IκBβ acts to inhibit and activate gene expression during the inflammatory response
The activation of pro-inflammatory gene programs by nuclear factor-κB (NF-κB) is primarily regulated through cytoplasmic sequestration of NF-κB by the inhibitor of κB (IκB) family of proteins 1. IκBβ, a major isoform of IκB, can sequester NF-κB in the cytoplasm 2, although its biological role remains unclear. Although cells lacking IκBβ have been reported 3 4, in vivo studies have been limited and suggested redundancy between IκBα and IκBβ 5. Like IκBα, IκBβ is also inducibly degraded; however, upon stimulation by lipopolysaccharide (LPS), it is degraded slowly and re-synthesized as a hypophosphorylated form that can be detected in the nucleus 6 7 8 9 10 11. The crystal structure of IκBβ bound to p65 suggested this complex might bind DNA 12. In vitro, hypophosphorylated IκBβ can bind DNA with p65 and c-Rel, and the DNA-bound NF-κB:IκBβ complexes are resistant to IκBα, suggesting hypophosphorylated, nuclear IκBβ may prolong the expression of certain genes 9 10 11. Here we report that in vivo IκBβ serves both to inhibit and facilitate the inflammatory response. IκBβ degradation releases NF-κB dimers which upregulate pro-inflammatory target genes such as tumour necrosis factor-α (TNF-α). Surprisingly, absence of IκBβ results in a dramatic reduction of TNF-α in response to LPS even though activation of NF-κB is normal. The inhibition of TNF-α messenger RNA (mRNA) expression correlates with the absence of nuclear, hypophosphorylated-IκBβ bound to p65:c-Rel heterodimers at a specific κB site on the TNF-α promoter. Therefore IκBβ acts through p65:c-Rel dimers to maintain prolonged expression of TNF-α. As a result, IκBβ−/− mice are resistant to LPS-induced septic shock and collagen-induced arthritis. Blocking IκBβ might be a promising new strategy for selectively inhibiting the chronic phase of TNF-α production during the inflammatory response.
To understand the biological function of IκBβ better, we studied mice lacking the IκBβ gene. Homologous recombination was used to delete most of the IκBβ coding sequences (30–308 amino acids) including elements essential for binding to NF-κB ( Supplementary Fig. 2) 6 12 13. Absence of IκBβ was confirmed by immunoblotting of mouse embryonic fibroblasts (MEFs; Supplementary Fig. 2). Although IκBβ is expressed broadly, including in haematopoietic organs ( Supplementary Fig. 3a), the IκBβ knockout mice breed and develop normally without any obvious phenotypic defects.
NF-κB and IκB proteins function in an integrated network. Hence reduced expression of one component may cause compensatory changes in levels of other proteins 14 15. However, expression levels of IκBα, IκBε, p65, RelB, c-Rel, p105 and p100 were unaffected in IκBβ−/− mice ( Supplementary Fig. 3b). Increased NF-κB activity has been observed in other IκB knockouts 16 17 18, and increased basal NF-κB reporter activity was observed in IκBβ−/− MEFs (Fig. 1a). Electrophoretic mobility shift assays (EMSAs) demonstrated increased basal NF-κB activity in IκBβ−/− cells (60%) ( Supplementary Fig. 3c). Conversely, overexpression of IκBβ inhibits NF-κB activation ( Supplementary Fig. 3d). Thus IκBβ inhibits NF-κB and degradation or loss of IκBβ contributes to NF-κB activity. NF-κB reporter assays reveal that absolute NF-κB activity in response to LPS, IL-1β or TNF-α is slightly higher in the IκBβ−/− than wild-type cells (Fig. 1a). However, the kinetics of NF-κB activation by EMSA, and the pattern of IκB degradation by immunoblotting, in cells stimulated with LPS, IL-1β or TNF-α were not demonstrably different in IκBβ−/− cells ( Supplementary Fig. 4). Thus, loss of IκBβ results in a modest elevation in basal NF-κB activity, whereas inducible NF-κB activation is relatively unaffected.
