Evolution of adaptive immunity: Implications of a third lymphocyte lineage in lampreys

Introduction

The adaptive immune system is a highly interconnected self-defense system that effectively attacks and eliminates a wide variety of pathogens and shows evidence of conservation from humans to sharks (jawed vertebrates). The main players in this system are immunoglobulin superfamily proteins, B-cell and T-cell receptors (TCR), and major histocompatibility complex (MHC) molecules. While B cells produce antibodies that bind to pathogens for elimination, T cells express TCRs, which bind processed antigens that are displayed on MHC molecules, and exert cellular immunity via cytokines and chemokines. The jawless vertebrates (lampreys and hagfish), the most phylogenetically distant vertebrates from mammals, possess an alternative form of adaptive immune system that is mediated by antigen receptors that contain a leucine-rich repeat (LRR) motif; these receptors are called variable lymphocyte receptors (VLRs) 1-4.

Three VLR genes (VLRA, VLRB, and VLRC) have been identified in lampreys and hagfish 1, 4-7, and all undergo somatic gene assembly through the insertion of several germline LRR cassettes into the incomplete germline VLR gene to create a high level of diversity 1, 5, 8, 9. The assembled VLRs show considerable diversity similar to that of the immunoglobulin repertoire, and VLRs bind antigen on the concave surface where the sequence is most variable 10-14. In lampreys, VLRA⁺ cells develop in a “thymus-like” region (the so-called “thymoid” at the tip of the gill filaments in the gill basket) 15. These cells do not secrete VLRA molecules upon antigenic stimulation, and show T-cell-like characteristics 16. VLRB⁺ cells develop in hematopoietic sites (typhlosole, an internal fold of the inner wall of the intestine) and differentiate into plasma cells that secrete soluble VLRB upon antigenic stimulation; thus, their function is essentially equivalent to that of B cells 1, 10, 16. A third gene, VLRC, has also been identified, and the organization of the VLRC gene locus, its assembly mechanism, and the characteristics of VLRC⁺ cells have been reported 6, 17, 18. These studies provide valuable evidence for the evolutionary development of the adaptive immune system.

Lampreys possess one B-cell-like and two T-cell-like lymphocytes

Kasamatsu et al. first identified VLRC in a unique population of lymphocytes from lampreys 6. Recently, Hirano et al. characterized VLRC⁺ cells by analyzing the sites at which they localized, their role in the immune response, and their gene expression profile. Flow cytometry analysis using monoclonal antibodies specific for VLRC confirmed that VLRC⁺ cells in lampreys do not express VLRA or VLRB 17. Thus, there appear to be three separate lymphocyte lineages that express VLRA, VLRB, or VLRC in a mutually exclusive manner. Immunofluorescence analysis revealed that VLRA⁺ and VLRC⁺ cells showed a similar distribution in the typhlosole, kidneys, and gills. On the other hand, digitate-shaped VLRC⁺ cells were also found in some areas, and they were numerically predominant in the epidermal region of the skin. In addition, the repertoire of VLRC genes within VLRC⁺ cells in the skin was very limited. The digitate-shape and conservative repertoire of epidermal VLRC⁺ cells are interesting because of the similarity with mouse γδ T cells 19.

The responses of VLRC⁺ cells to antigenic stimulation are similar to those of VLRA⁺ cells in that they respond to antigens and mitogens with similar morphological changes, but do not appear to differentiate into VLRC-secreting plasma cells 17. In terms of gene expression profiles, VLRA⁺ and VLRB⁺ cells showed a clear distinction; however, VLRC⁺ cells were, in some respects, shown to be similar to VLRA⁺ cells. Interestingly, VLRA⁺ cells show unique expression of molecules such as T-cell factor 1 20 and CTLA4 21, both of which are seemingly important for the development and function of αβ T cells. By contrast, VLRC⁺ cells uniquely express molecules such as the SRY-box containing gene 13, which is necessary for the development of γδ T cells 22, and the Toll-like receptor (TLR) 3, which possibly helps VLRC⁺ cells to respond to RNA viral infections. This is consistent with reports showing that TLR ligands modulate the functions of γδ T cells 23. Thus, the presence of VLRA⁺, VLRB⁺, and VLRC⁺ cells in lampreys could suggest a potential correspondence to αβ T cells, B cells, and γδ T cells in mammals, respectively.

