Discrete roles of canonical and non-canonical Wnt signaling in hematopoiesis and lymphopoiesis

The mechanisms that regulate proliferation, fate decisions and differentiation of hematopoietic stem cells (HSC) and thymic stem cells are highly complex. Several signaling pathways including Wnt signaling have important roles during these processes. Both canonical and non-canonical Wnt signaling are important in normal and malignant hematopoiesis and lymphoid development, yet their precise roles are controversial. In a side-by-side comparison, we investigated the roles of the canonical and non-canonical Wnt pathway in hematopoiesis and thymopoiesis. As complete loss-of-function models for non-canonical Wnt signaling are not yet available and highly complex for canonical Wnt signaling, we decided to use a gain-of-function approach. To this end, Wnt3a and Wn5a, two well-known prototypical canonical and non-canonical Wnt ligands were produced in hematopoiesis supporting stromal assays. High levels of Wnt3a signaling blocked T-cell development at early stages, whereas intermediate levels accelerated T-cell development. In contrast, Wnt5a signaling prompted apoptosis in developing thymocytes, without affecting differentiation at a particular stage. To explore the role of Wnt3a and Wnt5a in vivo, we transduced HSCs isolated from fetal liver, transduced with Wnt3a and Wnt5a vectors, and performed reconstitution assays in irradiated C57Bl/6 mice. Wnt3a overexpression led to increased lymphopoiesis, whereas Wnt5a augments myelopoiesis in the bone marrow (BM) and spleen. Thus, the canonical and non-canonical Wnt signaling have discrete roles in hematopoiesis and thymopoiesis, and understanding their right dose of action is crucial for prospective translational applications.

The development of blood and immune cells are highly complex and regulated processes. A wide variety of signaling pathways has been implicated in these processes. Several developmental signals have key roles in both the bone marrow (BM) and thymus, such as BMP, Wnt and Notch signaling. 1,2 Hematopoietic stem cells (HSC) are rare BM-residing cells with the capacity to self-renew and differentiate into all blood cell lineages. All blood cells, except T lymphocytes, develop within the BM. Different types of progenitor cells migrate from BM to thymus where they develop to mature T cells. 3,4 The nature of these cells is still subject of debate. As only few early thymic progenitors (ETPs) arrive in the thymus (o10/day), massive proliferation is necessary to establish a pool of T-cell progenitors. 5 During development, these immature thymocytes gradually lose their proliferative and multilineage potential, and initiate a T-cell developmental program, a process termed T-cell commitment. Notch signaling has been shown to have an important role during T-cell commitment by directly or indirectly upregulation of T-cell-specific genes including Ptcra, Cd3e and Zap 70. 6,7 Other soluble factors including Wnt ligands might also be crucial for T-cell proliferation and commitment. [8][9][10] Early stages of T-cell development are phenotypically characterized by absence of the mature T-cell markers CD4 and CD8 and referred to as double negative (DN). 11 DN stages are subdivided into four stages. DN1: CD44+ CD25-, DN2: CD44+ CD25+, DN3: CD44 − CD25+ and DN4: CD44-CD25-. 8,12 It is believed that T-cell commitment occurs at the transition of DN2 to DN3 stages. 13,14 Afterwards, thymocytes develop to the immature single positive (ISP) stage defined as CD3 − CD8+. Thymocytes with functionally rearranged T-cell receptors (TCRs) develop to next stage, which is double positive (DP) for CD4 and CD8 and finally they become mature single positives (SP) 12 either CD4 or CD8. [7][8][9] Thymic epithelial cells (TECs) provide a unique environment for ETPs to develop towards T cells. 15,16 TECs also express high levels of Notch ligands including Delta like ligands 1 and 4, soluble Wnt ligands and IL-7, which all are crucial for early stages of T-cell development. 17 The Wnt signaling pathway is subdivided into canonical (β-catenin dependent) and non-canonical (β-catenin independent) pathways. Binding of different Wnt proteins to frizzled (Fzd) receptors can trigger different Wnt pathways. The diversity of ligands and receptors makes the study of Wnt signaling from point of view of cell surface receptors and ligands challenging. Wnt proteins function as proliferationinducing growth factors but may also affect cell-fate decisions, apoptosis and quiescence. 