SEARCH

SEARCH BY CITATION

Keywords:

  • intermediate filaments;
  • Nestin;
  • Gfap;
  • Vimentin;
  • cytoskeleton;
  • glia

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Subcellular RNA localization plays an important role in development, cell differentiation, and cell migration. For a comprehensive description of the population of protrusion localized mRNAs in astrocytes we separated protrusions from cell bodies in a Boyden chamber and performed high-throughput direct RNA sequencing. The mRNAs with localization in astrocyte protrusions encode proteins belonging to a variety of functional groups indicating involvement of RNA localization for a palette of cellular functions. The mRNA encoding the intermediate filament protein Nestin was among the identified mRNAs. By RT-qPCR and RNA FISH analysis we confirmed Nestin mRNA localization in cell protrusions and also protrusion localization of Nestin protein. Nestin mRNA localization was dependent of Fragile X mental retardation syndrome proteins Fmrp and Fxr1, and the Nestin 3'-UTR was sufficient to mediate protrusion mRNA localization. The mRNAs for two other intermediate filament proteins in astrocytes, Gfap and Vimentin, have moderate and no protrusion localization, respectively, showing that individual intermediate filament components have different localization mechanisms. The correlated localization of Nestin mRNA with Nestin protein in cell protrusions indicates the presence of a regulatory mechanism at the mRNA localization level for the Nestin intermediate filament protein with potential importance for astrocyte functions during brain development and maintenance. GLIA 2013;61:1922–1937


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Astrocytes constitute the most abundant cell type in the CNS (Allen and Barres, 2009; Freeman, 2010; Sofroniew and Vinters, 2010). Astrocytes typically exhibit a highly polarized morphology extending multiple pseudopodial protrusions participating in (i) establishing scaffolds for crawling neurons during CNS development, (ii) establishing a part of the gliovascular structure and a part of the blood-brain-barrier, and (iii) mediating interactions with synapses aiding in optimal neuronal signal transduction (Allen and Barres, 2009; Morest and Silver, 2003; Sofroniew and Vinters, 2010). Many studies have been carried out in order to understand the dynamics of astrocyte protrusions. During cell differentiation and migration the astrocyte morphology changes dramatically by formation of pseudopodial protrusions driven by coordinated polymerization and depolymerization of the cytoskeleton (Etienne-Manneville, 2004). The mammalian cytoskeleton consists of three types of filaments: actin filaments, microtubules, and intermediate filaments (IFs). The highly diverse family of IF proteins are encoded by ∼70 genes, and the complexity of the IF family is increased by generation of multiple protein isoforms (Herrmann et al., 2009). Proteins of the IF family are subdivided into different classes due to sequence homology and capability to co-assemble into IFs (Eriksson et al., 2009; Herrmann et al., 2009). IF proteins have a common overall structure composed of an amino-terminal head domain and a carboxy-terminal tail domain linked together by a highly coiled α-helical rod domain. The rod domain is highly conserved among IF proteins whereas the head and tail domains exhibit a large degree of variability (Eriksson et al., 2009; Herrmann et al., 2009; Middeldorp and Hol, 2011). The coiled α-helical rod domain mediates generation of parallel homo-dimers, and the homo-dimers form anti-parallel tetramers that are linked head to tail. Eight tetramers associate into unit-length filaments by which the mature 7- to 11-nm thick IFs are comprised (Herrmann et al., 2009). Apart from providing static mechanical support to the cell structure, IF proteins are also involved in dynamic reorganization of the cell morphology during growth and migration (Eriksson et al., 2009; Michalczyk and Ziman, 2005).

Astrocytes of the mammalian brain express the IF proteins Nestin, Vimentin, and Glial fibrillary acidic protein (Gfap) (Middeldorp and Hol, 2011). Nestin is considered a marker for undifferentiated progenitor cells, down regulated in terminally differentiated cells, but become reactivated during injury responses in a process termed reactive gliosis characterized by cell hypertrophy and proliferation (Gilyarov, 2008). The Nestin protein has a short amino-terminal head domain of only six amino acids (Herrmann et al., 2009). Nestin is unable to form homomeric filaments but coassembles with Vimentin and Gfap (Eliasson et al., 1999; Herrmann et al., 2009; Michalczyk and Ziman, 2005). Nestin is believed to play a pivotal role in cell morphology changes during mitosis through the dynamic assembly and disassembly of Vimentin including IFs (Chou et al., 2003). Vimentin, like Nestin, is expressed early during brain development. Vimentin and Nestin becomes gradually replaced by Gfap in terminally differentiated astrocytes but Gfap expression has also been described in neuronal and astrocyte progenitor cells (Doetsch et al., 1999; Imura et al., 2006; Michalczyk and Ziman, 2005; Middeldorp and Hol, 2011; Middeldorp et al., 2010; Zhu and Dahlstrom, 2007). Gfap expression is upregulated along with Nestin and Vimentin during reactive gliosis after CNS injury (Pekny et al., 2007). Gfap mRNA is alternative spliced to generate protein isoforms whereas alternative splicing generating protein isoforms of Nestin and Vimentin is not yet described, but alternative Vimentin mRNA splicing exists (Blechingberg et al., 2007a; Condorelli et al., 1999; Middeldorp and Hol, 2011; Nielsen and Jorgensen, 2004; Nielsen et al., 2002; Quinlan et al., 2007; Zhou et al., 2010). Attention is emerging to understand how IFs are regulated during cell growth, migration and morphology changes.

RNA localization in cell protrusions was demonstrated in a considerable number of cell types including oocytes, crawling fibroblasts and in axons and dendrites of neurons (Doyle and Kiebler, 2011; Mili and Macara, 2009). RNA localization is important for development, cell differentiation and cell functionality (Mili and Macara, 2009). A limited number of studies have described IF mRNA localization in astrocytes. One study demonstrated subcellular localization of the Nestin mRNA in protrusions of radial glia cells (Dahlstrand et al., 1995). Gfap mRNA localization was demonstrated in the branch points and distal parts of astrocyte protrusions (Landry et al., 1994; Medrano and Steward, 2001). However, detailed studies of the subcellular mRNA localization patterns in astrocytes remains to be conducted. Comprehensive examinations of RNA molecules localized in fibroblast, neuronal and cancer cell protrusions were performed by microarray analysis of RNA purified from pseudopodial cell protrusions separated from cell bodies using the Boyden chamber assay (Feltrin et al., 2012; Mili et al., 2008; Shankar et al., 2010). In this study we took advantage of a recently presented high throughput next generation sequencing (NGS) method, termed single molecule direct RNA sequencing (DRS) (Ozsolak et al., 2009). DRS analyses were applied on mouse astrocyte RNA from protrusions and cell bodies isolated using a modified Boyden chamber assay. We identified numerous mRNAs with localization to cell protrusions including Nestin mRNA. We further analyzed Nestin mRNA and protein localization in astrocyte protrusions and the presented results indicate that the function of Nestin in reorganization of astrocyte morphology can be regulated through coordinated mRNA and protein localization.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Primary Cells, Cell Lines, and Brain Tissue

Primary astrocytes cultures were prepared as described (Andres-Barquin et al., 1994; Thomsen and Lade Nielsen, 2011). A pregnant mouse of the NMRI mouse strain was purchased from Taconic, Denmark and astrocytes were isolated from the cerebral cortex of newborn (P0) mice. After 8 days in vitro (DIV) cells showed a 95% confluence and >80% of the cell were staining positive for the astrocyte marker Gfap. Before experimental procedure cells were trypsinized using 0.5% trypsin-EDTA (GIBCO). Under these conditions, neurons, oligodendrocytes and microglia rapidly die or do not adhere (Imura et al., 2006). The type 2 astrocyte mouse cell line C8-S, the mouse embryo-derived teratocarcinoma cell line P19, the mouse neuroblastoma cell line N1E-115, and the mouse fibroblast cell line NIH/3T3 were purchased from the American Tissue Culture Collection and cultured in Dulbecco's Modified eagle Medium (DMEM) with 10% Fetal Bovine Serum (FBS), streptomycin, penicillin and glutamine, in a 5% CO2 humidified atmosphere at 37°C. P19 cells were, if indicated, incubated with retinoic acid (1 μM, R2625, Sigma) for 24 h. Whole brains were collected from P0 and 21-month-old (P21m) NMRI mouse.

