Mst1-FoxO Signaling Protects Naïve T Lymphocytes from Cellular Oxidative Stress in Mice

Background The Ste-20 family kinase Hippo restricts cell proliferation and promotes apoptosis for proper organ development in Drosophila. In C. elegans, Hippo homolog also regulates longevity. The mammalian Ste20-like protein kinase, Mst1, plays a role in apoptosis induced by various types of apoptotic stress. Mst1 also regulates peripheral naïve T cell trafficking and proliferation in mice. However, its functions in mammals are not fully understood. Methodology/Principal Findings Here, we report that the Mst1-FoxO signaling pathway plays a crucial role in survival, but not apoptosis, of naïve T cells. In Mst1−/− mice, peripheral T cells showed impaired FoxO1/3 activation and decreased FoxO protein levels. Consistently, the FoxO targets, Sod2 and catalase, were significantly down-regulated in Mst1−/− T cells, thereby resulting in elevated levels of intracellular reactive oxygen species (ROS) and induction of apoptosis. Expression of constitutively active FoxO3a restored Mst1−/− T cell survival. Crossing Mst1 transgenic mice (Mst1 Tg) with Mst1−/− mice reduced ROS levels and restored normal numbers of peripheral naïve T cells in Mst1 Tg;Mst1−/− progeny. Interestingly, peripheral T cells from Mst1−/− mice were hypersensitive to γ-irradiation and paraquat-induced oxidative stresses, whereas those from Mst1 Tg mice were resistant. Conclusions/Significance These data support the hypothesis that tolerance to increased levels of intracellular ROS provided by the Mst1-FoxOs signaling pathway is crucial for the maintenance of naïve T cell homeostasis in the periphery.


Introduction
Maintenance of T cell homeostasis is critical for normal functioning of the immune system. Fas, TNF, and ROS effectively promote the elimination of antigen-specific activated T cells and limit autoimmunity [1,2]. Intracellular redox status is a physiological regulator of T cell activation and apoptosis during normal T cell development in the thymus, and also regulates the immune response in peripheral lymphoid organs [3]. Recent studies have shown that the transcription factor, FoxO1, is critical for maintaining naïve T cells in peripheral lymphoid organs by virtue of its regulation of L-selectin (CD62L), CCR7 and interleukin 7 receptor a (IL-7Ra/CD127) expression [4,5]. However, naïve T cell homeostatic mechanisms in protective immunity are not fully understood.
In mammals, Mst1 (mammalian sterile 20-like 1) kinase, which is a key component of the ''Hippo'' signaling pathway, has been implicated in regulating the cell cycle, apoptosis and cellular responses to oxidative stress [6]. Recently, Mst1-deficient (Mst1 2/2 ) mice were shown to possess decreased numbers of peripheral T cells, mostly naïve T cells; those cells that remained showed increased proliferation after T cell receptor (TCR) ligation in vitro and impaired lymphocyte homing to the spleen and lymph nodes [7,8]. Additionally, defective egress of mature thymocytes from Mst1 2/2 thymus were also reported [7,9]. The Rassf adapter protein, Rapl (also named Nore1b), is known to activate Mst1 by regulating its localization and kinase activity during lymphocyte migration to lymphoid organs [10]. These observations suggest that the Rapl-Mst1 pathway can modulate cell proliferation and T-cell migration [7,8]. Consistent with these recent findings, we observed a similar phenotype in Mst1 2/2 mice. In addition, we identified a new and important role for the Mst1-FoxO signaling pathway in regulating the homeostasis of peripheral naïve T cells.

