SnoN: Bridging Neurobiology and Cancer Biology

Abstract

The transcriptional regulator SnoN has been the subject of growing interest due to its diverse functions in normal and pathological settings. A large body of evidence has established a fundamental role for SnoN as a modulator of signaling and responses by the transforming growth beta (TGFβ) family of cytokines, though how SnoN regulates TGFβ responses remains incompletely understood. In accordance with the critical and complex roles of TGFβ in tumorigenesis and metastasis, SnoN may act as a tumor promoter or suppressor depending on the stage and type of cancer. Beyond its role in cancer, SnoN has also been implicated in the control of axon morphogenesis in postmitotic neurons in the mammalian brain. Remarkably, signaling pathways that control SnoN functions in the divergent cycling cells and postmitotic neurons appear to be conserved. Identification of novel SnoN regulatory and effector mechanisms holds the promise of advances at the interface of cancer biology and neurobiology.

Role of SnoN in TGFβ Signaling

SnoN has an established role as a critical regulator of signaling by the transforming growth factor beta (TGFβ) family of secreted factors [1]. The TGFβs control a wide array of biological responses that are essential for normal development and homeostasis [2, 3]. TGFβ acts via specific cell surface serine/threonine kinase receptors that activate the TGFβ-regulated-Smad (R-Smad) signaling proteins [4]. Activated R-Smads form a heteromeric complex with the common partner Smad4 [5]. The R-Smad/Smad4 complex accumulates in the nucleus, where it associates with distinct transcriptional regulators to modulate the expression of diverse TGFβ-responsive genes. Together with transcriptional coactivators such as the histone acetyltransferase p300 and with corepressors such as the histone deacetylase HDAC1, the Smad complex stimulates or represses, respectively, the expression of TGFβ-target genes [6–8].

SnoN function has been intimately linked in the control of TGFβ-regulated transcription. Initial studies suggested that SnoN antagonizes TGFβ-induced transcription [9–11]. SnoN associates with the R-Smad/Smad4 complex and is recruited to TGFβ-responsive genes [1, 10, 12]. In turn, SnoN is thought to recruit HDACs which deacetylate histones and thereby repress transcription.

Accumulating evidence suggest that in addition to inhibiting TGFβ-induced transcription and responses, SnoN may also remarkably positively regulate TGFβ-signaling and responses [1]. Knockdown of SnoN by RNAi impairs TGFβ-induced transcription in distinct cell types, including epithelial cells and postmitotic neurons suggesting that endogenous SnoN may mediate TGFβ-dependent responses [13, 14]. Consistent with these observations, expression of SnoN at low levels in ovarian epithelial cells enhances expression of the TGFβ-responsive genes cyclin-dependent kinase inhibitor p21 and plasminogen activator inhibitor 1 (PAI-1), and promotes cell cycle arrest [15]. Genetic studies in drosophila have also indicated that dSno is required for the ability of the TGFβ pathway to signal in the nervous system, suggesting that the positive role of SnoN in TGFβ signaling is evolutionarily conserved [16]. Collectively, these data support the idea that SnoN facilitates TGFβ-dependent responses in specific biological contexts.

How does SnoN stimulate TGFβ-induced transcription?

A recent study has begun to shed light on this interesting feature of SnoN function. Shirin Bonni and colleagues have identified the inhibitor of growth (ING) family of proteins as playing a critical role in the ability of SnoN to promote TGFβ-induced transcription and responses [13]. The ING family comprises a group of chromatin remodeling proteins that share a highly conserved plant homeodomain (PHD) zinc finger domain at their C-terminus [17–19]. The PHD domain binds with high affinity to the chromatin protein histone H3 specifically when it is trimethylated at Lysine 4 [20–22]. The trimethylated histone H3 Lysine 4 marks active transcription start sites in genes [23]. The ING proteins are thus recruited to sites of active transcription in the genome, and through their association with transcriptional coactivators, such as p300, they can stimulate transcription of responsive genes [17, 24–28]. The ING family member ING2 was recently found to promote TGFβ-induced transcription and cell cycle arrest in epithelial cells [13]. ING2 upregulates TGFβ-induced transcription by enhancing the transcriptional activity of the R-Smads in a PHD-dependent manner [13]. Importantly, SnoN also forms a complex with ING2 in a PHD-dependent manner [13]. Further, SnoN promotes the formation a tripartite complex composed of ING2, R-Smads, and SnoN suggesting that SnoN may mediate the ability of ING2 to stimulate TGFβ-dependent transcription. Consistent with this idea, loss and gain of function studies suggest co-dependency of SnoN and ING2 for their abilities to upregulate TGFβ-mediated transcription [13]. Collectively, these results define a Smad-SnoN-ING2 signaling axis that mediates TGF-β-induced transcription [PART A and 13]. In future studies, it will be important to investigate the downstream mechanisms by which SnoN/ING2 complex controls TGFβ-dependent responses.

