Epigenetic and post-transcriptional regulation of somatostatin receptor subtype 5 (SST5) in pituitary and pancreatic neuroendocrine tumors

Epigenetic and post-transcriptional regulation of somatostatin receptor subtype 5 (SST 5 ) in pituitary and pancreatic neuroendocrine tumors Sergio Pedraza-Arevalo 1,2,3,4 , Alejandro Ib a~ nez-Costa 1,2,3,4 , Ricardo Bl azquez-Encinas 1,2,3,4 , Miguel R. Branco 5 ,MariC.V  azquez-Borrego 1,2,3,4 , Aura D. Herrera-Mart ınez 1,6 , Eva Venegas-Moreno 7 , Raquel Serrano-Blanch 1,8 ,  Alvaro Arjona-S anchez 1,9 , Mar ıa A. G alvez-Moreno 1,6 , Marta Korbonits 10 , Alfonso Soto-Moreno 7 , Manuel D. Gahete 1,2,3,4 , Marika Charalambous 11 ,Ra  ul M. Luque 1,2,3,4 and Justo P. Casta~ no 1,2,3,4 1 Maimonides Institute for Biomedical Research of Cordoba (IMIBIC), C ordoba, Spain 2 Department of Cell Biology, Physiology, and Immunology, University of C ordoba, C ordoba, Spain 3 CIBER Fisiopatolog ıa de la Obesidad y Nutrici on (CIBERobn), C ordoba, Spain 4 Reina Sofia University Hospital, C ordoba, Spain 5 Blizard Institute, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK 6 Endocrinology and Nutrition Service, Reina Sofia University Hospital, C ordoba, Spain 7 Metabolism and Nutrition Unit, Hospital Universitario Virgen del Roc  ıo, Instituto de Biomedicina de Sevilla (IBIS), Sevilla, Spain 8 Medical Oncology Service, Reina Sofia University Hospital, C ordoba, Spain 9 Surgery Service, Reina Sofia University Hospital, C ordoba, Spain 10 Centre for Endocrinology, William Harvey Research Institute, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK 11 Developmental Epigenetics group, Department of Medical and Molecular Genetics, King’s College of London, London, UK Keywords epigenetics; natural antisense transcript; neuroendocrine tumors; pancreas; pituitary; SST5 Correspondence J. P. Casta~ no, R. M. Luque and A. Ib a~ nez- Costa, Edificio IMIBIC, Avenida Men endez Pidal s/n, 14004, C ordoba, Spain Tel: (+34) 957 21 37 43 (J. P. C.); (+34) 957 21 37 40 (R. M. L.); (+34) 957 21 37 38 (A. I.-C.) E-mails: [email protected] (J. P. C.); [email protected] (R. M. L.); [email protected] (A. I.-C.) Alejandro Ib a~ nez-Costa, Ra ul M. Luque and Justo P. Casta~ no contributed equally to this article. (Received 12 September 2020, revised 19 August 2021, accepted 6 September 2021, available online 26 October 2021) doi:10.1002/1878-0261.13107 Somatostatin receptor subtype 5 (SST 5 ) is an emerging biomarker and actionable target in pituitary (PitNETs) and pancreatic (PanNETs) neuroendocrine tumors. Transcriptional and epigenetic regulation of SSTR5 gene expression and mRNA biogenesis is poorly understood. Recently, an overlapping natural antisense transcript, SSTR5-AS1, potentially regulating SSTR5 expression, was identified. We aimed to elucidate whether epigenetic processes contribute to the regulation of SSTR5 expression in PitNETs (somatotropinomas) and PanNETs. We analyzed the SSTR5/SSTR5-AS1 human locus in silico to identify CpG islands. SSTR5 and SSTR5-AS1 expression was assessed by quantitative real-time PCR (qPCR) in 27 somatotropinomas, 11 normal pituitaries (NPs), and 15 PanNETs/paired adjacent (control) samples. We evaluated methylation grade in four CpG islands in the SSTR5/SSTR5-AS1 genes. Results revealed that SSTR5 and SSTR5- AS1 were directly correlated in NP, somatotropinoma, and PanNET samples. Interestingly, selected CpG islands were differentially methylated in somatotropinomas compared with NPs. In PanNETs cell lines, SSTR5-AS1 silencing downregulated SSTR5 expression, altered aggressiveness features, and influenced pasireotide response. These results provide evidence that SSTR5 expression in PitNETs and PanNETs can be epigenetically regulated by the SSTR5-AS1 antisense transcript and, indirectly, by DNA methylation, which may thereby impact tumor behavior and treatment response. Abbreviations FBS, fetal bovine serum; NAT, natural antisense transcript; NET, neuroendocrine tumor; NP, normal pituitary; NTAT, nontumor adjacent tissue; PanNET, pancreatic NET; PitNET, pituitary NET; SSA, somatostatin analogue. 