Hierarchical and Ontogenic Positions Serve to Define the Molecular Basis of Human Hematopoietic Stem Cell Behavior

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

provided by Elsevier - Publisher Connector

Developmental Cell, Vol. 8, 651–663, May, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.devcel.2005.03.004 Hierarchical and Ontogenic Positions Serve to Define the Molecular Basis of Human Hematopoietic Stem Cell Behavior

Farbod Shojaei,1,2 Jennifer Trowbridge,1,2 define HSCs (Enver et al., 1998). Since transplantable Lisa Gallacher,1 Lou Yuefei,1 David Goodale,1 HSCs can give rise to progenitors detected in vitro, Francis Karanu,1 Krysta Levac,1 whereas in vitro progenitors lack multilineage repopu- and Mickie Bhatia1,2,* lating function in vivo, a hierarchical arrangement of 1Stem Cell Biology and Regenerative Medicine hematopoietic cells can be assembled based on these Robarts Research Institute functional assays in which repopulating HSCs are posi- 100 Perth Drive tioned at the apex, followed by more abundant progeni- London, Ontario N6A 5K8 tors with more restricted proliferative and develop- Canada mental potential (Orkin and Zon, 2002; Verfaillie, 2002). 2 Department of Microbiology and Immunology In addition to the functional relationships that define The University of Western Ontario progenitors and HSCs, the ontogenic origin of mamma- London, Ontario N6A 5K8 lian HSCs is also related to changes in overall HSC ca- Canada pacity (Geiger and Van Zant, 2002). Both proliferative ability and differentiative potential of HSCs decline with age, e.g., HSCs obtained from the fetus possess Summary greater proliferative and differentiative capacity than those isolated from adult mice (Geiger and Van Zant, The molecular basis governing functional behavior of 2002). Since hematopoietic function must be sustained human hematopoietic stem cells (HSCs) is largely un- for the lifetime of an organism and is therefore depen- known. Here, using in vitro and in vivo assays, we dent on HSCs, rigid control of HSC survival and differ- isolate and define progenitors versus repopulating entiation decisions throughout the aging and develop- HSCs from multiple stages of human development for ment process is critical (Geiger and Van Zant, 2002; global gene expression profiling. Accounting for both Verfaillie, 2002). Despite these biological observations the hierarchical relationship between repopulating and collective functional understanding of mammalian cells and their progenitors, and the enhanced HSC HSCs, the molecular mechanisms that tightly regulate function unique to early stages of ontogeny, the hu- HSC physiology (Murdoch et al., 2001; Orschell-Tray- man homologs of Hairy Enhancer of Split-1 (HES-1) coff et al., 2000; Domen et al., 2000) and cell fate remain and Hepatocyte Leukemia Factor (HLF) were iden- poorly understood (Lemischka, 1999). In the current tified as candidate regulators of HSCs. Transgenic hu- study, we use in vitro and in vivo assays, combined with man hematopoietic cells expressing HES-1 or HLF prospective isolation of cell populations defined to pos- demonstrated enhanced in vivo reconstitution ability sess properties of either progenitors or repopulating that correlated to increased cycling frequency and in- HSCs from multiple stages of human development, for hibition of apoptosis, respectively. Our report identi- global gene expression profiling. We reveal two regula- fies regulatory factors involved in HSC function that tory transcription factors involved in human HSC func- elicit their effect through independent systems, sug- tion that elicit their effect through independent mecha- gesting that a unique orchestration of pathways fun- nisms capable of controlling stem cell behavior in vivo. damental to all human cells is capable of controlling stem cell behavior. Results

Introduction Functional Characterization and Isolation of Human Hematopoietic Progenitors and Repopulating Stem Rare hematopoietic stem cells (HSCs) are responsible Cells during Human Development for production and maintenance of mature human Based on our current knowledge of human HSC char- blood cells (Morrison et al., 1995). HSCs are defined by acteristics, molecular profiling that discriminates genes two functional properties: vast proliferative self-renewal expressed in HSCs from those expressed in more ma- ability, and multilineage hematopoietic differentiation ture progenitors (hierarchical) and those genes that dis- potential (Morrison et al., 1995). In the human, cellular tinguish functionally unique HSCs at different stages of characteristics of primitive hematopoietic cells and pu- mammalian development (ontogenic) should be taken tative HSCs can only be experimentally investigated by into account to best define the molecular basis of HSC using surrogate approaches (Dick, 1996). In vitro pro- behavior. Using this premise, subsets of lineage- genitor assays are widely used to measure primitive depleted (Lin−) human hematopoietic cells with multi- colony-forming unit (CFU) (Sutherland et al., 1989) ac- lineage repopulating capacity (Lin−CD34+CD38−), and tivity, but this assay is not indicative of stem cell func- Lin− hematopoietic cells capable of in vitro progenitor tion since it is short term and limited to myeloid progen- function but devoid of in vivo repopulating ability itor detection. Only in vivo xeno-transplantation assays unique to HSCs (Lin−CD34+CD38+), were isolated. (Dick, 1996) are able to assess the multilineage differen- Lin−CD34+CD38− and Lin−CD34+CD38+ were purified tiation and proliferative self-renewing properties that from Lin− fractions of human fetal blood (FB), cord blood (CB), and mobilized peripheral blood (MPB) with *Correspondence: [email protected] more than 99% purity upon reanalysis (Figure 1A). Al- Developmental Cell 652

Figure 1. Functional Characterization of Purified Human FB, CB, and MPB Hematopoietic Subsets with In Vitro and In Vivo Assays (A) Isolation of Lin−CD34+CD38− and Lin−CD34+CD38+ populations from FB, CB, BM (not shown), and MPB. Lin− from the above-mentioned populations (biological and experimental replicates from each preparation are detailed in Table S1) were enriched and stained with anti- CD34-APC and anti-CD38-PE and were sorted according to the sorting gates assigned for CD34+CD38− (I) and CD34+CD38+ (III) populations. Reanalysis of sorted fractions (II and IV) verified purity of the populations of interest from each stage of ontogeny. (B) CFU capacity: 1200 cells from each population were plated in methyl cellulose supplemented with SCF, GM-CSF, IL-3, and EPO, followed by total colony counts after 14 days of incubation at 37°C. (C) Repopulating capacity: cells from each population were intravenously transplanted into NOD/SCID mice at a range of 5,000–10,000 Lin−CD34+CD38− cells, and from 10,000 to 200,000 Lin−CD34+CD38+ cells from all human ontogenic sources indicated. Human engraftment was evaluated 6 weeks posttransplant by staining cells harvested from BM of transplanted mice with monoclonal antibody against human CD45. The level of engraftment represents the percentage of human hematopoietic chimerism in the BM for recipient mice. (“**” and “*,” statistically different, p < 0.05). (D) BM cells of transplanted animals were stained with several cell surface markers to examine the percentage of myeloid (CD33, CD15), lymphoid (CD19, CD20), and primitive (CD34, CD38) cells indicative of multilineage human hematopoietic engraftment. Averages and SEM for each subset are provided in quadrants based on 4–8 independent experiments for FB, CB, and MPB repopulating HSCs respectively, n = 18. (E) The subsets indicated were stained with Ki67-FITC antibody as per the protocol (see Experimental Procedures) and analyzed by flow cytometry. “*,” statistically different, p < 0.01. (F) Proposed model to account for hierarchical and ontogenic position of human HSCs and nonrepopulating progenitors. Defining the Molecular Nature of Human HSCs 653

