Gigaxonin is an E3 ubiquitin ligase that plays a role in cytoskeletal stability. Its role in cancer is not yet clearly understood. Our previous studies of head and neck cancer had identified gigaxonin interacting with p16 for NFκB ubiquitination. To explore its role in cancer cell growth suppression, we analyzed normal and tumor DNA from cervical and head and neck cancers. There was a higher frequency of exon 8 SNP (c.1293 C>T, rs2608555) in the tumor (46% vs. 25% normal, P = 0.011) pointing to a relationship to cancer. Comparison of primary tumor with recurrence and metastasis did not reveal a statistical significance. Two cervical cancer cell lines, ME180 and HT3 harboring exon 8 SNP and showing T allele expression correlated with higher gigaxonin expression, reduced in vitro cell growth and enhanced cisplatin sensitivity in comparison with C allele expressing cancer cell lines. Loss of gigaxonin expression in ME180 cells through CRISPR-Cas9 or siRNA led to aggressive cancer cell growth including increased migration and Matrigel invasion. The in vitro cell growth phenotypes were reversed with re-expression of gigaxonin. Suppression of cell growth correlated with reduced Snail and increased e-cadherin expression. Mouse tail vein injection studies showed increased lung metastasis of cells with low gigaxonin expression and reduced metastasis with reexpression of gigaxonin. We have found an association between C allele expression and RNA instability and absence of multimeric protein formation. From our results, we conclude that gigaxonin expression is associated with suppression of epithelial–mesenchymal transition through inhibition of Snail.

Significance:

Our results suggest that GAN gene exon 8 SNP T allele expression correlates with higher gigaxonin expression and suppression of aggressive cancer cell growth. There is downregulation of Snail and upregulation of e-cadherin through NFκB ubiquitination. We hypothesize that exon 8 T allele and gigaxonin expression could serve as diagnostic markers of suppression of aggressive growth of head and neck cancer.

Giant axonal neuropathy (GAN) is a rare but severe neurodegenerative disorder caused by mutations of the GAN gene (1). This gene encodes the protein gigaxonin, which possesses E3 ubiquitin ligase properties and is involved in intermediate filament (IF) processing and degradation (1–3). Gigaxonin mutations cause accumulation of unprocessed IF which in neurons are visualized in larger (giant) appearing neurons on cytopathology (4). Among the identified partners of gigaxonin, class III and IV IFs such as peripherin, vimentin, and neurofilament proteins, as well as the autophagy protein ATG16L1 and Ptch receptor of Sonic hedgehock signaling molecule (Shh) signaling have been shown to be regulated by gigaxonin in GAN models (5–8). Degradation of IF is regulated through glycosylation of gigaxonin (9). Patients with GAN initially display peripheral motor and sensory disease symptoms at infancy, which progressively affects the central nervous system causing loss of mental function and seizures. Prognosis is poor as life expectancy is 30 years (10). Currently GAN gene therapy clinical trial is being conducted for the treatment of GAN in collaboration between National Institute of Neurological Disorders and Stroke (NINDS) and Taysha Gene Therapies, Inc. (Intrathecal Administration of scAAV9/JeT-GAN for the Treatment of Giant Axonal Neuropathy, ClinicalTrials.gov Identifier: NCT02362438).

As a distant relative of the BTB/kelch superfamily, gigaxonin is composed of a single BTB domain for interacting with Cul3 ubiquitin ligases and six kelch domains for interacting with proteins targeted for degradation (1). In patients with GAN, genomic mutations have no regional preference (6). Although GAN gene mutations are generally known for their manifestation in neural cells, they are found in primary human cancers at a low frequency. In the Catalogue Of Somatic Mutations in Cancer (COSMIC) database, there are 522 (1.4%) recorded cases of GAN gene mutations in different cancer types out of 38,170 cancer samples studied (11). Analysis of two other databases, Driver-DB and International Cancer Genome Consortium (ICGC) Data Portal, has shown cancer specific GAN gene mutations mostly in liver cancer (12–14).

The 1000 Genome Project has recorded exon 8 SNP, rs2608555 (c.1293C>T, p.Y431Y) in 27.4% of the world population (8.1%–45.6%) with the highest frequency (45.6%) in the European population (14–16). SNPs are often assumed to be non-functional, but there are examples of functional SNPs in the literature. Cystic fibrosis patients with the rs397508419 (c.2679G>T, p.G893G) polymorphism in exon 15 of the CFTR gene exhibit aberrant mRNA splicing, and SNPs in the MDR1 gene affect mRNA folding that prevents cytoplasmic translocation and translation (17, 18). In patients with Duchenne muscular dystrophy, a combination of SNPs in the SPP1 and LTBP4 genes are associated with age at loss of ambulation (19, 20). In patients with Alzheimer's disease, SNPs are associated with variations in amyloid β deposition (21). Although the clinical phenotype of GAN gene exon 8 SNP is not known, it is likely that a heterozygous SNP (T and C alleles) impacts gigaxonin expression as the gene seems to be inactivated by heterozygous allelic mutations (possibly loss of T allele expression) in patients with GAN (22).

Human papillomavirus (HPV) sequences detected in >90% of cervical cancer and 15%–20% of head and neck cancer show high p16 expression (23–27). While the p16 and p53 genes are not mutated in HPV-positive tumors, p53 and Rb proteins are inactivated through ubiquitination by E6 and E7 proteins of the HPV, respectively (26, 27). In HPV-negative tumors, p16 is inactivated through deletions and promotor methylation, and p53 is inactivated through point mutations (28, 29). Inactivation of the RB gene by the E7 protein results in the activation of the E2F1 transcription factor which in turn activates cell cycle proteins CDK4, CDK6, and cyclin D1. Thus, in the absence of functional p53 and Rb proteins, cancer cell regulation is disrupted in human cancer. We have previously shown that p16 in association with gigaxonin ubiquitinates NFκB, a major transcription factor involved in the development of head and neck cancer (30). It is not known whether other transcription factors are targeted directly or indirectly through NFκB by gigaxonin. To understand the role of gigaxonin in tumor development, we determined the prevalence GAN exon 8 SNP (rs2608555) in cervical and head and neck cancers. We then investigated cancer cell lines containing homozygous exon 8 SNP (T/T) or heterozygous alleles (C/T) in their genome, relationship of T versus C allele to gigaxonin expression and its effect on cancer cell growth including cisplatin sensitivity.

Primary Tumors and Cancer Cell Lines

Tumor samples were obtained with Institutional Review Board approval from VAGLAHS, UCLA, and City of Hope Medical Center. Cervical cancer cell lines were obtained from ATCC at different times and UM-SCC cell lines were obtained from The University of Michigan in 2000. Cell lines were used within 5 passages and were tested before their initial use for Mycoplasma. Latest testing on ME180 cells was done in 2021. Cell lines were grown as described earlier in RPMI medium containing PSF and supplemented with 10% FBS and ampules were stored in liquid nitrogen tank. Cell lines were tested for Mycoplasma contamination using the approved kit. Cell authentication was done using mitochondrial repeat PCR primers (refs. 31, 32; Supplementary Fig. S1). Cell morphology formed second method of cell authenticity (Supplementary Fig. S2). ATCC indicates HT3 cell line as HPV negative as well as containing HPV 30 sequences. These cells have homozygous mutation of TP53 gene. Because HPV positivity is related to wild-type p53, mutated p53 could be interpreted as HPV-negative cells.

Incucyte (Real-time Cell Imager) Cell Proliferation Assay

A total of 4,000 or 8,000 cells (ME180, GAN 15.1, Control LV 50, GAN LV 50) were seeded in quadruplicates into 96-well plates on day 0 and media was changed next day with respective treatment concentrations (cisplatin) After the media was changed, the plate was put into Incucyte imager (located at core facility of Jonsson Comprehensive Cancer Center) and the image was automatically captured every 2 hours as per machine routine setting. After the 4–5 days, the data were analyzed by the built-in software, replot and statistical analysis was performed by GraphPad Software (Prism, version 6). The experiments were performed thrice for each experiment.

Soft Agar Colony Assay

Cells were trypsinized, and suspended in minimum essential medium containing 0.1% lukewarm agar at a cell concentration of 5 × 103 cells/mL. The suspension was spread on top of 0.5% solidified agar plates. The agar plates were incubated for 12 to 30 days at 37°C. Colonies were stained with 0.001% crystal violet blue, counted, and photographed using a Zeiss microscope (30, 33). Soft agar colony assays were performed in triplicates and repeated thrice.

mRNA Stability Assay

Cancer cell lines (300,000 per well) were grown in 6-well tissue culture plates to 60% confluency (48 hours), serum-free media was added for 24 hours and then grown in serum plus media for 12 hours. Treated with actinomycin D (10 µg/mL, 8 µmol/L) and cells were isolated at different time periods for RNA and protein expression analyses. For the mRNA stability assay, total RNA (1 µg) was used for cDNA synthesis using the Applied Biosystems High-capacity cDNA reverse transcriptase kit and qPCR was performed using Applied Biosystems primers as described previously (34–37). Differential expression was calculated using house-keeping gene GAPDH as control and percent mRNA expression was determined for the different time periods. Percent remaining RNA was used against timepoints to derive 50% expression half-life of mRNAs. Increased γH2AX expression was used as a measurement of actinomycin D activity. Percent mRNA remaining was plotted on GraphPad Prism against the number of hours after actinomycin D treatment and analyzed with the built-in linear regression function. The equations determined by the linear regression analysis are as follows: Y = −15.52*X + 100.0 (HeLa), Y = −9.815*X + 100.0 (ME180), Y = −4.530*X + 100.0 (HT3), Y = −6.341*X + 100.0 (SiHa), Y = −3.650*X + 100.0 (HeLa: CCND1), Y = −9.851*X + 100.0 (ME180:CCND1), and Y = −9.111*X + 100.0 (HeLa: CDKN2A). Half-life (t1/2) was calculated by determining the X-axis value (hours after actinomycin D treatment) for 50% RNA in Y-axis. At least two different qRT-PCR experiments, and each time in quadruplicates, were carried out for all the cell lines. The studies were done at least thrice.

