While thymic development in most eutherian mammals is initiated prenatally and offspring are born relatively immunocompetent, marsupials are comparatively less mature at birth with an undifferentiated thymus and few or no circulating lymphocytes. Both cell-mediated and humoral immune responses are consequently absent in marsupials until at least 2 weeks postpartum, with offspring relying significantly on maternal immune protection for the first few weeks of life. The differences in immune development prepartum and postpartum and significant time since marsupials and eutherians shared a common ancestor (170–180 million years ago) have sparked the question of whether marsupial immunity may have evolved to possess unique differences from eutherians?

In 2007, Miller et al. discovered that marsupials indeed had a rather unusual T-cell receptor (TCR) chain in addition to the well-described TCR α, β, δ and γ chains that heterodimerize within αβ and γδ T cells of all mammals.¹ This chain, termed TCRμ, had a constant (Cμ) and variable (Vμ) domain like other TCR chains, but sandwiched between these domains was a second variable-like domain that was already rearranged in the germline (Vμj) (Figure 1a). In a recent Science paper,² Miller joins forces with Rossjohn and Le Nours to report that TCRμ pairs with TCRγ to create a heterodimeric γμTCR. Crystal structures of two examples from the gray short-tailed opossum show that Vγ and Vμ_j create a scaffold that supports the Vμ domain at the membrane distal region of the receptor. This novel research sets the scene to understand what roles these T cells have in the marsupial immune system, identify what type of ligands they recognize and discover why other mammals have lost this third type of TCR. In a world in which we are only just beginning to extend our own immune repertoire by co-opting other species immune recognition systems (e.g. camelid antibodies) and through de novo receptor design (e.g. chimeric antigen receptors), the marsuipial γμTCR provides a new source of nanobodies and its structure may inform receptor engineering designs in the future.

To define the architecture of a TCR containing TCRμ, Morrissey et al.² set about determining which TCR chain pairs with TCRμ with a single-cell transcriptome RNA sequencing campaign. After finding no TCRμ transcripts in peripheral blood T cells, they moved on to the spleen where 37% of cells contained TCRμ. The authors measured transcripts in these cells for the presence of TCRα, β, γ and δ chain transcripts. While the odd TCRμ⁺ T cell had transcripts of rearranged TCRβ and δ, all contained functional TCRγ transcripts confirming a pairing between TCRμ and TCRγ that was hypothesized because of TCRμ’s similarity to TCRδ.

Further mining of the single-cell data set enabled the authors to look for coreceptor expression and TCR signaling chains (summarized in Figure 1b). γμ T cells are mostly CD8αα⁺, with around a quarter having neither CD4 or CD8 transcripts, and γμ T cells contain the TCR machinery required for signaling (CD3 ε, γ, δ and ζ). The μ chain indeed contains the canonical basic residues required for productive CDδε and ζζ pairing,³ which strongly suggests the isolated γμ T cells are able to respond upon ligand recognition. That they are CD8αα⁺ suggests that they may have innate characteristics. CD8αα is a poor coreceptor for major histocompatibility complex class I compared with CD8αβ, indicating that like γδ T cells they may recognize antigen directly rather than in combination with major histocompatibility complex class I.⁴ It will be interesting to perform further analysis of the transcriptome data as it no doubt provides a wealth of information on the activation state of these cells, and what cytokines and chemokines they produce or respond to.

Single-cell transcriptomics also elucidates the nature of V (variable), D (diversity), J (joining) recombination to determine which splicing events are more heavily represented in both chains of the γμ TCR, and reveals the sequence of the CDR3 (complementary determining region 3) loop, which provides clues as to what type of antigen these receptors might respond to. This analysis suggests that antigen engagement is likely restricted to the N-terminal variable domain (Vμ) of the TCRμ chain as the Vμj domain is germline encoded and thus invariant, and the TCRγ chains that pair with TCRμ also seem to display less diversity than those paired with TCRδ in γδ T cells. By contrast, the CDR3 loop of Vμ is unusually long and most often incorporates three D segments (as opposed to the single D segments that are typical in αβ and γδ TCRs), providing a larger surface area for the direct engagement of antigen.

