Differential effect of DJ-1/PARK7 on development of natural and induced regulatory T cells

Regulatory T cells (Tregs) are essential for maintaining an effective immune tolerance and a homeostatic balance of various other immune cells. To manipulate the immune response during infections and autoimmune disorders, it is essential to know which genes or key molecules are involved in the development of Tregs. Transcription factor Foxp3 is required for the development of Tregs and governs most of the suppressive functions of these cells. Inhibited PI3K/AKT/mTOR signalling is critical for Foxp3 stability. Previous studies have suggested that DJ-1 or PARK7 protein is a positive regulator of the PI3K/AKT/mTOR pathway by negatively regulating the activity of PTEN. Thus, we hypothesised that a lack of DJ-1 could promote the development of Tregs. As a result, loss of DJ-1 decreased the total CD4+ T cell numbers but increased the fraction of thymic and peripheral nTregs. In contrast, Foxp3 generation was not augmented following differentiation of DJ-1-deficient naïve CD4+ T cells. DJ-1-deficient-iTregs were imperfect in replication, proliferation and more prone to cell death. Furthermore, DJ-1 deficient iTregs were less sensitive to pSmad2 and pStat5 signalling but had activated AKT/mTOR signalling. These observations reveal an unexpected differential role of DJ-1 in the development of nTregs and iTregs.


DJ-1 deficiency augments the percentage of nTregs both in the thymus and periphery, but decreases the total CD4 + T cell numbers.
To examine the role of DJ-1 protein in T cell development, we first characterised the development of T cells in thymocytes from DJ-1 −/− mice. The development of T cell receptors (TCRs) in the thymus was similar in both DJ-1 −/− and DJ-1 +/+ mice in the pre-TCR stage before CD4 + and CD8 + thymocyte maturation (data not shown). No significant change was also observed of single positive CD4 or CD8 thymocyte development in both strains of mice (Fig. 1). However, CD4 + CD25 + Foxp3 + thymocytes (thymic derived natural Tregs or nTregs) were significantly more abundant in DJ-1 −/− mice than in DJ-1 +/+ mice as measured by flow cytometry at percentage level (Fig. 1B). Further analysis revealed that due to less thymic cellularity the total number of CD4 + thymocytes and nTregs were significantly reduced in DJ-1 −/− mice compared with control (DJ-1 +/+ ) littermates (Fig. 1C). In search for any defect in expression of various other key Tregs markers, we characterised the nTregs using several known markers such as Helios (Ikzf2), Eos, CTLA-4, GITR, Neuropilin-1 (Nrp-1) in addition to CD25 and Foxp3. As a result, the percentage of DJ-1 −/− nTregs had a significantly higher expression level of Helios, GITR and significantly lower expression of Nrp-1. No significant change was observed in CTLA-4 and Eos in both strains of mice ( Fig. 2 and Suppl. Fig. 1). Again due to a reduced thymic cellularity, total numbers of various marker positive cells were significantly reduced in DJ-1 −/− nTregs compared to control group except for CTLA-4 marker ( Fig. 2 and Suppl. Fig. 1).
In DJ-1 deficient mice iTregs induction, cell cycle progression, cell survival and proliferation are compromised. DJ-1 is a negative regulator of PTEN and thus upregulates AKT and mTOR 12,20 , which have in turn been shown to inhibit the induction of Foxp3 in developing iTregs 10,13,14 . We thus presumed that deficiency of DJ-1 may affect the induction of Foxp3 in T cells and explored further whether deficiency of DJ-1 influences the induction of Foxp3 in developing iTregs. When we differentiated naïve CD4 + T cells into iTregs using TGF-β and IL-2, surprisingly, we found that instead of an increase in the induction of Foxp3, there was significantly reduced expression of Foxp3 in developing iTregs ( Fig. 6A-D). However, naïve CD4 + T cells were left stimulated in the presence of anti-CD3 and anti-CD28 (Th0 cells), and the Foxp3 + cell number was higher in DJ-1 −/− mice than in DJ-1 +/+ mice (Fig. 6A,B). These data suggested that in vitro DJ-1 −/− mice induce less Foxp3 in developing iTregs. To explore the possible cause of reduced induction of Tregs, we examined the cell cycle of Th0 and iTregs in DJ-1 −/− mice by Propidium Iodide (PI). PI staining suggested that DJ-1 Th0 cells had a significantly more percentage of G0-G1 and less percentage of S phase but no change in G2-M phase of the cell cycle (Fig. 6E,F) and similarly iTregs also had a significantly more percentage of G0-G1 and less percentage of S phase of the cell cycle (Fig. 6E,G). These results point to a proliferation defect in the absence of DJ-1 protein. However, it does not explain why DJ-1 −/− mice have less Foxp3 induction. Further, we checked the cell death analysis in iTregs and found that DJ-1 −/− developing iTregs were more prone to cell death compared with DJ-1 +/+ developing iTregs (Fig. 6I). In contrast, no difference between the genotypes was observed in Th0 cells (Fig. 6H). Furthermore, Foxp3 − iTregs were significantly more proliferative in DJ-1 −/− compared with DJ-1 +/+ mice, whereas no statistical difference was observed in Th0 cells between DJ-1 −/− and DJ-1 +/+ mice (Suppl. Fig. 4). These data suggested that in vitro DJ-1 −/− mice have reduced cell cycle progression and more cell death in developing iTregs.
