Sea squirts and immune tolerance Open Access

Once two colonies are united by a common vasculature, the stem cells that are responsible for regenerating new buds transplant between individuals and contribute to both somatic and germline development in the other genotype (Sabbadin and Zaniolo, 1979). In many cases, the two stem cell lineages will compete, with one lineage dominating the other and eventually replacing all germline and/or somatic tissues for the life of the chimera. This process, called stem cell parasitism (SCP) is a repeatable and heritable trait with winner and loser genotypes found in lab-reared and field colonies (Pancer et al., 1995; Stoner et al., 1999; Laird et al., 2005a). As an individual that loses its germline is dead in an evolutionary context, there is clearly tremendous selective pressure on both the creation of, and ability to correctly discriminate between, polymorphic histocompatibility ligands.

The natural transplantation of stem cells is not unique to Botryllus. In cattle, two or more fetuses in the same uterus can possess a common extra-embryonic vasculature, and this can result in adults that are somatic cell (blood) chimeras and that accept reciprocal skin transplants. The natural transplantation of long-term hematopoietic stem cells (HSCs) was the basis of the original concept of immunological tolerance (Owen, 1945; Billingham et al., 1953). Importantly, SCP is thought to be widespread and the driving force for the evolution of polymorphic allorecognition throughout the metazoa (Burnet, 1971; Buss, 1982; Buss, 1987).

The allorecognition system in Botryllus demonstrates remarkable specificity, and three observations provide clues to the possible mechanisms underlying this process. First, two B. schlosseri colonies sharing only a single fuhc allele are compatible, and the characteristics of the fusion reaction are indistinguishable from colonies that share both alleles (Sabbadin, 1962; Scofield et al., 1982). Second, a priori, it seems unlikely that specificity could be the result of homotypic interactions: this would predict that an fuhc allele could mutate randomly such that it could only bind to itself out of multiple competing specificities and, furthermore, that this has happened from hundreds to thousands of times. Finally, functional and genomic studies do not support the presence of recombination, somatic hypermutation or other characteristics of adaptive immunity, such as memory, in this reaction (reviewed by Saito et al., 1994; Dehal et al., 2002). B. schlosseri has a small genome (725 Mbp) (De Tomaso et al., 1998) and cannot somatically modify its germline-encoded genes, thus it seems unlikely that receptors exist that can specifically bind to non-self fuhc. This is consistent with the single-allele match rules of fusibility and, together, indicates the presence of an effector system that is based on recognition of self-fuhc alleles by non-rearranging, germline-encoded receptors that are analogous to the missing-self recognition in vertebrate NK cells (Karre et al., 1986). The discriminatory ability of this system is significant. Colonies from California reject those from Japan and the Mediterranean at high frequencies (>90%), suggesting that the Botryllus effector system can pinpoint a self-allele from hundreds, to potentially thousands, of competing specificities (Rinkevich et al., 1992; Rinkevich and Weissman, 1991).

Further insight into the functional aspects of histocompatibility comes from studies in xenorecognition. There are nine closely related botryllid species but, in most xenopairings, the two individuals do not react to each other, neither fusing nor rejecting (Hirose et al., 2002; Saito, 2003). This implies the presence of an activation signal that initiates the reaction, a process that is further supported by two other experiments. First, if two incompatible colonies are placed briefly into contact and then separated, scarring will often occur 24–48 hours later at the site where the ampullae were in contact, demonstrating that contact with a conspecific individual activates a downstream reaction that occurs even if the colonies have been separated (Tanaka, 1973). Second, ampullae do not react when they touch other objects. This suggests that allorecognition in Botryllus consists of an activation step that leads to a default rejection reaction; however, if the colonies are compatible, an inhibitory step later blocks rejection and initiates a remodeling step resulting in vascular fusion.

Using a forward genetic approach, a candidate fuhc gene was identified, the polymorphisms of which correctly predict the outcome of histocompatibility assays in both lab-reared and wild-type pairings. Although fuhc is a member of the immunoglobulin (Ig) superfamily, it has no orthologous relationship to any mammalian protein and is clearly not the forbearer of the MHC. Fuhc has no recognizable signaling domains, and the average difference between any two fuhc alleles is 40 amino acids, which are spread throughout the ectodomain. The latter is intriguing in the context of how this polymorphism is interpreted to provide specificity (De Tomaso et al., 2005).

In addition to the fuhc, another gene was identified within the locus, named fester, which probably represents an allorecognition receptor (Nyholm et al., 2006). fester encodes a membrane-bound protein with no homology to any vertebrate proteins, but is highly polymorphic; however, these polymorphisms do not contribute to histocompatibility outcomes. Fester also undergoes somatic diversification by alternative splicing: 64 alternative splice variants of the 11-exon gene have been identified. All individuals express a full-length version plus three splice variants. In addition, each individual expresses a unique subset of between eight and 24 different splice variants, alluding to an individual-specific function.

