Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Defining Transcriptional Regulatory Mechanisms for Primary let-7 miRNAs

  • Xavier Gaeta,

    Affiliations Broad Stem Cell Center, University of California Los Angeles, Los Angeles, California, United States of America, Molecular Biology Institute, University of California Los Angeles, Los Angeles, California, United States of America

    ⨯
  • Luat Le,

    Affiliation Molecular, Cell and Developmental Biology, University of California Los Angeles, Los Angeles, California, United States of America

    ⨯
  • Ying Lin,

    Affiliation Molecular Biology Institute, University of California Los Angeles, Los Angeles, California, United States of America

    ⨯
  • Yuan Xie,

    Affiliation Molecular, Cell and Developmental Biology, University of California Los Angeles, Los Angeles, California, United States of America

    ⨯
  • William E. Lowry

    blowry@ucla.edu

    Affiliations Broad Stem Cell Center, University of California Los Angeles, Los Angeles, California, United States of America, Molecular Biology Institute, University of California Los Angeles, Los Angeles, California, United States of America, Molecular, Cell and Developmental Biology, University of California Los Angeles, Los Angeles, California, United States of America, Jonsson Comprehensive Cancer Center, University of California Los Angeles, Los Angeles, California, United States of America

    ⨯

Abstract

The let-7 family of miRNAs have been shown to control developmental timing in organisms from C. elegans to humans; their function in several essential cell processes throughout development is also well conserved. Numerous studies have defined several steps of post-transcriptional regulation of let-7 production; from pri-miRNA through pre-miRNA, to the mature miRNA that targets endogenous mRNAs for degradation or translational inhibition. Less-well defined are modes of transcriptional regulation of the pri-miRNAs for let-7. let-7 pri-miRNAs are expressed in polycistronic fashion, in long transcripts newly annotated based on chromatin-associated RNA-sequencing. Upon differentiation, we found that some let-7 pri-miRNAs are regulated at the transcriptional level, while others appear to be constitutively transcribed. Using the Epigenetic Roadmap database, we further annotated regulatory elements of each polycistron identified putative promoters and enhancers. Probing these regulatory elements for transcription factor binding sites identified factors that regulate transcription of let-7 in both promoter and enhancer regions, and identified novel regulatory mechanisms for this important class of miRNAs.

Citation: Gaeta X, Le L, Lin Y, Xie Y, Lowry WE (2017) Defining Transcriptional Regulatory Mechanisms for Primary let-7 miRNAs. PLoS ONE 12(1): e0169237. https://doi.org/10.1371/journal.pone.0169237

Editor: Yun Zheng, Kunming University of Science and Technology, CHINA

Received: October 3, 2016; Accepted: December 13, 2016; Published: January 4, 2017

Copyright: © 2017 Gaeta et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The Chromatin-RNA seq data can be found in GSE32916. All other datasets are listed in supplemental tables 1 and 2.

Funding: This work was funded by the National Institutes of Health (NIH, P01GM99134) to WEL and an NRSA-NIH-GMS to XG. This research was also supported by the Allen Distinguished Investigator Program, through The Paul G. Allen Frontiers Group. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

The let-7 family of miRNAs were first identified in C. elegans as a single heterochronic factor controlling developmental timing[1, 2]. Since then, this family of miRNAs has been shown to play somewhat equivalent roles in all bilaterian organisms, and the let-7s were the first miRNAs identified in humans[1, 3]. The let-7s have now been implicated in differentiation and maturation of many tissues during development in vivo and in vitro[4–7]3–8. As with other miRNAs, the initial pri-let-7 transcripts are first transcribed by RNA polymerase II, then processed via the canonical pathway through the pre-miRNA stage generated by the action of Drosha/DGCR8. The pre-miRNA is then processed in the cytoplasm by Dicer to generate the mature version of the miRNA[8–10]. In addition, in the case of let-7 miRNAs, other processes such as uridylation are used to stabilize or destabilize miRNAs[11–13]. LIN28A and LIN28B are RNA binding proteins that regulate several of these processing steps to control levels of mature let-7 transcripts[14, 15]. Over evolution, let-7 isoforms have expanded such that the human genome contains 9 isoforms. The study of regulation of the let-7 family of miRNAs has focused on these processing steps, but less is understood about how the pri-let-7 transcripts are regulated by transcription prior to any processing.

Studies in C. elegans, where the activity and expression of let-7 is regionally and temporally constrained, have attempted to clarify transcriptional regulation from the single let-7 locus. Two regulatory regions upstream of the locus were identified as the temporally regulated expression binding site (TREB) and the let-7 transcription element (LTE), and many studies have tested the binding and transcriptional control exerted by several TFs including elt-1 and daf-12[2, 16–18]. These sequences are not present upstream of mammalian let-7 gene, and there are not similarly consistently present sequences near all the different let-7 loci. In higher organisms, a different system for regulating let-7 miRNA transcription must have been established.

