Effects of Overexpression of HBP1 Upon Growth and Differentiation of Leukemic Myeloid Cells

Leukemia (2005) 19, 1958–1968 & 2005 Nature Publishing Group All rights reserved 0887-6924/05 $30.00 www.nature.com/leu Effects of overexpression of HBP1 upon growth and differentiation of leukemic myeloid cells

CJ Yao1,2, K Works1, PA Romagnoli2,3 and GE Austin1,2

1Department of Pathology and Laboratory Medicine, Veterans Affairs Medical Center, Decatur, GA, USA; 2Emory University School of Medicine, Atlanta, GA, USA; and 3Department of Pediatrics, Veterans Affairs Medical Center, Decatur, GA, USA

HMG-box containing protein 1 (HBP1) is a member of the high sequences: an LXCXE site and an IXCXE site. This protein also mobility group (HMG) of chromosomal proteins. Since HBP1 contains a 28 amino-acid sequence that shows 86% homology exhibits tumor-suppressor activity in nonmyeloid tissues, we with a transcriptional activation domain previously demon- examined the effects of ectopic overexpression of HBP1 upon strated in rat HBP1.11 This activation domain is flanked by two the growth and differentiation of myeloid cells. We prepared transient and stable transfectants of the myeloblast cell line putative repressor domains (identified by comparison with the K562, which overexpress HBP1 mRNA and protein. HBP1 rodent HBP1 gene). The presence of these sequences suggests transfectants displayed slower growth in cell culture and that human HBP1 may activate and/or repress transcription of reduced colony formation in soft agar, retardation of S-phase other genes, in part through interaction with RB. progression, reduced expression of cyclin D1 and D3 mRNAs Information developed by ourselves and other investigators and increased expression of p21 mRNA. HBP1 transfectants indicates that HBP1 is an important regulator of the cell cycle, also underwent increased apoptosis, as demonstrated by cell cycle exit, apoptosis and terminal differentiation in a variety morphology and binding of Annexin V. Fas ligand mRNA levels 11–22 were increased in HBP1 transfectants, suggesting involvement of tissues and cell types. In specific nonmyeloid tissues it of the Fas/Fas ligand pathway. HBP1 overexpression enhanced partners with the tumor-suppressor RB and other pocket differentiation of K562 cells towards erythroid and megakar- proteins,11–13 inhibits the p47phox gene,22 targets downstream yocyte lineages, as evidenced by increased hemoglobin and factors such as N-Myc, Histone H1, and cyclin D1,23 and alters CD41a expression. Overexpression of HBP1 modulated mRNA chromatin structure by bending DNA and recruiting histone levels for myeloid-specific transcription factors C/EBPa, c-Myb, 17 c-Myc, and JunB, as well as lineage-specific transcription deacetylase cofactors. However, its role in myeloid cell factors PU.1, GATA-1, and RUNX1. These findings suggest that growth and differentiation remains to be elucidated. in myeloid cells HBP1 may serve as a tumor suppressor and a Mounting evidence from nonmyeloid tissues suggests that general differentiation inducer and may synergize with chemi- HBP1 may be a tumor suppressor. For example, HBP1 inhibits cal differentiating agents to enhance lineage-specific differen- G1 progression in fat and muscle cells12 and blocks the Wnt tiation. pathway,21 which is important in cancer. The HBP1 gene lies Leukemia (2005) 19, 1958–1968. doi:10.1038/sj.leu.2403918; within the 7q31 chromosomal region that is frequently deleted published online 22 September 2005 10 23 Keywords: HBP1; high mobility group proteins; transcription; or translocated in cancers. Yee et al have recently isolated myeloid differentiation; cell cycle variants of HBP1 in breast cancers and myeloid leukemias, suggesting a possible role of HBP1 in human neoplasia. Introduction However, the role of HBP1 in pathogenesis of myelodysplasias and myeloid leukemias remains to be elucidated. HMG-box containing protein 1 (HBP1) was first identified in rat To explore the role of HBP1 in regulating the growth of myeloid brain, on the basis of its ability to suppress the potassium cells and to study its possible tumor-suppressor activity we transport-defective phenotype of mutant Saccharomyces cere- overexpressed HBP1 in the myeloblast cell line K562 and studied visiae.1 HBP1 was recognized to be a member of the high its effect upon cell growth and differentiation. Our results show mobility group (HMG) protein family, since it contains a 70 that overexpression of HBP1 in myeloblasts leads to growth arrest amino-acid DNA-binding domain known as an HMG-box, and apoptosis and acts as a general inducer of differentiation. characteristic of these proteins.2 HMG proteins are transcription factors that are widely distributed among eukaryotic cells and Materials and methods have diverse functions.2–9 In the course of a yeast one-hybrid search for novel factors that bind to the MPO promoter and Materials modulate MPO promoter activity, we identified HBP1, as an 10 MPO promoter-binding protein. We demonstrated by trans- The human cell line K562 (erythroleukemic) was obtained from fection experiments that HBP1 enhances MPO promoter the American Type Culture Collection (Rockville, MD, USA). activity. We isolated and cloned the cDNA for the human Polyclonal rabbit antibody against HBP1 was prepared as HBP1 gene (GenBank Accession Number AF019214, October previously described.10 The pBudCE4.1 vector, Zeocin antibiotic, 2, 1997). The corresponding human HBP1 mRNA sequence and DMRIE-C transfection reagent were obtained from Invitrogen contains a single open reading frame, which codes for a protein (Carlsbad, CA, USA). Anti-CD41a (FITC-conjugated) and other composed of 514 amino acids. In addition to the HMG box-like antibodies used in flow cytometric analyses were obtained from sequence, which is the DNA-binding region, the amino-acid BD Biosciences (San Jose, CA, USA). SYBRs Green PCR Master sequence of HBP1 includes two pocket protein-binding Mix was from Applied Biosystems (Foster City, CA, USA).

