Microtubules are dynamic structures that are crucially involved in a variety of cellular activities. The dynamic properties and functions of microtubules are regulated by various factors, such as tubulin isotype composition and microtubule-binding proteins. Initially identified as a deubiquitylase with tumor-suppressing functions, the protein cylindromatosis (CYLD) has recently been revealed to interact with microtubules, modulate microtubule dynamics, and participate in the regulation of cell migration, cell cycle progression, chemotherapeutic drug sensitivity and ciliogenesis. These findings have greatly enriched our understanding of the roles of CYLD in physiological and pathological conditions. Here, we focus on recent literature that shows how CYLD impacts on microtubule properties and functions in various biological processes, and discuss the challenges we face when interpreting results obtained from different experimental systems.
Microtubules are filamentous cellular structures composed of α-tubulin–β-tubulin heterodimers and are responsible for diverse cellular activities, such as cell shaping, intracellular trafficking, cell division and cell motility. Microtubules can constantly switch between growth and shortening phases, a behavior termed dynamic instability. The dynamic properties of microtubules must be tightly regulated by various factors, such as tubulin isotype composition and microtubule-binding proteins (MBPs), to properly function in cellular processes. Canonical MBPs include microtubule-associated proteins (MAPs), microtubule-destabilizing proteins, microtubule plus-end-tracking proteins (+TIPs), microtubule minus-end-binding proteins, microtubule inner proteins (MIPs) and microtubule-dependent molecular motors. In addition, recent studies have illuminated an expanding list of signaling proteins that can bind to and modulate microtubules. One such MBP is the deubiquitylase cylindromatosis (CYLD).
Cyld was first identified as a gene associated with familial cylindromatosis, a condition involving multiple skin tumors that result from Cyld germline mutations (usually nonsense or missense mutations) associated with somatic mutations in dermal cells (loss of heterozygosity) (Bignell et al., 2000). This gene encodes a protein containing three N-terminal cytoskeleton-associated protein glycine-rich (CAP-Gly) domains and a C-terminal ubiquitin-specific protease (USP) domain (Fig. 1). As most of the tumor-associated mutations are found in the catalytic USP domain (Bignell et al., 2000; Massoumi, 2011), previous studies have focused primarily on the functions of its deubiquitylase activity. The crystal structure of the USP domain and results from in vitro deubiquitylation assays have demonstrated that CYLD has lysine-63 (K63)-linked polyubiquitin-chain-specific endodeubiquitylase activity (Komander et al., 2008, 2009; Sato et al., 2015). A series of studies has found that CYLD regulates multiple signaling pathways, including nuclear factor-кB (NF-кB), Wnt/β-catenin, c-Jun N-terminal kinase (JNK), p38 mitogen-activated protein kinase (p38 MAPK), transforming growth factor-β (TGF-β), Hippo and Notch signaling (Chen et al., 2014; Lim et al., 2012; Massoumi, 2010; Rajan et al., 2014; Welte et al., 2014; Zhao et al., 2011) (Fig. 1). Here, CYLD activity appears to regulate the assembly of signaling protein complexes, consistent with the findings that K63-linked polyubiquitylation primarily affects protein–protein interactions and protein localization (Komander and Rape, 2012).
Subsequent studies have suggested that CYLD might have additional functions that are mediated by parts of the protein outside of the catalytic domain. Loss of full-length CYLD frequently leads to different, or even contradictory, results compared to when only its deubiquitylase activity was inhibited. Moreover, Cyld-knockout mice have strikingly different phenotypes than transgenic mice that carry truncations in the C-terminal deubiquitylase domain. For example, Cyld-knockout mice are viable, whereas C-terminal truncations of CYLD lead to embryonic lethality (Eguether et al., 2014; Jin et al., 2008; Lim et al., 2007; Massoumi et al., 2006; Reiley et al., 2006; Trompouki et al., 2009; Wright et al., 2007; Yang et al., 2014b; Zhang et al., 2006). In agreement with these findings, several studies have shown that the N-terminal CAP-Gly domains of CYLD mediate its interaction with microtubules and that these domains are required for its physiological and pathological functions. In this Commentary, we discuss recent studies that highlight the role of CYLD in the regulation of microtubule dynamics and microtubule-related cellular activities.
