Hematopoietic and Leukemic Stem Cells Have Distinct Dependence on Tcf1 and Lef1 Transcription Factors*

Hematopoietic and leukemic stem cells (HSCs and LSCs) have self-renewal ability to maintain normal hematopoiesis and leukemia propagation, respectively. Tcf1 and Lef1 transcription factors are expressed in HSCs, and targeting both factors modestly expanded the size of the HSC pool due to diminished HSC quiescence. Functional defects of Tcf1/Lef1-deficient HSCs in multi-lineage blood reconstitution was only evident under competitive conditions or when subjected to repeated regenerative stress. These are mechanistically due to direct positive regulation of Egr and Tcf3 by Tcf1 and Lef1, and significantly, forced expression of Egr1 in Tcf1/Lef1-deficient HSCs restored HSC quiescence. In a preclinical CML model, loss of Tcf1/Lef1 did not show strong impact on leukemia initiation and progression. However, when transplanted into secondary recipients, Tcf1/Lef1-deficient LSCs failed to propagate CML. By induced deletion of Tcf1 and Lef1 in pre-established CML, we further demonstrated an intrinsic requirement for these factors in LSC self-renewal. When combined with imatinib therapy, genetic targeting of Tcf1 and Lef1 potently diminished LSCs and conferred better protection to the CML recipients. LSCs are therefore more sensitive to loss of Tcf1 and Lef1 than HSCs in their self-renewal capacity. The differential requirements in HSCs and LSCs thus identify Tcf1 and Lef1 transcription factors as novel therapeutic targets in treating hematological malignancies, and inhibition of Tcf1/Lef1-regulated transcriptional programs may thus provide a therapeutic window to eliminate LSCs with minimal side effect on normal HSC functions.

Hematopoietic stem cells (HSCs) 5 represent a rare population that self-renew and continuously replenish all blood cells throughout one's lifetime (1,2). HSCs are regulated by both intrinsic factors and extrinsic signals from the microenvironment to maintain a homeostatic HSC pool at steady state as well as in response to injuries such as radiotherapy or traumatic blood loss (3)(4)(5). Leukemic stem cells (LSCs) share similar properties with HSCs, and the self-renewing LSCs are responsible for initiation, maintenance, and propagation of the disease (6,7). Chronic myelogenous leukemia (CML) is initiated by reciprocal translocation of chromosomes 9 and 22, which generates a constitutively active fusion kinase, BCR-ABL (8). BCR-ABL tyrosine kinase inhibitors (TKIs) including imatinib are effective in inducing remissions and improving survival in CML patients at the chronic phase. Because CML LSCs are not sensitive to imatinib, continuous TKI treatment is necessary to prevent relapse. Molecules and/or pathways that are specifically utilized by the LSCs are of great therapeutic value for eradication of leukemias.
A plethora of data has implicated Wnt signaling pathway in regulation of HSC activities (9,10). Overexpression of Dkk1 or Wif1, which blocks the interaction between Wnt ligands and their receptors, diminishes HSC quiescence and its repopulation capacity (11,12). Wnt3a deficiency also greatly impaired HSC activity (13). Because Wnt signaling is involved in proper bone formation (14,15), blocking Wnt/receptor interaction or germline deletion of Wnt proteins may affect the HSC niche, hence indirectly impacting HSCs. Activation of the canonical Wnt signaling pathway leads to stabilization and nuclear translocation of ␤-catenin. The role of ␤-catenin in HSCs has been a highly contentious issue. Depending on the experimental systems utilized, ␤-catenin activation is reported to have detrimental (16,17), beneficial (18), or no effect on adult HSCs (19,20). A recent study reports that the magnitude of ␤-catenin activation matters, with a narrow window of active ␤-catenin positively regulating HSC repopulation capacity (21). With regard to necessity of ␤-catenin, it is essential for definitive hematopoiesis (by embryonic day 10.5) (22), but its requirement in adult HSCs is only evident after serial transplantation (23). Despite the reported discrepancies on its roles in normal HSCs, ␤-catenin has been consistently demonstrated to be critical for development and maintenance of LSCs in CML (23)(24)(25).
Although ␤-catenin itself does not have the ability to bind DNA after translocation into the nucleus, it interacts with Tcf/ Lef transcription factors to modulate gene expression (9). Tcf/ Lef factors contain 4 members, Tcf1, Tcf3, Tcf4, and Lef1, and all have a highly conserved high mobility group (HMG) DNA binding domain. Whereas the requirements for Tcf1 and Lef1 in blood cells, in particular T lymphocytes, have been well documented (9,26), none of the Tcf/Lef factors has been studied in HSCs. By focusing on Tcf1 and Lef1 in the study, we demonstrate that these two factors regulate the regenerative fitness of HSCs and self-renewal of LSCs in a pre-clinical model of CML.

EXPERIMENTAL PROCEDURES
Animals-Tcf7 Ϫ/Ϫ , Tcf7 fl/fl , and Lef1 fl/fl mice were described (27)(28)(29). Vav-Cre, C57BL/6 and B6.SJL mice were from the Jackson Laboratory, and Cre-ER T2 mice were kindly provided by Dr. Yingzi Zhang (30). All mouse experiments were performed under protocols approved by the Institutional Animal Use and Care Committee of the University of Iowa.