The study of mammalian pri-let-7 transcription is hampered by the relative scarcity of the transcript which is processed immediately in the nucleus and therefore difficult to detect. We previously took advantage of a method that allows for the capture of nascent RNA transcripts, which are still associated with the chromatin from which they are transcribed, to carefully annotate pri-let-7 transcripts[19, 20]. Another group later induced pri-let-7 accumulation in the context of DGCR8 knockout, and validated with RACE PCR that primary let-7 transcripts have multiple isoforms, some of which aligned nearly identically to our observed annotation patterns and varied in different cellular contexts[21]. From these annotations, it is clear that many let-7 family members are transcribed within very long (up to 200KB), often polycistronic transcripts[20, 21]. While some studies have identified transcriptional models of pri-miRNAs in higher organisms, the lack of proper annotation left the precise regulatory motifs for human let-7 transcripts undefined. Here, after complete annotation of let-7 transcripts, we attempt to define regulatory motifs for this family of miRNAs by taking advantage of Chromatin-associated RNA-seq and the latest genomic descriptions of chromatin states within let-7 loci. We model let-7 transcription in distinct neural paradigms to reveal subsets of let-7 family members that are transcribed constitutively versus dynamically regulated in particular contexts. Finally, by analyzing publically available data for let-7 loci, we identify transcription factors that appear to regulate let-7 transcription by acting at either promoter or enhancer elements enriched in dynamically regulated let-7 polycistrons.

Results

Identification of dynamics of let-7 polycistron transcription

As a first step to determine how let-7 miRNAs are transcriptionally regulated, we attempted to define developmental models that display dynamism of transcription. We previously identified dynamic transcriptional regulation of some let-7 family members between neural progenitors that represent distinct developmental stages[20]. This developmental system has been described in our previous studies[19, 20, 22], and has become routine in the field. Initially, human pluripotent stem cells are directed towards a neural fate by changing the media. Then, as neural rosettes are formed, they are manually isolated and expanded in neural proliferation media containing EGF and FGF. To further differentiate towards later lineages, we used growth factor withdrawal, where EGF and FGF are removed and the neural progenitors are forced to differentiate towards neurons and glia. We validated the identity of cells generated at each step by immunostaining for markers typical of each stage of specification (pluripotent: OCT4/NANOG, NPC: SOX2, SOX1, Neuron: MAP2, TuJ1), and consider the culture to be suitable homogenous if the cells are at least 90% positive for combinations of markers.

We previously showed that pri-let-7 transcripts can be identified by Chromatin-associated RNA-seq data[20](NIH GEO Dataset GSE32916). This method captures RNA still associated with chromatin, and therefore represents nascent messages[23]. Our previous analysis initially predicted that some pri-let-7s could be dynamically regulated, so we extended these analyses here. Here we show that there is dynamism of let-7 transcription as measured by Chromatin-associated RNA-seq as witnessed by the fact that the let-7a3/b locus is practically silent in pluripotent stem cells, and neural progenitors derived therein, but strongly expressed in tissue derived neural progenitors (Fig 1A). This is consistent with the idea that tissue derived progenitors represent a later stage of development than pluripotent derived progenitors[19, 20].

thumbnail
Download:
Fig 1. Dynamic transcriptional regulation of some pri-let-7 transcripts.

Chromatin-associated RNA-seq reads were mapped onto two distinct polycistronic let-7 loci. At left, the let-7a3/b locus, is dynamic, while at right, the let-7a1/d/f1 locus, is constitutively expressed. Reads are shown for ESC, iPSC, PSC-derived NPC, and neural tissue-derived NPC stages. These reads are aligned with validated primary miRNA transcripts from RACE PCR experiments in green and RefSeq annotated genes in blue25. Note that Chromatin-associated RNA-seq and RACE PCR annotated transcripts demonstrate the existence of longer transcripts from different transcriptional start sites than suggested by the RefSeq annotation. In the case of let-7a1/d/f1, this discrepancy extends to the strand from which initial transcription occurs.

https://doi.org/10.1371/journal.pone.0169237.g001

On the other hand, the same analysis for the let-7a1/d/f1 locus showed that this polycistron is constitutively expressed across all cell types assayed (Fig 1B). Furthermore, Chromatin-associated RNA-seq also allows for mapping reads which highlighted the fact that let-7 transcripts are long and sometimes polycistronic. Finally, this allowed for proper annotation of these polycistronic transcripts by actual measurement of message as opposed to the RefSeq annotations that were performed by localizing epigenetic markers. In so doing, we find that the RefSeq annotations underestimate the length of the let-7 polycistrons. The annotations resulting from the Chromatin-associated RNA-seq are then highly overlapping with those in another study, Chang et al[21]. A complete presentation of transcriptional data from the other let-7 loci as demonstrated by Chromatin-RNA-seq is in (S1 Fig).

Using RNA from cultures previously described [20], we performed Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) to show that as human pluripotent stem cells are specified to neural progenitors, and subsequently into neurons, some primary let-7 transcripts are strongly induced as measured by RT-PCR (Fig 2). Because pri-miRNAs are transcribed as longer messages, they can be specifically identified by designing at least one of the PCR primers to recognize strictly cDNA made from the portions of the pri-miRNA message not found in pre-mrRNA or mature miRNA. In both developmental scenarios, we observed that a subset of let-7 family members showed transcriptional induction over developmental time, while other members appeared to be constitutively transcribed (Fig 2A). Using primers that recognize the mature version of miRNA, RNA isolated in a manner that enriches for small RNA, and a specialized cDNA synthesis kit (miScript) we also specifically analyzed levels of mature let-7s. We found that the levels of all mature let-7 family members were strongly induced across development (Fig 2B).

thumbnail
Download:
Fig 2. Expression of pri-let-7 during neural specification.