There is a restricted usage of V(D)J segments in both the TCRμ and TCRγ chains found in γμ T cells, suggesting that these T cells are selected on antigen or that there are other factors at play that limit diversity. While the bulk of γμ T cells had TCRγ chains splicing Vγ2 and Jγ3, γδ T cells of the spleen mostly bore TCR with Vγ3.2 and Jγ1 splicing events, indicating that γδ and γμ T cells likely have distinct biological functions. The TCRμ gene locus contains five clusters in which V, D, J and C segments can be spliced; however, interrogation of about 30 γμ T cells only found evidence that clusters 3, 5 and 7 were used, with 57% using cluster 5, again hinting that these T cells recognize a restricted diversity of antigen. Care must be taken in extending these findings to all γμ T cells as only splenic γμ T cells were interrogated. Indeed γδ T cells tend to localize to different tissues depending on Vγ specificity⁵ and γμ T cells if found in other tissues may do the same.

How a TCR chain with two variable and one constant domain heterodimerizes with a TCR chain containing one variable and one constant domain is revealed by representative crystal structures of two γμ TCRs isolated from splenic γμ T cells. The structures show what looks like a typical TCR with a tethered Vμ domain extending from the membrane, positioning the CDR loops outward from the cell in an orientation consistent with their hypothesized role in antigen binding. The elongated CDR3 loop appears flexible, as density for some residues is weak. The two crystal structures orientate the Vμ domain differently atop the VμjVγ domain, but this may not be a physiological difference as each arrangement is likely to be enforced by crystal packing, which is very different between the two structures. The crystal structure that describes the more common Vγ2 γμ TCR heterodimer does show a contact between Vμ and Vγ2 in the form of a hydrogen bond, and a number of key contacts strengthen the interaction between Vμj and Vγ2 domains, perhaps explaining why this pairing is favored in the majority of γμ T cells. Coupled with the restricted diversity of both the Vμj and Vγ domains, it appears likely that neither domain contributes to antigen binding, but instead provides a scaffold to orient the tethered and variable Vμ domain.

Marsupials and monotremes are not the only species to evolve a small antigen-binding domain that is tethered to familiar structures used in antigen recognition. Cartilaginous fishes possess a similar TCR termed γδ-NAR with a very similar architecture.⁶ Indeed, a restricted γδ TCR pairing provides the ideal platform for the NAR domain which is spliced into the TCRδ transcript. Antibodies of camelids and sharks also have antigen recognition domains resembling the Vμ domain. Indeed, Vμ and Vμj are more structurally similar to antibody V domains than V domains from TCRδ. In line with the Vμ domain being solvent exposed, polar residues in place of hydrophobic residues that would usually be buried in heterodimeric interfaces are present and suggest that the Vμ domain could be released from its hinge as a soluble domain and used as a marsupial-derived nanobody.

Identifying both the pairing partner for TCRμ and the most common splicing arrangement of TCRγ (at least in the spleen) sets the scene to understand the nature of the ligand γμ T cells respond to and what their role might be in the marsupial/monotreme immune system. As TCRα and β chains are rearranged before the TCRγ chain in marsupials,⁷ it does not seem likely that they are examples of fetal T cells produced prior to the establishment of the mature adaptive immune system. One might speculate that γμ T cells play a specialized role in protecting marsupials while they develop in the pouch of their mother, but what of the similar receptor found in cartilaginous fishes? Curiously, what is the selective pressure for marsupials and monotremes to keep this fifth TCR chain—and are Eutherian mammals, including us, missing out?

