MicroRNA Expression Signature and Target Prediction in Familial and Sporadic ACTH-Independent Macronodular Adrenal Hyperplasia

Background: ACTH independent macronodular adrenal hyperplasia (AIMAH) is a rare disorder characterized by bilateral macronodular hyperplasia of the adrenal glands and increased cortisol production with subclinical or overt Cushing's syndrome. To screen and analyze the microRNA prole of AIMAH, so as to elucidate the possible pathogenesis of the disease. Methods: MiRNA microarray was used to test the tissue samples from familial AIMAH patients, sporadic AIMAH patients and normal controls of other non-tumor adrenocortical tissues, to identify characteristic microRNAs expression signatures. Some miRNAs were validated by RT-PCR. Furthermore, the key signal pathways and miRNAs involved in AIMAH pathogenesis were analyzed by gene ontology and pathway analysis. Results: Characteristic microRNA expression signatures were identied for familial AIMAH patients including 16 differentially expressed microRNAs and sporadic AIMAH patients including 8 differentially expressed microRNAs respectively. RT-PCR assay conrmed the chosed miRNAs expression, suggesting the high reliability of miRNA array. Pathway analysis showed that the most enriched pathway was renal cell carcinoma pathway. Overexpression of miR-17 (cid:0) miR-20a and miR-130b can inhibit glucocorticoid-induced apoptosis in the AIMAH pathogenesis. Conclusion: We have identied the miRNA signature in in Familial and Sporadic AIMAH patients, The differentially expressed miRNAs may be involved in the mechanisms of AIMAH pathogenesis. Specic miRNAs, such as miR-17 (cid:0) miR-20a and miR-130b may be new potential targets for further functional studies of AIMAH.

syndrome, most AIMAH patients presents in the fth and sixth decades, a later age of onset compared with unilateral adenomas or Cushing's syndrome [2,11]. Although reports for large kindred with AIMAH have suggested the phenotype is consistent with monogenetic, probable autosomal dominant inheritance [12], the genetic or epigenetic basis of familial AIMAH remains to be undetermined.
MicroRNA associated research has generated great interest owing to their extensively potential effects on the regulation of gene expression both on the level of transcription and that of post-transcription since the rst microRNA was identi ed [13]. MicroRNAs (miRNAs) are a group of single-stranded, non-coding small RNAs with the length of approximately 22 nucleotides [14,15], MiRNAs are rst transcribed by RNA polymerase in hairpin structures and then processed by RNase Drosha to become long pre-miRNAs. And then RNase Dicer is involved in the process from precusors to mature microRNAs, which could bind to 3' untranslated region (3'UTR) of their targeting messenger RNAs in a sequence-speci c manner completely or partially complementarily leading to degradation or translation repression, respectively [16]. Although only a small fraction of identi ed miRNAs have been eludicated for their biological functions, these miRNAs could involve into the essential regulation processes such as proliferation, differentiation, apoptosis and development [17]. Although microRNA expression pro les differed with each other in different cancer types, it was demonstrated that several microRNAs might play very speci c roles in tumourigenesis through regulating a series of pathways [18][19][20][21][22][23].
Although, AIMAH is rare and generally presents as a sporadic disease, the family clustering of AIMAH and normal controls of other non-tumor adrenocortical tissues is more infrequent. We have tried our best to nd such a familial affected pedigree and normal adrenocortical tissues with complete clinical information and successfully collect adrenalectomy tissue samples. The aims of the study were to identify via a high-throughput approach the differentially expressed miRNAs in the samples from familial AIMAH patients, sporadic AIMAH patients and normal controls of adrenal lipoma patients, and to conduct bioinformatics analysis to determine the potential functions of the differentially expressed miRNAs and possible pathogenesis of AIMAH through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses, respectively.

