| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Molecular & Cellular Proteomics 6:798-811, 2007.
© 2007 by The American Society for Biochemistry and Molecular Biology, Inc.
,**,
From the Department of Clinical Biochemistry, Rigshospitalet, || Institute of Molecular Biology, and Bioinformatics Centre, University of Copenhagen, DK-2100 Copenhagen, Denmark and ¶ Danish Fundamental Metrology, Danish Technical University, DK-2800 Lyngby, Denmark
 |
ABSTRACT |
|---|
 TOP ABSTRACT EXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Cytoplasmic mRNPs may become destined for local translation. In support of this possibility, RNAs have been found in large mRNP granules, which are transported along cytoskeletal structures and anchored at their final destination. Messenger RNA localization has mainly been examined in polarized oocytes and neurons, and it has been proposed that local postsynaptic protein synthesis is required for synaptic plasticity (4). Previous studies have identified neuronal Staufen (5) and FMRP granules (6, 7), containing mRNAs, small and large ribosomal subunits, translation initiation factors including eIF4E and eIF2
, and RNA-binding proteins (Refs. 8–11; for a review, see Ref. 12). The protein composition of neuronal mRNP granules is to some degree overlapping with stress granules and processing bodies (P-bodies). The hallmark of stress granules is the presence of stalled 48 S initiation complexes and stress-dependent RNA-binding factors such as G3BP (13, 14), whereas P-bodies contain components of the 5'–3' mRNA decay machinery and factors involved in nonsense-mediated decay (15).
The zip code-binding proteins IMP1, -2, and -3 (human), ZBP1 (chicken), Vg1-RBP/Vera (Xenopus), and coding region determinant-binding protein (mouse) are members of the same vertebrate RNA-binding protein family, consisting of two RNA recognition motifs and four K-homology domains (16–21). Various post-transcriptional functions of IMPs and their orthologues have been reported. The Xenopus IMP3 orthologue Vg1-RBP/Vera has been implicated in localization of Vg1 mRNA to the vegetal pole of the Xenopus oocyte (16, 18), and mouse coding region determinant-binding protein stabilizes the c-myc transcript (17). Chicken ZBP1 regulates localization of ß-actin mRNA to the leading edge of fibroblasts (22), and finally IMPs play a role in H19 and tau mRNA transport (23, 24) and IGF2 mRNA translation (19, 25). IMPs are expressed during early embryogenesis and at midgestation in the mouse, and disruption of IMP1 leads to dwarfism and impaired gut development (26). In the same vein, the IMP3 orthologue Vg1-RBP/Vera is required for normal pancreatic development (27) and promotes migration of neural crest cells (28). In the cytoplasm, IMPs are distributed in large 200–700-nm (optical diameter) RNP granules, and in motile cells granules are found around the nucleus and transported toward the leading edge where they dock at the cortical region of the lamellipodia. Granules travel at a speed of 0.2 µm/s, and cells are able to switch from a delocalized to a localized pattern within 15–20 min (29). In neuronal cells, ZBP1 granules localize to dendrites, and reduced levels of ZBP1 lead to impaired growth of dendritic filopodia (30).
The function of IMP1 and other RNP granules is incompletely understood. The number of different types of granules is unknown, and it is uncertain whether the particles are heterogeneous and how they are assembled and dismantled, although recent data have indicated that local signaling events may play a role in mRNA release (31). It is anticipated that associated mRNAs are translationally quiescent during transport, but there are limited data to support this assumption. With the exception of FMRP granules, which have been demonstrated to contain about 2% of the transcriptome (32), no global picture of human mRNA cargo is available. An intriguing proposal is that RNP granules, or more generally mRNP particles, represent post-transcriptional operons (33, 34), and the finding that RNA-binding Puf proteins in yeast associate with mRNAs encoding proteins with common functions and subcellular localization lends strong support for this proposal (35).
