Altered profiles of gene expression reflect the reprogramming of intestinal epithelial cells during their maturation along the crypt-luminal axis. To focus on genes important in this process, and how they in turn are regulated, we identified 14 transcripts commonly downregulated in expression during lineage-specific maturation of the immortalized cell lines Caco-2 (absorptive), HT29Cl16E (goblet), and HT29Cl19A (secretory) induced by contact inhibition of growth or the short-chain fatty acid butyrate. One such gene, Mybl2 (Myb-related protein B), has been linked to the stem cell phenotype, and we report is also markedly suppressed in maturing cells along the crypt-luminal axis in vivo. Mybl2 is not significantly downregulated transcriptionally during colon cell maturation, but we identified a potential micro-RNA (miRNA)-binding sequence in the Mybl2 3′-untranslated region that mediates reporter gene suppression in differentiating colon cells. Accordingly, miRNAs predicted to bind this functional target are upregulated in differentiating colon epithelial cells in vitro and in vivo; expression of one of these, hsa-miR-365 (but not hsa-324–5p), suppresses Mybl2 protein expression in proliferating Caco-2 cells. These data demonstrate that miRNA silencing plays an important role in regulating gene expression in maturing colon epithelial cells, and that utilizing a target-centered approach, rather than profiling global miRNA expression, can identify physiologically relevant, functional miRNAs.
- crypt-luminal axis
normal intestinal homeostasis is dependent on reprogramming of highly proliferative, multipotent stem/progenitor cells into enterocytic, goblet, and enteroendocrine cells as they migrate along the crypt axis toward the lumen and Paneth cells that remain at the bottom of the crypt. Stem and progenitor cells at the crypt base robustly express genes that promote cell cycling, driving high levels of cell division, while they suppress expression of phenotypic markers of mature states. As these cells migrate from the crypt toward the lumen, many signals and pathways regulate their reprogramming to exit the cell cycle and undergo lineage-specific differentiation, predominantly through changes in gene expression (42, 74, 75, 77). Therefore, identifying molecules that drive this maturation and understanding how they are, in turn, regulated will provide crucial insights into how colon epithelial cells acquire mature states and avoid acquisition of premalignant and malignant phenotypes.
Reprogramming of maturing colonic epithelial cells is recapitulated in some colon adenocarcinoma cell lines in culture. These can often be induced to undergo cell-cycle arrest and express differentiated phenotypes by the short-chain fatty acid butyrate, a product of dietary fiber fermentation present at concentrations approaching 20 mM in the colon and a physiological regulator of colonic cell maturation in vivo (4, 25, 55, 68). Human adenocarcinoma cell lines have also been isolated by selection for cells that undergo contact inhibition of growth, followed by lineage-specific differentiation along absorptive (Caco-2), goblet (HT29Cl16E), or secretory (HT29Cl19A) cell lineages. We and others have dissected changes in gene expression profiles of these cell lineages as they differentiate (20, 42, 77), and we have also identified RNA (44) and protein expression profiles (10) that characterize intestinal cell maturation in vivo by analysis of cells eluted sequentially from the crypt-villus axis of the mouse small intestine. The requirements for reprogramming these highly proliferative tumor cells into growth-arrested, differentiating cells has shed light on many mechanisms that drive maturation of normal colon epithelial cells (58, 72).
Transcriptional regulation is responsible for many gene expression changes in maturing colon epithelial cells, including Wnt target gene repression (43), Notch target gene de-repression (66), and activation of phenotypic markers, such as alkaline phosphatase (28) and dipeptidyl peptidase IV (45). Nevertheless, posttranscriptional mechanisms, in particular silencing by micro-RNAs (miRNAs), also play critical roles in reprogramming of cells. These small, noncoding RNAs direct an RNA-induced silencing complex to mRNA targets by complementary base pairing between the miRNA and sequences in the 3′-untranslated region (UTR) of the message, resulting in mRNA degradation and/or translational repression (29, 30, 73). Recently, miRNAs have also been demonstrated to relieve translational inhibition of certain genes by acting as decoys that sequester heterogeneous nuclear ribonucleoprotein E2 from target mRNAs (16). miRNAs frequently target genes in developmental pathways, and their aberrant expression has been implicated in the pathogenesis of several diseases, including cancer (27) and cardiomyopathy (76).