Mst1 Deficiency Causes Systemic Lymphopenia
To define the physiological functions of Mst1 in higher organisms, we generated mice lacking Mst1. Lymphoid tissues from Mst1 2/2 mice displayed a paucity of T and B cells in the spleen and reduced numbers of T cells in lymph nodes; in particular, a large number of cells in white pulp and T cell zones appeared to be absent ( Figure S1A and B). We confirmed that the numbers of T cells in spleen, lymph nodes, and peripheral blood of Mst1 +/+ and Mst1 +/2 mice were not different, indicating the haplosufficiency of Mst1 with respect to leukocyte generation ( Figure S2). Mst1 +/+ or Mst1 +/2 mice were thus used as controls. Consistent with recent findings from Mst1-deficient mice [7][8][9], we also observed the reduced numbers of both CD4 + and CD8 + peripheral T cells in Mst1 2/2 mice ( Figure S2A). Most of the absent T cells in peripheral blood, spleen, and lymph nodes were naïve T cells, not effector/memory T cells ( Figure S2B). Although the ratio of single-positive CD4 and CD8 cells appeared to be slightly increased in the thymus from Mst1 2/2 mice ( Figure S1E), the thymic architecture and the actual numbers of these cells were similar in Mst1 +/+ , Mst1 +/2 , and Mst1 2/2 mice ( Figure S1C and D). The thymic development of Mst1 2/2 T cells bearing TCRs was also normal, as was their environmental epithelium (unpublished data). Thus, we concluded that the reduced number of peripheral naïve T cells from Mst1 2/2 mice does not result from impaired T cell development or thymic development.

Mst1 Deficiency Causes Naïve T Cell Death
Mst1 promotes cellular oxidative stress and/or growth factor deprivation-induced apoptosis in neurons [6,11]. Therefore, we first tested whether Mst1 deficiency affected the survival of T cells. Interestingly, many of the peripheral CD4 + and CD8 + T cells in Mst1 2/2 mice rapidly underwent apoptosis ( Figure 1A, top). There was also a loss of mitochondrial transmembrane potential (y m ) in these Mst1 2/2 peripheral CD4 + and CD8 + T cells ( Figure 1A, bottom), confirming apoptosis. Thus, spontaneous cell death of peripheral Mst1 2/2 T cells might be responsible for the reduced numbers of peripheral T cells in Mst1 2/2 mice.
To explore the basis for this reduced cellularity and/or apoptosis of peripheral T cells in Mst1 2/2 mice, we first crossed Fas-and FasL-defective mice (lpr and gld, respectively) with Mst1 2/2 mice. Neither Mst1 2/2 ;Fas lpr/lpr nor Mst1 2/2 ;FasL gld/gld mice exhibited a restoration of peripheral T cells to normal levels ( Figure S3), indicating that the apoptosis of Mst1 2/2 T cells was independent of Fas and FasL. An examination of serum cytokines and chemokines revealed no significant differences between Mst1 +/2 and Mst1 2/2 mice. Notably, the levels of IL-7, which is critical for the maintenance of both naïve and memory T cell homeostasis [12], and tumor necrosis factor (TNF), which is associated with apoptosis of activated T cells [13], were similar in the sera of Mst1 +/2 and Mst1 2/2 mice ( Figure S4).

