|Home | About | Journals | Submit | Contact Us | Français|
The RNA-binding protein LIN28A regulates the translation and stability of a large number of mRNAs as well as the biogenesis of certain miRNAs in embryonic stem cells and developing tissues. Increasing evidence indicates that LIN28A functions as an oncogene promoting cancer cell growth. However, little is known about its molecular mechanism of cell cycle regulation in cancer. Using tissue microarrays, we found that strong LIN28A expression was reactivated in about 10% (7.1–17.1%) of epithelial tumors (six tumor types, n = 369). Both in vitro and in vivo experiments demonstrate that LIN28A promotes cell cycle progression in cancer cells. Genome-wide RNA-IP-chip experiments indicate that LIN28A binds to thousands of mRNAs, including a large group of cell cycle regulatory mRNAs in cancer and embryonic stem cells. Furthermore, the ability of LIN28A to stimulate translation of LIN28A-binding mRNAs, such as CDK2, was validated in vitro and in vivo. Finally, using a combined gene expression microarray and bioinformatics approach, we found that LIN28A also regulates CCND1 and CDC25A expression and that this is mediated by inhibiting the biogenesis of let-7 miRNA. Taken together, these results demonstrate that LIN28A is reactivated in about 10% of epithelial tumors and promotes cell cycle progression by regulation of both mRNA translation (let-7-independent) and miRNA biogenesis (let-7-dependent).
LIN28 is an evolutionarily conserved RNA-binding protein with two RNA-binding domains (a cold shock domain and a retroviral-type CCHC zinc finger motif), which was first characterized as a critical regulator of developmental timing in Caenorhabditis elegans (1, 2). The mammalian genome encodes two homologues of the C. elegans lin-28 genes (LIN28A and LIN28B) (3, 4), which may control gene expression by distinct molecular mechanisms (5). They are important in processes such as embryogenesis (6), skeletal myogenesis (7), germ cell development (8, 9), neurogliogenesis (10, 11), differentiation (12, 13), lymphopoiesis (14), and glucose metabolism (15). Genome-wide association studies have implicated the LIN28B locus in controlling both height and the timing of menarche in humans (16–20). This finding has been successfully phenocopied in a transgenic mouse model with an inducible LIN28A gene (21). Increasing evidence suggests that LIN28 may also be a master regulator controlling the pluripotency of embryonic stem (ES)4 cells (22–25). LIN28A, together with OCT4, SOX2, and NANOG (the “reprogramming factors”), can reprogram somatic cells to induced pluripotent stem cells (26). Several reports have demonstrated that LIN28A binds to mRNAs, regulating their translation (7, 23, 25, 27, 28) and/or stability (29). In addition, LIN28 can bind to the terminal loops of the let-7 miRNA precursor, thereby blocking the processing of let-7 into its mature form (10, 11, 30–39). Importantly, the expression of LIN28A/LIN28B is highly restricted to ES cells and developing tissues, and expression dramatically decreases as differentiation progresses (2, 3, 7, 8, 11, 21, 40, 41). In human tumors, LIN28A/LIN28B expression is up-regulated/reactivated (4, 42–51) and may function as an oncogene promoting malignant transformation (42, 43, 52, 53), inducing metastasis (43, 46, 52–54), regulating inflammation (5, 43), and maintaining the cancer stem cells (43, 45, 55–57). Clinical studies have indicated that higher levels of LIN28A/LIN28B expression are associated with poor clinical outcomes (44, 58, 59) and that LIN28 family polymorphisms may influence susceptibility to ovarian (60) and breast (61) cancers.
