Fresh WNT into the regulation of mitosis

Abstract

Canonical Wnt signaling triggering β-catenin-dependent gene expression contributes to cell cycle progression, in particular at the G1/S transition. Recently, however, it became clear that the cell cycle can also feed back on Wnt signaling at the G2/M transition. This is illustrated by the fact that mitosis-specific cyclin-dependent kinases can phosphorylate the Wnt co-receptor LRP6 to prime the pathway for incoming Wnt signals when cells enter mitosis. In addition, there is accumulating evidence that various Wnt pathway components might exert additional, Wnt-independent functions that are important for proper regulation of mitosis. The importance of Wnt pathways during mitosis was most recently enforced by the discovery of Wnt signaling contributing to the stabilization of proteins other than β-catenin, specifically at G2/M and during mitosis. This Wnt-mediated stabilization of proteins, now referred to as Wnt/STOP, might on one hand contribute to maintaining a critical cell size required for cell division and, on the other hand, for the faithful execution of mitosis itself. In fact, most recently we have shown that Wnt/STOP is required for ensuring proper microtubule dynamics within mitotic spindles, which is pivotal for accurate chromosome segregation and for the maintenance of euploidy.

Introduction

Wnt signaling can be classified into canonical (β-catenin-dependent) and non-canonical (β-catenin-independent) pathways, each directed by specific cysteine-rich and lipid-modified secreted glycoproteins, called WNTs, and their transmembrane receptors and co-receptors. The β-catenin-independent Wnt-signaling pathways (the Planar Cell Polarity/PCP- and the WNT-Ca²⁺-pathway) are mainly implicated in remodeling of the cytoskeleton and in cell migration, whereas the β-catenin-dependent signaling pathway can function in embryonic development and in regulating cell proliferation.^1,2

The canonical Wnt pathway mediates its cellular responses by Wnt-dependent activation of the transcription factor β-catenin, which acts in concert with the LEF/TCF family of transcription factors to trigger gene expression. In the absence of Wnt ligands, cytoplasmic β-catenin is continuously degraded by a multi-protein destruction complex, which consists of the scaffolding proteins Axin1 and Axin2 (conductin), the tumor suppressor adenomatous polyposis coli (APC), the casein kinase 1 gamma (CK1γ) and the glycogen synthase kinase 3β (GSK3β). β-catenin is sequentially phosphorylated at Ser45 and Thr41, Ser37, and Ser33 by CK1γ and GSK3β, respectively. Phosphorylation of β-catenin results in its recognition by the β-transducin repeat containing protein (βTrCP) dependent E3 ligase and its subsequent degradation by the ubiquitin-proteasome pathway. In the presence of Wnt ligands, degradation of β-catenin is prevented and expression of its target genes is induced.^1,2 In detail, Wnt ligands bind to the 7-pass transmembrane receptor Frizzled (Fzd) and its co-receptor low-density lipoprotein receptor-related protein-6 (LRP6) or its close relative LRP5. The ternary complex forms clusters with the scaffolding protein disheveled (Dvl) to form endocytic LRP6-signalosomes.³ Following receptor-complex internalization and clustering into early endosomes, LRP6 is phosphorylated on its intracellular domain (ICD), at multiple PPPSPxS motifs by several proline-directed kinases. PPPSP-phosphorylation of LRP6 involves GSK3,⁴ which primes the co-receptor for subsequent phosphorylation of an adjacent Ser/Thr site via casein kinase 1γ (CK1γ), thereby triggering the recruitment of the Axin/APC/GSK3β-destruction complex to the LRP6-signalosome.^3,5 Further fusion of endocytosed signalosomes into multivesicular bodies (MVPs) leads to sequestration of GSK3β rendering the kinase unable to phosphorylate cytosolic β-catenin, which prevents its recognition by the βTrCP-ubiquitin ligase.^6,7 In addition to GSK3β sequestration, inhibition of GSK3β by the dual phosphorylated PPPSPxS motifs of LRP6 keeps the destruction complex inactive.⁸ Likewise, a conserved tyrosine-motif within the cytoplasmic tail of LRP6 has recently been reported to be phosphorylated by the non-receptor tyrosine kinases Src and Fer to decrease formation of LRP6-signalosomes, thereby inhibiting Wnt/β-catenin-signaling.⁹ Additional studies suggested an association of the protein phosphatases PP1 and PP2A with Axin1 or β-catenin, which might counteract GSK3β- and/or CK1-mediated phosphorylation, thereby promoting dissociation of the destruction complex and reduced β-catenin degradation.^10,11 As a result, β-catenin becomes stabilized leading to its accumulation and nuclear translocation where it can trigger gene expression. Among relevant target genes, which are expressed upon Wnt signaling are MYC (encoding for c-myc) and CCND1 (encoding for cyclin D) indicating that Wnt signaling can directly influence the cell cycle progression, in particular at the G1/S transition.^1,2