Part A.

SnoN collaborates with ING2 to promote TGFβ-induced transcription and biological responses. SnoN acts as an adaptor to induce the assembly of a R-Smad-SnoN-ING2 multiprotein complex that enhances TGFβ-induced transcription and cell cycle arrest.

SnoN has a dual role in tumorigenesis

Research focusing on the role of SnoN in cancer suggest that SnoN may have complex roles in cancer. SnoN involvement in TGFβ signaling and responses is thought to provide a basis for its actions in cancer. One of the widely studied and important effect of TGFβ is its ability to inhibit cell proliferation [2, 29–31]. Another important response of TGFβ is epithelial-mesenchymal transition (EMT), a fundamental process whereby epithelial cells dedifferentiate to acquire a mesenchymal and more motile phenotype [30, 32–35]. Cellular hallmarks of cancer include the evasion of cell cycle arrest and induction of EMT [30, 32]. TGFβ is thought to play negative or positive roles in neoplastic disease depending on the stage of tumor development [36, 37]. At early stages, loss of cell responsiveness to the anti-proliferative effect of TGFβ may contribute to the pathogenesis of several epithelial tumors including colorectal, pancreatic, and mammary carcinomas [29, 30, 32]. Conversely, TGFβ-induced EMT is believed to contribute to tumor progression observed at later stages of many malignancies, including mammary, prostate, and colorectal carcinomas [30, 32, 35, 38–40]. Interestingly, similar to TGFβ, SnoN may have oncogenic or tumor suppressor roles as described below.

SnoN as an oncogene

Several lines of evidence strongly support the idea that high levels of SnoN may be oncogenic. Early studies showed that overexpression of SnoN in chicken embryo fibroblast triggers transformation [41, 42]. The negative regulation of the TGF-β pathway by overexpressed SnoN has been proposed to promote malignant cellular transformation at early stages of cancer [10–12]. The ability of overexpressed SnoN to negatively regulate TGFβ signaling and cell cycle arrest might provide a mechanism for the loss of responsiveness to TGFβ signaling [15, 43]. The high levels of SnoN may inhibit TGFβ repression of c-myc, which is critical for TGFβ-induced cell cycle arrest [44]. Thus, SnoN behaves as an oncogene in this context. Consistent with these observations, SnoN is amplified in several types of cancers, including melanomas, breast, and esophageal carcinomas [43–50]. Interestingly, SnoN maybe overexpressed only at early stages of some types of cancer such as the precancerous condition Barrett’s esophagus and colorectal carcinomas [51, 52]. On the other hand, SnoN levels correlate with depth of invasion, recurrence and poor prognosis in esophageal squamous cell carcinoma [53, 54]. Collectively, these data highlight the putative oncogenic role of SnoN, linked to its overexpression and negative regulation of TGFβ signaling, specifically at early stages of tumorigenesis [55].

Expression of SnoN might be induced as a result of chromosomal amplification as has been observed in esophageal squamous cell carcinoma and ovarian cancer cells [15, 53]. Other mechanisms may contribute to upregulated cellular SnoN levels. SnoN is increased at the protein level in colorectal cancer without a corresponding induction of its mRNA, suggesting that overexpression could result from increase in translation rate of mRNA or in stability of the protein [51]. Since the levels of SnoN protein are tightly regulated by the ubiquitin-proteasome pathway, the failure to degrade SnoN might result from reduced expression or dysfunction of E3 ubiquitin ligases that induce the ubiquitination of SnoN [1, 15, 51]. Changes in the subcellular localization of SnoN may also affect its levels, since cytoplasmic SnoN appears to be more resistant to TGF-β-dependent degradation than nuclear SnoN [56]. Although SnoN resides predominately in the nucleus, SnoN may also localize in the cytoplasm in some cell types, including cancer cells [42, 56–58]. Interestingly, SnoN is mostly cytoplasmic in Barrett’s esophagus [52]. Cytoplasmic SnoN may sequester Smads and block their accumulation in the nucleus, leading to the inhibition of TGFβ-induced transcription and cell cycle arrest [56, 57].

SnoN as a tumor suppressor

Several lines of evidence support the concept that in addition to its oncogenic role, remarkably SnoN also harbors a tumor suppressive function. Heterozygous sno^+/−-mice develop spontaneous tumors at low frequency and have an increased rate of tumor formation when challenged with chemical carcinogens [59]. In addition, as compared to wild type controls, sno^+/− -derived mouse embryonic fibroblasts,(MEFs) and B and T lymphocytes exhibit increased proliferation rates, and resistance to cell cycle arrest and apoptosis by specific stimuli [59]. Together with knockdown studies showing that endogenous SnoN can promote TGF-β responses including cell cycle arrest, these data suggest that at physiological levels SnoN may act as a tumor suppressor [1].

