Smoothening the Controversial Role of Hedgehog in Hematopoiesis

Abstract

The role of multiple developmental signaling pathways in the regulation of hematopoiesis has been controversial. In this issue of Cell Stem Cell, two separate reports (Hofmann et al., 2009; Gao et al., 2009) examine whether the Hedgehog signaling pathway modulates normal adult hematopoietic stem cells and blood formation.

The role of developmental signaling pathways such as Wnt, Notch, and Hedgehog (Hh) in adult tissue homeostasis and, in particular, the regulation of adult stem cells has been the subject of intense study. Hematopoiesis has served as a model system to study normal stem cells, and all three of these pathways have been found to affect the self-renewal and proliferation of hematopoietic stem cells (HSCs). However, these data have not been conclusive, and seemingly contradictory results exist in the literature. Now, two independent articles from the Gilliland and Aifantis laboratories report the unexpected finding that canonical Hh signaling is dispensable for normal adult hematopoiesis (Hofmann et al., 2009; Gao et al., 2009).

Hh signaling in mammals is initiated by the binding of one of three closely related ligands, Sonic Hh (Shh), Indian Hh (Ihh), and Desert Hh (Dhh), to the cell-surface receptor Patched (Ptch). Upon ligand binding, the inhibitory activity of Ptch on the positive transmembrane effector Smoothened (Smo) is lost and ultimately results in the modulation of activity of the three zinc finger transcription factors, Gli1, Gli2, and Gli3, at target promoters (Taipale and Beachy, 2001). Each component of this cascade is required for proper Hh signal transduction. In this issue, both Hofmann et al. (2009) and Gao et al. (2009) report that hematopoiesis, peripheral blood counts, colony formation in vitro, and stem and progenitor subset frequencies are all normal after conditional ablation of Smo. Furthermore, through competitive and serial bone marrow transplants, the two groups also demonstrate that Smo null HSCs exhibit no defects in homing, engraftment, and long-term self-renewal. Collectively, their findings indicate that Hh signaling is dispensable for hematopoietic function in mice.

In contrast, seemingly different results from previous reports have led to alternate conclusions. Initial studies reported that Hhligand induced the expansion of human HSCs in vitro (Bhardwaj et al., 2001) and is required for definitive hematopoiesis in zebrafish (Gering and Patient, 2005). A subsequent report found that HSCs with increased baseline Hh activity, modeled using Ptch heterozygote (Ptch^+/−) mice, displayed increased HSC proliferation resulting in eventual exhaustion, indicated by the loss of long-term engraftment after bone marrow transplantation (Trowbridge et al., 2006). However, using the same murine model, Dierks et al. recently reported enhanced engraftment of bone marrow from Ptch^+/− mice (Dierks et al., 2008). In the same paper, this group also reported no loss of long-term HSC engraftment in mice homozygous for a germline deletion of Smo. More recently, the opposite conclusion was drawn by Zhao et al., who observed a profound loss of HSC function after conditional deletion of Smo (Zhao et al., 2009).

Conflicting findings regarding the role of Hh in normal hematopoiesis could be due to interspecies differences or the specific component of the signaling pathway being modified, yet these explanations do not account for the variable results observed following Smo deletion in mice. Although germline loss of Smo is embryonically lethal, the mice live beyond the point at which definitive HSCs are specified in the fetal liver, which allows the cells to be removed and transplanted into adult hosts as carried out by Dierks et al. (2008). HSCs derived from Smo wild-type and null fetal liver exhibit identical function, proving that Smo is not required during the specification of definitive hematopoiesis in the embryo. Unlike Dierks et al., Zhao et al. and the two new papers from Gilliland and Aifantis all used conditional Smo alleles. Zhao et al. used the vav-Cre mouse to drive inactivation of Smo in hematopoietic tissues, given that the vav transcriptional elements are active in all hematopoietic and endothelial tissues from embryonic life onward. In contrast, both Hofmann et al. (2009) and Gao et al. (2009) used the interferon inducible Mx1-Cre allele to inactivate Smo in the hematopoietic tissues of adult mice. It is possible that Smo specifies certain HSC characteristics during embryonic development that have important consequences during adult life. This hypothesis might explain the results of Zhao et al. (2009), who observed a dramatic defect in HSC function, while the two current groups saw no differences after transplantation. However, if embryonic activity of Smo was critical for adult HSC function, one would expect that Dierks et al. would have also seen an HSC defect, since they used mice with a germline deletion of Smo.

Similarly discordant results have been reported following conditional deletion of β-catenin using vav-Cre or Mx1-Cre mice, where the former led to significant HSC defects and the latter revealed no functional deficits (reviewed in Malhotra and Kincade, 2009). These results suggest that the Mx1-Cre and vav-Cre models may not be comparable when studying HSC function. While both vav-Cre and Mx1-Cre appear to be active in the most primitive stem cell compartment, vav-Cre is also active in endothelial cells, which are found in close proximity to many primitive HSCs (reviewed in this issue by Garrett and Emerson, 2009). The two current papers from Gilliland and Aifantis show that Smo is not necessary for the cell-autonomous function of HSCs in mice, but it is possible that Hh may influence hematopoiesis through more-complicated cell-niche signaling interactions, as has been observed in other tissues (Takashima et al., 2008). It is likely that the inherent complexity of these pathways, promiscuous ligand receptor interactions, functional redundancy, and possible crosstalk between the signaling cascades also contribute to the range of experimental results obtained.

Similar to normal hematopoiesis, data regarding Hh signaling in various leukemias has also been inconsistent. Both Zhao et al. (2009) and Dierks et al. (2008) found that that Hh signaling is required for the development of a chronic myeloid leukemia-like disease induced by BCR-ABL. However, in the two reports presented in this issue, Smo was found to be dispensable for the development of acute leukemias induced by either the MLL-AF9 fusion gene or by an activated form of Notch-1. It is likely that the role of Hh in malignant hematopoiesis is highly contextual and dependent upon the specifics of the model being used. Furthermore, aberrant pathway activation, noncanonical signal transduction, and tumor-microenvironment interactions have all been observed, further complicating our understanding of Hh signaling in cancer.

A wide variety of human cancers display Hh pathway activity, and these findings have spurred the development of novel pathway inhibitors that antagonize Smo. Understanding Hh signaling in hematopoiesis will likely inform the optimal clinical use of these inhibitors in terms of both patient selection and the choice of other therapies to use in combination. However, it is also possible that the clinical use of potent SMO inhibitors will clarify the controversies surrounding the role of Hh in hematopoiesis, since their effects on human HSCs and progenitors can be evaluated directly in patients.