SET domain containing 1B gene is mutated in primary hepatic neuroendocrine tumors

Abstract

Primary hepatic neuroendocrine tumors (PHNETs) are extremely rare NETs originating from the liver. These tumors are associated with heterogeneous prognosis, and few treatment targets for PHNETs have been identified. Because the major genetic alterations in PHNET are still largely unknown, we performed whole-exome sequencing of 22 paired tissues from PHNET patients and identified 22 recurring mutations of somatic genes involved in the following activities: epigenetic modification (BPTF, MECP2 and WDR5), cell cycle (TP53, ATM, MED12, DIDO1 and ATAD5) and neural development (UBR4, MEN1, GLUL and GIGYF2). Here, we show that TP53 and the SET domain containing the 1B gene (SETD1B) are the most frequently mutated genes in this set of samples (3/22 subjects, 13.6%). A biological analysis suggests that one of the three SETD1B mutants, A1054del, promotes cell proliferation, migration and invasion compared to wild-type SETD1B. Our work unveils that SETD1B A1054del mutant is functional in PHNET and implicates genes including TP53 in the disease. Our findings thus characterize the mutational landscapes of PHNET and implicate novel gene mutations linked to PHNET pathogenesis and potential therapeutic targets.

Introduction

Neuroendocrine tumors (NETs) have an average annual incidence of 2 per 100,000 cases among all tumors of the gastrointestinal (GI) tract. NETs are derived from neuroendocrine cells that are capable of producing functional peptide hormones. Typically, these tumors are found in the GI tract (55%) and lungs (30%), although they can also arise in the pancreas (2%), reproductive system (1%), biliary tract (1%) and head and neck (0.4%), among other areas.¹ GI NETs frequently metastasize to the liver, but primary NETs of hepatic origin are extremely rare.² According to the 2010 World Health Organization (WHO) classification system, primary hepatic neuroendocrine tumors (PHNETs) are divided into three grades (1–3) based on their histological criteria. Ultrasound, CT and MRI are all typically employed to characterize these lesions, but their appearance on diagnostic imaging can be highly variable and often mimics the features of more common hepatic malignancies. Due to the rarity of PHNETs, their clinical features and treatment outcomes are largely unknown, but there remains a compelling need to extensively identify genomic abnormalities underlying PHNETs and to elucidate their molecular basis for the purpose of guiding effective clinical therapies.

Here, we present the mutational landscape of PHNETs using whole-exome sequencing (WES) to identify tumor recurrent mutations in multiple genes involving disrupted pathways. We report for the first time the identification of somatic mutations in the SET domain containing 1B gene (SETD1B) in PHNETs. One of the three SETD1B mutants, A1054del, confers a more aggressive PHNET phenotype. This is the largest genomic sequencing study of PHNETs and represents major progress toward the complete molecular characterization of this rare tumor, which may help improve its clinical management.

Materials and Methods

Clinical sample collection

PHNET tissue samples from the livers of confirmed PHNET patients were collected from Beijing 302 Hospital, Eastern Hepatobiliary Surgery Institute/Hospital, Chinese PLA General Hospital, Beijing 307 Hospital, Fuzhou General Hospital of Nanjing Military Command of Chinese PLA, Peking University Cancer Hospital & Institute, Cancer Hospital Chinese Academy of Medical Science and Beijing You’an Hospital. All patients included in our study were clinically diagnosed with PHNETs, and all tumors were confirmed based on pathology, MRI, CT or PET-CT. All of the samples used in our study were residual specimens collected after diagnosis, and none of the patients received treatment prior to sample collection. Matched normal tissues were collected from adjacent liver tissue 5 cm from the border of the surgical area. Informed consent was obtained from all study participants. All experimental protocols were approved by the Beijing 302 Hospital Ethics Committee.

All tumor and normal samples were subjected to hematoxylin staining and histopathological review to assess the presence of tumor cells, normal liver cells, lymphocytic infiltration and necrotic cells. Tumor cellularity was visually scored in a semiquantitative fashion and only tumors with >80% of malignant cells were used for sequencing. The tumor histology was centrally reviewed and validated according to the WHO guidelines. WES of the paired tumor and nonadjacent normal tissue samples from 22 patients were performed. Mass spectrometric genotyping was used to ensure that the tumor DNA and normal DNA were derived from the same patient.

