Imatinib Regulates miR-483-3p and Mitochondrial Respiratory Complexes in Gastrointestinal Stromal Tumors

Metabolic adaptation to increased oxidative phosphorylation (OXPHOS) has been found in gastrointestinal stromal tumor (GIST) upon imatinib treatment. However, the underlying mechanism of imatinib-induced OXPHOS is unknown. Discovering molecules that mediate imatinib-induced OXPHOS may lead to the development of therapeutic strategies synergizing the efficacy of imatinib. In this study, we explored the role of microRNAs in regulating OXPHOS in GIST upon imatinib treatment. Using a microarray approach, we found that miR-483-3p was one of the most downregulated miRNAs in imatinib-treated tumors compared to untreated tumors. Using an extended series of GIST samples, we further validated the downregulation of miR-483-3p in imatinib-treated GIST samples by RT-qPCR. Using both gain- and loss-of-function experiments, we showed that miR-483-3p could regulate mitochondrial respiratory Complex II expression, suggesting its role in OXPHOS regulation. Functionally, miR-483-3p overexpression could rescue imatinib-induced cell death. These findings provide the molecular link for imatinib-induced OXPHOS expression and the biological role of miR-483-3p in regulating cell viability upon imatinib treatment.


Introduction
Gastrointestinal stromal tumor (GIST) is the most common type of gastrointestinal mesenchymal tumor, primarily harboring gain of function mutations in the KIT or plateletderived growth factor receptor-alpha (PDGFRA) genes [1]. The tyrosine kinase inhibitor imatinib that blocks c-KIT and PDGFRA activation is the mainstay treatment for advanced GIST, which has led to improved survival with a median progression-free survival of 18-20 months and a median overall survival of 52-72 months [2][3][4][5]. However, most patients who initially respond to imatinib treatment develop acquired resistance and tumor progression after two years [6].
Deregulated cell metabolism has been recognized as a hallmark of cancer [7]. Although the Warburg effect is generally accepted as the synonym of metabolic reprogramming, recent emerging evidence disclosed that tumor cells could alternate their bioenergetics between glycolysis and oxidative phosphorylation (OXPHOS) in a context-dependent fashion [8]. Metabolic adaptation with shifting from glycolysis to OXPHOS has been We then performed hierarchical clustering analysis of the 34 GIST samples using these 14 significant differentially expressed miRNAs. The analysis classified the samples into two main clusters: all imatinib untreated samples but two (GIST1 and GIST7) were grouped together, while the majority of the treated samples (17 out of 19) were clustered in a separate group ( Figure 1A).
We then performed hierarchical clustering analysis of the 34 GIST samples using these 14 significant differentially expressed miRNAs. The analysis classified the samples into two main clusters: all imatinib untreated samples but two (GIST1 and GIST7) were grouped together, while the majority of the treated samples (17 out of 19) were clustered in a separate group ( Figure 1A).
To validate the miRNA array data, we performed RT-qPCR analyses of four selected miRNAs in an extended cohort of 62 samples. We found significantly higher expression of miR-320b (p = 0.001, Mann-Whitney U-test) and lower expression of miR-483-3p (p < 0.001, Mann Whitney U-test) in imatinib-treated GISTs, as compared to untreated GISTs ( Figure 1B). The expression of miR-320a and miR-193a-3p did not differ significantly between imatinib-treated and untreated tumors ( Figure 1B).   To validate the miRNA array data, we performed RT-qPCR analyses of four selected miRNAs in an extended cohort of 62 samples. We found significantly higher expression of miR-320b (p = 0.001, Mann-Whitney U-test) and lower expression of miR-483-3p (p < 0.001, Mann Whitney U-test) in imatinib-treated GISTs, as compared to untreated GISTs ( Figure 1B). The expression of miR-320a and miR-193a-3p did not differ significantly between imatinib-treated and untreated tumors ( Figure 1B).

miR-320b and miR-483-3p Expression in GIST Cells upon Imatinib Treatment
To determine whether miR-320b and miR-483-3p can be regulated by imatinib, we quantified their expressions in GIST cell lines after 24 h of imatinib treatment. Consistent with the observations in clinical samples, we found significant upregulation of miR-320b and downregulation of miR-483-3p in GIST 882 and GIST T1 upon treatment with 1 µM imatinib ( Figure 2).

miR-320b and miR-483-3p Expression in GIST Cells upon Imatinib Treatment
To determine whether miR-320b and miR-483-3p can be regulated by imatinib, we quantified their expressions in GIST cell lines after 24 h of imatinib treatment. Consistent with the observations in clinical samples, we found significant upregulation of miR-320b and downregulation of miR-483-3p in GIST 882 and GIST T1 upon treatment with 1 μM imatinib ( Figure 2). . miRNA expression was normalized to U48. Fold change refers to the normalized miRNA expression values relative to their respective untreated cells. Differences between groups were calculated using paired t-test. * p < 0.05, ** p < 0.01.

