Loss of smarcad1a accelerates tumorigenesis of malignant peripheral nerve sheath tumors in zebrafish

Abstract

Malignant peripheral nerve sheath tumors (MPNSTs) are a type of sarcoma that generally originates from Schwann cells. The prognosis for this type of malignancy is relatively poor due to complicated genetic alterations and the lack of specific targeted therapy. Chromosome fragment 4q22‐23 is frequently deleted in MPNSTs and other human tumors, suggesting tumor suppressor genes may reside in this region. Here, we provide evidence that SMARCAD1, a known chromatin remodeler, is a novel tumor suppressor gene located in 4q22‐23. We identified two human homologous smarcad1 genes (smarcad1a and smarcad1b) in zebrafish, and both genes share overlapping expression patterns during embryonic development. We demonstrated that two smarcad1a loss‐of‐function mutants, sa1299 and p403, can accelerate MPNST tumorigenesis in the tp53 mutant background, suggesting smarcad1a is a bona fide tumor suppressor gene for MPNSTs. Moreover, we found that DNA double‐strand break (DSB) repair might be compromised in both mutants compared to wildtype zebrafish, as indicated by pH2AX, a DNA DSB marker. In addition, both SMARCAD1 gene knockdown and overexpression in human cells were able to inhibit tumor growth and displayed similar DSB repair responses, suggesting proper SMARCAD1 gene expression level or gene dosage is critical for cell growth. Given that mutations of SMARCAD1 sensitize cells to poly ADP ribose polymerase inhibitors in yeast and the human U2OS osteosarcoma cell line, the identification of SMARCAD1 as a novel tumor suppressor gene might contribute to the development of new cancer therapies for MPNSTs.

1. INTRODUCTION

Cancer is essentially a genetic or genomic disease, as there are many genetic alterations in cancer cell genomes. ¹ Based on their functions, genes can be classified as either cancer driver genes or passenger genes. The cancer driver genes (i.e., oncogenes and tumor suppressor genes) are the genes whose mutations directly contribute to cancer initiation, development, and metastasis. In contrast, the genes whose mutations are not directly related to cancer are passenger genes. ² It is challenging to identify cancer driver genes even with high throughput microarray and genome sequencing technologies. ³ One of the most challenging types of genetic alterations is arm‐level copy number alterations (CNAs), which are synonymous with aneuploidy in terms of cytogenetics. ⁴ There are usually hundreds of genes on such an arm‐level CNA, and the gene dosage of each of these genes is affected by the large CNA. Unless the cancer driver genes also exhibit point mutations, there is almost no way to sort the driver genes apart from the passenger genes. Cross‐species comparative oncogenomics serves as one powerful solution for identifying cancer drivers on large CNAs due to genes conservative functions. ⁵ , ⁶ , ⁷ , ⁸

Malignant peripheral nerve sheath tumors (MPNSTs) are a rare type of sarcoma (∼2% of all sarcomas) originating from the neural crest or Schwann cell lineage with an incidence rate of five per million each year. ⁹ The prognosis for MPNSTs is generally poor due to their complex genetic changes within tumors, invasive growth nature, and insensitivity to chemo‐ and radiotherapies. ¹⁰ , ¹¹ , ¹² Targeted cancer therapy is one of the mainstays of the current cancer treatment regime because of its high specificity and reduced side effects. Unfortunately, there is no targeted therapy currently available for this type of malignancy. This is mainly caused by the lack of knowledge about the cancer driver genes of MPNSTs. Chromosome fragment 4q22‐23 is frequently deleted in MPNSTs and many other types of human cancers. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ However, the critical cancer driver genes on this CNA remain largely unknown.

Zebrafish have become a popular organism for modeling human cancer due to their large number of offspring, tractable genetics, and amenability to in vivo imaging as well as chemical screening. ²² Numerous zebrafish genetic models confirm that, between humans and zebrafish, there is functional conservation of known core cancer genes, such as tp53, pten, Myc, mutant KRAS, and mutant BRAF. ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Since human and zebrafish genomes share small syntenies, due to extensive reshuffling of genes' locations during evolution, we have demonstrated that zebrafish–human comparative cancer genomics is an effective approach for pinpointing cancer driver candidates on large CNAs in MPNSTs. ⁸ Utilizing the zebrafish–human comparative cancer genomics approach, we narrowed a limited number of tumor suppressor candidate genes in chromosome fragment 4q22‐23. ⁸ One of them is SMARCAD1 (SWI/SNF‐related, matrix‐associated actin‐dependent regulator of chromatin, subfamily a, containing DEAD/H box 1). SWI/SNF and other chromatin remodelers are frequently reported as cancer driver genes in many types of human cancers. ²⁸ , ²⁹ , ³⁰ A recent study revealed that one of the SWI/SNF family members, SMARCA2, is commonly mutated in human MPNSTs. ³¹ Thus, we decided to focus on the SMARCAD1 gene in the current study.

SMARCAD1 is a chromatin remodeler that belongs to the SNF2 helicase subfamily, which possesses a conserved ATP‐helicase domain. ³² Human SMARCAD1 was also reported to cooperate with the E1A oncogene to increase gene reactivation events by genomic rearrangement, suggesting its roles in maintaining genomic stability. ²¹ The homologous yeast gene, Fun30/Fft3, was found to play an essential role in the end resection of DNA double‐strand break (DSB) repair and heterochromatin maintenance. ³³ , ³⁴ , ³⁵ In human cells, SMARCAD1 is preferentially involved in homologous recombination after DNA DSB (double strand break) repair during S‐phase, and this process depends on ATM (ataxia‐telangiectasia). ³⁶ In addition, defects in FUN30 were reported to be sensitive to DNA topoisomerase 1 (TOP1) and poly ADP ribose polymerase (PARP) inhibitors in yeast and the U2OS cancer cell line. ³³ , ³⁷ Moreover, SMARCAD1 was recently found involved in mismatch repair, ³⁸ and endogenous retroviral silencing, ³⁹ both critical for genomic stability. A more recent study also backed up its critical roles in maintaining genomic integrity by stabilizing the replication forks. ⁴⁰ In humans, SMARCAD1 gene mutations were found associated with skin cancer susceptibility, adermatoglyphia, and Basan syndrome. ⁴¹ , ⁴² , ⁴³ However, the in vivo tumorigenic roles of the SMARCAD1 gene in MPNSTs remain unexplored.

Here, we provide evidence for the first time that zebrafish smarcad1a functions as a tumor suppressor gene in vivo. Moreover, we demonstrate that the human SMARCAD1 gene dosage is essential for MPNST cells to maintain genomic stability in response to DNA damage. Since it is known that the SMARCAD1 mutation causes yeast and U2OS osteosarcoma cell lines to become sensitive to PARP inhibitors, identifying SMARCAD1 as a tumor suppressor gene may provide us a new potential therapeutic target for some MPNST patients with currently available PARP inhibitors.

2. MATERIALS AND METHODS

2.1. Zebrafish lines, husbandry, and tumor onset analysis

Zebrafish were raised and maintained at the Purdue animal housing facility, which is approved by AAALAC. All experiments were carried out according to the protocols approved by the Purdue Animal Care and Use Committee (PACUC). The sa1299 fish were purchased from ZIRC (zebrafish international resource center). All the fish husbandry was carried out according to the zebrafish book. ⁴⁴ The tumor‐prone zebrafish line carrying the tp53 ^M214K/M214K point mutation has been described previously. ⁸ , ²⁵ After crossing, PCR genotyping was carried out at 6–8 weeks of age using previously published genotyping methods for tp53. ⁸ , ²⁵ Siblings of different genotypes were housed in adjacent tanks at similar densities to minimize environmental differences. During tumorigenesis monitoring, fish were euthanized at first observation of tumors or other signs of illness. The tumor types of the euthanized fish were confirmed by examining the hematoxylin and eosin (HE) stained tumor sections by a board‐certified veterinary pathologist blinded to the experimental groups.

