Schlafen2 mutation unravels a role for chronic ER stress in the loss of T cell quiescence

Immunologically naïve lymphocytes are kept in a quiescent state until antigen engagement. These quiescent immune cells are characterized by small cell size, lack of spontaneous proliferation and low metabolic rate. Lymphocyte quiescence is actively enforced condition which ensures the preservation of proper differentiation and proliferation capabilities of naïve and memory lymphocytes. Previously we described a chemically induced mutation in Schlafen2 (Slfn2), termed elektra, which breaks quiescence and compromises immunity. However, the mechanism by which Slfn2 maintains quiescence remains unknown. Here we demonstrate that elektra T cells display chronic ER stress under steady state conditions. Modulation of ER stress response by depletion of either UPR mediators XBP1 or CHOP, improved viability and partially corrected the developmental abnormalities and proliferation capabilities of elektra T cells. Altogether, our results demonstrate a functional connection between Slfn2 induced quiescence in T cells and ER homeostasis, clarifying a novel mechanism by which immune cell quiescence is maintained.


IntroductIon
The endoplasmic reticulum (ER) is the entry site of proteins into the secretory pathway and a central organelle for lipid synthesis. To acquire a native state in the ER, newly translated proteins undergo post translation modifications and are engaged by molecular chaperones. A disruption of ER homeostasis by physiological, pathogenic or chemical insults leads to the accumulation of unfolded or misfolded proteins, conditions termed ER stress [1]. ER stress elicits an unfolded protein response (UPR), a signaling pathway that emanates from the ER, aimed to restore ER function [2]. The UPR is mediated through the activation of three ER transmembrane stress sensors; pancreatic ER kinase (PKR)-like ER kinase (PERK), activating transcription factor 6 (ATF6) and inositol-requiring enzyme 1 (IRE1) [3]. These proteins are maintained in an inactive state through interaction with the ER chaperone, glucose regulated protein 78 (GRP78, also known as Bip). Unfolded proteins can interact directly with IRE1 to mediate its activation or trigger the release of GRP78 from the UPR sensors allowing their activation [4]. Activated IRE1, PERK and ATF6 promote adaptation to ER stress through the activation of transcriptional upregulation of key ER chaperones as well as by controlling protein synthesis through the phosphorylation of eIF2α. The transcriptional arm of the UPR adjusts ER folding capacity, enhances ER-associated degradation (ERAD) and drives the expansion of ER volume [5]. However, if ER stress is not alleviated in a timely manner, the UPR promotes apoptosis via multiple mechanisms. The switch from pro-survival to pro-apoptotic roles depends on the cell type and the type of insults [6]. One of the key factors in maintaining this balance is the activation of the transcription repressor CHOP, of which level is controlled by all UPR arms, primarily by PERK through ATF4 induction [7]. CHOP has been shown to alter the balance between pro-survival and pro-apoptotic Bcl-2 family members [8,9] and death receptor 5 (DR5) thus promoting apoptosis through the mitochondrial pathway [10]. Depletion of CHOP in both animal models and cell lines showed a significant reduction in ER stressinduced apoptosis [7]. IRE1 possess both pro-survival and pro-death roles. Dependent on its kinase activity, IRE1

