Activation of Wnt/β-Catenin Protein Signaling Induces Mitochondria-mediated Apoptosis in Hematopoietic Progenitor Cells*

Background: Wnt/β-catenin signaling plays a strong role in maintaining homeostasis of hematopoietic progenitor cells (HPCs). Results: Activated β-catenin deregulates the survival of HPCs by induction of the mitochondrial apoptotic pathway. Conclusion: Wnt/β-catenin signaling is critical for HPC pool maintenance. Significance: This study clarifies the role of the Wnt/β-catenin signaling in HPCs by showing that it has negative impact on the function and survival of the cells. The canonical Wnt/β-catenin signaling is activated during development, tumorigenesis, and in adult homeostasis, yet its role in maintenance of hematopoietic stem/progenitor cells is not firmly established. Here, we demonstrate that conditional expression of an active form of β-catenin in vivo induces a marked increase in the frequency of apoptosis in hematopoietic stem/progenitor cells (HSCs/HPCs). Activation of Wnt/β-catenin signaling in HPCs in vitro elevates the activity of caspases 3 and 9 and leads to a loss of mitochondrial membrane potential (ΔΨm), indicating that it induces the intrinsic mitochondrial apoptotic pathway. In vivo, expression of activated β-catenin in HPCs is associated with down-regulation of Bcl2 and expression of Casp3. Bone marrow transplantation assays reveal that enhanced cell survival by a Bcl2 transgene re-establishes the reconstitution capacity of HSCs/HPCs that express activated β-catenin. In addition, a Bcl2 transgene prevents exhaustion of these HSCs/HPCs in vivo. Our data suggest that activation of the Wnt/β-catenin pathway contributes to the defective function of HPCs in part by deregulating their survival.

The canonical Wnt/␤-catenin pathway plays an important regulatory role in hematopoiesis; however, its involvement in regulation of HSC 3 function is controversial. Mice with dele-tion of ␤-catenin (1) or ␤-catenin and ␥-catenin (2) had no obvious HSC defects. In contrast, other investigators showed the existence of hematopoietic abnormalities in mice with deletion of ␤-catenin from fetal life (3). Retroviral expression of activated ␤-catenin enhances the self-renewal of Bcl2 transgenic HSCs in vitro, and the Wnt signaling inhibitor Axin represses HSC proliferation and decreases reconstitution capacity of HSCs (4). However, others reported that purified Wnt protein in vitro (5) or inhibition of GSK3␤ in vivo (6) promotes the self-renewal of HSCs. Baba et al. (7,8) showed that constitutive activation of ␤-catenin impaired multilineage differentiation, whereas others found that expression of constitutively active ␤-catenin in hematopoietic cells in vivo leads to loss of HSC repopulation and multilineage differentiation block (9,10).
The role of ␤-catenin in the regulation of cell proliferation and apoptosis is cell context-dependent. ␤-Catenin is an enhancer of proliferation and survival in tumor cells (11,12) but can also induce apoptosis in a variety of cells. Forced expression of an activated of ␤-catenin lacking the NH 2 terminus promoted proliferation and apoptosis in mouse intestinal epithelial cells (13). However, in fibroblasts, the overexpression of ␤-catenin induced apoptosis independently of its transactivation function with LEF1 (14). In enzastaurin-treated multiple myeloma cells, it resulted in cell death-dependent up-regulation of JUN and TP73 (15).
Taken together, these contradictory reports emphasize the difficulty of understanding the role of the Wnt/␤-catenin pathway in HSCs/HPCs. Here, we have revisited this controversy and have examined the effects of activated ␤-catenin in hematopoietic progenitor cells in vitro and in vivo. In vitro our studies reveal that activation of the Wnt/␤-catenin signaling induces a mitochondria-dependent apoptotic pathway. In vivo we find that it promotes apoptosis of HPCs by suppressing Bcl2 and inducing Casp3. In addition, we show that an enhanced survival signal is required for maintenance and function of HSC/HPCs with activated ␤-catenin in vivo. Our studies suggest that the negative effect of Wnt/␤-catenin signaling in HPCs results from the deregulation of their survival.