Proteosome and Autophagy Inhibitor Assay

Cells (1 × 106 in 10 mm tissue culture dishes) were grown for 48 hours to 60% to 70% confluency, shifted to serum-free media for 24 hours and then to serum plus media for 24 hours. Autophagy inhibitor treatment with bafilomycin A (100 nmol/L) was carried out for 24 hours. Treatment with proteosome inhibitors MG132 and n-ethylmaleimide (10 µmol/L each) was performed for final 4 of 48 hours cell growth in serum plus media. Proteins isolated from the cells were subjected to SDS-PAGE analysis, transferred to nylan filters, and hybridized to antibodies using the established Western blot hybridization protocol (30, 38, 39). Inhibitor studies were done thrice for each experiment.

Boyden Chamber Migration and Matrigel Invasion Assays

Migration and Matrigel invasion assays were performed in triplicates using 8 µmol/L filters in 24-well plates (33). Twenty-five thousand cells were seeded onto Boyden chamber cell culture inserts of 24-well plates and grown in 0.5 mL of 2% FBS-RPMI medium. The bottom wells of the chamber received 0.6 mL of 10% FCS as a chemoattractant. For the Matrigel assay, filters contained 100 µL of 300 µg/mL Matrigel, dried in tissue culture incubator for 2.5 hours and 25,000 cells (0.5 mL 2% FBS-RPMI medium) were gently layered on top of the Matrigel. Chemo attractant in the bottom well was 10% FBS-RPMI medium. After 24 hours incubation in the tissue culture incubator, cells in the top surface were wiped out with cotton swabs, cells at the bottom surface were fixed with 4% paraformaldehyde (z-fix) for 12 hours, washed twice with PBS for 5 minutes each time, stained with 0.4% crystal violet for 12 hours. After two washes for 5 minutes with PBS, cells were fixed with 70% ethanol for 15 minutes. The blue cells were photographed using Thermo Fisher scientific EVOS XL cell imaging system at 10X and 20X magnification settings and plotted as cells per observed field. The studies were done at least thrice.

Lentiviral GAN Gene Transduction

GAN plasmid, a gift from Dr. Erik Lykken of Dr. Steven Gray's lab at University of Texas – Southwestern, Dallas, TX was used to amplify the full gigaxonin sequence (nt 1– 1794), and the product was cloned in the eukaryotic expressing vector JeT CMV-Myc. The derivative plasmid was digested with SAC II to get 1,840 bp fragment containing the GAN gene (cytomegalovirus promoter and Myc probing marker) and cloned into the SAC II site of pLEX-MCS lentiviral vector (Thermo Fisher Scientific). Transductin of lentiviral control vector and the vector containing the GAN gene preparations were performed using the established protocol. Lenti boost Kolliphor P338 (Sirion Biotech) was used at 1 mg/mL in the viral transductions. Clones were selected in puromycin (2 µg/mL) RPMI medium and single clones were isolated using cloning ring technique.

RNA Isolation, Library Construction, and RNA-sequencing Gene Differential Analysis

RNAs were isolated from control or siRNA transfected cells using a pure Link RNA mini kit from Ambion (12183018A; Life Technologies, Inc.) and the RNA libraries were prepared using the KAPA mRNA HyperPrep kit (Roche) according to the manufacturer recommendations. The final libraries were pooled at equimolar concentration and sequenced on an Illumina NovaSeq6000 sequencer (SP lane, 2 × 50 bp). First, reads were demultiplexed and RNA sequencing (RNA-seq) data were processed to remove adapter sequences and low-quality reads. Then, raw reads were mapped to the human genome (hg38/GRCh38) with comprehensive genome annotation from Genocode (version 36) by the STAR ultrafast universal RNA-seq aligner (version 2.7.3a; refs. 40–42). Uniquely mapped reads were used by Htseq-count to calculate the number of reads (43). The reads were then normalized using DESeq2’s median of ratios method and DESeq2 (version 1.4.5) was used to perform differential gene expression analysis (43). For the Gene Ontology (GO), rpkm.csv data were used to identify 50 genes in each signature pathway. log-transform, differential mean value as Input Matrix Transformation, and Euclidean Distance (Complete Linkage) for sample and signature clustering metric were used.

Whole Genome Library Preparation and Exome Sequence Analysis

Purified genomic DNA from 48 hours after serum starvation of ME180 and CRISPR-Cas9 derivative cell lines were quantified using Qubit dsDNA HS. Libraries were prepared using SparQ DNA Frag Library Prep kit (Quanta Bio) according to manufacturer's protocol. Briefly, 100 ng of DNA was fragmented and end-repaired for 15 minutes at 32°C followed by 30 minutes at 65°C. Ligation was performed using Illumina TruSeq Single Index Adapters (catalog no. 20015960 and 20015961 - Illumina). The final PCR was performed according to manufacturer's instructions for a total of five PCR cycles. Final libraries were sequenced as 1 × 50 on a HiSeq3000 (llumina; refs. 41, 42).

Genomic SNPs and insertions/deletions (InDel) variants were discovered using the GATK4 best practices workflows (https://gatk.broadinstitute.org/hc/en-us/sections/360007226651-Best-Practices-Workflows). Paired-end sequencing reads were aligned to the Homo sapiens genome assembly hg38 with the Burrows-Wheeler Aligner (BWA) program (43). Aligned sequencing output from BWA was sorted and converted to binary alignment map (BAM) using samtools version 1.9 (44). Optical and PCR duplicates in the aligned BAM were then marked using the Picard MarkDuplicates tool (http://broadinstitute.github.io/picard) and base quality score recalibration was performed and applied using the GATK BaseRecalibrator and ApplyBQSR tools. Variant calling was performed using the GATK HaplotypeCaller in GVCF mode followed by consolidation and genotyping using the GATK CombineGVCFs and GentotypeGVCFs tools (45). Variants were classified and split as SNPs or InDels with the GATK SelectVariants command and hard filtered with the parameters (QD < 2.0, QUAL < 30.0, SQR > 3.0, FS > 60.0, MQ < 40.0, MQRankSum < −12.5, and ReadPosRankSum < −8.0) for SNPs and (QD < 2.0, QUAL<30.0, FS > 200.0, and ReadPosRankSum < −20.0) for InDels. The variants passed filters were annotated with the Genetic variant annotation toolbox SnpEff (46). Variant allele frequency was calculated as the ratio of observed variant depth divided by the overall depth at the variant locus (47). Density of the variants was counted in 1 Mb interval across the genome using the bedtools 2.29.2 (48) and used for the creation of Circos plots. Circos plos were created using the RCircos Package (49) and chromosome plots were created using the ggplot2 library (50). Source code available on request.

Mouse Xenograft and Tail Vein Injection Studies

Animal experimental studies were carried out with approval from the Institutional Animal Care and Use committee of the VAGLAHS. Cells (2 × 106) were injected into the right flank for subcutaneous xenograft tumor formation or into the tail vein for lung metastasis studies (33, 51). Tumors were measured daily, and the volume was calculated using the formula Length × Width2/2, where length is the larger measurement. Animals were sacrificed when the control ME180 xenograft tumors reached a size of 1,000 mm3 or after 2 months for tail vein injected animals. Tumor tissues were fixed in z-fix (aqueous buffered zinc formalin, Anatech Ltd.) and 5 µm tissue slices were used for IHC studies.

Statistical Calculations

Fisher exact test, two sided, was used for calculating significance of SNP to primary tumors. For the qRT-PCR, statistical analysis of differential expression was performed by one-way ANOVA with multiple pairwise comparisons with Sidak correction. All the studies including growth suppression studies were carried out at least in three independent experiments and statistical significance was calculated using the Student t test. All the results are presented as means ± SD and a P value of <0.05 was considered significant.

MTT assay, PCR and qRT-PCR studies, CRISPR-Cas9 Oligo preparation and Cloning, siRNA, Western Blotting, Immunoprecipitation, IHC, and Immunofluorescence methods are included in the Supplementary Data.

Data Availability

All the data including those presented in Supplementary Materials and Methods, Tables and figures are available to other investigators. RNA and genome sequencing data are available in the following sites: CRISPR-Cas9 clone: RNA-seq submission: Gene Expression Omnibus Submission (GSE179424). CRISPR-Cas9 clone: Exome seq submission: SRA Submission (SRP327422). Data and code used to create Circos plots archived in UCLA Dataverse: https://doi.org/10.25346/S6/IMXOOK.