Flow Cytometry, Analysis of Apoptosis, Cell Cycle, and BrdU Incorporation-Single cell suspension from the bone marrow (BM) was prepared and surface-stained as previously described (31,32). Apoptotic cells were detected with the PE AnnexinV Apoptosis Detection Kit (BD Bioscience), and cell cycle analysis was performed by intracellular staining for Ki67 and Hoechst 33342 as described (31,32). For BrdU labeling, the mice were i.p. injected with 1 mg of bromo-2Ј-deoxyuridine (BrdU) and fed in drinking water (1 mg/ml) for 20 h. Cells were surfacestained, fixed, and permeabilized, and intracellularly stained with an anti-BrdU antibody using the APC BrdU Flow Kit (BD Biosciences). The stained cells were analyzed on FACSCalibur, FACSVerse, or Becton Dickinson LSR II or LSR Violet flow cytometers (BD Biosciences). The data were processed using FlowJo software (Version IX or X, TreeStar).
Cell Sorting and Quantitative RT-PCR-For quantification of Tcf7 and Lef1 transcripts in hematopoietic stem and progenitor cells, Flt3 Ϫ LSK, Flt3 ϩ LSK, and Lin Ϫ c-Kit ϩ cells were sorted on FACSAria, and the total RNA was extracted from sorted cells and reverse-transcribed as described (28,32). Plasmids containing Tcf7, Lef1, or Hprt coding sequence were used to generate standard curves for each transcript, and Tcf7, Lef1, or Hprt transcript in the sorted cells was determined by quantitative PCR. The copy numbers of Tcf7 and Lef1 were then calculated assuming that Hprt is expressed at 10,000 copies. The primers for Tcf7 are 5Ј-cccttcctgcggatatagac and 5Ј-ggtacaccagatcccagcat, those for Lef1 are 5Ј-tgagtgcacgctaaaggaga and 5Ј-ctgaccagcctggataaagc, and those for Hprt are 5Ј-gcgtcgtgattagcgatgatg, and 5Ј-ctcgagcaagtctttcagtcc.
Isolation of Lin Ϫ c-Kit ϩ BM Cells and Chromatin Immunoprecipitation (ChIP)-Total BM cells were isolated from WT C57BL/6 mice and depleted of lineage-positive cells. The Lin Ϫ BM cells were then incubated with mouse CD117 Microbeads (Miltenyi Biotec) for positive selection of c-Kit ϩ cells. The Lin Ϫ c-Kit ϩ BM cells were cross-linked for 10 min with 1% formaldehyde in medium, then processed using a truChIP Chromatin Shearing Reagent Kit (Covaris) and sonicated for 5 min on Covaris S2 ultrasonicator. The sheared chromatin fragments, in the range of 200 -500 bp, were immunoprecipitated with an anti-Lef1 (C12A5, Cell Signaling Technologies) or control rabbit IgG and was washed as described (29). The immunoprecipitated DNA segments were used for quantification by PCR. For calculation of enrichment of Lef1 binding, each anti-Lef1 ChIP sample was first normalized to corresponding IgG ChIP sample, and the signal at a target region was then normalized to that at the Hprt promoter region. The primers for detecting the Gapdh and Lef1 genomic regions were described (29), and those for Egr1, 5Ј-cctgctcagttctccctcac and 5Ј-tgccctctgtaggtctctgg; those for Tcf3, 5Ј-gggctctgggtgttatgtgt and 5Ј-ccatcgtttcctagcctctc; and those for Cdkn1c, 5Ј-aagagggactcaggtgagca and 5Ј-agagaactgcgcaggagaac. Note that one Tcf/ Lef consensus motif "CAAAG" is found within the PCR amplicon of the Tcf3 locus. Although no Tcf/Lef motif is within the PCR amplicon of the Egr1 locus, one motif is found at 52 bp upstream, two motifs are found at 55 bp and 216 bp downstream of the amplicon.
Bone Marrow Transplantation (BMT)-For competitive repopulation assays, the test BM cells were collected from CD45.2 ϩ gene-targeted mice, and the competitor BM cells were from CD45.1 ϩ B6.SJL mice. Both test and competitor BM cells were measured for LSK frequency, and the whole BM cells were mixed at 1:1 LSK ratio (each containing 3,000 LSKs) and then transplanted into lethally irradiated (1,050 rad) CD45.1 ϩ CD45.2 ϩ recipients. Sixteen weeks after the transplantation, contribution of test and competitor BM cells to different blood lineages was determined in peripheral nucleated blood cells (PBCs) by flow cytometry.
For serial transplantation assays, total BM cells (2 ϫ 10 6 ) from CD45.2 ϩ gene-targeted mice were transplanted into lethally irradiated CD45.1 ϩ primary recipients. Eight weeks later, BM cells from the primary recipients were transplanted into CD45.1 ϩ secondary recipients, and this process was repeated to obtain tertiary recipients. Sixteen weeks after each transplantation, PBCs were analyzed for the contribution from the original donors to different blood lineages.