Effects of miR-483-3p on OXPHOS Protein Expressions
Vitiello et al. and our previous study have shown upregulated OXPHOS proteins upon imatinib treatment [11,12]. Given that miR-483-3p is involved in the insulin pathway and in glucose metabolism [20], we further investigated whether miR-483-3p plays a role in the regulation of OXPHOS proteins ( Figure 3). In both GIST cell lines, we observed that inhibition of miR-483-3p, using miRNA inhibitor, increased protein expression of Complex II (SDHB), but not Complexes III (UQCRC2) or V (ATP5A) ( Figure 3A,B). Conversely, overexpression of miR-483-3p, using miRNA mimic, reduced Complex II (SDHB) protein levels in both cell lines, while no significant changes were observed for Complex III (UQCRC2) or V (ATP5A) ( Figure 3C,D). Notably, miR-483-3p overexpression or inhibition did not yield significant changes in SDHB mRNA levels ( Figure 3E).

Relationship between miR-483-3p and SDHB expression upon Imatinib Treatment
Given that imatinib decreased miR-483-3p expression, we evaluated the expression of the Complex II component SDHB in GIST cell lines with and without imatinib treatment. As shown in Figure 4A,B, we observed increased SDHB protein levels in both cell lines after imatinib treatment for 48 h and 96 h. To further establish the relationship between miR-483-3p and SDHB expression, we compared miR-483-3p and SDHB protein expression in GIST tumor samples. The results showed inverse correlations between miR-483-3p and SDHB expression for both untreated (r = −0.5801, p = 0.009) and imatinibtreated (r = −0.6270, p = 0.008) GISTs ( Figure 4C). and GIST 882 cell lines after imatinib treatment (1 µM, 24 h). miRNA expression was normalized to U48. Fold change refers to the normalized miRNA expression values relative to their respective untreated cells. Differences between groups were calculated using paired t-test. * p < 0.05, ** p < 0.01.

Effects of miR-483-3p on OXPHOS Protein Expressions
Vitiello et al. and our previous study have shown upregulated OXPHOS proteins upon imatinib treatment [11,12]. Given that miR-483-3p is involved in the insulin pathway and in glucose metabolism [20], we further investigated whether miR-483-3p plays a role in the regulation of OXPHOS proteins ( Figure 3). In both GIST cell lines, we observed that inhibition of miR-483-3p, using miRNA inhibitor, increased protein expression of Complex II (SDHB), but not Complexes III (UQCRC2) or V (ATP5A) ( Figure 3A,B). Conversely, overexpression of miR-483-3p, using miRNA mimic, reduced Complex II (SDHB) protein levels in both cell lines, while no significant changes were observed for Complex III (UQCRC2) or V (ATP5A) ( Figure 3C,D). Notably, miR-483-3p overexpression or inhibition did not yield significant changes in SDHB mRNA levels ( Figure 3E).

Relationship between miR-483-3p and SDHB Expression upon Imatinib Treatment
Given that imatinib decreased miR-483-3p expression, we evaluated the expression of the Complex II component SDHB in GIST cell lines with and without imatinib treatment. As shown in Figure 4A,B, we observed increased SDHB protein levels in both cell lines after imatinib treatment for 48 h and 96 h. To further establish the relationship between miR-483-3p and SDHB expression, we compared miR-483-3p and SDHB protein expression in GIST tumor samples. The results showed inverse correlations between miR-483-3p and SDHB expression for both untreated (r = −0.5801, p = 0.009) and imatinib-treated (r = −0.6270, p = 0.008) GISTs ( Figure 4C). Error bars refer to SD of two or three independent experiments. Differences between groups were calculated using paired t-test. * p < 0.05, ** p < 0.01, ns = not significant.  Error bars refer to SD of two or three independent experiments. Differences between groups were calculated using paired t-test. * p < 0.05, ** p < 0.01, ns = not significant.