Increased Frequency of the Exon 8 SNP (c.1293 C>T, rs2608555) in Primary Tumor versus Normal

The prevalence of GAN gene exon 8 SNP ranges from 8.1% in the East Asian population to 45.6% in the European population (15). This SNP has also been identified to varying degrees in human tumors in the COSMIC, DriverDB, and ICGC databases. However, the association to clinical phenotypes is not known. To determine the SNP frequency in primary tumors, 53 normal endometrial tissues adjacent to tumors, 70 head and neck, and 52 cervical cancers were studied. For 19 cervical cancer samples, we had adjacent normal endometrium. Rest of 34 normal DNAs were random samples derived from normal squamous epithelial cells. Ethnicity of the normal or tumor tissues were not known. RFLP analysis with the enzyme TspGW1 was performed using the PCR product of exon 8 primers (Supplementary Table S1). The SNP (T allele) is recognized by the presence of a 287 bp product while the wild-type (C allele) yields 155 and 132 bp products as shown in Supplementary Fig. S3A). Our analysis identified exon 8 SNP in 24.5% of normal samples, comparable to the frequency observed in the International HapMap Project (Supplementary Table S2). Tumor samples had 44.6% and 48% frequency in head and neck and cervical cancers, respectively. High frequency in tumor samples could be due to tumors belonging to subjects of European ethnicity, higher fraction treated at the three hospitals. There was no association to HPV in the tumor samples. However, there was a significantly higher SNP frequency in primary tumors versus normal tissues (13/53 normal vs. 56/122 tumors; P = 0.011). Analysis using Fisher two-sided comparison did not show a statistical significance to recurrence and metastasis (P = 0.171), to C/T or T/T alleles (P = 0.671 for C/T and P = 0.390 for T/T alleles (Supplementary Table S3). Both genomic (C/C, C/T, and T/T) and RNA expression (T vs. C allele) status of primary and recurrence/metastatic tumors including those of HPV positivity will be required to confirm and extend our findings on the relationship between gigaxonin expression and tumor development.

qRT-PCR analysis of RNAs isolated from FFPE samples available from one normal endometrium and six head and neck cancers were performed and differential cycle numbers with respect to GAPDH (ΔCt values) was used to calculate relative transcript levels using HeLa cell (although it is a tumor cell line control) GAPDH RNA-seq numbers (164,185 transcripts) as control. The results showed higher gigaxonin expression in two of three cancers (#17 and 18) containing GAN gene exon 8 T allele in their genome (Fig. 1A). There was higher expression of CDH1 (e-cadherin, a representative marker of epithelial cell phenotype) in these two primary tumors. A third primary tumor (#19) containing exon 8 T/T alleles had reduced gigaxonin and e-cadherin expression. The discordance between the presence of T allele and absence of gigaxonin expression may be related to RNA instability through other mutational events. One of the cancers (#8R), a recurrence tumor, containing C alleles was positive for CDKN2A (p16) by qRT-PCR and IHC. However, this tumor showed decreased expression of gigaxonin and e-cadherin. Of the two other tumors with C alleles, one of them (#12) had higher gigaxonin expression, but no e-cadherin expression. The other tumor (#11) was p16 positive but the expression of gigaxonin, and e-cadherin was absent. Although the results are preliminary, our data pointed to a direct relationship between the presence of T allele and expression of gigaxonin and e-cadherin in primary head and neck cancer.

Presence of Exon 8 SNP Correlates with Higher Expression of Gigaxonin in ME180 (HPV18-Positive) and HT3 (HPV-Negative) Cell Lines

To determine the association between GAN gene exon 8 SNP and gigaxonin expression, we analyzed five head and neck squamous cell carcinoma (HNSCC) and seven cervical cancer cell lines. The neuroblastoma cell line LAN6 was used as the HPV-negative control. These cell lines were authenticated using mitochondrial DNA sequencing and cellular morphology (Supplementary Figs. S1 and S2). The exon 8 genomic analysis revealed homozygous SNP (T/T) alleles in two cervical cancer cell lines (ME180 and HT3), heterozygous (C/T) alleles in four cell lines (three cervical cancer and one HNSCC) and homozygous wild-type (C/C) alleles in the other seven cell lines (Supplementary Table S4). Sequencing of the PCR and RT-PCR products derived using genomic and exon specific primers confirmed the presence of T allele in the six cell lines (two with T/T and four with C/T alleles). We did not find missense or stop codon mutations in any of the cancer cell lines.

RT-PCR and qRT-PCR were performed on the cell lines using genome specific primers (Supplementary Table S5). Again, differential cycle numbers with respect to GAPDH (ΔCt values) was used to calculate relative transcript levels using HeLa cell (tumor cell line control) GAPDH RNA-seq numbers (164,185 transcripts) as control. In addition to gigaxonin we focused our attention on the expression of CDKN2A (p16) and CCND1 (cyclin D1) as we have shown inverse relationship between the two proteins in our studies on head and neck cancer (28). Expression of e- and n-cadherins shown in epithelial versus mesenchymal phenotype and vimentin shown to be ubiquitinated by gigaxonin (3) were also studied. Analysis performed on RNAs from 24 hours after serum starved cervical cancer cell lines HT3, C33A, HeLa, ME180, and head and neck cancer cell line UM-SCC1 showed higher gigaxonin expression in all four cervical cancer cell lines (Fig. 1B). The cervical cancer cell lines were also positive for CDKN2A (p16 expression). However, HT3 and ME180 containing T alleles are the only cell lines with higher CDH1 (e-cadherin) expression pointing to a direct association between gigaxonin and e-cadherin expression. Vimentin expression was higher in HeLa cells. The results from the steady state (no serum starvation—heterogenous cell population) cell line samples showed CDKN2A expression in cervical cancer cell lines and a head and neck cancer cell line, CCL23 (Fig. 1C). There was higher GAN (gigaxonin) expression in ME180 and C33A cells and reduced expression in all other cell lines. Expression of e-cadherin was observed only in ME180 cells. Expression of mesenchymal marker n-cadherin was seen at a low level in two head and neck cancer cell lines, UM-SCC12 and CCL23. While higher CCND1 (cyclin D1) expression was seen in all the cell lines except SiHa, Vimentin was expressed in two each of cervical cancer (HeLa and C33A), and head and neck cancer (UM-SCC12 and CCL23) cell lines.

Time course expression analysis of CDKN2A and GAN at 0, 6, and 24 hours after a 24-hour serum starvation was performed to measure the relationship between SNP and stability of GAN transcripts during cell cycle. Fibroblast cell line GM05399 was used as normal cell line cell cycle control and GAPDH was used as housekeeping gene control. Similar pattern of CDKN2A expression, higher at 6 hours and reduced at 24 hours was seen in all four cervical cancer cell lines (Supplementary Fig. S3B). However, gigaxonin expression was stable in ME180 and HT3 in comparison with HeLa and C33A cell lines (Supplementary Fig. S3C). Expression of both CDKN2A and gigaxonin was not detected in head and neck cancer cell lines UM-SCC1 and UM-SCC14A.

RNA-seq analysis of ME180 and HeLa cell lines showed a 3-fold increased gigaxonin expression in ME180 in comparison with HeLa cells (Fig. 1D). The data further showed statistically significant (P < 0.05) upregulation of e-cadherin (CDH1), and CD24 genes and downregulation of Zeb1, and vimentin in ME180 in comparison with HeLa cells. Snapshot view (IGV: interactive genome browser) of exon 8 SNP area of RNA-seq showed T allele expression in ME180 and C allele in HeLa cells (Fig. 1E). There was an 11-fold increase in transcript reads (33:3) in ME180 in comparison with HeLa cells. Although both C and T alleles were present in the HeLa cell genome (Supplementary Table S4), expression was seen only for the C allele (vertical line in Fig. 1E). T allele expression was not observed pointing to lesser exon 8 expression in HeLa cells when the expression was from the C allele.

Western blot protein analysis showed highest expression of gigaxonin in ME180 cells containing T/T alleles (Fig. 1F). There was higher expression, but less than that of ME180 in HT3 cells. Cell line C33A is a faster growing cell line compared with HT3 both in the MTT and soft agar colony forming assays (Supplementary Fig. S4A–S4C). Although C33A cells had a similar GAN transcript level of ME180 in both the serum starved and non-serum starved steady state level conditions (Fig. 1B and C), protein expression was reduced because of expression of the C allele in the RNA sequence (Supplementary Fig. S4D). There was reduced expression in cell lines containing C/T alleles and minimal or absence of expression in C/C allele containing HNSCC cell lines. As observed in the RNA-seq, there was increased expression of e-cadherin in ME180 cells while the expression was absent in HeLa and C33A cell lines (Fig. 1G). SiHa, and head and neck cancer cell lines had reduced e-cadherin expression with the least expression in UM-SCC14A cells. High expression of Zeb1 was observed in HeLa and C33A cells and there was reduced expression in SiHa and UM-SCC12 cells. There was minimal or no Zeb1 expression in other cell lines. There was expression of Sox2 and Twist1 in some cell lines. Vimentin expression was highest in UM-SCC12 cell lines and a reduced expression in HeLa cells (Fig. 1H). Other cell lines, including ME180 and HT3, did not show vimentin expression.

