Forward genetic dissection of innate response to infection in inbred mouse strains: selected success stories

Introduction

The host genetic background plays an important role in determining onset, regulating progression and deciding the ultimate outcome of many human infections. Some of the clearest manifestations of this genetic component include the well-known examples of malaria resistance in individuals heterozygous for haemoglobinopathy alleles (HbS, thalassaemias) and human immunodeficiency virus (HIV) resistance in individuals carrying variants of the viral co-receptor chemokine receptor 5 (CCR5) [1–3]. Although researchers have identified single-gene effects in humans that cause either selective or generalized immunodeficiencies, expressed phenotypically as an innate susceptibility to certain types of bacterial and viral infections, candidate gene analyses and whole genome association studies have shown that susceptibility to infections in humans is often multi-factorial and genetically heterogeneous. In addition, the host genetic component is modified further by complex variable environmental factors such as diet and nutritional status, overall health status, living conditions, cross-immunity, microbe-associated effects, including virulence, and resistance to treatment [3–6].

The laboratory mouse provides a valuable complementary tool to identify major gene effects in such complex traits where confounding environmental factors can be controlled carefully. In particular, excellent experimental models of infection with murine pathogens have been developed that closely mimic pathogenesis and host response characteristics of the corresponding human infections. There are also excellent mouse reagents to study individual cell populations, physiological responses, biochemical pathways and individual genes and proteins that may regulate response to infections. There exist more than 200 commercially available, phylogenetically diverse inbred mouse strains that contain enough genetic diversity to identify individual strains showing major differences in response to a specific infection, and in which major gene effects can be identified by classical genetic analysis. These inbred strains are genetically pure (homozygous at all their loci) and can therefore be sampled repeatedly, facilitating multiple serial analyses of different aspects of pathogenesis and host response. Recent large-scale germline chemical mutagenesis programmes in the mouse have provided and will continue to provide an additional rich source of genetic variation to be mined. The availability of a complete and annotated mouse genome, high-density single nucleotide polymorphism (SNP) maps, and the possibility of germline modification (gene targeting/transgenesis) make the mouse an experimental model of choice for the genetic analysis of susceptibility and response to infection with bacterial, viral and parasitic pathogens. Genetic studies in these mouse models may identify genes, proteins and biochemical pathways central to the host defences, which can then be validated in parallel studies in human populations from areas of endemic disease. The study of differential susceptibility of inbred mouse strains to infection has been extremely successful in providing such candidates (reviewed extensively in [7]).

There are two major approaches to identify individual host genes and proteins that impact pathogenicity and host response to infections in mice. The first is so-called ‘reverse genetics’, in which individual genes are deleted from the germline by homologous recombination, followed by assessing the effect of the null mutation (knock-out mouse) in appropriate infection models. This strategy has been extremely successful, and much of our understanding of innate and acquired immunity in mice and humans derives from such studies. This field encompasses a large body of published work that will not be reviewed herein. However, reverse genetics carries some inherent limitations which include potential embryonic lethality of a null mutation, absence of a detectable phenotype due to genetic redundancy in the pathway targeted, compensatory mechanisms, and finally modulation of the mutation-associated phenotype by genetic background effects.

The second approach, known as ‘forward genetics’, is based on the identification of one or a few mouse strains that manifest an unusual or extreme response to infection, being either highly resistant or highly susceptible, or demonstrating a unique pathology. The genetic component of the trait is then analysed in standard informative crosses to determine inheritance (dominant, recessive) and complexity (monogenic versus polygenic). In general, a genome-wide scan is conducted to locate the gene (in case of monogenic trait) or major quantitative trait loci (QTL) influencing the trait (in the case of a complex trait). Such genes can then be identified by positional cloning, a tedious procedure that involves reduction of the genetic interval in additional crosses and systematic analysis of positional candidates, including validation in transgenic animals or in transfected cell lines or primary cells. While this approach is relatively straightforward for monogenic traits, gene identification is much more difficult and has rarely been successful for polygenic traits, although with a few exceptions [8]. In addition, the nature of the genetic lesion in a candidate gene may itself be difficult to decipher (gene copy number, coding polymorphisms, regulatory mutations, splicing variants, etc.), and complicated further by the presence of naturally occurring but functionally neutral polymorphic variants fixed during the inbreeding of the informative strain. These limitations are alleviated in the study of mouse mutants generated by chemical mutagenesis, where the phenotype-causing mutations can be detected more easily by virtue of their absence in the parental non-mutagenized strains [9,10].

