Biology of the Immune System

Home > Research > Research Areas > Biology of the Immune System

As our defense against the microbial world, the immune system must be capable of detecting and eliminating from the body any pathogen that breaches the epithelium’s physical barrier. Our studies aim to identify the genes necessary for both detection and elimination of microbes, and for preventing aberrant activation of the immune system in the absence of infection. Already hundreds of genes that affect immune function have been identified. We work to decipher the molecular events that mediate these processes.

Immune Deficiencies

Immune deficiency disorders prevent the immune system from defending the body against infections (such as those caused by bacteria and viruses) and other diseases (such as cancer). In contrast to acquired immune deficiencies, congenital or primary immune deficiencies are caused by heritable mutations in genes important for immune system function, and are therefore present from birth. Several phenotypic screens carried out in the Center for the Genetics of Host Defense are designed to identify primary immune deficiencies, which may affect either the innate or adaptive branches of the immune system.

Innate Immune Deficiencies

From the moment of exposure to a microbial pathogen, the innate immune system launches immediate defensive responses, which include inflammation, engulfment of pathogens, and recruitment of a variety of immune cells. These tactics, collectively termed the innate immune response, are employed against all classes of microbe irrespective of prior exposure, and are the vanguard of the full immune response. In the absence of the innate immune response, mice and humans are highly susceptible to all types of infection.

Approximately 25 known cell surface and cytoplasmic receptors are dedicated to sensing all microbes and initiating the innate immune response in humans and mice. These include the Toll-like receptors (TLR), NOD-like receptors (NLR), C-type lectin receptors (CLR), and the RIG-I-like receptors (RLR), which are highly expressed in innate immune cells such as macrophages and dendritic cells, where each receptor detects its specific microbial ligand(s) and triggers signaling leading to transcriptional activation of thousands of genes that coordinate an inflammatory response (Figure 1). Principal among these are genes encoding inflammatory cytokines such as tumor necrosis factor (TNF), interleukin (IL)-6, IL-12, the type I interferons, and IL-1β. To identify new genes necessary for innate immune responses, the Center for the Genetics of Host Defense performs phenotypic screens for impaired innate immune responses when mutagenized mice or cells derived from them are challenged with authentic pathogens such as viruses, or with known innate immune receptor ligands such as lipopolysaccharide (a TLR4 ligand found on the surface of gram-negative bacteria) and poly(I:C) (a TLR3 ligand that mimics double stranded RNA found in viruses). The mutant mouse phenotypes identified in these screens may define innate immune deficiency syndromes that are also present in humans. This occurred with the 3d phenotype (see Unc93b1), providing a molecular explanation for a primary immune deficiency with previously unknown etiology.

Figure 1. The TLR signaling pathways and ENU-induced mutations (red) identified in phenotypic screening for innate immune deficiencies. Anchored in the cell or endosome membrane, each TLR binds a distinct microbial ligand (see Table 1), activating intracellular signaling leading to the transcription of genes necessary for the innate immune response. TLRs function as homo- or heterodimers. In their active conformation, they recruit one or more of four adapter proteins (MyD88, TICAM1, TRAM, TIRAP). The adapter proteins further assemble multiple proteins into signaling complexes that activate transcription factors including NF-κB, CREB, AP-1, IRF7, and IRF3. Abbreviations: VSV-G, vesicular stomatitis virus glycoprotein G; LTA, lipoteichoic acid; LP2, lipopeptide 2; ssRNA, single stranded RNA; dsRNA, double stranded RNA. Pam3CSK4 is a synthetic triacylated lipopeptide. Phosphorylation events are represented by small yellow circles labeled with a “P”. This image is interactive. Click on the mutations in red to view them on Mutagenetix.

As key sensors of invading microorganisms including bacteria, fungi, protozoa and viruses, TLRs play an essential role in the innate immune response.  There are 12 TLRs in mice, and 10 in humans, and each receptor recognizes distinct microbial ligand(s) (Table 1).  TLR activation ultimately leads to the induction of thousands of genes necessary for the innate immune response.  Critical among them are genes encoding TNF, induction of which depends on NF-κB and MAP kinase signaling, and type I IFN, an antiviral cytokine that depends on activation of IRF3 and IRF7.  Humans carrying loss of function mutations in one or more TLRs display elevated susceptibility to a variety of infections.  Aberrant TLR signaling may promote autoimmune diseases such as systemic lupus erythematosus and autoimmune arthritis.

Table 1. Location and Ligands of TLRs in human and mice.

To identify new regulators of the TLR signaling pathways leading to TNF production, we stimulate peritoneal macrophages from mice carrying ENU-induced mutations with TLR ligands.  The 3d phenotype was characterized by impaired TNF and type I interferon (IFN) production by mutant macrophages in response to ligands for the nucleic acid-sensing TLRs (TLR3, TLR7, and TLR9; important for the detection of viruses), and by increased susceptibility of mutant mice to infections with the herpesvirus mouse cytomegalovirus and with Staphylococcus aureus.  The recessive 3d phenotype was ascribed to a mutation of Unc93b1, a gene encoding a 12 transmembrane spanning protein found to mediate translocation of TLR3, TLR7, and TLR9 from the endoplasmic reticulum to endolysosomes (Figure 2) (Brinkmann et al. J.Cell Biol. 177, 265-275; Kim et al. Nature 452, 234-238). Strikingly, mutations of UNC93B1 were found to underlie herpes simplex virus-1 (HSV-1) encephalitis in two human individuals that were otherwise healthy, with normal immunity to nine other viruses (Casrouge et al. Science. 314, 308-312).  Similar to mouse macrophages homozygous for the 3d mutation, peripheral blood mononuclear cells from these patients produced reduced amounts of type I IFN and TNF in response to ligands for nucleic acid-sensing TLRs (except TLR3), and to HSV-1.  UNC93B1 deficiency defined a new human primary immunodeficiency in which susceptibility to HSV-1 encephalitis is increased, despite normal immunity to most pathogens.