Functional Role of miR-483-3p in Regulation of Cell Viability upon Imatinib Treatment
To address the functional role of miR-483-3p in GIST cells, we assessed the effect of miR-483-3p inhibition or overexpression on the expression of cleaved PARP (cPARP), which is a marker for apoptosis. No significant changes were found on cPARP level upon inhibition or overexpression of miR-483-3p ( Figure 5A,B), as well as cell viability in miR-483-3p overexpressing cells ( Figure 5C). However, overexpression of miR-483-3p rescued cell death in imatinib treated cells, as assessed by trypan blue staining ( Figure 5C). Notably, overexpression of miR-483-3p also reversed imatinib-induced SDHB expression (Figure 5D). Together, these results support the involvement of miR-483-3p in imatinib-induced SDHB expression and the important role of this regulation in protecting cell death induced by imatinib treatment.

Functional Role of miR-483-3p in Regulation of Cell Viability upon Imatinib Treatment
To address the functional role of miR-483-3p in GIST cells, we assessed the effect of miR-483-3p inhibition or overexpression on the expression of cleaved PARP (cPARP), which is a marker for apoptosis. No significant changes were found on cPARP level upon inhibition or overexpression of miR-483-3p ( Figure 5A,B), as well as cell viability in miR-483-3p overexpressing cells ( Figure 5C). However, overexpression of miR-483-3p rescued cell death in imatinib treated cells, as assessed by trypan blue staining ( Figure 5C). Notably, overexpression of miR-483-3p also reversed imatinib-induced SDHB expression ( Figure 5D). Together, these results support the involvement of miR-483-3p in imatinib-induced SDHB expression and the important role of this regulation in protecting cell death induced by imatinib treatment.

Discussion
Cancer cells can exhibit metabolic adaptation in response to targeted therapy [21]. Several pieces of evidence have revealed that mitochondrial OXPHOS drives resistance to chemotherapy or targeted drugs in leukemia and solid tumors [22]. Similarly, GIST cells upregulate OXPHOS upon imatinib treatment [11,12]. However, the mechanism underlying such metabolic reprogramming is not fully understood. Here we propose the involvement of miR-483-3p in OXPHOS regulation in response to imatinib. miR-483 is located in intron 7 of the insulin-like growth factor 2 gene (IGF2) in chromosomal region 11p15. Two mature miRNAs, miR-483-3p and miR-483-5p, are processed from the stem-loop structure of the precursor and both are known to play important roles in cancer development [23]. miR-483-3p is often aberrantly overexpressed in cancers such as adrenocortical carcinoma [24], pancreatic cancer [25], Wilms' tumor [26], hepatocellular carcinoma [27], colon cancer [26], and prostate cancer [28]. In contrast, downregulation of miR-483-3p has been found in some other cancer types, including nasopharyngeal cancer [29], breast [30], and gastric [31] cancers. These findings suggest that miR-483-3p may function as an oncogene or tumor suppressor, depending on the cellular context. In GIST, overexpression of miR-483-3p was found in CD117 positive/KITmutated tumors as compared to CD117 negative/KIT-unmutated tumors [32]. Wild-type GIST is a distinct subgroup of GIST that lacks KIT and PDGFRA mutations, which can be further classified as SDH-deficient or SDH-competent. SDH-deficient GISTs commonly harbor inactivating mutations in SDH subunit genes (SDHA, SDHB, SDHC or SDHD) or promoter hypermethylation of SDHC [33,34]. Increased expression of the insulin-like