Tamoxifen Treatment-For induced deletion of Tcf7 FL/FL Lef1 FL/FL alleles using CreER T2 in the CML recipients, the mice were treated with tamoxifen at 0.2 mg/g body weight via oral gavage for 4 consecutive days. MAY 20, 2016 • VOLUME 291 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 11149
CML Model and Treatment with Imatinib-The pMIG-p210 BCR-ABL retroviral construct was provided by Dr. Warren Pear (34), and the retrovirus was packaged and used for infection of Lin Ϫ BM cells as previously described (32). Twenty-four hours after two rounds of spinofection, the infected Lin Ϫ BM cells containing 3,000 -6,000 GFP ϩ LSK cells along with 2 ϫ 10 5 protector BM cells were transplanted into irradiated B6.SJL recipients. CML progression in the recipients was monitored through regular evaluation of lethargy, weight loss, splenomegaly, and signs of morbidity or detection of GFP ϩ Mac1 ϩ leukemic cells in the PBCs. For treatment of CML with TKI, imatinib (Novartis) was obtained from the Pharmacy of University of Iowa Healthcare and Clinics and administered to CML recipients at 100 mg/kg body weight by oral gavage twice a day.
For secondary transplantation of LSCs, more than 5 primary recipients were sacrificed on day 15 post-BMT to harvest BM cells and sorted for GFP ϩ LSK cells. 10,000 -15,000 sorted LSCs were then transplanted into another cohort of CD45.1 ϩ synge-neic mice along with 2 ϫ 10 5 protector BM cells for CML propagation.
Genetic Complementation with Forced Expression of Egr1-The control and Egr1-pLZRS retroviral vectors were provided by Dr. David Wiest (33), and the retroviruses were packaged to infect Lin Ϫ BM cells as above. The infected cells were transplanted into irradiated B6.SJL mice at 5 ϫ 10 5 per recipient without additional protector BM cells. Six weeks later, CD45.2 ϩ GFP ϩ Lin Ϫ BM cells were sorted from the recipients for cell cycle analysis as described above. The sorting isolation of GFP ϩ cells was used to avoid quenching of GFP fluorescence during intracellular staining for Ki67.
Statistical Analysis-Unpaired comparison among different genetic groups was performed using the Student's t test with a two-tailed distribution assuming equal sample variance. For survival of CML recipients, statistical significance between different genetic and/or treatment conditions was assessed using log-rank test using Prism6 software.

Ablation of Tcf1 and Lef1 Causes Moderate Expansion of the HSC Pool-Tcf1
and Lef1 are abundantly expressed in T-lineage cells, and are readily detectable in Flt3 Ϫ lineagenegative Sca1 ϩ c-Kit ϩ cells (Flt3 Ϫ LSKs), which contain both long-term (LT) and short-term (ST) HSCs (Fig. 1A). Tcf1 expression (encoded by Tcf7) was sharply reduced upon differentiation of HSCs to Flt3 ϩ LSKs, the multipotent progenitors (MPPs). Both factors were further down-regulated in more differentiated Lin Ϫ c-Kit ϩ myeloid progenitors (Fig. 1A). These expression changes suggest a potential requirement for Tcf1 and Lef1 in HSCs.

Loss of Tcf1 and Lef1 Diminishes HSC Function under Repeated Regenerative Stress and Competitive Condition-
HSCs have self-renewal capacity to regenerate themselves under homeostatic conditions or in response to injuries. We used serial transplantation assays, the gold standard for measuring HSC self-renewal. Utilizing disparate CD45 allele markers, we transplanted the same numbers of CD45.2 ϩ BM cells from Lef1 Ϫ/Ϫ , Tcf7 Ϫ/Ϫ , Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ mice or their littermate controls into CD45.1 ϩ recipients. We then determined the contribution of donor BMs to blood reconstitution in peripheral blood mononuclear cells (PBCs) at 16 weeks after the BM transplantation (BMT), because this time point allows exclusive measurement of long-term blood reconstitution capacity of donor HSCs. Consistent with previous reports that Tcf1-deficient BM cells fail to generate T cells in the BM chimeras (35,36), there were few CD45.2 ϩ CD3 ϩ T cells produced from Tcf7 Ϫ/Ϫ or Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ BM cells (not shown). To avoid potential bias resulting from loss of T cells in the absence of Tcf1, we analyzed the non-T compartment in PBCs by focusing on CD3 Ϫ cells. After excluding T cells in the PBCs, we found that BM cells from all genotypes were able to restore CD3 Ϫ blood lineages similarly in these primary recipients ( Fig. 2A). This result also indicates that loss of Tcf1 and/or Lef1 does not affect HSC homing.