18 Canonical Wnt proteins bind to their receptors, thereby preventing proteosomal degradation of the Wnt-mediator β -catenin. Subsequently, β -catenin trans locates to the nucleus where it will form an active transcription complex with one of the four transcription factors downstream of the Wnt pathway: Tcf1, 3 or 4 (T-cell Factor 1,3,4) or Lef1 (lymphocyte-enhancer-binding factor). Upon transcriptional activation, several target genes will be activated including Axin2, c-fos, c-myc and many others, which are important for proliferation and/or cell-fate decisions. Non-canonical Wnt signaling involves recognition of distinct Wnt ligands by a cognate Frz-LRP receptor complex, heterotrimeric G protein activation of phospholipase C as well as the release of intracellular Ca 2+ ions. Non-canonical Wnt signaling also regulates cellular polarization and migration (the so-called planar-cell-polarity pathway). 9,10,[18][19][20][21] A large body of evidence has shown the significance of canonical Wnt signaling during T-cell development. Generation of Tcf1 KO mice provided the first evidence of a Wnt signaling effect during T-cell development. 22 Tcf1 deficiency partially blocks T-cell development at various early DN stages, resulting in fewer mature T cells and smaller thymus. In addition, Tcf1/Lef double KO mice have a complete block at the ISP stage, which indicates redundancy between these factors during thymocyte development. 23 Similar to Tcf1 deficiency, fetal thymic organ cultures using Wnt3a-deficient progenitors exhibited progressively  impaired T-cell development caused to an ISP block. 24 Several loss-of-function and gain-of-function studies have targeted the core Wnt-mediator β-catenin in thymocytes. Conditional β-catenin deletion using Lck-Cre impaired β-selection of TCR, 25 whereas β-catenin overexpression regulates positive selection and generation of SP CD4/ CD8. 26 Inhibiting the interaction between β-catenin and Tcf1 also blocks the DN to DP transition. 27 Wnt signaling is active at various stages of T-cell development but most predominantly at DN stages. Indeed, inhibiting Wnt signaling by using Dickkopf (DKK1) as Wnt sequestering molecule, blocks the development at the most immature DN1 stage. 21 Finally, genetic proof that canonical Wnt signaling is crucial for normal T-cell development stems from complementation studies in which only the large form of Tcf1 that can interact with β-catenin and transduce Wnt signals was capable to restore T-cell development, whereas the short form that lack the β-catenin domain could not. 28 On contrary, there are only a handful of studies focused on the role of non-canonical Wnt signaling in T-cell development. Liang et al. 29 showed that Wnt5a deficiency protects against apoptosis in DP stage, but it was proposed to be irrelevant at early stages of T-cell development. Another series of studies revealed that Wnt4 induces expansion of Lin-Sca1+ Kit+ (LSKs) in BM, which subsequently causes ETP expansion in thymus. 30 Wnt4 regulates ETP expansion via a TEC-dependent mechanism. 31 Whether Wnt4 functions as a canonical or noncanonical ligand is still debatable, in particular in gain-offunction studies in which thymopoiesis was increased. Perreault and colleagues showed that Wnt4 binds to Fzd6 and activates JNK kinase via PCP pathway, 32 strongly suggesting that Wnt4 signals in a non-canonical fashion.
Loss-of-function approaches using β-catenin and γ-catenin have often not revealed thymic phenotypes, 33 most likely because Wnt signaling is still present at appreciably levels in these models. 34 In addition, during T-cell development canonical and non-canonical Wnt signaling have only been studied in isolation but not together. Here, we aimed to sideby-side compare the two prototypical canonical and noncanonical Wnt ligands, that is, Wnt3a (canonical) and Wnt5a (non-canonical).

OP9-based in vitro assays for T-cell development.
To study the role of canonical and non-canonical Wnt signaling  Figure 1). We also generated another OP9-based assay in which we directly transduced OP9-Dl1-GFP cell line with Wnt3a and Wnt5a-tomato constructs, and assessed in vitro T-cell development ( Figure 1b). We quantified relative expression of Wnt3a, Wnt5a and Dl1 using Q-PCR. Wnt3a had over 1000fold higher expression compared with OP9-WT, whereas Wnt5a had~500-fold and Dl1 remained to be expressed at levels over 700-fold ( Figure 1c). In OP9-DLWnt3a-tomato (DLW3A) and OP9-DLWnt5a-tomato (DLW5A) Wnt3a and Wnt5a were expressed as high as in the OP9-Wnt3a-GFP and OP9-Wnt5a-GFP, respectively, and Dl1 expression was not altered compared with the OP9-Dl1-GFP ( Figure 1d).
