Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Traffic. Author manuscript; available in PMC 2011 Jul 1.
Published in final edited form as:
PMCID: PMC2906249

Differential subcellular distributions and trafficking functions of hnRNP A2/B1 spliceoforms


Trafficking of mRNA molecules from the nucleus to distal processes in neural cells is mediated by hnRNP A2/B1 trans-acting factors. Although hnRNP A2/B1 is alternatively spliced to generate 4 isoforms, most functional studies have not distinguished among these isoforms. Here we show, using isoform-specific antibodies and isoform-specific GFP-fusion expression constructs, that A2b is the predominant cytoplasmic isoform in neural cells, suggesting that it may play a key role in mRNA trafficking. The differential subcellular distribution patterns of the individual isoforms are determined by the presence or absence of alternative exons that also affect their dynamic behavior in different cellular compartments, as measured by fluorescence correlation spectroscopy. Expression of A2b is also differentially regulated with age, species and cellular development. Furthermore, co-injection of isoform-specific antibodies and labeled RNA into live oligodendrocytes shows that assembly of RNA granules is impaired by blockade of A2b function. These findings suggest that neural cells modulate mRNA trafficking by regulating alternative splicing of hnRNP A2/B1 and controlling expression levels of A2b, which may be the predominant mediator of cytoplasmic trafficking functions. These findings highlight the importance of considering isoform-specific functions for alternatively spliced proteins.

Keywords: mRNA trafficking, hnRNP A2/B1, alternative splicing, subcellular distribution


mRNA trafficking in neural cells is a highly dynamic and regulated process in which a subset of mRNAs is transported from the nucleus to the distal processes, where translation leads to localized protein expression (reviewed in (1, 2)). Heterogeneous nuclear ribonucleoprotein (hnRNP) A2/B1 was the first trans-acting factor to be described in neural mRNA trafficking (3), and is necessary for dendritic targeting of various mRNAs that are involved in myelination and synaptic regulation (4). hnRNP A2/B1 recognizes the A2 response element (A2RE) (5), a cis-acting signal present in certain trafficked mRNAs including those encoding myelin basic protein (MBP), CaMKII, neurogranin and Arc (4, 6). Aside from RNA trafficking, hnRNP A2/B1 participates in virtually all aspects of mRNA processing, including packaging of nascent transcripts, splicing of pre-mRNAs, and translational regulation (reviewed in (7, 8)). Thus, hnRNP A2/B1 isoforms play important roles in post-transcriptional regulation within both the nuclear and cytoplasmic compartments.

A2/B1 pre-mRNA is alternatively spliced to produce 4 isoforms: B1 (includes all exons), A2 (excludes exon 2), B1b (excludes exon 9) and A2b (excludes exon 2 and 9), as shown in Figure 1A (9, 10). Despite the high degree of sequence similarity between A2/B1 isoforms, they are differentially expressed in different tissue types (10) throughout the cell cycle (11, 12), and in various disease states (1316). In addition, the relative ratios of hnRNP A2/B1 transcripts vary between tissue types and at different ages (9). This differential expression may reflect isoform-specific functions and regulatory mechanisms. However, to date, most of the work done on hnRNP A2/B1 has not distinguished between its isoforms, or has focused only on A2, which is the major isoform in most tissues. Determining the subcellular localization of these isoforms is essential for understanding their relative contributions to nuclear (mRNA packaging and splicing) and cytoplasmic (mRNA trafficking and translational regulation) functions.

Figure 1
Differential intracellular distribution of endogenous hnRNP A2/B1 isoforms

Nucleocytoplasmic shuttling of hnRNP A2/B1 is regulated by the M9 nuclear localization signal (NLS) located within the C-terminal glycine-rich domain (GRD), which is present in all isoforms (17). B1 and B1b contain exon 2, which encodes a 12-amino acid sequence with a high proportion of charged residues. The C-terminal 2 residues of exon 2, in combination with the N-terminal 2 residues of exon 3, make up the sequence RKKR, which resembles a classical NLS (18, 19). B1 and A2 contain exon 9, which encodes a 40-amino acid glycine-rich sequence within the GRD, N-terminal to the M9 NLS. It contains a repeated motif of several glycines followed by a hydrophobic amino acid. Such motifs are predicted to adopt a secondary structure of glycine loops (known as Ω-loops), which may be involved in protein-protein interactions (20). We set out to investigate if these alternative exons constituted additional novel localization signals that influence the differential subcellular distributions of indivudal hnRNP A2/B1 isoforms.

In this study, we found that A2b was the predominant cytoplasmic isoform in rat neural cells. Fluorescence correlation spectroscopy (FCS) in live cells revealed different dynamic properties for different isoforms. Also, subcellular localization of hnRNP A2/B1 was dependent on RNA integrity and the inclusion of alternative exons 2 and 9. Furthermore, the transcript and protein levels of the different isoforms varied with developmental stage and species. Finally, we detected isoform-specific differences in cytoplasmic functions. This is the first study to establish that there are differences among the hnRNP A2/B1 isoforms in terms of their subcellular localizations, dynamic properties and functional roles in RNA trafficking. This will have important implications for understanding their differential roles in mRNA processing, with A2b being the major player in mRNA trafficking.


