The mechanism of action of the lymphocyte checkpoint protein LAG-3 was always rather mysterious. It now seems to operate at least in part by recognizing and suppressing responses to stable complexes of peptide and major histocompatibility complex class II.
LAG-3 was discovered almost 30 years ago by Triebel and colleagues1 in the form of a transcript expressed by a cytokine IL-2–dependent natural killer cell line that encoded a protein that looked a lot like the co-receptor CD4. The extracellular region of LAG-3 has four immunoglobulin superfamily domains, and although it shares only very limited similarity with CD4 (~26%), sequences next to the cysteine residues in the first and third domains, along with the chromosomal locations of the genes encoding these molecules, indicate that the genes encoding LAG-3 and CD4 share an ancient ancestor. The homology to CD4 immediately suggested that LAG-3 would bind major histocompatibility complex (MHC) class II, which was quickly confirmed by Triebel and colleagues in cell-binding assays2. However, compared with CD4, LAG-3 has always been a bit of an oddball. It is expressed not only on activated CD4+ T cells but also on activated CD8+ T cells, natural killer cells and myeloid cells, none of which would otherwise be interested in MHC class II (reviewed in ref. 3). Also in contrast to CD4, LAG-3 blockade prolongs rather than limits T cell responses4, and whether this is due to competition with CD4 for MHC class II has remained unclear. Finally, the signaling effects of LAG-3 rely on its cytoplasmic domain, a wholly unique structure completely lacking conventional signaling motifs5. New work by Maruhashi et al. now reveals LAG-3 to be even more idiosyncratic: it inhibits T cell responses, but only those driven by especially stable complexes of peptide and MHC class II6.
Maruhashi et al. set out to re-evaluate the binding specificity of LAG-3 using a pentameric form of the mouse receptor as a staining reagent in flow cytometry6. To their surprise, they find that although the binding of LAG-3 correlates to some extent with the expression of MHC class II, there are striking exceptions. Large fractions of conventional dendritic cells, most mature bone marrow–derived B cells and nearly all plasmacytoid dendritic cells bind little or no LAG-3, despite expressing substantial amounts of MHC class II. Maruhashi et al. then go on to show that cell lines presenting exogenous, classical MHC class II–binding peptides elicit comparable T cell responses regardless of whether they bind the LAG-3 pentamer or not, but only responses to those cell lines that bind the pentamer are inhibited by T cell–expressed LAG-36. This suggests either that there is a non–MHC class II ligand for LAG-3 or that the binding of LAG-3 to MHC class II relies on some co-factor.
Maruhashi et al. turn to expression cloning to search for the alternative ligand for LAG-3 or the binding-permissive co-factor. Tellingly, what they isolate is the gene Ciita. This gene encodes an interferon-γ (IFN-γ)-inducible transcriptional co-activator of not just MHC class II but also, among others, proteins involved in the assembly and peptide editing of MHC class II: Ii and DM. Although this does not completely exclude the possibility that there is a non–MHC class II ligand for LAG-3 still to be discovered (CRISPR-based searches7 are best for this), it is strong circumstantial evidence that MHC class II is the dominant player in interactions with LAG-3. Moreover, Maruhashi et al. are able to explain all their observations in terms of the role of CIITA in creating stable peptide–MHC class II (pMHCII) complexes6. They show that ectopic Ciita expression dramatically increases the binding activity of antigen-presenting cells that do not bind LAG-3 and restores their ability to suppress the activation of LAG-3-expressing T cells (Fig. 1a). Conversely, deletion of Ciita or the genes encoding Ii and DM completely eliminates the binding of LAG-3 while only partially reducing the expression of MHC class II. Finally, Maruhashi et al. establish a direct link between the binding of LAG-3 and the stability of pMHCII complexes by showing that MHC class II expressed with covalently attached peptides (i.e. super-stable pMHCII complexes) binds strongly to LAG-3 in a CIITA-independent manner, whereas complexes formed with variants of these peptides disrupted at their anchor residues, or complexes formed with other low-affinity peptides, do not6. In other experiments they produce the clearest evidence yet that LAG-3 does not block CD4–MHC class II or T cell antigen receptor–MHC class II binding and provide yet more evidence that LAG-3 relies on its cytoplasmic domain to suppress T cell activation.