Additional functions of Wnt pathway components during mitosis

In addition to regulating gene transcription, recent studies have illustrated an additional layer of how Wnt pathway components might regulate the cell cycle. In fact, several Wnt-signaling components were shown to be involved in the regulation of mitosis when gene transcription is suppressed.¹² In particular, mitotic centrosomes and spindles are predominant structures where various Wnt components, such as Dvl, Axin-1, Axin-2/conductin, GSK3β, β-catenin, and APC are localized and where they might exert specific functions. For instance, Dvl was shown to localize to centrosomes during mitosis and to the mitotic spindle where it can interact with the key mitotic kinase Plk1 to function in the regulation of spindle orientation and kinetochore-microtubule attachment.¹³ Axin-1 and Axin-2 also have been localized to centrosomes. Axin-1 might play a role for microtubule nucleation and microtubule stabilization,^14,15 whereas Axin-2 appears to be involved in the regulation of centrosome cohesion.¹⁶ Furthermore, GSK3 has been implicated in the regulation of microtubule dynamics and is required for proper chromosome alignment.^17,18 β-catenin also localizes to mitotic centrosomes and might be required for proper mitotic spindle assembly and timely centrosome separation.^19-21 Whether all these functions, which are still ill defined, are regulated by Wnt-signaling is currently not clear. However, at least for Dvl´s function in spindle orientation, a dependence on LRP6 and Fzd was demonstrated, suggesting that Wnt-dependent signaling (or Wnt/STOP, see below) might be involved in the regulation of spindle orientation during mitosis.¹³

A particular important function in mitosis, which is likely independent of WNTs is fulfilled by APC, which can act as a microtubule plus end binding protein.^22,23 Interestingly, APC interacts with the microtubule plus end binding protein EB1 and localizes preferentially to kinetochore microtubules (k-fibers ) where it might contribute to the establishment of proper kinetochore-microtubule interactions.^24,25 Evidence also exists showing that APC, together with EB1, might control some parameters of microtubule plus end dynamics,²⁴ but microtubule plus end growth is not influenced by EB1 and APC per se.²⁶ In addition to that, APC can also form complexes with spindle assembly checkpoint proteins such as Bub1 and Bub3 suggesting that APC might also be involved in the regulation of the mitotic spindle assembly checkpoint.²⁵ However, details about this possible regulation have not been reported yet. Nevertheless, it seems clear that APC fulfills Wnt-independent functions during mitosis required for faithful chromosome segregation and this is supported by the fact that cancer-associated truncations of APC severely interfere with chromosome alignment and result in chromosome missegregation and chromosomal instability.^27,28

Cell cycle regulation of Wnt signaling

As already mentioned above, Wnt- and β-catenin-mediated gene transcription requires phosphorylation of the 5 PPPSPxS motifs in the ICD of LRP6.^4,5 Interestingly, phosphorylation of Ser1490 within the PPPSP-motif was shown to peak at the G2/M transition of the cell cycle and is mediated by the mitosis-specific CDK14/CyclinY kinase complex.²⁹ In fact, CDK14/cyclin Y-mediated phosphorylation primes LRP6 in mitosis for subsequent CK1γ-mediated phosphorylation and thereby, sensitizes cells for incoming Wnt signals. These recent findings suggest that Wnt signaling is under cell cycle control peaking when cells enter mitosis. Interestingly, this is in line with the fact that Wnt regulated target proteins such as β-catenin and its transcriptional target Axin2/conductin also show highest expression at G2/M and during mitosis.^29-32 This, in turn, raises the question why Wnt signaling peaks in mitosis when the transcription machinery is shut off.

