Conditional Negative Selection of Gene-Modified Hematopoietic Stem Cells

In this issue of Molecular Therapy, Barese and colleagues from the National Institutes of Health present a landmark study conducted in a clinically relevant large-animal model to evaluate the utility of the herpes simplex virus thymidine kinase (HSVtk) gene for conditional elimination of transplanted long-term repopulating hematopoietic cells, potentially representing true hematopoietic stem cells (HSCs).¹ This paper provides proof of concept for this approach, is rich in highly relevant experimental findings, and presents a stimulating discussion that will certainly motivate the field to pay even more attention to the important principle of conditional negative selection of gene-modified cells.

Indeed, despite the remarkable success achieved with the negative selection of gene-modified T lymphocytes in clinical trials,^2,3,4 the application of this principle to gene-modified stem cells has been far less explored. Although some studies have addressed suicide gene transfer as a safety mechanism to eliminate potential tumors occurring following the transplantation of pluripotent stem cell progeny (see references in Barese et al.¹), studies addressing the utility of this principle for the more realistic and increasingly successful genetic modification of HSCs have been lacking. This is remarkable given that suicide gene transfer has been discussed over many years as a potential solution for the control of serious adverse events originating from the transformation of gene-modified hematopoietic cells since the first discovery of leukemias and premalignant clonal expansion induced by insertional mutagenesis.^5,6,7

One explanation for the lack of experimental studies testing the utility of the suicide principle for HSC modification may be related to negative expectations. The HSVtk-mediated mechanism of suicide was known to be largely restricted to proliferating cells; human HSCs, however, typically spend weeks in quiescence before undergoing cell division, at least in homeostatic conditions. In addition, ganciclovir (GCV), the prodrug used for suicide induction of HSVtk-expressing cells, is myelotoxic when administered for prolonged periods of time, thus potentially also eliminating untransduced hematopoietic cells. Moreover, studies with HSVtk-modified tumor cells had suggested that epigenetic and genetic escape mechanisms, such as vector silencing, rearrangement of the transgene or the flanking chromosomal sequences, or point mutations, may constitute major limitations for the complete elimination of the target population.^8,9

The new article from the Dunbar lab addressed all these limitations, with truly encouraging results.¹ First, this work revealed that the restriction of GCV-mediated cell killing to proliferating cells does not limit the potential to eliminate long-term repopulating hematopoietic cells, not even in the clinically relevant rhesus model that is far closer than any small-animal model to recapitulating the complex dynamics of human hematopoiesis, with coexistence of cycling and quiescent cells. The remarkable complete elimination of vector-transduced cell populations was assessed by very sensitive assays and persisted for at least 18 months after GCV application. These results imply the elimination of the pool of quiescent, long-lived lymphocytes, which must have originated from the transplanted HSCs and progenitor cells in the long time before administration of GCV. Whether GCV/HSVtk-mediated killing of quiescent hematopoietic cells is caused by mitochondrial damage, as speculated by Barese et al., remains to be addressed.

Second, this study identified at least two drug regimens that allow killing of long-term repopulating, HSVtk-transduced hematopoietic cells while avoiding major unspecific myelotoxicity. These drug regimens are another important component of this study, with direct translational implications. The good news is that a single course of GCV over a period of five days was sufficient for cell elimination. Another, more continued course lasting for three weeks was also well tolerated and highly efficient. Thus, even if elimination would be more difficult, as in the case of a larger population of transduced cells and/or malignant transformation, the therapeutic index of GCV might be sufficient to be clinically useful in patients receiving gene-modified HSCs.

Third, the authors of this study found no evidence for escape mechanisms by genetic instability, but did report evidence suggesting that the specific activity of the HSVtk has an impact on the completeness of cell killing in their model. Although not directly addressed, this implies that a given threshold of HSVtk expression and activity has to be overcome to exploit this suicide principle in HSCs and their progeny cells. Engineering the HSVtk expression cassette and coding sequences provides room for further improvements (e.g., Preuss et al.¹⁰). Furthermore, Barese and colleagues make the interesting and somewhat challenging suggestion that promoters be designed that work only in transformed cells, which would be of special interest when a malignant gene-modified clone is to be eliminated while sparing the untransformed yet gene-modified cell population.

Finally, Barese and colleagues also addressed whether cell elimination involved an immune response to the transgene products encoded by the gene vectors, for which they found no evidence. This was not unexpected, given that the transplantation of gene-modified primitive hematopoietic cells is known to potentially induce central tolerance to transgene products. In any case, this represents another important observation, as many suicide genes contain nonhuman sequences. In summary, the article by Barese et al. marks major progress on the path toward conditional stem cell therapy, relevant for numerous applications beyond hematopoiesis. This work will certainly revive the concept of conditional, optional suicide induction to control the fate of gene-modified stem cells and their progeny in vivo.