MicroRNAs (miRNAs) act at the posttranscriptional level to control gene expression in virtually every biological process, including oncogenesis. Here, we report the identification of a set of miRNAs that are differentially regulated in childhood adrenocortical tumors (ACT), including miR-99a and miR-100. Functional analysis of these miRNAs in ACT cell lines showed that they coordinately regulate expression of the insulin-like growth factor–mammalian target of rapamycin (mTOR)–raptor signaling pathway through binding sites in their 3′-untranslated regions. In these cells, the active Ser2448-phosphorylated form of mTOR is present only in mitotic cells in association with the mitotic spindle and midbody in the G2-M phases of the cell cycle. Pharmacologic inhibition of mTOR signaling by everolimus greatly reduces tumor cell growth in vitro and in vivo. Our results reveal a novel mechanism of regulation of mTOR signaling by miRNAs, and they lay the groundwork for clinical evaluation of drugs inhibiting the mTOR pathway for treatment of adrenocortical cancer. Cancer Res; 70(11); 4666–75. ©2010 AACR.
Adrenocortical tumors (ACT) in children may occur sporadically or in association with other types of neoplasms in the context of multiple neoplasia syndromes linked to germline tumor suppressor gene mutations (1). The incidence of these tumors is highest during the first 3 years of life and is several times more frequent in southern Brazil than in the rest of the world (2). In that geographic region, childhood ACTs are almost invariably associated with the germline R337H tumor protein p53 (TP53) mutation (3). These tumors are believed to be derived from the fetal adrenal cortex because of their age distribution, their pattern of hormone secretion, and their molecular phenotype (2, 4).
Although the most common genetic basis of childhood ACT is germline TP53 mutations with loss of heterozygosity (LOH) in the tumor, our knowledge of the molecular pathogenesis of these neoplasms is still limited. A very frequent feature is represented by LOH of the distal region of the short arm of chromosome 11, with preferential expression of the imprinted paternal allele and overexpression of the insulin-like growth factor-II (IGF-II; refs. 4, 5). IGF-II signaling through the IGF-IR is thought to represent an important mechanism for ACT growth and a relevant therapeutic target (6, 7). In addition, amplification and overexpression of the nuclear receptor steroidogenic factor-1 (NR5A1) is thought to play an important role in ACT pathogenesis (8, 9). Last, a study of protein-coding mRNAs found a distinct pattern of their expression in childhood ACT compared with normal adrenal cortex and also identified a set of transcripts whose expression is related to prognosis (10).
MicroRNAs (miRNAs) belong to a class of small noncoding RNAs of ∼21-nucleotide length that control gene expression at the posttranscriptional level. In their mature form, miRNAs recognize by base-pairing sequences in the 3′-untranslated region (UTR) of protein-coding transcripts, leading to translational repression or mRNA degradation (11). miRNAs are involved in virtually every biological process, from development to viral infection, and are also associated with oncogenesis (12). In addition, a great number of known miRNAs are located at fragile sites and genomic regions implicated in cancer (13).
As part of our effort to elucidate the genetic determinants of childhood ACT, in this study, we analyzed the miRNA expression pattern of childhood ACT. A group of miRNAs was found to be differentially expressed in ACT compared with normal adrenal. We focused our functional analysis on miR-99a and miR-100, which are significantly downregulated in ACT and share the same seed sequence. Among the predicted targets of these miRNAs, transcripts are found that encode key components of the IGF (IGF-IR) and mammalian target of rapamycin (mTOR and raptor) signaling pathways. Here, we show that mTOR signaling is activated in ACT and that miR-99a and miR-100 regulate expression of mTOR, raptor, and IGF-IR in adrenocortical cancer cells. Moreover, the specific mTOR inhibitor RAD001 (everolimus) potently suppresses adrenocortical cancer cell proliferation in vitro and when grown as xenografts in immunodeficient mice. These data reveal a novel layer of regulation of IGF and mTOR signaling by miRNAs and show that mTOR inhibition represents a potential new therapeutic tool in adrenocortical cancer.
