Th2 Skewing by Activation of Nrf2 in CD4⁺ T Cells

Thelper cell 1/Th2 differentiation of CD4⁺ T cells is a critical process in tailoring an adaptive immune response to a specific pathogen, which allows for flexibility in T cell function and downstream immune activity. However, there is also an inherent potential danger in pathological responses that are due to Th1/Th2 imbalance. Th1 responses are vital in cell-mediated immunity, which is important in host defense against numerous bacterial and viral pathogens (1). Conversely, Th2 responses are thought to be important in host defense against larger parasites, such as helminths. Th2 responses also play a major pathological role in the development of allergy and asthma (1, 2). Although the incidence of allergy and asthma has sharply increased in the United States and worldwide over the last several decades, the underlying cause is unknown (3–5).

NF erythroid 2-related factor 2 (Nrf2) is a basic leucine zipper transcription factor that is activated by cell stress from various sources, including oxidative and electrophilic stimuli (6, 7). In the absence of oxidants and electrophiles, Nrf2 is sequestered in the cytosol by its repressor protein, Kelch ECH associating protein 1 (Keap1), and is subsequently ubiquitinated and degraded (8, 9). During cell stress, ubiquitination of Nrf2 by Keap1 is disrupted, allowing Nrf2 to translocate to the nucleus and upregulate genes containing an antioxidant response element (ARE) in their promoter regions (10–12). Nrf2-regulated genes facilitate a variety of functions, including, but not limited to, antioxidative activity, detoxification and transport of xenobiotics, and glutathione homeostasis (13–15). In immune cells, Nrf2 was reported to upregulate numerous genes, such as Hmox1, Nqo1, and Gclc. Therefore, these genes are useful markers of Nrf2 activation in leukocytes (16, 17).

The synthetic antioxidants, tert-butylhydroquinone (tBHQ) and butylated hydroxyanisole (BHA), are widely used experimentally as Nrf2 activators. Specifically, it was shown that tBHQ modifies the thiol groups of cysteine-273 and cysteine-288 on the Keap1 protein, which ultimately inhibits the repression of Nrf2 by Keap1 and facilitates Nrf2 signaling (18). With respect to commercial use, both tBHQ and BHA are food preservatives used to stabilize oils and fats against oxidative deterioration (19). Although banned for use in food in Japan and other countries, tBHQ and BHA are used as preservatives in numerous types of food in the United States, including cereals, crackers, potato chips, noodles, grains, vegetable oils, margarine, and others.

Recent studies demonstrated anti-inflammatory effects of Nrf2 activation and conversely, the proinflammatory effects of Nrf2 deletion, in a number of models. For instance, Nrf2-null mice were shown to have increased mortality in models of sepsis (20), more severe pathology in experimental autoimmune encephalomyelitis (21), increased production of inflammatory mediators in LPS-induced neuroinflammation (22), and worsened injury in a model of T cell-mediated hepatitis (23). With respect to lung inflammation, Nrf2-null mice have increased susceptibility to OVA-induced asthma (24), allergic airway inflammation (25), and hyperoxia-induced acute lung injury (26). Accordingly, there has been considerable interest in developing anti-inflammatory therapeutics targeting Nrf2 (16).

Further evidence for an important role for Nrf2 in immune regulation is derived from the development of autoimmune disease in Nrf2-null mice. Several articles have characterized the development of multiorgan autoimmune inflammation in Nrf2-null mice, which resembles systemic lupus erythematosus (SLE) in humans (14, 27, 28). Like SLE in humans, the autoimmune disease in Nrf2-null mice is found predominantly in adult females, progresses with age, is characterized by the production of Abs against dsDNA, and ultimately results in glomerulonephritis. Moreover, it was reported that, in addition to the SLE-like disease, 14-mo-old Nrf2-null female mice develop hemolytic anemia, which often occurs in human SLE patients (29). Notably, a recently identified NRF2 polymorphism was shown to be associated with autoimmune nephritis in female SLE patients (30).