To deepen our understanding of human RNP granules or "locasomes" (36), we determined the molecular composition of IMP1-containing RNP granules. The results demonstrate that IMP1 granules represent a unique RNP entity containing untranslated mRNAs, corresponding to 
3% of the transcriptome. Messenger RNAs encoding proteins of the secretory pathway and ER-associated quality control, as well as ubiquitin-dependent metabolism, were enriched reinforcing the concept of post-transcriptional operons.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
HEK293 cells were transiently transfected with FuGENE 6 transfection reagent (Roche Diagnostics) according to the manufacturer's instructions. Briefly 20,000 cells/cm2 were seeded on fibronectin-coated glass plates or 10-cm plastic dishes 24 h prior to transfection. Cells were transfected with a 2 µg/ml concentration of the indicated plasmid and left for 48 h before they were examined. The distribution of CFP-IMP1, YFP-hStaufen, YFP-FMRP, YFP-IMP3, YFP-ELAV, and YFP in living and fixed cells (see below) was examined with a Zeiss LSM 510 confocal laser scanning microscope. The Drosophila S2 cells were maintained in Drosophila Schneider medium (Invitrogen) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin at 25 °C.
Immunocytochemistry—
HEK293 cells were seeded on fibronectin-coated (25 µg/cm2) glass bottom dishes. Cells were fixed with 4% formaldehyde for 15 min at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 3 min. After blocking with 1% BSA in PBS for 1 h at room temperature, the cells were incubated with primary antibody overnight at 4 °C, washed 3 x 5 min in PBS, and visualized with Alexa Fluor 488- or Alexa Fluor 555-conjugated anti-mouse or anti-rabbit antibodies.
Western Blot Analysis—
Protein extracts from HEK293 cells were separated in SDS-polyacrylamide gels and transferred to Hybond-P membranes (Amersham Biosciences). After blocking, membranes were incubated overnight with primary antibody in blocking solution at 4 °C before they were washed and incubated with horseradish peroxidase-conjugated anti-rabbit, anti-mouse, or anti-goat IgG for 1 h at room temperature. Immunoreactive proteins were detected with SuperSignal chemiluminescence reagents (Pierce) according to the manufacturer's instructions. Quantitation was performed on a LAS-1000 luminescence imager (Fuji) using Image Gauge 4.0 software (Fuji).
Isolation of 3xFLAG-IMP1 Granules—
Cells were lysed in ice-cold lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5 µg/ml PMSF, and 0.5% protease inhibitor mixture (Sigma-Aldrich)) by sonication. The cell lysate was cleared by centrifugation at 8200 x g for 10 min at 4 °C before 250 units of Protector RNase inhibitor (Roche Applied Science)/ml of lysis buffer was added. Immunoprecipitations were performed using FLAG-specific monoclonal antibody M2 covalently coupled agarose beads (Sigma) using 30 million cells/50 µl of beads. The beads were prewashed three times in TBS (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 0.5 µg/ml PMSF, and 0.5% protease inhibitor mixture) and resuspended in lysis buffer containing 25 µg/ml synthetic FLAG peptide (Sigma-Aldrich). The suspension was rotated for 1 h at 4 °C before the supernatant was exchanged with cleared cell lysate and rotated for an additional 2 h. The immunoprecipitate was washed four times in TBS, and the proteins were eluted by addition of 50 µg of synthetic FLAG peptide/ml of TBS followed by rotation for 2 h at 4 °C. Eluted proteins used for silver staining were precipitated by addition of 10% TCA followed by five washes in acetone. Purification of the 3xFLAG-IMP1 granule in the presence of Drosophila S2 cells was performed as described above with the exception that 1.7 times S2 cells were added to the HEK293 3xFLAG-IMP1 cells before lysis so equal amounts of RNA would be extracted.