Materials and Methods
All patients and normal subjects gave their informed consent to this study, which was approved by the Ethical Committees of Pequeno Principe and St. Jude Children's Research Hospitals. Patients' clinical data are reported in Supplementary Table S1. Normal adrenal glands were obtained with Institutional Review Board approval as discarded tissue from cases of Wilms' tumor. These patients, whose age ranged from 2 to 6 years, had not received chemotherapy before surgery. Normal adrenal cortex samples were isolated by an American Board–certified pathologist, snap frozen in liquid nitrogen, and processed for RNA extraction as described below.
miRNA expression studies
RNA was extracted from frozen tissue samples using the miRNeasy Mini kit (Qiagen). Total RNA (200 ng) from each sample was labeled and hybridized to human V2 microarrays (Agilent), following the manufacturer's procedures. The original expression data were first thresholded in such a way that any value less than the threshold was assigned to be equal to the threshold and any value above the threshold was left intact. The thresholded data were then log2 transformed and used for further analysis. We used the Kruskal-Wallis test to compare the median expression levels between the normal and ACT samples in each probe set. The probe sets with P value of <0.01 and signals present in at least 10 samples were used to make the heat map plot. The threshold used in this study is 1. All analyses were implemented in R version 2.6.2. Microarray data were deposited in the Gene Expression Omnibus database under record number GSE19856. Taqman reverse transcription-quantitative PCR (qPCR; Applied Biosystems) was used to confirm the expression levels of the miRNAs identified as differentially expressed by microarray plus the let-7a miRNA as a control. RNU48 was used as a reference gene for miRNA qPCR. For target gene identification, the Targetscan database (http://www.targetscan.org/) was interrogated.
Protein extracts were prepared by homogenization of tissues and cells in Laemmli buffer [50 mmol/L Tris-HCl (pH 6.8), 50% glycerol, 2% SDS, 0.02% bromphenol blue] containing 5% β-mercaptoethanol. Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane (Schleicher & Schuell). Primary antibodies used were anti-mTOR (Abcam), anti–phospho-mTOR (Ser2448; Cell Signaling Technology), anti-raptor (Cell Signaling Technology), anti-rictor (Cell Signaling Technology), anti–IGF-IRβ (Cell Signaling Technology), and anti-p42/p44 mitogen-activated protein kinases (Cell Signaling Technology). Immunoblot was performed using a chemiluminescence system for protein detection (GE Healthcare). Bands on blots were quantified using the ImageJ software (http://rsbweb.nih.gov/ij/).
mTOR activity enzyme immunoassay
Tissue lysates were prepared by homogenization in lysis buffer [50 mmol/L Tris-HCl (pH 7.4), 100 mmol/L NaCl, 50 mmol/L β-glycerophosphate, 10% glycerol, 1% Tween 20, 1 mmol/L EDTA, 20 nmol/L microcystin-LR, 25 mmol/L NaF] supplemented with protease inhibitor cocktail (Sigma-Aldrich). After centrifugation at 16,000 × g for 20 minutes to eliminate debris, total protein concentration in the lysates was adjusted to 3 mg/mL and 0.8 mL of the lysates was incubated with anti-mTOR agarose beads (Abcam) for 2 hours at 4°C. The immunoprecipitates were then washed twice with lysis buffer, and mTOR activity in the tissue lysates was measured with an immunoenzymatic assay (K-LISA mTOR Activity kit, Calbiochem) using a recombinant GST-p70S6K protein as a substrate, following the manufacturer's instructions.
Immunohistochemistry was performed on tumor paraffin sections after antigen retrieval using antibodies directed against phospho-mTOR (Ser2448) and phospho–ribosomal protein S6 (RPS6; Ser240/244; Cell Signaling Technology). Immunoreactivity was graded with scores from 0 to +++, which corresponded to negative, weak, intermediate, or strong staining intensity.
