A glance over the fence: Using phylogeny and species comparison for a better understanding of antigen recognition by human γδ T‐cells

Both, jawless and jawed vertebrates possess three lymphocyte lineages defined by highly diverse antigen receptors: Two T‐cell‐ and one B‐cell‐like lineage. In both phylogenetic groups, the theoretically possible number of individual antigen receptor specificities can even outnumber that of lymphocytes of a whole organism. Despite fundamental differences in structure and genetics of these antigen receptors, convergent evolution led to functional similarities between the lineages. Jawed vertebrates possess αβ and γδ T‐cells defined by eponymous αβ and γδ T‐cell antigen receptors (TCRs). “Conventional” αβ T‐cells recognize complexes of Major Histocompatibility Complex (MHC) class I and II molecules and peptides. Non‐conventional T‐cells, which can be αβ or γδ T‐cells, recognize a large variety of ligands and differ strongly in phenotype and function between species and within an organism. This review describes similarities and differences of non‐conventional T‐cells of various species and discusses ligands and functions of their TCRs. A special focus is laid on Vγ9Vδ2 T‐cells whose TCRs act as sensors for phosphorylated isoprenoid metabolites, so‐called phosphoantigens (PAg), associated with microbial infections or altered host metabolism in cancer or after drug treatment. We discuss the role of butyrophilin (BTN)3A and BTN2A1 in PAg‐sensing and how species comparison can help in a better understanding of this human Vγ9Vδ2 T‐cell subset.


| INTRODUC TI ON
The discovery that butyrophilins (BTNs) control the development and activity of human and mouse γδ T-cell subsets, 1-3 the increasing knowledge on γδ T-cell antigen receptor (TCR)-ligand interaction 4,5 together with the exploding interest in cell-based therapeutics stimulated a general interest in γδ T-cells during the last years. 6 One fascinating aspect of γδ T-cells is their Janus-faced nature. Cells with γδ TCRs are found in all classes of jawed vertebrates, yet they can be highly divergent between species and specialized within an organ- Some TCRs and their ligands seem to co-evolve and are conserved between some species but are completely lost in others and give examples for "Birth-And-Death Evolution" of multigene families. 8,9 Finally, although not subject of this review, some cells develop capabilities involved in processes such as sensing of mechanical stress, 10 thermoregulation, 11 or meningeal development 12 which raises the question on the role of their TCR in these processes. In other cases, TCRs seem to act as sensors of danger or alterations which indicate a need for surveillance and maintenance of tissue integrity. [13][14][15] This review describes some of the specialized features of γδ T-cells and other non-conventional T-cells and illustrates how a comparison between species and additional phylogenetic considerations may promote a better understanding of these cells. A special emphasis will be placed on Vγ9Vδ2 T-cells. They were originally discovered in humans and have been intensively studied for their role in infection and tumor surveillance and are promising targets for cellular therapies. The TCR of the Vγ9Vδ2 T-cell subset senses phosphorylated metabolites of isoprenoid synthesis found in microbial infections and in some host T-cells, especially in tumors or after treatment with drugs such as aminobisphosphonates (ABP). [16][17][18] After a short overview on how species comparison and the discovery on the first non-primate species possessing these cells helps to understand the molecular basis of PAg-sensing, we discuss the discovery of BTN2A1 as new player in PAg sensing by Vγ9Vδ2 T-cells and general implications on models of γδ TCR specificity and for the development of small animal models to study physiology and develop preclinical models of human γδ T-cells. gene families can also be subjected to homogenization events resulting from homologous recombination as part of concerted evolution. 19 Sometimes, it is difficult to discriminate between orthologs, meaning that genes have a direct common ancestor, and paralogs, defined as genes that were created by duplication which took place after speciation. Although of interest when discussing phylogenetic relationships, sometimes we may only use the term homolog, indicating that genes have a common ancestor.