Isolation of RNA from 3xFLAG-IMP1 Granules—
RNA isolation was performed as described above with a few modifications. Prior to the addition of the cleared lysate, 10 µg of yeast ribosomal RNA was added. The mRNA was isolated using oligo(dT)-agarose beads (Roche Applied Science). After washing of the beads mRNA was retrieved from the agarose beads by addition of 1 ml of TRIzol with 5 µg of yeast ribosomal RNA. Total RNA from HEK293 cells, HEK293 cells stably transfected with the 3xFLAG-IMP1 (no tetracycline), HEK293 cells stably expressing 3xFLAG-IMP1 by the inclusion of 1 µM tetracycline, or Drosophila S2 cells was isolated in TRIzol as described by the manufacturer (Invitrogen).
Protein Identification—
Immunoprecipitation (IP) was performed on 120 million HEK293 cells, HEK293 cells stably expressing 3xFLAG-IMP1, and HEK293 cells stably expressing FLAG-ERG1. After lysis in 3.6 ml of lysis buffer, 10 µl was removed and used as total protein extract of the cells. The IPs were TCA-precipitated and dissolved in 50 µl of 2x SDS loading buffer. The proteins were separated on a 10% SDS-polyacrylamide gel and silver-stained using Silverquest (Invitrogen) as described by the manufacturer. Slices were cut from the gel and digested with trypsin. The gel piece was cut into cubes and washed for 1 h in 100 mM NH4HCO3 before excess liquid was removed. Proteins were reduced for 30 min at 60 °C in 4 mM dithiothreitol in 100 mM NH4HCO3, cooled, and alkylated by addition of 100 mM iodoacetamide for 30 min in the dark. Excess liquid was removed, and the gel pieces were washed for 1 h in 50% acetonitrile in 100 mM NH4HCO3 followed by shrinkage in acetonitrile and vacuum centrifugation. The dry gel pieces were placed on ice and allowed to swell in 25 mM NH4HCO3 including 2 µg/100 µl modified trypsin (Promega). The supernatant was removed, and buffer was added to cover the gel pieces during overnight incubation at 37 °C. The next day, the supernatant was removed and combined with a second extract from the gel piece, obtained by incubation for 30 min in 50 µl of 0.5% formic acid. The combined extracts were loaded onto a C18 ZipTip (Millipore), washed with 0.5% formic acid, and eluted with 10 µl of 0.5% formic acid in 50% acetonitrile. A 0.5-µl aliquot was analyzed by MALDI-TOF using an AutoFlex instrument with TOF-TOF facility (Bruker Daltonics). An aliquot of 2 µl of the purified peptide mixture was introduced into a Q-TOF-2 tandem mass spectrometer (Micromass) using the nanospray interface and analyzed in MS mode as well as in MS/MS mode for a number of fragments to obtain sequence information. For the MALDI-TOF results, the flexAnalysis program (Bruker Daltonics) was used to create the peak lists, and internal calibration was performed using trypsin peptides. Profound version 4.10 was used for the database searches with a mass accuracy of 100–200 ppm. Mascot was used for database search on the MS/MS results. The nonredundant NCBI (NCBInr) database was searched, and the search results were accepted when the scores indicated identity or extensive homology with p values less than 0.05. Default values of the score threshold were used. When needed, keratin peaks were excluded prior to a new search. Often several names were indicated that appeared to be the same protein. Overall we used the Swiss-Prot names as the most appropriate.
Atomic Force Microscopy—
Atomic force microscopy (AFM) measurements were carried out using a MultiMode atomic force microscope on a Nanoscope III controller (Veeco). The atomic force microscope was calibrated in all axes using certified reference standards to an accuracy below 5%. For the measurements in liquid, silicon nitride cantilevers were used (OMCL-TR400PSA, Olympus), and for measurements in air, silicon cantilevers were used (Point-Probe-Plus, Nanosensors). For all samples mica was used as substrate (Moscovite Clear Ruby, Goodfellow). Mica pieces were glued onto metal discs using Araldite 2011 (Huntsman). 3xFLAG-IMP1 RNP and FLAG-ERG1 proteins were immunoprecipitated from 3 x 107 cells as described above. The eluted proteins were diluted 1000x in TBS, and 100 µl was pipetted onto the freshly cleaved mica surface. After 15 min of incubation, the samples were rinsed three time in TBS buffer. The samples were kept under TBS buffer at all times and mounted in the atomic force microscope. The dimerization of IMP1 on RNA was performed by mixing a 1 nM concentration of an RNA SELEX target with 2 nM recombinant IMP1 (rIMP1) in UV cross-linking buffer (24). The sample was incubated for 25 min at room temperature and stored at –20 °C. 50 µl of the rIMP1 SELEX sample was allowed to adsorb to freshly cleaved mica for 15 min. The samples were rinsed briefly in TBS and water, dried, and mounted in the atomic force microscope. Thermal equilibration of the atomic force microscope and sample was allowed for at least 1 h before imaging was started. All samples were investigated in at least three separate areas. Image analysis was done using Scanning Probe Image Processor (SPIP) (Image Metrology). All images were subjected to first order linewise leveling before further analysis. Heights of the various features in the images were determined using cross-section plots.