FoxO Pproteins Are Inactivated in Mst1 2/2 Peripheral T Cells
Mst1 phosphorylates and activates FoxO3a on serine 207, promoting cellular oxidative stress-induced apoptosis in neurons and longevity in Caenorhabditis elegans [6]. Phosphorylation of FoxO1 on serine 212 by Mst1 also regulates neuronal cell death under conditions of growth-factor deprivation [11]. Akt and Mst1 exhibit mutually antagonistic functions, and FoxO proteins can be inactivated by Akt [14][15][16]. On the basis of these reports, we examined the expression levels of FoxO1/3 proteins in peripheral CD4 + , CD8 + , naïve (CD62L hi CD44 lo ), and effector/memory (CD62L lo CD44 hi ) T cells. Interestingly, Mst1 2/2 T cells displayed reduced levels of FoxO3a and FoxO1 proteins, and the phosphorylated form of FoxO1/3 ( Figures 1B and S5). Thus, we hypothesized that Mst1 is required for maintaining normal levels of FoxO1/3 proteins, and further postulated that inactivation of these proteins might disrupt peripheral naïve T cell homeostasis.
In primary neurons, FoxO3a subcellular localization is reciprocally regulated by MST1, which promotes nuclear localization by phosphorylating FoxO3a on serine 207 in response to hydrogen peroxide (H 2 O 2 ), and by Akt, which phosphorylates FoxO proteins and induces their translocation to the cytoplasm where they are subsequently degraded [6,16]. Using immunostaining to examine the subcellular localization of FoxO3a in peripheral T cells, we found that FoxO3a was primarily localized to the nucleus in Mst1 +/2 T cells (,85%). By contrast, the majority of FoxO3a was localized in the cytoplasm of peripheral Mst1 2/2 T cells (,65%), and its staining intensity was significantly weaker than in Mst1 +/2 T cells ( Figure 1C). Consistent with this observation, FoxO1/3 was detected in an activated, phosphorylated state in a significant proportion (,37%) of Mst1 +/2 peripheral T cell nuclei, whereas negligible numbers of similarly stained T cells were detected in Mst1 2/2 mice ( Figure 1D). We were unable to specifically detect FoxO1 protein because FoxO1 antibodies suitable for immunostaining are unavailable. These data show that Mst1 modulates FoxO1/3 nuclear localization and activation in peripheral T cells.
Expression of Sod2 and Catalase Are Significantly Down-Regulated in Mst1 2/2 Peripheral T Cells Recent reports have shown that FoxO1 deletion in T cells induces a reduction in the number of peripheral naïve T cells due to a defect in the expression of lymphocyte-trafficking molecules (L-selectin and CCR7) and IL-7Ra [4,5], indicating a critical role for FoxO1 in homeostasis of naïve T cells. On the basis of these reports and our observation that FoxO1/3 expression was reduced in Mst1 2/2 T cells, we examined the expression of L-selectin, CCR7 and IL-7Ra in Mst1 2/2 naïve T cells. Although IL-7Ra expression was slightly reduced in peripheral blood, the cell surface levels of these proteins did not show statistically significant difference between Mst1 +/+ and Mst1 2/2 naïve T cells ( Figure 2). Thus, it is likely that a reduction in FoxO1 protein levels, unlike ablation of the FoxO1 gene, does not significantly affect the expression of these molecules in Mst1 2/2 mice.
FoxO family factors stimulate transcription of Sod2 and catalase (Cat), which are important in reducing cellular oxidative stress [17,18]. Thus, we then examined the expression levels of these FoxO targets in T cells. Consistent with a role for the Mst1-FoxO pathway in regulating Sod2 and Cat expression, quantitative PCR analyses showed that Sod2 and catalase mRNA levels were significantly decreased in T cells derived from Mst1 2/2 spleens and lymph node compared to those from Mst1 +/2 mice ( Figure 3). Consistent with reduced mRNA levels, their protein levels were also reduced in Mst1 2/2 T cells ( Figure S5) Thus, in the absence of Mst1, reducing and/or inactivating FoxO1/3 could result in decreases in Sod2 and catalase expression in CD4 + and CD8 + T cells.