Inverse Relationship Between Gigaxonin Expression and In Vitro Cell Growth

To examine the association between gigaxonin expression and cell growth, MTT and soft agar colony formation assays were performed. HeLa and ME180 contain HPV18, HeLa and SiHa contain HPV18 and HPV16 sequences, respectively. Studies were performed in all three cell lines at the same time. Because of differential growth of SiHa (higher growth) and ME180 (slower growth), data are presented as ME180 versus HeLa and HeLa versus SiHa cells. ME180 cells showed statistically significant (P < 0.0001 for 10,000 cells and 0.042 for 20,000, cells respectively) reduced cell growth and soft agar colony formation in comparison with HeLa cells (Fig. 2A and B). Representative soft agar colonies are shown in Supplementary Fig. S5A. The data further revealed reduced cell proliferation and increased cisplatin sensitivity of HeLa cells in comparison with SiHa cells (Fig. 2C and D; Supplementary Fig. S5B). Our results therefore indicated that HPV18 containing cells are less aggressive and more sensitive to cisplatin than HPV16 containing cervical cancer cell lines.

To further identify the relationship of gigaxonin expression to cisplatin sensitivity, ME180 cells with the highest gigaxonin expression was treated with 0.1 to 3.0 µg/mL (0.34 to 10 µmol/L concentration) of cisplatin, and cell growth and protein expression were measured. The Incucyte (Real-time Cell Imager) growth assay showed increased sensitivity with increased cisplatin concentration (Fig. 2E). Cell images acquired during Incucyte growth assay showed presence of live cells 72 hours after 1.0 and 3.0 µg/mL of cisplatin treatment (Supplementary Fig. S6). MTT assay performed 6 days after cisplatin treatment confirmed the presence of live cells in cisplatin-treated cells (Fig. 2F). Protein expression analysis of cisplatin-treated cells showed upregulation of NFκB at 3.0 µg/mL cisplatin correlating to the development of cisplatin resistance (Fig. 2G). There was downregulation of p16 and gigaxonin expression correlating to the inverse relationship between the two proteins to NFκB and sensitivity to cisplatin. While we have observed an inverse relationship between in vitro cell growth and GAN gene exon 8 SNP, other intrinsic properties of each cell line could play a role in the differential phenotype of cancer cell lines.

CRISPR-Cas9–mediated Conversion of Exon 8 SNP (T/T) to Wild-type C/C Alleles Results in Reduced Gigaxonin Expression, Increased In Vitro Cell Growth and Enhanced Epithelial–mesenchymal Transition Marker Expression

To explore the relationship further, we decided to convert T/T alleles of ME180 cells to C/C alleles using the CRISPR-Cas9 system. We prepared four different guide RNAs (gRNA) containing PAM sequences for the exon 8 SNP site (Fig. 3A; Supplementary Table S6). Exon 8 PCR products of the DNA were subjected to TspGW1 restriction enzyme analysis to identify the conversion of T to C allele (Fig. 3B). Puromycin-resistant single-cell clones were isolated using antibiotic selection and non-selection for 2 to 3 weeks. Clones of gRNA 1, C1.2, C1.2, and C1.3 had similar morphology to ME180 with the retention of the 287 bp exon 8 PCR product indicating the retention of the SNP (Supplementary Fig. S7). These clones were not pursued further. However, gRNA 4 clones C4.15.1, and C4.15.3 showed a flat cell-cell adherent mesenchymal cell morphology with low light transmission (Fig. 3C). Presence of C/C alleles in in these CRISPR-Cas9 clones (now on called GAN edited clones) was seen in the exon 8 PCR products (155/132 bp product in TspGW1 restriction enzyme analysis) which was confirmed by DNA sequencing (Fig. 3D).

Western blot analyses showed decreased gigaxonin expression in GAN edited cells indicating a direct relationship between T>C conversion and gene expression (Fig. 3E). Zeb1 expression was observed in cervical cancer cell lines HeLa, C33A, and SiHa, while it was absent in ME180 cells. We therefore searched for other epithelial–mesenchymal transition (EMT) markers and found ME180 cells to express Snail and Twist at basal levels and focused on these two proteins (Snail in later experiments) and cadherins in ME180 and GAN edited clones. There was enhanced expression of Snail, Twist1, and n-cadherin accompanied by reduced e-cadherin expression in the GAN edited clones in comparison with parental ME180 cells (Fig. 3E). Incucyte growth assay showed faster growth and increased resistance to cisplatin of GAN edited clones in comparison with ME180 cells (Fig. 3F).

Reversal of EMT Marker Phenotype with Lentiviral-mediated Re-expression of Gigaxonin in GAN Edited Cells

To validate the relationship between gigaxonin expression and tumor cell growth, we reintroduced lentiviral cloned GAN gene (Supplementary Fig. S8) using an established transduction protocol. A clone each of the control (LV Con 25, 50, and 250) and GAN (LV GAN 25, 50, and 250) transductions were analyzed. The aliquots 25, 50, and 250 correspond to 0.5 × 104, 1 × 104, and 2.5 × 104 lentiviral particles, respectively. Re-expression of gigaxonin in GAN edited cells showed an intermediate epithelial morphology with a brighter light transmission than GAN edited cells or control lentiviral clones (Fig. 4A).

Exome sequence analysis of DNA samples from the control vector and GAN virus transductions (50 µL aliquot corresponding to 1 × 104 virus particle transfected cells) showed identical peaks for chromosomal deletions (del plot), interchromosomal deletions (indel plots), and SNP plot in ME180, GAN edited and control or GAN vector transfected cells (Fig. 4B; Supplementary Figs. S9 and S10). The circos plots of the del, indel, and SNPs further demonstrated that GAN edited cells were derived from the parental ME180 cells (Supplementary Fig. S11A–S11C). Similarly, the lentiviral transfected cells were identical in genetic composition to the GAN edited C4.15.1 cells. The non-synonymous mutation analysis showed unique peaks in ME180 cells that were absent in other cell lines. A Venn diagram showed that two-thirds of sites (5,369 sites) were identical in all four cell lines and one-third (3,861 sites) was unique to ME180 (Supplementary Fig. S11D). Thus, the GAN edited cells were mostly identical for synonymous and non-synonymous chromosomal regions to the parental ME180 cells. GAN clone isolation from single-cell cloning through puromycin selection and non-selection and the isolation of cells with a morphology different from that of parental ME180 cells could explain genomic differences between the parental ME180 cells and the GAN clone ME180/C4.15.1. The difference between the GAN clone and lentiviral control or GAN lentiviral transfected cells was identical in (8,878 sites Supplementary Fig. S11D). Only five sites were unique to GAN edited ME180/C4.15.1 cells and 13 and 9 unique sites were present in control and GAN lentiviral transfected cell lines, respectively. Western blot studies revealed enhanced expression of gigaxonin and e-cadherin in all the GAN re-expression cells compared with control virus transfected clones (Fig. 4C). We also found decreased expression of Snail and n-cadherin indicating reversal of the EMT phenotype in gigaxonin re-expressing cells (Fig. 4C).

Differential expression analysis of the RNA-seq confirmed gigaxonin expression in GAN lentiviral transfected cells and partial reversion of genes of which SerpinB2 codes for a protein involved in senescence and autophagy (Supplementary Fig. S12). GO analysis data showed near reversal to ME180 signaling pathways including that of myogenesis and TNFα signaling via NFkB with the re-expression of gigaxonin (Fig. 4D; Supplementary Fig. S13). Thus, the RNA-seq data confirmed our earlier report on the effect of gigaxonin on NFκB (30) and suggested that NFκB might mediate downregulation of Snail. Both RNA sequence and Western blots suggested that many of the ME180 phenotypes are reversible with re-expression of gigaxonin.

Gigaxonin Re-expression Inhibits In Vitro Cell Proliferation and Matrigel Invasion

A snapshot view of the exon 8 RNA-seq data pointed to high expression of gigaxonin in ME180 and GAN lentiviral transfected cells (Supplementary Fig. S14). Red vertical bar indicating T allele expression correlating to higher gigaxonin expression was seen in ME180 cells. Deletion of exon 8 sequences (red horizontal lines) was observed in GAN edited and control lentiviral transfected cell lines.

Furthermore, the Incucyte growth assay showed statistically significant reduced growth of GAN re-expressing cells in comparison with control lentiviral vector cells (P < 0.001, Fig. 5A). The reduced growth seen in control vector transfected cells in comparison with that of parental GAN edited cells was not significant (P = 0.1088). Because GAN edited cells migrated through soft agar and did not form recognizable colonies, we performed migration and Matrigel invasion assays using 8 µmol/L cell culture filters. The analysis showed statistically significant increased migration (P < 0.0001) and Matrigel invasion (P < 0.0029) of GAN edited cells in comparison with ME180 cells (Fig. 5B–E). GAN lentiviral transfected cells showed statistically significant reduced migration (P < 0.0001) and Matrigel invasion (P < 0.0013) in comparison with control vector transfected cells. Control lentiviral cells had a higher migration rate, but a similar Matrigel invasion to that of the GAN edited cells.

To confirm that loss of gigaxonin expression leads to aggressive cell growth, another protocol, that of siRNA transfection was performed. GAN siRNA transfected cells had reduced gigaxonin expression and statistically significant increase in migration (P < 0.0014) and Matrigel invasion (P < 0.0008) in comparison with untreated, liposome alone and control siRNA treated ME180 cells (Supplementary Fig. S15A–S15E). From both the CRISPR-Cas9 editing and siRNA transfection studies, we have confirmed that the loss of gigaxonin expression leads to aggressive in vitro cancer cell line growth and growth suppression occurs with the re-expression of gigaxonin.

