T cell differentiation generates a wide range of specialized cell types; however, how the range of attainable cell states (i.e., phenotypes) is determined remains an open question. Given that T cell differentiation is tightly linked to TCR signal strength, the authors hypothesized that attenuation of downstream TCR signaling by CTLA-4 might also regulate differentiation. In order to verify this, the authors profiled lymphocytes derived from Ctla4-/- and littermate control mice via mass cytometry and subsequently used clustering methods to classify T cell populations. t-SNE visualization illustrated changes in T cell population frequencies and phenotypes in Ctla4-/- mice, but no differences between Ctla4+/+ and Ctla4+/- mice were apparent. Populations that were only observed in Ctla4-/- mice were entirely contained within the CD4+ compartment. These enriched CD4+ T cell populations expressed high amounts of CD44 and inducible co-stimulatory molecule (ICOS) but displayed variable levels of other activation markers and lineage transcription factors. For example, the frequency of terminally differentiated (TBET+EOMES+CD44+CD62L+) CD8+ T cells increased in Ctla4-/- mice. Similarly, an ICOS+ Treg cell population that is very minor in CTLA-4 competent mice expanded in frequency in Ctla4-/- mice, whereas other Treg cell populations remained stable or contracted. These findings contrasted with the effects on CD8+ T cells, in which loss of CTLA-4 modulated expression levels (e.g., EOMES, IRF4, and CXCR3), but no CD8+ T cell clusters, was detected only in Ctla4-/- mice. This suggests that CTLA-4 limits CD4+ T cell phenotypes but not those of CD8+ T cells. Next, the authors concurrently assessed changes in 5-iodo-deoxyuridine (IdU) incorporation and changes in frequency in T cell subsets due to loss of CTLA-4 which suggested that there was no correlation between the change in frequency of each T cell subset and the respective changes in IdU incorporation. The phenomenon revealed above raised a question that whether loss of CTLA-4-mediated negative feedback allows cells to occupy otherwise unattainable phenotypic states. Archetypal analysis, an unsupervised computational approach used to identify optimal phenotypes associated with specific biological phenomena at the organismal and single-cell level, helped researchers identified phenotypic boundary and reveals that loss of CTLA-4 expands CD4+ but not CD8+ T cell phenotypic space. Next, because ICOS was highly expressed by phenotypically expanded T cells and antibody blockade of ICOS attenuates T cell activity in Ctla4-/- mice, researchers tested whether ICOS is necessary for the generation of aberrant populations. As a result, all T cell populations identified in Ctla4-/-Icos+/- mice were present in Ctla4-/-Icos-/- mice, indicating that although ICOS is highly expressed on phenotypically expanded subsets, it is not required for their development.
immunized Ctla4-/- and littermate controls with ovalbumin in complete Freund’s adjuvant (CFA) indicated that extension of phenotypic limits derives directly from loss of CTLA-4 rather than indirect effects of its deletion. Then, analysis of TCRβ repertoire of thymic and LN tissue from Ctla4-/- and littermate control mice revealed that no major molecular defects in thymic T cell repertoire or the frequency of productive rearrangements were detected. In contrast, T cell clonality increased in Ctla4-/- LN tissue. Increased clonality was accompanied by decreased average CDR3 length, in part resulting from a decrease in N1 and N2 nucleotides inserted at V-D and D-J junctions. On the other hand, analysis of PD1 suggested that PD1 mainly affected CD8+ populations but the subtle expansion of CD8+ T cell phenotype limits due to loss of PD-1 appears to be age dependent and most likely relies on the prior induction of an immune response. Finally, the authors used mass cytometry data to reconstruct differentiation trajectories between the naive CD4+ T cell archetype and each differentiated CD4+ T cell archetype. Along the trajectory, the authors investigated that a subset of markers (such as PD1) expression continued to increase beyond the boundary which suggested that a primary mechanism by which CTLA-4 restrains T cell phenotypes is through imposing maximal limits on already active pathways.
In summary, this paper revealed that (1) T cell negative co-stimulation regulates the limits of T cell differentiation, a role related to but distinct from its regulation of T cell activation; (2) CTLA-4 enforces boundaries on CD4+ T cell phenotypes; (3) PD-1 subtly restrains CD8+ T cell phenotypes; and (4) CTLA-4 constrains T cell phenotypes by imposing maximal expression limits during differentiation. These finding support the “nuanced model” in which T cell subsets represent a continuum of interrelated phenotypic states, defined by ratios of co-expression of lineage transcription factors, rather than discrete lineages. It remains to be assessed whether the expanded phenotypes represent stable terminally differentiated states or rather transient intermediate states not normally attainable.