Genetic analysis of the innate immune responses in wild-derived inbred strains of mice

Authors

  • Kristin Stephan,

    1. Graduate Program in Immunology, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    Search for more papers by this author
  • Irina Smirnova,

    1. Department of Pathology, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    Search for more papers by this author
  • Berri Jacque,

    1. Graduate Program in Immunology, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    Search for more papers by this author
  • Alexander Poltorak Dr.

    Corresponding author
    1. Graduate Program in Immunology, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    2. Department of Pathology, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    3. Graduate Program in Genetics, Sackler School of Graduate Biomedical Sciences, Tufts University School of Medicine, Boston, USA
    • Department of Pathology, Tufts University School of Medicine, 150 Harrison Avenue, J-512, Boston, MA 02111, USA, Fax: +1-617-636-2990
    Search for more papers by this author

Abstract

The vertebrate immune system has evolved to recognize nucleic acids of bacterial and viral origin. Microbial DNA, as well as synthetic oligonucleotides based on these motifs, activates innate immune pathways mediated by the family of Toll-like receptors (TLR) initiating a cascade of signals in immune cells necessary for responses to pathogens. However, not all of the proteins that participate in TLR-mediated responses have been identified. In studies described herein, we observed significant variation in innate immune responses among selected wild-derived strains of mice. Specifically, we show that mice of MOLF/Ei, Czech/Ei, and MSM/Ms strains are hypo-responsive to polyinosinic-polycytidylic acid (poly(I:C)) because of a mutation in Tlr3. In addition, we discovered a hypo-response to cytosine guanine dinuleotide in MOLF/Ei mice and established that it is not linked to Tlr9, but to another locus. Further inquiry revealed that this hypo-response is transmitted as a monogenic dominant trait that can be mapped and cloned through positional cloning methods. These results suggest the existence of a novel molecule that can alter pro-inflammatory signals or activate additional signal transduction pathways. In addition, they support the wild-derived mouse strain as a forward genetic tool for the identification of novel immunological phenotypes.

Abbreviations:
Chr:

chromosome

Chy:

CpG hypo-response

cM:

centimorgan

LOD:

logarithm of odds

ODN:

oligodeoxynucleotide

Poly (I:

C): polyinosinic-polycytidylic acid

SSLP:

simple sequence length polymorphic

Introduction

The innate immune system of mammals is able to detect microbial pathogens via recognition of the molecules that are often referred to as pathogen-associated molecular patterns (PAMP) 1. The biological activity of some of these molecules, such as LPS, a component of Gram-negative bacteria, has been known for many decades 2, before the signaling pathway initiated by LPS was elucidated. Similarly, the activation of B cells by bacterial nucleic acids containing unmethylated CpG motifs was reported a long time before the mechanism of their recognition 3. We know now that these and other molecules of microbial origin are directly sensed by TLR 4, 5 and activate cells of the innate and adaptive immune systems, including B cells, natural killer cells, dendritic cells, and monocytes. In mice and humans, TLR2 is essential for the recognition of peptidoglycan, lipopeptides or lipoteichoic acid 6 while LPS activates a TLR4-mediated pathway 7. TLR3 mediates the recognition of double-stranded viral RNA 8. TLR9 recognizes unmethylated motifs of bacterial DNA referred to as CpG motifs 9. Many of the microbial components, including CpG and guanosine derivatives, are used as synthetic adjuvants for the adaptive immune response.

Despite significant progress understanding how the TLR network initiates a robust host response, the molecular basis of these activities is not completely understood. Specifically, very little is known about the mechanism that controls the magnitude of inflammation 10. It is well accepted that pro-inflammatory signaling downstream of TLR involves a family of five MyD88-like adaptor proteins 11, a family of IRAK proteins (IL-1 receptor-associated kinase), and the TRAF-6 (TNF receptor-associated factor 6) transcription factor 12. Further amplification of the signal leads to the activation of mitogen-activated protein kinases (MAPK), p38 kinase and the transcription factors activating protein-1 (AP-1) and nuclear factor-kB (NF-kB). Activated transcription factors translocate to the nucleus and activate many pro-inflammatory genes. It is also believed that several anti-inflammatory signaling pathways are involved in down-regulation of the inflammatory response. However, the precise mechanism and timing of the interplay between components of pro- and anti-inflammatory signaling pathways is unknown 13. Current evidence suggests that in addition to several key negative regulators of signaling 14, 15, there may be additional unknown components of pro-inflammatory signaling mediated by several TLR 16. Given the complexity of TLR-mediated signaling and high probability of synergistic interaction within the TLR-network 17, the identification of additional mouse models that reveal new components of signaling would greatly enhance our ability to investigate the mechanism and regulation of innate responses. For this reason, we initiated a screen for innate responses in wild-derived inbred mouse strains.

