The Single Nucleotide Polymorphism Mal-D96N Mice Provide New Insights into Functionality of Mal in TLR Immune Responses

Sepsis is a life-threatening condition that is characterized by an excessive inflammatory immune response triggered by an infection. Pathogens frequently isolated from sepsis patients include Gram-negative bacteria Escherichia coli, Klebsiella spp., and Pseudomonas aeruginosa; Gram-positive bacteria, such as Streptococcus pneumonia and Staphylococcus aureus (1); and viruses such as influenza A virus (IAV), dengue viruses, and rhinovirus (2). A growing awareness suggests sepsis involves the breakdown of the careful balance between the inflammatory and anti-inflammatory host responses, which may be seen as a pattern recognition receptor–mediated dysregulation of the immune system following infection (3, 4).

Pattern recognition receptors such as the TLRs are positioned at the forefront of innate immune responses to infection and are critical in recognizing danger signals, such as invading bacteria and initiating an inflammatory response. Ten TLRs have been found in humans (TLR1–TLR10) and twelve in mice (TLR1–9, TLR11–13) (5). TLRs are widely expressed and recognize constituents of bacterial membranes, pathogen-derived nucleic acids, and host danger signals in the form of products of damaged or deregulated cells (5). Critically in regard to diseases such as sepsis (3, 6), TLR2 recognizes Gram-negative and -positive lipopeptides, which are recognized through differential heterodimerization of TLR2 with TLR1 and TLR6 (4). TLR4 is best recognized for its role in sensing LPS from Gram-negative bacterial cell walls, which is often referred to as endotoxin because of its unique role in microbial physiology and its toxic role in the molecular pathogenesis of sepsis (3). The toxicity of LPS is directly related to its potent immunogenicity induced by the host response. Recent studies targeting TLR4 however have highlighted the critical role it plays in IAV pathogenicity induced by hyperinflammation (7, 8).

TLR signaling pathways are diversified and the subsequent inflammatory response directed via six known adaptor proteins: MyD88; MyD88 adaptor-like (Mal) also known as TIRAP; TIR domain-containing adaptor protein-inducing IFN-β (TRIF); TRIF-related adaptor molecule; sterile alpha and armadillo motif–containing protein (SARM); and B cell adaptor for PI3K (BCAP) (9–11). Each adaptor contains a Toll/IL-1R (TIR) domain capable of interacting with the cytoplasmic TIR domain of TLRs (12). Mal is a critical adaptor in these pathways and interacts with both TLR4 or TLR2 and MyD88 to act as a molecular “bridge,” facilitating signal transduction upon TLR4 or TLR2 activation (13–15). Recently, Mal interaction with MyD88 was also reported to be required to signal downstream of endosomal TLRs such as TLR9 in response to viral nucleic acids requisite to its subcellular localization and ability to bind multiple lipid species (16). Formation of the Mal:MyD88 signaling complex triggers downstream signaling cascades, involving MAP kinases, activation of transcription factors such as NF-κB (via p65 phosphorylation), and the initiation of proinflammatory responses (10, 17).

Genetic variations in many TLRs, TIR adaptors, and signaling components have provided key insights into the pathophysiology of infections. Advances in genotyping and bioinformatics reveal that genetic variations can play a vital role in the variability of an individual’s susceptibility to disease (18–20). Single nucleotide polymorphisms (SNPs) represent one such variation. The importance of the association of SNPs in TLRs, their adaptors, and disease susceptibility is in its early stages. Mal is the most polymorphic of all TLR adaptor proteins with eight nonsynonymous SNPs (nsSNPs) in its coding region. nsSNPs have been shown to alter both the characteristics of proteins and affect their ability to function. Several of these rare point mutations within Mal have been linked with an altering susceptibility to infectious diseases. Indeed, recent studies have found that genetic variation in Mal can have either detrimental or protective outcomes. The rs7932766 (A186A) SNP is associated with an increased susceptibility to tuberculosis (21), whereas heterozygous carriage of Mal nsSNP, rs8177374 (S180L), is associated with substantial protection against invasive pneumococcal disease, bacteremia, malaria, and tuberculosis in multiple populations (22). An association with systemic lupus erythematosus (23) but not rheumatoid arthritis (21, 24) has also been reported. Conversely, Mal-S180L has also been associated with more severe outcomes in sepsis patients, although usually in combination with functionally deficient TLR4 polymorphisms (25, 26). Overall, increasing evidence suggests that nsSNPs in Mal may be considered important genetic determinants for disease and provide critical insights in Mal functionality.

Recently two independent studies identified that the rare Mal nsSNP rs8177400 (D96N) leads to impaired TLR4 signaling with reduced NF-κB activation and cytokine production (27, 28). George et al. (27) demonstrate the frequency of Mal-D96N heterozygously in a white population to be 0.97%, with other ethnic groups (nonwhites) displaying a low frequency (<1%) in accordance with HapMap data (27, 29). The D96 residue is located on the surface of Mal both in the crystal (30) and solution (31) structures. The crystal structure of the Mal-D96N mutant shows that the mutation causes no significant structural changes, including no significant change in the distribution of electrostatic charge on the surface of the protein (32). The molecular basis of the effect of the mutation on the function of Mal in TLR signaling, therefore, remains unclear.