though MPB cells do not represent a normal stage of HSCs possess superior proliferative and self-renewal human hematopoietic ontogeny (since primitive cells capacity compared to their aged counterparts. Based are induced to the circulation via cytokine administra- on our functional observations, we suggest a model to tion), MPB represents an established and accessible position human HSCs that accounts for both hierarchi- source of adult HSCs for fundamental investigation. cal and ontogenic characteristics of the primitive Within each stage of ontogeny (fetal, neonatal, and hematopoietic compartment in the human (Figure 1F). adult), clonogenic progenitor capacity measured by the This model illustrates that human HSCs versus progeni- in vitro CFU assay (Sutherland et al., 1990) was similar tors are restricted to a specific population of hemato- between Lin−CD34+CD38− and Lin−CD34+CD38+ cells, poietic cells throughout life, suggesting that the hierar- with adult MPB sources possessing far less clonogenic chical arrangement of the hematopoietic system is progenitor capacity, independent of the population iso- conserved during development, but also accounts for lated (Figure 1B). However, in vivo functional analysis higher proliferative potential of fetal HSCs, and the by intravenous (IV) transplantation into immune-defi- gradual decline in this property within human HSCs to- cient recipient NOD/SCID mice demonstrated that only ward adult life. We suggest that the gene expression Lin−CD34+CD38− from human FB, CB, and MPB pos- profiles that regulate functional capacity of human sessed repopulating ability (Figure 1C). Therefore, these HSCs are more abundantly expressed in FB HSCs ver- purified populations represent a hierarchical distinction sus neonatal and adult sources, and that these profiles within the human hematopoietic system in which are restricted to HSCs versus progenitors at all stages Lin−CD34+CD38− repopulating cells include functional of human development. human HSCs (termed “repopulating cells” herein) and Lin−CD34+CD38+ are restricted to the progenitor com- Molecular Profiling of Purified Cells Comprising partment (termed “nonrepopulating progenitors” herein). the Human Hematopoietic Hierarchy This hierarchical relationship was conserved at all throughout Human Ontogeny stages of human ontogeny, indicating that the pheno- Utilizing the conserved hierarchical relationship be- type used to define putative human HSCs and re- tween repopulating HSCs and nonrepopulating progen- stricted progenitors is reliable and provides a useful itors, and the ontogenic relationships among human means to distinguish these subsets throughout human HSCs functionally defined and modeled in Figures 1A– development. 1F, we established a strategy to define the molecular In addition to the consistency of HSC and progenitor basis of human HSC behavior through global gene ex- function within these two subsets, detailed analysis of pression analysis of purified subsets by using microar- engrafted animals indicated a gradual decline in human ray. To provide an in depth assessment, several experi- HSC repopulating ability from FB to CB to MPB, where mental and biological replicates were used (Table S1, both the level and frequency of human engraftment in see the Supplemental Data available with this article FB was greater than CB and, in turn, greater than MPB online). To simultaneously account for hierarchy and (Figure 1C). Furthermore, the level and frequency of re- ontogenic positioning of HSCs, our strategy for molec- populating capacity are higher in FB HSCs versus CB ular analysis used several filters (Figure 2A). First, we and MPB HSCs, and, in turn, CB HSC capacity is higher applied hierarchical discrimination that identified all than that of MPB HSCs (Figure 1C). Other than a pro- genes expressed in repopulating Lin−CD34+CD38− pensity for more myeloid versus lymphoid differentia- fractions of FB, CB, bone marrow (BM), and MPB that tion arising from reconstituting FB HSCs, all sources were differentially expressed from nonrepopulating pro- of human HSCs possessed multilineage differentiation genitors at each stage of ontogeny. Adult BM was capacity in vivo (Figure 1D). The gradual reduction of reconstitution capacity from FB to CB to adult MPB added in these analyses to provide an additional HSCs prompted us to examine cell cycle status (Glimm source of adult HSCs. This filter provided genes et al., 2000) of Lin−CD34+CD38− cells and Lin−CD34+ uniquely enriched in repopulating human HSCs at all CD38+ progenitors from all stages of human ontogeny stages of human development and included 366 and by using Ki67 staining (Glimm et al., 2000)(Figure 1E). 329 genes up- and downregulated, respectively (Tables Higher cycling activity in all nonrepopulating progeni- S3 and S4). Self-organizing map (SOM) (Tamayo et al., tors compared to repopulating cells confirmed the 1999) clustering provided correlative nodes of dif- unique quiescent state associated with de novo iso- ferentially expressed genes identified (Figure S1). In the lated HSCs (Srour et al., 2001; Murdoch et al., 2001). second filter, ontogenic discrimination was applied to However, consistent with reconstituting capacity (Fig- identify HSC-enriched transcripts that were preferen- ure 1C), FB Lin−CD34+CD38− cells demonstrated greater tially expressed in early versus later stages of human cycling activity compared to similar cells from CB and development (Figure 2A). Third, we further selected MPB (p < 0.05, Figure 1E). Collectively, these data re- genes based on the hypothesis that transcription veal that differences between repopulating capacity of factors are likely to play a substantive role in HSC regu- HSCs isolated from different stages of human ontogeny lation (Ramalho-Santos et al., 2002). This bioinformat- correlate with differences in cell cycle activity that are ics strategy, unique to human hematopoietic cell mo- likely to affect the in vivo proliferative capacity of HSCs. lecular profiling, resulted in the identification of 16 Despite the fact that the HSCs have similar cell sur- candidate transcription factors (Figure 2B), of which face phenotypes and functional properties at various STAT-1 (Jenkins et al., 2002), Id3 (Cooper et al., 1997), stages of human development, our results demonstrate Ets1 (Dieterlen-Lievre et al., 1993), HLF (Crable and An- an ontogenic distinction of human repopulating HSCs, derson, 2003), and HES-1 (Kunisato et al., 2003) have but not of more mature progenitors: ontogenically early been previously implicated in primitive hematopoiesis. Developmental Cell 654 Defining the Molecular Nature of Human HSCs 655