764 Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 1. Introduction Neuroendocrine tumors (NETs) comprise a heterogeneous group of neoplasms, with rising incidence over the last decades [1–3]. These tumors arise from cells of (neuro)endocrine origin, which share common features like the synthesis, storage, and secretion of hormones and neurotransmitters. NETs can be widely distributed throughout the body, although they are more abundant in the gastrointestinal and respiratory tracts [1– 3]. Specifically, pancreatic NETs (PanNETs), which display one of the highest increases in incidence within the different types of NETs in the last 10 years [4], are associated with the endocrine compartment of the pancreas. In fact, PanNETs have been classically thought to be derived from hormone-producing cells of the pancreatic Langerhans’ islets [5], although recent evidence has arisen challenging this concept, and it is presently under debate whether NETs can in fact be originated from a common cell progenitor from the pancreas [6]. Genetic alterations contributing to Pan- NETs tumorigenesis include frequent mutations in MEN1,ATRX, or DAXX genes [7]. Additionally, tumors derived from the anterior pituitary have been classically termed adenomas due to their nonmetastatic behavior [8]. However, based on their potential aggressiveness and associated morbimortality, the International Pituitary Pathology Club recently proposed to reclassify this pathology and to name them as pituitary neuroendocrine tumors or Pit- NETs [9], although some controversies have arisen for this nomenclature [10,11]. Autopsy and imaging studies reveal that PitNETs are the most common intracranial neoplasms (prevalence 10–22%) [12]. PitNETs are primarily classified according to their size and accompanying hormonal hypersecretion [8]. Among them, somatotropinomas arise from somatotropes and oversecrete growth hormone (GH), causing gigantism (in children/adolescents) or acromegaly, characterized by extremity enlargement, facial and skeletal changes, and metabolic, gastrointestinal, cardiovascular, and respiratory complications [13,14]. A common feature shared by most NETs is the key role played by somatostatin and its receptors (SST 1 - SST 5 ) in their pathophysiological regulation and medical treatment, which is particularly relevant in Pit- NETs and PanNETs [13–17]. Indeed, somatostatin represents the main inhibitory signal for normal somatotropes and endocrine pancreatic cells, where it decreases hormone secretion [16,18,19]. Importantly, somatostatin also acts on tumor cells inhibiting hormone hypersecretion and cell proliferation, as reported in different tumor types including somatotropinomas, PanNETs, and thyrotropinomas, which abundantly express SSTs [13–17]. In general, SST 2 is the most expressed receptor in tumors, followed by SST 5 , with high tumor specificity [20]. Of note, the truncated SST 5 splicing variant, SST 5 TMD4, has also been found to be notably expressed in several endocrinerelated tumors, particularly PitNETs and PanNETs [21,22]. Therein, SST 5 TMD4 has been associated with tumorigenesis and malignancy features, likely by playing an inhibitory role over SST 2 and canonical, fulllength SST 5 [23–25]. Elucidating the regulation and interplay of SST 2 and SST 5 is particularly important given their key role in the NETs response to treatment with synthetic somatostatin analogues (SSAs) such as octreotide, lanreotide, or pasireotide [2]. First-generation SSAs (octreotide and lanreotide) preferentially target SST 2 , with less affinity to SST 5 — and, octreotide, SST 3 —and negligible binding for the other SSTs. These drugs have been widely used in the treatment of GH- and TSH-secreting PitNETs and also in PanNETs, to reduce hormonal secretion, control tumor volume, and improve patient symptoms [26–28]. Nonetheless, a substantial proportion of patients are or become resistant to these treatments [29,30]. Consequently, a second generation of SSAs with multireceptor binding affinity was developed, based on the idea that simultaneous targeting of several SST, like natural somatostatin, could improve effectiveness in unresponsive patients. From this group, the most widely used compound is pasireotide, showing high affinity to SST 5 , SST 2 , SST 3, and SST 1 [31,32]. However, SSAs actions do not only depend on their differential binding to specific SSTs. Actually, in somatotropinomas, although the complete set of factors defining SSA responsiveness is not yet fully defined, various specific tumor features and molecular markers have been shown to relevantly influence tumor response to SSAs, including granulation pattern, AIP and GNAS mutations, b-arrestin, filamin A, and E-cadherin expression, as well as, interestingly, SSTR2/SSTR5 expression balance and SST 5 TMD4 presence [13,16,33,34]. Thus, it is important to understand the mechanisms governing the expression of the SSTR5 gene and its resultant receptor variants (SST 5 , SST 5 TMD4, SST 5 TMD5), for they may impact NETs response to SSAs. Gene expression is known to be regulated by multiple factors, among which extrinsic factors, such as epigenetic mechanisms, have gained great attention in recent years. A prime epigenetic modification is DNA methylation, which is based on the addition of a 765Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies S. Pedraza-Arevalo et al.SST 5 regulation in neuroendocrine tumors