Increased E2F1 mRNA and miR-17-5p Expression Is Correlated to Invasiveness and Proliferation of Pituitary Neuroendocrine Tumours.

miR-17-5p and E2F1 have been described as deregulated in cancer, but they have scarcely been studied in pituitary neuroendocrine tumours (PitNETs). This study evaluates the relationship of E2F1 and miR-17-5p with the invasiveness and proliferation of PitNETs. In this cross-sectional descriptive study, we evaluated the expression of E2F1, MYC, and miR-17-5p by quantitative real time PCR analysis in 60 PitNETs: 29 gonadotroph (GT), 15 functioning somatotroph (ST), and 16 corticotroph (CT) tumours, of which 8 were silent (sCT). The clinical data were collected from the Spanish Molecular Register of Pituitary Adenomas (REMAH) database. We defined invasiveness according to the Knosp classification and proliferation according to a molecular expression of Ki-67 ≥ 2.59. E2F1 was more expressed in invasive than in non-invasive tumours in the whole series (p = 0.004) and in STs (p = 0.01). In addition, it was overexpressed in the silent subtypes (GTs and sCTs; all macroadenomas) and normoexpressed in the functioning ones (fCTs and STs; some microadenomas). miR-17-5p was more expressed in proliferative than in non-proliferative tumours (p = 0.041) in the whole series but not by subtypes. Conclusions: Our study suggests that in PitNETs, E2F1 could be a good biomarker of invasiveness, and miR-17-5p of proliferation, helping the clinical management of these tumours.


Introduction
One of the hallmarks of cancer in humans is the deregulation of E2F transcriptional activity through the alteration of the p16INK4a-cyclinD1-retinoblastoma protein (Rb) pathway. E2F is a family of transcription factors that controls the expression of important genes for the progression of the cell cycle, apoptosis, DNA repair, and cellular differentiation. E2F1 is a transcription factor belonging to

Samples
We selected 60 PitNETs from the biobank pituitary collection of the Alicante General University Hospital and 9 normal pituitary glands obtained from autopsies performed at the same centre. All specimens were preserved within 24 h to 48 h after surgery at 4 • C in RNAlater solution and frozen at −80 • C.

Clinical, Radiological, and Pathological Data of the Patients
Patients' gender, age, and clinical and radiological data were retrieved anonymously. Clinical information included: endocrine or neuro-ophthalmological symptoms and signs, and biochemical evaluation of the pituitary function.
Two investigators independently reviewed the magnetic resonance images (MRIs) of selected patients. All exams where performed on a 1.5 Tesla magnet Philips Intera (Philips, Amsterdam, Netherlands) according to a predefined protocol including contrast enhanced dynamic and delayed image acquisitions. From these images, tumours were classified according to invasiveness and maximum tumour diameter (MTD). In terms of invasiveness, tumours were classified as invasive (Knosp grades III-IV) or non-invasive (Knosp grades I-II) [18].
Pathological data included the measurement of the relative gene expression of the Ki-67, because it was not possible to centralise its immunohistochemical (IHC) evaluation. To define a Ki-67 gene expression fold change (FC) equivalent to the 3% IHC cutoff, we performed a receiver operator curve analysis (ROC analysis). As a result, we found that 2.59 FC Ki-67 gene expression provided the highest sensitivity (85.5%) and specificity (77.6%). Therefore, tumours were classified as proliferative (Ki-67 ≥ 2.59) or non-proliferative (Ki-67 < 2.59).