The mouse model has been instrumental in identifying the function of most members of the TLR family 18. Several inbred mouse strains are naturally deficient in their responses to LPS 19, 20. Gene targeting in the germ line helped to discover many components of innate signaling. In addition, germ-line mutagenesis in mice has recently emerged as the mainstay of efforts to assign gene functions 21. This now well-established technique yields robust data and has contributed significantly to our knowledge of immune functions 22. However, because of the relatively large size of the mouse genome and the low frequency of mutations, random mutagenesis in mice remains chiefly the province of large specialized laboratories and consortiums. With the number of genes in the mammalian genome (∼34 000) significantly exceeding the number of functions assigned to these genes, finding additional approaches to close the phenotypic gap is of great importance. One obvious possibility is to improve current phenotypic screens. Another is to employ a genetically diverse model, such as wild-derived strains of mice 23, which are an unlimited reservoir of genetic polymorphisms 24. They exhibit, on average, a single nucleotide polymorphism in every 100–200 bp when compared with classical laboratory strains 25.

In our attempt to find new components of TLR-mediated signaling, we initiated innate immune screens on a panel of wild-derived inbred mouse strains. We hypothesized that their greater genetic diversity would facilitate the identification of new phenotypes previously undescribed in laboratory mice. Some of the recently identified immune relevant wild-derived alleles include Mx126, which makes wild-derived mice resistant to influenza virus, and Wnv, 27, which plays a critical role in the pathogenesis of West Nile virus infection. In addition, wild-derived strains helped to identify new alleles conferring resistance to Salmonella infection 28 and to TNF-induced lethal shock 29.

We found that MSM/Ms mice are non-responsive to polyinosinic-polycytidylic acid (poly(I:C)) because of a mutation in Tlr3. In addition, we discovered a hypo-response to CpG in MOLF/Ei mice, another wild-derived strain. Further inquiry revealed that the CpG-response in MOLF mice is a dominant monogenic trait not linked to Tlr9 but to another locus provisionally termed Chy (CpG hypo-response).

Results

Molecular phylogeny of wild-derived mice at the Tlr4 locus

To validate our hypothesis regarding genetic variations in the wild-derived mouse strains, we sequenced the Tlr4 gene of six wild-derived strains of mice and performed phylogenetic comparison between laboratory inbred and wild-derived inbred strains of mice. In previous reports, the Tlr4 sequence has been extensively analyzed among inbred laboratory mouse strains 30. Therefore, Tlr4 exons from six different wild-derived inbred strains, CALB/RK, CAST/Ei, CZECH/Ei (CZECH), MOLF/Ei (MOLF), MSM/Ms (MSM), and SPRETUS/Ei (SPRETUS) were sequenced and aligned using the ClustalW multiple alignment program. The resultant phylogenetic tree was compared to the correspondent arrangement made previously for inbred laboratory strains 30 (Fig. 1). Among 35 inbred strains, only 9 alleles evolved from the ancestor allele that is common for the rest of the 26 strains. By contrast, all 6 wild-derived alleles independently evolved from a common ancestor's allele (Fig. 1A). Most importantly, there were only 2 common aa changes (in red, Fig. 1B) between wild-derived and laboratory inbred mice with 26 aa polymorphisms in wild-derived strains overall. Comparably, 35 inbred strains accumulated 10 aa polymorphisms. Thus, haplotype analysis of murine Tlr4 indicates that wild-derived mice exhibit a greater variety of natural genetic polymorphisms and suggests that such diversity might provide new immunologically relevant phenotypes.

Figure 1.

Genetic distance and probable ancestral relationships among Tlr4 genes of six wild-derived strains of mice (A). Numbers adjacent to arrows represent the number of mutations separating one strain from another and are proportional to the estimated distances between the strains (Clustal W). Distribution of coding mutations found in Tlr4 of six wild-derived strains of mice. Only coding mutations are shown. DNA samples for sequencing analysis were obtained from the Jackson Laboratory. (B) Distribution of the polymorphisms observed in wild-derived mice in respect to the exons of TLR4. Mutations common between wild and inbred strains are shown in red; mutations that are structurally significant, in blue.