| Variable lymphocyte receptors of jawless vertebrates
Cyclostomata such as lamprey (Petromyzontidae) or hagfish loop. In VLRs, they form a solenoid structure that binds antigen at its concave surface ( Figure 1A). The VLR loci contain a central LRR region with multiple exons encoding for VLR variable sequences (LLRV) which can be combined by gene conversionlike events mediated by activation-induced cytidine deaminases (AID), like cytosine deaminase (CDA). The combination of these central LRRVs that determine the antigen specificity with a Cand an N-terminal LRR finally build up an intron-less gene. [19][20][21] Cyclostomata possess B-like lymphocytes producing soluble VLRs (VLRBs) and T-like cells which mature in a thymus-equivalent expressing VLRAs and VLRCs and might be seen as analogs of αβ and γδ T-cells of jawed vertebrates. The genes encoding for VLRBs are assembled by CDA1 and the VLRA and VLRC genes of the T-like lymphocytes by CDA2. The T-cell like cells may provide "help" for B-like VLRB-expressing cells as indicated by the fact that they produce cytokines such as IL-17 while VLRBexpressing cells carry the IL-17 receptor. VLRC cells are mostly found in the epithelia, analogous to γδ T-cells of jawed vertebrates, and express genes such as Sox13 which has been connected to the development of subsets of IL-17-producing mouse γδ T-cells and genes for adhesion molecules that might be related to their epithelial localization. [20][21][22][23] Jawed vertebrates also possess three antigen receptor-defined lymphocyte lineages but these antigen receptors differ completely from VLRs and are discussed below. Nevertheless, although both types of vertebrates use vastly different genetic mechanisms to generate their three major lymphocyte antigen receptors with completely different structures, these mechanisms were preserved over a 450 million years after separation of the phylogenetic groups. 19,24 Such convergent evolution is common for diverging immune receptors as seen for rapidly evolving classes of mammalian NKcell receptors. These NK-cell receptors bind to polymorphic Major Histocompatibility Complex (MHC) molecules although they were observed to be very different in structure, for example murine Ly49 molecules which are C-type lectin-like type II membrane glycoproteins and primate killer inhibitory receptor (KIRs) which belong to the immunoglobulin superfamily (IgSF). 19,25

| The variable immunoglobulin-like domain (V-domain)
Like Cyclostomata, Gnathostomata also possesses three lineages of lymphocytes which are defined by their antigen receptors. The membrane-bound αβ or γδ TCR is expressed by αβ and γδ T-cells, respectively, and immunoglobulins which consist of an Ig-heavy and Ig-light chain and serve as antigen receptors of B-cells but are also secreted as antibodies. The antigen receptors belong to the IgSF and use a variable type immunoglobulin domain (V-domain) for antigen binding. The antigen receptors are single-pass membrane protein heterodimers consisting of a membrane-distal antigen-binding variable (V) domain and one or more constant (C) domains. These domains consist of two β-sheets, each made up of short anti-parallel β-strands usually joined by a disulfide linkage. V-domains form the antigen-binding site of antigen receptors but are also functional domains of other immunologically relevant molecules such as members of the B7 family, including butyrophilins. Figure 1B gives a general view of a single V-domain of an antigen receptor. It contains A to G β-strands forming two opposing surfaces; one composed by the A-B and E-D strands and the other by G-F, C, and Cʹ-C″ strands. The antigen-binding sites are formed by the complementary determining regions (CDRs). The loop connecting the B and C strand is called CDR1, the one linking Cʹ and C″ CDR2 and CDR3 connects F and G strand. The loop between the E-strand and F-strand is also referred to as the hypervariable loop 4 (HV4). Usually, two V-domains with altogether six CDRs form the antigen-binding site with the CDR3s positioned in the center. 19

