Introduction

Gigaxonin, also known as KLHL16, is a member of the BTB-KELCH family of proteins and is encoded by the GAN gene on chromosome 16q24 (Cavalier et al. 2000). Mutations of the GAN gene were found to be causative in the rare autosomal recessive neurodegenerative disorder giant axonal neuropathy (GAN, OMIM 256850) (Bomont et al. 2000). GAN is characterized by a progressive neuropathy affecting both sensory and motor nerves in both the central and peripheral nervous systems, and was first described in the 1970 (Asbury et al. 1972; Berg et al. 1972). Patients usually begin manifesting symptoms in childhood, and most patients die by the second or third decade of life (Johnson-Kerner et al. 2014). Histologically, neurons in GAN patients show large axonal accumulations, giving the disorder its name. Although gigaxonin is mutated in GAN, the precise molecular function of this ubiquitous protein is not yet fully understood. The normal physiological role of gigaxonin is in the maintenance of cytoskeletal structure in normal tissues that include processing of intermediate neural filaments in neural cells and vimentin in fibroblasts and endothelial cells (Donaghy et al. 1988; Bomont et al. 2000; Bomont and Koenig 2003; Mahammad et al. 2013). Many studies have shown that gigaxonin is involved in the ubiquitin–proteasome pathway, controlling the degradation of intermediate filaments and other cytoskeletal components (Ding et al. 2002; Allen et al. 2005; Mahammad et al. 2013). Here we summarize what is known about gigaxonin’s role in cellular functions, and point out gigaxonin mutations that may play a role in cancer development.

Gigaxonin and the ubiquitin proteasome complex

Gigaxonin is a 65 kilodalton protein composed of an N-terminal BTB domain followed by six KELCH repeats (Fig. 1). By northern blotting of the mouse tissues, Ding et al. (2002) have found ubiquitous expression of gigaxonin in all tissues, but more so in the brain, heart, and muscle. By western blotting, these authors have confirmed higher gigaxonin protein expression in the brain, heart and muscle. Ganay et al. (2011) have shown higher expression of gigaxonin protein in the tissues of the nervous system and lower expression in muscle, heart, kidney and liver. Differences in the pattern of protein expression in the same tissues could be due to differences in the mouse species studied and/or difference in the antibody used for the expression analysis. Since intermediate filaments are the hallmark of GAN patients, the role of gigaxonin in maintaining the cytoskeleton network has been studied by several investigators. Gigaxonin mutations were found to be associated with the accumulation of vimentin intermediate filaments (IFs), an important IF expressed in mesenchymal cells (Cleveland et al. 2009).

Fig. 1
figure 1

Gigaxonin mutations common to both giant axonal neuropathy (GAN) and cancer patients. a GAN mutations from GAN were compared with GAN mutations in cancers recorded in the catalog of somatic mutations in cancer (COSMIC). There are 9 discovered common mutations between the two pathologies: a single silent mutation is shown in green, 4 missense mutations are shown in blue and 4 nonsense mutations are shown in red. Nucleotide and protein numbers start from the coding region of gigaxonin, 148 bps from the exon 1 start site. Map is not drawn to scale. b Structures of gigaxonin Broad-Complex, Tramtrack and Bric a brac (BTB) and BTB and C-terminal Kelch (BACK) region homo dimer, and actin-binding protein Mayven (KLHL2) Kelch domain were obtained from Research Collaboratory for Structural Bioinformatics Protein Database (RCSB PDB). The predicted full structure was processed on the SWISS-MODEL Repository

Full size image

Extensive studies have shown that vimentin binding and processing is the major function of gigaxonin in the maintenance of normal cytoskeletal structures (Bomont et al. 2000; Bomont and Koenig 2003; Cleveland et al. 2009; Mahammad et al. 2013). Further, GAN gene replacement studies in vitro using AAV2 viral particles into GAN patient primary skin fibroblasts has shown that gigaxonin is involved in the processing of vimentin intermediate filaments (Mussche et al. 2013). Proteomic studies of the gigaxonin transfected GAN fibroblasts further showed reversal of protein accumulation in 32 proteins and three have previously been shown to be associated with the GAN phenotype.

