Forward genetic analysis of Toll-like receptor responses in wild-derived mice reveals a novel antiinflammatory role for IRAK1BP1

Wild-derived strains of mice are separated from common laboratory strains by significant evolutionary distance, but are still able to breed with inbred strains and can thus be used in forward genetic analyses (17). Accordingly, we screened the wild-derived strain MOLF/Ei against the classical strain C57BL/6J for cytokine responsiveness to various TLR family agonists. Peritoneal macrophages from MOLF/Ei mice exhibited increased levels of cytokine mRNA after stimulation with the TLR2 agonist lipoteichoic acid (LTA), the TLR3 agonist polyI:C, and the TLR4 agonist LPS (Fig. 1, A and B, and Fig. S1). The largest effect was observed for LTA- and LPS-stimulated IL-6 production. To confirm that MOLF/Ei mice were hyperresponsive compared with other classical inbred strains, we examined LTA-induced IL-6 production in CBA and BALB/c mice. These strains responded similarly to C57BL/6J, suggesting that intersubspecific differences (18) rather than more recently introduced mutations underlie the phenotype (Fig. 1 C). The trait was also observed at the level of secreted IL-6 measured by ELISA and was present in two other wild-derived strains, MSM/Ms and Czech/EiII (Fig. 1 D), again consistent with an intersubspecific origin.

To examine the pattern of inheritance of TLR hyperresponsiveness, Northern blot and quantitative RT-PCR analyses were performed using F1(C57BL/6JxMOLF/Ei) peritoneal macrophages. IL-6, but not TNF-α, was elevated in F1 compared with C57BL/6J macrophages after LTA stimulation (Fig. 1 E). We thus chose LTA-stimulated IL-6 production as a marker of TLR hyperresponsiveness to use in our mapping studies. To map genetic loci underlying the observed phenotype, 82 C57BL/6JxF1(C57BL/6JxMOLF/Ei) backcross mice were analyzed by ELISA for IL-6 secretion 6 h after stimulation with LTA. Genome-wide microsatellite marker coverage for the initial 41 mice analyzed by QTL analysis revealed two loci on chromosomes 6 and 9, with suggestive linkage to the trait (Fig. 1 E). These loci were named wild-derived hyperresponse (Why)1 and Why2, respectively. The remaining 41 mice were genotyped at these loci, and significant logarithm of the odds (LOD) scores of 3.69 (Why1) and 5.49 (Why2) were obtained. Heterozygosity at Why1 (D6MIT328) was concordant with the observation in parental strains and conferred a positive effect on IL-6 secretion, whereas a partially compensating negative effect was linked to heterozygosity at Why2 (D9MIT155). Furthermore, a significant epistatic interaction (LRS 48.1) was detected between the two loci. This interaction implies that both genes function in the same biological pathway, a feature commonly found between evolutionarily compensatory alleles (24).

These quantitative trait-mapping results indicate that the major underlying genetic basis of TLR hyperresponsiveness in MOLF/Ei mice is contained within the Why1 interval. Irak-2 is the only gene in this region with a known role in TLR signaling. Despite the presence of several polymorphisms between the MOLF/Ei and C57BL/6J Irak-2 alleles, we have thus far been unable to link any of these differences to the phenotype, though a serial outcross strategy has narrowed the interval even farther on either side of Irak-2 (unpublished data). In addition to revealing the Why1 locus, our mapping strategy also unveiled genetic mechanisms that wild-derived mouse strains have evolved to compensate for this trait and uncovered Why2 as a major negatively acting locus in MOLF/Ei mice. Although Why2 clearly does not act strongly enough to mask the hyperresponsive phenotype observed in MOLF/Ei mice, we reasoned that a regulatory allele was likely preserved at this locus under selective pressure from the overall proinflammatory bias conferred by Why1.

Although the genetically determined interval containing Why2 was relatively large (23 cM; Fig. 1 G), fine mapping of quantitative traits is often extremely difficult due to the effects of other loci with minor contributions to the phenotype. These undetected loci provide variability that prevents the use of single crossover events for defining interval boundaries. Thus, we combined a gene microarray approach (Fig. S2 and Table S1) with our mapping studies to identify potential candidate genes. Among genes differentially expressed in TLR-activated MOLF/Ei and C57BL/6J splenocytes from the Why2 locus (Table I), Irak1bp1 stood out both for the magnitude of its expression difference (∼20-fold) and for its known role in inflammatory signaling. Furthermore, this gene's previously defined ability to physically bind another IRAK family member, IRAK-1 (23), made it an even more attractive candidate given the detected epistasis with the locus containing Irak-2.

