Moving T memory stem cells to the clinic

In this issue of Blood, Cieri and colleagues elucidate the instructive signals that guide naive T lymphocytes to differentiate into memory stem cells (T_SCM) in vitro, paving the way for a rapid translation of T_SCM into new clinical trials.¹ 

Memory T lymphocytes play a critical role in the immune response against infectious disease and cancer.^2,3  The memory T-cell compartment is heterogeneous but it can be divided into 3 main subsets, with the recently identified T memory stem cells (T_SCM) joining the well-established central memory (T_CM) and effector memory (T_EM) T-cell subsets (see figure).^3-7  T_SCM are capable of reconstituting the full diversity of memory and effector T lymphocytes on serial transplantation in mice, indicating that these cells are endowed with the stem cell–like attributes of self-renewal capacity and multipotency.⁵  These qualities make T_SCM a particularly attractive subset to employ in adoptive immunotherapies for cancer as they might overcome current limitations, including inefficient T-cell engraftment, poor persistence, and inability to mediate a prolonged immune attack. Indeed, T_SCM displayed robust proliferative and survival capacities and were superior to other memory T-cell subsets in causing regression of large, established tumors in preclinical models.^5,6  While the instructive signals guiding the formation of T_SCM have just begun to be investigated,⁵  clinical-grade protocols to facilitate translation of T_SCM into the clinic remain undefined.

Cieri and colleagues now demonstrate that it is possible to generate, expand, and gene engineer T_SCM in vitro starting from naive precursors using reagents that are currently manufactured under GMP-quality conditions. The authors found that activation of naive T cells with anti-CD3 and anti-CD28 antibody-conjugated beads in the presence of low doses of IL-7 and IL-15 promoted the generation of CD45RA⁺CD62L⁺CCR7⁺ CD95⁺ T cells phenotypically resembling the T_SCM cell population recently described by our group (see figure).⁶  Cells generated under these conditions, however, also expressed CD45RO, a prototypical marker of memory and effector T lymphocytes, which is absent from the surface of T_SCM as originally reported.⁶  This phenotypic discrepancy likely resulted from differences in the expression of HNRPLL (encoding heterogeneous nuclear ribonucleoprotein L-like),⁶  a key regulator of the alternative splicing of the CD45 pre-mRNA required for efficient CD45RO expression. This difference places IL-7/IL-15–generated cells in a more differentiated position in the spectrum of differentiation (see figure). However, in vitro–generated T_SCM closely clustered with naturally occurring T_SCM when analyzed using a set of 65 genes differentially regulated between naive and memory T lymphocytes. Furthermore, in vitro–generated T_SCM displayed an enhanced proliferative capacity on adoptive transfer into immunodeficient mice, a finding consistent with those using naturally occurring T_SCM.⁶  Most importantly, Cieri et al in a brilliant set of experiments showed for the first time that T_SCM were the only T-cell subset capable of expanding and mediating GVHD on serial transplantation. These findings represent the most compelling evidence that T_SCM preferentially retain stem cell–like attributes among all human T lymphocytes and position researchers in the field to formally test the “stemness” of T_SCM in human clinical trials.

By dissecting the relative contribution of the biologic signals required for the in vitro generation of T_SCM, the authors found that IL-7 was indispensable for the formation of these cells, while IL-15 was necessary to sustain their expansion. In vivo, elevated levels of IL-7 and IL-15 are found in hosts receiving lymphodepleting conditioning regimens for hematopoietic stem cell (HSC) transplantation or adoptive T-cell immunotherapy as a result of transient eradication of cellular sinks for homeostatic cytokines.⁸  Consistent with this notion, Cieri and colleagues found that after allogeneic HSC transplantation, virtually all seemingly naive T cells were actually T_SCM, because CD45RA⁺CD62L⁺ T cells also expressed high levels of CD95. Although the instructive signals guiding T_SCM cell formation during physiologic immune responses are yet to be elucidated, these findings provide the first glimpse on how T_SCM can be formed in vivo in a clinically relevant setting.

The identification of clinically compliant conditions for the efficient generation and genetic manipulation of T_SCM also has important implications for the development of new T cell–based immunotherapies. Despite the potent anti-tumor activity of T_SCM in preclinical animal tumor models,^5,6  it is currently not feasible to treat patients with naturally occurring T_SCM because these cells represent only a small portion of circulating lymphocytes. Adoptive immunotherapy may require larger number of transferred cells than can be obtained from the naturally occurring T_SCM compartment. Therefore, the identification of strategies that generate, expand, and enable redirecting of T_SCM against cancer cells is crucial. We have previously shown that programming naive T cells in the presence of small molecules targeting the Wnt/β-catenin pathway, such as glycogen synthase-3β (GSK-3 β) inhibitors, promotes the generation of T_SCM (see figure).^5,6  Although inhibitors of GSK-3β are effective at arresting T-cell differentiation, they also inhibit T-cell proliferation.⁵  For this reason, finding alternative approaches that uncouple cell expansion and differentiation is desirable. Cieri and colleagues now describe how priming T cells in the presence of low doses of IL-7 and IL-15 can generate larger numbers of T_SCM than previously reported with GSK-3β inhibitors.

Finally, the findings reported here by Cieri et al provide new experimental evidence that helps to resolve the ongoing debates regarding the ontogeny of memory cells⁹  and which T-cell subset needs to be isolated to generate more potent anti-tumor T cells for human clinical trials.¹⁰  Whole genome profiling as well as phenotypic and functional data described in this issue of Blood¹  are consistent with a linear model of differentiation in which naive T cells differentiate first into T_SCM cells and then into T_CM and T_EM.^2,3  Consistent with this model and a progressive loss of therapeutic potential with differentiation (see figure),³  T cells expanded from sorted T_CM were less potent at mediating GVHD compared with naive-derived T cells and were unable to reconstitute immunodeficient animals on serial transplantation. These data indicate that T_CM-derived T cells are relatively ineffective and suggest that new adoptive immunotherapies will greatly benefit from the generation of tumor-reactive T_SCM from sorted naive precursors.