New Evidence Supporting Megakaryocyte- Erythrocyte Potential of Flk2/Flt3+ Multipotent Hematopoietic Progenitors

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Matters Arising

E. Camilla Forsberg,1,3,* Thomas Serwold,1 Scott Kogan,2 Irving L. Weissman,1 and Emmanuelle Passegue´ 1,3,4,* 1 Institute of Stem Cell Biology and Regenerative Medicine, Departments of Pathology and Developmental Biology, Stanford University School of Medicine, Stanford, CA 94305, USA 2 Department of Laboratory Medicine and Comprehensive Cancer Center, University of California, San Francisco, CA 94143, USA 3 These authors contributed equally to this work. 4 Present Address: Developmental and Stem Cell Biology Program, University of California, San Francisco, CA 94143, USA. *Contact: [email protected] (E.C.F.); [email protected] (E.P.) DOI 10.1016/j.cell.2006.06.037

SUMMARY model states that only the most primitive long-term reconstituting subset of HSC (LT-HSC) self-renew for life and A model of hematopoietic development wherein give rise to short-term reconstituting subsets of HSC multipotentiality is conserved until segregation (ST-HSC), with limited self-renewal activity, and then to of myeloid and lymphoid potential has recently multipotent progenitors (MPP) that do not self-renew been challenged, proposing that megakaryo- (Spangrude et al., 1988; Morrison et al., 1997; Christensen cyte/erythrocyte (MegE) potential is lost in and Weissman, 2001). In this model, differentiation prog- Flk2/Flt3-expressing early progenitors. Here, resses with segregation of myeloid and lymphoid lineages through the generation of common myeloid progenitors we used sensitive in vivo approaches to quanti- (CMP) (Akashi et al., 2000) and common lymphoid progen- tatively and kinetically assess the MegE poten- itors (CLP) (Kondo et al., 1997). Both CLP and CMP un- tial of hematopoietic stem cells and various dergo further lineage segregation, with CMP giving rise + Flk2 early progenitors and compared it with to myelomonocytic progenitors (GMP) and megakaryo- the MegE potential of downstream committed cyte/erythrocyte progenitors (MEP) that separate granulo- myeloid and lymphoid progenitors and with cyte and macrophage production from red blood cell and their ability to give rise to mature myelomono- platelet production (Akashi et al., 2000). cytic and lymphoid cells. We demonstrate that In the mouse, multipotent hematopoietic activity is con- Flk2+ early progenitors retain MegE potential in tained in a small fraction of the bone marrow that lacks vivo both at the population and clonal levels. expression of mature hematolymphoid-associated cell- These results indicate that Flk2 expression by surface markers (lineage or ‘‘Lin’’ markers) and displays high levels of the stem cell-associated markers c-Kit and early progenitors is not at the expense of full Sca-1 (LinÀ/c-Kit+/Sca-1+ or ‘‘KLS’’) (Spangrude et al., multipotency and support the current model of 1988; Ikuta and Weissman, 1992). The KLS fraction can hematopoietic development with segregation be further subdivided into functionally distinct subpopula- of myeloid and lymphoid lineages from multi- tions based on the expression levels of the Thy1.1 potent progenitors. (Spangrude et al., 1988), CD4 and Mac-1 (Morrison and Weissman, 1994), CD34 (Osawa et al., 1996), Flk2/Flt3 INTRODUCTION (Adolfsson et al., 2001; Christensen and Weissman, 2001), and the Slam family (Kiel et al., 2005; Forsberg Hematopoietic development is one of the most intensely et al., 2005) of surface molecules. Within the KLS fraction, studied systems of cellular differentiation, where multipo- loss of Thy1.1 and gain of Flk2 expression correlates with tent self-renewing hematopoietic stem cells (HSC) give decreased self-renewal ability (Adolfsson et al., 2001; rise to a hierarchy of progenitor populations with more Christensen and Weissman, 2001) and increased proli- restricted lineage potential, ultimately leading to the pro- feration (Passegue´ et al., 2005). Recently, it has been duction of all types of mature blood cells. Prospective reported that Flk2-expressing KLS cells might not be isolation of cellular subsets based on expression of cell- fully multipotent because they failed to generate signiﬁ- surface markers followed by functional characterization cant megakaryocyte (Meg) and erythrocyte (E) progenies using in vitro and in vivo assays has provided key insights in the assays used in those studies (Yang et al., 2005; into hematopoietic development (Weissman, 2000). One Adolfsson et al., 2005). Conclusions from those reports