MATERIALS AND METHODS
Mice-C57BL/6 and B6.SJL mice were purchased from Taconic (Hudson, NY) for breeding in the mouse facility at the University of Illinois at Chicago. The mice (6 -12 weeks) were used for bone marrow (BM) collection. All animal protocols were approved by the Animal Care and Use Committee of the University of Illinois at Chicago. Vav-Bcl2 transgenic mice were kindly provided by Dr. Jerry Adams (The Walter and Eliza Hall Institute of Medical Research, Bundoora, Victoria Australia). ␤-Catenin flox/flox mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Ctnnb1(Ex3) flox/flox mice have been described (16) and were obtained from Dr. Kathleen Goss (University of Chicago). Effective deletion of floxed exons was monitored by PCR. The sequence of the primers is listed in supplemental Table S1.
Quantitative Real-time PCR-To evaluate the expression of candidate genes deregulated in mutant ␤-catenin mice, HPCs were sorted from Mx1-cre Ctnnb1 (Ex3)⌬/ϩ and control mice (n ϭ 3) 4 days after three doses of pI-pC injection. RNA was isolated with TRIzol reagent (Invitrogen) and quantified. Relative abundance of specific candidate gene transcripts normalized to ␤-actin was determined by quantitative RT-PCR by SYBR Green incorporation. The primer sequence for each gene is listed in supplemental Table S1.
Retroviral Vectors and BM Infection-The pBabe-␤-cat-S33Y vector encoding an activated form of ␤-catenin was kindly provided by Dr. Eric Fearon (University of Michigan School of Medicine, Ann Arbor, MI). The BamHI-EcoRI DNA fragment encoding ␤-cat-S33Y was cloned into the MSCV-IRES-EGFP (MIGR1) vector. The Bcl2 cDNA was subcloned into MSCV-EYFP vector. Using Effectene transfection reagent (Qiagen), high titer retrovirus was produced by transient transfection of 293T cells with the pCL ecotropic packaging plasmid and the appropriate retroviral vectors (17). The MSCV-Cre-IRES-EYFP and MSCV-puro-Cre vectors were provided by Dr. Jiwang Zhang (Loyola University, Maywood, IL). The BM cells isolated from mice treated with 5-fluorouracil (150 mg/Kg) for 4 days, the lineage-negative BM cells, and the day 14.5 fetal liver cells were infected with the appropriate retrovirus by spinoculation as described previously (18).
Colony-forming Unit Assays-Isolated BM progenitor cells were infected with MIGR1 or ␤-catenin retrovirus. The EGFPpositive cells, isolated by flow cytometry, were plated in duplicate in methylcellulose medium (MethoCult M3434, StemCell Technologies) supplemented with IL-3, IL-6, erythropoietin and stem cell factor. The number of colonies was scored 10 days later.
Flow Cytometric Analysis-We used single-cell suspensions of BM and PB. Flow cytometric analysis of HSCs/HPCs has been described previously (19). Briefly, the infected BM cells were stained with biotinylated antibodies specific for lineage markers (CD3, B220, IgM, Ter119, Gr-1, Thy1.1, Il-7R, and CD19) followed by PE-Cy5-conjugated streptavidin, Kit-APC-eFluo780, Sca-PE, CD48-PE-Cy7, and CD150-APC staining. Flow cytometry was performed at the University of Illinois at Chicago facility using FACSCalibur or CyAn flow cytometers. Cell cycle analysis with DAPI staining was performed as described previously (20). Cell cycle analysis with Hoechst 3342 was performed as described (4). For detection of apoptosis, BM cells were cultured overnight at a density of 0.2 ϫ 10 6 /ml. The next day, the cells were stained with annexin V (BD Biosciences) and 7-AAD or DAPI following the manufacturer's instructions. Caspase 3 and caspase 9 staining was performed using kits from BD Biosciences and BioVision according to the manufacturer's instructions. All data were analyzed by the FlowJo software (TreeStar).
Transplantation Assays-BM cells and day 14.5 fetal liver cells from C57BL/6 or BM cells from Vav-Bcl2-transgenic mice were infected with MIGR1 or ␤-catenin retrovirus and were transplanted into lethally irradiated (960 radians) B6.SJL mice by retro-orbital injection. After transplantation, the mice were closely monitored. FACS analysis of EGFP-positive cells in PB was performed every month after transplantation.