Differential distribution of endogenous A2/B1 isoforms

As A2/B1 proteins participate in a variety of cellular processes that take place in either nuclear or cytoplasmic compartments, the nuclear/cytoplasmic distribution of specific isoforms is important in determining their involvement in particular processes. We therefore examined their intracellular distributions by immunostaining with isoform-specific antibodies (Figure 1B). In rat neural cells (hippocampal neurons, B104 neuroblastoma cells and oligodendrocytes), anti-A2/B1, which recognizes all isoforms, strongly stained nuclei with granular staining of cytoplasmic processes extending considerable distances along processes and often beyond several branch points. Anti-exon 8/10, which recognizes A2b and B1b, exhibited a similar staining pattern to that of anti-A2/B1, while staining with anti-exon 9, which recognizes A2 and B1 but not A2b or B1b, was restricted to nuclei. In contrast, human HeLa and SH-SY5Y cells, which express detectable levels of only the A2 and B1 isoforms, were not stained by anti-exon 8/10, and both anti-A2/B1 staining and anti-exon 9 staining was exclusively nuclear (Supplementary Figure 1). Thus, our results demonstrate that while all four isoforms are strongly expressed in nuclei, A2 and B1 are mainly localized to nuclei, and the predominant extranuclear isoforms in rat neural cells are A2b and B1b.

Differential distribution of exogenous A2/B1 isoforms

As both A2b and B1b share a common epitope recognized by anti-exon 8/10, we were unable to differentiate between the two isoforms based on immunostaining with antibodies. We therefore expressed individual exogenous GFP-fusion proteins for each of the A2/B1 isoforms in hippocampal neurons, and B104 neuroblastoma cells. Cells were immunostained for GFP to enhance the extranuclear signal. In hippocampal neurons expressing individual A2/B1 GFP-fusion proteins GFP fluorescence was localized predominantly to nuclei (Supplementary Figure 2), while control pEGFP was expressed throughout the entire cell (Figure 2A). There was significantly more GFP expression in processes of cells expressing A2b-GFP compared to A2-GFP, and A2b-GFP was often detected in multiple processes. While some cells expressing B1b-GFP also showed process staining, the difference compared to A2-GFP was not significant when analyzed by the Least Significant Differences method. The results for expression in B104 neuroblastoma cells were comparable to those for hippocampal neurons, except that significantly more cells expressing B1b-GFP had process staining compared to cells expressing A2-GFP (Figure 2B).

Figure 2Figure 2Figure 2Figure 2
Differential intracellular and intranuclear distribution of exogenous A2/B1-GFP fusion proteins in cells

Taken together, these findings indicate that in rat neural cells, while all A2/B1 isoforms are localized to nuclei, A2b (and also B1b in the case of B104 cells) is exported from the nucleus or retained in the cytoplasm to a greater extent than the other isoforms. The data also suggest that in the case of hippocampal cells, the predominant extranuclear isoform detected by immunostaining is likely to be A2b.

We had noted earlier that human HeLa and SH-SY5Y cells did not appear to express endogenous A2b or B1b (Figure 1B). To determine if extranuclear localization of exogenous A2b could take place in these cell types, they were also transfected with the A2/B1-GFP-fusion constructs. In HeLa cells, while pEGFP was expressed diffusely throughout the entire cell, GFP fusions of all A2/B1 isoforms, including A2b-GFP, were exclusively localized to nuclei (Figure 2C, 2D).. Interestingly, the intranuclear distribution of A2b-GFP in some HeLa cells was markedly different from that of the other isoforms. A2-, B1- and B1b-GFP were concentrated at the periphery of nucleoli (Figure 2C, arrows), in agreement with our previous findings (20), but A2b-GFP appeared to be concentrated at the nuclear envelope (Figure 2C, small arrowheads). These results suggest that the mechanism for extranuclear localization of A2b may be species-specific, and is absent or operates less efficiently in human compared to rodent cells (see also Figure 6A).

Figure 6
Expression of hnRNP A2/B1 transcripts at different developmental stages and in different cell types

Dynamic properties of A2/B1 isoforms in the nucleus and cytoplasm

To study the dynamic behavior of hnRNP A2/B1 isoforms in living cells, FCS was performed on B104 cells microinjected with plasmids expressing A2/B1-GFP fusion proteins. In FCS the fluorescence signal from a small (<1fl) observation volume is monitored over time. Autocorrelation analysis of the fluorescent fluctuations caused by movement of fluorescent molecules through the observation volume is used to determine concentrations and dynamic properties (diffusion coefficients, binding constants) for mobile molecules. Immobile fluorescent molecules located within the observation volume are gradually photobleached and analysis of photobleaching kinetics is used to determine the immobile fraction. Figure 3 shows FCS data indicating the proportions of freely diffusing, slow and immobile molecules for each of the A2/B1 isoforms in nucleus and cytoplasm of B104 cells.