Variations in innate responses in the wild-derived mouse strains

On a phenotypic level, we compared innate responses of the wild-derived strains against several agonists of TLR. In addition to LPS, poly(I:C) (ligand for TLR3), peptidoglycan (TLR2 stimuli), CpG (TLR9 agonist), and resiquimod (mouse TLR7 ligand) were chosen. The in vitro response was measured by the amount of TNF secreted by activated macrophages. The results of the screen are presented in Fig. 2 and reveal a significantly low secretion of TNF in response to poly(I:C) and CpG in MOLF mice. In addition, two other strains, MSM and CZECH, were hypo-responsive to poly(I:C). High levels of TNF in LPS-activated MOLF/Ei and MSM macrophages indicate that the low response to poly(I:C) and CpG is unlikely linked to a signaling molecule downstream of the TLR because most of the downstream components, like MyD88, TRIF, the downstream adaptors, and the kinases are shared by several innate response pathways, which are functional in these strains. Indeed, MSM, MOLF, and Czech mice are fully responsive to LPS (TLR4), peptidoglycan (TLR2), and resiquimod (TLR7). In addition to MyD88, signaling from TLR4 depends also on TRIF, a sole contributor to TLR3-mediated signaling. To confirm that the hypo-response to poly(I:C) in MOLF, Czech, and MSM and the hypo-response of MOLF/Ei mice to CpG is not related to a deficiency of TNF production, we analyzed the expression of other cytokines, including TNF, in RNase protection assays (RPA). In these experiments, the presence of mRNA corresponding to several different cytokines can be detected simultaneously (Fig. 3). LPS elicits robust and prompt responses from peritoneal macrophages of both strains of mice, MSM and MOLF. As a result, transcription of TNF and other inflammatory genes such as IL-1β, RANTES, IL-1α, and others is up-regulated, indicating that an abnormality in the TNF response is not responsible for the observed defects. Furthermore, the hypo-response to CpG in MOLF/Ei mice affects numerous cytokines, not only TNF. Because of the well-established role of Tlr3 as a recognition receptor for poly(I:C), we amplified and sequenced Tlr3 cDNA from all wild-derived strains and compared it with the wild-type allele of C57BL/6J mice. More than 30 polymorphisms have been found, 50% of them resulting in amino-acid changes. MSM, MOLF, and CZECH strains possess a common allele of Tlr3. Only one nucleotide substitution was found to segregate between poly(I:C) responsive and hypo-responsive mice: the substitution of a T instead of a C in the coding region of Tlr3 results in the substitution P(369)L in one of the LRR of Tlr3 (Fig. 4A). In addition, proline 369 is evolutionarily conserved among humans, rats, and mice. Altogether, our sequencing analysis suggested that the Tlr3 allele is a likely cause of the hypo-response to poly(I:C) observed in several strains of mice. In addition, cytokine profiling using RPA indicated that hypo-responses to poly(I:C) and to CpG are specific for TLR3 and TLR9 signaling and cannot be explained by general defects in the MyD88 pathway or in cytokine induction. Appearance of a common Tlr3 allele in three poly(I:C) hypo-responsive strains indicates that this defect might be linked to the TLR3 locus.

Figure 2.

Hypo-responsiveness to various TLR agonists is observed in several wild-derived mice. Peritoneal macrophages elicited from C57BL6/J, MOLF/Ei, MSM/Ms, Spretus/Ei, Czech/Ei, Cast/Ei, and Calb/Ei mice were activated with various TLR-specific stimuli and after 4 h of incubation the supernatant was collected and subsequently analyzed for the presence of TNF-α using ELISA (RαD). For elicitation of a CpG response HPLC-purified ODN 1668 was used with the following sequence 5′-TCCATGACGTTCCTGATGCT-3(Phosphorothioate). LPS was a gift of C. Galanos (Max Planck Institute, Freiburg, Germany), resiquimod from 3 M Pharmaceuticals; other stimuli were commercially available.

Figure 3.

The hypo-response phenotypes are not due to defective TNF activation. RNase protection assay. RNA isolated from macrophages of MSM and MOLF mice was hybridized with an anti-sense probe correspondent to several cytokines (on the left). Time of stimulation with agonist (top) is given in hours. L32 and GAPDH transcripts are provided as RNA loading controls (bottom).

Figure 4.

Tlr3 in wild-derived strains is responsible for the poly(I:C) hypo-response phenotype: (A) Sequence alignment of murine Tlr3 amplified from different strains of mice. Rectangle indicates a P(369)L substitution in one of the leucine-rich repeats (LRR) of the extracellular domain of TLR3. (B) High resolution map of Chr. 8 of the mouse. Units are in cM. Arrows show transcriptions of the genes. (C) Luciferase reporter assay. HEK293 cells were co-transfected with an NF-kB reporter construct and a construct expressing either mutant or wild-type Tlr3 alleles. Following a 7-h stimulation with poly(I:C) or LPS, luciferase activation was measured with Steadylite HTS. (D) TLR3 mutant MOLF/Ei mice fail to degrade IkB-α, the NF-kB negative regulator, upon stimulation with poly(I:C). Peritoneal macrophages (5 × 105) were stimulated with 100 μg/mL poly(I:C). Total cell lysates were harvested at the indicated time (minutes) and subjected to SDS-PAGE followed by Western blotting.