AKT/mTOR signalling is activated whereas TGF-β/IL-2 signalling is less responsive in DJ-1 deficient iTregs. In the preceding section, we have shown that DJ-1 deficiency compromises the production of iTregs. It is known that DJ-1 also modifies the PTEN/mTOR/Rictor 10 , which is activated by TCR and CD28 signalling. Therefore, we explored the signalling pathways in iTregs. To our surprise, we found that PTEN expression was lower in iTregs from DJ-1 −/− than in iTregs from DJ-1 +/+ mice, Difference between DJ-1 +/+ and DJ-1 −/− mice CD4 + T cells splenocytes were subtle but statistically significant (p = 0.036), however, no significant difference was observed between DJ-1 +/+ and DJ-1 −/− CD8 + T splenocytes. Left hand side shows the representative FACS plots for CD4 and CD8 staining and right hand side shows the mean ± SEM (n = 5-7 independent experiments). Data were obtained from 11 biological replicates. (B) Splenocytes were stained with CD4, CD8, CD25 and Foxp3 antibodies and splenocytes were gated on CD4 + T cells. CD4 + T cells were characterised for Tregs markers CD25 and Foxp3. As a result, CD25 + Foxp3 + Tregs are more abundant in DJ-1 −/− mice than in DJ-1 +/+ mice (p = 0.011). Left hand side shows the representative FACS plots for CD25 and Foxp3 staining gated on CD4 + T cells and right hand side shows mean ± SEM (n = 4 independent experiments). (C) Total splenocytes were counted and estimated CD4 + T cells as well as CD4 + CD25 + Foxp3 + T cells numbers, data were derived from 5 biological replicates.
an observation in sharp contrast to other cell types 11,20 (Fig. 7A). To confirm whether we had complete knock-down in DJ-1 deficient iTregs, we performed Western blotting against DJ-1 (Fig. 7A). In addition, both mTOR and Rictor were significantly higher in DJ-1 −/− than in DJ-1 +/+ iTregs (Fig. 7A). As a result of higher Rictor expression, phosphorylation of AKT was statistically higher in DJ-1 −/− than in DJ-1 +/+ iTregs (Fig. 7A,B). These results suggest that iTregs from DJ-1 deficient mice have higher AKT/Rictor signalling pathways which could lead to reduced iTregs development. We also characterised iTregs for TGF-β activated Smad2 protein levels and IL-2 receptor signalling, which could lead to Stat5 activation and to Foxp3 induction by binding to the transcription signalling site 10 . Indeed, phosphorylation of both signalling molecules was significantly less in DJ-1 −/− than in DJ-1 +/+ iTregs (Fig. 7C,D). Collectively, TGF-β and IL-2R signalling was impaired in in DJ-1 −/− as compared with DJ-1 +/+ iTregs leading to reduced Foxp3 induction.