CYLD interacts with microtubules
CAP-Gly domains are conserved regions implicated in numerous cellular processes and human diseases. They primarily mediate protein–protein interactions by binding to end-binding homology (EBH) domains, zinc-finger motifs, C-terminal glutamate-glutamate-tyrosine/phenylalanine (EEY/F) sequence motifs, and proline-rich sequences (Steinmetz and Akhmanova, 2008). EBH domains are found in end-binding proteins, and C-terminal EEY/F sequence motifs are present in α-tubulin and some MBPs, such as end-binding proteins and cytoplasmic linker protein 170 (CLIP-170, also known as CLIP1) (Steinmetz and Akhmanova, 2008). Thus, the presence of these N-terminal CAP-Gly domains suggests that CYLD might participate in microtubule-related activities by directly binding to microtubules or by interacting with other MBPs.
By using purified proteins, CYLD and its first two CAP-Gly domains have been shown to interact directly with tubulin and microtubules, and the first CAP-Gly domain exhibits the highest binding affinity (Gao et al., 2008; Wickström et al., 2010). A subsequent study has revealed that pretreatment of in-vitro-assembled microtubules with subtilisin (also known as PCSK9), a serine protease that removes the C-terminal tails of tubulin, blocks the interaction of purified CYLD with microtubules, suggesting that CYLD binds to the C-terminal tails of tubulin (Yang et al., 2015). However, it remains to be determined whether the C-terminal EEY/F sequence motifs of tubulin directly interact with the first two CAP-Gly domains of CYLD.
It is worth noting that the microtubule-binding affinity of CYLD might be lower than that of canonical MBPs. Two teams have found that CYLD colocalizes with microtubules (Gao et al., 2008; Wickström et al., 2010), whereas other teams have shown that CYLD localizes primarily to the cytoplasm and spindle poles and only weakly colocalizes with microtubules (Eguether et al., 2014; Gomez-Ferreria et al., 2012; Urbe et al., 2012). Furthermore, Weisbrich et al. have demonstrated that the conserved glycine-lysine-asparagine-aspartate-glycine (GKNDG) motifs of CAP-Gly domains are responsible for their binding to the C-terminal EEY/F sequence motifs (Weisbrich et al., 2007). However, the three CAP-Gly domains of CYLD possess divergent GKNDG motifs, which are glycine-phenylalanine-threonine-aspartate-glycine (GFTDG), glycine-asparagine-tryptophan-aspartate-glycine (GNWDG) and glycine-cysteine-threonine-aspartate-glycine (GCTDG), respectively. This might contribute to the low microtubule-binding affinity of CYLD.
Interestingly, CYLD has been demonstrated to bind to EB1 (also known as MAPRE1) (Li et al., 2014). Although EB1 contains a C-terminal EEY/F sequence motif, which is known to mediate binding of EB1 to CAP-Gly domains (Weisbrich et al., 2007), the interaction of EB1 with CYLD is in fact mediated by its deubiquitylase domain, and not any of its CAP-Gly domains (Li et al., 2014) (Fig. 1). The lack of an interaction between EB1 and the CAP-Gly domains of CYLD might also result from changes in the amino acids within the GKNDG motifs of the three CAP-Gly domains of CYLD. CYLD has also been shown to interact, through its first CAP-Gly domain, with histone deacetylase 6 (HDAC6) (Wickström et al., 2010) (Fig. 1), a microtubule-associated enzyme that deacetylates a variety of proteins including tubulin (Hubbert et al., 2002; Liu et al., 2015; Matsuyama et al., 2002; Zhang et al., 2003). Taken together, the above findings suggest that CYLD might interact with microtubules also indirectly, by binding to MBPs, such as EB1 and HDAC6, in addition to its direct binding to microtubules (Fig. 1).