Transfections and luciferase assay
mTOR (nucleotides 7782–8707 of sequence NM_004958), raptor (nucleotides 4957–5361 of sequence NM_020761), and IGF-IR (nucleotides 9543–9833 of sequence NM_000875) 3′-UTR sequences were PCR amplified from a HeLa cell cDNA library (Clontech) using the primers shown in Supplementary Table S2 and cloned in the NotI-XhoI sites of the double-luciferase psiCHECK-2 vector (Promega) appended to the Renilla luciferase gene. Deletions in the 3′-UTR–predicted miR-99a/miR-100 binding sites were made by QuikChange mutagenesis (Stratagene). HEK 293T cultured in DMEM (4.5 g/L glucose) supplemented with 10% FCS and antibiotics in 96-well white plates (Costar) were reverse transfected in triplicate with 3′-UTR reporter constructs using Lipofectamine 2000 (Invitrogen) and pre-miR-99a, pre-miR-100, or negative control #1 precursor molecules (Ambion). Renilla and firefly luciferase activities were measured with a Luminoskan Ascent (Thermo Labsystems) luminometer. For each sample, Renilla luciferase activity was normalized by firefly luciferase activity. Each experiment was repeated three times.
ACT cell line culture
H295R cells were cultured in DMEM/F12 supplemented with 2% NuSerum (Becton Dickinson), 1% ITS+ (Becton Dickinson), and antibiotics. SW-13 cells were cultured in DMEM/F12 medium supplemented with 10% FCS (Invitrogen) and antibiotics. HAC15 cells were cultured in DMEM/F12 supplemented with 10% cosmic calf serum (HyClone), 1% ITS+, and antibiotics. Primary cells isolated from a pediatric ACT were cultured as described (14). As a method of authentication, the karyotype of cell lines and the steroid secretion profile of H295R cells were periodically tested.
For endogenous miRNA knockdown, 1 × 105 cells per well were seeded in 12-well plates and transfected with miR-100–specific or scramble control miRCURY LNA knockdown probes (Exiqon) using Lipofectamine 2000. Forty-eight hours after transfection, levels of endogenous mTOR, raptor, and IGF-IR were analyzed by Western blotting.
Immunofluorescence was performed on formaldehyde-fixed H295R cells cultured on glass slides using the anti-mTOR (Cell Signaling Technology), anti–phospho-mTOR (Ser2448), and anti–β-tubulin (Sigma-Aldrich) as primary antibodies revealed with Alexa594- and Alexa488-labeled secondary antibodies (Invitrogen). 4′,6-Diamidino-2-phenylindole (DAPI) was used as nuclear counterstain. Images were acquired with a Zeiss Axioplan 2 fluorescence microscope coupled to a digital charge-coupled device camera and processed using ImageJ.
H295R cells (6 × 106) were inoculated s.c. into the right flank of 4-week-old female NOD/SCID/γcnull mice. Three weeks later, when palpable tumors appeared, injected mice were randomly assigned to three groups of eight animals treated with placebo or with a preparation of RAD001 for oral administration (3 and 10 mg/kg/d). Drugs were administered by gavage and tumor growth was monitored by measuring with a Vernier caliper and calculating tumor volume (length × width × height × π/6). After the end of treatment, animals were euthanized according to the institutional animal care and use committee protocol. Xenografts were excised, weighted, and either frozen in OCT or formaldehyde fixed and included in paraffin for histologic studies. Blood vessels were stained with Texas red–conjugated tomato lectin (Vector), and immunohistochemistry for phospho-RPS6 was performed as described before (see Immunohistochemistry section).
A set of miRNAs is differentially expressed in childhood ACT
We investigated miRNA expression in a series of 25 ACT (patients' clinical data are reported in Supplementary Table S1) and compared it with a group of 5 normal adrenocortical samples from age-matched subjects. Significant modulation of expression in ACT was found for 26 miRNAs with a P value of <0.01. These miRNAs are shown in the heat map in Fig. 1. The majority (77%) of the differentially expressed miRNAs was downregulated in ACT compared with normal adrenal cortex. Differentially regulated expression of all these miRNAs in ACT compared with normal adrenal cortex was confirmed by Taqman qPCR, whereas expression of the let-7a miRNA, which was not detected as differentially expressed in the microarray study, was not significantly different (Supplementary Fig. S1).
