CONFERENCE REPORT

Innate Immunity in the Aegean:
Ancient Pathways for Today's Survival, Part II.
Mykonos May 26-30th 2002

Louis Du Pasquier and John D. Lambris
University of Basel, Basel Switzerland, University of Pennsylvania, Philadelphia, PA

Invertebrate/vertebrate. Innate immunity/adaptive immunity. Often the two terms of such fractions are thought of as being equivalent, but here this is clearly not the case. Innate immunity is present in all phyla, whereas adaptive immunity, which depends on rearrangement, is present only in vertebrates. Compounds involved in innate immunity range from molecular cascade molecules such as complement (C') to those of the prophenoloxidase cascade and can be very diverse. Yet some strong conservation can be observed throughout many phyla; this is the case for the signalling cascade inside cells.

This second workshop has confirmed the importance of the relationships between the innate and adaptive immune systems in vertebrates and has called our attention to the potential involvement of these systems in other physiological situations, including regeneration and development. The workshop also addressed how one might want to circumvent some of the deleterious effects of certain components of innate immunity, and how the pathogens would love to circumvent the often too-efficient innate immunity. Finally, it allowed us to speculate on the evolution of these systems and their evolutionary relationship to adaptive immune systems. Some aspects of adaptive immunity are also present in vertebrates, where they precede the unfolding of the complex adaptive immune response, which depends on receptors generated by somatic rearrangement. An important question raised was: 'how does adaptive immunity complement and interact with pathways of innate immunity?'

The meeting in Mykonos was the second on the same subject in the Aegean conference series; the first is summarized in Vasta and Lambris (2002).

Local responses

Local Toll NF-kB responses

Human epithelial cells at the boundary between host and environment act as the first line of host defense against invading microbes. and are crucial for host survival. Intuitively, this situation could correspond to the phylogenetically primitive one encountered in histologically simple Metazoa, such as sponges or other diplobastic species. In this workshop primitive metazoa were not represented; in their absence the closest to a simple situation was that in human keratinocytes (Pivarcsi Andor, University of Szeged, Szeged, Hungary). Following direct interaction with pathogens (fungi and bacteria), keratinocytes can kill bacteria in an NF-kB-dependent manner. The expression of the Toll-like receptors 2 and 4 on keratinocyte surface (as detected with monoclonal antibodies), and their expression is modulated by the bacterial components. In terms of further interaction they do produce IL-8, which can start building a network of interactions leading to a systemic response. TLR2, although expressed at very low levels in unstimulated human epithelial cells (Shuto Tsuyoshi, Kumamoto, Japan), is greatly up-regulated by non-typeable Hemophilus influenzae (NTHi), an important human bacterial pathogen causing otitis media and chronic obstructive pulmonary disease. Activation of an IKKb-IkBa-dependent NF-kB pathway is required for TLR2 induction, whereas inhibition of the MKK3/6-p38a/b pathway leads to enhancement of NTHi-induced TLR2 up-regulation. Surprisingly, glucocorticoids, which are well-known potent anti-inflammatory agents, synergistically enhance NTHi-induced TLR2 up-regulation, probably via negative cross-talk with the p38 MAP kinase pathway.

Local production of complement

Local reactions that do involve lymphocytes or other cells of the hematopoietic system can also be mediated by direct synthesis of complement components involving local usage not only of the ancient alternative pathway but of other components, sometimes in a non-conventional manner. As Paul Morgan (UWCM, Cardiff, UK) reminded us, C'components of serum should have poor access to the brain because of the blood-brain barrier. However, many components of this system are found in the brain and are synthesized by brain cells: C3a, C5a, and C3b are produced locally. The C1q molecule of the classical pathway, with binding properties that enable it to participate in other recognition events, is also made locally. This molecule binds to neurons and not to astrocytes. Neuron themselves can activate the classical pathway in the absence of antibody. Locally, C' can play an immuno-regulatory role. For instance, in encephalitis, abnormal constituents activate the C' pathway, and recognition and amplification take place, followed by demyelinization. The components made by astrocytes can kill bacteria directly as soon as one day after infection. In this situation C' is suspected to play a role in the remodeling of the damaged tissue.