Clinical samples
Written informed consent for genetic studies was obtained prior to initiating this study in agreement with protocols approved by the institutional review board at the General Hospital of the People's Liberation Army. This research was reviewed and approved by the ethics committee of General Hospital of the People's Liberation Army. Adrenal nodule samples from familial and sporadic AIMAH patients were collected after the surgeries. The normal controls tissue were collected from the non-tumor part of large adrenal lipoma which were ≧ 2 cm apart from the lipoma after the surgeries. Moreover, the computed tomographic (CT) imaging technique and hematoxylinandeosin(H&E)staining methods were employed to identify the characteristics of nodular hyperplasia. For CT analysis, the images were reviewed by two experienced observers, who arrived at a consensus. Imaging characteristics were recorded. For H&E staining, all surgical specimens were xed in 10% buffered formaldehyde and routinely processed for histologic diagnosis. The tumor histology was independently con rmed by 2 pathologists (Fig. 1).
Samples were snap frozen in liquid nitrogen immediately after resection and stored at -20℃ until the extraction of RNA.

Total Rna Extraction
Total RNA from the removed adrenal gland tissue of familial and sporadic AIMAH patients as well as the normal controls was extracted from the removed adrenal gland tissue according to the standard protocol.

Microrna Assay
The Affymetrix GeneChip miRNA Array (Affymetrix, Santa Clara, CA) was applied to investigate microRNA differential expression pro le in familial and sporadic AIMAH compared with normal control samples. This microRNA array contains 46228 probes which comprise 7815 probe sets including controls, and covers 71 organisms including human, mouse, rat, canine and rhesus macaque. The array content is derived from the Sanger miRBase miRNA database v.11 (April 15, 2008). These probe sets targeting human snoRNAs and scaRNAs are derived from the snoRNABase and the Ensemble archives. In brief, 1.5 µg of total RNA extracted from different groups of samples is labeled using the 3DNA Array Detection FlashTag™ RNA Labeling Kit according to the recommendations from manufacturers. Firstly, addition of poly A tail was carried out at 37℃ for 15 min in a volume of 15 ml reaction mix, which contained 1 × Reaction Buffer, 1. Mirna Qrt-pcr miRNA reverse transcription was performed using the TaqMan microRNA Reverse transcription Kit (Applied Biosystems) at 16•C for 30 min, 42•Cfor 30 min, and denaturation of the enzyme at 85•C for 5 min. The RT reaction was performed at 37•C for 1 hour followed by 5 min at 95•C. TaqMan microRNA assays (Applied Biosystems) were performed using the 7500 Fast Real-Time PCR System at the 9600 emulation run mode. Ct values were converted into copy numbers (copy no. = 2(− Ct)) and normalized to RNU48.

Bioinformatics Analysis
The GO project provides a controlled vocabulary to describe gene and gene product attributes in any organism (http://www.geneontology.org). The ontology covers three domains: biological process, cellular component and molecular function. Fisher's exact test was used to determine if there is more overlap between the differentially expressed list and the GO annotation list than would be expected by chance. The P-value denotes the signi cance of GO terms enrichment in the differentially expressed genes. The lower the P-value, the more signi cant the GO term (P-value < 0.05 is recommended).
Pathway analysis is a functional analysis mapping genes to KEGG pathways. The P-value (EASEscore, Fisher-P value or hypergeometric-P value) denotes the signi cance of the pathway correlated to the conditions. Lower the p-value, more signi cant is the pathway (the recommend P-value cutoff is 0.05).

Statistical analysis
Results are expressed as means ± SE. Data were compared using paired t-test for hemodynamic parameters. Values of P < 0.05 were considered statistically signi cant (P < 0.05 for differentially expressed miRNAs).
When comparing the differentially expressed miRNA pro les between two groups, fold change and Pvalue were calculated and used to identify signi cant signi cantly differentially expressed miRNAs (based on all-isoform value). Differentially expressed miRNAs between two samples were ltered through fold change (based on all-isoform value), followed by hierarchical clustering. miRNA target prediction was performed by the GO and KEGG pathway analyses were performed based on the differentially expressed miRNAs.

