Dissecting innate immunity by germline mutagenesis

Abstract

The innate arm of our immune system is the first line of defence against infections. In addition, it is believed to drive adaptive immune responses, which help fight pathogens and provide long-term memory. As such, the innate immune system is instrumental for protection against pathogens that would otherwise destroy their host. Although our understanding of the innate immune components involved in pathogen sensing and fighting is improving, it is still limited. This is particularly exemplified by increased documentation of innate immune deficiencies in humans that often result in high and recurrent susceptibility to infections or even death, without the genetic cause being evident. To provide further insight into the mechanisms by which pathogen sensing and eradication occur, several strategies can be used. The current review focuses on the forward genetic approaches that have been used to dissect innate immunity in the fruit fly and the mouse. For both animal models, forward genetics has been instrumental in the deciphering of innate immunity and has greatly improved our understanding of how we respond to invading pathogens.

Introduction

The innate immune system (IIS) is of key importance for the early recognition and removal of a wide range of infectious agents. In insects, which are devoid of lymphocyte-based adaptive immunity, innate immunity is the sole system responsible for the eradication of pathogens, allowing insects to live long enough to ensure the survival of their species. Drosophila's immune system relies on two main arms: a humoral response (namely production of antimicrobial peptides and melanization) and a cellular response (i.e. phagocytosis of microorganisms and encapsulation of parasites) (Fig. 1a).¹

Figure 1.

Innate immune responses. (a) Drosophila develops humoral and cellular innate mechanisms of defence.¹ Upon infection, cells of the fat body (equivalent of the mammalian liver) produce antimicrobial peptides that are secreted in the haemolymph where they directly target microorganisms. The cellular response relies on plasmatocytes which phagocytose, coagulate and, together with lamellocytes, encapsulate foreign cells. In addition, encapsulation requires melanin secreted by cristal cells. (b) In vertebrates, the immune response is more complex. Innate mechanisms rely mainly on cells of the myeloid lineage (macrophages, neutrophils, dendritic cells) and natural killer (NK) cells that directly address the infection (phagocytosis, cytolysis, opsonization, reactive oxygen species (ROS) production). In addition, these cells produce proinflammatory cytokines and chemokines, which stimulate and direct the adaptive immune response (memory mainly relying on lymphocytes), preparing the host for a second encounter with the pathogen.²

In vertebrates (Fig. 1b), in addition to sensing and killing pathogens, the IIS is also believed to drive adaptive immune responses (AIR) and, to a certain extent, influence their nature (e.g. tolerance versus T helper type 1/type 2 activation).² The IIS acts rapidly; partly because innate cells are present at sites of potential encounter with pathogens (skin, lungs, mucosa), and partly because they constitutively express germline-encoded cytosolic or cell surface receptors that recognize conserved microbial molecules.^3,4 Unlike receptors of the adaptive immune system (AIS), these receptors do not require random rearrangement of gene segments to be functional. Once activated, cells of the IIS release proinflammatory molecules to recruit and activate the AIS. In addition, antigen-processing cells phagocytose and process pathogens, subsequently maturing and migrating to lymphoid organs, where they present foreign peptides to cells of the AIS, enhancing their activity and specificity.⁵

Although the essential role of the IIS in our survival remains often unnoticed (microbes are usually eradicated before disease symptoms are apparent), its importance is revealed by patients who have genetic aberrations in IIS genes that result in altered responses to pathogens. Currently, more than 120 Mendelian traits that confer predisposition or resistance to specific infections have been described at the molecular level in humans, and 38 of them exclusively affect cells and molecules of the IIS.⁶ There is, however, a clear discrepancy between immune-related diseases observed in humans and the underlying genetic deficiencies that cause them. To close this gap, a number of genetic approaches relying on the production of mutants with abnormal immune responses have been successfully used to identify genes essential for our IIS. As a result of the complexity of the immune system, these genetic approaches are best used in vivo, where all molecular and cellular interactions occur under physiologically relevant conditions.