Pluripotent stem cells were differentiated through the neural lineage to neural progenitor cells (NPCs) and then to neurons. Using RT-PCR with primers specific to the let-7 miRNAs at different stages of processing, we tested changes in expression of the pri-let-7s (A) and their mature forms (B). While all mature miRNAs increased over the course of differentiation, only a subset (marked with dotted lines), the dynamically regulated let-7s, also increased before processing, at the primary let-7 stage. RT-PCRs were also performed beginning with ES cells. (C) Graphic comparing the length of the RefSeq annotated let-7a3/b with our predicted transcript. Stars mark primer pairs for RT-PCR along the full transcript. (D) RT-PCR of pri-let-7a3/b transcript in tissue-derived NPCs, in which transcription is abundant. In control, siDGCR8 (to block Microprocessor function and pri-to-pre conversion), and siDICER (to block pre-to-mature conversion) conditions. When Microprocessor is disabled, the entire let-7a3/b transcript accumulates.

https://doi.org/10.1371/journal.pone.0169237.g002

As evidence for the long length of these transcripts, RT-PCR was performed using primers that recognize different regions of the predicted transcript from the let-7a3/b locus (Fig 2C). In addition, we posited that this transcript would accumulate in abundance if downstream processing by DGCR8 was inhibited. siRNA-mediated silencing of DGCR8 increased levels of the let-7a3/b transcript as measured by all the primers across the entire predicted polycistron. Silencing of DICER, necessary for the final step of miRNA processing, did not change the level of any portion of the let-7a3/b transcript (Fig 2D). As further evidence that let-7 transcripts are polycistronic, the data in Fig 1A and 1B on dynamic versus constitutive indeed showed a shared pattern for those let-7s that are in the same polycistron. For instance, the pattern of let-7a and let-7b was conserved and dynamic in both contexts, while let-7a1, let-7d and let-7f1, which are also polycistronically transcribed, were constitutively expressed in both contexts.

Identification of potential epigenetic regulation of let-7 polycistrons

We then sought to determine whether the dynamic versus constitutive let-7 polycistrons display distinct regulatory schemes. Using data from the Epigenetic Roadmap, we annotated the chromatin states across each polycistronic let-7 locus (Fig 3 and S2 Fig). The Roadmap database includes data from dozens of human cell types, including several of the neural lineage and pluripotent stem cells, both highly relevant to our current study[24, 25]. This analysis showed a clear distinction between the regulatory framework for the dynamically regulated let-7a3/b locus, versus that of the constitutively expressed let-7a1/let-7d/let-7f1locus. The clearest distinction comes in the form of location of epigenetic marks for enhancers and transcriptional start sites (TSSs), where the let-7a3/b locus contains several possible start sites and putative enhancers, while the let-7a1/let-7d/let-7f1 locus appears to only have one predicted start site and regulatory scheme. We posit that having multiple possible TSSs could indicate a message that is dynamically regulated in a variety of settings.

thumbnail
Download:
Fig 3. Dynamically and constitutively transcribed let-7 loci show distinct epigenetic signatures.

Computationally imputed chromatin states generated by the ChromHMM algorithm at the same let-7 loci. Each row represents one biological sample. These states show active transcriptional marks at the predicted TSS for let-7a1/d/f1 in multiple cell types. At the let-7a3/b locus, ES cells, iPS cells, and PSC-derived NPCs have marks consistent with poised promoters, but later in differentiation active TSS marks appear at the same sites, reflecting changes in epigenetic state during neural differentiation. Epigenetic marks in K562 leukemia cells show active transcription at the RefSeq annotated let-7a3/b locus.

https://doi.org/10.1371/journal.pone.0169237.g003

Using these data and the imputed chromatin state model in tamed, we clearly identified TSSs, promoters (active and poised), enhancers, and actively transcribed regions for two of the let-7 polycistrons (Fig 3). As further evidence for their polycistronic nature, these updated epigenetic data from a wide variety of primary cell types again predicted single, long transcripts across entire loci that encompass multiple let-7 family members, as opposed to older analyses on transformed cell lines upon which the RefSeq annotations were created. Importantly, some of the genomic state models predicted variation of states in distinct cell types. Notably, the predicted promoter of let-7a3/b was shown to be poised in hPSCs and hPSC-derived NPCs, and active and transcribed in later neural derivatives and in brain. This pattern is highly consistent with our own transcriptional data whereby the pri-let-7a3/b polycistron was not transcribed significantly until hPSC-derived NPCs were driven further to neurons (Fig 2A and 2B).

Functionally defining regulators of let-7 transcription

Globally, the utility of these analyses was to define more precisely the location of promoters and enhancers for each of the pri-let-7s. Taking advantage of the annotation of promoters, we attempted to identify mechanisms of transcriptional regulation of the dynamic versus constitutively regulated let-7 polycistrons. With a focus on let-7a3/b, we searched for transcription factors that could regulate this polycistron through interactions at the promoter. We first used transcription factor ChIP-Seq data from the ENCODE and the Epigenetic Roadmap datasets to detect transcription factor binding sites enriched in this promoter (Fig 4A). We then narrowed the list of candidates to include just those whose expression changes in contexts where let-7a3/b transcription also changes. This led to the identification of 10 TFs as defined by previous data showing their ability to both bind DNA and affect transcription of target genes (Fig 4B). To functionally determine whether any of these TFs can affect let-7a3/b transcription, we silenced some of them in tissue-NPCs (where transcription of let-7a3/b is high) and performed RT-PCR. In addition, we also targeted MYCN as a positive control because of its previously established ability to regulate let-7 transcription [1, 26, 27] and based on its expression pattern in our neurodevelopmental model. Silencing of N-MYC, AP2a, or EGR1 all appeared to lead to an increase in let-7a3/b transcription after just two days (Fig 4C and 4D), indicating a role for these TFs in transcriptional regulation of this polycistron.

thumbnail
Download:
Fig 4. Transcription factors predicted to bind to the let-7a3/b promoter regulate primary let-7a3/b transcription.

