Leukemia (2003) 17, 1157–1163 & 2003 Nature Publishing Group All rights reserved 0887-6924/03 $25.00 www.nature.com/leu Homeobox gene expression profile in human hematopoietic multipotent stem cells and T-cell progenitors: implications for human T-cell development T Taghon1, K Thys1, M De Smedt1, F Weerkamp2, FJT Staal2, J Plum1 and G Leclercq1

1Department of Clinical Chemistry, Microbiology and Immunology, Ghent University Hospital, Ghent, Belgium; and 2Department of Immunology, Erasmus Medical Center, Rotterdam, The Netherlands

Class I homeobox (HOX) genes comprise a large family of implicated in this transformation proces.14 The HOX-C locus transcription factors that have been implicated in normal and has been primarily implicated in lymphomas.15 malignant hematopoiesis. However, data on their expression or function during T-cell development is limited. Using degener- Hematopoietic cells are derived from stem cells that reside in ated RT-PCR and Affymetrix microarray analysis, we analyzed fetal liver (FL) in the embryo and in the adult bone marrow the expression pattern of this gene family in human multipotent (ABM), which have the unique ability to self-renew and thereby stem cells from fetal liver (FL) and adult bone marrow (ABM), provide a life-long supply of blood cells. T lymphocytes are a and in T-cell progenitors from child thymus. We show that FL specific type of hematopoietic cells that play a major role in the and ABM stem cells are similar in terms of HOX gene immune system. They develop through a well-defined order of expression, but significant differences were observed between differentiation steps in the thymus.16 Several transcription these two cell types and child thymocytes. As the most 17–21 immature thymocytes are derived from immigrated FL and factors have a crucial role in this developmental process, 22 ABM stem cells, this indicates a drastic change in HOX gene resulting in a complex regulatory network. Little is known, expression upon entry into the thymus. Further analysis of however, if HOX genes also play a role during T-cell HOX-A7, HOX-A9, HOX-A10, and HOX-A11 expression with development. A crucial role for HOX-A9 in this developmental specific RT-PCR in all thymocyte differentiation stages showed process was elucidated by the generation of HOX-A9À/À mice.23 a sequential loss of 30 region HOX-A cluster genes during Thymocytes from these animals are blocked at an early stage of intrathymic T-cell development and an unexpected expression 24 of HOX-A11, previously not recognized to play a role in differentiation. Other evidence for the role of HOX genes in hematopoiesis. Also HOX-B3 and HOX-C4 were expressed thymopoiesis came from a study in which enforced expression throughout thymocyte development. Overall, these data pro- of HOX-B3 resulted in skewing of T-cell progenitors towards the vide novel evidence for an important role of certain HOX genes gd T-cell lineage instead of the ab-lineage.25 in human T-cell development. Human CD34+ lineage-marker negative (LinÀ) FL stem cells Leukemia (2003) 17, 1157–1163. doi:10.1038/sj.leu.2402947 26 Keywords: HOX; T-cell development; human; transcription factors have been shown previously to generate T cells, while in contrast, ABM stem cells seem to have a strongly diminished capacity to generate these cells.27,28 In this study, we therefore compared the expression pattern of HOX genes between highly purified CD34+LinÀ cells from both sources to investigate if Introduction differences in HOX gene expression are present that can account for the difference in T-cell differentiation potential. In Class I homeobox (HOX) genes code for transcription factors addition, to have a better insight in which genes might be that are characterized by a highly conserved region of 60 amino involved in T-cell development, we also analyzed HOX gene acids that is responsible for DNA binding.1 Generally, this gene expression in the earliest multipotent progenitor cells in the human thymus and in the earliest T-cell committed progenitor family is divided into two different classes. Class I HOX genes 29 were discovered in Drosophila because of the appearance of the cells. We show that in terms of HOX gene expression, virtually so-called ‘homeotic mutations’, mutations that cause transfor- no difference could be detected between FL and ABM stem mation of one part of the body into another.2 In the genome of cells, but that important differences in the expression pattern of vertebrates, they are arranged in four different gene clusters and this gene family were apparent between thymic and nonthymic share a high degree of homology. In man, 39 HOX members precursor cells. Furthermore, specific RT-PCR reactions for have been identified so far.3 Class II homeobox genes are the selected genes showed an intriguing expression profile for HOX- divergent homeobox genes. These genes have less homology. A cluster genes during human thymocyte development and Originally identified as master regulatory genes during expression of HOX-B3 and HOX-C4 throughout this develop- embryogenesis, research over the last decade has provided mental pathway. ample evidence to postulate that the HOX gene family is also able to control proliferation and differentiation of hematopoietic cells.4–6 Various reports have indicated the involvement of HOX Materials and methods genes in leukemogenesis, underlining their role in normal hematopoiesis. Several HOX genes are involved in the Monoclonal antibodies transformation of hematopoietic cells as a result of translocation events, especially with NUP98, a nucleoporin gene.7–10 Mouse anti-human MoAbs used were CD1a (OKT6 fluorescein Furthermore, high-level expression of both HOX-A9 and HOX- isothiocyanate (FITC)), CD3 (OKT3 FITC), CD4 (OKT4 FITC), A10 is frequently detected in leukemic blasts from patients with CD7 (3A1 FITC), CD8a (OKT8 FITC) and CD19 (Leu12) from acute myeloid leukemia,11–13 and also HOX-B6 was recently American Type Culture Collection (ATCC, Rockville, MD, USA); CD3 (Leu-4 phycoerythrin (PE) or allophycocyanine (APC)), Correspondence: Dr G Leclercq, Ghent University Hospital, 4BlokA, CD4 (Leu-3a PE or APC), CD34 (HPCA-2 FITC, PE or APC) and De Pintelaan 185, Ghent B-9000, Belgium. Fax: +32 9 2403 659 CD69 (L78 PE) from Becton Dickinson Immunocytometry Received 11 September 2002; accepted 14 February 2003 Systems (BDIS, Mountain View, CA, USA); CD1a (T6 PE) and HOX gene expression in T-cell progenitors T Taghon et al 1158 TCR panab (BMA031 PE) from Coulter (Miami, FL, USA); immunomagnetic depletion was performed with unlabeled glycophorin-A (10F7MN) and CD8b (2ST8.5H7 unlabeled or CD8b antibody, and subsequently, cells were stained with FITC), a kind gift of Dr L Lanier (DNAX, Palo Alto, CA, USA) and CD8a-FITC, rat anti-mouse-IgG2a+b-PE (which detects the Dr E Reinherz (Dana-Farber Cancer Institute, Boston, MA, USA), unlabeled CD8b antibody) and CD3-APC to sort + À À + respectively. Rat anti-mouse-IgG2a+b (X57 PE) from BDIS was CD8a CD8b CD3 cells. All these cells are CD4 . For used to detect the unlabeled CD8b antibody. purification of CD4+CD8ab+CD3À, CD4+CD8ab+CD3+CD69À and CD4+CD8ab+CD3+CD69+ cells, thymocytes were labeled with CD8b-FITC, CD69-PE and CD4-APC. The Cell sources CD8b+CD69+CD4+ fraction contains only CD3+ cells and therefore represents the CD4+CD8ab+CD3+CD69+ cells, while FL tissues were obtained after legal termination of pregnancy, the CD8b+CD69ÀCD4+ fraction contains both the ABM was collected from healthy donors, and child thymus (CT) CD4+CD8ab+CD3À and CD4+CD8ab+CD3+CD69À popula- was obtained from children undergoing cardiac surgery. All tions. Therefore, sorted CD8b+CD69ÀCD4+ cells were labeled human tissues were obtained with informed consent and used with CD3-PE to allow subsequent sorting of both fractions. For according to the guidelines of the Medical Ethical Commission sorting of TCRab+CD4+CD8À and TCRab+CD8+CD4À cells, of the University Hospital Ghent. FL cells were isolated by thymocytes were labeled with CD8a-FITC, TCRab-PE and CD4- gentle disruption of the tissue in complete medium (Iscove’s APC. Cells were sorted on a FACS Vantage (BDIS) using modified Dulbecco’s medium/10% fetal calf serum (FCS), all CellQuest software (BDIS) and checked for purity, which was from Invitrogen, Carlsbad, CA, USA). Child thymocytes were always at least 98.5%. obtained by cutting the thymic tissue into small pieces of 0.5 cm  0.5 cm, which were then extensively teased apart with cataract knives in serum-free RPMI 1640 medium (Invitrogen). RNA isolation, RT-PCR, cloning of PCR products and Mononuclear cells from all tissues were obtained by density- sequence analysis gradient centrifugation over Lymphoprep (Nyegaard, Oslo, Norway) and were subsequently resuspended in 9 vol FCS RNA from purified progenitor cells and different thymocyte (Invitrogen) and 1 vol dimethyl sulfoxide (Serva, Heidelberg, subsets was isolated using Trizol (Invitrogen) according to the Germany), and cryopreserved in liquid N2 until use. instructions of the supplier, treated with DNAseI (Promega Coorporation, Madison, WI, USA) to avoid genomic DNA contamination, and cDNA was prepared using Superscript Purification of hematopoietic precursor cells and (Invitrogen) following the guidelines of the manufacturer. HOX thymocyte subsets homeodomain degenerated primers used for amplification were modified at the 50-end by providing restriction sites that were not After thawing and washing the cells, cells were labeled with present in any of the 39 homeodomains, to facilitate subsequent lineage-specific antibodies and precursor cells from all human cloning into the pUC18 plasmid (Invitrogen). A BamHI site was tissues were enriched by immunomagnetic depletion of Lin+ included in the forward primer: 50-GCGGATCCGA(AG)(TC)TI- cells (using sheep anti-mouse Ig-coated Dynabeads (Dynal AS, GA(AG)AA(AG)GA(AG)TT(CT)C(AT)IT(AT)(CT)A-30 and a Hin- Oslo, Norway) with a ratio cells/Dynabeads of 1/5), followed by dIII site in the reverse primer: 50-GCGAAGCTT(AG)(CT)IC(GT)- cell sorting. In case of FL and ABM, Lin+ cells were IC(GT) (AG)TT(CT)TG(AG)AACCAIA(CT)(CT)TT-30. Individual immunodepleted with antibodies against glycophorin A, CD19 clones were analyzed by running cycle sequencing products and CD7-FITC and remaining cells were subsequently labeled (ABI PRISM BigDye Terminator cycle sequencing kit, Applied with CD34-PE and CD1-, CD3-, CD4- and CD8a-FITC to allow Biosystems, Foster City, CA, USA) on an ABI PRISM 310 Genetic sorting of CD34+CD1ÀCD3ÀCD4ÀCD7ÀCD8aÀCD19À (called Analyzer (Applied Biosystems) to identify the expressed homeo- CD34+LinÀ) precursor cells. Human thymocytes were immuno- box genes. Specific primers for selected HOX genes are given in depleted with antibodies against CD3-, CD4- and CD8a-FITC Table 1. All reactions were performed using Taq Polymerase and subsequently labeled with CD34-APC and CD1-PE to allow (Perkin-Elmer, Emeryville, CA, USA) with an initial denaturation sorting of CD34+CD1ÀCD3ÀCD4ÀCD8aÀ (called CD34+CD1À) step of 2 min at 941C, except for HOX-A9 PCR for which we and CD34+CD1+CD3ÀCD4ÀCD8aÀ (called CD34+CD1+) T-cell used Taq Gold Polymerase (Perkin-Elmer) with an initial precursors. For purification of CD4+CD8aÀCD3À thymocytes, denaturation step of 12 min at 941C. Conditions were as follows: immunomagnetic depletion was performed on total thymocytes For hypoxanthine guanine phosphoribosyl transferase (HPRT) with FITC-labeled CD3 and CD8a antibodies, followed by and HOX-A7: as described for HPRT30; for HOX-A9: 40 cycles staining with CD4-PE and CD34-FITC to sort CD4+FITCÀ of 30 s at 941C, 40 s at 601C and 40 s at 721C; for HOX-A10: thymocytes. For purification of CD4+CD8a+CD8bÀ thymocytes, 35 cycles of 30 s at 941C, 30 s at 681C and 1 min at 721C; for