2.2. Sa1299 mutant fish genotyping and mRNA splicing analysis

For sa1299 fish, gDNA was prepared from caudal fin clips with the hotshot protocol ⁴⁵ with the following modifications: boiling in 100 μl NaOH (50 mM) at 95°C for 1 h, then neutralized with 10 μl 1 M Tris‐HCl pH 8.0 per sample. PCRs were performed using primers designed to distinguish between the mutant and wild‐type sequences. For wildtype alleles, primers sa1299‐F and sa1299‐WR were used, and the sa1299‐F and sa1299‐WR were used to detect mutant alleles. For both PCRs, mlEXON5.WT.FW and UMLEXON5.RV were included as internal controls. Both PCRs resulted in a 391 bp product for the smarcad1a allele and a 239 bp product for the internal control primers. Conditions for both PCRs were 95°C 2 min, (95°C 30 s, 65.5°C 30 s, and 72°C 60 s) × 35 cycles, 72°C 10 min, and 10°C 2 min. PCR products were analyzed on 2% agarose gels with ethidium bromide staining and imaged using GelDoc‐IT2 imaging system (UVP).

To analyze mRNA splicing, total RNAs were isolated from 1dpf homozygous and wildtype fish embryos using Trizol reagent. The RNAs were treated with DNAase I to remove possible gDNA contamination. The reverse transcriptions were performed with SuperScript III Reverse Transcriptase (Thermo Scientific) and Oligo(dT)₁₈ following manufactory protocol. Full length and partial cDNA (exon8‐exon10) were amplified with GXL Taq (Takara) using the cDNA as a template and primers listed in Table S1. The PCR products were then cloned into the pJET1.2 vector using the blunt ligation kit. Sanger sequencing was performed using the primer smarcad1a‐exon8‐F.

2.3. Smarcad1a CRISPR mutant generation

CRISPR gRNAs, smarcad1a‐CR383 (5′ GCGGATGGTCAGTTTCCATC 3′) and smarcad1a‐CR490 (5′ GCAGCATATAAACAAGGACA 3′), were designed against the exon3 and exon4 of ENSDART00000091409, respectively (Figure 6). Oligonucleotides were synthesized by IDT (Integrated DNA Technologies Inc.) and were cloned into DR274 vector following the published method. ⁴⁶ Briefly, both sense and antisense oligonucleotides were annealed to dsDNA, double‐stranded DNA (95°C 10 min, and then ramp down to 12°C at 5°C/min). DR274 plasmid was digested with BsaI (NEB) to generate compatible ends with the annealed oligonucleotides, and the dsRNAs were inserted into the DR274 with T4 DNA ligase (NEB). CRISPR clones were finally verified by Sanger Sequencing. To generate the gRNAs for injections, the DR274 vector was first linearized with DraI (NEB) and used as the template for the transcription reaction with T7 RNA polymerase (NEB). All the gRNAs were purified by RNA Clean & Concentrator kit (Zymo, #R1017). All injections were performed at the one‐cell stage of zebrafish embryo development. ⁴⁷ For each embryo 2 nl of the following mixture was injected; 12.5 ng/μl gRNA, 100 ng/μl Cas9 protein (NEB), and 0.4% phenol red (Sigma, P0290). The injected adult F₀ fish were outcrossed with wild‐type fish, and around 20 fish embryos from each pair were sacrificed for gDNA isolation. PCR assays were performed to screen out positive F₀ fish. Briefly, pooled fish embryos were put into 100 μl of DNA lysis buffer (50 mM NaOH) and heated at 95°C for 60 min, then allowed to cool to room temperate and 10 μl Tris‐HCl pH 7.9 was added to neutralized the lysis buffer. ⁴⁵ Then, PCRs were performed using the gDNA and smarcad1a‐F3 and smarcad1a‐R3 primers (Table S1). The wildtype allele yields a 539 bp PCR product, while a 296 bp PCR product is generated in the smarcad1a ^p403 mutant. Once a positive F₀ was identified, the rest of fish embryos from the same cross were raised to adulthood as F₁ founder fish. To clean out possible off‐target mutants, we continually out‐crossed the F1 founders with wildtype fish twice for F₂ and F₃. Then, F₃ were in‐crossed for generating homozygous F₄ smarcad1a mutants. All the experiments were performed on fish embryos from the F₄ fish cross.

2.4. Quantitative RT‐PCR

Total RNAs were isolated from cells using TRIzol reagent according to the manufacturer's instruction. For reverse transcription, 2 μg total RNA was used as a template, and cDNAs were synthesized using the Transcriptor First Strand cDNA Synthesis Kit (Roche). QRT‐PCRs were conducted using SYBR Green I Master Mix (Roche), following the manufacturer's instruction on Light Cycler 480. Primers (Table S1) for each gene were designed to cover all the transcripts, which are located on different exons to avoid potential genomic DNA contamination. PCRs were performed at the following condition: 95°C, 10 s; 60°C, 15 s; and 72°C, 20 s for 40 cycles. Results were analyzed using ΔCt method to calculate the relative gene mRNA level. ⁴⁸

2.5. SMARCAD1 phylogeny, synteny, and gene structure analyses

SMARCAD1 protein sequences were identified by a BLASTp search using the human SMARCAD1 sequence as a query in Ensembl and NCBI. The longest sequence was preferentially chosen when there were multiple sequences. Multiple protein sequences (Table S2) were aligned using the MUSCLE alignment program. ⁴⁹ The evolutionary model for phylogenetic analysis was identified using the best model test using maximum likelihood and default parameters in MEGA6. ⁵⁰ A minimum evolution phylogeny was generated in MEGA6 with complete deletion option: bootstrap = 10 000; gamma = 0.61. A maximum‐likelihood phylogenetic tree was constructed using JTT + G with 1000 bootstrap replicates with PhyML 3.1. ⁵¹ For Bayesian analysis (BP) phylogenetic analysis, 20 million generations were run using the following parameters in MrBayes 3.2.6: nruns = 2, nchains = 4, aamodel = fixed (Jones), rates = gamma ngammacat = 8, samplefreq = 500, burninfrac = 0.25. The final phylogenetic trees were viewed and generated with FigTree V1.4.2 (http://tree.bio.ed.ac.uk/software/figtree). Gene intron‐exon structures were analyzed using the longest transcripts in Ensembl.

Synteny analyses were performed with the teleost synteny database ⁵² and verified with Ensembl and UCSC human and zebrafish genome databases. The SMARCAD1 exon‐intron structure analysis was performed in the UCSC genome browser, and the Gene structure display server was used for visualization using the information of the smarcad1a gene downloaded from the table genome annotation in the UCSC zebrafish genome (GRCz11). SMARCAD1a protein domains were predicted by online SMART analysis using the default settings. ⁵³

2.6. Gene cloning, whole‐mount in situ hybridization, and imaging

Full‐length coding regions of zebrafish smarcad1 genes were amplified by RT‐PCR using gene‐specific primers designed according to the current DNA sequence in Ensembl. Phusion® High‐Fidelity DNA Polymerase master mix (New England Biolabs) was used for PCR amplification. PCR primers used here are listed in Table S1. The PCR products were purified using Zymo Gel Extraction Kit (Zymo Research) before they were cloned into the pJet1.2 vector using the CloneJET PCR Cloning Kit (Thermo Scientific). Gene inserts orientation was verified by Sanger sequencing. Riboprobes were synthesized through in vitro transcription using T7 DNA polymerase (New England Biolabs) and DIG RNA Labeling Mix (Roche). Then, the riboprobes were purified by SigmaSpin™ post‐reaction clean‐up columns (Sigma, S5059). Whole‐mount in situ hybridization was carried out according to our previously published method. ⁵⁴ , ⁵⁵ For histological analysis, post‐hybridization embryos were equilibrated in 15% sucrose, then 30% sucrose in 20% gelatin, after which they were embedded in 20% gelatin for cryosectioning (6–12 μm). Images were acquired using AxioCam MRc camera on Zeiss SteREO Discovery.V12 and Axio Imager 2 compound microscope.

