Different miRNAs Related to FBXW7 Mutations or High Mitotic Indices Contribute to Rectal Neuroendocrine Tumors: A Pilot Study

Recent studies suggest that miRNA may be involved in the development of rectal neuroendocrine tumors (NETs). We explored the frequency of clinicopathologically relevant mutations and miRNA expression in rectal NETs to examine molecular profiles related to prognosis and behavior. Twenty-four eligible specimens with endoscopically excised rectal NETs were selected. Next-generation sequencing and an miRNA expression assay were used to evaluate the expression profile relevant to common genetic mutations in rectal NETs. Kyoto Encyclopedia of Genes and Genomes analysis predicted that the possible target signaling pathways were correlated with dysregulated miRNAs. Nineteen rectal NETs harbored more than one mutation in the 24 cancer-related genes. Seven miRNAs (hsa-miR-769-5p, hsa-miR-221-3p, hsa-miR-34a-5p, hsa-miR-181c-5p, hsa-miR-1246, hsa-miR-324-5p, and hsa-miR-361-3p) were significantly down-regulated in tumors harboring the FBWX7 mutation. Unsupervised hierarchical clustering analysis showed that up-regulation of these seven miRNAs may result in high mitotic indices, indicating the role of miRNAs in tumor progression. Among the down-regulated miRNAs, hsa-miR-769-5p was strongly correlated with extracellular matrix–receptor interaction and lysine degradation. Among the clinicopathological factors, up-regulated hsa-miR-3934-5p was linked to an increased mitotic count. No change in miRNA expression was associated with a tumor size >1 cm, lymphovascular invasion, or Ki-67 index. In summary, we identified different miRNA signatures involved in FBXW7 mutations or high mitotic indices in rectal NETs, which may play a critical role in tumor behavior.


Screening of Candidate miRNAs Associated with Specific Genetic Mutations
We screened the expression of 584 miRNA relevant to the aforementioned 24 genetic mutations in mutated and wild-type tumors. The results revealed that FBXW7 was the only candidate gene relevant to miRNAs differentially expressed between mutated and wild-type NETs. When comparing the miRNA expression levels of 19 tumors harboring FBXW7 wild-type with 5 tumors carrying FBXW7 mutations, we observed seven downregulated miRNAs (hsa-miR-221-3p, hsa-miR-769-5p, hsa-miR-181c-5p, hsa-miR-34a-5p, hsa-miR-1246, hsa-miR-324-5p, and hsa-miR-361-3p) in FBXW7-mutant tumors compared to FBXW7-wild-type tumors (all p < 0.001, Table 2). Among them, hsa-miR-769-5p was the most correlated with FBXW7 mutation (false discovery rate [FDR] < 0.05). Then, unsupervised hierarchical clustering analysis was performed to compare the expression profiles of the dysregulated seven miRNAs (Figure 1). The analysis separated the tumors into two main clusters according to miRNAs being up-or down-regulated. The cluster with up-regulated miRNA expression tended to correlate with NETs with higher mitotic indices, grade 2, wild-type FBXW7, and wild-type TP53. In contrast, the cluster with down-regulated miRNAs tended to correlate with NETs with lower mitotic indices, grade 1, FBXW7 mutation, and TP53 mutation. These findings indicated distinct miRNA differential expression (hsa-miR-769-5p, hsa-miR-221-3p, hsa-miR-34a-5p, hsa-miR-181c-5p, hsa-miR-1246, hsa-miR-324-5p, and hsa-miR-361-3p) between the tumor groups with high mitotic indices, grade 2, wild-type FBXW7, and wild-type TP53 and other tumor groups with low mitotic indices, grade 1, FBXW7 mutation, and TP53 mutation.  Hierarchical clustering map of differentially regulated miRNAs between FBXW7 wildtype and mutated rectal neuroendocrine tumors. Each row represents one miRNA, and each column represents one sample. Red and light yellow indicate the high expression level above the mean and the low expression level below the mean.
When comparing the miRNA expression profiles between tumors with either low (0-1) or high (≥2) mitotic indices, we identified five differentially expressed miRNAs between the two groups: miR-3934-5p was significantly up-regulated, and miR-153-3p, miR-4286, miR-223-3p, and miR-129-5p were down-regulated in the high mitotic index group. After adjustment, only miR-3934-5p was statistically significant (Table 3). However, miR-3934-5p, related to a high mitotic index, was not correlated with the miRNA expression profiles related to FBXW7 mutations (p > 0.99999). Hierarchical clustering map of differentially regulated miRNAs between FBXW7 wild-type and mutated rectal neuroendocrine tumors. Each row represents one miRNA, and each column represents one sample. Red and light yellow indicate the high expression level above the mean and the low expression level below the mean.