We then isolated BM cells from the primary recipients followed by transplantation into secondary CD45.1 ϩ recipients, and repeated this process to generate tertiary recipients. Whereas BM cells from all genotypes remained capable of FIGURE 2. Ablating Tcf1 and Lef1 moderately compromises HSC functions under regenerative stress or competitive conditions. A, serial transplantation assay. Total BM cells (2 ϫ 10 6 ) from test animals (CD45.2 ϩ ) were transplanted into CD45.1 ϩ primary recipients. Eight weeks later, BM cells from the primary recipients were transplanted into CD45.1 ϩ secondary recipients, and this process was repeated to obtain tertiary recipients. The PBCs were analyzed at Ն 16 weeks after each transplantation to determine the contribution from the original donors. Data on the contribution to CD3 Ϫ non-T cells are shown to avoid T lineage-derived bias. Data were pooled from two experiments, with 4 -9 recipients analyzed at each transplantation stage. B and C, competitive repopulation assay. Test (CD45.2 ϩ ) and competitor (CD45.1 ϩ ) BM cells, each containing 3,000 LSKs, were mixed and transplanted into CD45.1 ϩ CD45.2 ϩ recipients. After 16 weeks, their contribution to CD3 Ϫ non-T cells was measured in PBCs. Representative contour plots (B) and cumulative data (C) from three independent experiments are shown, with 4 -12 recipients analyzed. **, p Ͻ 0.01; ***, p Ͻ 0.001 compared with control (Ctrl) mice. MAY 20, 2016 • VOLUME 291 • NUMBER 21 reconstituting all non-T blood lineages in the secondary recipients ( Fig. 2A), in the tertiary recipients, Lef1 Ϫ/Ϫ BM cells exhibited moderately reduced contribution, and loss of both Tcf1 and Lef1 significantly impaired blood reconstitution ( Fig.  2A). These results demonstrate that Tcf1 and Lef1 contribute to maintaining the regenerative fitness of HSCs.

Tcf1 and Lef1 Regulate HSC and LSC Self-renewal
As shown above, HSCs lacking Tcf1 and/or Lef1 were competent to reconstitute non-T blood lineages in both the primary and secondary recipients in the absence of competitors. We next investigated their competitive repopulation capacity. We obtained BM cells from the CD45.2 ϩ gene-targeted strains (as test HSCs) and those from CD45.1 ϩ WT mice (as competitor

Tcf1 and Lef1 Regulate HSC and LSC Self-renewal
HSCs) and determined the LSK frequency. Because the frequency of LSK cells was elevated in Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ BM cells ( Fig. 1 B), for a fair comparison among the test HSCs, we transplanted the same numbers of LSKs into lethally irradiated CD45.1 ϩ CD45.2 ϩ recipients, instead of using the same numbers of total BM cells. After 16 weeks, we assessed the relative contribution of test HSCs to non-T cell blood lineages in PBCs. The control test BM cells moderately outcompeted the competitor BM cells, constituting ϳ65% of the non-T blood cells (Fig. 2, B and C). Whereas Tcf7 Ϫ/Ϫ BM cells were similarly competent in reconstitution non-T blood cells, the contribution by Lef1 Ϫ/Ϫ BM cells was moderately decreased. On the other hand, the contribution by Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ BM cells was reduced to ϳ50% of that by control BM cells (Fig. 2, B and C). These observations indicate that Tcf1 and Lef1 have redundant roles in regulating HSC activities, similar to their functions in T lineage cells (26). These analyses collectively identified a requirement for Tcf1 and Lef1 in maintaining functional HSCs, especially under competitive conditions or repetitive regenerative stress.
Tcf1 and Lef1 Contribute to Maintaining HSC Quiescence-To elucidate the mechanisms by which Tcf1 and Lef1 regulate the HSC biological activities, we first measured its cell cycle status. By combination of Ki67 and Hoechst33342, which measures cellular proliferation and DNA content, respectively, we found that majority of WT CD34 Ϫ Flt3 Ϫ LT-HSCs are in Hoechst lo Ki67 Ϫ G0 phase, with a small portion in Hoechstlo Ki67 ϩ G1 phase and very few in Hoechst med-hi Ki67 ϩ S/G2/M phase (Fig. 3A). Although loss of Tcf1 or Lef1 alone did not have a discernible impact, Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ LT-HSCs showed modest but consistent reduction of the proportion in G0 phase, with corresponding increase of that in G1 phase (Fig. 3, A and B).  MAY 20, 2016 • VOLUME 291 • NUMBER 21
Tcf1 and Lef1 Act Upstream of Egr1 and Other HSC Regulators-We next aimed to gain molecular insights into the regulation of HSC quiescence and function by Tcf1 and Lef1. A number of factors have been identified to be responsible for maintaining HSC quiescence, and genetic ablation of these factors leads to moderate expansion of HSCs accompanied by functional defects due to exhaustion (5). These factors include transcription factors such as Egr1 (37) (45); and cyclin-dependent kinase inhibitors such as p27-Kip1 and p57-Kip2 (encoded by Cdkn1b and Cdkn1c, respectively) (46,47). We sort-purified Flt3 Ϫ LSK cells from Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ and control mice and scanned for the expression changes of these known HSC regulators. The expression of Egr1, Tcf3, and Cdkn1c was substantially diminished in Tcf1/Lef1-deficient hematopoietic stem/progenitor cells, whereas other genes were not affected or moderately reduced in expression (Fig. 4A).
