EQUAL DISTRIBUTION OF COMPETITIVE LONG-TERM REPOPULATING STEM-CELLS IN THE CD34(-) FRACTIONS OF THY-1(LOW)LIN(-() AND CD34()LOW)SCA-1(+) BONE-MARROW CELLS/

CD34 antigen is present on most, if not all, human hematopoietic stem cells (HSCs). Consistent with this pattern of expression, we recently reported that primitive murine HSCs defined as competitive long-term r epopulating units (CRUs) are highly enriched among CD34(+) bone marrow (BM) cells (one CRU/2500 cells). However, in agreement with one recen t report that some murine HSCs do not express CD34 (Science 273:242), we observed that 15% of phenotypically defined Thy-1(low)Lin(-/low)Sca -1(+) (TLS) stem cells were CD34(-) by fluorescence-activated cell sor ting. To examine further the nature of CD34 expression on murine hemat opoietic cells, we separated TLS cells into CD34(+) (0.022% of BM cell s) and CD34(-) (0.005% of BM cells) fractions, confirmed their phenoty pe by reverse transcriptase-polymerase chain reaction analysis of CD34 transcripts, and evaluated them in a variety of in vitro and in vivo assays. The CD34(+) TLS population contained most (93-95%) of the day 12 spleen colony-forming units (CFU-S) and in vitro colony-forming cel ls (CFCs). Cobblestone area-forming cells (CAFCs) able to proliferate on a murine bone marrow stromal cell line (SyS-1) represented one of e very 5 CD34(+) TLS and one of every 31 CD34(-) TLS cells. When lethall y irradiated mice were injected with 100 CD34(+) TLS cells, all animal s survived and began to recover circulating leukocytes, platelets, and erythrocytes by 15 days. In contrast, only 40% of mice injected with 100 CD34(-) TLS cells were radioprotected, and hematopoietic reconstit ution in surviving mice was not apparent until 21 days. The frequency of CRUs in CD34(+) and CD34(-) TLS cells was determined by injecting l imiting numbers of cells into lethally irradiated Ly-5 congenic hosts together with 10(5) ''compromised'' BM cells to provide radioprotectio n. CRUs able to regenerate and maintain lymphoid and myeloid cells for at least 6 months in primary and 5 months in secondary hosts represen ted one of every 156 CD34(+) TLS and one of every 35 CD34(-) TLS cells . However, when normalized for the proportion of TLS cells that were C D34(+) or CD34(-), it was determined that the recovery of CRU among CD 34(+) and CD34(-) TLS cells was equivalent (46% and 54%, respectively) . These data are consistent with the previous description of repopulat ing HSCs among CD34(-)c-kit(+)Sca-1(+)Lin(-) cells (Science 273:242, 1 996) and provide additional evidence that TLS cells are functionally h eterogeneous and can be further fractionated on the basis of CD34 expr ession. Overall, similar to 95% of CFCs, CFU-S, and CAFCs in the TLS p opulation were found to be CD34: whereas more primitive CRU were distr ibuted equally among CD34(+) and CD34(-) TLS cells. These results shou ld enable better characterization of the most primitive stem cells in murine BM.