Discussion
In this study, screening 584 miRNA expressions in rectal NETs revealed that FBXW7 and mitotic index were the sole candidate genes and clinicopathological features specifically relevant to miRNAs, respectively. FBXW7 acts as a potent tumor suppressor that may control the expression levels of multiple oncoproteins (i.e., mTOR, notch, cyclin E, c-Myc, and c-Jun) that engage in cellular signaling pathways by inducing them for ubiquitin-mediated proteasomal degradation, thereby protecting against cancer development [27,28]. Several studies have demonstrated that FBXW7 also significantly influences DNA damage response and repair processes, prime molecular mechanisms for maintaining cellular homeostasis, and genomic integrity [29][30][31][32]. Accordingly, FBXW7 dysregulation in this process may be a critical factor contributing to tumorigenesis [27,28]. A recent study has reported a high frequency of P53 (61%), APC (53%), FBXW7 (25%), and KRAS (25%) mutations in rectal NETs [15]. Likewise, in our previous study, FBXW7 (p.R465H) was found to be likely pathogenic in low-grade rectal NETs [18]. In this context, it may be important to focus on the epigenetic regulation of FBXW7 in rectal NETs, as FBXW7 is a critical regulator of several key signaling pathways involved in cell growth and proliferation [27,28]. Understanding the epigenetic regulation of FBXW7 in rectal NETs may provide insights into the mechanisms underlying tumor development and potentially identify novel therapeutic targets for treating these tumors.
Alterations in FBXW7 expression have been linked to the development and progression of several types of cancer [27,28]. Some FBXW7 regulators include miRNAs, such as miR-27 in lung cancer, miR-32 in breast cancer, miR-92a in cervical cancer, miR-223 in