Table 1 Specific primers used for RT-PCR (50-30)

Gene of interest Forward primer Reverse primer

HPRT TATGGACAGGACTGAACGTCTTGC GACACAAACATGATTCAAATCCCTGA HOX-A7 AGCCCCCTTTATCAGAGC TTCATCATCGTCCTCCTCG HOX-A9 TCGATCCCAATAACCCAGC CACTCGTCTTTTGCTCGG HOX-A10 AGAGCAGCAAAGCCTCGCCG AAGTTGGCTGTGAGCTCCCGGATCC HOX-A11 CGTGCGCGAAGTGACCTTCAGAGAGTAC CCTGCCCACGGTGCTATAGAAATTGGAC HOX-B3 GGTTGCCAATGTCCGGTCCTGAG GTGCGTGGCGGCCTGAAATG HOX-C4 TATAGCTGCACCAGTCTCC GACCTGACTTTGGTGTTGG

Leukemia HOX gene expression in T-cell progenitors T Taghon et al 1159 HOX-A11 and HOX-C4: 35 cycles of 30 s at 941C, 40 s at 591C and analyzed these populations for HOX gene expression. We, and 1 min at 721C; for HOX-B3: 35 cycles of 30 s at 941C, 30 s at therefore, used RT-PCR with degenerated primers that recognize 661C and 30 s at 721C. the most highly conserved region of the homeodomain, followed by cloning and sequence analysis. This approach has been previously shown to yield representative data.31 Microarray analysis In Table 2, an overview is given of the total number of clones that was sequenced for each cell population and of the number RNA was isolated using RNeasy columns as described by the of different HOX genes that were identified. The most manufacturer (Qiagen, Hilden, Germany). The integrity of the remarkable overall feature was the high frequency of HOX-A RNA was tested on 1% formaldehyde containing agarose gels. cluster genes compared to HOX-B, HOX-C and HOX-D cluster RNA (5 mg) was used to generate ds cDNA using superscript genes. This seems to indicate that this is the most active HOX reverse transcriptase and a T7-oligo dT primer. The resulting cluster during early human hematopoiesis. Especially HOX-A7, cDNA was used in an in vitro cRNA reaction using T7 RNA HOX-A9 and HOX-A10 were detected at high levels. The polymerase and biotinylated ribonucleotides employing an relative expression levels of all detected HOX genes within the ENZO kit (ENZO, Farmingdale, NY, USA). The biotinylated various cell populations are shown in Figure 1 (Figure 1a for cRNA was cleaned up using RNeasy spin columns (Qiagen) and HOX-A and HOX-D cluster genes, Figure 1b for HOX-B and quantified by spectrophotometric methods. An adjusted cRNA HOX-C cluster genes). yield was calculated to reflect carryover of unlabeled total RNA. As the used method of PCR amplification of RNA with Fragmentation of 20 mg cRNA was performed at 951C for 35 min. degenerate homeodomain primers might not be quantitative, we cRNA (10 mg) was subsequently hybridized for 16 h to a HG- also performed microarray analysis on sorted CD1À and CD1+ U133A or a HG-U133B microarray (Affymetrix, Santa Clara, subpopulations of CD34 progenitor thymocytes. Figure 2 shows CA, USA) at 451C. After washing and staining, the arrays were the results of two independent experiments. Although all the scanned in an HP/Affymetrix scanner at 570 nm. For all HOX genes are represented on the microarray chips, only 0 0 experiments, the 5 /3 ratios of GAPDH were between 0.9 and transcripts for seven HOX genes were detected. Comparable to 1.5. The scanned images were analyzed using Affymetrix Micro- array Suite 5.0 software, and the signal intensity of detected transcripts is shown.