Validation Of Microrna Arrays By Qrt-pcr
With the purpose of validating the results from microRNA microarrays, qRT-PCR was performed on total RNA extracted from familial and sporadic AIMAH surgical tissues compared with the normal controls respectively. We evaluated microRNA level of has-miR-20b that, based on our microarray data, was differentially expressed in tissues from familial AIMAH patients compared with normal controls. The expression level of has-miR-20b in familial AIMAH patients was 1.56 fold higher than normal controls with p-value of 0.01165 which was in accordance with the results from microRNA arrays (3.11 fold higher with p-value = 0.003463). The expression levels of has-miR-342-3p (0.68 fold downregulated with p-value = 0.005814) were determined by qRT-PCR respectively in sporadic AIMAH patients and normal control samples to validate the comparison of microRNA expression pro le between these 2 groups, which was consistent with the result from microRNA array (hsa-miR-342-3p was 0.21 fold downregulated in sporadic AIMAH patients versus normal controls). Similarly, hsa-miR-342-3p and hsa-miR-101 were selected to validate the different microRNA expression levels between familial and sporadic AIMAH patients, qRT-PCR results showed that, hsa-miR-342-3p was 1.58 fold upregulated while hsa-miR-101 was 0.40 fold downregulated in familial AIMAH patients compared to sporadic patients. These qRT-PCR test results were consistent with those from microRNA arrays in which hsa-miR-342-3p was 3.6 fold upregulated and hsa-miR-101 was 0.24 fold downregulated in familial AIMAH patients versus sporadic patients.

Hierarchical Clustering Of Micrornas With Different Expression Levels
In order to identify the special microRNAs expression signatures among different groups of patients, hierarchical clustering analysis was performed on all microRNAs expression levels in the form of signal intensities included in microarrays. This hierarchical clustering analysis clustered microRNAs together according to their expression levels and samples from different groups based on similarity of the investigated microRNA expression pro les [24]. Supervised hierarchical pairwise comparison among these 3 groups of samples, we could always identify a special group of microRNAs which could discriminate one group from the other. The 16 differential expressed microRNAs signature from the results in microarray could clearly discriminate between familial AIMAH patients and normal controls including 10 downregulated microRNAs and 6 upregulated microRNAs (Fig. 3A). Similarly, the microRNAs signature containing 8 differentially expressed microRNAs could discriminate between the sporadic AIMAH patients and normal controls with statistical signi cance p-value of 0.05 (Fig. 3B). The identi ed 17 differentially expressed microRNAs could well discriminate between familial and sporadic AIMAH patients.
Comprehensive target-network prediction of differential expressed microRNAs and Pathway analysis Functional analysis of the altered microRNAs between the AIMAH patients and normal controls revealed that some enriched pathways such as signal transduction, signaling molecules and interaction metabolic, were potentially associated with the pathogenesis of AIMAH. When comparison analysis was performed between familial AIMAH patients and normal controls, all of the 16 altered microRNAs were imported into KEGG Pathway analysis software to reveal that Circadian Rhythm pathway was the most affected pathway by these microRNAs with 6 genes predicted as the potential targets, including NPAS2, CRY2, BHLHE40, BHLHE41, CRY1 and CLOCK. At the same time, however, renal cell carcinoma pathway, mTOR signaling pathway, glioma pathway, pancreatic cancer pathway and endocytosis pathway were ranked top most signaling pathways affected by these altered microRNAs (Fig. 4A). When KEGG Pathway analysis was similarly performed on the differentially expressed 8 microRNAs between the sporadic AIMAH patients and normal controls, renal cell carcinoma pathway, dilated cardiomyopathy pathway, axon guidance pathway, Ubiquitin mediated proteolysis, endocytosis and MAPK signaling pathway were ranked top most affected pathways on the basis of predicted targets, while renal cell carcinoma pathway contained 5 genes which was predicted as the potential downstream targets for the altered microRNAs, including CDC42, CUL2, EP300, GRB2 and SLC2A1 (Fig. 4B).
Interestingly, the renal cell carcinoma pathway was ranked at top 2 most affected pathways among all associated pathways containing potential targets of the altered microRNAs which were differentially expressed in familial or sporadic AIMAH patients compared to normal controls. In comparison of altered microRNAs between familial AIMAH patients and normal controls, renal cell carcinoma pathway contained 25 genes which were targeted by these microRNAs( including EGLN3, EGLN2, EGLN1, PAK6, PAK7, CUL2, CDC42, RAC1, SOS2, GAB1, SLC2A1, TGFA, PAK1, PIK3R1, AKT3, PIK3R2, MAP2K1, MET, RAF1, MAPK1, HIF1A, CRKL, VEGFA, RAP1A and CRK), while 5 genes (including CDC42, CUL2, EP300, GRB2 and SLC2A1) were targeted by microRNAs which were differentially expressed between sporadic AIMAH patients and normal controls. These results indicated that renal cell carcinoma pathway might play an important role in the pathogenesis of AIMAH although AIMAH was a benign hereditary endocrine abnormality without malignant characters (Fig. 5).

