Wnt-Related Molecules and Signaling Pathway Equilibrium in Hematopoiesis

Complexity of Wnt Signaling Components and Pathways

Some 19 Wnt ligands interact with both secreted and membrane-associated proteins to initiate signaling processes (detailed in Staal et al. [2008]; references in Table 1; summarized in Figure 1). The list of Wnt binding proteins include at least ten seven-pass transmembrane Frizzled (Fzd) receptors, two low-density lipoprotein receptor-related proteins (LRP), and a number of extracellular Wnt-modulating proteins such as Kremlin, Dickkopf (Dkk), Wnt-inhibitory factor (WIF), secreted Fzds (SFRP), and Norrin. Depending on the type of ligand-receptor interaction, the availability of intracellular signaling components and the specific target cell, an astonishing ten Wnt signaling pathways have been identified (Semenov et al., 2007; Macdonald et al., 2007). The canonical pathway, arguably the most studied and best understood, results in stabilization and nuclear translocation of β-catenin (Figure 1). This molecule is normally targeted for degradation by a GSK protein complex, but canonical signaling stabilizes β-catenin by destruction of the degradation complex. Stabilized β-catenin then enters the nucleus and acts as a coactivator for transcription factors such as LEF and TCF. A related molecule, γ-catenin (plakogloblin), is at least partially redundant to β-catenin and can bind TCF/LEF. Simultaneous targeting of both catenin genes results in embryonic lethality (Haegel et al., 1995; Ruiz et al., 1996). Further, Wnt signaling continued in the combined absence of β/γ catenin after inducible deletions, suggesting involvement of unknown redundant proteins (Jeannet et al., 2007). Adenomatous polyposis coli (APC) is a multidomain protein that can bind GSK3b, Axin, and β-catenin to prevent nuclear localization of catenin, thus decreasing canonical Wnt signaling (Aoki and Taketo, 2007). Independently of Wnt signaling, APC is involved in cellular adhesion, microtubule stabilization, chromosomal segregation, and proliferation events.

Actions of the Wnt ligands are frequently opposed by other Wnts, soluble receptors, and antagonists (see below). Furthermore, biological responses are dictated by which of the alternate signaling mechanisms are utilized. Two noncanonical pathways, Wnt-Ca²⁺ and Wnt-JNK, do not (normally) stabilize β-catenin pools. Rather, Wnt-Fzd receptor interactions activate membrane-associated G protein complexes and Disheveled (Dsh) to either increase intracellular Ca²⁺ levels through inositol-3-phosphate (IP3) or induce the JNK pathway through Rho/Rac GTPases. As a result, noncanonical signals can influence actin-dependent cytoskeletal reorganization. Other studies suggest crosstalk/competition between canonical and noncanonical Wnt pathways. For example, activation of the canonical Wnt pathway is observed in mice deficient in the noncanonical Wnt5a ligand (Topol et al., 2003). Further, activation of noncanonical pathways can sometimes block canonical signaling (Nemeth et al., 2007).

Wnt Ligands Influence Stem and Progenitor Cells

The discovery that Wnt genes are expressed in hematopoietic tissues was soon followed by attempts to determine their contribution to blood cell formation (see Table 1, for a summary, and references). A variety of hematopoietic cell fractions were exposed in culture to recombinant proteins in conditioned medium, commercially obtained materials or purified ligands. Other experimental paradigms placed stem cells and progenitors in culture with Wnt producing stromal cells. Different forms of presentation, doses, and posttranslational modifications are likely to have affected potencies of specific ligands and their corresponding outcomes. Furthermore, it is becoming clear that the nature of the target cells is an important parameter in determining the specific response to a given ligand, either alone or in combination.