(A) Comparison of transcription factors with experimentally determined binding sites to the bona fide let-7a3/b promoter from the ENCODE database with genes differentially expressed between tissue-derived NPCs (in which let-7a3/b is abundantly transcribed) and PSC-derived NPCs (in which it is not). 10 genes were present in both sets, and are shown at right, ranked by their fold change of expression between tissue-derived and PSC-derived NPCs from microarray based gene expression measurements. We knocked down several of these candidate let-7 regulator transcription factors in tissue-derived NPCs. (B) Knockdown of the TFAP2C gene encoding the AP-2γ protein, and of the MYCN gene increase transcription of several let-7 genes. Data shown are representative of 3 independent experiments. (C) Knockdown of EGR1 increases transcription of primary let-7b and other let-7 genes. Error bars are ± SEM from n = 3 biological replicates.

https://doi.org/10.1371/journal.pone.0169237.g004

To focus on potential regulatory mechanisms at enhancers, we next looked for putative enhancers by looking for regions of enriched DNAse-hypersensitivity, peaks of H3K27ac, and peaks associated with p300 binding in the let-7a3/b locus (Fig 5A). Several predicted enhancers, outlined in red rectangles, were highly DNAse sensitive region in Fetal and Adult brain samples but not in PSCs or PSC-derived NPCs. This pattern correlates with the timing of increased pri-let7a3/b transcription. A different site, 10kb upstream of the newly annotated TSS, is outlined in a green rectangle, and instead showed DNAse sensitivity only in PSC-derived NPCs, and not in either the undifferentiated or fully differentiated cells in the database. All of these identified regions showed P300 binding, were surrounded by ChIP-seq peaks for acetylated H3K27 in neural samples, and were significant for a specific depletion of histone-associated ChIP-seq binding peaks right at the site of DNAse sensitivity.

thumbnail
Download:
Fig 5. FOX proteins are predicted to bind to putative let-7a3/b enhancer regions.

(A) The predicted existence of an upstream enhancer for the let-7a3/b locus was based on the epigenetic state at a region 10kb upstream of the TSS, outlined in green. In addition to being marked by H3K27Ac ChIP-Seq peaks with a localized dip in signal intensity, and peaks for the enhancer-associated histone acetyltransferase protein P300, this region showed dynamic changes in DNAse sensitivity. Note that a large DNAse sensitivity peak appears only in ES-derived NPCs, suggesting a differentiation state-specific chromatin opening at this region. At bottom, relative intensity of forkhead box protein ChIP-Seq from multiple cell types are pooled, with the darkest regions indicating intense FOX protein binding. Outlined in red are similar regions that show DNAse sensitivity beginning at the fetal brain stage that also colocalize with FOX protein binding. (B) A zoomed in view of the green region of increased DNAse sensitivity in PSC-derived NPCs. In blue are computationally predicted transcription factor binding sites from the ORCAtk database. The degree of genomic conservation along this region from the PhastCons64 database is shown in purple. At bottom are transcription factor ChIP-seq mapped peaks from the ENCODE database. The regions in green mark forkhead box transcription factor conserved motifs. Note that the forkhead box motifs co-localize with a region of highly conserved sequence, and the redundant binding of the forkhead box motif by many family members predicts that many such proteins can bind there.

https://doi.org/10.1371/journal.pone.0169237.g005

We used a similar approach to identify predicted and validated TF binding sites in the enhancer regions as on promoter sequences (Fig 5B). This analysis yielded a strong enrichment of binding by the forkhead box transcription factors (FOX proteins), all of which can bind the same motif: 5'-[AC]A[AT]T[AG]TT[GT][AG][CT]T[CT]-3'[28]. Note the increased intensity of FOX protein ChIP-seq signal within the putative enhancers in Fig 5A. The furthest upstream such region, outlined in green, is shown in more detail, with both predicted and experimental FOX protein binding localizing to one highly conserved area (Fig 5B). The forkhead box TFs contain winged helix domains, which contribute to the pioneer transcription factor activity of the entire family and in some settings could be responsible for observed changes in chromatin accessibility[29–32]. We compared this finding with expression data to subsequently determine which candidate forkhead box TFs could potentially be acting at the let-7a3/b enhancer during neural development (Fig 5B).

The forkhead box proteins FOXP2, FOXP1, FOXP4, FOXN2, FOXN3, FOXN4, and FOXG1 showed both high baseline expression and dynamic changes in expression over the course of nervous system development (Fig 6A). FOXP2’s role in brain development is linked closely to its involvement in diseases of speech and language[33, 34]. In the murine developing spinal cord and cortex, Foxp2 and Foxp4 are expressed in neural progenitor cells and increase in abundance during neuronal differentiation[33]. In Foxp4-/- mice, these NPCs fail to exit the progenitor stage and cause major disruptions in the developing neural tube. No function in the nervous system has been ascribed to either FOXN2 or FOXN3, but murine Foxn4 is expressed in the brain and retina, and is necessary for the specification of retinal amacrine cells[35]. Taken together, forkhead box proteins have the molecular components necessary to induce reorganizations of the epigenetic state, and some are expressed at anatomic locations and times that correlate with let-7 expression.

thumbnail
Download:
Fig 6. A role for FOX proteins in regulation of pri-let-7s.