Genetic strategies to assess gene function

Both reverse and forward genetic approaches have been successfully applied to reveal gene function in vivo (Fig. 2). Reverse genetics starts with a defined genetic change (usually the deletion of a gene by homologous recombination),⁷ followed by the quest for an eventual phenotype. As such, this approach is entirely hypothesis driven and the investigated gene is presumed to play a role in the phenomenon of interest. The success of this approach is therefore highly dependent on a correct guess of the alleged function and on the gene being dispensable for the organism’s development. Reverse genetics can be a relatively fast strategy, providing the instant knowledge of the molecular defect at the origin of the phenotype, but it presents the risk of deleting regulatory elements of unrelated genes that are located within the targeted sequence and of over-interpretation of phenotypes that are too small to be otherwise significant. In addition, one should be careful with the interpretations of phenotypes created in genetically mixed backgrounds. The use of conditional deletions⁸ (subtraction of the gene only in a population of cells, an organ or at a specific time-point) permits the study of otherwise essential genes without impairing the survival of the organism. Data from conditional deletions should, however, be considered carefully because they might only reflect part of the protein’s function. Reverse genetics will not be discussed further here, instead we will focus on forward genetics and how it has contributed to our understanding of innate immune functions in the fruit fly (Drosophila melanogaster) and the mouse (Mus musculus).

Figure 2.

Reverse and forward genetics are complementary strategies used to identify gene function. Reverse genetics starts with a known gene of interest. This gene is deleted based on the hypothesis that it plays a role in the phenomenon of interest. The mutant animals, which might not have any phenotypic defect, are then analysed. Forward genetics starts with alterations of the phenomenon of interest (mutant phenotypes) and walks back to the culpable mutation and gene. The forward approach does not make any assumption on gene identity or function.

Forward genetics

Forward genetics is driven by phenotypic observations.⁹ It is based on the existence of phenotypic differences that are either spontaneous or, for instance, introduced by random genomic mutations in germline cells. Forward genetics does not require any educated guess on the functions of genes, but instead monitors phenotypic deviances and aims to isolate the culpable mutation that causes them. This approach therefore requires the phenotype to be sufficiently strong and reproducible to distinguish between a wild-type and mutant animal.¹⁰ To date, mutagenesis is the most powerful way to identify yet unknown genes’ functions, to find new functions for known genes and to create allelic series that permit a fine-tuned analysis of gene function. A forward genetic approach using germline mutagenesis proceeds in three steps.

Mutagenesis

Multiple random mutations are created by use of mutagenic alkylating agents: ethylmethylsulphonate (EMS) for Drosophila¹¹ and ethylnitrosourea (ENU) for mice.¹² Both induce mainly point mutations^13,14 which can be recovered as heterozygote or homozygote traits using different breeding strategies.

In Drosophila, most immune screens require the infected animals to be killed to assess their phenotype. Consequently, mutant micropedigrees in which all animals originate from the same single heterozygote G1 father have to be established, and only homozygote G3 animals are screened (Fig. 3a). Several genetic tools (e.g. balancer chromosomes, which prevent recombination of visible genetic markers associated with the mutation, temperature-sensitive genes, etc.) are used to facilitate the breeding and the recovery of the mutations and to allow the selective screening of one of the four chromosomes of Drosophila at a time.

Figure 3.

Mutagenesis and breeding schemes. (a) Male flies are fed ethylmethylsulphonate (EMS) to generate random mutations in the germline cells. They are bred with wild-type (wt) females that carry a chosen balancer chromosome and genetic markers. Heterozygous G1 males are individually bred with wild-type females to establish independent micropedigrees. G2 animals are intercrossed to produce a G3 progeny composed of homozygous and heterozygous flies. Homozygotes are screened for mutations of interest, while heterozygotes for the mutagenized chromosome over balancer are used to establish the mutant strain. (b) Male mice are injected with ethylnitrosourea (ENU) and bred with wild-type females. Dominant mutations can be directly screened in the G1 progeny. Recessive mutations can only be identified in homozygous animals, thus requiring the generation of G3 mice either via backcross (left panel) or intercross (right panel) breeding. The latter strategy allows recovery of X-linked mutations.