Independent functional assays suggest that fester appears to play roles in both initiating the reaction as well as recognition of fuhc. For the latter, an anti-fester mono-clonal antibody (mAb) interfered with reactions in an allele-specific fashion. When added to a compatible pairing, there was no effect and colonies fused normally; however, in rejecting pairs, the presence of the mAb turned a rejection into a fusion, but only if both partners carried the fester allele that the antibody was specific for (Nyholm et al., 2006). This finding suggested that fester encodes an inhibitory receptor for fuhc, and that the mAb was mimicking ligand binding. These results also suggested that both ampullae must be stimulated for fusion to occur, as both individuals needed to carry the correct allele for the phenotype.

B. schlosseri is highly amenable to reverse genetics using RNA interference (RNAi) (Laird et al., 2005b). When all fester expression was knocked down, completely opposite results were observed from mAb interference: the colonies became unreactive, neither fusing nor rejecting. This suggests that fester is also involved in activating the rejection reaction. We speculate that the splice variants that are expressed by all individuals may be involved in activation, whereas the individual-specific repertoire may play a role in binding to fuhc and constitute an inhibitory receptor. Specificity may be achieved by an overall avidity integrated from multiple splice variants binding to the polymorphic regions of fuhc. Thus, mAb interference and fester knockdown results would be consistent because fester-deficient ampullae could neither initiate a rejection reaction in incompatible pairings, nor recognize the fuhc locus in compatible pairings. This also suggests that fusion and rejection are not mutually exclusive events. A recent unpublished study identified a new member of the fester family that is encoded near the fester locus, and functional studies support these conclusions.

The phylogenetic relationship of the ascidians and vertebrates predicted that fuhc-based allorecognition was controlled by an ancestral, MHC-type molecule (Scofield et al., 1982). However, there is clearly no orthologous relationship between the two systems. In addition, recent results in the cnidarian Hydractinia describe a candidate allorecognition gene, called alr-2, which has no relationship to the MHC or fuhc (Nicotra et al., 2009). Furthermore, the jawless vertebrates (lampreys and hagfish) have a completely unique adaptive immune system that is not based on Ig superfamily members, but rather on the recombination of genes encoding leucine-rich repeat proteins called variable lymphocyte receptors (VLRs) through mechanisms that are not based on recombination-activating gene (RAG) or activation-induced deaminase (AID); this system is probably responsible for the allorecognition reactions described in this species (Pancer et al., 2004). Thus, it appears that the proteins that control allorecognition are unique to each phylum and have evolved independently. Moreover, there can be completely unrelated mechanisms within a phylum. For example, ascidians are hermaphroditic and multiple species show self-sterility, an allorecognition event between egg and sperm. There are two candidate loci involved in this process that are unrelated to the fuhc locus (Harada et al., 2008). This rapid evolution is also found within the mammalian NK cell inhibitory receptors: different species use completely divergent proteins (C-type lectins or Ig superfamily members, or even a combination of both) to accomplish missing-self recognition of the MHC class I molecules (reviewed by Parham, 2005).

These recent discoveries demonstrating an absolute lack of conservation of allorecognition molecules amplify one of the most intriguing mysteries in evolution: the origins of the vertebrate adaptive immune system. Adaptive immunity appeared on the scene abruptly in nearly its present form in the jawed vertebrates (Flajnik et al., 2003), with no precursors of any of the major genes, including the MHC, TCR or B-cell receptor (BCR), found in any other non-jawed vertebrate genome sequenced to date. Thus, one of the most intricate and complex organs in the vertebrate body seems to have evolved incredibly fast and left no clue of its origins in extant organisms: a phenomenon that is completely at odds with the evolution of any other complex biological system. The same appears to be true for allorecognition between, and even within, each phylum.

If there is no conservation of the ligands and receptors used in these various polymorphic recognition systems, are there any commonalities? One mechanism that all of these systems absolutely require is the ability of effector cells to monitor the specificity of recognition events, a process referred to recently as quality control (Boehm, 2006). This phrase encompasses both the processes that create specificity during development, a mechanism that is also termed education, as well as its maintenance over the lifespan of the individual, or tolerance. However, from a functional standpoint, education and tolerance are probably two sides of the same coin. As outlined below, recent data suggest that it may be these quality-control processes that are conserved in the metazoa.

The idea that quality control, not ligands and receptors, is the conserved process of allorecognition is best seen within the vertebrates. Recognition of polymorphic ligands is the basis of both adaptive and some innate immune function, and is carried out by a diversity of receptors including the TCR, BCR, Ly49, killer cell Ig-like receptors (KIRs) and novel immune-type receptors (NITRs), among others (Parham 2005; Cannon et al., 2008). However, each one of these systems has a common theme: the presence of activating and inhibitory receptors, coupled to signaling through immunoreceptor tyrosine-based activation motif (ITAM) and immunoreceptor tyrosine-based inhibititory motif (ITIM) domains, respectively. In other words, evolutionarily unrelated ectodomains (many of which recognize the same ligand, the MHC), all converge onto equivalent signaling pathways. And, although no evidence has yet linked these signaling pathways to allorecognition in the non-vertebrates, the key genes are conserved. Membrane proteins with ITIM and ITAM domains, as well as signal transduction molecules Zap70, SHIP and a shp1/2 homolog, have been found in genome sequences throughout the metazoa, including marine sponges (B. Degnan, personal communication), cnidarians (Steele et al., 1999), sea urchins (Rast and Messier-Solek, 2008) and protochordates, including Botryllus (Azumi et al., 2003). Allorecognition has been described in species in each of these phyla.