Canonical Wnt3a overexpression blocks T-cell development at early stages, thereby favoring development of alternative lineages. We cultured fetal liver cells (as these provide a good source of both T-cell progenitors and stem cells) either with the mixture of OP9-WT/Dl1 in 1:1 ratio as a control, or with OP9-Wnt3a/Dl1 or OP9-Wnt5a/Dl1 with the same ratio for 14 days. Wnt3a overexpression blocked T-cell development at the DN1 stage (Figures 1e and f).
The mixture of OP9-WT and OP9-Dl1 provides a means to study development of B, T, NK and myeloid lineages simultaneously. 35 DN1 and DN2 immature thymocytes have the potential to develop towards other lineages. Consistently, in the Wnt3a-overexpressing cultures, non T cells, such as B and myeloid cells developed more efficiently compared with the control (Figures 2a and b). The difference was not only observed in proportion of each lineage, but also in the absolute numbers ( Figure 2c).
Similar data were observed when Wnt3a and Wnt5a were expressed in the OP9-Dl1 cells themselves (supplementary Figure 2). Therefore, we conclude that canonical Wnt3a signaling has the capacity to inhibit T-cell development at early DN2-DN3 stages, and to increase alternative (myeloid and B cell) development. Wnt5a overexpression in neither of assays Wnt5a overexpression results in increased apoptosis in developing thymocytes, whereas Wnt3a does not. Further analysis revealed that total percentage of lymphocytes was around fourfold lower in the Wnt3a-overexpressing group at various time points after co-culture (4, 7 and 14 days) (Figure 3a). This could be caused by increased apoptosis in the Wnt3a-expressing cultures. We performed apoptosis assays by using AnexinV and 7AAD in combination with the Thy1 marker to separate T cells from non-T cells. In the Wnt3a-overexpressing cultures total Thy1+thymocytes and specifically DN thymocytes were not undergoing any a significant level of apoptosis ( Figure 3a). There is an increase in apoptosis in non T cells, yet not to the extent that non T cells would decrease in numbers in comparison with thymocytes (see Figures 2b and c). Thus, we concluded that inhibition of differentiation toward T lineage due to Wnt3a overexpression causes low cellularity in this culture. Although the Wnt5a-overexpressing cultures did not show any phenotypic differences compared with the controls, the absolute numbers were around three-to fivefold lower (Figure 3b). Lack of Wnt5a promotes Bcl-2 expression and inhibits apoptosis of DP thymocytes. 29 Our analysis showed that both Thy1+ and Thy1-co-cultured cells were undergoing more apoptosis. Thus, this gain-of-function approach showing more apoptosis is consistent with the loss-of-function Wnt5a experiments reported before. The effect of apoptosis in thymocytes (DN and Thy1+) by Wnt5a was much stronger than Wnt3a, which was similar to controls. Interestingly, for non T cells, similarly high apoptosis was found. The strong  Optimal dosage of Wnt3a signaling accelerates T-cell development. Gene expression analysis revealed that Wnt3a was very highly expressed in our transduced OP9 cells (Figures 1c and d). This is 500-fold higher than the physiological level, for example, as found in fetal thymus, 21 which we used as comparison. We have previously shown that canonical Wnt signaling functions in a dosagedependent fashion during HSC reconstitution and T-cell development using Apc hypomorphic models. 19 We hypothesized that the same would hold true with the Wnt3a gain-offunction model. We modified the Wnt3a concentration in the culture by serially mixing OP9-DL with OP9-DLW3a at various ratios. After 14 days of co-culture, the majority of cells developed into DN3 and DN4 with 32% of DN4 thymocytes. Interestingly, the co-cultured cells with 1% DLW3A (which has fivefold higher overexpression relative to the physiological level, supplementary Figure 3), showed an accelerated T-cell development with 50% DN4 thymocytes development. However, T-cell development was inhibited again by increasing the concentration of DLW3A (10% and 50% Wnt3a), and was completely blocked at the DN3 stage with 100% OP9-DLW3A alone (Figure 4a).
To determine the Wnt signaling activity, we performed the same experiment using Axin2 LacZ heterozygous reporter cells, a well-established Wnt reporter mouse model. 36 Phenotypic analysis of the co-cultured cells exhibited similar T-cell development potential at different dosages of Wnt3a (data not shown). Importantly, the actual Wnt signaling activity of total Thy1+ cells correlated with the increasing concentration of Wnt3a (Figure 4b). Therefore, our data suggest that Wnt3a canonical Wnt signaling functions in a dosage-dependent fashion, in accordance with the differential Wnt signaling activity of the cells.

Wnt3a overexpression enhances B lymphopoiesis and Wnt5a overexpression augments myelopoiesis in vivo.