Dequeen et al. (2008) created a GAN exon 1 deletion mutant mice (GANΔexon1; Δexon1) as a model of the GAN disease. Although these mice contained neural IF defects, accumulation of NF-L filaments, they did not resemble the human disease phenotype of neurodegeneration. Similarly, GAN exon 3–5 deletion mutant mice showed neural cytoskeletal abnormalities and mild motor and sensory defects without neuronal degeneration of the human disease (Ganay et al. 2011). However, intracisternal injection of AAV9/GAN into GAN deletion mutant mice caused a complete clearance of peripherin, an IF aggregate of gigaxonin deficiency clearly pointing out that neural IFs are the direct targets of gigaxonin (Mussche et al. 2013). Another study has also shown normal processing of peripherin and neurofilament NF-L aggregates after GAN gene introduction through lentiviral or stable transgene into 3 different GAN negative-induced pluripotent stem cells (iPSCs) (Johnson-Kerner et al. 2015). This study further pointed out that overexpression of gigaxonin is not toxic to the iPSC motor neurons pointing to the possibility of gene therapeutic studies in GAN patients. The in vitro and in vivo animal model studies have, thus, confirmed the role of gigaxonin as an ubiquitin ligase adapter with a direct impact on the structural integrity and cytoskeletal structure of normal neuronal and non-neuronal cells. Studies have also shown that microtubule related proteins MAP1B, tubulin folding cofactor B (TBCB) and MAP8 to be binding partners of gigaxonin and ubiquitinated by gigaxonin (Wang et al. 2005; Dequeen et al. 2008). However, the in vitro and animal model studies have found that these proteins are not accumulated or microtubule disorganization is not seen in GAN mutant fibroblasts, iPSC derived motor neurons or GAN mutant mice indicating that microtubule disorganization do not play a role in the GAN disease (Cleveland et al. 2009; Ganay et al. 2011; Johnson-Kerner et al. 2015.

The ubiquitin–proteasome system is a highly evolutionarily conserved process critical for protein homeostasis and proper cell function, and involves tagging proteins with multiple ubiquitin molecules and subsequent degradation of the tagged protein via the proteasome complex (Glickman and Ciechanover 2002). The first of these steps, also called ubiquitination, is highly controlled process dependent on three enzymatic activities: ubiquitin activation (E1), conjugation (E2), and ligation (E3) (Furukawa et al. 2003). Malfunction at any step in this complex system can lead to a variety of human diseases, including malignancies, neurodegenerative disorders, immune and inflammatory disorders, and genetic diseases (Glickman and Ciechanover 2002). Allen et al. (2005) have found that gigaxonin binds to the ubiquitin-activating enzyme E1 via its N-terminal BTB domain while its C-terminal kelch domain interacts with the cytoskeletal protein MAP1B-LC, leading to degradation of this protein (Allen et al. 2005).