| Genes encoding the V-domain antigen receptors
The genes for the V-region of antigen receptors are encoded by two or three exons named variable (V), diversity (D), and joining (J) gene (segments) which are flanked by recognition sites for the recombination-activating gene products RAG-1 and RAG-2. During lymphocyte development in the primary lymphoid organs, they are merged by the RAG-dependent recombination machinery to single genes encoding for a V-domain which is later spliced to the respective constant domains (C) and finally translated to an antigen receptor chain. The loci for the antigen receptor chains contain the respective V-and C-domain-encoding genes. 26 The antigen-binding V-domains are encoded by V-, D-, and J-genes of the Ig-heavy chain locus IgH, of the TCR-β chain locus TRB and the TCR-δ chain locus TRD. The loci for the light chain pairing with the Ig-heavy chain loci are (IGL) for λ and (IGK) for κ. Those for TCR-α chain and TCR-γ chain pairing with TCR-β or TCR-δ chains are TRA and TRG. Their variable domains are encoded by V-genes and J-genes. Especially remarkable with respect to the phylogeny of these gene loci is that in all species (except those few which have two TRD loci) the TRD locus is inserted into the TRA locus with the consequence that the TRD locus is lost during recombination of TRA and that some of the V-genes can rearrange with TRDJ or TRAJ genes and consequently be part of either TCR-δ or TCR-α chains. 19 This is analogous to Cyclostomata, where the VLRC locus is located within the VLRA locus. 23 Peculiarities of TRG loci of unknown physiological relevance are their high variability in size and composition and the use of two separate TRG loci by some ruminants. 27 Many Gnathostomata, but not placental mammals, possess additional types of RAG-recombined IgSF antigen receptors which can be considered functional analogs to TCR or BCR and antibodies, respectively. Interestingly, some species such as Xenopus and chicken possess a second TRD locus containing IIGHV like variable genes. 28,29 Whether these TCR-δ-like chains pair with TCR-γ chains or are part of other antigen receptors is not yet known and also whether cells expressing these alternative types of receptors are functionally and developmentally different from typical B-and T-cells. 28 Rearranged Ig-heavy and Ig-light chains are further diversified by AID-mediated somatic hypermutation (SHM) during affinity maturation in the germinal centers. AID also increases diversity by gene conversion in chicken and in rabbit intestinal B-lymphocytes. For very few species, SHM has been described for TCR loci as well: TRA, TRG, and TRD of nurse sharks 30 and TRG and TRD of dromedary. 31 At least in nurse sharks, SHM happens in the thymus 32 and might serve as a further mechanism of diversity generation of naïve lymphocytes and not as part of an affinity maturation process as in the case of BCR or antibodies, respectively.

| Antigen-binding sites
Diversity and structure of the antigen-binding site is genetically determined. The CDR1s and CDR2s are encoded by the V-genes, and hence, their diversity is limited by the V-gene number while CDR3s correspond to a region generated by recombination of V-, (D-), and J-genes. The RAG-dependent recombination mechanism generates CDR3 variability by combining V(D)J-genes and additional junctional diversity by excision or insertion of variable numbers of nucleotides at the recombination sites and insertions of N-nucleotides by the terminal-deoxynucleotide transferase (TdT). The lengths of most αβ TCRs are rather restricted. This reflects the general conservation of the ligands for MHC-restricted T-cells whose TCRs bind MHCpeptide complexes. The peptides of these complexes bind to an antigen binding groove on the MHC molecule flanked by two α-helices.
TCRs bind with all six CDRs to the MHC-peptide complexes. 33,34 TCRs of non-conventional αβ T-cells bind to complexes of non-polymorphic MHC class I-like (class Ib) molecules and small non-peptide antigens. Those TCRs also cover the surface formed by antigen and α-helices but sometimes as in the case of invariant natural killer T-(iNKT-)cell TCRs, only a few CDRs contribute to ligand-binding. 34 The lengths of the CDRs in Ig can vary considerably which reflects the chemical and topological variety of antigens. The CDR3-length of the TCR-γ chains is rather homogenous while lengths of CDR3s of TCR-δ can vary significantly and be very long. Such long CDR3δs can result from varying numbers of D-genes which are used simultaneously by different species (eg, two in mouse, three in humans, four in cow) 35 which increases junctional diversity and potential diversity of antigen receptors in general 26,36 (Figure 2).

F I G U R E 2
Examples for contribution of CDRs to ligand binding in different classes of IgSF antigen receptors. For detailed discussion of interactions, see. 60 Upper panel: A, γδ TCR CDR3 of the δ-chain dominates interaction with its ligand while the other CDRs and HV4 of the δ-chain and CDR3γ make only minor contributions. B, requires the participation of all CDRs of both the TCR chains for the same. C, Comparable interaction was observed with some BCR as well. Noteworthy of all the above types of receptors, is their uniqueness in the mode of antigen recognition. Lower panel: A, G8 γδ TCR-T22 complex (adapted from PDB 1YPZ), CDR3δ loop is relatively longer than the rest and the one that extends into the groove formed between α1/α2 helixes of T22. B, the 2C αβ TCR interacts with H-2K(b)-dEV8 peptide complex (adapted from 2CKB), where CDR3α and CDR3β contact the peptide in the groove. CDR2α and CDR2β interact with α1/ α2 helixes, and CDR1α and CDR1β interact partly with α1/α2 helixes and peptide. C, Fab of antibody HyHEL complexed with hen egg white lysozyme, where antibody sits right on the antigen and CDRs had spread over the antigen (PDB: 1NDG)