Discussion
Cancer cells can exhibit metabolic adaptation in response to targeted therapy [21]. Several pieces of evidence have revealed that mitochondrial OXPHOS drives resistance to chemotherapy or targeted drugs in leukemia and solid tumors [22]. Similarly, GIST cells upregulate OXPHOS upon imatinib treatment [11,12]. However, the mechanism underlying such metabolic reprogramming is not fully understood. Here we propose the involvement of miR-483-3p in OXPHOS regulation in response to imatinib.
miR-483 is located in intron 7 of the insulin-like growth factor 2 gene (IGF2) in chromosomal region 11p15. Two mature miRNAs, miR-483-3p and miR-483-5p, are processed from the stem-loop structure of the precursor and both are known to play important roles in cancer development [23]. miR-483-3p is often aberrantly overexpressed in cancers such as adrenocortical carcinoma [24], pancreatic cancer [25], Wilms' tumor [26], hepatocellular carcinoma [27], colon cancer [26], and prostate cancer [28]. In contrast, downregulation of miR-483-3p has been found in some other cancer types, including nasopharyngeal cancer [29], breast [30], and gastric [31] cancers. These findings suggest that miR-483-3p may function as an oncogene or tumor suppressor, depending on the cellular context. In GIST, overexpression of miR-483-3p was found in CD117 positive/KIT-mutated tumors as compared to CD117 negative/KIT-unmutated tumors [32]. Wild-type GIST is a distinct subgroup of GIST that lacks KIT and PDGFRA mutations, which can be further classified as SDH-deficient or SDH-competent. SDH-deficient GISTs commonly harbor inactivating mutations in SDH subunit genes (SDHA, SDHB, SDHC or SDHD) or promoter hyper-methylation of SDHC [33,34]. Increased expression of the insulin-like growth factor 1 receptor (IGF1R) is commonly found in SDH-deficient GISTs but rarely in SDH-competent GISTs [35]. Of note, IGF2 is a ligand exerting its effects by binding to the IGF1R. Future studies are warranted to investigate the relationship between miR-483-3p, IGF2, and SDH.
Here we observed reduced expression of miR-483-3p upon imatinib treatment. Although it is unclear how imatinib regulates miR-483-3p expression, miR-483 transcription can be suppressed by glucose starvation and treatment with the glycolysis inhibitor 2deoxy-d-glucose [36]. Given that imatinib can inhibit aerobic glycolysis and decrease glucose uptake [37], it is tempting to speculate that the reduction of miR-483-3p is caused by imatinib-mediated inhibition of glycolytic activity. The mechanism has yet to be elucidated.
Importantly, we demonstrate that miR-483-3p regulates Complex II but not Complexes III and V, suggesting the specificity of miR-483-3p in the regulation of Complex II. However, it is still unclear whether the regulation is direct or indirect. Complex II or SDH is a hetero-tetrameric protein complex consisting of SDHA, SDHB, SDHC and SDHD. Loss of any component of the SDH complex can lead to destabilization of the SDH complex and rapid degradation of the SDHB protein. Thus, loss of SDHB expression is a reliable marker for SDH deficiency [38,39], which leads to the question of which SDH subunit is a target of miR-483-3p. The interaction between miR-483-3p and SDH subunits is unknown. Although SDHB is not predicted as a target of miR-483-3p, SDHC is predicted as a putative target of miR-483-3p by RNAhybrid v2.2. Further investigation is needed to determine which SDH subunit is the potential direct target of miR-483-3p. To date, two miRNAs, miR-147b [40] and miR-210 [13] have been reported to regulate SDHD. In our study, we did not find miR-210 to be differentially expressed between imatinib-treated and untreated GIST samples, however, miR-147b was not detected.
Our previous work has shown that imatinib could increase the expression of several OXPHOS proteins, including Complexes II, III and V [11]. Given that miR-483-3p only regulates Complex II, it is possible that other miRNAs or factors could regulate OXPHOS proteins. It is worth mentioning that miR-483-5p was also downregulated in imatinibtreated as compared to untreated GIST tumors in our microarray data, however the FDR was 28.8%, which is higher than the arbitrary cut-off (20%) that we set in this study. This miRNA is predicted to target multiple genes involved in OXPHOS, including Complex I (NDUFC1, NDUFA9, NDUFV3, NDUFA11, and NDUFA5), Complex III (UQCR11), Complex IV (COX19, COX6B1) and Complex V (ATP5S, and ATP5G1) by TargetScanHuman release 7.2 [41]. The other downregulated miRNA, miR-149, has been demonstrated to target poly(ADP-ribose) polymerase-2 that activates the sirtuin-1/peroxisome proliferatoractivated receptor γ coactivator-1α (PGC1α) pathway, leading to increased mitochondrial biogenesis [42]. The involvement of miR-149 in mitochondrial biogenesis may provide the explanation for the suppression of PGC1α-dependent mitochondrial biogenesis by imatinib treatment observed in our previous study [11].

Conclusions
In conclusion, our results reveal that specific miRNAs are responsive to imatinib treatment. We also show that miR-483-3p can regulate mitochondrial respiratory Complex II, suggesting that suppression of this miRNA could partly contribute to increased OXPHOS upon imatinib treatment.

Clinical Samples
A total of 64 snap-frozen tumors from 56 GIST patients were included in this study. Of these, 30 tumors were from 24 patients who had received neoadjuvant imatinib treatment, and 34 tumors were from 32 patients who had not been treated with imatinib prior to surgery (referred to as untreated tumors). For 36 of the samples (19 untreated and 17 imatinib-treated), SDHB protein expression levels were previously determined by Western blot analysis as published in our previous study [11].

