Modulation of Th1 Activation and Inflammation by the NF-κB Repressor Foxj1

See allHide authors and affiliations

Science  13 Feb 2004:
Vol. 303, Issue 5660, pp. 1017-1020
DOI: 10.1126/science.1093889


Forkhead transcription factors play key roles in the regulation of immune responses. Here, we identify a role for one member of this family, Foxj1, in the regulation of T cell activation and autoreactivity. Foxj1 deficiency resulted in multiorgan systemic inflammation, exaggerated Th1 cytokine production, and T cell proliferation in autologous mixed lymphocyte reactions. Foxj1 suppressed NF-κB transcription activity in vitro, and Foxj1-deficient T cells possessed increased NF-κB activity in vivo, correlating with the ability of Foxj1 to regulate IκB proteins, particularly IκBβ. Thus, Foxj1 likely modulates inflammatory reactions and prevents autoimmunity by antagonizing proinflammatory transcriptional activities. These results suggest a potentially general role for forkhead genes in the enforcement of lymphocyte quiescence.

The Fox (forkhead) family encompasses a growing number of helix-loop-helix transcription factors that participate in diverse biological systems (1). In the immune system, recent studies have identified roles for Foxp3 in regulatory T cells (2), the Foxo members in lymphocyte proliferation and apoptosis (3), and Foxn1 in thymic epithelial development (4). However, possible roles for the many other forkhead family members remain undefined (5). We previously observed that Foxj1 (HNF-3, forkhead homolog-4, HFH-4, or FKHL-13) was significantly down-regulated in lymphocytes from lupus-prone mice (6). However, other work has described Foxj1 only in the developmental regulation of ciliated epithelial cells (7, 8). Nonetheless, an examination of Foxj1 expression revealed prominent levels in naïve T and B cells; this was rapidly down-regulated during stimulation (figs. S1 and S2).

The vast majority of mice with targeted mutations in Foxj1 die in utero and/or in the early neonatal period because of hydrocephalus and/or heterotaxy (7, 8). We therefore pursued fetal liver chimerization of recombinase activating gene (Rag)–/– mice in order to generate animals possessing either a Foxj1+/+ and Foxj1–/– lymphoid system (9, 10). By 12 to 16 weeks postreconstitution, a large proportion of animals that had received Foxj1–/– fetal livers appeared moribund, with ruffled fur and a hunched posture (10 of 12 versus 0 of 10 of their Foxj1+/+ counterparts, P < 0.001). Upon histopathological examination, Foxj1–/– chimeras displayed evidence of systemic autoimmune inflammation, including moderate to severe infiltrates of the salivary gland, liver, lung, and kidney, which were absent in Foxj1+/+ counterparts (Fig. 1). Immunocytometric examination revealed these infiltrates to be a heterogeneous population of CD4+, CD8+, and B220+ cells, with few Gr-1+ or Mac-1+ cells, suggesting a lymphocytic etiology to the pathology (6).

Fig. 1.

Autoimmune inflammation in the absence of Foxj1. Note the inflammation of the organs in Foxj1–/– chimeric animals, in contrast to normal-appearing tissues in Foxj1+/+ chimeras. Each panel is from a different animal, representative of 10 animals examined for each genotype.

No evidence of systemic humoral autoimmunity, as judged by serum assays for antinuclear antibody, anti-DNA, and rheumatoid factor, was observed, suggesting the absence of B cell hyperactivation (6). Because other Fox transcription factors have principally been described to play immunoregulatory roles in CD4+ T cells (2, 3), we considered that an abnormality in this lymphocyte subset might account for our findings. Importantly, Foxj1–/– chimeras demonstrated grossly normal T and B cell development, as judged by flow cytometry of peripheral lymphoid organs (fig. S3). Additionally, total cellularity of spleens and lymph nodes were comparable between Foxj1+/+ and Foxj1–/– chimeras, although Foxj1–/– chimeras consistently possessed slightly higher proportions of peripheral CD4+ T cells (60 to 70% versus 50 to 56%). In addition, CD4+ cells in Foxj1–/– chimeras displayed surface phenotypes consistent with increased in vivo activation, including modestly higher proportions of cells bearing the CD44hiCD45RBloCD62LloCD25hi-activated phenotype (fig. S3).

Bulk CD4+ Th populations from Foxj1–/– chimeras stimulated with plate-bound anti-CD3 produced two- to fourfold more interleukin-2 (IL-2) and five- to sevenfold more interferon-γ (IFN-γ) than Foxj1+/+ counterparts and appeared to be significantly impaired in the production of IL-4 (Fig. 2A). This suggested either impaired Th2 or overactive Th1 differentiation. During primary stimulation, naïve Foxj1–/– T cells consistently produced elevated quantities of IL-2 and IFN-γ in response to CD3 ligation, often without requiring supplemental IL-2 and/or CD28 costimulation (Fig. 2B), and continued to do so during secondary stimulation (Fig. 2C). Indeed, during primary stimulation, naïve Foxj1–/– Th cells consistently upregulated the Th1 transcription factor T-bet out of proportion to the Th2 transcription factor GATA-3 such that their T-bet/GATA-3 ratio was somewhat higher than that of Foxj1+/+ counterparts (7.2 ± 1.0 versus 3.1 ± 0.3, P < 0.001), consistent with accentuated Th1 skewing (6, 11). However, Foxj1–/– Th cells were not impaired in Th2 development: In fact, upon in vitro stimulation under Th2 development conditions, they produced increased amounts of the Th2 cytokines IL-4 and IL-5, as well as IL-6 and IL-10 (fig. S4), compared to their Foxj1+/+ counterparts (Fig. 2C). We therefore concluded that Foxj1 most likely acts as a direct repressor of T cell activation; however, the overproduction of Th2 cytokines by Foxj1–/– T cells suggested that simple exaggerated skewing to the Th1 lineage was alone insufficient to explain our findings.

