The development of safe and effective gene therapy poses many formidable challenges. Some biomedical applications such as the hemoglobinopathies require that the gene product be expressed in a specific cell type, for example, erythroid progenitors. No such constraint pertains when the gene product circulates in the plasma. But in many cases, the level of the therapeutic protein needs to be tightly regulated. Perhaps the most compelling example is the hotly pursued goal of treating diabetes by a vector in which insulin gene expression responds appropriately to changes in blood glucose. The closest hematologic counterpart is the development of a vector in which erythropoietin (Epo) production is induced by subtle physiologic decreases in intracellular oxygen tension. Unlike the insulin gene, wherein the transcriptional response to glucose is complex and not well understood, the hypoxic induction by Epo depends upon a single crucial hypoxia inducible (transcription) factor (HIF) that is regulated by an elegant system of oxygen sensing and signal transduction, shared by all cells.

Binley and colleagues (page 2406) exploit this transcriptional servomechanism. They show quite convincingly that intramuscular injection of an adeno-associated virus harboring an EPOgene driven by a promoter containing HIF-response elements can cure severe anemia in mice in which endogenous Epo production has been nearly eliminated. Normal hematocrit levels are maintained for 7 months or more after initiation of treatment. The remarkably tight oxygen-dependent regulation of the transducedEPO gene is borne out by the total lack of change in hematocrit when this vector is injected into healthy animals.

This strategy could lead to a muchimproved way of administering recombinant human Epo (rhEpo) to patients. Long-term treatment with thrice-weekly rhEpo is very costly, particularly in patients who require relatively high doses. A key benefit of a physiologically regulated therapeutic gene is that enhanced production of Epo will continue until the anemia is corrected and the tissue depot in which the vector has been injected no longer senses hypoxia. But one note of caution seems warranted. Mother Nature placed the major site of Epo production in the subcortex of the kidney for a very good reason. At this site, wide fluctuations in oxygen tension are dampened, and therefore Epo production properly responds to total body hypoxia rather than to local vicissitudes. Intramuscular injection of a regulatableEPO gene could result in untoward increases in Epo expression from local hypoxia induced by exercise. Further studies in larger animal models are needed to investigate this potential problem and circumvent it.