methyl group to a cytosine preceding a guanine (CpG). CpG residues are enriched at CpG islands, regions of the genome frequently associated with promoter function. Likewise, noncoding RNAs may act as modular epigenetic regulators [35]. A particular type of noncoding RNAs comprise natural antisense transcripts (NATs) [36], that is, transcripts derived from the opposite strand to a protein-coding or sense gene, which can regulate the transcription of their corresponding sense genes. NATs importance is rising as sequencing technologies improve, and recent studies are deciphering NATs role in different diseases, including PitNETs [37], where they play distinct roles, like AFAP1-AS1, which influences tumor growth, or C5orf66-AS1, related to invasiveness. Recently, a NAT for SSTR5 was reported to be expressed in laryngeal squamous cell carcinoma, where it may act as tumor suppressor [38]. Nevertheless, its role in PitNETs and PanNETs has not been explored yet. Consequently, in this study we aimed to widen our still limited knowledge of the epigenetic mechanisms underlying the regulation of SSTR5 expression in NETs, specifically somatotropinomas and PanNETs, and to explore the functional and pathological implications of those epigenetic underpinnings in tumor behavior to better understand the role of this receptor. 2. Material and methods 2.1. Patients and samples This study was carried out within a project approved by the Research Ethics Committee of Co ordoba (Comit ede  Etica de la Investigaci on de C ordoba) and was conducted in accordance with ethical standards of the Helsinki Declaration of the World Medical Association. Written informed consent was obtained from each patient. Pituitary samples were collected during transsphenoidal surgery from 27 acromegaly patients and 11 normal pituitaries (NPs) by autopsy from donors and were stored frozen. Formalin-fixed paraffin-embedded samples (FFPE, n=15) were obtained from primary PanNETs; nontumor adjacent tissue, used as control, was extracted from the same piece and both tissues were separated by expert pathologists (patient features summarized in Table S1). 2.2. Cell culture and treatment Functional assays were performed in PanNET model cell lines BON-1 and QGP-1 [39–42], using passages lower than 25 in all cases. BON-1 cells were kindly provided by Dr. M.C. Zatelli and were cultured in DMEM-F12 (Life Technologies, Barcelona, Spain), whereas QGP-1 cells were kindly provided by Dr. K. € Oberg and were cultured in RPMI-1640 (Life Technologies), both supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, Madrid, Spain) and 0.2% antibiotic (Gentamicin/Amphotericin B; Life Technologies). Cell lines were grown at 37 °C, in a humidified atmosphere with 5.0% CO 2 and were verified for mycoplasma contamination by PCR with specific mycoplasma primers. To ensure the identity of the cells, we could not employ typical STR tests, as they are not available for these cell lines. Therefore, we use a different strategy, by measuring an ample set of genes typically expressed by the cell lines as previously reported [39,40], including SSTs, and secretory products (e.g., chromogranin, serotonin, or somatostatin). In addition, we have tested cell responses and behaviors after classic treatments, which closely resembled those described by original studies [41,42]. Pasireotide was provided by Novartis and administered at 100 nM, dissolved in sterile water, as previously reported [31,43], and 5-azacytidine (Sigma-Aldrich) was administered at different doses, based on the literature [44], also dissolved in sterile water. 2.3. Silencing of SSTR5-AS1 and SSTR5 expression BON-1 and QGP-1 cells were transfected with a specific shRNA targeting SSTR5-AS1, previously validated in our laboratory (Origene, Rockville, MD, USA), and selected with puromycin. On the other hand, SSTR5 was transiently silenced with a specific siRNA (Thermo Fisher, Waltham, MA, USA). Specifically, cells were seeded in 6-well culture plates and transfected with 1 lg of the small RNA, using Lipofectamine 2000 and Lipofectamine RNAiMAX Transfection Reagents (Thermo Fisher) for the shRNA and siRNA, respectively, during 6 h. Scramble shRNA/siRNA served as control. 2.4. DNA and RNA isolation and retrotranscription Total RNA from cell lines was isolated using TRIzol Reagent (Sigma-Aldrich) treated with DNase (Promega, Barcelona, Spain). In FFPE samples, RNA was isolated RNeasy FFPE Kit (Qiagen, Limburg, Netherlands). Particularly genomic DNA and RNA from fresh pituitary samples were extracted using AllPrep DNA/RNA/Protein Kit (Qiagen). Nucleic acid amount and quality was determined using NanoDrop2000 spectrophotometer (Thermo Fisher) and reversely 766 Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies SST 5 regulation in neuroendocrine tumors S. Pedraza-Arevalo et al. transcribed using random hexamer primers with the First Strand Synthesis Kit (Thermo Fisher). 