Identification of HES-1 and HLF as Candidate profiling studies (Georgantas et al., 2004; Ivanova et al., Regulators of Human Hematopoietic 2002). To functionally examine a potential role of HES-1 Repopulating Stem Cells and HLF in human HSCs, full-length cDNA coding se- Of the 16 candidates identified (Figure 2B), HES-1 and quences for each gene were subcloned into a retroviral HLF expression best mirrored the functional model we vector under the control of a constitutively active pro- predicted to be the most relevant to overall HSC beha- moter and upstream of the reporter gene-enhanced vior (Figure 1F). Quantitative real-time PCR (Q-RT-PCR) green fluorescent protein (GFP) (Figure 2F). Human CB using freshly isolated FB, CB, and adult BM and MPB Lin− cells containing both CD34+CD38− and CD34+ Lin−CD34+CD38− cells demonstrated a declining pat- CD38+ subsets were used to evaluate the role of HES-1 tern of expression from FB to neonatal to adult sources and HLF based on subsequent functional readout of (Figure 2C). In addition, the ratio of HES-1 and HLF ex- repopulating HSCs and nonrepopulating progenitors pression between Lin−CD34+CD38− repopulating cells overexpressing the transduced gene. Following trans- and Lin−CD34+CD38+ progenitors was consistently duction of primary human Lin− CB cells, the in vitro greater than 1 (i.e., HSC expression > progenitor ex- proliferative capacity, phenotype, progenitor potential, pression) at all stages of ontogeny, and this ratio declined and in vivo repopulating function of HES-1 and HLF from early to late stages of HSC/progenitor development transgenic human hematopoietic cells were compared (Figures 2D and 2E). This pattern of expression was to vector-transduced controls. Our experimental strat- unique to HES-1 and HLF-1 in comparison to other egy is illustrated in Figure 2G. HES-1, HLF, and vector genes examined in a similar manner (data not shown). alone retrovirally transduced hematopoietic cells were Accordingly, HES-1 and HLF were selected for further analyzed by flow cytometry for GFP and cell surface functional analysis. CD34 expression (Figure 2H). No difference in retroviral HES-1 is a basic-helix-loop-helix transcriptional re- transduction (number of GFP+ cells, Figure 2I) or total pressor that inhibits differentiation of many cell types number of primitive CD34+ cells was detectable (Figure (Ishibashi et al., 1994; Kageyama et al., 2000; Nakamura 2J). These transgenic human cells overexpressing HES-1 et al., 2000) and is a downstream target of the Notch and HLF were used in functional in vitro and in vivo signaling pathway (Karanu et al., 2000b, 2001, 2003; assays to evaluate the affects of these genes on pro- Kumano et al., 2003; Varnum-Finney et al., 2000). A re- genitor and HSC behavior, respectively. cent study has shown that HES-1 is able to maintain the reconstitution ability of mouse CD34−Lin−c-Kit+ HES-1 and HLF Augment Human Hematopoietic Sca-1+ HSCs (Kunisato et al., 2003). HLF is closely re- Stem Cell Activity In Vivo lated to zipper-containing transcription factors that Transduction of HES-1 or HLF had no effect on in vitro- have a role in developmental stage-specific gene ex- detected progenitors, suggesting that overexpression pression (Inaba et al., 1992), and it has recently been of these genes does not impact functional properties shown to activate the LMO2 promoter (Crable and An- of nonrepopulating cells. Transduced (GFP+) hemato- derson, 2003) necessary for initiation of mammalian poietic progenitor frequency or lineage composition embryonic hematopoiesis (Warren et al., 1994). In addi- were not affected by overexpression of the HES-1 or tion, HLF was found to be involved in acute human leu- HLF transgene (Figures 3A and 3B) since erythroid, kemias when fused to the E2A basic-helix-loop-helix macrophage, and granulocytic CFU subtypes were transcription factor (Honda et al., 1999; Inaba et al., equally identified. 1992) and was recently identified in two other molecular Similar numbers of HES-1-, HLF-, and vector-trans-

Figure 2. Identification of Candidate Transcription Factors Regulating Human HSCs with a Unique Strategy Based on Hierarchical and Onto- genic Positioning (A) Using global gene expression analysis of human Lin−CD34+CD38− from FB, CB, BM, and MPB and Lin−CD34+CD38+ progenitors, transcripts were interrogated by microarray analysis. Bioinformatics-based priority filters were applied as schematically illustrated and included hierarchical and ontogenic discrimination, followed by the selection of candidate DNA binding transcription factors. (B) A total of 16 transcripts were identified by using the strategy shown in (A). A complete list of differentially regulated genes is available as Table S3. Abbreviations: POLD4, polymerase delta4; TEF4, transcriptional enhancer factor4; HES-1, hairy enhancer of split; LIM, zinc binding domain present in Lin-11, Isl-1, Mec-3; HLF, hepatic leukemia factor; PAX8, paired box gene8; Id3, inhibitor of DNA binding3; HOXA6, homeobox A6; ETS1, erythroblastosis virus E26 oncogene homolog 1; STAT1, signal transducer and activator of transcription; ZNFP1&2, zinc finger protein; EST, expressed sequenced tag; MBLL, melanogaster muscle Blind B; PHRET1, PH domain containing protein in retina 1. (C) Expression profile of HES-1 and HLF in freshly isolated Lin−CD34+CD38− cells derived from all stages of human ontogeny using Q-RT-PCR. (D and E) Q-RT-PCR was used to analyze the ratio of (D) HES-1 and (E) HLF expression in Lin−CD34+CD38− to Lin−CD34+CD38+ from FB, CB, and MPB sources. All Q-RT-PCR analysis was normalized to GAPDH transcript, and differential expression was calculated by using the ⌬⌬Ct equation. “*,” statistically different than corresponding expression in Lin−CD34+CD38+ cells, p < 0.05. (F) Full-length cDNA of HES-1 was cloned in the EcoRI and XhoI sites, and HLF full-length cDNA was cloned in the EcoRI and SalI sites of the MIEV retrovirus vector as illustrated. (G) Schematic illustration of experimental design to functionally analyze the effect of HES1 and HLF overexpression by using retroviral gene reconstitution. (H) Lineage-depleted population from human CB was transduced with viral supernatant containing HES-1, HLF, and vector alone. Gene transfer efficiency was determined by the percentage of GFP+ cells 6 days posttransduction by using flow cytometry and primitive hematopoietic phenotype using CD34 expression. (I and J) Summary of the total number of (I) GFP+ and (J) CD34+GFP+ transgenic cells 6 days posttransduction. Developmental Cell 656

Figure 3. Functional Analysis of HLF and HES-1 Overexpression in Primitive Populations of Human Hematopoietic Cells (A) Number of clonogenic progenitors transduced with HES-1 (blue bars), HLF (yellow bars), and vector (black bars) by using the CFU assay. A total of 1000 transgenic (GFP+ sorted) cells were plated into methylcellulose supplemented with SCF, GM-CSF, IL-3, and EPO, and differential colony counts of CFU subtypes (erythroid, macrophage, and granulocyte) were determined and compared after 14 days. Defining the Molecular Nature of Human HSCs 657

duced cells were intravenously transplanted into recipi- HES-1 and HLF Act Independently to Enhance Cell ent NOD/SCID mice to investigate the in vivo effect of Cycle and Inhibit Apoptosis of Human these genes on repopulating human HSCs. Multilineage Hematopoietic Stem Cells, Respectively human reconstitution of HES-1 and HLF transgenic hu- Based on the robust effect of HES-1 and HLF in en- man HSCs did not differ from vector-transduced repop- hancing human HSC function independent of HSC- ulating cells, and both lymphoid and myeloid lineages, specific homing properties to the bone marrow in trans- along with primitive subpopulations of CD34+ cells, planted recipients, we sought to understand the were detectable at similar frequencies in engrafted re- underlying mechanism involved through the consider- cipients (Figure 3C). This suggests that HES-1 and HLF ation of fundamental pathways that impact all human do not affect lineage differentiation of human HSCs. cell types, such as cell cycle and apoptotic status. Brd-U However, mice transplanted with HES-1 or HLF trans- incorporation and 7-AAD staining of transgenic cells genic HSCs revealed a dramatic increase in the level of were used to determine the cell cycle status by analyz-