虽然大多数 Eutherian 哺乳动物的胸腺发育是在出生前开始的，并且后代出生时具有相对的免疫能力，但有袋动物在出生时相对较不成熟，胸腺未分化，循环淋巴细胞很少或没有。因此，有袋动物至少在产后 2 周内都不存在细胞介导的免疫反应和体液免疫反应，后代在生命的最初几周内显着依赖于母体的免疫保护。产前和产后免疫发育的差异以及自有袋动物和真人动物拥有共同祖先（1.70-1.8 亿年前）以来的重要时间引发了一个问题，即有袋动物的免疫是否进化为与真人动物具有独特的差异？

2007 年，米勒等人。发现除了在所有哺乳动物的 αβ 和 γδ T 细胞内异二聚化的 TCR α、β、δ 和 γ 链之外，有袋动物确实具有相当不寻常的 T 细胞受体 (TCR) 链。¹这条链，称为 TCRμ，与其他 TCR 链一样，具有恒定 (Cμ) 和可变 (Vμ) 域，但夹在这些域之间的是已经在种系 (Vμj) 中重排的第二个类可变域（图 1a） . 在最近的一篇科学论文中，² Miller 与 Rossjohn 和 Le Nours 联手报道了 TCRμ 与 TCRγ 配对以创建异二聚体 γμTCR。来自灰色短尾负鼠的两个例子的晶体结构表明 Vγ 和 Vμ _j创建一个支架，支持受体膜远端区域的 Vμ 结构域。这项新研究为了解这些 T 细胞在有袋动物免疫系统中的作用奠定了基础，确定了它们识别的配体类型，并发现为什么其他哺乳动物失去了这第三种类型的 TCR。在我们刚刚开始通过选择其他物种的免疫识别系统（例如骆驼抗体）和通过从头受体设计（例如嵌合抗原受体）来扩展我们自己的免疫库的世界中，有袋动物 γμTCR 提供了一个新的来源纳米抗体及其结构可能会为未来的受体工程设计提供信息。

为了定义包含 TCRμ 的 TCR 架构，Morrissey等人。²着手通过单细胞转录组 RNA 测序活动确定哪条 TCR 链与 TCRμ 配对。在外周血 T 细胞中没有发现 TCRμ 转录物后，他们转移到脾脏，其中 37% 的细胞含有 TCRμ。作者测量了这些细胞中是否存在 TCRα、β、γ 和 δ 链转录物的转录物。虽然奇怪的 TCRμ ⁺ T 细胞具有重排的 TCRβ 和 δ 的转录物，但都含有功能性 TCRγ 转录物，证实了 TCRμ 和 TCRγ 之间的配对，这是假设的，因为 TCRμ 与 TCRδ 相似。

对单细胞数据集的进一步挖掘使作者能够寻找辅助受体表达和 TCR 信号链（总结在图 1b 中）。γμ T 细胞主要是 CD8αα ⁺，大约四分之一既没有 CD4 也没有 CD8 转录本，γμ T 细胞包含信号所需的 TCR 机制（CD3 ε、γ、δ 和 ζ）。μ 链确实包含生产性 CDδε 和 ζζ 配对所需的典型碱性残基，³这强烈表明分离的 γμ T 细胞能够对配体识别做出反应。它们是 CD8αα ⁺表明它们可能具有先天特征。与 CD8αβ 相比，CD8αα 是主要组织相容性复合体 I 类的不良辅助受体，表明它们可能像 γδ T 细胞一样直接识别抗原，而不是与主要组织相容性复合体 I 类结合。⁴对转录组数据进行进一步分析会很有趣因为它无疑提供了关于这些细胞激活状态的大量信息，以及它们产生或响应的细胞因子和趋化因子。

单细胞转录组学还阐明了 V（可变）、D（多样性）、J（连接）重组的性质，以确定哪些剪接事件在 γμ TCR 的两条链中更显着，并揭示了 CDR3（互补确定区域 3) 环，它提供了关于这些受体可能对什么类型的抗原作出反应的线索。该分析表明，抗原参与可能仅限于 TCRμ 链的 N 端可变域 (Vμ)，因为 Vμj 域是种系编码的，因此是不变的，与 TCRμ 配对的 TCRγ 链似乎也显示出比那些更少的多样性与 γδ T 细胞中的 TCRδ 配对。相比之下，Vμ 的 CDR3 环异常长，通常包含三个 D 段（与 αβ 和 γδ TCR 中典型的单个 D 段相反），