2.5. Quantitative real-time PCR (qPCR) qPCRs were performed using Mx3000p system with the Brilliant III SYBR Green Master Mix (Stratagene, La Jolla, CA, USA) with specific primers (Table S2a) [45]. Results were validated as previously reported [46], adjusting gene expression with a normalization factor, calculated from values of ACTB,GAPDH,HPRT1, and/or RNA18S1 control genes. 2.6. Methylation assay DNA methylation of CpG islands overlapping SSTR5 and SSTR5-AS1 was evaluated in the PitNETs and normal pituitary cohort, as well as BON-1 and QGP-1 cell lines. One µg genomic DNA was used following a protocol previously reported [47] using EZ DNA methylation-Gold kit (Zymo, Irving, CA, USA). Primers were designed using PYROMARK software (Qiagen; Table S2b) for 300 bp amplicons, approximately. These primers included Illumina sequencing adaptors, used for a second-round PCR, which was then performed to index each pituitary sample. Samples were pooled, purified, and size selected with AmpPure beads (Beckman-Coulter, Brea, CA, USA) and sequenced using the Illumina MiSeq v2 300 cycle run kit. Pairedend reads were mapped using Bismark to a custom genome made up of the amplicon sequences. An R script was then used to extract average methylation values for each CpG position. Methylation levels from multiple CpGs were then averaged to produce a value per amplicon, excluding positions where mutations/ deletions at CpGs were frequently observed in patients; specifically, the first five CpGs were used for CpG1 and CpG2; the first eight CpGs in CpG4.1; and all CpGs in the remaining regions. 2.7. Proliferation, colony formation, and migration Proliferation, colony formation, and cell migration assays were performed as previously described [45,48]. Briefly, BON-1 proliferation and colony formation were performed by seeding 1000 cells in 6-well plates for 10 days. For proliferation, cells were treated 24 h after seeding and refreshed every 48 h; for colony formation, treatment was made only during 24 h prior to seeding. QGP-1 proliferation assay was performed using Alamar Blue Reagent (Bio-Source International, Camarillo, CA, USA), as previously reported [45]. Cell migration was evaluated by wound-healing assay, seeding cells in 24-well plates until maximum confluence. Then, we made a scratch in the middle on the well and took images of the scratch at 0 and 24 h. Wound healing was calculated as the uncovered area 24 h after the wound compared to the uncovered area just after wounding. Wound-healing assay is not feasible in QGP-1 cell line since these cells grow in clusters and do not migrate to fill out the empty space made on the plate surface. 2.8. Western Blot BON-1 and QGP-1 cells transfected and treated were lysed to analyze protein phosphorylation by western blot, following standard procedures [49], and using phospho- ERK (#4370S, Cell Signaling, Beverly, MA, USA), phospho-AKT (#4060S, Cell Signaling), AKT (#9272S, Cell Signaling), and ERK (sc-154, Santa Cruz Biotechnology, Dallas, TX, USA) antibodies and HRP-conjugated goat-anti rabbit (#7074s; Cell Signaling) secondary antibody. Primary antibodies were diluted 1 : 1000, and secondary antibody was used at 1 : 2000. Band densitometry analysis was performed with IMAGEJ software (Bethesda, MD, USA), using total protein as reference factor of corresponding phosphorylated protein. 2.9. Statistical analyses Statistical comparisons between groups were performed by unpaired parametric ttest and nonparametric Mann-Whitney Utest, according to normality (Kolmogorov-Smirnov test). Pearson’s or Spearman’s bivariate correlations were performed for quantitative variables. One-way ANOVA analysis was used for the statistical comparison between more than two groups, since all of them were normally distributed (Kolmogorov–Smirnov or Shapiro–Wilk tests). The P- values were two-sided, and statistical significance was considered when P<0.05. Statistical analyses were assessed using GRAPHPAD PRISM 7 (GraphPad Software, La Jolla, CA, USA). 3. Results 3.1. Role of DNA methylation and natural antisense transcript (NAT) in the regulation of SSTR5 transcription in somatotropinomas and PanNETs As an initial approach, we performed an in silico study of the structure of the SSTR5 gene (Fig. 1A, Fig. S1). 767Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies S. Pedraza-Arevalo et al.SST 5 regulation in neuroendocrine tumors The information obtained from the UCSC Genome Browser (version GRCh37/hg19) revealed the existence of an overlapping gene in humans, SSTR5-AS1, which encodes a long intergenic noncoding RNA, and could regulate SSTR5 expression, as has been shown for other NATs. Moreover, there are four CpG islands, named hereafter as CpG1-4, which are susceptible zones of methylation, along both genes, which could also regulate their expression. Some of those CpG islands are in sites of interest, for they could be important in the control of the expression of these genes. Specifically, CpG1 overlaps with the last exon of the NAT and CpG2 falls on the big intron of NAT. CpG3 coincides with the first exon of the SSTR5 gene, partially overlapping with its promoter, and with another part of the larger intron of the SSTR5-AS1. Besides, CpG4 was the largest region identified and was subdivided into three subzones for the purpose of the study: CpG4.1 overlaps with the start of the NAT, possibly with its promoter, and the intron of SSTR5; CpG4.2 falls in the exon of SSTR5 and coincides with the coding sequence of the canonical SST 5 ; CpG4.3 overlaps with the center of the large exon of SSTR5 gene, including its zone of alternative splicing, and the zone immediately previous to the SSTR5-AS1 gene. In the first experimental assay, we used bisulfite sequencing to measure the methylation levels of these four CpG zones (Fig. 1B) in a cohort of 11 normal pituitary (NP) samples and 27 samples of somatotrope tumors causing acromegaly (summarized in Table S1). Specifically, CpG1 was 20% more intensely methylated in somatotropinomas than in NP. In contrast, CpG3, which displayed levels of <5% of methylation in all the samples, exhibited a marginally lower, but significant, degree of methylation in somatotropinoma than in NP samples. In CpG4.1 and CpG4.2, methylation levels were between 10% and 20%, but no significant differences were observed; whereas, in CpG4.3 methylation levels showed a significant decrease of approximately 5% in somatotropinomas compared to NPs, albeit displaying very high levels in both cases. Similarly, CpG2 showed high methylation levels, although no significant differences were observed between groups. As a next step, we evaluated the RNA levels of the two genes of interest, SSTR5 and SSTR5-AS1, in the same cohorts of somatotropinoma and NP samples (Fig. 1C). Interestingly, SSTR5 was clearly overexpressed in somatotropinoma samples compared to NP tissues, whereas expression levels of SSTR5-AS1 gene showed a similar trend but did not exhibit a statistically significant change. Of note, the expression of both genes showed a direct correlation in both NP and somatotropinoma samples (Fig. 1D), which could suggest a possible functional association between these two genes. Conversely, no correlations were observed between the expression of the antisense gene and the SST 5 TMD4 truncated variant of the receptor (Fig. S2). Expression of SSTR5 and SSTR5-AS1 genes was next compared with methylation levels of the CpG islands overlapping them in the genome. Remarkably, expression of both genes was tightly and inversely correlated with methylation levels of CpG4.3 (Fig. 1E)in somatotropinoma but not in NP samples, whereas they did not show a significant correlation with methylation levels of any of the other CpG islands examined (Fig. S3). CpG4.3 overlaps two functionally relevant regions; the large exon of SSTR5 wherein noncanonical alternative splicing can take place, and the putative promoter of SSTR5-AS1. Therefore, the methylation at CpG4.3 could be related with the expression of these two genes in somatotropinomas, in a manner that might be relevant to their pathological context. Nonetheless, the methylation levels of this CpG island or any of the others measured in this work did not exhibit correlations with the expression levels of the truncated isoform SST 5 TMD4 (Fig. S4). In order to investigate whether the relationship between SSTR5 and its NAT SSTR5-AS1 is also present in other tumors where the somatostatin-SST system is important, we extended our study to PanNETs. To this end, expression of both genes was measured in a cohort of 15 PanNETs, comparing tumor tissue with their paired nontumor adjacent tissue (NTAT), used as reference. Results from this analysis revealed that, while SSTR5 expression did not differ between both regions, the levels of SSTR5-AS1 mRNA were significantly higher in tumor samples (Fig. 1F). By contrast, expression levels of these genes were directly and strongly associated in both tumor and nontumor tissue, reinforcing the idea of a functional link between them (Fig. 1G). Unfortunately, the methylation levels of these samples could not be measured due to the limited quality of the DNA from formalin-fixed paraffinembedded samples. 