To verify the microarray results in the conditions used during the mapping analysis, MOLF/Ei and C57BL/6J peritoneal macrophages were stimulated with LTA, and IRAK1BP1 mRNA was measured. Northern blotting revealed a significant increase in transcript level after LTA stimulation in MOLF/Ei, but not C57BL/6J, macrophages (Fig. 2 A). To determine whether the observed TLR inducibility of IRAK1BP1 mRNA in MOLF/Ei macrophages was linked to the detected locus on chromosome 9, allelic imbalance tests were performed. In F1(C57BL/6JxMOLF/Ei) macrophages, IRAK1BP1 mRNA was present almost exclusively from the MOLF/Ei allele 3 h after LTA stimulation, demonstrating the presence of cis-acting polymorphisms that make the gene transcriptionally responsive in MOLF/Ei mice (Fig. 2 B). Thus, the ability to up-regulate IRAK1BP1 mRNA after TLR stimulation is genetically linked to the inhibitory Why2 locus detected in the analysis of IL-6 hyperresponse.

Why2 was found in our mapping analysis as a compensating mutation in the presence of an overall hyperresponsive phenotype. We therefore reasoned that if IRAK1BP1 was responsible for the linkage to this region, two other wild-derived strains with the same phenotype, Czech/EiII and MSM/Ms (Fig. 1 D), would also be under selective pressure to induce expression of IRAK1BP1 in response to TLR stimulation. Northern blot analysis revealed that these strains up-regulate IRAK1BP1 mRNA after stimulation with LTA, whereas C57BL/6J and another classical inbred strain, A/J, were transcriptionally unresponsive (Fig. 2 C). Macrophages in the quiescent state did not show high levels of IRAK1BP1 mRNA in any of the five strains analyzed in Fig. 2 (A and C), although previous data have shown basal expression levels in other tissues (23). We further extended our transcriptional analysis of IRAK1BP1 to human PBMCs and showed that it was transcriptionally responsive to a variety of TLR agonists in these cells as well (Fig. 2 D). Thus, IRAK1BP1 may also function in the context of human TLR signaling biology, and this trait is well modeled in three wild-derived mouse strains.

We next assessed the functional implication of Irak1bp1 inducibility from the Why2 locus in the context of our genetic mapping data. Because the MOLF/Ei allele at the Why2 locus had an inhibitory effect on IL-6 production, and IRAK1BP1 was up-regulated in MOLF/Ei but not C57BL/6J mice, we predicted that IRAK1BP1 would down-regulate TLR-stimulated IL-6 production in macrophages. Previous studies have implied that IRAK1BP1 functions in the context of TNF-α signaling as a co-activator of RelA-mediated transcription (22, 23). However, these studies have been conducted using reporter plasmids, whereas our phenotype relied on measuring endogenously produced IL-6. Although ectopically introduced reporter assays are often used to assess the general function of inflammatory regulators, the intricate nature of site-specific chromatin remodeling and individual cytokine expression patterns is increasingly gaining appreciation (25, 26). Furthermore, reporter assays are not able to reveal all trans-acting regulatory effects that may be influencing the actual amount of a specific cytokine that is produced.

Given the complexity of cytokine gene regulation that is not incorporated into basic reporter assays, we sought to first characterize the effect of IRAK1BP1 in a more functional model of TLR signaling. The RAW264.7 mouse macrophage cell line is responsive to TLR activation and synthesizes cytokines from their endogenous genomic loci (27). To examine the role of IRAK1BP1 in a fully functional TLR pathway, we generated stably overexpressing RAW264.7 cells through retroviral transduction. IRAK1BP1-transduced cells produced significantly less IL-6 mRNA after LTA stimulation compared with empty vector–transduced cells. Furthermore, secreted IL-6 levels were decreased by ∼75% in IRAK1BP1-transduced cells (Fig. 3 A). Unexpectedly, in analyzing the levels of IRAK1BP1 mRNA in these cells, it was observed that RAW264.7 macrophages constitutively express high levels of transcript (Fig. S3). Therefore, we used a lentiviral-based short-hairpin RNA (shRNA) gene knockdown approach to confirm IRAK1BP1's inhibitory effect on IL-6 production. Expression of IRAK1BP1 was reduced >95% in knockdown cells compared with untransduced RAW264.7 cells and those receiving a control shRNA construct with no homologous sequence in the mouse genome. Consistent with its inhibitory role, IRAK1BP1 knockdown potently increased both LTA-induced IL-6 mRNA and protein expression compared with control and untransduced cell lines (Fig. 3 B).