We hypothesize that the core of polymorphic recognition strategies may reside in intracellular processes that integrate and respond to external stimuli, and have an early evolutionary origin. The mix and match of divergent ectodomains to common signaling domains, which in turn converge onto these unknown but conserved processes, would allow rapid evolution of the domains involved in recognition at the cell surface – a necessary characteristic of an immune system. This could also provide an explanation for the sudden appearance of adaptive immunity in the jawed vertebrates, as these quality-control mechanisms would have pre-dated the ability to diversify germline-encoded genes. This also explains the complete lack of conservation of ligands and receptors both between and within phyla.

Although it is clear that further complexity exists (e.g. Kharitonenkov et al., 1997; Carlyle et al., 2008), effector function depends ultimately on the correct interpretation of external stimuli, which requires cells to measure the strength of signaling pathways and compare that with a threshold response value. In turn, this threshold must be set during development in an education process, and may also be dynamic throughout the life of a cell (Grossman and Paul, 1992). The mechanisms that underlie integration and education are unknown, but are key to understanding clinical issues such as tolerance. Moreover, over the last few years, multiple studies have demonstrated the regulation of signaling strength in the immune system. Studies in T cells and macrophages have revealed that calcineurin (which acts through an unknown mechanism), miR-181 and miR-155 (which regulate the translation of phosphatases, thereby affecting inhibitory signaling) all function to modify the signaling strength of ligand engagement on the cell surface (Gallo et al., 2007; Li et al., 2007; O’Connell et al., 2009). Obviously, the ability to manipulate signaling strength or response thresholds may be excellent targets for future clinical intervention, such as inducing tolerance.

Botryllus provides an excellent model to study both the structural basis of innate recognition as well as these quality-control processes. Unlike other invertebrate models, allorecognition in Botryllus follows simple Mendelian genetics from which even wild-type crosses do not deviate. In addition, in laboratory and wild genotypes, the polymorphisms of a single protein within this locus determine the outcome, irrespective of genetic background. Recognition occurs on the surface of a single-cell layered epithelial sheet, on the tips of the ampullae, which is macroscopic and sits outside of the body and can be easily visualized and manipulated. Thus, B. schlosseri offers a model where the entire allorecognition system is segregated spatially and focused on recognition of polymorphisms of a single gene, with few inputs and a binary outcome. Studies can be done in a wild-type background comparing the outcomes from multiple fuhc alleles, allowing the sensitivity of the system to be tested directly. Botryllus uses germline-encoded receptors to achieve this specificity, analogous to NK cell recognition of polymorphic MHC, but how this occurs is unknown. The relative simplicity of this system will allow multiple in vivo and ex vivo approaches to dissect the structural and functional mechanisms that underlie specificity.

In addition, Botryllus juveniles are competent to undergo allorecognition immediately following metamorphosis; thus, education must occur during embryogenesis, which is a 1-week process in a tadpole larva that develops in a mosaic fashion and consists of around 5000 cells. In addition, our functional assay in adults is based on the surgical removal and regeneration of the ampullae, which re-develop with no loss in specificity (Nyholm et al., 2006). Together this will allow characterization and comparisons of gene expression and alternative splice variants throughout embryogenesis and regeneration, and this in turn can be compared between individuals that are homozygous for any fuhc allele in different genetic backgrounds.

Botryllus is also amenable to small-molecule screens. In proof-of-principle experiments, the vascular endothelial growth factor receptor 2 (VEGFR2) inhibitor PTK787 was used to block vascular regeneration, and this effect could be phenocopied by VEGFR2 knockdown (Tiozzo et al., 2008). Given the genetic and physical properties of this interaction, coupled to the ability to take multiple redundant approaches for functional analysis, perhaps the most exciting potential use of Botryllus is the ability to take a quantitative proteomic approach to dissect allorecognition responses. As outlined above, multiple aspects of this system make it amenable to quantifying the proteome, phosphoproteome and cell surface expression during embryogenesis, regeneration and allorecognition in vivo, and for comparing values between different genetic backgrounds. This will allow quantitative analysis of the regulation of specificity.

Besides the functional aspects described above, the relative genetic simplicity of this system will allow analyses of the evolution of polymorphism, both within B. schlosseri as well as among related botryllid species (Saito et al., 1994). Allorecognition proteins have also been hypothesized to have an effect on larval behavior, which can now be tested directly (Grosberg and Quinn, 1986). In summary, B. schlosseri has a life history that links together a number of components of allorecognition from disparate fields that are experimentally accessible and can be studied empirically.