BM reconstitution and number of HSCs, as indicated ing the frequency of primitive CD34+ dormant (G0–G1) by increased frequency of engrafted recipients, n = 35 and cycling (S and G2+M) cells (Figure 4A). While HLF (Figure 3D). Based on the levels of human chimerism, transgenic and vector control cells were similar to each human HSC capacity was enhanced by w6-fold upon other with respect to distribution among the different overexpression of HES-1 and w4-fold upon overex- stages of the cell cycle, a greater proportion of HES-1 pression of HLF compared to vector-transduced HSCs transgenic cells was actively cycling, and a corre- (Figure 3G). Therefore, despite the lack of effect on pro- spondingly smaller proportion was found to be quies- genitor capacity, HSC repopulation potential was dra- cent (Figure 4B). Based on these observations, we matically enhanced in transgenic cells overexpressing sought to evaluate the affect of HES-1 on cell cycle HES-1 or HLF. These functional analyses validate our activity on transgenic human HSCs compared to HLF- original strategy to identify key molecular regulators of and vector-transduced cells in vivo by Brd-U labeling. HSC behavior in vivo that are distinct from regulators Brd-U is passively taken up by cells and stably incorpo- of nonrepopulating hematopoietic progenitors. rated into the DNA of cells undergoing mitotic division, Since HSC capacity was evaluated in vivo by IV thereby providing a retrospective method to evaluate transplantation, increased repopulating capacity of actively dividing parent cells that give rise to Brd-U+ transgenic HES-1 and HLF human HSCs may be due to progeny. Recipient mice were transplanted with human enhanced ability to home to the recipient bone marrow HSCs overexpressing HES-1 or HLF and were adminis- compartment. Cell surface expression of proteins asso- tered Brd-U for 3 days prior to sacrifice between 5–6 ciated with increased homing capacity, CD44, CD49, weeks posttransplantation. Our experimental design is and CXCR4, were equally expressed on HES-1-, HLF-, illustrated in Figure 4C. Mice transplanted with trans- and vector-transduced cells (Figure 3E). To functionally genic human HES-1 cells demonstrated a larger num- evaluate the effect of HES-1 and HLF on homing ability ber of primitive CD34+ cells incorporating Brd-U as of HSCs, human cells overexpressing HES-1 and HLF compared to either HLF- or vector-transduced repopu- were seeded directly into the femoral bone marrow lating HSCs (Figure 4D). These data indicate that HES-1 space of recipient mice by using intrafemoral bone is able to augment proliferative capacity of transgenic marrow transplantation (IBMT) (Mazurier et al., 2003a). human HSCs in vivo, whereas HLF overexpression had HSC capacity of transgenic cells delivered by IBMT no influence on cell cycle status of human HSCs. was compared to HSCs transduced with vector alone. While HES-1 overexpression increases cycling of Similar to results observed after IV transplantation (Fig- primitive cells, HLF transgenic cells had a similar cell ure 3D), IBMT of transgenic HSCs demonstrated that cycling status profile to vector-transduced cells, sug- human HSC activity was equally enhanced by overex- gesting that HLF confers enhanced repopulation ca- pression of HES-1 and HLF,n=9(Figure 3F). These pacity on human HSCs via an alternative mechanism results indicate that HES-1 and HLF do not likely affect (Figures 4A–4E). Given previous observations of the homing capacity of human HSCs, and prompted eluci- ability of HLF to inhibit apoptosis (Hitzler et al., 1999), dation of alternative mechanisms. we investigated the apoptotic status of transgenic

(B) Representative examples HES-1, HLF, and vector alone transgenic progenitors detected in the CFU assay confirming transgene expression in multiple hematopoietic lineages. (C) Comparison of multilineage human hematopoietic reconstitution from HES-1, HLF, and vector transgenic repopulating HSCs in the BM of transplanted mice. Approximately equal numbers of transgenic (GFP+) cells (average of 75,000) were transplanted in each mouse by using a total of seven independent human CB samples, n = 35. Mouse BM was stained with anti-CD45-APC (to identify human cells); anti-CD10 and anti-CD19-PE (to identify B lymphoid lineage), anti-CD13, -CD14, -CD15, and -CD33-PE (to identify myeloid lineage); and anti-CD34 (to identify primitive human hematopoietic component) antibodies and analyzed by flow cytometry 6 weeks posttransplantation of human cells. (D) The level of human hematopoietic repopulating capacity was compared as a percentage of human GFP+ transgenic cells, and the frequency of engrafted mice intravenously transplanted with transgenic HES-1-, HLF-, and vector alone-transduced cells 6 weeks after transplantation is shown. “*,” the average level of engraftment is statistically different from the level of repopulation in mice transplanted with vector-transduced HSCs, p < 0.05. (E) Frequency of transgenic cells expressing cell surface molecules CD44, CD49, and CXCR4 associated with HSC homing to the bone marrow compartment. (F) The average level of human hematopoietic repopulation (inset) from HES-1, HLF, and vector alone transgenic HSCs delivered directly to the recipient bone marrow site by intrafemoral bone marrow transplantation (IBMT), n = 9. “*,” the average level of engraftment is statistically different from the level of repopulation in mice transplanted with vector-transduced HSCs, p < 0.05. Developmental Cell 658

Figure 4. Characterization of Cell Cycle and Apoptosis Regulation in HES-1-, HLF-, and Vector-Transduced Primary Human Hematopoietic Cells (A) Human CB CD34+ cells transduced with HES-1, HLF, and vector alone were pulsed with Brd-U for 2 hr, followed by fixation and washing, and were stained with anti-Brd-U-APC and 7-AAD to analyze the fraction of cells at different stages of the cell cycle, including G0+G1, S, and G2+M phases. (B) Quantitative summary of cycling and noncycling HES-1 (blue bars), HLF (yellow bars), and vector alone (black bars) transgenic cells. “*,” statistically different from HLF- and vector-transduced cells, p < 0.05. (C) Schematic illustration of the experimental design used to evaluate mitotically active cells repopulating recipient mice. Defining the Molecular Nature of Human HSCs 659

cells. Annexin-V staining for preapoptotic cells (Figure Several candidate regulators of human stem cell fate 4F) revealed a significantly smaller Annexin-V+ popula- have also been implicated or originally identified from tion in HLF transgenic CD34+ cells compared to either human cancers. As such, transformation events and HES-1- or vector control-transduced CD34+ cells (Fig- pathways influenced by transformation may be related ure 4G), suggesting that HLF transgenic cells are less to the regulators of human HSCs (Pardal et al., 2003). prone to apoptotic cell death. Since mechanisms and Genetic rearrangement of HLF with E2A has been ob- targets of apoptotic control are species and cell served in human leukemias and associated with re- context specific (Danial and Korsmeyer, 2004), we duced apoptosis in leukemic progenitors (Kurosawa et aimed to further characterize the potential target genes al., 1999). Using PCR amplification, we were unable to associated with apoptotic control via HLF that specifi- detect HLF-E2A transcripts in primary HLF-transduced cally affect human hematopoietic cells. Within the cells (Figure 5B), indicating that our observation of HLF context of hematopoietic cells, SODD (Takada et al., apoptotic inhibition is independent of primary human 2003), TOSO (Song and Jacob, 2005), and Bcl-2 (Cory cells harboring HLF-E2A mutations. Similarly, the Notch et al., 2003) have been identified as factors controlling pathway has been implicated in human leukemias, and apoptosis. Accordingly, primitive human hematopoietic it has been shown in nonhuman systems to affect HES-1 cells transduced with HES-1 or HLF were isolated and expression. To evaluate whether HES-1 could be examined quantitatively for changes in SODD, TOSO, targeted for regulation, human HSCs were purified and and Bcl-2 expression compared to vector-transduced treated with Jagged-1 (Karanu et al., 2000b). Short- cells (representative Q-RT-PCR analysis, Figure 4H). term stimulation of purified cells with Jagged-1 demon- Our results demonstrate that overexpression of HLF in strated an increase in HES-1 expression within 24 hr, primitive human hematopoieitc cells induces Bcl-2 ex- compared to protein control (IgG)-treated cells (Figure pression, but has no effect on the levels of either SODD 5C). Upregulation of HES-1 could be sustained by 72 or TOSO (Figure 4I). In contrast, cells’ overexpression hr, but it required the addition of Jagged-1 (Figure 5C), of HES-1 did not alter Bcl-2, SODD, or TOSO levels, suggesting that HES-1 could be upregulated but re- providing further evidence of the inability of HES-1, as quires continual stimulation similar to that achieved by compared to HLF, to influence apoptosis of transgenic overexpression of HES-1 in transgenic HSCs regulated cells (Figure 4I). We suggest that HLF targets antiapo- by retroviral vectors. We suggest that HES-1 and HLF ptotic pathways associated with complex Bcl-2 regula- enhance the reconstitution capacity of human HSCs tion in human HSCs, similar to effects observed by through independent systems, and that pathways gov- using Bcl-2 mutations in mouse HSCs (Domen et al., erning fundamental cell physiology, which include cell 2000). Taken together, our results reveal that HES-1 and cycle regulation and apoptotic control, are capable of HLF promote cell cycling and reduce apoptosis, modulating human HSC function. respectively, thereby enhancing repopulating capacity of human HSCs via independent pathways. To characterize HES-1 and HLF gene targets, and to Discussion further investigate potential downstream signaling of these two transcriptional regulators, transgenic cells The molecular basis that defines human HSCs has were isolated and analyzed by Q-RT-PCR. As would be been under intensive investigation (Georgantas et al., predicted, HES-1- and HLF-transduced cells demon- 2004; Ivanova et al., 2002; Ramalho-Santos et al., 2002; strated a 20- and 27-fold greater level of HES-1 and Steidl et al., 2002), but few insights into complex HSC HLF expression, respectively, compared to vector- behavior have emerged. The frequency of repopulating transduced cells (Figure 5A). Consistent with their ob- HSCs or progenitors enriched in all sources of human served independent mechanisms of action, neither Lin−CD34+CD38− and Lin−CD34+CD38+ cells sug- HES-1 nor HLF affected the expression of the other gests that the vast majority of cells within these popula- (Figure 5A), suggesting limited overlap between their tions are in fact not HSCs or progenitors, and therefore respective pathways. Similar to other studies (Ronchini these purified subsets represent extremely hetero- and Capobianco, 2001), HES-1 upregulated expression geneous populations not likely suitable for molecular of CDK-2 but had no effect on a recently identified pu- profiling (Iscove et al., 2002). Despite these concerns, tative target gene of HLF, LMO2 (Crable and Anderson, our unique strategy allowed for identification of two 2003). Conversely, HLF upregulation increased LMO2 candidates’ genes capable of regulating repopulating expression (Figure 5A) but had no effect on CDK-2 ex- HSCs and not progenitors, demonstrating the impor- pression. The distinct roles of HES-1 and HLF are fur- tance of these genes in functional features used to de- ther supported by the distinct clustering location of fine HSCs enriched in the Lin−CD34+CD38− popula- these two genes in SOM analysis (Figure S1). tion. Arrival at this result suggests that functional