In mice, for genome-wide mutagenesis, ENU-treated males are bred to wild-type females to produce a heterozygote G1 progeny that can be screened directly for dominant mutations, or can be used to generate G3 mice carrying homozygote recessive mutations (Fig. 3b). To this end, two common breeding strategies can be used: G2 daughters are backcrossed to their G1 father, or G1 and then G2 mutants are intercrossed, permitting the recovery of X-linked mutations.⁹ The standard treatment of male mice with ENU¹⁵ induces approximately one mutation per million base pairs (bp) of the 2600 × 10⁶ bp genome.¹⁶ Only 1, 6% of the genome has coding function¹⁷ and only 76% of random base pair changes within coding regions cause a change in the relevant protein, so each G1 animal will carry approximately 30 changes affecting coding regions, and each G3 animal will have three or four homozygous mutations. The mutation recovery rate will depend on the breeding scheme and on the number of mice screened per micropedigree.¹⁶

Screen

Animals carrying random mutations are monitored for a phenotype of interest, the underlying concept being that any mutation affecting a gene involved in the phenomenon will alter the observed phenotype. For example, a screen could assess the induction of antimicrobial peptide genes following an infection in the adult fly,¹⁸ or the in vivo resistance to viral infections in the mouse.¹⁹ The design of the screen dictates the success of a mutagenesis programme;¹⁰ the most crucial factor being the phenotypic difference between the wild-type and the mutant animals, which has to be significant enough to avoid false positives, establish a stable mutant strain and allow positional cloning.

Positional cloning

In Drosophila, two complementary approaches are used to map point mutations: by calculating meiotic recombination frequencies between the mutation and visible genetic markers (like eye or body colour, eye or wing shape, etc.), and/or by performing complementation tests between the mutation and chromosomal deletions positioned along the chromosome. In mice, mutant animals are outcrossed to a philogenetically divergent inbred strain (mapping strain), which provides heterogeneity for known markers spread all over the genome. The hybrid heterozygous animals are then either backcrossed to the mutant strain or intercrossed, a strategy that doubles the number of cross-over events.⁹ Meiotic recombination frequencies between the mutation and the markers are obtained, and a logarithm of odds (LOD) score is calculated, defining a chromosomal region of interest from which candidate genes are sequenced.⁹ Positional cloning can be strenuous and is often the bottleneck of a mutagenesis programme, but with the mouse genomic sequence readily available to the public and the continuous improvement of sequencing technology, soon this step will no longer be limiting.

Large-scale mutagenesis and the analysis of innate immunity in Drosophila

In the fruit fly, the use of forward genetic mutagenesis to decipher the IIS started in the late 1990s.^18,20 This strategy was inspired by the work of C. Nusslein-Volhard, E. Wieschaus and their colleagues, who identified the mechanisms of dorsoventral patterning in embryos, and for the first time demonstrated the feasibility of a large-scale EMS mutagenesis in Drosophila in vivo.²¹ Similar screens on maternal effect genes established that the nuclear translocation of the nuclear factor-κB (NF-κB) homologue Dorsal during patterning is controlled by the Toll pathway.²² In 1996, having demonstrated that the promoters of the genes coding for the antimicrobial peptides contained essential NF-κB response elements,^23–25in vivo infections of mutants of the Toll pathway established a role for Toll, spaetzle, tube and pelle in the antifungal IIR of the adult fly.²⁶ Interestingly, Dorsal and the cascade of proteases upstream of Spaetzle required during embryonic development were dispensable during the adult immune response. As a consequence, there were still many questions left unanswered. How does pathogen sensing occur? What activates Spaetzle? What is the transcription factor of the Toll pathway in adults? What is the identity of all the genes in the imd pathway? These initial successes encouraged the use of forward genetic approaches to decipher immunity and this resulted in the identification of a large number of the genes involved in both pathways controlling the humoral immune response of the fly (Fig. 4).¹ Besides the genetic dissection of gene function in Drosophila by classical EMS mutagenesis,^{18,20,27–33} the screening of libraries containing transposable elements (piggyBag transposons, P-elements) significantly contributed to the field.^34–38 P-elements can be mobilized in a transposase-dependent manner³⁹ to create new deletions and insertions (http://engels.genetics.wisc.edu/Pelements/index). Mutants can subsequently be analysed for mutations in genes of interest (reverse genetics) or can be used in phenotypic screens as random mutant strains (forward genetics). The combined screening of EMS and P-element mutants has been instrumental in the elucidation of pathogen sensing in Drosophila which, unlike vertebrates, does not depend on Toll receptors.

