm among total CD8 T cells derived from young DJ-1 KO and WT BM cells in blood (k) and spleen (l) of young WT recipients. Results represent two independent experiments. The P-values are determined by a two-tailed paired Student's t-test. n.s. or unlabeled, not significant, *P<=0.05, **P<=0.01 and ***P<=0.001.

Intracellular cytokine quantification
For intracellular cytokine measurement, cells (2E5) from spleen or pLNs were stimulated by 50 ng/ml PMA (Phorbol 12-myristate 13-acetate, Sigma-Aldrich, P8139) and 750 ng/ml ionomycin I0634) in the presence of Golgiplug (BD Biosciences, 555029) and Golgistop (BD Biosciences, 554724) for 5 h in 96-well plates. Following cell surface staining, cells were fixed and permeabilized with Cytofix/Cytoperm buffer (BD Biosciences, 554714). The cytokine antibodies diluted in permeabilized buffer were added and incubated at 4 °C for 30 min protected from light. Cells were acquired on a BD LSRFortessa TM and analyzed by Flowjo v10.

Bone marrow transplantation
Bone marrow (BM) cells pooled from the femurs and tibias of DJ-1 -/mice and age-, gender-matched CD45.1 B6 mice were isolated and 1:1 mixed in 100ul of cold PBS solution (Ca2+ and Mg2+ free; Lonza, BE17-516F). Gender-matched DJ-1 KO or WT mice aged 8-12 weeks as recipients were lethally irradiated from a gamma source (RS2000 X-Ray Biological Irradiator from Rad Source Technologies, two doses of 450 rads with 3 h resting period between the two doses) and 6 h later received 10E6 mixed donor BM cells by intravenous injection. For BM chimeras generated from young donor mice, blood was sampled at six and eight weeks post transplantation. Spleen was analyzed at eight weeks after transplantation. For BM chimeras generated from aged donor mice, both blood and spleen were analyzed at four months post transplantation.
Cells were acquired on a BD LSRFortessa TM followed by the analysis with Flowjo v10.
Adoptive transfer of naïve CD8+ T cells CD90.2 + cells from spleen and pLNs of DJ-1 -/mice and DJ-1 +/+ littermates aged 8-12 weeks or ~55 weeks were magnetically isolated and first enriched by using anti-mouse CD90.2 microbeads (Miltenyi Biotec, 130-049-101, also refer to Supplementary Microarray analysis of murine T cells CD25 -CD4 + Tconv cells from around 45-week-old DJ-1 -/mice and DJ-1 +/+ littermates were sorted using BD FACSAria TM III sorter. The cell pellets were immediately lysated with RLT buffer supplemented with 1% beta-mercaptoethanol (Sigma-Aldrich, 63689) and frozen at -80 °C for further analysis. RNA was extracted by using the RNeasy Mini Spin Kit (Qiagen, 74104) and genome DNA was removed. Samples were analyzed via RNA 6000 Pico kit (Agilent, 50671513) by using an Agilent Bioanalyzer 2100, ensuring that only the samples have RIN higher than 8.5 were further used for microarray measurement.
RNA samples were analysed with the Affymetrix mouse Gene 2.0 ST Array at EMBL Genomics core facilities (Heidelberg). The microrray data analysis was performed in the same way as described in our previous work 2 .
To ease the reading of this work, we described the major filtering steps here again. The expression signal at the exon level was summarized by the Affymetrix PLIER algorithm with DABG and PM-GCBG options by means of the sketch-quantile normalization approach (Affymetrix Expression Console v1.4). The corresponding probesets were considered differentially expressed if they passed the following combinatory filters 3 : (a) whether the change folds were >= 2 between the means of DJ-1 KO and WT Tconv; (b) whether the P-value, resulting from a two-tailed Student t-test, was <=0.05; (c) whether the cross-hyb type of the probeset was equal to 1; (d) whether the probeset with the highest expression level was higher than 100 (with the median value of ~90 for each our sample); (e) for a given group (e.g. WT) with the higher mean intensity value of the probeset, whether the probeset in all the replicates of the given group was detected as 'present' according PBMCs were isolated from the patient's blood by Ficoll gradient centrifugation using SepMate tubes (StemCell, 86450, also refer to Supplementary CD4 and CD8 T cells were sorted using the BD Aria III. DNA was extracted from the flash frozen pellets to perform TCR repertoire sequencing. Genomic DNA (gDNA) was extracted from the sorted naïve and memory CD4 and CD8 T cells using the QIAamp DNA Blood Mini Kit (Qiagen, 51104) following the manufacture's instructions. The gDNA was eluted in 55 ul RNase-and DNase-free water to match the volume and concentration requirements for survey analysis of TCR beta repertoire sequencing by ImmunoSEQ (Adaptive Biotechnologies). All the analyses (TCR richness estimation and clonality) were performed using the online tool of ImmunoSEQ Analyzer 3.0. The lower bound of TCR repertoire of CD4 Tn and CD8 by applying nonparametric statistics using the iChao1 estimator. The richness of TCR repertoire of CD4 Tn and CD8 Tn by applying a nonparametric empirical Bayes estimation using the Efron Thisted estimator (extrapolation value is 120K from ImmunoSEQ Analyzer 3.0). The sample clonality of TCR repertoire of CD4 Tn and CD8 Tn (Clonality is equal to 1 -normalized Shannon's Entropy).

CMV ELISA
The CMV seropositivity was measured in the plasma samples of the DJ-mutant patients using the anti-Cytomegalovirus (CMV) IgG Human ELISA Kit (Abcam, ab108724) following the manufacturer's instructions. The plasma was used at a 1:100 dilution.

Data Availability
The microarray data from human CD8 T cells of the DJ-1 mutation carrier family and aged murine CD4 Tcon and enter the token utirsmiaxzyvxkn into the box. Upon acceptance, we will make all those data including single-cell TCR repertoire sequencing data publically available.

Supplementary Tables
Supplementary Table 1. List of mouse-related antibodies used in this work.