Correspondence: Dr GE Austin, Atlanta VA Medical Center (113), 1670 Clairmont Road, NE, Decatur, GA 30033, USA; Plasmid construction Fax: þ 1 404 235 3007; E-mail: [email protected] Received 7 January 2005; accepted 30 June 2005; published online Wild-type HBP1 cDNA was subcloned into pBudCE4.1 22 September 2005 vector, which has dual promoters, pCMV and pEF1a, and a HBP1 and myeloid cell growth and differentiation CJ Yao et al 1959 Zeocin-resistance gene for Zeocin selection of stable cells. (30 mg total protein) were placed in SDS buffer and applied to Two types of constructs containing HBP1 were established 10% SDS-PAGE gels and subsequently transferred to nitrocellu- as follows: HBP1/CMV (with HBP1 driven by a CMV lose membranes. Membranes were incubated with the desired promoter) was prepared by ligating HBP1 to pBudCE4.1 at the primary antibodies, followed by suitable secondary antibodies. multiple cloning site of the CMV promoter at SalI and BamHI Antigen detection was performed with an ECL plus kit restriction sites. HBP1/EF1a (with HBP1 driven by an EF1a (Amersham Biosciences Corp., Piscataway, NJ, USA) promoter) was prepared by ligating HBP1 to modified pBudCE4.1 (cutting off the whole segment containing the CMV promoter) at the multiple cloning site of the EF1a promoter Cell proliferation assay at NotI and XhoI restriction sites. All of the constructs were verified by sequencing. The control vector employed was Cell proliferation was measured by a BrdU incorporation assay pBudCE4.1/LacZ/CAT (hereafter referred to as ‘CAT vector’). kit from BD Pharmingen (San Diego, CA, USA). Briefly, the cells The linearization sites of these constructs were: NheI for HBP1/ were pulse labeled with BrdU for 30 min and the levels of BrdU CMV, FspI for HBP1/EF1a, and BspHI for CAT vector. All incorporated into cellular DNA were quantified by anti-BrdU constructs were expanded in Escherichia coli DH5a and antibody; total DNA was stained by 7-AAD. The samples were plasmids were purified using a Maxi-prep kit (Qiagen, Valencia, analyzed by BD FACScan and data were evaluated using CA, USA). WinMDI 2.8 research software. This two-color flow cytometric analysis permits the enumeration and characterization of cells that are actively synthesizing DNA (BrdU incorporation) in Cell culture, establishment of stable cell lines and terms of their cell cycle position (ie, G0/1, S, or G2/M phases transfection (defined by 7-AAD staining intensities). For these and other assays, the statistical significance of the differences between K562 cells were maintained in RPMI 1640 (GibcoBRL, K562 cells and stable transfectants was assessed using a two- Gaithersburg, MD, USA) supplemented with 1% L-glutamine, tailed t-test at Po0.05 or Po0.005. 100 U/ml penicillin, 100 mg/ml streptomycin, and 10% heat-inactivated fetal calf serum, and grown in humidified 5% CO2 at 371C. Exponentially growing cells were used for all Apoptosis assay experiments. For determination of growth rate, cells were diluted at an initial concentration of 0.2 Â 106 cells/ml in Apoptosis assays were performed using an Annexin V-FITC complete medium and aliquots were removed daily for detection kit I (BD Biosciences Pharmingen). The cells were determination of cell concentration. In experiments addres- analyzed using a FACScan flow cytometer and data were sing induction of differentiation, cells at an initial concentra- evaluated using WinMDI 2.8 research software. The results were tion of 0.2 Â 106 cells/ml were incubated in complete medium expressed as percentage of apoptotic Annexin V-FITC positive with or without 30 mM hemin or 10 ng/ml TPA (Sigma Chemical cells with respect to total cells counted. Co, St Louis, MO, USA). For growth in soft agar, cells were cultured in RPMI-1640 medium containing 0.3% Bectar agar, and 20% fetal calf serum. After 10 days colonies containing May–Grunwald–Giemsa staining, and benzidine more than 50 cells were enumerated using an inverted staining microscope. To establish stable HBP1-overexpressing cell lines, 1.6 Â 107 Approximately 5 Â 104 cells were spun onto a microscope slide K562 cells were transfected by the linearized constructs (32 mg in a Cytospin 3 cytocentrifuge (Shandon, Thermo Electron each) with 48 ml DMRIE-C reagent (Invitrogen). Stably trans- Corporation, Pittsburgh, PA, USA). After air drying, slides were fected cells, except K562 cells only, were selected in culture stained with May–Grunwald–Giemsa stain (Sigma, St Louis, medium containing 200 mg/ml of Zeocin. Colonies (bulk MO, USA). cultures) of Zeocin-resistant cells were obtained approximately The percentage of cells staining for hemoglobin was estimated 24 14 days after selection and were propagated. HBP1-over- by staining with benzidine/H2O2 essentially as described. At expressing stable cell lines were screened by quantitative real- least 500 cells were counted and the number of cells containing time RT-PCR (QRT-PCR), and further confirmed by Western oxidized tetramethylbenzidine (visualized as bright blue-stained blotting. Control cell lines transfected with CAT vector were cells) was taken as indicative of peroxidase activity, thus screened using a chloramphenicol acetyltransferase (CAT) reflecting hemoglobin production. enzyme-linked immunosorbent assay kit (Boehringer Man- nheim), according to the manufacturer’s specifications. These bulk cultures of stably transfected cells were then subjected to Flow cytometric assessment of cell surface antigen limiting dilution to isolate individual clones. Eight clones were expression obtained from each of the bulk cultures. Two representative clones derived from each construct were fully analyzed and Expression of cell surface antigens in differentiating cells was described in this communication. These were designated HBP1/ determined by flow cytometry. K562 cells (106 cells/sample) CMV-1, HBP1/CMV-2, HBP1/EF1a-1, HBP1/EF1a-2, CAT-1, and were washed and resuspended in PBS. Fluorescent-conjugated CAT-2. antibody (10 ml) against the appropriate cell surface antigens (eg FITC-conjugated mouse antibody CD41a) or isotype-matched control IgG were added to the cell suspension to stain the cells. Preparation of cell lysates and Western blot analysis After incubation at room temperature for 30 min, the stained cells were washed and fixed with formaldehyde. Cell fluores- Total cell extracts from 107 K562 cells or stable transfectants cence was measured on a FACScan flow cytometer (BD were prepared as described by Shih et al.13 The cell lysates Biosciences).