In addition to its general anti-inflammatory effects, a number of studies suggested that Nrf2 activation may alter Th1/Th2 balance specifically. Expression of IL-4 and IL-13 is elevated in lungs of Nrf2-null mice in models of asthma and pulmonary fibrosis (24, 31). In addition, Nrf2-deficient lung dendritic cells were shown to have heightened oxidative stress that confers a Th2-like immunoresponsiveness upon stimulation with ambient particulate matter (32). With respect to Th1 responses, activation of Nrf2 by sulforaphane was shown to restore the age-related decrease in dendritic cell responses in a model of contact hypersensitivity (33). Collectively, these studies initiated the investigation of Nrf2 in Th1/Th2 responses and served as the basis for the current studies.

The present studies investigated the role of Nrf2 in T cell function and differentiation. These studies indicated that Nrf2 is activated by tBHQ in T cells, as evidenced by the binding of Nrf2 to its response element, ARE, and subsequent upregulation of Nrf2 target genes. The current studies also demonstrated that Nrf2 activation inhibits Th1 cytokine production and T-bet activity, while promoting Th2 cytokine production and GATA-binding protein 3 (GATA-3) binding activity. In addition, the Nrf2 activator, tBHQ, induced Il4-regulated firefly luciferase (Fluc) activity and Il13-regulated Renilla luciferase (Rluc) activity in CD4⁺ T cells isolated from Il4-Fluc/Il13-Rluc transgenic mice, suggesting that Nrf2 promotes transcriptional activity in both the Il4 and Il13 genes.

BHA, tBHQ, and all other reagents were purchased from Sigma-Aldrich (St. Louis, MO), unless otherwise noted.

Nrf2-null mice on a mixed C57BL/6 and AKR background were generated, as described previously, and received from Dr. Jefferson Chan at the University of California San Francisco, San Francisco, CA (6). The mice were subsequently back-crossed for eight generations onto the C57BL/6 background and were 99% congenic (analysis performed by The Jackson Laboratory, Bar Harbor, ME). Female mice, 8–14 wk of age, were used for the current studies. Wild-type female C57BL/6 mice, 8 wk of age, were purchased from Charles River Laboratories (Wilmington, MA). Mice were given food and water ad libitum. All animal studies were conducted in accordance with the Guide for the Care and Use of Animals, as adopted by the National Institutes of Health, and were approved by the Institutional Animal Care and Use Committee at the University of Kansas Medical Center.

Single-cell suspensions from spleens and lymph nodes were washed and filtered, after which CD3⁺ or CD4⁺ T cells were isolated with magnetic beads by negative selection, following the manufacturer’s instructions (Miltenyi Biotec, Auburn, CA). The total number of cells isolated varied, depending on the specific experiment. Cellular viability was >90%, as determined by trypan blue exclusion. The purity of the isolated cell populations was >90%, as determined by flow cytometric analyses.

Cells were cultured in DMEM (with l-glutamine, sodium bicarbonate, and d-glucose), supplemented with 100 U penicillin/ml, 100 U streptomycin/ml, 50 μM 2-ME, and 10% FCS. For most studies, cells were treated with either tBHQ or BHA at various concentrations for 30 min prior to T cell activation. The duration of T cell activation ranged from 6 h (mRNA quantification of Nrf2 target genes) to 96 h or more (cytokine analysis), depending on the specific end point measured (noted in the figure legends). T cells were activated with purified hamster anti-mouse CD3e (500A2, 1.5 μg/ml), purified hamster anti-mouse CD28 (37.51, 1.5 μg/ml), and an F(ab′)₂ fragment specific for anti-Syrian hamster IgG that was used to cross-link CD3 and CD28 to enhance activation. Anti-CD3 and anti-CD28 were purchased from BD Biosciences (San Jose, CA), and the F(ab′)₂ cross-linker was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

For studies not restricted by limited amounts of total RNA, branched DNA analysis was performed for detection of mRNA. Cells were treated, as described, after which they were lysed with diluted lysis buffer. The cell lysis and branched DNA assays were performed using a commercially available kit (Quantigene 1.0; Panomics, Fremont, CA), following the manufacturer’s protocol, as described previously (34). The specific oligonucleotide probe sets used for the quantification of Ifng are shown in Supplemental Table I. The probe sets were designed using ProbeDesigner software (Bayer, Diagnostics Division, Tarrytown, NY), as described previously (35). Probe sets were synthesized by Integrated DNA Technologies (Coralville, IA).

For single-plex analysis, IFN-γ protein was quantified by a sandwich ELISA method, using a commercially available kit and following the manufacturer’s protocol (eBioscience, San Diego, CA).