The present review will highlight some key examples of successful implementation of forward genetic strategies to identify novel genes and pathways influencing the host innate response to infection (Table 1). These examples have led to a number of sometimes unexpected discoveries that have increased our understanding of the early innate pathways involved in responding to microbial infections.

Pathogen sensing

Host cells harbour a complex array of cell surface and intracellular sensors that recognize pathogens or pathogen-infected cells. Engagement of such receptors results in immediate activation of complex cellular signalling cascades and transcriptional responses that not only activate microbicidal effector functions of these cells, but that result in secretion of specific cytokines to recruit additional effector cells to the site of infection. The critical importance of these pathways has been highlighted by the discovery that several ‘infection susceptibility’ genes and proteins, identified by forward genetic analysis in inbred mouse strains, participate in and/or regulate these key early host–pathogen interactions.

Pattern recognition receptors, including members of the Toll-like receptor (TLR) and non-obese diabetic (NOD)-like receptor (NLR) families, recognize and respond to small molecular motifs that are associated with microbes and damaged or infected cells. These are called pathogen-associated and danger-associated molecular patterns (PAMPs and DAMPs). Upon stimulation, TLRs and NOD1 and NOD2 signal through adaptor proteins in a cascade that leads ultimately to the activation of a transcriptional response (reviewed in [11]). Stimulation of other NLRs results in caspase 1 activation, leading to the release of interleukin (IL)-1 and IL-18 and stimulating cell death via pyroptosis (reviewed in [12]).

Almost 20 years ago, the C3H/HeJ mouse strain was observed to have altered response to bacterial lipopolysaccharide (LPS). These mice are susceptible to infection with Salmonella Typhimurium, are resistant to toxic effects of LPS in vivo, and their splenocytes fail to display mitogenic response upon LPS exposure [13,14]. Genetic analyses showed that these phenotypes were controlled by a single locus on chromosome 4, designated Lps[15]. Subsequently, it was determined that a relative of the Drosophila Toll gene (involved in dorsoventral patterning and later shown to be important for immunity), designated the Toll-like receptor 4 (Tlr4) gene, mapped within the minimal Lps interval [16]. Sequencing of Tlr4 in C3H/HeJ mice revealed a proline to histidine missense mutation at position 714, a domain shown later to be responsible for interaction with the signalling molecule myeloid differentiation primary response gene (MyD88) [17,18]. Analysis of Tlr4 in two other LPS hyporesponsive mouse strains, C57BL/10ScCr and C57BL/6.KB2-mnd, showed that these strains carry partial or complete deletions of Tlr4[17,18]. The cloning of Lps provided the first demonstration of the role of TLR proteins in response to Salmonella and Salmonella-derived products.