Discussion
Recent researches indicated that microRNA expression signatures might be useful for the characterization and prediction of some certain diseases especially for benign or malignant tumors [25,26], but the investigation of microRNA expression signatures on familial AIMAH and sporadic AIMAH patients whose incidence was very low remained to be unclear. In this study microRNA microarray analysis was applied to identify the differentially expressed microRNAs between familial or sporadic AIMAH patients and normal controls, our results revealed that a series of altered microRNAs found in comparison of different groups could contribute to characterization of AIMAH. Out of the differentially expressed microRNAs, hsa-miR-101, hsa-miR-20b and hsa-miR-342-3p were validated by qRT-PCR which showed directly correlated with our microRNA microarray data.
KEGG pathway analysis of the selected microRNAs and further validation by qRT-PCR revealed that these microRNAs had putative downstream targets involved into many important pathways including circadian rhythm pathway, renal cell carcinoma pathway, MAPK signaling pathway and some other signaling pathways.
our results showed that the renal cell carcinoma pathway was ranked top most affected signaling pathways in both familial AIMAH patients and sporadic AIMAH patients compared to normal controls although AIMAH was considered to be a cytologically benign disease without propensity for invasion and metastasis. CUL2, CDC42, GLUT1 and some other proteins, which had been demonstrated to be critical in malignant renal cell carcinoma, were also predicted to be downstream target proteins of microRNAs altered in AIMAH patients compared with normal controls. Our ndings in this study was consistent with the former researches from other groups, for instance, microRNA expression pro le for massive macronodular adrenocortical disease shared differentially microRNAs with renal cell carcinoma which were previously implicated in tumorigenesis or metastasis [27]. In another research, several aberrantly expressed genes which were demonstrated in oncogenic pathways were also identi ed by gene expression pro le in one AIMAH patient [28]. Although AIMAH were regarded benign for the reason that there had never been reported occurrence of metastasis and invasion in a long term post operative surveillance of patients [29], colossal increase in adrenocortical cell mass turned out be a typical characteristic for AIMAH. Based on results mentioned above, it was indicated that there was some kind of abnormality occurred on the control of proliferation in adrenocortical cells which was similar to some extent to the characteristics of malignant tumors. Potential mechanisms underlying abnormality in cell apoptosis might include the following possibilities at least partially: normal cell cycle progression was disrupted through the altered microRNAs.
MiR-17 and miR-20a which were signi cantly upregulated between familial AIMAH patients and normal controls were two of six miRNAs of the miR-17-92 cluster consists (miR-17, -18a, -19a, -19b-1, -20a, and 92a-1) generated from a single precursor RNA that is transcribed from chromosome 13 [30]. Both miR-17 and miR-20a was reported to contribute to glucocorticoid-induced apoptosis of chondrocytes in lymphoma cells [31][32][33]. MiR-17 was reported that involvement of miR17 pathway was in glucocorticoidinduced cell death in pediatric acute lymphoblastic leukemia [34][35][36]. And the NR3C1-targeting miR-130b was more strongly expressed in a glucocorticoid-insensitive multiple myeloma cell line, and its introduction to a glucocorticoid-sensitive line impaired cellular response to glucocorticoid treatment, including the induction of apoptosis [37]. So, we might predict that overexpression of miR-17,miR-20a and miR-130b can inhibit glucocorticoid-induced apoptosis in the AIMAH pathogenesis.
Hsa-miR-342-3p, hsa-miR-20b and hsa-miR-101 targeting CDC42, ERK and MEK in the MAPK signaling pathway, or abnormal expression of PI3K/AKT by the differentially expressed microRNAs might interfere the normal apoptosis progression in turn to promote cell survival of adrenocortical cells, in addition, TGFβsignaling pathway and VEGF signaling pathway were also involved into the pathogenesis of AIMAH.
In comparison between the familial AIMAH and normal controls, circadian rhythm pathway turned out to be the most affected signaling pathway according to downstream targets of the altered microRNAs. Circadian rhythm showed a universally 24-hour oscillation pattern in metabolic, physiological, endocrinal and behavioral functions of almost all species, core circadian genes such as PER, CRY and CLOCK seemed to be important for tissue homeostasis and tumorigenesis [38][39][40]. In the eld of endocrinal diseases, key components of circadian rhythm pathway could interact with the steroid hormone receptors to exert in uence on pathogenesis of some endocrinal tumors. Core circadian rhythm proteins such as PER2, CLOCK could be co-expressed with steroid hormone receptors and CLOCK associated genes could regulate glucocorticoid activity in almost all tissues through enhancing the transcriptional activity of glucocorticoid receptors [41,42]. One of the potential underlying mechanisms was that, the acetylation of several lysine residues of glucocorticoid receptors and concomitant attenuation of GR binding to glucocorticoid response elements [43]. Thanks to the fact that AIMAH patients were characterized by bilateral massive enlargement with dysfunctions of glucocorticoid hormone as well as their response receptors, circadian rhythm pathways might play an important role during the development of this disease through their interactions with glucocorticoid receptors.
Our results could only determine the differentially expressed microRNAs through comparing familial or sporadic AIMAH patients to normal controls, but the mechanism underlying the difference on microRNA expression pro les deserved to be further researched. We had collected clinical tissue samples from AIMAH affected family members and sporadic cases, which provided us with su cient materials to performed the next step of mechanism research. The microRNA expression signatures identi ed in comparison among different groups of patients need to be validated their effectiveness in other independent test sets of patients. In addition, it was remained to be determined whether the differentially expressed microRNAs in familial and sporadic AIMAH patients were directly involved into the pathogenesis. The microRNA expression pro les of familial AIMAH patients and sporadic AIMAH patients identi ed in this study turned out to be different, indicating that the mechanism underlying the pathogenesis familial AIMAH patients differed from that of sporadic AIMAH patients. However, results from functional pathway analysis found that renal cell carcinoma pathway was associated with both familial and sporadic AIMAH although different pathways were predicted to be involved into these two groups of patients respectively, for instance, circadian rhythm was ranked the rst among associated pathways in familial AIMAH patients while mTOR signaling pathway might take critical part in sporadic AIMAH pathogenesis.

