A Double Fail-Safe Approach to Prevent Tumorigenesis and Select Pancreatic β Cells from Human Embryonic Stem Cells

摘要

class="head no_bottom_margin" id="sec1title">IntroductionClinical trials have already been initiated for the treatment of type 1 diabetes with hESC-derived β cell progenitors (), and newer approaches with functional β-like cells are lining up (, ). However, for all their promise, hESCs are not without limitations. While the proportion of β-like cells obtained is relatively high (30%–40%) (, ), a significant percentage of undefined cell types are also generated (, ). Most importantly, non-differentiated escapees may produce teratomas upon transplantation (). Although recent protocol refinements have reduced this risk, the foreseeable implementation of hESC-based therapies in thousands of patients calls for caution. One single case of hESC-derived tumors may set the field back for years.To address these concerns, we set out to engineer hESC lines with control mechanisms allowing for the selective ablation of both tumorigenic cells and cells differentiated along undesirable fates. We hypothesized that constructs of our invention would impart such selectivity when stably integrated within the host genome. Given its clinical immediacy, we decided to focus on an insulin-producing cell differentiation model, even though our system could be customized for any cell type. In such a model, cells are engineered with two suicide gene cassettes. One of them (herpes simplex virus thymidine kinase, or HSV-TK) is activated only in cells that resume/continue tumorigenic proliferation. The second (nitroreductase [NTR]) is active in all cells unless they express insulin, in which case it is permanently excised out of the genome. HSV-TK and NTR impart sensitivity to ganciclovir (GCV) () and CB1954 (), respectively, which have been tested clinically (, ). In our design, the NTR cassette, flanked by loxP sites, is eliminated upon expression of Cre by the human insulin promoter (). Therefore, insulin-expressing cells are rendered insensitive to CB1954. HSV-TK is driven by the telomerase promoter, which is active only in undifferentiated cell types (). This makes proliferating cells sensitive to GCV. Thus, our method provides a double fail-safe control such that (1) only insulin⁺, non-proliferating cells survive selection; (2) cells that may de-differentiate after transplantation () (and in which NTR was lost with the onset of insulin expression) may still be selectively killed in vivo by GCV, leaving the rest of the graft intact; and (3) undifferentiated cells are sensitive to two pro-drugs, making it less likely for tumorigenic cells to survive in case one single drug was insufficient to destroy 100% of them, or if they became resistant to one pro-drug due to spontaneous mutations of the relevant suicide gene (href="#bib19" rid="bib19" class=" bibr popnode">Kotini et al., 2016). No other method reported thus far offers the same degree of safety and specificity, as conventional suicide gene-based strategies bring about the destruction of the entire graft or do not enrich for the cells of therapeutic interest. Our results offer proof-of-principle of this approach and open the door to the subsequent targeting of these constructs to specific “safe harbor” locations within the genome of clinical-grade hESCs.