An example of this context-dependent response of blood progenitors is found in an early report that showed that Wnt1, Wnt5a, or Wnt10b stimulated the expansion of fetal hematopoietic cells in culture (Austin et al., 1997). Most of the experiments were conducted with Wnt5a, and 2- to 4-fold increments in cell production were observed with addition of the factor or retroviral introduction of a Wnt5a cassette. In another study, human CD34⁺ progenitors responded to a similar degree when Wnt5a, Wnt2b, or Wnt10b were provided by transduced stromal cells, but it is noteworthy that recombinant human SCF also gave a similar result (Van Den Berg et al., 1998). Much more impressive results were reported by Reya, Weissman, and colleagues, where HSC numbers increased 10- to 50-fold on addition of purified Wnt3a to cultures that contained SCF (Willert et al., 2003). Although dramatic results were obtained in primary experiments conducted with stem cells from Bcl-2 transgenic mice, the authors provided more limited supplemental data generated using marrow from normal animals. While similar trends were observed, expansion of long-term repopulating stem cells from conventional mice was not as robust nor vigorously demonstrated (Reya et al., 2003). More recent studies concluded that Wnt3a was detrimental to stem cell survival and expansion in short-term cultures, even when Bcl-2 transgenic cells were used (Nemeth et al., 2007; Malhotra et al., 2008). Wnt5a was shown to function as an antagonist of canonical Wnt signaling in hematopoietic cells and caused some expansion of transplantable stem cells (Nemeth et al., 2007). Retroviral introduction of Wnt3a prevented reconstitution, but Wnt5a was without influence when expressed using the same approach. The negative result could be due to the use of a potentially toxic DsRed fluorochrome (Tao et al., 2007) as an expression marker. Alternatively, Wnt5a might normally be expressed in sufficient amounts in marrow niches such that elevated exposure to the ligand had no effect. However, injection of purified Wnt5a into NOD/SCID mice increased engraftment of human hematopoietic progenitors (Murdoch et al., 2003).

Canonical Wnt Signaling and Hematopoiesis

As discussed above, there is a multitude of Wnt signaling cascades, but the most well-known is the canonical β-catenin pathway, which has been implicated to play a regulatory role in hematopoiesis. In one dramatic example, HSC of BCL-2 transgenic mice were transduced with a constitutively active-β-catenin cassette, promoting their expansion almost 100-fold (Reya et al., 2003). In fact, even after a 1 week culture period, the transduced cells retained a robust ability to engraft the marrow of irradiated recipients and differentiated in all blood lineages. The culture conditions that were sufficient to achieve this result were remarkably simple, but it remains unclear whether the β-catenin gene provided was itself responsible for the observed growth promotion of long-term repopulating HSC. While this hypothesis is certainly an option, it is also conceivable that appropriate levels of canonical Wnt signaling may have simply preserved stem and/or multipotent progenitor integrity and maintained the cultured cells in an undifferentiated state (see additional comments below). We and others have since found that some, but not all, stem cell characteristics can be retained for at least 5 months when primitive hematopoietic cells were transduced with constitutively active β-catenin (Baba et al., 2006; Templin et al., 2008). That is, the cell lines were multipotent and self-renewing but unable to efficiently reconstitute the thymus or rescue lethally irradiated mice. It is noteworthy that both introduced and endogenous β-catenin levels declined whenever our multipotent cell lines acquired lineage-associated markers (Baba et al., 2006). Wnts form activity gradients, and signaling levels might have to be within a precise range to dictate stem cell renewal versus differentiation. Consistent with that interpretation, two lines of transgenic mice that overexpressed large amounts of β-catenin in hematopoietic cells experienced marrow failure (Kirstetter et al., 2006; Scheller et al., 2006). Canonical signaling was also increased in Hmgb3 knockout mice, where progression in hematopoietic lineages was again diminished (Nemeth et al., 2006).