CYLD regulates the dynamic properties of microtubules
Our studies have examined the functional consequences of the CYLD–microtubule interaction. In an in vitro tubulin polymerization assay, addition of CYLD lowers the critical concentration of tubulin required for the nucleation of microtubule assembly (Gao et al., 2008). Moreover, cold treatment of microtubules polymerized in vitro in the presence of CYLD has revealed that CYLD enhances microtubule stability and decreases the rate of cold-induced microtubule depolymerization (Gao et al., 2008). Results from these in vitro experiments suggest that CYLD promotes microtubule assembly and enhances microtubule stability through a direct interaction.
However, the in vivo regulation of microtubules by CYLD appears to be more complicated than in vitro studies suggest. Silencing of CYLD in human umbilical vein endothelial cells (HUVECs) dramatically inhibits microtubule dynamics, which is evident from the increased time microtubules spend in a paused state and the decreased rates of microtubule growth and shortening (Gao et al., 2010). In keratinocytes, CYLD inhibits the tubulin deacetylase activity of HDAC6, resulting in elevated levels of α-tubulin acetylation. Both in vivo and in vitro assays have revealed that the N-terminal region of CYLD (amino acids 1–212, containing the first CAP-Gly domain) is solely responsible for the inhibition of HDAC6 activity (Wickström et al., 2010). These findings indicate that, besides increasing the binding of α-tubulin–β-tubulin heterodimers to the microtubule polymers (Gao et al., 2008), CYLD can further enhance microtubule stability by inhibiting the deacetylase activity of HDAC6 and elevating the level of acetylated α-tubulin (Wickström et al., 2010) (Fig. 1). Another study in migrating cells has shown that CYLD regulates microtubule assembly and dynamics through an interaction with EB1 (Li et al., 2014) (Fig. 1). Moreover, the synergistic effects of CYLD and EB1 at the distal ends of microtubules are also likely to contribute to the regulation of microtubule dynamics (Li et al., 2014). In addition to these known interactions, other MBPs could also be involved in CYLD-mediated regulation of microtubule dynamics.
The finding that a fraction of CYLD localizes to centrosomes and basal bodies suggests that, in addition to regulating microtubule dynamics as is mentioned above, CYLD might regulate certain cellular events at centrosomes and basal bodies, such as the organization of surrounding microtubules (Eguether et al., 2014; Gomez-Ferreria et al., 2012; Urbe et al., 2012). Interestingly, CYLD also localizes to the midbody, another complex microtubule-based structure (Stegmeier et al., 2007; Wickström et al., 2010). Midbodies contain many proteins in addition to microtubules, and it remains unknown whether CYLD localizes to the midbody in a microtubule-dependent manner. Several studies have shown that CYLD localizes to the plus ends of microtubules, including growing microtubules and ciliary tips (Eguether et al., 2014; Li et al., 2014; Sun et al., 2010). Thus, CYLD could function by modulating protein complexes at microtubule plus ends to regulate interactions between microtubules and other cellular structures. This is possible given that CLIP-170, another CAP-Gly domain-containing protein, localizes to microtubule plus ends and connects microtubules to the cell cortex and kinetochores (Gouveia and Akhmanova, 2010). In addition, CYLD is enriched in the postsynaptic densities of neurons (Dosemeci et al., 2013; Thein et al., 2014a,b). Whether the localization and function of CYLD within these regions are dependent on microtubules remains to be investigated.
CYLD plays an important role in cell migration
Cell migration refers to the movement of cells in specific directions in response to external signals. Dysregulated cell migration is associated with impaired wound healing, compromised immune responses, vascular diseases and tumor metastasis. CYLD has been implicated in cell migration mechanisms associated with several physiological processes. Small interfering RNA (siRNA)-mediated inhibition of CYLD expression in HUVECs slows cell migration during in vitro wound healing (Gao et al., 2010). A three-dimensional capillary sprouting assay and an in vivo angioreactor implantation assay in mice has revealed that CYLD is required for angiogenesis, the generation of new blood vessels from preexisting ones (Gao et al., 2010). Mechanistic studies have indicated that CYLD regulates microtubule dynamics and promotes cell polarization, a crucial step during cell migration (Gao et al., 2010) (Fig. 2). CYLD also stimulates the activation of Rac1, a member of the Rho family of small GTPases, that promotes membrane ruffling and cell spreading during cell migration (Fig. 2). The above activities might contribute to the role of CYLD in mediating vascular endothelial cell migration and angiogenesis (Gao et al., 2010). CYLD also promotes the activation of RhoA, another small Rho GTPase involved in the regulation of tail retraction during cell migration, through deubiquitylation and activation of leukemia-associated Rho guanine-nucleotide exchange factor (LARG; also known as ARHGEF12) (Yang et al., 2013) (Fig. 2). In addition, the interaction between CYLD and EB1 has been shown to coordinate microtubule dynamics during cell migration, and disruption of this interaction impairs microtubule stabilization at the leading edge, as well as tail retraction (Li et al., 2014) (Fig. 2). Other CYLD-interacting MBPs, such as HDAC6, might also be involved in CYLD-mediated regulation of cell migration (Fig. 2).
