Ghrelin-O-Acyltransferase (GOAT) Enzyme as a Novel Potential Biomarker in Gastroenteropancreatic Neuroendocrine Tumors

Objectives The association between the presence and alterations of the components of the ghrelin system and the development and progression of neuroendocrine tumors (NETs) is still controversial and remains unclear. Methods Here, we systematically evaluated the expression levels (by quantitative-PCR) of key ghrelin system components of in gastroenteropancreatic (GEP)-NETs, as compared to non-tumor adjacent (NTA; n = 42) and normal tissues (NT; n = 14). Then, we analyzed their putative associations with clinical-histological characteristics. Results The results indicate that ghrelin and its receptor GHSR1a are present in a high proportion of normal tissues, while the enzyme ghrelin-O-acyltransferase (GOAT) and the splicing variants In1-ghrelin and GHSR1b were present in a lower proportion of normal tissues. In contrast, all ghrelin system components were present in a high proportion of tumor and NTA tissues. GOAT was significantly overexpressed (by quantitative-PCR (qPCR)) in tumor samples compared to NTA, while a trend was found for ghrelin, In1-ghrelin and GHSR1a. In addition, expression of these components displayed significant correlations with key clinical parameters. The marked overexpression of GOAT in tumor samples compared to NTA regions was confirmed by IHC, revealing that this enzyme is particularly overexpressed in gastrointestinal NETs, where it is directly correlated with tumor diameter. Conclusions These results provide novel information on the presence and potential pathophysiological implications of the ghrelin system components in GEP-NETs, wherein GOAT might represent a novel diagnostic biomarker.


Introduction
Neuroendocrine tumors (NETs) comprise a heterogeneous family of malignancies with complex clinical behavior and increasing incidence [1][2][3] . Primary tumor is identified only in 70% of patients 4 , while distant metastases are frequently found at diagnosis (27-73%), influencing the overall survival [5][6][7] . Despite that histological differentiation and Ki67 index are some prognosis factors 4 , well-differentiated low-grade tumors may behave aggressively 8 . Unfortunately, surgery is often not applicable since most tumors are diagnosed at advanced stage. For these reasons, the development of novel diagnostic markers has gained scientific and clinical interest 9,10 .
The ghrelin system is involved in the regulation of multiple (patho)-physiological functions, including hormonal secretion, β-cell survival or appetite and gastric motility [11][12][13][14] . Ghrelin must undergo a unique modification, consisting of the acylation of the third serine residue, which is catalyzed by the ghrelin-O-acyltransferase (GOAT) enzyme 14,15 . Acylated ghrelin (AG) represents the peptide binding and activating its canonical ghrelin receptor, GHSR1a. Interestingly, several ghrelin system variants, resulting from post-transcriptional modifications or alternative splicing, have been identified, including the In1-ghrelin 11,16 and a truncated receptor GHSR1b, with unknown ligand and function 11,16,17 .
Alterations in the expression of specific components of this system have been associated with the development/ progression of various neoplasms 16,[18][19][20][21] , including NETs, but the clinical-molecular correlations have not been elucidated 22,23 . Accordingly, in this study we aimed to: (1) analyze systematically the expression of different components of ghrelin system in gastroenteropancreatic-(GEP-)NETs compared to non-tumor adjacent (NTA) tissue and, most importantly, to normal control tissues by quantitative real-time PCR (qPCR); (2) correlate the expression of these components with clinical/histological characteristics; and (3) perform in vitro experiments to elucidate the potential pathophysiological role of GOAT enzyme as a key component particularly altered in our cohort of NET samples, using BON-1 and QGP-1 cell lines.

Patients and samples
This study was approved by the Ethics Committee of the Reina Sofia University Hospital (Cordoba, Spain), was performed according to the Declaration of Helsinki, and patients were treated following national and international clinical practice guidelines. A written informed consent was required before inclusion. Data from 42 patients with GEP-NETs were collected (demographic and clinical characteristics of the cohort are summarized in Table 1). Additionally, 14 normal control tissues from healthy donors were also included. Patients with hereditary endocrine syndrome were excluded. Clinical records were used to collect full medical history. GEP-NETs were classified according to histopathology features as welldifferentiated NETs (G1), moderately differentiated (G2), and poorly differentiated NETs (G3) 24 . Formalin-fixed paraffin-embedded (FFPE) samples were also collected (42 tumor samples, 42 NTA and 14 normal tissues).

