MYC Protein Inhibits Transcription of the MicroRNA Cluster MC-let-7a-1∼let-7d via Noncanonical E-box*

Background: The three microRNAs from miRNA cluster MC-let-7a-1∼let-7d are often down-regulated in human cancers. Results: We functionally characterized the promoter and investigated the transcriptional regulation of MC-let-7a-1∼let-7d. Conclusion: MYC oncoprotein could inhibit the transcription of MC-let-7a-1∼let-7d via binding to a noncanonical E-box. Significance: Understanding how MC-let-7a-1∼let-7d is regulated could shed light on identifying new molecular targets and designing novel treatment for cancers. The human microRNA cluster MC-let-7a-1∼let-7d, with three members let-7a-1, let-7f-1, and let-7d, is an important cluster of the let-7 family. These microRNAs play critical roles in regulating development and carcinogenesis. Therefore, precise control of MC-let-7a-1∼let-7d level is critical for cellular functions. In this study, we first showed that the expression of these three members was significantly reduced in human hepatocellular carcinoma HepG2 cells as compared with the immortalized human liver L02 cells. We demonstrated that the MC-let-7a-1∼let-7d cluster was encoded by a single polycistronic transcript driven by a 10-kb upstream promoter, with two MYC-binding sites. Importantly, MYC inhibited MC-let-7a-1∼let-7d promoter activity via binding to the noncanonical E-box 3 downstream of the transcription start sites, whereas it enhanced promoter activity by binding to the canonical E-box 2 upstream of the transcription start sites. We found that although the binding affinity of MYC to E-box 2 was stronger than E-box 3, the binding quantum of MYC to E-box 3 was significantly higher in cancerous HepG2 cells as compared with the noncancerous L02 cells. In addition, forced expression of let-7 could reverse the MYC-mediated cell proliferation. These findings suggested that in L02 cells with a low level of MYC, MYC binds mainly to E-box 2 to enhance MC-let-7a-1∼let-7d expression. However, in HepG2 cells with an elevated MYC, the extra MYC could bind to E-box 3 to suppress the transcription of MC-let-7a-1∼let-7d and thus enable HepG2 cells to maintain a high level of MYC and a low level of let-7 microRNAs simultaneously.


MicroRNAs (miRNAs)
are short noncoding RNAs (ϳ22 nucleotides) that regulate protein expression and control diverse aspects of biology, including carcinogenesis, development, and numerous cellular processes (1). In principle, miR-NAs expression can be regulated at any step during the maturation process. This process initiates from the transcription of primary transcripts, followed by post-transcription regulations, including several processes that generate mature miRNA from primary miRNA (pri-miRNA) and then precursor miRNA (pre-miRNA) (2,3). Because miRNA genes are frequently located at the chromosomal fragile sites of cancer genomes (4), miRNAs have been considered as novel classes of oncogenes and tumor suppressors (5)(6)(7). The altered expressions of several miRNAs (including let-7, miR-9, -21, -122, -151, -221, etc.) have been reported to play pivotal roles in carcinogenesis (8 -12).
Among the various families of miRNAs, the let-7 family of miRNAs has become a prototype for miRNAs that function as a tumor suppressor; this is because they inhibit the expressions of multiple oncogenes, including RAS and MYC (13)(14)(15). They are often down-regulated in cancer cells and are potential biomarkers as well as prognostic markers that predict disease progression and response to treatment (16,17).
A recent report suggested that MYC could induce LIN28B and LIN28 expressions and subsequently repress let-7 microR-NAs expressions post-transcriptionally (30 -40). In addition, it is also suggested that MYC may bind to a conserved site upstream of the let-7a-1/let-7f-1/let-7d cluster. However, the precise binding sites, the functions, and the mechanisms involved in the transcription regulation have not been characterized (41).
In this study, we investigated the transcriptional regulation of MC-let-7a-1ϳlet-7d. We predicted the promoter region and transcription start sites (TSSs) of MC-let-7a-1ϳlet-7d by bioinformatics and then verified in human liver L02 and hepatocellular carcinoma (HCC) HepG2 cells. We further showed that MYC could directly down-regulate the expression of MC-let-7a-1ϳlet-7d at the transcription level in HCC. Two MYC-binding sites were identified. One E-box element located downstream of the TSS is critical for the let-7 transcription repression in HepG2 cells and one upstream for stimulation. Taken together, these findings suggested that the MYC oncoprotein could either inhibit or stimulate the transcription of miRNA let-7 tumor suppressors depending on its intracellular levels.

EXPERIMENTAL PROCEDURES
Genomic Features of let-7 Family-We obtained genomic coordinates of the let-7 family from the miRBase microRNA sequence data base (version 16) (42). Because the size of miRNA primary transcripts is unknown, we clustered same strand miR-NAs at 50-kb thresholds according to the miRBase. Flanking sequence data, expressed sequence tags data (ESTs) and CpG islands were obtained from the University of California, Santa Cruz, Genome Browser by the ENCODE (ENCyclopedia Of DNA Elements) Project (43). Putative TSSs were predicted from the FANTOM (the Functional Annotation of the Mammalian Genome) web resource (44). Putative binding sites for transcription factors of conserved genomic regions were explored by conducting the UCSC Genome Browser JASPAR and TESS.
Cell Culture-Human HCC cell line HepG2, immortalized human liver cell line L02, and human embryonic kidney cell line HEK293T were obtained from ATCC and maintained in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supple-mented with 10% fetal bovine serum (FBS, Invitrogen), in a humidified atmosphere with 5% CO 2 at 37°C.
Genomic DNA Extraction, RNA Extraction, cDNA Synthesis, and Real Time PCR-For genomic DNA extraction, total cell genomic DNA was extracted by using DNeasy Blood and Tissue (Qiagen), following the manufacturer's instructions. For RNA extraction, the total RNA was extracted by using TRIzol (Invitrogen), in accordance with the manufacturer's instructions. For cDNA synthesis, 1 g of total RNA was reverse-transcribed by the SuperScript TM III First-Strand synthesis system (Invitrogen) after DNase I treatment (Invitrogen). Quantitative real time-PCR (qRT-PCR) was subsequently performed in triplicate with a 1:5 dilution of the resultant cDNA by using the Applied Biosystems 7500 real time PCR system (Applied Biosystems) with the Brilliant II SYBR Green qPCR master mix (Stratagene). All mRNA quantification data were normalized to GAPDH. The Ct value for each sample was calculated by the ⌬⌬Ct method, and the results were expressed as 2 Ϫ⌬⌬Ct (45).
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-The 5Ј-RACE, version 2.0 (Invitrogen), was used in accordance with the manufacturer's instructions. Briefly, cDNA was synthesized by using the MC-let-7a-1ϳlet-7d specific primers GSP1-1 or GSP1-2 (supplemental Table S2). Primary amplification was carried out with an abridged anchor primer (provided in the kit) and a GSP2-1 primer (supplemental Table S2). Nested PCR was performed with an abridged universal amplification primer (provided in the kit) and a GSP2-2 (supplemental Table S2) by using TaKaRa LA Taq with GC Buffer I (TaKaRa). The 5Ј-RACE PCR products were resolved on a 1% agarose gel and stained by EtBr. The bands were excised, cloned into pGL3 basic vector (Promega) via restriction endonuclease MluI sites, and sequenced.
Constructions of Promoter/Reporter and shRNA Expression Plasmids-A 1.9-kb MC-let-7a-1ϳlet-7d promoter was amplified by PCR by using PrimeSTAR TM HS DNA polymerase (TaKaRa) and cloned into pGL3 basic vector (Promega) between Xho I and Hind III sites by using primers PPR-1, as listed in supplemental Table S2. This plasmid was named PPR-1. Subsequent promoter truncations were generated from this template and cloned in a similar manner by using the primers in supplemental Table S2. To construct a MYC expression vector, a human MYC open reading frame (ORF) was amplified by using MYC primers, as indicated in supplemental Table S2, and inserted into the pcDNA3.1(ϩ) vector (Invitrogen) between Hind III and BamH I sites. To construct shRNA vectors for knocking down endogenous MYC, hairpin encoding oligonucleotides (sh-Myc, supplemental Table S2) were annealed and ligated into the pGE-1 shRNA expression vector (Stratagene) between BamH I and Xba I sites. The following targeting sequence was used, GATGAGGAAGAAATCGATG, in accordance with the literature published previously (46). The sequence and orientation of all inserts were confirmed by sequencing.
Mutagenesis-Mutations in the MYC-binding sites were made in the PPR-3 or PPR-10 construct by using two-step PCR (47). Briefly, two pairs of primers, including PPR-3 or PPR-10 sense primers and E-box antisense primers PPR-3 or PPR-10 antisense primers and E-box sense primers, were used in the first PCR step to generate intermediate PCR products. Then these two mutated products were denatured as a template for the second PCR step by using flanking primers PPR-3 or PPR-10. The final PCR products were digested with Xho I and Hind III and then inserted into pGL3 basic vector (Promega). The mutations were confirmed by sequencing.
Transient siRNA Transfection-The si-Lin28 and si-Lin28B were synthesized by GenePharma in accordance with the sequences reported previously (supplemental Table S2) (35). The transient transfection of L02 cells with control siRNA or siRNAs targeting LIN28 and LIN28B mRNAs was performed, as described previously (48).
Luciferase Reporter Assays-HepG2 cells, L02 cells, or HEK293T cells were plated at a density of 50,000 cells per well in a 24-well plate at 24 h before transfection. The cells were co-transfected with 250 ng of MC-let-7a-1ϳlet-7d promoter/ firefly luciferase reporter plasmids and 5 ng of pRL-TK Renilla plasmids (Promega) by using FuGENE HD transfection reagent (Roche Applied Science). For gain-and loss-of-function experiments, 750 ng of MYC expression plasmids (c-Myc), or knockdown MYC expression plasmids (sh-Myc), or their control vectors (pcDNA3.1 or pGE-1 negative control vector (pGE-NC)) were transfected. After 48 h post-transfection, the cells were either lysed in TRIzol (Invitrogen) for RNA extraction or in passive lysis buffer (Promega) for luciferase assay measured with the Dual-Luciferase reporter assay system (Promega) by using the TD-20/20 luminometer (Turner Designs). The relative luciferase activities were determined by calculating the ratio of firefly luciferase activities over Renilla luciferase activities.
Protein Extraction and Western Blot Analysis-Cells were seeded onto a 6-well plate at a density of 10 6 cells/well. The cells were lysed for 30 min on ice in RIPA lysis buffer (Thermo) with protease inhibitor mixture (Sigma) and phosphatase inhibitor (Pierce). Protein concentration was determined by a Bio-Rad protein assay kit (Bio-Rad). 50 g of proteins were separated by 12% SDS-polyacrylamide gels and transferred onto PVDF membranes. Membranes were probed with specific antibodies, including MYC (1:4000; Abcam) and ␤-actin (1:3000; Abcam), respectively. After washing, membranes were incubated with an HRP-conjugated secondary antibody (1:5000; Abcam). Membranes were exposed by using the enhanced chemiluminescence (ECL) substrate kit (Thermo) and then were photodocumented. The expression level of ␤-actin was also determined, and it served as an internal control.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts from HepG2 cells were prepared in accordance with the manufacturer's protocol (NE-PER Nuclear and Cytoplasmic Extraction Reagents, Pierce). Complementary oligonucleotide pairs corresponding to the three E-boxes embedded in the promoter region of MC-let-7a-1ϳlet-7d and the mutated E-boxes were synthesized and 5Ј end-labeled with biotin by Invitrogen. The oligonucleotide sequences are shown in supplemental Table S2. EMSA was performed by using a Light-Shift chemiluminescent EMSA kit (Pierce). Binding reaction with 10 g of HepG2 nuclear extracts and 100 fmol of 5Ј biotinlabeled oligonucleotide was carried out in accordance with the manufacturer's instructions. For the competition assay, a 100-fold molar excess of unlabeled oligonucleotide was added to the binding reaction mixture as a specific competitor. For antibody-supershift assay, a nuclear extract was preincubated with a 3 l MYC antibody (Abcam) before adding it to the binding reaction. DNA-protein complexes were separated on a preelectrophoresed 6% polyacrylamide gel in 0.5ϫ TBE, transferred to a nylon membrane, and cross-linked at 120 mJ/cm 2 for 1 min and detected by chemiluminescence, in accordance with the manufacturer's directions. Membranes were exposed by using the enhanced chemiluminescence (ECL) substrate kit (Thermo) and then were photo-documented.
Chromatin Immunoprecipitation (ChIP)-ChIP assays were carried out by using an EZ-ChIP assay kit (Upstate Biotechnology, Inc.), in accordance with the manufacturer's instructions. In brief, cells were grown to 90% confluence and added 1% formaldehyde at room temperature for 10 min. The cross-link reaction was quenched with 0.125 M glycine for 5 min at room temperature. Cells were then washed, scraped, and resuspended in 1 ml of lysis buffer. DNA was sonicated into around 400-bp pieces at 4°C by using Sonics Uibra Cells TM (XINCHEN). Supernatants were recovered by centrifugation and precleared for 1 h at 4°C with 60 l of protein G-agarose. Then 10 l (1%) of supernatant was removed as input. Immunoprecipitations were performed overnight with MYC antibody (Abcam, 9E11, ChIP Grade, 6 g) or IgG antibody (provided in kit, 1 g). The immune complexes were captured by incubation with 60 l of protein G-agarose for 2 h at 4°C. The immunoprecipitates were washed sequentially with wash buffers. After that, the immunoprecipitates were eluted from the protein G-agarose by incubating with elution buffer (1% SDS, 100 mM NaHCO 3 ). DNA-protein complexes were reversely cross-linked by a high salt solution at 65°C for 5 h. RNA and protein were eliminated by treating with 10 g of RNase A at 37°C for 30 min and then with protease K for 2 h at 45°C. Finally, DNA was purified by using the spin column provided in the ChIP kit and eluted with 50 l of elution buffer. qRT-PCR was performed by using the Brilliant II SYBR Green qPCR master mix (Stratagene) on the Applied Biosystems 7500 real time PCR system instrument. The primer pairs used for PCR analysis are shown in supplemental Table S2. All data were normalized to input (49).
MTT Analysis of Cell Proliferation-L02 cells were seeded at a density of 2500 cells per well onto a 96-well plate with the complete medium (100 l/well). After 24 h of incubation, cells were transfected with MYC, let-7a-1 expression plasmids (plllet-7), or their control vectors (pll3.7 or pcDNA3.1) and their respective negative control by using FuGENE HD. The cells were then incubated at 37°C in a humidified environment with 5% CO 2 . At days 1-6, supernatants were removed, and the cells were incubated with 20 l of 5 mg/ml MTT reagent (Sigma) at 37°C for 4 h. At the end of incubation, MTT reagent was removed from each well and replaced by 100 l of DMSO (Invitrogen). The plate was gently agitated for 10 min, and the absorbance (A) at 490 nm was determined by an ELISA reader (Bio-Rad).
Flow Cytometry-The collected cells were washed twice with PBS, fixed in 70% ethanol overnight, digested with RNase A (10 mg/liter) in PBS for 30 min, then stained with propidium iodide, and analyzed on a FACScalibur (BD Biosciences).
Statistical Analysis-Data were shown as means Ϯ S.D. Statistical analyses for detection of significant differences between the control and the experimental groups were carried out by using an unpaired two-tailed Student's t test (*, p Յ 0.05; **, p Յ 0.01; ***, p Յ 0.001).

RESULTS
Bioinformatic Analysis Predicted That Hsa-let-7a-1, let-7f-1, and let-7d Are Co-evolutionary Intergenic miRNAs-We analyzed the phylogenetic relationships of these three let-7 members. As shown in supplemental Table S3, they first emerged as a cluster at a very early stage during evolution and have never separated since then. Hence, we postulated that let-7a-1, let-7f-1, and let-7d were co-transcribed as a cluster MC-let-7a-1ϳlet-7d. Consistently, the UCSC Genome Browser also indicated that these three members are clustered within a 3-kb region on chromosome 9 (Fig. 1A). The distances between the encoding region of pre-let-7a-1, pre-let-7f-1 and pre-let-7f-1, pre-let-7d are 311 and 2401 bp, respectively. They are transcribed to the same orientation, and no gene is found to overlap with them in GenBank TM .
To predict the TSS and the promoter region of MC-let-7a-1ϳlet-7d, we first searched ESTs around their encoding region. As shown in Fig. 1A, EST BG326593 is overlapped with these three let-7 members. In addition, 39 scattered ESTs are found between the 5Ј boundary of EST BG326593 and MC-let-7a-1ϳlet-7 encoding region (data not shown). In addition, in the FANTOM data base (44), a subset of TSSs within 15 kb upstream of the MC-let-7a-1ϳlet-7d encoding region are found. All of the predicted TSSs are located in the region around the 5Ј end of EST BG326593 (chromosome 9: 95,967,500 -95,982,000). We also found that a CpG island overlaps with 5Ј boundary of EST BG326593 (Fig. 1A). In addition, analysis of the Pol2 (RNA polymerase II) ChIP-seq Signal/Histone Modifications ChIP-seq Signal and Transcription Factors ChIP-seq Signal within 15 kb upstream of MC-let-7a-1ϳlet-7d encoding region by the UCSC Genome Browser suggested that the TSSs of these three miRNAs were located about 10 kb upstream of their encoding region. Hence, these analyses suggest that the three members were encoded by a single polycistronic transcript MC-let-7a-1ϳlet-7d with a promoter located about 10 kb upstream of their encoding region (Fig. 1A).
Experimental Confirmation That Hsa-let-7a-1, let-7f-1, and let-7d Are Encoded by a Single Polycistronic Transcript-We performed strand-specific reverse transcription-PCR (RT-PCR) by using cDNA produced by HH antisense primer from HepG2 cells and L02 cells with primers matching segments along the length of the predicted human transcripts (Fig. 1A and supplemental Table S2), as described previously (50). Con-  has-let-7a-1, let-7f-1, and let-7d are encoded by a single polycistronic transcript MC-let-7a-1ϳlet-7d. A, schematic representation of the has-let-7a-1, let-7f-1, and let-7d gene location, putative promoter, and TSS. B, putative MC-let-7a-1ϳlet-7d primary transcript was confirmed by qRT-PCR in L02 cells. cDNA was synthesized by HH antisense primer. qRT-PCR products were obtained with primers located within the predicted transcript (BB, CC, DD, EE, FF, GG, and HH), but no product was obtained with primers located upstream of the predicted TSS (AA). As a control, genomic DNA was successfully amplified by all primers. M indicates DNA markers. C, qRT-PCR was performed. All the segments within the primary transcript produced less qRT-PCR product in HepG2 cells than that in L02 cells. *, p Յ 0.05; **, p Յ 0.01; ***, p Յ 0.001. sistent with the prediction, qRT-PCR products were obtained with primers located within the predicted transcript (BB, CC, DD, EE, FF, GG, and HH), but no product was obtained with primers located upstream of the predicted TSSs (AA). As a control, genomic DNA was successfully amplified by all primer sets (Fig. 1B). Furthermore, we quantified the expression level of the seven segments on the predicted primary transcript by qRT-PCR by using cDNA produced by random hexamers (Invitrogen) from HepG2 cells and L02 cells. As shown in Fig.  1C, all the segments within the primary transcript produced less qRT-PCR product in HepG2 cells than that in L02 cells. Taken together, these data suggested that let-7a-1, let-7f-1, and let-7d were encoded within a single independent non-proteincoding polycistronic transcript, which is transcribed 10 kb upstream from initiation sites.
Alternative Transcription Initiation Sites by HepG2 Cells and L02 Cells-To identify the initiation site of MC-let-7a-1ϳlet-7, we next performed 5Ј-RACE by using total RNA prepared from HepG2 cells and L02 cells. Two types of gene-specific primers, GSP1-1 and GSP1-2, were used for the synthesis of the first strand cDNA (Fig. 2A). After nested PCR by GSP2-1 and GSP2-2, the ϳ700-bp products were amplified in HepG2 and L02 (Fig. 2B).
We cloned the major band of 5Ј-RACE products and sequenced 10 clones for each sample. Sequence analysis of these 40 5Ј-RACE products (20 products for each cells) revealed that the products obtained from the HepG2 cells were about 30 bp smaller than those obtained from the L02 cells (Fig. 2C). Specifically, the major TSSs used in the L02 cells were located about 30 bp upstream of the TSSs used in the HepG2 cells.

Characterization of the MC-let-7a-1ϳlet-7d
Promoter-According to the preceding promoter analysis and the 5Ј-RACE results, the promoter of MC-let-7a-1ϳlet-7d should be located around the TSSs (Fig. 1A). Because the encoding region of MC-let-7a-1ϳlet-7d is highly conserved, we hypothesized that its promoters and regulating machinery would also be conserved. As shown in Fig. 3A, the genomic region proximal to the TSSs (from Ϫ540 to ϩ1317 with respect to the 5Ј end of EST BG326593 as ϩ1) shows high evolutionary conservation among different species. Therefore, we selected a 1.9-kb conserved region as the primary putative promoter to construct a promoter/luciferase reporter construct (PPR-1) and showed that PPR-1 could drive the luciferase expression in the L02 cells (Fig.  3B). The truncations of the promoter further revealed that the promoter (Ϫ540 to ϩ1 bp) exhibited the maximal promoter activity. Our results also showed that the truncation of the downstream region from ϩ1317 to ϩ1 dramatically increased the promoter activity (Fig. 3B), suggesting that the 3Ј end of PPR-1 contains suppression elements. In addition, truncation of the upstream region from Ϫ540 to Ϫ227 resulted in a loss of activity (Fig. 3B), thereby indicating that the elements between Ϫ540 and Ϫ227 contain an activator.
MYC Down-regulated the Expression of the Primary MC-let-7a-1ϳlet-7d Transcript-According to the whole genome ChIP-seq data on the UCSC Genome Browser, MYC has a very strong binding signal surrounding the MC-let-7a-1ϳlet-7d promoter in all of the cells (supplemental Fig. S1). Thus, we sought to determine whether MYC could regulate the MC-let-7a-1ϳlet-7d expression at the transcription level. We showed that overexpression of MYC in L02 cells caused a decrease in  -let-7a-1ϳlet-7d by 5-RACE. A, primers used for 5Ј-RACE. Two types of gene-specific primers (GSPs) GSP1-1 or GSP1-2 were, respectively, used to synthesize the first strand cDNA. Primary amplification was carried out with abridged anchor primer (AAP) and GSP2-1 primer. Nested PCR was performed with abridged universal amplification primer (AUAP) and GSP2-2. B, 5Ј-RACE results. The ϳ700-bp products were amplified in HepG2 cells and L02 cells by using both of the first cDNA strands derived from GSP1-1 and GSP1-2. C, sequence analysis of 20 products from each cell lines revealed that variable TSSs were used. The major TSSs in L02 cells were located about 30 bp upstream of TSSs in HepG2 cells.
the MC-let-7a-1ϳlet-7d primary transcript expression (Fig.  4A). Consistently, the knockdown of endogenous MYC by sh-Myc caused an increase (Fig. 4B). According to the published reports, LIN28 and LIN28B, which are induced by MYC, could mediate let-7 post-transcription repression in multiple human and mouse tumor models (30 -40). To exclude the possibility that repression of primary MC-let-7a-1ϳlet-7d is also mediated by LIN28 and LIN28B, we blocked the MYC/LIN28/let-7 cascade by siRNAs targeting LIN28 and LIN28B. As shown in Fig. 4C, the decrease of LIN28 and LIN28B could not eliminate the repression of primary MC-let-7a-1ϳlet-7d by MYC. These results suggested that the repression of primary MC-let-7a-1ϳlet-7d by MYC is LIN28-independent.

Endogenous MYC Regulated the MC-let-7a-1ϳlet-7d Expression by Two E-boxes Upstream and Downstream of TSSs-MYC
is known to bind to the canonical E-box sequence CACGTG, as well as to the noncanonical sequences, including CGCGTG (51,52). We noted two presumptive MYC-binding sites CACGTG located upstream and one noncanonical binding site CGCGTG located downstream of the MC-let-7a-1ϳlet-7d TSSs (Fig. 5A). According to the UCSC Genome Browser, these sites were conserved in vertebrates. Thus, we hypothesized that MYC might bind to these E-boxes to down-regulate MC-let-7a-1ϳlet-7d promoter activity.
To determine the ability of MYC binding to these three E-boxes, EMSAs and supershift analyses were performed with the promoter regions embedding the three E-boxes and the mutated E-boxes. As shown in Fig. 5C, E-box 2 and E-box 3 showed evident shift bands, whereas the E-box 1 could not bind to a nuclear extract. The formation of the complex E-box 2/protein or E-box 3/protein was inhibited by competition (1:100fold molar excess) with the unlabeled E-box 2 or E-box 3. Furthermore, mutation of the E-box 2-or E-box 3-binding site eliminated the shift bands. In addition, the specificity of MYC binding to the E-box 2 and E-box 3 was confirmed by the supershift of the E-box/protein bands. These results suggest that E-boxes 2 and 3 were necessary for the binding of the nuclear factors.
To assess the effects of these E-boxes in transcription regulation of MC-let-7a-1ϳlet-7d, co-expression studies were performed using MYC expression or knockdown vector and a PPR-10 promoter/reporter plasmid containing canonical MYC-binding motifs E-box 2 and a PPR-3 promoter/reporter plasmid containing both E-box 2 and the noncanonical E-box 3 (Fig. 6A). To our surprise, we found that the overexpression of MYC significantly activated luciferase activity of PPR-10 both in HepG2 and L02 cells (Fig. 6C), whereas knockdown of MYC resulted in a significant reduction of the PPR-10 activity (Fig. 6C). The positive regulation of PPR-10 (E-box 2) by MYC was further confirmed by E-box mutation assays. As shown in Fig. 6C (right panel), mutations of the E-box 2 reduced the promoter activity and abrogated MYC-mediated regulation of PPR-10 promoter activity. We then examined the regulation of PPR-3. Importantly, we found that overexpression of MYC significantly repressed the PPR-3 activity by 25-50%, whereas knock-  MC-let-7a-1ϳlet-7d promoter. A, plot generated from the UCSC Genome Browser depicts the evolutionary conservation (cons) of MC-let-7a-1ϳlet-7d encoding region and putative regulation region. Annotated 5Ј end of MC-let-7a-1ϳlet-7d is high evolutionary conservation among different species. B, schematic diagram of the reporter construct (left panel). Truncation of the downstream region to ϩ1 dramatically improved the promoter activity, whereas truncation of the upstream region to Ϫ227 resulted in progressive loss of activity.
down of MYC significantly increased PPR-3 activity (Fig.  6D). Furthermore, the mutation of E-box 2 in PPR-3 reduced promoter activity, but it could not eliminate the negative regulation by MYC. In contrast, the deletion of E-box 3 increased the promoter activity and also eliminated the negative regulation by MYC on PPR-3 activity (Fig. 6D). Taken together, our results showed that the E-box 3 located downstream of TSSs was the essential element required for MYCmediated repression and E-box 2 located upstream of TSSs for stimulation of the promoter activity of MClet-7a-1ϳlet-7d.
Consistently, upstream of the TSS, the ChIP-seq signal of the MYC binding partner MAX (supplemental Fig. S1) is as strong as MYC, whereas downstream of the TSS, it is much lower. These data further support the hypothesis that MYC regulates the transcription of MC-let-7a-1ϳlet-7d through these two E-boxes.
Binding of MYC with the Two MC-let-7a-1ϳlet-7d E-boxes in HepG2 and L02 Cells-To confirm whether MC-let-7a-1ϳlet-7d promoter is a direct MYC target, we further performed the ChIP assay by using a MYC-specific antibody in HepG2 cells and L02 cells. Because these two E-boxes are located within 650 bp, we sheared the fixed chromatin to an average length of 400 bp by sonication to enhance resolution (Fig. 7A). Real time PCR amplicons were designed to detect two putative E-box-binding sites in ChIP samples. Strong MYC binding with this locus was observed in both E-box 2 and E-box 3 but not in the control upstream AA site, which lacks the MYC binding E-box, or in the IgG negative control (Fig. 7, B and C). Importantly, we found that in the HepG2 cells, the binding quantum of MYC to E-box 3 is similar or slightly higher than that of the E-box 2 (Fig. 7C), and in L02 cells, the binding quantum of MYC to E-box 3 was significantly lower (50% lower) than that of the E-box 2.
We further analyzed the relative binding activity of E-box 2 and E-box 3 by EMSA. As shown in Fig. 7D, unlabeled E-box 3 could not compete with E-box 2 (1:200-fold molar excess), and 200-fold unlabeled E-box 2 could abrogate the shift band of the E-box 3-protein complex, suggesting that the binding activity of E-box 2 is stronger than that of E-box 3. These results suggest that the relative binding quantum of a MYC to E-box 3 and E-box 2 is at least partially responsible for the MYC-mediated decrease of the transcription of the MC-let-7a-1ϳlet-7d in the cancerous HepG2 cells. let-7 Could Partially Decrease MYC-induced Cell Proliferation-We overexpressed MYC in L02 cells and observed enhanced cell proliferation (Fig. 8A) and reduced expression of let-7 family members (Fig. 8B), and cell cycle distribution analysis by flow cytometry showed increased cell entry into S phase (Fig. 8C). Consistently, when we overexpressed let-7a-1 in L02 cells, we observed decreased cell proliferation rate (Fig. 8A), and a reversed MYC-mediated shift in cell distribution, with an accumulation of G 0 /G 1 -and G 2 /M-phase cells and a corresponding reduction of S phase cells (Fig. 8C).

DISCUSSION
Emerging evidence has shown that let-7 family members are tumor-suppressor miRNAs that are down-regulated in many types of cancers (13,18). Elegant works have been performed and showed that let-7 is regulated post-transcriptionally (30 -40). However, the regulation at the transcription level is not fully understood. In this study, we characterized the transcription machinery of MC-let-7a-1ϳlet-7d and revealed molecular mechanisms underlying the differential MYC-mediated transcription repression in human hepatocarcinoma HepG2 cells as compared with the noncancerous liver L02 cells. We showed that the binding quantum of MYC to the E-box 3 (suppress) relative to E-box 2 (activate) is significantly lower in noncancerous L02 cells than in those of HepG2 liver cancer cells, which may contribute to the differential transcriptional activity in these two cell lines.
The let-7 family accounts for a very high percentage of the miRNAs (21,53). The evolutionarily conserved E-box binding MYC is one of the most commonly activated oncoproteins associated with the pathogenesis of liver cancer (54,55). Overexpression of MYC can induce HCC with a high frequency in mice, whereas inhibition of MYC expression results in a loss of the neoplastic properties of carcinoma (56,57). This is consistent with the recent report suggesting that widespread miRNA repression by MYC contributes to tumorigenesis (41). MYC has been demonstrated to repress the expression of let-7 post-transcriptionally through LIN28B and LIN28 in multiple human and mouse tumor models (30 -40). The possibility that MYC could also regulate the transcription was first suggested by a report that MYC binds to a conserved site upstream of the microRNA let-7a-1/let-7f-1/let-7d cluster (41). However, to the best of our knowledge, the transcription machinery and the exact role of MYC in let-7 transcription regulation have not been reported.
In this study, we identified the TSSs of MC-let-7a-1ϳlet-7d, and demonstrated that multiple TSSs were used in HepG2 cells and L02 cells. An SNP site (rs11792471) was found in the 30-bp region. According to the UCSC Genome Browser, in this region the CTCF ChIP-seq signal was different in different cells. This raised the possibility that alternative transcription initiation sites might influence the transcription of MC-let-7a-1ϳlet-7.
The combination of computational prediction and strandspecific RT-PCR data suggested that the three let-7 family members in this cluster are encoded by a single polycistronic transcript with a promoter located about 10 kb upstream of their encoding region. The transcription activity of the core promoter (Ϫ540 to ϩ1 bp), with the maximal promoter activity, is almost equivalent to the SV40 promoter. The truncation of the downstream region from ϩ1317 to ϩ1 dramatically enhanced the promoter activity, suggesting the presence of suppression elements downstream of the transcription initiation site.
In fact, we found two presumptive canonical MYC-binding sites (CACGTG) located upstream and a noncanonical binding site (CGCGTG) located downstream of the MC-let-7a-1ϳlet-7d TSS. Our study revealed that MYC acts as an activator in binding to the canonical MYC-binding site upstream of the TSSs (E-box 2), although it acts as suppressor in binding to the noncanonical binding site downstream of the TSSs (E-box 3).
When MYC stimulates transcription, it dimerizes with its binding partner MAX and binds to genomic DNA directly (58,59). When it represses transcription, MYC does not directly bind to DNA, but it is recruited through protein-protein interactions to the promoters (60). As shown in supplemental Fig.  S1, the MAX ChIP-seq signal upstream of the TSSs, is strong, suggesting that MYC activates MC-let-7a-1ϳlet-7d promoter by directly binding and dimerizes with MAX to enhance MC-let-7a-1ϳlet-7d promoter activity through binding with E-box 2. In contrast, the MAX ChIP-seq signal downstream of the TSSs is much lower than MYC, suggesting that MYC does not dimerize with MAX, and this is consistent with its role in repressing MC-let-7a-1ϳlet-7d transcription through binding E-box 3.
Taken together, these results suggest that the ratio of MYC and MAX in the nucleus is the key factor controlling the tran-  -7a-1ϳlet-7d as a direct MYC target gene and binding depends upon cell type. A, agarose gel demonstrates sonication of chromatin from HepG2 and L02 cells to a length between 300 and 500 bp. B, chromatin from HepG2 and L02 cells was immunoprecipitated by using antibodies against MYC protein. PCR detection of the E-box sites showed binding of MYC to E-box 2. Antibody against IgG was used as a negative control. C, qRT-PCR assays were performed to quantify the MYC binding quantum. Schematic representation of the location of AA, E-box 2, and E-box 3 (upper panel) is shown. The results are summarized as relative binding (MYC ChIP relative to input ChIP). The enrichment of the E-box 3 was significantly less in comparison with the enrichment of the E-box 2 in L02 cells, whereas no significant difference was found in HepG2 cells. D, binding activity of MYC to E-box 2 is stronger than that to E-box 3. 200-Fold unlabeled E-box 3 could not compete 5Ј biotin-labeled E-box 2, whereas 200-fold unlabeled E-box 2 could compete 5Ј biotin-labeled E-box 3. *, p Յ 0.05; **, p Յ 0.01; ***, p Յ 0.001. scription of MC-let-7a-1ϳlet-7d. As we have demonstrated, the relative binding activity of MYC and E-box 2 is higher than that of MYC and E-box 3. Therefore, in normal cells, when MYC/MAX ratio is normal, MYC may dimerize with MAX and bind E-box 2 to enhance the transcription of MC-let-7a-1ϳlet-7d. The let-7 produced in turn binds to c-MYC mRNA and inhibits MYC translation. However, in cancer cells, when the MYC/MAX ratio is high, the extra MYC will bind to E-box 3 and repress the transcription of MC-let-7a-1ϳlet-7d. This will in turn decrease the let-7 level and abolish its inhibitory effect to MYC and ensure that the MYC protein will be translated. This mechanism may be partly responsible for the differential rate of MC-let-7a-1ϳlet-7d transcription in different cancer and noncancer cells. Further studies are also required to determine whether this transcription regulation is HCC specific or exists in other types of cancers.