It has been well demonstrated that LIN28A promotes ES cell proliferation (22, 23, 28) via regulation of the G2/M transition of the cell cycle (23). Recent studies in cancer also found that knocking down LIN28A expression in cancer cell lines can remarkably reduce cancer cell viability and inhibit cell growth in vitro (42, 45, 62). These findings led to a hypothesis that in cancer cells, LIN28A may have a similar function as in ES cells to regulate the cell cycle. However, primary transcripts of the let-7 family are present at high levels but are not processed to their functional mature forms in ES cells (63). Therefore, the function of LIN28A in cell cycle regulation of ES cells is probably miRNA-independent (23). In contrast, mature forms of the let-7 family are always at detectable levels in cancer cells, although they are dramatically down-regulated compared with the normal corresponding tissues. In the present study, we report that, as in ES cells, LIN28A promotes proliferation in cancer cells; however, this function was mediated by a distinct cellular mechanism (i.e. regulating the G0/G1 transition of the cell cycle). Genome-wide RNA-IP-chip experiments demonstrate that LIN28A binds to thousands of mRNAs, including a large group of genes involved in cell cycle regulation, such as CDK2. In addition, we also found that LIN28A regulates CCND1 and CDC25A expression and that this is mediated via the inhibition of let-7 miRNA biogenesis.
Breast tumor specimens (benign tumors, n = 14; ductal carcinomas in situ (DCIS), n = 14) were collected at the University of Turin (Turin, Italy). All tumors were from primary sites and were immediately snap-frozen and stored at −80 °C. Specimens were acquired and processed under procedures approved by the local institutional review board and were compliant with the Health Insurance Portability and Accountability Act.
Cancer cell lines were purchased from the American Type Culture Collection (ATCC) and the Division of Cancer Treatment and Diagnosis Tumor/Cell Line Repository. All cancer cell lines were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (Invitrogen). Four independent, immortalized, human ovarian surface epithelial (IOSE) cell lines were generously provided by Dr. Nelly Auersperg. IOSE cells were cultured 1:1 in Medium 199/MCDB 105 (Sigma), supplemented with 15% FBS. The mouse embryonic stem cell lines R1 (ATCC) and C57BL/6J-693 (Jackson Laboratory) were maintained on gelatin-coated flasks with mitomycin C-treated mouse embryo fibroblasts in DMEM containing 15% ES cell FBS (Invitrogen), 2 mm l-glutamine (Invitrogen), 1% minimum Eagle's medium nonessential amino acids (Invitrogen), 1% penicillin-streptomycin, 0.1 mm 2-mercaptoethanol (Invitrogen), and 1000 units/ml mouse leukemia inhibitory factor (Chemicon).
Tissue microarray slides were purchased from the Tissue Array Network. Six different human tumor tissue microarrays were used to characterize LIN28A expression in epithelial tumors. Each cancer patient was also represented with at least two core tissue biopsies.
Immunohistochemistry was performed using the VECTASTAIN ABC Kit as outlined by the manufacturer (Vector). The following primary antibodies were used in this study: rabbit anti-human LIN28A (1:4,000; Abcam); mouse anti-human Ki67 (1:400; DAKO); rabbit anti-human CDK2 (1:60; Cell Signaling); rabbit anti-human CDC2 (1:300; Cell Signaling); rabbit anti-human CDC20 (1:100; Cell Signaling); rabbit anti-human CCND1 (1:2,000; Cell Signaling); or rabbit anti-human CDC25A (1:200; Cell Signaling). Antibodies were incubated overnight at 4 °C, the immunoreaction was visualized with 3,3′-diaminobenzidine, and the images were collected and analyzed using Image-Pro Plus 4.1 software (Media Cybernetics).
Cells were lysed in mammalian protein extraction reagent (Pierce). After quantification using a BCA protein assay kit (Pierce), 15 μg of total protein was separated by 10% SDS-PAGE under denaturing conditions and transferred to PVDF membranes (Millipore). Membranes were blocked in 5% nonfat milk (Bio-Rad) and then incubated with an anti-LIN28A primary antibody (1:10,000; Abcam), CDK2 (1:1,000; Cell Signaling), CDC2 (1:1,000; Cell Signaling), CDC20 (1:500; Cell Signaling), CCND1 (1:1,000; Cell Signaling), or CDC25A (1:1,000; Cell Signaling), followed by incubation with a secondary antibody conjugated with horseradish peroxidase (HRP) (1:10,000; Amersham Biosciences) together with an HRP-conjugated primary antibody for β-actin (1:10,000; Sigma). Immunoreactive proteins were visualized using LumiGLO chemiluminescent substrate (Cell Signaling).