在 γμ T 细胞中发现的 TCRμ 和 TCRγ 链中，V (D) J 片段的使用受到限制，这表明这些 T 细胞是在抗原上选择的，或者存在其他限制多样性的因素。虽然大部分 γμ T 细胞具有 TCRγ 链剪接 Vγ2 和 Jγ3，但脾脏的 γδ T 细胞大多具有 Vγ3.2 和 Jγ1 剪接事件的 TCR，表明 γδ 和 γμ T 细胞可能具有不同的生物学功能。TCRμ 基因座包含五个簇，其中 V、D、J 和 C 片段可以剪接；然而，对大约 30 个 γμ T 细胞的询问仅发现使用第 3、5 和 7 组的证据，其中 57% 使用第 5 组，再次暗示这些 T 细胞识别的抗原多样性有限。在将这些发现扩展到所有 γμ T 细胞时必须小心，因为仅询问了脾 γμ T 细胞。⁵和 γμ T 细胞如果在其他组织中发现，也可能有同样的作用。

具有两个可变域和一个恒定域的 TCR 链如何与包含一个可变域和一个恒定域的 TCR 链异二聚化，这可以通过从脾脏 γμ T 细胞中分离的两个 γμ TCR 的代表性晶体结构来揭示。这些结构显示了一个典型的 TCR，具有从膜延伸的系链 Vμ 结构域，将 CDR 环从细胞向外定位，方向与它们在抗原结合中的假设作用一致。拉长的 CDR3 环看起来很灵活，因为一些残基的密度很弱。两种晶体结构在 VμjVγ 域顶部以不同的方式定向 Vμ 域，但这可能不是生理差异，因为每种排列都可能由晶体堆积强制执行，这在两种结构之间是非常不同的。更常见的 Vγ2 γμ TCR 异二聚体的晶体结构确实显示了 Vμ 和 Vγ2 之间以氢键形式的接触，并且许多关键接触描述了加强 Vμj 和 Vγ2 域之间的相互作用，这或许可以解释为什么这种配对受到青睐在大多数 γμ T 细胞中。再加上 Vμj 和 Vγ 域的多样性有限，这两个域似乎都没有对抗原结合有贡献，而是提供了一个支架来定向连接的可变 Vμ 域。

有袋动物和单孔目动物并不是唯一进化出一个小的抗原结合域的物种，该域被拴在熟悉的抗原识别结构上。软骨鱼具有类似的 TCR，称为 γδ-NAR，具有非常相似的结构。^第六名事实上，受限的 γδ TCR 配对为拼接到 TCRδ 转录本中的 NAR 域提供了理想的平台。骆驼科动物和鲨鱼的抗体也具有类似于 Vμ 域的抗原识别域。事实上，Vμ 和 Vμj 在结构上更类似于抗体 V 结构域，而不是来自 TCRδ 的 V 结构域。与暴露在溶剂中的 Vμ 结构域一致，存在极性残基代替通常埋在异二聚体界面中的疏水残基，这表明 Vμ 结构域可以作为可溶性结构域从其铰链中释放出来并用作有袋动物衍生的纳米抗体。

确定 TCRμ 的配对伙伴和 TCRγ 最常见的剪接排列（至少在脾脏中）为了解配体 γμ T 细胞响应的性质以及它们在有袋动物/单孔目免疫系统中可能扮演的角色奠定了基础由于 TCRα 和 β 链在有袋动物中在 TCRγ 链之前重新排列，⁷它们似乎不太可能是在成熟的适应性免疫系统建立之前产生的胎儿 T 细胞的例子。有人可能推测 γμ T 细胞在它们在母亲的育儿袋中发育时在保护有袋动物方面发挥着特殊的作用，但是在软骨鱼类中发现的类似受体是什么？奇怪的是，有袋动物和单孔目动物保持这第五条 TCR 链的选择压力是多少——包括我们在内的 Eutherian 哺乳动物是否错过了？