Local expression of innate immunity genes during regeneration and development

To a greater and greater extent, the elements of the innate immune system are being found to engage in functions that have less to do with defense than with development. C' components are a good example. A critical role for the anaphylatoxin C5a has been demonstrated during liver regeneration. Dimitrios Mastellos (University of Pennsylvania, Philadelphia, PA) reviewed the involvement of C' in lymphocyte proliferation, bone development, limb regeneration in urodele amphibians, apoptotic induction of neurons, and finally liver regeneration. C5-deficient mice have impaired liver regeneration, and blockade of the C5a receptor halts regeneration. Is this effect due to cytokine release? Is it due to an early need for C5 to control growth? Both explanations are apparently true. IL-6 production is impaired in anti-C5a-treated animals, and, in addition, C5aR seems to be involved in the control of the cell cycle and therefore growth. From regeneration to development the distance is not that great, and it was traversed by Maria Rosaria Pinto (Stazione Zoologica, Napoli, Italy), who showed expression of C3 during development of the tunicate, Ciona (a species with two C3 genes). By in situ hybridization she showed C3 expression in the mesenchyme cells and some neurectoderm cells of the embryo, without any hint as to function. Masuru Nonaka (University of Tokyo, Tokyo, Japan) also noticed unconventional localization of C3 in the amphioxus notochord. Perhaps an ancestral molecule had a dual role in development and defense.
Billie Swalla (University of Washington, Seattle, USA) showed us that C' components and many other genes specifically identified as part of the innate immune system are expressed during metamorphosis in the urochordate, Boltenia. The expression (of pentraxin, P-selectin and a cohort of other genes) was concentrated in the apical portion of the larva, where it adheres to the rock before metamorphosis. Large numbers of cells migrate there during this phase of development. Are the innate immunity genes used to control this migration? Is the innate immune system necessary for the restructuring of the developing animal? Is it simply the setting for the adult immune system? Among the genes identified first as components of innate immunity, the hemolin gene, a member of the Ig superfamily (Faye Ingrid, Stockholm University, Stockholm, Sweden), has now joined the group of those that are also expressed in the ovaries, in the eggs, and during development. Again the question arises whether this expression has a purpose related to immunity. Silencing of hemolin (induced by injection of double-stranded RNA into pupae and mating of the developing adults with wild-type) leads to death at the embryo stage. Whether the gene is needed for protection or for the developmental process itself remains to be investigated.

From local to systemic

Interactions between innate and adaptive immunity

The local interaction is not always sufficient, and in complex metazoa many other cell types and mediators reach a wound and can carry information and further amplify and diversify the response. For instance, interaction of neuropeptides with dendritic cells can influence the outcome of the T-cell response. Dendritic cells are among the first to encounter antigen in a vertebrate immune response. These cells mature upon exposure to pathogens; they produce cytokines that can interact with NK cells, and their behavior can be influenced by neutrophils. Negative regulation of TNFa and IL-12 was observed by the neuropeptides VIP and PACAP (Doina Ganea, Rutgers University, Newark, NJ). The neuropeptides act as anti-inflammatory agents and steer the adaptive response in the Th2 direction rather than the Th1. Interaction, that of DC with NK, also results in modulation of the acquired response, because NK cells can interact with mature or immature DC and the output is therefore not the same. NK cells can kill DC and therefore modulate the response via their NKP30 receptors. This Ig sf V domain encoded in the MHC appears early in ontogeny, together with NKp46 and before NKG2D and CD92 (Alessandro Moretta, University of Genova, Genova, Italy). Why are these cells not more lethal? Their activity is modulated by 2B4, a member of the CD2 family (resembling and coded by the same region as NTBA, which acts during this period as an inhibitory receptor). NK cells can also act a reservoir of HIV virus. HIV infection leads to various defects in innate immunity that could be due to direct infection of CD4+ NK by the virus, with the dreaded result that seronegative patients can harbor virus in their NK cells. NK cells appear as a double-edged sword (George Pavlakis, National Cancer Institute, Frederick, MA), a recurrent feature in biology.
At the outskirts of this system the mast cell (Stephen Galli, Stanford University, CA) is no longer thought to be exclusively or essentially linked to the IgE anti-parasite response. Mast cells can contribute to innate immunity. Using a mutant mouse model in which the mast cell population can be manipulated, he showed that in the absence of mast cells, host resistance is reduced and mast cells can indeed release TNF-a_ not only because of IgE exposure but after direct contact with LPS in a sepsis situation. Mice subjected to cecal ligation and puncture (CLP) do not die if their mast cell population is reconstituted. Similarly, in the absence of C3, mice are much more sensitive to CLP. It seems that the chief role of these cells is to detect the early presence of the pathogens and to monitor it during infection. They then play an amplification role in the inflammatory response.
Neutrophils, in addition to their role in phagocytosis, store cathelicidins, a family of antimicrobial peptides not known in invertebrates. In addition to their possible bactericidal activity, these molecules can attract cells of the adaptive immune system. With a two-hybrid system Margherita Zanetti (University of Udine, Udine, Italy) was able to identify putative interactors expressed in NK, T, and follicular dendritic cells. In parallel, another cathelicidin, bmap-28, induces apoptosis.

