Extrathymic T-cell differentiation in vitro

Sirs

The first 'Innovation' article in Nature Reviews Immunology¹ discussed the introduction of a technique for the extrathymic generation of T cells in culture², which was suggested as being a useful new tool for studying T-cell differentiation in a way that was not previously possible^1,3. The originators of this technique showed that extrathymic differentiation from bone-marrow progenitors in vitro can be accomplished using the OP9 stromal cell line, ectopically expressing the Notch ligand delta-like 1 (Refs 1,2). It was stated that this allowed T-cell-differentiation processes to be studied in a simple cell-culture system for the first time. Although this might be true for mouse cells, it is not the case for human cells. Some years ago, we showed that CD34⁺ lineage-negative human haematopoietic progenitors could acquire mature T-cell characteristics in a thymus-free culture system. This depended on the presence of peripheral-blood mononuclear cells (PBMCs) as feeder cells, cytokine cocktails and the use of serum-free media. Given the importance of delta-like 1, identified in the mouse system, and the fact that Notch ligands, including this one, are expressed by antigen-presenting cells⁴, one could ask why previous attempts by others to establish extrathymic T-cell-differentiation systems using PBMCs were less successful^5,6,7. We suggest that the reason for such failures was probably the choice of culture medium and, to a lesser extent, the choice of cytokine cocktail and feeder cells. We tested many different formulations over the years, at first containing human serum or fetal-calf serum, with little success. We also tested many serum-free formulations, with equal lack of success. However, the use of the then newly developed X-Vivo 10 medium from BioWhittaker allowed development, for the first time, of T cells in the absence of thymic components in a limiting-dilution cloning system. It was necessary to include stem-cell factor (SCF), fms-related tyrosine kinase 3 ligand (FLT3L), interleukin-2 (IL-2) and IL-3 in the cytokine cocktail. Cloning efficiencies were markedly increased by additional inclusion of oncostatin M or IL-7 (Ref. 8). Feeder cells consisted of irradiated (at a dose such that no cells escaped proliferation blockade after irradiation) PBMCs pooled from 20 different donors. During the cloning procedure, these cells would have interacted with each other as a multi-way mixed-lymphocyte culture, resulting in cytokine release, and macrophage and dendritic-cell activation. The purity of the starting haematopoietic-cell population and the high cloning efficiency that could be achieved ruled out the possibility that the T cells isolated were derived from contaminants in the starting population. Derived T cells were not autoreactive, suggesting that negative selection had taken place in a system where the feeder cells did express MHC class II or class I molecules, unlike the OP9 mouse system recently discussed¹. Derived clones were mostly CD4⁺ αβ-T-cell receptor (αβ-TCR)⁺, expressed a wide range of antigen-receptor clonotypes, responded to mitogenic stimulation by proliferation and cytokine release⁸, and had the cell-surface markers and growth characteristics⁹ typical of human CD4⁺ T-cell clones. Currently, human T-cell clones derived in this manner are being examined¹⁰ by European Union-supported consortia focused on ageing of the immune system, ImAginE and T-CIA (see further information for websites). Perhaps because of our focus on immunogerontology, and exclusively in humans, these previous findings might have been overlooked.