Fig. 2.

Exaggerated Th cytokine production in the absence of Foxj1. (A) Secreted cytokines produced by bulk CD4+ T cells from Foxj1–/– (black bars) and Foxj1+/+ (open bars) chimeras, stimulated with plate-bound anti-CD3. For all cytokines, Foxj1–/– T cells were significantly different than their Foxj1+/+ counterparts (P < 0.001). WT, wild type; KO, knockout. (B) Secreted cytokines produced by naïve CD4+ T cells from Foxj1–/– (circles) and Foxj1+/+ (squares) chimeric animals, stimulated with the indicated amounts of plate-bound anti-CD3 in the presence (black) or absence (open) of IL-2, with (lower graphs) or without (upper graphs) anti-CD28. For both cytokines, Foxj1–/– T cells were significantly different than their Foxj1+/+ counterparts (P < 0.001) at all anti-CD3 doses tested except for 0 μg/mL, as well as 0.1 and 0.3 μg/mL for anti-CD28 supplemented samples. (C) Cytokines secreted by Th1- or Th2-differentiated Th cells from Foxj1–/– (circles) and Foxj1+/+ (squares) chimeric animals, stimulated with the indicated amounts of plate-bound anti-CD3 in the presence (black) or absence (open) of IL-2. Where indicated, anti–IL-2 neutralizing antibody was substituted for IL-2. Foxj1–/– Th1 cells were significantly different than their –sufficient counterparts for both cytokines at all anti-CD3 doses above 0 μg/mL (P < 0.001); Foxj1–/– Th2 cells were significantly different than their Foxj1+/+ counterparts for IL-4 at 3 and 10 μg/mL anti-CD3, and for IL-5 at 10 μg/mL anti-CD3 (P < 0.01). Error bars indicate standard deviations. Asterisks indicate undetectable by assay (<1 ng/mL).

Instead, the overproduction of cytokines by Foxj1–/– Th cells likely reflected, in large part, a hyperproliferation of differentiated cytokine-producing cells (Fig. 3): Naïve Foxj1–/– Th cells showed elevated proliferation over their Foxj1+/+ counterparts, even in the absence of CD3 ligation (Fig. 3A). Furthermore, unlike Foxj1+/+ cells, naïve Foxj1–/– Th cells required only the presence of IL-2 in order to proliferate (Fig. 3A), suggesting that these T cells were autoproliferative, as also indicated by the development of autoimmune-like inflammation in vivo. Although naïve Foxj1–/– Th cells failed to proliferate when incubated in the presence of autologous antigen-presenting cells (APCs) (6), Foxj1–/– but not Foxj1+/+ Th cells that had been previously primed with anti-CD3 and anti-CD28 in vitro proliferated significantly (Fig. 3B). This effect was largely independent of IL-2, because treatment with neutralizing IL-2 antibody did not significantly diminish the level of autoinflammatory proliferation. Thus, Foxj1–/– T cells are prone to autoreactivity as well as hyperactivation in response to autologous or self targets.

Fig. 3.

T cell proliferation and autoproliferation in the absence of Foxj1. (A) Proliferative capacity of naïve CD4+ T cells from Foxj1–/– (circles) and Foxj1+/+ (squares) chimeric animals, stimulated with the indicated amounts of plate-bound anti-CD3 in the presence (black) or absence (open) of IL-2, with (lower graph) or without (upper graph) anti-CD28. (B) Autologous mixed lymphocyte reactivity of CD4+ T cells from Foxj1–/– (black bars) and Foxj1+/+ (open bars) chimeras. Th cells primarily stimulated for 6 days under Th-neutral conditions were incubated in the presence or absence of autologous splenic APCs, in the presence or absence of neutralizing IL-2 antibody. Both Foxj1+/+ and Foxj1–/– cells proliferated vigorously (optical density > 1.0) to concanavalin A (6).