The genomic organization and assembly of the VLRC locus is similar to that of VLRA and VLRB

The structure of the assembled VLRC gene is basically the same as that of VLRA and VLRB 1, 4-7, 18. It comprises an N-terminal LRR (LRRNT), LRR1, multiple other LRRs, a connecting peptide (CP), a C-terminal LRR (LRRCT), and an invariant stalk region (Fig. 1A). The genomic organization of the VLRC locus, its assembly, and the mechanisms underlying its diversification are also fairly similar to those of the VLRA and VLRB loci, albeit with some unique features 1, 4-9, 18 (Fig. 1A, B). The germline VLRC gene contains short stretches of sequence that encode mainly invariant regions of N- and C-termini and a portion of the LRR cassettes, with non-coding sequences in the middle; on the other hand, the short LRR-encoding sequences, encoding several kinds of LRR cassettes, are randomly scattered throughout the locus. There are >500 LRR cassettes in the VLRA locus, >800 LRR cassettes in the VLRB locus, and ∼200 LRR cassettes in the VLRC locus 5, 18. The difference in the number of germline LRR cassettes could easily affect the repertoire of assembled VLRs, although there is no apparent diversity difference between assembled VLRA, VLRB, and VLRC genes 6, 17, 18. Interestingly, TCRγ and TCRδ loci contain considerably fewer V gene segments than TCRα and TCRβ loci, although TCR δ chains show extensive diversity in complementarity-determining region 3, which forms the antigen binding site 24. In the case of VLRC loci, the LRR-encoding sequences identified thus far are all short, and do not encode an entire module 18. In addition, only two LRRCT sequences have been identified 6, 18. Sequence analysis of the VLRC locus suggested that there are several duplication events within the germline LRR cassettes. These events affect 1 to 15 LRR cassettes, either in tandem or separately, and one or more duplications 18.

VLR assembly occurs in a step-wise manner via short homologous sequences 5, 8, 9, 18. The intervening sequence is replaced by inserting LRR cassettes in various combinations; short homologous sequences guide the copying of flanking LRR sequences (Fig. 1B and C). The insertion of an LRR cassette starts either 5′ or 3′ of the germline VLR (Fig. 1B and C), although the exact points (areas) at which assembly is initiated have yet to be identified. Interestingly, although the germline C-terminal region containing LRR sequences is often retained in mature VLRs, it is sometimes replaced with other germline LRR cassettes 18. At this point, the insertion start point can be either within an intervening sequence or within the LRR-coding germline region.

VLRC⁺ lymphocytes represent a distinct lineage with some similarity to VLRA⁺ lymphocytes

VLRA⁺ and VLRB⁺ represent different cell lineages in both lampreys and hagfish 16, 25 and, as mentioned above, Hirano et al. identified VLRC⁺ cells as a third lineage in lampreys 17. So how do these cells develop into different lineages? Since VLRA⁺ and VLRC⁺ cells do not assemble or transcribe the VLRB gene, they are clearly distinct from VLRB⁺ cells. On the other hand, the relationship between VLRC⁺ and VLRA⁺ cells is more complicated. Hirano et al. reported that the gene expression profiles of VLRA⁺ and VLRC⁺ cells overlap to a certain extent, and that the VLRA and VLRC genes are both transcribed in VLRA⁺ and VLRC⁺ cells, indicating that they are transcriptionally regulated in the same way. Kasamatsu et al. used single cell PCR analysis to show mono-allelic assembly in the majority of VLRA⁺ and VLRC⁺ cells, although in one case both the VLRA and VLRC genes were assembled in the same cell 6. Taken together, these results suggest that VLRA⁺ and VLRC⁺ cells might go through the same stages of lymphocyte development.