(A) By filtering for FOX genes that are actively transcribed in our neural cells and differentially expressed between PSC-derived NPCs and Tissue-derived NPCs, we generated a list of candidate proteins that might mediate changes in let-7a3/b transcription. (B) Knockdown of FOX proteins, FOXP2, FOXP4 and FOXN3 in Tissue-derived NPCs show distinct effects on primary let-7 transcript levels. Statistics were performed across three independent experiments (two-tailed t-test, p < 0.05).

https://doi.org/10.1371/journal.pone.0169237.g006

FOXP2, FOXP4, and FOXN3 are all suppressed over the course of our neurodevelopmental model (Fig 6A). Silencing either FOXP4 or FOXP2 in human neural progenitors appeared to inhibit expression of the let-7a3/b, let-7e and let-7i loci, while having nearly the opposite effect on the let-7a1/d/f1 and let7c loci (Fig 6B). On the other hand, silencing FOXN3 did not affect let-7a3b, but did induce let-7c and let-7e (Fig 6B). The fact that the polycistronic let-7 pri-miRNAs appeared to be regulated in concert as a result of these manipulations is further evidence of the co-regulatory mechanisms used during cell fate decision-making. Together, these data demonstrate that proper annotation of let7 loci can facilitate prediction of regulatory elements that are bound by transcription factors with the ability to regulate let-7 transcription.

Discussion

Together, these analyses define contexts in which particular let-7 polycistrons are transcriptionally regulated, and identify TFs that play roles in this dynamism. This study is not the first to identify transcriptional mechanisms for let-7 family members, but previous studies from lower organisms did not take advantage of genome-wide analyses to systematically define regulatory modules or transcription factors that regulate them. The fact that let-7 miRNAs can be dynamically regulated at the transcriptional level has only recently been appreciated, but the relative contribution of this regulation relative to levels of mature let-7s remains undefined. This is potentially an important issue to resolve as recent evidence suggests that not all let-7 miRNAs are processed by the same machinery[36], and therefore, the level of mature let-7 might not simply be DICER dependent.

These issues bring to light an interesting question, why have mammals evolved to have so many let-7 isoforms in their genomes, and why do so in polycistronic fashion. Because all the let-7 family members have the same seed sequence, it seems redundant to express so many. Even in the early neural lineage where mature let-7s are scarce, some of the let-7 polycistrons are not transcribed, whereas others appear to be constitutively expressed. While we can only speculate, it is possible that both dynamic and constitutive let-7 transcription is a function of feed-back activity of let-7-target interactions. It is worth pointing out that some let-7 targets also regulate let-7 maturation, such as LIN28A, LIN28B and LIN41. Furthermore, it has been proposed that some let-7 target RNAs can act as ceRNA or sponges of mature let-7 to regulate their activity[37]. In addition, some of the TFs shown here and elsewhere to regulate let-7 transcription (e.g. N-MYC) are also let-7 target genes[27, 38, 39]. Perhaps, the constitutive transcription and maturation of small amounts of let-7 serves as something of a rheostat of developmental timing that is tuned as cells become more specified, leading to changes in let-7 targeted TFs that can then in turn regulate let-7 transcription, leading to even more mature let-7 through an additional feed-forward mechanism.

In C. elegans, where let-7s were first discovered, there is evidence for both transcriptional and maturation control despite the fact that all let-7 is transcribed from a single locus. In fact, there are two distinct transcriptional start sites and these are distinctly regulated by both cis and trans mechanisms. There is further evidence that let-7s play a more general role in miRNA biogenesis through an interaction with Argonaute[40, 41]. Therefore, sophisticated mechanisms for let-7 regulation have been preserved and expanded across evolution, perhaps pointing to their critical roles in both developmental timing and tumorigenesis. These issues are highly relevant to the study of cancer, where let-7 targets are strongly induced, consistent with a loss of mature let-7. It is possible that transcriptional induction of let-7 family members could be a strategy to drive a cascade of re-expression of let-7 in cancerous tissues, akin to the process which appears to happen during early human development.

Materials and Methods

Cell culture

Pluripotent stem cell culture and differentiation into NPCs and neurons was performed as previously described6. Briefly, PSCs were induced to differentiate along the neuroepithelial lineage by treatment with dual inhibitors of the SMAD signaling pathway, SB431542 (Sigma, 5 μM) and LDN193189 (Sigma, 50 nM). Neuroepithelial rosettes were manually picked and replated onto plates coated with ornithine and laminin. Cells were maintained and expanded in NPC media, containing DMEM/F-12 (Gibco), B27 supplement (Gibco), N2 (Gibco), EGF and bFGF. To induce differentiation, cells were fed with media lacking EGF and bFGF for 3 weeks. Tissue-derived NPCs were cultured and differentiated with the same reagents6.

siRNA transfection

Gene knockdowns were performed by transfecting cells with double stranded 27mer RNAs (OriGene) using the lipofectamine RNAiMAX reagent (Thermo Fisher) according to the protocol provided.

Measurements of gene expression by RT-PCR

Cells were lysed in Trizol lysis reagent (Thermo Fisher), and total RNA was purified from lysates using the QIAgen miRNeasy kit. cDNA was made by reverse transcription from mRNAs with the SuperScript III First Strand Synthesis system (Thermo Fisher), or from miRNAs with the miScript II RT Kit (Qiagen). Realtime PCR was performed on a Roche Lightcycler 480 instrument. For mRNA-derived cDNA, Roche 480 SYBR green I was used. For miRNA-derived cDNA, miScript SYBR (Qiagen) was used. To calculate relative amounts of transcripts, the Real-time data were calculated based on expression levels of housekeeping genes (GAPDH for mRNAs; U6 for miRNAs). miR-15 was included as a control in RT-PCR experiments as this has been proposed to be constitutively expressed in a variety of settings.