CYLD also regulates cell migration in tumor cells. Knockdown of CYLD in melanoma cells has been shown to decrease cell migration in an in vitro wound healing assay (Ishikawa et al., 2012). However, contradictory results have been reported. For example, exogenous expression of CYLD in melanoma cells inhibits cell growth, cell migration and tumor metastasis by decreasing the expression of cyclin D1 and N-cadherin and increasing the expression of E-cadherin (Ke et al., 2013; Massoumi et al., 2009). In addition, transgenic mice expressing a truncated form of CYLD that is catalytically deficient under control of the epidermal keratin 14 promoter are prone to chemically induced skin tumors that frequently metastasize to the lymph node through a JNK- and activator protein 1-dependent pathway, suggesting that CYLD inhibits tumor cell migration in a deubiquitylase-activity-dependent manner (Miliani de Marval et al., 2011). These apparent discrepancies with regard to the role of CYLD in cell migration could be attributable to the different experimental systems and techniques used in these studies. Although knockdown experiments suggest that endogenous CYLD is required for cell migration, overexpression experiments indicate that high levels of CYLD can suppress cell migration. Further studies are warranted to elucidate the precise mechanisms of how CYLD participates in cell migration.
It is worth noting that whether and how CYLD participates in the regulation of immune cell motility has not yet been reported, although the role of CYLD in immune system development, and immune and inflammatory responses has already been extensively studied (Mathis et al., 2015; Sun, 2010). Loss of CYLD has also been demonstrated to stimulate the formation of inflammation-associated cancers, such as colorectal, skin and liver cancers (Massoumi et al., 2006; Nikolaou et al., 2012; Zhang et al., 2006). The regulation of cell motility by CYLD might represent a potential mechanism that underlies some of its functions in immunity and inflammation.
CYLD regulates cell cycle progression
Given that CYLD is a well-established tumor suppressor, several groups have examined its role in cell cycle progression. A study in melanoma cells has demonstrated that CYLD overexpression induces a delay in the G1/S transition (Wickström et al., 2010). Interestingly, this effect requires both its deubiquitylase activity and the N-terminal region containing the CAP-Gly domains (Wickström et al., 2010). The deubiquitylase activity of CYLD is required for the inhibition of cyclin D1 expression (Massoumi et al., 2009), which might explain its role in the G1/S transition (Fig. 3). Moreover, it has been shown that growth-promoting signals, such as 12-O-tetradecanoylphorbol-13-acetate (TPA) and ultraviolet light, relocalize CYLD from the cytoplasm to the perinuclear region, in a manner that is possibly mediated by its inhibitory interaction with HDAC6, resulting in an increase in the level of acetylated microtubules around the nucleus (Wickström et al., 2010). This relocalization of CYLD promotes its binding to its substrate B-cell lymphoma 3 (Bcl-3), and CYLD-mediated deubiquitylation of Bcl-3 prevents its nuclear translocation and activation; both events are necessary for the G1/S progression as they promote cyclin D1 transcription (Massoumi et al., 2006).