Nucleic Acid Extraction
All experiments were performed equally on problem cases and normal pituitary glands. Tissue fragments, obtained from representative areas of the tumour, were efficiently disrupted, lysed, and homogenised using the TissueLyser (Qiagen, Hilden, Germany) and the corresponding buffer. Genomic DNA, RNA, and proteins were isolated from tumour and control tissues using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Hilden, Germany), followed by an extension of the protocol to rescue the miRNAs, according to the manufacturer's instructions. Total RNA was processed with RNase-Free DNase Set (Qiagen, Hilden, Germany) within the extraction protocol. We evaluated the quality and quantity of total RNA and miRNAs by microcapillary electrophoresis on the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

mRNA and miRNA Expression Analysis
Real time PCR (RT-PCR) was performed with the Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA, USA). To analyse miRNA expression, the RT-PCR was performed using the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) with the RT-Primers pool, according to the manufacturer's instructions.
The quantification of mRNA and microRNA expression was performed by quantitative real time PCR (qRT-PCR) with TaqMan technology. The TaqMan Gene Expression Assays and TaqMan MicroRNA Assays (Life Technologies, Carlsbad, CA, USA) chosen are detailed in Table 1. All analyses were done in duplicate and following the same protocol. Normal pituitary autopsy samples were mixed in equal parts into a pool that was analysed as a calibrator. The analyses were performed in 10-µL reaction volumes containing 5 µL of TaqMan Fast Advanced Master Mix (Life Technologies, Carlsbad, CA, USA), 0.5 µL of TaqMan Gene or miRNA Expression Assay (Life Technologies, Carlsbad, CA, USA), and 2.5 µL of template in the 7500 Fast Real Time PCR System (Applied Biosystems, Foster City, CA, USA), and were analysed using a SDS software (Applied Biosystems, Foster City, CA, USA). A relative quantification (RQ) was established based on the ∆∆Ct method in FC units (Applied Biosystems, Foster City, CA, USA).

Statistical Analysis
Molecular variables showed a non-normal distribution (Kolmogorov-Smirnov test). Associations between the molecular assays and the clinic-pathological features were calculated. Mann-Whitney U or Kruskal-Wallis tests were used for comparisons between two or more groups, as appropriate. The χ 2 test followed by Fisher's exact test when appropriate was used to identify the correlations between categorical variables. A non-parametric Spearman's correlation test was used to test the association between molecular variables and age or MTD. The effect of a variable was estimated using 95% confidence interval (CI). For multivariable analysis, we used multiple linear regression between continuous variables and unconditional logistic regression between categorical variables. P values of less than 0.05 were considered statistically significant. Statistical analysis was performed with the SPSS 15.0 software (IBM Software, Armonk, NY, USA).

Results
We studied 60 samples from the collection of PitNETs of the biobank of the Alicante Health and Biomedical Research Institute. The demographic, clinical, and radiological characteristics of the patients are shown in Table 2.

Relationships between Genes (E2F1/MYC) and miRNAs Studied and PitNET Proliferation
There was no correlation in the expression of E2F1 or MYC among the three categories of PitNETs in the whole series or by subtypes. When we grouped grades 1 and 2 (non-proliferative tumours) to compare them with grade 3 (proliferative tumours), the results did not change ( Figure 1A,B).

Relationships between Genes (E2F1/MYC) and miRNAs Studied and PitNET Proliferation
There was no correlation in the expression of E2F1 or MYC among the three categories of PitNETs in the whole series or by subtypes. When we grouped grades 1 and 2 (non-proliferative tumours) to compare them with grade 3 (proliferative tumours), the results did not change ( Figure  1A,B). By contrast, we found significant differences in the expression of miR-17-5p in the whole series, when we compared grade 3 (proliferative tumours) vs. grades 2 and 1 (non-proliferative or non-invasive tumours) (Figure 2A,B). This difference was not observed in the subtypes. Indeed, when we compared the expression of the previous genes and miR 17-5p in the tumours according only to their Ki-67 expression, only miR-17-5p was significantly higher in proliferative (Ki-67 ≥ 2.59) compared with non-proliferative tumours (Ki-67 < 2.59) (p = 0.05).