The discovery of the 3d phenotype and causative Unc93b1 mutation are reported in (Tabeta et al. Nat.Immunol. 7, 156-164).

Figure 2. Unc93b1 is necessary for TLR3, TLR7, and TLR9 trafficking from the endoplasmic reticulum (ER) to their resident location in the endosome. UNC93B1 facilitates loading of these TLRs into COPII vesicles, and traffics with them through post-Golgi sorting steps, which are different for TLR9 versus TLR7 (Lee et al. Elife. 2, e00291). Other proteins that are necessary for TLR maturation and trafficking are gp96 and PRAT4A (protein associated with TLR4). Gp96 is important for all TLR maturation, while PRAT4A is involved in trafficking of specific TLRs, including TLR7 and TLR9. In the endosome, TLR7 recognizes single-stranded RNA, TLR9 binds to CpG DNA, and TLR3 binds to double-stranded RNA. Proteolysis of both TLR7 and TLR9 is known to occur, and at least in the case of TLR9, is required for function. This image is interactive. Click on the mutations in red to view them on Mutagenetix.

The TLR signaling screen (testing TNF production by macrophages in response to TLR ligands) in which the Unc93b13d mutation was identified disclosed the importance of genes throughout the TLR signaling pathway (Figure 2), from the Tlr genes themselves down to Tnf.  Two of those genes were Hcfc2 and Rhbdf2 (described in the next section), which exert their effects through distinct mechanisms.

TLR3 expression is upregulated by viral infections and by IFN-α, resulting in an amplification of the TLR3-dependent antiviral response. We found that a mutation of Hcfc2, called feckless, abrogated TNF production and reduced IFN-β production by peritoneal macrophages stimulated with the TLR3 ligand poly(I:C), but not other TLR ligands. Mechanistic studies showed that, by promoting the binding of the transcription factors IRF1 and IRF2 to their target genes, HCFC2 is necessary for transcription of hundreds of interferon-regulated genes including Tlr3 (Figure 3). Thus, macrophages homozygous for the feckless mutation failed to express adequate amounts of TLR3 and therefore failed to respond to TLR3 ligands. As a result of this and the impaired expression of numerous IRF1/2-dependent genes, mice with Hcfc2 mutations failed to survive influenza or herpes simplex virus 1 infections. We suspect that mutations of Hcfc2 increase susceptibility to diverse infectious diseases.

The feckless phenotype and causative Hcfc2 mutation are reported in (Sun et al. J.Exp.Med. 214, 3263-3277).

Figure 3. HCFC2 is necessary for Tlr3 transcription by promoting the binding of IRF1 and IRF2 to the Tlr3 promoter. HCFC2 forms a complex with IRF1 and IRF2, which are required for induced and basal Tlr3 transcription, respectively. HCFC2 appears to dissociate from IRF1 and IRF2 upon their binding to their DNA targets. We hypothesized that HCFC2 may generate a conformational change in the IRF that favors DNA binding.

The sinecure mutation of Rhbdf2 caused mildly impaired TLR-dependent TNF production by peritoneal macrophages; the most pronounced defect was in the response to MALP-2, a ligand for the TLR2-TLR6 receptor dimer.  A catalytically inactive member of the rhomboid protease family, RHBDF2 is necessary for the function of the TNFα-converting enzyme (TACE, encoded by Adam17), which cleaves the active TNF ectodomain from its transmembrane precursor.  RHBDF2 binds TACE and promotes its exit from the endoplasmic reticulum and trafficking to the cell surface, where it cleaves the TNF precursor (Figure 4) (Adrain et al. Science. 335, 225-228; McIlwain et al. Science. 335, 229-232).  Although TACE also cleaves EGFR ligands, RHBDF2 was not required for the in vivo processing of EGFR ligands.

The sinecure phenotype and causative Rhbdf2 mutation are reported in (Siggs et al. Blood. 119, 5769-5771).

Figure 4. RHBDF2 is essential for TACE to exit the ER. RHBDF2 associates with TACE in the ER to facilitate the folding and/or trafficking of TACE to the Golgi. Within the Golgi, TACE is processed by furin to the mature form. At the plasma membrane, TACE cleaves the active TNF ectodomain from its membrane-bound precursor. This image is interactive. Click on the mutations in red to view them on Mutagenetix.Modified from Lichtenthaler. Science. 335, 179-180.

Activated by a wide range of microbial and endogenous ligands, signaling from the NLRs ultimately leads to the production of the inflammatory cytokines interleukin (IL)-1β and IL-18.  The NLR family includes NLRP1, NLRP3, and NLRC4, proteins which upon activation recruit a multiprotein complex called the inflammasome.  Inflammasomes serve as platforms for the activation of the cysteine protease caspase-1, leading to the processing and secretion of IL-1β and IL-18, and to the induction of pyroptosis, a form of programmed inflammatory cell death.  Numerous autoinflammatory disorders are associated with activating mutations in NLRP1 [e.g. multiple self-healing palmoplantar carcinoma (MSPC), vitiligo-associated multiple autoimmune disease susceptibility (VAMAS1), autoinflammation with arthritis and dyskeratosis (AIADK)], NLRP3 [e.g. neonatal onset multisystem inflammatory disease (NOMID), familial cold autoinflammatory syndrome 1 (FCAS1), Muckle-Wells syndrome (MWS)], and NLRC4 [e.g. familial cold autoinflammatory syndrome 4 (FCAS4), autoinflammation with infantile enterocolitis (AIFEC)].