DNA sample preparation

Frozen and formalin-fixed paraffin-embedded (FFPE) tumor tissue samples with matched adjacent tissues from 22 PHNET patients were obtained for WES. Genomic DNA was extracted from frozen samples with a DNeasy Blood and Tissue Mini Kit (Qiagen, Hilden, Germany), and genomic DNA was extracted from FFPE samples with a QIAamp DNA FFPE Tissue Kit (Qiagen) according to the manufacturer’s instructions. Sample testing included the sample concentration, integrity and purity. The sample concentration was detected by a fluorometer or Microplate Reader (Qubit Fluorometer, Invitrogen, Carlsbad, CA). Sample integrity and purity were detected by Agarose Gel Electrophoresis (concentration of agarose gel: 1%, voltage: 150 V, and electrophoresis time: 40 min).

Exome capture and sequencing

The qualified genomic DNA from 22 PHNET patients was fragmented by Covaris Technology (Woburn, MA) into library fragments that were 200–300 bp in size, and adapters were ligated to both ends of the fragments. The extracted DNA was amplified, purified and hybridized to the Nimble GenSeqCap EZ Exome 64M (V3) human exome array for enrichment. Both noncaptured and captured PCR products were subjected to real-time PCR to estimate the degree of enrichment. Each captured library was loaded on a HiSeq4000 platform. The raw data were processed using the base-calling module (version 1.7) for base calling, and the sequences of the samples from each individual were generated as 90-bp paired-end reads.

Reads mapping and variation calling

Sequence tags were mapped to the hg19 human genome reference sequence using the Burrows–Wheeler Aligner (BWA, version: 0.5.9-r16). Somatic point mutations were detected by VarScan2.2.5 using the following parameters: SAM tools (v0.1.18) mpileup −Q 0 && VarScan2.2.5 somatic –min-coverage10 –min-coverage-normal 10 –min-coverage-tumor 10 –min-var-freq 0.1 –min-avg-qual 0. Somatic indels were predicted with the GATK Somatic Indel Detector using default parameters. All high-confidence mutations were identified using our in-house pipeline coupled with visual inspection and then annotated with ANNOVAR (version: 2016Feb01).

Copy number alteration analysis

Copy number alterations (CNAs) were detected in the WES data by CONTRA.³ CNAs with a copy ratio of ≥1.25 were classified as amplification, and those with a copy ratio of ≤0.75 were classified as a deletion.

Amino acid sequence alignment of SETD1B across multiple species

The amino acid sequences of the SETD1B protein from Homo sapiens, Mus musculus, Gallus gallus and Xenopustropicalis were retrieved from Ensembl and aligned using ClustalW in the DNASTAR Lasergene software suite.

Computational modeling of SETD1B and validation of the model

The automated server program I-TASSER was employed with its default parameters to build the model according to the homology modeling method.^4,5 The stereochemical quality of the models was verified with the program PROCHECK to select the best model.⁶ The 3D profile of the residue was generated by the Verify3D Structure Evaluation Server and QMEAN.^7,8 The model was created using PyMOL (www.pymol.org).

Cell culture

The human embryonic kidney cell line HEK293T (American Type Culture Collection) was maintained in DMEM GlutaMAX (Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum and penicillin/streptomycin (Life Technologies) and incubated at 37°C in an environment containing 5% CO₂. We authenticated the source of the cell lines and tested for mycoplasma contamination; we did not observe cross-contaminated cell lines or mycoplasma contamination.

Plasmid construction

To construct the SETD1B wild-type plasmid, we cloned the SETD1B ultimate open reading frame (ORF) clone into a pCDF vector under the control of an hCMV promoter. The three SETD1B mutations (S487L, C931Y and A1054del) were generated by PCR mutagenesis using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA) on the pCDF-SETD1B plasmid according to the manufacturer’s instructions.