For studies for which limited total RNA was available, real-time PCR was used for mRNA quantification. Total RNA was isolated using TRI Reagent (Sigma-Aldrich), following the manufacturer’s protocol. The isolated RNA was then reverse transcribed, after which relative expression levels of the target genes were determined by SYBR Green real-time PCR. Relative mRNA expression for the target genes was normalized to ribosomal protein L13a and calculated using the DDCt method, as previously described (36). Primers for rpL13a were designed using Primer3 software (http://primer3.sourceforge.net) and were synthesized by Integrated DNA Technologies: rpL13a forward primer, 5′-ACAAGAAAAAGCGGATGGTG-3′ and rpL13a reverse primer, 5′-TTCTCCTCCAGAGTGGCTGT-3′. The primers for Nfe2l2 (Nrf2), Hmox1, Nqo1, and Gclc have been published (28, 37, 38).

Nuclear protein was isolated using the Nuclear Extract kit (Active Motif, Carlsbad, CA), following the manufacturer’s protocol. Protein concentrations were determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA), which is based on the Bradford method of protein quantification.

Nrf2 (ARE)-, T-bet–, and GATA-3–binding activities were quantified using TransAM assay kits (Active Motif), following the manufacturer’s protocol. The wild-type oligonucleotide (provided by the manufacturer) is an unlabeled free-consensus oligonucleotide that competes with the plate-bound oligonucleotide to confirm specificity. In addition to the wild-type oligonucleotide, a mutated oligonucleotide was used as a control. The Cos-7 (Nrf2 transfected), Cos-7 (T-bet transfected), and Jurkat nuclear extracts served as positive controls for the Nrf2-, T-bet–, and GATA-3–binding assays, respectively.

For multiplex cytokine analysis, cytokine concentrations were determined using a MilliPlex mouse cytokine kit for the detection of IFN-γ, TNF-α, IL-4, IL-5, IL-13, and IL-10, following the manufacturer’s protocol (Millipore, Billerica, MA). The cytokine concentrations were quantified using a Bio-Plex 200 System (Bio-Rad). PE and xMAP bead fluorescence was detected by a dual-laser detector (532 and 635 nm) and quantified by Bio-Plex Manager 5.0 software (Bio-Rad).

A 184-kb bacterial artificial chromosome (BAC) clone (RP24-259D13) that includes both the Il4 and Il13 genes from chromosome 5 was selected from a BAC library and used to generate the Il4-Fluc/Il13-Rluc transgenic mouse (manuscript in preparation). In addition to the Il4 and Il13 genes, the BAC clone includes the Il5 gene, as well as the proximal regulatory regions of Il5; thus, it includes the majority of the Th2 locus (it is missing distal regulatory regions for the Il5 gene, however). The first exon of the Il4 gene was replaced with Fluc, and the first exon of the Il13 gene was replaced with Rluc by homologous recombination. The purified BAC DNA was then introduced into fertilized murine oocytes by pronuclear microinjection. The fertilized oocytes were then transferred into pseudopregnant foster mothers. The copy number of the transgene in the pups was determined by quantitative real-time PCR, as previously described (39). A detailed description of the generation of Il4-Fluc/Il13-Rluc transgenic mice is being prepared for publication.

The mean ± SE was determined for each treatment group in the individual experiments. Homogeneous data were evaluated by one-way or two-way parametric ANOVA. When significant differences were observed, either the Dunnett (for one-way ANOVA) or the Holm–Sidak (for two-way ANOVA) post hoc test was used to compare treatment groups to the vehicle control using SigmaStat 3.01a software from Systat Software (Chicago, IL).