DNA Microarray Analysis—
2 µg of purified total RNA from HEK293 cells (HEK293 total), 0.3–1 µg of RNA from immunoprecipitated beads containing HEK293 cell lysates (HEK293 IP), 15 µg of RNA from immunoprecipitated beads containing lysates from HEK293 cells expressing FLAG-IMP1 (HEK293 FLAG-IMP1 IP), 2 µg of total RNA from Drosophila S2 cells (S2 total), 0.3–1 µg of RNA from immunoprecipitated beads containing Drosophila S2 cell lysates (S2 IP), and 15 µg of RNA from immunoprecipitated beads containing a mixture of lysates from HEK293 cells expressing FLAG-IMP1 and Drosophila S2 cells (HEK293 FLAG-IMP1/S2 IP) were used to synthesize double-stranded cDNA with the Superscript Choice system (Invitrogen). The cDNA was used as template in an in vitro transcription reaction to generate biotin-labeled antisense cRNA (BioArray high yield RNA transcript labeling kit, Enzo Diagnostics). After fragmentation at 94 °C for 35 min in 40 mM Tris, 30 mM magnesium acetate, 10 mM potassium acetate, 1–3 µg of labeled HEK293 IP cRNA, 15 µg of labeled HEK293 FLAG-IMP1 IP, or 15 µg of labeled HEK293 total cRNA was hybridized for 16 h to Affymetrix HG-U133plus2 arrays (Affymetrix Inc.) containing 54,613 probe sets, whereas 1–3 µg of labeled S2 IP cRNA, 15 µg of labeled HEK293 FLAG-IMP1/S2 IP cRNA, or 15 µg of labeled S2 total cRNA was hybridized for 16 h to Affymetrix Drosophila Genome 2.0 arrays (Affymetrix Inc.) containing 18,952 probe sets. The arrays were washed and stained with phycoerythrin-streptavidin using the Affymetrix Fluidics Station 450, and the arrays were scanned in the Affymetrix GeneArray 3000 scanner.
Three independent experiments with RNA from HEK293 total, HEK293 IP, HEK293 FLAG-IMP1 IP, S2 total, S2 IP, and HEK293 FLAG-IMP1/S2 IP were performed. The individual samples were analyzed using two different approaches. First, for HEK293 IP and HEK293 FLAG-IMP1 IP samples and S2 IP and HEK293 FLAG-IMP1/S2 IP samples, the raw intensity measures were obtained by scaling the expression profiles by a scaling factor equal to 1, which does not change the probe intensity measures. Based on these unscaled probe signals, expression values were calculated by the MAS5 algorithm using Affymetrix GCOS version 1.3. The expression values were imported into the dChip software package2 for further analysis. HEK293 FLAG-IMP1 IP samples were compared with HEK293 IP (unspecific pulldown) to detect mRNAs that are present in the IMP1 granules. Probe sets were identified as being differentially expressed if the comparison of groups yielded a -fold change higher than 100 (using the 90% lower confidence interval of the -fold change), a difference of group means higher than 100 expression units, and a p value below 0.05 (two-tailed, two-sample unequal variance t test, Welch t test).