RNA isolation and reverse-transcription
Total RNA from FFPE samples (n = 98) was isolated using the RNeasy-FFPE Kit (Qiagen, Limburg, Netherlands) according to the manufacturer's instructions. Quantification of the recovered RNA was assessed using NanoDrop2000 spectrophotometer (Thermo Scientific, Wilmington, NC). Total RNA was retrotranscribed to cDNA with the First-Strand Synthesis kit using random hexamer primers (Thermo Scientific) as previously reported [25][26][27][28] Quantitative real-time PCR (qPCR) cDNAs were amplified with the Brilliant III SYBR-Green Master Mix (Thermo Scientific) using the Stratagene Mx3000p system and specific primers for each transcript of interest. Specifically, expression levels (absolute mRNA copy number/50 ng of sample) of ghrelin, In1-ghrelin, GOAT enzyme, GHSR1a and GHSR1b, were measured using previously validated primers 21,29,30 . RNA expression was adjusted by 18S gene expression 28,31 .

Immunohistochemistry (IHC) analysis
IHC analysis of GOAT was implemented in all 42 FFPE samples (tumor and NTA regions) using standard procedures 32 . Optimum antibody concentration (1:300) using a commercially available antibody against human GOAT (AA257-287, Acris-antibodies, Herford, Germany) was selected by performing a series of antibody dilution tests in normal pancreas 33 . Two independent pathologists performed the IHC analysis following a blinded protocol. In the analysis, 0, 1 + , 2 + , 3 + stand for absent, low, moderate, and high staining intensities of GOAT enzyme in the tumor compared to the NTA region.

Cell proliferation assay in response to GOAT inhibitor
The only commercially available GOAT inhibitor (GOATi; GO-CoA-Tat; Ref: 032-37) was purchased from Phoenix Pharmaceuticals (Burlingame, CA). The final concentration (10 -5 M) was selected based on dose-response experiments performed in prostate celllines and on previous reports 37 . Cell proliferation was determined by using Alamar-blue assay (basal, 24, 48, and 72 h) as previously reported 21,22,32 . Cells were seeded per quadruplicate and assays were repeated four times. Paclitaxel (PAX; Sigma-Aldrich) was used as control for the inhibition of proliferation 27,30 .

Migration capacity assay
The ability of BON-1 cells to migrate after 24 h of treatment was evaluated by wound-healing technique 22,[38][39][40] . Briefly, stable cells were plated at subconfluence in 6-well plates. The wound was made on confluent cells using a 100 μl sterile pipette tip. Wells were rinsed in PBS and treated for 24 h in FBS-free medium. Wound-healing was calculated as the area of a rectangle centered in the picture 24 h after the wound vs. the area of the rectangle just after doing the wound. Three experiments were performed in independent days, in which three random pictures per well along the wound were acquired and, the mean area of these pictures was used for analysis. Images were analyzed using the ImageJ software 41 .

Statistical analysis
Paired t-test analysis was used to compare the expression levels between GEP-NETs samples and NTA tissue. Non-paired t-test analysis was used to compare the expression levels between normal tissue and GEP-NETs samples or NTA tissue. U-Mann-Whitney tests were used to evaluate clinical-molecular relations within GEP-NETs samples. Chi-squared test was used to compare categorical data. All statistical analyses were performed using SPSS and GraphPad Prism. Data are expressed as mean ± SEM. p-values < 0.05 were considered statistically significant. In functional experiments, results were expressed as percentage vs. control (vehicle-treated cells). Cell proliferation rate compared to control was assessed by multiple comparison test (two-way ANOVA followed by Newman-Keuls post-hoc test).

Results
Forty-two patients with GEP-NETs were included. Demographic/clinical features are summarized in Table 1

Histopathological characterization of GEP-NETs and NTA tissue
Primary tumor samples were delimited from the NTA tissues after the evaluation of two experienced pathologists using histology and immunohistochemistry, as previously reported 32 .

