A Proteomic Approach Reveals That miR-423-5p Modulates Glucidic and Amino Acid Metabolism in Prostate Cancer Cells

Recently, we have demonstrated that miR-423-5p modulates the growth and metastases of prostate cancer (PCa) cells both in vitro and in vivo. Here, we have studied the effects of miR-423-5p on the proteomic profile in order to identify its intracellular targets and the affected pathways. Applying a quantitative proteomic approach, we analyzed the effects on the protein expression profile of miR-423-5p-transduced PCa cells. Moreover, a computational analysis of predicted targets of miR-423-5p was carried out by using several target prediction tools. Proteomic analysis showed that 63 proteins were differentially expressed in miR-423-5-p-transfected LNCaP cells if compared to controls. Pathway enrichment analysis revealed that stable overexpression of miR-423-5p in LNCaP PCa cells induced inhibition of glycolysis and the metabolism of several amino acids and a parallel downregulation of proteins involved in transcription and hypoxia, the immune response through Th17-derived cytokines, inflammation via amphorin signaling, and ion transport. Moreover, upregulated proteins were related to the S phase of cell cycle, chromatin modifications, apoptosis, blood coagulation, and calcium transport. We identified seven proteins commonly represented in miR-423-5p targets and differentially expressed proteins (DEPs) and analyzed their expression and influence on the survival of PCa patients from publicly accessible datasets. Overall, our findings suggest that miR-423-5p induces alterations in glucose and amino acid metabolism in PCa cells paralleled by modulation of several tumor-associated processes.


Introduction
Prostate cancer (PCa) is one the most common cancers worldwide in males [1,2]. Despite screening based on the prostate-specific antigen (PSA), the diagnosis and treatment of PCa are frequently delayed because specific symptoms and biomarkers are lacking in the early stage [3]. Consequently, it is crucial to understand the molecular bases of PCa in order to develop novel strategies for both diagnosis and treatment.
MicroRNAs (miRNAs) are small non-coding RNAs highly conserved during evolution and key regulators of numerous genes involved in physiological processes and diseases

Proteomic Profiling of LNCaP Cells Overexpressing miR-423-5p
To investigate the effect of miR-423-5p on its intracellular targets, we performed a quantitative SWATH proteomic analysis of PCa LNCaP cells, both those expressing miR-423-5p mimic and the control transduced with the empty vector.
Differential expression analysis showed that 63 proteins were significantly differentially expressed in the LNCaP cells overexpressing miR-423-5p compared to control, of which 19 proteins were downregulated and 44 upregulated ( Figure 1A-C). A complete list of differentially expressed proteins (DEPs) is available in Table S1.
Pathway enrichment analysis of all up-and down-regulated proteins using the online software MetaCore (Thomson Reuters) identified several pathways, (Tables 1 and 2) modulated following miR-423-5p mimic expression. We found that most of the proteins, whose expression was downregulated, are mainly involved in cellular metabolism, such as IDO1 (Indoleamine 2,3-dioxygenase-1) effect on T cell metabolism, glycolysis and gluconeogenesis, and histamine, glutathione, tyrosine, catecholamine, and ascorbate metabolisms ( Figure 2). In fact, we observed a decreased expression of G6PD (Glucose-6-phosphate dehydrogenase), HX2 (Hexokinase 2), ALDH2 (Aldehyde dehydrogenase 2 family), and LDHA (Lactate dehydrogenase A) (Table 1, Figure 2). Other downregulated proteins are involved in immune response, related to Myeloid-derived suppressor cells and M2 macrophages in cancer, MIF-mediated glucocorticoid regulation, and MIF-JAB signaling. Proteins involved in transcription and hypoxia, such as HIF-1 targets' pathways and transcriptional assembly, were also significantly downregulated (Table 1, Figure 2). Conversely, proteins whose expression was upregulated were mostly associated with cell metabolism, such as galactose metabolism, neolacto-series GSL (Glycosphingolipid) metabolism, peptidase activity, and the lysosomal pathway but also with clathrin vesicles formation, G-protein signaling, and the HP1 role in transcriptional silencing (Table 2, Figure 2). Furthermore, when both downregulated and upregulated proteins were analyzed as process networks, those involved in the immune response through Th17-derived cytokines, inflammation via amphorin signaling, and ion transport were downregulated. Conversely, those related to the cell cycle S phase, chromatin modifications, apoptosis, blood coagulation, and calcium transport were upregulated (Tables 3 and 4, Figure 2). Although several pathways and processes were identified in response to miR-423-5p overexpression, cellular metabolism, immune response, and transcription appear to be the ones that are mainly affected as indicated by the pathway enrichment analysis of proteomic data (p-value ≤ 0.05). Proteinprotein interaction (PPI) networks of significant downregulated and upregulated proteins in LNCaP PCa cells expressing miR-423-5p mimic compared to control (empty vector) made by STRING analysis of proteomic data confirmed Metacore enrichment analysis [17]. As shown in Figure 3A, the majority of the proteins downregulated by miR-423-5p are involved in the pentose phosphate pathway, glycolysis, and carbon metabolism in cancer (G6PH and LDHA), as well as the biosynthesis of amino acids (SHMT2 and ASNS). Conversely, upregulated proteins in LNCaP PCa cells overexpressing miR-423-5p are involved in catabolic mechanisms related to amino acids metabolism and lysosome function (TPP1, DPP7 and, CTSH) ( Figure 3B).  Pathway enrichment analysis of all up-and down-regulated proteins using the online software MetaCore (Thomson Reuters) identified several pathways, (Tables 1 and 2) modulated following miR-423-5p mimic expression. We found that most of the     Quantitative SWATH-mass spectrometry data analysis performed with the SCIEX OneOmics cloud processing platform (n = 6 biological replicates).