Biochemical studies showed that siRNA transfected cells have increased expression of NFκB and Snail with the loss of gigaxonin expression. Gigaxonin reexpression through GAN cDNA showed reduced expression of NFκB and Snail (Supplementary Fig. S16A). Because HeLa cells do not express Snail, but express Zeb1, we re-expressed gigaxonin through GAN cDNA. There was reduced Zeb1 expression indicating downregulation of EMT markers by gigaxonin (Supplementary Fig. S16B).

Reduced Mouse Lung Metastasis with the Re-expression of Gigaxonin

To identify the effect of gigaxonin on in vivo tumor growth, ME180, GAN edited and gigaxonin re-expressing cell lines were used for mouse subcutaneous xenograft tumor formation. Injection of ME180 cells (2 × 106 cells per mouse) underneath the skin of immune compromised nude mice (NSG mice) showed xenograft tumor formation in 6 of 6 mice (Fig. 6A; Supplementary Fig. S17). Tumors grew to a size of 100 to 600 cmm in 4 weeks. Small tumors (60 cmm size) were seen in 4 of 6 mice with GAN lentiviral transfected cells. GAN edited cells formed a small tumor (50 cm) in one mouse. Tumor growth was not observed in the other 5 mice and in all 6 of 6 mice of control lentiviral transfected cells. Absence of tumor formation in GAN edited and control lentiviral transfected cells indicated that the cells did not attach to the skin.

To explore the possibility of in vivo invasion and metastasis, tail vein injections (2 × 106 cells per mouse) were carried out in NSG mice, and the mice were sacrificed after 2 months. Lung, liver, and colon were examined by hematoxylin and eosin (H&E). There was no liver or colon metastasis in any of the animals. While there were tiny tumors at the base of the tail (may be from a fraction of cells attaching to inner endothelial cell wall), there was no lung metastasis in ME180 cells (Fig. 6B). GAN edited and control vector transfected cells formed metastatic lung tumors. There was >50% infiltration of tumor cells into the lung in these two cell lines. However, the GAN lentiviral transfected cells showed reduced (>50% reduction in comparison with the control lentiviral cells) metastasis. These results showed that ME180 cells expressing higher gigaxonin (through exon 8 T allele) did not form distant metastasis. GAN edited cells with reduced gigaxonin expression correlated with the development of metastasis. Re-expression of the GAN gene leads to reduced metastasis indicating gigaxonin as a metastasis suppressor protein.

To determine the relationship between EMT and lung metastasis, paraffin fixed lung tissue slides were hybridized to e-cadherin and Snail antibodies. Prostate cancer with e-cadherin positivity, and Snail negativity and lobular breast cancer with negative e-cadherin and positive cytoplasmic Snail expression were used as controls (Supplementary Fig. S18). There was high e-cadherin expression in the normal lung epithelium of ME180 cell injected mice (Fig. 6C). Expression was reduced in GAN edited and control lentiviral tumor cells. Statistically significant enhanced e-cadherin expression was observed in GAN lentiviral tumor cells (P < 0.01, Fig. 6C and D). ME180 cells did not form metastasis. Thus, ME180 cell injected mice showed normal lung and Snail expression was seen in normal lung pneumocytes (Fig. 6E). GAN re-expression leads to 50% reduced metastasis in comparison with the control GAN edited and control lentiviral transfected cells. Thus, there was statistically reduced Snail expression in comparison with control lentiviral transfected cells (P < 0.01, Fig. 6E and F). The brown stains in the non-tumor regions of GAN re-expressing cells might reflect normal pneumocytes of the lung. Thus, both in vitro and mouse in vivo studies showed that gigaxonin expression has a direct relationship to e-cadherin and inverse relationship to Snail.

Indirect Mechanism of Snail Downregulation by Gigaxonin

To determine the effect of gigaxonin on exogenously expressed Snail and to identify the mechanism of gigaxonin-Snail relationship, transfection studies were performed using 5 µg each of FLAG-tagged Snail and GAN plasmid DNAs. Two sets of gene transfected cells, one split 1:2 after 48-hour transfection and grown for an additional 24 hours, and the second set grown without splitting for 72 hours after transfection, were analyzed.

Both the control and plasmid transfected cells showed reduced expression of the higher molecular weight gigaxonin in 48-hour split cells (Fig. 7A). There was increased FLAG and Snail (combined endogenous and exogenous Snail) expression in gene transfected cells in comparison with liposome transfected cells. The unsplit cells revealed higher gigaxonin expression and loss of Snail expression (using both the FLAG and Snail antibodies) confirming Snail degradation with the expression of gigaxonin. There was reduced actin expression in the unsplit cells that could be due to gigaxonin mediated regulation of actin or the effect of combined Snail and gigaxonin expression.

Western blots of immunoprecipitated proteins with FLAG and Snail antibodies hybridized to a protein with similar size as that of the control IgG immunoprecipitated proteins (Fig. 7B). This indicated that it is not possible to demonstrate Snail interaction with immunoprecipitation studies. It is also possible that the hybridizing protein is not Snail, but IgG. In the reverse experiment where the FLAG or Snail immunoprecipitated proteins were hybridized to gigaxonin, we did not observe hybridization to the 60 or 140 kD gigaxonin protein confirming that there is no direct interaction between gigaxonin and Snail.

Because Snail expression is lost with the overexpression of gigaxonin (Fig. 7A; Supplementary Fig. S16A), we wanted to find out whether Snail degradation occurs through ubiquitination. We performed expression studies on cell lines treated with inhibitors of proteosomal (MG132 and n-ethylmaleimide) and autophagy (bafilomycin A) pathways. As has been shown earlier for the activity of MG132/NEM and bafilomycin A on cyclin D1 (36, 37), we found increased cyclin D1 expression in the presence of proteosomal inhibitors (MG132 and NEM) and decreased expression in the presence of bafilomycin A confirming that the inhibitors were functionally active (Fig. 7C). Proteosomal inhibitors did not affect the expression of gigaxonin and p16. We observed a 50% reduced expression of p16 and gigaxonin with bafilomycin A indicating a direct association between these two proteins and cell cycle blockage at G1–S boundary by bafilomycin A.

With the loss of gigaxonin in GAN edited cells, there was increased expression of Snail in the presence of proteosomal inhibitors, but there was no effect of bafilomycin A in comparison with the untreated cells (Fig. 7C). Absence of increased Snail expression with proteosomal inhibitors in the control ME180 cells could be attributed to low level expression of Snail and robust expression of gigaxonin even in the presence of proteosomal inhibitors. We also found increased NFκB expression in GAN edited cells correlating to decreased NFκB ubiquitination in the absence of p16 and gigaxonin. Absence of an effect by bafilomycin A suggesting that NFκB degradation does not involve autophagy.

Immunofluorescence analysis was done on ME180 cells to identify cellular expression of gigaxonin and Snail. There was overlapping expression of the two proteins in the cytoplasm, surrounding the nucleus (Fig. 7D). There were also cells expressing gigaxonin or Snail alone. Nuclear expression of gigaxonin was observed in 10% to 20% of cells (indicated by arrows).

To determine whether Snail ubiquitination occurs correlating to gigaxonin-mediated downregulation of Snail, proteosomal inhibitor–treated ME180 and GAN edited cells were immunoprecipitated with Snail and hybridized to ubiquitin antibody. Higher molecular weight proteins were not seen in ME180 cells pointing to the absence of Snail ubiquitination (Supplementary Fig. S19A). To probe Snail downregulation further, control and GAN siRNA transfected and proteosomal inhibitor–treated ME180 cells were hybridized to NFkB. There was increased NFκB expression with GAN siRNA in comparison with control siRNA (Supplementary Fig. S19B). Ubiquitin hybridization showed the presence of higher molecular weight bands in ME180 cells (Supplementary Fig. S19C). Lighter intensity bands could also been in control liposome and siRNA-treated cells. These bands were absent in GAN siRNA treated cells indicating gigaxonin is involved in NFκB ubiquitination and loss of gigaxonin leads to loss of NFκB ubiquitination.

Loss of Exon 8 T Allele Expression is Associated with the Loss of Gigaxonin Multimeric Protein Formation and mRNA Instability

To identify the molecular mechanism of association between exon 8 T allele and gigaxonin expression, RNA stability assays were performed. Activation of DNA damage response represented by γH2AX expression was used as a marker of transcription inhibitor actinomycin D treatment in ME180 cells. Enhanced γH2AX expression was observed with different concentrations of actinomycin D (Supplementary Fig. S20). There was cell death by 24 hours even for treatment with 0.1 µg/mL of the drug. Maximal γH2AX expression observed in 7 hours for the 10 µg/mL concentration was used in subsequent experiments. Higher expression of γH2AX expression was observed in drug-treated HeLa and HT3 cells in comparison to that of ME180 and SiHa cells (Fig. 8A). Multimeric forms of gigaxonin observed in ME180, HeLa, and HT3 cells was absent in SiHa cells. Because GAN gene mutations are not present in SiHa cells, we suggest that loss of multimeric protein formation, a posttranslational event, could also be associated with the loss of T allele expression.