Immunohistochemistry

Paraffin-embedded tumor sections were deparaffinized using standard protocols. The sections were blocked in 10% goat serum in phosphate-buffered saline (PBS) containing 0.5% Triton X-100 for 30 min, incubated with an anti-SETD1B antibody (Abcam, Cambridge, UK, Cat. No: ab113984) and detected using the HRP Rabbit DAB Detection System. Images were acquired at 400× magnification and processed using an Olympus microscope system. The mitotic index of the tumors was recorded as the number of mitotic figures in 10 high-power fields.

Cell growth and colony formation assays

Anchorage-dependent cell growth was evaluated using a CCK-8 Kit (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer’s protocols. For the colony formation assay, transfected cells were cultured in six-well plates for 2 weeks, after which the colonies were fixed with 4% paraformaldehyde and stained with crystal violet for 30 min. The number of colonies with a diameter greater than 1.5 mm was counted.

Cell cycle analysis

Cell cycle analysis was conducted using flow cytometry as described previously.^9,10 Briefly, cells were fixed in 70% ethanol for at least 18 hr, washed with PBS, and incubated with RNase A (0.2 mg/ml) in PBS. Propidium iodide was then added to the cell suspension, and the samples were analyzed on a FACS Calibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ).

Apoptosis and flow cytometry analysis

An AnnexinV assay (BD) was performed according to the manufacturer’s protocols to assess apoptosis. Briefly, cells were harvested after transfection with either the wild-type or mutant SETD1B plasmids, washed twice with PBS, incubated with FITC-conjugated AnnexinV for 15 min, and measured by flow cytometry (FCM) using the FACScan flow cytometer (BD).

Cell migration and invasion assays

Wound healing assays were used to assess cell migration. Briefly, transfected cells were cultured in six-well plates until they formed confluent monolayers; the cell layer was then mechanically scratched using a 1-ml pipette tip to create a wound. The cells were washed with PBS to remove cellular debris and cultured for an additional 24 hr to permit wound healing. Cell invasion was determined by seeding cells into the upper chamber of a Transwell plate (Corning, NY) coated with Matrigel (BD Biosciences, San Jose, CA). After 18 hr, cells that had invaded through the Matrigel membrane were fixed with 4% paraformaldehyde and stained with crystal violet. The number of invading cells was counted in five randomly selected microscopic fields, and the fields were photographed.

Statistical analysis

Statistical analysis was performed with either the SPSS Version 20.0 statistical software package or GraphPad Prism 6 software. Continuous variables were analyzed using either Student’s t-test or the Mann–Whitney test. Categorical data were compared using either Fisher’s exact test or the χ² test. Overall survival (OS) was defined from the date of diagnosis to the date of death or to the date of the last follow-up. A value of p ≤ 0.05 was considered to indicate statistical significance.

Data availability

The raw sequencing data from our study have been deposited in the Genome Sequence Archive of the Beijing Institute of Genomics at the Chinese Academy of Sciences (http://gsa.big.as.cn/) with the accession number PRJCA000559. Data that support the findings of our study are available from The Cancer Genome Atlas (TCGA) database (http://cancergenome.nih.gov).

Results

Mutational landscape

To explore the molecular pathogenesis of PHNETs and gain a better understanding of the complete landscape of somatic mutations in PHNETs, we first performed WES of DNA from paired tumor/normal tissues from 22 PHNET patients. A summary of the clinical and demographic information of the patients included in the tumor cohort is provided in Supporting Information Table S1. The average sequencing depth was approximately 181-fold (145X–234X) and 98.0% (96.9–98.6%) of the target regions were covered at least 10-fold (Supporting Information Fig. S1a and S1b, Supporting Information Table S2). We identified a total of 777 somatic mutations involving 726 genes, including 194 synonymous, 419 missense, 37 nonsense, 7 splice-site and 120 insertion or deletion (indel) mutations, with a median of 1.07 mutations per megabase (Mb; range 0.15–4.42) in 22 PHNET patients (Supporting Information Tables S3 and S4). Then, we performed GO-Term and Pathway enrichment analysis of these mutant genes, p values were defined by the DAVID algorithm. The enriched GO terms included nuclear chromatin, protein phosphorylation and ATP binding, and so on (Supporting Information Table S5). The results of the KEGG enrichment analysis showed that mutant genes mainly enriched in central carbon metabolism in cancer and FoxO signaling pathway (Supporting Information Table S6). Among those mutations, SETD1B and TP53 represented the most frequently mutated genes (3/22 subjects for both genes, 13.6%; Supporting Information Table S7). We analyzed the somatic mutations of prominent genes using the MutSigCV method and found that none of them were significantly mutated genes with q < 0.1 (Supporting Information Table S8). The most common substitution among these samples was the C:G>T:A mutation, which is similar to the substitution patterns reported in most cancer types¹¹ (Supporting Information Fig. S2 and Table S9). The somatic nonsilent mutation load per individual varied remarkably and did not significantly correlated with the clinical stage of the disease, the hepatitis B virus (HBV) infection status, or the subject’s age. A high somatic mutation burden (≥1.5/Mb) was associated with poor prognosis in PHNET patients (p < 0.001, log-rank test; Supporting Information Fig. S3). Intriguingly, G3 grade tumors exhibited the highest percentage of mutations. A significant increase in tumor grade G3 was associated with the number of somatic mutations and correlated with a more aggressive phenotype.