To study the role of Wnt3a and Wnt5a signaling in vivo we carried out reconstitution assays in which we sorted CD45.2 LSK cells from fetal liver, transduced them with lentiviral vectors encoding Wnt3a, Wnt5a in combination with GFP, and transplanted them into irradiated CD45.1 B6-recipient mice (Figure 5a). At week 16 post transplantation we analyzed spleen, thymus and BM of the recipients. In the spleens of Wnt3a-overexpressing mice the percentage of B220+ CD19+ B cells was around twofold higher compared with the control mice within the GFP+ transduced compartment. In contrast, Wnt5a-overexpressing mice had around twofold higher percentage of CD11b+ Gr1+ myeloid cells relative to the control. Therefore, the ratio of B cells versus myeloid cells was in favor of B cells in the Wnt3a group, and in favor of myeloid cells in the Wnt5a group (Figures 5b and c).
In the BM, the ratio of B cells versus myeloid cells was around threefold lower in the Wnt5a group, whereas there was no statistically significant difference between the control and Wnt3a group. MPPs (LSK Flt3+) were around  (Figure 6a). In the thymus, T-cell development was clearly blocked at early stages in the transduced compartment of the Wnt3a group, and to some extend with Wnt5a (as characterized by higher percentage and MFI of GFP in DN and ISP stages). However, the blocks did not affect the absolute number of mature T cells or total thymic cellularity (Figures 6b and c). Thus, Wnt3a overexpression induces B lymphopoiesis in spleen and BM whereas Wnt5a overexpression induces increased myelopoiesis in these organs.

Discussion
Canonical Wnt signaling has a well-established role in T-cell development in the thymus; yet only a few reports deal with non-canonical Wnt signaling. A side-by-side comparison of the effects of these interacting pathways has not been performed, in contrast to for instance B-cell development 20 and HSC biology. 37 We therefore set out to compare the prototypical canonical Wnt ligand Wnt3a with the prototypical non-canonical Wnt ligand Wnt5a. We demonstrated that the induction of canonical Wnt signaling via Wnt3a is first, important for T-cell development, and second functions in dosage-dependent fashion. Although intermediate-to-low doses of canonical Wnt signaling is beneficial for thymopoiesis, higher doses support B lymphopoiesis in vivo. On the other hand, Wnt5a non-canonical Wnt signaling induces myelopoiesis in vivo and it does not appear to function in a strict dosage-dependent fashion. Moreover, Wnt5a signaling induces apoptosis in developing thymocytes, in line with the diminished apoptosis observed in Wnt5a-deficient thymi. Previous studies in our laboratory and many others revealed that canonical Wnt signaling is crucial for T-cell development, and it functions in a dosage-dependent fashion. 19 In the current study we used Wnt3a, a natural ligand of canonical Wnt signaling, and we obtained similar data. This shows that Wnt3a triggers canonical Wnt signaling via the β-catenin and TCF/LEF-dependent pathway in the thymus, 34 and that it is possible to modulate canonical Wnt signaling via differential concentration of Wnt3a physiologically. It is very likely that in the thymus, developing thymocytes are exposed to different types and concentrations of Wnt ligands. It is likely that thymocytes express different FZD receptors with various binding affinity to the existing Wnt proteins within the thymic microenvironment. Previous Q-PCR data suggest that this might be the case, 21 although experimental proof awaits the development of specific antibodies for each Frizzled receptor suitable for flow cytometry. As a result, developing thymocytes would undergo different levels of Wnt signaling owing to the accumulation of different amounts of β-catenin proteins in the cytoplasm.
Another possibility is that the interaction of canonical and non-canonical Wnt signaling might be important for control of β-catenin dosage in the cytoplasm. Sugimura et al. 37 elegantly described a situation where canonical and non-canonical Wnt signaling interact with each other in the hematopoietic system. Non-canonical Wnt maintains quiescent long-term HSCs through Flamingo and Frizzled 8 receptor on HSCs. Under stress, non-canonical Wnt is attenuated and canonical Wnt is enhanced, which results in the activation of HSC. The same might be true during T-cell development in the thymus under stress, which could be a fascinating issue for future studies.
