In Vivo Tracking of T Cells in Humans Unveils Decade-Long Survival and Activity of Genetically Modified T Memory Stem Cells

A deeper understanding of T lymphocytes survival and differentiation potential in humans is paramount for the development of effective gene/cell therapies based on T-cell engineering. We here performed a comprehensive study of T-cells dynamics and plasticity in humans by a unique combination of phenotypic/functional studies and high-throughput integration sites (IS) analyses. We analyzed samples from hematopoietic stem cells (HSC) (n=10) or mature lymphocytes (PBL) gene therapy (GT) (n=4) treated ADA (adenosine deaminase) deficient-SCID patients. For comparative analyses, we also collected data from pediatric (n=19) and adult (n=52) healthy donors (HD), and from bone marrow transplanted patients (BMT) with primary immunodeficiencies (n=10, 4 with ADA-SCID).

We observed that vector-positive CD62L+/CD45RA+ putative T naïve cells were detectable 12 years after last infusion of gene-corrected lymphocytes in peripheral blood of PBL-GT patients that lack the support of transduced lymphocytes precursors. We then unveiled that the vast majority of these CD62L+/CD45RA+ cells (80.3%) in PBL-GT patients could be actually classified phenotypically (CD95, IL2Rβ and IL7Rα surface expression) and functionally (IFNγ production and aCD3/aCD28 in vitro differentiation) as active long-lasting T memory stem cells (Tscm).

The peculiar Tscm frequency found in PBL-GT patients was most likely due to a combinatorial in vitro and in vivo effect. Indeed, by a series of in vitro assays, we showed that Tscm relative enrichment in CD45RA+CD62L+ compartment have occurred during the in vitro manipulation of T cells before infusion. Additionally, we found higher-then-normal Tscm contribution among CD45RA+/CD62L+ cells even in ADA-SCID patients receiving HSC-GT and BMT, suggesting a role of disease background on in vivo Tscm persistence. Analyzing our cohorts of healthy donors and treated individuals we were able to further correlate Tscm contribution in vivo with age, conditioning regimen, disease background, cell source, and long-term T-cell reconstitution.

One unique aspect of our study consisted in the opportunity to track Tscm clonal dynamics in vivo in humans since each gene-corrected cell infused in our GT patients is univocally and permanently tagged by a retroviral integration site.To perform in vivo molecular tracing of individual T-cell clones we sorted T naïve, Tscm, central memory and effector memory subtypes. We then collected from these subpopulations, by LAM-PCR+Illumina-Miseq sequencing, 2.584.137 integration sites (IS) sequences mapped to 1.746 unique chromosomal positions, corresponding to 910 integrations from 5 HSC-GT patients in vivo, 79 integrations from 2 PBL-GT samples of transduced cell products prior to infusion and 754 integrations from 4 PBL-GT patients in vivo.

Firstly, to establish a relationship between precursors and terminally differentiated T cells we searched for the presence of identical insertion sites detected in multiple T-cell subtypes, applying stringent analytical filters for cross-contaminations. Strikingly, the level of shared integrations in each subtype was directly correlated to its stage of differentiation with Tscm, isolated from PBL-GT patients, showing the highest proportion of integration sites shared with the other T-cell subsets. Importantly, the results of the same analysis performed on HSC-GT patients were outstandingly coherent with the progressive developmental model of memory T-cell differentiation.

We then assessed the survival of individual Tscm clones by performing a longitudinal IS analysis of different T-cell subtypes isolated from 3 PBL-GT patients over a 2 to 5 years timeframe up to 12 years after last infusion. We were able to formally prove the persistence of individual Tscm by re-capturing identical IS tagging specific Tscm clones in two independent timepoints in a 5- years window. Importantly, the same IS were also detected in multiple T-cell subtypes, representing the best indirect evidence that these clones were endowed with long-term precursor activity. We also documented, by IS sequencing reads, the long-term polyclonal composition of each subtype and we did not observe enrichment for IS flanking proto-oncogenes. Overall, this study validates, for the first time in humans, the safe and functional decade-long survival of engineered Tscm, paving the way for their future application in clinical settings.