The NLRP3 inflammasome is activated in macrophages by a two-step process that involves priming through activation of NF-κB-activating pathways prior to or simultaneously with exposure to a second NLRP3-specific trigger such as extracellular ATP, alum, or the pore-forming toxin nigericin.  To identify regulators of NLRP3-dependent inflammatory signaling, we stimulate peritoneal macrophages from mice carrying ENU-induced mutations with lipopolysaccharide (LPS) followed by nigericin (NLRP3 inflammasome screen).  A mutation of Nek7, called cuties, resulted in reduced IL-1β secretion by macrophages primed with LPS and treated with nigericin.  Mice homozygous for the cuties mutation were protected against experimental autoimmune encephalitis (EAE), an inflammatory disease mediated by IL-1.  We found that by binding to NLRP3, NEK7 is necessary for activation of the NLRP3 inflammasome; specifically, NEK7 is required for formation of a complex containing NLRP3 and the adaptor ASC, oligomerization of ASC, and activation of caspase-1 (Figure 5).  NEK7 is a serine/threonine kinase that is also necessary for progression through mitosis.  We made the remarkable discovery that NLRP3 inflammasome activation and mitosis do not occur simultaneously.  NEK7 enforces this restriction because the quantity of NEK7 present in macrophages is sufficient for only one or the other of these processes. 

NEK7 binds directly to NLRP3, an interaction dependent on the LRR domain of NLRP3 and required for inflammasome assembly. Notably, several mutations within the LRR domain of NLRP3 have been linked with the autoinflammatory disease NOMID. Our findings suggest that the aberrant activation of NLRP3 inflammasomes in myeloid cells of patients with such mutations may stem from an increased association between NLRP3 and NEK7. Targeting this interaction may represent an alternative to neutralizing IL-1β for the treatment of NLRP3-mediated autoinflammatory diseases.

The Cuties phenotype and causative Nek7 mutation are reported in (Shi et al. Nat.Immunol. 17, 250-258).

Figure 5. NEK7 binds to NLRP3 and promotes inflammasome assembly and caspase-1 activation. During interphase, NEK7 is necessary for formation of a complex containing NLRP3 and the adaptor ASC, oligomerization of ASC, and activation of caspase-1. During mitosis, all available NEK7 is involved with mitotic spindle formation, separation of centrosomes, and with abscission during cytokinesis. Thus, NLRP3 inflammasome activation is blocked in mitotic cells. This image is interactive. Click on the mutations in red to view them on Mutagenetix.

In humans and mice, immunoglobulins exist in five forms, or isotypes (IgM, IgD, IgG, IgA, IgE), that differ in functional specialization, for example recruitment of macrophages or activation of the complement cascade.  Antigen-naïve mature B cells co-express IgM and IgD, which are derived from alternative splicing of a single mRNA species.  During an antibody response, B cells can switch their immunoglobulin isotype through the process of class-switch recombination (CSR): after activation by antigen, B cells typically undergo classical CSR (cCSR) mediated by DNA-editing enzymes and DNA double-strand break (DSB) repair proteins, in which the Igh Cμ or δ exon is replaced with Cγ, Cε, or Cα, altering the Ig isotype (Figure 6). Less frequently, activated B cells undergo class switching to IgD, which occurs through recombination targeting the switch region preceding Cμ (Sμ) and a noncanonical switch-like region 5′ to Cδ known as σδ.  The unique physiological functions of IgD, which shares antigenic specificity with IgM, are not fully understood.  Secreted immunoglobulins, also known as antibodies, are produced by plasma cells that have differentiated from antigen-activated B cells, and are therefore produced in the form of the new isotype generated by CSR; antibodies circulate in the blood and help to confer long-lasting protection against their targets.

We use a flow cytometry screen of peripheral blood cells from mice carrying ENU-induced mutations to discover genes important for lymphocyte development.  The lentil phenotype, defined by elevated serum IgD, was identified in this screen and attributed to a mutation in the DNA repair enzyme Trp53bp1.  We showed that excessive CSR to IgD occurred in homozygous lentil mice despite defective cCSR to other isotypes, and was restricted to B cells present in mucosa-associated lymphoid tissue.  IgD overproduction was dependent on hematopoietic expression of the TLR adapter protein MyD88 and an intact microbiome, against which circulating IgD, but not IgM, was reactive.  Analysis of the hyper-IgD phenotype of lentil mice led to the discovery that IgD CSR in wild type mice is regulated in a similar manner, thus revealing previously unknown fundamental characteristics of IgD CSR (Figure 7).  Our findings suggest that by targeting commensal microbes, IgD may contribute to homeostatic regulation of the microbial community.  They also raise the possibility that some human diseases associated with perturbed commensal homeostasis may be due in part to dysregulated IgD. 

The lentil phenotype and causative Trp53bp1 mutation are reported in (Choi et al. Proc.Natl.Acad.Sci.U.S.A. 114, E1196-E1204).

Figure 6. Classical CSR and IgD CSR. (Top) The heavy chain locus in naïve B cells following V(D)J recombination. Naïve B cells express transcripts encoding IgM. (Lower left) In activated B cells that undergo classical CSR, the Cμ exon is replaced with another constant region gene (Cγ1 shown). (Lower right) In activated B cells that undergo IgD CSR, the Cμ exon is replaced with the Cδ exon. IgD CSR occurs by targeting the switch region preceding Cμ (Sμ) and a noncanonical switch-like region 5′ to Cδ known as σδ. In both cases, deleted DNA is released as circular DNA. In the absence of 53BP1 classical CSR is impaired and all CSR is diverted to the IgD CSR pathway, which becomes overactivated.

Figure 7. IgD CSR is activated in a TLR-dependent manner and requires the presence of an intact microbiome in both wild type and 53BP1-deficient mice. The nature of the molecular signature that activates TLR-dependent IgD CSR remains to be determined, as does the mechanism by which signaling from MyD88 promotes IgD CSR. This image is interactive. Click on the mutation in red to view it on Mutagenetix.