CYLD has also been implicated in mitosis to act as a regulator of mitotic entry, spindle formation and orientation, and cytokinesis (Fig. 3). CYLD has been shown to be upregulated during mitosis (Stegmeier et al., 2007; Yang et al., 2014a). In HeLa cells, CYLD depletion delays mitotic entry in a deubiquitylase-activity-dependent manner, likely due to decreased activation of polo-like kinase 1 (PLK1) (Stegmeier et al., 2007) (Fig. 3). Centrosomal protein of 192 kDa (Cep192) has been shown to inhibit the deubiquitylase activity of CYLD to ensure bipolar spindle formation, suggesting that CYLD has an inhibitory effect on spindle assembly (Gomez-Ferreria et al., 2012) (Fig. 3). Depletion of CYLD leads to spindle misorientation due to defects in the stability of astral microtubules and formation of the dishevelled (DVL), nuclear mitotic apparatus protein (NuMA) and dynein–dynactin complex at the cell cortex (Yang et al., 2014a) (Fig. 3). Loss of CYLD also leads to the hyperactivation of Aurora B and accompanying defects in chromosome segregation and cytokinesis (Fig. 3). Here, CYLD depletion disrupts the binding of Aurora B to the protein phosphatase-2A (PP2A) complex, which is responsible for dephosphorylating Aurora B and inhibiting its activity (Sun et al., 2010). Consistent with these observations, knockdown of CYLD results in defects in chromosome segregation and cytokinesis due to decreased activation of RhoA (Yang et al., 2013) (Fig. 3). Interestingly, overexpression of CYLD also causes failure of cytokinesis in HeLa cells and human osteosarcoma U2OS cells, which is evident from the accumulation of multinucleated cells (Stegmeier et al., 2007; Sun et al., 2010). Similarly, CYLD overexpression induces a delay in cytokinesis in keratinocytes, and this delay is almost completely dependent on the CAP-Gly domains of CYLD (Wickström et al., 2010).
Taken together, these results suggest that CYLD has important roles throughout the cell cycle. Moreover, the microtubule-regulatory activity of CYLD appears to be essential for its functions in cell cycle progression, especially in the regulation of the G1/S transition, mitotic spindle assembly and orientation, and cytokinesis (Fig. 3).
CYLD modulates the efficacy of microtubule-targeted drugs
Rapid proliferation of tumor cells is highly dependent on microtubule dynamics, which are essential for cell division. Because of this requirement, microtubule-targeted agents, such as paclitaxel, which stimulates microtubule polymerization and stabilizes microtubules, and vinblastine, which inhibits microtubule polymerization, are widely used as cancer chemotherapeutics (Dumontet and Jordan, 2010). However, the effectiveness of these drugs is seriously compromised by the variable drug sensitivity of tumor cells. Accumulating evidence suggests that the sensitivity of tumor cells to microtubule-targeted drugs can be modulated through synergic or antagonistic effects that result from alterations in specific MBPs (Bauer et al., 2010; Luo et al., 2014; Mohan et al., 2013; Rouzier et al., 2005; Sun et al., 2012; Wang et al., 2009; Xie et al., 2016). For example, elevation of CLIP-170 expression in breast cancer cells promotes paclitaxel sensitivity by increasing its binding to microtubules (Sun et al., 2012). By contrast, downregulation of tau, another MBP, enhances the sensitivity of breast cancer cells to paclitaxel, because microtubules formed in the presence of tau bind less paclitaxel (Rouzier et al., 2005).
Cyld gene deletion or mutation, downregulation of CYLD protein and inactivation of its deubiquitylase activity have been reported in multiple human cancers. Loss of CYLD function has also been linked to increased malignancy of tumors (Massoumi, 2011). Thus, understanding the role of CYLD in the efficiency of chemotherapy will benefit cancer patients. For example, inhibition of NF-кB signaling reverses the sensitivity of HeLa cells to tumor necrosis factor-α (TNF-α) and cycloheximide (CHX)-induced apoptosis, suggesting that CYLD-mediated regulation of NF-кB signaling might affect the sensitivity of tumor cells to apoptosis-inducing chemotherapeutic agents (Brummelkamp et al., 2003). However, another study performed with primary mouse embryonic fibroblasts that had been isolated from Cyld-knockout and wild-type mice showed no difference in TNF-α or CHX-induced apoptosis (Liang et al., 2011), suggesting that the effect of CYLD in sensitizing cells to apoptosis might be cell type specific.