In this way, Maruhashi et al. convincingly show that LAG-3 preferentially suppresses the activation of T cells and, presumably, other LAG-3-expressing cells, including CD8+ T cells, by binding to MHC class II proteins presenting only stably bound peptides6. The questions now are as follows: how does LAG-3 distinguish between stably bound peptides and unstably bound peptides, and why? Peptide-dependent conformational variation in MHC class II proteins was first detected as changes in their migration during SDS-PAGE8. Comparisons of stable HLA-DR–peptide complexes with DM-stabilized HLA-DR in complex with a low-affinity peptide9 have revealed that stable peptide binding triggers conformational changes at one end of the peptide-binding helix of HLA-DRα and the floor of the peptide-binding groove next to it. By binding to this part of MHC class II, LAG-3 might be able to ‘read’ complex stability. This region is opposite the CD4 binding site (Fig. 1b), which perhaps explains the ability of CD4 and LAG-3 to bind simultaneously to MHC class II, as shown by Maruhashi et al.6. It is nevertheless surprising that these molecules can do this, given their shared history. This is because it implies either that one of the orthologs lost binding of MHC class II to the shared site and regained it ab initio at the second site or that it somehow worked its way around to the other side of the molecule, two scenarios that seem equally unappealing. The structure of the LAG-3–MHC class II complex, when it is known, will be very interesting.
Understanding why LAG-3 is ligand selective is also something for the future. The IFN-γ- and CIITA-dependent generation of stable MHC class II complexes ensures that LAG-3 only ever exerts its inhibitory effects in the context of ongoing immune responses. However, this requirement is already fulfilled by LAG-3’s expression being activation dependent, like that of other immune checkpoint molecules. It can also be argued from a structural viewpoint that it is not so surprising that LAG-3 is restricted to binding to the final, stably folded form of MHC class II. Most protein interactions rely on interfaces formed by large arrays of atoms positioned stably in three-dimensional space. Simply by producing folding intermediates capable of reaching the cell surface, MHC class II’s singular maturation pathway might allow LAG-3 to exhibit apparent selectivity. However, none of this is to say that the selective ligation of MHC class II by LAG-3 will not have important implications for immunological function. Maruhashi et al. propose that autoimmunity could be elicited by T cells that escape negative selection mediated by DM-edited peptides in the thymus and cannot be suppressed by LAG-3 in the periphery because they react with unstable pMHCII complexes6. In support of that idea, the authors note that CD4+ T cells capable of responding to unstable peptides are known to be highly diabetogenic in non-obese diabetic mice10.
There is now great interest in LAG-3, which, along with the receptors TIM-3 and TIGIT, forms part of a second wave of immune checkpoint targets because it is expressed alongside the immunoregulatory receptor PD-1 on tumor-infiltrating lymphocytes and is associated with T cell exhaustion3. Will the work of Maruhashi et al.6 alter LAG-3’s status as a target for immunotherapy? Probably not: the 24 ongoing or completed trials of LAG-3-based biologicals (as of 23 September 2018; https://clinicaltrials.gov/) will settle this matter. Partly due to the success of antibodies to immunoregulatory receptors PD-1 and CTLA-4, but mostly because there are now so many trials (over 1,100 in the case of PD-1), many other immunotherapy targets and combinations are unlikely ever to be triaged in this way. Careful studies of immune checkpoints, especially the strange ones, in the manner of Maruhashi et al.6 will need to underpin candidate selection, with the bonus that there will be yet more surprises.
Triebel, F. et al. J. Exp. Med. 171, 1393–1405 (1990).
Baixeras, E. et al. J. Exp. Med. 176, 327–337 (1992).
Le Mercier, I., Lines, J. L. & Noelle, R. J. Front. Immunol. 6, 418 (2015).
Huard, B., Tournier, M., Hercend, T., Triebel, F. & Faure, F. Eur. J. Immunol. 24, 3216–3221 (1994).
Workman, C. J., Dugger, K. J. & Vignali, D. A. J. Immunol. 169, 5392–5395 (2002).
Maruhashi et al. Nat. Immunol. https://doi.org/10.1038/s41590-018-0217-9 (2018).
Sharma, S., Bartholdson, S. J., Couch, A. C. M., Yusa, K. & Wright, G. J. Genome Res. 28, 1372–1382 (2018).
Sadegh-Nasseri, S., Stern, L. J., Wiley, D. C. & Germain, R. N. Nature 370, 647–650 (1994).
Pos, W. et al. Cell 151, 1557–1568 (2012).
Mohan, J. F., Petzold, S. J. & Unanue, E. R. J. Exp. Med. 208, 2375–2383 (2011).
Wang, X. X. et al. Proc. Natl Acad. Sci. USA 108, 15960–15965 (2011).
The authors declare no competing interests.
About this article
Cite this article
Lui, Y., Davis, S.J. LAG-3: a very singular immune checkpoint. Nat Immunol 19, 1278–1279 (2018). https://doi.org/10.1038/s41590-018-0257-1
Expert Opinion on Pharmacotherapy (2021)
Immune profiling of pituitary tumors reveals variations in immune infiltration and checkpoint molecule expression
OncoTargets and Therapy (2020)
Mechanisms of resistance to immune checkpoint inhibitors and strategies to reverse drug resistance in lung cancer
Chinese Medical Journal (2020)
The EMBO Journal (2020)