We then investigated if Tcf1 and Lef1 directly regulate Tcf1/ Lef1-dependent genes. To this end, we searched the proximal regulatory regions, i.e. Ϫ1.5 kb to ϩ0.5 kb sequence flanking the transcription start site (TSS) of genes of interest for consensus Tcf/Lef binding motif "CAAAG." We found three motifs at Ϫ1,378, Ϫ1,124, and Ϫ963 bp upstream of the Egr1 TSS, one at Ϫ180 bp upstream of the Tcf3 TSS, and one at Ϫ310 bp upstream of the Cdkn1c TSS. We then performed chromatin

Tcf1 and Lef1 Regulate HSC and LSC Self-renewal
immunoprecipitation (ChIP) on primary murine Lin Ϫ c-Kit ϩ BM progenitor cells using an anti-Lef1 antibody. The progenitor cells were used instead of primary HSCs because HSCs constitute only a rare population in mouse bone marrow but ChIP assay for transcription factor binding usually requires relatively large numbers of purified cells to obtain reliable results. The use of Lef1 antibody is because Lef1 protein was more readily detectable than Tcf1 in Lin Ϫ c-Kit ϩ BM cells (not shown). Lef1 itself is a direct target gene of the Tcf/Lef factors (28) and used as a positive control. We found that Lef1 exhibited enriched binding at the 5Ј-regulatory regions of Egr1 and Tcf3 but not that of Cdkn1c (Fig. 4B). These data suggest a direct regulatory role of Lef1, and likely Tcf1 as well, in the expression of key HSC molecules in the hematopoietic progenitor cells.
The phenotypic and functional HSC impairments are strikingly similar between Egr1 Ϫ/Ϫ and Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ mice, including expansion of the HSC pool and loss of HSC quiescence (37). We next investigated if genetic complementation of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ cells with Egr1 restores HSC quiescence (33). We used retroviral transduction to achieve forced expression of Egr1 in control or Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ Lin Ϫ BM cells and then transplanted into CD45.1 ϩ hosts to make BM chimeras. The bicistronic pLZRS retroviral vector uses GFP as an expression indicator, and we focused our cell-cycle analysis on sorted CD45.2 ϩ GFP ϩ Flt3 Ϫ LSK cells in the BM chimeras. In the empty pLZRS vector-transduced group, Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ HSCs continued to exhibit reduced portion in G0 phase with a moderate increase in S/G2/M phases (Fig. 4, C and D). Whereas forced expression of Egr1 did not detectably alter the cell cycle status of control HSCs, Egr1-complemented Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ HSCs showed increase in quiescent G0 phase and substantial decrease in S/G2/M phases (Fig. 4, C and D). In steady state, Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ HSCs showed an increase in G1 phase compared with control HSCs (Fig. 3, A and B); however, the difference in G1 phase was not evident between Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ and control HSCs in the BM chimeras, whether transduced with the pLZRS vector or Egr1-expressing retroviruses (Fig. 4, C and D). FIGURE 6. Tcf1 and Lef1 are intrinsically required for LSC self-renewal. A, experimental design. Lin Ϫ BM cells from WT or CreER T2 -Tcf7 FL/FL Lef1 FL/FL mice were used to establish CML in the primary recipients, which were untreated or treated with Tamoxifen during days 12-15 post-BMT. One day later, BM cells from the primary recipients were lineage-depleted, and the Lin Ϫ BM cells containing 12,000 -16,000 of CD45.2 ϩ GFP ϩ Sca1 ϩ cells were transplanted into CD45.1 ϩ secondary recipients along with 2 ϫ 10 5 protector BM cells. B, leukemia burden in the secondary recipients, as determined in PBCs on day 25 post-BMT. Cumulative data from two independent experiments are shown. ns, not statistically significant; ***, p Ͻ 0.001 compared with WT recipient mice. C, Kaplan-Meier survival curves of the secondary recipients, pooled from two experiments. MAY 20, 2016 • VOLUME 291 • NUMBER 21

Tcf1 and Lef1 Regulate HSC and LSC Self-renewal
A likely explanation is that when the BM chimeras were analyzed at 6 weeks after HSC transplantation, similar amounts of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ and control HSCs remained in a cycling-ready state during the process of active reconstitution of hematopoiesis. Collectively, these data support the notion that Tcf1 and Lef1 act, at least in part, through transcriptional activation of Egr1 to maintain HSC quiescence.