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1960 Quantitative real-time RT-PCR

Total RNA was isolated using the Trizol reagent (Invitrogen), and 1 mg of RNA was subjected to reverse transcription using ThermoScriptt RT (Invitrogen). Specific primers were designed using Primer3 software (MIT Whitehead Institute, http:// www.basic.nwu.edu/biotools/Primer3.html) (see Supplementary Information). In total, 40 cycle, two-step, real-time PCR was performed on an iCycler iQ multicolor real-time PCR detection system (Bio-Rad). Each sample was run in triplicate. Cycle threshold (Ct) values were obtained and relative RNA expression was determined using the formula Relative Expression ¼ (DDCt) 2 Â 100, where DDCt ¼ (Ct gene of interestCt GAPDH in experimental sample)(Ct gene of interestCt GAPDH in untransfected, untreated K562 cells sample). Amplicon size and Figure 1 HBP1 expression vectors. (a) HBP1 with CMV promoter n reaction specificity were confirmed by agarose gel electrophoresis. (HBP1/CMV); (b) HBP1 with EF1a promoter (HBP1/EF1a). CMV promoter excised by cleavage with SpeI and XbaI and ligation of the ends. Co-immunoprecipitation

K562 cells were transiently cotransfected with pRC/RB and HBP1/EF1a. Cell lysates from 107 cells were prepared as described above. After measuring the protein concentration, 1 mg of cell lysate was precleared using protein A- or protein G-Sepharose beads (Santa Cruz) before antibodies were added. Anti-HBP1 or anti-RB (or other) antibody and beads were added to the precleared cell lysate and incubated at 41C overnight with gentle agitation. Bound proteins were washed, centrifuged, the supernatant removed and the immunocomplexes resuspended in sample buffer containing b-mercaptoethanol; subsequently 200 mg aliquots of proteins were analyzed by Western blotting. K562 cells transfected with only the CAT vector served as the negative control.

Results

Establishment of stable transfectants of K562 cells that overexpress HBP1

We have previously transiently transfected myeloid cell lines with HBP1-expressing plasmids and have demonstrated increased HBP1 RNA and protein levels in the transfected cells.10 However, in order to study the effect of overexpression of HBP1 on myeloid cell growth and differentiation, stable transfectants were prepared. To prepare clones with limited overexpression of Figure 2 HBP1 mRNA and protein in clones of stably transfected HBP1 we took advantage of our observation that the CMV K562 cells. (a) Relative expression of HBP1 RNA determined by QRT- promoter produced lower levels of HBP1 expression than did PCR. The ratios between mRNA levels were calculated from the (DDC ) the EF1a promoter. Hence, we prepared four sets of parallel 2 t formula (see Materials and methods) and expressed as 7 cultures of K562 cells, as descried in Materials and methods: mean s.d. (n ¼ 12). (b) Western blot showing levels of HBP1 and b-actin protein in clones of transfected K562 cells, and in untrans- (1) cells transfected with HBP1/CMV; (2) cells transfected with fected K562 cells and CAT-transfected cells. HBP1/EF1a; (3) cells transfected with CAT vector; and (4) untransfected K562 cells. Figure 1 shows diagrams of the HBP1 expression vectors HBP1/CMV and HBP1/EF1a. Figure 2b illustrates Western blot data showing increased Bulk populations of transfected cells were prepared and HBP1 expression in single clones derived from HBP1-trans- individual clones were isolated through single cell limiting fected cells. HBP1 protein expression was greatest in the clones dilution from these bulk populations. Expression of HBP1 was derived from HBP1/EF1a-transfected cells. Expression of the screened by Western blotting and QRT-PCR. Eight clones were b-actin protein was essentially identical in all clones. selected from each bulk population and two representative clones prepared from each of the constructs were fully studied and described in this communication. Overexpressed HBP1 protein interacts with RB HBP1 RNA expression in the HBP1/CMV clones (Figure 2a) in K562 cells was increased by about 40–50-fold compared with that in untransfected K562 cells, while that in the HBP1/EF1a clones Since HBP1 protein contains pocket protein-binding motifs, and was increased by 80–100-fold. since previous studies in fat and muscle cells had suggested that

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1961 HBP1 in those tissues may interact with RB and other factors HBP1 inhibits cell cycle progression that regulate the cell cycle and cell differentiation, we investigated the interactions between HBP1 and RB in K562 Cell proliferation is determined, in part, by the relative balance cells that were overexpressing these proteins. When those between cell cycle progression and apoptosis. To address the cells were transiently cotransfected with RB and HBP1, effect of overexpressed HBP1 in the context of cell cycle immunoprecipitation of RB co-immunopurified HBP1, as regulation, the cell cycle profiles were determined by a BrdU demonstrated by a Western blot with an HBP1 antibody incorporation assay. K562 cells overexpressing HBP1 showed (Figure 3a, lane 2), while immunoprecipitation with a control an increase in cells residing in the G0/G1 phase (41–44%) in antibody (against AP-1) did not bring down HBP1 under these HBP1-overexpressing cells compared with 28–31% in cells conditions (Figure 3a, lane 3). HBP1 migrates in gels at a transfected with the CAT vector or non-transfected K562 position equivalent to 80 kDa, which is higher than expected cells (Table 1). Likewise, HBP1-overexpressing cells showed from its molecular weight, but similar to the figure observed by a decrease in percentage of cells in S phase (40–46%) other investigators.12 Likewise, when K562 cells were cotrans- compared with 55–59% in CAT-transfected and nontransfected with RB and HBP1, immunoprecipitation of HBP1 co- fected K562 cells, suggesting that the rate of BrdU incorporation immunopurified RB, as recognized by a Western blot with an RB in S phase was also decreased (Table 1). These results indicate antibody (Figure 3b, lane 1), and immunoprecipitation with a that HBP1 inhibits cell cycle progression in G1 phase in K562 control antibody (against AP-1) did not bring down RB under cells. these conditions (Figure 3b, lane 3). (The antibody employed precipitates total RB with a molecular weight of 110 kDa.) These data confirm that RB and HBP1 can form a specific complex in HBP1 decreases the expression of specific cyclins which K562 cells. regulate G0/G1-to-S progression