Figure 3
Dynamics of A2/B1-GFP fusion proteins in B104 cells

In most cells, A2/B1 isoforms were predominantly immobile in the nucleus (Figure 3A) and predominantly freely diffusing in cytoplasm (Figure 3B), suggesting that the majority of A2/B1 molecules in the nucleus are bound to relatively immobile binding partners such as RNA transcripts or telomeres, whereas the majority of A2/B1 molecules in the cytoplasm are freely diffusing. There were also differences in dynamic behavior among A2/B1 isoforms. B1 had significantly higher immobile and lower freely diffusing faction (p<0.001) in both the nucleus and cytoplasm compared to the other isoforms suggesting that B1 binds to certain immobile partners to which the other isoforms do not bind. In the nucleus (Figure 3A), the freely diffusing fraction is significantly increased for A2b and B1b compared to A2 and B1 (p<0.05), suggesting that the presence of exon 9 in A2 and B1 increases association with immobile binding partners in the nucleus. In the cytoplasm (Figure 3B), the freely diffusing fraction is significantly increased for A2 and A2b compared to B1 and B1b (p<0.05), suggesting that the presence of exon 2 in B1 and B1b increases association with immobile binding partners in the cytoplasm.

In addition, the overall cytoplasmic/nuclear partition coefficient is greater for A2b than for the other isoforms (Fig 3C), which is consistent with whole cell immunofluorescent staining of endogenous A2/B1 isoforms and exogenous A2/B1 GFP fusions. This may reflect the reduced association of A2b with immobile binding partners in the nucleus, which would allow for increased export of freely diffusing A2b to the cytoplasm. In summary, the FCS data provides an explanation for the differential subcellular distribution patterns of A2/B1 isoforms in terms of differential binding of exons 2 and 9 to immobile binding partners in the nucleus and cytoplasm.

Distribution of A2/B1 isoforms is dependent on alternative exons and RNA integrity

Thus far, we have shown that the cytoplasmic distribution and mobility characteristics of A2b are different than the other isoforms. As A2b is the isoform that lacks exons 2 and 9, we next investigated the effect of these specific exons on localization of GFP constructs. To prevent leaky scanning due to the short length of exon 2 and exon 9, which could result in translation of GFP alone, their sequences were cloned into mGFP, which is identical to pEGFP-N1 except that the start codon immediately upstream of the GFP sequence has been mutated. We also cloned in A2ex2scr a scrambled version of exon 2 with the same overall distribution of amino acids (and therefore charges), and A2ex9mut, which is identical to exon 9 except that tyrosines, which are thought to contribute to the formation of Ω-loop secondary structures (20), are mutated to glycines (see Materials and Methods for more details). In addition, the M9 sequence in A2 was also cloned into mGFP as a positive control.

The constructs were transfected into HeLa cells (Figure 4). To ensure consistency in image collection, all cells were imaged using the same confocal settings, with the focal plane positioned midway between the top and bottom of the cells along the z-axis. In cells transfected with pEGFP-N1, GFP was expressed throughout the cytoplasm and nucleus. In contrast, M9-mGFP had markedly higher levels of expression in the nucleus than in the cytoplasm, as did both ex2-mGFP, ex2scr-mGFP and A2ex9-mGFP. Removal of the tyrosines in ex9mut-mGFP partially reduced nuclear accumulation. However, it should be noted that the differences in localization patterns were subtle, and even for M9-mGFP, which contains a well-characterized nuclear localization signal, nuclear accumulation was incomplete.

Figure 4
Localisation of exon 2 and 9

We had found previously that nuclear localization of anti-A2/B1 in HeLa cells was dependent on RNA integrity (21). Therefore, we treated hippocampal neurons with RNase in combination with detergent extraction to examine the effect on anti-A2/B1 staining in processes. At the RNase concentration used, almost all RNA was removed from cells (Figure 5, top row), and transfected GFP was completely removed from cells under the detergent extraction conditions used (data not shown). In cells treated with detergent only, anti-A2/B1 staining remained in the nucleus and in granules along cell processes (Figure 5, arrows). In contrast, almost all staining with anti-A2/B1 was removed in cells treated with RNase.

Figure 5
hnRNP A2/B1 localisation in processes and nuclei is dependent on RNA integrity

These results indicate that exon 2 and exon 9 promote nuclear localization, and that the localization of A2/B1 isoforms in the processes and nuclei of hippocampal cells is dependent on RNA integrity. In addition, while rearrangement of the amino acid sequence of exon 2 did not seem to have affect localization, removal of the tyrosines in exon 9 did affect its localization properties.