Introduction of Wnt-mediated stabilization of proteins (Wnt/STOP)

A plausible answer to this question came recently when it was recognized that Wnt signaling not only mediates the stabilization of β-catenin leading to an induction of gene expression, but also of a large number of additional proteins unrelated to transcription. In fact, bioinformatics analyses have shown that up to 20% of the proteome contain GSK3 phospho-degrons required for subsequent ubiquitin-dependent proteasomal degradation. Thus, Wnt signaling associated with inhibition of GSK3β might potentially stabilize all of these proteins.^6,33 Interestingly, most recently it has been demonstrated that Wnt signaling in the presence of exogenous Wnt3a can cause rapid and GSK3β-dependent accumulation of poly-ubiquitinated proteins into endolysosomes in interphase providing a new link between poly-ubiquitination and the lysosomal degradation pathway. Moreover, sequestering of poly-ubiquitinated proteins into vesicles might contribute to a transient depletion of free ubiquitin, which, in turn, might influence protein stability of other proteins.³⁴ Along this line, Acebron et al. recently proposed that Wnt-dependent stabilization of proteins, referred to as Wnt/STOP, is restricted to the G2 and M phase of the cell cycle and occurs independently of β-catenin-mediated gene transcription.³⁵ Indeed, treatment with Wnt3a was shown to stabilize a subset of proteins specifically at G2/M. Vice versa, inhibition of Wnt signaling, e.g. after treatment with Dkk1, caused ubiquitin-proteasome-mediated proteolysis of various proteins containing Gsk3β−degron motifs.³⁵ Since Wnt/STOP might affect a large number of proteins it was suggested that it could contribute to the maintenance of a critical cell size essential for cell division in mitosis. However, measurable differences in Wnt-dependent cell sizes were not immediately obvious at the G2/M transition when cells enter mitosis, but only observed upon prolonged treatment with Wnt3a.^34,35

Since canonical Wnt signaling and Wnt/STOP seem to employ the same set of signaling components, it is difficult to dissect these 2 modes of action. One way to do this is the analysis of Xenopus laevis oocytes, which are naturally arrested in metaphase II and thus, transcriptionally silenced. Indeed, Xenopus oocytes show high levels of LRP6 phosphorylation indicative for active Wnt signaling and reduced polyubiquitination and stabilization of putative GSK3β target proteins.³⁶ Additional experiments showed that Wnt/STOP in oocytes is required for subsequent mitotic cell divisions upon fertilization.³⁶ Thus, Wnt/STOP appears to be active in metaphase II arrested Xenopus oocytes in the absence of any gene transcription and involved in stabilization of proteins involved in cell division.

A key mitotic function of Wnt/STOP in human somatic cells

Analyses of the role of Wnt/STOP in human somatic cells is difficult since β-catenin-mediated Wnt signaling, which is usually present in those cells, might interfere with Wnt/STOP. Thus, in order to rule out an involvement of gene transcription it is required to analyze exclusively mitotic cells, in which transcription is turned off. Our most recent work has employed analyses of single mitotic cells and demonstrated that Wnt/STOP is required for the maintenance of proper microtubule plus end dynamics during a normal mitosis.³⁷ In fact, we have previously shown that abnormally increased microtubule plus end assembly rates within mitotic spindles can cause chromosome missegregation by inducing transient spindle geometry and positioning defects.³⁸ This was found to be a wide-spread phenotype in chromosomally instable cancer cells, which are characterized by an increased rate of chromosome missegregation during mitosis. Upon inhibition of basal Wnt signaling in mitosis, e.g., upon repression of LRP5/6 or DVL or upon treatment with Dkk1 or sFRPs, we detected an increase in microtubule growth rates in living mitotic cells. This, in turn, led to abnormal spindle formation and resulted in the generation of so-called lagging chromosomes during anaphase, which are the result of erroneous microtubule-kinetochore attachments. Importantly, spindle malfunction as well as chromosome missegregation were suppressed when normal microtubule dynamics were restored indicating that increased microtubule plus end growth is indeed responsible for the mitotic defects.^37,38 Thus, the inhibition of basal mitotic Wnt signaling induces an increase in microtubule plus end dynamics, which results in chromosome missegregation and aneuploidy (Fig. 1). Importantly, neither mitotic defects nor chromosome missegregation were observed upon loss of β-catenin indicating an involvement of β-catenin-independent Wnt signaling in the regulation of mitotic microtubule dynamics. In fact, proof for a requirement of Wnt-mediated stabilization of proteins in the regulation of mitotic microtubule dynamics came from additional experiments, in which we restored normal microtubule plus end assembly and faithful chromosome segregation by inhibition of the proteasome or by co-depletion of components of the Wnt-destruction complex including Axin-1 and APC. Both means clearly suppressed the abnormally increased microtubule plus end assembly rates and restored faithful chromosome segregation demonstrating that Wnt/STOP is indeed required for the proper regulation of microtubule plus end assembly and faithful chromosome segregation (Fig. 1).³⁷ It is important to note that short-term proteasome inhibition in mitotic cells was shown to be sufficient to restore proper microtubule growth rates, which indicates that protein degradation in response to Wnt/STOP inhibition occurs specifically during mitosis.³⁷ In sum, our recent results demonstrate that Wnt/STOP, which requires basal Wnt signaling, rather than canonical and β-catenin-dependent Wnt signaling, is essential for the faithful execution of mitosis and for the maintenance of euploidy by controlling mitotic microtubule assembly. Although this is the first defined mitotic function of Wnt/STOP reported, it seems likely that Wnt/STOP might control also other regulatory steps during mitotic progression. These new roles of Wnt/STOP remain to be elucidated.