Research Paper: Immunology
Oncotarget 39397 www.impactjournals.com/oncotarget activates TNF receptor associated factor 2 (TRAF2) and JNK, which ultimately signals to apoptosis. This activity of IRE1 occurs in parallel to its nuclease activity which mediates the splicing of XBP1 mRNA to generate the transcriptionally active form of XBP1 protein. Depending on the stress strength and persistency as well as on the cell type, active XBP1 can drive either pro-survival or pro-death transcriptional networks [11][12][13][14][15]. XBP1 mRNA is not the only substrate for IRE1 nuclease activity. A growing list of mRNA and miRNA molecules was found to be targeted for degradation by IRE1, in a manner critical to cell survival [16]. Importantly, constitutive ER stress conditions, termed chronic ER stress, usually arise on the background of severe disturbance of homeostasis and are rarely seen physiologically.
Survival of naive newly thymic emigrants T cells is dependent on their ability to maintain quiescence. Quiescent immune cells are characterized by small cell size, lack of spontaneous proliferation and low metabolic rate [17][18][19]. Acquisition of this state is important to provide stress free environment and resistance to apoptosis while keeping the naïve T cells responsive to activating stimuli [17,[19][20][21][22].
Previously, we described a mouse strain bearing a Slfn2-mutated allele, named elektra. The elektra mutation causes an isoleucine-to-asparagine substitution of residue 135 of the 278 amino acid of SLFN2 protein [23]. Using these mice we demonstrated an essential regulatory role for SLFN2 in both innate and adaptive immune responses [23].
In elektra mutant mouse, naïve newly thymic emigrant (CD44 lo ) fail to maintain quiescence and instead acquire a semiactivated phenotype characterized by activation of part of JNK and p38, higher propensity to enter cell cycle as well as downregulation of IL7Ra and CD62L [23]. As a result, upon maturation (CD44 hi ) or activation signals, elektra T cells fail to acquire memorylike phenotype and to engage pro-survival machinery leading to premature apoptosis. In addition to T cells, inflammatory monocytes are also affected by the elektra mutation, exhibiting similar fragility in the face of signals of proliferation or activation [23]. A recent study from our group showed an essential role for Slfn2 in the progression of T cell malignancies such as T-ALL and lymphoma as well as in other diseases evolving aberrant T cell development [24]. These findings highlight the great potential in targeting Slfn2 and other family members for therapeutic purposes, either to manipulate specific immune responses or to suppress blood borne malignancies. However, the mechanism by which Slfn2 maintains quiescent, stress-free environment in T cells is still unknown.
In the present study, we demonstrate that elektra monocytes and T cells exhibit chronic ER stress conditions. By partially preventing the engagement of the UPR response either by CHOP or XBP1 depletion, viability of elektra cells was restored and proliferation capabilities of elektra T cells were improved. These results establish for the first time a functional connection between the loss of quiescence in Slfn2-deficiency to chronic unresolved ER stress.

results er stress regulated genes are elevated in elektra monocytes
In elektra cells both JNK and the p38 pathways are constitutively active without the activation of the ERK1/2 pathway. This phenotype is typical to a variety of stress conditions, such as starvation, ER stress, DNA damage and oxidative stress [23].
To identify which of the stress conditions is responsible for the aberrant activation of the MAPK pathway in elektra cells, we performed an unbiased transcriptome profiling. To avoid possible secondary defects, such as activation of apoptotic signaling pathway mediated by the elektra mutation, we decided to analyze monocyte precursors (CD11b + /ly6C hi ) from the bone marrow. These cells are phenotypically normal and viable in elektra mice [23]. The gene expression profile of the BM monocytes precursors clearly shows elevated levels of cell stress related genes, particularly ER stress, in elektra cells as compare to cells from wild-type mice. Among these are genes coding for members of activating transcription factors/cAMP response element binding protein (ATF/CREB) family; ATF3, ATF4 and ATF5 [25,26]. Up-regulation of these genes has been strongly related to cellular stresses, survival and cell death. Additionally, components of ER stress mediated apoptosis pathway i.e. C/EBP homologous protein (CHOP/DDIT3/GADD153) [8] and TRIB3 [27], an Akt inhibitor, were found to be significantly enriched in elektra cells. Interestingly, during ER stress, CHOP and TRIB3 are induced by ATF4 [27] which is also induced in elektra cells as mentioned above. Furthermore, we also observed up regulation of several chaperones; Hspa5 (encodes for the ER chaperone Bip), Hspb7, Hsph1 and the co-chaperone Dnaja1 (Hsp40) in elektra cells, emphasizing up regulation of the UPR and ER stress [28]. Finally, our results show elevated level of the protein synthesis regulator, Eif2ak2, which phosphorylates and inhibits the translation initiation factor eIF2α leading to translation inhibition, an essential process in UPR [29]. Microarray results were validated by real time PCR ( Figure 1B).
The splicing of XBP1 mRNA is a hallmark of ER stress. Semi quantitative real time-PCR analysis for XBP1 splicing ( Figure 1B), which was also confirmed by PCR analysis for the spliced and unspliced forms of XBP1 ( Figure 1C), demonstrated a constitutive level of the www.impactjournals.com/oncotarget spliced form, condition that is rarely seen for unstimulated cells. These results demonstrate that the elektra mutation in Slfn2 leads to the unabated activation of stress response in BM monocytes.