Lack of DJ-1 leads to enhanced ROS production and higher Sgk1 expression. DJ-1 regulates the production of reactive oxygen species (ROS) and acts as an anti-oxidant 18,29 . We measured ROS production utilizing the 2, 7-dichlorodihydrofluorescein diacetate (DCFDA) dye in iTregs by flow cytometry. As illustrated in Fig. 8A, DJ-1 −/− iTregs produce more ROS than DJ-1 +/+ iTregs. Further, we also found that isolated pTregs from DJ-1 −/− mice also produce a subtle but statistically significant amount of ROS (Fig. 8A). We previously showed that ROS can be induced by lipopolysaccharide (LPS) in dendritic cells (DCs) and that AKT1 deficient DCs produce less ROS compared to control DCs in presence of LPS 30 . Thus, we deduced that LPS treatment could enhance ROS and reduce the Foxp3 induction. We treated developing iTregs with LPS and found that LPS treatment could also inhibit the induction of iTregs in both wild type and DJ-1 deficient mice (Fig. 8B), in keeping with recently published findings 31 . It has previously been described that ROS down-regulates NFκ B signalling in the nucleus 32 leading to less Foxp3 induction. To test this hypothesis, we used 50 nM Wogonin (NFκ B inhibitor) 33 , on developing iTregs, we confirmed that Wogonin treatment can reduce Foxp3 induction in DJ-1 deficient mice compared with control animals (Fig. 8C). Additionally, previous studies suggested that oxidative stress increases the expression of Sgk1 mRNA 34,35 . We thus explored Sgk1 mRNA levels in DJ-1 −/− mice using qRT-PCR. Sgk1 expression was significantly lower in Th0 cells from DJ-1 −/− mice however, contrasting results were observed in iTregs. mRNA expression was higher in DJ-1 −/− iTregs than in DJ-1 +/+ iTregs (Fig. 8D,E). Thus, it appears that a higher expression of Sgk1 correlates with less Foxp3 expression in DJ-1 mice (Fig. 8F). To investigate this, we used a Sgk1 pharmacological inhibitor (10 nM-10 μ M; GSK650349) 36 and found that the Sgk1 inhibitor suppressed Foxp3 induction in DJ-1 −/− and DJ-1 +/+ mice (Fig. 8E). The loss was not as high as in DJ-1 −/− mice but it did not enhance Foxp3 expression. Taken together, Sgk-1 and ROS could contribute to lower in vitro expression of iTregs in DJ-1 deficient mice.

Inhibition of mTOR by rapamycin augments the iTregs induction. As aforementioned,
DJ-1-deficient mice have significantly higher AKT and Rictor expression in iTregs, we then explored whether attenuation of AKT or mTOR can lead to enhanced induction of Foxp3 in developing iTregs. We used pharmacological inhibitors for AKT (0.3 μ M; iAKT1/2) and mTOR (25 nM; Rapamycin) 13 , to determine the pathway required for enhancing iTregs induction in the absence of DJ-1. The data in Fig. 9 point towards a general improvement of iTregs induction after mTOR inhibition and appears to be independent of DJ-1.

Discussion
For effective immunotherapeutic treatment during infections, autoimmunity, neuro-inflammation and cancer, it is essential to understand Tregs development as these cells are pivotal in the maintenance of DJ-1/PARK7 influences the clinical course of Parkinson´s disease (PD) 15,17 , which is characterized by progressive loss of dopaminergic neurons in the substantia nigra pars compacta and their terminal connections in the striatum 15,22,37 . Previous findings suggest that microglia neuro-inflammatory responses boost the neurodegeneration in PD 38 . Further studies suggested that the 1-methyl-4-phenyl-1, 2,3,6-tetrahydropyridine (MPTP)-induced mouse model of PD can be protected by adoptive transfer of Tregs 39 . DJ-1 mRNA expression is enhanced in aged human pancreatic islets cells and reduced in type 2 diabetes mellitus (T2DM) 40 . In aged mice, there was significant increase in the nTregs number 41 and Tregs deficiency heading to destruction of islet cells and immune-pathology [42][43][44] . This finding suggests that DJ-1 affects the development of Tregs, which could be instrumental in understanding the pathology of the neuro-immunology of PD and other autoimmune responses. Various drugs such as PI3K inhibitors, AKT inhibitors and Rapamycin enhance whilst cyclosporine A can attenuate the development of iTregs 14,45,46 . Our data suggest that nTregs development was enhanced in the thymus as well as periphery and percentage of Tregs increases in DJ-1 deficient mice. However, total numbers of Tregs were almost equal in peripheral organs. These mice develop low levels of immunopathology when fed on high fat diet and CD4 + T cells were prone to differentiate into Th1 and Th2 cells 47 .