| MHC-restricted cells and other cells with highly variable TCRs
Jawed vertebrates express polymorphic MHC class I and II molecules selecting MHC class I-restricted CD8+ T-cells which are precursors of cytotoxic T-lymphocytes and MHC class II-restricted CD4+ T-cells which can differentiate into helper and regulatory T-cells.
Although αβ TCR genes are likely to have co-evolved with MHC class I/II molecules, no evidence can be found for an association of certain TRA or TRB genes with MHC class I or II isotypes or alleles and TCR gene usage does not allow to predict antigen specificity, homing preference, or effector function. As discussed later, this is a principal In summary, the genes defining "invariant" αβ T-cells can be highly conserved among species but can also be lost. Thus the numbers of these genes have limited predictive value for frequency and function of these cells. 46 Furthermore, invariant T-cells share some features of innate γδ T-cells such as a specific homing to non-lymphoid organs or expression of transcription factors such as PLZF (encoded by ZTB16) or positive selection by double-positive thymocytes. 42,47 At least in mice, (subsets of) innate αβ and γδ T-cells have a different common intra-thymic precursor than MHC-restricted αβ T-cells. 48

| Adaptive vs. innate lymphocytes
Lymphocytes can be roughly described as adaptive vs. innate subpopulations. Hallmarks of adaptive immunity are antigen specificity and a memory based on antigen-driven clonal expansion of cells. An important feature of "classical" antigen-specific lymphocytes is high specificity and variability of their clonally expressed antigen receptors. Nevertheless, some antigen receptor-bearing cells also have features of innate immune cells, and especially, γδ T-cells (or subsets thereof) have been described as innate or bridging innate and adaptive immunity, a term which sometimes means that they share certain phenotypes or functional features with innate immune cells or that they support both types of immunity, for example, by acting as antigen-presenting cells (innate immune cells) which support the adaptive immune response. 49,50 During the last years, the borders between innate and adaptive became less sharp and some lymphocytes originally considered as genuinely innate, for example, NK-cells, show adaptive features such as "antigen specificity" and memory for haptens and/or viruses like Cytomegalovirus (CMV). Nevertheless, the range of specificities is so far rather restricted. The "antigen receptors" are either not known as in the case of hapten-specific NK-cells and if identified they are germline-encoded pattern recognition receptors (PRRs) and variable specificity is generated by varied expression of different receptors by the same cell and therefore fundamentally different from clonally expressed highly specific antigen receptors of classical B-or T-lymphocytes. 51,52 αβ T-cells can also adopt features common to "innate" lymphocytes, either as a result of certain types of antigen-independent activation as in the case of cytokine-induced killer T-cells 53 or as a consequence of their intra-thymic differentiation. 47 This is clearly distinct from MHC-restricted αβ T-cells and does not result in gene silencing leading to the "naïve" phenotype typical for MHC-restricted cells before thymic egress. 50 Furthermore, germline-encoded parts of certain receptors or special V(D)J-rearrangements can be specific for molecular patterns and serve as "innate" PRRs as described for B1 B-lymphocytes or unconventional αβ T-cells and many γδ T-cells. 47