In the data from 22 patients with PHNETs, we performed a somatic CNA analysis and identified acquired genomic alterations that showed considerable variability across the individual tumors. Among the minimal common regions, amplifications in chromosomes 5p, 17p, 17q, 19p, 19q, 20p and 20q as well as copy number deletions in chromosomes 1p, 10q, 11q and 13q were most frequently observed (Supporting Information Fig. S4 and Table S10). Furthermore, we observed the deletion of CDKN2C, NRAS, TLX1, RHOC, RUNX3 and E2F2 and the amplification of ERBB2, GNA11, MAFB and TCF3, suggesting that alterations in the expression of these genes might be a common event related to PHNETs.

SETD1B mutations in PHNETs

We sought to clarify the pathogenic roles of genes and better define their mutation prevalence among individuals with PHNETs. In total, 22 recurring mutated genes were identified; of these, TP53 and SETD1B were the most frequently mutated genes among the 22 PHNET patients (Fig. 1). Somatic mutations in TP53 (E214X, E154K and 83del) were identified in three out of the 22 samples and were mainly located in the functional domain of p53. Notably, the frequency of TP53 mutations among PHNET patients was much lower than that among hepatocellular carcinoma (HCC) and intrahepatic bile duct carcinoma patients.^12–14 Interestingly, these mutations were associated with poor prognosis in PHNET patients (p < 0.001, log-rank test; Supporting Information Fig. S5).

Figure 1.

Whole-exome sequencing and mutated genes in 22 patients with PHNETs. Each column represents an individual tumor, and each row denotes a single gene. The rates of somatic mutations in each examined tumor sample are expressed as the number of mutations per megabase (Mb) of the covered target sequence (top). Also shown are the key clinical parameters (middle) and the significantly mutated genes (SMG) colored by the type of mutation.

The multiple endocrine neoplasia type 1 (MEN1) gene encodes menin, a putative tumor suppressor associated with multiple endocrine neoplasia type 1. Previous studies have shown that MEN1 is an important molecule in NETs that is frequently mutated in pancreatic neuroendocrine, parathyroid and pituitary tumors.^15,16 In two out of 22 PHNET patients, we identified two novel MEN mutations (R420X and 128del) in loci that are distinct from previously identified loci of MEN1 gene mutations; these two novel mutations were located in the menin domain (Supporting Information Fig. S6), suggesting that they may be related to PHNET pathogenicity.