The role of CYLD in microtubule regulation suggests that CYLD might affect the sensitivity of cancer cells to microtubule-targeted drugs. We have evaluated the sensitivity of acute lymphoblastic leukemia (ALL) cells to noscapine, a microtubule-targeted agent currently under investigation in phase I and II clinical trials for the treatment of leukemia and lymphoma (Rida et al., 2015). Our results showed that CYLD promotes noscapine-induced mitotic arrest and subsequent apoptosis in ALL cells (Yang et al., 2015). Deleting the first CAP-Gly domain of CYLD abolishes this effect, suggesting that CYLD regulates noscapine sensitivity in a microtubule-dependent manner. However, CYLD expression does not correlate with sensitivity towards paclitaxel or vinblastine (Yang et al., 2015). Indeed, our subsequent experiments showed that CYLD promotes the binding of noscapine, but not of paclitaxel or vinblastine, to microtubules (Yang et al., 2015). It is likely that CYLD exerts structural and/or allosteric effects on microtubules, and these changes in microtubule structure might modulate noscapine binding. Analyzing the effects of CYLD on the sensitivity of microtubule-targeted cancer chemotherapeutic drugs might help to improve therapeutic outcomes.
CYLD is crucially involved in ciliogenesis
Cilia are microtubule-based protrusions present on the surface of most eukaryotic cells and have a variety of physiological functions. Defects in ciliary assembly or function are associated with a wide spectrum of diseases, collectively referred to as ciliopathies (Yu et al., 2016). Intriguingly, several symptoms observed in ciliopathies, including male infertility, impaired lung maturation and osteoporosis, have been observed in Cyld-knockout mice or in mice expressing truncated CYLD (Jin et al., 2008; Trompouki et al., 2009; Wright et al., 2007). Recently, two independent studies using different mouse models have shown that CYLD is required for ciliogenesis. We have found that Cyld-knockout mice exhibit polydactyly, a cilium-associated symptom (Malik, 2014), and defective ciliogenesis in multiple organs, including the skin, kidney, trachea and testis (Yang et al., 2014b). Transmission electron microscopy showed that CYLD is required for the anchorage of the basal body and proper organization of the basal body and axoneme (Fig. 4). Our further analysis revealed that CYLD deubiquitylates centrosomal protein of 70 kDa (Cep70) and thereby increases Cep70 localization at the basal body, promoting basal body organization and anchorage to the plasma membrane through unclear mechanisms (Fig. 4). In addition, we found that CYLD-mediated inactivation of HDAC6 enhances tubulin acetylation and thereby stabilizes axonemal microtubules, facilitating axoneme organization (Yang et al., 2014b) (Fig. 4). Consistent with this observation, pharmacological inhibition of HDAC6 activity partially rescues the ciliogenesis defects in Cyld-knockout mice and cultured cells (Ran et al., 2015; Yang et al., 2014b).
The second study also showed that CYLD is essential for basal body anchorage during ciliogenesis (Eguether et al., 2014). Here, centrosome-associated protein 350 (CAP350) was found to be required to localize CYLD to centrosomes and basal bodies through an interaction with the C-terminal domain of CYLD (Fig. 4). Importantly, a disease-associated CYLD truncation that lacks the last 24 C-terminal amino acid residues lost its ability to bind CAP350 and to mediate ciliogenesis in ependymal cells (Eguether et al., 2014). Taken together, these findings suggest that both the deubiquitylase activity and the microtubule-targeting of CYLD are essential for its role in promoting ciliogenesis. In this context, the deubiquitylase activity of CYLD appears to be important for regulating basal bodies and docking them to the plasma membrane, whereas the role of CYLD in regulating microtubules is primarily required for axonemal organization (Fig. 4).