Results and discussion

To identify HOX genes that have a potential role in T-cell development, we purified hematopoietic stem cells from different sources as well as T-cell progenitors from the thymus,

Table 2 Expression of HOX genes in purified human precursor cell populations from various sources FL BM CT CD34+LinÀ CD34+LinÀ CD34+CD1À CD34+CD1+

HOX-A1 1a 51 2 HOX-A3 8 4 0 1 HOX-A4 10 11 0 1 HOX-A5 10 10 14 16 HOX-A6 9 5 2 9 HOX-A7 28 42 16 23 HOX-A9 61 41 3 16 HOX-A10 6 5 83 49 HOX-A11 0 0 2 3 HOX-B1 1 5 0 0 HOX-B3 0 3 3 2 HOX-B4 8 9 0 1 HOX-B5 3 1 0 0 HOX-B6 3 4 0 0 HOX-B7 1 2 1 1 HOX-B9 1 0 0 0 HOX-C4 0 1 2 0 HOX-C6 1 0 0 0 HOX-D10 0 0 1 1 HOX-D11 0 0 3 0 Figure 1 HOX gene expression profile in human hematopoietic T- cell progenitors. Frequency of HOX-A and HOX-D (a) and HOX-B and Total 151 148 131 125 HOX-C (b) cluster gene clones identified in the various hematopoietic progenitor cell populations, as indicated. Scales on the Y-axis of aNumbers represent the total number of clones identified. panels a and b are different.

Leukemia HOX gene expression in T-cell progenitors T Taghon et al 1160 become expressed,31 in accordance with the colinear expres- sion pattern that is observed during embryonic development.3 In the case of FL CD34+LinÀ cells, 36% of the clones analyzed contained 30 region HOX genes, while in the case of the ABM cells with the same phenotype, this was 39%. Therefore, this indicates the presence of primitive progenitor cells in both populations. Surprisingly, the expression frequency of HOX-A1 and HOX-B1 was higher in ABM-derived progenitors (Figure 1). This was unexpected as generally FL cells are regarded as more primitive cells than ABM cells. One small difference between both cell fractions was the expression of HOX-B3, which was detected in ABM cells but not in FL progenitors (Figure 1b). In terms of T-cell differentiation, it has been shown that enforced expression of this gene in mice bone marrow cells results in a reduced development of TCRab T cells.25 As ABM CD34+LinÀ cells have a reduced capacity to generate these cells,27,28 one might correlate the expression of this gene in ABM progenitors with this feature. However, the detection rate of this gene was Figure 2 Microarray analysis of HOX gene expression profile in human thymic progenitor cells. HOX gene expression in CD1À and too low to give convincing evidence and suggests that only a CD1+ subpopulations of CD34 thymic progenitor cells was analyzed minor fraction of the ABM cells expresses HOX-B3. Further- on microarray chips. The signal intensity of detected HOX transcripts more, a specific RT-PCR showed expression of this gene in two independent experiments is shown. throughout all stages of TCR-ab T-cell development (Figure 3b), indicating a role for this gene during development of this T-cell the method of PCR amplification, the HOX-A cluster seemed to lineage as discussed further on. be over-represented, with HOX-A5, HOX-A7, HOX-A9, and HOX-A9 is expressed at a relative higher level in FL compared HOX-A10 transcripts detected in both populations and in both to ABM cells (Figure 1a) and has previously been shown to be 24 experiments except for the HOX-A7 transcript in the CD1+ required for T-cell development, but other data have CD34 progenitor cells which was only detected once. In implicated this gene in differentiation of other hematopoietic 23 addition, HOX-C4 was detected once in the CD1À CD34 lineages as well and possibly also in stem cell self-renew- 33,38 progenitor cells, whereas HOX-C8 and HOX-D8 were detected al. Therefore, it is impossible to associate the higher once in the CD1+ CD34 progenitor cells. Whereas these data expression level of HOX-A9 in FL progenitors compared to confirm that mainly HOX-A cluster genes are expressed in ABM precursor cells with the increased ability of the former to hematopoietic stem/progenitor cells, it is also clear that generate T cells. microarray analysis is less sensitive (although a sensitivity of 1:300 000 can be obtained) and does not detect low expressed HOX genes (which are apparently expressed below the thresh- a old mentioned). Nevertheless, the microarray data provide CD34+ CD34+ CD4+ CD4+ CD4+ CD4+ CD4+ CD1- CD1+ CD3- CD3- CD3- CD3+ CD3+ CD4+ quantitative confirmation of the high expression of HOX-A CD8- CD8α+ CD8αβ + CD8αβ + CD8αβ+ CD8- genes, as initially found by RT-PCR. Other reports confirm that CD8β- CD69+ 8 CD3+ NK TCR αβ+ mainly HOX-A cluster genes are expressed in hematopoietic 1 2 3 4 5 6 7 DC CD8+ stem cells; in one report the PCR amplification and sequencing T cell lineage 9 CD4- 31 committment approach was used, in two more recent reports microarray CD3+ analysis was used.32,33 Proliferation/survival TCRαβ+ dependent on SCF/IL-7 signals provided by epithelial cells Fetal liver and adult bone marrow progenitors show b little difference in HOX gene expression profile 1 2 3 4 5 6 7 8 9 HPRT CD34+LinÀ stem cells from FL and ABM have previously been shown to have different characteristics in their ability to HOX-A7 generate human T cells.27,28 Therefore, we investigated if HOX-A9 differences in HOX gene expression could account for this HOX-A10 difference. However, virtually no differences were observed (Figure 1). In both fractions, high levels of HOX-A7 and HOX- HOX-A11 A9 were detected, which is in accordance with the data from HOX-B3 Sauvageau et al.31 However, in contrast to the same report, but HOX-C4 in agreement with recent data,33 we were also able to detect HOX-B4 expression with the degenerated primers, which Figure 3 Specific RT-PCR for selected HOX genes during human correlates with the primitive differentiation stage of these thymocyte development. (a) Schematic overview of human intrathy- cells.34,35 Indeed, FL and ABM CD34+LinÀ cells have been mic T-cell development. (b) Specific RT-PCR for HPRT, HOX-A7, shown to be multipotential, reminiscent of a stem cell HOX-A9, HOX-A10, HOX-A11, HOX-B3 and HOX-C4 in purified 36,37 human thymocyte subsets, ordered according to their maturation state. phenotype. It has been postulated that in hematopoiesis The numbers indicate the stage of development, in accordance with 0 especially 3 region HOX genes (characterized by the numbers the numbers in (a). HPRT was used as a control for RNA extraction and 1–6) are expressed in the most primitive cells, while at later conversion to cDNA. Data are representative for three independent stages of differentiation more 50 region HOX genes (7–13) experiments with three different donors.

Leukemia HOX gene expression in T-cell progenitors T Taghon et al 1161 Overall, however, the data show that there is no major that this gene is required for the development of this T-cell difference between CD34+LinÀ precursor cells from FL and lineage as well. This might indicate that slight changes in HOX- ABM in terms of HOX gene expression, and therefore it seems B3 expression can be decisive for the TCR-ab-vs TCR-gd-lineage that this gene family is not responsible for the difference in T-cell choice. differentiation potential . HOX-C4 expression has previously been detected in T-cell leukemias and in normal peripheral T cells42 and it is shown here that also ABM and CT progenitor cells express this gene HOX gene expression during human T-cell (Figure 1b). We further investigated the expression of this gene development during human thymopoiesis and it is shown that HOX-C4 is expressed at all stages of T-cell development (Figure 3b). To have a better insight in which HOX genes might be involved Combined with the expression in peripheral T cells,42 this result in T-cell differentiation, we also analyzed the expression of this suggests a potential role of this gene during normal development gene family in the earliest T-cell precursors within the human of T cells. Recently, the effect of HOX-C4 overexpression in thymus, namely in CD34+CD1À cells, which are still multi- human CD34+ cells was analyzed, but no data on T-cell potential, and in CD34+CD1+ cells, which are committed to the development were given.43 T-cell lineage. Larger differences in terms of HOX gene The most obvious feature of CT precursor cells is the high expression were observed when comparing FL and ABM expression level of HOX-A10 compared to FL and ABM populations with both CT populations (Figure 1). As the former CD34+LinÀ cells, especially within the CD34+CD1À CT fraction populations are the precursor cells that migrate towards the (63% of all clones analyzed, Figure 1a). These data would thymus to enable T-cell development, it seems that upon thymic therefore suggest a role for HOX-A10 in T-cell development. To entry a substantial change in HOX gene expression occurs. further investigate this, we developed a specific RT-PCR that Overall, the expression of 30 region genes was reduced within shows that HOX-A10 is expressed in immature thymocytes but these CT populations (17% in CD34+CD1À cells and 26% in is absent in single positive CD4 and CD8 mature thymocytes CD34+CD1+ cells) compared to FL (36%) and ABM (39%) (Figure 3b). This correlates with the observation that enforced CD34+ LinÀ cells, suggesting that upon entry of progenitor cells expression of this gene in CD34+LinÀ cord blood cells severely within the thymus, differentiation is already occurring, even in inhibits final differentiation of human T cells.4 The observed the most immature CD34+CD1À CT cells. In accordance, HOX- inhibitory effect of overexpression of HOX-A10, however, does B4, a gene that has been associated with stem cell self- not imply that this gene is not involved or required for the early renewal,34,35 is virtually absent within both CT populations stages of T-cell development. (Figure 1b). This is in agreement with previous reports that We were also able to detect HOX-A11 within both CT discussed the absence of true stem cell characteristics within populations (Figure 1a). This gene was not detected in FL or human CD34+ thymocytes compared to similar cell fractions ABM precursor cells and has not been reported previously to be from other sources.39,40 Surprisingly, the relative expression expressed in cells of hematopoietic origin. Further analysis with level of HOX-A9 is severely reduced upon thymic entry specific RT-PCR showed that HOX-A11 was expressed at all (CD34+CD1À cells), despite the fact that this gene is essential major stages of T-cell development (Figure 2b), indicating that during T-cell development.24 However, once the cells become this HOX gene might have an important role in this develop- T-cell committed (CD34+CD1+ cells), HOX-A9 expression is mental pathway. upregulated again (Figure 1a). Owing to the fact that in HOX- HOX-A7 was one of the few genes that was expressed at a A9À/À mice the block in T-cell development is after entry of comparable level in all analyzed stem cell and progenitor cell precursor cells into the thymus,24 this might indicate that HOX- populations (Figure 1a). A specific RT-PCR in all thymocyte A9 has no crucial role in the earliest multipotential progenitor differentiation stages showed that this gene was only expressed cells in the thymus, but only becomes important once these cells in the very early stages of T-cell development (Figure 2 and 3b). are developing towards the T-cell lineage. To investigate this in A remarkable observation concerning the expression profile more detail, we analyzed HOX-A9 expression in cell popula- of the analyzed HOX-A cluster genes (HOX-A7, HOX-A9, tions that represent the different stages of human T-cell HOX-A10 and HOX-A11, Figure 3b) is the fact that all these development. This shows that only in the early stages of T-cell genes are expressed during the early stages of T-cell develop- development HOX-A9 is expressed, compatible with its ment, but more importantly, that further maturation is accom- essential role during early differentiation (Figure 3b). HOX-A9- panied by the sequential downregulation of the most 30 region deficient murine thymocytes display a reduced expression of the gene of this cluster. HOX-A7 is expressed in developing receptor for IL-7,24 a necessary cytokine during the early phases thymocytes until the CD4+CD8a+CD8bÀ stage, HOX-A9 until of T-cell development.41 As the expression profile of HOX-A9 the CD4+CD8b+CD3À stage, HOX-A10 until the (Figures 2 and 3b) corresponds remarkably well with the IL-7- CD4+CD8+CD3+CD69+ and HOX-A11 throughout all stages. dependent stages of thymocyte differentiation (Figure 3a), this This phenomenon corresponds with the general idea that 30 suggests that also in human this gene is involved in the region HOX genes are expressed in the most primitive cells, regulation of the IL-7 receptor. The lack of HOX-A9 expression while 50 region genes are present in more mature cell types.31 at later stages of TCR-ab T-cell development (Figure 3b) seems This also suggests that the human T-cell development by itself to indicate that this gene is dispensable for the final maturation should be regarded as a separate developmental system, in of human T cells. which HOX genes play a similar well-organized role as during HOX-B3 was also detected within the CT populations, but at a embryonic development.44 low frequency (Figure 1b). As HOX-B3 overexpression mainly In conclusion, we have shown that upon differentiation results in TCR-gd T-cell development,25 the expression of this towards the T-cell lineage, hematopoietic progenitor cells gene in early T-cell development might reflect their potential to undergo important changes in terms of HOX gene expression differentiate towards this lineage. However, upon further and provide evidence that these genes are involved in the analysis, we observed that HOX-B3 is expressed at all stages complex regulatory network of transcription factors that control of human TCR-ab T-cell development (Figure 3b), suggesting human T-cell development. Our results should incite further