NF-κB regulates the expression of many genes, in particular those involved in inflammation and immune responses 19. To determine whether IκBβ has a role in the inflammatory response, IκBβ−/− and IκBβ+/+ mice were challenged with LPS. Surprisingly, IκBβ−/− mice were significantly resistant to the induction of shock (Fig. 1b). We therefore examined the serum levels of the key acute phase cytokines TNF-α, IL-1β and IL-6 (ref. 20) after LPS injection. In wild-type mice TNF-α production peaked 1 h after LPS injection, whereas IL-6 and IL-1β production peaked around 2 h, in agreement with previous studies 21. Although serum IL-6 and IL-1β were reduced (approximately 25%) in the IκBβ−/− mice, the reduction of TNF-α levels (greater than 70%) was more striking (Fig. 1c). As the peak of serum TNF-α precedes that of IL-1β and IL-6, it is likely that the reduction of IL-1β and IL-6 is secondary. As monocytes and macrophages are major sources of systemic TNF-α, we analysed LPS-induced cytokines in thioglycollate-elicited peritoneal macrophages (TEPMs). Although equivalent macrophage populations were obtained from the mice ( Supplementary Fig. 5a), production of TNF-α, but not IL-6, was drastically reduced in IκBβ−/− TEPMs (Fig. 1d).
To understand how IκBβ affects TNF-α synthesis we examined each step of TNF-α production. Secreted TNF-α was detectable by enzyme-linked immunosorbent assay (ELISA) after 2 h of LPS stimulation and by 4 h was significantly impaired in IκBβ−/− TEPMs (Fig. 2a). IL-6 production was equivalent (Fig. 2a). We examined the level of pro-TNF-α by intracellular fluorescence-activated cell sorting and found there was very little pro-TNF-α detected in the IκBβ−/− TEPMs, even after 8 h of LPS stimulation (Fig. 2b). The average amount of pro-TNF-α produced was two- to threefold higher in wild-type than IκBβ−/− TEPMs (Fig. 2c). Consistent with this difference in protein levels, steady-state TNF-α mRNA was decreased two- to sixfold in the IκBβ−/− TEPMs compared with wild-type cells (Fig. 2d). Although TNF-α mRNA is known to be regulated post-transcriptionally 22 23, there was no difference in TNF-α mRNA stability between wild-type and IκBβ−/− TEPMs ( Supplementary Fig. 5b). Therefore, IκBβ promotes TNF-α transcription.
To understand how IκBβ affects TNF-α transcription, we investigated which NF-κB subunits were associated with IκBβ in macrophages. It is known that IκBβ associates with p65:p50 and c-Rel:p50 complexes 24 through direct binding to p65 and c-Rel but not p50 (ref. 6). However, we found that IκBβ could be immunoprecipitated only with p65 and c-Rel, but not p50 (Fig. 3a). Both immunoprecipitations with anti-p65 and anti-c-Rel antibodies pull down IκBβ, IκBα and p50. Thus, there are p65:p50 and inducible c-Rel:p50 complexes that are associated with IκBα or other IκBs, but not IκBβ. Reciprocal immunoprecipitation of p65 with c-Rel and both p65 and c-Rel with IκBβ suggests a p65:c-Rel heterodimer associated with IκBβ (Fig. 3b). To demonstrate the association of IκBβ with p65:c-Rel, we performed sequential immunoprecipitations by first immunoprecipitating IκBβ and then immunoprecipitating the eluted IκBβ complexes with anti-c-Rel antibody. The presence of p65 in the anti-c-Rel immunoprecipitate confirms the presence of an IκBβ:p65:c-Rel complex (Fig. 3c). The IκBβ:p65:c-Rel complex was found in nuclear extracts, which suggests that this could be a transcriptionally active complex. We had previously reported 10 that IκBβ exists in two phosphorylation states: a hyperphosphorylated state in quiescent, unstimulated cells, and a hypophosphorylated newly synthesized state in LPS-stimulated cells ( Supplementary Fig. 6a). In the co-immunoprecipitation experiments shown here we found that both forms of IκBβ can bind p65 and c-Rel, although the hypophosphorylated form predominates in the IκBβ:p65:cRel complex after LPS stimulation.