(D and E) (D) Representative analysis and (E) summary of human hematopoietic cells repopulating the bone marrow of recipient NOD/SCID that have incorporated Brd-U after in vivo administration. “*,” statistically different from HLF- and vector-transduced cells, p < 0.05. (F and G) (F) HES-1-, HLF-, and vector-transduced CD34+ cells were fixed and stained with Annexin-V-APC, and the percentage of preapoptotic cells is summarized in (G) and expressed as the percentage of apoptotic cells relative to vector control-transduced cells. “*,” statistically different from HES-1- and vector-transduced cells, p < 0.05. (H and I) (H) Representative Q-RT-PCR of BCL-2, SODD, and TOSO expression in human hematopoietic cells transduced with HES-1, HLF, or vector alone as indicated, and the (I) average level of gene expression summarized. GAPDH was used as a housekeeping control to normalize cDNA input. The average level of gene expression was based on the results of four independent experiments. “*,” the BCL-2 gene expression level is statistically different from that in HES-1- and vector-transduced cells, p < 0.05. Developmental Cell 660

Figure 5. Target Gene Regulation in HES-1-, HLF-, and Vector-Transduced Primary Human Hematopoietic Cells (A) Q-RT-PCR of HES-1, HLF, CDK-2, and LMO2 genes with cDNA from primary human hematopoietic cells transduced with HES-1 or HLF as indicated. “*,” statistically different, p < 0.05. (B) Conventional PCR of cDNA. Lanes: 1, molecular marker; lanes 2, 3, and 4, primary human CB cells transduced with HLF; lane 5, positive control using diagnosed human acute lymphoblastic leukemia sample containing HLF-E2A rearrangement; lane 6, No-RT control. (C) Regulation of HES-1 in purified Lin−CD34+CD38− cells evaluated by Q-RT-PCR in response to Jagged-1 treatment at 24 and 72 hr as compared to control-treated cells. “*,” statistically different, p < 0.05. assays used to determine the frequency of human re- capable of regulating human HSC capacity via modula- populating cells in purified subsets underestimate HSC tion of the cell cycle involving upregulation of CDK-2 content, and/or these purified populations are more gene expression. Unexpectedly, our analysis also iden- molecularly homogeneous than originally predicted. tified the DNA binding transcription factor HLF as a Since our study identifies two candidate genes that uniquely associated factor involved in maintaining the modulate the repopulation capacity of human HSCs, repopulating capacity of human HSCs by conferring our unique strategy that accounts for both hierarchical antiapoptotic effects and preventing premature cell and ontogenic positioning of human HSCs themselves death that are mediated in part by Bcl-2 regulation. The may have enabled these findings. By first defining hu- idea that HES-1 and HLF impact HSC function via two man HSCs and progenitors via prospective purification different mechanisms, and that these mechanisms act and use of established functional assays, our approach through fundamental pathways (cell cycle and apopto- to classify genes involved in HSC regulation could be sis) common to all human cells, suggests that specific tested via gene reconstitution of identified candidates HSC genes may not molecularly define the basis of by using the same functional criteria of HSC repopula- HSC behavior. Alternatively, a unique combination of tion on which the study was founded. We believe that more general signaling pathways active in the majority our report represents a demonstration of the power of of specialized cell types may define the molecular sig- molecular profiling when based on functional criteria nature of human stem cells (Pyle et al., 2004) (e.g., and biological outcomes. genes required to establish HSCs during development Identification of HES-1 as one of the upregulated are not required to sustain or molecularly define HSCs). transcripts in both hierarchical and ontogenic compari- However, independent categories that delineate signal- sons underscores the role of the Notch pathway in hu- ing pathways and transcription factors that are “essen- man HSC regulation (Karanu et al., 2000b; 2001, 2003; tial” versus those that “augment” existing HSC function Ohishi et al., 2002; Varnum-Finney et al., 2000). Al- are emerging. These categories may be related, but though the importance of Notch signaling to HSCs was they may allow access to unique components of HSC previously appreciated, details of the cellular mecha- behavior. Those that fall in the category of essential are nism were unknown. Our current report provides inde- likely to disrupt HSC function upon inactivation and, as pendent evidence for HES-1 as a putative Notch target recent studies indicate, do not seemingly include Jag- Defining the Molecular Nature of Human HSCs 661

ged1-mediated Notch signaling (Mancini et al., 2004) Molecular Analysis with Microarray or β-catenin based signaling induced by Wnt pathway Total RNA was extracted and amplified as shown previously (Wang activation (Cobas et al., 2004). Despite these observa- et al., 2004) from each purified population. Amplified-labeled RNA was hybridized to HG-U133AB chips (Affymetrix, USA) by using tions, both Notch (Karanu et al., 2000a; Varnum-Finney standard protocols at the Ottawa Health Research Institute (OHRI; et al., 2000) and Wnt signaling (Murdoch et al., 2003; Ottawa, Canada) and Microarray Facility of the Robarts Research Reya et al., 2003) have been identified to expand or Institute (London, Ontario, Canada). The quality of the data was augment baseline HSC function. assessed by examining several parameters, including the 3#:5# ra- Based on the fundamental role of human HSC regu- tio of some housekeeping genes such as β-Actin and β-Glucuroni- lating genes identified here, the search for genes that dase, the percentage of present and absent calls, and background noise. The processes of data analysis, including normalization and delineate “stemness” (Ivanova et al., 2002; Ramalho- clustering, were performed by using GeneSpring 6.0. The data were Santos et al., 2002) may amount to diminutive changes normalized per array and per gene by using algorithms in Gene- in gene expression effecting universal signaling path- Spring 6.0. For per array normalization, data in each array were ways that synergize to influence functional properties normalized to the 50th percentile of the measurements taken from that define tissue-specific stem cells once they have that specific array. Per gene normalization was performed by divid- developed in utero. Whether global gene analysis will ing the mean of the signal intensity from each gene in the selected sample by the mean of the same gene in the control sample. Genes aid in providing new insights into the concept of stem- that were significantly (p < 0.05) differentially (more or less than two ness or in identifying all of the genes that impact stem times) regulated in each comparison and flag present in at least cell behavior remains to be determined. one of the experimental replicates were selected for further analysis by methods used previously (Wang et al., 2004).