Expression of components of the ghrelin system in control and GEP-NETs samples
Ghrelin system components were present at variable proportions in normal GEP samples, as determined by qPCR. Ghrelin and its native receptor GHSR1a were expressed in more than 80% of healthy controls (34/42 and 39/42, respectively), while their splicing variants In1ghrelin and GHSR1b were expressed in about 40% of the samples (17/42 and 19/42, respectively). In contrast, expression of GOAT enzyme was only detected in less than 20% (7/42) of normal samples (Supp. Fig. 1). Ghrelin and GHSR1a were also present in a high proportion (more than 60%) of the NTA and tumor samples (32/42 and 29/ 42, respectively); while GOAT enzyme and the splicing variants In1-ghrelin and GHSR1b were present in more than 40% of the samples (25/42, 21/42 and 17/42, respectively; Supp. Fig. 1). Of note, ghrelin expression levels were decreased in NTA and tumor tissue compared with normal samples, with a slightly but not significantly increased expression in tumor compared with NTA tissue (Fig. 1). A similar observation was found for GHSR1a expression, while GOAT enzyme was clearly overexpressed in tumor tissues compared with NTA regions and normal tissues, wherein it was virtually absent (Fig. 1). Finally, In1-ghrelin was more expressed in tumor tissues than in control samples but these differences were not statistically significant, while no significant changes were found in the case of the splicing variant GHSR1b (Fig. 1).
In terms of tumor grade, no significant differences in the expression of any of the ghrelin system components analyzed were found between differentiated (G1/G2) and non-differentiated (G3) GEP-NET (Supp. Fig. 2). However, we found that the expression of GOAT enzyme and GHSR1a in GI-NETs was markedly higher than in PNETs, while the expression of ghrelin was lower in GI-NETs compared to PNETs (Supp. Fig. 3). Additionally, ghrelin expression levels correlated with those of In1-ghrelin (R 2 = 0.532; p < 0.01) and GOAT (R 2 = 0.422; p < 0.05) in tumor samples, while the expression of GHSR1a was correlated with GHSR1b (R 2 = 0.444; p < 0.05).

Immunohistochemistry analysis of the presence of GOAT enzyme
Based on the marked overexpression of GOAT enzyme, as well as on previous reports 16,18 , we also sought to analyze its presence at the protein level. IHC analysis of tumor tissue revealed that GOAT enzyme was present in the vast majority of tumor cells compared with NTA tissue (Fig. 2a), with different grades of staining. Indeed, in our cohort, 86% of the tumor samples (36/42) evaluated were positive for the presence of GOAT enzyme by IHC (Fig. 2b), wherein 40% of the tumor cases (17/42) presented a strong staining (2 + or 3 + ) for GOAT compared to NTA tissue (Fig. 2c, d). Of note, mRNA expression levels in tumors samples correlated with GOAT expression by IHC (Fig. 2e). Additionally, strong staining (2 + or 3 + ) for GOAT was correlated to increased age at diagnosis (62.5 ± 4 years) compared to those tumors with absent or lower staining (0 or 1 + ; 51 ± 2 years; p > 0.05).

Correlations between the expression levels of ghrelin system components and clinical-histological characteristics in GEP-NETs
Epidemiological data revealed that patients with tobacco exposure exhibited higher expression of ghrelin and In1ghrelin (Fig. 3a). Moreover, patients with family history of tumor disease had a lower expression of ghrelin (Fig. 3b). Conversely, sex, personal history, previous neoplasm history, clinical symptoms, or other histological parameters (vascular/peritumoral invasion, lymph node metastasis) were not associated with the expression of any of the components of the ghrelin system. Expression of some ghrelin system components was also associated to tumor characteristics, invasion capacity and prognosis in GEP-NETs. Specifically, functioning tumors presented higher levels of GHSR1a (Fig. 3c), while lower expression levels of this receptor were associated to the presence of affected surgical borders and mortality (Fig. 3c). Tumors with necrosis had lower GOAT mRNA levels and those with liver metastasis had decreased expression levels of In1-ghrelin (Fig. 3d). Interestingly, functionality was also associated with increased expression of GHSR1b (Fig. 3e). Finally, tumor diameter was directly correlated to GOAT expression (R = 0.33; p < 0.05). Remarkably, no further associations were found between expression levels of ghrelin system components and clinical/histological characteristics when considering separately PNETs and GI-NETs (data not shown).