GIST Cell Lines and Imatinib Treatment
Two imatinib-sensitive GIST cell lines were used, i.e., GIST 882 and GIST T1. GIST 882 was kindly provided by Dr. Jonathan Fletcher (Brigham and Women's Hospital, Boston, MA, USA). This cell line was cultured in RPMI 1640 medium containing 15% fetal bovine serum, and 2 mM L-glutamine at 37 • C with 5% CO 2 . GIST T1 was purchased from Cosmo Bio Co., Ltd. (Tokyo, Japan) and cultured in DMEM supplemented with 10% fetal bovine serum. The authenticity of these cell lines was previously confirmed by short tandem repeat (STR) profiling [11]. GIST cell lines were treated with 1 µM imatinib for the indicated time periods. This imatinib concentration was according to previous publication [11].

RNA Extraction and Quantification
Total RNA was extracted with mirVana miRNA Isolation Kit (Thermo Fisher Scientific, Waltham, MA, USA) and RNA concentrations were measured using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

miRNA Expression Profiling
Genome-wide miRNA expression profiles were determined for 19 imatinib-treated and 15 untreated tumors. Results for 17 of the imatinib-treated tumors have been previously reported [43]. All samples were analyzed using the human miRNA microarray system (Agilent, Santa Clara, CA, USA), which contains 903 human miRNAs (miRbase release 14). Array hybridizations and data analysis were according to the previously described methodology [44]. The data were normalized and centered to median using the Cluster 3.0 software (http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm, accessed on 2 November 2014). Normalized miRNAs with less than 20% of missing values were further analyzed using Significance Analysis of Microarrays (SAM) to identify differentially expressed miRNAs. Java TreeView software (http://jtreeview.sourceforge.net, accessed on 2 November 2014) was used to visualize the data from hierarchical clustering.

Reverse Transcription Quantitative PCR (RT-qPCR)
Four selected mature miRNAs, including miR-320a (ID 002277), miR-320b (ID 002844), miR-193a-3p (ID 002250) and miR-483-3p (ID 002339), were quantified using TaqMan-based RT-qPCR assays (Thermo Fisher Scientific, Waltham, MA, USA). To obtain cDNA, total RNA (20 ng) was reverse-transcribed using TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA). RT-qPCR was performed using the TaqMan Universal PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) in the ABI Prism 7900HT Real-time PCR System (Thermo Fisher Scientific, Waltham, MA, USA). RNU48 (ID 001006) was used as an endogenous control. For quantification of SDHB mRNA, cDNA was synthesized from 100 ng total RNA using High Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific) and RT-qPCR was performed using TaqMan pre-designed gene expression assays SDHB (Hs01042482_m1) and ACTB (Hs01060665_g1). All reactions were performed in triplicate, and relative expression levels were calculated with the delta C T method (2 −∆CT ). For imatinib treatment experiments, the 2 −∆CT values of treated cells were normalized to the untreated control cells.

In Vitro Transfection of GIST Cells and Cell Viability Analysis
mirVana miR-483-3p inhibitors and mimics, and their respective negative controls (mir-Vana miRNA mimic and inhibitor negative control#1), were purchased from Thermo Fisher Scientific (Waltham, MA, USA). A total of 3 × 10 6 GIST T1 and 882 cells were transfected with 50 pmol of miRNA mimics or inhibitors using Amaxa Nucleofector (Lonza, Basel, Switzerland), and program X01 and T20, respectively, according to the manufacturer's instructions. For Western blotting, cells were harvested after 72 h of transfection.
For cell viability analysis, cells at 24 h post-transfection were treated with or without 1 µM imatinib for another 24 h, followed by staining with 0.4% trypan blue stain (Thermo Fisher Scientific, Waltham, MA, USA). Viable cells were counted using the TC10 automatic cell counter (Bio-Rad).

Statistical Analysis
All statistical analyses were performed using Graphpad Prism 8.0 (Graphpad software, La Jolla, CA, USA). Independent Student's t-test or Mann-Whitney U-test was performed to compare the expression of miRNA and protein in different patient groups and cell lines. Paired t-test was used to compare two different conditions between the experimental groups. Correlation between miR-483-3p and SDHB protein levels was assessed by Spearman's Rank Order Correlation. All p-values were based on two-tailed tests, and a p-value < 0.05 was considered as statistically significant.  Data Availability Statement: The microarray data are publicly available at the NCBI Gene Expression Omnibus (GSE63159).