Given the increasing evidence indicating mutations within Mal are linked to altered susceptibility to infectious disease, we sought to clarify the role of Mal-D96N in an endogenous setting. As human and mouse Mal are highly similar (∼71% identity), we generated mice on a C57BL/6 background that carry the human Mal-D96N mutation. In this study, we have assessed the role of Mal-D96N at a molecular level in vitro and in vivo during TLR activation and pathogen-induced sepsis. This study clarifies the interaction between endogenous Mal-D96N and MyD88 and for the first time, to our knowledge, identifies that the presence of the N allele of the mutation homozygously or heterozygously confers protection against LPS-induced sepsis and infection with E. coli and IAV in mice.

Genomic DNA was extracted from wild-type (WT) mice and mice homozygous and heterozygous for Mal-D96N. The Mal gene was amplified by PCR using primers 5′-TGTTGAAGAAGCCCAAGAAGA-3′ and 5′-GCCTCTGCCATCCACATAGTA-3′ generating a 578-bp product. PCR products were then purified using a PureLink PCR Purification Kit (Life Technologies) and sequenced using the above primers. Alignment of the sequences to identify bp changes was performed using Vector NTi AlignX (Invitrogen).

Bone marrow cells from WT (WT/WT), homozygous (Mal-D96N/Mal-D96N), and heterozygous (WT/Mal-D96N) mice for the D96N SNP were differentiated for 7 d in DMEM supplemented with 10% (v/v) FCS and 1% (v/v) penicillin/streptomycin solution and M-CSF (20%[v/v] L929 mouse fibroblast supernatant). For experiments, differentiated bone marrow–derived macrophages (BMDMs) were seeded at 5 × 10⁵ cells/ml. Immortalized macrophage cell lines were generated with a published J2 recombinant retrovirus [carrying v-myc and v-raf(mil)] oncogenes (33), as previously described (34). In brief, primary bone marrow cells were incubated in L929 mouse fibroblast-conditioned medium for 3–4 d for the induction of macrophage differentiation. Subsequently, cells were infected with J2 recombinant retrovirus. Cells were maintained in culture for 3–6 mo and slowly “weaned off” L929 supernatant until they were growing in the absence of conditioned medium. Macrophage phenotype was verified by surface expression of the markers CD11b (M1/70; BD Biosciences) and F4/80 (BM8; eBioscience) as well as a range of functional parameters, including responsiveness to TLR ligands. Immortalized lines were generated from WT (C57BL/6), Mal-D96N/Mal-D96N, and WT/Mal-D96N mice.

For cytokine measurements, BMDMs were seeded at 5 × 10⁵ cell/ml in a 96-well plate and stimulated in triplicate. Cells were stimulated as indicated and supernatants removed and analyzed for IL-6, TNF-α (both from BD Biosciences), and RANTES (R&D Systems) using ELISA kits according to the manufacturers’ instructions.

RNA was purified from WT and Mal-D96N BMDMs using the RNeasy column purification kit (QIAGEN); cDNA synthesis was prepared using SuperScript III First-Strand cDNA kit (Invitrogen) and random hexamers (Invitrogen), following the manufacturer’s protocol. Quantitative RT-PCR was performed on an Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems) using SYBR reagents (Applied Biosystems); amplification was directed by the forward and reverse primer pair Mal forward: 5′-TGTTGAAGAAGCCCAAGAAGA-3′, Mal reverse: 5′-GCCTCTGCCATCCACATAGTA-3′. All experiments were carried out with biological and technical triplicates (except where stated) with data normalized relative to the expression of 18S and transformed using the ΔΔ cycle threshold method.

Primary BMDMs from WT and Mal-D96N mice (1.5 × 10⁶ cells/well of a six-well plate) were stimulated with LPS as outlined in the figure legends. Cells were washed twice in ice-cold PBS and lysed in 100 μl of low-stringency lysis buffer (50 mM HEPES [pH 7.5], 100 mM NaCl, 10% glycerol [v/v], 0.5% Nonidet P-40 [v/v], 1 mM EDTA, 1 mM sodium orthovanadate, 0.1 mM PMSF, 1 μg/ml aprotinin, and 1 μg/ml leupeptin). The cell lysates were centrifuged at 13,000 rpm × g for 10 min, supernatants collected, and the protein concentration of each was determined using a bicinchoninic acid protein assay (Pierce Biotechnology) according to manufacturer’s instructions. Samples containing equal protein concentrations were generated using 5× SDS sample loading buffer (125 mM Tris-HCl [pH 6.8], 15% glycerol [v/v], 2% SDS [v/v], and 10 mg/ml bromophenol blue) containing 50 mM DTT. Normalized samples were then analyzed by SDS-PAGE and immunoblotted for the phosphorylation of p38, phosphorylation of JNK, and pERK1/2 following the manufacturer’s instructions. Membranes were stripped with stripping buffer before being reprobed for β-actin as a loading control.

Primary BMDMs from WT/WT, Mal-D96N/Mal-D96N, and WT/Mal-D96N mice were plated at 1 × 10⁶/ml. Cells were stimulated with LPS (100 ng/ml) for 0–120 min. Nuclear extracts were prepared by the replacement of media from cells with ice-cold hypotonic buffer. The subsequent cell pellet was lysed and placed on ice for 10 min. Nuclear proteins were extracted and maintained on ice for 20 min. Following centrifugation, the supernatant was mixed with storage buffer and used immediately or frozen at −80°C. Protein concentrations were determined using a bicinchoninic acid protein assay (Pierce Biotechnology). In the EMSA, 4 μg of nuclear extracts were incubated for 30 min with 10,000 cpm of a 22-bp DNA fragment oligonucleotide containing the NF-κB consensus sequence, which had previously been labeled with [γ-³²P]ATP and 10× binding buffer. Incubated mixtures were subjected to electrophoresis on native 6% (w/v) polyacrylamide gels, which were subsequently dried and autoradiographed.