Previously, Malhorta et al. 20 used a similar OP9-based Wnt3a/Wnt5 gain-of-function approach to study B lymphopoiesis. There are several differences between these two studies. (i) Their study is restricted to lympho-hematopoiesis of the BM and not thymopoiesis in the thymus. 34 Their study was   40 have demonstrated that human aged HSCs have a reduced canonical Wnt signaling activity, and as a result impaired or delayed T-cell development, which is restricted to the T-cell progenitors, indicating the significance of Wnt signaling in human T-cell development during senescence. The notion that non-canonical Wnt signaling mimics murine HSC ageing has been put forward by Florian et al. 41 They reported that a shift from canonical to noncanonical Wnt signaling occurs during HSC ageing. Wnt5a treatment of young HSCs induced ageing associated stem cell apolarity, and an lymphoid to myeloid differentiation skewing, which is observed normally during ageing. We performed similar experiments in which we transduced HSC with Wnt3a and Wnt5a, rather than ex vivo treatment. Similarly, myelopoiesis was enhanced in Wnt5a-overexpressing HSC, and B lymphopoiesis in Wnt3a-overexpressing HSC (Figures 5 and  6). Therefore, Wnt3a treatment of stem cells before transplantation could be considered a promising approach to improve lymphopoiesis.
A series of studies in the laboratory of Perreault and coworkers [30][31][32] suggest that Wnt4 is a non-canonical Wnt ligand, enhances MPP expansion in the BM and ETP proliferation in the thymus, which results in increased thymic cellularity. We have also shown that Wnt5a overexpression induces MPP expansion, and DN expansion in thymus ( Figure 6). However, this did not affect the total cellularity of the organs. This suggests specific roles of various Wnt family members during different stages of thymopoiesis, probably due to their differential expression throughout the thymus and/ or differential responsiveness of developing thymocytes. For instance, Wnt4 has been proposed to function as a canonical Wnt in inducing FoxN1 expression in TEC, 42 but seems to act as a non-canonical Wnt in studies aimed at improving thymic reconstitution from hematopoietic rather than thymic stromal cells. 30 Previous studies indicate that deregulation of Wnt signaling occurs in leukemia. 43 We have also shown that deregulation of Wnt signaling due to the absence of Tcf1, induces lymphomas in mice. 44 Independent of our study, Yu et al. 11 also reported similar results and showed similarities of Tcf1 lymphomas with human ETP-ALL cases. In addition, Martins et al. 45 recently suggested that natural cell competition between young and old thymic progenitors is crucial for inhibition of T-ALL development. Therefore, thymic progenitor fitness is necessary for normal T-cell development. Our gain-of-function Wnt3a model preserves thymocytes in an immature state without inducing any malignancy. It is therefore intriguing to speculate that canonical Wnt signaling may be involved in regulating stemness of thymic stem cells.
Concluding, our work and that of others referred to above show discrete effects of Wnt3a and Wnt5a treatment on hematopoietic cells, both in vitro and in vivo. These attempts might lead to application of Wnt ligands as therapeutic candidates to improve HSC repopulation and T-cell reconstitution after SCT. However, the challenge and focus of future studies should be on determining the 'right concentration' of Wnt proteins to avoid deregulated Wnt signaling.

Materials and Methods
Some methods are described in the supplementary data.
Lentiviral production and transduction. Wnt3a and Wnt5a gene transfer plasmids individually cloned into the multiple cloning sites of pRRL-SFFV-IRES-GFP lentiviral vector by restriction digestion and ligation reactions. 293 T cells were transiently transfected with either the genes transfer or empty control constructs together with helper plasmids using X-TremeGENE9 transfection reagent (Roche, Basel, Switzerland). Virus containing supernatants were harvested 20 h and 40 h after transfection in Iscove's Modified Dulbecco's Medium supplemented with 10% fetal bovine serum, L-glutamin, 100 U/ml penicillin and 100 mg/ml streptomycin and used immediately for transduction, or stored at -80°C for later use.
For transduction of OP9 stromal cells, 150 000 OP9-WT or OP9DL1 cells were seeded into a well of six-well plates with alpha MEM (Lonza, Verviers, Belgium) supplemented with 20% FCS-HI, L-glutamin and penicillin/streptomycin. The following day, OP9 cells were transduced with fresh or frozen viral supernatant containing 4 µg/ml proteamine sulfate (Sigma-Aldrich, St. Louis, MO, USA). After few days of culture transduced GFP+ cells were sorted, and stable OP9-pRRL-GFP (empty vector transduced), OP9-(DL1)-W3A (Wnt3a-transduced) and OP9-(DL1)-W5A (Wnt5atransduced) cell lines were generated. Overexpression of the genes were then confirmed by Q-PCR gene analysis. The overexpression of DL1 gene was not altered in the transduced cell lines compared with untransduced or transduced with empty vector.