Adenomatous polyposis coli (APC) is another essential component of the canonical Wnt pathway signaling, functioning in the APC-Axin-GSK3-β complex that causes β-catenin degradation. A recent study used interferon-triggered deletion to assess the contribution of APC to hematopoiesis (Qian et al., 2008). The targeted knockout resulted in marrow failure within 2 to 3 weeks, marked by greatly reduced numbers of the stem cell-containing LSK subset as well as more mature progenitors in all lineages. Moreover, the APC-deficient HSCs were intrinsically defective and unable to sustain any type of blood cell formation when transplanted. The analyses revealed abnormally increased proliferation of stem and progenitor rich subsets as well as a tendency to undergo apoptosis. As would be expected, levels of β-catenin were very high in APC-deficient cells, and most of the abnormalities were consistent with extraordinary high levels of canonical Wnt signaling (Kirstetter et al., 2006; Scheller et al., 2006). It is noteworthy that APC is also involved in Wnt-independent functions, and the authors speculated that β-catenin driven phenomena might not alone account for the rapid apoptosis and stem/progenitor depletion they observed. This hypothesis is similar to the Wnt pathway-independent proapoptotic function proposed for axin (Hsu et al., 2001). In any case, this report again demonstrates that stem cell function requires canonical Wnt signaling to be kept within a critical physiological range.

Glycogen synthase kinase 3 (GSK3) is another key component of the canonical signaling pathway. GSK3 is a serine-threonine kinase that phosphorylates cytosolic β-catenin and prevents it from translocating to the nucleus (Figure 1). Inhibition of GSK3 by treatment with a low MW compound increased the repopulation ability of murine as well as human HSC in immunodeficient NOD/SCID mice (Trowbridge et al., 2006). The treatment also expanded stem cells in vitro. In contrast to this result, another group recently found that GSK3 inhibition delayed ex vivo expansion of human CD34⁺ cord blood cells, while preserving a stem-like phenotype (Holmes et al., 2008). It is important to note that, in addition to Wnt, GSK3 is involved in other signaling pathways such as Insulin, Hedgehog and Notch, some of which have consequences for hematopoiesis (Kockeritz et al., 2006) (Doble and Woodgett, 2003).

All of these experimental approaches caused canonical Wnt signaling to be abnormally elevated, and reciprocal results have not always been obtained in knockout studies. Targeted deletion of either β- or γ- catenin genes results in embryonic lethality, so several studies made use of inducible gene deletions. Hematopoietic cells deficient in one (Cobas et al., 2004) or both (Koch et al., 2008; Jeannet et al., 2007) catenin genes had no obvious defects. These impressive studies were hailed by many as definitive demonstrations that the canonical pathway is not essential for hematopoiesis, but Held and colleagues subsequently showed that TCF/LEF-dependent transcription continued in the combined absence of β- and γ-catenin (Jeannet et al., 2007). It has been speculated that partially functional β-catenin protein might persist in these animals (Staal and Sen, 2008). As another possibility, the reporters used to assess Wnt-dependent signaling in the Jeannet study could have been triggered by non-Wnt ligands such as TGF-β (Barolo, 2006). Furthermore, each of the conditional deletion experiments discussed above utilized an interferon-inducible Cre model to induce gene deletion in adults. In contrast, Zhao et al. subsequently used Vav-cre for β-catenin gene deletion from fetal life and observed significant hematopoietic abnormalities (Zhao et al., 2007). Therefore, the combined set of experiments beg the question of whether the unique proliferative activity and expansion demands placed on fetal stem cells reveal a requirement for Wnt signaling that is otherwise obscured in adults and during homeostasis. As with other models of Wnt blockage or deficiency (Fleming et al., 2008; Luis et al., 2008), stem cell renewal defects were only obvious in secondary transplantation assays. Thus, Wnt-dependent influences on hematopoietic cells can be subtle and only appreciated when very rigorous tests are performed.

Competition among Wnt Family Molecules

Responses to Wnts are buffered by the complexity of interacting ligands and antagonists. Also, there are many examples of crosstalk and competition between Wnt signaling pathways (Mikels and Nusse, 2006; Nemeth et al., 2007; Baksh et al., 2007). As described above, HSCs respond to Wnt5a, and the same ligand also promotes lineage progression of primitive lymphoid progenitors. Wnt5a is usually considered to be a noncanonical Wnt ligand and can antagonize Wnt3a signals via GSK-3-independent β-catenin degradation. For that reason, results obtained with Wnt5a^−/− mice could reflect the absence of Wnt5a, an increase in canonical Wnt signaling, or both. As an example of that complexity, transition of HSC into early lymphoid cells is inhibited by the canonical Wnt3a ligand and promoted by Wnt5a (Malhotra et al., 2008). Environmental context and the nature of target cells are important because Wnt5a can be an inhibitor of later stages of lymphoid differentiation (Liang et al., 2003).