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 415 challenged the current fate map of hematopoietic devel- RESULTS opment and the existence of MPP and CMP populations by suggesting that early progenitor populations might se- Tracking the Lineage Potential of Stem quentially lose lineage potential as they differentiate from and Progenitor Cells LT-HSC, starting with loss of MegE potential (Adolfsson Due to differences in self-renewal ability and proliferation et al., 2005; also see commentaries by Akashi et al., rates, stem and progenitor cells differ considerably in their 2005; Hock and Orkin, 2005). ability to expand both in vitro and in vivo. To assay the An important test for HSC function is their ability to expansion capacity of LT-HSC, ST-HSCF, MPPF, CMP, reconstitute the blood system of an irradiated host, includ- GMP, and MEP in vitro, cells were grown in liquid media ing the HSC pool. Commonly, donor cells are distin- containing cytokines and growth factors promoting mye- guished from host cells by transplantation of purified pop- loerythroid proliferation and differentiation (Figure 1A). ulations into conditioned hosts with an allelic difference in For the first 4 days of culture, MPPF displayed the greatest the hematopoietic cell-surface marker CD45, followed by rate of expansion, while by days 6 and 8 ST-HSCF and allele-specific antibody labeling (Spangrude et al., 1988). LT-HSC, respectively, accumulated to a larger extent. While all mature myelomonocytic and lymphoid cells ex- Among the myeloid progenitors, CMP displayed a kinetics press robust CD45 levels, mature cells of the MegE line- of expansion very similar to MPPF, while GMP and MEP age lack CD45 expression, making it difficult to assess proliferated poorly throughout the culture period. After directly whether the transplanted cells are capable of 8 days, LT-HSC had undergone an expansion of 1600- forming platelets and red blood cells in vivo. Furthermore, fold, ST-HSCF of 1200-fold, MPPF of 350-fold, CMP while single-cell assessment of lineage potential is feasi- of 250-fold, MEP of 81-fold, and GMP of 65-fold. ble for LT-HSC (Smith et al., 1991; Wagers et al., 2002; These results indicate an inverse correlation between Osawa et al., 1996; Adolfsson et al., 2001), progenitors the burst size of each progenitor population and its degree with lower self-renewal potential give rise to insufficient of maturation, which has to be taken into account when numbers of progeny for robust single-cell detection investigating lineage potential in vivo. The differentiation in vivo. Thus, as self-renewal ability declines, increasing potential of each population was also investigated at the numbers of input cells are needed to allow detection of single-cell level in methylcellulose media and further con- progeny. In addition, mature myeloid cells are short lived firmed the direct correlation between the time course of (days), while a significant fraction of mature lymphocytes differentiation and the developmental stage of the starting are long lived (months to years), enabling easier detection population (Figure S1). of lymphoid versus myeloerythroid progeny at late time To circumvent the lack of CD45 expression by mature points. Together, these limitations have made it difficult to cells of the MegE lineage, we used b-actin GFP transgenic assess in vivo the full lineage potential of early progenitor mice (Wright et al., 2001) as a source of donor cells and cells. transplanted purified stem and progenitor populations at Here, we have used sensitive in vivo approaches to various doses into sublethally irradiated (475 rad) con- quantitatively and kinetically assay the MegE potential genic wild-type recipient mice (Figure 1B). For all experi- of the KLS fraction of mouse bone marrow, divided ments cells were double or triple sorted to ensure into LT-HSC (Thy1.1int/Flk2À KLS), short-term HSC maximum purity. Since all platelets (Plt) (Nakorn et al., (ST-HSCF: Thy1.1int/Flk2+ KLS), multipotent progenitor 2003) and nucleated myelomonocytic and lymphoid cells population (MPPF: Thy1.1À/Flk2+ KLS) (Christensen and in the peripheral blood (PB) of b-actin GFP mice are Weissman, 2001), or ‘‘lymphoid-primed’’ multipotent pro- positive for GFP (Figures 1C and 1D), this assay enabled genitors (LMPP: 25% brightest Flk2+ KLS) (Adolfsson direct and quantitative in vivo tracking of mature cell pro- et al., 2005), and compared it with the MegE potential of duction by each transplanted population. The use of sub- downstream myeloid (CMP, GMP, and MEP) and lym- lethal irradiation, as opposed to lethal irradiation, allowed phoid (CLP) progenitors and with their ability to give rise us to follow the kinetics of mature cell production with to mature myelomonocytic and lymphoid cells in vivo. high resolution, starting within days after transplantation. We demonstrate that Flk2+ ST-HSCF, MPPF, and LMPP retain MegE potential at the population and clonal levels Platelet Production in vivo, readily detectable when sufficient numbers of To determine whether Flk2 expression identifies a KLS input cells are provided and when progeny are investi- population lacking MegE potential, we purified in parallel gated at the early and late time points. We also estab- LT-HSC, ST-HSCF, MPPF (Christensen and Weissman, lished the timecourse of mature cell production by each 2001), and LMPP (Adolfsson et al., 2005) from b-actin of these progenitor populations, hence providing the tim- GFP mice (Figure 2A). The relationship between these ing required for these early differentiation events in vivo. populations is shown in Figure S2. At 12 days posttrans- These results demonstrate that Flk2 expression is not at plantation, LT-HSC provided the highest levels of GFP+ the expense of full multipotency and support the current Plt at every cell dose transplanted (Figure 2B). Although model of hematopoietic development with segregation with lower efficiency than equivalent numbers of of myeloid and lymphoid lineages from multipotent LT-HSC, ST-HSCF, MPPF, and LMPP produced robust, progenitors. dose-dependent numbers of GFP+ Plt. Importantly, Plt

416 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. Figure 1. Experimental Strategy Used to Investigate the Line- age Potential of Stem and Progenitor Cells (A) Comparison between the burst size of LT-HSC, Flk2+ early progenitors (ST-HSCF, MPPF) and committed myeloid progenitors (CMP, GMP, and MEP) in liquid culture. Five hundred purified cells were seeded in IMDM media containing SCF, Flt3-L, IL-11, IL-3, Tpo, Epo, and GM-CSF, and total numbers of live cells were counted after the indicated days of culture. (B) Schematic representation of the in vivo tracking strategy used to investigate lineage potentials of stem and progenitor cells. Donor cells are purified from b-actin GFP mice and transplanted at various cell doses into sublethally irradiated (475 rad) wild-type recipient. Trans- planted mice are bled at frequent intervals to follow with high resolution the kinetics of lineage repopulation in the peripheral blood (PB). (C) FACS analysis of PB platelets (Plt) defined as FSClow/Gr-1À/Ter- 119À/CD61+ cells, illustrating that 100% of them are GFP+ in b-actin GFP mice. (D) FACS analysis of PB nucleated cells (Ter-119À), illustrating that 100% of them are GFP+ in b-actin GFP mice. GFP+ nucleated cells can be further subdivided into myelomonocytic cells (Gr-1/Mac-1+) and lymphoid B220+ B cells or TCR-b+ T cells. production from those Flk2+ KLS cells was more efficient than from equivalent numbers of myeloid committed pro- Figure 2. Platelet Production by Stem and Progenitor Cells genitors CMP/MEP isolated and transplanted together. In Vivo Neither GMP nor CLP, even used at the highest cell (A) Gating strategies used to purify LT-HSC, ST-HSCF, MPPF,and dose tested (10,000 cells), gave rise to any detectable LMPP from c-Kit-enriched b-actin GFP transgenic C57Bl/6-Thy1.1 mouse bone marrow. Plt (Figures 2B and S3). These results demonstrate that F F (B) Dose dependency of donor-derived platelet production. FACS ST-HSC , MPP , and LMPP are capable of thrombopoie- analysis at 12 days posttransplantation of GFP+ donor-derived Plt sis in vivo but require higher cell doses than LT-HSC, al- present in the PB of sublethally irradiated mice transplanted with the though lower than CMP/MEP, to provide robust platelet indicated doses for each subpopulation (k = 103 cells).

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 417 Figure 3. Multilineage Differentiation Kinetics of Stem and Progenitor Cells In Vivo (A) Quantitative and kinetic analysis of GFP+ donor-derived Plt production in PB of sublethally irradiated mice transplanted with the indicated doses for each subpopulation (n = 5 per subset and cell dose, k = 103 cells). Plt generation from MPPF, LMPP, and CMP/MEP are compared on a logarithmic scale for clarity (bottom graph). (B) Quantitative and kinetic analysis of GFP+ donor-derived nucleated cell production in PB of the same mice as in (A). The two bottom graphs show the contribution of Gr-1/Mac-1+ and B220+ cells as percent of total donor- derived cells. The shaded area indicates back- ground detection level. Results are expressed as average ± standard deviation (error bars). (C) FACS analysis of total GFP+ donor-derived nucleated cells, and its further subdivision into myeloid (Gr-1/Mac-1) and lymphoid (B220/ TCR-b) components, in the PB of mice transplanted with 500 LT-HSC, MPPF, or LMPP at day 16 (top panels) and day 28 (bottom panels) posttransplantation.