3.2. SSTR5-AS1 and SSTR5 expression levels are interrelated and may be altered by demethylases To better understand the potential functional role of SSTR5-AS1 in NETs, we performed a stable silencing of this NAT using a specific shRNA and interrogated its possible link with the SSTR5 gene. For this and the ensuing assays, the PanNET model cell lines BON-1 and QGP-1 were used, also due to the lack of suitable 768 Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies SST 5 regulation in neuroendocrine tumors S. Pedraza-Arevalo et al. 769Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies S. Pedraza-Arevalo et al.SST 5 regulation in neuroendocrine tumors human cell models for somatotropinomas. After silencing, cells were treated with pasireotide, a secondgeneration SSA with high affinity for SST 5 , in order to test whether SSTR5-AS1 may impact in the cell response to this treatment. Interestingly, the first observation was that SSTR5-AS1 silencing by 30%, concomitantly decreased SSTR5 expression in BON-1 cells (Fig. 2A), and, while not reaching a significant difference, it caused a similar trend to decrease in QGP-1 cells. The relation between the expression of these two genes seems to be reciprocal, working in both directions, in that silencing of SSTR5 with a specific siRNA also decreases the expression of SSTR5-AS1 (Fig. S5). Treatment with pasireotide (100 nM; 24 h) increased the expression levels of both SSTR5 and SSTR5-AS1 only in BON-1 cells, suggesting the existence of a positive feedback regulatory mechanism linking SST 5 activation and the expression of this receptor, which may also involve NAT. Intriguingly, whereas silencing of SSTR5-AS1 fully abrogated the stimulatory effect of pasireotide on the expression of this NAT, the same did not occur with SSTR5, rather, pasireotide also tended to elevate SSTR5 expression under NAT silencing. Next, to further study the possible role of methylation in the expression of SSTR5 and SSTR5-AS1 genes, basal methylation levels of CpG islands of interest were measured in BON-1 and QGP-1 cells. Interestingly, both cell lines exhibited similar levels of methylation in all the CpG islands evaluated (Fig. 2B). In fact, their levels were comparable to those observed for the human samples (Fig. 1B), except for CpG1, which displayed higher methylation levels in both cell lines than in human samples; these findings also indicated that this particular zone was more methylated in somatotropinoma samples than in NP. To explore this issue in more detail, cells were treated for 48 h with different doses of the demethylating agent 5- azacytidine (Fig. S6A). The highest effects were observed with 5 lM5-azacytidine, which acted oppositely in both genes, decreasing SSTR5-AS1 and increasing SSTR5 expression levels (Fig. 2C). This finding contrasts with the direct correlation of the expression levels of both genes observed in the previous measurements and may unveil a potential for a distinct epigenetic regulation for each gene. However, despite the ability of 5-azacytidine treatment to clearly influence gene expression, no specific alterations were found in the methylation of the CpG islands studied (Fig. S6B). These results may suggest that the changes observed are not a direct consequence of a demethylation of SSTR5/SSTR5-AS1 but may reflect off-target effects of the dose of 5-azacytidine used or may be mediated by an indirect influence of trans-regulatory elements, such as transcription factors. In any case, our findings in the cell lines suggest that DNA methylation may not be a direct regulatory mechanism for the expression of SSTR5/SSTR5-AS1 but may influence it indirectly. 3.3. Decrease in SSTR5-AS1 expression promotes aggressiveness features in vitro To further examine the functional role of SSTR5-AS1, we tested whether the presence of this NAT influences tumor aggressiveness features in vitro using the BON-1 and QGP-1 cell models. Specifically, proliferation was measured in these cell lines, while colony formation and migration were measured in BON-1, under SSTR5-AS1 silencing and pasireotide treatment. This approach first showed that NAT silencing clearly increased cell proliferation under basal culture conditions. Conversely, pasireotide did not alter proliferation under basal conditions, while it seemingly blunted the effect of NAT silencing (Fig. 3A). Interestingly, colony formation was also elevated after SSTR5-AS1 silencing, as compared to its scramble control, further suggesting the ability of this NAT to influence malignancy features of NET cells. Conversely, pasireotide did not alter colony formation under control conditions, while, again, blunting the stimulatory action of NAT silencing (Fig. 3B). In contrast with the above, SSTR5-AS1 silencing did not increase but decreased cell migration, compared to scramble shRNA, thereby suggesting a disconnection between the actions of SSTR5-AS1 on these distinct