| Mono-or oligoclonal expansion: Sufficient evidence for antigen-specific immunity?
Hallmarks of an adaptive immune response are clonal diversity of antigen receptors and expansion of unique clones as a consequence of antigen recognition. The emergence of high-throughput and single-cell sequencing technologies allows a rather simple and reliable determination of clonal diversity and outgrowth of single clones. 54 In line with earlier reports of dominance of certain TCR-defined γδ T-cell populations in CMV-infected transplant patients, 55 a change in the γδ TCR repertoire could be determined in detail in patients with reactivation of CMV infection after bone marrow transplantation. These cells were part of the Vδ1 subset of human γδ T-cells, which is clonally diverse 56 and in contrast to innate Vγ9Vδ2 T-cells has a phenotype typical for naïve cells which changes to an effector phenotype in the expanding clones. 57 Altogether, these findings are usually interpreted as an indication of an antigen-specific TCR activation, for example, by viruses.
Alternatively, such expansions might reflect a "pseudo-antigen-specific response"; for example, if single clones are activated by antigen-independent signals and expanded by a growth-promoting micro-milieu. In this case, the antigen receptor would just identify a cell clone comparable to a barcode. Clonal differences between activating and inhibitory NK-cell receptors could have a similar effect and even "unspecific" signaling via TCRs might lead to pseudo-adaptive responses if only a few clones are activated beyond a minimal threshold. 58 Such "unspecific" TCR signals could be similar to that of activation by superantigens or other TCR ligands like butyrophilins, as discussed later in this review.
Therefore, in an ideal case, defining an immune response as antigenspecific should be demonstrated by testing the TCRs for specificity to the presumed antigen or microorganism initiating a clonal expansion.

| Sometimes γδ T-cells may be conventional
The lengths of CDRs reflect at least to some extent the nature of the antigen, and it has been noted quite early that the three lineages of antigen receptors differ in their CDR length, especially the CDR3.
TCRs of MHC-restricted αβ T-cells show a highly limited length distribution of α-and β-chain CDRs which fits well to the physical constraints of peptide-MHC binding and a contribution of all six CDRs. 36 In the case of immunoglobulins, the CDR3s of both chains vary considerably consistent with the wide array of ligands and the recognition of conformational epitopes. The antigen-binding sites (paratopes) of murine and human antibodies binding to folded proteins tend to have a flat to concave surface. As a result of convergent evolution, the CDR3s of single heavy chain antibodies of camelids and the New Antigen Receptors (NAR) of sharks can form binding sites that can fit into pockets of proteins and can be very long. 59 The CDR3s of the TCR-δ chain can also be very long, and this length is incompatible with a mode of interaction analogous to the typical ligand binding of MHC-restricted αβ T-cells. 59 The CDR3δmediated mode of membrane protein binding has been elucidated for a γδ TCR specific for the non-conventional MHC class Ib T22 and T10 molecules. Around 0.1% to 1% of murine γδ T-cells have such specificity and express TCRs which bind these molecules with a rather high affinity. T10 and T22 molecules do not bind antigenic peptides, but the structure of the γδ TCR G8 complexed with T22 reveals that the CDR3δ binds in an autonomous mode, so that the other CDRs are not involved. Instead, the CDR3δ nuzzles into the molecule similar to the insertion of antigenic peptides into the binding groove of classical MHC molecules [60][61][62] (Figure 2). The CDR3δ and its flanking regions have also been reported to be essential for binding of human γδ TCRs to other molecules such as MutS and ULBP4. 63 Quite interesting are also other cases of γδ TCR binding without the involvement of "presenting" molecules. Well documented is the binding of mouse and human γδ TCRs to typical B-cell antigens such as the fluorescent algae protein phycoerythrin (PE) 64 and TCRs with specificity for smaller fluorescent molecules like cyanine and the hapten 4-hydroxy-3-nitrophenylacetyl. 65 This may illustrate that γδ TCRs can "recognize" many types of ligand but the physiological meaning is sometimes difficult to demonstrate.
However, the ectopic expression of the mitochondrial F1-ATPase by some tumors could indicate involvement in tumor surveillance of the TCR G115 which binds this molecule. 66 The same might apply for specificity to heat-shock proteins such as a GroEL-like protein in Daudi cells. 67 The most convincing disease association is that of the phosphoantigen-unreactive human Vγ3Vδ2 TCR M88 which was isolated from a muscle lesion of a patient with polymyositis. This TCR binds as a single Fv-fragment to a structure found in several amino-acyl tRNA synthases which is remarkable, as one of them is the histidyl tRNA synthase, a target of Jo-1 autoantibodies also detected in myositis patients. This conformational epitope was also found on other structures such as the short helical loop in the elongation initiation factor 1 of E. coli. 68 This wide spectrum of antigens and the link of TCRs to autoimmunity urges questions on central tolerance, γδ T-cell maturation, and control of auto-or cross-reactivity. 69 Most mouse and human γδ T-cells are usually coreceptor-negative or express the CD8αα coreceptor which is fully consistent with their lack of MHC restriction. Therefore, it was surprising that about 85% of rat splenic γδ T-cells express CD8 and that lymphocyte kinase (Lck) association of the CD8αβ heterodimers of γδ T-cells was indistinguishable from that of CD8αβ-expressing αβ T-cells.
Nevertheless, analysis of CDR3δ lengths of TCRs expressed by the different T-cell types (CD8αβ-positive vs. CD8-negative) showed no difference in length and the average length was even higher than for homologous mouse TCR-δ chains, indicating that these TCRs cannot act in an αβ TCR-like MHC-restricted manner. 70 More related to classical MHC-restricted αβ T-cells are recently discovered γδ T-cells with specificity for melanoma antigen recog-