Figure 1.

Schematic view on the role of Wnt pathways in the regulation of the cell cycle. Canonical Wnt signaling resulting in accumulation of β-catenin contributes to the G1/S progression by supporting the expression of key cell cycle regulators such as c-myc or cyclin D. In contrast, Wnt/STOP resulting in the stabilization of various proteins other than β-catenin is required for the G2/M transition and during mitosis. It can contribute to the maintenance of a critical cell size required for mitotic cell division and is also required for specific functions during mitosis. One key function of Wnt/STOP is to ensure proper microtubule plus end assembly, which is pivotal for accurate mitotic spindle assembly and function and essential for faithful chromosome segregation during mitosis. Additional requirements for Wnt/STOP for the progression of mitosis await to be uncovered.

Outlook

The layers of cell cycle regulation provided by different Wnt signaling pathways are highly complex. In addition to canonical Wnt signaling, which regulates the G1/S transition of the cell cycle by driving β-catenin-dependent gene expression, Wnt/STOP regulates the G2/M transition and mitosis by specifically stabilizing proteins. In addition, also non-canonical Wnt pathways might contribute to cell cycle regulation by affecting cytoskeletal organization and a further layer of regulation might be provided by significant cross-talks of the different Wnt signaling pathways during the cell cycle.

For the regulation of mitosis, Wnt/STOP might be the major regulatory Wnt signaling pathway. It is expected that Wnt/STOP specifically stabilizes proteins at G2/M that are otherwise targeted for degradation after GSK3β-mediated phosphorylation. Significantly, large numbers of proteins have been identified that are subject to polyubiquitination in a GSK3β-dependent manner in vitro.^6,33,35,36 Whether all these proteins are indeed stabilized in response to Wnt/STOP in a cellular context remains to be determined. It will be of particular importance to identify the relevant subset of proteins, which are stabilized by Wnt/STOP and which are physiologically important to ensure proper progression of mitosis. Based on our results, those proteins would also include key regulators of mitotic microtubule plus end assembly, whose depletion after loss of Wnt/STOP results in the generation of aneuploidy. Since aneuploidy is often associated with diseases such as cancer or neurodegenerative diseases,^39,40 one might speculate that impairment of Wnt/STOP might also be disease-relevant. So far, there is no indication that Wnt/STOP is impaired in human cancer, but it is premature to exclude this possibility, since the pathway is yet not fully elucidated and the list of signaling components required for fully active Wnt/STOP is not complete. However, it is interesting to note that deficiency of LRP6-mediated Wnt signaling was recently shown to contribute to the pathology of Alzheimer´s disease, where aneuploidy is highly prevalent.⁴¹ Whether aneuploidy in cells from Alzheimer brains is a consequence of loss of Wnt/STOP will be important to be investigated.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Gerald Wulf and Heike Krebber for critically reading the manuscript.

Funding

This work was supported by the Deutsche Forschungsmeinschaft (DFG) and by the FOR942.