Several tumors, including mammary and lung carcinomas, show downregulation of ING expression, suggesting that the INGs may act as tumor suppressors in certain types of cancers [18, 60–63]. In view of the recent findings that ING2 may mediate TGF-β-induced transcription and cell cycle arrest, it is conceivable that loss of ING2 may allow cancer cells to evade the tumor-suppressive effects of TGFβ [13]. As SnoN is important for the ability of ING2 to mediate TGFβ-induced transcription and responses, this may thus define a mechanism by which SnoN acts as a tumor suppressor [Part A and 13].

SnoN has also been suggested to inhibit tumor invasion at later stages of cancer. This may involve SnoN’s ability to suppress TGFβ-induced EMT in cancer cells [55]. The TGFβ-regulated Smad pathway contributes to the ability of SnoN to inhibit EMT [55]. The SnoN-related protein Ski was recently demonstrated to block metastasis in breast and cancer cells, and TGFβ-mediated degradation of Ski was suggested to enhance metastasis in late stage tumors [64]. Thus, it is conceivable that TGFβ may also enhance SnoN degradation in later stages of cancer in order to overcome SnoN inhibition of EMT and thus promote metastasis.

The tumor suppressor role of SnoN may also involve other signaling pathways. A recent study using Smad-binding defective SnoN knockin mice suggested that SnoN induces cellular senescence via stabilization of p53 in PML bodies in MEFs [65]. In contrast, although transiently expressed SnoN was reported to localize in PML bodies in large T antigen immortalized human normal ovarian epithelial cells in which p53 is inactivated, stably expressed SnoN appears to promote senescence independently of PML levels or p53 status in these cells [15]. Whether the discrepancy in the findings of these two studies is due to differences in cell type or levels of SnoN remains to be investigated. Interestingly, ING1 and ING2 can promote senescence in a p53-dependent and independent manner [24, 27, 66, 67]. That SnoN functions together with ING2 to activate TGFβ-induced transcription suggests that SnoN’s role in senescence may also involve its association with ING2. Further research is required to determine whether downstream targets of the ING-SnoN-Smad complex may be involved in establishing senescence [13].

Overall, these data illustrate that SnoN’s role in tumorigenesis is strongly dependent on its expression level and the stage of cancer development at which SnoN is expressed. In addition, SnoN function in TGFβ signaling appears to largely contribute to SnoN’s roles in cancer.

Novel functions for SnoN in postmitotic neurons

In addition to the critical roles of SnoN in cycling cells, novel functions of SnoN have been uncovered in postmitotic neurons. Knockdown of SnoN in primary cerebellar granule and hippocampal neurons dramatically inhibits the growth of axons [68–72]. In addition, SnoN knockdown in postnatal rat pups profoundly impairs the morphogenesis of parallel fiber axons in the cerebellar cortex in vivo [70, 72]. These observations suggest that SnoN may play a pivotal role in the development of axons in the brain.

The mechanisms that regulate SnoN function in axon morphogenesis are beginning to be elucidated. The major E3 ubiquitin ligase Cdh1-anaphase promoting complex (Cdh1-APC) stimulates the ubiquitination and consequent proteasome-dependent degradation of SnoN in neurons [72]. By inducing the degradation of SnoN, Cdh1-APC thus restricts the growth of axons in neurons [72].

The identification of the Cdh1-APC/SnoN link as a cell intrinsic pathway regulating axon morphogenesis in neurons mirrors the characterization of SnoN as a substrate of Cdh1-APC in the control of G1 phase of the cell cycle in proliferating cells. Notably, R-Smads recruit Cdh1-APC to stimulate SnoN ubiquitination in proliferating cells [9, 73]. Remarkably, this mechanism is conserved in neurons. Accordingly, TGF-β-Smad signaling stimulates the Cdh1-APC/SnoN pathway and thus limits the growth of axons in neurons [71].

The steps that operate downstream of SnoN in neurons have been recently characterized [70]. Microarray analyses in SnoN knockdown neurons have been performed to identify the SnoN-regulated program of genes that promote axon growth. Strikingly, a large fraction of genes that are altered in SnoN knockdown neurons are downregulated [70]. These data suggest that endogenous SnoN may activate transcription of a subset of target genes in neurons. Consistent with this observation, SnoN interacts with the transcriptional co-activator p300. Loss of function of p300 phenocopies SnoN depletion on axon growth in neurons [70]. In addition, p300 knockdown blocks the ability of a constitutively active form of SnoN to stimulate axonal growth. These data suggest that p300 may mediate the ability of SnoN to activate the expression of genes that promote axon growth [Part B].