Relationships between Genes (E2F1/MYC) and miRNAs with the Invasiveness of the Tumours by Magnetic Resonance Imaging
To evaluate the association between the expressions of the genes and miRNAs studied with the invasiveness of PitNETs, we compared invasive tumours (grades 3 and 2) with non-invasive tumours (grade 1).
In the whole series, invasive tumours showed higher E2F1 gene expression than non-invasive ones (p = 0.004; Figure 3A). By subtypes, this finding was only maintained in STs (p = 0.001; Figure   Figure 1. Association between E2F1 (A) and MYC (B) fold change expression and grades of invasiveness and proliferation (grade 3 vs. grade 2 and 1) in the whole series. Grade 3: proliferative and invasive tumours; grades 2 and 1: non-proliferative and non-invasive tumours. ns, non significant; FC, fold change.
By contrast, we found significant differences in the expression of miR-17-5p in the whole series, when we compared grade 3 (proliferative tumours) vs. grades 2 and 1 (non-proliferative or non-invasive tumours) (Figure 2A,B). This difference was not observed in the subtypes.

Relationships between Genes (E2F1/MYC) and miRNAs Studied and PitNET Proliferation
There was no correlation in the expression of E2F1 or MYC among the three categories of PitNETs in the whole series or by subtypes. When we grouped grades 1 and 2 (non-proliferative tumours) to compare them with grade 3 (proliferative tumours), the results did not change ( Figure  1A,B). By contrast, we found significant differences in the expression of miR-17-5p in the whole series, when we compared grade 3 (proliferative tumours) vs. grades 2 and 1 (non-proliferative or non-invasive tumours) (Figure 2A,B). This difference was not observed in the subtypes. Indeed, when we compared the expression of the previous genes and miR 17-5p in the tumours according only to their Ki-67 expression, only miR-17-5p was significantly higher in proliferative (Ki-67 ≥ 2.59) compared with non-proliferative tumours (Ki-67 < 2.59) (p = 0.05).

Relationships between Genes (E2F1/MYC) and miRNAs with the Invasiveness of the Tumours by Magnetic Resonance Imaging
To evaluate the association between the expressions of the genes and miRNAs studied with the invasiveness of PitNETs, we compared invasive tumours (grades 3 and 2) with non-invasive tumours (grade 1).
In the whole series, invasive tumours showed higher E2F1 gene expression than non-invasive ones (p = 0.004; Figure 3A). By subtypes, this finding was only maintained in STs (p = 0.001; Figure   Figure 2. (A) Percentage of tumours with miR-17-5p overexpressed, normoexpressed, or repressed depending on grade of invasiveness and proliferation in the whole series; (B) Differences in miR-17-5p fold change expression between proliferative and invasive tumours (grade 3) and non-proliferative or non-invasive tumours (grades 2 and 1) in the whole series. * Mann-Whitney U, p = 0.041.

Relationships between Genes (E2F1/MYC) and miRNAs with the Invasiveness of the Tumours by Magnetic Resonance Imaging
To evaluate the association between the expressions of the genes and miRNAs studied with the invasiveness of PitNETs, we compared invasive tumours (grades 3 and 2) with non-invasive tumours (grade 1).
In the whole series, invasive tumours showed higher E2F1 gene expression than non-invasive ones (p = 0.004; Figure 3A). By subtypes, this finding was only maintained in STs (p = 0.001; Figure 3B). MYC also showed higher expression in invasive STs than in non-invasive ones (p = 0.018; Figure 4A), but not in the whole series. Indeed, the expression of MYC was lower in invasive than in non-invasive GTs (p = 0.033; Figure 4B). Diagnostics 2020, 10, x FOR PEER REVIEW 6 of 13 3B). MYC also showed higher expression in invasive STs than in non-invasive ones (p = 0.018; Figure  4A), but not in the whole series. Indeed, the expression of MYC was lower in invasive than in non-invasive GTs (p = 0.033; Figure 4B).  Unlike E2F1 and MYC, there were no differences in the expression of pri-miR-17-92 and mature miRNAs and invasiveness behaviour in the whole series or by subtypes.