2.7. Cell culture, stable cell lines, and cell growth assays

All experimental protocols using cell lines and plasmid constructs were approved by the Purdue University institutional biosafety review board. The human MPNST cell lines were authenticated by ATCC using short tandem repeat profiling. Cells were cultured in DMEM with 10% heat‐inactivated fetal bovine serum, penicillin (100 IU/ml), and streptomycin (100 μg/ml). All cell cultures were carried out at 37°C in a humidified 5% CO₂ atmosphere. A full‐length sequence of the SMARCAD1 gene was amplified from HEK293T cells and was confirmed by Sanger sequencing. The SMARCAD1 gene was subsequently cloned into pLIX408, a T2A bicistronic lentiviral vector modified from pLIX402 (Addgene #41394). For SMARCAD1 gene knockdown, shRNAs against human SMARCAD1 gene and non‐targeting shRNA control were purchased from TransOMIC technologies Inc. (ULTRA‐3351709, ULTRA‐3351712, ULTRA‐3351713). Stable cell establishment, western blots, MTT, and plate agar assays were conducted as described previously. ⁵⁶ For Western blot, a commercial anti‐SMARCAD1 antibody (Bethyl, A301‐593A 1:2000 dilution) was chosen. Statistical analyses were performed using GraphPad Prism 6.0h. Data were analyzed using the unpaired student t test. p < 0.05 was considered to have a statistically significant difference.

2.8. Immunohistochemistry

Zebrafish with tumors were fixed with 10% neutral buffered formalin (VWR). Dehydration, paraffin section, and H&E staining were performed in the Histology Research Laboratory, College of Veterinary Medicine, Purdue University. Slides with adjacent sections were dewaxed with xylene (2 × 10 min) to remove paraffin and followed by rehydration with a series of diluted ethanol (100%, 95%, 80%, 50%, 5 min each) and water rinse. Next, heat‐mediated antigen retrieval was performed to unmask the antigenic sites: all the sections were incubated at 95°C for 10 min in Tris/EDTA buffer (pH 9.0). Then, 3% hydrogen peroxide was used to block endogenous peroxidase in tissue for 10 min, followed by 5 min tap‐water rinsing and 2.5% goat serum blocking for 2 h. The primary antibody (anti‐S100: Dako #IS504; anti‐H3K27me3: Millipore #07‐499, 1:500 dilution in 2.5% goat serum) was incubated with sections in a 4°C humidity chamber overnight. On the second day, sections were washed with Tris‐buffered saline (TBS) 3 times. Then the primary antibody was visualized with the avidin/biotin‐based peroxidase system (VECTASTAIN Elite ABC‐HRP kit, #PK‐6101; DAB Substrate Kit, #SK‐4100, Vector Laboratories, CA, USA) by following the manufacturer's instructions. Briefly, sections were incubated with biotinylated goat anti‐rabbit antibody for 30 min, followed by three times TBS wash. Then all the sections were incubated with ABC reagent for 30 min. After three times wash with TBS, 3,3′‐diaminobenzidine (DAB) substrate was used to label primary antibody binding tissues. Next, sections were counterstained with hematoxylin and dehydrated with gradient dilutions of ethanol (80%, 95%, 100%, and 100%). Sections were then cleared by xylene and mounted with Cytoseal 60 mounting medium (Epredia™ 83 104). Images of stained sections were taken with Zeiss Axio imager A2.

2.9. DNA damage repair response

Human cells with stable SMARCAD1 overexpression and knockdown (both doxycycline induced and un‐induced) were treated with X‐ray irradiation treatment. They were then harvested and lysed using RIPA buffer with 0.5 mM PMSF (phenylmethylsulfonyl fluoride). The indicated doses of radiation were administered using the X‐RAD 320 biological irradiator device (PXi Precision X‐Ray, North Branford, CT, USA). Radiation was produced via an X‐ray tube radiation source with a dose rate of ~1 Gy/25 s. Samples were irradiated in Petri dishes at room temperature (~25°C), and non‐irradiated samples were mock irradiated.

Zebrafish embryos were collected and raised in 10 cm diameter Petri dishes with system water (0.6 g/L aquarium salt in RO water) and 0.01 mg/L methylene blue until they reached 1 day old. Chorions were removed with pronase (1 mg/mL) in the fish system water and rinsed at least 3 times with calcium‐free Ringer's solution (116 mM NaCl; 2.9 mM KCl; 5 mM HEPES, pH 7.2). Then, the fish embryos were transferred into 1.5 ml Eppendorf tubes to remove the yolks from fish embryos, according to the zebrafish book. ⁴⁴ Briefly, after adding 1 ml deyolking buffer (calcium‐free Ringer's solution with 0.3 mM PMSF and 10 mM EDTA), yolks were separated by triturating the fish embryos with 1000 ml pipette tips. Then, the Eppendorf tubes were centrifuged at 200 rpm for 30 s, and the supernatant with yolk components was removed. Lastly, fish embryos were rinsed three times with ice‐cold calcium‐free Ringer's solution before adding RIPA Lysis Buffer with a complete EDTA‐free protease inhibitor cocktail (5892791001) and PhosSTOP phosphatase inhibitor (4906837001). Total protein concentration was measured using the Bradford method. Approximately 30 μg total protein was run on a 15% SDS‐PAGE. Anti‐phospho‐histone H2A.X (Ser139) (Millipore #05‐636, 1:2000 dilution) was used for detecting broken double‐strand DNAs using western blot. Beta‐actin (ACTB, Santa Cruz, sc‐47 778, 1:2000 dilution) was used to ensure equal protein loading. Band/protein intensity was quantified using Image Lab™ (Bio‐rad, Hercules, CA, USA).

2.10. Data availability

Reagents are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

3. RESULTS

3.1. SMARCAD1 is a tumor suppressor candidate on the chromosome fragment 4q22‐23

Deletions of 4q22‐23 are frequent in MPNSTs and many other types of human cancers, suggesting tumor suppressor genes may be located in this region. SMARCAD1 is one of the potential tumor suppressor gene candidates (Figure 1A–C). Given the importance of chromatin remodelers in cancer biology, especially the recent report on the frequent loss of SMARCA2 in MPNSTs, ³¹ we further examined SMARCAD1 gene mutations in TCGA data through cBioportal. ⁵⁷ We found it is frequently mutated in various human cancers (uterine, skin, colorectal, sarcoma, etc.), although human MPNST data are not currently available (Figure 1D,E). We reasoned that SMRACAD1 expression might be lower if it is a tumor suppressor gene in human MPNSTs. To test this, we examined six human MPNST cell lines and found SMARCAD1 protein levels are indeed relatively lower (three significantly, and three mildly) compared to a schwannoma cell line, HEI‐193 (Figure 1F). Altogether, these results suggest the SMARCAD1 could be a tumor suppressor gene.

FIGURE 1.