Hemophagocytic lymphohistiocytosis (HLH) is a severe inflammatory disease in which uncontrolled activation and proliferation of lymphocytes and macrophages leads to excessive cytokine production; macrophages invade the bone marrow and lymphoid organs. Fever, splenomegaly, anemia, thrombocytopenia, and multi-organ failure ensue, resulting in a fatal outcome if untreated. Mutations of several genes including UNC13D, PRF1, STX11, and STXBP2, are known to cause familial HLH, and we encountered one such mutation in the mouse orthologue of UNC13D. The encoded protein is necessary for the priming step of lymphocyte cytolytic granule secretion prior to vesicle fusion with the membrane (Feldmann et al. Cell. 115, 461-473). The jinx mutation of Unc13d caused dramatically elevated susceptibility to mouse cytomegalovirus (MCMV) infection stemming from a failure of NK cells to degranulate. CD8 T cells also failed to degranulate, although both NK and CD8 T cells produced IFN-γ at high levels. Infection with MCMV at low doses did not cause death nor HLH. However, when infected with typically non-lethal doses of lymphocytic choriomeningitis virus (LCMV), homozygous jinx mice developed HLH-like disease in which excessive IFN-γ production and activation of myeloid and lymphoid cells occurred, yet the proliferation of LCMV continued uncontrolled.

The jinx phenotype provided significant support for an infectious etiology of FHL3 (familial HLH caused by UNC13D mutation), although no specific pathogen has been linked to the disease in humans. By initiating an uninhibited cytokine production loop consisting of activated antigen presenting cells and CD8 T cells, an infectious agent may elicit myeloid expansion that drives lymphoid cell proliferation and activation in an uncontrolled manner in individuals homozygous for UNC13D mutations, or in other susceptible individuals (Figure 8).

The jinx phenotype and causative Unc13d mutation are reported in (Crozat et al. J.Exp.Med. 204, 853-863).

Figure 8. The Unc13d mutation results in an uninhibited cytokine loop initiated by infection with LCMV, which cannot be eradicated due to the failure of lymphocyte degranulation. HLH-like disease develops in homozygous jinx mutant mice. Figure modified from Beutler. Immunol Rev. 227, 248-63.
Adaptive Immune Deficiencies

Immunity works best when the innate and adaptive immune systems work together. The cytokines produced by activated innate immune cells promote inflammation and help stimulate the adaptive immune response carried out by T and B lymphocytes. Some innate immune cells, particularly dendritic cells, take up microbial antigens and present them to T cells to initiate the adaptive immune response. Only T cells dedicated to responding to the specific antigen become activated, and in turn activate antigen-specific B cells. These B cells proliferate and differentiate into antigen-specific antibody-secreting cells, which undergo antibody affinity maturation and immunoglobulin class switch recombination, processes which modify the specificity and functionality of antibodies. Some microbial antigens can directly activate B cells and antibody production in the absence of T cell help; this T cell-independent antibody response is also part of the adaptive immune response. The adaptive immune response develops days after initial exposure to a pathogen and is targeted against the specific pathogen that has invaded the body.

Immune deficiency disorders can involve specific cellular components of the adaptive immune system, such as T cells or B cells, or multiple components, as in severe combined immunodeficiency. Humans with adaptive immune deficiencies typically have a history of recurrent infections; susceptibility to particular classes of pathogen can indicate specific cellular deficiencies or functional defects. The Center for the Genetics of Host Defense uses a flow cytometric screen of peripheral blood cells to identify deficiencies of lymphocyte subsets. Another phenotypic screen measures antigen-specific antibody responses in mice immunized with the T cell-dependent model antigens ovalbumin or β-galactosidase, or with the T cell-independent antigen NP-Ficoll. Still another tests the competence of cytotoxic T cell (CTL) responses, wherein T cells directly and specifically attack antigenically foreign cells introduced into the host. The mutations responsible for immune deficiencies identified in these screens illuminate the normal pathways of lymphocyte development and host defense against infectious diseases.

B cell development gives rise to three major B cell populations: marginal zone B cells that reside in the splenic marginal zone and respond to T cell-independent antigens in the blood; B-1 B cells that reside primarily in the pleural and peritoneal cavities, respond to T cell-independent antigens, and produce natural serum IgM; and follicular B cells, also known as B-2 B cells, which respond to protein antigens in a T cell-dependent manner. In mice and humans, the development of B cells from progenitor cells begins in the fetal liver. After birth, follicular B cell development switches to the bone marrow, where these cells must be produced continuously throughout life. In contrast, marginal zone B cells and B-1 B cells self-renew in their respective locations in the periphery.

We identified a deficiency of follicular B cells, but not marginal zone B cells, in mice with a homozygous mutation of Atp11c (designated emptyhive). B-1 B cells were also reduced in number in emptyhive mice. T cell numbers were normal. Whereas fetal liver B cell development was normal, mutant progenitors from either fetal liver or bone marrow failed to differentiate in the adult bone marrow, indicating a requirement for ATP11C in B cell development specifically in the context of adult bone marrow (Figure 9). This defect was consistent with the failure of B cell differentiation by emptyhive B cell progenitors stimulated with IL-7, which is required for adult but not fetal B cell development. In addition, expression of EBF, an early transcriptional regulator of B cell differentiation whose expression is lost in the absence of IL-7R signaling, was reduced in emptyhive B cell progenitors, although at least one of its targets (Igll1) was expressed normally at the mRNA and protein levels.

Members of the P4 subset of P-type ATPases function as inwardly-translocating lipid flippases thought to enrich aminophospholipids at the cytoplasmic leaflet of lipid bilayers. ATP11C was found to function as a phosphatidylserine flippase specifically in pro-B cells (Yabas et al. Nat.Immunol. 12, 441-449), although excess phosphatidylserine was not observed on the exoplasmic leaflet of emptyhive B cell progenitors. Neither phosphatidylinositol nor the lipid raft component ganglioside GM1 appeared to be disrupted in emptyhive B cell progenitors. The mechanism by which ATP11C promotes IL-7 responsiveness and EBF expression in B cell progenitors, and a full understanding of thedependence of these properties on membrane phosphatidylserine, remains to be determined. Nonetheless, the emptyhive phenotype defined a new type of B cell deficiency, which is cell-intrinsic yet dependent upon the bone marrow microenvironment.