chronic er stress of elektra t cells
Next we aimed to confirm that the ER stress response is activated also in elektra T cells. For this purpose T cells were isolated from elektra spleens and the expression of ER stress genes was measured in comparison to wt cells.
ER stress response genes were indeed elevated in elektra T cells (Figure 2A), and the ratio between the spliced and unspliced transcripts of XBP1 was elevated ( Figure 2B). In addition, ER tracker staining shown to be decreased in elektra T cells as compare to wt T cells ( Figure 2C) suggesting a defect in ER structure. These data confirm that similarly to elektra monocytes, ER stress response is activated also in elektra T cells. array shows differentially expressed genes in C57BL6/J and Slfn2 eka/eka bone marrow CD11b + ly6C hi sorted monocytes. P-values (highly significant are marked by red font) and average fold of change are included. b. Gene expression analyzed by semi-quantitative RT-PCR of ER stress response genes in C57BL6/J and Slfn2 eka/eka bone marrow CD11b + ly6c hi monocytes (n=4). c. Representative result of XBP1 splicing in Slfn2 eka/eka monocytes. Total RNA extracted from sorted monocytes and subjected to RT-PCR analysis with XBP1 primers.

Elektra t cells die via er stress induced apoptosis
Elektra mice overexpressing BCL2 in the T cell compartment, BCL2(Tg)/Slfn2 eka/eka have normal numbers of T cells. In addition, Bcl2 is downregulated in elektra T cells [23,24]. These results demonstrate that elektra T cells undergo apoptosis via the intrinsic apoptotic pathway [23,24]. Since chronically activation of ER stress promotes apoptosis in several cell types [30], we postulated a contribution of this pathway in the demise of elektra T cells.
To test this possibility, we examined whether elektra T cell death is indeed mediated by chronic ER stress by evaluating whether modulation of the ER stress response by either down regulation of XBP1 or CHOP can rescue the elektra T cell phenotype. To this end, Lck-Cre XBP1 lox/ lox mice, in which XBP1 is knocked out specifically in the T cell compartment, or CHOP knockout mice, CHOP -/-, were crossed to elektra mice to generate Lck-Cre XBP1 lox/ lox / Slfn2 eka/eka or CHOP -/-/Slfn2 eka/eka transgenic mice respectively. The percentages of CD4 + and CD8 + T cells from spleens of Lck-Cre XBP1 lox/lox /Slfn2 eka/eka or CHOP -/-/Slfn2 eka/eka transgenic mice were markedly elevated as compare to those from elektra mice, in fact resembling those of wild-type mice ( Figure 3A and 3B). In addition and in line with these results, BCL2 expression levels were restored in CD8 + T cells from CHOP -/-/Slfn2 eka/eka mice ( Figure 3C). Our data implicate the XBP1/CHOP pathway as mediator of elektra T cell death.
blocking er stress response partially restores elektra t cell semiactivated phenotype Viability is not the only defect of elektra T cells. Previously we showed that elektra T cells exist in a semiactivated state, characterized by down regulation of the IL7Ra (CD127) and the homing receptor L-Selectin (CD62L) on both naïve newly thymic emigrant (CD44 lo ) and on naïve mature (CD44 hi ) T cells. In addition, the genes in naïve T cells purified from C57BL6/J and Slfn2 eka/eka splenocytes (n=4 each, 96-98% and 95-96% purity of wt and elektra T cells respectively). b. XBP1 splicing in C57BL6/J and Slfn2 eka/eka T cells (5 different mice from each genotype). c. Scatter plot presenting densitometry of the bands in B. d. ER tracker blue white dpx staining of both CD4 + and CD8 + T cells obtained from C57BL6/J or Slfn2 eka/ eka mice (n=3 each).
BCL2 overexpression [23] or p53 deficiency [24] rescue elektra T cell death, but does not affect their semiactivated phenotype, and thus does not correct their proliferation defect [23]. These results illustrate the complexity of the elektra phenotype and indicate that Oncotarget 39401 www.impactjournals.com/oncotarget apoptosis of the elektra T cells is a consequence of their acquisition of a semiactivated state.
Therefore, we next assessed whether manipulating the ER stress response by CHOP or XBP1 deficiency also impinges on the acquisition of the semiactivated state of the elektra T cells and their proliferation capabilities. To this end T cells from Lck-Cre XBP1 lox/lox /Slfn2 eka/eka and CHOP -/-/Slfn2 eka/eka mice were further immunophenotyped by flow cytometry.