Additional insights into TLR signalling pathways came from the study of ethyl-nitrosourea (ENU)-induced mouse mutants with abnormal responses to TLR ligands. The Lps2 ENU mutant was identified by the inability of its peritoneal exudate cells to produce inflammatory cytokines in response to known ligands of TLR-3 and TLR-4, double-stranded RNA (dsRNA) and LPS, respectively [19]. The defect in these mice was identified as a non-sense mutation in the Toll/IL-1 receptor (TIR)-domain-containing protein TRIF/Toll/IL-1 receptor domain (TIR)-containing adapter molecule 1 (TICAM-1); these studies led to the recognition that TRIF/TICAM-1 can transduce TLR-dependent signals in a MyD88-independent fashion. Further analysis indicated that TRIF is critical to a common signalling pathway shared by TLR-3 and TLR-4, and is absolutely required for any signalling downstream of TLR-3 [19]. The oblivious ENU mutant shows a reduced response to TLR-2 and TLR-6 ligands, namely lipotechoic acid and MALP2, and displays increased susceptibility to infection with Staphylococcus aureus[20]. The oblivious locus was mapped to chromosome 5 and the defect identified as a non-sense mutation in CD36 [20]. Although CD36 was already known to function as a scavenger receptor, these new results indicated that CD36 could act as a cell surface co-receptor for TLR-2 and TLR-6 for some ligands. Studies of the heedless ENU mutant revealed that the TLR-4 co-receptor CD14 is absolutely required for LPS-induced activation of the TRIF–TRAM pathway, but not for the MyD88-dependent pathway, indicating that TLR-4 signalling can be fine-tuned by CD14 [21]. The heedless mouse also displays a dampened response to TLR-2 and TLR-6 ligands, indicating that CD14 may modulate signalling from these receptors as well [21]. The 3d ENU mutant displays a diminished response of TLR-3, TLR-7 and TLR-9 to nucleic acids ligands which is caused by a mutation in the endoplasmic reticulum (ER) multi-spanning membrane protein Unc93b1 [22]. Unc93b1 interacts directly with TLR-3, -7 and -9, and the missense mutation found in 3d mice (H412R) abrogates this interaction [23]. Unc93b1 appears essential for the delivery of several TLRs from the ER into endosomes and this translocation does not occur in the 3d mutant [23]. 3d mice also have defects in exogenous antigen presentation to T cells [22], highlighting a potential link between TLR-3, -7 and -9 and antigen presentation, the details of which are not currently known. The important role of TLR pathways in human infections has now been demonstrated in studies of human susceptibility to infectious disease (reviewed in [24]).

Forward genetic analyses of inbred mouse strains also provided a first clear insight into the critical role of NLR proteins in recognition of PAMPs/DAMPs, and early activation of innate immunity in general. As opposed to human macrophages that readily support intracellular Legionella replication, mouse macrophages are generally restrictive. One exception is macrophages from the A/J mouse strain, which cannot contain Legionella replication. The A/J defect was mapped to the central portion of chromosome 13, a region of complex and repetitive DNA that harbours several complete and incomplete copies of the Neuronal inhibitor of apoptosis (Naip) gene [25–27]. Functional complementation studies in transgenic mice identified Naip5 as the gene affecting Legionella replication in A/J macrophages [28,29]. The realization that Naip proteins bore some resemblance to the newly discovered NLR proteins led to the proposal that Naip5 was involved in sensing and responding to Legionella products in macrophages [29–31]. This hypothesis was validated subsequently in studies which showed that Naip5 acts as a cytosolic sensor for Legionella flagellin following translocation of this protein into the cytosol by the Legionella Type IV secretion system [32,33]. Upon engagement of bacterial flagellin, Naip5 induces caspase-1 activation leading to IL-1β release and cell death, a protective response attenuated in A/J macrophages.

Forward genetics demonstrated the role of another NLR in susceptibility to bacterial toxin-mediated cell death. Anthrax lethal toxin (LeTx) is an important virulence factor of Bacillus anthracis, and is believed to cause death in systemic anthrax infections. Macrophages from inbred mouse strains show differences in susceptibility to LeTx-induced cell death. Genetic crosses from susceptible and resistant strains mapped a locus on mouse chromosome 11 containing 14 known genes, including the NOD-like receptor locus Nalp1. Reminiscent of the Naip locus, the region encompassing Nalp1 is complex and contains tandem gene repeats including three paralogues, Nalp1a, b and c. Nalp1b was found expressed in toxin-sensitive macrophages, and its causative role was confirmed by demonstration that transgenic expression of the allele from susceptible mice in resistant macrophages conferred toxin susceptibility [34]. LeTx triggers the formation of a membrane-bound inflammasome complex in macrophages leading to caspase 1-dependent cell death [35]. The nature of the PAMP or DAMP detected by Nalp1B is undetermined, but the fact that the protease activity of the toxin is required for cell death suggests that recognition of a toxin-cleaved host cell protein may be involved. In humans, NOD-like receptor pathways (in particular NOD1/NOD2) have been associated directly with susceptibility to inflammatory and infectious diseases, including Crohn's disease (reviewed in [36]) and mycobacterial infections [37,38].