The emptyhive phenotype and causative Atp11c mutation are reported in (Siggs et al. Nat.Immunol. 12, 434-440).

Figure 9. Effect of the emptyhive mutation on B cell development. The emptyhive mutation of Atp11c, encoding a phosphatidylserine flippase of pro-B cells, results in a block in B cell differentiation from the pro-B to pre-B cell stage exclusively in adult bone marrow. Although B cell differentiation occurred normally in homozygous emptyhive fetal liver, mutant fetal liver cells failed to reconstitute the B cell population when transferred to irradiated adult wild type mice.

Mice homozygous for the bumble mutation of Nfkbid exhibited a failure of antibody responses to T cell-independent antigens, and the absence or reduction of the B cell subsets that carry them out.  Specifically, bumble mice displayed a selective defect in T-independent antibody responses of extrafollicular antibody-secreting cells (ASC), which are thought to arise primarily from B-1 and marginal zone B cells.  Thus, bumble mice failed to mount antigen-specific IgM or IgG responses, respectively, to immunization with NP-Ficoll or rSFV-βGal (β-galactosidase encoded by a recombinant Semliki Forest Virus vector), antigens which elicit extrafollicular antibody responses.  In addition, the extrafollicular IgM response to NP-chicken gamma globulin was absent in bumble mice, despite the development of high-affinity IgG antibodies in germinal centers that is also induced by this antigen.  Consistent with these functional defects, bumble mice lacked splenic marginal zone B cells and the frequencies of peritoneal B-1 B cells were greatly reduced; of the B-1 B cells that existed, almost all were B-1b B cells.  Natural IgM antibody levels were reduced in bumble mice.  However, frequencies of peripheral blood lymphocytes were normal in bumble mice; as were pro-/pre-, immature, and mature B cells in the bone marrow; immature and mature follicular B cells in the spleen; double-negative (DN), double-positive (DP), and CD4 and CD8 single-positive (SP) thymocytes; and CD44hi and CD44lo subsets of CD4+ and CD8+ T cells in the spleen.

Further studies showed that Nfkbid is required for differentiation of B-1a B cells.  Development of homozygous bumble B-1a B cells from a neonatal transitional splenic intermediate cell population (TrB-1a; IgM+CD93+B220loCD43+CD5+) was blocked prior to the TrB-1a stage in a cell-intrinsic manner, even though splenic transitional (TrB; IgM+CD93+B220+CD5-) cells were able to differentiate to B-2 B cells (Figure 10) (Pedersen et al. Proc.Natl.Acad.Sci.U.S.A. 111, E4119-26).  Interestingly, heterozygous bumble mice displayed an intermediate antibody response to T cell-independent (type 2) antigens despite normal frequencies of B-1 and marginal zone B cells, indicating a requirement for Nfkbid expression from two functional alleles for normal antibody responses (Pedersen et al. Front.Immunol. 7, 65).

Nfkbid encodes the IκB-like protein IκBNS, which is localized in the nucleus and regulates NF-κB-dependent gene expression.  The targets of IκBNS regulation and the mechanism by which they affect the differentiation of TrB-1a cells and marginal zone B cells have yet to be determined.

The bumble phenotype and causative Nfkbid mutation are reported in (Arnold et al. Proc.Natl.Acad.Sci.U.S.A. 109, 12286-12293).

Figure 10. Effect of the bumble mutation on B cell development. (A) Mice with the homozygous bumble mutation lack splenic marginal zone B cells and reduced frequencies of peritoneal B-1a B cells. (B) Proposed developmental stages for B-1a B cells and point of IκBNS requirement. Early B-1p cells exist in 9-d-old fetal yolk sac and splanchnopleura (Godin et al. Nature. 364, 67-70; Yoshimoto et al. Proc.Natl.Acad.Sci.U.S.A. 108, 1468-1473). Markers for these cells have not been identified to date. From day 11 of gestation, B-1p cells can be identified in the fetal liver as LinCD93+CD19+B220lo/− (Montecino-Rodriguez, Leathers, and Dorshkind. Nat.Immunol. 7, 293-301). Cells with a similar phenotype can be found at low frequencies in the neonatal and adult bone marrow and spleen (Ghosn et al. Proc.Natl.Acad.Sci.U.S.A. 108, 2879-2884; Montecino-Rodriguez, Leathers, and Dorshkind. Nat.Immunol. 7, 293-301). B-1p cells undergo transitional development into mature B-1a cells, initiating CD5 expression in the neonatal spleen and possibly in other secondary lymphoid organs. The B220+CD5 neonatal TrB-cell population gave rise to mostly B-2 cells but also some B-1 cells. Herein, we refer to this neonatal TrB-cell population as “TrB” to reflect their apparent capacity to give rise to both B-1a and B-2 cells, suggesting that they are a heterogeneous population. As indicated by the stippled arrow, it is likely that the TrB cells develop into B-1a cells via the TrB-1a intermediate. The red arrow indicates that IκBNS is required for B-1p cells to progress to the stage of TrB-1a cells. Figure adapted and legend reproduced from (Pedersen et al. Proc.Natl.Acad.Sci.U.S.A. 111, E4119-26).