Targeting Tcf1 and Lef1 Impairs Self-renewal of LSCs-We next investigated a requirement for Tcf1 and Lef1 in LSCs. Because a moderate impact on HSC functionalities was only evident in the absence of both factors, we focused on Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ cells. We used a preclinical CML model, where we infected Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ or control BM cells with a bicistronic retrovirus expressing the p210 BCR-ABL fusion protein along with GFP, followed by transplantation into irradiated CD45.1 ϩ congenic recipients (Fig. 5A) (34). Forced expression of p210 BCR-ABL resulted in malignant transformation of hematopoietic stem/progenitor cells and accumulation of GFP ϩ Mac1 ϩ myelogenous leukemic cells. The leukemic cells were readily detectable in the PBCs, and all the recipients succumbed to the disease by day 40 post-BM transplantation (post-BMT) (Fig. 5, B and C). Accumulation of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ GFP ϩ Mac1 ϩ leukemic cells was not significantly different from that of control leukemic cells during the early stages of disease development (days 10 and 14 post-BMT), albeit it was lower than control on day 18 post-BMT (Fig. 5B). The survival of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ CML recipients was marginally extended compared with that of control recipients (Fig. 5C). These data suggest that Tcf1/Lef1 deficiency only minimally affected the onset and progression of CML, mirroring their role in HSCs at the steady state.
We next sort-purified GFP ϩ LSKs from the BM cells of primary recipients as LSCs and transplanted the same numbers of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ or control LSCs into secondary recipients (Fig.  5D), which is the most stringent test for LSC self-renewal. The control LSCs were able to propagate CML in the secondary hosts, with leukemic cells readily detected in PBCs on day 14 post-BMT; in contrast, Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ LSCs gave rise to substantially lower numbers of leukemic cells (Fig. 5E). Consistent with the difference in leukemia burden, all the secondary recipients of control LSCs succumbed to the disease by day 60 post-BMT, but all those of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ LSCs survived beyond day 80 post-BMT (Fig. 5F). We demonstrated above that the repopulation capacity of Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ HSCs was only evidently diminished in the tertiary recipients. These data in CML suggest that LSCs are more dependent on Tcf1 and Lef1 for self-renewal than HSCs.
Tcf1 and Lef1 Are Intrinsically Required for LSC Self-renewal-Because Tcf7 was targeted in germline in the Tcf7 Ϫ/Ϫ model, the Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ BM cells may have been exposed to Tcf1-deficient niche environment or have other intrinsic defects, leading to unwanted changes in Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ LSCs. To address these potential caveats, we took advantage of a recently established Tcf7 FL/FL strain (29) and generated Tcf7 FL/FL Lef1 FL/FL mice. To achieve inducible inactivation of both floxed genes, we used a CreER T2 strain in which the fusion protein of Cre recombinase and estrogen receptor (ER) was driven by the ubiquitously expressed Rosa26 locus (30). In the resulting CreER T2 Tcf7 FL/FL Lef1 FL/FL mice, treatment with tamoxifen for 4 consecutive days effectively ablated both Tcf7 and Lef1 genes in hematopoietic cells as determined in splenic T cells (not shown). We used BM cells from CreER T2 Tcf7 FL/FL Lef1 FL/FL mice to establish CML so that the LSCs were generated without pre-existing conditions. We left one cohort of primary recipients untreated, and treated another cohort with tamoxifen to induce Tcf7 and Lef1 gene deletion, followed by transplantation of Lin Ϫ BM cells containing the same numbers of GFP ϩ Lin Ϫ Sca1 ϩ cells into secondary recipients (Fig. 6A). This approach of using the GFP ϩ Lin Ϫ Sca1 ϩ phenotype to enumerate LSCs has been shown to have similar capacity as using sorted GFP ϩ LSKs to propagate CML in secondary or tertiary recipients (25), with the advantages of obtaining sufficient LSCs from fewer donor mice and avoiding additional stress on LSCs due to cell sorting. In parallel, we used Tcf1/Lef1-sufficient WT BM cells to establish CML coupled with tamoxifen treatment as another control group. LSCs from tamoxifen-treated WT mice and those from untreated CreER T2 Tcf7 FL/FL Lef1 FL/FL mice had similar capacity to generate leukemic cells and led to death of the secondary recipients by day 50 post-BMT (Fig. 6, B and C). This observation indicates that tamoxifen treatment per se had little impact on CML propagation. In contrast, very few leukemic cells were generated from LSCs derived from tamoxifen-treated CreER T2 Tcf7 FL/FL Lef1 FL/FL mice, and all the secondary recipients of these LSCs survived beyond day 80 post-BMT. These data demonstrate an intrinsic requirement for Tcf1 and Lef1 in maintaining self-renewal capacity of LSCs. It has been proposed that self-renewal of stem cells is determined by a balance among three major cellular events, prolif-eration, survival, and differentiation (48). To further investigate the impact of Tcf1/Lef1 deficiency on self-renewing LSCs, we used Lin Ϫ BM cells from CreER T2 Tcf7 FL/FL Lef1 FL/FL or Cre-ER T2 Tcf7 FL/ϩ Lef1 FL/ϩ mice to establish CML, treated the primary recipients with tamoxifen and transplanted enriched LSCs into secondary recipients as in Fig. 6A. Fifteen days post-BMT, we harvested BM cells from the secondary recipients and found that the total BM cells and GFP ϩ LSKs from Tcf1/Lef1deficient LSC recipients were ϳ29 Ϯ 5.4% and 19.2 Ϯ 3.8% of those from control LSC recipients, respectively. Cell cycle analysis in GFP ϩ LSK cells in the secondary recipients revealed that both Tcf1/Lef1-sufficient and -deficient LSCs showed similar frequency in the quiescent G0 state or in the actively cycling S/G2/M state (Fig. 7, A and B). On the other hand, Tcf1/Lef1deficient LSCs in the secondary recipients were severely compromised in survival compared with control LSCs, showing ϳ3-4-fold increase in Annexin-V ϩ Hoechest33342apoptotic population (Fig. 7, C and D). Because of the extreme paucity of Tcf1/Lef1-deficient LSCs in the secondary recipients, it was not feasible to perform colony formation assays which required sorting separation. Nonetheless, the GFP ϩ Lin Ϫ c-Kit ϩ transformed myeloid progenitor cells, which are derived from GFP ϩ LSKs, exhibited similar frequency in the BM of Tcf1/Lef1-deficient or control LSC recipient mice (Fig. 7E). We infer that loss of Tcf1 and Lef1 did not potently affect LSC differentiation. These data collectively indicate that the impaired self-renewal  MAY 20, 2016 • VOLUME 291 • NUMBER 21 of Tcf1/Lef1-deficient LSCs can be largely attributed to a strong dependence of CML LSCs on Tcf1 and Lef1 for survival.