contribution, consistent with their more limited expansion and disappeared within 16 to 28 days for every cell dose capacities. tested (Figure 3A). LT-HSC were the only cells able to provide sustained Kinetics of Mature Cell Production production of multilineage donor-derived nucleated cells, To establish the kinetics of mature cell production from which emerged in the PB around day 12 and continued to each population, we followed over time the production increase beyond 1 month posttransplantation. In contrast, of donor-derived Plt and nucleated Gr-1/Mac-1+ myelo- all progenitor populations produced a transient wave of monocytes, B220+ B cells, and TCR-b+ T cells in the PB mature cells, which depended on for its amplitude, emer- of the transplanted animals (Figures 3 and S3). Donor- gence, and disappearance time the differentiation stage derived Plt were ﬁrst detected at 7–9 days for all trans- of the input population and on the type of mature cell pro- planted populations, with Plt production from LT-HSC duced (short-lived myeloid cells or long-lived lymphoid increasing over a 30-day period and then remaining cells) (Figures 3B, 3C and S3). Hence, ST-HSCF-derived stable (Figure 3A). In contrast, Plt production from cells emerged by day 9 and reached a plateau by days ST-HSCF, MPPF, and LMPP peaked at days 16–20 and 16–20 before gradually declining, while both MPPF and then declined, resulting in gradually decreasing levels of LMPP-derived cells emerged (day 7) and peaked (days ST-HSCF-derived Plt over a period of 3 months, while 12–16) a few days earlier before declining. In all cases, MPPF- and LMPP-derived Plt reached undetectable myelomonocytic maturation occurred ﬁrst (Figure 3B). levels by 4 weeks posttransplantation. Importantly, While LT-HSC, and to a lower extent ST-HSCF, sustained CMP/MEP-derived Plt production was optimal at day 9 durable production of myelomonocytic cells, production

418 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. from MPPF and LMPP rapidly declined, reaching unde- spectively), while mice transplanted with either 100 or tectable levels within 4 weeks posttransplantation, a tim- 500 MPPF or 500 LMPP had by day 10 readily detectable ing matching Plt disappearance in the same animals CFU-S, which were more numerous and larger by day 12 (Figure 3A). Myelomonocytic cells derived from trans- (Figure 5A and data not shown). MPPF gave rise to CFU-S planted CMP/MEP (peaking at day 12) and GMP (peaking with a frequency of 1/78 and LMPP with a frequency at day 9) also did not accumulate and disappeared within of 1/230 in our assay, which is below the frequency of 16 to 20 days (Figures 3B and S3). Donor-derived B cells 1/85 previously reported for this population at day 11 emerged in the PB of MPPF- and LMPP-transplanted (Yang et al., 2005). The lower frequency of CFU-S in mice around day 9, while ST-HSCF required 16 days LMPP than in MPPF could represent lower efficiency of and LT-HSC 20 days for significant B lymphocyte pro- spleen homing by clonogenic precursors, or higher het- duction. T cell emergence also occurred earlier from erogeneity in the LMPP population with some precursors transplanted MPPF and LMPP compared to LT-HSC and failing to produce detectable myelo-erythroid spleen colo- ST-HSCF (24–28 days versus 40–49 days). From trans- nies. Flow cytometric analysis of splenocytes after RBC planted CLP, donor-derived B cells emerged as early as lysis indicated the presence of numerous Ter-119+ cells, day 7 and increased until the emergence of T cells by either GFPÀ (nucleated erythrocytes) or GFPdull (erythrocyte day 20, before gradually declining without detectable my- precursors), in every spleen harboring CFU-S (Figure 5B). eloid readout (Figure S3). Lymphopoiesis, like myelopoie- Such cells were absent from spleens of lethally irradiated sis, was not sustained in progenitor-transplanted mice, nontransplanted control mice, demonstrating that nucle- but in contrast to short-lived myeloid cells, long-lived lym- ated Ter-119+ cells are of donor origin despite the lack phoid cells persisted for several months in these animals, of GFP expression. Immunofluorescence analysis of day although with declining levels. 12 spleen sections showed large CFU-S colonies, all con- Mature red blood cells (RBC) from b-actin GFP trans- taining nucleated Ter-119+ cells interspersed with donor- genic mice do not express GFP, hence precluding direct derived GFP+ cells, consistent with the clonal origin of the analysis of donor contribution to mature erythrocytes. cells within a colony (Figure S5). Radioprotection analysis However, immature erythrocyte precursors (FSChigh/ also indicates that ST-HSCF radioprotected as efficiently CD45int/Gr-1À/Mac-1À/CD71+/Ter-119+)(Socolovsky et al., as LT-HSC at the 500 cell dose, while MPPF and LMPP, 2001) in the bone marrow (BM) and spleen (Sp) of these though with decreasing potency, were more efficient mice express low levels of GFP (Figure 4A). At day 16 post- than CMP/MEP at providing radioprotection (Figure S6). transplantation, both BM and Sp populations of GFPdull Together, these results demonstrate that Flk2+ KLS cells donor-derived erythrocyte precursors were observed in are capable of robust erythropoiesis and megakaryopoie- mice transplanted either with 500 LT-HSC, ST-HSCF, sis in vivo both at the population and clonal levels. MPPF, or LMPP and were absent from mice transplanted with either 10,000 GMP (data not shown) or CLP Short-Term Repopulation Kinetics (Figure 4B). While these erythrocyte precursors stably We also followed the emergence of GFP+ donor-derived persisted in LT-HSC- and, for shorter times, ST-HSCF- cells in the BM, Sp, and PB of the lethally irradiated transplanted mice, they were only transiently found in mice transplanted with 100 LT-HSC or ST-HSCF and MPPF-and LMPP-transplanted mice and disappeared by 500 MPPF or CMP/MEP used for the CFU-S assays. At day 49 (Figure S4). Together, these results demonstrate day 8, LT-HSC-transplanted mice only showed the pres- that Flk2+ KLS cells are fully multipotent and have MegE ence of low numbers of GFP+ cells in the BM, while potential in vivo, readily detectable when sufficient num- ST-HSCF-transplanted mice displayed GFP+ cells in bers of input cells are provided and when progeny are both BM and PB, and MPPF-transplanted mice had even investigated at the proper time points. They also indicate higher numbers of GFP+ cells in all three compartments that on a cell-by-cell basis ST-HSCF, MPPF, and LMPP (Figure 5C). At this time, CMP/MEP-transplanted mice are less efficient than LT-HSC in producing every type of already displayed large numbers of GFPdull nucleated mature cell, including B and T lymphocytes, consistent erythrocytes in the Sp but very few GFP+ cells in the BM with their more advanced maturation stage. and the PB. By day 10, the readout from LT-HSC-transplanted mice was very similar to day 8 ST-HSCF, with Erythrocyte Production from Clonogenic Cells GFP+ cells only in the BM and PB, while the readout of To investigate directly the ability of Flk2+ ST-HSCF, MPPF, ST-HSCF-transplanted animals closely resembled day 8 and LMPP to give rise to erythrocytes in vivo, we used MPPF except for higher numbers of donor-derived cells conventional spleen colony-forming-unit (CFU-S) assays generated. At this time, MPPF-transplanted mice dis- in lethally irradiated mice. Similar to previous results (Na played massive expansion of GFP+ cells in all three Nakorn et al., 2002), only mice transplanted with 500 compartments and appearance of GFPdull nucleated ery- CMP/MEP had visible CFU-S by day 8 (1/24 frequency), throcytes in the Sp, which correlate with the early emer- while mice transplanted with 500 GMP did not develop gence of CFU-S in this population (Figure 5A). At day 12, spleen colonies at any time point (Figure 5A). Mice trans- all three populations showed expansion of donor-derived planted with 100 LT-HSC or ST-HSCF had no obvious cells in all compartments and presence of large numbers CFU-S until day 12 (1/33 and 1/18 frequencies, re- of GFPdull nucleated erythrocytes in the Sp. Together,