Mitochondrial Membrane Potential (⌬⌿ m ) Measurement-Tetramethylrhodamine methyl ester perchlorate (TMRM) (Sigma) was used as an indicator of ⌬⌿ m . Live BM cells were washed once with PBS and stained with freshly prepared 10 M TMRM for 30 min at 37°C. The cells were then washed twice with PBS and suspended in PBS for flow cytometric analysis.
Luciferase Reporter Assay-The luciferase assay was performed with the Dual-Glo TM luciferase assay system (Promega).
Statistical Analysis-Statistical significance was calculated using the two-tailed Student's t test (Excel, Microsoft, Redmond, WA).

Activation of ␤-Catenin Induces Apoptosis in Vivo-
To determine whether in vivo activation of ␤-catenin induces apoptosis of HSCs/HPCs, we used the Cre-loxP system to activate ␤-catenin in hematopoietic stem/progenitor cells in vivo.
We generated Mx1-cre ϩ Ctnnb1 (Ex3)fl/ϩ mice by crossing Ctnnb1 (Ex3)fl/fl mice with transgenic mice expressing cre recombinase under control of the type I interferon-inducible Mx1 promoter. In Mx1-cre ϩ Ctnnb1 (Ex3)fl/ϩ mice, the exon 3 of Ctnnb1 is flanked by loxP sites. Induction of exon 3 deletion by injection of interferon-␣ inducer pI-pC leads to expression of activated ␤-catenin. By PCR analysis, Mx1-cre ϩ Ctnnb1 (Ex3)fl/ϩ mice treated with pI-pC had Cre-mediated deletion of exon 3 of Ctnnb1 in the majority of BM cells (Fig. 1A). After 4 days of treatment with three doses of pI-pC every other day, we characterized the HPCs and Lin Ϫ c-Kit ϩ Sca ϩ cells (LSKs, stem-enriched population) in these mice. The frequency (Fig. 1B, left panel) as well as the total number (Fig. 1B, right panel) of HPCs and LSKs were decreased 3-10-fold in BM cells from Mx1-cre Ctnnb1 (Ex3)⌬/ϩ when compared with control mice. The LSKs and HPCs expressing activated ␤-catenin displayed a 3-4-fold increase in apoptosis when compared with control cells (Fig.  1C). Our data suggest that in vivo activation of Wnt/␤-catenin by stabilized ␤-catenin induces apoptosis in HPCs and LSKs, leading to the depletion of these cells.
To gain insight into the molecular mechanism underlying the increased apoptosis of HPCs that express the mutant ␤-catenin, we examined several genes regulating the apoptotic and survival pathways. HPC cells were isolated from the Mx1cre Ctnnb1 (Ex3)⌬/ϩ and control mice 4 days after three doses of pI-pC injection. In sorted Ctnnb1-mutant HPCs, Bcl2 is down-regulated (Ͼ3-fold, Fig. 1D, black bars) whereas Casp3 is upregulated (Ͼ3-fold, Fig. 1D, black bars) when compared with HPCs of control mice (Fig. 1D, gray bars). However, the expression of other genes involved in cell apoptosis and survival was not significantly altered.
Activation of Wnt/␤-Catenin Induces Apoptosis in Hematopoietic Progenitors in Vitro-To determine whether these effects are also seen in primary BM cells expressing the active form of ␤-catenin (␤-cat-S33Y) in vitro, we cloned ␤-cat-S33Y in the retroviral vector MIGR1 and expressed the protein in primary BM cells (supplemental Fig. S1A). As a control for endogenous expression of the protein at a physiological level, we used Apc Ϫ/Ϫ BM cells. Apc is an essential component of ␤-catenin destruction complex, and its loss leads to retention of the protein in the cytoplasm (11). Western blot analyses showed no significant difference between ␤-cat-S33Y-transduced Lin Ϫ cells and Apc Ϫ/Ϫ Lin Ϫ cells, indicating that the retroviral expression of ␤-cat is within a normal range (supplemental Fig. S1B). To confirm that ␤-cat-S33Y activates the Wnt/␤-catenin signaling, we performed luciferase reporter assays. The Wnt signaling was stimulated by this mutant ␤-catenin (supplemental Fig. S1C). In addition, we found that in ␤-cat-S33Y-EGFP ϩ HPCs (Lin Ϫ Kit ϩ ) or Lin Ϫ cells, the frequency of early apoptotic cells ( Fig. 2A) and the level of caspase 3 (Fig. 2B) were increased significantly when compared with MIGR1-transduced HPCs and Lin Ϫ cells. The Wnt/␤-catenin pathway can be activated by the addition of LiCl, which inhibits GSK-3␤, a component of the degradation complex, thereby stabilizing ␤-catenin. In methylcellulosebased medium, treatment of BM cells with LiCl, but not with equimolar amounts of NaCl, repressed colony formation (supplemental Fig. S2A), and the number of early apoptotic BM cells was significantly increased (supplemental Fig. S2B). Similarly, in liquid culture, the early apoptotic Lin Ϫ Kit ϩ progenitor cells (Fig. 3A) and progenitor cells with an elevated caspase 3 activity (Fig. 3B) were increased significantly after LiCl treatment when compared with NaCl treatment. The frequency of apoptosis was increased more significantly at high concentration of LiCl than at low concentration of LiCl, suggesting that LiCl-induced apoptosis is dosage-dependent (Fig. 3A).