Part B.

SnoN stimulates axon growth in postmitotic neurons in the mammalian brain. SnoN associates with the histone acetyltransferase p300 to induce the transcription of actin-binding and signaling scaffolding gene Ccd1. The protein Ccd1, which localizes to axon terminals, plays a critical role in the growth of axons. The ability of the SnoN-p300-Ccd1 pathway to stimulate axon growth is restricted by the TGFβ-Smad pathway which recruits the E3 ubiquitin ligase Cdh1-APC to SnoN leading to its ubiquitination and consequent degradation.

The actin-binding protein Ccd1 has been identified as a critical downstream gene target of SnoN in neurons [70]. SnoN and p300 are required for Ccd1 transcription in neurons [70]. Ccd1 contains coiled coil domain and a DIX (Dishevelled-Axin) domain [74, 75]. Ccd1 localizes to axon terminals, suggesting a potential role in axon growth [70]. Knockdown of Ccd1 inhibits axon growth and prevents the ability of SnoN to stimulate the growth of axons. Importantly, Ccd1 is required for the development of granular neuron parallel fiber axons in the cerebellar cortex [70]. These results suggest a requirement for Ccd1 in SnoN-dependent axon morphogenesis in the brain [70]. Collectively, studies of SnoN in neurons have defined an elaborate cell-intrinsic pathway that orchestrates the growth of axons.

PERSPECTIVES

Recent studies suggest that SnoN plays important roles in both cancer biology and neurobiology. Expression analyses and functional studies indicate that SnoN may promote or suppress tumorigenesis depending on the stage and type of cancer. However, the mechanisms that regulate these opposing functions of SnoN in cancer biology have remained largely unexplored. Thus, future studies should focus on investigating potential upstream regulators and downstream mediators that might orchestrate the dual role of SnoN in cancer. Identification of novel SnoN-interacting proteins will be fruitful toward this goal. Functional studies of SnoN and SnoN-interacting proteins in cancer pathogenesis should include xenograft and transgenic in vivo models of cancer. Most of the studies thus far have focused on the tumorigenic role of SnoN as an intrinsic factor within cancer cells. Increasingly, the tumor microenvironment, immune system, and vascular system appear to also influence cancer development and metastasis. Therefore, future studies should also investigate whether SnoN, its regulators, and effector mechanisms modulate the ability of TGFβ to control tumorigenesis in a non-cell autonomous manner.

The defined function of SnoN in axon growth in postmitotic neurons supports the concept that proteins with key roles in cycling cells regulate fundamental aspects of neuronal morphogenesis. The transcriptional activating function of SnoN plays a crucial role in the promotion of axon growth in neurons, and Ccd1 appears to be a critical target gene that operates downstream of SnoN in this function. Whether other SnoN-dependent genes participate in axon morphogenesis remains to be investigated. In addition, whether SnoN harbors a transcriptional repressive activity in neurons and the potential functional consequences of such an activity in brain development remain to be explored. Beyond the implications of SnoN function in axon morphogenesis in the developing brain, studies of SnoN raise the intriguing question of whether expression of exogenous SnoN might stimulate the growth of axons in the adult brain following injury.

Studies of SnoN in the context of cancer biology bear important implications for our understanding of SnoN biology in neurons. For example, the role of ING2 in SnoN-induced transcription in proliferating cells raises the question of whether ING2 forms a complex with SnoN in neurons and thereby contributes to SnoN’s ability to promote the growth of axons. Likewise, the recent characterization of the PIAS1/SnoN sumoylation pathway in proliferating cells leads to the question of whether sumoylation influences SnoN function in neurons [1, 76]. Conversely, SnoN studies in neuronal morphogenesis provide interesting leads for improved understanding of SnoN functions in proliferating cells. For example, identification of Ccd1 as a physiologically relevant target of SnoN in axon growth raises the question of whether Ccd1, which harbors actin binding and signaling scaffold activities, might contribute to SnoN functions in TGFβ-induced cell cycle arrest and EMT. Besides Ccd1, it will be interesting to determine whether other genes identified as SnoN targets in microarray analyses of SnoN knockdown neurons are also regulated by SnoN in proliferating cells. The TGFβ-response genes thrombospondin 2 and Zeb1, which have been implicated in tumor progression and EMT respectively, were downregulated in SnoN knockdown neurons [70, 77–81]. It will be interesting to determine whether thrombospondin 2 and Zeb1 operate downstream of SnoN in the control of tumorigenesis and metastastis. Studies of SnoN will likely continue to provide novel insights in cancer biology and neurobiology for years to come.