Leukemia HOX gene expression in T-cell progenitors T Taghon et al 1162 study into the role of HOX genes in both normal and leukemic 17 Ting CN, Olson MC, Barton KP, Leiden JM. Transcription factor hematopoiesis. GATA-3 is required for development of the T-cell lineage. Nature 1996; 384: 474–478. 18 Oosterwegel M, van de Wetering M, Dooijes D, Klomp L, Winoto Acknowledgements A, Georgopoulos K et al. Cloning of murine TCF-1, a T cell-specific transcription factor interacting with functional motifs in the CD3- We thank Christine Brunaud, Mario Vaneechoutte, Chris epsilon and T cell receptor alpha enhancers. J Exp Med 1991; 173: Verhofstede and Thierry De Baere for their excellent scientific 1133–1142. advice, and Ilse Swennen, Leentje Van Simaey, Nancy De 19 Tomita K, Hattori M, Nakamura E, Nakanishi S, Minato N, Cabooter and Jose´ De Bosscher for their technical assistance. Kageyama R. The bHLH gene Hes1 is essential for expansion of early T cell precursors. Genes Dev 1999; 13: 1203–1210. This work was supported by grants from Ghent University, 20 Schilham MW, Moerer P, Cumano A, Clevers HC. Sox-4 facilitates Belgium; the Flanders Institute for Biotechnology (VIB), Belgium; thymocyte differentiation. Eur J Immunol 1997; 27: 1292–1295. and the Fund for Scientific Research Flanders (FWO-Vlaanderen), 21 Spain LM, Guerriero A, Kunjibettu S, Scott EW. T cell development Belgium. in PU.1-deficient mice. J Immunol 1999; 163: 2681–2687. 22 Rothenberg EV, Anderson MK. Elements of transcription factor network design for T-lineage specification. Dev Biol 2002; 246: References 29–44. 23 Lawrence HJ, Helgason CD, Sauvageau G, Fong S, Izon DJ, 1 Kornberg TB. Understanding the homeodomain. J Biol Chem 1993; Humphries RK et al. Mice bearing a targeted interruption of the 268: 26813–26816. homeobox gene HOXA9 have defects in myeloid, erythroid, and 2 Lawrence PA, Morata G. Homeobox genes: their function in lymphoid hematopoiesis. Blood 1997; 89: 1922–1930. Drosophila segmentation and pattern formation. Cell 1994; 78: 24 Izon DJ, Rozenfeld S, Fong ST, Komuves L, Largman C, Lawrence 181–189. HJ. Loss of function of the homeobox gene Hoxa-9 perturbs early 3 Krumlauf R. Hox genes in vertebrate development. Cell 1994; 78: T-cell development and induces apoptosis in primitive thymo- 191–201. cytes. Blood 1998; 92: 383–393. 4 Taghon T, Stolz F, De Smedt M, Cnockaert M, Verhasselt B, Plum J 25 Sauvageau G, Thorsteinsdottir U, Hough MR, Hugo P, Lawrence et al. HOX-A10 regulates hematopoietic lineage commitment: HJ, Largman C et al. Overexpression of HOXB3 in hematopoietic evidence for a monocyte-specific transcription factor. Blood 2002; cells causes defective lymphoid development and progressive 99: 1197–1204. myeloproliferation. Immunity 1997; 6: 13–22. 5 Magli MC, Largman C, Lawrence HJ. Effects of HOX homeobox 26 Plum J, De Smedt M, Defresne MP, Leclercq G, Vandekerckhove genes in blood cell differentiation. J Cell Physiol 1997; 173: 168– B. Human CD34+ fetal liver stem cells differentiate to T cells in a 177. mouse thymic microenvironment. Blood 1994; 84: 1587–1593. 6 Lawrence HJ, Sauvageau G, Humphries RK, Largman C. The role 27 Robin C, Bennaceur-Griscelli A, Louache F, Vainchenker W, of HOX homeobox genes in normal and leukemic hematopoiesis. Coulombel L. Identification of human T-lymphoid progenitor cells Stem Cells 1996; 14: 281–291. in CD34+ CD38 low and CD34+ CD38+ subsets of human cord 7 Sato Y, Abe S, Mise K, Sasaki M, Kamada N, Kouda K et al. blood and bone marrow cells using NOD-SCID fetal thymus organ Reciprocal translocation involving the short arms of chromosomes cultures. Br J Haematol 1999; 104: 809–819. 