Frequent mutations of SMARCAD1 in humans and zebrafish suggest it is a tumor suppressor. (A) Heat map of human chromosome 4. SAMRCAD1 is located on this chromosome. The locus of SMARCAD1 (green line) was found to be underrepresented in ~26% (16 out of 61) human MPNSTs. ⁸ (B) Zebrafish smarcad1a is located on chromosome 8, which is underrepresented in about 70% (103 out of 147) of zebrafish MPNSTs. (C) Zebrafish smarcad1b is located on chromosome 10, which is overrepresented in about 60% (88 out of 147) of zebrafish MPNSTs. Samples are sorted top‐to‐bottom by decreasing deletion amplitude at the respective SAMRCAD1/smarcad1 locus, indicated by a green line. Blue and red bars on the right side of each panel indicate samples with SAMRCAD1/smarcad1 losses (blue) or gains (red). Color densities are corresponding to the degree of loss and gain as previously described. ⁸ (D) SMARCAD1 mutation frequency of human cancers in TCGA database. (E) SMARCAD1 mutation types and positions found in human cancers in TCGA database. (F) SMARCAD1 expression level of 6 human MPNST cell lines (all NF1 mutant except STS26T) and HEI‐193, immortalized human schwannoma

3.2. Zebrafish have two smarcad1 orthologues

To examine the tumor suppressor function of the smarcad1, we decided to take advantage of the zebrafish MPNST model. In zebrafish, there are two smarcad1 genes on different chromosomes: smarcad1a and smarcad1b. To reveal the genetic relationships between zebrafish smarcad1 and other vertebrate SMARCAD1 genes, we first performed multiple phylogenetic analyses by Bayesian analysis [BP], maximum likelihood [ML], and minimum evolution [ME] method, respectively). Our results consistently support that smarcad1a and smarca1b are orthologous to the human SMARCAD1 gene (Figure 2A; Figures S1 and S2). Since most teleosts have two smarcad1 genes, our results suggest that the two genes resulted from teleost‐specific whole‐genome duplication. ⁵⁸ , ⁵⁹ , ⁶⁰ To further validate whether the two genes are orthologous to the human SMARCAD1 gene, we performed syntenic analysis using the established zebrafish–human synteny database. ⁵² The smarcad1a gene is located on zebrafish chromosome 8, and it shares a conserved synteny with the human SMARCAD1 gene, which is located on human chromosome 4 (Figure 2B). Zebrafish smarcd1b is located on chromosome 10, and no neighbor genes of the conserved synteny are found around the gene locus (Figure 2B). However, the nearby regions of zebrafish chromosome 10 share conserved syntenies (LIN54‐PLAC8 and ANX3‐RASGEF1B) with human chromosome 4 (Figure 2B), suggesting that the smarcad1b‐linked neighbor genes were lost during evolution.

FIGURE 2.

Zebrafish have two smarcad1 genes. (A) Extended majority‐rule consensus tree for the Bayesian phylogenetic analysis of SMARCAD1 proteins. Numbers at each node indicate posterior probability (pp) values based on 20 million runs. Branch lengths are proportional to the means of the pp densities for their expected replacements per site. The ME and ML phylogenetic trees (Figures S1 and S2) were generally in agreement with the BP phylogeny: most of the metazoan species have one SMARCAD1 gene, while there are usually two smarcad1 genes in teleost genomes. The two smarcad1s most likely resulted from teleost‐specific whole‐genome duplications, as each formed a distinct clade. The tree was rooted with yeast. (B) Syntenic relationship between human and zebrafish chromosomes. Zebrafish chromosome 8 (Dre8) that contains smarcad1a and three other genes (FAM190A, GRID2, and ATOH1, bottom row) are orthologous to, and in the same order as, genes in the portion of human chromosome 4 (Hsa4) that contains SMARCAD1 (middle row). A portion of Dre10 contains smarcad1b, but not the other three directly linked genes. However, this part of Dre10 is co‐orthologous to the portion of Hsa4 that contains SMARCAD1, as they share other syntenies (top row). Orthologous genes are indicated with colored lines

3.3. The zebrafish smarcad1 genes share overlapping and distinct expression patterns during early development

Paralogous genes, resulting from whole‐genome duplication, may have undergone neofunctionalization (new functions), or subfunctionalization (split functions of their parental gene) according to the duplication‐degeneration‐complementation (DDC) model. ⁶¹ To explore the functional relationship of two zebrafish smarcad1 genes, we examined gene expression patterns using whole‐mount in situ hybridization in the zebrafish's early developmental stages. We found that both genes are expressed at the 12S (12‐somite) stage in the brain, optical vesicle, neural tube, and tailbud (Figure 3A,E). The expression extended to the endoderm at the 20‐somite stage (Figure 3B,F). At 24 hpf (hours post‐fertilization), both genes are expressed mainly in the head region, neural tube, and intermediate cell mass (Figure 3C,D,G,H). The differential expression patterns of the two genes become evident at 48hpf (Figure 3I–P). The smarcad1a gene is mainly expressed in the eye (retina and ciliary marginal zone), brain, and pharyngeal arches (Figure 3I–L), while the smarcad1b gene is primarily expressed in the brain (Figure 3M–P). At 72hpf, smarcad1a expression in the brain region is decreased, but its expression in the ciliary marginal zone of the eye and pharyngeal arches remains (Figure 3Q,R). Cross‐sections revealed that smarcad1a is also expressed in the gut and dermomyotome around the pectoral fin (Figure 3S,T). In contrast to smarcad1a, smarcad1b is mainly expressed in the brain but also found in the retina, gut, and dermomyotome (Figure 3U–X). The overlapping and different spatiotemporal expression of the two zebrafish smarcad1 genes suggests that there was a sub‐functionalization after the teleost whole‐genome duplication. Moreover, these similar expression patterns indicate that the two genes may possess overlapping functions during zebrafish development and normal physiology.

FIGURE 3.

The two smarcad1 zebrafish genes are expressed in both overlapped and distinct regions during embryonic development. Whole‐mount in situ hybridization of zebrafish embryos at stages 12S (A, E), 20S (B, F), 24 hpf (C, D, G, H), 48 hpf (I–P), and 72 hpf (Q–X). Anterior is to the left in all whole‐mount images, and dorsal is to the top in all transverse sections. Gene names are labeled at the upright corner in the panels of whole‐mount images. (A–D, I–L, and Q–T): gene expression of smarcad1a. (E–H, M–P, and U–X): gene expression of smarcad1b. (A and C) Lateral view of the expression of smarcad1a at 12S, 20S, and 24dpf, respectively. (D) Transverse section through the trunk region of the embryos in panel C. smarcad1a is mainly expressed in the neural tissues, endoderm, and intermediate cell mass at these stages. (E and G) Lateral view of the expression of smarcad1b at 12S, 20S, and 24dpf, respectively. (H) Transverse section through the trunk region of the embryos in panel G. The expression patterns of smarcad1b are very similar to smarcad1a at these early stages. (I, J) Lateral and dorsal view of the expression of smarcad1a at 48 hpf. (K and L) Transverse section through the head and pectoral fin regions. (M and N) Lateral and dorsal view of the expression of smarcad1b at 48 hpf. (O and P) Transverse section through the head and pectoral fin regions. (Q and R) Lateral and dorsal view of the expression of smarcad1a at 72 hpf. (S and T) Transverse section through the head and pectoral fin regions. (U and V) Lateral and dorsal view of the expression of smarcad1b at 72 hpf. (W and X) Transverse section through the head and pectoral fin regions. The dashed lines indicate the positions of sections. The letters below the dashed lines correspond to the panels. de, diencephalon; dm, dermomyotome; end, endoderm; fb, forebrain; hb, hindbrain; icm, intermediate cell mass; ir, iris; le, lens; mand, mandibular cartilage; mb, middle brain; mhb, midbrain‐hindbrain boundary; mz, ciliary marginal zone; n, notochord; nt, neural tube; op, optical cup; pf, pectoral fin; pa, pharyngeal; quad, quadrate cartilage; ret, retina; tb, tail bud