Here we identify a subset of genes commonly downregulated in four independent colon cell models of cell maturation: butyrate-induced growth arrest and differentiation, and spontaneous differentiation along the absorptive, goblet, and deep-crypt secretory cell lineages (Caco-2, HT29Cl16E, and HT29Cl19A cells, respectively), triggered by contact-mediated growth arrest. One of these genes, Mybl2 (Myb-related protein B), belongs to a family of transcription factors linked to both cell cycling and differentiation in multiple systems (reviewed in Refs. 53, 60), and importantly, is a transcriptional regulator highly expressed in a number of pluripotent stem cells (47) and is critical to maintenance of the stem cell state (9, 69). We demonstrate that Mybl2 is indeed regulated in expression along the crypt-villus axis in vivo and that the downregulation during maturation is driven not by transcriptional regulation, but at least in part by a miRNA-mediated mechanism involving miR-365.
MATERIALS AND METHODS
The Caco-2 human adenocarcinoma cell line was maintained as described (42) in medium containing 20% fetal bovine serum. Spontaneous differentiation: cells were grown to confluence (day 0) and harvested at various time points thereafter (medium changes every 2 days). For butyrate induction, cells were treated with 5 mM sodium butyrate when ∼95% confluent (day 0) and harvested at various time points thereafter.
Caco-2, HT29Cl16E, and HT29Cl19A RNA isolation, probe synthesis, array hybridization and scanning, and data analysis have been reported (42, 77), utilizing cDNA arrays prepared by Albert Einstein College of Medicine microarray facility (13). For each time point, arrays were performed on two independent chips: two with 8,063 sequences (16E, 19A), or one with 8,064 and the second with 9,216 (Caco-2). Criteria for altered expression was at least 1.5-fold change in day 14–15 spontaneously differentiating, or day 2 butyrate-induced, cells relative to day 0 cells. Functional group classifications utilized gene ontology annotations (http://www.geneontology.org/).
Caco-2 cells, harvested by scraping from six-well plates, were sonicated on ice for two 5-s pulses with a 30-s pause (setting 2, 7.5% output; Branson 450 Sonifier, Fisher Scientific). Caco-2 lysate (20 μg), 100 μg mouse small intestine (encompassing duodenum, jejunum, and ileum) crypt-villus tissue (44), or 30 μg human colon crypt-luminal tissue was analyzed by Western blotting, as described (52) except protease inhibitor cocktail was from Sigma, and blocking and antibody dilution buffer was 5% milk/Tris-buffered saline-Tween (0.05% Tween 20, 100 mM Tris pH 7.5, 150 mM NaCl). Blots were developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific), with signal detected using film or a Kodak 4000R Image Station, and analyzed with Kodak Molecular Imaging Software, version 4.0 (Eastman Kodak). Primary antibodies were as follows: 1:10 mouse anti-Mybl2 (R. Watson, Imperial College School of Medicine, London, UK), 1:1,000 rabbit anti-alkaline phosphatase (NB 600–588, Novus Biologicals), 1:4,000 mouse anti-α-tubulin (T5168, Sigma), 1:500 mouse anti-villin (610358, BD Transduction Laboratories), 1:1,000 rabbit anti-histone H2A (07–146, Millipore; gift of Dr. Richard Chawan), or 1:1,000 rabbit anti-GAPDH (Abcam; gift of Dr. Art Skoultchi). Secondary antibodies were 1:2,000 goat anti-mouse or -rabbit IgG-horseradish peroxidase (Santa Cruz).
Cells were collected in Trizol (Invitrogen), and total RNA was prepared as described by the manufacturer. cDNA was synthesized from 1 μg total RNA using the iScript cDNA Sythesis Kit (Bio-Rad). Quantitative RT-PCR (qRT-PCR) utilized SYBR Green PCR Master Mix (Applied Biosystems) in 30 μl with 50 ng cDNA and 100 nM primers, with a 7900HT Sequence Detection System (Applied Biosystems). Cycling conditions were as follows: 50°C for 2 min; 95°C for 10 min; 40 cycles of 95°C for 15 s, melting temperature (Tm) for 30 s, 60°C for 30 s. For pre-mRNA and mRNA quantification, Tm = 60°C. For miRNA quantification, Tm was specific for each primer set (see miRNA quantification below). Specificity of amplification was confirmed by single peaks in dissociation curves and/or single bands on agarose gel electrophoresis. Analysis utilized SDS version 2.3 software (Applied Biosystems). GAPDH-normalized mRNA expression in differentiating cells was calculated relative to that in proliferating cells by the 2−ΔΔCt method (38). Pre-mRNA quantification utilized primers spanning intron 2/exon 3 of Mybl2, exon 4/intron 4 of cyclin D2, and intron 1/exon 2 of c-myc. Primers are listed in Supplemental Table S1. (The online version of this article contains supplemental data.)
Subconfluent/proliferating or day 10 or 12 spontaneously differentiating Caco-2 cells were transfected in 12-well plates with 1 μg/well luciferase reporter + 1 μg/well pRL-TK renilla (Promega, Madison, WI) for 48 h using Lipofectamine Plus (Invitrogen), as described by the manufacturer. The ratio of renilla-normalized luciferase activity in differentiating vs. proliferating cells was determined. Reporters were as follows: pGL2 Basic with luciferase expression driven by nucleotides −910 to −102 of the human Mybl2 promoter [R. Watson (35)] and TOP flash (33) driven by three T-cell factor binding sites upstream of a minimal c-fos promoter. Lysate (20 μl) was assayed for luciferase and renilla activity on the LMaxII dual-injection microplate reader (Molecular Devices). Data analysis utilized SoftMax Pro 5 software (Molecular Devices).
Determination of potential miRNA targets in the Mybl2 3′-UTR.
miRNAs predicted to bind the human Mybl2 3′-UTR were identified using miRBase Targets version 5 (http://microrna.sanger.ac.uk/targets/v5/), and predicted targets were compiled into 15 nonidentical sequences. RNA secondary structures were computationally modeled using GeneBee (http://www.genebee.msu.su/services/rna2_reduced.html) and Vienna (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi) algorithms. Targets forming secondary structures exhibiting negative free energies were considered further.
Assay for functional miRNA targets.
A double-stranded oligo consisting of recognition sites for the restriction endonucleases XbaI-SpeI-EcoRI-XbaI was subcloned into the XbaI site of pGL3 Control (Promega), two residues downstream of the luciferase stop codon, in forward and reverse orientations. miRNA targets, synthesized as single-stranded complementary oligos (Sigma Genosys), consisted of a dimerized Mybl2 3′-UTR potential miRNA binding site (see Fig. 6), separated by [residue A-recognition site for BamHI-residue A] and flanked by SpeI (5′) and EcoRI (3′) recognition sites. Annealed, double-stranded oligos were subcloned into pGL3 Control (forward and reverse) vectors digested with SpeI and EcoRI. The ratio of renilla-normalized luciferase activity of miRNA target (forward or reverse)-harboring vector vs. pGL3 control (forward or reverse) alone was calculated. miRNA targets suppressing parental pGL3 control activity by ≥50% were considered functional.
Cells were harvested in Trizol (Invitrogen), and total RNA, including those <200 nucleotides, was isolated using the mirVana miRNA Isolation Kit (Ambion), according to the manufacturer. cDNA was synthesized from 2 μg total RNA using primer RTQ, as described (57) with minor modifications. cDNA (∼10 ng) was used for qRT-PCR with a miRNA-specific primer and the universal primer RTQ-UNIr using an annealing temperature specific to each set of primers (miR-365, 588, 130a, 143, 145: Tm = 48°C; miR-30a, 362–5p, 10a, 205, 221: Tm = 50°C; miR-331–3p, 34a and let-7a1, 5S: Tm = 60°C; miR-324–5p: Tm = 75°C), and miRNA expression was normalized to that of 5S rRNA. Primers for miRNAs 588, 145, and 324–5p amplified a pre-miRNA as well as a mature form. Primer sequences are listed in Supplemental Table S1.
Proliferating, subconfluent Caco-2 cells (∼8,333 cells/cm2) were transfected for 48 h with 100 nM miRNA (miRIDIAN miRNA mimics, Thermo Scientific Dharmacon) using Oligofectamine (Invitrogen), as described (52).
Mechanical isolation of human colon tissue along crypt-luminal axis.
Tissue sample preparation was essentially as described (34). Briefly, tissue from a colectomy specimen removed at surgery and forwarded to Pathology at Montefiore Medical Center was obtained within ∼45 min to 1 h after resection. Four sections of morphologically normal colon mucosae, each 2.5 cm × 0.5 cm, were dissected free from underlying muscularis propria, embedded in OCT medium (Sakura Finetek) luminal face down, and frozen in liquid nitrogen. With the luminal aspect facing the blade, 10-μm-thick horizontal sections were cut sequentially from the luminal surface, such that early fractions contained cells at the top of the crypt, and later sections contained cells at the base of the crypt. Sections from ∼50-μm intervals were processed for hematoxylin and eosin staining to confirm morphology and position along the crypt-luminal axis (see Fig. 3), with intervening sections collected into 1.5-ml microfuge tubes (1 or 2 sections/tube for RNA or protein, respectively) and kept frozen at −80°C. Sections were processed for total RNA isolation using the mirVana kit, and unamplified RNA was used for qRT-PCR and miRNA quantitation, as described above. Protein lysates, prepared by sonicating two sections on ice in 250 μl at 7.5% output for two 5-s pulses with a 30-s pause, were analyzed by Western blotting, as described above. This protocol was classified as exempt, per federal regulations 45 CFR 46.101 (b), by the Montefiore Medical Center Institutional Review Board.
Genes commonly regulated in colon epithelial cell differentiation.
Using our laboratory's published databases (Refs. 42, 77, available at www.augenlichtlab.com), gene expression profiles were compared at intermediate time points of differentiation of colon epithelial cell lines to identify sequences that potentially regulate colon cell maturation rather than those that arise as downstream markers of a mature state. Therefore, data were analyzed at 14–15 days of spontaneous differentiation in Caco-2, HT29Cl16E, and HT29Cl19A cells, which fully differentiate, respectively, along the absorptive, goblet, and deep-crypt secretory cell lineages in 20–21 days, and at 2 days of butyrate treatment of Caco-2 cells, which require 5–7 days to fully express absorptive cell markers. Comparison of >8,000 overlapping sequences in these databases identified 14 significantly downregulated (reduced ≥1.5-fold relative to proliferating cells) in all differentiating cells (Table 1). These sequences encompass functional groups involving chromosome organization, transcriptional activity, DNA replication and repair, cell cycle, RNA processing and transport, translation, protein folding and modification, and cell adhesion. Each sequence was altered during maturation of each of three distinct colon epithelial cell lineages induced by two different methods and, therefore, likely functions in the complex process of colon cell maturation.
Mybl2 is suppressed in differentiating colon cells in vitro and in vivo.
One of the 14 downregulated genes (Table 1) is Mybl2, a sequence that encodes a transcription factor with an established role in promoting cell proliferation (3, 23, 36, 37, 49, 56, 59, 63) and a putative function in regulating differentiation of a number of cell types (8, 17, 22, 54, 56). As expected from analysis of the expression array databases (above), Mybl2 conformed to this expression profile, with Mybl2 RNA and protein robustly expressed in proliferating Caco-2 cells, but significantly suppressed in cells undergoing differentiation induced by either contact-mediated growth arrest (Fig. 1A) or butyrate treatment (Fig. 1B). Importantly, Mybl2 has been identified as a marker of stemlike, pluripotent cells (9, 47).
Two methods were used to determine the pattern of expression of Mybl2 along the crypt-luminal axis. First, with mouse tissue, we used sequential elution of cells from the intestinal mucosa that, as we and others have extensively validated, isolates cells as a function of their position along the crypt-villus axis of the mouse small intestine in vivo (19, 44, 64, 79). Second, we sectioned fresh frozen human mucosal samples from the luminal surface downward toward the crypts (see materials and methods). These approaches demonstrate that mouse (Fig. 1C) and human (Fig. 2B) Mybl2 protein is most highly expressed in the stem/progenitor cell compartment at the bottom of the crypt and decreases in a gradient to undetectable levels as intestinal epithelial cells undergo maturation along the crypt-luminal axis. This decreasing Mybl2 expression in these fractions as cells mature along the crypt-villus axis is similar to that of proteins, such as cyclin-dependent kinase 2 and proliferating cell nuclear antigen, which drive proliferation and is in contrast to that of differentiation markers, such as alkaline phosphatase (Fig. 1C) or villin (Fig. 2B) (44, 64). Similarly, RNA for Mybl2, the cell-cycle promoting factors cyclin B2 and cyclin D2, and the stem cell marker Lgr5 are highly expressed in human stem and progenitor cell populations at crypt bottoms and decrease markedly in differentiating cells near the lumen, whereas RNA for the differentiation-specific markers gut enriched Kruppel-like factor-4 and alkaline phosphatase exhibit the opposite expression pattern and are enriched in differentiated cells near the lumen (Fig. 2A). Hematoxylin and eosin staining confirms an ordered architecture of cells in these human colon tissues, with well-organized crypts surrounded by lamina propria throughout the crypt-luminal axis (Fig. 3). Mybl2's localization to the nuclei of these human mucosal cells (Fig. 4), similar to that in Caco2 cells (Fig. 4), confirms its role as a transcriptional regulator of key genes in the colon.
Transcriptional repression of Mybl2 is not significant during colon cell maturation.
Surprisingly, there was no difference in transfected Mybl2 promoter activity in proliferating vs. spontaneously differentiated Caco-2 cells (Fig. 5A). Note the data are corrected for transfection efficiency by cotransfection with pRL-TK renilla (materials and methods), eliminating the possibility that these, as well as other, results (below) are influenced by differential transfection efficiency of cells in different physiological states. As a further control, differentiating Caco-2 cells did significantly suppress the TOP FLASH reporter [driven by a T-cell factor-responsive promoter, (33)], to 31% of the activity in proliferating cells (Fig. 5A). As confirmation of these findings, levels of Mybl2 pre-mRNA transcripts, consisting of unspliced intron/exon boundary sequences for which expression indicates ongoing transcription, are not significantly altered in day 14 spontaneously differentiated Caco-2 cells compared with proliferating cells (Fig. 5B), whereas, in contrast, pre-mRNA transcripts of cyclin D2 and c-myc, Wnt pathway target genes transcriptionally downregulated during colon cell maturation, are decreased 76% (cyclin D2) and 55% (c-myc), respectively (Fig. 5B). Therefore, transcriptional repression cannot account for the dramatic changes in expression of Mybl2 during colon cell differentiation.
Potential miRNA binding sites in the Mybl2 3′-UTR suppress reporter gene activity.
Among potential nontranscriptional mechanisms mediating Mybl2 repression in differentiating colon cells, the contribution of miRNA-mediated repression is of particular interest, since global expression profiles of miRNAs suggest that miRNA expression is closely associated with differentiated states (39). Moreover, specific miRNAs are reported to be fundamental in differentiation of hematopoietic (12, 18, 39), neuronal (41), muscle (32, 81), and adipocyte (67) cells.
Using a functional approach, rather than global screening of miRNA expression, we identified potential miRNA targets in the Mybl2 3′-UTR by searching the Sanger database, a regularly updated repository of known miRNAs. Many predicted target regions overlapped; these were combined into 15 nonidentical potential miRNA binding sites, ranging from 17 to 47 nucleotides in length and exhibiting a range of predicted free energies, as modeled using GeneBee (http://www.genebee.msu.su/services/rna2_reduced.html) and Vienna (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi) algorithms. We hypothesize that sequences exhibiting negative free energies are more stable and likely form bona fide miRNA targets. Thus, 11 of 15 Mybl2 3′-UTR sequences predicted to form secondary structures with negative free energies (Fig. 6A) were further analyzed.
We investigated functionality of these putative miRNA targets by testing their abilities to mediate miRNA suppression of a reporter gene in spontaneously differentiating Caco-2 cells. We initially tested no. 5 and no. 7, sequences that form structures with low free energies [−6.4 (no. 5) or −5.4 (no. 7) kcal/mol] and similar lengths [47 (no. 5) or 42 (no. 7) residues], but different locations in the Mybl2 3′-UTR [10 bp (no. 5) or 296 bp (no. 7) downstream of the translational stop codon; see Fig. 6B]. Each sequence was inserted downstream of the luciferase coding region in the pGL3 control vector in either the forward or, as a negative control, reverse orientation (see materials and methods), and its ability to suppress luciferase gene expression assayed, again corrected for transfection efficiency by cotransfection of pRL-TK renilla. During spontaneous Caco-2 cell differentiation, sequence no. 5 decreased reporter gene expression by ∼60% in proliferating Caco-2 cells and by a similar, but slightly greater, extent in differentiating cells (Fig. 7). In each case, insertion of this sequence in reverse orientation either did not alter, or decreased by <50%, reporter expression (Fig. 7). In contrast, sequence no. 7 did not significantly decrease reporter expression under any condition. Thus, sequence no. 5, but not no. 7, may be a bona fide miRNA target in the Mybl2 transcript that acquires a proper configuration to bind relevant miRNAs and thus contribute to Mybl2 regulation (see discussion).
Differentiating colon cells upregulate expression of known miRNAs targeting functional Mybl2 3′-UTR sequence no. 5 in vitro and in vivo.
To determine whether known miRNAs contribute to Mybl2 suppression in maturing colon cells, we used qRT-PCR to assess expression of relevant miRNAs. In a subset of miRNAs predicted to bind functional Mybl2 3′-UTR sequence no. 5 (miR-30a, 362–5p, 10a, and 365), 30a and 365 substantially increased (mean 2.5-fold) in day 14 spontaneously differentiating Caco-2 cells compared with proliferating cells (Fig. 8A), with the largest increase of 3.8-fold seen for 365. There was a somewhat smaller increase in 10a, and a marginal increase in 362–5p. In addition, several miRNAs predicted to bind nonfunctional sequence no. 7 (miR-205, 221, 324–5p, 588, and 331–3p) also increased, but to a lower mean level of 2.1-fold over the level in proliferating cells (Fig. 8A). Differentiated cells along the human colon crypt-luminal axis in vivo also exhibit an increase in miR-362–5p (1.9-fold), 365 (3.8-fold), 324–5p (6.0-fold), and 588 (4.6-fold) relative to proliferating cells at the crypt bottom (Fig. 8B). Other miRNAs reported to play important roles in the colon, including miR-34a, 130, 145, and let-7a1, are also increased in differentiating Caco-2 cells (Fig. 9A). Interestingly, mature cells along the human crypt-villus axis exhibit increased let-7a1 but decreased miR-34a, 130, 143, and 145 expression (Fig. 9B). Importantly, transfection of miR-324–5p (targeting nonfunctional site no. 7) into subconfluent, proliferating Caco-2 cells had no effect on Mybl2 protein expression (Fig. 10). In contrast, miR-365 (targeting functional site no. 5) suppressed Mybl2 protein level ∼40% compared with that of mock-transfected cells (P < 0.0005, Fig. 10). Therefore, the functional miRNA binding site we identified in the Mybl2 3′-UTR indeed regulates levels of Mybl2 by one of its cognate miRNAs.
Exit of cells from the proliferative/progenitor cell compartment in intestinal crypts is coupled to commitment to specific lineages of differentiation. Numerous genes and pathways, including c-myc, cyclin D2, p27Kip1, Wnt signaling, and Notch signaling, regulate these transitions to mature intestinal epithelial cells. Correct functioning and homeostasis of the intestinal mucosa, including proper cessation of cell growth and commitment to differentiation, requires coordination and integration among these genes and pathways that likely contribute distinct as well as overlapping functions and regulation.
Using databases that analyze model systems of differentiation in colon epithelial cells, we identified 14 sequences commonly downregulated during maturation of intestinal epithelial cells, one of which, Mybl2, we showed to be downregulated during maturation along the crypt-luminal axes of both the mouse small intestine and the human colon in vivo (Figs. 1 and 2). Because Mybl2 is also important in differentiation of other cell types, including myeloid (8, 17, 22, 56) and neuroblastoma cells (54), and is characteristically expressed in stem/pluripotential cells (9, 47, 69, 70), it is likely that its downregulation is fundamental to intestinal cell maturation. Our recent data has, in fact, demonstrated that Mybl2 is important in linking regulation of progression through the cell cycle with commitment to differentiation (51), suggesting that, after commitment, Mybl2 suppression is important for acquisition of a fully differentiated phenotype. Therefore, in these studies, we addressed the mechanism of Mybl2 regulation, demonstrating that, during in vitro colon cell maturation, Mybl2 expression is not substantially downregulated transcriptionally (Fig. 5), consistent with the report that a posttranscriptional component contributes to regulation of this gene (56). While not excluding other possible mechanisms, we demonstrate that a potential miRNA binding site we identified in the 3′-UTR can modulate Mybl2 expression levels, that specific miRNAs predicted to recognize this site are upregulated during maturation of colon cells both in vitro and in vivo (Fig. 8), and that overexpression of one of these, miR-365, downregulates Mybl2 expression by ∼40% (Fig. 10).
Interestingly, the pattern of regulation of certain miRNAs in human colon in vivo reflects that of differentiating human colon cells in vitro. Day 14 of spontaneous induction of Caco-2 cell maturation represents an intermediate time point in the full 21-day time course of differentiation in vitro. At this time point, relative to proliferating and day 2 or 7 differentiating cells, expression of most miRNAs in Caco-2 cells is at or near their highest levels. Expression patterns of several miRNAs, including 362–5p, 365, 324–5p, 588, and let-7a1, in human colon in vivo exhibit a biphasic profile, with high levels in tissue isolated from intermediate fractions along the crypt-luminal axis (in particular fraction no. 8), likely representing colon cells at an intermediate phase of maturation, as well as in cells nearest the lumen (fraction no. 1), representing fully differentiated colon cells (Figs. 8B and 9B). These results suggest that regulation of at least certain miRNAs in Caco-2 cells in vitro reflects that of human colon cells in vivo and that miRNAs play critical roles in promoting aspects of colon cell differentiation at intermediate phases, as well as in maintaining a fully differentiated phenotype. Differences between expression patterns of miRNAs in Caco-2 cells in vitro and colon tissue in vivo, in particular for miR-30a, 10a, 205, and 331–3p, may be due to the fact that colon tissue in vivo is composed of several distinct lineages of epithelial cells, which, as a whole, may modulate individual miRNA expression differently than Caco-2 cells, which represent only the absorptive lineage. Taken together, these results suggest that there are similarities as well as differences in regulation of miRNAs in different colon cell lineages.
miRNAs likely play critical roles in colon cells, particularly in regulating pathways involved in the balance between normal cell growth and tumorigenesis. Several miRNAs, including let-7 (1, 14, 78), miR-34 (11, 26, 71), miRs −143 and −145 (2, 6, 14, 62), and miR-130a (21, 39) are repressed in colorectal tumors and human colorectal cancer cell lines relative to normal mucosa or can suppress tumor growth. Furthermore, several reports have suggested potential targets of these miRNAs in colon cells. These include Ras and c-myc for let-7 (1); cyclin E2, cyclin-dependent kinase-4, hepatocyte growth factor receptor, and E2F-1 and -3 proteins for miR-34a (26, 71); ERK-5 for miR-143 (2); and insulin receptor substrate-1 for miR-145 (62). miRNAs 143 and 145 have also been suggested to target other genes, including those involved in growth-promoting signaling pathways, by homology to known mRNA sequences (46). Moreover, a recent report indicates that miR-7 modulates the expression of the transmembrane glycoprotein CD98 during colon epithelial cell differentiation (48). These studies suggest that miRNAs play important roles in colon epithelial cells and that they may be involved in driving colon cell maturation.
Our data showing that multiple miRNAs predicted to target the functional 3′-UTR miRNA target site in Mybl2 are increased in expression during colon cell maturation are consistent with observations that miRNAs individually exert modest effects (5, 61) and act combinatorially, with a number of different miRNAs targeting the same gene. Moreover, the data support the hypothesis that factors in addition to primary sequence, including the stability of the 3′-UTR target, may influence their efficacy. Importantly, these studies demonstrate that expression profiles of miRNAs alone do not reveal their functionality. Although miR-365 and miR-324–5p are both predicted to target Mybl2 and are both increased in expression during colon epithelial cell maturation, only miR-365, which, by our analysis, is predicted to target a functional miRNA binding site in the Mybl2 3′-UTR, is able to suppress Mybl2 protein expression. miR-324–5p, predicted to bind a nonfunctional sequence in the Mybl2 3′-UTR, does not suppress Mybl2. These findings, therefore, highlight the importance of utilizing a target-centered approach to not only profile miRNA expression, but to also determine whether miRNA targets are physiologically relevant.
Several studies have documented the involvement of miR-365 in disease states, but, unlike the data herein demonstrating that miR-365 suppresses Mybl2 expression, these reports have not identified functional targets for this miRNA. miR-365 is upregulated in UVB-irradiated NIH3T3 cells (24) and human breast cancer tissue (80), and downregulated in human lupus nephritis renal tissue (15), rat vascular walls following balloon injury (31), human psoriatic skin (65), and quiescent as well as senescent (relative to replicating) human lung cells (40). miR-365 was also identified as one of 14 miRNAs upregulated in ectopic vs. eutopic endometrial tissue and predicted to target mRNAs in the c-Jun, CREB (cAMP-response element binding protein) binding protein, protein kinase b (Akt), and cyclin D1 signaling pathways (50). These studies suggest that miR-365 may play a role in fine-tuning expression of critical molecules under a number of conditions, and our data demonstrate that this encompasses Mybl2 expression during colon cell maturation.
Important to consider, however, is that the Mybl2 suppression mediated by miR-365 accounts for only part of the Mybl2 downregulation during Caco-2 cell differentiation. Because miRNAs function combinatorially, as discussed above, other upregulated miRNAs may also play a role. In this context, we have identified a number of miRNAs with documented involvement in colon cells, including miR-34a, miR-130a, miR-145, and let-7a1, as upregulated during Caco-2 cell differentiation (Fig. 9A). In particular, let-7a1 may play an important role in human colon cell differentiation, because it is also enriched in differentiating human colon tissue in vivo (Fig. 9B). Expression patterns of miR-34a, 130a, 143, and 145 in Caco-2 (absorptive) cells are very different from those (all lineages) in vivo and may indicate, as suggested above, that these miRNAs are regulated differently in distinct colon cell lineages. These, and other as yet unidentified miRNAs, may play a combinatorial role in regulating Mybl2 and other genes during colon cell maturation.
Suppression of key regulatory molecules such as Mybl2 by miRNAs suggests that expression of miRNAs themselves is subject to fine control, providing a level of fine tuning in cellular reprogramming. Indeed, we have reported that Mybl2 appears to function in fine tuning progression through the cell cycle in preparing colon cells to differentiate (51). It will, therefore, be important to determine whether expression of miR-365 and other critical miRNAs is regulated by the same molecules and pathways that drive colon epithelial cell maturation, or whether distinct mechanisms are involved. Such regulation could occur at a number of levels, potentially in distinct cellular contexts: pri-miRNA transcription, nuclear processing by the Drosha protein complex, cytoplasmic processing by Dicer, incorporation into the RNA-induced silencing complex, recognition of and binding to mRNA targets, miRNA turnover, and RNA editing of miRNAs. In addition, we hypothesize that secondary structure of the miRNA target may determine whether a potential miRNA target is functional and can bind miRNAs. It is likely that the finely regulated balance in cell proliferation and allocation of cells to different lineages that establish intestinal mucosal homeostasis will be found to be dependent on multiple levels of regulation that include miRNA and other noncoding RNA components.
This work was supported by National Cancer Institute Grant CA137508 to M. Papetti and U54 CA100926, CA114265, CA123473, and P013330 to L. H. Augenlicht.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Dr. Katheryn E. Tanaka and Navjot Kaur for human colon patient samples, Georgia Corner for assisting with gene functional group classifications, Dr. John Mariadason for Caco-2 cell microarray data and mouse small intestine crypt-villus protein fractions, Dr. Anna Velcich for HT29 microarray data, and Elena Dhima for technical assistance.
- Copyright © 2011 the American Physiological Society