Gigaxonin in oncogenesis

Multiple genes and proteins involved in neuromuscular disorders have been implicated in the oncogenesis of a variety of human cancers. For example, mutations in mitofusin-2 (MFN2), a mitochondrial protein, were found to be causative of Charcot-Marie-Tooth disease type 2A (Züchner et al. 2004). It was recently found that MFN2 is a key player in the cell cycle and cell invasion in lung adenocarcinoma (Lou et al. 2015). Similarly, Parkin, a mitochondrial protein is shown to be involved in the development of Parkinson’s disease as well as in human cancers (Shimura et al. 2000; Matsuda et al. 2015. Therefore, given the role of gigaxonin in the ubiquitin proteasome complex, it is possible that it could play a role in oncogenesis. A recent study by Veena et al. (2014) found that gigaxonin directly interacts with NF-κB, and this interaction is dependent on the interaction between p16, also known as cyclin-dependent kinase inhibitor 2A (CDKN2A). It is well known that p16 is an important tumor suppressor gene that renders cancer cells susceptible to cisplatin treatment (Veena et al. 2014). The putative role of gigaxonin in this interaction is that it helps the ubiquitination of p16 bound NF-κB, and thus the downregulation of NF-κB in turn makes tumors more susceptible to cisplatin treatment (Fig. 2). It is also possible that patients with one mutated copy of gigaxonin may not exhibit GAN, but may be more susceptible to cancer development if the ubiquitin pathway is affected by the heterozygous gigaxonin mutations.

Fig. 2
figure 2

Model of gigaxonin’s effect on NF-κB. In association with p16, gigaxonin is shown to ubiquitinate NF-κB leading to reduced expression of growth factors in cisplatin mediated cancer cell senescence

Full size image

Presence of gigaxonin mutations in cancer

Although gigaxonin mutations were originally discovered by their involvement in GAN, many cancers contain similar mutations (Fig. 1a). The reported gigaxonin mutations are spread throughout the protein structure but all result in similar symptoms of giant axonal neuropathy (Supplementary Table 1) (Boizot et al. 2014; Johnson-Kerner et al. 2014). Through postmortem pathology, a wide range of central nervous system degeneration was shown to be caused by these mutations (Johnson-Kerner et al. 2014).

Although gigaxonin mutations are seen throughout the protein, mutations are also found clustered in the BTB-BACK domain of the protein (15/54: 28 %), which could result in instability and impairment of homodimerization (Supplementary Table 1; Fig. 1b) (Cullen et al. 2004; Boizot et al. 2014). Without the ability to homodimerize, gigaxonin loses its function as an ubiquitin ligase adaptor. It has been shown that gigaxonin expression is severely decreased in these patients due to mRNA and protein instability. Further, Boizot et al. (2014) using 3D models of GAN mutations has suggested that mRNA instability could be due to nonsense-mediated decay and protein instability is related to impaired folding of the BTB-KELCH domain and instability in the binding of gigaxonin to substrates or cullin-E3 ligase. These studies, therefore, point out that nonsense and missense mutations play an important role in the functional inactivation of gigaxonin.

Although most prominent in neural cells, gigaxonin mutations are observed in several primary human cancers. There are 95 recorded cases of gigaxonin mutations in 19626 unique samples of cancer in the Catalog of Somatic Mutations in Cancer (COSMIC) database and these mutations can be found in http://cancer.sanger.ac.uk/cosmic/gene/analysis?ln=GAN#histo. These mutations are mostly localized to the large intestine (colon cancer), stomach, endometrium, lung, and skin tumors (Table 1). Although there are common mutations in GAN and cancers, there are missense and nonsense mutations found only in cancer patients (Supplementary Table 1). Most of these mutations occur in the BTB-KELCH region in exons 3–5 and 8–11 similar to the mutations in GAN. However, studies are not available relating these mutations to the gigaxonin function of IF degradation seen in GAN patients.

Table 1 Distribution of GAN mutations in COSMIC database by tissue
Full size table

The COSMIC cell line mutations for the GAN gene can also be found at: http://cancer.sanger.ac.uk/cell_lines/gene/analysis?ln=GAN#hist. Of the tissues that have over 40 cell lines in COSMIC, the most common tissues are in cell lines of the bone (6.82 %), large intestine (14.28 %), hematopoietic/lymphoid (5.14 %) and lung tumors (2.79). These cell line and primary tumor mutations show that mesenchymal cells are more frequently mutated in cancer cells. Transcription studies of the Broad Institute Cancer Cell Line Encyclopedia (http://www.broadinstitute.org/ccle/home) have shown that gigaxonin mRNA is highly expressed in the cancers of the intestine, breast, and lung. It is interesting to note that the tissues with the highest mutation frequencies are also the tissues with highest expression of gigaxonin.