VLRC assembly is prone to generating non-functional genes

Hirano et al. further examined gene assembly in sorted VLRA⁺, VLRB⁺, and VLRC⁺ cells 17. VLRA⁺ cells contained equal proportions of assembled and germline VLRA genes (∼50/50), suggesting that only one allele is assembled per cell; however, VLRC⁺ cells contained more assembled VLRC genes than germline VLRC genes (∼65/35), although 15% of the assembled VLRCs were non-productive. If one assumes feedback regulation, wherein the generation of a functional VLR prevents further gene assembly, then the failure to assemble one allele would precede the assembly of the other. This would result in diallelic assembly and only one productively assembled VLR allele 25. The fact that VLRC⁺ cells appear to contain many diallelically assembled VLRC genes, many of which are non-productive, suggests that the VLRC assembly process is somehow prone to generating non-functional genes. The reason for this is not yet clear, although it may be due to the structures of LRR cassettes; they can be short, and quite often contain an internal stop codon or show highly diverse sequence at the 5′ or 3′ regions 17.

VLR assembly status suggests the existence of bi-potent progenitors that differentiate into either VLRC⁺ or VLRA⁺ lymphocytes

Hirano et al. also reported that some VLRA⁺ cells contain a fully assembled VLRC gene (indeed, the majority is non-functional), whereas VLRC⁺ cells rarely contain an assembled VLRA gene (those that are assembled are non-functional). The authors went on to propose a model for the assembly of VLRA and VLRC genes, and suggested that if both VLRC alleles within a cell fail to assemble, then VLRA assembly may occur (Fig. 2A). At this point, it is possible to assume that VLRA and VLRC assembly is not mutually exclusive (Fig. 2B), as in the case of V(D)J recombination at TCR β, γ, and δ loci 19. The progenitors can begin assembly on either the VLRA or the VLRC gene and, even when the first assembly generates a non-functional VLRC (or VLRA) gene, the next can occur on either VLRA or VLRC (Fig. 2B). This would generate VLRA⁺ or VLRC⁺ cells that harbor a non-productively assembled VLRC (or VLRA) gene. As mentioned above, the generation of many VLRA⁺ cells harboring a non-productively assembled VLRC gene may be due to the fact that VLRC assembly is somehow prone to generating non-functional genes. Further detailed analyses of gene assembly in individual cells are needed to clarify these issues.

Taking the above into account, the following scenario is possible: VLRA/C⁺ and VLRB⁺ cells are derived from the same lymphoid progenitor, although there is no evidence for this at present. The VLRA/VLRC progenitors move to the thymoid, which is located in the gill, whereas VLRB progenitors move to the typhlosole in the gut. The cells then undergo gene assembly. Those that successfully assemble and express functional VLRs are selected and move to the periphery, whereas those that generate a non-functional VLR gene are subjected to another round of assembly. This may continue until a functional gene is assembled. If gene assembly fails to produce a functional VLR gene, cells will probably die through apoptosis. Before cells enter the periphery, they would need to undergo some form of selection to ensure that they express a functional non-self-reactive antigen receptor.

Did three types of lymphocyte exist before the development of an adaptive immune system?

Since three different lymphocyte lineages, VLRB⁺, VLRA⁺, and VLRC⁺ (possibly equivalent to mammalian B, αβ T, γδ T cells) exist in lampreys, Hirano et al. suggested that three primordial lymphocyte lineages may have existed and served functions in a primordial “non-adaptive” immune system before the rearrangement of antigen receptor genes evolved 17. This is quite important if we are to fully understand the evolution of adaptive immunity.

The hagfish is interesting from an evolutionary perspective because it belongs to a group of jawless vertebrates that is both anatomically and developmentally different from the lamprey; indeed, their evolutionary relationship with the lamprey is still controversial, although recent molecular analysis has shown that hagfishes and lampreys are monophyletic with respect to vertebrates 26-29. Until recently, only VLRA and VLRB were reported in hagfish; however, Li et al. have now identified the third VLR gene 7. By comparing the structures of the three hagfish VLRs with those of the lamprey, they concluded that the third hagfish VLR is the counterpart of lamprey VLRA, and that the previously identified hagfish VLRA is actually the counterpart of lamprey VLRC (Fig. 1A). The diversity of the mature third hagfish VLR was comparable with that of the other VLRs. Further study of the hagfish lymphocytes that express the third VLR will advance our understanding of the evolution of VLR-based adaptive immunity.

What is the origin of VLR gene assembly?