Epigenome characterization and candidate TF prediction

Summary ENCODE and Roadmap gene expression data, ChIP-Seq mapping data, and ChromHMM chromatin state prediction were accessed and visualized using the UCSC genome browser and the WashU Epigenome Browser43,44. These tools were also used to import and visualize Chromatin-associated RNA-seq reads from Patterson et al. and miRNA gene transcripts in cells lacking DGCR8 from Chang et al6,25.Transcription factor binding site predictions were performed with the ORCA Toolkit web server, and with the MEME suite of motif analysis applications45,46.

Expanded method

The ENCODE and Epigenetic Roadmap datasets, with accession numbers listed in Tables 1 and 2, contain mapped reads from various gene expression and ChIP-sequencing experiments performed on a panel of cell lines and cell types isolated from primary human tissue. These datasets were accessed and imported using the These datasets were accessed and imported using the UCSC genome browser and the WashU Epigenome Browser to co-register and visualize enrichment of epigenetic characteristics across the genome. In figures utilizing Roadmap expression data, track intensity is a logarithmic graph of p-value signal.

thumbnail
Download:
Table 1. Data from Roadmap project in GEO database.

https://doi.org/10.1371/journal.pone.0169237.t001

thumbnail
Download:
Table 2. Data from ENCODE project in GEO database (or DCC Accession #).

https://doi.org/10.1371/journal.pone.0169237.t002

The genome regions surrounding known let-7 gene locations were surveyed for the presence of histone modifications and open chromatin in cell types representative of the stages of differentiation from PSCs to NPCs and neurons. We also imported miRNA gene transcripts in cells lacking DGCR8 from Chang et al, which are accessible using the NIH SRA database, accession SRP057660. Together, these data allowed us to identify primary let-7 transcripts, based on their expression in our Chromatin-associated RNA-seq samples and in DGCR8-/- RNA-seq samples, even when they disagreed with RefSeq-annotated MIRLET7 genes. Hypothesized regulatory regions were assembled by searching 20 kilobases upstream and downstream of each transcript for colocalization of H3K27Ac, H3K4me3, and DNAse sensitivity in samples known to express let-7 primary transcripts, and H3K27me3 or H3K9me3 in samples without appreciable primary let-7 transcripts. These regions were frequently annotated as Active TSS, Flanking active TSS, Bivalent/Poised TSS, Enhancer, Genic Enhancer, and Bivalent Enhancer in the ChromHMM chromatin state prediction algorithm performed on Roadmap datasets.

These hypothesized regulatory regions were then queried for transcription factor binding sites, based on known and predicted TF binding sites and motifs. We used both the ORCA Toolkit web server and the MEME suite of motif analysis applications to isolate highly conserved (>80% phastCons score) sub-regions within these hypothesized regulatory regions, and then queried those conserved sub-regions for TF motifs. These motifs were assembled from the JASPAR motif database as well as a small list of manually curated motif sequences. Where possible, experimental ChIP-Seq validated TF binding sites from ENCODE and ROADMAP dataset were used as secondary validation of predicted TF binding sites. Accession numbers for these datasets are also found in Tables 1 and 2.

Supporting Information

S1 Fig. Complete annotation of let-7 miRNA transcripts and regulation in human PSCs and NPCs by Chromatin RNA-seq.

Shown are each of the let-7 family member transcripts, including polycistrons. The top of the graphic shows the genomic locus. The middle section are data from the Chromatin RNA-seq described in Fig 1. Below in green are the annotations for let-7 miRNAs described in Cheng et al in the indicated cell types. Below in blue are the annotations according to public genome browsers.

https://doi.org/10.1371/journal.pone.0169237.s001

(PDF)

S2 Fig. Annotation of epigenetic marks at two let-7 polycistronic loci.

Epigenetic marks from the Roadmap Epigenomics project at the dynamic (let-7a3/b) and constitutive (let-7a1/d1/f1) polycistronic loci. At top are the Chromatin-associated RNA-Seq peaks and RefSeq annotations of the primary let-7 transcripts, and below are the relative intensities of DNAse sensitivity or histone modification ChIP-Seq peaks at those loci.

https://doi.org/10.1371/journal.pone.0169237.s002

(PDF)

S3 Fig. Complete annotation of let-7 miRNA transcripts and summary of available data on epigenetic marks across various cell types.

Shown are the let-7 genomic loci with accompanying epigenetic marks as identified by ChIP-seq data available from the epigenetic roadmap across the indicated cell types. The bottom portion also includes available ChIP-seq data on the indicated transcription factor binding patterns at these same loci.

https://doi.org/10.1371/journal.pone.0169237.s003

(PDF)

Acknowledgments

We highly appreciate the outstanding technical support given to this work by Jessica Cinkornpumin. This work was funded by the National Institutes of Health (NIH, P01GM99134) to WEL and an NRSA-NIH-GMS (GM113641-01) to XG. This research was also supported by the Allen Distinguished Investigator Program, through The Paul G. Allen Frontiers Group

Author Contributions

  1. Conceptualization: WL XG.
  2. Data curation: XG.
  3. Formal analysis: XG.
  4. Funding acquisition: WL.
  5. Investigation: XG LL YL YX.
  6. Methodology: XG WL YL.
  7. Project administration: WL.
  8. Resources: WL.
  9. Software: XG.
  10. Supervision: WL.
  11. Validation: XG LL.
  12. Visualization: XG WL.
  13. Writing – original draft: XG.
  14. Writing – review & editing: XG WL.