To explore the mechanisms by which HBP1 overexpression HBP1 inhibits cell proliferation in K562 cells slows cell cycle progression, we examined the expression of multiple cell cycle regulatory genes in HBP1-overexpressing To analyze whether expression of HBP1 affects the proliferation cells. Genes examined include cyclins D1, D3, and E as well as of K562 cells, all clones were plated and allowed to grow in CDK2 and CDK4, CDK inhibitors p21, p27, and p16. Total standard media for 4 days. The HBP1-expressing clones showed RNAs from HBP1-overexpressing cell lines were analyzed by a reduced proliferation rate compared with nontransfected cells QRT-PCR. Overexpression of HBP1 decreased the expression of or the CAT-transfected clone, as measured by total number of mRNAs for cyclin D1 and D3, but not the E cyclins and cells/ml (Figure 4a). Cells with the highest HBP1 expression increased the message level of p21 (Figure 5). These findings (HBP1/EF1a clones) showed the slowest growth rate compared suggest that HBP1 effects upon cell cycle progression may be with the other clones. Changes in bulk culture growth rates mediated by changes in cyclin expression during G1. No followed similar kinetics as did the corresponding single clones significant changes were seen in mRNAs corresponding to (data not shown). These findings suggest that HBP1 inhibits cell CDK2, CDK4, p53, p16, or p27 or in the expression of the proliferation in K562 cells. housekeeping gene GAPDH. All cells overexpressing HBP1 K562 cells and all clones derived from them were also tested showed a marked increase in expression of p21 mRNA for growth in a colony forming soft agar assay. K562 cells and following TPA treatment (data not shown). This observation is CAT clones formed more and larger colonies while HBP1 consistent with prior evidence that p21 enhances differentiation 25 overexpressing clones formed fewer and smaller colonies in K562 cells. (Figure 4b and c). HBP1 overexpression induces apoptosis

K562 cell clones overexpressing HBP1 showed morphologic changes indicative of apoptosis in increased numbers of cells. To more accurately quantitate the percentage of apoptotic cells, apoptosis was measured by Annexin binding, an early apoptotic event. The percentage of untransfected K562 cells and of K562 cells transfected with the CAT clone undergoing apoptosis, as measured by Annexin-V, ranged from 6 to 7%, whereas for the HBP1 stably transfected cells, the percentage of cells undergoing apoptosis ranged from 12% in HBP1-CMV clones to about 16–18% in HBP1/EF1a clones (Po0.05) (Figure 6a). Represen- tative dot plots showing percentages of Annexin positivity exhibited by each of the clone types are shown in Figure 6b-e. Figure 3 Immunoprecipitation/Western blots showing interaction of HBP1 and RB in transiently cotransfected K562 cells. (a) K562 cells These results indicate that higher expression of HBP1 can were cotransfected with pRC/RB and HBP1/EF1a (lanes 1–3), or CAT induce apoptosis in K562 cells. vector (lane 4), and immunoprecipitated with antibodies to HBP1 One pathway for the induction of apoptosis involves the (lanes 1, 4, and 5), RB (lane 2), AP-1 (lane 3). The pull-down protein interaction of Fas (CD95) with its ligand FasL (CD95L). As a ﬁrst complex is subjected to Western blotting with HBP1 antibody. Lane 5, step to exploring this pathway we examined the effect of HBP1 lysis buffer as negative control. (b) K562 cells were transiently overexpression on mRNA levels for FasL. Figure 7 shows data cotransfected with pRC/RB and HBP1/EF1a (lanes 1–3), or CAT vector only (lane 4), and immunoprecipitated with antibodies to HBP1 (lane comparing the levels of mRNA for FasL in the four groups of 1), RB (lanes 2, 4 and 5), or AP-1 (lane 3). The pull-down protein K562 cell clones. High level HBP1 overexpression (HBP1/EF1a complex is subjected to Western blotting with RB. Lane 5, lysis buffer. clones) induces an increase in FasL mRNA (normalized to

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1962

Figure 4 Overexpression of HBP1 in K562 cells inhibits cell growth and soft agar colony formation. (a) Cell growth. K562 cells and single clones of stable cells overexpressing HBP1, at an initial concentration of 0.2 Â 106/ml, were harvested at time points indicated and counted. Shown are mean values7s.d. from two independent experiments, each performed in triplicate. (b) Soft agar colony assay. K562, CAT clones, and HBP1-overexpressing clones were seeded in triplicate into 0.3% agar in six-well plates at 5 Â 103/ml. After 10 days, the plates were photographed. K562 and multiple transfectant clones were analyzed. Representative results from three independent experiments are shown. (c) Soft agar colony assay. After 10 days of culture, colonies containing more than 50 cells were enumerated. Results are mean7s.d. for three separate experiments.

Table 1 Cell cycle distribution of stably transfected or nontransfected K562 cells

G0+G1 S G2

K562 30.872.1 54.873.1 14.573.7 CAT-1 31.971.9 55.674.1 12.571.8 CAT-2 28.271.9 58.673.2 13.273.4 HBP1/CMV-1 41.172.8n 44.971.8n 14.071.8 HBP1/CMV-2 42.374.1n 45.573.3n 12.374.4 HBP1/EF1a-1 42.973.1n 39.575.2n 17.672.6 HBP1/EF1a-2 43.774.8n 41.673.8n 14.775.1 BrdU incorporation assay showing percentage of cells in different phases of the cell cycle. Values show percentage of viable cells. Mean values are from three Figure 5 Expression of cell cycle-related genes in K562 cells and independent experiments. stably transfected clones, relative to expression levels of the GAPDH n gene measured by QRT-PCR. Values are based on three independent Po0.05 compared with untransfected K562 cells. n experiments with triplicate measurements for each. Po0.05 compared with untransfected K562 cells. CCND1 ¼ cyclin D1; CCND3 ¼ cyclin D3; CCNE1 ¼ cyclin E1.