Figure 4.

Several strategies used to unravel the signalling pathways involved in the humoral immune response of Drosophila. Most mutagenesis screens have addressed the humoral immune response to bacterial and fungal infections. Ethylmethylsulphonate (EMS) and P-elements mutagenesis, sequence homology and RNA interference studies have permitted the discovery of the majority of genes in both the Toll and the imd pathways. References and abbreviations: Necrotic (NEC),²⁹ Persephone (PSH),²⁸ Gram-negative binding protein-3 (GNBP3),³⁴ GNBP1,³⁵ Peptidoglycan recognition protein-SA (PGRP-SA),²⁷ PGRP-SD,³⁶ Grass, Sphinx1/2, Spirit, Spheroid,⁹⁵ Spaetzle processing enzyme (SPE),¹⁰⁰ Spaetzle (SPZ), Toll, Tube, Pelle, Cactus,^26,101 Myeloid differentiation factor-88 (MyD88),¹⁰² Dorsal-related immunity factor (DIF),^18,103 dorsal,¹⁰⁴ PGRP-LE/PGRP-LC,^37,94,105 immune deficiency (IMD),¹⁰⁶ FADD,^107,108 DREDD,¹⁰⁹ inhibitor of apoptosis protein-2 (IAP2),³⁸ transforming growth factor-β activated kinase-1 (TAK1),³³ TAK1 binding protein-2 (TAB2),¹¹⁰ IκB kinase-β (IKKβ),³² IKKγ,³¹ janus kinase/janus kinase kinase (JNK/JNKK),¹¹¹ ubiquitin conjugating enzymes (Ubc13 and Uev1a),¹¹² Relish¹¹³.

PGRPs, GNBPs and pathogen sensing in Drosophila

In Drosophila, the mechanisms of pathogen sensing remained elusive until the isolation of semmelweis (seml) by EMS mutagenesis.²⁷seml reduces the expression of Toll-induced antimicrobial peptides in response to some Gram-positive bacteria (GPB), but not Gram-negative bacteria (GNB) or fungal infections.²⁷ The mutation affects peptidoglycan recognition protein-SA (PGRP-SA), a gene coding for an extracellular circulating protein that binds peptidoglycan from GPB Micrococcus luteus.^40–42seml was the first pattern recognition molecule found to be essential for Drosophila resistance to in vivo GPB infection. Shortly after its isolation, a mutation inducing a similar phenotype, osiris (osi), was isolated in a semi-random P-element screen.³⁵ The P-element was located in a gene called gram-negative binding protein-1 (GNBP1), encoding a protein originally described in vitro as binding GNB lipopolysaccharide and fungal β-(1,3)-glucan, but not GPB peptidoglycan.⁴³ The susceptibility of osi mutant flies to GPB, but not GNB, was therefore quite surprising and established that in vivo, GNBP1 participates in the immune signalling against GPB only³⁵ by processing GPB lysine-type peptidoglycan for recognition by PGRP-SA.⁴⁴ Finally, the function of other members of the PGRP family of genes was tested. For one of them, PGRP-SD, the targeted excision of a closely located P-element created a deletion of the entire coding region and phenotypic analysis revealed that it mediates Toll activation by a specific subset of GPB, in partial redundancy with the PGRP-SA/GNBP1 complex.³⁶