Discussion
In this study, screening 584 miRNA expressions in rectal NETs revealed that FBXW7 and mitotic index were the sole candidate genes and clinicopathological features specifically relevant to miRNAs, respectively. FBXW7 acts as a potent tumor suppressor that may control the expression levels of multiple oncoproteins (i.e., mTOR, notch, cyclin E, c-Myc, and c-Jun) that engage in cellular signaling pathways by inducing them for ubiquitinmediated proteasomal degradation, thereby protecting against cancer development [27,28]. Several studies have demonstrated that FBXW7 also significantly influences DNA damage response and repair processes, prime molecular mechanisms for maintaining cellular homeostasis, and genomic integrity [29][30][31][32]. Accordingly, FBXW7 dysregulation in this process may be a critical factor contributing to tumorigenesis [27,28]. A recent study has reported a high frequency of P53 (61%), APC (53%), FBXW7 (25%), and KRAS (25%) mutations in rectal NETs [15]. Likewise, in our previous study, FBXW7 (p.R465H) was found to be likely pathogenic in low-grade rectal NETs [18]. In this context, it may be important to focus on the epigenetic regulation of FBXW7 in rectal NETs, as FBXW7 is a critical regulator of several key signaling pathways involved in cell growth and proliferation [27,28]. Understanding the epigenetic regulation of FBXW7 in rectal NETs may provide insights into the mechanisms underlying tumor development and potentially identify novel therapeutic targets for treating these tumors.
Here, we noted that hsa-miR-769-5p was most correlated with FBXW7 mutation, further linked to ECM-receptor interaction and lysine degradation via the KEGG analysis of target signaling pathways. The ECM is a central element of both normal and tumor tissues, consisting of a complex texture of cross-linked proteins to play a role in resident cells' hydration, elasticity, and structural organization [38]. It importantly helps in neoplastic proliferation, angiogenesis, progression, and tumor survival [39]. Specific ECM alterations regulating tissue biochemical and biomechanical properties have been described in NETs [40], which may imply that several ECM glycoproteins may contribute to the tumorigenesis of NETs. In a recent study, the up-regulation of six fibrogenic genes (COL5A2, COL1A2, COL3A1, ITGA5, ITGB1, and ITGB1) in pulmonary neuroendocrine carcinoma and the down-regulation of pulmonary neuroendocrine neoplasms were strongly related to no metastasis and overall survival [41]. Only one study has shown that the activation of the lysine degradation pathway injures tumor cell proliferation [42], which is also related to miR-769-5p tumor suppression, as shown in subsequent studies. miR-769-5p/miR-769-3p-mediated inhibition of proliferation causes apoptosis in gastric cancer cells [43]. Another study has shown that miR-769-5p inhibited non-small cell lung cancer proliferation and invasion, indicating that this type of tumor could benefit from miR-769-5p as a diagnostic and prognostic biomarker [44]. hsa-miR-769-5p has also been described to be expressed at low levels in papillary thyroid carcinomas, and the overexpression of hsa-miR-769-5p highly affected the formation of the tumor microenvironment [45], which appears to be consistent with our findings.
Up-regulated miR-3934-5p and down-regulated miR-4286, miR-223-3p, miR-129-5p, and miR-153-3p were characterized in tumors with high mitotic indices, in which miR-3934-5p was the only one statistically significant after adjustment. Further clustering analysis separated two main groups: (1) high mitotic index, grade 2, wild-type FBXW7, and wild-type TP53 and (2) low mitotic index, grade 1, FBXW7 mutation, and TP53 mutation. This may indicate that these miRNA contributions were mutually exclusive between the high mitotic index and FBXW7 mutation in rectal NETs. The KEGG pathway maps also confirmed the different pathway involvement in different miRNAs between FBXW7mutated and highly mitotic-active tumors. hsa-miR-3934-5p may be closely implicated in a high mitotic index, eventually contributing to higher grade 2 tumors in rectal NETs.
In contrast, hsa-miR-769-5p related to FBXW7 mutation may be involved in grade 1 tumors and is less likely to be involved in tumor aggressiveness. Indeed, a recent study has discovered that WHO grade 3 gastro-pancreatic NETs tend to harbor NETs with lower FBXW7 mutation frequency than grade 1 and grade 2 NETs [15]. Comparisons of normal pancreatic islets with their matched neoplastic groups disclosed changes in ECM compositions, which also differed depending on histological grade [40,46]. This may explain the finding that the ECM-receptor interaction KEGG pathway (hsa-miR-769-5p) was significantly associated with FBXW7 mutations and was more likely involved in grade 1 tumors. hsa-miR-769-5p appears to be a possible regulator of FBXW7-mutated rectal NETs. These differentially expressed miRNAs could be potential tissue biomarkers for aggressive behavior in rectal NETs.
Our study was limited by a relatively small number of rectal NETs. We did not perform the validation step using quantitative PCR. However, we utilized the advantages of Nanostring technology to identify 24 cancer-related miRNAs out of 584 miRNAs without requiring the amplification of transcripts. We believe that our approach was able to identify low-abundance transcripts that may have been difficult to validate through PCR [47]. Our study focused on identifying candidate miRNAs. In addition, we did not analyze the difference between plasma and tissue miRNAs. miRNAs from tumors may be secreted within membrane vesicles (exosomes) or directly into the blood, indicating that miRNAs play a key role in intercellular communication [48]. Plasma miRNAs communicate the most with the pericardial, adipose, liver, and spleen tissues [49]. As rectal NETs are submucosal tumors, tissue-based analysis of our miRNAs may be more appropriate as they reflect the origin of these tumors. Because some may metastasize, we aim to further investigate the differences in plasma miRNAs between localized rectal NETs and those with metastasis to advance our understanding of these tumors in future studies.
Nevertheless, some conclusions emerged from our results. To our knowledge, this is the first study to identify a correlation between FBXW7 mutations and miRNAs in rectal NETs. hsa-miR-769-5p was most correlated with FBXW7 mutation and further linked to ECM-receptor interaction and lysine degradation via a KEGG analysis of target signaling pathways. hsa-miR-3934-5p may be strongly implicated in a high mitotic index, which may eventually contribute to higher grade 2 tumors in rectal NETs, whereas hsa-miR-769-5p related to FBXW7 mutation may be involved in grade 1 tumors and is likely to play a role in suppressing rectal NET progression.

Patients
The institutional review board of Hallym University Sacred Heart Hospital approved this study (HALLYM 2022-11-006-002). All participants provided written informed consent for the procedure, and the study protocol adhered to the guidelines and regulations of the ethics committee of Hallym University. A search in the electronic database of the pathology department of Hallym University Sacred Heart Hospital disclosed 69 consecutive patients with primary rectal NETs who underwent endoscopic resection in the hospital between 1 June 2005, and 31 July 2015. These patients were included in previous studies [6,50]. We excluded 45 patients whose formalin-fixed paraffin-embedded (FFPE) tissue blocks contained too few tumor cells for the assay or whose FFPE tumor blocks were inadequate for molecular analysis.

Histologic Evaluation
All H&E-stained slides were reviewed by a gastrointestinal and molecular pathologist (M.J.K.) to confirm the diagnosis and evaluate histopathological characteristics. The results of mitotic rates, LVI, and proliferative index established in the previous studies were applied in the present study [6,50].