RNA expression levels as measured by qRT-PCR showed half-life of GAN mRNA to be 3.22 hours in HeLa cells (P = 0.0060) while it was 5.09 hours for ME180 (P = 0.0081) and 7.88 hours for SiHa cells (P = 0.0003; Fig. 8B). The mRNA was stable in HT3 cells. While ME180 cells had increased mRNA instability than SiHa cells, there was an association between loss of exon 8 T allele expression and mRNA instability in HeLa cells in comparison with T allele expressing ME180 and HT3 cell lines. As a confirmation of actinomycin D effect, we found that the half-life of HeLa cell p16 to be 5.49 hours (P = 0.0016) and cyclin D1 RNA to be stable (P = 0.0160; Fig. 8C). The half-lives of the two proteins matched with half-lives (4.16 hours for p16 and stable expression for cyclin D1) reported in the literature (36, 37). Our results therefore showed that actinomycin D treatment was functional in the four cancer cell lines and that absence of T allele expression was associated with gigaxonin mRNA instability in HeLa cells.

To determine whether NFκB is responsible for the upregulation of Snail that in turn results in downregulation of e-cadherin, qRT-PCR analysis was performed on ME180 cells transfected with RelA cDNA or TNFα (20 µg/mL) up to 1 hour. There was activation of NFκB in RelA transfected cells in comparison with that of control cDNA transfected cells (Fig. 8D). There was enhanced NFκB expression 15 minutes after exposure to TNFα in comparison with untreated control cells. Snail was transcriptionally activated with RelA cDNA and at 30 and 60 minutes after exposure to TNFα. Time delay between NFκB and Snail activations could be attributed to a delay in NFκB-mediated signaling. Downregulation of e-cadherin coincided with Snail activation pointing to transcriptional downregulation of e-cadherin by Snail confirming the direct relationship between NFκB and Snail and inverse relationship between Snail and e-cadherin as was reported by others (52–57). We have also observed transcriptional downregulation of GAN after NFκB activation, as was seen with cisplatin treatment (see Fig. 2G) pointing to a feedback regulation between these two genes. From our results, we propose a mechanism where gigaxonin regulates NFκB through ubiquitination and NFκB mediated transcriptional activation of Snail results in transcriptional downregulation of e-cadherin (Fig. 8E).

GAN gene mutations have not been reported in head and neck cancer. Also, there is no information in databases for exon 8 SNP and head and neck cancer survival. However, the Stanford University cancer and survival database https://precog.stanford.edu shows expression of p16, gigaxonin, and e-cadherin (CDH1) to be associated with better overall survival and expression of Snail, Twist1, and n-cadherin (CDH2) to be associated with poor overall survival of head and neck cancer (Supplementary Fig. S21). We hypothesize that exon 8 T allele expression showing a direct relationship to gigaxonin expression and inverse relationship to EMT markers, plays a role in the inhibition of aggressive tumor cell growth, thereby improving survival of cancer subjects.

Snail is a transcription factor that inhibits e-cadherin expression accompanied by progression to the EMT phenotype. Snail is reported to be overexpressed in aggressive head and neck and other human cancers (52, 57). It has also been shown that the loss of Snail through inhibitors leads to the reversal of EMT phenotype. Conversely, overexpression of Snail is shown to convert less aggressive head and neck cancer to aggressive cancer, strongly correlating Snail expression to tumor cell metastasis (57–60). While stabilization of Snail through BDR4 or TGFβ-activated USP27X deubiquitinase has been shown, destabilization has not yet been demonstrated (57). In the absence of GAN expression in head and neck cancer cell lines, we used ME180, a cervical cancer cell line with higher levels of GAN and basal Snail expression to study the relationship between these two proteins. We show that Snail deregulation occurs with gigaxonin expression resulting in the inhibition of EMT. Our studies indicate that this inhibition is due to gigaxoin-mediated ubiquitination of NFκB.

Although we isolated T>C converted faster growing GAN edited cells from slow growing ME180 cells, the reverse isolation of C>T of HeLa cells was not successful. This could be due to fast growing parental HeLa cells masking the isolation of slow growing GAN edited cells. It is to be noted that HeLa and C33A cells contain both C and T alleles and yet express only the C allele, indicating loss of slow growing T allele containing cells during the isolation of these cell lines.

Our previous investigation (30) and current results (see Fig. 7C; Supplementary Fig. S19B and S19C) have shown downregulation of NFκB through ubiquitination. Protein interactome studies have shown that NFκB interacts with several EMT signaling proteins including Snail, Twist, and CBP/p300: Protein–Protein Interactions » NFκB Transcription: https://www.bu.edu/nf-kb/physiological-mediators/interacting-proteins. These studies have also shown that there is an interaction between NFκB and actin, possibly explaining reduced β-actin expression in some gigaxonin-overexpressing cells. Because, equal concentration of proteins was loaded in Western blots, loading is not the reason for the actin downregulation shown in Fig. 7A. Tubulins have also been shown to be downregulated by gigaxonin making it difficult to use other loading controls (61). It has also been shown by Barberà and colleagues (52) that NFκB regulates Snail expression through transcriptional control. We have data pointing to transcriptional upregulation of Snail by NFκB which in turn leads to transcriptional downregulation of e-cadherin. We therefore suggest that gigaxonin downregulates Snail expression through downregulation of NFκB (Fig. 8E).

We have observed expression of Twist1 in GAN edited cells in the absence of p16 and gigaxonin (see Fig. 3). We have observed cervical cancer cell lines to have basal expression of Snail and Twist1, and head and neck cancer cell lines exhibiting higher Twist1 expression indicating a distinct biology governing cervical and head and neck cancers. In support of our data pointing to the relationship between p16, GAN, and Twist1, TCGA (http://ualcan.path.uab.edu/cgi-bin/TCGA-survival1.pl?genenam=TWIST11&ctype=HNSC) and (http://www.progtools.net/gene/index.php database; this later database is cited in https://bmccancer.biomedcentral.com/articles/10.1186/1471–2407–14–970) shows that, overexpression of CDKN2A (p16; P = 0.0756202), and GAN (P = 0.0108553) are associated with better overall survival of patients with metastatic head and neck cancers (Supplementary Fig. S22). The combined effect of CDKN2A and GAN expression for better overall survival is statistically significant (P = 0.023389), although GAN alone (P = 0.0108553) has a better statistical significance. Furthermore, recurrence of primary laryngeal cancers is directly associated with over expression of Twist1 (P = 0.0093843). We suggest therefore that, in addition to Snail, expression of Twist1 is also a prognostic marker of metastatic head and neck cancer.

Gigaxonin has many partners, probably >100 (8). For now, all known partners are shown to be regulated directly: Gigaxonin interacts with the rod domain of all IF proteins (explaining why all IF are aggregating in patients and animal models), and with the WD40 domain of ATG16L1. These findings have been shown by direct (immunoprecipitation), and indirect (proximity ligation assay staining) binding studies and reveal a direct interaction between gigaxonin and its partner proteins. We show in the present investigation that downregulation of Snail by gigaxonin is indirect. We demonstrate that downregulation of NFκB leads to downregulation of Snail which in turn leads to increased expression of e-cadherin and maintenance of epithelial phenotype. Loss of gigaxonin leads to activation of NFκB and Snail and inactivation of e-cadherin resulting in the development of mesenchymal phenotype. In addition, our results suggest a feedback regulation between gigaxonin and NFκB that would require extensive future studies to confirm and extend the findings. Our data also provide opportunities for preclinical therapeutic studies for GAN activation using NFkB inhibitors.

There are multiple neurogenerative diseases that are implicated in human cancer (Supplementary Table S7). Depending on the type of protein that is accumulated in the disease, the relationship to cancer is altered. In the gradual neurodegenerative diseases Alzheimer's and Parkinson's, there is a lower cancer incidence. However, more aggressive neurodegenerative diseases such as Ataxia-telangiectasia seems to have increased cancer incidence through gene mutations, leading to inhibition of DNA damage repair pathway (Supplementary Table S7). In conclusion, we show for the first time that GAN gene product gigaxonin involved in the control of neuronal maintenance is also a human metastasis suppressor protein. We conclude that (i) GAN gene exon 8 SNP T allele expression is associated with RNA stability, (ii) loss of T allele results in the loss of gigaxonin expression, (iii) activation of NFκB, (iv) upregulation of Snail, (v) downregulation of e-cadherin, and (vi) development of chemoradiation resistance and faster tumor cell growth.

No disclosures were reported.

M.S. Veena: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing-original draft, writing-review and editing. J.J. Gahng: Conceptualization, data curation, software, formal analysis, validation, investigation, methodology, writing-original draft, writing-review and editing. M. Alani: Data curation, validation, investigation. A.Y. Ko: Data curation, investigation, methodology. S.K. Basak: Data curation, formal analysis, validation, methodology. I.Y. Liu: Data curation, formal analysis, investigation, methodology. K.J. Hwang: Data curation, formal analysis, investigation, methodology. J.R. Chatoff: Data curation, formal analysis, investigation. N. Venkatesan: Data curation, formal analysis, validation, investigation, methodology. M. Morselli: Data curation, investigation, methodology. W. Yan: Data curation, investigation, methodology. I. Ali: Data curation, investigation, methodology. K.E. Kaczor-Urbanowicz: Data curation, validation, investigation, methodology. B.S. Gowda: Data curation, formal analysis, validation, investigation, methodology. P. Frost: Data curation, investigation, methodology. M. Pellegrini: Data curation, formal analysis, validation, investigation, methodology, writing-original draft. N.A. Moatamed: Data curation, validation, visualization. S.P. Wilczynski: Resources, investigation, methodology. P. Bomont: Resources, validation, methodology. M.B. Wang: Resources, data curation, funding acquisition, validation. D.S. Shin: Resources, data curation, funding acquisition, validation, investigation, methodology, writing-review and editing. E.S. Srivatsan: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

We thank Dr. Jin Zhong, Chief Pathologist at VAGLAHS, West Los Angeles for helping us with lung metastatic tumor H&E pictures. RNA and DNA sequences were performed at the UCLA Jonsson Comprehensive Cancer Center (JCCC) sequencing Core Facility. Reagents for the CRISPR-Cas9 cloning and lentiviral vectors were received from Dr. Donald B. Kohn's laboratory at UCLA. We thank Dr. Steven Gray of UT Southwestern Medical Center for providing us the GAN plasmid. We thank Dr. Dennis Ruenger of David Geffen School of Medicine at UCLA for statistical evaluation of SNP in normal and primary tumors. This work was supported by the VA Greater Los Angeles Healthcare System, West Los Angeles Surgical Education Research program to E.S. Srivatsan. Support was also provided by the Academic Senate Grant of David Geffen School of Medicine at UCLA to M.B. Wang. D.S. Shin was supported by Career Development Award from Conquer Cancer Foundation/American Society of Clinical Oncology (ASCO). BX004280 of the VA Merit grant “Targeting the in Vivo Hypoxic Microenvironment of Multiple Myeloma as an Anti-Tumor Strategy” to P. Frost.