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 419 Figure 4. Generation of Erythrocyte Precursors from Stem and Progenitor Cells In Vivo (A) Erythrocyte precursor analysis in bone marrow (BM) and spleen (Sp) of b-actin GFP mice. Gating strategy used for the FACS analysis and sorting of FSChigh/CD45int/GFPdull/Gr-1À/Mac-1À/CD71+/Ter-119+ erythrocyte precursor subsets. Shown are May-Gru¨ nwald-Giemsa stains confirming immature erythrocyte precursor morphology in purified subsets. (B) Erythrocyte precursor generation in transplanted mice. Sublethally irradiated (475 rad) mice were transplanted with either 500 LT-HSC, ST-HSCF, MPPF, and LMPP or 10,000 CLP purified from b-actin GFP transgenic mice and analyzed at day 16 posttransplantation for the presence of erythrocyte precursors in the BM and Sp. The numbers indicate percentages of total live cells. these observations suggest that LT-HSC need 2 days to out and then another day for a CMP/MEP-equivalent produce a ST-HSCF-equivalent readout, which required readout, thereby providing an estimate of 4 days for 1–2 days more to give rise to an MPPF-equivalent read- transit through the KLS pool.

420 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. Figure 5. Erythroid Cell Production and Short-Term Kinetics of Repopulation (A) CFU-S assay. Spleens of mice transplanted with either no cells (control) or the indicated populations (*100/500 cells) were isolated at days 8, 10, and 12 post-lethal irradiation (950 rad) and visually inspected for presence of macroscopic colonies before or after ﬁxation in Tellyesniczky’s solution (TF). The frequency of CFU-S [1/(numbers of injected cells divided by numbers of observed spleen colonies)] is given at day 8 for CMP/MEP and day 12 for LT-HSC, ST-HSCF, MPPF, and LMPP. n.d.: not determined; d.: dead. (B) Generation of nucleated erythrocytes. FACS analysis (after RBC lysis) of day 12 untransplanted (no cells) and MPPF-transplanted spleens showing Ter-119 levels in live nucleated Gr-1À/Mac-1À cells separated into three subsets based on GFP expression levels. Note that all Ter-119+ cells in MPPF spleen are GFP negative to dull and are absent in control spleen. The overlay histogram (right side) compares Ter-119 levels by Gr-1À/Mac-1À live cells (regardless of GFP expression) in spleen of mice transplanted with the indicated populations. (C) FACS analysis following over time the production of total GFP+ donor-derived cells in the bone marrow (BM), spleen (Sp), and peripheral blood (PB) in mice from (A). The boxes give the percentages of GFP+ and GFPdull donor-derived cells.

EpoR, TpoR, and IL3R Gene Expression were found in MEP, with LT-HSC displaying an 5-fold We then analyzed these populations by quantitative RT- lower level, similar to levels reported in CMP (Terskikh PCR for the expression levels of the two major receptors et al., 2003). Although expressed at lower levels than in associated with MegE development: the erythropoietin LT-HSC, EpoR mRNA was detectable in ST-HSCF (14- receptor (EpoR) and the thrombopoietin receptor (TpoR) fold lower), MPPF (45-fold lower), and LMPP (65-fold (Figure 6). The highest levels of EpoR mRNA expression lower) and to levels much higher than in populations that

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 421 Figure 6. Multipotency and Expression of the MegE and GM Differentiation Gene Programs (A) Real-time quantitative RT-PCR analysis (qRT-PCR) for erythropoietin receptor (EpoR), thrombopoietin receptor (TpoR), and b-actin mRNA expression levels in LT-HSC (gray), ST-HSCF (orange), MPPF (red), LMPP (black), MEP (magenta), and Gr-1+/Mac-1+ granulocytes (light blue) and B220+ splenic B lymphocytes (green). (B) Fold differences. Summary of the qRT-PCR analysis of EpoR, TpoR, and interleukin-3 receptor (IL3R) mRNA expression. Each value has been standardized for b-actin expression levels and the relative expression level of each gene for each subpopulation is expressed as fold differences compared to the levels detected in LT-HSC (set to 0). Results are given as the average from two to three independent experiments. nd: not detectable. did not undergo MegE differentiation such as mature loss of MegE potential. Here, we demonstrate that multi- granulocytes (220-fold lower) and B lymphocytes (unde- potentiality is retained in the Flk2/Flt3-positive KLS frac- tectable). In agreement with previous reports, LT-HSC ex- tion, and that loss of lineage potential occurs later during pressed TpoR mRNA at higher levels than MEP (Terskikh differentiation into the oligolineage-committed progenitor et al., 2003). While slightly decreased in ST-HSCF (3-fold populations. Moreover, by uncovering differences in the lower), MPPF (5-fold lower), and LMPP (7-fold lower) timecourse of mature cell production from LT-HSC and compared to LT-HSC, TpoR expression was much higher progenitor cells, we provide a temporal component that in those populations than in granulocytes (700-fold is of critical importance for the investigation of the lineage lower) and B lymphocytes (undetectable). Similarly, potential of these populations in vivo and for comparing ST-HSCF, MPPF, and LMPP expressed higher levels of ﬁndings between laboratories (Figure 7). GM-associated receptors IL3R, GM-CSFR, G-CSFR, Our results differ from those of the Jacobsen group who and M-CSFR than mature cell populations, even to slightly concluded that the 25% brightest Flk2+ KLS cells (LMPP) higher levels than in LT-HSC (Figure 6B and data not lack MegE potential (Adolfsson et al., 2005). It is unlikely shown). Together, these results indicate that Flk2+ KLS that the difference in the assessed ability of LMPP to pro- cells express at the population level the appropriate duce erythrocytes and platelets is due to subtle variations myeloid gene programs to undergo both MegE and GM in mouse strains or in sorting strategies, as the Jacobsen differentiation, and while the transcript levels of some group previously reported erythroid readout from LMPP in of the MegE regulators are lower than in LT-HSC, they CFU-S assays (Yang et al., 2005). Rather, we believe that appear to be sufﬁcient for their in vivo MegE potential. this discrepancy is due to several important differences in the interpretation of the results and in experimental de- DISCUSSION sign, such as the choice of cell doses and timing for the in vitro and in vivo analyses. To obtain a comprehensive A thorough understanding of the lineage potential of each assessment of mature cell production in vivo, we com- subset of hematopoietic stem and progenitor populations pared the engraftment levels of multiple progenitor popu- is critical in establishing an accurate map of cell-fate de- lations with different lineage potential and burst sizes and termination during hematopoietic development. A contro- titrated the cell dose required to detect their mature prog- versy exists as to whether multipotentiality is conserved enies. We found a direct correlation between the number until segregation of myeloid and lymphoid potential, or of cells needed and the maturation stage of the starting whether progenitor populations sequentially lose lineage population, indicating that engraftment and lineage poten- potential when differentiating from LT-HSC, starting with tial of progenitor cells have to be investigated with greater