Two individual lentiviral shRNA clones targeting human LIN28A (pLKO.1; TRCN0000102576 and TRCN0000102579) were purchased from Open Biosystems. Enhanced GFP shRNA (SHC005) and non-target shRNA (SHC002) were used as a control. The pSin-EF2-LIN28A lentiviral expression vector was purchased from Addgene, and the pSin-EF2 empty vector was used as control. Lentiviral vector and packaging vectors were transfected into the packaging cell line 293T (ATCC) using the FuGene6 transfection reagent (Roche Applied Science). The medium was changed 8 h post-transfection, and the medium containing lentivirus was collected 48 h later. Tumor cells were infected with lentivirus in the presence of 8 μg/ml Polybrene (Sigma). Puromycin (Sigma) was used to select stable cell clones.
Total RNA was isolated from 100–500 mg of frozen tissue or 1 × 106 cultured cells, using TRIzol reagent (Invitrogen). The quality and quantity of the isolated RNA was analyzed using a Bioanalyzer 2100 system (Agilent).
Total RNA was reverse-transcribed using a High Capacity RNA-to-cDNA kit (Applied Biosystems) according to the manufacturer's instructions. cDNA was quantified by real-time PCR on an ABI Prism 7900 Sequence Detection System (Applied Biosystems) using the primers listed in Table 1. PCR was performed using SYBR Green PCR Core reagents (Applied Biosystems) according to the manufacturer's instructions. PCR amplification of the housekeeping gene GAPDH was performed for each sample as a control for sample loading and to allow normalization across samples.
Cells (2 × 103) were resuspended in RPMI 1640 containing 10% FBS with 0.3% agarose (Invitrogen) and layered on top of 0.4% agar in medium supplemented with 20% FBS on 6-well plates. The plates were incubated at 37 °C in a humid atmosphere of 5% CO2. Culture medium was replaced weekly. After 3–4 weeks, cell colonies were stained by crystal violet (Sigma) and counted under a microscope. Only cell colonies containing more than 50 cells were counted.
6–8-week-old female nude mice were obtained from Jackson Laboratory. Subconfluent A2780 cells were trypsinized and suspended in phosphate-buffered saline (PBS). A total volume of 0.1 ml containing 2.5 × 106 cells was injected subcutaneously into the mouse flank. Approximately 10 days later, tumors were detectable, and tumor size was measured using a Vernier caliper. Tumor volumes were calculated using the formula, V = ½ (L × W)2, where L is length (longest dimension) and W is width (shortest dimension) of the tumor. Moribund animals were euthanized according to the protocols of the University of Pennsylvania. Tumor growth rates in the xenograft experiment were evaluated by fitting a linear mixed effects model on the log10-transformed tumor volume with days, experiment indicator (LIN28A shRNA versus control shRNA), and interaction between days and experiment included as independent variables.
Cell cycle analysis was performed using a BrdU cell proliferation kit (Roche Applied Science) according to the manufacturer's instructions. BrdU-positive cell populations were analyzed by flow cytometry. The data were analyzed using FlowJo software (Tree Star).