A number of secreted molecules modulate Wnt functions in positive and negative ways (Kawano and Kypta, 2003). The Dickkopf (Dkk) proteins may inhibit proximal events in Wnt responses by binding to LRP coreceptors, and this property was recently exploited to implicate Wnt signaling in stem cell maintenance (Fleming et al., 2008). Transgenic expression of Dkk1 with a collagen promoter directed it to subendosteal stem cell niches, reducing canonical Wnt pathway signaling in hematopoietic cells at least 100-fold. While numbers of hematopoietic cells were marginally affected, there were dramatic and durable defects in transplantable stem cells. That is, fewer HSCs were quiescent and bone marrow function was compromised with respect to reconstitution. Remarkably, even temporary residence in Wnt-deficient marrow caused HSC abnormalities.

Wnt equilibrium in normal hematopoietic tissues could theoretically involve one or more of the Dkk proteins. Dkk1 is essential to normal bone development and bone pathology results when it is elevated (Mukhopadhyay et al., 2001; Pinzone et al., 2008). Experimental manipulation of inhibitor levels in the stem cell niche could indicate how Wnts in that site are normally countered by Dkks (Fleming et al., 2008).The Sox17 transcription factor is expressed only during fetal and neonatal life, and this disparity may account for some fetal/adult stem cell differences observed in the hematopoietic system (Kim et al., 2007). It is interesting that Dkk1 was dramatically upregulated in embryos whose Sox17 gene was knocked out, given that HSC also failed to develop in these animals. If Dkk1 normally contributes to Wnt-dependent homeostasis in adult bone marrow, there must be Sox17-independent mechanisms for regulating its expression. Again, it must be considered that requirements for particular Wnt family molecules in fetal and adult life may be different.

In addition to Dkk proteins, several other soluble molecules are important modulators of Wnt functions. For example, secreted Frizzled-related proteins (sFRPs) can bind Wnt ligands, prevent their interaction with transmembrane receptors, and thus, function as antagonists to Wnt signaling (Uren et al., 2000). However, the impact of this class of inhibitors is highly dose dependent, and small amounts of sFRP1 actually enhance Wnt stimulation. We recently discovered that stromal cells release sFRP1 when exposed to estrogen, and this production accounts in part for hormone-mediated suppression of B lymphopoiesis (Yokota et al., 2008). That is, sFRP1 could function similar to Wnt3a as a canonical ligand, at least at a given concentration. However, Wnt3a and sFRP1 were mutually antagonistic when added together at an appropriate ratio. Wnts are known to be matrix bound, while the stability and availability of sFRP1 is heparin dependent (Uren et al., 2000; Zhong et al., 2007). Therefore, proteoglycan-tethered gradients of these molecules could be important for determining stem and progenitor cell functions. The probability of a given stem cell responding via a particular signaling pathway within bone marrow is, therefore, likely to be determined by multiple parameters (Figure 2).

Retention and Migration of HSC in Response to Wnts?

Wnts were discovered within the context of cellular patterning/movement, and β-catenin has well-known roles in cell adhesion junctions (Daugherty and Gottardi, 2007). Thus, it is interesting that canonical Wnt signaling elicited by Wnt3a or introduction of a constitutively active β-catenin gene dramatically reduced expression of cell adhesion molecules (Malhotra and Kincade, 2008). VCAM-1 was diminished in stromal, endothelial, and hematopoietic cells. This adhesion molecule is important for the homing of HSC to bone marrow and is thought to help retain progenitor cells in appropriate niches (Ulyanova et al., 2007; Leuker et al., 2001; Koni et al., 2001). Consistent with the requirement on stromal rather than hematopoietic cells, HSC that had been exposed to Wnt3a and as a result no longer expressed VCAM-1 were still able to engraft transplant recipients. ICAM-1 levels were also reduced on Wnt-stimulated stromal cells (Malhotra and Kincade, 2008). Proximity of hematopoietic cells to appropriate Wnt concentrations is likely to be a critical parameter, and Wnts can control this access by modulating cell adhesion and migration-related molecules. For example, one can imagine a stem cell approaching a given niche and then being encouraged to stay or exit because of the particular Wnt gradient present in that location (Figure 2).