Genetic mapping of hypo-response to poly(I:C) in MSM/Ms mice

To perform linkage analysis, we crossed C57BL/6 and MSM/Ms mice to obtain F1 hybrids. The F1 were further intercrossed and the resultant F2 progeny were analyzed for a response to poly(I:C). A cohort of 20 F2 animals was phenotyped for the production of TNF and genotyped for the Tlr3 locus. The deviation between observed frequencies and expected frequencies (given the hypothesis that there is linkage) was estimated using Chi-square analysis. Accordingly, we mapped the non-responsive phenotype of the MSM strain to the Tlr3 locus (logarithm of odds, LOD = 30.6, with p <0.000001) by using a panel of 122 intercross animals and measuring the amount of secreted TNF as a quantitative trait. The poly(I:C) non-response behaved as a monogenic trait and was circumscribed to a position within 2 centimorgan (cM) between the D8MIT227 and D8MIT297 markers (Fig. 4B). Genetic analysis was carried out over an interval, and several genes, including Tlr3, were identified within the genetic region. The role of TLR3 in mediating the signal from poly(I:C) is well established 8. In addition, there were no other authentic immune-specific transcripts in the vicinity of the Tlr3 locus. Altogether, this suggests that the hypo-response to poly(I:C) in some wild-derived strains is due to a defective Tlr3 allele. We also performed haplotype analysis of the critical region using 20 genetic markers for the three poly(I:C) hypo-responsive strains, MSM/Ms, MOLF/Ei, and CZECH/Ei. MSM/Ms and MOLF/Ei strains were virtually identical through the entire area flanked by D8MIT227 and D8MIT297 markers. However, CZECH/Ei, with the exception of the Tlr3 locus, had a different haplotype, revealed by both SNP and simple sequence length polymorphic (SSLP) analysis. At the same time, the Tlr3 haplotype in all poly(I:C) non-responsive mice was identical. Taken together, the mapping data and the haplotype analysis in the Tlr3 locus provide sufficient evidence that the non-response to poly(I:C) is conferred by a defective Tlr3 allele.

Functional analysis of the P369L mutation on TLR3 signaling

To establish if the P369L mutation was responsible for the poly(I:C) hypo-response phenotype, we investigated the functionality of two late signaling events downstream of TLR3. First, we investigated if the presence of the mutant P369L allele was sufficient to abrogate NF-kB transcription upon stimulation with poly(I:C) (Fig. 4C). By co-expressing mutant MSM/Ms Tlr3 and NF-KB luciferase reporter constructs in HEK 293 cells, we found that the MSM/Ms derived P369L allele of Tlr3 failed to activate NF-kB upon poly(I:C) addition. In contrast, the wild-type C57BL/6J allele of Tlr3 strongly induced NF-kB reporter expression. As additional evidence that the P369L mutation prevents signal transduction of the TLR3 pathway, we analyzed the degradation of IkB-α, the negative regulator of NF-kB, following stimulation of peritoneal macrophages from MOLF/Ei and B6 mice with poly(I:C) (Fig. 4D). In contrast to wild-type B6 macrophages, MOLF/Ei peritoneal macrophages fail to degrade IkB. In agreement with the luciferase reporter assay, these data demonstrated that the P369L mutation prevents TLR3 signal transduction.

Sequencing analysis of the Tlr9 allele in different wild-derived mouse strains

Peritoneal macrophages elicited from MOLF/Ei were hypo-responsive to CpG, which was revealed by the low amount of TNF produced in response to CpG. In contrast, when all other wild-derived strains were stimulated with CpG, TNF levels were significantly elevated. CpG motifs are recognized both in humans and in mice by TLR9, which is expressed primarily on cells of myeloid origin and on B cells 31. MOLF/Ei macrophages have a normal response to LPS and peptidoglycan, indicating that the mutation does not reside in any of the signaling molecules downstream of Tlr9 because these components are also utilized by TLR4 (receptor for LPS) and TLR2 (receptor for peptidoglycan). To exclude the possibility that the CpG defect could be related to Tlr9, we analyzed the sequence of Tlr9 on the genomic level in all wild-derived strains. Tlr9 sequencing revealed numerous polymorphisms that would result in aa substitutions in all strains. Most differences affected the extracellular domain of TLR9 with a few substitutions in the transmembrane and TIR domains. Among all changes, there was only one that was specific for the MOLF/Ei strain, an A(406)V substitution in the extracellular domain. Sequencing data together with the normal responsiveness of MOLF mice to other TLR ligands, such as LPS and peptidoglycan, suggest that the mutation in Tlr9 could result in a low CpG-response. To address this question, we performed linkage analysis on a panel of intercross mice.