Immortalized BMDMs from WT/WT, Mal-D96N/Mal-D96N, and WT/Mal-D96N mice were seeded at 1 × 10⁶ cells/ml in 10-cm dishes. Cells were left untreated or stimulated with LPS (100 ng/ml) for 10 and 30 min. Cells were harvested in low-stringency lysis buffer, and 50 μl of each sample was removed as whole-cell lysate. Coimmunoprecipitations were performed with goat MyD88 Ab (R&D Systems) or IgG control for 2 h at 4°C while rotating. Following this, the lysate and beads were centrifuged at 2200 × g for 3 min at 4°C; the supernatant was removed, and the beads were washed three times in 1 ml of wash buffer. Immune complexes were eluted by the addition of 50 μl of SDS/Laemmli buffer and boiling the samples. Samples were analyzed by SDS-PAGE and Western blotting.

Duolink Proximity Ligation Assays (PLA) were performed according to manufacturer’s protocols (Sigma-Aldrich). Briefly, cells were seeded in an eight-well ibidi chamber slide and left to rest overnight. Cells were then left untreated or stimulated with LPS (100 ng/ml) for 10, 20, and 30 min followed by fixation with 4% paraformaldehyde (pH 7.4) in PBS for 15 min at room temperature (RT). Cells were washed in PBS, and permeabilization was performed by the addition of 0.2% Triton X-100 diluted in PBS for 10 min at RT. Cells were washed with PBS and then blocked in Duolink blocking solution for 40 min at 37°C. Following this, primary Abs, anti-Mal (rabbit; Cell Signaling Technology), and anti-MyD88 (goat; R&D Systems) were added and incubated overnight at 4°C in Duolink Ab Diluent. Cells were washed two times for 5 min in wash buffer A, and then secondary Abs conjugated with oligonucleotides (PLA probes, anti-rabbit [Mal], and anti-goat [MyD88], Duolink; Sigma-Aldrich) were added and incubated for 1 h at 37°C. For the remainder of the assay, Duolink In Situ Detection Reagents Red were used. Cells were washed two times for 5 min in wash buffer A. After washing, the ligation solution was added together with ligase for 30 min at 37°C. Cells were washed two times for 5 min in wash buffer A, and amplification solution was added together with polymerase for 90 min at 37°C. Slides were washed two times for 10 min in 1× wash buffer B at RT followed by 0.01× wash buffer B for 1 min. Finally, slides were incubated with Hoechst 33342 nuclear stain for 5 min at RT followed by PBS washes three times for 5 min before mounting with Dako Fluorescent Mounting Medium. Cells were imaged via the DeltaVision confocal microscope (GE Healthcare). Representative images show the three-dimensional deconvolution of multiple z-stacks. Finally, images were analyzed using the ImageJ software. Interactions were quantified by counting the number of dots per cell. In the figures, each bar (mean ± SEM) represents the mean obtained from the quantification of signals observed in five different fields with ∼150 cells per field chosen randomly. The experiment in the figure is representative of three independent experiments.

For NF-κB nuclear localization microscopy images, BMDMs were seeded twice at 104 cells per well in 10 chamber slides (Life Technologies). Cells were treated with 100 ng/ml of LPS for the indicated times and cotreated with Hoechst 33342 (Life Technologies, Melbourne, VIC, Australia) for 10 min prior to harvesting. After washing, cells were fixed using 10% formalin for 10 min and permeabilized with 0.1% saponin and 2.5% FCS in PBS for 30 min. Cells were stained with anti-p65 mAb (F6: Santa Cruz Biotechnology) for 120 min and counterstained with Alexa Fluor 488 (Life Technologies) for 120 min. Cells were washed in 0.1% saponin and 2.5% FCS in PBS between each step. Cells were imaged on an Olympus FV1200 inverted microscope with a 40× oil objective.

Using Fiji image processing software, background fluorescence was removed from the DAPI and Alexa Fluor 488 channels by applying a rolling ball with a 50-pixel radius and a Gaussian blur filter with a 1-pixel diameter. A mask of the nuclear region was created by applying the Li threshold algorithm to the DAPI channel. The resulting mask was transferred to the Alexa Fluor 488 channel, and the measure function in the region of interest manager was used to determine the mean intensity of p65 in the nucleus.

Six- to eight-week-old female and male C57BL/6 mice were maintained in the Specific Pathogen-Free Physical Containment Level 2 Animal Research Facility at the Monash Medical Centre. All experimental procedures were approved by the Monash Medical Centre Animal Ethics Committee, and experimental procedures were carried out in accordance with approved guidelines. Age-matched male and female WT, Mal-deficient (Mal^−/−), Mal-D96N/Mal-D96N, and WT/Mal-D96N mice were injected i.p. with 10 mg/kg of LPS (E. coli O111:B4) and mortality assessed over 24 h. To detect cytokines, peritoneal fluid and sera were collected 12 h following i.p. injection of LPS and stored at −80°C. IL-1β was quantified by ELISA according to manufacturer’s instructions (BD Biosciences). Levels of IL-6, IL-10, IL-12p70, and TNF-α proteins were determined by cytokine bead array, mouse inflammation, and flex kits (BD Biosciences).