ACT and normal samples were clustered into two distinct groups by unsupervised analysis, except for a single ACT sample whose miRNA expression profile clustered together with the normal group. Further analysis showed that inside the ACT cluster, two subclusters were present (T1, more distant from the normal group, and T2, closer to the normal group; see Fig. 1). Probability of relapse was significantly associated with localization in the T1 subcluster (P = 0.002, Fisher's exact test), whereas histologic type (adenoma or carcinoma) was not significantly associated with any group of samples (P = 0.35, Fisher's exact test). The expression of no single miRNA could perfectly discriminate cases with relapse from cases without relapse.
mTOR, raptor, and IGF-IR are overexpressed in ACT and upregulated by miR-100 knockdown in ACT cell lines
Among the most significantly downregulated miRNAs in ACT were miR-99a and miR-100, which share the same seed sequence and are predicted to target the 3′-UTRs of pivotal components of IGF (IGF-IR) and mTOR [FRAP1 (mTOR) and RPTOR (raptor)] signaling (Supplementary Fig. S2). Although the importance of IGF signaling in ACT is well known (6, 7, 15), no data exist yet about the involvement of the mTOR pathway in the pathogenesis of ACT. We studied the expression of mTOR (and its active phosphorylated form at Ser2448), raptor, and IGF-IR proteins in a series of ACT samples and compared it with normal adrenal cortex samples. mTOR, phospho-mTOR (Ser2448), raptor, and IGF-IR protein levels were significantly higher in ACT (Fig. 2A; Supplementary Fig. S3). Conversely, the other mTOR-associated protein rictor was not detectable either in normal adrenal samples or in ACT (data not shown). mTOR activity was also significantly higher in ACT compared with normal adrenocortical tissue (Fig. 2B). Finally, phosphorylation of mTOR at Ser2448 and of RPS6 as markers of active mTOR signaling was detected in ACT by immunohistochemistry (Fig. 2C). Quantification of immunohistochemical results is shown in Supplementary Table S3. Taken together, these results show that mTOR signaling is active in ACT. To investigate whether downregulation of endogenous miR-100 can modulate the levels of mTOR, raptor, and IGF-IR proteins, we transfected a specific miR-100 knockdown probe or a control probe into two different human adrenocortical cell lines (H295R and SW-13) and detected a dose-dependent increase in mTOR, raptor, and IGF-IR protein expression only after miR-100–specific knockdown (Fig. 2D).
miRNAs of the miR-100 family regulate the expression of mTOR, raptor, and IGF-IR through their 3′-UTRs
With the purpose to assess whether miR-99a and miR-100 can directly regulate the expression of mTOR, raptor, and IGF-IR, we fused portions of the 3′-UTR sequences of these genes harboring predicted binding sites for miR-99a/miR-100 to the luciferase reporter gene and performed transfections in HEK 293 cells in the presence of synthetic miR-99a, miR-100, or control miRNA precursors. As shown in Fig. 3A, the addition of mTOR and IGF-IR 3′-UTR sequences conferred high expression to the luciferase gene, whereas the raptor 3′-UTR had only a modest effect on luciferase activity. Although the expression of all constructs, including the empty vector, was insensitive to the effect of a high concentration (25 nmol/L) of the control miRNA precursor, miR-99a and miR-100 efficiently repressed the expression of luciferase fused to the mTOR and IGF-IR 3′-UTRs in a dose-dependent fashion, starting from a concentration of 5 nmol/L. They also repressed the expression of luciferase-raptor 3′-UTR to baseline levels. Importantly, even the highest concentrations of miR-99a and miR-100 precursors (25 nmol/L) had no effect on the expression of luciferase, indicating that the miRNA effect was specifically mediated by the sequences appended to the 3′ end of the luciferase coding region (Fig. 3A). Moreover, IGF-IR, mTOR, and raptor 3′-UTRs deleted of their predicted miR-99a/miR-100 binding sites were insensitive to repression by a high concentration (50 nmol/L) of miR-100 precursor that efficiently repressed the wild-type constructs (Fig. 3B).
