Whether this mismatch represents focal allelic variation, or alternatively, an artifact introduced during cloning, is controversial. However, in an in vivo study utilizing rfEPO protein coding for glycine, the feline genomic DNA of 12 recipient cats was amplified and sequenced to establish evidence for allelic variation. The nucleotide sequence at this locus in all 12 cats was determined to be 44GAG, coding for glutamic acid. In a database search of kinase inhibitors mammalian erythropoietin cDNA sequences, over 50 different species of mammals demonstrate the GAG codon at this site with none coding for GGG. In aggregate, these findings suggest that the “polymorphism” identified at the 44th codon of the feline erythropoietin cDNA most likely represents an artifact introduced during the initial cloning procedures. Importantly, modification of a single amino acid residue in human EPO has been shown to affect the biological function dramatically. For in vivo applications, the problem of rEPO immunogenicity is not the only important consideration that needs to be addressed. Regulation of EPO transgene dosage may play a critical role in the future success or failure of rfEPO gene therapy attempts. Transgene regulation can be dramatically affected by the function of the promoter driving expression of the rfEPO transgene. Multiple researchers have chosen to use a powerful constitutive promoter such as the cytomegalovirus immediate-early promoter. However, In two different studies, healthy cynomologous macaques treated with macaque-derived EPO-expressing AAV vectors developed supraphysiologic levels of EPO and polycythemia. In an experiment where the recombinant EPO cDNA was expressed constitutively from a cytomegalovirus promoter, severe anemia developed in some animals and polycythemia required repeated therapeutic phlebotomies to maintain nontoxic hematocrits. Profound anemia has also been reported in macaques treated with an AAV vector expressing EPO cDNA driven from a doxycycline-regulated promoter. As a result of studies like these, researchers have developed naked DNA-based gene therapy protocols in rodent models. These studies have resulted in correction of anemia without triggering excessive hematopoiesis. Researchers have attempted to regulate the mammalian host hematocrit level by controlling the dosage of the administered recombinant viral vector. Unfortunately, this technique has resulted in an all-or-none phenomenon rather than the clinically desired dose-dependent increase in hematocrit. Excision of the intramuscular rAAV2-rfEPO vector injection site has also proved unreliable for abrogating pathological erythrocytosis of rfEPO. Although the therapeutic window of rfEPO in anemic cats is expected to be large, many researchers now feel that clinically useful gene expression of rfEPO should include a mechanism to regulate gene expression to avoid life-threatening anemia or pathological erythrocytosis. An oxygen-regulated gene therapy approach has been successfully employed in erythropoietin-deficient mice. In the same study, mice treated with CMV-EPO constructs acquired fatal polycythemia. The main physiologic stimulus for erythropoietin synthesis is renal hypoxia. When hypoxia is present, hypoxiainducible factor is free to bind to hypoxia–response elements of oxygen regulated genes. These response elements control the erythropoietin gene within the kidneys, and binding of HIF-1a stimulates an increase in production of erythropoietin.