Expression of A2/B1 transcripts and proteins differ between species and developmental stage

The different expression patterns of A2/B1 isoforms in rat and human cell lines (Figure 1B) led us to investigate if their expression levels differed between these species. In addition, since transcript levels vary with age (9), we also examined expression levels at different developmental stages. Protein and RNA were extracted from the brains of day 21 postnatal (P21) and newborn rats (NB), and also from hippocampal neurons (rat), B104 neuroblastoma cells (rat), HeLa cells (human) and SH-SY5Y neuroblastoma cells (human). RT-PCR of the RNA samples was performed using isoform-specific primer pairs (Figure 6A). The same primer sets were also used to perform PCR on adult human brain cDNA. The results indicate that expression of the B1 and A2 transcripts is similar across developmental time points and cell types. However, levels of B1b and A2b transcripts are markedly lower in P21 than NB rat, and in human (HeLa and SH-SY5Y) cells compared to rat (hippocampal and B104) cells. There is no obvious difference in expression patterns between epithelial HeLa and neuronal SH-SY5Y cells. Neither B1b nor A2b transcripts are detected in human brain cDNA. Similarly, Western blots of whole cell lysates (WCL) of HeLa and SH-SY5Y cells show that A2b protein is not detectable in these human cell lines (data not shown). Taken together, these results indicate that splicing out of exon 9 is less frequent in human than in rat cells and tissues, and is less frequent in P21 rat brain than in newborn rat brain or rat cells in culture. Thus, our results demonstrate that the splicing pattern of A2/B1 is developmentally regulated and differs between species.

We next looked at whether extranuclear localization of A2b was developmentally regulated. Hippocampal neurons were differentiated in 10% fetal bovine serum for 0–21 days, before immunostaining for A2/B1 isoforms. Treatment with serum caused the cells to express neurofilaments, and develop extended and highly branched processes (Figure 7, insets). In cells that had been differentiated for 6 days or less, there was very little cytoplasmic immunostaining with anti-A2/B1 (i.e. all isoforms). However, the level of cytoplasmic anti-A2/B1 staining increased from day 8 up to day 14 (Figure 7A,B). This increase continued up to day 21, at which point the image analysis software was no longer able to differentiate individual granules due to their density (data not shown). There was no significant difference between the density of staining with anti-exon 8/10 and that of anti-A2/B1, whereas there was very little cytoplasmic staining with anti-exon 9. While anti-exon 8/10 does not distinguish between A2b and B1b, based on our earlier results, the cytoplasmic staining is likely to be mostly due to A2b. Thus, cytoplasmic expression of A2b appears to increase with cellular differentiation and maturation of neuronal processes.

Figure 7
Temporal regulation of cytoplasmic localisation

Assembly of A2RE mRNAs into cytoplasmic granules is specifically mediated by the A2b isoform

Our immunostaining and FCS data demonstrate that the cytoplasmic/nuclear partition coefficient for A2b is greater than that of other A2/B1 isoforms, which suggests that the A2b isoform is the predominant mediator of cytoplasmic A2/B1 functions. One of the cytoplasmic functions of hnRNP A2/B1 is assembly of A2RE RNAs into granules for trafficking to dendrites and distal processes by the A2 pathway (4). Binding of hnRNP A2/B1 isoforms to TOG protein is believed to mediate assembly of RNA granules in the cytoplasm. If a particular exon-specific antibody blocks binding of a specific hnRNP A2/B1 isoform to TOG protein this may inhibit granule assembly. To determine if exon-specific antibodies block binding of hnRNP A2/B1 isoforms to TOG protein, surface plasmon resonance (SPR) was used to measure binding of hnRNP A2 or hnRNP A2b to full length TOG in the presence and absence of either exon 9 antibody (which binds to hnRNP A2) or exon 8–10 antibody (which binds to hnRNP A2b). SPR measurements were performed with full length TOG-GFP covalently immobilized on the chip as ligand and hnRNP A2-GFP or hnRNP A2b-GFP mixed with either exon 9 antibody or antibody exon 8–10 antibody as analyte. If antibody binding to a specific hnRNP A2 isoform does not affect binding to TOG the amplitude of the spectrogram should be increased due to additional antibody mass associated with the hnRNP A2 analyte. If antibody binding to a specific hnRNP A2 isoform blocks binding to TOG the amplitude of the spectrogram should be decreased due to less analyte binding. Antibody binding to hnRNP A2 isoform analyte may not completely inhibit binding to TOG because the antibody in solution is competing for binding to the specific hnRNP A2 isoform with immobilized TOG protein, which is present at very high concentration on the chip. The concentration of antibody in solution may be insufficient to completely eliminate binding of hnRNP A2 isoforms to TOG. Spectrograms for different analyte/antibody combinations are shown in Supplementary Figure 4 panel A. Exon 9 antibody did not affect binding of hnRNP A2b but reduced binding of hnRNP A2 to TOG indicating that antibody binding blocks the TOG binding site on hnRNP A2. Exon 8–10 antibody did not affect binding of hnRNP A2 but reduced binding of hnRNP A2b to TOG indicating that antibody binding blocks the TOG binding site on hnRNP A2b. These results demonstrate that exon 9 antibody and exon 8–10 antibody can be used as function-blocking antibodies to disrupt hnRNP A2::TOG and hnRNP A2b::TOG interactions, respectively.