Wnt Control of Stem Cell Proliferation?

Wnts initially appeared to regulate hematopoiesis in a similar fashion as other established growth and differentiation factors. For example, exposure to Wnt proteins potentiated colony formation in cultures that had also been supplemented with colony-stimulating factors (Austin et al., 1997). One might reach the same conclusion based on reports that (1) HSC could be extensively expanded under the influence of recombinant Wnt3a and (2) that the stem/progenitor compartment was abnormally cycling in transgenic mice overexpressing β-catenin (Willert et al., 2003; Scheller et al., 2006; Kirstetter et al., 2006; Qian et al., 2008). However, fetal hematopoietic cells that were already rapidly dividing were used in some of the situations where Wnt family molecules were said to stimulate proliferation (see Table 1). Also, colony numbers in progenitor assays were usually not higher than what could be obtained with combinations of other factors (Van Den Berg et al., 1998). In our hands, exposure of HSC to Wnt3a-producing stromal cells limited their expansion (Malhotra et al., 2008). Furthermore, HSC mitotic activity was high in mice with a Wnt deficiency caused by a Dkk-1 transgene (Fleming et al., 2008). While it remains possible to attribute these discrepancies to varying dose and context of Wnt presentation, we consider it more likely that the primary influence of Wnts is on the differentiation of hematopoietic cells rather than replication. Of particular interest is the possibility that Wnts control the direction of differentiation.

Maintaining and Regaining an Undifferentiated State

Differentiation from long-term repopulating HSC to short-term multipotent and restricted progenitor cells is thought to be unidirectional under normal circumstances. In fact, the best examples of dedifferentiation of blood cells involve malignancies and genetic engineering (Cobaleda and Busslinger, 2008). Similarly, introduction of constitutively active β-catenin had a dramatic effect on partially differentiated hematopoietic cells (Baba et al., 2005). That is, lymphoid committed progenitors could give rise to myeloid cells, and myeloid progenitors could differentiate into lymphoid cells under the influence of canonical, β-catenin-dependent signaling. We speculated that excessive β-catenin caused epigenetic changes that, in turn, diminished lineage-restricting mechanisms. Cobaleda and Busslinger recently noted that “lineage infidelity” has been well documented in follicular lymphoma and suggest that changes in levels of stabilizing transcription factors could be involved (Cobaleda and Busslinger, 2008). Similarly, Lef-1 is a key transcriptional mediator in Wnt signaling, and its deregulation produces tumor stem cells with potential for lymphoid and myeloid differentiation (Petropoulos et al., 2008). An imbalance between Wnt family molecules occurs in some human leukemias (Nemeth and Bodine, 2007).