3.4. Sa1299 is a smarcad1a loss‐of‐function mutant

To investigate the tumor suppressor roles of the smarcad1a gene, we acquired a currently available splicing site mutant, sa1299, which was generated with ENU (N‐ethyl‐N‐nitrosourea) by the Zebrafish Mutation Project (ZMP). ⁶² We first confirmed that there is a T to G mutation at the essential spicing site of intron 9 by PCR and Sanger sequencing (Figure 4A). Based on bioinformatics analysis, this mutation only changes one amino acid in the short transcript (ENSDART00000139029) but may lead to open reading frameshifting or premature truncation for the longer transcript (ENSDART00000091409). Since the essential splice site mutation may lead to intron retaining, exon skipping, or cryptic splicing, we sequenced the region between exon 7 and exon 10 of the mature mRNA from sa1299 homozygous 1dpf (day post‐fertilization) zebrafish embryos. When compared to the wildtype smarcad1a mRNA, we found that the mutant mRNA is spliced through a proximal cryptic splicing site 11 bps downstream of the mutant site, and this led to the insertion of 13 bps (GGAAGGATCTGCT) in intron 9 between exon 9 and exon 10 (Figure 4B and Figure S3). This insertion resulted in a premature stop codon and a truncated protein of about 379 amino acids (Figure 4C). In zebrafish, mutant genes with point mutations are usually found down‐regulated through nonsense‐mediated mRNA decay. ⁶³ , ⁶⁴ We then tested this possibility by examining the smarcad1a gene expression levels in 1dpf sa1299 homozygous fish embryos with quantitative RT‐PCR (Figure 4D). Indeed, the overall expression was reduced by more than 50% compared to wild‐type zebrafish embryos. Furthermore, the decrease of smarcad1a gene expression levels was confirmed by whole‐mount in situ hybridization in 1–3 dpf fish embryos (Figure 4E). Thus, sa1299 is a smarcad1a loss‐of‐function mutant. The smarcad1a gene is dispensable since both heterozygous and homozygous sa1299 fish can survive to adulthood without any evident sickness or morphological abnormalities. This might be due to the functional overlap with the smarcad1b gene since both share close expression domains during zebrafish early development. To test this, we examined the smarcad1b gene expression in sa1299 homozygotes. We found this gene is expressed in similar domains with smarcad1a (Figure 4F), supporting our hypothesis.

FIGURE 4.

Sa1299 is an essential splicing mutant of the smarcad1a gene in zebrafish. (A) Schematic illustration of smarcad1a transcription and T to G single nucleotide mutation in the 5′ end essential spicing site of intron 9. The transcription information is based on the GRCz10 in Ensembl. (B) Normal and aberrant splicing are illustrated in the top row. Proximal cryptic splicing resulted in 13 bps from intron 9 retained in the mRNA. T > G mutation is highlighted blue, and cryptic GT is highlighted with red letters. This result was confirmed by Sanger sequencing (bottom row). (C) Predicted functional domains from full‐length protein and truncated proteins (without the functional domains) resulted from premature stop codon (asterisk) due to the proximal cryptic splicing. Amino acid numbers are indicated in the diagram. (D) The mRNA level of ENSDART00000091409 is decreased by RT‐PCR in 1dpf sa1299 homozygous embryos, most likely caused by nonsense‐mediated mRNA decay. (E) Decrease of smarcad1a ENSDART00000091409 mRNA also is detected by whole‐mount in situ hybridization in 1–3 dpf zebrafish embryos. (F) The smarcad1b gene is not affected and still expressed in similar expression domains with smarcad1a in sa1299 homozygotes. The sense probe controls are included in the first two vertical panels

3.5. Loss of smarcad1a in zebrafish accelerates tumorigenesis of MPNSTs

Zebrafish smarcad1a is located on the underrepresented chromosome 8 in our zebrafish MPNST CNA analysis (Figure 1A). Thus, we hypothesized that smarcad1a is a tumor suppressor gene in this zebrafish. Zebrafish MPNSTs can be initiated by either tp53 or ribosomal protein (rp) gene mutations. ²⁵ , ⁶⁵ It has been demonstrated that zebrafish MPSNTs mimic their human counterpart on multiple levels, from histology to transcriptomes. ⁸ , ⁶⁶ We reasoned that loss of smarcad1a might cooperate with initiating tp53 mutations to promote MPNST development in zebrafish through a synthetic genetic effect if smarcad1a is a novel tumor suppressor gene. To test this hypothesis, we crossed the sa1299 mutant with the tp53 ^M214K zebrafish line. As we expected, double heterozygotes (smarcad1a ^sa1299/wt; tp53 ^M214K/wt) developed MPNST tumors significantly faster than the sibling single heterozygotes (tp53 ^M214K/wt) (Figure 5A). However, we did not observe tumors from the smarca1a ^sa1299/wt heterozygous fish in the same time window, implying that the tumor suppressor function of smarcad1a is relatively weak compared to the tp53 gene. We also did not detect tumor spectrum shift like the zebrafish reck gene, ⁶⁷ and all the tumors are MPNSTs. To confirm the identity of MPNSTs, we examined human MPNST markers S100 and H3K27me3 in 4 tumors of tp53 ^M214K/wt and smarcad1a ^sa1299/wt; tp53 ^M214K/wt, respectively. All these tumors are stained with S100, but not H3K27me (Figure 6). These results are consistent with previous reports that zebrafish MPNSTs have high S100 expression ⁶⁸ and human MPNSTs loss of H3K27me3 expression. ⁶⁹

FIGURE 5.

Smarcad1a was able to accelerate tumorigenesis in a zebrafish MPNST model initiated by loss‐of‐function tp53. (A) Kaplan–Meier survival curve showing tumor‐free survival of cohorts of single (black line) and double heterozygotes (red line) derived from smarcad1a ^sa1299/WT; tp53 ^M124K/WT and tp53 ^M124K/WT. Fish were genotyped by PCR for each relevant mutation at 6–8 weeks of age and housed segregated by genotype. The numbers of fish of each genotype and the p values between the tp53 single heterozygote and the double mutants are shown in the figure. (B) Kaplan–Meier curve showing tumor‐free survival of cohorts of single (black line) and double heterozygotes (red line) derived from smarcad1a ^p403/p403; tp53 ^M124K/M124K and tp53 ^M124K/M124K. (C) Typical histology of zebrafish MPNST by hematoxylin–eosin staining: spindle cells and swirling structure (white triangle). (D) Heterogonous regions without typical spindle cells from the same tumor. Black arrows indicate necrosis

FIGURE 6.

Confirmation of zebrafish MPNSTs with S100 and H3K27me3. Four random tumors from tp53 heterozygotes and double heterozygotes (tp53 and sa1299) were stained with S100 and H3K27me3 antibodies. (A–C, G–I, M–O, S–U): tp53 heterozygotes. (D–F, J–L, P–R, V–X): double heterozygotes. HE (hematoxylin and eosin) staining: A, D, G, J, M, P, S, V. IHC with S100 antibody: B, E, H, K, N, Q, T, W. IHC with H3K27me antibody staining: C, F, I, L, O, R, U, X. Some stained cells are non‐Schwann cells in these sections, but they serve as a positive control for the antibody specificity. All the images were taken at the same magnification. Scale bar = 100 μm in panel A

To further confirm the tumor suppressor roles of the smarcad1a gene, we created another smarcad1a loss‐of‐function mutant, smarcad1a ^p403, by using CRISPR technology. We targeted exons 3 and 4 to disrupt all the 3 known zebrafish mRNA transcripts using two CRISPR gRNAs (Figure 7A). One of the knockout mutant lines, p403, deleted 258 bps and gained a 15 bp insertion. Thus, this mutation yielded a truncated protein (about 55–58 amino acids) by introducing a premature stop codon (Figure 7B,C). Like the sa1299 mutant, the homozygous p403 fish could grow to adulthood without any apparent morphological defects. Furthermore, this smarcad1a ^p403 mutant also showed accelerated tumorigenesis in the tp53 null background (Figure 5B). Histologically, the tumors from the p403 mutant are similar to those from the sa1299 mutant; both possess the typical spindle cell characteristics and tumor cell morphological heterogeneity (Figure 5C,D). In addition, we did not notice any morphological difference between tumors from both tp53 heterozygotes and homozygotes, consistent with our past studies. ⁸ , ⁶⁶

FIGURE 7.