Boyden Chamber Isolation of Cell Protrusions

Cell protrusions were isolated using a modified Boyden chamber assay as previously described (Thomsen and Lade Nielsen, 2011). To obtain sufficient protein and RNA, six 9.6 cm2 cell culture inserts (BD Falcon) with a 1-μm pore size polystyrene membrane, and three 75 cm2 growth flasks of cell culture were used for each experiment. Cell culture inserts were coated with extra cellular matrix (ECM) protein Collagen type-I (Sigma; C7661) (final concentration 10 μg mL−1 in phosphate buffered saline (PBS)) at 37°C for 2 h. The cell cultures inserts were afterward directly transferred to a six-well tissue culture dish containing serum free DMEM. Cells were grown to 90% confluence and medium changed to serum free medium with antibiotics and glutamine before cells were grown for additional 24 h. Cells were detached using 0.5% Trypsin-EDTA, and the trypsin was inactivated by dissolving cells in DMEM with 10% FBS. Cells were pelleted and re-dissolved in serum free DMEM, and 2 × 106 cells were seeded for each cell culture insert. Cell protrusions were growing through the membrane for 24 h at 37°C. Inserts were washed in PBS and the cells were gently scraped off the membrane. First the cell protrusion fraction (PF) was made by scraping the bottom side of the membrane. Cells were lysed by carefully washing the cell scraper in 1 mL TRI-Reagent (Sigma) for RNA or in 1 mL 1x protein gel loading buffer (Fermentas) for protein isolation. Afterwards, the cell body fraction (CF) was made by scraping the upper side of the membrane and dissolving the cell material in TRI-reagent or 1x protein gel loading buffer.

RNA Purification, cDNA Synthesis and Real Time Quantitative PCR (RT-qPCR)

RNA from cells and tissues was purified by standard TRI-reagent protocol (Sigma) using 1 μg of glycogen for precipitation (Thomsen and Lade Nielsen, 2011). cDNA was made using an iScript cDNA synthesis kit (Bio-Rad). Nearly 1 μL of total RNA solution was used per reaction. RT-qPCR was performed using SYBR Green 480 master mix (Roche). RT-qPCR analyses were performed using a Roche LightcyclerTM 480, with a primer annealing temperature of 58°C. Primers were designed as intron spanning and PCR amplicons were verified by gel electrophoresis and melting curve peak analysis. Primer sequences are shown in Supporting Information Table 1.

Direct RNA Sequencing (DRS), Data Processing and Analysis

A modified Boyden chamber assay was used for isolation of cell protrusions and cell bodies as previously described (Thomsen and Lade Nielsen, 2011). Single RNA molecules were sequenced by DRS (Ozsolak et al., 2009) using the Helicos Biosciences platform (Helicos Biosciences, Boston, MA). Sequencing reads were mapped to a 2 kb region surrounding the 3′ distal part of 27,131 genes for poly-adenylated RNA of the mm9 version of the mouse genome. Bioinformatics raw data analysis and sequence alignment was made by Helicos Biosciences, and transcript reads were presented as an excel spread sheet. Details of the RNA preparation and sequencing can be found at http://www.helicosbio.com/. Expression values were processed as RNA transcripts per million reads (tpm). To avoid including false positives due to stochastic counts in the lower range, transcripts exhibiting counts of <5 tpm in both PF and CF were excluded. The RNA enrichment in protrusions was presented as the relative expression value in PF compared to CF [tpm PF/tpm CF]. A list of the 250 transcripts with the highest ratio were submitted for ordination of gene name, location, and type, followed by functional annotation analysis using the IPA ingenuity online platform, URL: http://www.ingenuity.com. For increased stringency a direct relationship analysis was performed including pathways described both human rat and mouse. Results were presented with P values calculated by the Benjamini-Hochberg method, to control for multiple testing. RT-qPCR amplifications were made in triplicates for each gene and Ct values were converted into linear values using the Xo method (Thomsen et al., 2010). The mRNA localization ratio was calculated as the ratio between the mean expression level in CF and PF and afterwards normalized to the determined mRNA localization ratio for Arpc3. Differences in localization ratios were analyzed by a Student's unpaired two tailed t test. All experiments were repeated three times.

siRNA Transfections

For siRNA experiments 1.000.000 C8-S cells were immediately before the transfection plated into 10 cm dishes in DMEM with 10% FBS. In 640 μL serum free medium was mixed siRNA to a final concentration of 2 μM and incubated for 5 min. Nearly 13 μL Dharmafect was mixed with 1267 μL serum free medium and incubated 5 min. The two solutions were mixed and incubated 20 min, added to the cells, and incubated 24 h. The medium was changed to serum free medium and cells incubated for further 24 h. The cells were used in a standard Boyden chamber assay with 1 μm membranes and RNA purified from three membranes for each transfection and pooled. siRNA sequences Fmr1−1063: GGAUCAAGAUGCAGUGAAA; Fmr1−447: GUGAUGAAGUUGAGGUUUA; Fxr1−219: GAGAUGAAGUAGAGGUAUA; Fxr1−560: GCAACUGUGAAGAGAGUAA; Fxr2−1269: GGAAAGAACGGGAAAGUGA; Fxr2−1336: GAGAUAACGACAAGAAGAA; non-specific control: AGGUAGUGUAAUCGCCUUG.

Cloning of 3′-UTR Reporter Constructs and Transfections

The 3′-UTR sequences were amplified from mouse C8S cDNA (primer sequences are shown in Supp. Info. Table 1) and purified bands digested with XmaI and XbaI. The pcDNA5-beta-globin-6xMS2-SV40-LpA vector was XmaI and XbaI digested and after ligation with UTR inserts and E. coli transformation, positive constructs were verified by sequencing. To remove the 6xMS2-sites the vectors were digested with NotI and SmaI and by Klenow polymerase treatment blunt-ended. After ligation and E. coli transformation constructs were verified by sequencing. The constructs were pooled and transfected into NIH/3T3 cells using 2 μg total DNA mixture. In parallel was transfected the control construct pTAG4 (Blechingberg et al., 2007b). For each transfection were used 150,000 cells in six-well plates using 200 μL serum-free medium and 3 μL Xtreme Gene 9 DNA transfection reagent version 03 (Roche). Cells were incubated 24 h before a medium shift to serum free medium and a subsequent incubation for 24 h. The cells were used in a standard Boyden chamber assay with 1 μM membranes and RNA purified from three membranes for each transfection and pooled.

Western Blot Analysis

Proteins were extracted for Western blot analysis by the Boyden chamber assay. To normalize the protein amount from CF and PF α-Tubulin and Actb were used as loading controls. Proteins were separated by SDS-PAGE in a 4–15% gradient polyacryl amide gel (Bio-Rad). Proteins were transferred to a nitrocellulose membrane and analyzed with following primary antibodies: rabbit anti α-Tubulin (Rockland), rabbit anti β-Actin (Actb) (Sigma; A2103), rabbit anti Histone H3 (Abcam; ab1791), mouse anti Nestin (Millipore; MAB253), Mouse monoclonal anti Vimentin (Abcam; ab20346), and goat anti Gfap (Santa Cruz; sc-6170). Nestin antibody was diluted 1:1,000 and all other antibodies were diluted 1:10,000. Horse radish peroxidase conjugated anti mouse, anti rabbit and anti goat secondary antibodies (DAKO) were used for detection.

Single RNA Molecule FISH and ISH

Single molecule RNA fluorescence in situ hybridization (FISH) was essential done as described (Femino et al., 1998). Probes consisting of 50-mer single stranded DNA oligonucleotides were synthesized and labeled with 4–5 Cy3 fluorophores. A total of eight various oligonucleotides were hybridized to each target mRNA. Cells were seeded onto 0.17-mm-thick coverslips (Marienfeld) either coated with Collagen type 1 or uncoated and cultured in DMEM with 10% FBS, penicillin, streptomycin, and glutamine. At ∼60% confluence cells were fixed in 4% paraformaldehyde for 20 min at room temperature, and washed and stored in phosphate buffered saline (PBS) at 4°C. Before hybridization, cells were permeabilized using 0.5% triton X-100 in PBS for 10 min at room temperature, washed in PBS, and then incubated in pre-hybridization solution: (50% formamide (Sigma; F4761) and 2 × SSC (Ambion)) for 15 min. at room temperature. The probes were hybridized in prehybridization solution supplemented with 2 mg mL−1 BSA (Roche), 0.2 mg mL−1 E. coli tRNA (Roche), and 0.2 mg mL−1 sheared salmon sperm DNA (Sigma; D7656) for 3 h at 37°C. About 10 ng DNA probe was used per coverslip. Cells were washed twice with pre-hybridization solution for 20 min at 37°C, then 10 min in 2× SSC at room temperature, and in PBS for 10 min at room temperature. Cell nuclei were counterstained with DAPI (0.5 mg L−1 in PBS). After a final wash in PBS, coverslips were rinsed in double distilled water to remove excess salt, dried and mounted using ProLong gold (InVitrogen). Actb probes were a kind gift from Dr. Robert H. Singer. Probe sequences for Nestin and Gfap mRNA are shown in Supporting Information Table 2. SSA4 oligonucleotides were previously described (Jensen et al., 2001).