The adaptive immune deficiency observed in mice homozygous for the elektra mutation of Slfn2 was characterized by enhanced susceptibility to both bacterial (Listeria monocytogenes) and viral infections (mouse cytomegalovirus and lymphocytic choriomeningitis virus).  Normal serum cytokine levels in response to MCMV infection indicated intact innate immune activation.  However, frequencies of CD4+ and CD8+ T cells were reduced in the spleen, lymph node, and blood, despite normal thymocyte development.  Analysis of activation markers expressed on the surface of T cells in elektra mice showed that these cells exist in a semi-activated state (Figure 11).  Moreover, in response to a variety of activation signals including IL-7/anti-IL-7 mAb complexes (which mimics lymphopenia and elicits homeostatic expansion), both CD4+ and CD8+ T cells underwent massive apoptosis via the intrinsic apoptotic pathway (mediated by Bcl-2 family proteins).  Further work suggested that the partial activation of elektra T cells results from a loss of cellular quiescence and predisposes them to apoptosis in response to activation stimuli.

The mechanism of Slfn2 function is not fully understood, although biochemical analysis of T cell receptor (TCR) signaling demonstrated elevated p38 and JNK activation.  TCR-stimulated calcium influx and signaling through the NFAT, NF-κB, Erk, and Akt pathways were similar in wild type and elektra T cells.  Slfn2 is one of ten Schlafen (Slfn) family members in mice, and possesses two protein domains of unknown function.  Identified on the basis of sequence similarity, the divergent AAA domain and COG2685 domain reside within a protein with no sequence similarity to other known proteins.  No structural information has been reported for any Slfn protein.  Thus, questions remain to be answered concerning Slfn2: why do T cells and inflammatory monocytes (which are also susceptible to apoptosis following activation) selectively require Slfn2, while B cells and NK cells do not?  With which proteins does Slfn2 interact and signal to support quiescence under unstimulated conditions and survival following activation?  Although mouse Slfn2 has no human orthologue, might an unrelated human gene serve an orthologous function in human immune cells?

The elektra phenotype and causative Slfn2 mutation are reported in (Berger et al. Nat.Immunol. 11, 335-343).

Figure 11. The semi-activated phenotype of homozygous elektra T cells predisposes them to apoptosis in response to activation or expansion signals. Unstimulated homozygous elektra T cells display a unique phenotype in which CD44low T cells express low amounts of CD62L and IL-7Rα, and CD44high T cells express no CD62L, IL-7Rα, or CD5, indicative of a semi-activated phenotype. When stimulated with anti-CD3ε and anti-CD28, a combination of PMA and ionomycin, or with IL-2 in vitro, most elektra CD8+ T cells die. In vivo, lymphopenic conditions that induce homeostatic expansion result in apoptosis of elektra CD8+ T cells. Low expression of the anti-apoptotic protein Bcl2 in CD44high elektra T cells also contributes to their propensity to apoptosis (not shown).

Innate Immune Drug Screening

In collaboration with the Boger Laboratory (The Scripps Research Institute), we are developing chemical compounds for the ability to activate or inhibit the innate immune response. Harnessing this response through the use of synthetic compounds will permit regulated and powerful immune activation, which may be coupled with the highly specific antibody response to target, for example, cancer cells. Conversely, chemical compounds that inhibit the innate immune response may be useful in the treatment of inflammatory diseases such as systemic lupus erythematosus, rheumatoid arthritis, and psoriasis. To identify active compounds, we screen chemical libraries generated in the Boger Laboratory for the ability to activate TNFα biosynthesis in wild type mouse peritoneal macrophages in vitro. The discovery and characterization of Neoseptin-3, an activator of TLR4 signaling, supports the validity of this strategy.

The strongest stimulatory activity in a screen of ~90,000 synthetic peptidomimetic compounds for activation of TNF production by mouse macrophages was attributed to a single compound, named Neoseptin-1.  Chemical modification of Neoseptin-1 combined with structure–activity relationship (SAR) studies yielded Neoseptin-3 (Figure 12), which was structurally simpler but more potent than Neoseptin-1 in its ability to induce TNF.  In addition to TNF, Neoseptin-3 also activated IL-6 and IFN-β production in a dose-dependent manner.  The molecular target of Neoseptin-3 is the mouse TLR4/MD-2 receptor complex, which was identified by genetic studies in which TNF production was tested in macrophages with deficiencies of TLRs or TLR signaling components.  TLR signaling induced by Neoseptin-3, including activation of NF-κB, MAP kinase, and IRF signaling pathways, was closely similar to that induced by lipopolysaccharide (LPS), the canonical microbial ligand for TLR4.  However, in contrast to mouse macrophages, a human macrophage cell line (THP-1) failed to respond to Neoseptin-3; the reason for this is not known.

Figure 12. Chemical structures of Neoseptin-1, Neoseptin-3, and Neoseptin-4. Chemical modification of Neoseptin-1 combined with SAR studies produced the structurally simpler, much stronger, and approximately equipotent agonists Neoseptin-3 and Neoseptin-4.

The crystal structure of the mouse (m) TLR4/MD-2/Neoseptin-3 complex at 2.57 Å demonstrated that Neoseptin-3 induces the formation of a dimer consisting of two mTLR4/MD-2 heterodimers arranged symmetrically in an “m” shape, representing the active, signaling-competent receptor complex (Figure 13A).  Neoseptin-3 binds as an asymmetrical dimer within the hydrophobic pocket of MD-2 (Figure 13B).  Neoseptin-3 and lipid A, the active moiety of LPS, are structurally dissimilar, and consequently make distinct contacts with mTLR4/MD-2 in the receptor complex.  Yet the two ligands induce an identical active receptor conformation through similar local conformational changes in the MD-2 binding pocket and a nearly identical dimerization interface between the two mTLR4/MD-2 heterodimers.  These findings support the hypothesis that ligands structurally unrelated to LPS, including endogenous host-derived molecules, may activate the TLR4/MD-2 complex.

Identification of TLR4/MD-2 as the target of Neoseptin-3, characterization of signaling induced by Neoseptin-3, and the mTLR4/MD-2/Neoseptin-3 complex structure are reported in (Wang et al. Proc.Natl.Acad.Sci.U.S.A. 113, E884-93). Current work aims to optimize Neoseptin-3 for human TLR4/MD-2.