GAPDH mRNA) compared with that seen in untransfected K562 (data not shown). These data suggest that increased apoptosis cells or cells transfected with the CAT vector. This was true caused by HBP1 overexpression may involve the Fas/FasL either in the presence or absence of induction by TPA or hemin pathway.

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1963

Figure 6 HBP1 induces apoptosis in stable transfectants of K562 cells. The cells were labeled with FITC-Annexin V and propidium iodide and n analyzed by ﬂow cytometry. (a) Percentage of apoptotic cells (mean7s.d. of three separate experiments). Po0.05 compared with untransfected K562 cells. (b–e). Dot plots of analysis of cells. (b) K562; (c) CAT-1; (d) HBP1/CMV-1; (e) HBP1/EF1a-1. Percentages in right lower quadrant indicate cells in early apoptosis; percentages in right upper quadrant indicate cells at late stage of apoptosis. The results shown are representative of three different experiments.

Wright stain. Compared with untransfected or CAT transfected cells, HBP1 overexpression resulted in decreased mitotic rate and an increased number of multinucleated cells with promi- nent dense granules, suggestive of megakaryocytes (data not shown). Careful examination at a higher magnification also indicated the presence of increased numbers of small viable cells in the HBP1-overexpressing cultures. Since some of these cells stained positive with benzidine (see below) their presence may indicate differentiation toward the erythroid lineage. To confirm that HBP1 overexpression induced increased differentiation of K562 cells toward the erythroid lineage we carried out benzidine staining to demonstrate the presence of hemoglobin in the cells. While only about 2% of untransfected Figure 7 Levels of FasL mRNA in HBP1-overexpressing clones K562 cells or CAT clones showed erythroid differentiation, as and CAT clones compared with untransfected K562 cells measured by evidenced by benzidine positivity, about 30% of HBP1-over- QRT-PCR. All results are shown relative to the expression levels of the expressing cells were benzidine positive even in the absence of GAPDH housekeeping gene. Values are based on three independent n the inducing agent hemin (Figure 8). In this respect there was no experiments with triplicate measurements for each. Po0.05 com- significant difference among the various HBP1-overexpressing pared with untransfected K562 cells. clones (HBP1/CMV-1 or 2, and HBP1/EF1a-1 or 2). HBP1- overexpressing cells also showed increased hemoglobin expres- Overexpression of HBP1 enhances differentiation of sion after hemin treatment compared with untransfected K562 K562 cells toward erythroid or megakaryocytic lineages cells treated with the same agent. Specifically, about 50–60% of HBP1-overexpressing K562 cells treated for 24 h with hemin HBP1 has been postulated to have a role in the differentiation of were benzidine positive (Figure 8), compared with 20–25% for a number of tissues, such as liver, muscle and fat.12,13 To untransfected K562 cells treated with hemin. Interestingly, while determine whether or not HBP1 plays a role in the differentia- increasing the expression of megakaryocytic markers (see data tion of myeloid cells, we examined the effects of overexpression below), TPA treatment tended to reduce hemoglobin expression of HBP1 on differentiation of steady-state cultures of K562 in HBP1-overexpressing K562 cells compared with uninduced cells. To begin with, morphologic changes in HBP1-over- cells (Figure 8), although the latter change was not statistically expressing cultures and clones were compared with the significant. Cells examined after 48, 72, or 96 h of hemin or TPA morphology of untransfected K562 cells using a modified treatment showed hemoglobin expression patterns similar to

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1964 the blast marker CD34. On the other hand, increased expression of the erythroid marker CD71 (in the absence of CD33 expression), and of platelet markers CD41a and CD61 was observed. Figure 9 shows CD41a expression in the different clones. HBP1 induced an increase in CD41a expression in the absence of TPA treatment and caused higher levels after TPA treatment than were seen in the absence of HBP1 overexpression. Interestingly, even hemin treatment caused some increase in CD41a expression and this was enhanced by HBP1 overexpression (data not shown). This observation presumably reﬂected the multipotential nature of K562 cells, which have the ability to differentiate along both erythroid and megakaryocytic lineages.

Figure 8 Effects of HBP1 on erythroid differentiation of K562 cells, analyzed by benzidine oxidation test, reflecting hemoglobin content. HBP1 influences expression of lineage-specific Cells were seeded at a concentration of 0.2 Â 106/ml and incubated transcription factors involved in differentiation with or without hemin or TPA for 4 days. Cells were subjected to benzidine oxidation test daily. The percentage of benzidine-oxidizing The process of hematopoietic differentiation is orchestrated at cells after 24 h hemin or TPA treatment is shown. Results are mean n the molecular level by a complex network of transcription values from three independent experiments. Po0.05 compared with factors that act by regulating the expression of sets of target untransfected K562 cells; nnP 0.005. o genes. In order to investigate the relationship of HBP1 with other key transcription factors in regulating erythroid and megakaryocytic development, we used QRT-PCR to analyze expres- Table 2 Expression of surface markers in K562 cells and stable sion of a number of such regulatory genes to determine if they transfectants are upregulated or downregulated as a result of HBP1 overexpression. Figure 10a shows the normalized expression of Markers K562 CATa HBP1/CMVa HBP1/EF1aa mRNAs for GATA-1, RUNX1, PU.1 in the four groups of clones