There are four κB sites upstream of TNF-α coding region, three of which are crucial for NF-κB-dependent TNF-α expression 25. Therefore, we performed chromatin immunoprecipitation with anti-p65, anti-c-Rel and anti-IκBβ antibodies in RAW264.7 cells and monitored the region encompassing these three κB sites. After LPS stimulation, TNF-α promoter region DNA is enriched by p65, c-Rel and IκBβ antibodies by 56-, 70- and 7-fold respectively (Fig. 3d). In contrast, IκBβ is not recruited to the IL-6 promoter after LPS stimulation whereas p65 and c-Rel are recruited as expected (Fig. 3d). Recruitment of p65, c-Rel and IκBβ to the TNF-α promoter was also confirmed in wild-type bone-marrow-derived macrophages (BMDMs; Fig. 3e). In the IκBβ−/− BMDM, both p65 and c-Rel are recruited normally to the TNF-α promoter. However, when we performed immunoprecipitation with anti-p65, c-Rel and IκBβ are pulled down in wild-type but not IκBβ−/− BMDMs (Fig. 3f). Therefore, p65 and c-Rel fail to form a stable complex in IκBβ−/− cells. Thus, the p65 and c-Rel recruited to the TNF-α promoter in IκBβ−/− cells are not a p65:c-Rel complex. These data suggest that optimal TNF-α transcription requires a ternary complex of IκBβ:p65:c-Rel binding to the TNF-α promoter.
To identify the κB site for p65:c-Rel binding we performed EMSAs using the three κB sites from the TNF-α promoter as probes (κB2, κB2a and κB3; Supplementary Fig. 6b). We identified two distinct gel-shift patterns. κB3 and κB2a show two major bands (only κB3 is shown in Fig. 3g) whereas κB2 shows three major inducible shift bands. The components of the bands were identified by super-shift assay (Fig. 3g, right panel). The top band in the κB2 gel-shift is mostly p65:c-Rel. Interestingly, the κB2 site possesses features predicted to favour p65:c-Rel binding ( Supplementary Fig. 6c). Similar κB binding sites in the CD40 and CXCL1 promoters also demonstrated coordinate recruitment of IκBβ, p65 and c-Rel ( Supplementary Fig. 6d). Furthermore, deletion of the κB2 site from a TNF-α promoter reporter abrogated IκBβ-dependent reporter gene expression ( Supplementary Fig. 7). In IκBβ−/− BMDMs, the p65:c-Rel complex binding to the κB2 in EMSA assays is missing (Fig. 3h), in agreement with the immunoprecipitation result. Therefore optimal TNF-α transcription requires a p65:c-Rel complex, stabilized by hypophosphorylated IκBβ, binding to the κB2 site in the TNF-α promoter.
To identify other genes affected by IκBβ deficiency, we examined gene expression profiles in wild-type and IκBβ−/− BMDMs (Fig. 4a). As expected, TNF-α and IκBβ are among the genes whose expression is affected by IκBβ deficiency whereas IL-6 and IL-1β are not affected (Fig. 4b). Of the genes whose expression is reduced in the IκBβ−/− cells, we identified 14 with expression patterns resembling TNF-α (Fig. 4b). The expression of these genes was also reduced in p65, c-Rel or p65/c-Rel knockout fetal liver macrophages, which suggests that LPS-induced expression of these genes might depend on a mechanism similar to TNF-α (data not shown). The expression of TNF-α, IL-1α, IL-6 and IL-1β in response to LPS was further examined by RNase protection (Fig. 4c) and reverse transcription with quantitative real-time PCR (qRT–PCR) ( Supplementary Fig. 8), which demonstrated that the reduction in persistent expression of TNF-α in IκBβ−/− cells is unique. Reduced IL12b mRNA and protein secretion in the knockout TEPMs was confirmed by qRT–PCR (Fig. 4d) and ELISA (Fig. 4e). Notably, transcription of IL12b, which has a κB site similar to κB2 of TNF-α ( Supplementary Fig. 6c), has previously been shown to require c-Rel and be partly dependent on p65 (ref. 26). Thus, only a select group of NF-κB-dependent genes are diminished similarly to TNF-α upon IκBβ deletion. As TNF-α plays a key role in inflammation, we wanted to test whether IκBβ−/− deletion would affect the course of inflammatory diseases.
Rheumatoid arthritis is a common inflammatory disease with morbidity resulting from ongoing release of pro-inflammatory cytokines, including TNF-α, and consequent destruction of joint tissue 27. Previous studies have shown that NF-κB plays a key role in mouse models of arthritis and that blocking NF-κB has a dramatic effect in preventing disease 28 29. Rheumatoid arthritis can also be effectively treated by anti-TNF-α therapies, although there are significant side-effects 30. The ability to block only persistent TNF-α expression would be therapeutic without blocking beneficial TNF-α responses, including the expression of innate immune response genes. We therefore tested whether the lack of IκBβ altered the course of collagen-induced arthritis, a well-characterized mouse model of rheumatoid arthritis.