Forward genetic analysis of innate immunity in the mouse

In mice, large-scale mutagenesis programmes started in the early 1990s^45–54 focusing on the analysis of a wide variety of phenotypes including innate immunity.⁵⁴ The innate sensing of pathogens involves specific receptors including Toll-like receptors (TLRs). Their function in pathogen recognition was first established for the lipopolysaccharide receptor TLR4 by classical positional cloning of a spontaneous phenotype: the lack of tumour necrosis factor-α (TNF-α) production in lipopolysaccharide-stimulated macrophages from C3H/HeJ mice⁵⁵, opening other members of the TLR family to investigation by gene targeting. Shortly hereafter, additional genes involved in TLR signalling were found by structure and sequence homology searches and/or gene targeting. In terms of in vivo resistance to pathogenic infections, for example to murine cytomegalovirus (MCMV) infections, classical genetic approaches had isolated some of the sensing molecules involved and had established the importance of specific cells of the IIS, but major questions remained unanswered.⁵⁶ Forward genetics therefore provided a new way of interrogating the IIS on a genome-wide basis without any prior knowledge of gene sequence, cellular expression profile or suspected function. This approach allowed the identification of new genes (Lps2: a mutation in Trif,⁵⁷3d: a mutation in Unc93b1⁵⁸) and the discovery of new functions for known genes (Oblivious: a mutation in CD36⁵⁹). It helped to define gene function (the Heedless mutation in CD14,⁶⁰CpG1 in TLR9⁶¹ or PanR1 in TNFα⁶²), to refine mechanisms of receptor/adaptor interactions (Insouciant in TLR6, Lackadaisical and Pococurante representing mutations in MyD88⁶³) and to establish new mouse models for human diseases (i.e. the Jinx mutation in Unc13d⁶⁴).

Identification of Unc93b1 as a key component of innate and adaptive immunity

Although the era of discovering ‘novel’ genes has virtually come to an end in the sense that all genes have been sequenced and described to some extent, a key quest for the near future is to assign gene function. The identification of the 3d mutation is a prime example of how forward genetics can reveal gene functions that would otherwise have been hard to predict. The 3d mutation revealed a phenotype in which macrophages and/or dendritic cells were unresponsive to TLR3, TLR7 and TLR9.⁵⁸ These TLRs are expressed in endoplasmic reticulum and endosomes^65–70 and are the main TLRs involved in the sensing of viruses via double-stranded RNA (TLR3),⁷¹ single-stranded RNA (TLR7)^72–74 and/or unmethylated DNA (TLR9).⁷⁵ Consistent with a reduced viral sensing, 3d mutant mice were found to be highly susceptible to MCMV infections.⁵⁸ In addition, the mutation interfered with exogenous antigen presentation,⁵⁸ a process independent from TLR signalling.⁷⁶ Positional cloning revealed that the 3d mutation resides in Unc93b1, a gene originally identified in Caenorhabditis elegans.⁷⁷ The prototypic protein of the Unc93 family in C. elegans is part of a tripartite potassium channel consisting of sup9, sup10 and unc-93, and targeted knockouts for each sub-component revealed an uncoordinated body movement phenotype.^78,79 In mice, Unc93b seems to directly interact with TLR3, -7 and -9 transmembrane domains⁸⁰ where its potential function (e.g. as a coreceptor, trafficking partner or docking platform for adapter molecules) has still to be elucidated. The 3d mutation was shown to abolish these interactions, correlating with the TLR phenotypes observed in vivo, but it does not seem to affect the maturation or stability of major histocompatibility complex–peptide complexes, which is a separate function in need of further investigation.

At the same time, the group of J. L. Casanova identified two human patients who showed a remarkable similarity with the phenotype observed in 3d mutant mice. The patients developed recurrent herpes simplex virus encephalitis in response to herpes simplex virus infections and showed impaired TLR7, TLR8 and TLR9 signalling.⁸¹Unc93b1 was sequenced in both patients⁸¹ and aberrant mutations were found to be causative for the disease. These findings present the first monogenic cause of herpes simplex virus encephalitis in humans and helped in our understanding of the underlying cause of observed immune deficiencies in these patients. They may ultimately lead to improved treatment, for instance via recombinant interferon-α therapy.