n n derived from K562 cells. GATA-1 is a gene which is activated CD45 8276b,c 79786375 6077 7 7 7 n 7 n very early in erythroid development and plays a key role CD13 92 11 85 10 62 5 61 5 26 CD34 91712 86766977n 6377n in erythroid differentiation. Enforced GATA-1 expression in CD71 617467748379n 9176n myeloid cell lines promotes erythroid, megakaryocytic, or CD61 627565758578n 91711n combined differentiation, depending on the cell type.27 n Po0.05 compared with untransfected K562 cells. GATA-1 is scarcely expressed in quiescent erythroid progeni- aRepresentative data for two clones. tors, but is rapidly induced when these cells are induced to bFlow cytometry showing percentage of cells expressing surface erythroid differentiation by erythropoietin and then progres- markers (mean7s.d.). sively accumulates during erythroid maturation, being abun- c Mean values are from three independent experiments. dantly expressed during all stages of erythroid maturation. We found that GATA-1 was upregulated to a small extent in HBP1- overexpressing cells. GATA-1 levels increased in untransfected those of cells examined after 24 h of these treatments (data not K562 cells after hemin or TPA treatment. However, in HBP1- shown). These findings indicate that HBP1 overexpression in overexpressing cells GATA-1 levels increased to a greater extent K562 cells induced differentiation toward the erythroid lineage after the cells were induced with either hemin or TPA (data not and promoted or enhanced hemoglobin production induced by shown). hemin in these cells. RUNX1, a transcription factor involved in megakaryo- To further characterize the ability of HBP1 to induce cytic differentiation27 was upregulated as well in HBP1- differentiation of K562 cells along either erythroid or mega- overexpressing cells (Figure 10a). RUNX1 mRNA levels karyocyte lineages, four-color immunofluorescence flow cyto- increased to a greater extent after the cells were induced metry was carried out using fluorescent conjugated mAbs with TPA (but not with hemin) (data not shown). Interestingly, specific for expression of surface markers. HBP1 overexpression the mRNA level of PU.1, an important transcription factor produced a number of effects consistent with differentiation of involved in granulocyte differentiation, was slightly reduced these cells toward erythroid and megakaryocytic lineages and by HBP1 overexpression (Figure 10a). Hemin treatment reduced away from granulocyte/monocyte lineages. These changes PU.1 expression more in HBP1-overexpressing cells, while (Table 2) included a loss of granularity of the cells, a progressive TPA treatment restored PU.1 expression to a level similar to that loss of CD45 expression (CD45 expression is low in erythroid in K562 cells or the CAT control (data not shown). Transcripts and platelet precursors and high in cells of the granulocytic of C/EBPa, which is thought to be involved mainly in myeloid lineage), and a decrease in the granulocytic marker CD13 and in cell differentiation,28 were increased, surprisingly, up to 10- or

Figure 9 Flow cytometric measurement of megakaryocytic-speciﬁc surface antigen CD41a in stably transfected and nontransfected K562 cells with or without TPA treatment. White histograms, staining with the FITC-conjugated isotype control mAb; black histograms, staining with FITC- conjugated anti-CD41a mAb. The marker placed to the right of white histogram designates positive events. The mean value ¼ mean ﬂuorescence intensity of cells stained with CD41a mAb. (a) Nontransfected K562 cells; (b) K562 þ TPA treatment; (c) CAT-1; (d) CAT-1 þ TPA; (e) HBP1/CMV-1; (f) HBP1/CMV-1 þ TPA; (g) HBP1/EF1a; (h) HBP1/EF1a þ TPA. The histograms are from a representative experiment of three separate experiments each run in duplicate.

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1965

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1966 D1, cyclin D3, and c-Myc. Also, expression of the antiproli- ferative cyclin-dependent kinase inhibitor p21WAF1/CIP1 was increased in HBP1-overexpressing cells. p21 inhibits G1 progression by blocking the activity of cyclin-dependent kinases, providing an additional mechanism by which HBP1 may retard G1 progression. Our findings concerning the inhibition of cyclin D1 and c-Myc expression by HBP1 in K562 cells are in agreement with those of Sampson et al21 who showed that in the human embryonic kidney cell line HEK293, HBP1 overexpression repressed Wnt-b-catenin activation of the cyclin D1 promoter as well as endogenous expression of cyclin D1 and c-Myc genes, both of which are targets of the Wnt pathway. Our morphologic evidence that HBP1 overexpression in K562 cells increases the number of apoptotic cells, is supported by data showing an increase in Annexin V positivity in cells overexpressing HBP1. Levels of FasL were increased in HBP1- overexpressing cells, suggesting that binding of FasL to Fas may trigger apoptosis in these cells.32 However, expression of other preapoptotic factors examined was not consistently altered in HBP1-overexpressing cells. Thus, further work will be necessary to define the molecular mechanisms by which HBP1 stimulates apoptosis in K562 cells. HBP1 overexpression also appears to induce differentiation of Figure 10 Expression of differentiation-related transcription fac- K562 cells. Whereas most pro-differentiation factors promote tors in HBP1-overexpressing cells compared with K562 cells, differentiation toward only one lineage, HBP1 appears to measured by QRT-PCR. (a) GATA-1, PU.1, RUNX1, (b) C/EBPa, JunB, c-Myb, and c-Myc. All measurements are shown relative to the possess the unusual property of inducing differentiation along expression levels of GAPDH. Values are based on three independent both erythroid and megakaryocytic pathways. Furthermore, n experiments each performed in triplicate. Po0.05 compared with HBP1 also enhances lineage-specific differentiation induced nn untransfected K562 cells; Po0.005. by chemical agents such as TPA or hemin. However, although HPB1 induces differentiation of K562 cells along both erythroid 12-fold in high HBP1-overexpressing cells (HBP1/EF1a-1 or 2) and megakaryocytic pathways, it is not simply a non-specific and to a lesser extent in HBP1/CMV clones. Expression of inducer of differentiation, since it does not appear to promote another transcription factor, JunB, which is known for its differentiation of K562 cells toward granulocytes or monocytes. regulation of myeloid differentiation29 was also increased Thus, overexpression of HBP1 in K562 cells results in loss of cell significantly in HBP1-overexpressing cells (Figure 10b). JunB is granularity (determined by side scatter), as well as loss of surface known to negatively regulate cell proliferation, in part by CD45 expression and decrease in expression of the myeloid decreasing cyclin D1 expression.30 After hemin or TPA marker CD13 (data not shown). While the mechanisms by treatment, mRNAs for C/EBPa and JunB were upregulated to a which HBP1 induces bilineage differentiation remain to be dramatic extent (data not shown). Expression levels of the elucidated, Kohmura et al33 recently demonstrated that treat- transcription factors c-Myb31 and c-Myc are known to decline ment of K562 cells with the specific tyrosine kinase inhibitor during myeloid differentiation. The normalized levels of the STI571 (an inhibitor of Abl kinase) induced multilineage mRNA for both of these genes were significantly reduced in differentiation as evidenced by increased expression of mega- HBP1-overexpressing cells both in the absence (Figure 10b) and karyocytic markers CD41a and CD42, the erythroid marker in the presence (data not shown) of inducing agents TPA and glycophorin A, and myeloid markers CD11b and CD13. This hemin. agent appeared to act in part by inducing phosphorylation of p38 mitogen-activated protein kinase (MAPK) and dephosphor- ylation of extracellular signal-related kinase (ERK) in K562 cells. Discussion Xiu et al20 recently demonstrated that MAPK stabilizes HBP1, suggesting that HBP1 expression might be one factor responsible Our studies demonstrate that overexpression of HBP1 in K562 for multilineage differentiation in both systems. cells slows cell growth in standard tissue culture and soft agar, While the molecular mechanisms by which HBP1 induces increases apoptosis, and induces the cells to differentiate along erythroid and megakaryocytic differentiation of K562 cells both erythroid and megakaryocytic pathways. remain to be determined, our data indicate that HBP1 over- Inhibition of cell growth appears to result from arrest of the expression induces significant changes in levels of mRNAs for a cell cycle at the G0/G1 phases, resulting in reduced percentage number of important transcription factors involved in differ- of cells in S phase. This appears to be a general property of entiation of myeloid cells. mRNAs for C/EBPa, JunB, GATA-1, HBP1, since Tevosian et al12 observed a similar effect in fat and and RUNX1 are increased in K562 cells overexpressing HBP1, muscle cells in which HBP1 was overexpressed and Shih et al13 whereas levels of mRNAs for c-Myb, c-Myc, and PU.1 are showed that transgenic mice in which HBP1 is modestly decreased. Most of these changes agree with expectations, overexpressed showed delayed progression of liver cells through based upon previous literature regarding myeloid differentia- G1 to S following partial hepatectomy. We found that inhibition tion. However, an exception is C/EBPa. Radomska et al34 have of growth of K562 cells by HBP1 is accompanied by, and previously shown that overexpression of this factor in K562 presumably mediated, in part, by reduction in mRNA levels for cells induces granulocytic differentiation, whereas our findings specific modulators of the G1-to-S transition, including cyclin indicate that HBP1 increases C/EBPa mRNA levels while