5 × 106 A2780 or ES cells were lysed for 15 min on ice in a polysome lysis buffer containing 100 mm KCl, 5 mm MgCl2, 10 mm HEPES, pH 7.0, 0.5% Nonidet P-40 detergent supplemented with fresh 1 mm dithiothreitol (DTT), 100 units/ml RNase Out (Invitrogen), 400 μm vanadyl-ribonucleoside complex (New England Biolabs), and a protease inhibitor mixture (Sigma). The cell lysate was further diluted (1:10) with NT2 buffer containing 50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1 mm MgCl2, 0.05% Nonidet P-40 supplemented with fresh 200 units/ml RNase Out, 400 μm vanadyl-ribonucleoside complex, 1 mm DTT, 20 mm EDTA, and a protease inhibitor mixture. The insoluble particles in the lysate were removed by centrifugation at 15,000 × g for 15 min at 4 °C. LIN28A antibody (1:125; Abcam) or control IgG was added to protein A-Sepharose beads (Sigma), which were preincubated in 5% bovine serum albumin (BSA)-NT2 buffer for 1 h at 4 °C. After binding antibody to the protein A-Sepharose beads by gentle rotation for 4 h at 4 °C, the beads were washed four times in cold NT2 buffer and added to cell lysates (10 μl of beads/ml of lysate). Immunoprecipitation was performed by gentle rotation overnight at 4 °C, and then the immunoprecipitated complex was washed four times in NT2 buffer and resuspended in 100 μl of NT2 buffer containing 30 μg of proteinase K (Qiagen) to release the ribonucleoprotein complex. TRIzol reagent (Invitrogen) was used to extract RNA from the immunoprecipitates.
Total RNA (300 ng) was amplified using an Ambion WT expression kit (Ambion) according to the manufacturer's instructions. Reverse transcription was performed using 10 μg of RNA, and 5.5 μg of the resulting cDNA was hybridized to a Human Gene Array 1.0 ST array (Affymetrix) according to the manufacturer's protocol. After hybridization overnight, the microarray chips were terminally labeled and scanned using an Affymetrix GeneChip 3000 7G scanner. The normalized microarray data were analyzed using GeneSpring software (Agilent), Microarray Software Suite 4, BRB-ArrayTools version 3.8, and GenePattern software.
LIN28A siRNA and control oligonucleotides were purchased from IDT, and pre-miR miRNA precursor and control oligonucleotides were purchased from Ambion. Transfections were performed using the LipofectamineTM RNAiMAX transfection reagent (Invitrogen) following the manufacturer's instructions, and then cells were incubated in the medium containing the transfection mixture for 72 h.
FuncAssociate was used to search for gene ontology (GO) attributes, and TargetScan was used to predict let-7 targets. Ingenuity pathway analysis software was used to analyze molecular pathways and networks from the microarray data.
Statistical analysis was performed using the SPSS and SAS statistical software packages. All results were expressed as mean ± S.D., and p < 0.05 indicated statistical significance.
Expression of LIN28A is highly restricted to ES cells and developing tissues and dramatically decreases as differentiation progresses (2, 3, 7, 8, 11, 21, 40, 41). It has been reported that both expression of LIN28A and LIN28B are reactivated in human cancers (4, 42–47). However, the expression and distribution of LIN28A in different types of epithelial tumors is still unknown. In this study, we chose tumor tissue arrays from six types of common epithelial cancers (breast (n = 69), lung (n = 70), ovarian (n = 70), colon (n = 70), liver (n = 70), and pancreatic cancer (n = 20)) to explore LIN28A expression. Strong LIN28A expression was detected in ~10% (7.1–17.1%) of cases in all six tumor types (Fig. 1): breast (10.1%), colon (8.6%), liver (7.1%), lung (7.1%), ovarian (17.1%), and pancreatic cancer (15.0%). LIN28A-positive cells were only found in tumor islet and not in host stroma cells. To confirm that LIN28A was indeed expressed in tumor cells, we examined LIN28A expression in a collection of human breast (n = 17) and ovarian (n = 19) cancer cell lines by real-time RT-PCR and Western blots. Consistent with the immunohistochemical staining results in tissue arrays, strong LIN28A mRNA and protein expression was detected in 2 of 17 (11.8%) breast cancer cell lines (HCC202 and T47D) and 2 of 19 (10.5%) ovarian cancer cell lines (A2780 and IGROV1) (Fig. 2A). Importantly, LIN28A was not detectable in cultured normal human breast epithelial cells (n = 2) or in normal ovarian surface epithelial cells (n = 4). To further confirm this finding, we examined LIN28A mRNA and protein expression in benign breast tumors (n = 14) and DCIS (n = 14) by real-time RT-PCR and Western blots. LIN28A expression was detected in 2 of 13 DCIS specimens but not in any benign breast tumor specimens (Fig. 2B). Taken together, these data show that expression of LIN28A was reactivated in ~10% epithelial tumors and that LIN28A expression did not show any tumor type specificity among the common human epithelial cancers investigated in this study.
