| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Center for Anatomy, Institute of Cell Biology and Neurobiology, Charité-Universitätsmedizin Berlin, Berlin, Germany;
National Center for Documentation and Evaluation of Alternative Methods to Animal Experiments (ZEBET), Federal Institute for Risk Assessment (BfR), Berlin, Germany; and
Max-Delbrück-Centrum für Molekulare Medizin, Berlin-Buch, Berlin, Germany
2Correspondence: Center for Anatomy, Institute of Cell Biology and Neurobiology, Charité-Universitätsmedizin Berlin, Schumannstrasse 20–21, 10098 Berlin, Germany. E-mail: gregory.wulczyn{at}charite.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: embryonic stem cells • neurogenesis • fragile X mental retardation protein • miRISC • P/GW bodies
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
22 nt) RNA molecules that act as antisense regulators of mRNA utilization. Various mouse and human tissues and cell lines have been shown to express distinct and characteristic repertoires of miRNAs, and it is apparent that miRNAs significantly influence protein expression patterns during mammalian development. The salient features of miRNA genetics, biogenesis, and function have been presented in several comprehensive reviews (1
2
3)
. The mechanisms responsible for tissue and developmental specificity of miRNA expression have yet to be defined. In the first step of miRNA biogenesis, primary nuclear transcripts (termed pri-miRNA) are synthesized by RNA-polymerase II. The pri-miRNA are then processed by a nuclear protein complex containing the Drosha RNase (reviewed in ref. 4
). Nuclear cleavage releases an 70 nt hairpin precursor (pre-miRNA) that is exported to the cytoplasm (reviewed in ref. 4
). Understanding of cytoplasmic events is rapidly evolving. The pre-miRNA is subject to a second round of processing by the RNase Dicer, acting in concert with a dsRNA binding protein (TRBP in mammals) and an Argonaute protein (5
6
7)
. The mature miRNA is incorporated into an effector complex alternately referred to as the miRNP, RISC, or miRISC, reflecting the considerable functional and biochemical overlap between the miRNA and RNAi pathways (miRNA biogenesis is reviewed in refs. 1
, 3
, 8
). Recently, trafficking of miRNP components to processing bodies (also called GW bodies or P/GW bodies) has been shown to be an essential feature of the pathway (9
10
11
12
13
14
15)
. P/GW bodies are cytoplasmic subcompartments involved in mRNA metabolism, degradation, and translational control.
The original C. elegans miRNAs, lin-4 and let-7, control the developmental timing of progenitor cell maturation (reviewed in refs. 2
, 16
), a finding that has been extended to additional let-7 family members (17)
. Johnson et al. have demonstrated widespread and evolutionarily conserved regulation of the RAS family of growth control proteins by let-7 (18)
. In addition, let-7 and lin-4 cooperate in the post-transcriptional regulation of the C. elegans hunchback homologue hbl-1 (17
, 19
, 20)
, a key gene in the specification of neural progenitor cell fate. Premature let-7 expression in zebrafish and Xenopus led to a generalized developmental arrest and prominent failure of eye development (21)
.
In humans and mice, the let-7 family consists of 12 precursor genes that encode nine distinct, mature 22-nucleotide sequences. The mature forms (designated let-7a through let-7i together with the highly related mir-98) differ at one to four positions from the canonical let-7a sequence. Lagos-Quintana et al. were the first to demonstrate high representation of let-7 family members in miRNA populations from mouse brain (22)
, findings that were subsequently extended to primates (23)
. Expression of let-7 in the developing nervous system of zebrafish has recently been characterized by ISH (24)
, and the technique has been extended to mouse embryogenesis (25)
. However, the function of let-7 during brain development has not been specifically addressed. Given increasing evidence for the importance of miRNA in pathways controlling cell specification and patterning in the nervous system (reviewed in refs. 26
27
28
), we have investigated the control of let-7 expression in the embryonic mouse brain and differentiating ES cells. We present evidence for an important post-transcriptional component in the regulation of let-7 during neural differentiation and lineage specification.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 57)
; details are provided in the Supplemental material.
Cell culture
Protocols for maintenance and differentiation of EC and ES cells, as well as primary cell culture, have been recently described (29)
; a summary is provided in Supplemental material.