Tcf1 and Lef1 Regulate HSC and LSC Self-renewal
Targeting Tcf1 and Lef1 Synergizes with TKI in CML Therapy-The TKI therapy is effective in controlling the growth of CML leukemic blasts but lacks strong effect on quiescent LSCs and thus cannot prevent the disease relapse (49). Given the potent impact of targeting Tcf1 and Lef1 on LSC self-renewal, we investigated if the combination of Tcf1/Lef1 deficiency with imatinib could better control CML in the primary recipients (Fig. 8A). After establishing CML using Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ or control BM cells, we treated the recipients with imatinib for 3 weeks. By the end of treatment, all the CML recipients were protected from the lethal disease and survived beyond day 50 post-BMT. Both control and Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫderived LSCs engrafted recipients similarly during the early stage of CML onset, as measured by leukemia burdens in the recipient PBCs (Fig. 5B). To determine if imatinib treatment affects LSC persistence, we detected GFP ϩ LSKs as LSCs in the BM from imatinib-treated recipients and found that Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ -derived LSCs were greatly diminished in frequency and numbers (Fig. 8, B and C). Continuous imatinib treatment further prolonged survival of recipients of p210 BCR-ABL -infected control BM cells, albeit majority of them eventually succumbed to CML (Fig. 8, A and D). In contrast, recipients of p210 BCR-ABL -infected Tcf7 Ϫ/Ϫ Lef1 Ϫ/Ϫ BM cells were better protected by imatinib treatment, with more than 80% of the recipients survived at the end of observation period (Fig. 8D). These findings suggest that targeting Tcf1 and Lef1 has synergistic effect with TKI treatment to better control CML progression and relapse.

Discussion
In this study, we identified differential requirements for the Tcf1 and Lef1 transcription factors for the normal HSCs and transformed LSCs. For the HSCs, there is a modest requirement for Tcf1 and Lef1. Loss of both factors resulted in moderately diminished quiescence of HSCs without discernible effect on HSC survival, which leads to an expansion of the HSC pool. Despite the increase in phenotypic HSCs in Tcf1/Lef1-deficient mice, they are functionally inferior, especially under competitive conditions or regenerative stress. HSCs lacking Lef1 alone also showed a moderate reduction in reconstitution of multiple blood lineages under competitive condition. This is line with a recent report that shRNA-mediated knockdown of Lef1 in mobilized BM progenitor cells diminishes spleen-colony formation capacity on day 12 post-BMT (50). Nonetheless, the HSC impairments were more evident in the absence of Tcf1 and Lef1, indicating a functional redundancy. HSCs lacking both Tcf1 and Lef1 were able to successfully repopulate all non-T blood lineages in primary and secondary hosts, but their capacity for blood regeneration was compromised in the tertiary hosts. This functional deficiency remarkably resembles that observed in transgenic expression of Wnt inhibitors such as Dkk1 and Wif1 and in Wnt3a-deficient fetal liver cells (11)(12)(13). In those studies where Wnt activity is manipulated, a general concern is that such intervention has a side effect on bone structure and/or stromal cells and thus may have altered HSC niches. Our data on Tcf1 and Lef1, the nuclear effector proteins downstream of the canonical Wnt pathway, provide supporting experimental evidence to the conclusion that Wnt ligands have direct impact on HSCs per se to modulate their regenerative capacity.