| Butyrophilins enter the stage
Butyrophilins (BTNs) have originally attracted immunologists´ attention for their immunomodulatory capacity, 73-76 but in recent years, they gained interest for their contribution to γδ T-cell biology. 75,76 The name-giving protein of the gene family is BTN1A1 which is involved in the transport of fat droplets to maternal milk, 77 but has genetic homologs in all jawed vertebrates, for example, in birds where the BTN1A1 homolog Tvc acts as receptor for subgroup C avian sarcoma and leukosis viruses. 78 BTN1A1 and many other BTN and BTN-like (BTNL) molecules modulate activity of immune cells. 75,76,79 In humans, the BTN1A1 gene is part of the BTN gene cluster localized at the telomeric end of the MHC on the short arm of human chromosome 6 (Chr:6p). 80 BTNs are members of the extended B7 family which is named after the costimulatory molecules CD80 (B7- ing away from the membrane. 114 In contrast, introduction of a disulfide bridge between the C-domains, which fixes the V-conformation, diminished ABP-induced stimulation but did not affect ABP-induced mobility reduction in these BTN3A1 constructs, casting doubts on the significance of mobility changes in ABP-induced activation. 115 The B30.2 domain also forms dimers, the symmetric dimer A (or type II) and the asymmetric dimer B (type I) ( Figure 4B). In dimer A, While there is no doubt on the existence of BTN3A1-ED and BTN3A1-B30.2 dimers, the intact molecule might behave differently. 102 Purified BTN3A1 does not form homodimers but co-expression with BTN3A2 facilitates heterodimers which adapt a V-shaped conformation if inserted in artificial membranes (nano-disks). 115 A preference for heteromers was also found in co-immunoprecipitation experiments as well as different cellular trafficking of the isoforms. 120 Cells expressing these heteromers were far better in promoting PAg-mediated stimulation than cells expressing BTN3A1 alone. 102

| Vγ9Vδ2 TCRs and BTN3s in different species
Vγ9Vδ2 T-cells have long been considered to be of primate ori- The armadillo genome revealed ORFs for TRGV9, TRDV2, and

| Alpaca (Vicugna pacos): The first non-primate species with PAg-reactive cells
Genomic data and access to fresh blood allowed a more detailed analysis of alpaca as candidate species for PAg-reactive Vγ9Vδ2 T-cells.
Analysis of PBMCs confirmed database sequences and the expres- To directly test for a BTN3-dependent PAg response of alpaca Vγ9Vδ2 T-cells, 121  from the specific CDR3s of these alpaca Vγ9Vδ2TCRs or is common to alpaca Vγ9Vδ2 TCRs in general. 121 The differential specificity of alpaca Vγ9Vδ2TCR transductants which recognize PAg in the context of alpaca as well as human BTN3, and human TCR MOP transductants which respond exclusively to cells expressing human BTN3 should allow identification of BTN3 and TCR regions controlling PAg-mediated activation by comparing interspecies chimeras or single aa mutants. 121 This TCR-dependent differential response to vpBTN3 vs. human BTN3A argues, at least for this experimental system, against a role of BTN3 as PAg-sensing ligand whose interaction with a counter-receptor on the reporter cells mediates PAg reactivity of the Vγ9Vδ2 TCR as recently suggested by the Kuball group. 109

| Alpaca BTN3: An All-in-One Solution
In contrast to humans with several BTN2 and BTN3 genes, the alpaca BTN cluster carries only single copies of BTN1, BTN2, and BTN3.
The singleton nature of vpBTN3 implies that the capacity of PAg-