It is intriguing that Cyld-knockout mice and mice carrying CYLD truncations exhibit different phenotypes (Eguether et al., 2014; Jin et al., 2008; Lim et al., 2007; Massoumi et al., 2006; Reiley et al., 2006; Trompouki et al., 2009; Wright et al., 2007; Yang et al., 2014b; Zhang et al., 2006). All of the Cyld-knockout mouse lines reported so far are viable. The Cyld-knockout mouse line generated in a C57BL/6×DBA/2 mixed genetic background shows multiple developmental defects, including ciliopathies associated with several organs (Yang et al., 2014b), but those generated in a C57BL/6 genetic background are completely normal without any major phenotype (Lim et al., 2007; Massoumi et al., 2006; Zhang et al., 2006). Both C57BL/6 and DBA/2 are inbred mouse strains, but they have substantially different genetic backgrounds. Because C57BL/6×DBA/2 mice are derived from a cross of C57BL/6 and DBA/2 strains, their genetic background is mixed. The difference between these genetic backgrounds might contribute to the different phenotypes seen in the Cyld-knockout mice.
Meanwhile, mice harboring the truncated version of CYLD (lacking the last 24 C-terminal amino acid residues) have defects in both primary and motile cilia, and they die at birth due to impaired lung maturation (Eguether et al., 2014). A different mouse line, which expresses CYLD with a deletion of exon 9 also dies shortly after birth due to impaired lung maturation (Trompouki et al., 2009). It is therefore highly surprising that Cyld-knockout mice are viable, when deubiquitylase-dead mutations of CYLD result in lethality. A possible explanation is that there might be compensatory mechanisms in Cyld-knockout mice, whereas CYLD truncations could act as dominant-negative isoforms that bind to CYLD deubiquitylase substrates and so interfere with any compensatory activities. The microtubule-binding N-terminal CAP-Gly domains of CYLD might be important for its deubiquitylase-activity-dependent regulations. Further studies for instance with CAP-Gly-deficient mice are required to fully understand this question.
Conclusions and perspectives
Since the identification of Cyld as a tumor suppressor gene (Bignell et al., 2000), we have witnessed great advances in our understanding of its functions in pathological and physiological processes, including in tumor initiation and progression, immune system development and responses, spermatogenesis, lung maturation and osteogenesis (Hellerbrand and Massoumi, 2016; Massoumi, 2011; Mathis et al., 2015; Sun, 2010). Despite these important findings, we are only just beginning to understand the underlying mechanisms by which CYLD regulates these diverse processes. Since we first demonstrated that CYLD directly binds to tubulin and microtubules (Gao et al., 2008), accumulating evidence continues to suggest that CYLD regulates a variety of biological processes by targeting microtubules.
In this Commentary, we have discussed the involvement of CYLD in cell migration, cell cycle progression, chemotherapeutic drug sensitivity and ciliogenesis through various mechanisms that are largely dependent on its regulation of microtubules. In addition, CYLD might also participate in other microtubule-dependent activities, such as intracellular transport, because CYLD inhibits the deacetylase activity of HDAC6 and HDAC6-mediated α-tubulin deacetylation is likely to affect intracellular transport along microtubules (Dompierre et al., 2007; Kaul et al., 2014; Reed et al., 2006; Walter et al., 2012). CYLD also regulates the activation of small Rho GTPases, such as RhoA and Rac1, suggesting that it might play a role in coordinating the actions of microtubules and actin filaments (Gao et al., 2010; Yang et al., 2013).
At present, the interplay between the microtubule-regulatory activity and the deubiquitylase activity of CYLD remains unclear. The deubiquitylase domain of CYLD has been shown to interact with EB1, thereby participating in the regulation of microtubule dynamics (Li et al., 2014). Similarly, the CAP-Gly domains of CYLD might be involved in the regulation of its deubiquitylase activity by interacting with certain proteins. Given that microtubules are crucially involved in various signaling pathways and that many microtubule-binding proteins undergo ubiquitylation (Etienne-Manneville, 2010), it is tempting to speculate that these two activities of CYLD might mutually affect each other or even act in concert to control various biological processes. Gaining a holistic view of the integrated functions of the different domains of CYLD could thus provide substantial insight into the roles of this protein in both pathological and physiological processes.
This work was supported by grants from the National Natural Science Foundation of China [grant numbers 31471262, 31130015, 30825022, 30600313]; and the National Basic Research Program of China [grant numbers 2012CB945002, 2010CB912204] to J.Z.
The authors declare no competing or financial interests.