Zebrafish smarcad1a mutant created by CRISPR. (A) Transcripts of the smarcad1a gene, according to Ensembl (GRCz11). (B) Diagram of consequences of the p403 mutant on the SMARCAD1A protein. The top is the intact protein, and the bottom is the truncated protein. Protein domains were predicted by SMART domain. (C) Sequencing result of the smarcad1a p403 allele compared with wildtype (WT). The PAMs were highlighted in cyan, and the positions of CRISPR gRNAs are underneath the PAM sequences. The yellow highlighted nucleotides are the insertion. There is a 329 bps deletion from position 114 to 382 in the diagram. The tildes (~) represent deleted sequences. Asterisk (*) and hashtag (#) indicate the gRNA CR383 and CR490, respectively

3.6. DNA damage repair is compromised in zebrafish smarcad1a mutants

The yeast SMARCAD1 homologous gene, Fun30, was reported to be required for end resection in the process of DNA DSB repair. ³³ , ³⁴ , ³⁵ Recently, SMARCAD1 was reported to be involved in DNA DSB repair by reducing homologous recombination in an ATM‐dependent manner and the mismatch repair (MMR) system, which is mediated by the Msh2 gene ³⁸ in human and Xenopus cells, respectively. It has been demonstrated that phosphorylated H2AX (pH2AX) foci duration is elongated in SMARCAD1 depleted cells. In addition, phosphorylated RPA, 53BP1, BRCA1, and RAD51 foci were decreased in SMARCAD1 depleted cells. ³⁶ Moreover, DNA DSB repair deficiency was also confirmed in human primary fibroblasts and keratinocytes by phosphorylated pH2AX after DSB DNA damage was induced by irradiation. ⁴³ As DNA damage repair is an essential pathway for tumorigenesis, we hypothesize that the zebrafish smarcad1a mutant, sa1299, might have DNA damage repair defects. Since pH2AX Ser139 was used as a rough readout of DNA damage repair efficiency, ³³ , ³⁶ , ⁴³ we chose to examine pH2AX Ser139 in our experiments. If the DNA damage repair machinery is compromised, the presence/duration of pH2AX will be elongated. To test this hypothesis, we treated 1dpf wildtype and sa1299 mutant fish embryos with X‐ray irradiation and examined the expression of pH2AX level after irradiation. Indeed, the level of pH2AX decreased slower in both sa1299 and p403, compared to wildtype during the process of recovery after irradiation (Figure 8A,B). This suggests that the DNA damage repairing machinery is compromised in both zebrafish smarcad1a mutants.

FIGURE 8.

DNA damage response is compromised in zebrafish embryos. (A) Representative western blot showing protein expression or phosphorylation of H2AX ser139 upon the indicated treatment in wildtype, smarcad1a ^{sa1299−/−}, and smarcad1a ^p403−/− 1dpf zebrafish embryos. For irradiated samples (IR+), samples were harvested at the indicated hour post 15 Gy IR. (B) Quantification of protein expression via western blotting from panel A. For each biological replicate, values were normalized to the value for “IR−” to calculate the fold change in phosphorylation of H2AX upon the indicated treatment. Bars in (B) are the mean of three independent experiments. Statistical analysis comparing experimental to the control (“IR−”) was performed using Welch's t test (*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; NS p > 0.05)

3.7. The dosage of SMARCAD1 is critical for double‐strand DNA repair in human neurofibroma and MPNST cells

The SMARCAD1 gene was found to be involved in the double‐strand DNA repair in human U2OS and HeLa cell lines, ³³ , ³⁶ although this remains unknown in MPNST cells. Many DNA damage genes are found to have evolutionarily conserved cellular functions. Thus, based on our results of zebrafish smarcad1 DNA damage repair defect (Figure 8), we reasoned that human SMARCAD1 might have similar functions in DNA damage repair in human MPNST cell lines. To test our hypothesis, we first created a tetracycline‐inducible SMARCAD1 knockdown cell line using shRNA in a premalignant neurofibroma cell line, HEI‐193 (Figure S4A,B), since it has a relatively high expression (Figure 1F). Then, we examined the repair of DNA DSBs with these cell lines. We found the level of pH2AX Ser139 diminished slower in the SMARCAD1 knockdown cells after both 10 and 20 Gy irradiation treatment, comparing to the control cells (Figure 9A,B).

FIGURE 9.

Human cells show a defect in double‐strand DNA damage repair. (A, B) Double‐strand DNA damage repair was measured by Western blots with anti‐histone H2AX S139ph (phospho Ser139) in SMARCAD1 knockdown schwannoma cell line HEI‐193 after X‐ray irradiations (10 and 20 Gy, respectively). ULTRA‐3351712 knockdown cell was used for this experiment. Doxycycline (final concentration: 0.1 μg/ml) was added 48 h ahead of the experiments. (C, D) Double‐strand DNA damage repair in SMARCAD1 overexpression MPNST cells (STS26T) by H2AX S139ph. Doxycycline (final concentration: 1 μg/ml) was added 24 h ahead of the experiments. Both cell lines were harvested at 1, 5, and 10 h post‐irradiation along with untreated samples. un, untreated sample. Beta‐actin (ACTB) was used as a loading control. The densitometry ratios of H2A X S139ph over ACTB was listed underneath the blots

In our zebrafish CNA analysis, smarcad1a was underrepresented in more than half of the samples. However, the smarcad1b showed the opposite (Figure 1B,C). In addition, SMARCAD1 amplifications and deletions are also frequently present in some human cancers such as sarcoma and ovarian (Figure 1D). Considering smarcad1a and smarcad1b have potentially similar functions, we hypothesized that the gene dosage (increase or decrease of gene function) of SMARCAD1 is imperative for maintaining genomic stability in response to double‐strand DNA damage repair. To test this hypothesis in human cells, we created another tetracycline‐inducible SMARCAD1‐overexpressing human MPNST cell line, STS26T (Figure S4C). We choose this cell line because the SMARCAD1 expression is relatively lower than in schwannoma cell HEI‐193 (Figure 1F), and it may represent sporadic MPNSTs that are consistent with our zebrafish MPNST model. Indeed, the expression of pH2AX decreased slower after both 10 and 20 Gy X‐ray irradiation in doxycycline‐treated samples (Figure 9C,D). This result suggests that SMARCAD1 plays a role in DNA damage repair in a gene dosage‐sensitive manner. This dosage sensitivity could be caused by the stoichiometric effect with its partner proteins found in some protein‐complex. ⁷⁰ To further investigate the tumorigenic impacts of SMARCAD1, we performed MTT and soft agar colony formation assays in both SMARCAD1 knockdown and overexpression cell lines. We found that cell growth rate and anchorage‐independent growth decreased in both cases (Figure 10A–D). These results suggest that the proper gene dosage or amount of SMARCAD1 protein might be required for maintaining normal cell growth.

FIGURE 10.