SETD1B, a SET domain containing 1B gene, is a component of the histone methyltransferase complex that produces trimethylated histone H3 at Lys4 and has been implicated in multiple biological processes.¹⁷ Given that we identified SETD1B gene abnormalities as the most frequent somatic mutations in PHNET patients, we hypothesized that this gene plays an essential role in the pathogenesis of PHNETs. Three SETD1B somatic mutations (S487 L, C931Y and A1054del) were identified in three out of the 22 patient samples (Supporting Information Fig. S7). It is interesting that these mutations are not located in the domain regions (Fig. 2a); however, when we assessed the evolutionary conservation of two single-nucleotide variations (SNVs) and one nonframe shift deletion of SETD1B, we found that these mutations are located in domains in which amino acid sequences are highly conserved across species (Fig. 2b). Next, we investigated the effects that these mutations might exert on the structure of SETD1B and found that the predicted mutant SETD1B structure exhibits great changes; thus, a SETD1B model was constructed and validated using the homology modeling method and the I-TASSER server. We speculated that the SETD1B mutations are closely related to the orientation of the C-terminal domain of the protein and that these mutations most likely alter protein function. Changes in the C-terminal orientation of SETD1B may affect its interaction with other proteins (Fig. 2c). In addition, we found that SETD1B had a mutation rate greater than 5% in colon adenocarcinoma, malignant lymphoma, ovarian cancer and endocrine pancreatic cancer using the International Cancer Genome Consortium (ICGC) database (Fig. 2d). To explore the clinical significance of SETD1B mutations, the patients in the cohort were divided into wild-type and mutant subgroups. Our analysis showed that subjects with SETD1B mutations had a worse prognosis than subjects without mutations in this gene (p = 0.046, log-rank test) as reflected by OS (Fig. 2e). To verify this association, we analyzed 8,872 samples from TCGA database with complete survival information and found that the association between SETD1B mutations and prognosis was also significant (p < 0.001, log-rank test; Fig. 2f). The clinical information of 56 “Mut SETD1B” cases in TCGA was shown (Supporting Information Table S11). We also observed that there were significant differences in overall mutation burden between the “Mut SETD1B” and the “WT SETD1B” in TCGA cases (Supporting Information Fig. S8). In addition, SETD1B mutations may correlate with tumor grade in PHNET patients (Supporting Information Fig. S9). Taken together, the correlation of SETD1B mutations with advanced disease stage and poor clinical outcome might explain the aggressive behavior of PHNETs with mutated SETD1B.

Figure 2.

SETD1B mutations in PHNETs. (a) Schematic representation of the domain structure and locations of the somatic mutations within SETD1B observed in our study and collected from the ICGC database. (b) Sequence alignment of the SETD1B protein across distinct species. (c) Prediction of the structures of the SETD1B mutants. (d) Frequency of samples from the ICGC database with SETD1B mutations. (e) Overall survival (OS) of patients with PHNETs according to the mutation status of SETD1B (p = 0.046, log-rank test). (f) Overall survival (OS) of patients in TCGA database according to the mutation status of SETD1B (***p ˂ 0.001, log-rank test). Survival analysis was conducted based on results of Broad Institute TCGA Genome Data Analysis Center (2015): Firehose stddata__2015_11_01 run. Broad Institute of MIT and Harvard. doi: 10.7908/C1571BB1. The mutation and clinic data were downloaded from the Broad TCGA GDAC site (http://gdac.broadinstitute.org/runs/stddata__2015_11_01/data/).

Functional analysis of SETD1B alterations

To gain further insight into the role of SETD1B mutations in PHNETs, we evaluated SETD1B levels by immunohistochemistry in formalin-fixed, paraffin-embedded PHNET samples. PHNET tissues that expressed wild-type (WT) SETD1B showed robust staining for SETD1B, whereas tumor cells from PHNETs with the SETD1B A1054del mutation showed significantly decreased SETD1B expression (Fig. 3a). However, tumors with the SETD1B S487L and C931Y mutations did not exhibit obvious differences in SETD1B expression compared to tumors that express WT SETD1B protein. Similarly, the SETD1B expression levels of paired adjacent tissues in the mutant group (S487L and C931Y) were similar to those of paired adjacent tissues in the WT SETD1B group (Supporting Information Fig. S10). Immunohistochemical staining of SETD1B in the tumors and in paired adjacent noncancerous hepatic tissues was compared to that in HCC tissues as a control (Supporting Information Fig. S11). To explore the functional consequences of the identified SETD1B mutations, we constructed vectors expressing either wild-type or mutant SETD1B protein (S487L, C931Y or A1054del) and transiently transfected HEK293T cells with these vectors. All primer sequences are provided in Supporting Information Table S12. By observing the anchorage-dependent growth of HEK293T cells, we found that HEK293T cells transfected with the SETD1B A1054del mutant grew much faster than cells transfected with either the SETD1B S487L or C931Y mutant or cells expressing wild-type SETD1B (Fig. 3b). To elucidate the mechanism through which SETD1B promotes cell growth, we investigated the effects of SETD1B on the cell cycle using flow cytometry. Compared to HEK293T cells expressing wild-type SETD1B, the percentage of HEK293T cells expressing the SETD1B A1054del mutant in G0/G1 phase (from 51.51% to 42.03%) and G2/M phase (from 36.13% to 21.72%) was remarkably reduced, whereas the percentage of HEK293T cells expressing the SETD1B A1054del mutant in S phase was increased (from 12.36% to 36.25%). In contrast, neither the S487L nor the C931Y mutation in SETD1B significantly changed the percentage of cells in G0/G1, G2/M or S phase (Fig. 3c). Moreover, colony formation assays revealed that the colony number and colony size were larger in cells transfected with the SETD1B A1054del mutant than in cells expressing the SETD1B S487L or C931Y mutations or wild-type SETD1B (Fig. 3d). In addition, the three SETD1B mutants did not appear to regulate apoptosis compared to wild-type SETD1B (Supporting Information Fig. S12).