7 and 11, t(7pÀ;11p+), associated with myeloid leukemia with 28 Galy AH, Cen D, Travis M, Chen S, Chen BP. Delineation of T- maturation. Blood 1987; 70: 1654–1658. progenitor cell activity within the CD34+ compartment of adult 8 Borrow J, Shearman AM, Stanton VP, Becher R, Collins T, Williams bone marrow. Blood 1995; 85: 2770–2778. AJ et al. The t(7;11)(p15;p15) translocation in acute myeloid 29 Spits H, Blom B, Jaleco AC, Weijer K, Verschuren MC, van leukaemia fuses the genes for nucleoporin NUP98 and class I Dongen JJ et al. Early stages in the development of human T, homeoprotein HOXA9. Nat Genet 1996; 12: 159–167. natural killer and thymic dendritic cells. Immunol Rev 1998; 165: 9 Fujino T, Suzuki A, Ito Y, Ohyashiki K, Hatano Y, Miura I et al. 75–86. Single-translocation and double-chimeric transcripts: detection of 30 Vanhecke D, Verhasselt B, De Smedt M, Leclercq G, Plum J, NUP98-HOXA9 in myeloid leukemias with HOXA11 or HOXA13 Vandekerckhove B. Human thymocytes become lineage com- breaks of the chromosomal translocation t(7;11)(p15;p15). Blood mitted at an early postselection CD69+ stage, before the onset of 2002; 99:1428–1433. functional maturation. J Immunol 1997; 159: 5973–5983. 10 Dash A, Gilliland DG. Molecular genetics of acute myeloid 31 Sauvageau G, Lansdorp PM, Eaves CJ, Hogge DE, Dragowska WH, leukaemia. Best Pract Res Clin Haematol 2001; 14: 49–64. Reid DS et al. Differential expression of homeobox genes in 11 Kawagoe H, Humphries RK, Blair A, Sutherland HJ, Hogge DE. functionally distinct CD34+ subpopulations of human bone Expression of HOX genes, HOX cofactors, and MLL in phenoty- marrow cells. Proc Natl Acad Sci USA 1994; 91: 12223–12227. pically and functionally defined subpopulations of leukemic and 32 Ramalho-Santos M, Yoon S, Matsuzaki Y, Mulligan RC, Melton normal human hematopoietic cells. Leukemia 1999; 13: 687–698. DA. ‘Stemness’: transcriptional profiling of embryonic and adult 12 Lawrence HJ, Sauvageau G, Ahmadi N, Lopez AR, LeBeau MM, stem cells. Science 2002; 298: 597–600. Link M et al. Stage- and lineage-specific expression of the 33 Ivanova NB, Dimos JT, Schaniel C, Hackney JA, Moore KA, HOXA10 homeobox gene in normal and leukemic hematopoietic Lemischka IR. A stem cell molecular signature. Science 2002; 298: cells. Exp Hematol 1995; 23: 1160–1166. 601–604. 13 Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, 34 Sauvageau G, Thorsteinsdottir U, Eaves CJ, Lawrence HJ, Largman Mesirov JP et al. Molecular classification of cancer: class discovery C, Lansdorp PM et al. Overexpression of HOXB4 in hematopoietic and class prediction by gene expression monitoring. Science 1999; cells causes the selective expansion of more primitive populations 286: 531–537. in vitro and in vivo. Genes Dev 1995; 9: 1753–1765. 14 Giampaolo A, Felli N, Diverio D, Morsilli O, Samoggia P, Breccia 35 Antonchuk J, Sauvageau G, Humphries RK. HOXB4-induced M, Lo CF, Peschle C, Testa U. Expression pattern of HOXB6 expansion of adult hematopoietic stem cells ex vivo. Cell 2002; homeobox gene in myelomonocytic differentiation and acute 109: 39–45. myeloid leukemia. Leukemia 2002; 16: 1293–1301. 36 Peault B, Weissman IL, Baum C, McCune JM, Tsukamoto A. 15 Bijl JJ, van Oostveen JW, Walboomers JM, Horstman A, van den Lymphoid reconstitution of the human fetal thymus in SCID mice Brule AJ, Willemze R et al. HOXC4, HOXC5, and HOXC6 with CD34+ precursor cells. J Exp Med 1991; 174: 1283–1286. expression in non-Hodgkin’s lymphoma: preferential expression of 37 Miller JS, McCullar V, Punzel M, Lemischka IR, Moore KA. Single the HOXC5 gene in primary cutaneous anaplastic T-cell and oro- adult human CD34(+)/LinÀ/CD38(À) progenitors give rise to gastrointestinal tract mucosa-associated B-cell lymphomas. Blood natural killer cells, B-lineage cells, dendritic cells, and myeloid 1997; 90: 4116–4125. cells. Blood 1999; 93: 96–106. 16 Res P, Spits H. Developmental stages in the human thymus. Semin 38 Thorsteinsdottir U, Mamo A, Kroon E, Jerome L, Bijl J, Lawrence HJ Immunol 1999; 11: 39–46. et al. Overexpression of the myeloid leukemia-associated Hoxa9

