Identification of human T-lymphoid progenitor cells in CD34 þ CD38 low and CD34 þ CD38 þ subsets of human cord blood and bone marrow cells using NOD-SCID fetal thymus organ cultures C ATHERINE ROBIN,A NNELISE B ENNACEUR -G RISCELLI ,FAWZIA L OUACHE ,WILLIAM VAINCHENKER AND L AURE C OULOMBEL INSERM U362, Institut Gustave Roussy, Villejuif, France Received 6 July 1998; accepted for publication 11 December 1998 Summary. In contrast to myeloid and B-lymphoid differ- entiation, which take place in the marrow environment, development of T cells requires the presence of thymic stromal cells. We demonstrate in this study that human CD34 þ , CD34 þ CD38 þ and CD34 þ CD38 low cells from both cord blood and adult bone marrow reproducibly develop into CD4 þ CD8 þ T cells when introduced into NOD-SCID embryo- nic thymuses and further cultured in organotypic cultures. Such human/mouse FTOC (fetal thymic organ culture) thus represents a reproducible and sensitive system to assess the T-cell potential of human primitive progenitor cells. The frequency of T-cell progenitors among cord-blood-derived CD34 þ cells was estimated to be 1/500. Furthermore, the differentiation steps classically observed in human thymus were reproduced in NOD-SCID FTOC initiated with cord blood and human marrow CD34 þ cells: immature human CD4 low CD8 ¹ sCD3 ¹ TCRab CD5 þ CD1a þ T cells were mixed with CD4 þ CD8 þ cells and more mature CD4 þ CD8 ¹ TCRab þ cells. However, in FTOC initiated with bone marrow T progenitors, <10% double-positive cells were observed, whereas this proportion increased to 50% when cord blood CD34 þ cells were used, and most CD4 þ cells were immature T cells. These differences may be explained by a lower frequency of T-cell progenitors in adult samples, but may also suggest differences in the thymic signals required by bone marrow versus cord blood T progenitors. Finally, since cytokine-stimulated CD34 þ CD38 low cells retained their ability to generate T cells, these FTOC assays will be of value to monitor, when combined with other biological assays, the influence of different expansion protocols on the potential of human stem cells. Keywords: T lymphopoiesis, stem cells, NOD-SCID mice, thymus, cord blood. Evaluation of the haemopoietic potential of human donor cells in transplantation protocols is usually based on the quantitation of colony-forming cells (CFC) and long-term culture initiating cells (LTC-IC) (Sutherland et al, 1989) in vitro, and more recently human cells reconstituting NOD- SCID recipients (Cashman et al, 1997; Conneally et al, 1997; Pflumio et al, 1996; Wang et al, 1997). Except for the in vivo NOD-SCID assay, which detects both B-lymphoid and myeloid progenitors, most of the in vitro assays identify an exclusive myeloid differentiation, and only during the past years have in vitro conditions been designed which success- fully allow human CD34 þ CD38 low cells to differentiate into CD19 þ pro-B cells (Berardi et al, 1997; Hao et al, 1998; Galy et al, 1995a; Rawlings et al, 1997), and CD56 þ NK cells (Carayol et al, 1998). However, the identification of the T-cell potential of human pluripotent stem cells has remained a major obstacle both in vitro and in vivo. Feeders of thymic stromal cells do not reproducibly support T-cell differentia- tion in vitro (Rosenzweig et al, 1996) and the most successful strategy so far is based on the colonization by test cells of intact human or murine embryonic thymic lobes, which are then either kept in in vitro organotypic culture (fetal thymic organ culture; FTOC) (Jenkinson & Anderson, 1994; Merkenschlager et al, 1992) or grafted into a SCID recipient (Barcena et al, 1994; Galy et al, 1995b; Pe ´ault et al, 1991). Both human (Barcena et al, 1994; DiGustio et al, 1994; Galy et al, 1994, 1995b; Pe ´ault et al, 1991) and murine (Blom et al, 1997; Plum et al, 1994; Verhasselt et al, 1998; British Journal of Haematology , 1999, 104, 809–819 809 1999 Blackwell Science Ltd Correspondence: Dr Laure Coulombel, INSERM U362, Institut Gustave Roussy, 39 Av. Camille Desmoulins, 94800 Villejuif, France. e-mail: laurec@igr.fr.