Using two independent lentiviral shRNA vectors, LIN28A was specifically knocked down in A2780, IGROV1, and T47D cell lines (Fig. 3A). We found that blocking endogenous LIN28A expression significantly inhibited cell growth in these cells under normal culture conditions (Fig. 3B). We also examined the effect of knocking down LIN28A on anchorage-independent growth using a soft agar colony-forming assay. We found that the colony forming ability of A2780 and T47D cells with knocked down LIN28A was significantly reduced compared with control cells (Fig. 3, C and D). IGROV1 cells could not form colonies in soft agar. Finally, we transplanted both knocked down LIN28A and control A2780 cells into nude mice. The xenograft tumors were followed for 25 days, and the function of the shRNA in vivo was confirmed by Western blots and real-time RT-PCR (Fig. 4, A and B). We found that knocking down LIN28A significantly inhibited tumor growth in vivo compared with control cells (n = 8, p = 0.0038, Fig. 4C). Thus, these studies demonstrate that LIN28A promotes tumor cell growth both in vitro and in vivo.
To determine if LIN28A affects the cell cycle in cancer cells, we chose two ovarian cancer cell lines, A2780 and IGROV1, to perform functional studies. Using a BrdU incorporation assay, we found that stably knocking down LIN28A expression in both A2780 and IGROV1 cell lines significantly promoted cell cycle progression (Fig. 5, A and B). To exclude the possibility that these results were due to off-target effects of the shRNA, the above experiment was repeated using two independent siRNA sequences, and similar results were observed (Fig. 5C). The proliferation status in LIN28A knockdown and control A2780 xenograft tumors was also examined by immunohistochemical staining of Ki-67, which is present during active phases of the cell cycle but is absent from resting cells. We found that the proliferation fraction in LIN28A knockdown tumors was remarkably reduced compared with control tumors (Fig. 5D). Finally, control and LIN28A knockdown cells (IGROV1) were arrested from cell cycle by serum starvation and then released into the cell cycle by the addition of 10% serum. Both RNA and protein samples were harvested at 0 and 19 h after the serum addition. We found that both mRNA and protein levels of CCND1 were significantly lower in LIN28A knockdown cells compared with control cells at 0 and 19 h after serum addition (supplemental Fig. S1). Importantly, during cell cycle re-entry (e.g. G0/G1 transition), the acceleration rate of CCND1 in both mRNA and protein levels in LIN28A knockdown cells was significantly slower than in control cells (supplemental Fig. S1). Taken together, the above results suggest that LIN28A promotes proliferation by regulating cell cycle progression, particularly at the G0/G1 transition phase in tumor cells.
