Unraveling the effect of proliferative stress in vivo in hematopoietic stem cell gene therapy mouse study

The hematopoietic system of patients enrolled in hematopoietic stem cells (HSC) gene therapy (GT) treatments is fully reconstituted upon autologous transplantation of engineered stem cells. HSCs highly proliferate up to full restoration of homeostasis and compete for niche homing and engraftment. The impact of the proliferation stress in HSC on genetic instability remains an open question that cured patients advocate for characterizing long-term safety and efficacy. The accumulation of somatic mutations has been widely used as a sensor of proliferative stress. Vector integration site (IS) can be used as a molecular tool for clonal identity, inherited by all HSC progeny, to uncover lineage dynamics in vivo at single-cell level. Here we characterized at single-clone granularity the proliferative stress of HSCs and their progeny over time by measuring the accumulation of mutations from the DNA of each IS. To test the feasibility of the approach, we set-up an experimental framework that combines tumor-prone Cdkn2a-/- and wild type (WT) mouse models of HSC-GT and molecular analyses on different hematopoietic cell lineages after transplantation of HSCs transduced with genotoxic LV (LV.SF.LTR) or GT-like non-genotoxic LV (SIN.LV.PGK). The Cdkn2a-/- mouse model provided the experimental conditions to detect the accumulation of somatic mutations, since the absence of p16INK4A and p19ARF enhances the proliferative potential of cells that have acquired oncogenic mutations. As expected, mice transplanted with Cdkn2a-/- Lin- cells marked with LV.SF.LTR (N=24) developed tumors significantly earlier compared to mock (N=20, p<0.0001), while mice treated with SIN. LV.PGK (N=23) did not. On the other side, mice that received WT Lin- cells treated with LV.SF.LTR (N=25) or SIN.LV.PGK (N=24) vector have not developed tumors. Given this scenario, we expect that Cdkn2a-/- Lin- cells transduced with LV.SF.LTR are associated with higher mutation rates compared to the SIN.LV.PGK group and wild type control mice. The composition of peripheral blood, lymphoid (B and T) and myeloid compartments was assessed by FACS on samples collected every 4 weeks and IS identification. More than 200,000 IS have been recovered. To identify the presence of somatic mutations, the genomic portions of sequencing reads flanking each different IS were analyzed with VarScan2. The accumulation rates of mutations have been evaluated by our new Mutation Index (MI) which normalizes the number of mutations by clones and coverage. Considering that a large portion of IS has been discarded since not covered by a minimum number of 5 unique reads (genomes), the remaining number of IS contained >90% of reads in each group. The MI increased over time in both LV.SF.LTR groups, with higher values for the Cdkn2a-/-. On the other hand, treatment with SIN.LV.PGK resulted in lower MI in both groups compared to LV.SF.LTR groups, reflecting the higher clonal composition of the cells treated with the SIN.LV.PGK and the phenomenon of insertional mutagenesis in the LV.SF.LTR. Moreover, the higher MI values of the SIN.LV.PGK Cdkn2a-/- group compared with the WT group proved the induction of DNA fragility. Our results showed that the analysis of the accumulation of somatic mutations at single clone unraveled HSC proliferation stress in vivo, combining for the first time the analysis of acquired mutations with IS. We are now applying our model to different clinical trials, and studying HSCs sub- clonal trees by symmetric divisions, previously indistinguishable by IS only. Our study will open the doors to in vivo long-term non-invasive studies of HSC stability in patients.