We next sought to confirm the role of IRAK1BP1 in the observed linkage to the Why2 interval (Fig. 1, F and G) by examining its role in primary macrophages from the strains used in the mapping study. We used the ability of the lentiviral shRNA system to infect primary mammalian cells as a means to inhibit LTA-inducible expression of IRAK1BP1 from the MOLF/Ei allele. Transduction and selection for stable integration of the shRNA construct in bone marrow–derived macrophages (BMDMs) from MOLF/Ei mice decreased the amount of IRAK1BP1 mRNA in these cells after LTA stimulation. Consistent with its effect in RAW264.7 cells and the mapping data, inhibition of IRAK1BP1 in MOLF/Ei BMDMs reduced IL-6 mRNA and protein levels after TLR activation (Fig. 4 A).

To assess the allelic specificity of IRAK1BP1's effect directly on the phenotype used in the mapping of Why2, we introduced the shRNA construct into both MOLF/Ei and C57BL/6J peritoneal macrophages. Although these cells are terminally differentiated and nondividing, thus decreasing efficiency of infection and preventing selection of a pure population of transductants, we were still able to significantly impair the ability of MOLF/Ei and F1(C57BL/6JxMOLF/Ei) peritoneal macrophages to induce IRAK1BP1. Furthermore, this inhibition directly affected the phenotype of LTA-stimulated IL-6 secretion in MOLF/Ei and F1, but not C57BL/6J, macrophages, implying a strain-specific effect on the phenotype (Fig. 4 B). Finally, to directly demonstrate that IRAK1BP1 was contributing to the linkage of Why2 to the trait, we chose a small panel of 21 N2 mice from the mapping experiments to receive IRAK1BP1 shRNA. Introduction of this construct into peritoneal macrophages decreased the LOD score for this group of mice by 0.67 compared with introduction of control shRNA into cells from the same mice (Fig. 4 C). Our inability to completely abolish linkage to Why2 in the presence of IRAK1BP1 shRNA may be due to a low efficiency of transduction or incomplete knockdown, allowing residual contributions to persist. Nonetheless, these data demonstrate that inhibition of LTA-induced IRAK1BP1 expression decreases linkage to the Why2 locus and implies a role for expression of this gene in the context of the overall phenotype.

The unique transcriptional requirements of individual cytokines are key mechanisms by which immune cells direct the nature of an inflammatory response. To characterize the breadth of cytokines affected by IRAK1BP1, we used the ability of RAW264.7 macrophages to activate transcription of endogenous TLR-responsive genes, using all of the subtle modifications required at each regulatory region. As shown in Fig. 5 (A and B), shRNA and overexpression studies demonstrated that IRAK1BP1 inhibits the ability of LTA-stimulated macrophages to transcribe the proinflammatory cytokines IL-12p40 and RANTES (CCL5). However, no effect was observed on the ability to transcribe TNF-α. As TNF-α is known to be an “early response” cytokine (25), we confirmed that IRAK1BP1 was not mediating effects before the time points used in our assay by assessing mRNA levels at 30, 60, and 90 min after stimulation with LTA and still observed no significant difference in mRNA levels between control and IRAK1BP1-overexpressing or knockdown cell lines (Fig. S4). To further expand the panel of proinflammatory cytokines affected by IRAK1BP1, we conducted RNase protection assays using untransduced and stably expressing control shRNA and IRAK1BP1 shRNA RAW264.7 cells after stimulation with LTA. In addition to confirming the presence of an effect for RANTES and the lack of an effect for TNF-α, these results showed that IRAK1BP1 also inhibits the mRNA accumulation of IL-1α, IL-1β, and to a lesser degree MIP-1α, whereas exerting only a minimal effect on IP-10 (Fig. 5 C).