Prediction of Functional Targets by Integrative Analysis of miRNA Predicted Targets and Protein Expression Data
Previous studies show that miR-423-5p is involved in the downregulation of lncRNA MALAT1 and promotes cell proliferation, migration, and invasion in PCa cells [16]. Moreover, the overexpression of this miRNA in PCa cells is associated with increased survival of PCa patients and reduced metastasis formation in mouse models [16]. To evaluate the targets involved in the role played by miR-423-5p in PCa, a target prediction analysis was performed. By using four target prediction tools, a total of 6695 miRNA target genes (miTGs) were identified of which 77 (1.2%) were represented in all databases (Table 5, Figure 4A) [18][19][20][21]. From the output list, miRNA targets predicted by two or more prediction tools were considered for further analysis. Then, an integrated analysis between predicted miRNA targets and DEPs identified in this study was performed. As shown in the Venn diagram, seven targets were found commonly represented between miR-423-5p target genes and DEPs in LNCaP cells overexpressing miR-423-5p ( Figure 4B). A detailed summary of the seven genes/proteins commonly represented in the prediction analysis results and proteomic data is given in Table 6. Table 3. MetaCore enrichment analysis of proteomics data in LNCaP PCa cells overexpressing miR-423-5p. Significant process networks modulated by downregulated proteins in LNCaP PCa cells expressing miR-423-5p mimic compared to control (empty vector) (p-value ≤ 0.05). Quantitative SWATH-mass spectrometry data analysis performed with the SCIEX OneOmics cloud processing platform (n = 6 biological replicates).
To exploit the expression of protein targets of miR-423-5p in the LNCaP PCa cell line, we performed a protein expression analysis using the Expression Atlas for the previously identified seven targets in the LNCaP cell line [28]. All protein targets showed a high expression in the LNCaP cell line (High: >1.000 Parts per billion) as reported in Table 7. Moreover, in order to extend the analysis to data from patients, a gene expression analysis was performed using GEPIA2 based on The Cancer Genome Atlas-Prostate Adenocarcinoma (TCGA-PRAD) and the Genotype-Tissue Expression (GTEx) project data [29].
Finally, the analysis of the expression of IMDH1 (gene IMDPH1) and SPEE (gene SRM), which were downregulated in LNCaP cells overexpressing miR-423-5p from the proteomic analysis, showed an opposite trend at the transcriptomic level using the TCGA-PRAD and GTEx datasets, suggesting that the lowered expression of these target proteins is mediated by the overexpression of miR-423-5p in PCa cells (Table 7). This evidence is supported by the real-time expression profiling (RT-qPCR) of miR-423-5p in several human PCa cell lines ( Figure S1). In fact, we observed a lower expression of miR-423-5p in LNCaP cells that was at the limit of detection if compared to DU-145 and PC-3 cells. Moreover, to determine the prognostic significance of miR-423-5p, an overall survival analysis using the ENCORI Pan-Cancer platform was performed [30]. Two clinical subgroups that consisted of patients with high and low levels of miR-423-5p were obtained but no significant correlations between the expression levels of miR-423-5p and prognosis in PRAD groups was found (log-rank p = 0.96) ( Figure S2). [29].
Comparison of mRNA expression in prostate adenocarcinoma (Tumour, T = 492) versus normal prostate tissue (Normal, N = 152) showed a significant difference for microtubule-associated protein 1B (MAP1B). Conversely, the other evaluated and validated targets CBX5/HP1A, NIT1, STK39, IMPDH1, ADGRG6, and SRM did not show any significant differences ( Figure 5). Finally, the analysis of the expression of IMDH1 (gene IMDPH1) and SPEE (gene SRM), which were downregulated in LNCaP cells overexpressing miR-423-5p from the proteomic analysis, showed an opposite trend at the transcriptomic level using the TCGA-PRAD and GTEx datasets, suggesting that the lowered expression of these target proteins is mediated by the overexpression of miR-423-5p in PCa cells (Table 7). This evidence is