Expression Profiles of E2F1, MYC Genes, and Pri-miR-17-92 by Subtypes
As shown in Table 3, E2F1 was overexpressed in the silent (GTs and sCTs) and normoexpressed in functioning PitNET subtypes (fCTs and STs). Contrarily, MYC was normoexpressed only in the tumours of the corticotroph line (both functioning and silent corticotroph tumours) and repressed in the other subtypes (GTs and STs).  Diagnostics 2020, 10, x FOR PEER REVIEW 6 of 13 3B). MYC also showed higher expression in invasive STs than in non-invasive ones (p = 0.018; Figure  4A), but not in the whole series. Indeed, the expression of MYC was lower in invasive than in non-invasive GTs (p = 0.033; Figure 4B).  Unlike E2F1 and MYC, there were no differences in the expression of pri-miR-17-92 and mature miRNAs and invasiveness behaviour in the whole series or by subtypes.

Expression Profiles of E2F1, MYC Genes, and Pri-miR-17-92 by Subtypes
As shown in Table 3, E2F1 was overexpressed in the silent (GTs and sCTs) and normoexpressed in functioning PitNET subtypes (fCTs and STs). Contrarily, MYC was normoexpressed only in the tumours of the corticotroph line (both functioning and silent corticotroph tumours) and repressed in the other subtypes (GTs and STs).  Unlike E2F1 and MYC, there were no differences in the expression of pri-miR-17-92 and mature miRNAs and invasiveness behaviour in the whole series or by subtypes.

Expression Profiles of E2F1, MYC Genes, and Pri-miR-17-92 by Subtypes
As shown in Table 3, E2F1 was overexpressed in the silent (GTs and sCTs) and normoexpressed in functioning PitNET subtypes (fCTs and STs). Contrarily, MYC was normoexpressed only in the tumours of the corticotroph line (both functioning and silent corticotroph tumours) and repressed in the other subtypes (GTs and STs).  As shown in Table 4, Pri-miR-17-92 was repressed in all PitNET subtypes. However, mature miRNAs (miR-20a and miR-17-5p), similarly to E2F1, appeared slightly overexpressed in the silent subtypes (GT and sCT) and normoexpressed in functioning PitNET subtypes (fCTs and STs). Finally, there was a negative correlation between E2F1 and mature miRNAs expression in the whole series. Only pri-miR-17~92 correlated positively with E2F1 in fCTs, but not in sCTs.