NF-AT and NF-κB transcription factor pathways both play principal roles in T cell receptor–induced helper T cell differentiation (12, 13), and accumulating evidence has revealed a pivotal role for NF-κB in the specific regulation of Th1 development (14). Given precedents for other Fox family members as transcriptional repressors (1, 15), Foxj1 might inhibit NF-AT and/or NF-κB activity. Foxj1 inhibited spontaneous NF-κB activity within the M12 lymphoma cell line, as assessed with the use of a NF-κB-luciferase reporter, but failed to significantly affect spontaneous NF-AT or T-box activity in the same cell line (Fig. 4A). These findings were confirmed by the ability of Foxj1 to inhibit inducible NF-κB activity in 293-T epithelial cells (Fig. 4B and fig. S5A). Indeed, Foxj1–/– Th cells demonstrated increased spontaneous NF-κB but equal or lower NF-AT activity (Fig. 4C). In addition, these cells contained elevated levels of the proproliferative NF-κB target genes cyclin D1 (16) and GADD45β (17) as well as inflammatory genes like IL-2 (fig. S5B). Finally, blockade of NF-κB activity inhibited the hyperactivation of Foxj1–/– Th cells (fig. S5C), suggesting that excessive NF-κB activation indeed accounted for our findings.

Fig. 4.

NF-κB suppression by Foxj1. (A) Spontaneous activities of NFAT-, NF-κB, and T-box–luciferase reporters in M12 cells were assessed in the presence of control (pCDNA, open bars) or Foxj1-expressing (pCDNA-Foxj1, black bars) expression vectors. Error bars indicate the standard deviations of three simultaneously performed samples, representative of three experiments. (B) Inducibility of indicated amounts of a NF-κB–luciferase reporter construct was assessed in 293-T cells in the presence (right two bars of each group) or absence (left two bars of each group) of TNF-α, in the presence of control (pCDNA, open and vertically hatched bars) or Foxj1-expressing (pCDNA-Foxj1, black and gray bars) expression vectors. Shown is one experiment, representative of five. (C) NF-κB– and NF-AT–luciferase activity was assessed in primary CD4+ T cells derived from Foxj1+/+ versus Foxj1–/– chimeras. (D) Nuclear translocation of RELA in 293-T cells was assessed after retroviral tranduction with a control (pMX-IRES-GFP, where IRES is internal ribosomal entry site and GFP is green fluorescent protein) or Foxj1-expressing (pMX-Foxj1-IRES-GFP) retrovirus. Cells were stimulated with TNF-α for 45 min and stained for RELA by immunohistochemistry (red). Shown are groups of GFP-positive cells. Percentages indicate the number of cells (per 100 GFP-positive cells) that demonstrated nuclear RELA staining. (E) Induction of IκB family members was examined by Western blotting on 293T cells transfected with pcDNA or pcDNA-Foxj1. Arrow indicates IKB-ϵ (F) Foxj1 regulates IκBβ in vivo. Levels of IκBβ and RELA mRNA were assessed by real-time polymerase chain reaction in Th cells from Foxj1–/– or Foxj1+/+ chimeras. Error bars indicate the standard deviations of three simultaneously performed samples, representative of two experiments.

We examined the ability of Foxj1 to affect NF-κB cytoplasmic-nuclear translocation, which corresponds directly with NF-κB activation (13). Whereas control 293 cells exhibited clear nuclear translocation of the NF-κB component RELA in response to tumor necrosis factor–α (TNF-α), Foxj1-treated cells were consistently impaired in this activity (Fig. 4D). The activation and nuclear translocation of NF-κB is largely regulated by the IκB proteins, leading us to hypothesize that Foxj1 might regulate one or more IκB proteins. Although Foxj1 had no effect on IκBϵ expression and caused only a modest up-regulation of IκBα expression, IκBβ was clearly up-regulated in Foxj1-transduced cells, in contrast to control cells (Fig. 4E). The expression pattern of IκBβ correlated well with that of Foxj1, diminishing particularly in response to anti-CD3 or IL-2 stimulation (fig. S1B). In addition, transduction of primary Th cells by Foxj1 increased IκBβ expression (fig. S1C), and Foxj1 could transactivate the IκBβ promoter (fig. S1D). Finally, Foxj1–/– T cells contained diminished levels of IκBβ mRNA (Fig. 4F). These findings strongly suggest that Foxj1 antagonizes NF-κB activity at least in part by inducing and/or maintaining IκB activity. As a consequence of this, deficiency in Foxj1 leads to spontaneous NF-κB activation and subsequent immune dysregulation.

We suggest that Foxj1 regulates early Th activation, enforcing T cell quiescence by regulating NF-κB activity, in part via IκB (fig. S6). However, recent studies suggest that NF-κB may preferentially play a role in later, postcommitment phases of Th1 development and proliferation (14), suggesting that Foxj1 may also enforce quiescence by modulating the activity of another class(es) of transcriptional regulators and/or may have as-yet undefined direct effects on T cell differentiation genes, such as LKLF (18) or Tob (19). Regardless, the present findings are consistent with prior observations demonstrating dysregulated NF-κB activity in both human (20) and murine (21, 22) lupus and inflammatory phenotypes of animals deficient in IκB activity (23) and furthermore are consistent with the significantly reduced expression of Foxj1 in murine lupus Th cells (fig. S1). Thus, studies to define further the target genes regulated by Foxj1 as well as other Fox transcription factors will undoubtedly shed insight into the regulation of Th differentiation and immune tolerance.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

References and Notes

References and Notes

Stay Connected to Science

Navigate This Article