To induce collagen-induced arthritis, we immunized DBA/1J mice with bovine type II collagen. IκBβ−/− mice displayed delayed onset, lower incidence and decreased severity of collagen-induced arthritis (Fig. 4f and Supplementary Fig. 9). Inflammation in the wild-type mice extended from the paws and digits to the ankle joints and distally through the limb (data not shown). In contrast, IκBβ−/− mice showed minimal visual signs of paw and joint swelling ( Supplementary Fig. 9c). Serum TNF-α was markedly decreased in IκBβ−/− mice whereas other pro-inflammatory cytokines were not significantly affected (Fig. 4g and Supplementary Fig. 10). Therefore the absence of IκBβ limits the progression and severity of arthritis by reducing the chronic production of TNF-α.
The results presented above demonstrate a dual role for IκBβ: during the early stages of LPS stimulation, NF-κB complexes released by IκBβ degradation contribute to the initial expression of TNF-α ( Supplementary Fig. 1). Then, newly synthesized hypophosphorylated IκBβ facilitates the formation of IκBβ:p65:c-Rel complexes, which selectively bind to the κB2 site in the TNF-α promoter, augmenting transcription. As shown in the gene chip and RNase protection assays, this is a relatively selective function and IκBβ−/− mice are, therefore, otherwise normal. Hence targeting IκBβ might be a promising new strategy to treat chronic inflammatory diseases such as arthritis.
IκBβ-deficient mice were generated by standard homologous recombination in the CJ7 ES cell line using a targeting construct that replaced exons 2 to 5 with a G418-resistance gene. Screened ES cell clones were injected into blastocysts derived from C57BL/6 mice to give rise to IκBβ−/+/IκBβ+/+ chimaeras. Germline transmission of the disrupted allele was obtained and verified by Southern blotting and PCR, and mice were backcrossed at least ten generations onto the C57BL/6 background. Mice were backcrossed at least eight generations onto the DBA background for collagen-induced arthritis experiments. Mice were maintained in pathogen-free animal facilities at Yale Medical School.
Wild-type and IκBβ knockout MEFs were generated from embryos at embryonic day 12.5 after timed breeding of IκBβ+/− animals. TEPMs were obtained from 6- to 8-week-old littermate mice 3 days after intraperitoneal injection with thioglycollate. BMDMs were collected by standard protocols and differentiated with 30% L929 supernatant-conditioned media.
Cell fractionation, western blotting, EMSA, and immunoprecipitations were performed as previously described unless otherwise indicated 6.
LPS-induced shock was tested by intraperitoneal injection of 50 μg g−1 body weight LPS and monitoring for survival. In a separate identical experiment, the mice were bled at 1 h and 2 h after LPS treatment and the concentrations of TNF-α, IL-6 and IL-1β in the serum were measured by ELISA.
Intracellular cytokine analysis
Pro-TNF-α levels were analysed in TEPMs after LPS stimulation and brefeldin-A treatment. TNF-α was detected after cell permeabilization by using standard intracellular cytokine staining and flow cytometry.
RNA expression was quantified by two-step SYBR qRT–PCR, and relative mRNA levels were obtained by normalizing the readout for each specific gene by that of β-actin.
Microarrays for gene expression analyses were performed on BMDMs stimulated with LPS and Affymetrix Mouse Genome 430A 2.0 arrays as per the manufacturer’s protocol.
Full methods accompany this paper on the web.
P.R. characterized the mice and performed most of the experiments, M.S.H. performed the immunoprecipitation experiments and helped in writing the paper, M.L. performed collagen-induced arthritis experiments, D.Z. and A.P.W. performed generation of BMDM cells, A.O. performed some experiments, M.L.S. and D.B. generated the knockout mice, C.L. and A.H. performed the RNAse protection assays, and S.G. conceived the study and wrote the paper.
We thank A. Lin at the Yale W.M. Keck Biostatistics Resource for analysis of microarray data. S.G. was supported by grants from the National Institutes of Health (R37-AI03343).
Downloaded by firstname.lastname@example.org on 25th Aug 2010 from 126.96.36.199
- Posted using BlogPress from my iPhone