Although Nrf2 has been described as anti-inflammatory in various in vivo models of inflammation, its role in T cells remains largely uncharacterized. Accordingly, the effects of Nrf2 activation on T cell activity were investigated. In our initial studies, a mixed splenocyte population was used in combination with a T cell-specific activator, anti-CD3/anti-CD28. A marked concentration-dependent inhibition of IFN-γ production by the Nrf2 activator tBHQ was observed in CD3/CD28-activated splenocytes (Fig. 1A). It is notable that even low concentrations of tBHQ (0.01 μM) substantially decreased IFN-γ secretion (58%), whereas higher concentrations completely abolished IFN-γ production. Treatment with another Nrf2 activator, BHA, also significantly reduced IFN-γ production by T cells, causing a marked decrease in IFN-γ secretion at 1 μM (data not shown). Neither tBHQ nor BHA significantly affected viability (data not shown). Consistent with its effect on protein secretion, tBHQ also inhibited IFN-γ mRNA expression in wild-type splenocytes, whereas splenocytes from Nrf2-null mice were relatively refractory to this effect (Fig. 1B). In addition, markedly higher mRNA expression of IFN-γ was observed in anti-CD3/anti-CD28–activated splenocytes derived from Nrf2-null mice relative to wild-type mice (Fig. 1C), suggesting Nrf2 plays an endogenous role in the regulation of IFN-γ. There was no difference in the ratios of CD4⁺, CD8⁺, CD19⁺, CD4⁺CD25⁺CD62L^hiCD44^lo (naive), CD4⁺CD25⁺CD62L^loCD44^hi (effector memory), or CD4⁺CD25⁺ populations (Supplemental Figs. 1, 2). These studies demonstrated that activation of Nrf2 can inhibit IFN-γ production by activated T cells and that Nrf2 plays a critical role in the endogenous regulation of IFN-γ.

Because activation of Nrf2 inhibited IFN-γ production by anti-CD3/anti-CD28–activated splenocytes (Fig. 1), the expression of Nrf2 and its ability to be activated was assessed in the pan T cell population. Similar to other cell types, Nrf2 expression in T cells is upregulated after treatment with the Nrf2 activator tBHQ (40), suggesting that Nrf2 upregulates its own expression (Fig. 2A). In addition, treatment of wild-type T cells with tBHQ resulted in increased binding of Nrf2 to the ARE, whereas decreased binding of Nrf2 to the ARE was observed in T cells derived from Nrf2-null mice (Fig. 2B). The increased binding of Nrf2 to the ARE correlated with increased mRNA expression of the Nrf2-target genes, Hmox1, Nqo1, and Gclc, in tBHQ-treated wild-type T cells. In contrast, treatment of Nrf2-null T cells with tBHQ did not upregulate Nqo1 or Gclc and only modestly induced Hmox1 expression (Fig. 2C–E). Collectively, these data indicated that Nrf2 is expressed, inducible, and capable of gene transactivation in wild-type T cells, as evidenced by the detection and induction of Nrf2 mRNA levels, binding of Nrf2 to the ARE, as well as the Nrf2-dependent upregulation of the target genes Hmox1, Nqo1, and Gclc.

The early effects of tBHQ on IFN-γ production prompted an investigation into the effects of Nrf2 activation in the CD4⁺ T cell polarization. Similar to the early effects shown in Fig. 1, treatment of CD4⁺ T cells with tBHQ over 96 h resulted in inhibition of IFN-γ secretion, whereas substantially increased IFN-γ production was observed in CD4⁺ T cells from Nrf2-null mice (Fig. 3A). In correlation, tBHQ treatment increased IL-4 secretion in wild-type CD4⁺ T cells, whereas decreased IL-4 production was observed in CD4⁺ T cells from Nrf2-null mice (Fig. 3B). Treatment with tBHQ also increased IL-5 and IL-13 secretion by CD4⁺ T cells, and, accordingly, the absence of the Nrf2 gene resulted in decreased IL-5 and IL-13 production (Fig. 3C, 3D). The decreased IFN-γ production and increased IL-4, IL-5, and IL-13 production by tBHQ were also observed in CD4⁺ T cells from AhR^d mice, which have a defective aryl hydrocarbon receptor, suggesting the effects are independent of the aryl hydrocarbon receptor (Supplemental Fig. 3). There was little difference in TNF-α and IL-10 secretion between the Nrf2-null and wild-type genotypes, suggesting that the effects of Nrf2 activation are relatively selective for Th1/Th2 cytokines (Fig. 3E, 3F). Furthermore, the effects of Nrf2 activation at 96 h were also observed in restimulated CD4⁺ T cells in which wild-type CD4⁺ T cells treated with tBHQ exhibited decreased IFN-γ production and increased IL-4, IL-5, and IL-13 secretion (Fig. 4). Conversely, IL-4, IL-5, and IL-13 levels were low in restimulated CD4⁺ T cells from Nrf2-null mice. Collectively, the data suggested that activation of Nrf2 promotes Th2 differentiation of CD4⁺ T cells, while inhibiting Th1 differentiation.