The LIN28A protein contains two RNA-binding domains, which bind to target mRNAs and regulate their translation (7, 23, 25, 27, 28) and/or stability (29). A recent deep sequencing study indicated that LIN28A binds to more than 1,000 transcripts and regulates translation of these mRNAs in human ES cells (28). To understand the mechanism by which LIN28A regulates cell cycle progression in tumor cells, we performed an RNA-IP-chip analysis. The LIN28A-RNA complexes were immunoprecipitated using a LIN28A-specific antibody, and a gene expression microarray was used to profile the bound transcripts in A2780 cells (Fig. 6A). RNAs immunoprecipitated by IgG were used for the control microarray. Based on previously published studies from other groups who performed similar experiments in human ES cells (28), we chose a 4-fold cut-off threshold for our A2780 microarray study. We found that 1,707 transcripts were enriched in the LIN28A RNA-IP microarray compared with the control IgG microarray (Fig. 6B and supplemental Table S1). We also performed similar RNA-IP microarrays using ES cells, where LIN28A is expressed at physiological levels and conditions. When we combined the above two microarray results, we found that about half of the transcripts (n = 801; Fig. 6B and supplemental Table S2) identified by the A2780 RNA-IP microarray were also enriched in the ES cell RNA-IP microarray. This indicates that LIN28A binds to a similar pattern of mRNAs in tumor cells and ES cells, although the gene expression background in these two cell types is quite different. Using these results and GO analysis, we explored signal pathways that may be regulated by LIN28A in tumor cells. As expected, one of major pathways identified was the cell cycle-associated pathway (Fig. 6C and supplemental Table S3). Interestingly, when we used the combination gene list (n = 801 transcripts enriched in both A2780 and ES cells) to analyze signal pathways by GO, the rank of cell cycle-associated pathways was considerably higher (Fig. 6C and supplemental Table S4). This suggests that one of the major functions of LIN28A shared by both tumor and ES cells is regulation of the cell cycle. A total 15 genes directly associated with the cell cycle, including cyclins, cyclin-dependent kinases (CDK), and cell division control proteins (CDC), were identified in the A2780 RNA-IP microarray (Fig. 6D), and nine of these were shared by the ES RNA-IP microarray (shown in green in Fig. 6D). Most importantly, CDK2, a master regulatory protein in cell cycle progression, was identified in both our A2780 and ES RNA-IP microarrays. These high throughput microarray results were confirmed by real-time RT-PCR performed in two tumor cell lines, including the A2780 cell line. All five cell cycle-associated mRNAs identified by the microarrays (CDK2, CDC2, CDC20, CCNA2, and CCNB2) were validated by real-time RT-PCR in both A2780 and IGROV1 cells (Fig. 7A). Because it has been demonstrated that LIN28A binds to polysomes and stimulates the translation of LIN28A-binding mRNAs (7, 23–25, 27–29), we also tested whether the increase in LIN28A led to an increase in the level of these cell cycle-associated proteins, by comparing the expression of CDK2, CDC2, and CDC20 proteins between control and LIN28A knockdown cells. Using Western blots, we showed that blocking endogenous LIN28A expression in A2780 and IGROV1 cells remarkably reduced the levels of CDK2, CDC2, and CDC20 proteins (Fig. 7B). We also examined CDK2, CDC2, and CDC20 expression in our LIN28A shRNA knockdown xenograft model. We found that knocking down LIN28A dramatically decreased CDK2, CDC2, and CDC20 expression in vivo (Fig. 7C). Taken together, this indicates that LIN28A binds to and regulates a large number of cell cycle-regulating proteins in tumor cells.