To assess the role of Mal-D96N in host defense against live bacteria, we used E. coli strain American Type Culture Collection 25922. This strain has been previously used to examine innate immune responses in the lungs of mice (35, 36). A single colony of bacteria was grown for 6 h to midlogarithmic phase at 37°C in 8 ml of lysogeny broth while shaking at 200 rpm × g. The bacteria were then resuspended in PBS at a concentration of 10⁷ CFU/50 μl/mouse. For infection studies, groups of six to eight C57BL/6, Mal-D96N/Mal-D96N, and WT/Mal-D96N mice were anesthetized and infected with 10⁷ CFU intranasally in 50 μl PBS. Mice were weighed and assessed for visual signs of clinical disease, including inactivity, ruffled fur, labored breathing, and huddling behavior. Animals that displayed severe clinical signs of disease were euthanized.

Six- to eight-week-old male C57BL/6 mice were maintained in the Specific Pathogen Free Physical Containment Level 2 Animal Research Facility at the Monash Medical Centre. All experimental procedures were approved by the Monash Medical Centre Animal Ethics Committee, and experimental procedures were carried out in accordance with approved guidelines. IAV strain HKx31 (H3N2) was used in this study which is a high-yielding reassortant of the A/PR/8/34 (H1N1) strain that carries the surface glycoproteins of A/Aichi/2/1968 (H3N2). Viruses were grown in 10-d-old embryonated chicken eggs by standard procedures and titrated on Madin–Darby canine kidney cells as described previously (37).

For virus infection studies, groups of C57BL/6 WT and Mal D96N mice were anesthetized and infected with 10⁵ PFU of HKx31 (H3N2) intranasally in 50 μ l PBS as previously described (38). Mice were weighed daily and assessed for visual signs of clinical disease, including inactivity, ruffled fur, labored breathing, and huddling behavior. Animals that displayed severe clinical signs of disease were euthanized. Titers of infectious virus in lung tissue homogenates were determined by standard plaque assay on Madin–Darby canine kidney cells.

In separate experiments, bronchoalveolar lavage (BAL) fluid was obtained from euthanized mice following 6 h of infection with E. coli via flushing the lungs three times with 1 ml of PBS. For flow cytometric analysis, BAL cells were treated with RBC lysis buffer (Sigma-Aldrich), and cell numbers and viability were assessed via trypan blue exclusion using a hemocytometer. BAL cells were incubated with Fc block (BD Biosciences), followed by staining with mAbs to Ly6C, Ly6G, CD11c, and I-A^b (BD Biosciences). Neutrophils (Ly6G⁺), airway macrophages (CD11c⁺ I-A^{d low}), dendritic cells (CD11c⁺ I-A^{d high}), and inflammatory macrophages (Ly6G⁻ Ly6C⁺) were quantified by flow cytometry, as described previously (35). Live cells (propidium iodide negative) were analyzed using a BD FACSCanto II flow cytometer (BD Biosciences), and total cell counts were calculated from viable cell counts performed via trypan blue exclusion. Similarly, to detect cytokines, BAL and sera were collected 6 h following infection and stored at −80°C. IL-1β was quantified by ELISA according to manufacturer’s instructions (BD Biosciences). Levels of IL-6, MCP-1, and TNF-α proteins were determined by cytokine bead array, mouse inflammation, and flex kits (BD Biosciences).

When comparing three or more sets of values, a one-way ANOVA was used with Tukey post hoc analysis. A Student t test was used when comparing two values (two-tailed, two-sample equal variance). Survival proportions were compared using the Mantel–Cox log-rank test. A p value < 0.05 was considered statistically significant.

To study Mal-D96N in an endogenous setting, we generated a C57BL/6 mouse strain harboring the human D96N mutation. Confirmation of D96N SNP variant expression in mice homozygous and heterozygous for the mutation was performed by DNA sequencing (Fig. 1A). Mice were born in the expected Mendelian ratios and developed no overt phenotype. Body weight and organ weights of mice homozygous for the mutation (D96N/D96N) were comparable to those of WT mice (Supplemental Fig. 1A, 1B). In addition, analysis of the levels of Mal by real-time PCR indicates commensurate levels of Mal expression as compared with WT mice (Supplemental Fig. 1C). We next sought to investigate the effect of the Mal-D96N on TLR signaling in vitro.

Given that previous studies demonstrated loss-of-function effects of Mal-D96N in response to TLR activation in overexpression systems (27, 28), we initially examined the effect of Mal-D96N in TLR-mediated cytokine production in primary BMDMs from mice homozygous and heterozygous for Mal-D96N compared with WT mice in vitro. Mal has a critical role in mediating TLR4-induced inflammation (13, 14, 39, 40), whereby its presence is required to facilitate the recruitment of MyD88 to the active TIR domains of TLR4 to allow signal transduction (41). Consistent with this, we found significantly suppressed levels of IL-6 in Mal-D96N BMDMs in response to TLR4 ligand LPS (Fig. 1B). Conversely, no significant changes in TNF-α or RANTES were observed between WT and D96N/D96N or WT/D96N macrophages stimulated with LPS (Fig. 1C, 1D).

