To determine whether the sensitivity of the human long-term culture-initiating cell (LTC-IC) assay could be increased, we have evaluated a spectrum of different fibroblast cell lines for their abilities to influence the number of cells detectable as LTC-IC, to influence LTC-IC maintenance, and/or to influence LTC-IC differentiation into colony- forming cells (CFC) in cocultures containing various sources of LTC-IC. In a series of initial experiments with highly purified subpopulations of CD34+ cells from normal human marrow, no significant difference could be found between any of 3 different murine stromal fibroblast cells in terms of their support of either LTC-IC detection (CFC production) or maintenance (over a 6-week period), and all were equivalent to primary human marrow feeders (HMF). On the other hand, murine M2–10B4 fibroblasts engineered to produce high levels of both human granulocyte colony-stimulating factor (G-CSF) and interleukin-3 (IL-3; 190 and 4 ng/mL, respectively), either alone or mixed 1:1 with SI/SI fibroblasts engineered to produce high levels of soluble Steel factor (SF), with or without production of the transmembrane form of SF (60 and 4 ng/ mL, respectively), stimulated the production of up to 20- fold more CFC in LTC of cells from normal human marrow, G-CSF-mobilized blood or cord blood when compared with parallel cocultures containing HMF. Limiting dilution analysis of the CFC output from all three sources of LTC-IC showed that most of this increase was due to an ability of the engineered feeders to increase the plating efficiency of the LTC-IC assay (approximately 14-fold for marrow LTC-IC and approximately 4-fold for cord blood or mobilized blood LTC-IC). Analysis of the phenotype of these additionally recruited LTC-IC from marrow showed they had the same primitive CD34+CD45RA-CD71-phenotype as conventionally defined LTC-IC. The limiting dilution studies also showed that the average number of CFC produced per LTC-IC was additionally and independently increased to yield values of 18 CFC per LTC-IC in marrow, 28 for LTC-IC in cord blood, and 25 for LTC-IC in G- CSF-mobilized blood. Replating of cells from primary LTC with different feeders into secondary LTC-IC assays containing the best combination of engineered feeders showed that LTC-IC maintenance could be significantly enhanced (up to 7-fold as compared with primary cocultures containing HMF). However, this enhancement was still not sufficient to amplify the number of LTC-IC present after 6 weeks above the input value. Thus, engineering murine fibroblasts to produce sufficient SF, G-CSF, and IL-3 can markedly enhance the detection as well as the maintenance in vitro of a very primitive population of human progenitor cells present in normal adult marrow, mobilized blood, and cord blood by providing the most sensitive assay conditions thus far described. The present findings also provide new evidence of biologic heterogeneity between different cell populations that can be operationally identified as LTC-IC, thus re-emphasizing the importance of limiting dilution analyses to distinguish between quantitative and qualitative effects on these cells.