Note: Supplementary data for this article are available at Cancer Research Communications Online (https://aacrjournals.org/cancerrescommun/).

1.
Bomont
P
,
Cavalier
L
,
Blondeau
F
,
Hamida
CB
,
Belal
S
,
Tazir
M
, et al
.
The gene encoding gigaxonin, a new member of the cytoskeletal BTB/kelch repeat family, is mutated in giant axonal neuropathy
.
Nat Genet
2000
;
26
:
370
4
.
2.
Bomont
P
,
Koenig
M
.
Intermediate filament aggregation in fibroblasts of giant axonal neuropathy patients is aggravated in non dividing cells and by microtubule destabilization
.
Hum Mol Genet
2003
;
12
:
813
22
.
3.
Mahammad
S
,
Murthy
SNP
,
Didonna
A
,
Grin
B
,
Israeli
E
,
Perrot
R
, et al
.
Giant axonal neuropathy-associated gigaxonin mutations impair intermediate filament protein degradation
.
J Clin Invest
2013
;
123
:
1964
75
.
4.
Berg
BO
,
Rosenberg
SH
,
Asbury
AK
.
Giant axonal neuropathy
.
Pediatrics
1972
;
49
:
894
9
.
5.
Bomont
P
.
The dazzling rise of neurofilaments: physiological functions and roles as biomarkers
.
Curr Opin Cell Biol
2021
;
68
:
181
91
.
6.
Lescouzères
L
,
Bomont
P
.
E3 ubiquitin ligases in neurological diseases: focus on gigaxonin and autophagy
.
Front Physiol
2020
;
11
:
1022
.
7.
Arribat
Y
,
Mysiak
KS
,
Lescouzères
L
,
Boizot
A
,
Ruiz
M
,
Rossel
M
, et al
.
Sonic Hedgehog repression underlies gigaxonin mutation-induced motor deficits in giant axonal neuropathy
.
J Clin Invest
2019
;
129
:
5312
26
.
8.
Scrivo
A
,
Codogno
P
,
Bomont
P
.
Gigaxonin E3 ligase governs ATG16L1 turnover to control autophagosome production
.
Nat Commun
2019
;
10
:
780
.
9.
Chen
P-H
,
Hu
J
,
Wu
J
,
Smith
TJ
,
Pan
S
,
Bisnett
BJ
, et al
.
Gigaxonin glycosylation regulates intermediate filament turnover and may impact giant axonal neuropathy etiology or treatment
.
JCI Insight
2020
;
5
:
e127751
.
10.
Johnson-Kerner
BL
,
Roth
L
,
Greene
JP
,
Wichterle
H
,
Sproule
DM
.
Giant axonal neuropathy: an updated perspective on its pathology and pathogenesis
.
Muscle Nerve
2014
;
50
:
467
76
.
11.
Cheng
W-C
,
Chung
I-F
,
Chen
C-Y
,
Sun
H-J
,
Fen
J-J
,
Tang
W-C
, et al
.
DriverDB: an exome sequencing database for cancer driver gene identification
.
Nucleic Acids Res
2014
;
42
:
D1048
54
.
12.
Zhang
J
,
Bajari
R
,
Andric
D
,
Gerthoffert
F
,
Lepsa
A
,
Nahal-Bose
H
, et al
.
The International Cancer Genome Consortium Data Portal
.
Nat Biotechnol
2019
;
37
:
367
9
.
13.
Liu
S-H
,
Shen
P-C
,
Chen
C-Y
,
Hsu
A-N
,
Cho
Y-C
,
Lai
Y-L
, et al
.
DriverDBv3: a multi-omics database for cancer driver gene research
.
Nucleic Acids Res
2020
;
48
:
D863
-
70
.
14.
Kang
JJ
,
Liu
IY
,
Wang
MB
,
Srivatsan
ES
.
A review of gigaxonin mutations in giant axonal neuropathy (GAN) and cancer
.
Hum Genet
2016
;
135
:
675
84
.
15.
1000 Genomes Project Consortium
,
Auton
A
,
Brooks
LD
,
Durbin
RM
,
Garrison
EP
,
Kang
HM
, et al
.
A global reference for human genetic variation
.
Nature
2015
;
526
:
68
74
.
16.
Yates
AD
,
Achuthan
P
,
Akanni
W
,
Allen
J
,
Allen
J
,
Alvarez-Jarreta
J
, et al
.
Ensembl 2020
.
Nucleic Acids Res
2020
;
48
:
D682
8
.
17.
Faa′
V
,
Coiana
A
,
Incani
F
,
Costantino
L
,
Cao
A
,
Rosatelli
MC
.
A synonymous mutation in the CFTR gene causes aberrant splicing in an Italian patient affected by a mild form of cystic fibrosis
.
J Mol Diagn
2010
;
12
:
380
3
.
18.
Kimchi-Sarfaty
C
,
Oh
JM
,
Kim
I-W
,
Sauna
ZE
,
Calcagno
AM
,
Ambudkar
SV
, et al
.
A “silent” polymorphism in the MDR1 gene changes substrate specificity
.
Science
2007
;
315
:
525
8
.
19.
van den Bergen
JC
,
Hiller
M
,
Böhringer
S
,
Vijfhuizen
L
,
Ginjaar
HB
,
Chaouch
A
, et al
.
Validation of genetic modifiers for Duchenne muscular dystrophy: a multicentre study assessing SPP1 and LTBP4 variants
.
J Neurol Neurosurg Psychiatry
2015
;
86
:
1060
5
.
20.
Flanigan
KM
,
Ceco
E
,
Lamar
K‐M
,
Kaminoh
Y
,
Dunn
DM
,
Mendell
JR
, et al
.
LTBP4 genotype predicts age of ambulatory loss in Duchenne muscular dystrophy
.
Ann Neurol
2013
;
73
:
481
8
.
21.
Vacher
M
,
Porter
T
,
Villemagne
VL
,
Milicic
L
,
Peretti
M
,
Fowler
C
, et al
.
Validation of a priori candidate Alzheimer's disease SNPs with brain amyloid-beta deposition
.
Sci Rep
2019
;
9
:
17069
.
22.
Boizot
A
,
Talmat-Amar
Y
,
Morrogh
D
,
Kuntz
NL
,
Halbert
C
,
Chabrol
B
, et al
.
The instability of the BTB-KELCH protein Gigaxonin causes giant axonal neuropathy and constitutes a new penetrant and specific diagnostic test
.
Acta Neuropathol Commun
2014
;
2
:
47
.
23.
zur Hausen
H
.
Papillomaviruses in the causation of human cancers – a brief historical account
.
Virology
2009
;
384
:
260
5
.
24.
Slama
J
,
Sehnal
B
,
Dusek
L
,
Zima
T
,
Cibula
D
.
Impact of risk factors on prevalence of anal HPV infection in women with simultaneous cervical lesion
.
Neoplasma
2015
;
62
:
308
14
.
25.
Lin
K
,
Lu
X
,
Chen
J
,
Zou
R
,
Zhang
L
,
Xue
X
.
E6-associated transcription patterns in human papilloma virus 16-positive cervical tissues
.
Oncol Lett
2015
;
9
:
478
82
.
26.
Tian
Q
,
Li
Y
,
Wang
F
,
Li
Y
,
Xu
J
,
Shen
Y
, et al
.
MicroRNA detection in cervical exfoliated cells as a triage for human papillomavirus-positive women
.
J Natl Cancer Inst
2014
;
106
:
dju241
.
27.
Tang
C
,
Fuller
CD
,
Garden
AS
,
Awan
MJ
,
Colen
RR
,
Morrison
WH
, et al
.
Characteristics and kinetics of cervical lymph node regression after radiation therapy for human papillomavirus-associated oropharyngeal carcinoma: quantitative image analysis of post-radiotherapy response
.
Oral Oncol
2015
;
51
:
195
201
.
28.
Namazie
A
,
Alavi
S
,
Olopade
OI
,
Pauletti
G
,
Aghamohammadi
N
,
Aghamohammadi
M
, et al
.
Cyclin D1 amplification and p16(MTS1/CDK4I) deletion correlate with poor prognosis in head and neck tumors
.
Laryngoscope
2002
;
112
:
472
81
.
29.
Yip
HT
,
Chopra
R
,
Chakrabarti
R
,
Veena
MS
,
Ramamurthy
B
,
Srivatsan
ES
, et al
.
Cisplatin-induced growth arrest of head and neck cancer cells correlates with increased expression of p16 and p53
.
Arch Otolaryngol Head Neck Surg
2006
;
132
:
317
26
.
30.
Veena
MS
,
Wilken
R
,
Zheng
J-Y
,
Gholkar
A
,
Venkatesan
N
,
Vira
D
, et al
.
P16 and gigaxonin are associated with the ubiquitination of NFκB in cisplatin induced senescence of cancer cells