Transfections and immunofluorescence microscopy
Details of sensor plasmid construction and use in stable and transient transfection asays have been described (29)
and are provided in the Supplemental material. For immunohistochemistry, EC and HEK293 cell were transfected with Fugene6 reagent (Roche, Nutley, NJ, USA) using 250 ng of FLAG-tagged Ago1, Ago2, or FMRP expression plasmids. MOV10 and TNRC6B constructs, described in (12)
, were Myc-tagged and visualized with rabbit anti-Myc antibody (Sigma, St. Louis, MO, USA). Twenty-four to 48 h after transfection, cells were fixed and stained with the following primary antibodies: mouse anti-Flag (M2, Sigma), mouse anti-GW182 (4B6, Abcam, Cambridge, MA, USA), rabbit anti-Ago1 (Upstate, Charlottesville, VA, USA), rabbit anti-Ago2 (Upstate), or mouse anti-FMRP (1C3, Euromedex, Strasbourg, France). Secondary antibodies were conjugated to Alexa-488 and –568 goat (Molecular Probes, Carlsbad, CA, USA). Images were obtained by confocal microscopy (Leica TCS SL) and analyzed with Leica confocal software, Volocity, and ImageJ software packages.
Northern blot analysis
Oligonucleotide probes for Northern blots correspond to miRNA sequences at the miRNA registry (http://www.sanger.ac.uk/Software/Rfam/mirna/index.shtml). Preparation of the mouse brain Northern blot has been described (29)
; see also Supplemental material. The miRNA Northern blot procedure, including hybridization conditions and end-labeling with T4 polynucleotide kinase, followed published protocols (29
, 58)
. The same procedure was followed for LNA probes, hybridization temperature was increased to 55°C and washes were at 65°C.
In vitro miRNA processing assay
The following primer pairs were used to generate templates for in vitro transcription: 5'TAATACGACTCACTATAGGAGGTAGTAGGTTGTATAG3' and 5'GGAAAGACAGTAGATTGTA3' (let-7a); 5'TAATACGACTCACTATAGGAGGTAGGAGGTTGTATAG3' and 5'GGAAAGCTAGGAGGCCGTA3' (let-7e); 5'CGTAATACGACTCACTATAGGGGGGCCGATGCACTGTA3' and 5'GAAAGAGACCGGTTCACTG3' (mir-128). The polymerase chain reaction (PCR) products were purified and used in in vitro transcription reactions with T7 RNA polymerase in the presence of 32P 
-uridine triphosphate. Cytoplasmic extracts for the in vitro reaction were prepared in lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 2 mM MgCl2, 10 mM NaF, 1 mM DTT, 20% glycerol, and protease inhibitors). For processing reactions, 10 µg protein was incubated for 90 min with 20,000 cpm RNA in 75mM NaCl, 20 mM Tris-HCl, pH 7.5, 3.0 mM MgCl2, and 0.1 U/µl RNAsin. Digestion with recombinant Dicer (0.1 U/reaction) was performed according to the manufacturer’s recommendations (Ambion, Austin, TX, USA). Reaction products were resolved by denaturing electrophoresis; gels were stained with ethidium bromide to visualize size standards, dried, and visualized by autoradiography.
EMSA
Cell extracts (2 µg protein) were incubated with 20,000 cpm 32P-labeled precursor RNAs for 30 min at 37° in 20 mM Tris-HCl, pH 8.0, 100 mM KCl, 1.5 mM MgCl2, and 1 mM DTT. Ficoll was added to 2%, followed by electrophoresis on a 5% polyacrylamide gel. Anti-FMRP monoclonal antibody (mAb) 7G1–1 and anti-Rip1 were obtained from the Developmental Studies Hybridoma Bank as concentrated hybridoma supernatant.