Discussion
In this paper, we try to find biomarkers of proliferation and invasiveness in PitNETs and report the possible implication of E2F1, MYC, pri-miR-17~92, miR-20a, and miR-17-5p in pituitary tumourigenesis. The strength of the paper is the size of the sample studied, which allows the analysis of different subtypes of PitNETs, and the scarcity of other published data. On the other hand, one limitation of the study is that we report correlations, which allow the generation of hypotheses, but not definitive conclusions on the relationship between E2F1, MYC, and pri-miR-17-92 in the tumourigenesis of PitNETs.
PitNETs are among the most common intracranial tumours and are usually benign neoplasms [21]. However, 30% to 40% of them invade surrounding structures, and relapses or re-growth during follow-up are frequent [22]. In the absence of more specific markers, IHC assessment of either p53 or the Ki-67 labelling index has been traditionally used as marker of aggressiveness. However, many researchers have raised doubts about their usefulness [12,[23][24][25]. Indeed, some years ago, Trouillas et al. [19,20] suggested a classification of the behaviour of pituitary tumours taking into account not only markers of proliferation, but also the IHC type, the invasiveness of the tumours, and their ability to relapse. To the best of our knowledge, E2F1 and MYC have never been studied as possible biomarkers of PitNETs.
E2F1 is a transcription factor with activities related to either inhibiting or stimulating the development of the tumours; therefore, it could be considered a modulator of cell growth and death [1][2][3]. Recently, Fu et al. showed that E2F1 was implicated in the overexpression of CCAT2, which, by interacting with PPTG1, could stimulate the development and progression of PitNETs [26]. In addition, Araki et al. also demonstrated that E2F1 could play a role in pituitary tumourigenesis, through its induction of POMC and ACTH in fCT [27,28]. More recently, Metin-Armagan et al. have suggested the involvement of E2F1 and other genes and proteins, such as CHEK2 and p73, in the tumourigenesis process of non-functioning and functioning (Cushing Disease and Acromegaly) PitNETs [29]. Indeed, in the present study, we observed higher expression of E2F1 in invasive versus non-invasive tumours in the whole series and in the ST subtype. However, we did not find differences in the expression of E2F1 between proliferative (Ki-67 ≥ 2.59) and non-proliferative tumours. Thus, E2F1 could be a potential biomarker of PitNET invasiveness, but not of proliferation.
MYC is an oncogene that participates in the tumourigenesis of most cancers and is associated with poor prognosis [30,31]. In PitNETs, Liu et al. found that the increase of MYC expression could predict the aggressiveness of silent tumours [32]. On the other hand, other authors found protein expression of MYC in just 27% of PitNETs [33]; thus, it can hardly be considered a biomarker of tumour behaviour.
In our study, MYC was repressed in all PitNET subtypes, and there was no association between MYC and proliferation in the whole series or for subtypes. However, similarly to E2F1, we found a positive relationship between MYC expression and tumour invasiveness, although only in the ST subtype. The opposite occurred in the GT subtype. In conclusion, and like E2F1, MYC could not be considered a biomarker for PitNET proliferation. On the other hand, more studies are needed to establish a relationship between MYC expression and PitNET invasiveness, because its behaviour could depend on the tumour subtype.
miRNAs are a large group of small RNAs that do not code for proteins and are a fundamental component of gene regulation through the regulation of post-transcriptional events [34,35]. Di Ieva et al. have highlighted the possible role of miRNAs as circulating biomarkers of pituitary tumours, given the high vascularisation of the pituitary gland [11]. Moreover, a recent review performed by Wierinckx et al. highlights the relevant role of miRNAs in tumorigenesis and aggressiveness in PitNETs through their upor down-regulation and their gene targets [36]. For instance, some years ago, down-regulation of a subset of miRNAs that target HMGA1 and HMGA2 (proteins overexpressed in malignant neoplasms [37]) were identified in PitNETs [38]. Similarly, D'Angelo et al. also observed a down-regulation of a set of miRNAs that target genes (such as HMGA1, 2, and E2F1) whose overexpression or activation plays a critical role in tumorigenesis in pituitary tumours of the somatotroph line [39,40]. Finally, the expressions of miR-23b and miR-130b, which target HMGA2 and CCNA2 genes [41], have been reported as down-regulated in somatotroph, gonadotroph, and silent PitNETs, while miR-183, which targets KIAA0101 [42], appears to be down-regulated in aggressive lactotroph tumours.
In the present paper, we observed that miR-17-5p was overexpressed in proliferative tumours. Wei et al. [43] also have demonstrated higher miR-17-5p expression in pituitary carcinoma and atypical PitNETs compared with typical PitNETs. In their study, they suggest that miR-17-5p is involved in pituitary carcinoma metastasis, attenuating target genes such as PTEN, a tumour suppressor gene, and TIMP2, involved in inhibiting the degradation of the extracellular matrix and basement membrane. Therefore, miR-17-5p could be a good biomarker of proliferation in PitNETs.
Therefore, the higher expression of E2F, MYC, and miR-17-5p in invasive and proliferative tumours (grade 3) than in grade 1 (non-invasive tumours) confirms that both invasiveness and proliferation must be taken into account in PitNET classification [20]. In addition, the high expression of these genes and miRNA, implicated in the cancer, confirms that an invasive and proliferative tumour is not a benign adenoma as a non-invasive tumour, with or without proliferation [20].
In addition to the possible implication of E2F1, MYC, and miRNAs in the behaviour of PitNETs, they could also participate in their functionality. However, there is very little information in the literature. Only Tani et al. looked for differences in the expression of E2F1 between functioning and silent subtypes, without finding any changes [44]. Contrarily, we found that E2F1 was overexpressed in the silent subtypes (GTs and sCTs) and normoexpressed in functioning PitNET subtypes (fCTs and STs). Nonetheless, functioning PitNETs comprised both macroadenomas and microadenomas, whereas all operated silent tumours were macroadenomas. Therefore, there is an important bias of size when trying to interpret the role of E2F1 in the functionality of PitNETs.
As previously noted, E2F1 is under the transcriptional regulation of MYC [7] and of some miRNAs. In the present paper, E2F1 was overexpressed while MYC was repressed in GTs, which suggests a negative feedback between E2F1 and MYC in the regulation of cellular cycle in this subtype. The negative feedback between E2F1 and MYC has been previously described by Zhang et al. in human oesophageal fibroblast (HOF) cells [45,46], through the negative effect of E2F1 on hTERT transcription [45]. However, positive feedback between E2F1 and MYC has also been reported [47,48]. Moreover, it has been recently suggested that the miR-17~92 cluster also participates in the regulation of MYC and E2F1 networks [14,[49][50][51]. Indeed, in our study, miR-17-5p was also overexpressed in GTs. The possible interrelations among E2F1, MYC, and miRNAs are outlined in Figure 5, which shows that the negative effect of MYC on E2F1 translation, mediated by miR 17-5p, would surpass its positive effect on E2F1 transcription. A similar relationship between MYC/E2F1 and miR-17-5p was observed in sCTs, but not in the functioning variants of PitNETs (fCTs and STs.) As all non-functioning PitNETs (GTs and sCTs) were macroadenomas, while some functioning PitNETs (fCTs and STs) were microadenomas, the effect of E2F1 on the growth and invasiveness of pituitary tumours could be mediated by its interaction with MYC and miR-17-5p.
subtype. The negative feedback between E2F1 and MYC has been previously described by Zhang et al. in human oesophageal fibroblast (HOF) cells [45,46], through the negative effect of E2F1 on hTERT transcription [45]. However, positive feedback between E2F1 and MYC has also been reported [47,48]. Moreover, it has been recently suggested that the miR-17~92 cluster also participates in the regulation of MYC and E2F1 networks [14,[49][50][51]. Indeed, in our study, miR-17-5p was also overexpressed in GTs. The possible interrelations among E2F1, MYC, and miRNAs are outlined in Figure 5, which shows that the negative effect of MYC on E2F1 translation, mediated by miR 17-5p, would surpass its positive effect on E2F1 transcription. A similar relationship between MYC/E2F1 and miR-17-5p was observed in sCTs, but not in the functioning variants of PitNETs (fCTs and STs.) As all non-functioning PitNETs (GTs and sCTs) were macroadenomas, while some functioning PitNETs (fCTs and STs) were microadenomas, the effect of E2F1 on the growth and invasiveness of pituitary tumours could be mediated by its interaction with MYC and miR-17-5p.