.
J Biol Chem
2014
;
289
:
34921
37
.
31.
Andrews
RM
,
Kubacka
I
,
Chinnery
PF
,
Lightowlers
RN
,
Turnbull
DM
,
Howell
N
.
Reanalysis and revision of the Cambridge reference sequence for human mitochondrial DNA
.
Nat Genet
1999
;
23
:
147
.
32.
Kalous
O
,
Conklin
D
,
Desai
AJ
,
O'Brien
NA
,
Ginther
C
,
Anderson
L
, et al
.
Dacomitinib (PF-00299804), an irreversible Pan-HER inhibitor, inhibits proliferation of HER2-amplified breast cancer cell lines resistant to trastuzumab and lapatinib
.
Mol Cancer Ther
2012
;
11
:
1978
87
.
33.
Soh
H
,
Venkatesan
N
,
Veena
MS
,
Ravichandran
S
,
Zinabadi
A
,
Basak
SK
, et al
.
Cystatin E/M suppresses tumor cell growth through cytoplasmic retention of NFκB
.
Mol Cell Biol
2016
;
36
:
1776
92
.
34.
Song
S
,
Krishnan
K
,
Liu
K
,
Bresalier
RS
.
Polyphenon E inhibits the growth of human Barrett's and aerodigestive adenocarcinoma cells by suppressing cyclin D1 expression
.
Clin Cancer Res
2009
;
15
:
622
31
.
35.
Yan
Y
,
Jiang
K
,
Liu
P
,
Zhang
X
,
Dong
X
,
Gao
J
, et al
.
Bafilomycin A1 induces caspase-independent cell death in hepatocellular carcinoma cells via targeting of autophagy and MAPK pathways
.
Sci Rep
2016
;
6
:
37052
.
36.
Zhang
X
,
Liu
Z
,
Yi
J
,
Tang
H
,
Xing
J
,
Yu
M
, et al
.
The tRNA methyltransferase NSun2 stabilizes p16INK4 mRNA by methylating the 3′-untranslated region of p16
.
Nat Commun
2012
;
3
:
712
.
37.
Bracken
CP
,
Wall
SJ
,
Barré
B
,
Panov
KI
,
Ajuh
PM
,
Perkins
ND
.
Regulation of cyclin D1 RNA stability by SNIP1
.
Cancer Res
2008
;
68
:
7621
8
.
38.
Veena
MS
,
Raychaudhuri
S
,
Basak
SK
,
Venkatesan
N
,
Kumar
P
,
Biswas
R
, et al
.
Dysregulation of hsa-miR-34a and hsa-miR-449a leads to overexpression of PACS-1 and loss of DNA damage response (DDR) in cervical cancer
.
J Biol Chem
2020
;
295
:
17169
86
.
39.
Basak
SK
,
Bera
A
,
Yoon
AJ
,
Morselli
M
,
Jeong
C
,
Tosevska
A
, et al
.
A randomized, phase 1, placebo-controlled trial of APG-157 in oral cancer demonstrates systemic absorption and an inhibitory effect on cytokines and tumor-associated microbes
.
Cancer
2020
;
126
:
1668
82
.
40.
Dobin
A
,
Davis
CA
,
Schlesinger
F
,
Drenkow
J
,
Zaleski
C
,
Jha
S
, et al
.
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
.
41.
Anders
S
,
Pyl
PT
,
Huber
W
.
HTSeq—a Python framework to work with high-throughput sequencing data
.
Bioinformatics
2015
;
31
:
166
9
.
42.
Anders
S
,
Huber
W
.
Differential expression analysis for sequence count data
.
Genome Biol
2010
;
11
:
R106
.
43.
Li
H
,
Durbin
R
.
Fast and accurate short read alignment with Burrows-Wheeler transform
.
Bioinformatics
2009
;
25
:
1754
60
.
44.
Li
H
,
Handsaker
B
,
Wysoker
A
,
Fennell
T
,
Ruan
J
,
Homer
N
, et al
.
The sequence alignment/map format and SAMtools
.
Bioinformatics
2009
;
25
:
2078
9
.
45.
McKenna
A
,
Hanna
M
,
Banks
E
,
Sivachenko
A
,
Cibulskis
K
,
Kernytsky
A
, et al
.
The genome analysis toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data
.
Genome Res
2010
;
20
:
1297
303
.
46.
Cingolani
P
,
Platts
A
,
Wang
LL
,
Coon
M
,
Nguyen
T
,
Wang
L
, et al
.
A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3
.
Fly
2012
;
6
:
80
92
.
47.
Strom
SP
.
Current practices and guidelines for clinical next-generation sequencing oncology testing
.
Cancer Biol Med
2016
;
13
:
3
11
.
48.
Quinlan
AR
.
BEDTools: the swiss-army tool for genome feature analysis
.
Curr Protoc Bioinformatics
2014
;
47
:
11.12.1
34
.
49.
Zhang
H
,
Meltzer
P
,
Davis
S
.
RCircos: an R package for Circos 2D track plots
.
BMC Bioinformatics
2013
;
14
:
244
.
50.
Villanueva
RAM
,
Chen
ZJ
.
ggPlot 2
.
Elegant graphics and data analysis
2019
;
17
:
160
7
.
51.
Sun
Q
,
Yang
Z
,
Li
P
,
Wang
X
,
Sun
L
,
Wang
S
, et al
.
A novel miRNA identified in GRSF1 complex drives the metastasis via the PIK3R3/AKT/NFκB and TIMP3/MMP9 pathways in cervical cancer cells
.
Cell Death Dis
2019
;
10
:
636
.
52.
Barberà
MJ
,
Puig
I
,
Domínguez
D
,
Julien-Grille
S
,
Guaita-Esteruelas
S
,
Peiró
S
, et al
.
Regulation of Snail transcription during epithelial to mesenchymal transition of tumor cells
.
Oncogene
2004
;
23
:
7345
54
.
53.
St. John
MA
,
Dohadwala
M
,
Luo
J
,
Wang
G
,
Lee
G
,
Shih
H
, et al
.
Proinflammatory mediators upregulate snail in head and neck cell squamous carcinoma
.
Clin Cancer Res
2009
;
15
:
6018
27
.
54.
Wu
Y
,
Deng
J
,
Rychahou
PG
,
Qiu
S
,
Evers
BM
,
Zhou
BP
.
Stabilization of snail by NF-kappaB is required for inflammation-induced cell migration and invasion
.
Cancer Cell
2009
;
15
:
416
28
.
55.
Zhang
K
,
Zhaos
J
,
Liu
X
,
Yan
B
,
Chen
D
,
Gao
Y
, et al
.
Activation of NF-B upregulates Snail and consequent repression of E-cadherin in cholangiocarcinoma cell invasion
.
Hepatogastroenterology
2011
;
58
:
1
7
.
56.
Li
H-M
,
Bi
Y-R
,
Li
Y
,
Fu
R
,
Lv
W-C
,
Jiang
N
, et al
.
A potent CBP/p300-Snail interaction inhibitor suppresses tumor growth and metastasis in wild-type p53-expressing cancer
.
Sci Adv
2020
;
6
:
eaaw8500
.
57.
Lambies
G
,
Miceli
M
,
Martínez-Guillamon
C
,
Olivera-Salguero
R
,
Peña
R
,
Frías
C-P
, et al
.
TGFβ-activated USP27X deubiquitinase regulates cell migration and chemoresistance via stabilization of Snail1
.
Cancer Res
2019
;
79
:
33
46
.
58.
Lee
SY
,
Ju
MK
,
Jeon
HM
,
Jeong
EK
,
Lee
YJ
,
Kim
CH
, et al
.
Regulation of tumor progression by programmed necrosis
.
Oxid Med Cell Longev
2018
;
2018
:
3537471
.
59.
Ota
I
,
Masui
T
,
Kurihara
M
,
Yook
J-I
,
Mikami
S
,
Kimura
T
, et al
.
Snail-induced EMT promotes cancer stem cell-like properties in head and neck cancer cells
.
Oncol Rep
2016
;
35
:
261
6
.
60.
Mendelsohn
AH
,
Lai
CK
,
Shintaku
IP
,
Fishbein
MC
,
Brugman
K
,
Elashoff
DA
, et al
.
Snail as a novel marker for regional metastasis in head and neck squamous cell carcinoma
.
Am J Otolaryngol
2012
;
33
:
6
13
.
61.
Wang
W
,
Ding
J
,
Allen
E
,
Zhu
P
,
Zhang
L
,
Vogel
H
, et al
.
Gigaxonin interacts with tubulin folding cofactor B and controls its degradation through the ubiquitin-proteasome pathway
.
Curr Biol
2005
;
15
:
2050
5
.
This open access article is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Supplementary data