References

  1. 1. Pasquinelli AE, Reinhart BJ, Slack F, Martindale MQ, Kuroda MI, Maller B, et al. Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature. 2000;408(6808):86–9. Epub 2000/11/18. pmid:11081512
  2. 2. Reinhart BJ, Slack FJ, Basson M, Pasquinelli AE, Bettinger JC, Rougvie AE, et al. The 21-nucleotide let-7 RNA regulates developmental timing in Caenorhabditis elegans. Nature. 2000;403(6772):901–6. Epub 2000/03/08. pmid:10706289
  3. 3. Schulman BR, Esquela-Kerscher A, Slack FJ. Reciprocal expression of lin-41 and the microRNAs let-7 and mir-125 during mouse embryogenesis. Developmental dynamics: an official publication of the American Association of Anatomists. 2005;234(4):1046–54. Epub 2005/10/26.
  4. 4. Cimadamore F, Amador-Arjona A, Chen C, Huang CT, Terskikh AV. SOX2-LIN28/let-7 pathway regulates proliferation and neurogenesis in neural precursors. Proc Natl Acad Sci U S A. 2013;110(32):E3017–26. Epub 2013/07/26. pmid:23884650
  5. 5. Wong SS, Ritner C, Ramachandran S, Aurigui J, Pitt C, Chandra P, et al. miR-125b promotes early germ layer specification through Lin28/let-7d and preferential differentiation of mesoderm in human embryonic stem cells. PLoS ONE. 2012;7(4):e36121. Epub 2012/05/01. pmid:22545159
  6. 6. Kawahara H, Imai T, Okano H. MicroRNAs in Neural Stem Cells and Neurogenesis. Frontiers in neuroscience. 2012;6:30. Epub 2012/03/15. pmid:22416227
  7. 7. Bussing I, Slack FJ, Grosshans H. let-7 microRNAs in development, stem cells and cancer. Trends Mol Med. 2008;14(9):400–9. Epub 2008/08/05. pmid:18674967
  8. 8. Tokumaru S, Suzuki M, Yamada H, Nagino M, Takahashi T. let-7 regulates Dicer expression and constitutes a negative feedback loop. Carcinogenesis. 2008;29(11):2073–7. Epub 2008/08/14. pmid:18700235
  9. 9. Banerjee D, Slack F. Control of developmental timing by small temporal RNAs: a paradigm for RNA-mediated regulation of gene expression. BioEssays: news and reviews in molecular, cellular and developmental biology. 2002;24(2):119–29. Epub 2002/02/09.
  10. 10. Grishok A, Pasquinelli AE, Conte D, Li N, Parrish S, Ha I, et al. Genes and mechanisms related to RNA interference regulate expression of the small temporal RNAs that control C. elegans developmental timing. Cell. 2001;106(1):23–34. Epub 2001/07/20. pmid:11461699
  11. 11. Hagan JP, Piskounova E, Gregory RI. Lin28 recruits the TUTase Zcchc11 to inhibit let-7 maturation in mouse embryonic stem cells. Nat Struct Mol Biol. 2009;16(10):1021–5. Epub 2009/08/29. pmid:19713958
  12. 12. Lehrbach NJ, Armisen J, Lightfoot HL, Murfitt KJ, Bugaut A, Balasubramanian S, et al. LIN-28 and the poly(U) polymerase PUP-2 regulate let-7 microRNA processing in Caenorhabditis elegans. Nat Struct Mol Biol. 2009;16(10):1016–20. Epub 2009/08/29. pmid:19713957
  13. 13. Heo I, Joo C, Cho J, Ha M, Han J, Kim VN. Lin28 mediates the terminal uridylation of let-7 precursor MicroRNA. Mol Cell. 2008;32(2):276–84. pmid:18951094
  14. 14. Piskounova E, Viswanathan SR, Janas M, LaPierre RJ, Daley GQ, Sliz P, et al. Determinants of microRNA processing inhibition by the developmentally regulated RNA-binding protein Lin28. The Journal of biological chemistry. 2008;283(31):21310–4. Epub 2008/06/14. pmid:18550544
  15. 15. Viswanathan SR, Daley GQ, Gregory RI. Selective blockade of microRNA processing by Lin28. Science. 2008;320(5872):97–100. Epub 2008/02/23. pmid:18292307
  16. 16. Hammell CM, Karp X, Ambros V. A feedback circuit involving let-7-family miRNAs and DAF-12 integrates environmental signals and developmental timing in Caenorhabditis elegans. Proc Natl Acad Sci U S A. 2009;106(44):18668–73. Epub 2009/10/16. pmid:19828440
  17. 17. Grosshans H, Johnson T, Reinert KL, Gerstein M, Slack FJ. The temporal patterning microRNA let-7 regulates several transcription factors at the larval to adult transition in C. elegans. Dev Cell. 2005;8(3):321–30. Epub 2005/03/02. pmid:15737928
  18. 18. Johnson SM, Lin SY, Slack FJ. The time of appearance of the C. elegans let-7 microRNA is transcriptionally controlled utilizing a temporal regulatory element in its promoter. Dev Biol. 2003;259(2):364–79. Epub 2003/07/23. pmid:12871707
  19. 19. Patterson M, Chan DN, Ha I, Case D, Cui Y, Handel BV, et al. Defining the nature of human pluripotent stem cell progeny. Cell research. 2012;22:178–93. Epub 2011/08/17. pmid:21844894
  20. 20. Patterson M, Gaeta X, Loo K, Edwards M, Smale S, Cinkornpumin J, et al. let-7 miRNAs Can Act through Notch to Regulate Human Gliogenesis. Stem cell reports. 2014. Epub 2014/10/16.