In vitro analysis of the role of GOAT in PNETs cell lines
We decided to further investigate the pathophysiological role of GOAT enzyme using the only available GOATi in PNETs cell lines. However, GOATi did not affect cell proliferation in BON-1 and QGP-1 cells (Fig. 4a) or the migration capacity of BON-1 cells (Fig. 4b).

Discussion
This study aimed at evaluating systematically the expression of various components of the ghrelin system in an ample series of clinically well-characterized GEP-NETs, and to compare these expression levels with those in the corresponding adjacent non-tumor tissues and in normal control tissues. Previous studies have reported certain components of the ghrelin system in GEP-NETs 22,42-44; however, to our knowledge, this is the first study that comprehensively characterizes these components in tumor samples compared to their corresponding adjacent non-tumor regions, as well as with normal tissue samples. Moreover, we analyzed the demographic, epidemiological, and clinical characteristics as well as the disease progression and prognosis after 2-10 years of the patients with GEP-NETs. Overall, our results revealed that most of the components of the ghrelin system exhibit a distinctive expression in tumor and peritumoral tissues compared to normal tissue samples. Indeed, specific components of the ghrelin system, and especially GOAT, displayed remarkable alterations and clinical-histological correlations in tumor tissues, suggesting their potential value as novel biomarkers in GEP-NETs.
Also in our cohort, GEP-NETs exhibited a substantial molecular heterogeneity and variability 22,32,45 . Our results are consistent with previous reports showing that different components of the ghrelin system are present in tumor and non-adjacent tissues, and that, some of these components can be overexpressed in tumor samples compared to the surrounding tissue 22,[42][43][44] . Differences among these studies may be related to the differences among patient cohorts.
Ghrelin system regulates key bodily functions, such as hormonal secretion and cell proliferation, in both normal and tumor cells [11][12][13]46,47 . In this context, our and other studies support the notion that the dysregulation of ghrelin system components observed in NETs could be pathologically relevant and may participate in tumor progression. The diverse localization and morphology of ghrelin-producing cells in the GI tract, and their implications on metabolic/endocrine functions, might suggest a role of this component in the regulation of GEP-NETs pathophysiology 12,13,42 , and could also explain the ample molecular heterogeneity found herein in the expression of ghrelin in different normal control tissues. Moreover, it could also be related to the overall overexpression of ghrelin in normal tissues compared to tumor samples. Although expression of canonical ghrelin has been described in various tumor types, its potential role in cancer is still controversial 48,49 . Ghrelin has been described in NETs using immunohistochemistry and In the analysis, 0, 1 + , 2 + , 3 + stand for absent, low, moderate, and high intensities of the tumor region staining compared to the adjacent region with non-tumor tissue (3D1, 3D2, 3D3, 3D4, respectively). This analysis revealed that GOAT was present in the vast majority of tumor cells compared with non-tumor adjacent tissue, with different grades of staining. e Correlation between the absolute mRNA expression of GOAT determined by qPCR in GEP-NETs samples (values are adjusted by 18S expression) and the intensity of GOAT staining qPCR 22,25,42,43 , in our cohort ghrelin was expressed in NET samples, albeit in substantially lower amounts than in normal tissues. In addition, in our cohort ghrelin expression levels were higher in PNETs than in GI-NETs, which is consistent with previous evidence 43 but differed from other reports 44 . At variance with previous studies that did not find any clinical correlation between ghrelin expression and clinical features 22,42 , we observed here that ghrelin expression was higher in patients without tumor family history. Similarly, the in vitro effects of ghrelin on cell proliferation are also controversial 48,50-56 and some studies have reported an association between ghrelin and poor survival in renal cell carcinoma patients 57,58 . Altogether, these data reinforce the notion that NETs are highly heterogeneous tumors, wherein the particular ghrelin expression profile and its clinical implications may depend on the type of tumor and the particular cohort of patients analyzed.
Expression of the canonical ghrelin receptor GHSR1a has been described in tumors including NETs 22,25,48 . Here, GHSR1a expression was highly variable in normal control samples, but tended to be overexpressed in tumor samples compared to adjacent non-tumor tissue, which is consistent with our previous study in a different cohort 22 . The relation between GHSR1a, functionality and mortality invites to explore further the potential relationship of this receptor with tumorigenesis, and its putative value as a molecular prognostic marker in NETs.