Conclusions
Our study suggests that in PitNETs, E2F1 could be a good biomarker of invasiveness, and miR-17-5p of proliferation, helping the clinical management of these tumours. In contrast, MYC's role in PitNET behaviour could be subtype-dependent. The expression of these genes must be evaluated in recurrent/progressive or aggressive PitNETs to precise if they are prognostic markers of tumour behaviour.
Despite some differences in E2F1 expression between functioning (ST and fCT) and non-functioning (GT and sCT) PitNETs, it is difficult to establish a relationship between E2F1 expression and functionality, because all silent operated tumours were macroadenomas, while most STs and fCTs were microadenomas. Finally, the effect of E2F1 on the growth of PitNETs could be mediated by a complex interaction between MYC and miR-17-5p.

Conclusions
Our study suggests that in PitNETs, E2F1 could be a good biomarker of invasiveness, and miR-17-5p of proliferation, helping the clinical management of these tumours. In contrast, MYC's role in PitNET behaviour could be subtype-dependent. The expression of these genes must be evaluated in recurrent/progressive or aggressive PitNETs to precise if they are prognostic markers of tumour behaviour.
Despite some differences in E2F1 expression between functioning (ST and fCT) and non-functioning (GT and sCT) PitNETs, it is difficult to establish a relationship between E2F1 expression and functionality, because all silent operated tumours were macroadenomas, while most STs and fCTs were microadenomas. Finally, the effect of E2F1 on the growth of PitNETs could be mediated by a complex interaction between MYC and miR-17-5p.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.