Increased ROS in Mst1 2/2 Peripheral T Cells Induces Apoptosis
Because Sod2 and catalase were down-regulated in Mst1 2/2 T cells, we tested whether the levels of intracellular ROS were increased in T cells from Mst1 2/2 mice, and thus might account for the increase in apoptosis. Although the overall levels of intracellular ROS in thymic T cells were similar in Mst1 +/2 and Mst1 2/2 mice, intracellular ROS levels increased gradually during the progression from double-positive to single-positive T cells in the thymus (unpublished data). Interestingly, the peripheral CD4 + and CD8 + T cells in Mst1 2/2 mice showed elevated ROS levels compared to Mst1 +/2 mice ( Figure 4A). Increased ROS can Mitochondrial transmembrane potential (y m ) of CD4 + and CD8 + T cells in peripheral blood from Mst1 +/2 and Mst1 2/2 mice was determined by staining with Rhodamine 123. CCCP-treated cells were used as a control (grey). FACS profiles shown are representative of three independent experiments. (B) FoxO1/3 protein levels and phosphorylation were decreased in Mst1 2/2 T cells. CD4 + /CD62L hi or CD4 + /CD62L lo T cells from spleens were purified by MACS. Cell lysates were then analyzed by immunoblotting. (C) Immunostaining of FoxO3a in peripheral blood T cells from Mst1 +/2 and Mst1 2/2 mice. Lymphocytes were stained for CD4 (green) and FoxO3a (red), or CD8 (red) and FoxO3a (green); nuclei were stained with DAPI (blue). (D) Immunostaining of p-FoxO1/3 in peripheral blood T cells from Mst1 +/2 and Mst1 2/2 mice. Lymphocytes were stained for CD4 (green) and p-FoxO1/3 (red), or CD8 (red) and p-FoxO1/3 (green); nuclei were stained with DAPI (blue). doi:10.1371/journal.pone.0008011.g001 produce DNA double-strand breaks (DSBs) in mammalian cells [19]. We therefore examined cH2AX, an established measure of DSBs, in T cells. As expected, CD4 + and CD8 + T cells from Mst1 2/2 mice showed an increase in cH2AX signals compared with Mst1 +/2 T cells ( Figure S6), indicating the presence of DSBs. Intracellular ROS can regulate Bcl-2 levels in activated T cells [2], and the Bim/Bcl-2 balance is critical for maintaining naïve and memory T cell homeostasis [20]. Therefore, we also investigated  Bcl-2 involvement in the T cell death associated with increased ROS in Mst1 2/2 mice. Bcl-2 mRNA expression was decreased about 2-to 3-fold in Mst1 2/2 compared with Mst1 +/2 T cells; however, the expression level of Bim EL , an antagonist of Bcl-2, in Mst1 2/2 T cells was increased compared with controls ( Figure 4B). Thus, elevated intracellular ROS in Mst1 2/2 T cells ultimately decreased Bcl-2 levels, creating a disruption in the balance between Bcl-2 and Bim EL that might contribute to the apoptosis of peripheral Mst1 2/2 T cells.
To determine whether the increased apoptosis of Mst1 2/2 T cells was primarily due to ROS, we incubated Mst1 2/2 T cells from peripheral blood with the antioxidant, N-acetyl-L-cysteine (NAC). At an NAC concentration of 10 ng/ml, intracellular ROS was reduced ( Figure 4C, left) and apoptosis was decreased in Mst1 2/2 CD4 + and CD8 + T cells ( Figure 4C, right), suggesting that the elevated levels of ROS in peripheral Mst1 2/2 T cells resulted in cell death. Although NAC treated Mst1 +/2 T cells had lower ROS levels than untreated cells, apoptosis in these cells was not affected by NAC treatment. Although NAC treated Mst1 2/2 CD8 + T cells have modestly lower levels of ROS compared with untreated Mst1 2/2 CD8 + T cells, it is also noted that ROS levels in NAC treated Mst1 2/2 CD8 + T cells were not reduced as much as in CD4 + T cells. These differences in response to NAC treatment between CD4 + and CD8 + T cells may account for smaller change in Mst1 2/2 CD8 + T cell death compared to CD4 + T cell death. Taken together, these results suggest that Mst1deficient naïve T cells fail to activate FoxO proteins, resulting in a failure to induce Sod2 and catalase expression, the accumulation of intracellular ROS, and increased apoptosis in single-positive T cells.
We next generated Mst1 wild-type transgenic (Mst1 Tg) and kinase-dead transgenic (Mst1KD Tg) mice, then tested whether the observed defects were corrected in the Mst1 Tg;Mst1 2/2 or Mst1KD Tg;Mst1 2/2 progeny. In Mst1 Tg;Mst1 2/2 mice, intracellular ROS levels in CD4 + and CD8 + T cells were decreased to levels comparable to those found in Mst1 +/2 peripheral T cells ( Figure 5A), and the numbers of peripheral naïve T cells were recovered to an even greater extent in Mst1 Tg;Mst1 2/2 mice ( Figure 5B). In Mst1 Tg;Mst1 2/2 mice, FoxO3a, Sod2, and catalase protein levels in T cells were also recovered to levels comparable to those in Mst1 +/2 T cells ( Figure  S5). Notably, expression of kinase-dead Mst1 in Mst1KD Tg;Mst1 2/2 mice failed to restore normal numbers of peripheral blood T cells ( Figure 5B), indicating that Mst1 kinase activity toward FoxO1/3 is critical for peripheral T cell homeostasis.

Expression of FoxO3a-S207D Rescues the Mst1-Null Phenotype
To determine whether Mst1-mediated phosphorylation of FoxO is sufficient to promote the neutralization of elevated intracellular ROS in peripheral T cells, we infected Mst1 2/2 bone marrow cells with a retrovirus expressing a constitutively active FoxO3a-S207D mutant (FoxO3a-SD), and then transplanted infected bone marrow into lethally irradiated recipient mice. This construct conferred a 3-to 4-fold survival advantage on peripheral Mst1 2/2 CD8 + T cells, examined 4 weeks after transplantation, compared to those transplanted with cells transduced with a control pMSCV-IRES-GFP (Mig) virus ( Figure 5C; p = 0.014). In contrast, transplanted bone marrow cells transduced with a nonphosphorylatable S207A FoxO3a mutant (FoxO3a-SA) were much less effective in rescuing the population of Mst1 2/2 CD8 + T cells than were FoxO3a-SD-transduced cells ( Figure 5C; p = 0.067). Interestingly, the Mst1 2/2 CD4 + T cell population showed maximum survivability in peripheral blood after trans- plantation with bone marrow cells transduced with wild-type FoxO3a. The responsiveness of FoxO3a-SD could be different in peripheral CD4 + and CD8 + T cells in the context of oxidative stress.