| BTN2A1: A Vγ9-binding molecule as player in PAg sensing
One drawback of the original report of BTN3A1 as a key compound in PAg sensing and γδ T-cell stimulation was that ABP-pulsed rodent T-cells transduced with BTN3A1 did not stimulate human Vγ9Vδ2 T-cells. 101 However, such a negative result could reflect impaired costimulatory signals or cell-cell adhesion as a consequence of the species barrier between stimulating and responding cells. As discussed in greater detail elsewhere, 102 our murine reporter cell system allows to (partially) overcome this obstacle. Using these lines as reporter cells and various human Chr:6-containing cell lines (rodent-human hybridoma) as presenters, we showed that in addition to BTN3A1, other gene(s) on human Chr:6 are mandatory for HMBPP-and Zoledronate-but not for mAb 20.1-induced stimulation. 135 Furthermore, unpublished data showed that these molecules were not BTN3A2 and BTN3A3, albeit expression of all molecules increased the PAg-independent "background" stimulation (Paletta, To identify these gene(s), we used another cytogenetic method exploiting species differences, the so-called radiation hybrids (RH).
RH are hybrids of irradiated (human) cells and a HAT-sensitive (rodent) cell line. 136,137 In these cells, parts of the irradiation-induced   with binding studies testing BTN2A1 tetramer-binding to a number of TCR mutants. 139 Importantly, mutating the Vγ9 CDR3 and Vδ2 CDR2 and CDR3 had no effect on BTN2A1 binding but stimulation by Zoledronate or HMBPP was lost completely. 119,139 Both groups demonstrated that no ABP was required for BTN2A1 interaction with BTN3A1 as shown by confocal microscopy and FRET, 139

| A composite ligand model of PAg recognition and its physiological consequences
When discussing binding partners of the TCR, it is of interest that BTN2A1-and BTN2A1 + BTN3A1-transduced murine or hamster cells show a robust and statistically significant background stimulation of 1%-10% of the maximum response to HMBPP. This was completely abolished by mutations of the Vγ9-HV4 but also by mutations/deletions in the CDR2δ and CDR3δ 119 despite that these regions of the TCR-δ chain are not involved in BTN2A1 binding. 119,139 This is fully consistent with mutagenesis studies on the ABP and PAg response which showed involvement of the three CDRs of both TCR chains. 141 We suggest that these TCR regions may interact or bind to other proteins than BTN2A1 required for full activation of the Vγ9Vδ2 T-cell ( Figure 6) and hypothesize a composite ligand model of PAg recognition. This ligand could be a highly conserved molecule(s) and binding simultaneously with BTNs.
An additional interaction partner would be sterically possible and may also contribute to PAg-sensing or PAg-independent activation reported for other ligands (Figure 7). Some of these ligands may have an affinity to the Vγ9Vδ2TCR that is too low to trigger a response on their own and may need to associate with BTN2 or BTN2-BTN3 complexes to be sensed by the Vγ9Vδ2TCR. Others may have a higher intrinsic affinity for specific TCR clonotypes and might be able to interact with the TCR even in the absence of PAginduced cellular changes 119 but in some cases still need "help" by the superantigen-like binding of BTN2A1. 58 In any case, it is worth noting that overexpression of BTN2A1 leads only to "background" stimulation of TCR transductants despite binding of TCR tetramers to these cells was substantially higher than to ABP-pulsed, untransduced cells whose activating properties were far superior. 119

ACK N OWLED G EM ENTS
Open access funding enabled and organized by Projekt DEAL (https://www.proje kt-deal.de/about -deal/).

CO N FLI C T S O F I NTE R E S T
TH obtained a lecture honorarium by Boehringer Ingelheim Vienna. which would probably interact with the CDR2 and CDR3 of Vδ2 chain iii) PAg-BTN3A1 may interact with a counter-receptor of the T-cell, however, the necessity and contribution for activation remains to be shown but would be compatible. The structure of G115 TCR adapted from 1HXM was modified with UCSF Chimera 1,14. CDRs are color-coded, BTN2A1 interacting residues of Vγ9 chain are shown as ball and stick. BTN2A1 was adapted from 119 and the composite ligand model scheme was generated with Inkscape