In addition to pattern recognition receptors, host cells also detect infection through the recognition of non-self antigens presented in the context of major histocompatibilty complex (MHC) molecules. Presentation of viral antigens on MHC class I can initiate direct killing of virally infected cells by cytotoxic T lymphocytes and natural killer (NK) cells. The critical importance of the MHC class I–NK cell interaction in viral immunity has been demonstrated elegantly in the genetic analysis of inbred mouse strains innately resistant and susceptible to infection with murine cytomegalovirus (MCMV) (reviewed in [39]). Host response to MCMV in mice, expressed phenotypically as the extent of viral replication in spleen and liver, is regulated by both MHC (H2) and non-MHC linked genes. Strains bearing an MHC H2^k haplotype are resistant to infection, while those with H2^b and H2^d haplotypes are susceptible [40,41]. However, C57BL/6 mice are resistant to infection, despite a ‘susceptible’H2^b haplotype. Analyses of the genetic basis for this resistance pinpointed the Ly49 family of 14 highly related genes encoding NK cell receptors that recognize MHC class I on target cells [42,43]. Activating Ly49 receptors up-regulate NK cell-mediated killing of target cells, whereas inhibitory receptors down-regulate NK cell-mediated cell lysis. One Ly49 receptor, Ly49h, is present in C57BL/6 mice and absent in susceptible strains [44]. During MCMV infection, Ly49h interacts directly with m157, a viral protein that structurally mimics MHC class I, leading to an up-regulation of NK cell-mediated killing of virally infected cells [45]. A clue as to why the virus would encode a protein so deleterious to its survival in the host comes from another strain of mice (129), which encode an inhibitory Ly49 receptor that binds m157 leading to an inhibition of NK cell function [45]. The relationship between Ly49 receptors and MCMV is even more complex: resistance of MA/My mice to MCMV infection is controlled by the specific interaction of the Ly49p activating receptor with virally infected cells of H2^k haplotype, and PWK/Pas mice encode an as yet uncharacterized activating Ly49 receptor that specifically recognizes virally infected cells, although the molecular details of this receptor have not yet been characterized [46,47]. The key role of NK cells in response to MCMV infection has been corroborated in a recent ENU-screen, where a large number of MCMV-susceptible mouse strains were found to have defects in NK cell biogenesis or NK cell effector mechanisms such as degranulation [48]. Although humans do not encode Ly49 receptors per se, their functional counterparts in humans, the KIR receptors, have been shown to play an important role in viral defence (reviewed in [49]).

