Micro-RNA Regulation of the Mammalian lin-28 Gene during Neuronal Differentiation of Embryonal Carcinoma Cells

DNA and RNA used for transfection.

Reporter plasmid pCMV-Luc, which encodes an mRNA comprising a firefly luciferase translational unit fused to the simian virus 40 (SV40) late 3′ UTR, was constructed from pGL3-control (Promega) by replacing the SV40 promoter with a cytomegalovirus (CMV) immediate-early promoter. In plasmid pCL-L28, the SV40 3′ UTR of pCMV-Luc was replaced with the 3′ UTR of human lin-28. Plasmids pCL-L28Δ1, pCL-L28Δ2, pCL-L28Δ0, pCL-L28Δ1+2, pCL-L28Δ0+1+2, pCL-L28ΔL7. and pCL-L28Δ0+1+2+L7 were derived from plasmid pCL-L28 by deleting a 15-bp segment encoding miRE1 (CACTGTGTTCTCAGG), an 18-bp segment encoding miRE2 (CATGAGCAATCTCAGGGA), a 13-bp segment encoding miRE0 (CATGTATCTCAGG), and/or a 27-bp segment encoding L7 (GAGTGCACAGCCTATTGAACTACCTCA). In other derivatives of pCL-L28, lin-28 3′ UTR segments with imperfect complementarity to miR-9 (TTTACTGCTAAAAACCAAAG), miR-30 (TCCGTGTTCTTTGGGGGTTTTGTTTACA), or miR-128 (GAGATCACCGCAAACCTACCTTACTGTG) were deleted. Plasmid pCL was constructed from pCMV-Luc by inserting a 0.11-kb spacer derived from the human immunodeficiency virus type 1 env gene between the luciferase stop codon and an XbaI site located just downstream of it. Plasmids pCL-2E1, pCL-4E1, and pCL-6E1 were constructed by inserting two, four, or six tandem copies of human lin-28 miRE1 (TCCTGCACTGTGTTCTCAGGTACAT) into the XbaI site of pCL. Plasmids pCL-2E2, pCL-4E2, pCL-6E2, pCL-2E0, pCL-4X, pCL-6X, and pCL-2Y were constructed similarly, except that multiple copies of miRE2 (AGGTACATGAGCAATCTCAGGGATAGCC), miRE0 (ACTGCCATGTATCTCAGGCTTGG), element X (CCAAATGCAAGTGAGGGTTCTGGGGGCAACC), or element Y (TTCTGTGGAAGGAGATCTCTCAGGAGTAA) were inserted into pCL. Plasmids pCL-BGH-con, pCL-BGHmut-con, pCL-BGH-6E1, and pCL-BGHmut-6E1 were constructed by inserting a 0.23-kb pcDNA3 gene fragment (Invitrogen) containing a wild-type (AATAAA) or mutated (TTCTTT) bovine growth hormone polyadenylation signal into pCL-con or pCL-6E1 at an EcoRI site located 10 bp downstream of the luciferase coding region. Plasmid pCL-6E1+hp was derived from pCL-6E1 by inserting the sequence CGGGGCGCGTGGTGGCGGCTGCAGCCGCCACCACGCGCCCCGGACGCGTAGCT at a HindIII site 30 bp upstream of the luciferase initiation codon. Plasmid pBC21/CMV/β-Gal (56), which encodes β-galactosidase, was used as an internal standard.

Plasmid pMIR125a, which was used to transfect 293T cells, was constructed from pCMV-Luc by replacing the luciferase translational unit between the CMV promoter and the SV40 3′ UTR with a 1.13-kb fragment of human chromosome 19 that encodes miR-125a flanked upstream by 166 nt and downstream by 941 nt. In plasmid pMIR125aΔ, a 58-bp segment of pMIR125a that encodes most of the pre-miR-125a stem-loop was deleted. Plasmid pMIR125b, also used to transfect 293T cells, was constructed from pCMV-Luc by replacing the luciferase translational unit with a 0.77-kb fragment of human chromosome 21 that encodes miR-125b flanked upstream by 484 nt and downstream by 270 nt. In plasmid pMIR125bΔ, a 59-bp segment of pMIR125b that encodes most of the pre-miR-125b stem-loop was deleted. Plasmid pMIR125b-U1A was derived from pMIR125b by introducing a T→A substitution at the position corresponding to the 5′-terminal nucleotide of miR-125b. Plasmid pSH-MIR125b, which was used to transfect P19 cells, was constructed by inserting a 0.08-kb gene fragment encoding pre-miR-125b downstream of the U6 promoter of pSHAG-1 (41). For use as a negative control, plasmid pCMV-con, which encodes no miRNA-related transcript whatsoever, was constructed from pCMV-Luc by deleting the luciferase translational unit.

The synthetic siDCR RNA duplex used to knock down Dicer synthesis was prepared by annealing complementary oligonucleotides that had been chemically synthesized (5′ AAAGGACCCAUUGGUGAGGAA 3′ and 5′ CCUCACCAAUGGGUCCUUUCU 3′; Dharmacon). The siGL2 RNA duplex used as a negative control comprised two complementary strands having the sequence 5′ CGUACGCGGAAUACUUCGAdTdT 3′ and 5′ UCGAAGUAUUCCGCGUACGdTdT 3′ (Dharmacon). The 2′-O-methyl oligonucleotide used to inhibit miR-125b function in transfected in P19 cells (UCACAAGUUAGGGUCUCAGGGA; Integrated DNA Technologies) and the nonspecific 2′-O-methyl oligonucleotide used as a negative control (AAGCGAAGCAGUGCGUCAAGUA) were purified by polyacrylamide gel electrophoresis before transfection.

Cell culture and transfection.

293T human embryonic kidney cells were grown in Dulbecco's modified Eagle's medium (GIBCO) supplemented with 10% fetal bovine serum. Transient transfection of 293T cells with DNA was performed for 12 h using GeneJuice (Novagen), following the manufacturer's protocol except that 1.5 ml of culture medium was used per 35-mm well. Transfected DNA mixtures (1 μg) contained a plasmid encoding a pri-miRNA (965 ng), a reporter plasmid (25 ng), and a plasmid encoding β-galactosidase (10 ng).

P19 mouse embryonal carcinoma cells were cultured and induced to differentiate into neurons essentially as described previously (16). In brief, undifferentiated P19 cells were grown in α-minimal essential medium (α-MEM) (GIBCO) supplemented with 10% fetal bovine serum. Differentiation of P19 cells was induced by treatment with retinoic acid (1 μM) for 4 days in 10-cm bacteriological petri dishes, which were coated with 0.2% agarose to prevent the cells from adhering. The cells were then disaggregated by treatment with trypsin, replated on poly-d-lysine-coated tissue culture dishes, and cultured in either α-MEM medium containing 10% fetal bovine serum or, in one instance, Neurobasal medium (GIBCO) containing B27 supplement (GIBCO) and glutamine (0.5 mM).

Transient transfection of undifferentiated P19 cells with DNA, siRNA, and/or 2′-O-methyl RNA was performed overnight by using Lipofectamine 2000 (3 to 5 μl; Invitrogen; see manufacturer's protocol) in tissue culture dishes with 12 23-mm wells. Transient transfection of differentiated P19 cells was carried out overnight using cells that had been treated with retinoic acid for 4 days. The cells were trypsinized, cultured for 24 h in tissue culture dishes with 12 23-mm wells, and transfected with DNA in the presence of Lipofectamine 2000 (2 to 3 μl). The transfection medium was replaced with fresh culture medium after 6 h, and cell extracts were prepared 30 h later.

Reporter assays.

Reporter lysis buffer (Promega) was used to prepare cell lysates for reporter expression assays. Luciferase activity was measured in a Tecan SpectraFluor Plus instrument by using the Bright-Glo luciferase assay system (Promega) according to the manufacturer's instructions. β-Galactosidase activity was assayed with o-nitrophenylgalactoside by spectrophotometry, as previously described (46), or with the Galacto-Light Plus chemiluminescent reporter assay system (Applied Biosystems). Each measurement of the ratio of luciferase to β-galactosidase activity was determined with lysates derived from at least three to four independent transfections that usually were performed on more than one day.

RNA assays.

For analysis of miRNA, total cellular RNA was extracted by using Trizol reagent (Invitrogen), according to the manufacturer's instructions. Equal amounts of total RNA (10 μg) were denatured, fractionated by electrophoresis on a 15% polyacrylamide-8 M urea gel, and electroblotted and cross-linked onto a Hybond-XL nylon membrane (Amersham). The blots were probed at 38°C in PerfectHyb Plus hybridization buffer (Sigma) with terminally radiolabeled DNA oligonucleotides complementary to miR-125a or miR-125b and then washed with 0.5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate (SDS) buffer at 38°C.

For luciferase, β-galactosidase, and lin-28 mRNA analysis, cytoplasmic RNA was purified from cells 36 h after transfection, as previously described (49), except that Trizol reagent (Invitrogen) was used to isolate RNA from cytoplasmic extracts. Equal amounts of cytoplasmic RNA (10 μg) were denatured by glyoxal treatment (46), fractionated by electrophoresis on a 1% agarose gel, and blotted and cross-linked onto a Hybond-XL nylon membrane (Amersham). The blots were probed at 68°C in PerfectHyb Plus hybridization buffer (Sigma) with radiolabeled luciferase DNA, β-galactosidase DNA, or lin-28 DNA prepared by random primer labeling using a High Prime DNA labeling kit (Roche) and then washed with 0.5× SSC-0.1% SDS buffer at 68°C.

Immunoblot assays.

Cells were harvested and lysed with RIPA lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing Complete protease inhibitor (Roche). Equal amounts of each lysate were separated by electrophoresis on a 6% polyacrylamide-SDS gel and electroblotted onto a Hybond ECL nitrocellulose membrane (Amersham). Each blot was blocked with 5% dry milk in phosphate-buffered saline containing 0.05% Tween 20, probed at 4°C overnight with antibodies against human Lin-28 (polyclonal rabbit antiserum, diluted 1:1,500) or actin (1:3000, Sigma), incubated with a secondary antibody conjugated to horseradish peroxidase (1:20,000; Bio-Rad), and developed with an Immun-Star HRP chemiluminescence kit (Bio-Rad).

miR-125-responsive elements in human lin-28 mRNA.

The nearly identical sequence of the 22-nucleotide human miRNA miR-125b suggests that it is the human homolog of C. elegans lin-4 (Fig. 1A) (23). A second human miRNA, miR-125a, differs from miR-125b at only four positions. The presence in humans of a gene encoding a protein homologous to the C. elegans lin-28 gene product raised the possibility that this human gene might be downregulated by miR-125a and/or miR-125b, much as the C. elegans lin-28 gene is repressed by lin-4. Consistent with this hypothesis, miR-125b levels rise and Lin-28 protein levels fall during mammalian embryonic development (21, 53; other data not shown). To determine whether either of these miRNAs regulates lin-28 gene expression in humans, we first developed a system for expressing miR-125a or miR-125b in a human cell line. Transient transfection of 293T human embryonic kidney cells with either of two human genes encoding primary RNA transcripts (pri-miR-125a or pri-miR-125b) that can be processed in vivo to produce miR-125a or miR-125b generated both the mature forms of these two miRNAs and their partially processed precursors (pre-miRNAs) at a significantly higher concentration than in untransfected cells (Fig. 1B). No such increase in abundance was observed for these two miRNAs in cells transfected with deletion mutants of the pri-miRNA genes from which the 58- to 59-bp segment encoding the immediate RNA precursor of miR-125a or miR-125b (the pre-miR-125a or pre-miR-125b stem-loop) had been removed.

To test whether the human lin-28 gene can be repressed by miR-125a and/or miR-125b, we constructed a reporter in which the 2.7-kb 3′ UTR of human lin-28 (Fig. 2A) was fused to a translational unit encoding firefly luciferase (Luc-lin 28). Expression of this reporter in transfected 293T cells was reduced by more than 80% upon cotransfection with a human gene encoding either miR-125a or miR-125b but not with otherwise identical miRNA genes lacking the pre-miR-125a or pre-miR-125b segment (Fig. 2B). These results indicate that the human lin-28 3′ UTR can function as a target for repression by both miR-125a and miR-125b.

Examination of the 3′ UTR of human lin-28 mRNA revealed countless segments with imperfect complementarity to these two miRNAs. The great abundance of these potential sites of regulation complicated the identification of the true regulatory elements. However, since human miR-125a and miR-125b are identical to their mouse counterparts, it seemed reasonable to assume that their miRE targets would also be conserved. By focusing on the 3′ UTR regions that are identical in human and mouse lin-28, we were able to identify two adjacent elements (miRE1 and miRE2) with the potential to base pair with miR-125a or miR-125b in a manner reminiscent of the imperfect duplexes predicted for lin-4 paired with its target elements in the C. elegans lin-14 and lin-28 mRNAs (9, 35, 52) (Fig. 2A). In both humans and mice, these two lin-28 elements are directly adjacent to one another. (On the basis of sequence examination, Moss and Tang [36] have also proposed that miRE2 might be a regulatory target of miR-125.) Deletion of both miRE1 and miRE2 from the 3′ UTR of the luciferase reporter nearly abolished repression by miR-125a and miR-125b, whereas deletion of either element alone reduced but did not eliminate repression (Fig. 2B). Whatever modest responsiveness remained in the reporter lacking both miRE1 and miRE2 was abolished when a third, weaker element (miRE0) (Fig. 3) was removed as well (because of its minimal efficacy, deleting miRE0 alone from the lin-28 3′ UTR had little influence on repression). Thus, miRE1 and miRE2 appear to be the principal elements within the human lin-28 3′ UTR that mediate repression by miR-125a and miR-125b, and both are necessary for efficient downregulation.

Additional experiments showed that miRE1 and miRE2 are alone sufficient to direct repression by miR-125a and miR-125b when each is inserted in multiple copies into another luciferase reporter bearing a 3′ UTR derived from SV40. The degree of repression was similar for miRE1 and miRE2 and was dependent on the number of miRE copies inserted (Fig. 2C) but not on their sequential order (1-2-1-2 versus 2-2-1-1 [data not shown]). By contrast, inserting multiple copies of other lin-28 3′ UTR elements with the potential to form thermodynamically favorable duplexes with miR-125a and miR-125b (element X or element Y) had no such effect, and insertion of miRE0 alone had only a small effect (Fig. 2C and 3). The repressive influence of miRE1 and miRE2 on reporter gene expression was also observed in mouse NIH 3T3 fibroblasts, which naturally produce miR-125b at a concentration similar to that in transfected 293T cells (data not shown).

Mutational analysis of miR-125-responsive elements.

That miRE1 is fairly effective at mediating the inhibitory effect of these miRNAs was somewhat surprising in view of previous reports that base pairing of the 5′-terminal segment of miRNAs with their target elements, especially miRNA nucleotides 2 to 8, is very important for repression, as is the overall free energy of duplex formation (3, 6, 20, 28). miRE1 can form consecutive Watson-Crick base pairs only with nucleotides 3 to 9 of miR-125a or miR-125b, unlike miRE2, which can base pair with nucleotides 1 to 9 (Fig. 2A). Furthermore, the duplex structures formed by miRE1 are expected to be of lower thermodynamic stability than those formed by miRE2. To investigate the basis for the efficacy of miRE1, we modified the 3′ portion of this element to alter its base-pairing potential with the 5′ segment of miR-125 and then tested the effect of these mutations on repression of a luciferase reporter containing two such elements (Fig. 4).

Our data indicate that the presence of an adenosine at the 3′ end of miRE1 (position 1) is important for the function of this RNA element. Without this adenosine residue (M1), miRE1 became significantly less active. The contribution of the adenosine at position 1 was also evident for an miRE1 variant (M2) whose efficacy had been increased by changing the adjacent nucleotide to allow base pairing at position 2; in this context as well, an A→U substitution at position 1 (M3) was deleterious. Among the miRE1 variants that could form seven consecutive base pairs with the 5′ segment of miR-125b, pairing with nucleotides 2 to 8 (M4) or 3 to 9 (M1) was more effective than pairing with nucleotides 1 to 7 (M5).

Additional experiments showed that the contribution of the 3′-terminal adenosine residue to miRE1 function is not a consequence of its complementarity to the uridine at position 1 of miR-125b. This was determined by changing the first nucleotide of miR-125b from U to A (miR-125b-U1A) and measuring the efficacy of the modified miRNA in downregulating the same set of reporters, each of which contains a pair of miREs that end with either an adenosine or a uridine residue. To within experimental error, the mutant and wild-type forms of miR-125b were indistinguishable in their ability to repress every sequence variant of miRE1, regardless of the base-pairing potential of the 3′-terminal miRE residue (Fig. 4B). That an adenosine at this position can enhance miRE1 function irrespective of its ability to base pair may also help to explain the lesser activity of miRE0 and the inactivity of element Y, both of which can pair with nucleotides 3 to 9 of miR-125b but lack a terminal adenosine residue opposite nucleotide 1 (Fig. 3).

Influence of miR-125 on mRNA abundance.

Gene regulation by miRNAs that are imperfectly complementary to their mRNA targets is generally thought to be achieved via a translational repression mechanism (5, 38, 40, 52, 56), whereas silencing by perfectly complementary siRNAs typically involves accelerated mRNA degradation (33, 51). To determine whether repression mediated by miRE1 and miRE2 occurs only at the level of translation, the cytoplasmic concentration of luciferase reporter mRNAs bearing six copies of miRE1 or miRE2 was compared in the presence or absence of miR-125a or miR-125b. The significant decline in luciferase protein synthesis observed in cells that produced either of these miRNAs was accompanied by a smaller but significant decrease in the concentration of the luciferase reporter mRNA (Fig. 5). In each case (and in the case of a reporter containing the entire lin-28 3′ UTR; data not shown), the calculated reductions in mRNA abundance and translational efficiency (protein yield per mRNA molecule) were approximately equal in magnitude. No such change in either mRNA or protein concentration was observed for a luciferase control gene that lacked the lin-28 miREs but was expressed under the control of the same promoter.

In mammalian cells, siRNAs can reduce the concentration of targeted mRNAs not only by base pairing with perfectly complementary RNA elements to direct mRNA cleavage but also by acting on complementary DNA elements to cause transcriptional silencing (17, 34). If the miRE-dependent reduction in mRNA concentration that is mediated by miR-125a and miR-125b occurs via a mechanism unrelated to transcription initiation, then this reduction should be promoter independent (as is observed; see below) and should require that the miREs be present in the mRNA and not merely encoded in the DNA. To address the latter point, a 0.23-kb DNA fragment encoding an intact or mutated polyadenylation signal (AAUAAA or UUCUUU) was inserted into a luciferase reporter gene between the coding region and six copies of miRE1 present in the 3′ UTR. The two resulting reporters, which were almost identical in DNA sequence but encoded mRNA transcripts that either contained or lacked the miREs depending on the site of polyadenylation, were tested in 293T cells for their responsiveness to miR-125b (Fig. 6). Only the reporter that bore the defective polyadenylation signal and therefore encoded a longer mRNA containing the miREs was repressed, as evidenced by an miR-125b-dependent reduction in both luciferase protein synthesis and reporter mRNA abundance. In contrast, the presence of an efficient polyadenylation site upstream of the miREs abolished repression and rendered the reporter mRNA concentration insensitive to miR-125b. Together, these findings suggest that the interaction of miR-125a and miR-125b with miRE1 and miRE2 of human lin-28 represses gene expression by a dual mechanism involving both an impediment to translation and a posttranscriptional reduction in mRNA abundance.

Further investigation revealed that the observed reduction in mRNA concentration was not a secondary consequence of the decrease in translational efficiency caused by interaction of miR-125 with the lin-28 miREs. A large (42-nucleotide) stem-loop structure was inserted into the 5′ UTR of a luciferase reporter transcript containing six copies of miRE1 in its 3′ UTR. By blocking ribosome scanning, the stem-loop alone virtually abolished translation (99% inhibition of luciferase synthesis), yet its presence had no effect on either the abundance of the reporter mRNA in the absence of miR-125b or the ability of this miRNA to cause a marked reduction in the cellular concentration of the reporter message (Fig. 7). Thus, impaired translation per se is not sufficient to cause the concentration of this mRNA to fall, nor is translation required for miR-125b to mediate a decline in reporter mRNA abundance. We conclude that miR-125b reduces the concentration of mRNA containing the lin-28 miREs by a posttranscriptional mechanism that is independent of its inhibitory effect on translation.

Role of miR-125 in downregulating lin-28 during neuronal differentiation.

Although 293T cells can be genetically engineered for use in studies of repression by miRNAs, this cell line does not normally produce lin-28 mRNA, miR-125a, or miR-125b. To examine the role of miR-125a and miR-125b in regulating expression of the endogenous lin-28 gene in mammalian cells, we instead chose P19 embryonal carcinoma cells. This mouse cell line has been widely used as a model system for investigating the differentiation of nerve cells in mammals, since it can be induced to differentiate into neurons, astrocytes, and fibroblast-like cells upon treatment with retinoic acid (16); in addition, it is known to express the lin-28 gene (36). Prior to induction of P19 cells, miR-125a and miR-125b are virtually undetectable, whereas the mRNA and protein products of the endogenous lin-28 gene are quite abundant (Fig. 8). Over a period of 4 to 5 days following induction with retinoic acid, the cellular concentration of miR-125b and, to a lesser extent, miR-125a increases significantly, while Lin-28 protein and mRNA levels decline markedly (Fig. 8) (48, 53). (The abundance of miR-125b in P19 cells treated with retinoic acid for 7 days is even higher than that in 293T cells transfected with a gene encoding this miRNA.) Thus, lin-28 gene expression in differentiating P19 cells is inversely correlated with the cellular concentrations of miR-125a and miR-125b, a finding consistent with a natural role for these miRNAs in repressing lin-28 expression.

To test the validity of this inference, we impeded miRNA production in P19 cells by using RNA interference to knock down expression of Dicer, the RNase responsible for the last step in miRNA processing. Transfection of retinoic acid-treated P19 cells with Dicer siRNA reduced miR-125b levels by 72% and increased the cellular abundance of the Lin-28 protein by almost a factor of two versus cells transfected with a control siRNA (data not shown). We conclude that the repression of lin-28 following induction of P19 cells is, at least in part, miRNA dependent.

That miR-125b in particular likely contributes to lin-28 repression during differentiation of P19 cells was implied by experiments showing that this miRNA is able to downregulate expression of the endogenous lin-28 gene in undifferentiated P19 cells engineered to produce miR-125b at a physiological concentration. Uninduced P19 cells were transfected with a plasmid encoding pre-miR-125b so as to raise the cellular abundance of the corresponding miRNA to a level comparable to that in retinoic acid-treated cells (Fig. 9) and in rat hippocampal neurons 3 days before birth (data not shown). The increased concentration of miR-125b in these cells resulted in a marked reduction in Lin-28 protein synthesis compared to that in P19 cells transfected with a control plasmid that did not encode the miRNA (Fig. 9). This reduction in protein synthesis was accompanied by a significant decrease in the cytoplasmic concentration of lin-28 mRNA, as observed in 293T cells for reporter mRNAs bearing the lin-28 miREs but transcribed under the control of a different promoter. These findings indicate that even in the absence of differentiation, miR-125b is alone sufficient to cause significant repression of lin-28 gene expression in P19 cells, a conclusion consistent with a similar role for this miRNA in differentiating P19 cells.

To demonstrate directly that miR-125b helps to downregulate lin-28 gene expression in P19 cells that are undergoing differentiation, we selectively inhibited the function of this miRNA by transfection with a complementary 2′-O-methyl oligonucleotide (13, 32). In uninduced P19 cells, this complementary oligonucleotide interfered with the capacity of plasmid-encoded miR-125b to repress the expression of a cotransfected luciferase reporter bearing six copies of miRE2 (Fig. 10A). In P19 cells induced to differentiate, the presence of the complementary oligonucleotide impaired the ability of endogenous miR-125b to repress Lin-28 protein synthesis (Fig. 10B).

To confirm that the miR-125-responsive elements in the lin-28 gene contribute to its downregulation in differentiating P19 cells, we compared the expression of a luciferase reporter gene bearing the entire lin-28 3′ UTR (Luc-lin 28) to that of an otherwise identical reporter from which these elements (miRE0, miRE1, and miRE2) had been deleted. In P19 cells treated with retinoic acid to induce miR-125 production, expression of the reporter bearing an intact lin-28 3′ UTR was about half that of the reporter lacking the three miREs (Fig. 11A). By contrast, little if any decrease in Luc-lin 28 expression was observed in uninduced cells devoid of miR-125 (Fig. 11A). Similarly, compared to a reporter that lacked miREs, the expression of luciferase reporters containing six copies of miRE1 or miRE2 was reduced by 60% in retinoic acid-treated cells and was diminished to an even greater extent when the induced P19 cells were cultured in a medium (Neurobasal/B27 [4]) that favored neuronal cells over other differentiated cell fates (Fig. 11B; other data not shown). These results imply that the miR-125-responsive elements in lin-28 help to downregulate this gene in P19 cells undergoing differentiation into neurons.

Additional segments within the lin-28 3′ UTR bear significant complementarity to other miRNAs that increase in abundance when P19 cells are treated with retinoic acid (let-7, miR-9, miR-30, and miR-128), raising the possibility that lin-28 may be repressed by multiple miRNAs (14, 27, 36, 38, 48). However, among these potential regulatory elements, only deletion of the segment complementary to let-7 (L7) had any influence on expression of the Luc-lin 28 reporter in retinoic acid-treated cells, increasing luciferase synthesis by ∼40% relative to that of a reporter with an intact 3′ UTR (Fig. 11A). The regulatory effects of this element and the miR-125-responsive elements within the lin-28 3′ UTR were additive. We conclude that let-7 makes a smaller, but significant, contribution to lin-28 repression in differentiating P19 cells.