Leukemia HOX gene expression in T-cell progenitors T Taghon et al 1163 gene in bone marrow cells induces stem cell expansion. Blood 42 Bijl J, van Oostveen JW, Kreike M, Rieger E, van der Raaij-Helmer 2002; 99: 121–129. LM, Walboomers JM et al. Expression of HOXC4, HOXC5, 39 Res P, Martinez-Caceres E, Cristina JA, Staal F, Noteboom E, and HOXC6 in human lymphoid cell lines, leukemias, and Weijer K et al. CD34+CD38dim cells in the human thymus can benign and malignant lymphoid tissue. Blood 1996; 87: 1737– differentiate into T, natural killer, and dendritic cells but are 1745. distinct from pluripotent stem cells. Blood 1996; 87: 5196–5206. 43 Daga A, Podesta M, Capra MC, Piaggio G, Frassoni F, Corte G. The 40 Kerre TC, De Smet G, De Smedt M, Offner F, De Bosscher J, Plum J retroviral transduction of HOXC4 into human CD34(+) cells et al. Both CD34+38+ and CD34+38À cells home specifically to induces an in vitro expansion of clonogenic and early progenitors. the bone marrow of NOD/LtSZ scid/scid mice but show different Exp Hematol 2000; 28: 569–574. kinetics in expansion. J Immunol 2001; 167: 3692–3698. 44 Duboule D, Morata G. Colinearity and functional hierarchy among 41 Plum J, De Smedt M, Leclercq G, Verhasselt B, Vandekerckhove B. genes of the homeotic complexes. Trends Genet 1994; 10: 358– Interleukin-7 is a critical growth factor in early human T-cell 364. development. Blood 1996; 88: 4239–4245.

Leukemia