HEK293 FLAG-IMP1/S2 IP samples were compared with S2 IP samples (unspecific pulldown) hybridized to Drosophila Genome 2.0 arrays to detect potential unspecific mRNAs that may be present in the IMP1 granule pulldown. Probe sets were identified as being differentially expressed if the comparison of groups yielded a -fold change higher than 50 (using the 90% lower confidence interval of the -fold change), a difference of group means higher than 50 expression units, and a p value below 0.05 (two-tailed, two-sample unequal variance t test, Welch t test). The 50-fold change cutoff for the Drosophila S2 samples was chosen to reduce stringency and because only about half the amount of cRNA was hybridized to the Drosophila Genome 2.0 arrays as compared with the HG-U133plus2.0 arrays where we used a 100-fold change cutoff. Second, total RNA samples and IMP1-FLAG bead samples with and without Drosophila mRNA that were hybridized to the HG-U133plus2 arrays were normalized/scaled to a target intensity of 100 using GCOS version 1.3, and expression values were calculated by the MAS5 algorithm implemented in GCOS version 1.3. The resulting expression matrix for each type of array, the HG-U133plus2 array and Drosophila Genome 2.0 array, were analyzed further after import into dChip2 where the expression profiles were used for detection of genes present in the IMP1 granules. Three comparisons were performed: HEK293 FLAG-IMP1 IP expression profiles against HEK293 total expression profiles, HEK293 FLAG-IMP1/S2 IP against HEK293 total for the HG-U133plus2 hybridizations, and HEK293 FLAG-IMP1/S2 IP against S2 total for the Drosophila Genome 2.0 array samples. First, for the human samples without Drosophila mRNA, only the 569 genes/probe sets that were selected as being present in the HEK293 FLAG-IMP1 IP versus HEK293 IP comparison of non-scaled data were used in this analysis. The 569-probe set list was transformed to scaled expression values, and probe sets that showed above a 3-fold change between HEK293 FLAG-IMP1 IP versus HEK293 total samples as well as a difference of group means higher than 100 expression units and a p value below 0.05 (two-tailed, two-sample unequal variance t test, Welch t test) were selected as being present in the IMP1 granules. Second, for the human samples with S2 Drosophila mRNA, only the 1469 genes/probe sets that were selected as being present in the HEK293 FLAG-IMP1/S2 IP versus HEK293 IP comparison of non-scaled data were used in this analysis. The 1469-probe set list was transformed to scaled expression values, and probe sets that showed above a 1.5-fold change (using the 90% lower confidence interval of the -fold change) and a difference of group means higher than 100 expression units between HEK293 FLAG-IMP1/S2 IP and HEK293 total samples were selected as being present in the IMP1 granules. Third, for the HEK293 FLAG-IMP1/S2 IP against S2 total, only the six genes/probe sets that were selected as being present in the S2 IP versus HEK293 FLAG-IMP1/S2 IP comparison of non-scaled data were used in this analysis. The six-probe set list was transformed to scaled expression values, and probe sets that showed above a 1.5-fold were selected as immunoprecipitated along with the IMP1 granules. Microarray data can be accessed at the European Bioinformatics Institute ArrayExpress database (www.ebi.ac.uk) under accession number E-MEXP-841.
Identification of IMP1 Binding Sites in Target mRNAs—
352 genes with a 3-fold enrichment or more were selected as a positive set. 1020 genes that were selected from the bottom of the list with a –8.2- to –200-fold enrichment were selected as a negative set. The mRNA sequences were retrieved from NCBI Entrez using the identities. Some were not found, and others were excluded because they contained more than 1% unknown bases. Sequences longer than 5000 bases were also excluded so they would not dominate the analysis. The final positive set contained 307 sequences, and the negative set contained 857 sequences. All subsequences of length 7 (words) were counted in the positive and negative sets. For each word, the over-representation in the positive set was calculated as the observed number divided by the expected from the negative set. If Np(w) and Nn(w) are the number of occurrences of word w in the positive and negative sets, respectively, and Np and Nn denote the total number of words in the two sets, the overrepresentation is Np(w)Nn/(Nn(w)Np). A 
2 p value was calculated with a Bonferroni correction for multiple testing (multiplying the p value with the total number of words). Fisher's exact test was also used, but it resulted in the same word list. For each word the coverage, that is the fraction of sequences in the positive set containing the word, was also calculated. The same analysis was done for words of size 5, 6, and 8.