In conclusion, as illustrated by the 3d mutation, random mutagenesis combined with the analysis of a poorly defined biological phenomenon, is an efficient means of assigning sometimes unexpected gene functions.

Jinx/Unc13d: a new model for human haemophagocytic lymphohistiocytosis

Forward genetics is most effective when applied to a broad genetic footprint that is poorly defined. For instance, it is likely that we require a considerable number of genes to resist lethal infection upon viral exposure. This has been previously referred to as the ‘resistome’ and was probed by testing mutagenized mice for their resistance to MCMV infections.⁵⁶ This has led to the identification of several mutants,¹⁹ including Jinx.⁶⁴

When infected with a sublethal dose of MCMV, Jinx mutant mice are unable to contain viral replication and die within 6 days post-infection.⁶⁴ Further characterization of the Jinx phenotype revealed a failure of natural killer cells and cytotoxic T lymphocytes (CTL) to degranulate, whereas cytokine responses appeared normal, indicating that the sensing of virus was unaffected. The mutation was genetically mapped to Unc13d, the orthologue of the human MUNC13-4 gene. Mutations in MUNC13-4 cause type three familial haemophagocytic lymphohistiocytosis (HLH),⁸² a lethal innate immune syndrome with defects in natural killer cell and CTL degranulation, severe cytopenias, massive hepatosplenomegaly with overwhelming proliferation of CTL and macrophages, and sustained cytokine production.^82,83 In humans, the pathogenesis of HLH is not well established. Viral (Epstein–Barr virus, lymphocytic choriomeningitis virus) or intracellular bacterial infections have been suggested to be at the origin of the disease,⁸⁴ but no real aetiology has been defined to date. Interestingly, the complete HLH phenotype (e.g. hyperactivation of CTL and macrophages and sustained cytokine production correlating with an inability to control the infection) is conditional in Jinx mice, developing only with lymphocytic choriomeningitis virus, but not with MCMV or Listeria monocytogenes.⁶⁴ As a consequence, Jinx mutant mice are a good model to probe the aetiology of HLH in humans, potentially providing further insight, which may lead to better treatments.

Concluding remarks

To date, most screens in Drosophila have focussed on the humoral part of the IIS. This is most probably because of the homology of the signalling cascades Toll and imd with their mammalian counterparts TLRs and TNF-α.⁸⁵ Mutagenesis programmes have successfully participated in unravelling the molecular pathways regulating the expression of antimicrobial peptides, with contributions made both at the intra- and extracellular levels.¹ Mechanisms of cellular immunity, however, are still largely unknown, providing tempting forward genetics opportunities.

Drosophila has already been a proven model for the study of conserved features of the IIS related to pathogen sensing in human diseases (for example Candida albicans,⁸⁶Mycobacterium marinum,⁸⁷L. monocytogenes^88,89 and Plasmodium gallinaceum⁹⁰). Currently, new approaches are being developed using RNA interference (RNAi) to target genes ex vivo^89,91–94 and in vivo,^95,96 ultimately resulting in a genome-wide library of transgenic Drosophila lines targeting more than 12 000 genes.⁹⁶ In addition to the existing libraries, these offer exciting opportunities for the scientific community to further probe gene function in the fly.

In mammals, it is clear that the functions of many genes remain poorly defined. A forward genetic approach using ENU mutagenesis may greatly help to fill in (part of) the phenotype gap and with the ongoing efforts to generate one mutant for each gene of the mouse genome^97,98 this will soon be reality. In addition, great efforts should be taken to expand the number of standardized phenotypic screens to further annotate the functional mouse genome.⁹⁹ These developments should make for an exciting future and, as is well illustrated by the 3d mutation, sometimes uncovering gene function is best achieved by following an unbiased non-hypothesis-driven approach.