In order to determine if LIN28A can regulate gene expression at the transcriptional level in cancer cells, we performed gene expression microarrays to compare the transcriptional differences between LIN28A knockdown and control tumor cells (supplemental Fig. S2A). However, we found that very few mRNAs (>1.5-fold, n = 152; supplemental Table S5) differed between these two groups (LIN28A was decreased 2.4-fold in LIN28A knockdown cells compared with control cells, shown by a blue spot in supplemental Fig. S2A). It has been reported that LIN28A can bind to the terminal loops of the let-7 miRNA precursor, thereby blocking the processing of let-7 into its mature form (10, 11, 30–39). In addition, several let-7 target genes, such as CCND1 (64), CCND2 (65), CCND3 (65), CCNA2 (64), CDK4 (64, 65), CDC25A (65), and CDC34 (66), have been shown to function as regulators of the cell cycle. Therefore, we explored whether LIN28A could regulate the cell cycle in a let-7-dependent manner. We first compared down-regulated genes (n = 153) in our microarray with the potential let-7 targets predicted by TargetScan (supplemental Table S6 (n = 819) and supplemental Fig. S2, B and C). We found seven genes that were shared by these two lists, including CCND1, another key regulator in cell cycle progression. The repression of transcription by miRNA is typically relatively mild, and some miRNAs may repress gene expression mainly at the level of translation. Also, gene expression microarrays may not be sensitive enough to detect mild mRNA expression changes. Therefore, although only CCND1 was identified here, there could be other let-7 targets that are regulated by LIN28A in tumor cells. To further test whether LIN28A can regulate the cell cycle via a let-7-dependent mechanism, we selected two genes for further study: 1) CCND1, which was identified in our microarray study and also has been reported by others as a let-7 target (64), and 2) CDC25A, which, although not identified in our original microarray study, has been reported to be repressed by let-7 (65). We examined CCND1 and CDC25A mRNA expression in LIN28A knockdown and control cells (A2780 and IGROV1), using real-time RT-PCR, and found that blocking LIN28A reduced CCND1 and CDC25A mRNA expression (Fig. 8A). This indicates that 1) LIN28A does indeed regulate let-7 targets in tumor cells, and 2) gene expression microarrays may not be sensitive enough to detect regulation at the mRNA level when the effects are modest. Next, we transfected let-7 and control mimics into the above cell lines and found that enforced let-7 expression remarkably decreased CCND1 and CDC25A expression (Fig. 8B). The expression levels of CCND1 and CDC25A affected by LIN28A shRNA and the let-7 mimic were examined using Western blots (Fig. 8C). Finally, the changes in the levels of the CCND1 and CDC25A proteins were also confirmed in vivo by immunohistochemistry (Fig. 8D). Taken together, these results indicate that LIN28A directly regulates the translation of cell cycle-associated genes (such as CDK2) via a let-7-dependent mechanism, which is mediated by at least CCND1 and CDC25A in tumor cells.
Expression of LIN28A is highly restricted to ES cells and developing tissues and dramatically decreases as differentiation progresses (2, 3, 7, 8, 11, 21, 40, 41). In adult mammalian tissues, we detected strong LIN28A mRNA expression mainly in the testis (data not shown), which is consistent with recent studies demonstrating that LIN28A plays a critical role in male germ cell development (8, 9). In the present study, we observed the reactivation of LIN28A expression in ~10% of the epithelial tumors we examined. Notably, in six types of common epithelial cancers investigated in this study, we observed a reactivation/up-regulation of LIN28A expression that was not tissue-specific. Therefore, LIN28A could serve as a common biomarker or a therapeutic target for a subset of epithelial cancers. This also suggests that there may be similar mechanisms by which LIN28A expression is reactivated during epithelial tumorigenesis. Studies have excluded the possibility that LIN28A overexpression in tumors is due to an alteration of the LIN28A DNA copy number (42). Based on recent studies (42, 67), there are at least two mechanisms that can lead to the repression of LIN28A expression in adult tissues that may be partially lost in human tumors: epigenetic silencing (42) and miRNA suppression (67). In fact, we have previously identified four LIN28A regulatory miRNAs that were globally down-regulated in human tumors (67). In addition, overexpression of LIN28 activators, such as the Myc oncogene (47, 54, 68), may serve as another mechanism to deregulate LIN28A expression in epithelial tumors.