On the mechanistic side, the expression of Egr1, Tcf3 and Cdkn1c depended on Tcf1 and Lef1 in hematopoietic stem/ progenitor cells. In addition, the 5Ј-regulatory regions of Egr1 and Tcf3 were bound by Lef1 in BM progenitor cells, albeit we cannot exclude the possibility that Cdkn1c may be regulated by Tcf1/Lef1 factors through distal elements. Egr proteins are the immediate early response transcription factors and are involved in stress responses in many tissues. The molecular connection between Tcf1/Lef1 and Egr proteins has been observed in immature thymocytes as well (51). By genetic complementation, we showed that forced expression of Egr1 was sufficient to rectify the diminished quiescence in Tcf1/Lef1deficient HSCs. These data suggest that Egr1 is one of the key target genes regulated by Tcf1 and Lef1 in preventing excessive HSC cycling. Cell cycle regulators are essential for maintaining HSC quiescence (52). Among the Cip/Kip family of cyclin-dependent kinase inhibitors, only genetic targeting of Cdkn1c (the p57-Kip2) caused evident loss of HSC quiescence, albeit it does have cooperative roles with other family members, p21-Cip (encoded by Cdkn1a) and p27-Kip1, in the maintenance of HSCs (46,47). The role of Tcf1 and Lef1 in HSC quiescence may be also mediated by regulation of the p57-Kip2, suggesting a multifaceted regulatory circuit under the control of Tcf/Lef factors. Future studies should address a possible redundancy of Tcf1/Lef1 with Tcf3 and Tcf4 factors in the same family, which will further elucidate their downstream genes that contribute to regulation of HSC homeostasis.
␤-Catenin stabilization and nuclear translocation are key events upon activation of the canonical Wnt pathway. The role of ␤-catenin in HSCs has been a highly contentious issue. Induced deletion of ␤-catenin alone or in combination with its homologue ␥-catenin did not have a detectable impact on HSCs and blood lineages under homeostatic conditions (19,20). Under competitive conditions, the ␤-catenin-deficient BM cells showed diminished contribution to all blood lineages, and their contribution was further reduced when transplanted into secondary recipients (23). Deletion of both Tcf1 and Lef1 showed similar effect, with no obvious impact on HSCs at the steady state but diminishing their competitive repopulation capacity. Our new observations on Tcf1/Lef1 deficiency reiterate the intrinsic requirements of Wnt-␤-catenin pathway in HSCs. Our data are also in line with a recent finding that ␤-catenin is required for HSC regeneration and BM recovery after radiation or chemotherapy (53).
Unlike the moderate impact on normal HSCs, Tcf1 and Lef1 deficiency more potently affected LSCs in the preclinical CML model. Imatinib treatment prolonged the survival of primary CML recipients and hence facilitated detection of LSCs at a later time point after BM transplantation. In this context, targeting Tcf1 and Lef1 greatly diminished numbers of LSCs in the BM. Furthermore, Tcf1/Lef1-deficient LSCs failed to propagate the disease after transplantation into the secondary recipients. Mechanistically, the self-renewing CML LSCs critically depend on Tcf1 and Lef1 for survival, whereas their proliferation and differentiation were not significantly altered due to Tcf1/Lef1 deficiency. Therefore, Tcf1 and Lef1 are essential for maintaining the pool size and self-renewal capacity of LSCs. In support of this notion, it was previously reported that forced expression of Lef1 in BM progenitors causes B lymphoblastic and acute myeloid leukemia, which are likely propagated by LSCs with lymphoid characteristics (54). An important goal in chemotherapy is to eliminate malignant cells and minimize damage to normal cells. Our studies show that HSCs and LSCs have differential requirements for Tcf1/Lef1 in their self-renewal capacity. Tcf1 and Lef1 can therefore be exploited as novel therapeutic targets in eradicating LSCs in hematological malignancies with less adverse effect on normal hematopoiesis.
Aberrant activation of ␤-catenin is a hallmark event in tumor initiation, growth and metastasis, and has been a well soughtafter drug target in cancer therapy (55). It has been demonstrated that genetic ablation or pharmacological inhibition of ␤-catenin expression synergizes with imatinib in improving survival of CML mice; however, this effect is only observed after the LSCs were serially transplanted into tertiary recipients (25). In contrast, the synergy of Tcf1/Lef1 deletion and imatinib was evident in the primary recipients where the CML was first established. These observations imply that the role of Tcf1 and Lef1 in LSCs may not be mediated exclusively by ␤-catenin. In fact, our recent studies showed that Tcf1 and Lef1 directly interact with Runx3 transcription factor and this interaction contributes to Cd4 gene silencing in CD8 ϩ T cells (29). Another study mapped genome-wide binding locations of Tcf1 and Runx1 transcription factors in a mouse EML (erythroid, myeloid, and lymphocytic) multipotent hematopoietic precursor cells (56). Significantly, the Tcf1 and Runx1 binding peaks overlap at more than 40% of all their binding locations. The differential use of cofactors (␤-catenin versus Runx proteins) by Tcf1 and Lef1 merits further investigation. It is currently recognized that transcriptional programs can be druggable targets (57). Detailed delineation of Tcf1/Lef1-dependent transcriptional program in HSCs and LSCs will facilitate identification of small molecules to antagonize Tcf1/Lef1 activities in LSCs and further in vivo assessment of their therapeutic potentials in treating leukemia.
Author Contributions-S. Y. and H. H. X. designed the experiments and wrote the manuscript. S. Y., F. Y., and S. X. performed the experiments and analyzed the data. W. P. directed the ChIP studies. T. Z. assisted the overall experiments.