blocking er stress response partially restores elektra t cell proliferation capacity
Next we examined the impact of ER stress on the proliferative capacity of elektra T cells. Initially, we examined the proliferation capacity of CHOP -/-/Slfn2 eka/ eka or Lck-Cre XBP1 lox/lox /Slfn2 eka/eka T cell in response to activation signals. For this end, wt, CHOP -/-/Slfn2 eka/ eka , Lck-Cre XBP1 lox/lox /Slfn2 eka/eka and elektra splenic T cells were labeled with the cytosolic dye Cell Trace and were stimulated for 72 h in vitro with a combination of antibodies to CD3ε and CD28. In line with our published data [23], elektra T cells completely failed to proliferate in response to activation stimuli ( Figure 5A and 5B and, Supplementary Figure 2A and B). However, as opposed to elektra, both CD8 + and CD4 + T cells from CHOP -/-/ Slfn2 eka/eka and Lck-Cre XBP1 lox/lox /Slfn2 eka/eka had undergone proliferation, although less than T cells from CHOP -/and Lck-Cre XBP1 lox/lox that proliferated comparably to the T cells from wt mice ( Figure 5A and 5B and, Supplementary  Figure 2A and B). These results demonstrate that the proliferation defect of elektra T cells upon activation ex vivo is mediated, at least in part, by the chronic ER stress.
To further test the impact of the elevated ER stress on the proliferative capacity of elektra T cells, we examined CHOP -/-/Slfn2 eka/eka and Lck-Cre XBP1 lox/ lox /Slfn2 eka/eka T cell response to homeostatic proliferation signals. For this end, we labeled wild-type, Slfn2 eka/eka , CHOP -/-/Slfn2 eka/eka and Lck-Cre XBP1 lox/lox /Slfn2 eka/eka (CD45.2) splenocytes with the cytosolic dye Cell Trace and adoptively transferred them into sublethally irradiated wild-type (CD45.1) recipients. As expected wild-type T cells underwent proliferation and we detected no elektra T cells in the spleen of recipient mice 7 d after infusion ( Figure 5C, left panels and Supplementary Figure 3). T cells from both CHOP -/-/Slfn2 eka/eka and Lck-Cre XBP1 lox/ lox /Slfn2 eka/eka failed to proliferate, however were still detectable in a pretty high percentages ( Figure 5C, right panels and Supplementary Figure 3). These results demonstrate that in vivo, under lymphopenic environment, elektra T cell death is mediated by the XBP1/CHOP pathway. However, the proliferation defect of elektra T cells cannot be attributed only to these proteins.