The CpG-hypo-response is conferred by the Chy locus

To clarify the role of Tlr9 in the CpG hypo-response of MOLF/Ei, we generated F1 (C57BL/6J × MOLF/Ei) hybrids and produced F2 intercross animals. F2 progeny, together with F1 hybrids and parental strains, were analyzed on their response to CpG. The TNF production in F1 mice was severely impaired compared to the C57BL/6J strain (Fig. 5A), indicating that CpG hypo-responsiveness in the MOLF strain is conferred by a dominant allele. Next, we established a null hypothesis that this trait is linked to the A(406)V Tlr9 mutation in MOLF mice. Similar to the mapping of the poly(I:C) hypo-response in MSM mice, we determined the phenotype of F2 (C57BL/6J × MOLF/Ei) progeny and then analyzed genotypes using markers proximal to murine Tlr9 from both sides of the locus. The difference between the numbers of expected and observed phenotypes for all groups of F2 animals with different genotypes was used to calculate the chi-square value according to the following formula: χ2 = Σ (obsi - expi)/expi. We obtained χ2 = 9.5, a value indicating that our hypothesis is wrong and that the hypo-response to CpG and the Tlr9 locus segregate independently. Therefore, we conclude that the phenotype does not result from a structural defect in MOLF TLR9 and is caused by an independent locus. The phenotypic analysis of F1 and F2 mice indicates that the CpG defect is a dominant transmittable trait. The 3:1 mode of distribution of phenotypes evident in F2 animals (Fig. 5B) suggests that mutation is encoded by a single genetic locus provisionally termed Chy.

Figure 5.

A phenotypic difference between C57BL/6J and MOLF permits genetic mapping of the locus that confers the hypo-response to CpG. (A) CpG response in peritoneal macrophages of B6, MOLF, and F1 hybrids; (B) phenotypic analysis of F2 mice. Bimodality was confirmed with the likelihood ratio test (p <0.0001, likelihood of unimodality).

Phenotypic characterization of the hypo-response to CpG in MOLF/Ei mice

The initial conclusion that the signaling molecules downstream of TLR9 are intact was based on the fact that MOLF macrophages produce TNF in response to LPS. However, this response was examined only for a relatively high concentration of LPS, 100 ng/mL. Therefore, to exclude the possibility that MOLF cells may have a different threshold for LPS activation and thus a deficiency in signaling could be common for the TLR4 and TLR9 pathways, we compared the LPS response in B6 and MOLF macrophages over a wide range of LPS concentrations (Fig. 6). When measured by the amount of secreted TNF, the MOLF macrophage response to LPS is even more vigorous than B6 macrophages at all tested concentrations of LPS, allowing us to dismiss the possibility of defective LPS-signaling in MOLF mice. These results suggest that the hypo-response to CpG by MOLF mice is unlikely to be explained by a defect in any known molecules downstream of TLR9, which are shared by TLR4 and TLR9. Furthermore, CFSE-labeled MOLF/Ei splenic B cells are capable of robust proliferation in response to CpG (Fig. 7), a response known to require TLR9 9. This provides additional evidence that the TLR9 protein is functional in MOLF/Ei mice.

Figure 6.

MOLF peritoneal macrophages do not display altered sensitivity to LPS. Cells (50 000/well) in 96-well plate were incubated with the indicated LPS concentration for 4.5 h. Supernatants were subjected to TNF-α ELISA.

Figure 7.

MOLF/Ei TLR9 is functional. CD19+ B cells were purified from the spleens of MOLF/Ei mouse and labeled with 2 μM CFSE. After stimulation for 48 h with 200 nM CpG or medium only, proliferation was assessed by flow cytometry.

To further investigate the pattern of CpG-initiated pro-inflammatory signaling, we followed the kinetics of TNF mRNA accumulation in B6 and MOLF macrophages (Fig. 8). The cellular responses to LPS were used for comparison with CpG activation. As expected, priming of MOLF and B6 macrophages with LPS elevated TNF mRNA to similar levels (Fig. 8A). Activation with CpG resulted in dramatic differences in expression patterns of TNF. The failure to induce TNF in MOLF macrophages in response to CpG was confirmed. These experiments confirm that the defect in MOLF macrophages is specific for CpG and is linked to impaired production of pro-inflammatory cytokines.

Figure 8.

CpG-specific impairment of TNF expression in MOLF macrophages. (A) Peritoneal cells were plated at a density of 5 × 105/well and activated with either LPS or CpG for the indicated time (Hr, at the top). The blot was hybridized with a TNF-specific riboprobe and exposed to X-ray film for 5 h. Arrows indicate TNF mRNA and total RNA (28S and 18S). (B) Western blot of IkB degradation in peritoneal macrophages of B6 and MOLF mice upon stimulation with 1 μM CpG.

Based on the current understanding of TLR signal transduction, the activation of NF-kB requires the inducible phosphorylation and subsequent degradation of IkB that, in the absence of stimuli, blocks nuclear import of p65 or c-Rel NF-KB subunits. Accordingly, we analyzed the kinetics of IkB degradation following CpG priming of macrophages and found no difference between MOLF/Ei and wild-type cells (Fig. 8B). Thirty minutes after stimulation with CpG of either peritoneal or bone marrow-derived macrophages, we observed degradation of IkB-α followed by its re-synthesis at later time points. Thus, TLR9-mediated signaling upstream of NF-kB appears to be normal in MOLF/Ei mice.