Although the role of Mal in TLR4 signaling is well established, its role in TLR2 signaling is not as clear. Initial studies suggested Mal was clearly required for TLR2-mediated inflammation; however, further studies suggested Mal was primarily required for IL-6 and NF-κB induction upon TLR1/2 activation only at low concentrations of the ligand (42, 43), suggesting a role for Mal in “sensitizing” signaling by TLR2 at low levels of pathogen challenge. At high concentrations of ligand or high multiplicity of infection with Salmonella enterica, Mal deficiency was redundant (43, 44). In agreement with this, we observed suppressed levels of IL-6 in macrophages from mice both homozygous and heterozygous for Mal-D96N when stimulated with low concentrations of ligand (Pam₃Cys 5 and 10 ng/ml, respectively) (Fig. 1E). Unexpectedly, however, at higher concentrations (Pam₃Cys 100 ng/ml), we consistently observed a significant increase in IL-6 in cells both homozygous and heterozygous for the mutation (Fig. 1E). A similar trend was observed for TNF-α in response to high and low doses of the ligand (Fig. 1F). Interestingly, we also observed significantly enhanced levels of RANTES in response to high concentrations of Pam₃Cys and only in BMDMs homozygous for Mal-D96N (Fig. 1G).

Given the emerging role of Mal in myddosome formation following activation of endosomal TLRs (16), we also examined the effect of Mal-D96N on TLR7 and TLR9-induced cytokine secretion. Interestingly, BMDMs both heterozygous and homozygous for Mal-D96N displayed significantly enhanced levels of IL-6, TNF, and RANTES secretion compared with WT cells in response to both TLR7 (Fig. 1H–J) and TLR9 activation (Fig. 1K–M). It is noteworthy that the alterations in cytokines were more pronounced in BMDMs homozygous (Mal-D96N/Mal-D96N) than those heterozygous (WT/Mal-D96N) for the mutation across stimulation of TLRs (TLR4, TLR2, TLR7, and TLR9) upon their stimulation, suggestive of a “gene-dosage” effect of the Mal-D96N SNP.

As Mal is required to bridge MyD88 to the TLR4 receptor complex, we next wished to determine if reduced TLR4 signaling in Mal-D96N macrophages may be because of dysregulated interaction between Mal and MyD88. We therefore generated immortalized WT, WT/Mal-D96N, and Mal-D96N BMDMs to conduct interaction studies. As can be seen in Fig. 2A, for the first time to our knowledge, we could observe endogenous immunoprecipitation of Mal with MyD88 within 10–30 min post-LPS challenge. However, Mal-D96N macrophages demonstrated reduced or delayed association between Mal and MyD88 at 10 and 30 min post-LPS, respectively, whereas WT/Mal-D96N macrophages demonstrate similar LPS-induced interaction between Mal and MyD88 as WT cells.

To support these findings, we further analyzed the physical proximity of WT Mal and Mal-D96N with MyD88 in situ by using a quantitative PLA. The analysis revealed a low-level PLA signal between MyD88 and Mal in resting cells (Fig. 2B, 2C). As expected, the number of close proximity sites was markedly enhanced in WT cells, reaching significance at both 20 and 30 min of stimulation with LPS (Fig. 2C, Supplemental Fig. 2) (**p = 0.0037 and ****p < 0.001, respectively). However, no significant change in the number of PLA signals per cell was observed in Mal-D96N cells following treatment with LPS (Fig. 2C). These results support our previous results obtained by endogenous immunoprecipitation that the ability of Mal-D96N to interact with MyD88 is impeded following the engagement of TLR4 in comparison with WT Mal.

As Mal-D96N macrophages display reduced recruitment of MyD88 following TLR4 activation, we next wished to determine if this dysregulated interaction affects LPS-induced nuclear translocation of NF-κB. As can be observed in Fig. 3, WT BMDMs display substantial LPS-induced nuclear localization of NF-κB in a time-dependent manner. The observation of delayed NF-κB nuclear localization was supported by densitometric analysis of multiple NF-κB EMSAs (Fig. 3B). This observation was confirmed by examining p65 NF-κB nuclear localization by fluorescent microscopy. As can be seen in Fig. 3C, we observed significantly reduced p65 nuclear localization in Mal D96N macrophages as compared with WT macrophages 60 min after LPS challenge. This delayed NF-κB nuclear translocation in Mal D96N macrophages was observed in a time-dependent manner (Fig. 3D, Supplemental Fig. 3).

Taken together, these results suggest that Mal-D96N macrophages display reduced or delayed NF-κB nuclear translocation kinetics in the early time course following LPS stimulation, demonstrating commensurate localization by 120 min posttreatment, consistent with the phenotype of delayed recruitment of MyD88 described above.

Our preceding results suggest that Mal-D96N affects the kinetics of Mal interaction with MyD88 and subsequently dysregulates MyD88-dependent signaling. We therefore wished to further investigate the signaling effects of the Mal-D96N mutation on MAP kinase signaling pathways. Primary WT and Mal-D96N BMDMs were stimulated with LPS over indicated times and examined for their ability to induce phosphorylation of MAP kinases p38, JNK, and ERK1/2. As shown in Fig. 4A, within 10 min of LPS stimulation, all three MAP kinases tested were activated in WT macrophages, with sustained phosphorylation for 30 min post-LPS. In contrast, Mal-D96N macrophages displayed reduced levels of LPS-induced activation for all MAP kinases (Fig. 4A).

We have previously reported that Mal is required for TLR-induced transactivation of NF-κB (45) and that Mal is crucial for Ser536 phosphorylation of the p65 subunit of NF-κB (46). Because transactivation of NF-κB is a critical process in the transcriptional regulation of the inflammatory response (47), we next investigated whether Mal-D96N affected this response. As can be observed in Fig. 4B, WT macrophages display p65 Ser536 phosphorylation within 5 min of LPS challenge, and the phosphorylation is significantly increased by 10 min. Consistent with Mal-D96N dysregulated signaling, Mal-D96N macrophages display delayed Ser536 phosphorylation from 10 min with a reduced intensity over the time course investigated.