Figure 3.

Functional analysis of the SETD1B mutations. (a) Representative micrographs of SETD1B protein expression as assessed by immunohistochemistry in PHNET specimens expressing the A1054del mutant and wild-type SETD1B (***p < 0.001, t-test). Scale bars, 500 μm. (b) HEK293T cells transfected with either wild-type or mutant SETD1B (S487L, C931Y and A1054del) were grown in regular medium and harvested at the indicated times. The cell number was determined using CCK-8 assays when compared to WT SETD1B (*p < 0.05 and **p < 0.01, respectively, t-test). (c) Flow cytometry analysis of the cell cycle in HEK293T cells transfected with either wild-type or mutant SETD1B (S487L, C931Y and A1054del). (d) Colony formation assays using HEK293T cells transfected as described in b, when compared to WT SETD1B (***p < 0.001, t-test). All values are presented as the mean ± SD of triplicate measurements. (e) Migration of HEK293T cells transfected with wild-type or mutantSETD1B (S487L, C931Y and A1054del) was determined using the wound healing assay (**p < 0.01, t-test). The images shown here are representative of the three experimental replicates for each condition. (f) Cell invasion of HEK293T cells transfected with wild-type or mutant SETD1B (S487L, C931Y and A1054del) was assessed using a Matrigel-coated invasion chamber. Invading cells were fixed and stained with crystal violet when compared to WT SETD1B (*p < 0.05, t-test). Scale bar: 100 μm. Error bars represent the standard deviation (SD) from at least three independent experiments.

To determine the effects of the SETD1B mutations on cell migration and invasion, wound healing and Transwell invasion assays were performed. The wound healing assay demonstrated that the SETD1B A1054del mutation increased cell migration, but marked changes in motility were not observed in cells transfected with either the SETD1B S487L or C931Y mutant compared to cells expressing wild-type SETD1B (Fig. 3e). Similarly, the Transwell invasion assay revealed that the SETD1B A1054del mutation significantly enhanced the number of invading cells, whereas no differences in the number of invading cells were observed among cells expressing the SETD1B S487L or C931Y mutant compared to those expressing wild-type SETD1B (Fig. 3f). These data suggest that the SETD1B deletion mutation contributes to the tumorigenesis of PHNETs by promoting cell proliferation, migration and invasion and that it may be a potential prognostic marker and/or a therapeutic target for PHNETs.

Dysregulated pathways in PHNETs

We analyzed all of the nonsilent mutations from the 22 PHNET patients and identified genomic alterations that were relevant to several important processes, including epigenetic modifications, the cell cycle and neural development. The changes we observed indicate that abnormalities in these processes might be involved in PHNET tumorigenesis (Fig. 4). Interestingly, we found that epigenetic modifications in the tumor cells were mostly associated with the activity of H3K4 methyltransferases (HMTs), including SETD1B and WDR5; nucleosome remodeling factors (NURF), including BPTF, RBBP4 and HIST1H3A; and histone deacetylase complexes (HDACs), including HDAC1. MEN1 acts as a corepressor of JunD transcriptional activity via HDAC recruitment.¹⁸ HDAC1 has been coprecipitated with WDR5, which has been shown to directly regulate histone methylation and recruit HDACs to histone-bound DNA. Genetic alterations of the cell cycle in PHNETs primarily manifest as distinct defects in the G1/S transition, as demonstrated by the presence of TP53 mutations in three out of 22 cases, ATM, ATAD5, MED12 and DIDO1 mutations in two cases, and PTEN and BRD4 mutations in one case. Neural development is another process in which the mutated genes in PHNETs, including UBR4, MEN1, GLUL, MECP2 and GIGYF2, displayed prominent enrichment. Mutations associated with these three processes were observed in 12 out of the 22 PHNET cases (Supporting Information Table S13). Altogether, these results highlight the potential cooperative roles of these processes in PHNET pathogenesis.