. Given the multiplicity of its components and its expression in many tissues, it is perhaps not surprising to see the complement again involved in the modulation of immune cells, especially the dendritic cells. Anna Erdei (Eotvos L University, God, Hungary) is studying the important role played by the activation of C' fragments interacting with DC via covalent binding, which leads to an enhancement of T-cell proliferation, the T cells expressing the CR1 and CR2 receptors upon triggering. In mice the B cells express these receptors known to be involved in the antibody response, but in humans the situation is different: CR1 and CR2 are distinct receptors, and CR1 binding (unlike CR2) inhibits B-cell proliferation. Matyas Sandor (University of Wisconsin, Madison, WI) showed that C3 contributes to T-cell epitope selection in the course of a viral infection. Mice infected with LCMV control the infection with a good acquired immune response. CD8 T cells are generated and antibodies are produced, yet when the animals are injected intracerebrally the infection is lethal and correlates with an increased level of C3; C3-deficient animals survive intracerebral infection. The most interesting aspect of these studies was that the peptide specificity of the CD8 T cells against the virus was not the same as in the C3-deficient animals. How could this epitope specificity be explained? By differences in affinity? By an effect on DC via the CR3 or the CR1/2? Does C3 bind to viral proteins and influence the proteasome activity, and therefore apparently the peptide?
It is perhaps not surprising to see that C' is involved in yet another function, namely hematopoiesis, since most of the C' components are synthesized by bone marrow cells. Ryan Reca (University of Louisville, Louisville) identified such pathways between the receptor of the G protein complex, CXCR4, and the C3a R receptor. Chemokines and C' are natural constituents of the bone marrow and the thymic microenvironment, and the way these receptors devoted to innate immunity influence the homing of T cells and other CD34 lympho-hematopoietic cells was stressed.

Inappropiate effects

With this richness in components and multiplicity of involvement,one could have predicted that the activation of the C' pathways could have unfavorable effects. For instance, an increase in C5a and upregulation of its receptor will lead to an adverse response, as studied in CLP sepsis models (Peter Ward, University of Michigan Medical School, Ann Arbor, MI). Liver fibrosis has been suspected to be linked to a C' regulation disorder because the Küpffer cells and the hepatic stellar cells (HSC) express the anaphylatoxin receptor, and since HSC expresses profibrotic protein, it is tempting to make this link. However, Otto Goetze (Georg-August University Goettingen, Goettingen, Germany) showed that after incubating HSC with C5a alone, fibronectin was up-regulated, and the role of C5a therefore appears to be minor. 
All these effects produced by inappropriate C' activation and dangerous by-products have led scientists to look for C' inhibitors to target particular C' components, such as C3. This is what John Lambris and his colleagues (University of Pennsylvania, Philadelphia, PA) have done in designing more active analogs of Compstatin, one of the most efficient and smallest C' inhibitors.

From the perspective of the pathogen, the effects of C' are indeed very inappropropriate, and these organisms have developed evasion strategies to escape being killed. Harvey Friedmann (University of Pennsylvania, Philadelphia, PA) reminded us of viruses incorporating C' regulator of C' activity, of molecules interfering with C3. He has continued his investigation of two glycoproteins of HVS1 that inhibit C' activation, gC and gE, using virus mutants one with C3b binding defects, one unable to block IgG Fc-mediated activities, and the double mutant. This model showed a reduction in virulence. These studies in C3 KO mice confirmed that the defect was due to a C' effect, and they suggest that the gC and gE proteins could be appropriate targets for vaccination.

