Recent studies have shown efficient gene transfer to primitive progenitors in human cord blood (CB) when the cells are incubated in retrovirus-containing supernatants on fibronectin-coated dishes. We have now used this approach to achieve efficient gene transfer to human CB cells with the capacity to regenerate lymphoid and myeloid progeny in nonobese diabetic (NOD)/severe combined immunodeficiency (SCID) mice. CD34(+) cell-enriched populations were first cultured for 3 days in serum-free medium containing interleukin-3 (IL-3), IL-6, granulocyte colony-stimulating factor, Flt3-ligand, and Steel factor followed by two 24-hour incubations with a MSCV-NEO virus-containing medium obtained under either serum-free or serum-replete conditions. The presence of serum during the latter 2 days made no consistent difference to the total number of cells, colony-forming cells (CFC), or long-term culture-initiating cells (LTC-IC) recovered at the end of the 5-day culture period, and the cells infected under either condition regenerated similar numbers of human CD34(+) (myeloid) CFC and human CD19(+) (B lymphoid) cells for up to 20 weeks in NOD/SCID recipients. However, the presence of serum increased the viral titer in the producer cell-conditioned medium and this was correlated with a twofold to threefold higher efficiency of gene transfer to all progenitor types. With the higher titer viral supernatant, 17% +/- 3% and 17% +/- 8%, G418-resistant in vivo repopulating cells and LTC-IC were obtained. As expected, the proportion of NEO + repopulating cells determined by polymerase chain reaction analysis of in vivo generated CFC was even higher (32% +/- 10%). There was no correlation between the frequency of gene transfer to LTC-IC and colony-forming unit-granulocyte-macrophage (CFU-GM), or to NOD/SCID repopulating cells and CFU-GM (r2 = 0.16 and 0.17, respectively), whereas values for LTC-IC and NOD/SCID repopulating cells were highly and significantly correlated (r2 = 0.85). These findings provide further evidence of a close relationship between human LTC-IC and NOD/SCID repopulating cells (assessed using a textgreater/= 6-week CFC output endpoint) and indicate the predictive value of gene transfer measurements to such LTC-IC for the design of clinical gene therapy protocols.