Conclusion
In this study we used microRNA microarray to identify characteristic microRNA expression signatures for familial AIMAH patients including 16 differentially expressed microRNAs and sporadic AIMAH patients including 8 differentially expressed microRNAs respectively which were followed by validation of qRT-PCR. Hierarchical clustering analysis using these differentially expressed microRNAs could discriminate familial and sporadic AIMAH patients from normal controls therefore characteristic microRNA expression signatures for familial and sporadic AIMAH patients were identi ed. Pathway analysis on target proteins of these altered microRNAs indicated that renal cell carcinoma pathway and circadian rhythm might be involved into the pathogenesis of AIMAH. Further elucidation of the exact roles that these altered microRNA, such as miR-17,miR-20a and miR-130b, play in the pathogenesis and progress of AIMAH would contribute not only to our better understanding of this benign disease but also to development of new therapeutic and preventative strategies for AIMAH. Volcano plot demonstrating differences in expression levels of microRNAs in 2 groups of AIMAH patients and normal controls based on the microarray study. The log-fold changes were plotted based on the log odds of differential expression. The microRNAs with signi cant differences in expression levels (p 0.05) between familial AIMAH patients and normal controls after Benjamini-Hochberg correction are indicated in A. The microRNAs with signi cant differences in expression levels (p 0.05) between sporadic AIMAH patients and normal controls after correction are indicated in B. The microRNAs with signi cant differences in expression levels (p 0.05) between familial and sporadic AIMAH patients after correction are indicated in C. KEGG pathway analysis on diffrentially expressed microRNAs between familial or sporadic AIMAH and normal controls. A. KEGG Pathway analysis was performed on 16 altered microRNAs to reveal that Circadian Rhythm pathway was the most affected pathway. B. KEGG Pathway analysis was performed on 8 altered microRNAs to reveal that renal cell carcinoma pathway was ranked the most affected pathway.