To test the hypothesis that A2RE RNA granule assembly is mediated specifically by the A2b isoform, oligodendrocytes were co-injected with fluorescently-tagged A2RE RNA (MBP RNA) and either anti-exon 9 antibody (recognizes B1 and A2) or anti-exon 8/10 antibody (recognizes B1b and A2b). Assembly of fluorescent RNA into granules in injected cells was analyzed by confocal microscopy. Since RNA and antibodies were injected directly into the perikaryon, this experiment specifically assays cytoplasmic functions without the confounding effects of prior nuclear events. We found that in most cells injected with anti-exon 9 or water, MBP RNA was assembled into granules. In contrast, in most cells injected with anti-exon 8/10, MBP RNA was diffuse throughout the perikaryon (Figure 8). Our results demonstrate that assembly of A2RE RNA into granules is inhibited by anti-exon 8/10 but not by anti-exon 9, which is consistent with the hypothesis that assembly of A2RE RNA into granules in the cytoplasm is predominantly mediated by ex 8/10-containing isoforms (A2b or B1b) but not by exon 9 containing isoforms (B1 and A2).

Figure 8
A2RE RNA granule assembly is blocked by isoform-specific antibodies


We have shown, using isoform-specific antibodies and GFP fusion proteins, that subcellular distributions and dynamic properties differ among the hnRNP A2/B1 isoforms. Inclusion of exon 2 and/or exon 9 promotes nuclear retention, with the result that A2b (which lacks exons 2 and 9) is the predominant cytoplasmic isoform. In addition, A2b levels are developmentally regulated and species-dependent, which suggests that regulation of its expression could be a means to modulate cytoplasmic mRNA trafficking.

Differential distribution and expression of hnRNP A2/B1 spliceoforms

One current explanation for the multitude of isoforms generated by alternative splicing of hnRNP A2/B1, and all hnRNPs in general, is that this allows for regulatory control via coordinated binding of structurally related proteins in a combinatorial model (22). In such a model, small changes in the relative proportions of individual isoforms with slightly different binding affinities could significantly alter the functional characteristics of the resultant complex, thus allowing cells to modulate hnRNP function by regulating isoform availability in a tissue- or cell-specific manner. Here, we highlight an additional role of alternative splicing wherein hnRNP A2/B1 function is modulated through differential inclusion of alternative exons that affect the subcellular distribution of isoforms. Hence, alternative splicing leads to expression of isoforms with distinct localization characteristics, which determine their contribution to cellular activities, such as mRNA trafficking, that are confined to specific subcellular compartments. This theme of regulation by localization can be seen in a variety of alternatively spliced proteins including membrane-associated receptors, nuclear transcription factors, and even proteins sublocalized within organelles (reviewed in (23)).

Control of protein expression via alternative splicing has been extensively characterized in the nervous system, where it mediates cell differentiation, axon guidance and synaptogenesis (24). Our findings suggest that regulation of production of hnRNP A2/B1 isoforms with different localization properties may also provide a means of modulating mRNA trafficking. Our results demonstrate that expression of A2/B1 spliceoforms is affected by developmental age, species and cellular differentiation. In addition, cytoplasmic levels of hnRNP A2/B1 in rat oligodendrocytes are temporally correlated with MBP mRNA levels (25), although these experiments did not distinguish between different isoforms. In light of our results, it is probable that cytoplasmic hnRNP A2/B1 staining observed in previous studies was predominantly due to A2b. Also, A2b and B1b protein and mRNA levels are higher in newborn and 5-day-old mice than in 3-week- and 7-week-old mice (data not shown), which suggests that their expression levels are upregulated during periods of rapid neuronal development when RNA trafficking is most active. Thus, cells may adjust levels of A2b through alternative splicing in response to mRNA trafficking requirements.

Mechanisms and dynamics of hnRNP A2/B1 localization

hnRNP A2/B1, and hnRNPs in general, play a crucial role in coupling post-transcriptional regulation of nuclear and cytoplasmic mRNA metabolism (Shyu and Wilkinson, 2000), and the nucleocytoplasmic distribution of hnRNP A2/B1 isoforms could have a significant impact on this function. A previous study found that an A2 construct missing the M9 nucleocytoplasmic shuttling signal was not localized to the nucleus, indicating that the M9 sequence is necessary for nuclear localization (26). However, it should be noted that this construct did not contain exon 2, and also that the M9 sequence had been removed along with the adjoining C-terminal half of exon 9. Thus, such a construct could not distinguish whether the presence of exon 2 or exon 9 affects nuclear localization. The results in Figure 4 suggest that these exons may contribute to nuclear localization, although the effect is partial. This could be because the inserted sequences did not adopt the same secondary structure in the mGFP constructs as in their native configurations. Since shuttling proteins such as A2/B1 constantly move between the nucleus and cytoplasm, differences in export and import rates can produce a predominantly nuclear or cytoplasmic distribution at steady state (27). The binding and dynamic properties of A2/B1 isoforms revealed by our FCS experiments suggest that differences in dynamic properties due to the inclusion or exclusion of exons 2 and 9 may influence their cellular distribution.

Our results in Figure 5 agree with previous findings that intranuclear hnRNP A2/B1 localization is primarily dependent on interactions with RNA (21), and that RNA granules are sensitive to RNase (28). A2 possesses at least 2 binding sites for RNA – one that binds only A2RE-containing oligonucleotides, and a second non-specific site that does not discriminate between sequences (it should be noted that hnRNP A2/B1 isolated from rat brain does not appear to bind RNA non-specifically, but this may be due to saturation of the non-specific site with endogenous RNA) (29). It is likely that the specific site is required for trafficking of A2RE-containing mRNAs (4) and telomere maintenance (30), while the non-specific site participates in intranuclear RNA packaging (31). In addition, our data indicate that binding of hnRNP A2/B1 to RNA may play an important role in its retention in the nucleus or assembly into cytoplasmic RNA granules.

A2RE RNA granule assembly is isoform-dependent

Our microinjection results demonstrate that assembly of A2RE RNA into granules is impaired by blockade of A2b and B1b function by antibody to exon 8/10 epitope but not by blockade of A2 and B1 function by antibody to exon 9 epitope. This effect is specific to A2RE RNAs, as non-A2RE RNAs are not affected by antibody microinjection (data not shown). Assembly of A2RE RNA into granules involves two separate molecular interactions of hnRNP A2/B1 molecules - binding to A2RE sequences in different RNAs, and binding to tumor overexpressed gene (TOG) protein, which serves as a scaffold to link multiple A2/B1-A2RE RNA complexes together in RNA granules (32). Since binding sites for A2RE RNAs are located in the N terminal part of the hnRNP A2/B1 molecule, binding to A2RE RNAs may not be affected by antibody binding to exon8/10 epitope, which is located in the C terminal part of the molecule. A more likely explanation is that antibody binding to the exon 8/10 epitope interferes with binding of A2b and/or B1b molecules to TOG protein, thereby preventing assembly of A2RE RNAs into granules. In this regard, studies in Drosophila indicate that hrp48 (the Drosophila orthologue of hnRNP A2) is specifically required for a distinct step in osk mRNA localization (33). Mutations in a region of hrp48 corresponding to the region of the ex8/10 junction in hnRNP A2b disrupt assembly of oskar RNA into granules, This suggests that this region of the protein is important for granule assembly in both hnRNP A2b in rodent neural cells and in hrp48 in Drosophila embryos. To date, functional studies have not distinguished between the A2/B1 isoforms, or have focused on A2, the most abundant isoform. Here we have shown for the first time that hnRNP A2/B1 functions may be isoform-specific, as A2b appears to be the predominant mediator of cytoplasmic RNA granule assembly. Isoform-specific characteristics should be taken into consideration in the design of future functional studies.

Our results illustrate how developmentally regulated alternative splicing can influence protein localization by modulating the inclusion of domains that participate in key protein-protein or protein-nucleic acid interactions. Due to this isoform-specific localization, minor isoforms such as A2b may play a disproportionately large role in compartmentalized cellular functions, such as mRNA trafficking. As A2b expression is undetectable or very low in adult human brain and human cell lines, respectively, human cells may utilize different mechanisms for mRNA trafficking compared to rodent cells. Furthermore, the splicing pattern of hnRNP A2/B1 is species-specific, and can be linked to splice site strength and the presence of splicing regulatory elements (manuscript in preparation). More research will be required to fully investigate the implications of species-specific splicing on protein localization and function of hnRNP A2/B1.

Materials and Methods

Cell Culture

Rat hippocampal progenitor cells were isolated (Brewer, 1999) and maintained in Neurobasal A supplemented with 2% B27, 50μM Glutamax (all from Gibco, Victoria, Australia) and 5 ng/ml bFGF (Sigma, Sydney, Australia) in poly-D-lysine-coated 6-well plates. In some cases, cells were differentiated in Neurobasal A/Glutamax supplemented with 10% fetal bovine serum. HeLa cells were maintained in DMEM (Gibco) containing 10% newborn calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and HEPES at pH 7.4. B104 neuroblastoma cells were maintained in DMEM-F12 containing 5% newborn calf serum, Antibiotic-Antimycotic and 2 mM glutamax (all from Gibco). SH-SY5Y cells were maintained in DMEM-F12 (Sigma) containing 10% fetal bovine serum. All cells were maintained at 37°C in a humidified 5% CO2 atmosphere, and half the culture media were changed every 3–4 days.