The observed differences in mRNA levels for the cytokines described could be due to transcriptional inhibition or to an increase in the rate of mRNA degradation mediated by IRAK1BP1. To discriminate between these scenarios, a nuclear runoff assay was performed for IL-6 in IRAK1BP1-overexpressing or empty vector–transduced cell lines. At 6 h after stimulation with LTA, IRAK1BP1-overexpressing RAW264.7 cells had noticeably less IL-6–nascent transcript, indicating that IRAK1BP1 exerts its effect at a stage preceding the transcription of inflammatory cytokines rather than through posttranscriptional regulatory mechanisms (Fig. 5 D). Finally, we sought to determine whether IRAK1BP1 acts at the level of proximal signaling events specific for TLR pathways or has a more general inhibitory role in inflammatory gene transcription. To address this question, mouse embryonic fibroblasts (MEFs) from C57BL/6J mice were transduced with IRAK1BP1 or empty vector controls and stimulated with IL-1β, TNF-α, or various TLR agonists. Although varying degrees of inhibition were observed, all agonists consistently exhibited decreased IL-6 production in the presence of overexpressed IRAK1BP1 (Fig. 5 E). Collectively, these studies show that the presence of IRAK1BP1 has an overall transcriptionally inhibitory effect on a subset of proinflammatory cytokines that is exerted in the presence of several receptor signaling pathways.

We have used QTL analysis of LTA-stimulated IL-6 production in wild-derived mice combined with microarray studies to elucidate a novel antiinflammatory function of IRAK1BP1. The ability of IRAK1BP1 to inhibit inflammatory cytokine production was demonstrated using ectopic expression as well as shRNA knockdown of IRAK1BP1 in a mouse macrophage cell line, two types of primary mouse macrophages, and MEFs. Although previous studies have examined this gene in the context of TNF-α–stimulated RelA co-activation, the functional data presented here are the first to show its effects on specific cytokines transcribed from their endogenous loci. IRAK1BP1 could be mediating its effects in a trans fashion by preferentially activating antiinflammatory NF-κB–dependent genes, which subsequently repress inflammatory cytokines. Alternatively, it could function as a “molecular switch,” activating gene transcription from some promoters while inhibiting others. The importance of a balanced and precisely coordinated inflammatory response is increasingly recognized for its contribution to immune-mediated disease. As greater numbers of diagnostic and therapeutic techniques rely on specific biological activities, the regulatory function of IRAK1BP1 in TLR-mediated inflammation described here extends our understanding of how these processes are controlled.

Parental mouse strains were obtained from The Jackson Laboratory. Peritoneal macrophages were elicited from N2 mice and phenotyped as described below, and genome-wide scanning was performed according to standard procedures, using two to three known polymorphic microsatellite markers per chromosome. Both genotypic and phenotypic data were analyzed using QTX software for genetic mapping of quantitative trait loci. 21 of the N2 mice used in the screen were also phenotyped after transduction with the pLKO.1 knockdown construct (described below), and linkage was assessed using the same methods. All experiments with mice were performed in accordance with the regulations and with the approval of Tufts/New England Medical Center Institutional Animal Care and Use Committee. All mice were maintained in a pathogen-free mouse facility at Tufts University School of Medicine.

Primary mouse macrophages were elicited by intraperitoneal injection of 3% thioglycollate in 6–8-wk-old mice. 3 d later, the peritoneal cavity was washed with PBS, cells were resuspended in DMEM, 10% FBS, and plated. Assays were performed 6–16 h after plating or cells were transduced for 24 h as described below. Bone marrow macrophages were isolated from femurs of 6–8-wk-old MOLF/Ei mice as described previously (28). MEFs were isolated from 14-d embryos by trypsinization after dissection of the head and fetal liver. All experiments with MEFs were performed between passages two and five. Human PBMCs were obtained from healthy anonymous donors and were used with approval of the Institutional Review Board at Tufts University School of Medicine. PBMCs were prepared by Ficoll-Hypaque centrifugation on a Lymphoprep gradient. Cells were plated and stimulated with TLR agonists 6 h after plating. All cell lines were obtained from American Type Culture Collection (ATCC) and grown in DMEM, 10% FBS.

IRAK1BP1 cDNA was reverse transcribed and amplified from RAW264.7 RNA and cloned into the BamHI restriction site of the PINCO vector (provided by G. Nolan, Stanford University, Stanford, CA) (29) for transduction of RAW264.7 or pLEX (Open Biosystems) for transduction of MEFs. Orientation of the insert was verified by sequencing. All shRNAs were expressed in the pKLO.1 vector and obtained from Open Biosystems. Plasmid DNA for psPAX2 and pMD2.G vectors was from Addgene.