Mst1 Confers ROS Resistance in Lymphocytes
Lymphocytopenia can be induced by a variety of cytotoxic stresses, including chemotherapy and radiation. Radiation generates oxygen-derived free radicals in peripheral lymphocytes [21]. To address the capacity of Mst1 in T cells to protect against irradiation-induced lymphopenia, we examined peripheral T cell viability after exposure to c-irradiation (5 Gy). Seventy-two hours after c-irradiation, we found dramatic increases in the viability of peripheral CD4 + (,2.5-fold) and CD8 + (,4-fold) T cells from Mst1 Tg;Mst +/2 mice compared to Mst1 +/2 T cells ( Figure 5D, left). Using sublethal doses (5 mg/kg) of N,N'-dimethyl-4,4'bipyridinium dichloride (paraquat), a herbicide that induces the formation of ROS [22], we found that intracellular ROS levels were dramatically increased in peripheral CD4 + and CD8 + T cells of Mst1 +/2 and Mst1 2/2 mice within 24 hours (unpublished data). However, overexpression of Mst1 (in Mst1 Tg;Mst +/2 mice) effectively neutralized intracellular ROS in peripheral T cells in association with approximately a 3.5-fold increase in T cell numbers 6 days after injection ( Figure 5D, right).
These results suggest that Mst1 plays a role in promoting survival of naïve T cells by detoxifying ROS through the Mst1-FoxO-Sod2/catalase pathway ( Figure 6). How does regulating intracellular ROS maintain peripheral naïve T cell homeostasis? The peripheral T cell population needs to be maintained at a nearconstant number. Naïve T cell maintenance in the periphery depends on the appropriate stimulation of TCR, whereas effector/ memory cells can survive in the absence of TCR ligation [23][24][25]. Consistent with recent reports [7,8], we confirmed that Mst1 is activated by TCR ligation in T cells ( Figure S7). Importantly, we observed that FoxO protein levels and phosphorylation of FoxO proteins were reduced in Mst1 2/2 T cells, and found that T cells from Mst1 2/2 mice exhibited higher levels of apoptosis (Figures 1  and 4C, right). Our findings thus suggest that the Mst1-FoxO-Sod2/catalase pathway might tightly modulate intracellular ROS in CD4 + and CD8 + T cells to maintain levels optimal for cell survival.
Importantly, two recent studies showed that FoxO1 is critical for the maintenance of naïve T cell homeostasis by controlling the expression of IL-7Ra, L-selectin, and CCR7 [4,5]. FoxO1deficient T cells have defects in both survival and homing to secondary lymphoid organs. Mst1 2/2 T cells also have defects in homing capacity [7,9], and the Rap1-Rapl-Mst1 pathway has been previously suggested to play a role in controlling T-cell trafficking [10,26]. However, unlike FoxO1-deficient T cells, Mst1-deficient T cells appeared to show insignificant changes in the expression of IL-7Ra, L-selectin, CCR7, but Sod2 and catalase were significantly reduced (Figures 2, 3 and S5). It is possible that although the residual levels of FoxO proteins in Mst1 2/2 T cells are too low to promote sufficient expression of Sod2 and catalase to protect cells from oxidative damage, they are high enough to maintain expression of other FoxO transcriptional targets involved in T cell trafficking and survival. Thus, increased ROS in Mst1 2/2 naïve T cells might also result in a homing defect to secondary lymphoid organs. In addition to the Rap1-Rapl-Mst1 pathway [7,10,26], the Mst1-FoxO pathway might also be important for T cell trafficking. Unlike Mst1 2/2 mice, which show a deprivation of peripheral naïve T cells, Rapl-deficient mice exhibit concentrations of peripheral blood lymphocytes comparable to those of wild-type mice [26]. We also found that Nore1/ Rassf5-deficient mice, which are deficient for both Nore1A and Nore1B/Rapl splice variants, exhibited normal numbers of peripheral blood lymphocytes, even though the numbers of T cells in lymph node and spleen were reduced ( Figure S8A). It is also reported that Mst1 2/2 thymocytes have defects in egress from thymus, which result in lymphopenia in peripheral lymphoid organs [7,9]. However, we did not observe the accumulation of CD4 or CD8 single positive