422 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. Figure 7. Kinetics of Blood Lineage Development As LT-HSC differentiate into ST-HSCF and then to MPPF, they progressively lose their self-renewal ability, increase their proliferation, and change their gene expression programs to leave the stem cell niche and undergo lineage differentiation while remaining multipotent. Within this scheme, LMPP represent a more committed subset of MPPF. We estimated the duration of the transit through the KLS pool to be 4 days. The simultaneous emergence of B lymphocytes and Gr, MF, and Plt as soon as 7 days posttransplantation of MPPF indicates that generation of CLP and CMP, and the further separation of CMP into GMP and MEP, occurs with similar kinetics. The accumulation of Plt for up to 16 days, the maturation of nucleated E after 12 days, and the progressive increase in B cell production likely result from differences in proliferation rates of GMP, MEP, and CLP (Passegue´ et al., 2005). The late emergence of T lymphocytes may relate to obligatory maturation steps in the thymus before release in the PB. Without constant resupply from self-renewing LT-HSC, MPPF or LMPP production of short-lived Gr, MF and Plt become exhausted within a month while long-lived T and B lymphocytes accumulate and persist for several months. Gr: granulocytes; MF: macrophages; Plt: platelets; E: erythrocytes; B: B lymphocytes; T: T lymphocytes. cell numbers than LT-HSC. Furthermore, without constant from more primitive cells located developmentally up- resupply, as for the self-renewing LT-HSC, all mature cells stream of these progenitors. Also as expected, we found derived from progenitors were exhausted with kinetics that Flk2+ KLS cells have lower MegE potential than that directly depended on the half-life of the mature cells. LT-HSC on a cell-by-cell basis, which was considered We found that short-lived myelomonocytic cells, platelets, the result of contamination by low numbers of LT-HSC and erythrocyte progenitors disappear within 1 month in the report by Adolfsson et al. (2005). By investigating posttransplantation, while long-lived lymphoid cells that the MegE potential of these progenitors within a develop- can give rise to self-renewing progeny (such as memory mental series (including both less [LT-HSC] and more T and B cells) accumulate and persist for several months. [CMP/MEP] committed progenitors), we show that this Thus, in contrast to LT-HSC, investigation of lineage po- readout indicates true MegE potential from a low burst tential of progenitor cells has to be performed at several size population. We ruled out contamination by using distinct time points to allow optimal detection of their highly puriﬁed, double or triple sorted populations for all progenies (before 3 weeks for myelopoiesis, after 3 weeks of our assays. While very low numbers of contaminating for lymphopoiesis). LT-HSC could be detected in ST-HSCF (<1/800 cells on By isolating stem and progenitor cells from b-actin GFP average), no contaminating LT-HSC or ST-HSCF were transgenic mice and by transplanting them at various cell observed in MPPF or LMPP by reanalysis of the sorted doses into sublethally irradiated mice, we were able to populations (Figure S7). The differences observed for all track directly all donor-derived cells (including platelets of the investigated populations in repopulating kinetics and erythrocyte precursors, with the exception of RBC) (emergence, peak, and disappearance) also make con- and to follow with high resolution their kinetics of produc- tamination an unreasonable explanation for our results. tion, starting within days after transplantation. Using this Furthermore, the ability of ST-HSCF, MPPF, and LMPP sensitive approach, we demonstrate that Flk2+ ST-HSCF, (Yang et al., 2005) to form CFU-S shows that erythropoietic MPPF, and LMPP have robust MegE potential in vivo output is a constant feature of these populations. This also and produce more mature cells of the MegE lineage than ruled out the possibility that rare-contaminating LT-HSC committed myeloid progenitors CMP/MEP, as expected provide the erythropoietic readout as MPPF, LMPP, and