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1967 inducing erythroid and megakaryocytic differentiation and protein that enhances myeloperoxidase (MPO) promoter activity. reducing expression of myeloid markers. Clearly, the overall Leukemia 2001; 15: 601–612. context of the cell and the global changes induced by HBP1 11 Lavender P, Vandel L, Bannister AJ, Kouzarides T. The HMG-box rather than the level of one particular transcription factor transcription factor HBP1 is targeted by the pocket proteins and E1A. Oncogene 1997; 14: 2721–2728. determine the direction of differentiation. 12 Tevosian SG, Shih HH, Mendelson KG, Sheppard KA, Paulson KE, An important aspect of HBP1 is its potential tumor-suppressor Yee AS. HBP1: a HMG box transcriptional repressor that is activity. As mentioned above, both our data and those of targeted by the retinoblastoma family. Genes Dev 1997; 11: Sampson et al21 indicate that HBP1 inhibits G1 progression and 383–396. blocks the Wnt pathway, which is important in cancer. The fact 13 Shih HH, Tevosian SG, Yee AS. Regulation of differentiation by that the HBP1 gene lies at a chromosomal location 7q31.1, HBP1, a target of the retinoblastoma protein. Mol Cell Biol 1998; 10 18: 4732–4743. which is frequently deleted in cancer, provides further 14 Gartel AL, Goufman E, Tevosian SG, Shih H, Yee AS, Tyner AL. suggestive evidence for a tumor-suppressor function of HBP1. Activation and repression of p21(WAF1/CIP1) transcription by RB Recent data that the HBP1 gene is frequently mutated in breast binding proteins. Oncogene 1998; 17: 3463–3469. cancers23 and leukemias35 further support this possibility. Our 15 Zhuma T, Tyrrell R, Sekkali B, Skavdis G, Saveliev A, Tolaini M data showing reduced colony formation in soft agar also et al. Human HMG box transcription factor HBP1: a role in hCD2 supports the possible tumor-suppressor activity of HBP1. LCR function. EMBO J 1999; 18: 6396–6406. Furthermore, the ability of HBP1 to partner with RB and to 16 Lemercier C, Duncliffe K, Boibessot I, Zhang H, Verdel A, Angelov D et al. Involvement of retinoblastoma protein and HBP1 increase the activity of p21 provides additional mechanisms by in histone H1(0) gene expression. Mol Cell Biol 2000; 18: which HBP1 may exert tumor-suppressor activity. 6627–6637. Knowledge of how HBP1 acts to modulate tissue-specific 17 Swanson KA, Knoepfler PS, Huang K, Kang RS, Cowley SM, gene expression, to regulate cell proliferation, and to control Laherty CD et al. HBP1 and Mad1 repressors bind the sin3 terminal myeloid differentiation should provide insights into the corepressor PAH2 domaine with opposite helical orientations. Nat overall control of growth and maturation of normal and Struct Mol Biol 2004; 11: 738–746. 18 Shih HH, Xiu M, Berasi SP, Sampson EM, Leiter A, Paulson KE leukemic myeloid cells, and may ultimately lead to better et al. HMG box transcriptional repressor HBP1 maintains a therapies for leukemias and other proliferative disorders of white proliferation barrier in differentiated liver tissue. Mol Cell Biol blood cells. 2001; 21: 5723–5732. 19 Smith JM, Bowles J, Wilson M, Koopman P. HMG box transcription factor gene HBP1 is expressed in germ cells of the developing Acknowledgements mouse testes. Dev Dyn 2004; 230: 366–370. 20 Xiu M, Kim J, Sampson E, Huang CY, Davis RJ, Paulson KE et al. This work was supported by a VA Merit grant to GEA. The transcriptional repressor HBP1 is a target of the p38 mitogen- activated protein kinase pathway in cell cycle regulation. Mol Cell Biol 2003; 23: 8890–8901. Supplementary Information 21 Sampson EM, Haque ZK, Ku MC, Tevosian SG, Albanese C, Pestell RG et al. Negative regulation of the Wnt-beta-catenin pathway by the transcriptional repressor HBP1. EMBO J 2001; 20: 4500–4511. Supplementary Information accompanies the paper on the 22 Berasi SP, Xiu M, Yee AS, Paulson KE. HBP1 repression of the Leukemia website (http://www.nature.com/leu). p47phox gene: cell cycle regulation via the NADPH oxidase. Mol Cell Biol 2004; 24: 3011–3024. 23 Yee AS, Paulson EK, McDevitt MA, Rieger-Christ K, Summerhayes References I, Berasi SP et al. The HBP1 transcriptional repressor and the p38 MAP kinase: unlikely partners in G1 regulation and tumor 1 Lesage F, Hugnot J-P, Amri E-Z, Grimaldi P, Barhanin J, Lazdunski suppression. Gene 2004; 336: 1–13. M. Expression cloning in K+ transport defective yeast and 24 Bergh G, Ehinger M, Olofsson T, Baldetorp B, Johnsson E, Brycke distribution of HBP1, a new putative HMG transcriptional H et al. Altered expression of the retinoblastoma tumor-suppressor regulator. Nucleic Acids Res 1994; 22: 3695–3688. gene in leukemic cell lines inhibits induction of differentiation but 2 Landsman D, Bustin M. A signature for the HMG-1 box DNA- not G1-accumulation. Blood 1997; 89: 2938–2950. binding proteins. Bioessays 1993; 15: 539–546. 25 Munoz-Alonso MJ, Acosta JC, Richard C, Delgado MD, Sedivy J, 3 Grosschedl R. Higher-order nucleoprotein complexes in transcrip- Leon J. p21Cip1 and p27Kip1 induce distinct cell cycle effects and tion: analogies with site-specific recombination. Curr Opin Cell differentiation programs in myeloid leukemia cells. J Biol Chem Biol 1995; 7: 362–370. 2005; 280: 18120–18129. 4 Tjian R, Maniatis T. Transcriptional activation: a complex puzzle 26 Labbaye C, Valtieri M, Barberi T, Meccia E, Masella B, Pelosi E with few easy pieces. Cell 1994; 77: 5–8. et al. Differential expression and functional role of GATA-2, 5 Wolffe AP. Architectural transcription factors. Science 1994; 264: NF-E2, GATA-1 in normal adult hematopoiesis. J Clin Invest 1995; 1100–1101. 95: 2346–2358. 6 Arie T, Christiansen SK, Yoder OC, Turgeon BG. Efficient cloning 27 Elagib KE, Racke FK, Mogass M, Khetawat R, Delehanty LL, of ascomycete mating type genes by PCR amplification of the Goldfarb AN. RUNX1 and GATA-1 coexpression and cooperation conserved MAI HMG Box. Fungal Genet Biol 1996; 21: 118–130. in megakaryocytic differentiation. Blood 2003; 101: 4333–4341. 7 Sugimoto A, Lino Y, Maeda T, Watanabe Y, Yamamoto M. 28 Keeshan K, Santilli G, Corradini F, Perrotti D, Calabretta B. Schizosaccharomyces pombe ste11+ encodes a transcription Transcription activation function of C/EBPalpha is required factor with an HMG motif that is a critical regulator of sexual for induction of granulocytic differentiation. Blood 2003; 102: development. Genes Dev 1991; 5: 1990–1999. 1267–1275. 8 Van De Wetering M, Oostewegel M, Dooijes D, Clevers H. 29 Yang MY, Liu TC, Chang JG, Lin PM, Lin SF. JunB gene expression Identification and cloning of TCF-1, a T lymphocyte-specific is inactivated by methylation in chronic myeloid leukemia. Blood transcription factor containing a sequence-specific HMG box. 2003; 101: 3205–3211. EMBO J 1991; 10: 123–132. 30 Bakiri L, Lallemand D, Bossy-Wetzel E, Yaniv M. Cell cycle- 9 Alexander-Bridges M, Dugast I, Ercolani L, Giere L, Nasrin N. dependent variations in c-Jun and JunB phosphorylation: a role Multiple insulin-response elements regulate transcription of the in the control of cyclin D1 expression. EMBO J 2000; 19: GAPDH gene. Adv Enzyme Regul 1992; 32: 149–159. 2056–2068. 10 Lin KM, Zhao WG, Bhatnagar J, Zhao WD, Lu JP, Simko S et al. 31 Rosson D, O’Brien TG. Constitutive c-myb expression in K562 Cloning and expression of human HBP1, a high mobility group cells inhibits induced erythroid differentiation but not tetradeca-