Experimental Procedures Conventional and Quantitative Polymerase Chain Reactions Human Cell Purification Expression of some of the candidate genes identified in the array Samples of human fetal blood (FB) from early gestation (16–22 analysis was quantified by quantitative real-time polymerase chain weeks), CB from the umbilical vein of neonates, BM from the iliac reaction (Q-RT-PCR) (MX4000 Stratagene CA) by using SYBRGREEN crest of healthy donors, and peripheral blood mobilized by granulo- (Stratagene CA) DNA binding dye (Bustin, 2000) as previously cyte colony-stimulating factor (G-CSF) were obtained in conjunc- shown (Shojaei et al., 2004). A complete list of primers used can tion with local ethical and biohazard authorities of the University of be found in Table S2. The amplified products were size verified by Western Ontario and London Health Sciences Center. Samples 2% agarose gel fractionation and were also sequence verified. To α were diluted (1:4) with -minimal essential medium (GIBCO-BRL, confirm that HLF-E2A rearrangements were absent in primary hu- Grand Island, NY) or PBS, and mononuclear cells (MNCs) were iso- man cells used to create transgenic human cells, conventional PCR lated by using Ficoll-Paque (Pharmacia) centrifugation (Gallacher was performed by using specific primers for detection of HLF-E2A et al., 2000). Subsets of Lin−CD34+CD38− and Lin−CD34+CD38+ chimeric transcripts as shown previously (Devaraj et al., 1994) and cells from FB, CB, BM, and MPB were isolated from the lineage- shown in Table S2. depleted population (Lin−) as previously shown (Gallacher et al., 2000). To increase efficiency, CD38 and CD33 antibodies were added to a lineage depletion cocktail to remove CD38+ and CD33+ Retroviral Gene Transfer of HES-1 and HLF into Primitive cells when the Lin−CD34+CD38− population was targeted for isola- Human Hematopoietic Cells tion. Lin− cells were stained with anti-human CD34-allophyco- Primary human Lin− CB cells from four independent samples were cyanin (APC) and anti-human CD38-phycoerythrin (PE; Beckton cultured and exposed to retrovirus as shown previously (Murdoch Dickinson, San Jose, CA) antibodies and sorted by fluorescence- et al., 2001). PA317 cells were transiently transfected with recombi- activated cell sorting (FACS) by using a FACSVantage SE (Beckton nant HES-1 (MIEV-HES-1), HLF (MIEV-HLF), and vector alone Dickinson) with the sorting gates illustrated. (MIEV) retrovirus, and viral supernatant of PA317 cells was used for stable transduction into PG13 cells, leading to the generation of In Vitro and In Vivo Assays packaging cell lines producing recombinant, HES-1, HLF, and sham De novo-isolated and -sorted GFP+ transgenic cells were plated in vector viral particles. CB Lin− cells were exposed to appropriate MethoCult H4230 (Stem Cell Technologies, Vancouver, BC) to assay retroviral particles by using established culture protocols (Murdoch for CFU potential as described previously (Murdoch et al., 2001; et al., 2001) and were harvested after 6 days of culture for subse- 2002a, 2002b). For the same experiments, unsorted cells exposed quent analysis. to retrovirus were harvested from cultures and transplanted via tail vein injection or intrafemoral bone marrow transplantation into the Staining of Cultured Cells for Brd-U, Annexin-V, femur (Mazurier et al., 2003b) of sublethally irradiated (350 cGy and Cell Surface Markers 137Cs) nonobese, diabetic, severe combined immune-deficient The Brd-U Flow Kit (BD Pharmingen; USA) was used to analyze the (NOD/SCID) mice (Gallacher et al., 2000). Mice were analyzed for cycling status of CB Lin− cells transduced with HES1, HLF, and human chimerism and multilineage hematopoietic composition as vector alone. Cells were pulsed with Brd-U for 2 hr, followed by shown previously (Gallacher et al., 2000; Murdoch et al., 2001). For several steps of fixation and washing according to the protocol. in vivo Brd-U incorporation, transplanted mice were administered The final stage included staining with anti Brd-U-APC antibody and Brd-U (200 ␮l, 6.25 ␮M concentration) intraperitoneally for 3 con- 7-AAD to identify fractions of cells in G0/G1, S, and G2+M phases secutive days prior to analysis at 6 weeks posttransplantation of of the cell cycle. The apoptotic status of transduced cells was eval- human HSCs. HES1, HLF, and vector transgenic (GFP+) cells was uated by using the Annexin-V Apoptosis Detection Kit (BD Phar- analyzed by gating on human CD45-expressing cells and/or mingen; USA). Cells were washed with cold PBS twice and stained CD45+CD34+ cells in chimeric animals. with Annexin-V-PE and 7-AAD according to the protocol provided by the manufacturer (BD Pharmingen; USA). The GFP+CD34+ pre- Ki-67 Staining apoptotic cells in each treatment were detected by gating the An- Purified cells were incubated for 30 min at 4°C in 0.4% para-formal- nexin-V+7-AAD− fraction by FACSCalibur. Cell surface expression dehyde with subsequent addition of 0.2% Triton X-100, followed by of homing molecules was analyzed by using CD44, CD49d (VLA4), overnight incubation, washing, and staining with Ki67-FITC mono- and CXCR4 PE-conjugated antibodies (BD Pharmingen; USA). clonal antibody (Immunotech, Burlington, Ontario) for 1 hr at 4°C. Transgenic (GFP+) cells were gated based on coexpression of cell The cycling status was analyzed by using a FACSCalibur and Cell surface CD34-APC, followed by evaluating expression of CD44, Quest software (Becton-Dickinson, San Jose, California). CD49d, and CXCR4 as indicated. Developmental Cell 662