Supplemental Data—
The supplemental data include two figures, four tables, and additional experimental procedures.
|
RESULTS |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
72 kDa corresponding to endogenous IMP1 and FLAG-IMP1 RNP, respectively, were apparent (Fig. 1D).
|
IMP1 Granules in Relation to hStaufen and FMRP Granules, P-bodies, and Stress Granules—
To determine whether IMP1 granules were distinct from hStaufen and FMRP granules, YFP-tagged hStaufen or FMRP as well as IMP3 and HuB, the latter being a member of the ELAV family recently shown to interact with IMP1 (23), were transfected into HEK293 FLAG-IMP1 or FLAG-ERG1 cells. Following immunoprecipitation of the FLAG-tagged proteins, the YFP-tagged factors were identified by Western analysis with anti-green fluorescent protein antibody (Fig. 2A). Whereas HuB and IMP3 co-immunoprecipitated, there was no evidence of any association with YFP-hStaufen or -FMRP. The results were substantiated by confocal microscopy of the living cells. This corroborated the presence of HuB in IMP granules and showed that FMRP exhibited a distinct cytotopical localization. The cellular localization of hStaufen was similar to that of CFP-IMP1, but on closer inspection there was no obvious overlap at the granular level (Fig. 2B). To establish whether IMP1 granules were associated with mRNA-degrading P-bodies or stress granules, we examined whether eIF4E, DCP1a, or G3BP were enriched in the pelleted IMP1 granules (Fig. 2C). All factors were detected in the supernatant of the lysates, but none was present in the pelleted RNP. Taken together, we conclude that IMP1 RNP granules are distinct entities that are different from hStaufen and FMRP RNP granules, P-bodies, and stress granules.
|
|
and components of the exon junction complex, UPF3B, Aly/REF, Y14, and eIF4AIII, were examined by Western blot analysis (Fig. 4B). Whereas the three initiation factors were excluded from the particles, the four examined exon junction complex components were detected in the immunoprecipitate. In addition, the presence of CBP80 was corroborated, and ribosomal protein S6 was added to the list of small ribosomal subunit proteins. Moreover capillary electrophoresis firmly established the presence of 18 S rRNA and the absence of 28 S rRNA in the IMP1-containing RNP granules (Fig. 4C). The absence of crucial initiation factors and 60 S ribosomal subunits indicates that embodied mRNAs are untranslated. In addition, the presence of CBP80, PABP2, and the exon junction complex components suggests that at least some of the transcripts have not undergone a pioneering round of translation (38).
|
Compared with the bead controls, enrichment varied from 100- to 2478-fold, and compared with the total RNA pool the enrichment was 3–18-fold. The enrichment was greatest for the least abundant mRNAs, and none of the 100 most abundant transcripts in HEK293 cells were enriched in the particles. From the relative signal values of the pelleted mRNAs, we estimated that the fraction of bound transcript ranged from 15% to almost 100%. Because reassociation of RNA-binding proteins and mRNAs has been described to occur after cell lysis, leading to immunoprecipitation of transcripts that are not physiologically relevant (39), we designed a global control experiment to determine the significance of this artifact during isolation of IMP1 granules. Briefly tetracycline-induced FLAG-IMP1 cells were mixed with Drosophila S2 cells, corresponding to a 1:1 ratio of total cellular RNA, before they were lysed, and FLAG-IMP1 particles were isolated. Immunoprecipitated mRNAs were subsequently analyzed on Affymetrix human 133 Plus 2.0 arrays as well as Drosophila Genome 2.0 arrays and compared with bead controls and total RNA as described above, but the stringency was lowered so transcripts only had to be 50-fold enriched compared with bead controls and 1.5-fold compared with the total RNA pool to be included. Whereas 671 human transcripts were found to be enriched in the particle, no Drosophila mRNAs were found to be enriched according to the criteria used. We therefore infer that reassociation is unlikely to invalidate the identification of target mRNAs. To examine whether the enriched transcripts represented direct RNA targets, we performed a UV cross-linking experiment with eight granule-enriched RNAs and 10 transcripts that were not enriched. Supplemental Fig. S2 shows the cross-linking pattern of three representative transcripts from each category. Whereas all granule-enriched transcripts demonstrated significant cross-links, corresponding to both the 69-kDa endogenous IMPs and the tetracycline-inducible 72-kDa FLAG-IMP1, there was no significant cross-linking to any of the transcripts that were not found to be enriched in the granules.