Cell proliferation and independent growth were inhibited in both SMARCAD1 knockdown and overexpression. Proliferation was measured by MTT assay in SMARCAD1 knockdown HEI193 cells (A) and overexpression STS26T cells (B). Growth differences were evident from day 3 in both cases. Anchorage‐independent growth was examined by soft‐agar assay in SMARCAD1 knockdown HEI193 cells (C) and overexpression STS26T cells (D). All the assays were done with three biological replicates. Asterisk (*) indicates statistical significance, p < 0.05

4. DISCUSSION

As cancer is essentially a genetic or genomic disease, one of the goals of current cancer research is to identify cancer driver genes that can be used as targets for cancer therapy and/or markers for diagnosis and prognosis. It remains challenging to identify cancer driver genes on large CNAs or aneuploid chromosomes in the era of massively parallel sequencing technology, simply because they are usually altered as a unit that carries many genes. Unless the cancer driver genes also have point mutations, it is very challenging to tell the driver genes apart from the passenger genes along with the altered chromosome fragments. Here, we focused on one candidate gene, SMARCAD1, on human chromosome 4q22‐23, due to the prevalence of chromatin remodelers as cancer drivers. We demonstrated that this gene is a bone fide tumor suppressor gene in zebrafish MPNSTs. Moreover, our experiments in human MPNST cells suggest that SMARCAD1 is involved in the DNA DSB repairing response in zebrafish embryos and human MPNST cell lines. These results are consistent with previous reports on other human cells. ³³ , ³⁶ , ⁴³ Given the sensitivity of SMARCAD1 mutant cells to PARP inhibitors, the demonstration of smarcad1a as a tumor suppressor gene might provide a new potential target for MPNST therapy.

4.1. Loss of smarcad1a accelerates MPNST tumorigenesis, suggesting it is a novel tumor suppressor gene in zebrafish

The SMARCAD1 gene encodes a member of the SWI/SNF (switching defective/sucrose non‐fermenting) complex, a well‐characterized machinery that affects chromatin structure that is frequently mutated in many human cancers. ²⁹ , ³⁰ , ⁷¹ , ⁷² The yeast orthologous FUN30/FFT3 has been found to play essential roles in regulating and maintaining silent chromatin domains, repairing DNA DSBs, preserving genome stability, and facilitating polymerase II transcriptional elongation. ³³ , ³⁴ , ³⁶ , ⁴³ , ⁷² , ⁷³ However, until now, there was no clear in vivo evidence to support that SMARCAD1 is a tumor suppressor gene. Here, we report the first in vivo genetic evidence that the zebrafish smarcad1a gene is a novel tumor suppressor gene, as both sa1299 and p403 mutant fish showed increase tumorigenesis with tp53 mutant‐initiated MNPSTs. Furthermore, we have demonstrated that double‐strand DNA break repair is compromised in both mutants, suggesting that smarcad1a is required to maintain genomic integrity. Thus, defects in DNA repair upon the misregulation of this gene may be one of its tumorigenic mechanisms. In the future, it will be interesting to investigate the mechanistic role of smarcad1 genes in DNA damage repair pathways in vivo using zebrafish models. In addition, it is worth noting that Smarcad1 null mice were reported to die from gastrointestinal tumors, although the skeletal defects were the primary phenotype in live mice from an early knockout mouse study. ⁷⁴ Along this line, human SMARCAD1 mutations were found to cause Huriez syndrome (sclerotylosis), Basan syndrome (adermatoglyphia), and are susceptible to skin cancers. ³⁶ , ⁷⁵ Altogether, these data suggest that SMARCAD1 may be an evolutionarily conserved tumor suppressor gene in vertebrates. Future conditional knockout mouse models will be helpful to address whether Smarcad1 is also involved in mouse MPNSTs since the new conditional mice were just created recently. ⁷⁶

In our experiments, both SMRACAD1 knockdown in HEI‐193 and overexpression in STS26T led to cell growth inhibition. This phenomenon may be due to more SMARCAD1 protein than physiologically needed, leading to lengthened resection tracks, which would also be deleterious to genome stability and indirectly cause cell growth inhibition. Thus, as tumor suppressor gene involved in DNA damage repair, either gain or loss of function may lead to cellular misfunction, similar to BRCA1 and BRCA2. The knockout of Brca1 and Brca2 resulted in embryonic lethality and cell proliferation defects in mice, ⁷⁷ while overexpression of BRCA1 and BRCA2 also causes anti‐proliferative effects in human cells. ⁷⁸ , ⁷⁹ This could be a common feature of tumor suppressor genes that are all involved in DNA damage repair. Therefore, we should not exclude SMARCAD1 as a tumor suppressor because the knockdown of this gene in human cells slows cell growth. Since both STS26T and HEI‐193 are not NF1 mutants, it will be informative to examine NF1 mutant Schwann cell lines and NF1‐related MPNSTs in the future since the schwannoma cell HEI‐193 possesses an NF2 mutation.

For the molecular mechanism of smarcad1a as a tumor suppressor, there could be a few possibilities. First, SMARCAD1 and its orthologues were repeatedly reported to be essential for DNA damage repair in yeast and human cell lines. ³³ , ³⁴ , ³⁶ It was known that BRCA1‐BARD1's function in homologous recombination requires SMARCAD1. ⁸⁰ In addition, SMARCAD1 is required by Msh2 for mismatch repairs. ³⁸ Our pH2AX expression results of zebrafish embryos, human neurofibroma, and MPNST cells after irradiation support that this gene is needed for DNA damage repair, although the detailed roles in different repair pathways need further investigation. Second, SMARCAD1 was reported to play important roles in maintaining genome stability through heterochromatin silencing, endogenous retrovirus inhibition, interacting with E1a oncoprotein, and replication fork stability. ²¹ , ³⁹ , ⁴⁰ , ⁷² Third, as a chromatin remodeler, SMARCAD1 was also reported to regulate gene expression through transcription. ⁸¹ , ⁸² Another chromatin remodeler, SMARCB1/INI1 was reported to regulate RB1, MYC, SHH, and WNT signaling pathways. ⁸³ Thus, SMARCAD1 mutation or misexpression may directly lead to other oncogene activation and tumor suppressor gene inactivation, which may cause cancer. Based on these possibilities, we propose a model of SMARCAD1 as a tumor suppressor in Figure 11.

FIGURE 11.

Model of SMARCAD1 tumor suppressor mechanisms. Three mechanisms (DNA damage response, heterochromatin maintaining, and transcription regulation) were proposed based on current knowledge. ³³ , ³⁴ , ³⁶ , ³⁸ , ⁷² , ⁷³ , ⁸⁰ , ⁸¹ The bidirectional arrows indicate protein–protein interactions. The arrows imply cellular consequences. The dashed arrows show a possible result. The involved genes and pathways are solely based on loss‐of‐function studies due to the lack of overexpression reports

4.2. Developmental roles of SMARCAD1

Mouse Smarcad1 (a.k.a. Etl1) was first reported from a LacZ enhancer trap line. It was expressed in the central nervous system, mesenchyme of the maxillary and mandibular arches, limb bud ectodermal cells around the snout and limb buds, liver, and spinal ganglia of mouse embryos. ⁸⁴ The mouse neural expression is similar to what we found in zebrafish embryos, suggesting its roles during neural development are evolutionarily conserved. Smarcad1 loss‐of‐function mice were created through recombination by inserting a selectable marker (lacZ‐neo or hygromycin B‐phosphotransferase) in frame to the Smarcad1 coding sequence immediately downstream of the first ATG to preclude the generation of any full‐length protein. ⁷⁴ About half of the homozygous Smarcad1 inactivated mice (48%) were able to survive to adulthood but exhibited variable skeletal dysplasia and reduced body weight. ⁷⁴ It is worth mentioning that gastrointestinal tumors were noticed in some of the lost homozygous mutants by autopsy. ⁷⁴ In contrast, our zebrafish smarcad1a sa2199 and p403 adult homozygotes did not have any noticeable morphological abnormalities or postnatal mortality. One likely explanation is the functional gene compensation. This idea is supported by the zebrafish smarcad1a and smarcad1b genes sharing overlapped expression during embryogenesis (Figure 3), as well as smarcad1b expression not being affected in the sa1299 null mutant embryos (Figure 4F). Future studies with the inactivation of the smarcad1b gene will be interesting to test this possibility.