The phosphorylated, active form of the mTOR kinase (phospho-Ser2448) is specifically enriched in mitotic tumor adrenocortical cells
Considering the potential effect of miRNAs of the miR-100 family in modulating the expression of proteins involved in mTOR signaling, we explored the role of this pathway in regulating the proliferation of ACT cells. We started by studying the cellular localization of mTOR and its Ser2448-phosphorylated, active form. Whereas mTOR is distributed in the cytoplasm of ACT H295R cells, phospho-mTOR is strikingly enriched in mitotic cells (Supplementary Fig. S4). In prophase, a bright phospho-mTOR staining appeared among condensed chromosomes, which at metaphase partly colocalized with the mitotic spindle, being also present in a bright dot-like pattern in the cytoplasm of mitotic cells (Fig. 4). Starting from anaphase, the phospho-mTOR signal moved to the midzone and progressively concentrated at the midbody in the cleavage furrow during telophase and cytokinesis (Fig. 4).
Blockage of mTOR activity inhibits ACT cell proliferation in vitro and xenograft growth
Because of the effects of mTOR signaling on cell growth and proliferation, mTOR inhibitors derived from the macrolide rapamycin are being used in the chemotherapy of different types of cancer (16). Because mTOR signaling is activated in ACT, we studied the effect of the mTOR inhibitor RAD001 (everolimus) on the proliferation of ACT cells H295R and SW-13. The drug significantly inhibited proliferation of both cell lines, showing a more potent effect on SW-13 than on H295R cells (IC50, 10−8.9 mol/L versus 10−8.1 mol/L; Fig. 5A). RAD001 also inhibited the proliferation of primary childhood ACT cells (14), with an IC50 of 10−9.2 mol/L, and of the HAC15 cell line, derived from a pediatric adrenocortical carcinoma (17), with an IC50 of 10−8.6 mol/L (Fig. 5A).
We then measured the capacity of RAD001 to inhibit the growth of H295R cells injected as xenografts into nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice. The drug (10 mg/kg/d) significantly inhibited xenograft growth compared with placebo treatment, without inducing detectable side effects on mice (Fig. 5B). A lower RAD001 dose (3 mg/kg/d) was not active (data not shown). Tumor weight and the number of mitoses per high-power microscopic field were also significantly lower in RAD001-treated animals (Supplementary Fig. S5A and B). As expected, RAD001 treatment potently reduced phospho-RPS6 expression in the xenografts and also reduced blood vessel number and extension (Supplementary Fig. S5C). In addition, we could detect thrombogenesis in tumor vasculature (Supplementary Fig. S5C), as described previously in other animal tumor models (18).
Here, we have shown that the expression of a distinct set of miRNAs is differentially regulated in childhood ACT compared with normal adrenal cortex. Interestingly, unsupervised clustering revealed that miRNA profiles could distinguish between two groups of ACT in our series, one of which (T1 in Fig. 1) was associated with the risk of relapse. These data need to be validated on a larger series of cases. Consistently with previous results in other types of cancer (19), the majority of the differentially expressed miRNAs in ACT was downregulated compared with normal adrenal. Conversely, one miRNA that was found to be highly upregulated in ACT was miR-483-3p. The gene encoding this miRNA lies within an intron of the IGF-II gene in 11p15, which is overexpressed at high frequency in childhood ACT (4, 5). Further studies are required to assess the potential role of miR-483-3p overexpression in ACT pathogenesis. Three among the differentially expressed miRs (miR-503, miR-214, and miR-375) were also identified in a study of the miRNA expression profiles in adult ACT, with miR-503 and miR-375 displaying a significantly different expression in carcinomas compared with functioning adenomas (20). Moreover, another recent study described downregulation of miR-195 (which is also downregulated in ACT in our series) and upregulation of miR-483-5p (which is derived from the same precursor as miR-483-3p) as potential prognostic markers in adult ACT (21).
We focused our functional analysis on miR-99a and miR-100, which are among the most highly downregulated miRNAs in ACT. They share the same seed sequence, which suggests that they can regulate a common set of target mRNAs. Here, we show that in ACT cells these miRNAs regulate IGF-IR and mTOR signaling cascades at multiple levels through modulation of the expression of key proteins implicated in those pathways (Fig. 6). Previous reports have shown the importance of the IGF–IGF-IR pathway in the regulation of adrenocortical cancer cell proliferation and the efficacy of targeting this pathway in preclinical models of the disease (6, 7, 15). Here, we have shown that IGF-IR levels are upregulated in ACT and that miR-99a/miR-100 can regulate its expression acting on a target site in the 3′-UTR of its mRNA. These data are consistent with previous findings showing overexpression of IGF-IR mRNA in ACT (6). In these tumors, very high IGF-II mRNA expression often does not translate into the production of a biologically active protein (10). High IGF-IR expression is then likely to play a pivotal role in the activation of the IGF pathway in these tumors.
