Unconventional T Cells: Already Armed, and No MHC Required

Most immunology is written as though a T cell must first be told what to look for. Three lineages did not read that memo. γδ T cells, invariant NKT cells and MAIT cells leave the thymus with their effector programme already written, their transcription factors already committed, and their receptors pointed at things that are not peptides at all — phosphoantigens, lipids and a vitamin B2 degradation product. They are between 1 and 10 per cent of blood T cells and up to half the T cells in liver and gut mucosa, and in the first hours of an infection they are usually the ones producing the cytokine. If your model of T-cell biology starts with antigen presentation and clonal expansion, these cells will keep producing results you cannot place.

The three unconventional T-cell lineages: their non-MHC restriction molecules, the transcription factors that fix their fate in the thymus, the shared cytokine-receptor route that activates them without any antigen at all, and the common effector output. Open the interactive version to click any protein for its role and the matching validated reagent.

Three restriction molecules, none of them MHC

The defining feature of this group is what presents to them. BTN3A1 (CD277) is not a presenting molecule in the classical sense at all: it binds pyrophosphate metabolites — microbial HMB-PP, or the host isopentenyl pyrophosphate that accumulates when the mevalonate pathway is dysregulated in tumour cells — on its intracellular B30.2 domain, and reports that binding through a conformational change on the outside of the cell. The Vγ9Vδ2 T-cell receptor reads the shape of BTN3A1, not the antigen. This is why aminobisphosphonates such as zoledronate expand Vγ9Vδ2 cells: they block farnesyl pyrophosphate synthase, IPP piles up inside the cell, and an otherwise unremarkable target starts signalling that its isoprenoid metabolism is wrong.

CD1d is a genuine presenting molecule with a deep hydrophobic groove built for lipids rather than peptides. Its canonical ligand, α-galactosylceramide, is a marine-sponge glycolipid that no mammal makes, which is a good reminder that the physiological ligand set — microbial diacylglycerols, self glycosphingolipids, lysophospholipids — is still incompletely mapped. MR1 is the strangest of the three: it presents 5-OP-RU, an unstable adduct formed between a riboflavin biosynthesis intermediate and a host aldehyde. Because only bacteria and yeast make riboflavin de novo, MR1 is effectively a sensor for "a live microbe with an active vitamin B2 pathway is here". MR1 is also almost monomorphic across humans, so unlike everything downstream of HLA, a MAIT-cell experiment reproduces across donors.

The practical consequence of all three is the same, and it is the reason this map exists: none of these recognition systems is defeated by MHC loss. A tumour that has deleted β2-microglobulin to escape CD8 T cells and checkpoint blockade is still fully visible to a Vγ9Vδ2 cell reading its isoprenoid metabolism.

The fate is fixed before the cell has ever seen an antigen

Conventional T cells acquire their effector identity in the periphery, days after activation, shaped by whatever cytokines were present. These three lineages acquire theirs in the thymus. PLZF (ZBTB16) is the master regulator of the innate-like programme — forced PLZF expression is enough to give a conventional T cell an effector-memory phenotype and immediate cytokine capacity — and it is expressed by iNKT cells and by MAIT cells, and by the T-bet-high and RORγt-high subsets that mirror Th1 and Th17 within the γδ compartment. What you get is a cell that is transcriptionally poised: the Ifng and Il17a loci are already in a permissive chromatin state, so protein appears within one to two hours of stimulation rather than the two to three days a naive conventional T cell needs.

The nomenclature follows the same split. NK1.1 and NKG2A on iNKT cells, CD161 (KLRB1) at very high density on MAIT cells, CXCR6 and CCR6 directing them to liver and mucosa respectively. This is why these populations are defined by surface phenotype and tissue address rather than by specificity — their specificity is invariant, so it carries no information about the individual cell.