DNA Extraction and NGS
A tissue microtome (Leica Biosystems, Wetzlar, Germany) was used to obtain two 10-µm sections from every FFPE tissue block on the glass slides. Tumor areas on the glass slides were manually macrodissected from the unstained tissue sections to enrich tumor cell populations > 50%. DNA was extracted and purified using the Ion AmpliSeq Direct FFPE DNA Kit (Thermo Fisher Scientific, Waltham, MA, USA) and QIAamp DSP DNA FFPE Tissue Kit (Qiagen, Hilden, Germany), respectively. The yield of purified genomic DNA was estimated using a Qubit 2.0 Fluorometer and Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific) prior to library preparation for sequencing.
The NGS platform was an Ion Personal Genome Machine Sequencer (Thermo Fisher Scientific). Library preparation for each sample was conducted using the Ion AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific) and Ion AmpliSeq Cancer HotSpot Panel v2 (Thermo Fisher Scientific), following the manufacturer's protocols, as previously explained [18]. The Ion Torrent platform-specific pipeline software was employed via the variant calling process. Annotation was performed using the Variant Effect Predictor. To compare our data with findings from other cancer genome studies and check for hotspot mutations, we downloaded all cancer genome studies from cBioPortal using the CGDS-R package (R foundation for Statistical Computing, Vienna, Austria). To identify confident putative somatic variants, annotated raw variants were filtered according to the following criteria: (1) non-synonymous singlenucleotide variant or short insertion or deletion in coding regions; (2) coverage ≥ 50× and variant allele frequency (VAF) ≥ 5% or coverage ≥ 1000× and VAF ≥ 3%; and (3) minor allele frequency < 0.1% in the gnomAD and 1000 Genomes Project. The resulting list of variants was manually reviewed and visually confirmed using the Integrated Genomics Viewer (http://www.broadinstitute.org/igv/, accessed on 1 October 2022).

miRNA Extraction and Nanostring nCounter miRNA Expression Assay
The miRNAs were isolated from FFPE tissues using the miRNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. The concentration was assessed using a Nan-oDrop spectrophotometer (Thermo Fisher Scientific) [51]. nCounter human v3 miRNA expression assays designed and synthesized by NanoString Technologies (NanoString, Seattle, WA, USA) were performed in this study. The miRNA panel included oligonucleotide tags for 798 human miRNAs (from miRBase v21) and 5 housekeeping mRNAs for reference (ACTB, B2M, GAPDH, RPL19, and RPLP0). Twenty-five control probes that recognized either synthetic mRNA or miRNA targets were used to monitor the efficiency and specificity of each reaction step. Hybridization was performed with 150 ng of total RNA (16 h at 65 • C) with probe pairs consisting of reporter probes carrying the signal at their 5 end and capture probes carrying biotin on their 3 ends. After hybridization, sample cleanup and digital report counts were performed according to the manufacturer's instructions. Raw data were analyzed using the NanoString nSolver software (version 4.0). Internal positive controls normalized raw data to remove variability in hybridization, and negative controls to subtract the background. In addition, miRNA input levels were normalized using the spike-in normalization method according to the manufacturer's instructions. The minimum threshold to consider an miRNA for further analysis was 20 counts in at least 10% of the samples. This threshold allowed us to identify 584 miRNAs suitable for analysis.

DNA Intelligent Analysis (DIANA)-miRPath for miRNA Pathway Analysis
Pathway analysis using significantly up-regulated and down-regulated miRNAs was performed using DIANA-miRPath v3.0 [52]. This software allowed us to link miRNAs to experimentally validated target genes from Tarbase, v7.0/microT-CDS v5.0 (https:// dianalab.e-ce.uth.gr/, accessed on 1 October 2022) and identifies the putative targeted molecular 'KEGG' pathways. As our analysis was hypothesis-free, we used the pathway union option of the miRPath software. The pathway analysis was performed using miRPath v3.0 with an FDR < 0.05, using the Benjamini and Hochberg methods.

Statistical Analyses
The differentially expressed mRNAs and significant pathways were selected based on the p-value and FDR analysis [53]. The miRNA expression levels were compared to each clinicopathological parameter using the Mann-Whitney U test, calculated using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA). The p-values were obtained through Fisher's exact test and χ2 test as enrichment analysis methods. The FDR was estimated using the two-stage process of Benjamini, Krieger, and Yekutieli using FDR < 0.05 and p < 0.001 as the significance threshold [26]. Hierarchical clustering was performed using the R package, version 4.13, using Pearson's correlation.

Conclusions
Different miRNAs related to high mitotic indices or FBXW7 mutations may contribute to tumor behaviors in rectal NETs. Specifically, FBXW7 mutations are likely to play an important role in inhibiting rectal NET progression, whereas mitosis may be the only significant histological risk factor related to miRNA expression. Furthermore, differentially expressed miRNAs may be used as potential tissue biomarkers for possible prognostic indicators in rectal NETs.