Figure 4.

Somatically altered pathways in PHNETs. Proteins associated with somatic mutations are affiliated with (from left to right) epigenetic modification, cell cycle or neural development. The frequencies of alterations are expressed as a percentage of all patient samples.

In an effort to identify therapeutic targets for PHNETs, we correlated the identified genomic mutations with novel targeted therapeutic approaches that have been approved for clinical use or are currently undergoing clinical trials. In this manner, we identified 13 mutated genes that can potentially be targeted for the treatment of PHNETs¹⁹ (Supporting Information Tables S14 and S15). These genes and their products represent druggable candidate targets and warrant further investigation.

Discussion

To the best of our knowledge, this is the first study in which WES was used to identify somatic alterations in PHNETs. In this report, we analyzed 22 PHNET patients and categorized 22 recurring mutated genes, including TP53, SETD1B and MEN1. Among these genes, SETD1B and TP53 were shown to have the highest frequency of recurring mutations. Of note, the association between SETD1B and PHNETs is a novel finding. Additional immunohistochemical and biological analyses suggested the functional relevance of SETD1B in PHNETs. More importantly, we determined that multiple genes involved in epigenetic modification, the cell cycle and neural development are dysregulated in PHNETs.

MEN1 encodes the histone modifier menin, and its inactivation drives various phenotypes, including widespread transcriptional dysregulation via histone modification. In vitro studies have shown that menin localizes to the nucleus, possesses two functional nuclear localization signals and inhibits JunD-associated transcriptional activation. Most genomic alterations of MEN1 affect one or all of the domains of menin that interact with JunD, Smad3 and NF-κB, three major effectors of transcription and cell growth regulation.²⁰ A previous study identified several germline mutations in the MEN1 gene among Japanese MEN1 families²¹ and among patients with glucagonoma and obesity.²² Moreover, inactivation of SETD1B, which has a broad range of functions, directly influences these three key processes: epigenetic modification, cell cycle and neural development (Fig. 4). SETD1B, encoding a histone 3 lysine 4 methyltransferase, is a component of the SET1 complex (SET1C)/COMPASS complex that participates in many life processes, such as the oncogenic gene expression program and Spemann’s organizer gene activation in Xenopus.¹⁷ SETD1B is a maternal effect gene required for the oogenic gene expression program²³ and a novel recurrently mutated gene in chronic myeloid leukemia.²⁴ SETD1B is essential for hematopoietic stem and progenitor cell homeostasis in adult mice.²⁵ De novo variants in SETD1B are associated with intellectual disability, epilepsy and autism.²⁶ Tang Z’s group reported robust SET1C-mediated H3K4 trimethylation that is dependent upon p53- and p300-mediated H3 acetylation. They also established a mechanism by which SET1C and p300 act cooperatively, through direct interactions and coupled histone modifications, to facilitate the function of p53.²⁷ Previous studies have shown that SETD1B associates with a 450-kDa complex that contains all five noncatalytic components of the SET1A complex: CFP1, Rbbp5, Ash2, WDR5 and Wdr82.²⁸ Concomitant SET1A and SETD1B localization appears as largely nonoverlapping puncta along the euchromatin in the nucleus, suggesting that each protein binds to a unique set of target genes, thereby making non-redundant contributions to the epigenetic control of chromatin structure and gene expression.^29,30 The mixed lineage leukemia (MLL) family of proteins (including MLL1-MLL4, SET1A and SETD1B) specifically methylatesLys4 of histone H3 and plays a pivotal role in the transcriptional regulation of genes involved in hematopoiesis and development.^31,32 Recent studies have provided evidence that TP53 is the target of SET1/COMPASS and have suggested that these genes may cooperate and function as a tumor suppressor.^27,28 Furthermore, the acetylation-regulated interaction between p53 and SET is part of a widespread regulatory process.³³ Li et al. revealed a synergistic role of hSET1 and NURF in regulating the USF-bound barrier insulator that prevents erythroid genes from encroaching upon heterochromatin.³⁴ Chromatin remodeling related genes (KMT2D, KMT2C, SETD1B and BCOR) is frequently mutated in endometrioid endometrial carcinoma.³⁵ Collectively, we concluded that gene-oriented (SETD1B) and pathway-oriented networks are dependent and cooperative and require further investigation in PHNETs. In addition, the main downstream target of SETD1B and how this SETD1B mutation regulates its function warrant to be elucidated in subsequent studies.