Amplification and integration of pathogen sensing signals

The engagement of innate pathogen sensing mechanisms results in the activation of signalling cascades and transcriptional pathways in corresponding cell types (macrophages, dendritic cells, NK cells). This signal amplification not only activates intrinsic microbicidal mechanisms of these cells, but also causes secretion of soluble mediators (inflammatory cytokines) which amplify the initial signal in situ, recruit additional cell types to the site of infection and can also dictate differentiation (polarization) of the immune response. For example, engagement of TLRs via PAMP recognition results in activation of NF-κB and interferon (IFN) regulatory factor (IRF) transcription factors, causing production of key cytokine IL-12p40, as well as type I and type II interferons; these in turn bind to receptors of macrophage and T cells, causing signal transducer and activator of transcription-1 (STAT1) activation, and IFN production thereby activating macrophage effector mechanisms while amplifying the initial signal. This early inflammatory/innate immune response [so-called T helper type 1 (Th1) polarization] has been shown to play a critical role in defences against certain bacterial infections, including mycobacterial infections such as tuberculosis. The role of this early response loop was established initially using mouse mutants bearing loss of function alleles in the genes involved and was validated subsequently in rare human patients suffering from Mendelian susceptibility to mycobacterial diseases (MSMD) and bearing mutations in the same genes (reviewed by [50]). Indeed, the genetic analysis of uniquely susceptible individuals, together with studies of human populations from endemic areas of tuberculosis, have validated the role of macrophages, dendritic cells and T cells, as well as IL-12, IFN-γ, tumour necrosis factor (TNF)-α and other cytokines in protection from tuberculosis (TB). In MSMD patients, Casanova and colleagues have identified mutations in IL-12p40 (IL12B) and its receptor (IL12RB1), the receptor for IFN-γ (IFNRG1; IFNRG2), the IFN-RG signalling partner STAT1 and the NF-κB regulator Iκκγ (NEMO) (reviewed in [50]). Positional cloning of the Myls mutation of BXH2 mice revealed that the transcription factor Irf8 is a master regulator of innate immunity. BXH2 mice are highly susceptible to infection with Mycobacterium bovis bacilli Calmette–Guérin (BCG), despite the presence of resistance at the Nramp1 host resistance locus. A mutation at position 294 (R294C) of Irf8 causes a complex phenotype in BXH2 which includes (a) susceptibility to a number of intracellular pathogens including M. tuberculosis, S. Typhimurium and Plasmodium chabaudi and (b) a myeloproliferative syndrome reminiscent of chronic myeloid leukaemia (CML) in humans [51–53]. Irf8 heterodimerizes with members of the IRF and ETS family members and, upon stimulation with IFN-γ, regulates expression of genes coding for effector function of macrophages, IL-12p40 and IFN-γ. Irf8 plays a critical role in maturation of the myeloid compartment, and Irf8 mutant mice have no CD8α⁺ and plasmacytoid dendritic cells [54].

Like BXH2, C3HeB/FeJ mice are highly susceptible to pulmonary TB following infection with M. tuberculosis. This susceptibility maps to the sst1 locus on chromosome 1 which was identified by positional cloning as the Ipr1 gene that codes for the Ifi75 protein (IFN-induced protein 75) [55]. Ipr1 is expressed in an infection-inducible manner in macrophages, but is not expressed in C3HeB/FeJ macrophages. Expression of an Ipr1 transgene in susceptible macrophages restricts multiplication of not only of M. tuberculosis, but also of Listeria monocytogenes[55]. The Ifi75 protein appears to switch the cell death pathway of infected macrophages from necrosis to apoptosis, contributing to restriction of the spread of M. tuberculosis infection [55]. Mutations in a human homologue of Ifi75, called SP110, are associated with immunodeficiency and hepatic veno-occlusive disease [56] and have also been associated with susceptibility to TB in some studies (reviewed in [56,57]), while other studies have failed to reproduce such associations [58,59].

Osteopontin, a secreted phosphoglycoprotein (Spp1) also known as early T cell activation protein 1 (Eta1), was identified originally as a bone-specific cDNA in rat osteosarcoma cells [60]. It was later detected as a positional candidate for a locus affecting susceptibility to Orientia tsutsugamushi, suggesting a role in inflammation and immunity [61]. Osteopontin appears to have pleiotropic effects in migration, activation and effector function in macrophages and dendritic cells, and has also been implicated in Th1 and Th17 polarization (reviewed in [62]). Osteopontin is secreted and can interact with integrins, but an intracellular form has also been described. The intracellular form of osteopontin has been detected in dendritic cells and macrophages which, upon interaction with MyD88, leads to activation of Irf7, IFN-α expression and polarization of Th1 and Th17 T cell responses (reviewed in [62]). Additional recent work has implicated osteopontin in response and pathogenesis of a number of bacterial and viral infections and inflammatory diseases [62–68]. Although the evidence is now strong for a role in infection and immunity, the candidacy of osteopontin as the gene responsible for Orientia tsutsugamushi susceptibility remains to be proved formally.