Of the nine common gigaxonin mutations present in GAN and cancer in the Cosmic database, large intestine was the primary tissue location of a missense mutation and all 4 nonsense mutations (Table 2). The other tissue of interest seems to be endometrium with two missense mutations. Complete transcriptional profile available for large intestinal cancer cell lines KM12 and DLD-1 containing GAN mutations (exon 4 nonsense mutations in KM12 cells and exon 1 missense mutation in DLD-1 cells) has shown gigaxonin expression to be in the average range (Table 3). While DLD-1 cells show diminished expression of CDKN2A and VIM genes pointing to a direct relationship between the expression of gigaxonin and CDKN2A and VIM genes, KM12 cells show an increased expression of CDKN2A and VIM genes indicating an inverse relationship to gigaxonin expression (Klijn et al. 2015). Further, the presence of GAN mutations (Table 2) in addition to the well-known overexpression of vimentin in colon cancers (Satelli and Li 2011; Todosi et al. 2012) indicate that a detailed analysis of genomic status and expression of gigaxonin is required to delineate the significance, if any, of GAN mutations to the development of colon cancer.

Table 2 Gigaxonin mutations in cancer and giant axonal neuropathies (GAN)
Full size table
Table 3 Gene expression (RNA seq−log2 values) in colon cancer cell lines
Full size table

Analysis of two other databases, Driver-DB (http://driverdb.tms.cmu.edu.tw/driverdbv2/gene_data_p.php?genename=GAN&geneproteinid=&submit=submit and ICGC Data Portal (https://dcc.icgc.org/genes/ENSG00000261609) have shown GAN mutations in 247 cancer samples (Supplementary Table 2). Location of these mutations in the different tumor samples is summarized in Table 4. In this dataset, liver tumors show the highest number (73) of nonsense (stop codon) mutations. These mutations are present mostly in exons 4, 7 and 9–11 with an indication that preponderance of these mutations are at the c-terminus of the protein. Missense mutations are found in skin (13), lung, head and neck and stomach (12 each), uterus (11), colorectal (10), breast, cervix and pancreas and gall bladder (7 each), kidney (6), ovary and bladder (5 each), cervix and pancreas (5 each), CNS (4) and in prostate (3) cancers. Silent mutations are present in 13 uterine and 9 head and neck cancers. While colon cancer had the highest number of mutations in the cosmic database, the driver-DB and ICDC data portal show the highest number of mutations in the liver cancer.

Table 4 Summary of GAN mutations on cancer databases DriverDB and ICGC data bank
Full size table

Function of gigaxonin protein has also been linked to the ubiquitination of NF-κB in cisplatin-induced senescence of cancer cells (Fig. 2). The results suggested that gigaxonin could be involved in the regulation of NF-κB in normal cell cycle. As human cancers are affected by the activation of NF-κB, the absence of NF-κB degradation due to gigaxonin mutations could lead to NF-κB-mediated oncogenic signaling in cancer cells. While cellular phenotypes are vastly different in GAN and cancer, it remains to be seen whether gigaxonin mutations cause vimentin IF aggregation in cancer cells (O’Neill and Kaltschmidt 1997). What is known is overexpression of vimentin in chemo-radiation resistant cancer cells indicating the possibility that gigaxonin indeed might be playing a role in the regulation of vimentin expression during normal cell growth and development. Similarly, maintenance of transcriptional regulation by gigaxonin could be important for the prevention of tumor development. Investigation of gigaxonin mutations using in vitro cell line and xenograft animal tumor models will, therefore, be valuable for the elucidation of gigaxonin function in normal cell growth and development.