Mouse P4 sagittal section mRNA in situ hybridization (ISH) data for Nestin, Gfap, Actb, CamKIIα and Tubb3 were extracted from Allen Developing Mouse Brain Atlas (http://developingmouse.brain-map.org) and used for publication with permission.

Immunofluorescence

Cells were grown on 0.17 mm coverslips until 60% confluence, then fixed in 4% paraformaldehyde for 20 min at room temperature, and washed and stored in PBS at 4o C. Cells were permeabilized using 0.5% triton X-100 in PBS for 10 min at room temperature. A blocking step was made using 1% BSA in PBS for 30 min at room temperature. Primary antibodies were dissolved in blocking buffer and incubated for 1 h at room temperature. Cells were washed three times in PBS and incubating with secondary antibody dissolved in blocking buffer. After three washes in PBS double immunofluorescence was performed as described above with a second treatment of primary and secondary antibodies. After final secondary antibody incubation cell were washed two times and cell nuclei were stained with DAPI and washed once in PBS. Coverslips were rinsed in double distilled water to remove salt, dried, and mounted with ProLong gold. Primary antibodies used: Mouse anti Nestin (Milipore; MAB 253), mouse anti Vimentin (Abcam; ab20346), rabbit anti Gfap (DAKO), rabbit anti α-Tubulin (Rockland), and rabbit anti Actb (Sigma; A2103). All primary antibodies were diluted 1:500. Secondary antibodies were Alexa 488 conjugated goat anti rabbit IgG (Invitrogen A11034) and Alexa 555 conjugated goat anti mouse IgG (Invitrogen A21127) both diluted 1:2000.

Microscopy and Image Processing

All images were made on a Zeiss axiovert 200m microscope, with a plan apochromatic 63× 1.4 NA objective, a HBO 100 W mercury light source, and a CoolSNAP-HQ camera (Roper Scientific), operated by the MetaMorph® software. Filters were from Chroma, Cy3 (41003), FITC (41001) and DAPI (31000). For RNA FISH analyses were used a z-stack of 20 z-sections with 0.2-μm step size and 500 msec exposure. For FISH images z-stacks were collapsed to a 2D maximum intensity projection and for immunofluorescence a single 2D image was selected from a 20 section z-stack. Images were processed by background subtraction and normalization using the open source software Image J (url: rsbweb.nih.gov/ij). To calculate the mRNA localization ratio FISH images were processed by subtracting the mean background value and the coordinates of the cell center was determined manually in the DAPI channel, and the coordinates of the apex of the cell protrusion was determined in the Cy3 channel. The intensities of each pixel and their respective coordinates together with the coordinates of the cell center and protrusion were imported into a custom made computer program. In this program, a line was drawn from the center of the cell to the apex of the protrusion, and each point was perpendicularly projected to this line. A pixel was categorized as localized if the position was more than 2/3 of the total length of the line toward the apex of the protrusion. A localization score was finally calculated by dividing the sum of the intensities of all localized pixels by the sum of the intensities of all pixels projected on the line.

Statistical Analysis

RT-qPCR amplifications were made in triplicates for each gene and Ct values were converted into linear values using the Xo method (Thomsen et al., 2010). All the relative expression levels were normalized to Arpc3 (for mRNA localization) or Gapdh (for mRNA expression). Differences in the localization ratios and expression levels were analyzed by a Student's unpaired two tailed t test. All experiments were repeated three times.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Identification of Localized mRNAs in Mouse Primary Astrocyte Protrusions by DRS

RNA localization has shown to play pivotal roles in cell signaling, morphology, and migration during both embryonic development, brain maintenance and in cancer metastasis (Mili and Macara, 2009). Whereas mRNA localization in neurons is extensively described, RNA localization in astrocytes is rather uncharacterized. To identify on a genome wide scale mRNA species localized in astrocyte protrusions we took advantage of the Boyden chamber cell fractionation method to separate cell protrusions from cell bodies (Thomsen and Lade Nielsen, 2011). Mouse primary astrocyte cultures were established from P0 mouse cortices and grown for 8 DIV. The mRNA expression pattern in the mouse primary astrocytes and two brain samples was by RT-qPCR examined for expression of cell type specific markers for microglia (Aif1), endothelial cells (Pecam1), oligodendrocytes (Mbp and Cnp), neurons (Nptx1), and astrocytes (Aldh1l1 and Gfap) (Imura et al., 2006; Stahlberg et al., 2011) (Supp. Info. Fig. 1). The expression analysis supported an in majority astrocyte lineage content of the primary astrocyte cultures and this was further substantiated by immunofluorescence staining showing that more than 80% of the cells were GFAP positive as also previously described (Imura et al., 2006; Pekny et al., 1998; Stahlberg et al., 2011; Thomsen and Lade Nielsen, 2011). It should be emphasized that the heterogeneity of the primary astrocyte cultures still could result in identification by the Boyden chamber approach of mRNA species with expression and RNA localization which cannot be confined to astrocytes. Compared to Stahlberg et al. (2011) presenting data for 10–12 DIV primary astrocytes we note relative more Nestin mRNA expression in our 8 DIV primary astrocyte cultures (Supp. Info. Fig. 1). In accordance, by RT-qPCR analysis of primary astrocyte cultures grown further DIV we observed a decrease in Nestin mRNA expression (Supp. Info. Fig. 2). We note a decrease in cell culture growth capability and lack of efficient plating in the Boyden chamber setting by using older DIV primary astrocyte cultures and selected eight DIV astrocytes for the subsequent analysis.

By RT-qPCR purification by the Boyden chamber assay of protrusion and cell body RNA from eight DIV mouse primary astrocytes was consolidated as described (Thomsen and Lade Nielsen, 2011). Absence of nuclear contamination of the PF was further controlled by the lack of DAPI staining and lack of histone proteins. Reproducible NGS analysis by single molecule DRS using minute RNA quantities were recently presented (Ozsolak et al., 2010). In the DRS procedure cDNA synthesis and amplification are evaded and NGS analysis can be performed directly on polyadenylated RNA (Ozsolak et al., 2009). RNA samples from primary mouse astrocyte PFs and CFs isolated by the Boyden chamber assay were analyzed by DRS. The resulting output of 36 bases average size sequences were mapped to a 2 kb region surrounding the distal 3′ end of 27131 genes of the mm9 mouse genome. The total number of mapped sequence counts in primary astrocytes was 5444770 for the CF and 2147050 counts for the PF. The individual sequence number for each transcript was normalized to transcripts per million reads (tpm). The tpm values represent the relative abundance of a given mRNA in the total population of mRNA molecules present in the CF or the PF. In all subsequent analyses we included transcripts with count numbers ≥5 tpm in both CF and PF to avoid insignificantly expressed mRNAs and false positives due to eventual stochastic fluctuations in the lower range sequence number. The total number of transcripts having ≥5 tpm in both CF and PF was 8894. The localization ratio was calculated as the ratio between the tpm values from the PF and CF, and 2298 mRNAs were determined to have a localization ratio >1 (Fig. 1A). Complete list is shown in Supporting Information Table 3A and the 250 mRNAs with highest localization ratios are shown in Supporting Information Table 4. The localization ratio for a given mRNA will be a variable depending on cell morphology and experimental settings, but the hierarchical order for localization ratios is envisaged to be relative independent of these variable factors. Moreover we note that a mRNA localization ratio > 1 is not describing that a majority of this mRNA is localized in protrusions but instead describes the relative enrichment in protrusions and can thereby represent localization of only a minor fraction of the total amount of a particular mRNA within the cell. DRS determined mRNA localization ratios for Rab13 mRNA (ratio 16.7), p0071/Pkp4 mRNA (ratio 15.7), and Kank2/Ankrd25 mRNA (ratio 10.9) are in accordance with previous mRNA localization observations (Mili et al., 2008; Thomsen and Lade Nielsen, 2011). These results demonstrate that the Boyden chamber assay combined with single molecule DRS are applicable to identify localized RNAs on a genome wide scale. Comparing the 250 most localized mRNAs with mRNA expression data from in vivo mouse P7 astrocytes (Cahoy et al., 2008) confirmed expression of the localized mRNAs in astrocytes in vivo (Data not shown). Annotation analysis revealed that the most localized mRNAs encode a broad variety of protein families (Supp. Info. Tables 4, 5 and Fig. 3). The most localized mRNAs encode proteins with a localization pattern strikingly similar to the total population of expressed mRNAs and in this line we notice that a significant fraction of protrusion localized mRNAs encode nuclear proteins.