Taken together, these results suggest that Mal-D96N reduces the transcriptional inflammatory response following TLR4 stimulation via a disrupted interaction with MyD88, resulting in reduced MAP kinase activity and NF-κB translocation and transactivation.

The excessive production of proinflammatory cytokines as a result of excessive TLR activated inflammatory responses is detrimental to host survival during pathogenic infections. Sepsis is a complex clinical disorder in which severe inflammatory responses, including “cytokine storm,” may culminate in multiorgan failure and death (48). Initial studies by us and Weber and coworkers (27) suggested the Mal-D96N SNP may confer either a detrimental or protective sepsis phenotype, respectively (28). We therefore wished to establish the physiological relevance of the effects of Mal-D96N in an in vivo model of LPS-induced sepsis in mice. As Mal^−/− mice have previously been demonstrated to be protected (39, 40), these were included as controls. Consistent with these reports, Mal^−/− mice were protected from LPS-induced lethality (Fig. 5A). Critically, mice homozygous for Mal-D96N were protected from LPS-induced endotoxemia compared with WT mice (WT versus Mal-D96N/Mal-D96N, ***p < 0.0001) (Fig. 5A), whereas mice heterozygous for Mal-D96N displayed an intermediary protective phenotype (WT versus WT/Mal-D96N, ***p < 0.0002) (Fig. 5A). These results were concomitant with reduced levels of IL-6, TNF-α, IL-1β, MCP-1, and IFN-γ in the serum of mice homozygous for Mal-D96N 12 h following LPS challenge compared with WT mice (Fig. 5B–F). Furthermore, no significant differences were detected in IL-10 levels in the serum between genotypes (Fig. 5G). In addition, levels of proinflammatory IL-1β in the peritoneum of Mal-D96N mice were significantly reduced compared with WT mice (Fig. 5H).

These results confirm a protective phenotype for Mal-D96N in vivo as mice homozygous for the mutation were completely protected from LPS-induced sepsis, whereas the intermediate protection observed in mice heterozygous for Mal-D96N mirrored our initial in vitro observations indicative of a gene-dosage effect of the mutation.

We next sought to extend our findings on the protective effect of Mal-D96N in vivo by looking at an acute bacterial model of infection, which potentially involves multiple TLR ligands. Innate immune responses are regularly demonstrated to be a double-edged sword. Excessive responses can lead to host injury, and an insufficient response can result in the overwhelming burden of the host by pathogen. Previous reports demonstrate that Mal signaling is critical for LPS-induced pathological immune responses to E. coli in the lung; this observation was validated with the use of a cell-permeable Mal blocking peptide in vivo (35), which exhibited attenuated immune responses, increased bacterial load, and early mortality.

To examine the pathophysiological outcomes of Mal-D96N genetic variation during disease, we challenged WT mice and mice homozygous or heterozygous for Mal-D96N with E.coli by intranasal inoculation. As expected, a high mortality rate was observed in WT mice within 20 h following bacterial challenge, with ∼75% (Fig. 6A) of mice requiring euthanasia. This was mirrored by heterozygous WT/Mal-D96N mice. Critically, homozygous Mal-D96N mice exhibited a significantly enhanced survival compared with WT and heterozygous mice with a mortality rate of 25% (WT versus Mal-D96N, *p < 0.0118) (Fig. 6A). The enhanced survival observed in Mal-D96N mice was associated with reduced inflammatory cytokine production in the BAL fluid of these mice compared with WT mice (Fig. 6B–E). Interestingly, survival was not associated with a reduction in the recruitment of inflammatory cells to the lung, including neutrophils, macrophages, inflammatory macrophages, and dendritic cells (Fig. 6F–I). Furthermore, we found no difference in lung bacterial load (Fig. 6J) and could not detect systemic bacterial dissemination, suggesting the putative effect of Mal-D96N lies in its ability to ablate inflammatory cytokine production or the cytokine storm associated with mortality during this infection rather than effecting cellular recruitment. This is supported by earlier studies that demonstrate the necessity of Mal for neutrophil recruitment into the airspace and lung parenchyma and survival in this model of infection (35).

To further examine the protective effects of the Mal D96N SNP from a pulmonary infection, we examined the response of Mal D96N mice to severe IAV infection. Pathogenic IAV infections are characterized by excessive pulmonary inflammation and cellular influx. Recent studies have highlighted the critical role of TLR4 signaling in IAV pathogenesis and that therapies targeting TLR4 or Mal signaling are protective in mouse models of influenza (7, 8, 49). We therefore challenged WT and Mal D96N mice with a lethal dose of HKx31 (H3N2) strain of IAV and observed that Mal D96N displayed reduced weight loss (Fig. 7A) and a significant delay of up to 2 d in HKx31-associated lethality (p < 0.001). Significantly, serum levels of IL-1β (Fig. 7C) and pulmonary concentrations of IL-1β, TNF-α, and IL-6 (Fig. 7D) were significantly reduced in Mal D96N mice as compared with WT concentrations on day 3 post-IAV infection. Importantly, there was no difference in viral loads (Fig. 7E) or lung cellular infiltrates (data not shown) between both genotypes, suggesting that the reduced pathogenesis was because of dysregulated Mal signaling rather than differential cellular recruitment or viral load.