functional cell features. Fig. 1. Expression of SSTR5 is regulated by DNA methylation and NAT. (A) Cartoon representation of SSTR5-AS1 and SSTR5 loci, based on Genome Browser information. (B) Comparison of methylation levels between somatotropinoma (SOMAT) and normal pituitary (NP) samples, expressed as percentage, under ttest. (C) Expression levels of SSTR5 and SSTR5-AS1 (ttest) and (D) correlations (Pearson correlation) between them in somatotropinomas and NPs, measured by qPCR and normalized by ACTB. (E) Correlations (Pearson correlation) between methylation levels of CpG4.3 and expression levels of SSTR5 and SSTR5-AS1 in somatotropinoma samples. (F) Expression levels of SSTR5 and SSTR5-AS1 (Mann–Whitney Utest) and (G) correlations (Spearman correlation) between them in PanNETs and nontumor adjacent tissue (NTAT), measured by qPCR and normalized by RNA18S1. Asterisks (*P<0.05; ***P<0.001) indicate values that significantly differ from control. In all cases, data represent median and interquartile range of 27 somatotropinomas, 11 NPs and 15 PanNETs with their NTAT. 770 Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies SST 5 regulation in neuroendocrine tumors S. Pedraza-Arevalo et al. Of note, pasireotide, while, as in the previous parameters measured, did not alter migration under control conditions (scramble shRNA), surprisingly increased migration when SSTR5-AS1 was silenced (Fig. 3D). We were also able to evaluate cell proliferation on QGP-1 cells, and we observed that NAT silencing also increased cell proliferation under basal conditions (Fig. 3F); moreover, after NAT silencing, pasireotide exerted an additional stimulatory effect in this cell line, which is reminiscent of the results found in migration studies on BON-1 cells. These observations highlight the relevance of the consequences that changes in SSTR5-AS1 expression may impact on the function of SSTR5 gene; in fact, proliferation assays performed after SSTR5 silencing resulted in similar, consistent increases in both cell lines (Fig. S7). In line with this, we finally evaluated the impact of SSTR5-AS1 on the activation of key proteins within typical signaling pathways regulated by SST 5 . Thus, activation of AKT and ERK were assessed after SSTR5-AS1 silencing and after 10 min of pasireotide treatment. Results obtained showed that NAT silencing decreased both AKT and ERK activation, compared to scramble shRNA (Fig. 4). Interestingly, pasireotide treatment exerted a slight but significant effect decreasing both AKT and ERK phosphorylation in BON-1 and QGP-1 under control conditions (scramble shRNA). Furthermore, pasireotide was unable to appreciably modify their phosphorylation levels after SSTR5-AS1 silencing. 4. Discussion There is now ample evidence that the somatostatin system plays a key pathophysiological role in various tumors, particularly in NETs, where detection of specific SSTs and use of synthetic SSAs provide valuable diagnostic and therapeutic tools [16]. SSAs are currently used to control tumor growth and/or hormone secretion in somatotropinomas (and other PitNETs) and in PanNETs, when surgery is not amenable [17,28,50]. SSAs action in these tumors requires sufficient SSTs expression, particularly SST 2 , the primary target of first-generation SSAs, lanreotide and octreotide [16]. Unfortunately, an appreciable proportion of patients are unresponsive to SSAs or develop Fig. 2. Interrelation of SSTR5-AS1 and SSTR5 expression and regulation by 5-azacytidine. (A) Expression levels of SSTR5-AS1 and SSTR5 were evaluated in BON-1 and QGP-1 cell lines after SSTR5-AS1 silencing (striped bars) and 24-h treatment with pasireotide 100 nM(Pas, blue), and were measured by qPCR, and adjusted by normalization factor (NF) with ACTB,GAPDH and HPRT housekeeping genes. (B) Basal methylation levels of BON-1 and QGP-1 in SSTR5 and SSTR5-AS1 loci, expressed as percentage. (C) RNA expression of SSTR5-AS1 and SSTR5 after treatment with 5-azacytidine demethylase. Asterisks (**P<0.01; ***P<0.001) indicate values that significantly differ between groups under one-way ANOVA; # symbol indicates values that significantly differ from control under ttest. In all cases, data represent mean SEM of n=3 and five independent experiments for BON-1 and QGP-1, respectively. 771Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies S. Pedraza-Arevalo et al.SST 5 regulation in neuroendocrine tumors resistance [16,29]. However, although it might apparently represent a survival disadvantage for the tumor, NETs also express high levels of other SSTs, especially SST 5 , which would enable the use of alternative treatments. Indeed, although first-generation SSAs bind SST 5 with high affinity, this receptor is a better target for the second-generation SSA pasireotide [51]. In fact, this SSA is currently used for the treatment of certain patients in different types of NETs [52,53]. The biology of SST 5 differs substantially from that of SST 2 or the other SSTs, and is still far from being fully understood [16,54]. High SST 5 /SST 2 ratio has been linked to SSAs resistance in acromegaly [33,55,56]. Likewise, human SSTR5 is the only gene of the SSTR family that, despite lacking typical introns in its coding sequence, can generate aberrant splice variants, for example, SST 5 TMD4, which are overexpressed in NETs and have been linked to oncogenic processes and SSAs resistance [16,25,33]. This underscores the importance of advancing in our understanding of the mechanisms regulating SSTR5 expression and the biogenesis of SST 5 , and to identify putative factors controlling its functioning in NETs. In this scenario, we initially applied an in silico analysis of the SSTR5 gene region that revealed the existence of a natural antisense transcript (NAT) overlapping in the genome with SSTR5 gene, which had already been named, accordingly, SSTR5-AS1, but whose role or regulation had not yet been reported. A closer analysis revealed that, distributed along the loci of these two genes, there are four CpG islands which could be targets for DNA methylation. We then analyzed in detail these two original features of SSTR5 in NETs. Specifically, presence and relative abundance of SSTR5-AS1 with respect to SSTR5 was examined in somatotropinomas and PanNETs, whereas methylation levels of the different islands were measured in two PanNET cell lines and in the cohort of somatotropinomas. Results from this latter approach revealed, for the first time, that some of Fig. 3. Alteration of aggressiveness features after SSTR5-AS1 silencing in BON-1 and QGP-1. (A) Proliferative rate of BON-1 cells after 10 days of silencing (striped bars) and/or pasireotide treatment (Pas, blue), represented as area covered in the well. (B) Capacity to form colonies under SSTR5-AS1 silencing (striped bars) and/or 24 h of pretreatment with pasireotide (blue), measured by number of colonies after 10 days of incubation. (C) Representative pictures of proliferation (top) and colonies formation (bottom) assays in BON-1 cells. (D) Migration rate under SSTR5-AS1 silencing (striped bars) and/or pasireotide treatment (blue), after 24 h of the wound, represented by healed area. (E) Representative pictures of migration assay in BON-1, scale bars represent 500 lm. (F) Proliferative rate of QGP-1 cells after 48 and 72 h. Asterisks (*P<0.05; **P<0.01; ***P<0.001) indicate values that significantly differ between groups under one-way ANOVA; # symbols indicate values that significantly differ from control under ttest. In all cases, data are presented as percentage of control and represent mean SEM of n=5 independent experiments for proliferation and 6 for colonies formation and migration. 772 Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies SST 5 regulation in neuroendocrine tumors S. Pedraza-Arevalo et al. pituitary tumor cell migration and invasion through cytoskeleton protein recruitment. Int J Cancer 142, 1842–1852. Supporting information Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. UCSC Genome Browser (version GRCh37/ hg19) representation of SSTR5-AS1 and SSTR5 loci. Fig. S2. Correlations of SSTR5-AS1 and SST 5 TMD4 expression in NP and somatotropinoma samples, measured by qPCR and normalized by ACTB. Fig. S3. Correlations between methylation levels of CpGs and expression levels of SSTR5 and SSTR5- AS1 in NP and somatotropinoma samples. Fig. S4. Correlations between methylation levels of CpGs and expression levels of SST 5 TMD4 in NP and somatotropinoma samples. Fig. S5. RNA expression of SSTR5 and SSTR5-AS1 after SSTR5 silencing compared to scramble siRNA (100%). Fig. S6. A. RNA expression of SSTR5 and SSTR5- AS1 after treatment with different doses of 5-azacy- tidine in BON-1 and QGP-1. B. Methylation levels of CpGs in cell lines treated with 5-azacytidine, compared to nontreated control. Asterisks (*,p<0.05; **, p<0.01; ***,p<0.001) indicate values that significantly differ from control under ANOVA analysis. Fig. S7. Proliferation assay after SSTR5 silencing in BON-1 and QGP-1 cell lines, performed with Alamar Blue. Asterisks (*,p<0.05; **,p<0.01) indicate values that significantly differ from control under ttest. Data are presented as percentage of control. Table S1. Summary of clinical parameters of somatotropinoma and PanNETs patients. Table S2. Details of primers used for quantitative PCR (a), as well as methylation assays (b). 779Molecular Oncology 16 (2022) 764–779 ª2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies S. Pedraza-Arevalo et al.SST 5 regulation in neuroendocrine tumors