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 423 CMP/MEP (an unlikely contaminant) are the only popula- Flk2+ multipotent progenitors with additional surface tions with CFU-S at day 10. ST-HSCF, MPPF, and LMPP markers, such as VCAM1 as recently reported (Forsberg also provide radioprotection following lethal irradiation at et al., 2005; Lai et al., 2005), will also be needed to appre- a much lower cell dose than CMP/MEP, thereby emphasiz- ciate how lineage segregation and/or lineage exclusion ing the robustness of their MegE potential in vivo. Then, we occurs from these multipotent cells. are left with the question: what are the LMPP? They are We and others have previously described KLS subsets multipotent (at least at the population level), they represent of stem and multipotent cells with different marker combi- 1/3 of the total MPPF population (Figure S2), and they nations (Osawa et al., 1996; Goodell et al., 1996; Morrison have lower burst size and overall lineage production than et al., 1997; Ivanova et al., 2002), and the relationships be- MPPF. Based on these results, we conclude that a signifi- tween those populations and the ones described here cant proportion of them are a more mature subset of MPPF have not yet been fully investigated. Thus, there are a vari- that are still located developmentally upstream of CMP ety of multipotent cells from which progeny could undergo and CLP (Figure 7). It is yet to be determined whether lineage segregation into lymphoid and myeloid pathways. they are heterogeneous and include a subset of more Some latent but normally unused lineage plasticity re- committed lymphoid precursors that lack CFU-S activity, mains at these early stages of lineage commitment as and if so, whether at the clonal level such precursors retain CLP, pro-T, and some early pro-B cells are able to revert myelomonocytic potential yet have lost MegE potential. to myeloid fates upon specific transgene expression and As we show here that Flk2+ KLS cells are fully multipo- transcription factor inactivation or overexpression (Nutt tent, Flk2 expression per se does not appear to be a et al., 1999; Kondo et al., 2000; King et al., 2002; Xie marker of lymphoid commitment. Flk2+ KLS cells are pro- et al., 2004). CMP also display low B cell production miscuous in terms of gene expression at the population in vivo (Akashi et al., 2000; Figure S3), but further subdivi- level and coexpress, though at different levels, master sion of this population with additional surface markers, transcriptional regulators, signaling molecules, and cell- such as Flk2 (D’Amico and Wu, 2003; Nutt et al., 2005), surface receptors that are known to be specific for and/ should provide a refined phenotypic definition of these or to specify myeloid or lymphoid fates (Forsberg et al., progenitors. Importantly, we never observed MegE poten- 2005). This type of promiscuous gene expression might tial without GM potential and vice versa from any of the be a defining characteristic of multipotency (Miyamoto progenitor populations investigated (except from MEP et al., 2002). However, it is possible that Flk2 signaling and GMP), indicating that segregation of MegE and GM facilitates lymphoid differentiation prior to commitment potential mainly occurs downstream of the CMP stage to the lymphoid lineage, which is then clearly marked by as shown previously (Akashi et al., 2000). cell-surface expression of the interleukin-7 receptor As erythropoiesis (in contrast to lymphopoiesis) is indis- (IL7R) (Kondo et al., 1997). This would be consistent with pensable for the survival of the animal, it is possible that the phenotype of Flk2 null mice, which have defects pri- multiple mechanisms exist for rapid production of MegE marily in lymphoid cell development but also in myeloid cells and that LT-HSC give rise to mature RBC and plate- reconstitution (Mackarehtschian et al., 1995). In aged lets without first differentiating through some or all of the wild-type mice, KLS cells have defective (but not absent) progenitor intermediates. Although this is an important lymphoid potential upon transplantation and a correlative question, we are not aware of assays that allow a rigorous defect in the numbers of Flk2+ KLS cells (Rossi et al., assessment of this possibility in vivo. The kinetics of line- 2005). As there are lineage preferences, even at the level age repopulation in both sublethally and lethally irradiated of the LT-HSC, during ontogeny (Ikuta et al., 1990; Cu- mice argues, at least, for differentiation occurring with se- mano and Godin, 2001) and aging (Rossi et al., 2005), it quential maturation. In addition, we found a direct correla- is possible that Flk2+ hematopoietic progenitors readout tion between the degree of commitment of the progenitor lineage preferences rather than lineage exclusions; how- cells and their rapidity to give rise to mature myelomono- ever, this is difficult to assess as production of different cytic, B and T cells. The only exception to this temporal types of mature progeny peak at distinct time points. It pattern is the simultaneous detection of platelets from all is interesting to note that 500 transplanted MPPF or the populations with MegE potential. We are currently un- LMPP produced GFP+ donor-derived cells in the same able to determine whether this is due to ‘‘short-cutting’’ range as 10,000 transplanted CMP/MEP at day 9 (myeloid through differentiation steps as proposed by others readout) and 10,000 transplanted CLP at day 20 (lymphoid (Takano et al., 2004); to lack of sensitivity of our assay in readout) (Figure S3), suggesting that about equal numbers detecting very small numbers of donor-derived platelets of CMP and CLP are generated from Flk2+ KLS at the pop- at earlier time points when the levels of host-derived ulation level. Hence, the proposed ‘‘lymphoid preference’’ platelets are still high; to differential responses to signals of Flk2+ KLS cells may mainly reflect easier detection and/ mediating the response to irradiation; or to differential or accumulation of their relatively long-lived lymphoid trafficking to niches supporting megakaryopoiesis. Future progeny. A better understanding of Flk2 function in lym- technical developments allowing the imaging of LT-HSC phoid differentiation and fate determination will require cell division and subsequent determination of cell fate further investigation of the activation level of Flk2 signaling will be needed to test these alternative models. As the im- pathways downstream of the receptor. Separation of mediate microenvironment plays critical roles in cell-fate

424 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. determination, to address these questions it will be im- In Vitro Assays portant to continue to improve the ability to determine For proliferation analysis, cells were cultured in Iscove’s modified Dul- the localization of stem and progenitor cells both under becco’s media (IMDM) containing 10% FCS, 13 penicillin/streptomycin, 2 mM GlutaMax-1, 0.1 mM nonessential amino acid, 1 mM sodium pyru- steady-state conditions (Kiel et al., 2005) and upon trans- vate, 50 mM 2-mercaptoethanol and supplemented with SCF (25 ng/ml), plantation (Cao et al., 2004). Flt3 ligand (25 ng/ml), IL-11 (25 ng/ml), IL-3 (10 ng/ml), Tpo (25 ng/ml), In conclusion, our results support a model of hemato- Epo (2.5 U/ml), and GM-CSF (10 ng/ml) (all purchased from Peprotech poietic development with segregation of myeloid and and R & D). At the indicated time point, aliquots were sampled to count lymphoid lineage potential from multipotent progenitors the total number of live cells capable of excluding trypan blue. (Figure 7). This model is, by definition, a simplification of Gene Expression Analysis complex biological processes, but at the moment it does Total RNA was isolated using Trizol reagent (InVitrogen) from equivalent account for most, if not all, of the differentiation events. numbers of cells, digested with DNaseI to remove DNA contamination, It tolerates hematopoietic plasticity in lineage segregation and used for reverse-transcription according to the manufacturer’s in- at early stages of commitment, and it accommodates structions (SuperScript II kit, InVitrogen). QRT-PCR primers were de- intrinsic lineage preferences during ontogeny and aging. signed using Primer Express software (Applied Biosystems): EpoR This model will likely evolve further when as of yet un- (fwd: GGC AGG AGG GAC ACA AAG G; rev: GCA GGT TGC TCA solved issues and critical questions will be addressed GAA CAC ACT CA), TpoR (fwd: CAA CCA CTC CTA CCT ACC ACT AAG C; rev: AGG AAT GTA TAG GTC TGC AGT AGC A), IL3R (fwd: through technical advances allowing the in vivo analysis TGT CTC CAT CCC CCA CTT CTG; rev GGG TAA ACC TCT GAC of clonal progeny for all progenitors, the in vivo tracking CTC GAC TTG), and b-actin (fwd: GAC GGC CAA GTC ATC ACT ATT of early differentiation events, the identification and char- G; rev: GAC GGC CAG GTC ATC ACT ATT G). All reactions were per- acterization of the niches where HSC and progenitors dif- formed in an ABI-7000 sequence detection system using SYBR Green ferentiate, and the analysis at the single-cell instead of PCR Core reagents and cDNA equivalent of 300 cells per reaction as population level of lineage priming and fate determination. previously described (Forsberg et al., 2005). Expression of the b-actin gene was used to normalize the amount of the investigated transcript.