Statistics Enver, T., Heyworth, C.M., and Dexter, T.M. (1998). Do stem cells Statistical analysis was performed by using T test comparison with play dice? Blood 92, 348–351. GraphPad Prism (GraphPad Software, Inc., San Diego, CA) or Excel Gallacher, L., Murdoch, B., Wu, D., Karanu, F., Fellows, F., and Microsoft Software. Values were compared as means ± standard Bhatia, M. (2000). Identification of novel circulating human embry- error of the mean, and p < 0.05 was considered statistically signifi- onic blood stem cells. Blood 96, 1740–1747. cant and is indicated by asterisks in the figures. Geiger, H., and Van Zant, G. (2002). The aging of lympho-hematopoietic stem cells. Nat. Immunol. 3, 329–333. Supplemental Data Georgantas, R.W., 3rd, Tanadve, V., Malehorn, M., Heimfeld, S., Supplemental Data detailing the full array of genes identified are Chen, C., Carr, L., Martinez-Murillo, F., Riggins, G., Kowalski, J., available at http://www.developmentalcell.com/cgi/content/full/8/ and Civin, C.I. (2004). Microarray and serial analysis of gene ex- 5/651/DC1. pression analyses identify known and novel transcripts overex- pressed in hematopoietic stem cells. Cancer Res. 64, 4434–4441. Glimm, H., Oh, I.H., and Eaves, C.J. (2000). Human hematopoietic Acknowledgments stem cells stimulated to proliferate in vitro lose engraftment potential during their S/G(2)/M transit and do not reenter G(0). Blood 96, We gratefully acknowledge the postdoctoral fellows in the Krembil 4185–4193. center, Drs. L. Wang, K. Vijayaragavan, and P. Menendez, for critical Hitzler, J.K., Soares, H.D., Drolet, D.W., Inaba, T., O’Connel, S., Ro- review of the manuscript; D. Sheerar for cell isolation; the nurses senfeld, M.G., Morgan, J.I., and Look, A.T. (1999). Expression pat- of the labor and delivery division of St. Joseph’s Health Care and terns of the hepatic leukemia factor gene in the nervous system of London Health Sciences Center for collection of samples; and Dr. developing and adult mice. Brain Res. 820, 1–11. A. Xenocostas for providing adult mobilized peripheral blood and Honda, H., Inaba, T., Suzuki, T., Oda, H., Ebihara, Y., Tsuiji, K., Naka- bone marrow samples. Funding for this research was provided by a hata, T., Ishikawa, T., Yazaki, Y., and Hirai, H. (1999). Expression of research grant from the National Cancer Institute of Canada (NCIC), E2A-HLF chimeric protein induced T-cell apoptosis, B-cell matura- Canadian Institutes of Health Research (CIHR), the National tion arrest, and development of acute lymphoblastic leukemia. Centres of Excellence (NCE) Program Stem Cell Network-Stem Cell Blood 93, 2780–2790. Genomics/Genome Canada, the Krembil Foundation, Canada Re- search Chair in Stem Cell Biology and Regenerative Medicine to Inaba, T., Roberts, W.M., Shapiro, L.H., Jolly, K.W., Raimondi, S.C., M.B., and a postgraduate scholarship award from Ontario Graduate Smith, S.D., and Look, A.T. (1992). Fusion of the leucine zipper gene Society (OGS) to J.T. HLF to the E2A gene in human acute B-lineage leukemia. Science 257, 531–534. Iscove, N.N., Barbara, M., Gu, M., Gibson, M., Modi, C., and Wine- Received: September 13, 2004 garden, N. (2002). Representation is faithfully preserved in global Revised: January 27, 2005 cDNA amplified exponentially from sub-picogram quantities of Accepted: March 1, 2005 mRNA. Nat. Biotechnol. 20, 940–943. Published: May 2, 2005 Ishibashi, M., Moriyoshi, K., Sasai, Y., Shiota, K., Nakanishi, S., and References Kageyama, R. (1994). Persistent expression of helix-loop-helix factor HES-1 prevents mammalian neural differentiation in the Bustin, S.A. (2000). Absolute quantification of mRNA using real- central nervous system. EMBO J. 13, 1799–1805. time reverse transcription polymerase chain reaction assays. J. Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A., Mol. Endocrinol. 25, 169–193. and Lemischka, I.R. (2002). A stem cell molecular signature. Sci- Cobas, M., Wilson, A., Ernst, B., Mancini, S.J., MacDonald, H.R., ence 298, 601–604. Kemler, R., and Radtke, F. (2004). Beta-catenin is dispensable for Jenkins, B.J., Quilici, C., Roberts, A.W., Grail, D., Dunn, A.R., and hematopoiesis and lymphopoiesis. J. Exp. Med. 199, 221–229. Ernst, M. (2002). Hematopoietic abnormalities in mice deficient in Published online January 2004. gp130-mediated STAT signaling. Exp. Hematol. 30, 1248–1256. Cooper, C.L., Brady, G., Bilia, F., Iscove, N.N., and Quesenberry, P.J. Kageyama, R., Ohtsuka, T., and Tomita, K. (2000). The bHLH gene (1997). Expression of the Id family helix-loop-helix regulators during Hes1 regulates differentiation of multiple cell types. Mol. Cells 10, growth and development in the hematopoietic system. Blood 89, 1–7. 3155–3165. Karanu, F.N., Murdoch, B., Gallacher, L., Wu, D.M., Koremoto, M., Cory, S., Huang, D.C., and Adams, J.M. (2003). The Bcl-2 family: Sakano, S., and Bhatia, M. (2000a). The notch ligand jagged-1 rep- roles in cell survival and oncogenesis. Oncogene 22, 8590–8607. resents a novel growth factor of human hematopoietic stem cells. Crable, S.C., and Anderson, K.P. (2003). A PAR domain transcrip- J. Exp. Med. 192, 1365–1372. tion factor is involved in the expression from a hematopoietic-spe- Karanu, F.N., Murdoch, B., Gallacher, L., Wu, D.M., Koremoto, M., cific promoter for the human LMO2 gene. Blood 101, 4757–4764. Sakano, S., and Bhatia, M. (2000b). The notch ligand jagged-1 rep- Danial, N.N., and Korsmeyer, S.J. (2004). Cell death: critical control resents a novel growth factor of human hematopoietic stem cells. points. Cell 116, 205–219. J. Exp. Med. 192, 1365–1372. Devaraj, P.E., Foroni, L., Sekhar, M., Butler, T., Wright, F., Mehta, A., Karanu, F.N., Murdoch, B., Miyabayashi, T., Ohno, M., Koremoto, Samson, D., Prentice, H.G., Hoffbrand, A.V., and Secker-Walker, M., Gallacher, L., Wu, D., Itoh, A., Sakano, S., and Bhatia, M. (2001). L.M. (1994). E2A/HLF fusion cDNAs and the use of RT-PCR for the Human homologues of Delta-1 and Delta-4 function as mitogenic detection of minimal residual disease in t(17;19)(q22;p13) acute regulators of primitive human hematopoietic cells. Blood 97, lymphoblastic leukemia. Leukemia 8, 1131–1138. 1960–1967. Dick, J.E. (1996). Normal and leukemic human stem cells assayed Karanu, F.N., Yuefei, L., Gallacher, L., Sakano, S., and Bhatia, M. in SCID mice. Semin. Immunol. 8, 197–206. (2003). Differential response of primitive human CD34− and CD34+ Dieterlen-Lievre, F., Pardanaud, L., Godin, I., Garcia-Porrero, J., Cu- hematopoietic cells to the Notch ligand Jagged-1. Leukemia 17, mano, A., and Marcos, M. (1993). Developmental relationships be- 1366–1374. tween hemopoiesis and vasculogenesis. C. R. Acad. Sci. III 316, Kumano, K., Chiba, S., Kunisato, A., Sata, M., Saito, T., Nakagami- 892–901. Yamaguchi, E., Yamaguchi, T., Masuda, S., Shimizu, K., Takahashi, Domen, J., Cheshier, S.H., and Weissman, I.L. (2000). The role of T., et al. (2003). Notch1 but not Notch2 is essential for generating apoptosis in the regulation of hematopoietic stem cells: overex- hematopoietic stem cells from endothelial cells. Immunity 18, pression of Bcl-2 increases both their number and repopulation 699–711. potential. J. Exp. Med. 191, 253–264. Kunisato, A., Chiba, S., Nakagami-Yamaguchi, E., Kumano, K., Defining the Molecular Nature of Human HSCs 663