It is widely accepted that V(D)J recombination originated from the accidental insertion of a transposon into a primordial receptor gene 30. V(D)J recombination is the mechanism that ensures the diversity of immunoglobulins and TCRs in jawed vertebrates. So what are the origins of VLR gene assembly? During VLR assembly, only the short homologous sequences are used to incorporate the LRR cassettes 5, 9, 18. VLR assembly differs from ordinary gene conversion in which overall sequence similarity is required along the entire converted region 31. Therefore, we have proposed a mechanism that involves a process called “copy choice” or “template switching” 9, in which DNA polymerase switches templates using short homologous sequences (Fig. 1C).

The activity of activation-induced cytosine deaminase (AID), which belongs to the AID-apolipoprotein B mRNA-editing, enzyme-catalytic, polypeptide-like (APOBEC) family of cytosine deaminases (CDAs) 32, is required for gene conversion of antibody genes in birds, rabbits, and cattle 33, 34. Two genes, CDA1 and CDA2, which are similar to one of the AID-APOBEC CDAs, APOBEC3 32, have been detected in the lamprey 5, and it has been suggested that CDA1 and CDA2 are involved in the assembly of VLRA/C and VLRB, respectively 15, 16, although their actual functions have not been demonstrated. To initiate VLR assembly, the germline VLR gene needs to be cleaved at or near the constant region. Other factor(s), in addition to CDA1 and CDA2, may be involved in this process. To understand the origins of VLR assembly, we need to identify the underlying mechanism(s). For example, which enzymes are responsible and how does the reaction proceed during lymphocyte development?

How do jawless vertebrates discriminate self from non-self?

Before the evolution of antigen receptor gene rearrangement, the genes encoding receptors that were reactive with “self” molecules were probably lost from the genome over time. The need to eliminate self-reactive receptors emerged when antigen receptor genes gained the ability to diversify through gene rearrangement. Both V(D)J recombination and VLR gene assembly are capable of producing over 10¹⁴ different antigen receptors, which are not directly encoded by the genome 8. This creates a new risk: the accidental production of receptors that recognize self-antigens. Therefore, a system that eliminates or inactivates such potentially harmful antigen receptors is required. Importantly, lampreys and hagfish do not appear to possess MHC genes. It would be interesting to examine how VLR-based adaptive immune systems eliminate self-reactive receptors. As for the hagfish VLRB, self-reactive receptors are likely to be eliminated, although the mechanism by which this is achieved has yet to be identified 35.

In the mammalian immune system, if a rearranged antigen receptor is reactive to a self-antigen, the variable region can be replaced with another sequence 36. What happens in jawless vertebrates? It is conceptually possible that a fully assembled VLR gene undergoes gene conversion using other LRR cassettes. Even if this reaction happens, it will be difficult to identify because the initial assembly process and secondary editing via gene conversion would yield indistinguishable VLR sequences.

Conclusions

The identification and analysis of VLRC antigen receptor in the lamprey has made a significant contribution to our understanding of the evolution of adaptive immunity. Hirano et al. analyzed a third lymphocyte lineage, VLRC⁺ cells, and suggested that they could be equivalent to mammalian γδ T cells. The existence of three major lymphocyte lineages and their development patterns suggest that these three cell lineages were present in the common vertebrate ancestor before the advent of antigen receptor gene rearrangement. Although both the antigen receptors and the mechanisms of gene rearrangement are completely different between jawed and jawless vertebrates, they share many of the basic features, including the diversification of antigen receptor genes via gene rearrangement, the expression of only one type of antigen receptor per lymphocyte, and three different lineages of lymphocytes expressing three different antigen receptors. And yet, many questions remain. For instance, what are the functions of the VLRs and how do VLRA and VLRC recognize antigens? Do VLRA⁺ and VLRC⁺ cells provide cellular immunity and interact with VLRB⁺ cells? Does the lamprey possess other types of immune cell, such as B1-like cells 37 or invariant natural killer T (iNKT)-like cells 38? Do lampreys and hagfish eliminate self-reactive antigen receptors, and if so, how? Do hagfish possess three types of lymphocyte? If so, are they functionally equivalent to those in lampreys? What is the mechanism underlying VLR gene assembly and what are the functions of CDA1 and CDA2? Further study of the convergent evolution of adaptive immunity will improve our fundamental understanding of this elegant system.