Overall Survival Analysis Based on the Expression of miR-423-5p Target Genes
To evaluate the clinical significance of the modulation of the miRNA targets in PCa, we conducted a Kaplan-Meier analysis of the overall survival using the GEPIA2 online tool [29]. Cohorts of 246 patients were assigned to low and high expression levels of mRNA for the targets found differentially expressed and no significant differences were detected in the overall survival for these targets ( Figure 6).

Overall Survival Analysis Based on the Expression of miR-423-5p Target Genes
To evaluate the clinical significance of the modulation of the miRNA targets in PCa, we conducted a Kaplan-Meier analysis of the overall survival using the GEPIA2 online tool [29]. Cohorts of 246 patients were assigned to low and high expression levels of mRNA for the targets found differentially expressed and no significant differences were detected in the overall survival for these targets ( Figure 6).

Discussion
In the present manuscript, we investigated downstream pathways that are affected by miR-423-5p expression using mass-spectrometry analysis. We found that cellular metabolism, amino acid metabolism, and transcription are the main affected pathways. IDO1 (Indoleamine 2,3-dioxygenase-1) effect on T cell metabolism, glycolysis and gluconeogenesis, and histamine, glutathione, tyrosine, catecholamine, and ascorbate metabolisms, were the metabolic pathways affected by downregulated proteins. In these pathways, a decreased expression of G6PD, HX2, ALDH2, and LDHA was found. Other downregulated proteins are associated with immune response, including those related to Myeloid-derived suppressor cells and M2 macrophages in cancer, MIF-mediated Figure 6. Kaplan-Meier plots for mRNA expression derived from the online GEPIA2 tool for sets of high (red) and low (blue) expression level cohorts. 'N' represents the size of the patient cohorts involved in the study with high and low expression of mir-423-5p targets.