Effector functions

Amplification of the early detection signals emanating from pathogen sensing mechanisms results in activation of effector functions of macrophages, recruitment of cytotoxic T cells and priming of the adaptive immune responses. All these pathways have been detected in genetic studies of differential susceptibility of inbred mouse strains to different types of infections. A few examples are reviewed below.

Engulfment of pathogens by professional phagocytes is a primary line of defence against invading microbes. Phagosomes undergo a maturation process fusing with endosomes and lysosomes which deliver microbicidal molecules leading ultimately to destruction of the internalized microbe. However, many intracellular pathogens have evolved mechanisms to avoid the cytotoxic environment of the phagolysosome, such as escape from phagosomes or redirection of membrane traffic to modulate the fusogenic properties of the phagosome. Nramp1 is one of the first ‘host resistance’ genes identified through positional cloning, and underlies susceptibility of mice to infection with the unrelated intracellular pathogens S. Typhimurium, M. bovis BCG and Leishmania donovani[69]. Susceptible mouse strains bear a loss of function G169D allele at Nramp1 leading to a complete absence of mature Nramp1 protein in macrophages [70]. In macrophages and neutrophils, Nramp1 is localized to the membrane of late endosomes/lysosomes and is recruited rapidly to the membrane of microbe-containing phagosomes [71]. Nramp1 acts to remove divalent cations from the phagosomal space, starving microbes for these essential ions [72,73]. Studies of pathogen-containing phagosomes from resistant and susceptible mice indicate that Nramp antagonizes the ability of pathogens to modulate phagosome maturation, resulting in increased fusion with lysosomes in cells with functional Nramp1 [74–76]. Recent studies with Leishmania parasites showed that Nramp1-mediated iron depletion from the phagosomal space results in activation of high-affinity metal acquisition systems of the internalized parasites, identifying iron as a key molecule at the interface of host–pathogen interaction [77].

As described above, the NLR protein Naip5 is critical to restrict Legionella replication in macrophages. Although the primary effect of Naip5 on inflammasome-dependent induction of caspase-1 activation and induction of pyroptosis has been well established, Fortier et al. demonstrated that Naip5 also influences trafficking of the Legionella-containing vacuole [78]. Naip5 promotes the rapid fusion of Legionella-containing vacuoles with the lysosomes, whereas in the absence of Naip5, the bacteria co-opt membrane traffic from the ER to build a replicative niche that avoids lysosomal fusion [78]. In addition, PAMP sensing by Naip5 also results in activation of Irf-dependent transcription. For example, phagocytosis of Legionella by macrophages causes rapid induction of Irf1 mRNA expression. In addition, macrophages from Irf-1 or Irf8 mutant mice are permissive to Legionella replication, and macrophages from Naip5^+/–:Irf8^+/– double heterozygotes are also susceptible, indicating a strong genetic interaction between these two genes [79].

The intracellular pathogen Chlamydia trachomatis resides within membrane-bound vacuoles termed inclusions that are formed by an active process requiring the bacterially driven interception of host Golgi-derived vesicles [80]. Resistant C57BL/6J mice are able to control growth of intracellular Chlamydia in a process that is dependent on IFN-γ, whereas susceptible C3H/HeJ mice cannot [81]. Genetic analyses pinpointed the C3H/HeJ defect to a DNA segment harbouring two members of the IFN-γ -inducible p47 GTPase family, Igtp and Irgb10[82]. A defect leading to reduced Irgb10 expression was identified as the probable cause of C3H/HeJ susceptibility [82]. p47 GTPases comprise a family of 21–23 proteins in mice, and their expression in phagocytic cells is induced by IFN. p47 GTPases associate with and promote the interaction of pathogen-containing phagosomes or inclusions with degradative compartments such as lysosomes. The mechanism underlying promotion of fusion to lysosomes is not understood fully, but may vary depending on the p47 family member and the pathogen involved. Notably, the p47 family is much smaller in humans and the human genes are not IFN-γ-inducible. However, a recent study indicates that the human guanylate-binding protein (GBP) family may serve an analogous function. These proteins localize to the Chalmydial inclusion in human cells where they are involved in IFN-dependent chlamydial growth restriction [83].