Figure 13. Crystal structure of the mTLR4/MD-2/Neoseptin-3 complex. (A) View of the overall structure of the complex. (B) Enlarged view of the binding of the asymmetrical Neoseptin-3 dimer in the hydrophobic pocket of MD-2. Image based on PDB 5IJC.

Inflammatory Bowel Disease

Chronic inflammation of the gastrointestinal tract occurs in inflammatory bowel disease (IBD), a term that encompasses several disorders including ulcerative colitis and Crohn’s disease.  In humans and mice, IBD may cause diarrhea that can be bloody, abdominal cramps and pain, weight loss, fever, and anemia (from blood loss in stool).  The loss of tolerance to commensal enteric bacteria underlies IBD, and both environmental and genetic factors predispose individuals to dysregulated (innate and adaptive) immune responses against intestinal microbes.  It remains unclear whether the etiology of IBD also includes an autoimmune component; the prevalence of autoimmune diseases, particularly ankylosing spondylitis, psoriasis, and multiple sclerosis, is elevated in individuals with IBD.

Animal models facilitate mechanistic studies, and our laboratory has implemented a phenotypic screen for mutations that increase susceptibility to colitis when mice are treated with the sensitizing agent dextran sodium sulfate (DSS).  DSS physically damages epithelial integrity to a degree similar to natural insults, such as colonization with toxigenic E. coli or Shigella.  Our studies and others using mouse models have demonstrated that a discrete set of genes, some acting within the epithelial compartment and some within cells of hematopoietic origin, are necessary to restore the status quo ante after injury to the epithelium occurs.  This process entails containment and elimination of microbes that have breached the barrier, and repair of the barrier itself  (Arnold et al. BMC Res.Notes. 5, 577; Brandl and Beutler. Curr.Opin.Immunol. 24, 678-685) .  Elevated endoplasmic reticulum stress (Bertolotti et al. J.Clin.Invest. 107, 585-593; Brandl et al. Proc.Natl.Acad.Sci.U.S.A. 106, 3300-3305; Heazlewood et al. PLoS Med. 5, e54), decreased autophagy (Cadwell et al. Nature. 456, 259-263; Murthy et al. Nature. 506, 456-462), inadequate supply of the high-energy molecule ATP (Turer et al. Proc.Natl.Acad.Sci.U.S.A. 114, E1273-E1281), and abnormal protein trafficking (Brandl et al. Proc.Natl.Acad.Sci.U.S.A. 109, 12650-12655) may all result in chronic intestinal inflammation after transient injury with DSS (Figure 14).  The importance of many of these processes to intestinal homeostasis has been confirmed in humans. 

Figure 14. Pathways necessary to repair intestinal injury, and contain and eliminate microbes after DSS treatment. The DSS-induced colitis screen disclosed numerous mutations (red) that impaired these pathways, leading to chronic intestinal inflammation and consequent body weight loss. Some of the affected genes act within the epithelial compartment (e.g. intestinal epithelial cells [IEC] and goblet cells) and some within cells of hematopoietic origin (e.g. dendritic cells, macrophages, and neutrophils).

In mice, DSS induces colitis characterized by weight loss, bloody diarrhea, intestinal ulcerations, and infiltrations with granulocytes.  However, at a sub-pathologic concentration of 1% DSS in drinking water, wild type mice show no significant weight loss.  In contrast, we found many mice with ENU-induced mutations that showed significant weight loss after seven days of treatment with 1% DSS.  These include mutations in Adam17, Egfr, and Gatm (discussed below).

While it was known that a deficiency of TLR signaling results in increased susceptibility to DSS-induced colitis (Rakoff-Nahoum et al. Cell. 118, 229-241), we showed that the requirement for TLR signaling is restricted to intestinal epithelial cells rather than cells of the hematopoietic compartment.  TLR signaling induces intestinal epithelial cell expression of Ereg and Areg, genes encoding epiregulin and amphiregulin, respectively, protein ligands belonging to the epidermal growth factor (EGF) family (Figure 15).  Consistent with these findings, a dominant inhibitory mutation of the EGF receptor itself (Velvet) and a mutation of ADAM17 (wavedX) both caused severe DSS-induced colitis.  By activating the EGF receptor, epiregulin and amphiregulin promote intestinal epithelial cell proliferation, which is necessary for repair of the epithelial barrier.  ADAM17, a matrix metalloproteinase also known as TNF-α converting enzyme (TACE), is necessary for post-translational processing of EGF family ligands, including epiregulin and amphiregulin.  As a result of impaired shedding of these ligands, homozygous wavedX mice are unable to restore the intestinal epithelial barrier.

The importance of ADAM17 for protection against inflammatory bowel disease has been confirmed in humans.  Blaydon et al. identified a loss-of-function mutation in ADAM17 as a cause of inflammatory skin and bowel disease in two of three children born to consanguineous parents (Blaydon et al. N.Engl.J.Med. 365, 1502-1508).  Based on the phenotype of wavedX mice, we proposed that a “trigger” may have initiated intestinal inflammation in these children, although no specific trigger has been identified.

The DSS-induced colitis phenotypes of Egfr+/Velvetand Adam17wavedX/wavedXmice are reported in (Brandl et al. Proc.Natl.Acad.Sci.U.S.A. 107, 19967-19972).

Figure 15. TLR signaling in intestinal epithelial cells protects against DSS-induced colitis. TLR signaling via MyD88 induces expression of Areg and Ereg. The pro-domains of AREG and EREG must be cleaved by the protease ADAM17 to release the active signaling molecules, which activate the EGF receptor (EGFR). EGFR signaling leads to epithelial cell proliferation needed for repair of the intestinal epithelial barrier.