The PINCO retroviral system was used as described previously (29). Viral supernatants were harvested from PHOENIX-AMPHO HEK293 cells (ATCC) 3 d after transfection (Fugene; Roche) and added to dividing RAW264.7 cells for 24 h. 3 d later, transductants were selected by FACS for GFP expression. shRNA was expressed in the pLKO.1 plasmid (Open Biosystems). Viral supernatants were generated by transfecting 293-T cells with the shRNA construct and packaging vectors pSPAX2 and pMD2.G (Addgene). Stably expressing RAW264.7 knockdown clones were selected by growth in puromycin for 2 wk. BMDMs were infected with the pLKO.1 shRNA construct for 24 h on day 4 of maturation. Cells were washed twice with RPMI, 10% FBS, and were further matured in 3 μg/ml puromycin for 3 d. Peritoneal macrophages were infected 3 h after elicitation for a 24-h period. Cells were washed twice with DMEM, 10% FBS, and recovered for 48 h before assays were performed. MEFs were transduced with IRAK1BP1 using the pLEX retroviral vector packaged using pSPAX2 and pMD2.G as described above, and stable transductants were selected in 5 μg/ml puromycin.

Salmonella minnesota Re595 LPS was purchased from Qbiogene. Poly I:C was obtained from GE Healthcare. All other agonists were obtained from InvivoGen. After stimulation, supernatants were analyzed using IL-6 ELISA (R&D Biosystems), and mRNA was prepared with Trizol (Invitrogen). mRNA was subsequently analyzed by Northern blot hybridization, RNase protection (BD Biosciences), or quantitative RT-PCR using Taqman gene-specific probes (Applied Biosystems).

PCR products were amplified from genomic DNA or cDNA. Fragments were sequenced bidirectionally with internal primers. cDNA was synthesized from purified RNA (Trizol) using either random nanomer or oligo-dT primers and M-MLV reverse transcription (New England Biolabs, Inc.).

Nuclei were isolated in Tween lysis buffer after 6 h of stimulation of cells with LTA. If not used immediately for transcription, nuclei were frozen in liquid nitrogen. 200 μl of nuclei suspension in 1× runoff buffer (25 mM Tris-HCl, pH 8, 12.5 mM MgCl2, 750 mM KCl, and 1.25 mM each of ATP, CTP, and GTP), 2 μl of RNase inihibitors (1 U/ml; New England Biolabs, Inc.), 200 mCi of a-[³²P]UTP, and 3,000 Ci/mmol (PerkinElmer) were added to 220 liters of nuclei suspension and incubated at 30°C for 30 min. Elongated transcripts were then isolated with Trizol, denaturated at 65°C for 5 min, and hybridized at 42°C for 48 h to 5 μg of denatured DNA immobilized on nylon filters in a few milliliters of hybridization solution (200 mM NaHPO4, pH 7.2, 1 mM EDTA, pH 8, SDS 7%, deionized formamide 45%, 250 mg/ml Escherichia coli tRNA). Filters were then washed once or twice at 37°C for 20–30 min in 40 mM NaHPO4-SDS 1% and exposed for autoradiography.

Total cellular RNA from individual samples was isolated with Trizol and hybridized for 16 h with custom-designed probes (BD Biosciences) according to the manufacturer's instructions. Nonhybridized RNA was digested with RNase, and remaining hybrids were run on a 5% PAAG gel together with probe. The gel was dried and exposed for 30 min to x-ray film.

Fig. S1 shows an extended panel of cytokine differences in C57BL/6J and MOLF/Ei. Fig. S2 depicts the microarray data discussed. Fig. S3 shows constitutive expression of IRAK1BP1 mRNA in RAW264.7 cells. Fig. S4 shows no effect of IRAK1BP1 on TNF-α production at early time points after LTA stimulation. Table S1 shows genome-wide results from the microarray experiment.

We thank Dr. P. Brodeur and Dr. B. Huber for comments on the manuscript, and Dr. D. Thorley-Lawson for help with PBMCs. We thank Dr. H. Wortis for continuous support.

This work was supported by grant AI56234 from the National Institutes of Health (to A. Poltorak) and start-up funds from Tufts University (to A. Poltorak).

The authors have no conflicting financial interests.