Indomethacin is a Cox inhibitor that blocks the Wnt signaling pathway by repressing Ctnnb (22). To confirm that the effects of LiCl are dependent on ␤-catenin, we treated HPCs with LiCl alone or with indomethacin. The results showed that LiCl-induced apoptosis is reduced by indomethacin (supplemental Fig. S3). Next, we isolated BM cells from ␤-catenin fl/fl mice in which exons 2-6 can be excised by Cre expression (23) and confirmed the deletion by PCR (supplemental Fig. S4). After treatment with LiCl, apoptosis in ␤-catenin fl/fl HPCs was comparable with that of control HPCs (Fig. 3C), suggesting that LiCl specifically affects ␤-catenin and that the activation of Wnt/␤-catenin either by expression of active ␤-catenin or by LiCl leads to apoptosis in HPCs.
Wnt/␤-Catenin Signaling Induces Apoptosis in Vitro through the Intrinsic Mitochondrial Apoptotic Pathway, and Apoptosis Is Inhibited by Bcl2-There are two major apoptotic pathways, the mitochondrial "intrinsic" and transmembrane "extrinsic" pathway. To determine whether the mitochondrial apoptotic pathway is activated by Wnt signaling, we examined the activity of caspase 9 in HPCs that expressed ␤-cat-S33Y or that were treated with LiCl. We found that caspase 9 was activated in ␤-catenin-positive HPCs (Fig. 4A, black bar) and that LiCl treatment resulted in a higher activity of caspase 9 than what was observed in the control HPCs (supplemental Fig. S5). The proper function of the mitochondria was tested with TMRM, which detects the mitochondrial membrane potential (⌬⌿ m ). HPCs treated with LiCl and TMRM showed a significant drop in fluorescence with increasing concentration of LiCl (Fig. 4B, black bars), indicating that a loss of ⌬⌿ m occurred in the LiCltreated HPCs. To determine whether expression of Bcl2 protects HPCs from apoptosis, Lin Ϫ BM cells isolated from Mx1cre Ϫ Ctnnb1 (Ex3)fl/fl mice were infected with MSCV-puro and MSCV-EYFP, MSCV-puro-Cre and MSCV-EYFP, or MSCVpuro-Cre and MSCV-EYFP-Bcl2. The excision of exon 3 of Ctnnb1 in Mx1-cre Ϫ Ctnnb1 (Ex3)fl/fl BM cells was induced by Cre recombinase expressed in a retroviral vector. As we expected, Lin Ϫ cells expressing activated ␤-catenin (Fig. 4C, middle bar) displayed a frequency of apoptosis higher than the control cells in vitro. However, Bcl2 could repress apoptosis (Fig. 4C, right bar). To confirm the role of Bcl2 in blocking apoptosis induced by activated ␤-catenin, Lin Ϫ BM cells, isolated from Vav-Bcl2 mice, were infected with the ␤-cat-S33Y or the MIGR1 retrovirus. Flow cytometric analysis of the cells revealed that the frequency of annexin V ϩ DAPI Ϫ cells and the activity of caspase 3 were slightly lower in ␤-cat-S33Y-positive Lin Ϫ Kit ϩ BM cells than in MIGR1-transduced Lin Ϫ Kit ϩ BM cells (Fig. 4, D and E) and that LiCl treatment did not affect apoptosis in HPCs that expressed the Bcl2 transgene (Fig. 4F). Together, our data suggest that Bcl2 can reduce apoptosis induced by activated ␤-catenin.