Several studies have reported that LIN28A may serve as an oncogene, promoting malignant transformation (42, 43, 52, 53), inducing metastasis (43, 46, 52–54), regulating inflammation (5, 43), and maintaining the cancer stem cells (43, 45, 55–57). However, the function and molecular mechanism of LIN28A in cell cycle regulation of cancer cells is still largely unknown. Because an important function of LIN28A in ES cells is cell cycle regulation (23), we examined whether LIN28A could promote cancer cell growth mediated by cell cycle regulation. We first demonstrated that blocking endogenous LIN28A expression did significantly inhibit tumor cell proliferation in vitro or in vivo. Furthermore, we demonstrated that LIN28A dramatically promoted the G0/G1 transition in tumor cells but probably not the G2/M transition, although the latter is the major regulatory function of LIN28A in ES cells (23). This suggests that LIN28A may control different cell cycle regulators in cancer cells compared with ES cells. Using RNA-IP microarrays, CDK2 was identified as one of the targets of LIN28A. We also demonstrated that LIN28A binds to CDK2 mRNA and stimulates its translation in tumor cells. Finally, we demonstrated that LIN28A regulates CCND1 and CDC25A levels to control the cell cycle via the suppression of let-7 expression, and therefore, LIN28A regulates cell cycle progression in cancer cells via a let-7-dependent mechanism. The different molecular function of LIN28A in cell cycle regulation between ES and cancer cells may be due to the following: 1) unlike cancer cells, ES cells lack an R point because pRB is constitutively inactivated; 2) in ES cells, primary transcripts of the let-7 family are present at high levels but are not processed to its functional mature forms (63). It is still unclear how LIN28A regulates cell cycle in differentiated adult tissues, such as testis.
In this study, we observed that LIN28A can bind to a large number of mRNA transcripts in cancer cells, suggesting that LIN28A may have multiple functions in the development and progression of human cancers, which is in alignment with its suggested roles in malignant transformation, metastasis, and maintenance of cancer stem cell populations, discussed above. Although the detailed molecular mechanisms underlying the functions of LIN28A in cancer cells are not yet defined, increasing evidence suggests that the functions of LIN28A in tumors are mediated by its ability to bind to RNA and regulate translation or stability as well as miRNA biogenesis. In ES cells and under some physiological conditions, it has been demonstrated that LIN28A binds to mRNA and transfers it to polysomes to stimulate translation or regulate mRNA stability, and in agreement with these observations, we found that 1) LIN28A binds to ~1,000–2,000 mRNA transcripts in tumor cells (similar to what has been observed in ES cells), and 2) knocking down the expression of LIN28A did not significantly affect mRNA transcription. Notably, we found that about half of the LIN28A-binding mRNAs identified by our microarray studies in tumor cells were the same as the LIN28A-binding mRNAs found in ES cells (Fig. 6C), suggesting that LIN28A may regulate a large group of genes that control the cell cycle, self-renewal, and differentiation in tumor cells. It is likely that the function of LIN28A depends on its cellular context because, in some tumors, the functions of LIN28A after genetic or epigenetic reactivation are largely dependent on the background mRNA expression in each type of cancer.
Similar to its function in ES cells, in cancer cells, LIN28A can specifically bind to the precursors of selected miRNAs, such as the let-7 family, and regulate their biogenesis, as discussed above. let-7 is one of the first miRNAs to be identified as a tumor suppressor (69), and it targets numerous oncogenic proteins, including some associated with the cell cycle (64–66). In this study, we found that LIN28A indirectly regulates CCND1 and CDC25A, two key regulators of the cell cycle. Because the potential regulatory circuitry afforded by specific miRNAs may be enormous, with one miRNA potentially regulating hundreds of transcripts, we believe that the downstream targets of the LIN28A/let-7 loop are probably not restricted to CCND1 and CDC25A. For example, Johnson et al. (65) have reported that other cell cycle regulators, such as CCND2 and CCND3, are controlled by let-7 in lung cancer (65).
*This work was supported, in whole or in part, by National Institutes of Health, NCI, Grant R01-CA142776 (to L. Z.) and Ovarian Cancer SPORE P50-CA83638–7951 Project 3 (to L. Z.). This work was also supported by Department of Defense Grant W81XWH-10-1-0082 (to L. Z.) and the Ovarian Cancer Research Fund Tilberis Scholar Award (to L. Z.).
This article contains supplemental Tables S1–S6 and Figs. S1 and S2.
4The abbreviations used are: