Innate Immunity: The Virtues of a Nonclonal System of Recognition

The immune system evolved under selective pressure imposed by infectious microorganisms. As a result, all multicellular organisms have developed various defense mechanisms that have the capacity to be triggered by infection and to protect the host organism by destroying the invading microbes and neutralizing their virulence factors. These phylogenetically ancient defense mechanisms, also known as the innate immune system, use germline-encoded receptors for the recognition of microbial pathogens. This feature distinguishes the innate immune system from the other component of immunity, the adaptive immune system, found only in vertebrates. Adaptive immunity is based on receptors that are generated by somatic mechanisms during the ontogeny of each individual organism. These mechanisms generate a diverse repertoire of antigen receptors with random specificities, which are clonally distributed on two types of lymphocytes: T cells and B cells. Consequently, the specificity of the receptors expressed on each lymphocyte is not predetermined, and neither is the response that can be induced in lymphocytes upon ligation of their receptors by antigen. However, induction of an immune response is only appropriate if the antigen recognized is derived from, or belongs to, a pathogen. Activation of lymphocytes specific for self antigens, or innocuous persistent environmental antigens, may result in autoimmune disorders and deleterious hypersensitivity reactions, respectively. Moreover, protection from different groups of pathogens may require induction of quite different types of effector responses. Again, activation of inappropriate effector responses that do not provide protection against the relevant pathogen often leads to various immunopathologies. Therefore, the adaptive immune response requires signals that provide information about the origin of the antigen and the type of response to be induced. Evidence has accumulated over the past few years to suggest that these signals are provided by the innate immune system (for review, see7Janeway Jr., C.A. Cold Spring Harbor Symp.Quant. Biol. 1989; 54: 1-13Crossref PubMed Google Scholar, 4Fearon D.T. Locksley R.M. Science. 1996; 272: 50-54Crossref PubMed Scopus (1378) Google Scholar, 13Medzhitov R.M. Janeway Jr., C.A. Curr. Opin. Immunol. 1997; 9: 4-9Crossref PubMed Scopus (1140) Google Scholar). Understanding the molecular mechanisms responsible for the generation of these signals would be crucial for the development of new approaches to vaccine formulation and immunotherapy. Immune recognition is unique in that the immune system constantly selects against the targets that it has evolved to recognize. While most other types of molecular recognition involve the products of the same genome (e.g., hormone and its receptor, transcription factor and its specific promoter), immune recognition is mediated between the products encoded in different genomes. Moreover, the selective advantage imposed by immune recognition on the genome of the host usually implies a selective disadvantage to the genome of the pathogen. We suggest that this conflict of interests directs the evolution of innate immunity toward recognition of invariant molecular constituents of the infectious agents. Molecular structures that are essential for the survival of the microbes are not subject to variability in the sense that mutations affecting these structures are lethal for the microbes. Conservation of these molecular structures implies that they are shared by large groups of pathogens. For example, the general structure of iopolyaccharide (LPS) is shared by all gram-negative bacteria. A receptor that would recognize the conserved lipid A of LPS would thus be able to detect the presence of any gram-negative bacterium. This example illustrates another feature of innate immune recognition: the targets of recognition represent molecular patterns, called PAMPs for athogen-ssociated olecular atterns, rather than particular structures (7Janeway Jr., C.A. Cold Spring Harbor Symp.Quant. Biol. 1989; 54: 1-13Crossref PubMed Google Scholar). Host organisms have developed a set of receptors that can specifically recognize PAMPs and are referred to, therefore, as attern ecognition eceptors (PRRs). This evolutionary strategy of the host both prevents the generation of microbial escape mutants and allows a limited number of germline-encoded receptors to recognize a great variety of molecular structures associated with pathogens. The other factor that must have played a role during the evolution of innate immune recognition is related to the effect of immune recognition—destruction of the target. Conceivably, the structures selected for immune recognition must have been absolutely distinct from the self antigens to avoid the damage to self cells and tissues. The consequence of this requirement is the absolute ability of the innate immune system to discriminate between self and nonself (8Janeway Jr., C.A. Immunol. Today. 1992; 13: 11-16Abstract Full Text PDF PubMed Scopus (992) Google Scholar). Indeed, all of the known microbial products that have immunostimulatory activity represent invariant structures shared by large groups of microorganisms, and these are absolutely essential for the microbe's survival: teichoic acids and LPS are common components of gram-positive and gram-negative bacteria, respectively; double-stranded RNA is a structural signature of several groups of RNA viruses; and mannans are conserved components of yeast cell walls. None of these compounds are produced by the host organism, and all of them are essential for the physiology and survival of the respective microbes (7Janeway Jr., C.A. Cold Spring Harbor Symp.Quant. Biol. 1989; 54: 1-13Crossref PubMed Google Scholar). Bacteria cannot survive, for example, without peptidoglycan (one of the PAMPs found in all bacteria), as demonstrated by the lethal effect of penicillin. It should be emphasized that recognition of PAMPs not only allows the adaptive immune response to discriminate between self and nonself, but also between innocuous nonself and pathogen-associated nonself. As mentioned above, an immune response to innocuous nonself antigens may be deleterious to the host organisms, especially if the antigen is persistent in the environment. Likewise, artificial nonself antigens are not recognized by the innate immune system and are usually nonimmunogenic unless they are designed to mimic the activity of PAMPs (see below). Innate immune recognition is mediated by a structurally diverse set of receptors that belong to several different protein families. Some of these receptors (PRRs) recognize PAMPs directly (e.g., CD14, DEC205, collectins), while others (e.g., complement receptors and Toll) recognize the products generated by PAMP recognition (compare Figure 1 and Figure 2). Importantly, whether PAMPs are recognized directly or indirectly, engagement of the receptors can signal the presence of pathogens due to the nature of their ligands. Functionally, PRRs can be divided into three types: humoral proteins circulating in the plasma, endocytic receptors expressed on the cell surface, and signaling receptors that can be expressed either on the cell surface or intracellularly (for review, see4Fearon D.T. Locksley R.M. Science. 1996; 272: 50-54Crossref PubMed Scopus (1378) Google Scholar, 13Medzhitov R.M. Janeway Jr., C.A. Curr. Opin. Immunol. 1997; 9: 4-9Crossref PubMed Scopus (1140) Google Scholar).Figure 2PRRs Recognize PAMPs and Translate Them into a Set of Endogenous Signals that Induce an Adaptive Immune Response and Direct the Differentiation of Lymphocytes into Particular Types of Effector CellsShow full caption(A) Various PAMPs recognized by cognate PRRs expressed on APCs induce the expression of B7 molecules, thus signaling the presence of pathogens and allowing activation of lymphocytes specific for antigens derived from the pathogens.(B) PRRs strategically expressed on the effector cells of the innate immune system (here shown as cell type A and cell type B) induce the expression of corresponding sets of effector cytokines, which in turn direct the induction of the appropriate effector mechanisms in the adaptive immune response. Induction of distinct effector cytokines can be either due to the recognition of different PAMPs, or recognition of the same PAMP on different cell types. IL-12 and IL-4 exemplify prototype effector cytokines controlling distinct effector functions.View Large Image Figure ViewerDownload Hi-res image Download (PPT) (A) Various PAMPs recognized by cognate PRRs expressed on APCs induce the expression of B7 molecules, thus signaling the presence of pathogens and allowing activation of lymphocytes specific for antigens derived from the pathogens. (B) PRRs strategically expressed on the effector cells of the innate immune system (here shown as cell type A and cell type B) induce the expression of corresponding sets of effector cytokines, which in turn direct the induction of the appropriate effector mechanisms in the adaptive immune response. Induction of distinct effector cytokines can be either due to the recognition of different PAMPs, or recognition of the same PAMP on different cell types. IL-12 and IL-4 exemplify prototype effector cytokines controlling distinct effector functions. Cellular PRRs are expressed on effector cells of the innate immune system, including cells that function as professional ntigen-resenting ells (APC) in adaptive immunity, and also on cells that are the first to encounter pathogens during infection, such as surface epithelia. This expression profile allows PRRs to directly induce innate effector mechanisms, and also to alert the host organism to the presence of infectious agents by inducing the expression of a set of endogenous signals, such as inflammatory cytokines and chemokines. This later function allows efficient mobilization of effector forces to combat the microbial invaders. With the development of adaptive immunity, however, the signals induced by nonclonal receptors acquired new functions, namely, the control of activation and differentiation of lymphocytes bearing clonally specific antigen receptors. An example illustrating this point is the conserved Toll/NFκB host defense pathway found in both invertebrates and vertebrates. In adult Drosophila, Toll has been shown to induce antifungal and antibacterial peptides upon infection (9Lemaitre B. Nicolas E. Michaut L. Reichhart J.M. Hoffmann J.A. Cell. 1996; 86: 973-983Abstract Full Text Full Text PDF PubMed Scopus (2832) Google Scholar), while in mammals the activation of Toll results in induction of cytokines and costimulatory molecules required for the activation of the adaptive immune response (14Medzhitov R.M. Preston-Hurlburt P. Janeway Jr., C.A. Nature. 1997; 388: 394-397Crossref PubMed Scopus (4208) Google Scholar) (Figure 1). Interestingly, a homologous pathway is also found in the plant defense system (for review, see5Hammond-Kosack K.E. Jones J.D.G. Annu. Rev. Plant Physiol. Plant Mol. Biol. 1997; 48: 575-607Crossref PubMed Scopus (866) Google Scholar), further suggesting an ancient origin of nonclonal immunity. Antigen receptors expressed on lymphocytes have randomly generated specificities that for this reason cannot determine the origin or biological context of their ligands. Signaling through an antigen receptor is therefore insufficient on its own to induce the activation of lymphocytes or their differentiation into appropriate effector cells. Activation of T lymphocytes, in particular, requires a so-called second or costimulatory signal in the form of B7 molecules provided by the same cells that present an antigen to T cells (11Liu Y. Janeway Jr., C.A. Proc. Natl. Acad. Sci. USA. 1992; 89: 3845-3849Crossref PubMed Scopus (199) Google Scholar; for review, see10Lenschow D.J. Walunas T.L. Bluestone J.A. Annu. Rev. Immunol. 1996; 14: 233-258Crossref PubMed Scopus (2276) Google Scholar). Furthermore, activation of B cells requires the help of activated CD4 T cells in most cases, and therefore, induction of the adaptive immune response is largely controlled by the expression of B7 molecules on APCs. Activation of lymphocytes, however, is only appropriate if their receptors are specific for an antigen derived from a pathogen, rather than self antigens or innocuous nonself antigens. It seems clear, therefore, that the function of costimulatory activity is to signal the presence of pathogens, and, consequently, that the expression of costimulation must be controlled by pathogen recognition (7Janeway Jr., C.A. Cold Spring Harbor Symp.Quant. Biol. 1989; 54: 1-13Crossref PubMed Google Scholar). As discussed above, PRRs are specific for PAMPs and can therefore signal the presence of a pathogen, and do so by induction of B7 expression on APCs that provides information concerning the microbial origin of the antigen to antigen-specific T cells (Figure 2A). The lack of B7 expression on an APC, on the other hand, indicates that the antigens presented by the APC are either self or are of nonpathogen origin. The presentation of antigen in the absence of costimulation leads to permanent inactivation of T cells specific for the antigen in question. Thus, this mechanism prevents activation of autoreactive T cells and T cells specific for innocuous environmental antigens while allowing activation of lymphocytes specific for pathogen-derived antigens. Unlike the effector cells of the innate immune system, conventional naive lymphocytes are not preprogrammed for a particular effector response. Differentiation of pluripotent naive lymphocytes into effector cells depends on external cues (for review, see1Abbas A.K. Murphy K.M. Sher A. Nature. 1996; 383: 787-793Crossref PubMed Scopus (3760) Google Scholar). Several distinct effector mechanisms have evolved in host organisms to efficiently eliminate or neutralize different types of pathogens. The signals that control the induction of any particular effector function should therefore convey information about the features of the pathogen so that it can stimulate an adaptive immune response that chooses the relevant effector mechanism. Effector cytokines, such as IL-12 and IFN-γ in the case of proinflammatory cytokines, and IL-4, IL-6, IL-10, and TGF-β in the case of anti-inflammatory cytokines, are believed to play a major role in the control of CD4 T cell differentiation. These signals, however, are induced upon innate immune recognition of pathogens rather than during the course of an adaptive immune response, and therefore provide the cues needed to produce the correct adaptive immune response (Romagnani, 1991; for review, see17Scott P. Science. 1993; 260: 496-497Crossref PubMed Scopus (475) Google Scholar, 4Fearon D.T. Locksley R.M. Science. 1996; 272: 50-54Crossref PubMed Scopus (1378) Google Scholar, 1Abbas A.K. Murphy K.M. Sher A. Nature. 1996; 383: 787-793Crossref PubMed Scopus (3760) Google Scholar) (Figure 2B). While all multicellular organisms appear to have some form of nonclonal host defense system, only vertebrates have developed the adaptive immune system (for review, see18Thompson C.B. Immunity. 1995; 3: 531-539Abstract Full Text PDF PubMed Scopus (196) Google Scholar). The innate immune system has long been thought of as an evolutionary rudiment whose only function is to contain the infection until the "real" or adaptive immune response can be induced. As discussed in this paper, accumulating evidence suggests that the innate immune system is also essential for the activation of the adaptive immune response and its direction into a particular effector type. Appreciation of this function of innate immunity leads us to reevaluate several issues concerning the evolution of immunity. One reason for the development of adaptive immunity in vertebrates is thought to be the relative "inflexibility" of the nonclonal receptors used by the innate immune system that are unable to cope with the high mutational rate and molecular heterogeneity of pathogenic microbes. However, there are several reasons to believe that this is a misconception. First, as discussed above, innate immune recognition is directed toward molecular targets (PAMPs) that are invariant and shared by many different groups of microorganisms. In fact, most known mechanisms used by pathogens to avoid immune recognition are actually selected to avoid recognition by the specific antigen receptors of the adaptive immune system. Second, if pathogens were able to avoid innate immune recognition, it would have allowed them also to avoid the induction of adaptive immune responses. Finally, invertebrates have very efficient immunity based entirely on a nonclonal recognition system (for review, see Hoffman, 1995), even though the pathogens infecting invertebrates can, of course, be as variable as the ones infecting mammals. It thus appears that the mutational variability of microorganisms, if anything, was not the major factor in the development of adaptive immunity. Adaptive immunity, however, does provide significant selective advantages, such as the generation of long-lasting pathogen-specific memory. The question then is, why have invertebrates not developed an adaptive immune system? The simple answer is that what is advantageous for vertebrates is not necessarily advantageous for invertebrates. More generally, vertebrates and invertebrates are products of the two different types of evolutionary selection strategies, referred to as r selection and K selection strategies (12MacArthur R.H. Wilson E.O. The Theory of Island Biogeography. Princeton University Press, Princeton, NJ1967Google Scholar). While there are different degrees of r or K selection, in general, vertebrates are relatively K-selected and invertebrates are relatively r-selected (15Pianka E.R. Am. Nat. 1970; 104: 592-597Crossref Google Scholar), and the two types of immunity fit readily with other correlates of the two selection strategies (Table 1). Considering short life span, single reproduction early in ontogeny, and other correlates of r selection characteristic of invertebrates, it seems clear that adaptive immunity would be of little, if any, advantage, even if invertebrates could afford it energetically or genetically. On the other hand, strong intra- and interspecific competition, longer life span, and delayed, repeated reproductions characteristic of vertebrates would certainly benefit from long-lasting immunological memory and efficient distribution of the