thymocytes in thymus from Mst1 2/2 mice ( Figure S1D). It is interesting that Rapl-deficient mice also have defects in thymocyte egress and ratio of CD4 or CD8 single positive thymocytes were increased, but actual cell numbers of these cells and numbers of T cells from peripheral blood were apparently normal [26]. Thus, the Rap1-Rapl-Mst1 pathway could be responsible for controlling the T cell trafficking to secondary lymphoid organs. By controlling intracellular ROS levels, the Mst1-FoxO pathway is likely only important for the survival of peripheral blood naïve T cells, a function that is independent of the Rap1-Rapl-Mst1 pathway. We also found that apoptosis was increased in splenic CD4 + and CD8 + T cells ( Figure  S9A), but their ROS levels were not different from those of Mst1 +/+ ( Figure S9B). However, these cells showed decreased protein levels of FoxO1/3 ( Figures 1B and S5), Catalase and Sod2 ( Figure S5). Therefore, environmental differences between spleen and peripheral blood might have differentially affected ROS levels in splenic and peripheral blood T cells of Mst1 2/2 . In addition, Mst1 2/2 CD4 + T cells were rescued by FoxO3a wild-type and FoxO3a-S207A mutant but not by FoxO3a-S207D mutant ( Figure 4C). Based on these results, we can not exclude the possibility that there might be still other functional mechanisms for regulating T cell survival by FoxO1/3 and/or Mst1 can regulate T cell homeostasis through other unknown factors.
Previous studies in yeast and mammalian cells have shown that Mst1 promotes cell death in response to oxidative stress by phosphorylating histone H2B [27,28]. In primary neurons, Mst1 enhances cell death by directly activating FoxO1/3 [6,11]. In sharp contrast, our in vivo mouse study demonstrates for the first time that Mst1 is crucial for cell survival rather than cell death in response to intracellular ROS, at least in naïve T cells, and it accomplishes this function by regulating FoxO proteins. Thus, Mst1 appears to act in a cell context-dependent manner to differentially regulate cell death and survival by modulating In peripheral T cells, Mst1 is activated after TCR ligation, followed by FoxO1/3 activation, which is important for ROS regulation and survival of naïve T cells. Mst1 might also be activated by Rapl through TCR ligation during migration to lymphoid organs, resulting in LFA-1 clustering, as recently reported by others [7,8]. Arrow size indicates intensity of activation of its target. doi:10.1371/journal.pone.0008011.g006 FoxO1/3 function. It will be interesting to determine whether the loss of Mst1 promotes cell survival in response to oxidative stresses in primary neurons.
The Hippo signaling pathway restricts cell proliferation and promotes apoptosis in Drosophila [29]. Consistent with this, epithelial tissues of WW45 2/2 mice also displayed hyperproliferation and defects in apoptosis [30]. Hippo signaling has also been proposed to regulate mammalian organ size [31], but neither Mst1 2/2 nor Mst2 2/2 mice showed any epithelial hyperplasia or increase in the size of any organs (unpublished data). Notably, Mst2 2/2 T cells showed no reduction in the numbers of T cells in peripheral lymphoid organs ( Figure S8B), suggesting that Mst2 is dispensable for maintaining naïve T cells homeostasis. In addition, Mst1 2/2 ;Mst2 2/2 mice died at about E8.5 with severe growth retardation, failed placental development, impaired yolks sac and embryo vascular patterning, and increased apoptosis in placentas and embryos [32]. Thus, although Mst1 and Mst2 kinases are functionally similar, both play essential roles in early mouse development. Taken together, these observations suggest that Mst1, but not Mst2, plays a distinct role in maintaining naïve T cells homeostasis.
Finally, we showed that overexpression of Mst1 rescues T cells from apoptosis caused by abnormally elevated intracellular ROS during exposure to anticancer treatments. Thus, Mst1 may represent a potential target of therapeutic strategies to ameliorate lymphopenic side effects in patients undergoing radiation treatment or chemotherapy.