Chromatin immunoprecipitation assay (ChIP)
RNA-pol-ChIP experiments were performed essentially as described in ref. 32
using the ChIP assay kit (Upstate) according to the manufacturer’s recommendations. PCR primers are described in Supplemental methods.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. To extend these studies to let-7, we employed recently developed protocols for ISH of miRNAs (24)
on early mouse embryos. The LNA-modified probes used have been shown to be sensitive to single-nucleotide exchanges (25)
, minimizing potential cross-hybridization. For three let-7 isoforms tested (let-7a, let-7c, and let-7e), similar expression patterns were obtained in whole mounts of E9.5 embryos. Staining was observed primarily in the neuroepithelium of the brain and spinal cord, the first and second branchial arches, and the forelimb primordia (Fig. 1
A). In the case of let-7a but not let-7c or let-7e, most embryos demonstrated weaker signal in the developing heart and vasculature. The staining pattern obtained was strikingly similar to that observed with the brain-specific miRNAs mir-124 and mir-128 (Fig. 1A
). To demonstrate specificity of the hybridization conditions, mir-1 strongly and specifically stained the heart and major vessels but not the neuroepithelium; no signal was obtained for mir-140, a cartilage-specific miRNA (Fig. 1A
). These results are consistent with an extensive ISH study in zebrafish (24)
and with many expression studies in the mouse using Northern blot or microarray technologies.
|
Tissue specificity was examined in sections of a representative E9.5 embryo stained for let-7 (let-7e). The section in Fig. 1B
provides an overview of the heart, branchial arches, and cranium. The neuroepithelium of the midbrain and forebrain are uniformly stained, as are the mandibular and maxillary parts of the first branchial arch. let-7e expression was not detected in heart muscle or vasculature. Figure 1C
provides a higher magnification of brain and branchial arch staining. For each let-7 probe tested, staining was evenly distributed throughout the hindbrain, midbrain, and forebrain, as well as the otic and optic vesicles (data not shown). Staining of the branchial arches is consistent with the initiation of expression in cells of the migrating neural crest. At the level of the forelimb bud (Fig. 1D
), let-7 was expressed in the limb bud mesoderm as well as in the ventral and dorsal neural tube. Within the somites, staining was most intense in the dorsal and lateral cell layers of the dermamyotome. Prominent staining in the developing nervous system encouraged us to conduct additional experiments at later time points.
To follow expression after E9.5, we employed ISH with 35S-labeled LNA probes. Staining of a sagittal section of an E13.5 embryo for let-7a revealed widespread expression, most prominently in the neuroepithelium of the brain and spinal cord, and in the cranial and dorsal root ganglia (Fig. 1E
). Expression in the limb buds was also conspicuous, particularly in the cartilage. Organs of the gut stained less intensely, and the heart and lungs showed the lowest levels of expression.
In general, in sections of the developing brain we observed high levels of expression in the embryonic hippocampal formation and cortical layers; expression in the granular layer of the cerebellum increased postnatally (A. Rybak and F. G. Wulczyn, unpublished results). In the adult brain, let-7 expression was reduced in comparison to the neonate, with high levels of expression in the neuron-rich layers of the hippocampus and cerebellum. Representative results for let-7a, let-7c, and the mir-1 control are shown in Fig. 1F
. These results are consistent with a Northern blot analysis of let-7 expression during brain development (see Supplemental Fig. 1A), in which 22 nt forms of let-7 increase in parallel with the major periods of neurogenesis in the cortex between E13 and E18 and in the postnatal cerebellum.
Comparison of primary, precursor, and mature let-7 expression in embryonic stem cells
In principle, signals obtained in ISH could represent primary transcripts, hairpin precursor forms, or the mature 22 nt let-7 miRNA. At the earliest stages of development, Northern blots reveal that levels of the 22 nt mature forms are low and the ratio of precursor to mature RNA is high (Supplemental Fig. 1A). We therefore turned to two well-established model systems: neural differentiation of embryonic stem (ES) cells and the P19 embryocarcinoma (EC) cell line to study early regulation of the primary transcript, precursor and mature forms during neurogenesis. We began by inducing ES cell differentiation either in the absence of exogenous inducer (cardiomyocyte differentiation) or in the presence of retinoic acid (RA) as an inducer of neural cell phenotypes. In agreement with published reports (30
, 31)
, mature 22 nt forms of let-7 (let-7a,c,e) were strongly induced during either cardiomyocyte or neural differentiation protocols (Fig. 2
A). No expression of mature forms was detected in either undifferentiated cells or at the embryoid body stage. As a control for RNA integrity, the blot was also probed for mir-294, a miRNA specifically expressed in undifferentiated cells and embryoid bodies (30)
.
|
In contrast to the mature miRNAs, hybridization signals representing 70 nt cytoplasmic precursor forms were clearly detected in undifferentiated as well as differentiated cell samples for let-7a, let-7c, and let-7e (Fig. 2A
). To verify probe specificity, all results obtained with DNA probes were confirmed using LNA probes (see Materials and Methods). In multiple experiments, precursor bands detected with the individual probes could be distinguished on the basis of mobility. This is most apparent in a similar experiment performed with EC cells (Fig. 2B
). In EC cells, precursor molecules of distinct mobility were also readily detected in undifferentiated as well as differentiated EC cell samples, and accumulation of mature let-7 forms was dependent on cell differentiation (Fig. 2B
).
We next assayed for primary transcripts, again comparing undifferentiated and differentiated EC cells. We performed semiquantitative RT-polymerase chain reaction (RT-PCR) using primer pairs specific for the let-7a-1 and let-7f-1 pair from chromosome 13 (pri-let-7a) and a three miRNA cluster from chromosome 17 containing mir-99b, let-7e, and mir-125a (pri-let-7e). Control amplifications were performed using mock cDNAs prepared in parallel from each RNA sample. Amplification of the 6-integrin mRNA served as a control for differentiation and cDNA integrity (Fig. 2C
). A representative amplification from three independent experiments is shown in Fig. 2C
.
Both the pri-let-7a and the pri-let-7e transcripts were detected in amplifications of cDNA from undifferentiated cells, consistent with constitutive expression of the hairpin precursor observed in Northern blots. Levels of primary transcripts in cDNA from differentiated cells were increased. Similar results were obtained for ES cell differentiation (data not shown). The degree of transcriptional activation was quantified by real-time PCR using the GAPDH mRNA as standard. Levels of pri-let-7e were 4- and 7-fold higher and levels of pri-let-7a 6- and 10-fold higher after differentiation in ES and EC cells, respectively. This relatively modest degree of transcriptional activation is unlikely to fully account for the strong induction of let-7 family members during differentiation.
To independently verify transcriptional activity in undifferentiated cells, we next assayed for RNA-polymerase II association with let-7 genes using the RNA-pol-ChIP technique (32)
. In this assay, chromatin was cross-linked to associated proteins and then fractionated. The resulting complexes were immunoprecipitated with an RNA-polymerase II-specific antibody; after elution coprecipitated DNA was amplified with primers corresponding to the pri-let-7a and pri-let-7e transcripts. As shown in Fig. 2D
, the efficiency of product amplification was similar in immunoprecipitates from undifferentiated and differentiated cells, confirming RNA-polymerase II association with the two let-7 transcription units in undifferentiated cells. As a control, transcriptional down-regulation of the stem cell marker Oct4 after differentiation was verified in samples processed in parallel.
To test the relevance of these results for brain development, we determined primary transcript levels in RNA samples from embryonic brain. Primary transcripts were detected in cDNA from each of five time points sampled, but not in mock cDNA (Supplemental Fig. 1C). Both primary transcripts tested were modestly increased at E16 and P0 compared with E13, and levels declined somewhat perinatally. Like ES and EC cells, these results indicate that both let-7 loci tested are transcriptionally activated during the major phase of embryonic neurogenesis; however, the relative magnitude of miRNA accumulation compared with transcriptional up-regulation suggests a significant contribution from post-transcriptional mechanisms.
Differential regulation of let-7 expression in neural lineages
We previously described distinct profiles of miRNA expression in embryonic neurons and astrocytes. Comparing let-7 expression between the two lineages, mature forms of let-7isforms accumulated to markedly higher levels in neurons (Fig. 2E
). There was a clear disparity between accumulation of the precursor and mature forms, as precursor forms were detected in both cell types at similar levels (Fig. 2E
, data shown for let-7c and let-7e). To address the question of regulation, we also compared levels of the primary transcripts pri-let-7a and pri-let-7e. Despite the differential expression of the mature miRNAs, primary transcript levels were similar (Fig. 2F
). This result was confirmed by quantitative real-time PCR (data not shown). Although this experiment provides only a snapshot of let-7 expression in the two lineages, which may change during development, it is consistent with the ISH results presented in Fig. 1E
.
We used a functional assay to confirm the Northern blot analysis of let-7 expression in differentiating EC cells and in primary cultures. We assayed let-7 activity using a sensor construct containing let-7 binding sites derived from the C. elegans lin-41 mRNA linked to eGFP. As expected, expression of the sensor construct was powerfully suppressed in EC-derived neurons but not in undifferentiated EC cells (Supplemental Fig. 2A, B). Differentiation had no effect on antisense or point mutant control constructs. Similarly, let-7-mediated suppression was substantially greater after transfection of the sensor into primary neurons than in astrocytes, reflecting differential let-7 expression in the two lineages (Supplemental Fig. 2C). Representative flow cytometry plots for these experiments are provided in Supplemental Fig. 3.
Developmental regulation of miRNA processing activity
The observation that precursor forms are present in undifferentiated ES and EC cells in the absence of mature let-7 prompted us to examine the efficiency of precursor processing as a function of cellular differentiation. To determine whether let-7 processing is developmentally regulated, we examined processing activity in vitro (Fig. 3
). Incubation of an internally labeled pre-let-7a RNA designed to contain a 2 nt overhang with cytoplasmic extracts from differentiated EC cells led to the release of a 22 nt RNA that comigrated with the digestion product produced by recombinant Dicer (Fig. 3A
), and with a synthetic 22 ntRNA oligoribonucleotide run in parallel (not shown). let-7 processing activity was very low in extracts from undifferentiated EC cells (lane 2) and was strongly increased 12 days after induction of neural differentiation by RA treatment (lane 4).
|
To determine whether the increase in processing activity was specific for pre-let-7a, we repeated the experiment with the neuron-specific mir-128 precursor as substrate (Fig. 3B
). mir-128 processing activity also showed a clear correlation with the differentiation state of the cells. Incubation of pre-mir-128 with extracts from day 8 or 12 cells (lanes 3 and 4), but not undifferentiated cell extracts (lane 2), led to release of a 22 nt RNA that comigrated with the digestion product produced by recombinant Dicer (lane 1).
Analysis of pre-miRNA binding complexes
We reasoned that differential processing activity might be reflected in differences in pre-let-7 binding activity, and next examined precursor binding activity in an EMSA. We first tested differentiated and undifferentiated EC cell extracts using pre-let-7a as substrate (Fig. 4
A). We observed one binding activity common to all extracts (designated complex A). Apart from complex A, neural differentiation after stimulation with RA was accompanied by a shift in the pattern of RNA binding. Undifferentiated extracts contained high levels of a high mobility complex designated complex C. Formation of complex C decreased during the course of neural differentiation and was supplanted by an additional complex with slower mobility, designated B (Fig. 4A
, lanes 1–4). For comparison, extracts prepared from cells induced under conditions favoring differentiation to cardiomyocyte-like cells were also tested. In contrast to neural cells, these cells displayed a more modest shift from complex C to B over the course of the experiment (Fig. 4A
, lanes 5, 6).
|
We next examined binding to the related pre-let-7e precursor and observed a similar pattern of binding to pre-let-7a (Fig. 4B
, lanes 1–4). Complex C was the predominant activity in undifferentiated cells. Differentiation was accompanied by a strong reduction in complex C and a parallel increase in complex B. In competition experiments, all three complexes were inhibited by addition of a 10-fold excess of unlabeled pre-let-7e RNA, but not by an unrelated RNA fragment transcribed from the eGFP gene (data not shown). To further characterize the binding activities and their specificity, we determined the sensitivity of binding to sense and antisense 2'O-methyl-modified oligoribonucleotides (2'OM-ORN). We first tested the interactions of each competitor with the let-7 precursor in the absence of protein extract (Fig. 4C
, lanes 1–3). Addition of the let-7a 2'OM-ORN had no effect on the migration of the labeled precursor (lane 2). In contrast, addition of the anti-let-7a 2'OM-ORN led to formation of a band with reduced mobility, indicating that the antisense molecule could invade the stem and compete with the let-7a* sequence for let-7a (lane 3).
The sense and antisense 2'OM-ORN had differential effects on the RNA binding complexes B and C. Addition of the let-7 surrogate eliminated complex C formation without reducing complex B (lanes 5 and 10). The interference was sequence specific, as anti-let7 did not affect complex C and a control 2'OM-ORN complementary to mir-128 displayed only partial inhibition (lanes 6 and 7). This result suggests that complex C makes specific contacts to the let-7 sequence in the context of the precursor RNA.
The pattern of inhibition observed for complex B was the inverse of complex C. Neither the let-7 surrogate (lanes 5 and 10) nor anti-mir-128 (lanes 7 and 11) significantly interfered with complex B formation. However, addition of anti-let-7 strongly reduced complex B formation (Fig. 4C
, lane 12). This suggests that complex B requires an intact hairpin, since the anti-let-7 2'OM-ORN disrupts the hairpin structure (Fig. 4C
, lane 3). The results are also consistent with a functional role for complex B in processing, as antisense but not sense 2'OM-ORN strongly inhibit miRNA function in vivo (33
, 34)
.
To directly link the precursor RNA binding activity to the miRNA biogenesis pathway, we challenged neural complexes with antibodies to several known miRNP components (Fig. 4D
). Addition of protein A purified anti-Argonaute1 antibody, a core miRNA protein (Ago1, lane 3), reproducibly reduced formation of complex B. As a control, binding was unaffected by addition of either an unrelated antibody preparation (protein A purified anti-H3) or an antibody specific for the RNA helicase Gemin3 (Fig. 4D
, lanes 4, 5). Furthermore, a mAb specific for the miRNA-associated fragile X mental retardation protein (FMRP, lane 8) reproducibly reduced formation of complex B. Results are shown for day 8 and day 12 EC extracts (Fig. 4D
, lanes 8, 12). Control reactions contained isotype-matched antisera that did not affect binding (lanes 9 and 13).
We next compared pre-let-7 binding activity in EC-derived neurons to primary cortical neurons (Fig. 4E
). In primary neurons we observed two RNA binding activities with mobility indistinguishable from complexes A and B. As expected, the embryonic complex C was not present. Preincubation with anti-FMRP antibody strongly and specifically inhibited formation of the neuronal complex B compared with control (Fig. 4E
, lanes 5, 6), consistent with the presence of FMRP in a neuron-enriched pre-let-7 binding complex.
Localization of miRNA processing proteins in stem cells and neurons
With the exception of germ cells, which possess a distinct complement of miRNP proteins, the miRNA pathway is generally thought to be ubiquitous. However, the possibility of developmental regulation has not been directly addressed. In Supplemental Fig. 4A, we confirmed that mRNA levels for most of the known components of the miRNP (Dicer, Ago1, Gemin3, Gemin4, Mov10, TNRC6B, and TRBP2) were relatively constant in the cell models used in our study. An interesting exception was FMRP, which is known to be preferentially (though not exclusively) expressed in neurons. We found that the FMRP mRNA was up-regulated during neural differentiation of both ES and EC cells (Supplemental Fig. 4B). In differentiated EC cells, FMRP immunoreactivity was elevated in cells expressing neuronal markers and protein levels were clearly elevated in primary neurons compared with astrocytes (Supplemental Fig. 4C, D). Finally, higher levels of the FMRP-associated complex B were observed by EMSA in extracts from embryonic neurons compared with astrocytes (Supplemental Fig. 4E).
We next examined localization of FMRP in comparison to the miRNA components Ago1, Ago2, and the P/GW body marker GW182. In agreement with several reports (10
11
12
, 14)
, we observed colocalization of Ago1 and Ago2 with GW182 to discrete, brightly staining cytoplasmic foci in HEK293 cells. The results for transfected Ago1 are presented in Fig. 5
A. Similar results were obtained after staining for endogenous Ago1 (Supplemental Fig. 5). Given the known association of FMRP with the miRNP, we performed cotransfections with either Ago1 or Ago2. Both Argonautes displayed partial overlap with FMRP in distribution and colocalization to cytoplasmic foci; results are shown for Ago1 (Fig. 5B
).
|
We subsequently investigated subcellular localization of Ago1, Ago2, and FMRP in ES and EC cells. In contrast to HEK293 cells, we observed diffuse cytoplasmic staining for each of the three proteins without discernible concentration at defined subcellular structures. Results were confirmed using polyclonal sera specific for Ago1 or Ago2, the FMRP-specific mAb 1C3, as well as epitope tag-specific anti-Flag (Ago1, Ago2, FMRP) and anti-hemagglutinin (Ago1, Ago2) antibodies (data not shown). Representative plates for Ago1 in ES and EC cells, and FMRP in EC cells, using the anti-Flag antibody, are presented in Fig. 5C
. To determine whether ES and EC cells possess P/GW bodies, we tested ectopically expressed MOV10 and TNRC6B, two P/GW body proteins that physically associate with Argonautes (12)
. Both localized to cytoplasmic foci in EC and HEK293 cells (data not shown). In cotransfection experiments, MOV10 and TNRC6B were able to relocalize either Ago1 or Ago2 to P/GW bodies in ES and EC cells. Representative results for Ago1 are shown in Fig. 5D, E
and suggest that the signals governing localization of Argonaute in stem cells may be overridden or saturated by ectopic expression of MOV10 or TNRC6B. In the same assay, the integral P/GW body protein TNRC6B was able to attract FMRP to cytoplasmic foci (Fig. 5F
). The effect was specific, as MOV10 had no obvious effect on FMRP distribution in either stem cell line (data not shown).
These results suggest that regulatory interactions may control miRNP organization during differentiation. Given the evidence implicating FMRP and MOV10 in miRNA-mediated neuronal translation (35
, 36)
, we repeated the transfection experiments in cultured hippocampal neurons. Staining was observed either in bright foci in the neuronal cell bodies (Ago1, Ago2, TNRC6B) or in a punctate distribution in the dendritic arbors (Ago1, Ago2, FMRP, MOV10). Colocalization of Ago1 and MOV10 and of FMRP and MOV10 to dendritic foci is shown in Supplemental Fig. 6A, B. To better visualize the MOV10 localization in relation to dendritic structures, such as branch points and spines, MOV10 and eGFP were cotransfected (Supplemental Fig. 6C). The two proteins did not colocalize, and MOV10 staining can be seen to frequently underlie dendritic spines. A different pattern was obtained with TNRC6B, which was found primarily in bright foci in the soma and proximal dendrites (Supplemental Fig. 6D). It will be interesting to functionally characterize the discreet compartments occupied by miRNA pathway components in neurons.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In zebrafish, the let-7 family is characterized by prominent expression in nervous tissue (24)
. Mansfield et al. used a sensor strategy to describe let-7 expression in mouse development, but did not focus on the CNS (37)
. Using ISH, we found that the distribution of three let-7 family members at E9.5 closely resembled that of the paradigm brain-specific miRNA mir-124 and brain-enriched mir-125 and mir-128. It is interesting to note the lack of either dorsal/ventral or anterior/posterior polarity in the CNS expression pattern at E9.5 or E13.5. This is consistent with genetic studies of let-7 function that point to a role in the temporal as opposed to spatial coordination of gene expression (17
, 38)
. The timing of let-7 expression triggers terminal cell differentiation in the C. elegans heterochronic gene networks (39)
, and it is thought that miRNA may be involved in regulating either the timing, outcome, or maintenance of cell fate decisions (reviewed in refs. 2
, 16
). It will be interesting to learn whether let-7 plays a similar role in the timing of cell fate decisions in the mammalian CNS.
In ES cells and embryonic neural progenitors, the transition from the pluripotent, self-renewing state to cell-fate restriction is accompanied by a profound shift in the miRNA populations present in the cell. Little is known about the regulatory signals directing this shift. Several studies have begun to address transcriptional control mechanisms underlying specificity in miRNA expression (40
41
42
43)
. We present evidence for both transcriptional and post-transcriptional control mechanisms in the induction of let-7 family members during neural differentiation.