Progenitor-like cystocytes in the ovary and spermatogonial cells in the testis can dedifferentiate to become germline stem cells in Drosophila (Brawley and Matunis, 2004; Kai and Spradling, 2004). While described in this model as an emergency stem cell source, it was long ago proposed that progenitor-to-stem cell retrodifferentiation might be possible for hematopoietic cells (Schofield, 1978). The findings described above indicate that Wnt signaling can be engineered to achieve a degree of reprogramming (Baba et al., 2005). However, it was not clear if dedifferentiation and progression within an alternate lineage would ever occur naturally, and if so, whether it could be observed in noninvasive experimental circumstances. We recently found that exposure to Wnt3a caused committed lymphoid cells to reacquire some stem cell-like characteristics (Malhotra et al., 2008). In this example, early lymphoid progenitors (ELP) were isolated from marrow of RAG-1/GFP knockin mice and placed in culture with Wnt-producing stromal cells. RAG-1 transcripts were extinguished within 18 hr, while Thy1.1 was reacquired and changes in differentiation potential occurred within the next few days. That is, the previously lineage-biased lymphoid cells retained or regained some myeloid colony-forming capacity during the period of culture. Downregulation of erythroid differentiation potential is thought to be one of the earliest events in the stem to progenitor transition (Adolfsson et al., 2005). Thus, it is significant that ELP became responsive to erythroid differentiation signals during exposure to Wnt3a. Pro-lymphocytes appear further along in the lymphoid lineage differentiation pathway, and yet, some of these relatively mature cells even acquired several phenotypic characteristics (Lin⁻Sca-1⁺c-Kit^HiThy1.1^LoRag-1⁻) of stem cells following exposure to Wnt3a. In functional assays, responsiveness to Wnt3a progressively declined with stages of lymphoid differentiation. These remarkable findings indicate that at least some early events in blood cell formation are reversible. It will be interesting to determine whether functional HSCs can be regenerated by manipulation of Wnt signals.

As recently reviewed by Graf and Stadtfeld (2008), HSC are heterogeneous and may oscillate through multiple states. For example, clonal analyses have revealed that individual stem cells exhibit lymphoid/myeloid biases that are heritable through multiple rounds of transplantation (Sieburg et al., 2006; Dykstra et al., 2007). As an example of a different type of heterogeneity, a continuum of HSC subsets was recently proposed to move from “reserved” to “primed” states (Haug et al., 2008). Changes in both directions were noted in this study by staining for expression of N-cadherin. Stem cell dormancy may also be distinguished by lack of CD34 and underappreciated because of experimental artifacts (Wilson et al., 2008). This reserved status might explain why subsets of human stem cells oscillate in terms of marker expression and require considerable time to expand following transplantation (McKenzie et al., 2006, 2007). Furthermore, the engraftment potential of murine HSC fluctuates as they move through the cell cycle (Habibian et al., 1998). We speculate that such reversible transitions are regulated in part by Wnt family molecules. In fact, the multipotent EMCL1 cell line was shown to spontaneously undergo fluctuations in gene expression and differentiation competence (Chang et al., 2008). Canonical Wnt pathway stimulation by introduction of active β-catenin blocked the myelopoietic potential of these cells (Baba et al., 2006). Similarly, manipulation of canonical Wnt signaling has been effectively used to maintain pluripotency of embryonic stem cell (ESC) lines and augment the reprogramming of fibroblasts to create induced ESCs (Ying et al., 2008; Marson et al., 2008).

Wnt proteins are likely to be present in both “osteoblastic” and “vascular” niches in bone marrow (Duncan et al., 2005; Zhang and Li, 2008; Wilson et al., 2007; Celso et al., 2008). However, the net biological activity in any location must be an extremely complex function of overlapping gradients of multiple Wnt ligands and secreted antagonists. Doses and combinations of Wnts, proximity to those molecules, receptors expressed, and cross-competition with other environmental cues are all certain to be important determinants of stem cell responses (Figure 2).

Concluding Remarks

The evidence is compelling that Wnts are required to maintain the integrity of HSC. However, induction of hematopoietic demand by serial transplantation or myeloablation has been required to clearly see abnormalities in HSC exposed to Wnt deficient environments. Thus, Wnts may operate in cooperation with many other components of stem cell niches and contribute most notably under conditions of disease, stress, or aging. On the other hand, at least some ligands for canonical Wnt signaling repress and even reverse differentiation of HSC and ESCs. As suggested above, refluxing between stem and progenitor cell states could be partially regulated by Wnts. Continued assault on this complex group of ligands, receptors, and coreceptors is certain to be useful as well as informative. For example, it may eventually be possible to expand and reprogram normal stem and progenitor cells by exploiting the Wnt family of proteins. Such an approach could improve the ability of HSCs to rebound following myeloablation or engraft following transplantation, and converse treatments may be found to deprive leukemia cells of essential Wnt stimulation.