Three lines of inquiry

For two decades the effector CD4 compartment was described as a pair. T helper 1 cells, making interferon-γ, dealt with intracellular pathogens; T helper 2 cells, making interleukin-4, dealt with parasites and allergy. Autoimmune inflammation was assigned to the first of these almost by default. The assignment never fit especially well, and in 2005 this laboratory reported why: a third population of effector cells that made neither cytokine, produced interleukin-17 instead, and arose by a developmental route independent of both established lineages.

What followed was a decade of asking how such a cell is made. The inducing signal turned out to be transforming growth factor-β, a cytokine better known for restraint than for inflammation — a finding that placed the Th17 lineage and the regulatory T-cell lineage on adjacent branches of the same developmental tree, sharing an inducer and diverging on the strength of additional signals.1 Later work traced the transcriptional and chromatin logic underneath: how interleukin-6 signaling must persist for a Th17 cell to retain its identity, how the transcription factor Batf pioneers the reorganization of chromatin in developing effector cells, and how cis-regulatory elements scattered across the Ifng and Il17 loci are used differently depending on the stimulus.

A lineage, it emerged, is not a destination. Th17 cells are unusually plastic: under the right pressure they convert to interferon-γ-producing cells, and in models of colitis it is those converted cells — Th1 cells of Th17 origin — that are required for pathology. The instability is not a laboratory artifact but a property of the loci themselves, which remain epigenetically poised rather than locked. This is why the lab tends to speak of programs rather than types, and why so much of its method is built to follow single cells over time instead of classifying them once.

The same instinct led to the finding that a cell's own interleukin-2 — not merely the interleukin-2 it receives — helps decide whether it becomes a follicular helper cell or an effector destined for tissue. Fate, in these systems, is rarely imposed from outside. It is negotiated.

The intestine is the body's largest immune organ, and the only one obliged to remain tolerant of the population it contains. Effector and regulatory CD4 T cells develop there under the influence of the microbiota, of dietary and microbial metabolites, and of nuclear receptors that read them: retinoic acid, vitamin D, bile acids sensed through the constitutive androstane receptor. The lab's interest is less in cataloguing these signals than in learning what they are for.

The clearest answers have come from a single, well-behaved model of infection. Citrobacter rodentium is a natural mouse pathogen that attaches to the distal colonic epithelium by the same lesions enteropathogenic E. coli uses in humans.2 It is cleared, over weeks, by a type 3 immune response — which makes it an unusually legible timeline for asking what each cell contributes, and when.

Read across that timeline, the cytokines of the type 3 response look less like alarms than like instructions addressed to the epithelium. Interleukin-22 from T cells proved non-redundant in the antibacterial defense of colonic crypts: innate sources could not substitute for it, and without it the crypts were breached. Interleukin-17 and interleukin-22 are, in this reading, a vocabulary for telling a barrier what to do.

The reply is the more surprising half. Colonocytes attacked by C. rodentium recruit T-cell help specifically to themselves — a degree of address that diffusion alone cannot explain, and that implies the epithelial cell is a participant in the exchange rather than its subject. Whether the epithelium presents antigen to CD4 T cells on its own class II MHC, under what conditions it does so, and what that presentation accomplishes, is an open question the laboratory is pursuing with conditional genetics and single-cell transcriptomics across the timeline above.3

Memory is the last chapter of the same story. Some of the cells raised against an intestinal pathogen remain in the tissue afterward, resident and locally poised, and a colon that has been infected is not immunologically identical to one that has not. Following those cells is what the day-42 column of the schedule is for.

Nothing in the biology above is exclusive to health. Th17 cells that defend the colon against infection are the cells implicated in inflammatory bowel disease; the interleukin-10 that restrains them is, when lost, the difference between a healthy gut and a colitic one; the chronic inflammation that follows can end in colorectal cancer. The laboratory's disease work is not a separate program but the same program run in failure conditions.

Three threads run through it. The first is conversion: the observation that pathology in colitis requires Th17 cells to become Th1-like, which reframes therapeutic targeting as a question of stability rather than of abundance. The second is restraint — the regulatory arm, and the T-cell-derived interleukin-10 that proves necessary for immune surveillance of the small intestine and for the control of adenomatous polyps. The third is the beginning of life: dysbiosis of the neonatal intestinal microbiome precedes late-onset sepsis in newborns, and preventing it protects the animal.4

The ambition throughout is precision rather than suppression. Broad immunosuppression works, at a cost measured in infection and malignancy; it works because it silences a conversation the tissue needs. The alternative is to understand the conversation well enough to change one sentence of it — to know which signal, at which moment, decides whether a T cell defends the border or destroys it.