Discussion
In the present manuscript, we investigated downstream pathways that are affected by miR-423-5p expression using mass-spectrometry analysis. We found that cellular metabolism, amino acid metabolism, and transcription are the main affected pathways. IDO1 (Indoleamine 2,3-dioxygenase-1) effect on T cell metabolism, glycolysis and gluconeogenesis, and histamine, glutathione, tyrosine, catecholamine, and ascorbate metabolisms, were the metabolic pathways affected by downregulated proteins. In these pathways, a decreased expression of G6PD, HX2, ALDH2, and LDHA was found. Other downregulated proteins are associated with immune response, including those related to Myeloidderived suppressor cells and M2 macrophages in cancer, MIF-mediated glucocorticoid regulation, and MIF-JAB signaling. Upregulated proteins were mostly associated with cell metabolism, such as galactose metabolism, neolacto-series GSL (Glycosphingolipid) metabolism, peptidase activity, and the lysosomal pathway but also with clathrin vesicles formation, G-protein signaling, and the HP1 role in transcriptional silencing. miR-423-5p's role in the regulation of metabolic pathways is still under-investigated. We provide for the first time, at least to our knowledge, that miR-423-5p regulates cellular metabolism in cancer cells. However, it was reported that miR-423-5p induces inhibition of neoglucogenesis, hyperglycemia, and liver deposition in an animal model of diabetes [31]. These data were confirmed in a model of gestational diabetes in which the repression of miR-423-5p ameliorated hepatic gluconeogenesis and insulin resistance [32]. G6PD, another target of miR-423-5p involved in the pentose phosphate pathway (PPP) and glucose metabolisms, was also downregulated in a model of preeclampsia and gestational hypertension [33]. In cancers, PPP has been upregulated if compared to normal counterparts becoming a possible target for therapy [34][35][36][37][38]. The involvement of G6PD in drug resistance in patients affected by acute myeloid leukemia was reported by Gregory et al. and, more recently, similar results were found in patients with multiple myeloma [39,40]. Interestingly, we have previously reported that G6PDH inhibition blocks proliferation and metastases of head and neck squamous cancer cells in vivo, and this effect was paralleled by enhanced autophagy [41,42]. These effects are in agreement with our findings about the involvement of miR-423-5p in the autophagic processes in hepatocellular cancer cell lines [43]. Conversely, no previous reports exist on the role of miR-423-5p in the regulation of amino acid metabolism. The HIF-1 pathway was also downregulated by miR-423-5p. This result is again in agreement with our previously reported findings on the anti-angiogenic effects of miR-423-5p in PCa cell lines [16]. In fact, in PCa cell lines, we have found a strong downregulation of both vascular endothelial growth factor (VEGF) B and C that are, in turn, induced by the HIF-1 pathway. It has also to be considered that miR-423-5p was reported to have pro-apoptotic effects in both cardiocytes and renal proximal tubular epithelial cells during hypoxia/reoxygenation, confirming its role in the regulation of cellular hypoxic homeostasis [44,45]. Although miR-423-5p involvement in the transcriptional silencing of HP1A is highly likely due to its post-transcriptional regulatory roles, its role in clathrin vesicle formation, and G-protein signaling, is not known, thus, requiring further investigations. We have also found an increase in the activity of lysosome formation. In this light, we have previously reported that the overexpression of miR-423-5p in hepatocellular cancer cells induced an increased activation of the autophagic process with the involvement of lysosome and the formation of abundant autophagic vacuoles [43]. Among the protein targets of miR-423-5p, we have found SPEE, which is involved in both amino acid metabolism and cancer [46]. In fact, spermidine has a role in the regulation of cancer cell proliferation and is involved in anti-cancer immune surveillance [46]. Other protein targets were MAP1B, CBX5, NIT1, STK39, IMDH1, and AGRG6. Interestingly, five out of the seven targets, namely, Micro-tubule-associated protein 1B (MAP1B), Chromobox protein homolog 5 (CBX5/HP1A), Inosine-5 -monophosphate dehydrogenase 1 (IMPDH1), Adhesion G-protein coupled receptor G6 (ADGRG6), and Spermidine synthase (SRM), are reported at different tissue levels, including cervix, kidney and bone marrow, at different experimental conditions. Integrating the results from the proteomic analysis in LNCaP cells and the gene expression profiling for the miR-423-5p targets in PRAD, we found that IMDPH1 and SRM showed an opposite trend at transcriptomic level, indicating that miR-423-5p could interact on these miTGs in PCa. Since we previously reported that the overexpression of miR-423-5p was able to inhibit MALAT1-mediated proliferation, migration, and invasion of PCa cells, we evaluated the clinical significance of miR-423-5p in the TCGA-PRAD dataset using the ENCORI Pan-Cancer analysis platform and no significant correlations between the expression level of miR-423-5p and prognosis in PRAD groups was detected [16]. Furthermore, the analysis of miTGs with GEPIA2 based on The Cancer Genome Atlas-Prostate Adenocarcinoma (TCGA-PRAD) and the Genotype-Tissue Expression (GTEx) project data demonstrated only for MAP1B a statistically significant higher expression in tumors if compared to normal counterpart in PCa patients. These results were likely due to the limited size of the samples analyzed and included in the publicly available dataset. There are no previous reports on MAP1B in PCa, but it was demonstrated to be associated with several processes in other cancers. MAPs are proteins that participate in the organization of cell cytoskeletons taking part in the microtubule assembly that is essential for the maintenance of cellular structure during cell division [47,48]. MAPs have been also correlated to microtubule dynamics required for cell motility and metastatization [49]. MAP1B was expressed to different extents in multiple cancers with a higher degree in central nervous system tumors and its levels were found to predict shorter survival and higher grade in urothelial cancers [50].