Antibodies and Immunostaining

Antibodies were raised against peptides that recognized sequences that were shared between all isoforms (GGNFGFGDSRGGC-NH2, A2/B1) (34), present in exon 9 (GGNFGGSPGYGGGRG-NH2, A2 and B1), or flanking exon 9 (NGYGGGPGGNYGSG-NH2, A2b and B1b).

Hippocampal progenitor cells, B104 or SH-SY5Y cells were plated onto poly-D-lysine coated coverslips at 10,000 cells per coverslip. In some cases, hippocampal progenitor cells were differentiated in serum-containing media for 1–21 days. HeLa cells were plated at ~30% confluency. Cells were then fixed and immunostained using isoform-specific primary antibodies raised in our laboratory (rabbit polyclonal A2/B1 (recognizes all A2/B1 isoforms) 1:200, exon 9 (recognizes A2 and B1) 1:200, and exon 8/10 (recognizes A2b and B1b) 1:25). The specificity of these antibodies was verified by peptide blocking experiments (Supplementary Figure 3.). Hippocampal progenitor cells were also immunostained using mouse anti-neurofilament (Sigma, 1:100) to confirm neuronal phenotype, and to visualize processes. Cells were then immunostained with Alexafluor 546 anti-mouse IgG and Alexafluor 488 anti-rabbit IgG (both from Molecular Probes, 1:2,000) and stained with Hoechst 32258 (Molecular Probes, 1:10,000). Finally, coverslips were mounted in Prolong Gold Antifade (Molecular Probes) and viewed under a LSM 510 Meta confocal microscope (Zeiss Inc, Oberkochen, Germany).

For quantification of granule density in processes, hippocampal cells were randomly selected and imaged (Plan-Apochromat 100x/1.4 oil lens, Zeiss) with confocal settings adjusted to ensure consistent exposure and background across images, which were then processed using NIH Image Analyzer. Image grayscale was inverted and images were smoothed, and the threshold adjusted to a constant value before particles within a defined size range were counted by the software. The results were analyzed by Student’s t-test.


Coding regions of the A2/B1 isoforms were amplified from mouse cDNA and cloned into the Xho I/Age I site of pEGFP-N1 (Clontech, CA, USA).

All primers and oligonucleotides used for cloning exons 2 and 9 are listed in Table S1. mGFP was made by site-directed mutagenesis of pEGFP-N1 with GFPmutq1 and GFPmutq2. This converted the Kozak consensus sequence and start site from GCCACCATG to GGCGGCGTG.

A2ex2 and A2ex2scr were made by annealing A2ex2F and A2ex2R, and A2ex2scrF and A2ex2scrR, respectively. These oligonucleotides were designed such that the annealed product would be flanked by a 5′ BglII site and start codon and a 3′ AgeI site with sticky ends. A2ex2 codes for ME-KTLETVPLERKK-RE, which corresponds to exon 2 and the last and first 2 amino acids of exons 1 and 3, respectively. A2ex2scr codes for METKELVTRLKPKERE. Note that the NLS-like RKKR sequence is no longer present in this sequence.

A2ex9 and A2M9 were generated by PCR from rat hippocampal progenitor cell cDNA with primers (A2ex9F and A2ex9R, A2M9F and A2M9R, respectively) that introduced a 5′ BglII site and start codon and a 3′ AgeI site. A2ex9mut was made by annealing A2ex9mut-1 and A2ex9mut-2, after which a fill-in reaction was performed using Platinum Taq polymerase (Invitrogen). A2ex9 and A2ex9mut code for MGGNFGGSPGYGGGRGGYGGGGPGYGNQGGGYGGGYDNYGG and MGGNFGGSPGGGGGRGGGGGGGPGGGNQGGGGGGGGDNYGG respectively, where the tyrosines that have been converted into glycines are underlined. A2M9 codes for MQQPSNYGPMKSGNFGGSRNMGGPYGGGNYGPGGSGGSGGYG.

A2ex9, A2M9 and A2ex9mut were then digested and ligated into digested mGFP, while A2ex2 and A2ex2scr were directly ligated into digested mGFP.


Hippocampal cells and SH-SY5Y cells were plated onto coated coverslips at 20,000 cells per coverslip and transfected with plasmids expressing A2/B1-GFP constructs using Lipofectamine LTX (Invitrogen) according to the manufacturer’s instructions. 200 ng of DNA was added per well, and the transfection media was replaced with conditioned media 4–6 hours later. Empty pEGFP-N1 (Clontech) was used as a control. HeLa cells at ~30% confluency were also transfected as above, while B104 were plated at 10,000 cells per coverslip and transfected using Lipofectamine 2000 (Invitrogen) instead. 6 sets of transfections were performed for each construct in hippocampal progenitor cells and B104 cells, while 4 sets of transfections were performed in HeLa cells and SH-SY5Y cells.