Studies analyzing other types of tumors revealed that miR-99a and/or miR-100 are also significantly downregulated in ovarian and prostate cancer and in head and neck carcinomas (22–28). In addition, the gene encoding miR-99a lies in a region in chromosome 21q21 harboring a putative tumor suppressor gene in lung cancer (29). It is then tempting to speculate that those miRNAs play a role in the modulation of IGF-mTOR signaling also in other types of tumors.
mTOR signaling is closely interconnected with the IGF pathway because it can be activated by upstream IGF receptor signaling. mTOR activity has an essential role in the regulation of various essential cellular processes (30, 31). The protein kinase mTOR (FRAP1) has been identified as a target for rapamycin bound to FKBP12. It exists in the cell in two distinct complexes: mTORC1 and mTORC2. The two complexes have different downstream effectors and only mTORC1 is directly sensitive to rapamycin inhibition. Nevertheless, it is known that in some cell lines prolonged treatment with rapamycin also perturbs mTORC2 assembly and inhibits Akt activity (32). The relationship between the mTOR pathway and cancer is complex because, depending on the cellular context, rapamycin treatment may either inhibit cell growth or activate the oncogenic Akt kinase (33, 34). In any case, mTOR inhibition seems particularly promising as a therapeutic tool for cancers characterized by an activated Akt pathway and a relevant angiogenic component, such as ACTs (7, 35, 36). Our data show that everolimus, a rapamycin analogue, efficiently inhibits ACT cell proliferation in vitro and in vivo. In the clinical setting, it will be interesting to test the efficacy of treatments combining IGF-IR and mTOR inhibitors for the therapy of advanced adrenocortical cancer.
Here, we have shown for the first time that the mTOR and raptor proteins are direct targets for miR-99a/miR-100 inhibition in cancer cells. Interestingly, an inhibitory effect of these miRNAs on mTOR and raptor expression was also shown during cytomegalovirus infection (37). In addition, we have revealed an unexpected mitotic localization of the active phosphorylated mTOR form (Ser2448) in adrenocortical cancer cells. Phospho-mTOR staining starts to be dramatically increased at prophase among condensing chromosomes and transfers at the mitotic spindle during metaphase, being localized at the midbody during telophase and cytokinesis. Recently, a similar pattern of subcellular localization of phospho-mTOR has been described in ovarian granulosa (38) and breast cancer cells (39). These findings suggest that active mTOR signaling may have a previously unrecognized role in the regulation of mitosis and cytokinesis through phosphorylation of still undefined substrates. This hypothesis is supported by the finding that in yeast the TOR protein has been shown to affect microtubule stability and morphology and function of the mitotic spindle (40).
In the adrenal gland, activation of the mTOR pathway has also been described in a particular form of benign adrenocortical neoplasm, primary pigmented nodular adrenocortical disease (PPNAD; ref. 41). Interestingly, a recent analysis of miRNA expression in PPNAD showed that miR-100 is one of the most significantly downregulated miRNAs (42). These data suggest that a link between activated mTOR signaling and miR-100 downregulation may also exist in other types of adrenocortical neoplastic diseases.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank B. Mari and S. Demolombe for discussions and the gift of material, C. Ferre for help with microarray scanning, W. Rainey for the gift of HAC15 cells, F. Aguila for artwork, Novartis for the gift of RAD001, and TrGET preclinical assay platform (Canceropole PACA) for the xenograft assays.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Grant Support: NIH/National Cancer Institute grants CA21765 and GM083159 and American Lebanese Syrian Associated Charities (G.P. Zambetti) and Institut National du Cancer, Agence Nationale de la Recherche, Conseil Général 06, CNRS (LIA NEOGENEX) and FP7 ENS@T-CANCER (E. Lalli).
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received October 29, 2009.
- Revision received March 3, 2010.
- Accepted March 23, 2010.
- ©2010 American Association for Cancer Research.