EXPERIMENTAL PROCEDURES Supplemental Data Supplemental Data include experimental procedures, seven figures, Mice and one table and can be found with this article online at http:// Four- to eight-week-old b-actin GFP C57Bl/6-Thy1.1 transgenic mice www.cell.com/cgi/content/full/126/2/415/DC1/. (Wright et al., 2001) were used as donors and 8- to 10-week-old con- genic C57Bl/6 wild-type mice were used as recipients. All mice were maintained in Stanford University’s Research Animal Facility in accor- ACKNOWLEDGMENTS dance with Stanford University guidelines. We thank D. Bhattacharya for help with FACS sorting and critical read- Flow Cytometry ing of the manuscript, C. Jamieson and D. Bryder for help with meth- Staining and enrichment procedures for LT-HSC, ST-HSCF, MPPF, ylcellulose assays, L. Jerabek for lab management, C. Richter for an- LMPP, CMP, GMP, MEP, and CLP cell sorting and platelet analyses tibody preparations, and L. Hidalgo and B. Lavarro for animal care. were performed as previously described (Akashi et al., 2000; Christen- E.C.F. was a fellow of the Cancer Research Institute and is now sup- sen and Weissman, 2001; Nakorn et al., 2003; Adolfsson et al., 2005). ported by NIH grant 1F32 DK72620-01A1. T.S. is supported by NIH Further information on cell preparation and staining procedures can be grant 1F32 AI058521-01A1. S.K. is a Scholar of the Leukemia and found in Table S1. Cell sorting from b-actin GFP C57Bl/6-Thy1.1 trans- Lymphoma Society and is supported by NCI grant U01-CA84221. genic mice was performed on a modified three-laser (488 nm argon, E.P. was a fellow of the Jose Carreras International Leukemia Founda- 598 nm dye, 407 nm Krypton) MoFlo hybrid controlled by digital Diva tion (FIJC-01/EDTHOMAS). This work was supported in part by NIH electronics (Cytomation, Becton Dickinson). Each subpopulation grants DK053074, CA086065, AI47458, CA86017, and a De Villier was either double or triple sorted and reanalyzed to ensure maximum Award from the Leukemia and Lymphoma Society to I.L.W. I.L.W. purity. Donor-derived GFP+ cells were analyzed on a two-laser (488 nm has stock in Amgen and is a cofounder and member of the Board of argon, 633 nm HeNe) LSR1A (Becton Dickinson). PB GFP+ Plt were Directors of Cellerant Inc. and Stem Cells Inc. The other authors analyzed using a gain of 8 and a threshold of 0 for the FSC channel. have no financial interests to disclose.

In Vivo Assays Received: December 19, 2005 Transplantations were performed by retro-orbital injection of purified Revised: March 12, 2006 cells. Recipient mice were either sublethally irradiated (475 rad) or Accepted: June 5, 2006 lethally irradiated (950 rad, delivered in split dose 3 hr apart) using an Published: July 27, 2006 X-ray source irradiator and given antibiotic-containing water for at least 6 weeks postirradiation. PB was obtained from tail vein bleeding REFERENCES and collected either into K2EDTA-coated collection tubes (Becton Dickinson) for blood parameter analysis or in 4 ml of ACK (150 mM Adolfsson, J., Borge, O.J., Bryder, D., Theilgaard-Monch, K., Astrand- NH4Cl/10 mM KHCO3) containing 10 mM EDTA for flow cytometry analysis. Complete blood counts were done using a CELL-DYN 3500 Grundstrom, I., Sitnicka, E., Sasaki, Y., and Jacobsen, S.E. (2001). Up- analyzer (Abbot Diagnostic) calibrated for mouse blood samples. regulation of Flt3 expression within the bone marrow Lin(À)Sca1(+)c- CFU-S assays were performed as described (Morrison and Weissman, kit(+) stem cell compartment is accompanied by loss of self-renewal 1994) and spleens were fixed in Tellyesniczky’s solution (4 parts water: capacity. Immunity 15, 659–669. 13 ethanol: 2 formaldehyde: 1 glacial acetic acid) to count the nodules. Adolfsson, J., Mansson, R., Buza-Vidas, N., Hultquist, A., Liuba, K., CFU-S per cell population is given as frequency 1/(numbers of injected Jensen, C.T., Bryder, D., Yang, L., Borge, O.J., Thoren, L.A., et al. cells divided by numbers of observed spleen colonies). (2005). Identification of adult Flt3+ lympho-myeloid stem cells lacking

Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc. 425 erythro-megakaryocytic potential: A new road map for blood lineage Morrison, S.J., and Weissman, I.L. (1994). The long-term repopulating commitment. Cell 121, 295–306. subset of hematopoietic stem cells is deterministic and isolatable by Akashi, K., Traver, D., Miyamoto, T., and Weissman, I.L. (2000). A phenotype. Immunity 1, 661–673. clonogenic common myeloid progenitor that give rise to all myeloid Morrison, S.J., Wright, D.E., and Weissman, I.L. (1997). Identification lineages. Nature 404, 193–197. of a lineage of multipotent hematopoietic progenitors. Development Akashi, K., Traver, D., and Zon, L.I. (2005). The complex cartography of 124, 1929–1939. stem cell commitment. Cell 121, 160–162. Na Nakorn, T., Traver, D., Weissman, I.L., and Akashi, K. (2002). Cao, Y.A., Wagers, A.J., Beilhack, A., Dusich, J., Bachmann, M.H., Myeloerythroid-restricted progenitors are sufficient to confer radio- Negrin, R.S., Weissman, I.L., and Contag, C.H. (2004). Shifting foci protection and provide the majority of day 8 CFU-S. J. Clin. Invest. of hematopoiesis during reconstitution from single stem cells. Proc. 199, 1579–1585. Natl. Acad. Sci. USA 101, 221–226. Nakorn, T.N., Miyamoto, T., and Weissman, I.L. (2003). Characteriza- Christensen, J.L., and Weissman, I.L. (2001). Flk2 is a marker in hema- tion of mouse clonogenic megakaryocyte progenitors. Proc. Natl. topoietic stem cell differentiation: a simple method to isolate long-term Acad. Sci. USA 100, 205–210. stem cells. Proc. Natl. Acad. Sci. USA 98, 14541–14546. Nutt, S.L., Heavey, B., Rolink, A.G., and Busslinger, M. (1999). Com- Cumano, A., and Godin, I. (2001). Pluripotent hematopoietic stem cell mitment to the B-lymphoid lineage depends on the transcription factor development during embryogenesis. Curr. Opin. Immunol. 13,166–171. Pax5. Nature 401, 556–562. D’Amico, A., and Wu, L. (2003). The early progenitors of mouse dendritic Nutt, S.L., Metcalf, D., D’Amico, A., Polli, M., and Wu, L. (2005). Dy- cells and plasmacytoid predendritic cells are within the bone marrow namic regulation of PU.1 expression in multipotent hematopoietic hemopoietic precursors expressing Flt3. J. Exp. Med. 198, 293–303. progenitors. J. Exp. Med. 201, 221–231. Forsberg, E.C., Prohaska, S.S., Katzman, S., Heffner, G.C., Stuart, Osawa, M., Hanada, K., Hamada, H., and Nakauchi, H. (1996). Long- J.M., and Weissman, I.L. (2005). Differential expression of novel poten- term lymphohematopoietic reconstitution by a single CD34-low/nega- tial regulators in hematopoietic stem cells. PLoS Genetics 1, e28. tive hematopoietic stem cell. Science 273, 242–245. 10.1371/journal.pgen.0010028. Passegue´ , E., Wagers, A.J., Guiriato, S., Anderson, W., and Weiss- Goodell, M.A., Brose, K., Paradis, G., Conner, A.S., and Mulligan, R.C. man, I.L. (2005). Global analysis of proliferation and cell cycle gene (1996). Isolation and functional properties of murine hematopoietic expression in the regulation of hematopoietic stem and progenitor stem cells that are replicating in vivo. J. Exp. Med. 183, 1797–1806. cell fates. J. Exp. Med. 202, 1599–1611. Hock, H., and Orkin, S.H. (2005). Stem cells: the road not taken. Nature Rossi, D.J., Bryder, D., Zahn, J.M., Ahlenius, H., Sonu, R., Wagers, 435, 573–575. A.J., and Weissman, I.L. (2005). Cell intrinsic alterations underlie hema- Ikuta, K., and Weissman, I.L. (1992). Evidence that hematopoietic stem topoietic stem cell aging. Proc. Natl. Acad. Sci. USA 102, 9194–9199. cells express mouse c-kit but do not depend on steel factor for their Smith, L.G., Weissman, I.L., and Heimfeld, S. (1991). Clonal analysis of generation. Proc. Natl. Acad. Sci. USA 89, 1502–1506. hematopoietic stem-cell differentiation in vivo. Proc. Natl. Acad. Sci. Ikuta, K., Kina, T., MacNeil, I., Uchida, N., Peault, B., Chien, Y.H., and USA 88, 2788–2792. Weissman, I.L. (1990). A developmental witch in thymic lymphocyte Socolovsky, M., Nam, H., Fleming, M.D., Haase, V.H., Brugnara, C., maturation potential occurs at the level of hematopoietic stem cells. and Lodish, H.F. (2001). Ineffective erythropoiesis in Stat5a(À/À) Cell 62, 863–874. 5b(À/À) mice due to decreased survival of early erythroblasts. Blood Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A., 98, 3261–3273. and Lemischka, I.R. (2002). A stem cell molecular signature. Science Spangrude, G.J., Heimfeld, S., and Weissman, I.L. (1988). Purification 298, 601–604. and characterization of mouse hematopoietic stem cells. Science 241, Kiel, M.J., Yilmaz, O.H., Iwashita, T., Yilmaz, O.H., Terhorst, C., and 58–62. Morrison, S.J. (2005). SLAM family receptors distinguish hematopoi- Takano, H., Ema, H., Sudo, K., and Nakauchi, H. (2004). Asymmetric etic stem and progenitor cells and reveal endothelial niches for stem division and lineage commitment at the level of hematopoietic stem cells. Cell 121, 1109–1121. cells: inference from differentiation in daughter cell and granddaughter King, A.G., Kondo, M., Scherer, D.C., and Weissman, I.L. (2002). Line- cell pairs. J. Exp. Med. 199, 295–302. age infidelity in myeloid cells with TCR gene rearrangement: a latent Terskikh, A.V., Miyamoto, T., Chang, C., Diatchenko, L., and Weiss- developmental potential of proT cells revealed by ectopic cytokine man, I.L. (2003). Gene expression analysis of purified hematopoietic receptor signaling. Proc. Natl. Acad. Sci. USA 99, 4508–4513. stem cells and committed progenitors. Blood 102, 94–101. Kondo, M., Weissman, I.L., and Akashi, K. (1997). Identification of Wagers, A.J., Sherwood, R.I., Christensen, J.L., and Weissman, I.L. clonogenic common lymphoid progenitors in mouse bone marrow. (2002). Little evidence for developmental plasticity of adult hematopoi- Cell 91, 661–672. etic stem cells. Science 297, 2256–2259. Kondo, M., Scherer, D.C., Miyamoto, T., King, A.G., Akashi, K., Suga- Weissman, I.L. (2000). Stem cells: units of development, units of regen- mura, K., and Weissman, I.L. (2000). Cell-fate conversion of lymphoid- eration, and units of evolution. Cell 100, 157–168. committed progenitors by instructive actions of cytokines. Nature 407, Wright, D.E., Cheshier, S.H., Wagers, A.J., Randall, T.D., Christensen, 383–386. J.L., and Weissman, I.L. (2001). Cyclophosphamide/granulocyte Lai, A.Y., Lin, S.M., and Kondo, M. (2005). Heterogeneity of Flt3- colony-stimulating factor causes selective mobilization of bone expressing multipotent progenitors in mouse bone marrow. J. Immu- marrow hematopoietic stem cells into the blood after M phase of the nol. 175, 5016–5023. cell cycle. Blood 97, 2278–2285. Mackarehtschian, K., Hardin, J.D., Moore, K.A., Boast, S., Goff, S.P., Xie, H., Ye, M., Feng, R., and Graf, T. (2004). Stepwise reprogramming and Lemischka, I.R. (1995). Targeted disruption of the flk2/flt3 gene of B cells into macrophages. Cell 117, 663–676. leads to deficiencies in primitive hematopoietic progenitors. Immunity Yang, L., Bryder, D., Adolfsson, J., Nygren, J., Mansson, R., Sigvards- 3, 147–161. son, M., and Jacobsen, S.E. (2005). Identification of Lin(À)Sca1(+)- Miyamoto, T., Iwasaki, H., Reizis, B., Ye, M., Graf, T., Weissman, I.L., kit(+)CD34(+)Flt3-short-term hematopoietic stem cells capable of and Akashi, K. (2002). Myeloid or lymphoid promiscuity as a critical rapidly reconstituting and rescuing myeloablated transplant recipients. step in hematopoietic lineage commitment. Dev. Cell 3, 137–147. Blood 105, 2717–2723.

426 Cell 126, 415–426, July 28, 2006 ª2006 Elsevier Inc.