The pathophysiological implications of the ghrelin system have been recently expanded with the discovery of new molecular components 11,16,17,29 , which have been found to be overexpressed in several tumors 16,22,27 and associated to relevant clinical parameters 22 . Herein we found comparable tendencies in the expression of these variants; however, these differences did not reach statistical significance. Nevertheless, in the present cohort, >40% of tumor samples presented detectable levels of In1ghrelin and GHSR1b, while in the previous study, >80% of the tumor samples exhibited detectable levels 22 . These differences could likely reduce the statistical power of the comparisons and correlations, and, again, would illustrate the elevated heterogeneity of NETs. The most novel and relevant finding of this study is the marked overexpression of GOAT in NET samples. Whereas the expression of this enzyme was almost absent in control tissues, it was present in adjacent non-tumor tissue and notably overexpressed in tumor tissues. These, together with previous results showing a similar, remarkable overexpression of GOAT in breast and pituitary tumors 11,21 provide suggestive evidence for a striking dysregulation of this enzyme in endocrine-related tumors. The expression levels GOAT does not always correlate with those of ghrelin, whereas they do parallel more consistently the expression levels of In1-ghrelin, suggesting the existence of additional targets for GOAT enzyme 11 . In NETs, GOAT levels have been correlated with those of In1-ghrelin, and associated with worse outcome 22 , these findings were not reproducible in our cohort, which may be explained by the tumor heterogeneity and the limited number of tumor samples. Despite this, in the present study, GOAT expression is associated to larger tumors, especially in GI-NETs, reinforcing the notion of a possible association between the dysregulation of this enzyme and the pathophysiology of NETs. This is the first study that demonstrates an intense overexpression of GOAT enzyme by IHC in GEP-NETs tissues compared to non-tumor adjacent tissues; however, its functional implications should be precisely defined. Here, a GOAT inhibitor administered on two NET cell lines, BON-1 and QGP-1, did not show relevant changes in cell proliferation or migration in vitro. Thus, future studies should explore this further, using novel inhibitors or other inhibiting/silencing approaches.
Notwithstanding this, our current and previous 22 studies provide compelling evidence that certain components of the ghrelin system, and specially GOAT enzyme, are clearly overexpressed in NETs, suggesting their potential value as diagnostic and/or prognostic biomarkers for this pathology. In support of the present finding in NETs, GOAT has been also recently reported as non-invasive plasma biomarker in prostate cancer 18 . Additionally, the association between GHSR1a and GHSR1b with the functionally of these tumors and the mortality of these patients further supports this notion and emphasizes the importance of exploring the modulation of this receptor for improving patient outcome. Therefore, although it is b Cell migration rate in BON-1 after 24 h of treatment with GOAT inhibitor by wound-healing assay. Cell proliferation rate compared to control was assessed by multiple comparison tests while migration was assessed by U-Mann-Whitney test. Values represent the mean ± SEM of at least three individual experiments. Asterisks indicate significant differences (*p < 0.05; ***p < 0.001) compared with control (set at a 100%). Legend: ns means nonsignificant difficult to predict the specific clinical impact of these findings, taken together, all these results invite to analyze in more detail the putative utility of GOAT overexpression as a diagnostic biomarker in NETs.
In summary, we present the first systematic characterization of the components of the ghrelin system, including splicing variants, in GEP-NETs tissues in comparison with their adjacent non-tumor regions, and also with normal tissue samples. Our results demonstrate that key components of this system are markedly dysregulated in GEP-NETs and associated to key clinical parameters, suggesting the interest of further studying these molecular targets, especially GOAT, as putative diagnosis and/or prognostic markers in GEP-NETs.

Study highlights
What is current knowledge -Some components of ghrelin system could be altered in neuroendocrine tumors

What Is New Here
-Key components of ghrelin system are markedly dysregulated in GEP-NETs and associated to key clinical parameters. -Changes in the expression of ghrelin system components are associated with the development and/or progression of GEP-NETs. -These molecular targets, especially GOAT, may represent putative diagnosis and/or prognostic markers in GEP-NETs.