dIscussIon
The outcome of an immune response is dependent on the balance between stimulatory and inhibitory signals. This delicate balance is critical for the ability of the immune system to defend the host, while maintaining tolerance to self so to prevent autoimmunity and immunopathology. To achieve this, the immune system maintains a vast repertoire of immunologically naïve lymphocytes in a quiescent state characterized by inactive cell cycle (arrest in G0), small cell size and relatively low basal metabolic activity [17]. Intuitively, it may be expected that elimination of a regulator of quiescence would have led to hyperactive immune responses and even autoimmunity. However, several recent findings demonstrated that the exact opposite can also happen; disruption of quiescence leads to immunodeficiency due to loss of proliferation capabilities and susceptibility to cell death of T cells [23,[31][32][33][34]. These results suggest that quiescence programming has a broader role in immunity than previously recognized, and that quiescence rather than a passive process is an active process that enforces passiveness. Nevertheless, most of the factors and pathways that maintain immune cell quiescence have yet to be identified.
Naïve quiescent T cells favor energy production over biosynthesis. To accommodate for this need, naïve T cells rely predominantly on the high-energy-yielding mitochondrial metabolism in which metabolites are oxidized via the TCA cycle [18,35]. Upon activation, T cells undergo dramatic shift in cell metabolism switching from oxidative metabolism to aerobic glycolysis, to support their expansion and effector functions [18,35].
Oncotarget 39404 www.impactjournals.com/oncotarget accompanied by development of stress conditions to be met by compensatory responses. This balance can have a substantial influence on cell fate decisions.
Here we showed that elektra T cells display chronic ER stress under steady state conditions ( Figure  2). Modulation of the ER stress response by depletion of either XBP1 or CHOP restored the viability (Figure 3) and partially improved the developmental abnormalities and proliferation capabilities of elektra T cells ( Figures  4 and 5). These results establish a functional connection between the loss of quiescence in elektra T cells and chronic unresolved ER stress.
Previous attempts to rescue elektra T cell viability either by over-expression of BCL2 [23] or p53 deficiency [24] demonstrated that although the propensity to undergo apoptosis was reduced, elektra T cells displayed a semiactivated phenotype and were unable to proliferate. These results strongly indicate that the elektra T cell death is a consequence of loss of quiescence. The results presented here demonstrate that knockout of CHOP or XBP1 not only prevented elektra T cell death, but also partially prevented the semiactivated phenotype and partially restored the T cell proliferative capacity upon activation (Figures 3 and 5).
Deletion of the tuberous sclerosis complex 1 (Tsc1) gene leads to a substantial reduction in peripheral T cell numbers [32,37], which correlates with increased propensity to apoptosis that can be rescued by expression of Bcl-2 [37]. Moreover, Tsc1 deficiency in T cells leads to acquisition of a semiactivated phenotype caused by loss of quiescence [37]. This phenotype highly resembles the phenotype observed in elektra T cells.
Tsc1, together with its partner Tsc2, restrains the activation of mammalian target of rapamycin (mTOR), an important regulator of metabolism and translation that is typically induced following PI3K signaling [38]. In the context of tumors and liver cells, ER stress and activation of the UPR pathway are important pathological features of TSC deficiency; TSC1 and TSC2 deficiency induces ER stress and activates the UPR [37]. The resulting ER stress increases the vulnerability of various cell lines and tumors to apoptosis [37]. Although not validated in TSC1 KO T cells, ER stress may also contribute to the quiescence defect in this model.
The functional connection between the loss of quiescence in Slfn2-deficiency and chronic unresolved ER stress is not known. As noted above, the elektra phenotype is caused by a point mutation in Slfn2 gene. Therefore, one may argue that the elektra mutation possibly results in misfolding and aggregation of the mutated Slfn2 protein leading to the observed ER stress. However, elektra heterozygous mice do not demonstrate any of the defects observed in elektra homozygous mice [23]. In addition, other cells than T cells and monocytes, which also express Slfn2 (e.g. granulocytes, B cells and NK cells), are not influenced by the elektra mutation [23].
Furthermore, Slfn2 overexpression in elektra homozygous completely rescues the entire phenotypes of the elektra mice [23]. Finally, knockdown of Slfn2 by shRNA recapitulate the elektra phenotype in EL4 cells [24]. These evidences strongly suggest that the ER stress, as well as other aspects of the elektra phenotype, is caused by the loss-of-function of the Slfn2 rather than by a gain-offunction caused by misfolding of the protein. Therefore, we propose two options for the underlying mechanism; one possibility suggests that Slfn2 directly regulates ER homeostasis. According to this possibility the UPR should be a main defect in elektra cells. In this regard, Slfn2 can play a role in protein folding in the ER (chaperone), protein translocation into the ER and protein trafficking within and downstream to the ER. A second possibility is that the observed chronic ER stress in elektra cells is a consequence of an overall much broader phenotype caused by Slfn2 deficiency. For example; an unbalanced translational regulation [39], impaired lipid and cholesterol biosynthesis and trafficking [40], disturbances of redox homeostasis [41], glucose deprivation [42] and aberrant Ca +2 regulation [43]. All of these conditions also result in ER stress; however, UPR is only a part of the effect. Interestingly, the mechanism by which TSC deficiency results in ER stress induced cell death was attributed to increased mTOR activity [37], which triggers uncontrolled protein synthesis. Moreover, Li, M. et al. demonstrated that the human SLFN family member, Slfn11, inhibits retroviral protein translation in a codon-usage dependent manner [44]. Considering these results and our findings, it is appealing to suggest that in elektra cells the chronic ER stress is mediated due to loss of protein translational control which leads among several other defects also to inability to match protein load to the folding capacity of the ER, supporting the second option. This scenario suggests a potential regulatory role for Slfn2 in protein translation. Therefore, models that directly explore the role of protein translation in T cell quiescence should be developed.