Endocytosis of CpG is a well-established requirement for macrophage activation by CpG-oligodeoxynucleotides (ODN). Therefore, to exclude the possibility that the hypo-response to CpG is associated with the general deficiency in the uptake of nucleic acids, we further studied the DNA uptake in MOLF/Ei and B6 mice using Alexa-488-labeled CpG. The uptake of the label was assessed by flow cytometry (Figure 9). Mac1+ macrophages are shown on histograms on the right from the corresponding dot plots. After incubation for 45 s with 0.1 μM labeled CpG (or DMEM alone), unincorporated CpG was washed away before cells were stained for FACS analysis. The number of CpG-stained cells for the B6 BMDM was slightly reduced (61%) compared to MOLF/Ei BMDM (70%), suggesting that uptake of nucleic acids is unlikely the cause of CpG hypo-responsiveness in MOLF mice.

Figure 9.

CpG uptake in MOLF and B6 macrophages. Cells were loaded with Alexa-488 labeled CpG (open) or with medium (filled) and gated for Mac-1 marker (y-axis). The histogram overlays (right) are gated on MAC-1+ cells.

Genetic characterization of the hypo-response to CpG

The hypo-response to CpG in MOLF mice is a dominantly inherited trait. Transmissibility of the trait allowed us to used meiotic recombination to determine the chromosomal location of the gene that confers the trait. As previously mentioned, the phenotypic effect exerted by a defective allele in MOLF/Ei mice is dominant, which is revealed by the fact that F1 hybrids produced in a B6 × MOLF/Ei cross were phenotypically identical to the MOLF/Ei parental strain (Fig. 5A). To reveal the differences in CpG responsiveness, F1 hybrids were crossed back to the B6 parental strain. The resultant cohort of 62 mice was used to perform a genome-wide scan of chromosomes (Chr) 1 through 19 and Chr X. Haplotype data for the mapping were generated with 83 SSLP markers spaced at approximately 20-cM intervals. The LOD score of the CpG response as a function of position in the mouse genome has been calculated for all markers. The results of this linkage are presented in Fig. 10. We have mapped the Chy locus to Chr19 (LOD score of 4.77).

Figure 10.

Transgenomic log likelihood (LOD score) analysis of the CpG hypo-response in N2(B6 × MOLF) mice. Sixty-two meioses were genotyped for 83 markers. The LOD was calculated according to Lg(L(Θ)/LG(0.5)), where LΘ is the probability of the event.

Discussion

The TLR protein family is essential for the recognition of pathogens by cells of the innate immune system. Among different TLR, TLR3 recognizes dsRNA, which is produced by many viruses during their replication cycle as an essential intermediate for RNA synthesis. In addition, some viruses encode RNA species, which have considerable ds structures. Specificity of TLR3 towards dsRNA was originally evidenced in TLR3-/- mice, which lost their pro-inflammatory response to poly(I:C), a synthetic analog of dsRNA 8.

In this report, we describe a new allele of TLR3, which confers hypo-responsiveness to poly(I:C) in several wild-derived inbred strains of mice. The identification of this allele was made possible via positional cloning on a panel of F2(B6 ×x MSM) mice followed by haplotype analysis in the vicinity of Tlr3 in three poly(I:C) hypo-responsive wild-derived strains. We found that (i) the phenotype is mapped to a relatively deserted 2-cM region containing several genes with TLR3 among them; (ii) haplotype analysis of the critical region for the poly(I:C) non-responsive strains allowed us to further narrow the area leaving Tlr3 as the only candidate; and (iii) nucleotide differences conferring a defective Tlr3 allele segregate between poly(I:C) responsive and non-responsive mice.

In all poly(I:C) hypo-responsive strains, a mis-sense mutation results in the substitution of an evolutionary conserved Proline to Leucine in one of the extracellular domain LRR of TLR3. Despite the recent crystallization of the extracellular domain of human TLR3 32, 33 the definitive poly I:C-binding site has yet to be identified. However, by comparing the mouse TLR3 sequence to the human TLR3 sequence, we are able to predict the consequence of P(369)L on ligand binding. According to Bell et al., this proline, which is conserved in mice, lies within a predicted ligand-binding groove in human TLR3. Therefore, we expect the harsh proline to leucine mutation in this position to disrupt the conformation of the ligand-binding site. Systematic screening of the laboratory inbred mouse strains on their response to various molecular TLR agonists has yet to be performed, but no other alleles of TLR conferring hypo-responsiveness to any of these agonists have been identified. Recently, ten different, underrepresented inbred strains of mice have been tested on their innate responses towards several TLR ligands and no significant variations in their inflammatory responses were observed (Smirnova and Poltorak, unpublished). In this regard, finding a poly(I:C) hypo-responsive Tlr3 allele that is shared by three different wild-derived strains of mice warrants further exploration of the genetic diversity among the wild-derived strains. While the genetic and phenotypic diversity of these strains has never been challenged, their utility in mapping is frequently criticized because of their poor breeding capabilities and lower than average recombination rate with the genome of laboratory inbred mice. In contrast to the latter, positional cloning of the mutation in Tlr3 was achieved on a relatively small panel of mice in a short period. Therefore, our preliminary data show that certain wild-derived mouse strains allow vigorous genetic analysis.