The characteristics of the IMP1 mRNA targets with respect to length of the 5'-UTR, coding region, and 3'-UTR are shown in Fig. 5. IMP1 target mRNAs exhibited shorter 5'-UTRs and coding region lengths but had longer 3'-UTRs compared with the entire mRNA transcriptome (40). Previous attempts to identify common denominators of IMP1 attachment sites in the RNA sequences have been hampered by the small number of reported RNA targets. To identify common sequence motifs or "words" in the granule-enriched mRNAs, sequences were retrieved from NCBI Entrez. Some were absent, and others were excluded because they contained more than 1% unknown nucleotides. Sequences longer than 5000 bases were also excluded so they would not skew the analysis. The final positive set of transcripts contained 307 sequences, and the negative set contained 857 sequences. In these two sets, all subsequences of length 7 (a "word") were counted. Words of length 5, 6, and 8 were counted in the same manner and yielded similar results (data not shown). The most over- and underrepresented words (p < 0.0001) are shown in Fig. 5. Taken together the analysis shows that IMP1 targets are enriched in CCYHHCC (Y = C or U and H = C or U or A)-rich motifs and lack purine-rich stretches.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The proteome analysis of the isolated IMP1 granules revealed that the most abundant group of proteins was the hnRNPs. However, from a stoichiometric point of view the major components of the granules were IMPs, YB1, and 40 S ribosomal subunits. The majority of the hnRNPs were probably enriched in the granules by virtue of their binding to the associated mRNAs, but it is possible that some are not only passive bystanders, having previously been implicated in mRNA localization and translational control. Central issues concerning RNP granules have been whether associated mRNAs are translationally repressed and at which stage the transcripts are packaged into the particles. The presence of CBP80, PABP2, and factors belonging to the exon junction complex implies that at least some of the embodied mRNAs have not been subjected to a pioneering round of translation (1, 2), and the absence of eIF4E, eIF4G, and 60 S ribosomal subunits further supports that the granular mRNAs are translationally inactive during transport. This inference is strengthened by the abundance of YB1, which is known to be a predominant component of translationally inactive mRNPs. YB1 has also been reported to displace eIF4E and eIF4G from the cap structure (41–43).
IMP1 granules have so far only been visualized by confocal microscopy so to obtain a closer view they were examined by AFM. The spherical granules exhibited a smooth surface with a diameter in the range of 100–300 nm, which was about half the size range of the optical diameter in cultured cells but similar to the range observed in developing brain (44). IMPs have been shown previously to associate with RNA by a sequential dimerization mechanism, and dimers were visible in AFM pictures of recombinant IMP1 bound to an RNA SELEX target (37) where each subunit in the recombinant IMP1 dimer was roughly 30 nm in diameter. This means that a 300-nm particle with a volume of 14 x 106 nm3 leaves space for about 680 molecules roughly the size of IMPs, YB1, and 40 S subunits. The space occupied by associated RNAs is likely to be small because a 2000-nucleotide RNA strand has a length of 600 nm and a width of 1.5 nm (volume, 1 x 103 nm3) (45). A calculation of the number of transcripts per particle, based on either the volume of the granule and the number of proteins in the granule or the estimated number of target copies and the number of granules in the cells, reveals that each granule contains 10–30 mRNA transcripts. The granules are therefore likely to be heterogeneous because IMPs associate with several hundred different transcripts. This number of transcripts per granule is similar to what has been determined previously for the core hnRNP A2 complex, which was reported to contain 29 gag and vpr RNAs on average (46).