Exon 8 polymorphism, rs2608555, in the general population

HapMap studies have identified a number of intronic mutations of the gigaxonin genomic sequence in the normal population. However, only a single silent mutation, rs2608555 or c.1293C>T, is reported for the exonic sequence of the gigaxonin genome. This polymorphism exists in 22 % of the world population ranging from ~3 to 52 % in different populations (Table 5). The Caucasian population has the highest frequency (44.25 %) and the native Chinese population has the lowest frequency (2.75 %) of this polymorphism.

Table 5 HapMap distribution of GAN rs2608555 mutation
Full size table

Veena et al. have shown that expression of gigaxonin and p16 is related to the ubiquitination of NF-κB. While p16 is inactivated through mutations and promotor hypermethylations, GAN gene contains mutations and c. 1293C>T polymorphism (silent mutation) in GAN patients (Bomont et al. 2000). Some GAN patients carrying the silent mutation could have normal gigaxonin mRNA levels possibly due to transcription upregulation of the other mutant allele as observed in patients carrying compound nonsense or deletion mutations (Boizot et al. 2014). We, therefore, hypothesize that the stability of the mRNA could depend on RNA binding miRNAs and/or secondary structures (Gao et al. 2013; Werk et al. 2014). While the polymorphism is currently assumed to be benign, it could play a role in mRNA folding and stability leading to differential expression in cancer cells.

The study of Bomont et al. (2000) has shown that the gene is inactivated by heterozygous allelic mutations in many GAN patients. In two patients, they have reported a missense mutation and c. 1293C>T polymorphism. They have also shown that heterozygous alleles are present in one of these samples. It is likely therefore that in addition to the missense mutation, c. 1293C>T polymorphism plays a role in the functional inactivation of the GAN gene. Therefore, investigation of GAN c. 1293C>T polymorphism in cancer is warranted to determine whether a relationship exists between GAN polymorphism and chemosensitivity and/or tumor aggressiveness.

Examples of silent mutations and mRNA instability include silent mutations in cystic fibrosis that exhibit mRNA instability and aberrant splicing due to c.2811G>T polymorphism in exon 15 of the CFTR (cystic fibrosis transmembrane conductance regulator) gene (Faa’ et al. 2010). There are also silent polymorphisms in the Multidrug Resistance 1 (MDR1) gene known to affect the mRNA folding (Kimchi-Sarfaty et al. 2007). A single change in the mRNA alters its folding thereby preventing cytoplasmic translocation and translation. Thus, we believe that in vitro studies need to be performed to investigate the role of exon 8 polymorphism of the GAN gene in RNA stability and tumor development. Details of the GAN single nucleotide polymorphism (SNP) rs2608555 including its chromosomal location can be found at: http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=2608555.

Conclusion

Gigaxonin, an E3 ligase adaptor protein, is shown to be involved in the processing of neural intermediate filaments in neurons and cytoskeletal proteins such as vimentin in fibroblasts. The protein has also been found to be involved in the degradation of proteins such as NF-κB. These studies demonstrate that gigaxonin plays a crucial role in cell cycle growth and cytoskeletal homeostasis. Therefore, the inactivation of gigaxonin through mutations would result in unrestricted growth signaling leading to unrestricted cell growth. There is also an exonic polymorphic mutation that is prevalent in the Caucasian population which may play a role in the development of neural diseases and/or cancers in this population. The direct relationship in the reduced expression of p16 and gigaxonin in cancer cell lines suggests that the two genes might be co-regulated in the regulation of the cell cycle. Since p16 expression is related to senescence of aging cells, it is likely that gigaxonin expression might also be elevated leading to reduced growth of cells in the aging individuals. We could hypothesize therefore that gigaxonin mutation profile and/or expression could be a diagnostic marker of a subset of cancers that might have reduced p16 expression. Finally, p16 expression being used as a surrogate marker of HPV positivity and chemo-radiation sensitivity, we believe that gigaxonin expression could be added to this list of cancer diagnostic markers.