host organism's resources due to clonal expansion of pathogen-specific lymphocytes that proved to be useful in a given environment.Table 1Comparison of r and K Selectionr SelectionK SelectionMortalityNondirected, density-independent, often catastrophicMore directed, density-dependentLength of lifeShort, usually less than 1 yearLonger, usually more than 1 yearFecundityUsually very highUsually lowSelection favors:Rapid developmentSlower development, greater competitive abilityEarly reproductionDelayed reproduction, neotenySmall body sizeLarger body sizeSingle reproductionRepeated reproductionLeads to:ProductivityEfficiencyImmunityInnateInnate and adaptiveBehaviorInnateInnate and adaptiveSome of the correlates of r and K selection (adapted from15Pianka E.R. Am. Nat. 1970; 104: 592-597Crossref Google Scholar, with modifications and additions from the authors). Behavior and immunity here refer to invertebrates and vertebrates only, whereas other correlates, while at extreme points are applicable to these two subphyla, are not necessarily restricted to them. Open table in a new tab Some of the correlates of r and K selection (adapted from15Pianka E.R. Am. Nat. 1970; 104: 592-597Crossref Google Scholar, with modifications and additions from the authors). Behavior and immunity here refer to invertebrates and vertebrates only, whereas other correlates, while at extreme points are applicable to these two subphyla, are not necessarily restricted to them. One of the predictions of the clonal selection theory was the ability of mature peripheral lymphocytes to discriminate between self and nonself. All self-reactive lymphocytes were thought to be deleted during maturation, and any molecule presented to the mature lymphocytes would be regarded as nonself and should therefore induce an adaptive immune response. Indeed, early experiments with model antigens, such as simple chemicals conjugated to self serum proteins and injected with adjuvants showed that responses to such antigens were the norm rather than the exception. However, if unmodified self antigens, and even nonself proteins, were administered alone in the absence of adjuvant, they gave rise to tolerance rather than an immune response. Immune responses specific even for self antigens could be induced, however, if the self antigen was mixed with adjuvants. Although adjuvants have a long history, so far they have only been defined operationally as any substance that increases the immunogenicity of admixed antigens. The reason for the failure of pure antigens to induce immune response is now known—the antigens used in these studies failed to induce the costimulatory signal necessary for the activation of lymphocytes. Accordingly, the mechanism of adjuvant activity has been proposed to be due to the induction of costimulatory signals by microbial constituents carrying PAMPs that are routinely mixed in adjuvants (7Janeway Jr., C.A. Cold Spring Harbor Symp.Quant. Biol. 1989; 54: 1-13Crossref PubMed Google Scholar). Recognition of these PAMPs by PRRs is suggested to induce the signals necessary for lymphocyte activation (such as B7) and differentiation (effector cytokines; see above). In other words, adjuvants induce the innate immune system to produce the signals that are required for activation of an adaptive immune response. While adjuvants are potent immunostimulators, most of them cannot be used in the clinic because of unwanted side effects, and much adjuvant research has been directed toward identification of the active components of the adjuvants and their subsequent modification to minimize side effects (for review, see2Audibert F.M. Lise L.D. Immunol. Today. 1993; 14: 281-284Abstract Full Text PDF PubMed Scopus (199) Google Scholar). Characterization of the nonclonal receptors of the innate immune system responsible for the adjuvant activity, and, evidently, for the associated side effects, would provide a powerful alternative approach, which would ultimately allow one to target these receptors directly. In one example of such a rational approach, fusion of an antigen with C3dg, a product of complement activation, resulted in dramatic potentiation of a specific immune response to the fused antigen (3Dempsey P.W. Allison M.E. Akkaraju S. Goodnow C.C. Fearon D.T. Science. 1996; 271: 348-350Crossref PubMed Scopus (972) Google Scholar). Rational approaches to vaccine design based on the detailed understanding of the molecular mechanisms of innate immunity should ultimately allow one not only to induce a specific adaptive immune response but also the desired effector mechanism without the accompanying damage to the host tissues.