Ethics Statement
Animal care and experimentation were performed in accordance with procedures approved by the KAIST (Korea Advanced Institute of Science and Technology) Animal Care and Use Committee (KACUC).

Generation of Antibodies
An antibody to Mst1 was generated as previously described [30]. An antibody to p-FoxO1 (S212)/p-FoxO3 (S207) was generated by injecting rabbits with the phosphopeptide antigen, C-SAGWKNpSIRHNLS. Antibodies were affinity purified from immune sera using the appropriate antigens.

Flow Cytometric Analysis
Thymus, spleen, and inguinal lymph nodes were filtered through a 40-mm Cell Strainer (BD Falcon), and the resulting lymphocytes were isolated by centrifugation over Lymphocyte Separation Media (Mediatech Cellgro). Peripheral blood was lysed with Ack lysis buffer (0.15 M NH 4 Cl, 10 mM KHCO 3 , 0.1 mM Na 2 EDTA, pH 7.2) to obtain leukocytes. Isolated cells were stained with antibodies against CD4 (GK1.5), CD8a (53-6.7; Southern Biotech), CD44 (IM7), CCR7 (EBI-1), CD127 (SB/199; eBioscience), and CD62L (MEL-14; BD Pharmingen). Intracellular ROS levels were assessed by prestaining 10 6 cells with antibodies against CD4 and CD8, followed by staining with DCF-DA (Invitrogen) at a concentration of 20 mM for 30 min at 37uC. Apoptosis was determined in freshly isolated lymphocytes by first prestaining for CD4 and CD8, after which lymphocytes were stained with Annexin V staining kit (BD Pharmingen). The mitochondrial transmembrane potential (y m ) of lymphocytes was determined by staining 10 6 lymphocytes (prestained for CD4 and CD8) with Rhodamine 123 at a final concentration of 4 mg/ml for 20 min at 37uC. Cells treated with 20 mM CCCP for 20 min before Rhodamine 123 staining were used as a control. Analyses were performed on a triple-laser LSRII flow cytometer with DiVA software (Beckton Dickinson).

Magnetic-Activated Cell Sorting
Lymphocytes were isolated from spleen and lymph nodes. CD4 + and CD8 + T cells were purified using CD4 and CD8a microbeads, respectively, and CD4 + /CD62L hi T cells were purified using CD4 + CD62L + T Cell Isolation Kit II with a VarioMacs magnetic cell sorter (Miltenyi Biotec).

In Vitro NAC Treatment
For in vitro NAC experiments, peripheral blood was collected and lysed with Ack lysis buffer containing NAC (10 ng/ml, Sigma). The concentration of NAC was maintained in all reagents used in subsequent steps of DCF-DA staining and Annexin V experiments.

Quantitative RT-PCR
Total RNA from freshly isolated or magnetically sorted lymphocytes from each organ was isolated using TRIzol Reagent (Intron) according to the manufacturer's instructions and treated with DNaseI. RNA was reverse transcribed using Superscript III (Invitrogen). Quantitative RT-PCR reactions were performed in triplicate using iQ SYBR Green Supermix and the iQ5 Multicolor Real-Time PCR Detection System, and data were analyzed using iQ5 optical system software (Bio-Rad). Gene-specific primer sequences were obtained from Primer Bank (http://pga.mgh. harvard.edu/primerbank). PCR primers used were 59-CAG ACC TGC CTT ACG ACT ATG G-39 and 59-CTC GGT GGC GTT GAG ATT GTT-39 for Sod2; 59-AGA GAG CGG ATT CCT GAG AGA-39 and 59-ACC TTT CCC TTG GAG TAT CTG G-39 for catalase; 59-ATG CCT TTG TGG AAC TAT ATG GC-39  and 59-GGT ATG CAC CCA GAG TGA TGC-39 for Bcl-2; 59-GAC AGA ACC GCA AGG TAA TCC-39 and 59-ACT TGT  CAC AAC TCA TGG GTG-39 for Bim EL ; and 59-AGG TCA  TCA CTA TTG GCA ACG A-39 and 59-CAC TTC ATG ATG  GAA TTG AAT GTA GTT-39