Leukemia HBP1 and myeloid cell growth and differentiation CJ Yao et al 1968 noyl phorbol acetate-induced megakaryocytic differentiation. Mol 34 Radomska HS, Huettner CS, Zhang P, Cheng T, Scadden DT, Cell Biol 1995; 15: 772–779. Tenen DG. CCAAT/enhancer binding protein alpha is a regulatory 32 Clement M-V, Hirpara JL, Chawdhury S-H, Pervaiz S. Chemopre- switch sufﬁcient for induction of granulocytic development ventive agent Resveratrol, a natural product derived from grapes, from bipotential myeloid progenitors. Mol Cell Biol 1998; 18: triggers CD95 signaling-dependent apoptosis in human tumor 4301–4314. cells. Blood 1998; 92: 9996–1002. 35 Koike M, Tasaka T, Spira S, Tsuruoka N, Koefﬂer HP. Allelotyping 33 Kohmura K, Miyakawa Y, Kawai Y, Ikeda Y, Kizaki M. Different of acute myelogenous leukemia: loss of heterozygosity at 7q31.1 roles of p38 MAPK and ERK in STI571-induced multi-lineage (D7S486) and q33-34 (D7S498, D7S505). Leuk Res 1999; 23: differentiation of K562 cells. J Cell Physiol 2004; 198: 370–376. 307–310.

Leukemia