image

Figure 1. Transcriptome analysis by DRS. (A) Graphical summary of the Boyden chamber assay for isolation of cell protrusions. ECM indicates coating of the membranes with Collagen type-1 prior to the seeding of cells to stimulate protrusion migration through the 1 μm pore size membrane. Protrusions were growing for 24 h before isolation. The protrusions and cell bodies were isolated in two separate fractions before subsequent analysis. (B,C) DRS transcriptome analysis of Boyden chamber purified RNA from mouse primary astrocyte protrusions (B) and mouse C8-S cell protrusions (C). Bar plots displaying the RNA localization ratios between the normalized numbers of DRS identified transcripts in protrusions over normalized number of transcripts in cell bodies in a hierarchical order. RNAs with ≥5 tpm in both protrusions and cell bodies are included. (D) Plot of the RNA localization ratio in primary astrocytes relative to the RNA localization ratio in C8-S cells. For visualization a red line indicates equal localization ratios, a dashed green line indicates the 26 relative most localized RNAs in C8-S cells and a dashed yellow line indicating the 28 relative most localized RNAs in primary astrocytes. (E) RT-qPCR analysis of the RNA localization ratio in C8-S cells compared with the RNA localization ratio in primary astrocytes. RNA localization ratios are normalized to Arpc3. P values were calculated by an unpaired student's t test. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

image

Figure 2. Nestin mRNA localization in cell protrusions from mouse primary astrocytes and C8-S cells. (A) Localization ratios determined by DRS for mRNA for IF proteins and controls. (B,C) RT-qPCR analysis of RNA localization ratios in primary astrocytes (B) and C8-S cells (C). cDNA was made from RNA from the protrusion and the cell body fractions. The ratios are normalized to Arpc3. (D) RT-qPCR analysis of RNA localization ratios in C8-S cells after growth in Boyden chamber assays for the indicated times. The RNA localization ratios are normalized to Arpc3 and subsequently to the RNA localization ratio after 24 h growth time given the value 100. (E) RT-qPCR analysis of RNA localization ratios in C8-S cells after growth in Boyden chamber assays with Laminin and Fibronectin membrane coating. (F) RT-qPCR analysis of RNA localization ratios in P19 cells without (−RA) and with (+RA) retinoic acid incubation for 24 h and the localization ratios normalized to Arpc3.

Download figure to PowerPoint

mRNA Localization Analysis in the Mouse C8-S Astrocyte Cell Line

We next included the mouse astrocyte like cell line C8-S that originally was isolated from the cerebellum of a post natal (P8) mouse and resembles type II astrocyte Bergmann glia cells (Alliot and Pessac, 1984). C8-S is a homogenous and non-cancerous cell line with a highly polarized morphology making it a qualified model for RNA localization studies. We note that the RNA localization analysis in mouse primary astrocytes represent an analysis of a heterogeneous cell population whereas C8-S RNA localization analysis would be confined to a more homogenous cell type of astrocyte lineage. To identify mRNAs localized in C8-S cells, we again separated protrusions from the cell bodies by the Boyden chamber assay and purified total RNA for DRS. Sequences were mapped to a 2 kb region of the 3′ distal part of 27131 genes from the mm9 mouse genome sequence. The total counts number for CF was 2269453 and for PF 2090020, and 7697 transcripts had read numbers ≥5 tpm in both PF and CF. The normalized reads were plotted as the ratio between PF and CF (Fig. 1C). The number of transcripts with a localization ratio > 1 was 1436. The 250 mRNAs with highest localization ratios are shown in Supporting Information. Table 6 and the complete list in Supporting Information Table 3B. Annotation analysis revealed that the most localized mRNAs encode a broad variety of protein families (Supp. Info. Table 7 and Fig. 3).

Comparative mRNA Localization Analysis of Mouse Primary Astrocytes and C8-S Cells

A total of 7119 mRNA species had transcript reads ≥5 tpm in PF and CF for both C8-S cells and mouse primary astrocytes. To examine whether we could identify cell type specific localized mRNAs, we plotted the localization ratio of primary astrocytes against C8-S cells (Fig. 1D). The analyses revealed that majority of the localized mRNAs in mouse primary astrocytes also localized in C8-S cells but also identified variability in localization patterns (Supp. Info. Tables 8 and 9). Moreover, we note that the annotation analysis of localized mRNAs also revealed differences for C8-S and primary astrocytes (Supp. Info. Fig. 3, Supp. Info. Tables 5 and 7).

To verify observed mRNA localizations in primary astrocytes and C8-S cells we conducted RT-qPCR analysis of representative candidates using independent biological RNA samples from Boyden chambers. We selected Tensin3 (Ten3) mRNA as it is highly expressed in both primary astrocytes and C8-S cells but exhibited a larger RNA localization ratio in primary astrocytes than C8-S cells, sperm flagellar 1 (Spef1) mRNA which is highly expressed in both primary astrocytes and C8-S cells but with a higher RNA localization ratio in C8-S cells than primary astrocytes, cytochrome c oxidase subunit IV isoform 1 (Cox4i1) mRNA which is highly expressed but displays no RNA localization in primary astrocytes and C8-S cells, and Rab13 and p0071/Pkp4 mRNAs which are localized in both cell types. The RNA localization ratios were normalized to Arpc3, a component of the Arp2/3 complex (Mili et al., 2008). Arpc3 mRNA showed localization in neither the DRS analysis and nor in RT-qPCR analyses using NIH/3T3 cells, primary astrocytes and C8-S cells (Feltrin et al., 2012; Mili et al., 2008; Thomsen and Lade Nielsen, 2011). RT-qPCR analysis revealed a significant localization of Tensin3 mRNA in primary astrocytes compared to C8-S cells and a significant localization of Spef1 mRNA in C8-S cells compared to primary astrocytes (Fig. 1E). Cox4i1 mRNA had no RNA localization in primary astrocytes and C8-S cells, whereas Rab13 and p0071 mRNAs have RNA localization in both cell types (Fig. 1E). The results of the RT-qPCR analysis were similar to the DRS analysis, supporting that the Boyden chamber method combined with a DRS analysis also is applicable for identification of specific RNA localization patterns between different cell types.

Next we searched for candidate mRNAs having protrusion localization and in addition being enriched for astrocyte expression. A comprehensive identification of mRNAs relatively enriched in in vivo mouse astrocytes, mature astrocytes, and developing astrocytes compared to other brain cell types was described by Cahoy et al. (Cahoy et al., 2008). The study also included identification of mRNAs enriched in in vitro grown astroglia cells compared to in vivo astrocytes (Cahoy et al., 2008). Of the 250 most localized mRNAs in mouse primary astrocytes 28 were also identified to be enriched in at least one of these four defined astrocyte populations (Supp. Info. Table 10). Twenty two of the mRNAs have enriched expression in astrocytes in vivo and 15 have enriched expression in astroglia cells grown in vitro (Supp. Info. Table 10). We note that all identified mRNAs to be both localized in astrocyte protrusions and expression enriched in developing astrocytes in vivo also are relatively enriched in primary astroglia cells in vitro [Supp. Info. Table 10 and (Cahoy et al., 2008)]. Moreover, the comparative analysis pointed that albeit some mRNAs with localization in astrocyte protrusions have enriched expression in astrocytes only few were assigned to be astrocyte specific (Dio2, Ppp1r3c and Gfap). Recently, Feltrin et al. described 80 mRNAs localized in mouse N1E-115 neuroblastoma cells (Feltrin et al., 2012). Of the 25 most protrusion localized mRNAs in mouse primary astrocytes and C8-S cells 13 mRNAs were also identified to be significantly localized in N1E-115 cells and of the 80 most protrusion localized mRNAs in mouse primary astrocytes and C8-S cells, 28 and 26, respectively, were also significantly localized in N1E-115 cells (Supp. Info. Tables 11 and 12) (Feltrin et al., 2012). Such mRNAs could represent a group of commonly expressed and protrusion localized mRNAs and several of these were also identified to be localized in mouse fibroblast protrusions (Mili et al., 2008). Of the 80 N1E-115 localized mRNAs 4 were overlapping with the astrocyte enriched and localized mRNAs (Cyb5r3, Ddr2, Arhgap11a, and Kctd10) (Supp. Info. Table 10).