In addition to a defective poly(I:C) response, MOLF mice were also hypo-responsive to CpG. The normal response of MOLF macrophages to other TLR agonists, such as LPS, peptidoglycan, and loxoribine, indicate that the defect is unlikely to be linked to a downstream component of TLR9 signal transduction, as many of these components are shared by different TLR. However, we cannot rule out a defect in a novel CpG-specific signaling protein. Linkage analysis helped us to dismiss the polymorphism in Tlr9 as a possible cause of the defect. Phenotypic analysis of the components positioned downstream of TLR9 provided additional information regarding the defect in CpG-signaling in MOLF mice. First, transcriptional activation of TNF in MOLF macrophages and B cell proliferation upon CpG treatment demonstrates there is signal transduction by the TLR9 pathway. However, the low abundance of TNF transcripts in MOLF explain the minute amounts of TNF protein detected in the supernatants of CpG-stimulated macrophages. Thus, the initial activation steps of the pro-inflammatory response in MOLF macrophages were very similar to the wild-type cells. Indeed, the kinetics of phosphorylation of the NF-kB inhibitor, IkB, in MOLF and B6 cells was identical, suggesting TLR9 signal transduction is functional. Further inquiry into TLR4-mediated signaling in MOLF macrophages did not reveal any defects in transcriptional activation upon LPS stimulation and showed even higher amounts of secreted TNF than B6 macrophages. Thus, the TLR4 pathway is intact in MOLF mice, suggesting that any differences between processing the signal from LPS and CpG might be important for analysis. Despite similarities in Th1-biased cytokine induction, the ability to induce T cell-independent proliferation of B cells, and the induction of antibody production, these two agonists have substantial differences. LPS is recognized on the surface of the cell by TLR4, the main transducer of signals from LPS, the GPI-linked receptor CD14, and a small, exteriorized adaptor known as MD-2 34. In contrast to that, CpG nucleic acids get endocytosed, reach early endosomes and within 10 min enter the late endosomes 35, 36. Two hours after addition of CpG, the size and the number of CpG containing late endosomes increase significantly. Accordingly, we assessed the uptake and endosomal localization (data not shown) of CpG in MOLF macrophages and did not find any difference compared to B6. Additional evidence of normal nucleic acid trafficking is the ability of MOLF macrophages to secrete TNF at levels comparable to wild-type macrophages upon stimulation with loxoribine. Loxoribine is a guanosine analog derivatized at position N7 and C8 recognized by Z7 and is believed to have similar requirements for endocytosis and endosomal maturation 37. The normal response to loxoribine by MOLF macrophages indicates that these mice do not have general deficiency neither in the uptake of nucleic acids nor endosomal maturation. Thus, in MOLF cells, despite reaching subcellular compartments and achieving activation of NF-kB, CpG is not capable of up-regulating the pro-inflammatory response to the levels of wild-type cells. Further inquiry into NF-kB activation would be useful for evaluation of the defect on the transcriptional level in MOLF mice. However, these experiments should be considered in the context of possible differences in signaling initiated by CpG and other agonists of TLR, to which MOLF mice appear to be fully responsive. Especially interesting would be to examine CpG-initiated signaling that could be independent of TLR9 or MyD88. In this respect, it is noteworthy that AKT is reportedly activated downstream of TLR9 36. However, there is also a contradictory report that TLR9 is not required for activation of AKT by CpG-ODN 38.

In addition to the phenotypic analysis of the hypo-response to CpG, we have expanded the cohort of backcross animals that was initially used for linkage analysis and obtained a chromosomal localization for the gene that confers CpG hypo-responsiveness in MOLF. The gene is contained within the locus that we provisionally termed Chy at the central part of mouse Chr 19. To improve the linkage data, we plan to expand the panel of backcross mice and to increase the density of markers surrounding the critical area. Such a strategy will allow us to proceed with high-resolution mapping of Chy. Finding the gene encoded by Chy could represent a major advance in our understanding of the normal response to bacterial DNA. Failure to mount such a response may result in susceptibilities towards numerous bacterial infections. In this regard, it is noteworthy that the MOLF strain exhibits resistance to infection with Salmonella typhimurium, which was shown to be inherited as a polygenic trait 28. It was confined to three independent loci within Chr 1 and 11 with no linkage to Chr 19 being found. In a separate study, it was shown that MOLF mice are extremely susceptible to another Gram-negative microorganism, Salmonella enterica39. It is plausible that Chy contributes to the immune response in vivo that is mounted against these infections. The possible involvement of Chy in mounting a host response to these infections could be further facilitated by the production of congenic lines carrying Chy on a resistant background.