In humans, SMARCAD1 gene mutants and variants were reported to be linked to Basan syndrome, an autosomal‐dominant adermatoglyphia, which is characterized by rapid healing congenital acral bullae, congenital milia, and lack of fingerprints. ⁴² , ⁸⁵ In addition, SMARCAD1 was also linked to a rare dominant disease, Huriez syndrome, which is characterized by congenital palmoplantar keratosis, scleroatrophic changes of the hands and feet, and an increased risk for cutaneous squamous cell carcinoma. ⁴³ Detailed examination of zebrafish or mouse embryonic development might shed light on these specific genetic diseases in humans.

4.3. Cellular functions of SMARCAD1 and future cancer therapy development

SMARCAD1 is a member of the SNF2 helicase subfamily. The members of this subfamily, such as ION80 and SWR1 genes, are known to play essential roles in the DNA damage repair processes. ⁸⁶ , ⁸⁷ Here, we have confirmed this DNA damage repair function in zebrafish smarcad1a mutants and human MPNST cell lines. More importantly, we demonstrated that the dosage of this gene, either overexpressed or downregulated, is critical for DNA DSB responses induced by X‐ray irradiation. The sensitivity of SMARCAD1 mutant cancer cells to DNA enzyme topoisomerase I inhibitor, CPT (camptothecin), and PARP inhibitors is beneficial for developing new therapies for MPNSTs given there is no effective treatment for this type of malignancy. ¹² Similar to BRCA1 and BRCA2 genes, the SMARCAD1 gene is also involved in the DNA end‐resection process and interacts with KAP1, BRCA1, and PARP1. ⁷² , ⁸⁰ Thus, this gene might provide us another targetable tumor suppresser gene with PARP inhibitors. Although we demonstrated the tumor suppressor activity of smarcad1a in MPNSTs here, its potential application with PARP inhibitors should not be limited to this type of cancer. This is evident with the SMARCAD1‐knockdown osteosarcoma cell line, U2OS, which showed a similar response to CPT and PARP inhibitors. ³³ Of course, this idea needs more experimental support and SMARCAD1 mutation characterization from human MPNST studies. There is still a long way to go to prove this idea.

Interestingly, human SMARCAD1 variants were also reported to be correlated with fluorouracil (5‐FU) sensitivity by GWAS, suggesting that SMARCAD1 might also be a useful marker for cancer treatment decisions. ⁸⁸ Additionally, it is worth noting that homozygous Smarcad1 mouse mutant embryonic stem cells did not show differences in their survival rates after gamma or UV‐radiation treatment. ⁷⁴ However, consistent evidence from yeast, zebrafish, and human cells suggests DNA damage repair is an evolutionarily conserved function of SMARCAD1. One possible explanation for this difference could be incomplete inactivation of the mouse Smarcad1 gene (a.k.a. ETL1), or irradiation dosage threshold. Future studies with Smarcad1 are needed to clarify the discrepancy with the reported mouse mutant.

4.4. Future of zebrafish–human comparative cancer genomics and the zebrafish cancer model

Cross‐species comparative oncogenomics serves as one of the solutions for identifying cancer drivers on large CNAs due to genes conserved functions. ⁵ , ⁶ , ⁷ , ⁸ We have found that zebrafish–human comparative cancer genomics with CNAs is an effective way to narrow down the cancer driver candidate genes on aneuploid chromosomes. ⁸ , ⁶⁶ In this report, we demonstrated that the smarcad1a mutants, sa1299 and p403, accelerated tumorigenesis initiated by tp53. Thus, smarcad1a is a novel tumor suppressor gene in zebrafish MPNSTs. Moreover, our results also demonstrated that the combination of zebrafish–human comparative cancer genomics and functional genetic studies in zebrafish is a powerful approach for a long‐lasting challenge for identifying novel cancer driver genes on large CNAs/aneuploid chromosomes. Considering the plethora of available zebrafish mutants from large‐scale forward genetic screens and convenient reverse genetics, such as TALEN and CRISPR, ⁸⁹ , ⁹⁰ , ⁹¹ , ⁹² a relatively large number of candidate cancer driver genes can be functionally validated in vivo with a similar approach as demonstrated for the smarcad1a gene here.

CONFLICT OF INTEREST

All authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

GuangJun Zhang designed the project and coordinated research; Han Han, Guangzhen Jiang, Rashmi Kumari, Martin R. Silic, Jake L. Owens, and GuangJun Zhang performed experiments; Han Han, Rashmi Kumari, Guangzhen Jiang, Martin R. Silic, Chang‐Deng Hu, Suresh K. Mittal, and GuangJun Zhang analyzed the results; GuangJun Zhang wrote the original draft of the manuscript, all authors reviewed and edited the final draft of the manuscript.

Supporting information

Figure S1 Maximum likelihood phylogeny of SMARCAD1 proteins, as obtained with JTT plus gamma distances. Numbers around each node indicate bootstrap values based on 1000 replicates. Branch lengths are proportional to expected replacements per site. The overall cladistics of each group is consistent with the BP and ME phylogenies. The tree was rooted with yeast SMARCAD1/FUN30.

Figure S2. Minimum evolution phylogeny for SMARCAD1 proteins, as obtained with JTT plus gamma distances (α = 0.61). Numbers indicate bootstrap scores for each node, based on 10 000 replicates, whereas branch lengths are proportional to expected replacements per site. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates are collapsed. The analysis involved 48 amino acid sequences. All positions containing gaps and missing data were eliminated. There were a total of 532 positions in the final dataset. Evolutionary analyses were conducted in MEGA6. This tree is rooted by yeast SMARCAD1. Constant with BP and ML, the ME phylogeny also cluster the teleost SMARCAD1 into two distinct clades, supporting that they likely resulted from bony fish‐specific whole‐genome duplication event.

Figure S3. The annotation on proximal critical spicing of zebrafish smarcad1a sa1299 mutant. (A) Normal splicing of the smarcad1a ENSDART00000091409 transcript in wildtype fish. (B) proximal cryptic splicing in the sa1299 mutant. The essential splice site GT is mutated into GG in sa1299. The proximal cryptic site is 11bps downstream of the mutation site in intron 9 (red box). Introns are indicated by black bars. The retained part of intron 9 is highlighted in blue. The sizes of introns and exons are not proportional to their actual lengths.

Figure S4. Creation and validations SMARCAD1 human cell lines. (A) The Human SMARCAD1 gene can be knocked down by 3 independent shRNAs in the neurofibroma cell line, HEI‐193, using a tetracycline‐inducible lentiviral system. 3351712 has the highest efficiency. (B) The knockdown efficiency of 3351712 can be modulated with different dosages of doxycycline. (C) SMARCAD1 STS26T MPNST stable cell line was verified by western blot. The overexpression can also be modified by different dosages of doxycycline. ACTB is a loading control for all the western blots.

Table S1. Primer sequences used in this article.

Table S2. SMARCAD1 protein sequences used for phylogenetic analysis.

Han H, Jiang G, Kumari R, et al. Loss of smarcad1a accelerates tumorigenesis of malignant peripheral nerve sheath tumors in zebrafish. Genes Chromosomes Cancer. 2021;60(11):743‐761. 10.1002/gcc.22983

DATA AVAILABILITY STATEMENT

Reagents are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.