mRNA Localization for the IF Protein Nestin

We observed that the mRNA for the IF protein Nestin had a high localization ratio in both mouse primary astrocytes and in C8-S cells and a preferential expression in developing astrocytes (and other neural cell type progenitors in vivo) (Fig. 2A, Supp. Info. Tables 4, 6, 10) (Cahoy et al., 2008). One of our long term research aims is to identify mechanisms for IF regulation and we accordingly focused the subsequent studies on Nestin mRNA. In primary astrocytes Nestin mRNA had a localization ratio of 42 and was identified as the relatively most protrusion localized mRNA. In C8-S cells the localization ratio was 19. The RNA localization ratios for the two Nestin related IF proteins in astrocytes, Gfap and Vimentin, were 5 and 0.8 in primary astrocytes and 1 and 0.7 in C8-S cells, respectively. The RNA localization outcome of the DRS analysis was confirmed by RT-qPCR analyses using independent biological samples including primers against Nestin, Vimentin, and Gfap cDNA. As references we included analyses of Actb and Arpc3 cDNA. The RNA localization ratios were normalized to Arpc3 given the value 1 (Fig. 2B). The RT-qPCR results from mouse primary astrocytes showed localization of Nestin mRNA in the cell protrusions with a localization ratio of 130. Likewise, we noted that the Gfap mRNA was localized in the protrusions with a 13 times higher ratio than Arpc3. The Vimentin mRNA had no localization, with a ratio of 0.6, which was concurrent to the ratios of Arpc3 and Actb mRNAs. In a similar RT-qPCR experimental setting using RNA material from C8-S cells Nestin mRNA showed a significant localization in cell protrusions, whereas neither Vimentin nor Gfap mRNAs displayed localization (Fig. 2C). We note that the Gfap mRNA expression in C8-S cells was approximately 500-fold lower than for primary astrocytes. In a time course experiment for C8-S cell protrusion growth Nestin RNA localization was at maximum after 24 h (Fig. 2D). Nestin RNA localization was also observed in C8-S cells after coating membranes in the Boyden chamber with the extracellular matrix proteins Laminin or Fibronectin (Fig. 2E), whereas absence of coating hindered quantitative experiments because only few cell protrusions penetrated the membrane.

To examine if the observed Nestin mRNA localization was restricted to cells of astrocyte lineage we next examine mouse embryo-derived teratocarcinoma P19 cell line in the Boyden chamber setting. P19 cells are pluripotent and can differentiate into cell types of all three germ layers and in response to retinoic acid treatment P19 cells first become neural stem-like cells with Nestin expression and finally differentiate to neural cells (neuron, glia, etc.) (Jones-Villeneuve et al., 1982; Tan et al., 2010). After the Boyden chamber assay P19 cells without incubation with retinoic acid had a low level of Nestin mRNA expression but RT-PCR analyses were able to determine Nestin mRNA localization to the P19 cell protrusions (Fig. 2F). After retinoic acid incubation for 24h Nestin mRNA expression was increased ∼25-fold (data not shown) and again Nestin mRNA localization to cell protrusions was determined (Fig. 2F). We were unable to detect Gfap mRNA expression to a significant level in the P19 cell samples (data not shown). We conclude that the Nestin mRNA present in mouse P19 pluripotent embryonic carcinoma cells also have protrusion localization capability.

Cis-elements determining mRNA localization are typically enclosed in the 3'-UTR (Mili and Macara, 2009). To determine if the 3′-UTR was involved in mediating localization of Nestin mRNA to cell protrusions we cloned the 419 bp 3′-UTR sequence including the intrinsic poly-adenylation signal into a modified pcDNA5-beta-globin-6xMS2-SV40-LpA vector downstream of the spliced β-globin transcription unit. We also cloned the Rab13 3′-UTR for positive localization control and the Vimentin 3′-UTR for negative control together with the vector without an UTR insert. We were unable to transfect mouse primary astrocytes and C8-S sufficiently for quantitative ectopic RNA localization measurements and instead used mouse NIH/3T3 cells which previously were used as model in mRNA localization studies (Mili and Macara, 2009). After cell transfection and Boyden assay purified RNA was examined by RT-qPCR for expression of the ectopic mRNAs by a forward primer recognizing β-globin exon 1 included in the chimeric mRNAs and a reverse primer corresponding to the inserted UTR fragments or a specific primer for the vector without an UTR insert. cDNA from the PF and the CF from cells transfected with the control vector, pTAG, resulted in no significant RT-qPCR amplification. Insertion of the Nestin 3′-UTR resulted in an approximately fourfold increase in localization ratio compared to the Vimentin 3′-UTR and vector control (Fig. 3A). The Rab13 3′-UTR resulted in a further fourfold increase in localization ratio (Fig. 3A). From the transfection experiments we conclude that the Nestin 3′-UTR includes sequence determinants which alone can mediate RNA localization.

image

Figure 3. Cis- and trans-factors involved in Nestin mRNA localization. (A) The Nestin 3′-UTR include cis-sequences for mRNA localization to protrusions. The 3′-UTR sequences of Nestin, Vimentin and Rab13 were transfected as a pool into mouse NIH/3T3 cells together with the expression vector without insert. RNA fractions from protrusions and cell bodies were purified by the Boyden chamber assay. By RT-qPCR the RNA localization ratios were determined and normalized to the localization ratio for the vector without insert (control) given the value 1. (B) Involvement of the FMRP protein family for Nestin mRNA localization. In C8-S cells Fmr1, Fxr1, and Fxr2 were either depleted together using a pool of siRNA (sifpool, left panels) or by individually siRNA (right panels). At 24 h after transfection C8-S cells were used in Boyden chamber assay and Nestin RNA localization ratio determined and normalized to 100 for the control siRNA transfection. The efficiency of siRNA treatments were measured by RT-qPCR analysis of Fmr1, Fxr1, and Fxr2 and normalized to 1 for the control siRNA transfection.

Download figure to PowerPoint

We noted the presence of a high Guanine (G) content and algorithm predicted putative G-quadruplex motifs in the Nestin 3′-UTR (http://bioinformatics.ramapo.edu/QGRS). The fragile X mental retardation protein family composed of Fmrp, Fxr1, and Fxr2 can through the RGG-box associate with G-quadruplex motifs and mediate transport to neuronal dendrites (Bagni and Greenough, 2005; Darnell et al., 2001; Mili et al., 2008; Schaeffer et al., 2001). To determine if these factors are involved in Nestin mRNA localization we depleted C8-S cells of these factors individually or in combinations by transient siRNA treatments and by Boyden chamber following measured Nestin mRNA localization. Semi-quantitative measurements showed that the Fxr2 mRNA expression level in C8-S was several fold lower than Fxr1 (data not shown). Fmr1 mRNA, which encodes the Fmrp protein, was also lower expressed than Fxr1 (data not shown). Fmr1, Fxr1 and Fxr2 siRNA mediated co-depletion resulted in an approximately twofold reduction in the Nestin RNA localization ratio (Fig. 3B). For the individual siRNAs we observed that depletion of Fmr1 and Fxr1 resulted in an approximately twofold reduction in the Nestin RNA localization ratio supporting that the proteins, directly or indirectly, are involved in Nestin mRNA localization (Fig. 3B).

FISH Analysis of Nestin mRNA

To substantiate the Nestin mRNA localization studies we examined for Nestin mRNA localization in mouse primary astrocytes by FISH. For optimal cytoplasmic RNA detection we took advantage of the single RNA molecule FISH technique (Femino et al., 1998). We designed probes complementary to mRNAs of Nestin and Gfap. For the FISH assay we used 50-mer DNA oligonucleotide probes each labeled with five covalently coupled Cy3 fluorophores and single mRNA molecules were detected by hybridizing a pool of eight different probes targeting each mRNA species. For control we included probes against Actb mRNA. Primary astrocytes (8 DIV) plated on Collagen ECM protein coated coverslips were used for analysis and the results of the FISH assays were monitored by blinded cell counting. Approximately 50% of the primary astrocytes were scored positive for Nestin mRNA expression due to the number of Nestin mRNA FISH signals. We note that Nestin mRNA FISH signals also could be detected in cells initially scored negative for Nestin mRNA expression by the blinded cell counting indicating that the actual number of Nestin expressing cells is higher than 50%. In approximately half of the positively scored cells numerous Nestin mRNA FISH signals could be detected in cell protrusions in accordance with Nestin mRNA protrusion localization (Fig. 4A). FISH analysis showed that Gfap mRNA was more uniformly distributed in the cytoplasm compared to Nestin mRNA. However, Gfap mRNA could be observed in the outermost regions of the cytoplasm in a majority of the Gfap positive cells, whereas Actb mRNA predominantly was confined to the peri-nuclear part of the cells (Fig. 4A).

image

Figure 4. Subcellular detection of Nestin mRNA by single molecule RNA FISH. (A) Single molecule RNA FISH analysis of endogenous Nestin, Actb and Gfap mRNAs in cultured primary astrocytes isolated from P0 mice (left panels). A region of interest (ROI) is highlighted and magnified by a 5× zoom factor (middle panels). Cell nuclei were stained by DAPI (right panels). Scale bar = 10 μm. (B) Single molecule RNA FISH of Nestin and Actb mRNAs in C8-S cells (left panels). SSA4 served as negative control. Middle panels show a ROI representing the tip of a cell protrusion enlarged five times (middle panels). Cell nuclei were stained by DAPI (right panels). Scale bar = 20 μm. (C) Heat plot displaying C8-S Nestin and Actb mRNA localization analysis output (right panels) and their respective raw images (left panels). Cell center is marked by a red spot en the length of the cell protrusion is marked by a green line. Examples of cells with Actb mRNA signals (upper two panels) and Nestin mRNA signals (lower two panels) are shown. (D) The outcome of a paired two tailed t test for Nestin and Actb mRNA localization according to the procedure in (C). The cell counting numbers were for Actb n = 9 and for Nestin n = 10. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

We also examined subcellular RNA localization in C8-S cells by RNA FISH, using probes against Nestin, Actb and mRNA for the S. cereviseae heat shock protein SSA4 as negative control. Gfap mRNA expression is very low in C8-S cells and accordingly not included in the analysis. C8-S cells were plated on collagen ECM protein coated coverslips and the results of the FISH assay were analyzed by blinded cell counting. Most C8-S cells (>99%) were scored positive due to the number of Nestin mRNA FISH signals. In approximately half of the positively scored cells numerous Nestin mRNA FISH signals could be detected in cell protrusions in accordance with protrusion localization of Nestin mRNA (Fig. 4B, upper panels). Actb mRNA was predominantly confined to the perinuclear region of the cell body with a low amount of mRNA also observed in cell protrusions (Fig. 4B, middle panels). The negative control probe showed no signal (Fig. 4B, lower panels). To determine if there is an effect of ECM components on Nestin mRNA localization we performed a similar RNA FISH analysis using C8-S cells plated on ECM uncoated coverslips (Supp. Info. Fig. 4). We observed no changes in Nestin and Actb mRNA localization patterns (Supp. Info. Fig. 4). To further examine the difference of mRNA localization between Nestin and Actb in C8-S cells, we randomly selected 10 cells with Nestin mRNA signals and 9 cells with Actb signals (Fig. 4C). Signal intensities were summarized and signals that displayed a relative distance more than two thirds from the center of the nucleus were regarded as localized. Localized signal intensities were divided by the total signal intensities within the cell to obtain a localization ratio. The ratios were statistically analyzed by a student's t test, which showed that the localization of Nestin mRNA was significantly higher than Actb (Fig. 4D). In conclusion, the observations from the FISH assays were in concordance with the biochemical analysis showing localization of a fraction of the Nestin mRNA in cell protrusions.

Nestin Protein Localization in Astrocyte Protrusions

We examined whether the Nestin protein display localization analogous to the mRNA. In this line we note that it was previously shown that mouse Nestin protein localizes in growth cones of P19 derived neurons and cerebellar granule cells (Yan et al., 2001). We isolated protein extracts from the PF and CF of mouse primary astrocytes using the Boyden chamber assay and subsequently performed Western blot analysis. To normalize the protein content between the PF and the CF we included analysis of α-Tubulin and Actb representing controls for uniform protein distribution. Histone H3 was included to control that cell nuclei were confined to the CF. Western blot analysis showed that Nestin and Vimentin were relatively more present in the PF whereas Gfap was equally detected in the CF and the PF (Fig. 5A). Western blot analysis also showed that in the C8-S cell PF Nestin and Vimentin proteins were relatively enriched (Fig. 5B). Notably, the absence in the presented Western blot analysis of detectable Vimentin and Nestin in the CF fraction shall not be interpreted as absence of the proteins in this fraction but solely reflects a relative increased localization in protrusions compared to α-tubulin and Actb.

image

Figure 5. Nestin protein localization analyses. (A) Western-blot analysis of protein isolated from the cell body fraction (CF, Boyden chamber upper side) and cell protrusion fraction (PF, Boyden chamber lower side) using mouse primary astrocytes. Proteins are detected by Western blotting using primary antibodies against Nestin, Vimentin, Gfap, α-Tubulin, Actb, and Histone H3. α-Tubulin and Actb served as load controls and Histone H3 to control for the lack of cell body migration through the Boyden chamber membrane. (B) Western blot analysis of protein isolated from CF and PF by the Boyden chamber assay using C8-S cells. (C,D) Double immunofluorescence analysis of Vimentin and Gfap (C) and Nestin and Gfap (D) in mouse primary astrocytes. Proteins were detected using primary antibodies against Nestin, Vimentin and Gfap. Nuclei were stained with DAPI and shown in merged pictures (Gfap, green; Nestin, red; Vimentin, red; DAPI, Blue) of representative cells. Scale bar = 20 μm. (E) Immunofluorescence analysis of C8-S cells showing subcellular protein localization of Nestin, Vimentin, Actb and α-Tubulin. Cell nuclei were stained with DAPI (blue). Scale bar = 20 μm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

To examine subcellular protein distributions in closer details we performed immunofluorescence analysis. Nestin and Vimentin were detected using mouse primary antibodies and an Alexa 555 conjugated secondary antibody and Gfap was detected using a rabbit primary antibody and an Alexa 488 conjugated secondary antibody. In primary mouse astrocytes in the order of 80% of the cells were scored positive for Gfap, Vimentin and Nestin but we note very heterogeneous expression levels. Most Gfap positive cells showed filamentous Gfap distribution throughout the cytoplasm (Fig. 5C,D). The largest fraction of the Vimentin positive cells has filamentous staining throughout the cytoplasm (Fig. 5C, upper panels) but some cells (∼30%) have Vimentin accumulation in cell protrusions with simultaneous Vimentin filaments mostly in the central part of the cell (Fig. 5C, lower panels). In Vimentin and Gfap positive cells we observed co-localization including colocalization in protrusions (Fig. 5C). Most Nestin positive cells showed filamentous Nestin distribution throughout the cytoplasm including the cell protrusions (Fig. 5D, upper panels). In Nestin and Gfap positive cells we observed co-localization including colocalization in protrusions. Some Nestin positive cells (∼20%) showed more pronounced Nestin accumulation in protrusions with Gfap colocalization and simultaneous Nestin in the central part of the cell (Fig. 5D, lower panels). In C8-S cells Gfap expression could not be detected whereas Vimentin and Nestin were detected in nearly all cells (Fig. 5E). In ∼50% of the C8-S cells Nestin accumulation was present throughout the protrusions (Fig. 5E). In nearly all C8-S cells Vimentin accumulation was present throughout protrusions (Fig. 5E). The control proteins α-Tubulin and Actb were uniformly dispersed throughout the cytoplasm (Fig. 5E). Thus, both in primary astrocytes and C8-S cells a fraction of the Nestin protein, as well as the Nestin mRNA, was present in protrusions.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

The number of mRNAs that exhibit a distinct subcellular localization pattern has increased dramatically as genome wide technologies are improved (Lecuyer et al., 2007). More than 1,000 mRNAs are identified to be relatively localized in neuronal protrusions (Eberwine et al., 2001; Feltrin et al., 2012; Taylor et al., 2009; Willis and Twiss, 2010; Willis et al., 2007). Our presented DRS-based results strongly indicate that numerous polyadenylated RNAs also are localized in protrusions of primary astrocytes and the astrocyte cell line C8-S. The heterogeneity of the used primary astrocytes could in principle result in identification of mRNA localization which cannot be confined to astrocytes but may represent expression or localization in a minor cell subpopulation of different lineage within the cell culture. We note that non-Gfap positive cells within primary astrocyte cultures most probably are primarily meningeal cells of fibroblast lineage (Imura et al., 2006). However, comparing the 250 most localized mRNAs from primary astrocytes with expression data for in vivo astrocytes (Cahoy et al., 2008) we could verify astrocyte expression of the identified localized mRNAs. Annotation analysis revealed that the 250 most localized mRNAs in primary astrocytes and the C8-S astrocyte cell line encode a broad variety of protein families. Moreover, both the 50 and the 250 most localized mRNAs encode proteins which exhibit a localization pattern similar to the total population of expressed mRNAs (Supp. Info. Fig. 3). In this line, a significant fraction of the protrusion localized mRNAs encode nuclear proteins, which indicates that RNA localization to protrusions might not necessarily be predictive for a direct function of the encoded protein in cellular protrusions or that such protein has moonlighting capacity. A previous study showed that a majority (∼70%) of all expressed RNAs are localized in Drosophila embryos (Lecuyer et al., 2007) and points toward the notion that RNA localization is a norm rather than the exception. We determined a higher number of localized mRNAs in primary astrocytes than in C8-S cells, a result that could be dependent on the different morphologies of the cells. To further verify the determined RNA localization patterns we conducted RT-qPCR analysis of representative mRNAs. The RT-qPCR analyses verified that Tensin3 mRNA displays a larger RNA localization ratio in primary astrocytes than C8-S cells, and Spef1 mRNA shows a higher RNA localization ratio in C8-S cells than primary astrocytes (Fig. 1D). The result of the RT-qPCR analysis was similar to the DRS analysis, further supporting that the Boyden chamber assay combined with a DRS analysis is applicable to compare and identify both commonly and cell type specific localized RNAs. The Spef1 protein was previously determined to have a developmental dependent subcellular localization in mouse sperm cells (Chan et al., 2005). Tensin proteins act as mediators between the extracellular matrix and the cytoskeleton, and studies of human cancer cell lines showed that Tensin3 act as a negative regulator of cell migration (Martuszewska et al., 2009). The results suggest that the two examined astrocyte cell types have different intrinsic capacity for Spef1 and Tensin3 mRNA localization which could be mediated through nonidentical expression patterns of mRNA localizing trans-factors or inclusion of different yet unidentified cis-sequences in the transcripts through alternative RNA processing.

Directional transport by cytoskeletal motors is the predominant mechanism for delivering mRNA to the destination (Bullock, 2011). mRNA and associated protein trans-factors are cotransported as messenger ribonucleoprotein (mRNP) particles (Mili and Macara, 2009). The mRNP transport is facilitated by myosin motor proteins on actin filaments or via kinesin and dynein motor proteins on microtubules (Shav-Tal and Singer, 2005). Different motors can be active in sorting mRNP in the same cell at the same time leading to differential patterns of mRNA localization (Bullock, 2011). Assembly of the cytoskeletal filaments used for mRNP transport can be self-regulated through mRNA localization dependent mechanisms which are well established for actin filaments (Shav-Tal and Singer, 2005). Actb mRNA is localized to the leading edge of fibroblasts in an actin filament dependent manner and disruption of Actb mRNA localization results in slower cell motility, loss of directionality, delocalization of actin polymerization and altered adhesion dynamics (Katz et al., 2012; Mingle et al., 2005; Shestakova et al., 2001). The actin-related protein 2/3 (Arp2/3) complex is a crucial actin polymerization nucleator and is localized to the leading protrusions of migrating fibroblasts. mRNAs for the seven subunits of the Arp2/3 complex (Arpc1a, Arpc2, Arpc3, Arpc4, Arpc5, Actr2, and Actr3) are localized to fibroblast protrusions in both actin filaments and microtubules dependent manners supporting that the Arp2/3 complex is targeted to the site of function by mRNA localization (Mingle et al., 2005). In the Boyden chamber approach we neither observed mRNA localization for Actb mRNA nor the seven mRNAs for the Arp2/3 complex (Supp. Info. Table 3). This observation is in accordance with other reports describing the lack of mRNA localization for these actin filament components and may reflect the use of different experimental settings to detect mRNA localization (Feltrin et al., 2012; Mili et al., 2008).

Only few studies have addressed mRNA localization in astrocytes in relation to IFs. In this study we have identified mRNA localization of Nestin. Compared to control Arpc3 mRNA, Nestin mRNA exhibited a significant localization in astrocyte protrusions as well as in protrusions of P19 embryo-derived teratocarcinoma cells. Gfap mRNA displayed a more moderate localization. Vimentin mRNA exhibited no significant localization, neither by RT-qPCR nor DRS analysis. For detailed subcellular analysis of Nestin mRNA we performed single RNA molecule FISH. This revealed that Nestin mRNA was localized to protrusions in ∼50% of the primary astrocytes and C8-S cells scored positive for Nestin mRNA expression. Moreover, when we applied a computer program to analyze FISH images of Nestin mRNA positive cells and compared these with Actb mRNA positive cells, we found that Nestin mRNA had significantly more localization to cell protrusions than Actb mRNA. The Nestin mRNA localization results sustain previous results obtained from the developing mouse brain indicating that Nestin mRNA is localized in columnar neuroepithelial cells and radial glial cells (Dahlstrand et al., 1995), and studies made on tissue sections of chicken brain and cell cultures demonstrating mRNA localization for Transitin, a Nestin like IF protein, in developing radial glia cells from chickens (Lee and Cole, 2000). To examine if Nestin mRNA localization also could be detected in vivo we analyzed mouse P4 brain sagittal section mRNA in situ hybridization (ISH) data for Nestin, Gfap, Actb, CamKIIα, and Tubb3 (extracted from Allen Developing Mouse Brain Atlas) (Supp. Info. Figs. 5 and 6). Nestin ISH signals were sparsely detected and based on the data we were not able to conclusively determine if Nestin mRNA in vivo is localized in astrocyte protrusions.

The presented Nestin mRNA localization observations by DRS, RT-qPCR and FISH in astrocyte cells are concordant with the results showing presence of Nestin protein in the cell protrusions (Fig. 5). Together the data supports that Nestin mRNA can be localized and most likely also locally translated in astrocyte protrusions. Although local proteins synthesis was not shown directly, our results suggest that Nestin protein is subcellular localized, at least partly, as a consequence of local mRNA translation. Changes of cell morphology depend on coordinated assembly and disassembly of cytoskeleton proteins. Notably, it has been demonstrated that Nestin can inhibit Vimentin filament formation in a concentration dependent manner (Steinert et al., 1999). Thus, localization and local translation of the Nestin mRNA can be used by the cell to create the necessary local environment to provide optimal conditions for IF modulation in the early onset of protrusion formation. The notion that local assembly of Nestin and Vimentin could participate in modulation of astrocyte morphology is further indicated by the results showing significant Vimentin protein localization in protrusions. Interestingly, we showed no significant Vimentin mRNA localization and Vimentin is accordingly most likely localized by protein transport mechanisms. This is supported by numerous studies demonstrating mictrotubule and dynein dependent transport of non-filamentous IF protein particles (Perlson et al., 2005; Prahlad et al., 1998; Yoon et al., 1998). Moreover, studies of migrating endothelial cells have shown Vimentin localization to focal adhesions sites (Tsuruta and Jones, 2003). Gfap expression is graduate up-regulated as Nestin and Vimentin expression decreases and it is proposed that Vimentin and Nestin filaments are scaffolds for the establishment of long term Gfap filaments (Dahlstrand et al., 1995). Our data indicates that different localization of Vimentin, Gfap, and Nestin mRNAs and the resulting proteins altogether participates in the control of IF dynamics in cell protrusions.