Mechanisms
The elucidation of the mechanisms involved in those interactions was the subject of several talks, including one addressing the involvement of reactive oxygen species (ROS) in the activation of interferon regulatory genes by lipopolysaccharide (LPS) (Lorena Navarro, UCSD, La Jolla, CA). Engagement of Toll-like receptors by LPS triggered the nuclear translocation of IRF3, followed by its DNA binding and the subsequent induction of several interferon-regulated genes. Navarro has now demonstrated that ROS participates in several signaling pathways: A redox-sensitive pathway is involved in the LPS-mediated IRF3 activation.
The classical C' pathway is essential to mediating bactericidal killing of Neisseria gonorrhoeae. Peter Rice (Boston University Medical Center, Boston, MA) tried to improve on nature by elucidating the acceptor molecule for C4b on the Neisseria surface. This molecule is a lipooligosaccharide, phosphoethanolamine (PEA), which forms an amide with C4b. A mutant of Neisseria that lacked PEA was more effective in binding IgM and C4b. It is as if PEA covered the binding sites for natural antibodies.
"Two-edged sword" is a term that often came to mind when discussing effector mechanisms. In the case of interactions between microbial virulence factors and the host response (Yaffa Mizrahi Nebenzahl, Ben Gurion University, Beer Sheva, Israel), the outcome of the disease is determined by the pattern of activation of the innate immune response. Serotype 3 induces sepsis and death, whereas serotype 14 induces only diffuse pneumonia that is followed by recovery. This difference in outcome is due to significant production of proinflammatory cytokines in the first case and to simultaneous production of immunosuppressive cytokines in the second.
The binding of the Staph A superantigen (SA) (responsible for many pathologic conditions, including food poisoning and toxic shock syndrome) to MHC class II-TcR leads to excess cytokine production and massive T-cell proliferation, as monitored by cytokine release. In order to understand the early events in this process, Goutam Gupta (Los Alamos National Lab, Los Alamos, NM) has defined a set of markers of the early host response that were used to design receptor mimics to block the generation of pathologies. These receptors were made of a DRa_linked to TCRvb_ representing the docking material for SA, and achieving blockage of SA pathogenesis at nanomolar concentrations.

Evolution

As mentioned earlier, the kinetics of the innate local>innate systemic> acquired adaptive immunity transition that takes place during the course of a response can somehow mimic the evolution of immunity in metazoa. Several evolutionary issues were addressed.
Innate immunity is phylogenetically older than acquired immunity, and self-non -elf discrimination depends in some cases on the density or clustering pattern of the ligand. The possibility of specific recognition by the horseshoe crab tachylectin, is reinforced by the short distance between its individual binding sites (Shun-ichiro Kawabata, Kyushu University, Fukuoka, Japan). Non-self-recognizing proteins involved in innate immunity often recognize sugars with low structural flexibility on the surface of pathogens. The way it helps scavenge pathogens is mysterious, because these molecules do not have neither a collagenase domain nor a serine protease and they cannot activate the C' (thus far missing from this phylum.)
The multiple components that make up the C' system seem to have developed from a simpler system through duplications during evolution (Nonaka). These duplications seem to have occurred in the early jawed vertebrates, meaning that earlier species such as deuterostome invertebrates have retained the simple structure that would help us understand the evolution of the system. In the prochordate amphioixus (one of the closest relatives of the vertebrates) and urochordates such as Halocynthia, Nonaka has identified a large number of C'-like components, some belonging (as expected) to the alternate pathway, but also others such as cC1q, C6, C3, factor B, C'receptor CR3, membrane attack complex, and regulatory SCR-containing components. The original system looks as if it were derived from an alternative/lectin-like pathway. In addition, the strange features of some of the genes, such as MASP and Bf, suggest retrotransposition and other events more complex than simple gene duplication to explain the evolutionary patterns.
The multiplicity of C' components parallels the multiplicity of molecules binding to each of its constituents. In the case of C1q, we have seen binding to AgAb complexes and to the cell surface. We now see MASP binding to MBL, its official ligand, and also with reasonable affinity to ficolin, an oligomeric lectin with a collagen-like domain associated with a fibrinogen domain. In vivo these two molecules may actually compete for binding to MASPs (Nicole Thielens, Institut de Biologie Structurale, Grenoble France). Such competition and alternative pathways could have played an important role in the evolution of the C' system.