Because cytokine analyses suggested that tBHQ promotes Th2 differentiation, the effect of Nrf2 activation on transcription factors involved in Th1/Th2 differentiation (T-bet and GATA-3) was assessed. Consistent with the cytokine analyses, activation of Nrf2 suppressed binding of T-bet to its consensus sequence and concurrently potentiated GATA-3 binding to its consensus sequence (Fig. 5). Furthermore, CD4⁺ T cells from Nrf2-null mice exhibited significantly increased T-bet DNA binding and suppressed GATA-3 DNA binding. Collectively, these studies provided further evidence that Nrf2 positively regulates Th2 differentiation, which is associated with enhanced GATA-3–binding activity.

Recently, we engineered a transgenic mouse model with which to study the molecular events that occur during Th2 differentiation (manuscript in preparation). A 184-kb BAC clone (RP24-259D13) from mouse chromosome 5 with Il4 and Il13 at center and flanked by kif3a and rad50 genes was used to generate BAC transgenic mice. The BAC clone was modified by excising the first exon of the Il4 gene and replacing it with the Fluc gene. In addition, the first exon of the Il13 gene was replaced with the Rluc gene. The modified Th2 locus was then inserted into the mouse genome to create the transgenic mouse. This unique model allows for focused investigation into the molecular modifications that occur in the Th2 locus during CD4⁺ T cell differentiation. Treatment of CD4⁺ T cells from this transgenic mouse with tBHQ caused a robust increase in both Fluc and Rluc activities, suggesting increased transcriptional activity in both the Il4 and Il13 genes (Fig. 6).

Previously published studies demonstrated that Nrf2 plays a significant immunomodulatory role in a number of models of inflammation (20, 22, 24). Although many of these studies suggested that Nrf2 can affect various aspects of T cell biology, its specific role in T cell function is incompletely understood (17, 21, 23, 24, 33). Accordingly, the purpose of the current study was to determine the role of Nrf2 in peripheral T cell differentiation.

Our data demonstrated that Nrf2 is expressed and is able to be activated by tBHQ in T cells, as determined by binding of Nrf2 to the ARE and induction of Nrf2 target genes. Importantly, these studies indicated that activation of Nrf2 inhibits production of the Th1 cytokine IFN-γ, while concurrently promoting the secretion of Th2 cytokines by wild-type CD4⁺ T cells during both primary and secondary activation. Conversely, CD4⁺ T cells from Nrf2-null mice produce increased IFN-γ and decreased IL-4, IL-5, and IL-13 compared with wild-type T cells. In addition, the present studies demonstrated that tBHQ suppressed T-bet DNA binding and enhanced GATA-3 DNA binding in wild-type, but not Nrf2-null, CD4⁺ T cells. In Il4/Il13-regulated luciferase transgenic mice, tBHQ treatment induced Fluc, regulated by the Il4 gene, as well as Rluc, regulated by the Il13 gene. Collectively, these studies indicated that Nrf2 activation in CD4⁺ T cells results in Th2 skewing, whereas lack of functional Nrf2 results in Th1 skewing. Overall, these studies suggested that the Nrf2-signaling pathway represents a novel regulatory mechanism in Th1/Th2 differentiation that warrants further investigation.

It is interesting to note that Nrf2-null female mice develop an autoimmune disease that resembles SLE in humans (27, 28). The autoimmunity in the Nrf2-null mice is characterized by production of autoantibodies against dsDNA; deposition of IgM, IgG, and C3 in the glomeruli; and the development of glomerulonephritis. Although autoantibody production in SLE may suggest a Th2 response, there is considerable evidence to indicate that IFN-γ contributes to the pathology of lupus nephritis in both humans and mice (41–48). Neither IFN-γ^−/− MRL/lpr nor IFNγ^+/− MRL/lpr mice develop the overt renal disease that is typically associated with the MRL/lpr model of SLE (41). In addition, SLE patients with diffuse proliferative lupus nephritis have increased numbers of IFN-γ–producing cells (42, 43). Increased IFN-γ sera concentrations were also reported in SLE patients (44, 45). A higher incidence of single nucleotide polymorphism mutations in both IFN-γR1 and IFN-γR2 has been found in SLE patients (46). Importantly, patients who were treated with IFN-γ for rheumatoid arthritis or myeloproliferative disorder developed a severe SLE-like disease, suggesting that IFN-γ plays a causative role in SLE in some patients (47, 48). Consistent with the aforementioned studies, our data demonstrated that CD4⁺ T cells derived from the Nrf2-null mice, which develop an SLE-like autoimmunity, produced markedly increased amounts of IFN-γ compared with wild-type.