Mice homozygous for the mrbig mutation of Gatm displayed increased susceptibility to DSS-induced colitis.  Increased cell death and reduced proliferation of the intestinal epithelium was observed in DSS-treated Gatm-deficient mice.  Colonocytes from Gatm-deficient mice expressed elevated levels of metabolic stress markers in response to DSS treatment.  Gatm encodes glycine amidinotransferase (GATM), which catalyzes the rate-limiting step in the biosynthesis of creatine, a nitrogenous organic acid that replenishes cytoplasmic ATP at the expense of mitochondrial ATP via the phosphocreatine shuttle.  In addition to its biosynthesis, creatine is replenished from the diet, and supplementation of homozygous Gatm mutants with exogenous creatine abrogated the colitis phenotype.  Immune cells did not appear to contribute to DSS-induced colitis, since normal peripheral immune cell populations and antibody responses to T cell-dependent and T cell-independent antigens were detected in Gatm-deficient mice.  These findings establish the importance of proper energy metabolism for the repair of the intestinal epithelial barrier after acute injury.  In particular, maintenance of the cytoplasmic ATP pool by the creatine pathway is necessary for epithelial regeneration.

The mrbig phenotype and causative Gatm mutation are reported in (Turer et al. Proc.Natl.Acad.Sci.U.S.A. 114, E1273-E1281).



Although most B cells and T cells with high affinity receptors reactive against the body’s own cells and tissues are eliminated during development, some complete development to express receptors of lower affinity for self antigens. Autoimmune disease results from an adaptive immune response specific for self antigens. Antigen specific T cells are thought to initiate the autoimmune response just as they do in the adaptive immune response against foreign antigens. However, whereas the outcome in protective immunity is clearance of foreign antigen from the body, in autoimmunity the antigen is an intrinsic component of the body and typically cannot be eradicated. The immune response therefore persists, and continual exposure to autoantigen amplifies it. The effector mechanisms directed against microbial pathogens are the same as those directed against self antigens in autoimmunity, and therefore mediate tissue damage in autoimmune diseases. For example, cytotoxic T cells specific for self peptides can directly damage tissue, or can activate macrophages to induce inflammation. Self-reactive B cells can produce harmful autoantibodies. The particular autoantigens and the effector mechanisms that target them determine the clinical manifestations of an autoimmune disease. In most cases, multiple effector mechanisms contribute to disease pathogenesis; this is the case for systemic lupus erythematosus, rheumatoid arthritis, and type I diabetes, in which both T cell and B cell responses mediate tissue injury.

The environment and genetics both contribute to an individual’s susceptibility to autoimmune disease. Although the factor(s) that initiate autoimmune disease are usually not known, in some cases an autoimmune response may be triggered by infection or by the recognition of commensal microbes, as described below for mice with a mutation of Ptpn6. To identify genes that provide a safeguard against systemic autoimmune disease, the Center for the Genetics of Host Defense tests blood from mutagenized mice for the presence of antibodies reactive to double stranded DNA, which are found only in the autoimmune state.

Mice carrying a homozygous mutation of Ptpn6 were initially recognized because they developed chronic inflammatory lesions on their feet.  The phenotype was named spin for “spontaneous inflammation” and was further characterized by inflammatory infiltrates in the salivary glands and lungs, and the presence of anti-chromatin antibodies in the blood.  TNF and IL-1 production by peritoneal macrophages from spin mice was normal in response to TLR ligands, and resistance to Listeria monocytogenes infection was increased relative to that of wild type mice.  Normal resistance to mouse cytomegalovirus was also observed in spin mice.  Both the chronic foot inflammation and the presence of anti-chromatin antibodies in spin mice were completely suppressed by derivation of the mice into a germ-free environment.  They were also suppressed by a deficiency of MyD88 signaling or the IL-1 receptor.  Our data showed that inflammation and autoimmunity in Ptpn6spin/spin mice are not endogenously driven but instead depend on the presence of microbes introduced during postnatal life.  These microbes drive inflammatory signaling via the IL-1 receptor (through MyD88), resulting in inflammation and autoimmunity in the absence of Ptpn6 (Figure 16).

More severe loss-of-function mutations of Ptpn6 in mice result in immunodeficiency, systemic inflammation, and autoimmunity, marked by alopecia, glomerulonephritis, dermatitis, inflammation of the paws, and interstitial pneumonitis, which ultimately causes death by about two months of age (Green and Shultz. J.Hered. 66, 250-258; Shultz et al. Am.J.Pathol. 116, 179-192).  In contrast, spin mice do not develop immunodeficiency or lethal pneumonitis, and typically live for over one year.  Ptpn6 encodes the cytoplasmic protein tyrosine phosphatase SHP1, which may negatively regulate signaling from cytokine receptors, B and T cell receptors, chemokine receptors, receptor tyrosine kinases, and integrins.  Although the relevant targets of its catalytic activity are yet unknown, our findings demonstrate that SHP1 suppresses microbe-dependent IL-1 signaling that causes chronic inflammation and autoimmune disease.

The spin phenotype and causative Ptpn6 mutation are reported in (Croker et al. Proc.Natl.Acad.Sci.U.S.A. 105, 15028-15033).

Figure 16. Putative mechanism of the Ptpn6spin mutation. (A) The IL-1 signaling loop is initiated by microbial stimulation of TLRs, which activate the MyD88-dependent pathway leading to production of pro-IL-1β. Microbial activation of the NLRP3 inflammasome may result in caspase-1 cleavage of pro-IL-1β to generate mature IL-1β, which is secreted and activates its own receptor. The IL-1R complex becomes competent for signaling upon recruitment of IL-1RAcP, and utilizes TLR pathway components to produce IL-1β. SHP1 inhibits IL-1 signaling by an unknown mechanism. (B) Impaired SHP1 function in Ptpn6spin cells results in a failure to downregulate IL-1 signaling, which then continuously feeds forward to produce more IL-1β. Uninhibited IL-1 signaling results in chronic inflammation and autoimmunity in spin homozygotes. This image is interactive. Click on the mutations in red to view them on Mutagenetix.