Historically, the wild-derived mouse strains are underrepresented in forward genetic studies of the immune system. Here, we present evidence that wild-derived mice exhibit a greater variety of natural genetic polymorphisms, which might provide new immunologically relevant phenotypes. Furthermore, we were able to identify a new allele of Tlr3, which confers a hypo-response to dsRNA in several wild-derived mouse strains. In addition, we identified a new locus, Chy, which confers a hypo-response to CpG in the MOLF/Ei strain. Altogether, these findings describe the wild-derived mouse strains as very appropriate models for studies of innate immune functions and the identification of new genes that contribute to the host response to bacterial and viral infections.

Materials and methods

Mice

C57BL/6J, MOLF/Ei, CALB/RK, CAST/Ei, CZECH/Ei, MSM/Ms, and SPRETUS/Ei mice were obtained from the Jackson Laboratories. All experiments were carried out in compliance with the rules of the Institutional Animal Care and Use Committee at Tufts-New England Medical Center.

Genetic mapping, linkage analysis and sequencing

SSLP analysis was performed according to standard procedures using high molecular weight genomic DNA. MAP MAKER was used to construct a genetic map. For linkage analysis, chi-square test for intercross data with three degrees of freedom was used. Sequencing was carried out on PCR amplified DNA fragments with internal primers.

Reagents

S. minnesota LPS was obtained from Alexis, peptidoglycan and loxoribin were purchased from Invivogen, resiquimod was obtained from 3 M Corporation. Unmethylated DNA oligonucleotide 1668 was synthesized by Integrated DNA Technologies. Antibodies to AKT and IkB-α were purchased from Cell Signaling. Antibodies to CD19 and B220 were from PharMingen. M-CSF was produced by L929 cells (gift of Dr. R. Isberg, Tufts University).

RNase protection assay

Custom-made multiple templates probe was purchased from PharMingen and was transcribed with T7 RNA polymerase in the presence of 32P incubated overnight at 56°C with the RNA samples, and treated sequentially with RNase and proteinase K. After phenol/chloroform extraction, the samples were loaded on a polyacrylamide gel and run for 45 min with 0.5X TBE as the running buffer at 50°C. Gel was dried and exposed to X-ray film.

Isolation of peritoneal macrophages

Mice were i.p. injected with 2–3 c.c. 3% thioglycollate medium. Three days post-injection, enriched macrophages were harvested by peritoneal lavage with sterile PBS.

Generation of bone marrow-derived macrophages

Bone marrow femurs were rinsed with RPMI 1640. Cells were cultured on sterile, bacteriological petri dishes for 7 days in bone marrow macrophage medium (RPMI 1640 + 2% pen/strep + 30% L cell sup + 20% FBS). After 7 days culture, the adherent macrophages were harvested with ice-cold PBS.

B cell proliferation

B cells were purified from the spleens of 8-week-old mice using Easy Sep® negative selection magnetic beads (Stem Cell Technologies). Purity was consistently assessed to be >94% by CD19+B220+ flow cytometry. B cells (5 × 106) were labeled with 2 μm CFSE (Molecular Probes) 10 min with shaking at 37°C, then washed three times in PBS containing 3% BSA. The labeled cells were plated at 1 X 106 cells/well in a 24-well dish in 1 mL complete medium with or without 200 nM CpG. After 48 h, CFSE intensity was assessed by flow cytometry using a FACScalibur (Becton Dickinson) and FlowJo software.

Northern blotting

RNA was prepared from thioglycollate-elicited peritoneal macrophage using Trizol (Invitrogen), with 5 × 105 cells being used for a single time point. RNA was re-dissolved in 5 µL of formamide, mixed with 15 µL of RNA buffer (Ambion), denatured, and run on a 1% agarose gel. After capillary transfer (20x SSC) onto Nytran membrane (Amersham), the blot was hybridized with gene-specific riboprobes in 50% formamide at 60°C, washed according to standard procedures, and exposed with X-ray film or in Phospho-Imager.

Luciferase reporter assay

HEK 293 cells (ATCC, Rockville) were cultured in DMEM supplemented with 10% FBS. Cells were plated in 10-cm tissue culture plates at a density 1 × 106 cells/plate and 12 h later were transfected with TLR3 cDNA or vector DNA (topoA, Invitrogen) and 100 ng of NF-kB-luc (kind gift of Dr. S. Bunnell) using Fugene (Roche). All cells were also transfected with renilla control plasmid for normalizing transfection efficiencies. Twenty-four hours after transfection the cells were washed plated in 96-well plates at 50 000 cells/well and activated with poly(I:C), LPS, or left untreated. Seven hours later, cells were lysed and assayed for luciferase activity using the Steadylite HTS kit (Perkin Elmer). All measurements were done in triplicates.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

  7. 7

    WILEY-VCH

  8. 8

    WILEY-VCH

  9. 9

    WILEY-VCH

  10. 10

    WILEY-VCH

Ancillary