The food additive tBHQ belongs to a family of phenolic antioxidant preservatives, which also includes BHA and butylated hydroxytoluene, all of which are added to a variety of foods to avert and/or slow spoilage. Because there has been considerable concern that many of the phenolic antioxidants may cause cancer, the primary toxicological focus has been the carcinogenicity of these chemicals (19). In this regard, the phenolic antioxidants have been widely studied and appear to be noncarcinogenic to humans. In contrast, the immunotoxicity of the phenolic antioxidants is not well characterized. It was reported that BHA inhibits IL-2 production by T cells (49); however, to our knowledge, the current study is the first to report the mechanism by which this may occur. This is also, to our knowledge, the first report of the effects of tBHQ on T cell function and differentiation.

The effects of the food preservatives tBHQ and BHA on CD4⁺ T cell function occur at concentrations that are of physiological relevance to humans. Concentrations of tBHQ as low as 10 nM produce significant inhibition of IFN-γ secretion, whereas BHA inhibits IFN-γ at 1 μM (Fig. 1, data not shown). Notably, human volunteers administered tBHQ orally (100–150 mg) had serum concentrations of tBHQ ranging from 24 to 222.5 μM, concentrations ≥2–20-fold higher than those used in the current study (50). Thus, the modulation of Th1/Th2 differentiation by tBHQ occurs at concentrations to which humans are likely exposed. Given that humans are exposed to tBHQ by ingestion, and the present studies suggested that tBHQ can enhance production of Th2 cytokines by CD4⁺ T cells, it seems possible that ingestion of tBHQ and potentially other phenolic antioxidants may create an environment in the gut that is permissive to food allergy or other Th2-skewed imbalances. Further studies are needed to investigate the role of tBHQ in oral tolerance and atopic responses.

Inhibition of IFN-γ transcription by tBHQ appears to be largely Nrf2 dependent. However, at higher concentrations, tBHQ can also suppress IFN-γ production in a Nrf2-independent fashion (Fig. 1B). This suppressive activity may be due to impaired calcium influx, as was described to occur with di-tBHQ, a structurally related congener of tBHQ (51).

The current studies suggested that activation of Nrf2 in CD3/CD28-stimulated CD4⁺ T cells skews differentiation toward a Th2 response. This effect was unexpected, because previous studies suggested that Nrf2 may serve to limit or regulate Th2 responses in models of asthma (24, 32, 52). The reason for the differential effects of Nrf2 on Th2 cytokine responses is unclear, but it may due to a number of factors. The current studies assessed the effects of Nrf2 activation on isolated CD3/CD28-stimulated CD4⁺ T cells, whereas asthma and experimental models of asthma are complex, with numerous cell types involved. In addition, the environment of the lung itself appears to be particularly conducive to Th2 responses, whereas the present studies are likely to be more neutral with respect to Th1/Th2 differentiation.

An important implication of these studies is that activation of Nrf2 by food additives, such as tBHQ and BHA, may compromise cell-mediated immunity by impairing the production of the signature Th1 cytokine, IFN-γ. Furthermore, induction of the Th2 cytokines, IL-4, IL-5, and IL-13, may create an environment that is conducive for the development of atopy or other Th2-mediated conditions. Because humans are exposed to tBHQ through ingestion of food, the development of food allergies may be of particular concern. Notably, there has been an increase in reports of food allergy that seems to correlate with the increased use of tBHQ and other phenolic antioxidants as food preservatives (53–55). Collectively, the current findings suggested that tBHQ and other food preservatives may modulate immune responses, which warrants further investigation.

We thank Dr. Jefferson Chan for generously providing the Nrf2-null mice and Dr. Scott Reisman, Dr. Jie Liu, and Kai Wu for technical assistance with the animals used in the current study. In addition, Dr. Joyce Slusser and Michael Stewart provided technical assistance with the flow cytometry studies. We also thank Dr. Bryan Copple for kindly providing materials used in these studies as well as the post-docs and students of the Klaassen laboratory for careful reading of the manuscript.

The authors have no financial conflicts of interest.