A Bcl2 Transgene Rescues the Defects of HSCs/HPCs Expressing ␤-cat-S33Y-We hypothesized that apoptosis induced by activated ␤-catenin contributes to the defective function of HSCs/HPCs with active Wnt/␤-catenin signaling. If this is the case, then Bcl2 expression should rescue the function of these HSCs/HPCs. Therefore, we isolated CD45.2 ؉ BM cells from C57BL/6 donors 4 days after treatment with 5-fluorouracil; cells were infected with ␤-cat-S33Y or the empty MIGR1 retrovirus and transplanted into lethally irradiated syngeneic mice (CD45.1 ϩ ). More than 90% of the PB cells from the recipients were donor-derived (CD45.2 ϩ ), indicating a successful reconstitution (data not shown). Five weeks after transplantation, the percentage of EGFP ϩ cells in total BM cells was determined by flow cytometry. Before transplantation (in vitro) and 5 weeks after transplantation, in the control mice, the percentage of MIGR1-transduced BM cells was comparable (Fig. 5A, gray  bars). In contrast, before transplantation, the percentage of ␤-cat-S33Y-EGFP-positive cells was about 14%, but the cells were hardly detected in the recipient mice at 5 weeks after transplantation (Fig. 5A, black bars). To confirm these results, we infected day 14.5 fetal liver cells with ␤-cat-S33Y and MIGR1 retroviruses followed by transplantation. The percentage of MIGR1-transduced fetal liver cells was comparable before or after transplantation, but again, although the percentage of ␤-cat-S33Y-EGFP-transduced fetal liver cells was about 27% before transplantation, no EGFP ϩ cells were detected in the PB of the recipient mice at 5 and 14 weeks after transplantation (Fig. 5B, black bars). These data suggest that retroviral expression of the active form of ␤-catenin disrupts the reconstitution capacity of HSCs/HPCs. To determine whether Bcl2 rescues the defects in the function of HSCs/HPCs expressing activated ␤-catenin, we isolated BM cells from Vav-Bcl2 transgenic mice (24), infected them with the MIGR1 vector expressing either ␤-cat-S33Y-EGFP or EGFP, and transplanted the cells into lethally irradiated syngeneic recipients. As expected, the percentage of control cells at time of transplantation and at 7-8 weeks after transplantation did not decrease (Fig. 5C, gray  bars). However, in contrast to the previous results, the percentage of ␤-cat-S33Y-Bcl2-EGFP ϩ cells was also comparable in vitro before transplantation and in total PB and BM cells at 7-8 weeks after transplantation (Fig. 5C, black bars). We characterized the GFP-positive HSC and HPC compartments in both groups of recipients. The number of LSK cells and CD48 Ϫ CD150 ϩ LSK cells was slightly lower in the GFP ϩ BM cells of Bcl2 transgenic/␤-cat-S33Y mice than in the control animals (Fig. 6A), suggesting that the expression of Bcl2 partially rescued the reconstitution capacity of shortterm HSCs and HPCs expressing activated ␤-catenin. Because others reported that in vivo expression of ␤-cat-S33Y leads to BM failure within 2-3 weeks (9), we monitored a cohort of chimeric mice expressing both activated ␤-catenin and Bcl2 transgene. The EGFP ϩ cells in PB were stabilized at 6 months and were almost doubled at 10 months after transplantation (Fig. 6B), indicating that the expression of Bcl2 rescues and may enhance the repopulating capacity of long-term HSCs expressing activated ␤-catenin. Of note, the apoptosis of Bcl2 transgenic LSKs expressing activated ␤-catenin was slightly higher than the LSKs expressing control vector, whereas the Bcl2 transgenic HPCs expressing activated ␤-catenin or control vector displayed a comparable frequency of apoptosis (Fig. 6C). These data suggest that Bcl2 can repair the hematopoietic defects of HSCs/HPCs that express activated ␤-catenin.