real time Pcr
Total RNA from purified monocytes or T cells was extracted with Direct-zol RNA MiniPrep Plus (Zymo Research). cDNA was synthesized using ProtoScript First Strand cDNA Synthesis Kit (Neb). Quantitative realtime PCR was then performed using QuantStudio 12K Flex Real Time PCR system with a Power SYBR green PCR master mix kit (Applied Biosystems). The primers sequences used in this study are previously described [46].

Flow cytometry
Spleen cells were stained with various conjugated mAbs against cell-surface markers. In some of the experiments, following the cell surface staining, cells were fixed, permeabilized and stained with intracellular antibody using Leucoperm-kit (AbD Serotec) according to manufacturer instructions. ER tracker staining was performed by incubating cells with 1µM ER-Tracker blue-white DPX (Molecular Probes) diluted in Hankʼs Balanced Salt Solution for 30 minutes in 37˚C. Stained cells were analyzed by Gallios flow cytometer with Kaluza software (Beckman Coulter).

Inflammatory monocytes and T cells isolation
For monocytes isolation total bone marrow cells were harvested and inflammatory monocytes were isolated by EasySep mouse monocyte isolation kit (Stem Cell Technologies). For further purification, CD11b + Ly6C high monocytes were sorted by flow cytometry. Total T cells were isolated from spleen by EasySep mouse T cell isolation kit (Stem Cell Technologies).

t cell proliferation assay
A total of 2.5 × 10 6 cell-trace violet-labeled spleen cells were activated in 24-flat-well plates previously bounded with anti-CD3ε (2µg/ml) and anti-CD28 (1µg/ ml) and IL-2 (20ng/ml). The cells were analyzed 72 h after activation by flow cytometry as described in the adoptive transfer section.