NETs are a rare type of malignancy. Understanding the molecular basis of PHNETs at the genome level will be very important for therapeutic development. This report represents the first and largest genome-wide DNA sequencing study of PHNETs ever performed. However, given the number of tumors and normal tissue pairs, we analyzed and the observed mutational frequency in PHNETs, our study had sufficient power to identify only genes with high-frequency mutations (i.e., >10%). Interestingly, the SETD1B mutations were present in all G3 stage tumor samples, in which the mutation burden is highest, and we also found that individuals with SETD1B mutations in TCGA had a much worse prognosis. Due to the high prognostic relevance of the mutation rate and the cooccurrence of two out of three SETD1B mutations with this increased mutation burden, whether SETD1B missense mutations are only passengers needs to be validated in much more PHNETs cohorts. Currently, there are no cell lines with which to perform functional work on this subtype of NETs, and the functional studies undertaken using HEK293T cells are limited in that HEK293T cells are not a primary liver NET cell line. However, a PHNET cell line is a high challenge question and remains to be investigated in a suitable experimental setting. Herein, our data showed that the novel SETD1B A1054del mutation promoted cell proliferation, migration and invasion compared to the wild-type protein. The functional analysis presented in this article seems to support a functional relevance of the A1054del mutations, while the two missense mutations only display no or insignificant effects. The likely reason is A1054del site is close to C-terminal which has complete opposite orientation in “Mut SETD1B” and “WT SETD1B” (Fig. 2c). Importantly, we analyzed the clinical characters of PHNETs with A1054del mutation. Despite the inframe indel A1054del appears in a sparsely mutated sample, this patient had shorter OS time than other patients with two missense mutations. To this end, we concluded that the somatic SETD1B mutation encoding A1054del is likely an activating mutation that contributes to PHNET tumorigenesis. Hence, additional studies should be performed to provide further insight into the biology of PHNETs and other somatic alterations involved in the tumorigenesis and malignant transformation of PHNETs. In addition to the fact that the low incidence of PHNETs has made it difficult to identify statistically significant associations between specific mutations and the disease, many cases of PHNETs have clinically aggressive features, which suggests that interactions between somatic changes and prognosis in this and other rare cancers require further investigation. Of course, since PHNETs may arise from different tissues and tumor stages, it is necessary to consider the heterogeneous features of these tumors and the overall incidence of mutations. In addition to the identification of a specific mutation in SETD1B, our analysis identified many other potential diagnostic and therapeutic approaches for patients with PHNETs, which will offer additional opportunities to identify and treat this typically rare tumor.

What’s new?

Neuroendocrine tumors often metastasize to the liver but primary hepatic neuroendocrine tumors are extremely rare. Here the authors sequenced tumors from 22 patients in China and provide a first look at the mutational landscape of primary hepatic neuroendocrine tumors. The study points to a possible etiological role of the SET domain containing 1B gene (SETD1B) as it was among the most frequently mutated genes, a role that needs to be further explored.

Abbreviations

CNAs

copy number alterations

FFPE

formalin-fixed paraffin-embedded

GI

gastrointestinal

NET

neuroendocrine tumors

OS

overall survival

PHNETs

primary hepatic neuroendocrine tumors

SETD1B

SET domain containing 1B gene

SNVs

single-nucleotide variations

WES

whole-exome sequencing