The activation route most people miss: no antigen required

This is the single most important thing on the map for anyone designing an experiment. IL-12 plus IL-18 will drive full IFN-γ production from MAIT and iNKT cells with the T-cell receptor entirely uninvolved. IL-23 plus IL-1β does the same for the IL-17A programme. Because IL-18R1 is constitutively high on these cells and the transcription factors are already in place, cytokine alone is a complete activation signal.

Two things follow. First, a positive readout in a co-culture does not demonstrate antigen recognition, and reporting it as such is one of the commoner errors in this literature: you need a blocking antibody against the restriction molecule, or a receptor-null control, to separate TCR-dependent from cytokine-driven activation. Second, it explains the clinical pattern — these cells are activated in sepsis, in severe COVID-19 and in autoimmune flares where no cognate antigen is plausible, purely as a function of the inflammatory cytokine milieu. MAIT-cell "activation" in a patient cohort is frequently a measurement of that milieu. The shared surface machinery — CD3, CD2, CD28 and CD127 — is real and functional, but it is not the only way in.

One effector output, two directions of clinical travel

Downstream the three lineages converge. Granzyme B and perforin for cytotoxicity; IFN-γ, TNF, IL-17A, IL-22 and GM-CSF as the soluble output; IL-10 as the counter-regulatory arm. The interesting asymmetry is that the same cell type is protective or pathogenic depending almost entirely on which of two cytokine programmes it runs. IFN-γ-producing γδ T cells are associated with better outcomes in most solid tumours; IL-17-producing γδ T cells recruit neutrophils, promote angiogenesis and correlate with worse ones. It is the same lineage, and the switch is upstream in the cytokine environment.

They exhaust, too, and faster than is often appreciated. Repeatedly stimulated Vγ9Vδ2 cells upregulate PD-1 and lose cytotoxic function, which is the standing explanation for why the early zoledronate-plus-IL-2 expansion trials produced large numbers of circulating γδ T cells and very few responses. The current generation of protocols therefore measures the checkpoint alongside the cytokine rather than counting cells.

That asymmetry is what makes these cells attractive therapeutically and difficult to deploy. Because their receptors are invariant and not HLA-restricted, allogeneic γδ and iNKT cell products can be manufactured from healthy donors and given off the shelf without the graft-versus-host risk that limits allogeneic conventional T cells — and γδ T cells do not cause GvHD even in a fully mismatched setting. The hard part is not making them. It is keeping them on the IFN-γ arm once they are in a tumour that is signalling IL-23 and IL-1β.

What to measure, and in what order

The In Vivo angle

Sixteen of the 36 nodes on the interactive map carry a functional-grade In Vivo antibody — low or ultra-low endotoxin, and formulated for administration rather than staining. That includes the receptors you would use to deplete or block a lineage (pan-TCRγδ, Vγ2, the Vα24-Jα18 iNKT receptor, NK1.1, NKG2A), the shared costimulatory and cytokine-receptor set (CD3, CD2, CD28, CD25, CD127), the CD4/CD8α subset tools, and IL-2, TNF and GM-CSF for neutralisation. Endotoxin control matters more here than almost anywhere else in immunology: contaminating LPS drives IL-12, IL-18, IL-23 and IL-1β from any myeloid cell in the well, and those four cytokines are precisely the antigen-independent activation route described above. A research-grade blocking antibody will manufacture the exact readout you are trying to abolish.

The one-line version

These are not oddities at the edge of T-cell biology. They are a parallel, pre-committed, MHC-independent arm of it that responds in hours rather than days, dominates the T-cell compartment in the tissues where infections actually begin, and can be activated by cytokine alone. Design the cytokine-only control first; everything else in the experiment depends on it.

For the IL-23-driven side of the IL-17 programme in conventional T cells, see the Th17 / IL-23 axis map; for the tissue-residency and mucosal addressing that governs where these cells sit, see mucosal homing and tissue-resident memory.

For research use only. Not for use in diagnostic or therapeutic procedures.