Saito, T., Masuda, S., Yamaguchi, T., Osawa, M., Kageyama, R., cycle disruption in transformation by Notch(ic). Mol. Cell. Biol. 21, Nakauchi, H., et al. (2003). HES-1 preserves purified hematopoietic 5925–5934. stem cells ex vivo and accumulates side population cells in vivo. Shojaei, F., Gallacher, L., and Bhatia, M. (2004). Differential gene Blood 101, 1777–1783. expression of human stem progenitor cells derived from early Kurosawa, H., Goi, K., Inukai, T., Inaba, T., Chang, K.S., Shinjyo, T., stages of in utero human hematopoiesis. Blood 103, 2530–2540. Rakestraw, K.M., Naeve, C.W., and Look, A.T. (1999). Two candidate Song, Y., and Jacob, C.O. (2005). The mouse cell surface protein downstream target genes for E2A-HLF. Blood 93, 321–332. TOSO regulates Fas/Fas-ligand induced apoptosis through its Lemischka, I. (1999). Searching for stem cell regulatory molecules. binding to FADD. J. Biol. Chem. 4, 1152–1167. Some general thoughts and possible approaches. Ann. N Y Acad. Srour, E.F., Jetmore, A., Wolber, F.M., Plett, P.A., Abonour, R., Yoder, Sci. 872, 274–287. M.C., and Orschell-Traycoff, C.M. (2001). Homing, cell cycle kinet- Mancini, S.J., Mantei, N., Dumortier, A., Suter, U., Macdonald, H.R., ics and fate of transplanted hematopoietic stem cells. Leukemia and Radtke, F. (2004). Jagged1 dependent Notch signaling is dis- 15, 1681–1684. pensable for hematopoietic stem cell self-renewal and differentia- Steidl, U., Kronenwett, R., Rohr, U.P., Fenk, R., Kliszewski, S., tion. Blood 18, 56–66. Maercker, C., Neubert, P., Aivado, M., Koch, J., Modlich, O., et al. Mazurier, F., Doedens, M., Gan, O.I., and Dick, J.E. (2003a). Rapid (2002). Gene expression profiling identifies significant differences myeloerythroid repopulation after intrafemoral transplantation of between the molecular phenotypes of bone marrow-derived and NOD-SCID mice reveals a new class of human stem cells. Nat. circulating human CD34+ hematopoietic stem cells. Blood 99, Med. 9, 959–963. 2037–2044. Mazurier, F., Doedens, M., Gan, O.I., and Dick, J.E. (2003b). Rapid Sutherland, H.J., Eaves, C.J., Eaves, A.C., Dragowska, W., and myeloerythroid repopulation after intrafemoral transplantation of Lansdorp, P.M. (1989). Characterization and partial purification of NOD-SCID mice reveals a new class of human stem cells. Nat. human marrow cells capable of initiating long-term hematopoiesis Med. 9, 959–963. in vitro. Blood 74, 1563–1570. Morrison, S.J., Uchida, N., and Weissman, I.L. (1995). The biology Sutherland, H.J., Lansdorp, P.M., Henkelman, D.H., Eaves, A.C., of hematopoietic stem cells. Annu. Rev. Cell Dev. Biol. 11, 35–71. and Eaves, C.J. (1990). Functional characterization of individual hu- Murdoch, B., Gallacher, L., Awaraji, C., Hess, D.A., Keeney, M., Jay, man hematopoietic stem cells cultured at limiting dilution on sup- K., Chadwick, K., Fowley, S.R., Howson-Jan, K., Chin Yee, I., et al. portive marrow stromal layers. Proc. Natl. Acad. Sci. USA 87, (2001). Circulating hematopoietic stem cells serve as novel targets 3584–3588. for in utero gene therapy. FASEB J. 15, 1628–1630. Takada, H., Chen, N.J., Mirtsos, C., Suzuki, S., Suzuki, N., Wake- ham, A., Mak, T.W., and Yeh, W.C. (2003). Role of SODD in regula- Murdoch, B., Gallacher, L., Chadwick, K., and Bhatia, M. (2002a). tion of tumor necrosis factor responses. Mol. Cell. Biol. 23, 4026– Characterization of retroviral gene transfer into highly purified hu- 4033. man CD34(−) cells with primitive hematopoietic capacity. Mol. Ther. 5, 635–643. Tamayo, P., Slonim, D., Mesirov, J., Zhu, Q., Kitareewan, S., Dmi- trovsky, E., Lander, E.S., and Golub, T.R. (1999). Interpreting pat- Murdoch, B., Gallacher, L., Chadwick, K., Fellows, F., and Bhatia, terns of gene expression with self-organizing maps: methods and M. (2002b). Human embryonic-derived hematopoietic repopulating application to hematopoietic differentiation. Proc. Natl. Acad. Sci. cells require distinct factors to sustain in vivo repopulating func- USA 96, 2907–2912. tion. Exp. Hematol. 30, 598–605. Varnum-Finney, B., Xu, L., Brashem-Stein, C., Nourigat, C., Flowers, Murdoch, B., Chadwick, K., Martin, M., Shojaei, F., Shah, K.V., Gal- D., Bakkour, S., Pear, W.S., and Bernstein, I.D. (2000). Pluripotent, lacher, L., Moon, R.T., and Bhatia, M. (2003). Wnt-5A augments re- cytokine-dependent, hematopoietic stem cells are immortalized by populating capacity and primitive hematopoietic development of constitutive Notch1 signaling. Nat. Med. 6, 1278–1281. human blood stem cells in vivo. Proc. Natl. Acad. Sci. USA 100, 3422–3427. Verfaillie, C.M. (2002). Hematopoietic stem cells for transplantation. Nat. Immunol. 3, 314–317. Nakamura, Y., Sakakibara, S., Miyata, T., Ogawa, M., Shimazaki, T., Weiss, S., Kageyama, R., and Okano, H. (2000). The bHLH gene Wang, L., Li, L., Shojaei, F., Levac, K., Cerdan, C., Menendez, P., hes1 as a repressor of the neuronal commitment of CNS stem cells. Martin, T., Rouleau, A., and Bhatia, M. (2004). Endothelial and J. Neurosci. 20, 283–293. hematopoietic cells fate of human embryonic stem cells originates from primitive endothelium with hemogenic properties. Immunity Ohishi, K., Varnum-Finney, B., and Bernstein, I.D. (2002). The notch 21, 31–41. pathway: modulation of cell fate decisions in hematopoiesis. Int. J. Hematol. 75, 449–459. Warren, A.J., Colledge, W.H., Carlton, M.B., Evans, M.J., Smith, A.J., and Rabbitts, T.H. (1994). The oncogenic cysteine-rich LIM Orkin, S.H., and Zon, L.I. (2002). Hematopoiesis and stem cells: domain protein rbtn2 is essential for erythroid development. Cell plasticity versus developmental heterogeneity. Nat. Immunol. 3, 78, 45–57. 323–328. Orschell-Traycoff, C.M., Hiatt, K., Dagher, R.N., Rice, S., Yoder, M.C., and Srour, E.F. (2000). Homing and engraftment potential of Sca-1(+)lin(−) cells fractionated on the basis of adhesion molecule expression and position in cell cycle. Blood 96, 1380–1387. Pardal, R., Clarke, M.F., and Morrison, S.J. (2003). Applying the principles of stem-cell biology to cancer. Nat. Rev. Cancer 3, 895– 902. Pyle, A.D., Donovan, P.J., and Lock, L.F. (2004). Chipping away at “stemness.” Genome Biol. 5, 235. Published online July 2004. Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C., and Melton, D.A. (2002). “Stemness”: transcriptional profiling of embryonic and adult stem cells. Science 298, 597–600. Reya, T., Duncan, A.W., Ailles, L., Domen, J., Scherer, D.C., Willert, K., Hintz, L., Nusse, R., and Weissman, I.L. (2003). A role for Wnt signalling in self-renewal of haematopoietic stem cells. Nature 423, 409–414. Ronchini, C., and Capobianco, A.J. (2001). Induction of cyclin D1 transcription and CDK2 activity by Notch(ic): implication for cell