We identified about 300 different mRNAs in the granules corresponding to 3% of the HEK293 mRNA transcriptome. In agreement with the lack of common elements among previously reported targets, the comprehensive analysis in this study did not identify a single common IMP1 binding element. Granule-associated mRNAs exhibited enriched CCYHHCC elements suggesting that IMP1 has a preference for unstructured guanosine-poor and cytosine-rich stretches. HEK293 cells do not express IGF2, H19, or c-MYC target mRNAs, so these transcripts were obviously not included in the granules. ß-actin mRNA was only enriched about 2-fold in agreement with the low binding affinity of IMPs toward the zip code (47). IMP1 granules encompassed an abundant representation of mRNAs encoding proteins involved in ER and Golgi function. Moreover components of the ubiquitin-dependent protein degradation machinery were extensively represented (Table I) (for a review, see Ref. 48). The two pathways are functionally connected because misfolded proteins, which are dislocated from the ER to the cytoplasm through the Sec61 translocon, are destined for ubiquitin-mediated degradation by the proteasome (49). Fig. 7 gives an overview of the function of the granule-associated transcripts from the two pathways. Control of protein secretion is required during development where extensive protein synthesis may generate a "constitutive" stress situation (50, 51). In this scenario, embryonal IMP1 granules would facilitate a coordinated production of factors involved in the secretory pathway to avoid improper processing and overloading. With respect to the ubiquitin operon, we speculate that IMP1-mediated transport may be responsible for the local ubiquitination at the synapses (52, 53).
|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Published, MCP Papers in Press, February 7, 2007, DOI 10.1074/mcp.M600346-MCP200
1 The abbreviations used are: CBP, cap-binding protein; ER, endoplasmic reticulum; RNP, ribonucleoprotein; IMP, insulin-like growth factor II mRNA-binding protein; hnRNP, heterogeneous nuclear ribonucleoprotein; ZBP1, zip code-binding protein 1; AFM, atomic force microscopy; G3BP, Ras GTPase-activating protein Src homology 3 domain-binding protein; eIF, eukaryotic translation initiation factor; ELAV, embryonic lethal abnormal vision; ERG1, early growth response 1; FMRP, fragile X mental retardation protein; NCBI, National Center for Biotechnology Information; IP, immunoprecipitation; CFP, cyan fluorescent protein; YFP, yellow fluorescent protein; PABP, poly(A)-binding protein; PIG, phosphatidylinositol glycan; SCAMP, secretory carrier membrane protein; VPS, vacuolar protein sorting; RING, really interesting new gene; RNF, RING finger; PSM, proteasome activator complex; NFAR, nuclear factors associated with double-stranded RNA; YB1, Y box-binding protein 1; HEK, human embryonic kidney; mRNP, messenger ribonucleoprotein; P-bodies, processing bodies; RBP, RNA-binding protein; rIMP1, recombinant IMP1; SELEX, systematic evolution of ligands by exponential enrichment; GCOS, GeneChip Operating Software; UTR, untranslated region; E2, ubiquitin carrier protein; E3, ubiquitin-protein isopeptide ligase. 
2 C. Li and W. Wong, unpublished software.
* This work was supported by The Danish Natural Science and Medical Research Councils, the Danish Cancer Society, the Novo Nordisk Foundation, and the Toyota Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
S The on-line version of this article (available at http://www.mcponline.org) contains supplemental material.
** Both authors contributed equally to this work.
To whom correspondence should be addressed. Tel.: 45-3545-2223; Fax: 45-3545-4640, E-mail: FCN{at}rh.dk
|
REFERENCES |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