In mice, susceptibility to infection with the malarial parasite P. chabaudi is regulated by a number of Char (Chabaudi resistance) loci [84]. The Char9 locus affects the extent of blood-stage replication of the parasite, and was identified by positional cloning as a loss-of-function haplotype at the Vnn1/Vnn3 locus [85]. Vnn genes encode metabolic pantetheinases that convert pantetheine into pantothenic acid (vitamin B5) and cysteamine. Direct administration of cysteamine to pantetheinase-deficient mice was shown to restore resistance to infection with P. chabaudi, as measured by extent of blood parasitaemia and survival from acute infection, identifying cysteamine as a natural aminothiol with robust anti-malarial properties [85]. In addition, cysteamine is used currently in the clinic for the life-long treatment of the rare kidney disorder nephropatic cystinosis. In recent preclinical studies, daily doses used in the treatment of nephropatic cystinosis were shown to (a) have anti-malarial activity and (b) strongly potentiate the anti-malarial activity of artemisinin, the standard of care recommended by the World Health Organization for the clinical treatment of malaria [86,87].

Genetic studies in mice also led to the discovery of important effector mechanisms of the IFN-dependent anti-viral response. While most mouse strains are susceptible to infection with mouse-adapted influenza virus, the A2G strain is resistant by virtue of a single autosomal dominant gene [88]. Proteomic analysis of IFN-treated cells from resistant and susceptible mice uncovered a 72 kDa protein present exclusively in IFN-stimulated resistant cells, Mx1 (myxovirus resistance 1) [89]. The gene encoding this protein was subsequently cloned and shown to be deleted or mutated in influenza-susceptible mouse strains [90]. Mx1 is a nuclear dynamin-like GTPase that inhibits primary viral transcription, by a mechanism that is still not characterized completely. A second Mx gene in mice, Mx2, is a cytoplasmic protein that inhibits the replication of vesicular stomatitis virus, suggesting that the anti-viral role of these proteins may be more generalized [91]. Both Mx genes are conserved in humans, where Mx1 has been shown to have broad anti-viral activity [92].

Another important type I IFN-dependent anti-viral defence mechanism was identified in genetic studies of flavivirus resistance in mice. 2′,5′-Oligoadenylate cyclase 1b (Oas1b) confers resistance to infection with flaviviruses (West Nile virus and Murray Valley encephalitis virus) in mouse strains with an intact copy of the gene [93,94]. Mouse strains that are susceptible to these infections have a premature stop codon, leading to an absence of mature protein [93,94]. Oas1b is up-regulated by type I IFNs and when the protein is activated by dsRNA it produces 2′,5′-oligoadenylates which activate RNase L leading to the degradation of viral and cellular RNAs. There is some evidence that an Rnase L-independent pathway may be involved in Oas1b-mediated viral resistance, the details of which have not been characterized [95]. In humans, differential 2′,5′-oligoadenylate synthetase (OAS) activity has been correlated with differential response to yellow fever vaccination [96], and genetic polymorphisms in the OAS gene have been shown to correlate with differential enzyme activity [97], providing evidence that polymorphisms in OAS are likely to impact response to viral immunization or infection in humans.

Conclusion and future perspectives

During the past 20 years, the study of inbred mouse strains by a forward genetic approach has shed considerable light on some of the most critical pathways involved in host–pathogen interactions, including pathogen sensing and early response to infection. Although successful positional cloning has been limited largely to monogenic traits, a few examples of gene identification in polygenic controls have been reported. New insights may be forthcoming as these tools evolve and are applied to other infection models. Furthermore, large-scale programmes in ENU mutagenesis of the mouse germline are certain to generate a wealth of novel genotypes and associated phenotypes that will continue to foster important discoveries on the proteins and pathways involved in innate response to infections.

Disclosure

The authors have no conflicts of interest to declare.