We were particularly intrigued by the temporal disparity in the appearance of primary transcripts and cytoplasmic precursors compared with the mature 22 nt form during stem cell differentiation. To account for the lack of mature let-7 in undifferentiated cells, we assayed precursor processing activity and found that it increased in parallel with neural differentiation. The deficiency in let-7 processing in undifferentiated cells might reflect either a lack of an essential component of the processing machinery or suppression of the pathway early in differentiation. However, both ES and EC cells possess an intact RNAi pathway, and accumulate a unique class of stem cell-specific miRNA (30)
. Consistent with this, we detected mRNA for all known components of the precursor processing pathway using RT-PCR in undifferentiated stem cells. We therefore favor a model in which let-7 precursor RNAs fail to engage the processing machinery. This model is consistent with the distinct precursor binding complexes present in undifferentiated and differentiated cells and with the evidence for differing subcellular organization of pathway components in self-renewing pluripotent cells.
In the precursor RNA binding assay, a prominent complex (complex B) was shared by stem cell-derived and embryonic neurons. Antibodies against the core processing complex protein Ago1, as well as the miRNP-associated factor FMRP, specifically interfered with complex formation. Besides their essential function in the miRNP effector complex, Argonaute proteins participate in a core ternary complex with Dicer and TRBP in precursor processing (5
6
7
, 44)
. Furthermore, immobilized anti-let-7 2'OM-ORN has been used to precipitate Ago1 and Ago2 from let-7-primed RISC complexes (33)
. Unlike Argonautes, FMRP is not required for precursor processing in vitro (5
, 45
46
47)
, but it may help to link the miRNP and translational control machinery. Several groups have shown that Drosophila and mouse FMRP proteins associate with miRNAs, polyribosomes, and RISC activity (reviewed in refs. 3
, 35
). The neuron-enriched complex we identified in the RNA binding assay was present at reduced levels in undifferentiated stem cells, which contained instead an alternative complex that was specifically inhibited by a let-7 analog. Further characterization of these binding activities should afford new insights into the regulation of let-7 during stem cell differentiation.
Current models for miRNA processing involve initial precursor binding by a miRNA loading complex and subsequent transfer or rearrangement to an activated effector complex (5
, 6)
(reviewed in refs. 48
, 49
). Localization of the effector complex to cytoplasmic P/GW bodies is thought to be essential for mRNA silencing (9
, 11
, 13)
, but it has not yet been established that P/GW bodies are the sites of precursor processing. In undifferentiated ES or EC cells, neither endogenous nor ectopically expressed Argonaute or FMRP were organized in definable cytoplasmic foci. We believe this observation is related to the inefficient processing of let-7 in these cells. It is possible that additional factors, perhaps related to the stem cell-specific pre-let-7 binding complex we have described, act as escorts regulating compartmentalization of the processing machinery. Ectopic MOV10 or TNRC6B presumably override these signals and tether Argonautes and FMRP to P/GW bodies by direct protein-protein interactions.
Our results also strengthen the evidence linking FMRP and the miRNA pathway. We identified FMRP as part of a neuronal pre-let-7 binding complex and found that a portion of ectopically expressed FMRP colocalizes with argonaute proteins to cytoplasmic foci in the HEK293 model. We also found extensive colocalization of Ago1 and Ago2, MOV10, and FMRP in cultured neurons (see Supplemental Fig. 6). These results are consistent with the well-established role for FMRP in the regulation of synaptic translation (reviewed in refs. 35
, 50
), recruitment of FMRP to stress granules (51)
, and identification of FMRP as a miRNP/RISC-associated protein (reviewed in ref. 35
).
P/GW bodies, as sites of mRNA metabolism, are dynamic structures that are modified in the cellular response to stress and undergo remodelling during the cell cycle (52)
. Alteration of cell cycle control is also a hallmark of stem cell differentiation (reviewed in ref. 53
). Since many miRNAs, including let-7, are thought to target regulators of the cell cycle and proliferation and to promote terminal differentiation, a mechanism to delay miRNA accumulation early in development may be necessary. This scenario is also consistent with the phenotype of mutations in the miRNA biogenesis pathway, which tend to affect development after completion of axis and pattern formation (54
55
56)
. Our results reveal novel aspects of early developmental regulation of the miRNA pathway during embryonic stem cell differentiation and neurogenesis.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication March 22, 2006. Accepted for publication September 22, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