Taken together, these results suggest that the Mal D96N SNP results in reduced viral sepsis and that dysregulated Mal signaling is protective against hyperinflammatory disease via reduced inflammatory cytokine burden.

In this study, we show, using mice expressing human Mal-D96N, that this common Mal nsSNP significantly attenuates responses to TLR4 activation with subsequent physiological protective phenotype in models of infection that result in damaging excessive inflammation. We have further found that the D96N mutation reduces the capacity of Mal to interact with MyD88, dysregulating canonical MyD88-dependent signaling and subsequent NF-κB activation. Although Mal is the most polymorphic of the TLR adaptor proteins, only two of its eight nonsynonymous mutations have been linked to functional consequences, S180L and D96N. Given increasing evidence for the association of SNPs in TLRs and their adaptors in disease, it is important we assess the mechanisms by which they alter signaling to help inform targeted treatments and therapeutics.

Extensive studies have previously demonstrated that Mal-S180L is associated with protection against invasive pneumococcal disease, bacteremia, and tuberculosis in multiple populations. Yet, the full extent of the mechanisms underlying the role of Mal-S180L in infection and disease have not been elucidated. Recent progress has been made regarding its role in Mycobacterium tuberculosis infection, in which the polymorphism was shown to attenuate IFN-γ signaling, impairing the response to the bacterium. In the context of Mal-D96N, only two studies have looked at its role in TLR signaling, our own previous work and that of Weber and colleagues (27, 28). Both studies demonstrated impaired TLR4 signaling with reduced NF-κB activation and cytokine production in the presence of exogenous Mal-D96N but differ in agreement on the precise interaction or lack thereof between Mal-D96N and MyD88. Importantly, a limitation of both studies was the requirement of ectopic expression of either Mal or MyD88 to facilitate and observe interaction, which may have led to nonspecific interactions. Indeed, both studies reported conflicting findings regarding the effect of the D96N mutation on the Mal:MyD88 interaction. In this study, we have generated a unique mouse expressing human Mal-D96N, demonstrating a protective phenotype presumably because of reduced inflammation. The presence of Mal-D96N in macrophages from these mice significantly inhibits IL-6 production via reduced phosphorylation and the initiation of MAP kinase pathways and NF-κB transactivation in response to LPS. Our studies suggest this dysregulated signaling is because of disruption of the recruitment of MyD88 by Mal, leading to altered signaling and NF-κB translocation kinetics. The observation that Mal-D96N did not affect TNF-α expression may involve other pathways, such as TRIF-dependent signaling, which has previously been suggested to occur via IFN regulatory factor-3 signaling and signaling kinetics postactivation (50). Indeed, we have previously observed this disparity in Mal-mediated induction of IL-6 versus TNF-α (45).

Interestingly, we noted that the Mal D96N SNP appears to have disparate effects upon cytokine responses to different TLR agonist-treated BMDMs, displaying both inhibitory and potentiating effects. Previous studies comparing WT and Mal^–/– macrophages proposed that Mal was dispensable for TLR2 induction of IL-6 at high agonist concentrations but increases the efficacy of TLR2 signaling at lower ligand concentrations (43). This would suggest that Mal acts as a sensitizing factor in responding to low levels of pathogen infections. Intriguingly, recent studies by Bryant and colleagues (51) proposed that the strength of TLR4 signaling depends on the number and size of myddosomes formed and the kinetics of this oligomerization. Given we noted a disruption in the time course and abundance of Mal:MyD88 interactions (Fig. 2), it may be that the efficacy and kinetics of Mal:MyD88 interaction, disrupted by the D96N SNP, effect the robustness of not only TLR4 but also other TLR pathways that use Mal. Conversely, the delay in signaling caused by the Mal D96N SNP and subsequent NF-κB translocation may result in extended or enhanced signaling as observed at higher concentrations with TLR2 agonists and endosomal signaling, which may contain higher concentrations of TLR ligands as proposed by Bonham et al. (16). Indeed, the studies by O’Neill and colleagues (43) identifying the sensitizing role for IL-6 expression in regard to TLR2 signaling may have identified only the reduction in responses to low concentrations of ligand because of the absence of Mal in deficient macrophages but not the enhancement of IL-6 responses at higher concentrations. Intriguingly, the authors noted that endosomal TLR3 signaling was enhanced, which may again suggest a divergent role for Mal in endosomal-mediated TLR signaling. The results using our Mal D96N macrophages may provide further mechanistic insights into the role of Mal in coordinating the efficacy of MyD88 recruitment and myddosome formation that regulates the strength of TLR signaling and cytokine responses. Immune responses during pathogen infection may therefore constitute balances between responses to low and/or high pathogen agonist concentrations detected either systemically or locally. Given we observed enhanced cytokine responses to high concentrations of TLR7- and TLR9-stimulated Mal D96N macrophages, future studies may investigate the subcellular localization of Mal and its ability to bind multiple lipid species (16) upon the number and kinetics of Mal-mediated myddosome formations in response to plasma membrane– and endosomal-induced TLR signaling.