Several studies have demonstrated that mRNA localization depends on cis-elements that typically, but not exclusively, are enclosed in the 3′-UTR and associated RNA binding trans-factors (Mili and Macara, 2009). In a reporter assay we determined that the Nestin mRNA 3′-UTR was sufficient to mediate localization to cell protrusions whereas the Vimentin mRNA 3′-UTR lacked this functionality. Fmrp and the autosomal paralogues, Fxr1 and Fxr2, compose a family of functional homologous RNA-binding proteins including two ribonucleoprotein K homology domains and a cluster of arginine and glycine residues in the RGG box (Bassell and Warren, 2008; Tan et al., 2009). These domains are important for RNA binding and polyribosome association. The FMRP-family has an important role in translation control, both in vivo and in vitro, and FMRP regulates protein synthesis at sites where mRNAs are locally translated (Kindler and Kreienkamp, 2012). The FMRP-family shuttles between the nucleus and the cytoplasm and after mRNA association in the nucleus forms a ribonucleoprotein complex transported to dendrites and spines (Kim et al., 2009; Kindler and Kreienkamp, 2012). An mRNA target sequence for the FMRP-family consists of a G-quadruplex motif recognized by the RGG box (Melko and Bardoni, 2010). Isolation of FMRP containing ribonucleoprotein complexes from mouse brains identified 432 mRNA species whereof 70% included putative G-quadruplex motifs (Brown et al., 2001). Moreover, 30% of an examined group of mRNA localized in neuronal dendrites, such as Psd95 and CamkIIα, contain G-quadruplex motifs in the 3′-UTR (Subramanian et al., 2011). We showed that Fmrp and Fxr1, either directly or indirectly, are involved in Nestin mRNA localization. The Nestin mRNA contains several putative G-quadruplex motifs in the 3′-UTR which could indicate that these elements in association with Fmrp and Fxr1 participates in Nestin mRNA localization. In genome wide identification analysis of mRNAs associating with Fmrp the Nestin mRNA does not appear as a highly significant Fmrp target (Ascano et al., 2012; Brown et al., 2001; Darnell et al., 2011). It should be emphasized that Nestin mRNA expression is relatively low in the examined cell lines and mouse brain samples which could be hindering identification of Fmrp and Nestin mRNA interactions. In this line Darnell et al. found interactions between Fmrp and Nestin mRNA but the interaction was not scored significantly positive (Darnell et al., 2011).

In summary, we identified that IFs potentially can be regulated at the level of mRNA localization mechanisms. We demonstrated that Nestin mRNA and Nestin protein can have localization in cell protrusions proposing that at least some Nestin protein can be localized as a consequence of local translation of the Nestin mRNA. The Vimentin mRNA displayed no significant protrusion localization whereas the Vimentin protein is present. Finally, Gfap mRNA is moderately localized in protrusions whereas the Gfap protein is distributed relatively evenly in the cytoplasm. The identified mRNA localization patterns might reflect that IF proteins use different sets of localization mechanisms with potential to regulate astrocyte morphology and migration during brain development and maintenance.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

The Lundbeck Foundation; Fonden til Lægevidenskabens Fremme; NANONET—COST Action BM1002; The Health Faculty, Aarhus University, Denmark.

The authors thank Robert H. Singer, Albert Einstein College of Medicine, Bronx, New York, USA, for introducing them to the FISH technique and the donation of FISH control probes. The grant funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. The authors declare no conflicts of interests.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
glia22569-sup-0001-suppinfoFig.1.eps1357KSupp. Info. FIGURE 1: RT-PCR gene expression profiles of cell cultures and brain samples. Markers for microglia (Aif1/Iba1), endothelial cells (Pecam1), oligodendrocytes (Mbp and Cnp), neurons (Nptx1) and astrocytes (Aldh1l1 and Gfap) were analyzed. The markers were selected as representatives from the comprehensive analysis presented by Stahlberg et al. Additionally was included Spef1, Ten3, and Cox4i representing mRNA identified in the RNA localization analysis in this report and the control mRNA Arpc3. All expression values were normalized to Gapdh expression given the value 1 in PA. Values represent three independent RT-PCR analyses. N1E (N1E-115), mouse neuroblastoma cell line; P19, mouse embryonic teratocarcinoma cell line; C8-S, mouse cerebellum Golgi-Bergmann astrocyte cell line; PA, mouse primary astrocytes (DIV 8); P0, mouse P0 brain; P21m, mouse 21 month old brain.
glia22569-sup-0002-suppinfoFig.2.eps976KSupp. Info. FIGURE 2: RT-PCR gene expression profile of Nestin mRNA in PA cell cultures. Mouse primary astrocytes were grown for different number of days in vitro (DIV) and purified RNA analyzed for expression of Nestin, Gfap, and Arpc3. All expression values were normalized to Gapdh expression given the value 1 in PA. Values represent three independent RT-PCR analyses.
glia22569-sup-0003-suppinfoFig.3.eps2169KSupp. Info. FIGURE 3: (A) Family annotation analysis of the 250 RNAs with highest localization ratio in mouse primary astrocytes. (B) Annotation analysis of the subcellular protein localization for the 50 and 250 most localized RNAs compared to the total group of RNAs that are expressed with count number ≥ 5 t.p.m. in both cell bodies and cell processes. RNAs are grouped in accordance with the subcellular localization of the protein encoded. Annotations were made using the IPA Ingenuity platform (www.ingenuity.com). (C) Family annotation analysis of the 250 RNAs with highest localization ratio in mouse C8-S cells. (D) Annotation analysis of the relative subcellular protein localization encoded by the 50 and 250 most localized RNAs encode compared to the total group of RNAs that are expressed with count numbers ≥ 5 t.p.m. in both cell bodies and cell processes. Annotations were made using the IPA Ingenuity platform (www.ingenuity.com). (E) Functional annotation analysis displaying examples of the gene ontology terms annotated for the 250 most significantly localized RNAs in cell processes of primary astrocytes and C8-S cells. The p-value indicated was corrected for multiple testing by the Benjamini Hochberg method. Annotations were made using the IPA Ingenuity platform (www.ingenuity.com).
glia22569-sup-0004-suppinfoFig.4.tif5672KSupp. Info. FIGURE 4: Subcellular detection of Nestin mRNA in mouse primary astrocytes grown on uncoated coverslips by single molecule RNA FISH. Single molecule RNA FISH analysis is performed on endogenous Nestin and Actb mRNA in cultured primary astrocytes (8 DIV) isolated from P0 mice by fluorescence microscopy. Images are maximum projection of a 20 frames z-stack of 0.2 μm step size, showing the Cy3 signal of single RNA molecules. Cell nuclei were stained by DAPI and the merged pictures shown. Scale bar = 20 μm.
glia22569-sup-0005-suppinfoFig.5.tif9082KSupp. Info. FIGURE 5: mRNA ISH in mouse P4 sagittal brain sections for hippocampus region. mRNA ISH data from mouse P4 brain was extracted from Allen Brain Atlas. ISH data for Nestin, Actb (control), Gfap (astrocyte marker), and CamKIIa and Tubb3 (neuronal markers) are shown. Tissue sections were DNA counter stained with Feulgen-HP yellow to visualize nuclei. Scale bar = 100 μm.
glia22569-sup-0006-suppinfoFig.6.tif8351KSupp. Info. FIGURE 6: mRNA ISH in mouse P4 sagittal brain sections for cortex. mRNA ISH data from mouse P4 brain was extracted from Allen Brain Atlas. ISH data for Nestin, Actb (control), Gfap (astrocyte marker), and CamKIIa and Tubb3 (neuronal markers) are shown. Tissue sections were DNA counter stained with Feulgen-HP yellow to visualize nuclei. Scale bar = 100 μm.
glia22569-sup-0007-suppinfo1.docx16KSupporting Information Table 1
glia22569-sup-0008-suppinfo2.docx14KSupporting Information Table 2
glia22569-sup-0009-suppinfo3.docx1452KSupporting Information Table 3
glia22569-sup-0010-suppinfo4.docx85KSupporting Information Table 4
glia22569-sup-0011-suppinfo5.docx18KSupporting Information Table 5
glia22569-sup-0012-suppinfo6.docx76KSupporting Information Table 6
glia22569-sup-0013-suppinfo7.docx17KSupporting Information Table 7
glia22569-sup-0014-suppinfo8.docx30KSupporting Information Table 8
glia22569-sup-0015-suppinfo9.docx23KSupporting Information Table 9
glia22569-sup-0016-suppinfo10.docx33KSupporting Information Table 10
glia22569-sup-0017-suppinfo11.docx53KSupporting Information Table 11
glia22569-sup-0018-suppinfo12.docx32KSupporting Information Table 12

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.