Liquid Chromatography with Tandem Mass Spectrometry (LC-MS/MS) Analysis
Samples (~50 µg protein) were reduced/alkylated and digested as described previously [51]. Then, samples were dried and resuspended in 5% (v/v) acetonitrile + 0.1% (v/v) formic acid and transferred to an HPLC vial for MS analysis in both SWATH (Sequential Window Acquisition of all Theoretical Mass Spectra) and IDA (information-dependent acquisition) modes. Each sample was analyzed on a SCIEX TripleTof 6600 mass spectrometer coupled in line with an Eksigent ekspert nano LC 425 system running in microflow as described previously with minor modifications [52]. For IDA (Information-Dependent Analysis) to generate a spectral library, 8 µL of pooled sample in triplicate were injected by autosampler (Eksigent nanoLC 425 LC system, Dublin, OH, USA) in microflow at 5 µL/min directly onto a YMC Triart-C18 column (15 cm, 3 µm, 300 µm i.d.) using gradient elution (2-40% Mobile phase B, followed by washing at 80% B and re-equilibration) over 87 min. For SWATH/DIA (Data-Independent Analysis), 3 µL was injected on the same gradient elution profile over 57 min. Mobile phases consisted of A: 0.1% formic acid; B: acetonitrile containing 0.1% (v/v) formic acid. IDA analysis was carried out in positive ion mode with a 250 ms survey scan, m/z range 400-1250; Top 30 peaks selected for fragmentation, accumulation time 50 ms per experiment, cycle time 1.8 s. SWATH analysis used 100 variable windows, 25 ms per window, 100-1500 m/z using the SCIEX Duospray source with a 50 µm electrode at 5500 V.

Library Generation, Spectral Alignment, and Fold Change Analysis
Spectral libraries were constructed from three IDA runs of pooled samples spiked with HRM-kit retention time peptides (iRT, Biognosys AG, Schlieren, Switzerland), searched using ProteinPilot 5.0 (SCIEX, Macclesfield, UK) against the Swissprot human database (January 2016). The resulting ion library file was aligned against the iRT peptides, and the SWATH data was obtained using the OneOmics cloud processing platform (SCIEX, Warrington, UK) with the following parameters: six peptides per protein, six transitions per peptide, XIC width set at 75 ppm, 5 min extraction window. Analysis of six biological replicates per group was carried out in Protein Expression Workflow within OneOmics to identify significantly differentially expressed proteins, and a 75% confidence limit and a minimum of two peptides per protein were used as criteria for assessing fold change. Differentially expressed proteins (DEPs) were defined as downregulated if fold change (FC) ≤ −1.5. and upregulated if fold change (FC) ≥ 1.5.

Bioinformatic Analysis
The pathway enrichment analysis on proteomics data was performed using META-CORE (version 6.34.69200, Clarivate Analytics, London, UK), available online: https: //portal.genego.com (accessed on 11 April 2018). Protein-protein interaction (PPI) network analysis was performed with the search tool for retrieval of interacting genes/proteins database (STRING, version V11.0), available online: http://stringdb.org (accessed on 19 October 2022). Medium confidence was applied for the analysis and the Markov Cluster Algorithm (MCA) performed to extract clusters of densely connected nodes from biological networks.
Overall Survival plots were obtained using the "Survival Analysis" module of GEPIA2 with default parameters including a 95% confidence interval, "Median" Group cut off and Hazard Ratio (calculated based on the Cox PH model) using both the TCGA-PRAD and GTEx prostate datasets.

Conclusions
In conclusion, our data indicate that miR-423-5p could induce a metabolic switch inhibiting essential energetic processes such as glycolysis and interfering with the biosynthetic amino acidic pathways in PCa cells. These effects are paralleled by the interference with angiogenetic processes as previously reported by us in PCa cells and in agreement with our previous reports on the autophagic pathway activation mediated by miR-423-5p in other cancer models. Further in vitro and in vivo experiments are needed in order to validate the identified miR-423-5-p targets. These data can open a scenario of intervention based on targeting of the newly identified miR-423-5p interactors.