It was also important to note that the reduction observed in IL-6 was more pronounced in BMDMs homozygous for Mal-D96N than those heterozygous for the mutation. Similarly, mice homozygous for Mal-D96N displayed a better survival advantage compared with heterozygous mice. This points toward a possible dosage effect of the variant. Certainly, earlier work by George et al. (27) suggests that, depending on the relative abundance of Mal-D96N compared with WT Mal, Mal-D96N may be able to modulate Mal WT–dependent signaling. A precedent exists in the case of Mal-S180L, in which a dosage effect was observed in monocyte-derived macrophages from human volunteers expressing the variant in response to both TLR2 and TLR4 activation (44). Indeed, the study by Ni Cheallaigh et al. (44) established that Mal plays a crucial role in mediating IFN-γ signaling and is protective in a mouse model of tuberculosis, and as such, it will be interesting to investigate the potential role of Mal-D96N in IFN-γ signaling and tuberculosis.

Recently, we described the formation of a filamentous structure by Mal and its signaling partners in the TLR4 pathway (41). This sequential and cooperative mechanism of signaling proffers a maximal response to low sensitivity recognition of TLR4 responses (41). The D96 residue is not found at any of the interfaces of the Mal protofilament and, therefore, would not be predicted to affect assembly. This is consistent with the observation that the D96A mutation does not affect Mal filament formation (41), previous reports that the TIR domains with D96N and D96A mutations can bind MyD88 based on GST pull-down assays (32), and that Mal-D96N can interact with MyD88 based on coimmunoprecipitation assays and confocal immunofluorescence microscopy (27). However, as conformational changes in the loop regions are required for Mal to assemble in the filamentous signalosome, the D96N mutation could influence these transitions, although the position of residue D96 does not change compared with the crystal and solution structure of Mal TIR domain. Interestingly, the D96 residue is located close to C91, which has been found to be glutathionylated in the cell, and this modification was found to be required for signaling, possibly by affecting the conformation of the nearby BB-loop (31). The D96N mutation could therefore affect the conformation of the BB-loop either directly in the Mal structure or indirectly by influencing the efficiency of C91 glutathionylation.

Given that we have demonstrated delayed kinetics of MAP kinase and NF-κB nuclear translocation because of reduced MyD88:Mal interaction, this disruption to signaling does not appear to be because of structurally altering protofilament formation of the Mal:MyD88 interface. Interestingly, Latty et al. (51) also reported that single-cell analysis of LPS-challenged macrophages demonstrated that NF-κB–dependent gene expression significantly correlated with the speed of NF-κB nuclear translocation; slow NF-κB translocation correlated with reduced gene production. These findings would appear to support our in vitro and in vivo findings that disrupted TLR4 signaling does not ablate immune responses but rather results in reduced immune responses potentially protective against septic shock. Given that we have observed differences in TLR2- and TLR4-mediated responses in Mal-D96N cells, this may suggest there are differences in how TLR2 and TLR4 employ Mal to recruit MyD88 and initiate downstream signaling.

The physiological relevance of reduced signaling kinetics and signal transduction is demonstrated by our in vivo models of LPS-, bacterial-, and viral-induced sepsis and endotoxemia. Importantly, we found that Mal-D96N mice display reduced inflammation and are protected from both acute endotoxemia- and hyperinflammatory-induced lethality. The production of inflammatory cytokines was significantly reduced in the serum of Mal-D96N mice, compared with WT mice, following challenge with LPS. Compellingly, Mal-D96N had a survival advantage in all models of infection. Our findings demonstrate that although Mal-D96N dysregulates Mal interaction with MyD88, signaling is not ablated or nonfunctional but rather diminished, conveying protection from LPS-, E. coli–, and IAV-induced inflammation and lethality. This suggests Mal is a critical checkpoint in TLR4-induced inflammation, moderating a “Goldilocks” model of TLR signal transduction proposed by Bernard and O’Neill (15), in which Mal may mediate a “just right” strength of signaling that is beneficial to the host. It has been proposed that the human Mal-S180L heterozygotes are protected from bacteremia, tuberculosis, malaria, and pneumococcal disease as the protective heterozygous state is likely to be associated with reduced signaling and NF-κB activation (22). Indeed, our observation that heterozygous Mal-D96N mice and macrophages display a gene-dosage partial phenotype may suggest that heterozygous carriers of Mal-D96N may be protected from pathogenic challenges that induce excessive inflammation, allowing enough inflammation to be protective without being excessive enough to cause detrimental pathologic outcomes.

This study offers insights into to how the D96N mutation affects immune responses and suggests that the targeting of the D96 residue in Mal may provide a novel therapeutic strategy for the treatment of infections and, perhaps more specifically, for infections mediated by Gram-negative bacteria and IAV, a major causative agent of sepsis and pulmonary pathogenesis, respectively. Toshchakov and colleagues (52) have previously demonstrated that targeting TLR2- and TLR4-induced signaling with Mal-derived decoy peptides could inhibit signaling at micromolar concentrations with most successful decoy peptides targeting the regions of mouse Mal aa 152–163 and 168–182. Additional studies from this group also demonstrated that a decoy peptide designed to disrupt Mal recruitment to TLRs was also protective in an IAV model of infection (49), further highlighting the inflammatory regulatory properties of Mal and the therapeutic potential of targeting Mal-mediated signaling to reduce immunopathologies during infectious disease. Our studies suggest that targeting the human Mal D96 region may also provide reduced TLR4-mediated inflammation and disease outcomes.

Collectively, these studies highlight the critical role Mal plays in moderating TLR-mediated inflammatory signaling and disease outcomes in relation to TLR4-induced disease and highlights the potential therapeutic advantages of targeting Mal recruitment of MyD88 to the receptor complex and subsequent downstream inflammation.

The authors have no financial conflicts of interest.