The in vitro experiments yielded the surprising result that DJ-1-deficient mice are unable to induce Foxp3 in developing iTregs. This is in contrast to the nTregs development and also the notion that DJ-1 deficiency enhances PTEN expression, thus attenuating the PI3K/AKT pathway and iTregs development. However, iTregs from DJ-1 −/− mice have less PTEN activity compared with control iTregs, a property in sharp contrast to observations in other cells types 11,20 . To resolve the disparity between the differential development of iTregs and nTregs, we measured cell proliferation and cell cycle assays and found that DJ-1-deficient iTregs are more proliferative and have poor in cell division compared with WT counterparts. Furthermore, iTregs from DJ-1 deficient mice are more prone to cell death. DJ-1 deficiency enhanced the death of developing iTregs but not of Th0 cells. Previous studies suggested that DJ-1 regulates the production of ROS and acts as an anti-oxidant 18,29 . Therefore, we also characterised the ROS production in developing iTregs and found that ROS production was severely enhanced in iTregs from DJ-1 deficient mice, which could lead to the enhanced death of these cells. However, previous studies suggested that ROS is upregulated in nTregs compared to Tcon cells and enhances the suppressive properties of Tregs 48 . In our experiments, it appears that enhanced ROS production lead to cell death and thus results in fewer iTregs. Previous studies have suggested that ROS can enhance Rictor/mTOR signalling 49 . Enhanced Rictor leads to AKT phosphorylation which affects the Foxp3 stability. Thus, ROS could be a mechanism conferring of Foxp3 stability. TLR4 activation by LPS also affects Foxp3 in a dose dependant manner 31 . ROS has dual effects on NFκ B, i.e. activation of cytoplasmic and inhibition of nuclear protein 32 . According to our observations NFκ B inhibition leads to further iTregs reduction and is consistent with previous findings 50 . In addition, a recent study showed that ROS production is reduced in AKT1 deficient dendritic cells 30 . As AKT deficiency leads to higher Foxp3 induction 14 , it could affect Foxp3 stability in DJ-1 deficient mice. Our data also revealed less Smad2 phosphorylation in DJ-1 deficient iTregs compared with littermate control iTregs, an observation suggesting that DJ-1 is also involved in IL-2 receptor signalling. DJ-1 −/− iTregs also shows less pStat5 activity compared with DJ-1 +/+ iTregs. It appears that in vitro DJ-1 −/− iTregs are less sensitive to TGF-β /IL-2 signalling due to less pSmad2/pStat5 phosphorylation. Furthermore, we speculate that TCR activation by anti-CD3/anti-CD28 leads to the activation of the AKT/mTOR pathway. It appears to be more dominating than TGF-β /IL-2 signalling in DJ-1 −/− developing iTregs and TGF-β /IL-2 signalling communication may be outcompeted during this process due to highly proliferative nature of Foxp3 -T cells and thus, could lead to lower Foxp3 induction. Nonetheless, an actual contribution of both signalling pathways will require further research. In summary, DJ-1 affects the in vitro development of iTregs and switching the signalling from in vivo developed nTregs. However, further studies are warranted to find the exact molecular mechanism for ex vivo isolated nTregs and iTregs.
Enhanced ROS production may increase Sgk1 expression 34,35 . Sgk1 could in turn lead to the phosphorylation of FoxO1/3a protein which could destabilise the Foxp3 protein 46,51 . However, inhibition of mTOR/Sgk1 is required for the development of iTregs in DJ-1-deficient mice but not in wild type mice. Therefore, it appears that a balanced Sgk1 activity is required for iTregs development. If the expression level reaches a threshold, it may lead to the induction of pathogenic Th-17 cells 52,53 . DJ-1-deficient mice showed a low-level inflammatory response in the high-fat diet-induced obesity model 47 . In contrast infiltration of T cells occurs in multiple sclerosis and PD, and levels of IL-1β , IL-6 and TNF-α are elevated in the cerebrospinal fluid of PD patients 54 . A recent study suggested that DJ-1 deficient CD4 + T cells are also able to differentiate into more Th1 and Th17 phenotypes in an atherosclerosis model 23 . Therefore, during inflammation in DJ-1 −/− mice iTregs presumably fail to develop adequately thus influencing disease severity.
In conclusion, our study identified a novel role of DJ-1 in differential regulation of nTregs and iTregs via regulating mTOR/Sgk1 axis as well through ROS/NFκ B signalling pathways. DJ-1 thus participates in the regulation of the immune response.

Materials and Methods
Mice. DJ-1 +/+ and DJ-1 −/− (10-12 week-old) mice were used for the experiments as described earlier 18 and kept under standard conditions. All experiments were performed according to the EU Animals Scientific Procedures Act and the German law for the welfare of animals. The procedures were approved by the authorities of the state of Baden-Württemberg.
Flow cytometry. Thymocytes and splenocytes from wild type and DJ-1-deficient mice were characterised by using surface and intracellular staining with relevant antibodies. In brief, thymocytes and splenocytes were collected and used for surface staining for anti-CD4, anti-CD8a, anti-CD25, anti-CD62L, anti-CD44 (eBioscience, Germany) or other antibodies depending on the experiment and washed with PBS. Cells were fixed with Foxp3 fixation/permeabilization buffer (eBioscience, Germany) for intracellular staining and incubated for 30 minutes. After incubation, cells were washed with 1x permeabilization buffer, exposed to added intracellular monoclonal antibodies for Foxp3 and incubated for additional 30 minutes. Cells were washed again with permeabilization buffer and PBS was added to acquire the cells on a flow cytometer (FACS-calibur from Becton Dickinson; Heidelberg, Germany).