DISCUSSION
Whether ␤-catenin acts as a proapoptotic factor in hematopoietic cells has not been conclusively defined. Previous studies revealed that expression of activated ␤-catenin did not alter the frequency of apoptosis in HSC population (9, 10). However, Li and colleagues (25) showed that activated ␤-catenin enhances apoptosis of HSCs in vitro. The goal of this study was to help to solve, if possible, the existing discrepancies on ␤-catenin in HSCs/HPCs. Our results include strong data, obtained in vitro and in vivo, which support the apoptotic effect that this protein exerts on HPCs. In addition, our results show the molecular mechanisms by which this occurs and indicate that regulation  of survival by ␤-catenin is critical for the maintenance of functional HSCs/HPCs. By using an Mx1-cre/loxP system to induce the deletion of exon 3 of Ctnnb1 gene, leading to activated ␤-catenin in vivo, we showed that a significant decrease in the number of HPCs and LSKs and a markedly increased frequency of apoptosis occur in these cells. Consistently, Cre-dependent activation of ␤-catenin in Ctnnb1 (Ex3)fl/fl HPCs enhances apoptosis in vitro. The result was confirmed by retroviral expression of activated ␤-cat-S33Y in HPCs. However, the frequency of apoptosis is induced by activated ␤-catenin more significantly in vivo than it is in vitro. In our opinion, it is likely that the in vivo microenvironment could contribute to the survival of HPCs. Bcl2 is a survival factor, and ectopic expression of Bcl2 prevents apoptosis in HSCs (26). Caspase 3 plays a central role in the execution of cell apoptosis (27). Down-regulation of Bcl2 and induced expression of Casp3 occur in Ctnnb1 mutant HPCs in vivo. The expression of p53 is not altered in the mutant HPCs (Fig. 1D). Thus, increased apoptosis in Ctnnb1 mutant HPCs may be a consequence of the combined deregulation of Bcl2, Casp3, and perhaps other genes involved in apoptosis and survival of HPCs, which are still unidentified and which could be regulated directly or indirectly by ␤-catenin. Our in vitro study suggests that activation of Wnt/␤-catenin signaling activates caspase 3 and caspase 9 and reduces the mitochondrial membrane potential, leading to a mitochondria-mediated apoptosis. We previously showed that loss of Apc induces apoptosis in HSCs/HPCs (19). Apc is a multifunction, cell context-dependent protein (28), and it is still unclear whether ␤-catenin is a critical function mediator of Apc. Our data suggest that ␤-catenin is likely to mediate, at least in part, the function of Apc in regulating the survival of HPCs.
Although previous studies suggested that the retroviral expression of active ␤-catenin in Bcl2 transgenic HSCs enhanced their self-renewal, there was no indication on the potential pathways utilized by ␤-cat-S33Y (4). It was therefore unclear whether the effects of ␤-catenin on self-renewal depended on Bcl2 expression, and the mechanism by which Bcl2 cooperates with ␤-catenin in maintenance of HSCs/HPCs remained not well defined. In this study, we show that retroviral expression of active ␤-catenin disrupts the reconstitution capacity of HSCs/HPCs but that Bcl2 transgenic HSCs/HPCs expressing activated ␤-catenin maintain their reconstitution capacity. We further show that a Bcl2 transgene inhibits the apoptosis induced by activated ␤-catenin in vitro and prevents the exhaustion of HSCs/HPCs in vivo. We believe that taken together, our studies clarify the existing contradictory data on the effects of activated ␤-catenin in HSCs/HPCs not only by supporting the notion that activation of Wnt/␤-catenin negatively impacts the function of HSCs/HPCs, but more importantly, by strongly indicating that an enhanced survival signal is required for the survival of these cells. It is very likely that regulation of the function of HSCs by Wnt/␤-catenin is dosage-dependent (29). Our study suggests that the negative impact of Wnt/␤-catenin in HSCs/HPCs is due, at least in part, to an increased apoptosis induced by high dosage of Wnt signaling.
Finally, the activation of Wnt/␤-catenin signaling and the overexpression of Bcl2 have been frequently associated with acute myeloid leukemia (30 -34). Our data suggest that Wnt/␤catenin may cooperate with survival signaling in promoting the self-renewal or proliferation of leukemia stem/progenitor cells in some acute myelogenous leukemia patients.