Heat shock proteins, another molecular family with multiple members and diverse roles, can elicit in mammals a potent adaptive T-cell response by providing a reloading of the class I pathway without using the biosynthetic pathway. Because the heat shock proteins are extremely conserved, it was of interest to see which pathway of utilization is conserved in lower vertebrates (Jacques Robert, University of Rochester, Rochester). The first T-cell response could be demonstrated in Xenopus using hsp70 and gp96 molecules, with their peptides derived from minor h antigens. He then demonstrated that in the adult, gp96 and hsp70 could generate potent immunity against the tumors from which they originated. This reactivity was peptide-specific but MHC-independent, as if the heat shock protein were an evolutionary bridge between innate and adaptive immunity. In tadpoles an anti-tumor immunity is also generated, but it is not peptide-specific. These findings point to a likely complexity of mechanism here. 
John Marchalonis (University of Arizona, Tucson) reminded us of the abrupt apparent horizontal transfer of genes, enabling somatic rearrangement with the rapid coevolution of TCR MHC and Ig, the hall marks of the adaptive immune system The IgM of the shark repertoire encoded by this newly arrived system, not surprisingly if one think ofv the selection pressures applied to the sysytem, covers the same epitopes as are dealt with by the innate immune system. They encompass C-reactive proteins and endogenous self-antigens.
Coevolution issues are among the most fascinating in the field of immunity, and Peter Parham (Stanford University, Stanford, CA) approached this question in the case of NK cells and MHC. MHC class I and II molecules are involved in mechanisms leading to the elimination of pathogens, and pathogens always try to down-regulate these molecules. Fortunately, NK cells are available to take care of these infected cells and to survey the MHC landscape of the individual, down-regulating their killing capacity when the proper molecule is present. HLA-C in humans is one of the loci that gives a predominantly inhibitory signal to NK cells. Diversity in NK cells is achieved by selective expression of a variety of surface molecules. There are lectin-like NKG2 and CD94 or ig sf (KIR) molecules that include receptors for the MHC class I and II molecules. The process of receptor expression and selection ensures that all circulating NK cells express one or more inhibitory receptors for MHC class I. Within a population the KIR family is so diverse that unrelated individuals always have different KIR genotypes. NKG2 and CD94 are, on the contrary, poorly polymorphic and well conserved, a trend seen in all primates. The repertoire of NK cells within an individual does not change with time or with the history of an individual. Constant and rapid (by evolutionary standards) selection taking place under pathogen pressure results in situations very different among mammals and even among primates, where the Ly49 l gene active in mouse is inactive in humans but active in another primate (the orangutan). Is the repertoire put together by the equivalent of positive and negative selection? In fact, HLA has a small effect on the selected KIR pattern -just subtle modifying effects, unlike its effect on T cells.
A slide of Peter Parham's showed unambiguously that if we mammals do not have good innate immunity, we can die of a disease before our adaptive immunity comes into play. Therefore, innate immunity is essential,and one can anticipate many evolutionary attempts at diversifying its components leading to improvement "Improving" innate immunity could mean using other mechanisms than that identified thus far. This is what Bruce Beutler (Scripps Clinic, La Jolla, CA) is trying to do: identify new genes involved in innate immunity in mice. Even though the core sensing receptors the Toll like receptors have been identified, there must be many additional accessory molecules that are required for TLR functioning, as well as other yet unknown mechanisms. To discover these new genes an ambitious mutagenesis program has been initiated, with the production of about 160,000 mutants a year. The screen used thus far is the ability of the mouse mutant to contain a Pseudomonas infection; 28 variants have been isolated in which transmittability of the mutation has been established. Another 54 mouse mutants have been detected with modifications in macrophage function, also with some instances of transmission. The hope now is to identify the many as yet unknown proteins involved in innate immunity, proteins which will be a major focus of the next workshop on the subject. It will be interesting to see whether these new genes are conserved.

One classical question, often raised by innate immunity specialists, remains: "Why did the vertebrates selected an adaptive system when innate immunity is so diverse and so efficient?" Let us first remind that not all invertebrates have been studies and that other adaptive solutions are perhaps possible. Otherwise one argument is classical: Adaptive immunity provides each individual with the capacity to deal with the whole antigen spectrum. In the view of the diversity and increasing complexity of the innate immunity mechanisms across the animal kingdom, another argument might be economy: The somatic rearrangement not only gives each individual the possibility to adapt to any environment but also saves on genetic material. The introduction of somatic rearrangement might have therefore affected deeply the evolution of innate immunity in vertebrates. The divergence of the evolutionary pathways of innate immunity components in the presence or the absence of somatic diversification could turn out to be an interesting coevolutionary issue.

Vasta, G.D. and J.D. Lambris (2002) Innate Immunity in the Aegean: ancient pathways for the today's survival. Dev. Comp. Immunol. 26:217-225.