  21. 21. Chang TC, Pertea M, Lee S, Salzberg SL, Mendell JT. Genome-wide annotation of microRNA primary transcript structures reveals novel regulatory mechanisms. Genome Res. 2015;25(9):1401–9. pmid:26290535
  22. 22. Xie Y, Zhang J, Lin Y, Meng M, Wisidagamad D, Cinkornpumina J, et al. Defining the role of oxygen tension in human neural progenitor fate. Stem cell reports. 2014;in revision.
  23. 23. Bhatt DM, Pandya-Jones A, Tong AJ, Barozzi I, Lissner MM, Natoli G, et al. Transcript dynamics of proinflammatory genes revealed by sequence analysis of subcellular RNA fractions. Cell. 2012;150(2):279–90. Epub 2012/07/24. pmid:22817891
  24. 24. Mullard A. The Roadmap Epigenomics Project opens new drug development avenues. Nat Rev Drug Discov. 2015;14(4):223–5. pmid:25829267
  25. 25. Amin V, Harris RA, Onuchic V, Jackson AR, Charnecki T, Paithankar S, et al. Epigenomic footprints across 111 reference epigenomes reveal tissue-specific epigenetic regulation of lincRNAs. Nat Commun. 2015;6:6370. pmid:25691256
  26. 26. Guan H, Zhang P, Liu C, Zhang J, Huang Z, Chen W, et al. Characterization and functional analysis of the human microRNA let-7a2 promoter in lung cancer A549 cell lines. Molecular biology reports. 2011;38(8):5327–34. Epub 2011/03/03. pmid:21365266
  27. 27. Chang TC, Zeitels LR, Hwang HW, Chivukula RR, Wentzel EA, Dews M, et al. Lin-28B transactivation is necessary for Myc-mediated let-7 repression and proliferation. Proc Natl Acad Sci U S A. 2009;106(9):3384–9. pmid:19211792
  28. 28. Hosseinpour B, Bakhtiarizadeh MR, Khosravi P, Ebrahimie E. Predicting distinct organization of transcription factor binding sites on the promoter regions: a new genome-based approach to expand human embryonic stem cell regulatory network. Gene. 2013;531(2):212–9. Epub 2013/09/18. pmid:24042128
  29. 29. Soufi A, Donahue G, Zaret KS. Facilitators and impediments of the pluripotency reprogramming factors' initial engagement with the genome. Cell. 2012;151(5):994–1004. Epub 2012/11/20. pmid:23159369
  30. 30. Sekiya T, Muthurajan UM, Luger K, Tulin AV, Zaret KS. Nucleosome-binding affinity as a primary determinant of the nuclear mobility of the pioneer transcription factor FoxA. Genes Dev. 2009;23(7):804–9. pmid:19339686
  31. 31. Zaret KS, Watts J, Xu J, Wandzioch E, Smale ST, Sekiya T. Pioneer factors, genetic competence, and inductive signaling: programming liver and pancreas progenitors from the endoderm. Cold Spring Harb Symp Quant Biol. 2008;73:119–26. pmid:19028990
  32. 32. Cuesta I, Zaret KS, Santisteban P. The forkhead factor FoxE1 binds to the thyroperoxidase promoter during thyroid cell differentiation and modifies compacted chromatin structure. Mol Cell Biol. 2007;27(20):7302–14. pmid:17709379
  33. 33. Tang BL. Molecular genetic determinants of human brain size. Biochem Biophys Res Commun. 2006;345(3):911–6. pmid:16716254
  34. 34. Enard W, Przeworski M, Fisher SE, Lai CS, Wiebe V, Kitano T, et al. Molecular evolution of FOXP2, a gene involved in speech and language. Nature. 2002;418(6900):869–72. pmid:12192408
  35. 35. Li S, Mo Z, Yang X, Price SM, Shen MM, Xiang M. Foxn4 controls the genesis of amacrine and horizontal cells by retinal progenitors. Neuron. 2004;43(6):795–807. pmid:15363391
  36. 36. Triboulet R, Pirouz M, Gregory RI. A Single Let-7 MicroRNA Bypasses LIN28-Mediated Repression. Cell Rep. 2015;13(2):260–6. pmid:26440890
  37. 37. Kallen AN, Zhou XB, Xu J, Qiao C, Ma J, Yan L, et al. The imprinted H19 lncRNA antagonizes let-7 microRNAs. Mol Cell. 2013;52(1):101–12. Epub 2013/09/24. pmid:24055342
  38. 38. Nadiminty N, Tummala R, Lou W, Zhu Y, Zhang J, Chen X, et al. MicroRNA let-7c suppresses androgen receptor expression and activity via regulation of Myc expression in prostate cancer cells. J Biol Chem. 2012;287(2):1527–37. Epub 2011/12/01. pmid:22128178
  39. 39. Helland A, Anglesio MS, George J, Cowin PA, Johnstone CN, House CM, et al. Deregulation of MYCN, LIN28B and LET7 in a molecular subtype of aggressive high-grade serous ovarian cancers. PLoS ONE. 2011;6(4):e18064. Epub 2011/05/03. pmid:21533284
  40. 40. Zisoulis DG, Kai ZS, Chang RK, Pasquinelli AE. Autoregulation of microRNA biogenesis by let-7 and Argonaute. Nature. 2012;486(7404):541–4. pmid:22722835
  41. 41. Hammell CM. The microRNA-argonaute complex: a platform for mRNA modulation. RNA biology. 2008;5(3):123–7. Epub 2008/08/14. pmid:18698153