Cell cycle analysis. Activated Th0 and iTregs were washed 1x with PBS at 600xg for 5 minutes at RT. After discarding the supernatant, 1 ml of − 20 °C cold ethanol:PBS mix (3:1) was added during swirl mixing of the cells. The preparation was kept at − 20 °C overnight to maximum 4 days depending on the experiment. After incubation at − 20 °C, cells were washed 1× with PBS again and 250 μ l of PI mix were added [50 μ g/ml PI (Sigma, Germany) and 100 μ g/ml RNase A (Qiagen, Germany)]. The cells were kept for 30 minutes at RT. Then, the cells were acquired by flow cytometry for cell cycle and death analysis. CFSE staining and cell proliferation. CD4 + T cells (5 × 10 6 ) were washed 1× with PBS (Sigma, Germany) and stained with 2 μ M CFSE (eBioscience, Germany) for 15 minutes at RT in the dark and washed 2× with RPMI-1640 medium as described earlier 55 . Stained cells were cultured for 3 days and after 3 days of culture; cells were stained with Foxp3 antibody and acquired on flow cytometry. For suppression assay conventional T cells and nTregs (DJ-1 +/+ and DJ-1 −/− ) were added into 1:1 ratio and run on the flow cytometry to measure the proliferation of conventional cells as described previously 55 . Immunoblotting. Naïve T cells (1 × 10 6 cells) from DJ-1 +/+ and DJ-1 −/− were activated in the presence of anti-CD3::anti-CD28 (1:2) and differentiated into iTregs in presence of TGF-β and IL-2 and after 3 days of culture, cells were lysed using Lamelli buffer and subjected to immunoblotting. Differentiated iTregs were activated for 30 minutes with respective conditions and then washed once with PBS and added the equal amount of H 2 O and 2X Lammelli's Buffer for cell lysis and proteins was denatured at 95 °C for 5 minutes and stored at − 20 °C. Sample proteins were loaded on 8% or 10% gel depending on proteins size and run for 80-120 mV for 90-100 minutes. Proteins were electrotransferred onto PVDF membrane (GE healthcare, USA). Membranes were probed with the indicated primary antibodies (Pten, pAktS473, mTOR, Rictor, pSmad2, pStat5, DJ-1 and GAPDH; Cell Signalling, USA) followed by HRP-conjugated secondary antibodies (Cell Signalling, USA). Membranes were washed and visualized with enhanced chemiluminescence detection system (ECL; peqLab, Germany). q-RT-PCR. Total mRNA was isolated from Th0 and iTreg cells using the mRNAeasy isolation kit (Qiagen, Germany) as described by the manufacturer. 1.0 μ g mRNA was converted into cDNA using the Superscript III cDNA synthesis kit (Invitrogen, Germany). Briefly, in 10.0 μ l reactions, 10.0 ng cDNA, 2X SYBR green Master-mix (KAPA SYBR green, peqLab, Germany) and 250 nM primers were used for q-RT-PCR reactions. q-RT-PCR and data analysis were performed as described previously 56 for Foxp3 (F primer: 5′ -GGTACACCCAGGAAAGACAG-3′ and R primer: 5′ -ATCCAGGAGATGATCTGCTTG-3′ ), Sgk-1 (F primer: 5′ -TGAAACAGAGAAGGATGGGC-3′ and R primer: 5′ -GAACTTCAGCGTGTTTGCAT-3′ ) using universal cycling conditions (95 °C for 10 minutes, 95 °C for 15 seconds and 60 °C for 1 minute for 40 cycles followed by melting curve analysis). All the primers were purchased from Sigma, Germany. ROS production. ROS production in iTregs was measured by 2′ ,7′ -dichlorodihydrofluorescein diacetate (DCFDA). Briefly, 1 × 10 6 cells were taken in a 24 well plate and DCFDA (Sigma, Germany) was added to the cell suspension at a final concentration of 10 μ M. After 30 minutes of incubation in the dark at RT, cells were centrifuged and the pellet was washed twice with 1x PBS. The pellet was then resuspended in FACS buffer and the fluorescence was analysed with a flow cytometer. DCFDA fluorescence intensity was measured in FL-1 with an excitation wavelength of 488 nm and an emission wavelength of 530 nm.

Statistics. Figures were made in Excel
Microsoft office software. Arithmetic mean values are presented ± standard error of mean (SEM) and n represents the number of independent biological experiments. GraphPad Prism and Excel were used for statistical analyses. Student's unpaired t-test was used