24 hours after transfection, cells were fixed and immunostained with rabbit anti-GFP (kind gift from Dr. Pam Silver, Harvard University; 1:15,000) followed by Alexafluor 546 anti-rabbit IgG (Molecular probes, 1:2,000) to enhance the extranuclear signal. They were then imaged at the same exposure settings (for each cell type and each set of transfection) using a LSM 510 Meta confocal microscope (Zeiss Inc, Oberkochen, Germany). B104, hippocampal, SH-SY5Y and HeLa cells were imaged using Plan-Apochromat 10x/0.45 dry, 20x/0.75 dry, 40x/1.3 oil and 100x/1.4 oil lenses (all lenses from Zeiss), respectively.

Images were assessed by an observer blinded to the identities of the transfected construct. Positive process staining was defined to be significant fluorescence beyond at least one branch point in the cell processes for hippocampal cells. For B104 cells, positive process staining was defined to be significant fluorescence that extended into at least one process, where processes were defined as projections from the cell body that were less than half the diameter of the nucleus. As the data consisted of proportions, variances were stabilized using the transformation X′ = 2 arcsin √X (35). One-way ANOVA was performed on the transformed data, followed by Least Significant Difference t-tests. The data was then transformed back to the original units for presentation purposes.

For mGFP constructs, HeLa cells were transfected with 100 ng of construct using Fugene HD (Roche, Sydney, Australia) according to the manufacturer’s instructions. Cells were fixed 24 hours post-transfection and viewed by confocal microscopy (Plan-Apochromat, 60x/1.4 oil lens, Zeiss). Images were taken at the same exposure settings. Using the z-stack function, the top and bottom of the cells along the z-axis were marked, and images were taken at the midpoint of the stack.

Fluorescence Correlation Spectroscopy

B104 cells were grown on 35 mm glass bottom petri dishes and microinjected with GFP fusion plasmids at concentrations of 0.1 μg/μl using the Eppendorf Micromanipulator 5171 system (Hamburg, Germany). Injected cells were incubated for approximately three hours post-injection to allow time for expression of GFP fusion proteins.

FCS measurements were carried out using a Confocor 3 system with C-Apochromat 40x, NA 1.2 water immersion objective (Zeiss). The FCS observation volume was positioned in either the nucleus or the cytoplasm. Fluorescence fluctuation data was collected for 5×10 seconds at each location. FCS data was analyzed using Zeiss Confocor software. Autocorrelation functions were calculated and fitted to a three dimensional diffusion model with two components.


Where N is the molecular concentration, R is the ratio of the axial over lateral dimensions of the observation volume, τD1 and τD2 are the diffusion times for the faster and slower components, respectively, Y is the fraction of particles with diffusion time τD2, and f(T) is the function used for fitting the triplet characteristics τT and % τT of the fluorophore (GFP). Concentrations of mobile fluorescent molecules [M] were determined from the amplitude of the autocorrelation function (1+1/N). Immobile components (I) were determined from photobleaching kinetics in the first trace. The concentration of immobile fluorescent molecules [I] was calculated as:


One-way ANOVA was performed on the data, followed by Least Significant Difference t-tests.

Detergent Extraction and RNase Treatment

Unfixed serum-differentiated hippocampal progenitor cells on coverslips were washed in CSK buffer (100 mM NaCl, 300 mM sucrose, 10 mM PIPES at pH 7.4, 3 mM MgCl2, 1 mM EGTA, 1 mM PMSF). Detergent extraction was immediately performed by permeabilizing the cells in the same buffer containing 0.2% Triton-X100 (CSK-TX), which removes the soluble protein within the cells. Some cells (RNase-treated) were then treated with 100 μg/ml RNase A (Fermentas, Ontario, Canada) followed by a second incubation in CSK-TX to remove solubilized protein. Other cells (detergent-extracted) were treated with CSK buffer in place of RNase A. Coverslips were then fixed and immunostained for A2/B1 (1:200) and α-tubulin (Sigma, 1:200) as described above. Coverslips were also stained with Hoechst 32258 (Molecular Probes, 1:10,000) followed by Pyronin Y (Sigma, 1:15,000) to visualize DNA and RNA respectively, so as to determine the extent of RNA degradation after RNase treatment. Cells were then viewed using confocal microscopy as described above.


RNA was extracted from P21 and NB rat brain, and hippocampal progenitor, B104, HeLa and SH-SY5Y cells, using Trizol (Invitrogen, Victoria, Australia) according to the manufacturer’s instructions. The concentration of the RNA was measured on a Nanodrop ND-1000 (BioLab, Victoria, Australia), and 2 μg of each RNA was used for cDNA synthesis using SuperScript III (Invitrogen). Adult human brain cDNA was a kind gift from Dr. Peter Dodd (School of Chemistry and Molecular Biosciences, University of Queensland). 1 μl of each cDNA was amplified using isoform-specific primers (see Table S2; Geneworks, SA, Australia) using the following conditions: 94°C for 5 minutes, 30 repeats of 94°C for 30 seconds, 60°C for 30 seconds, 72°C for 1 minute, followed by 72°C for 7 minutes. Amplification using β-actin primers was carried out under the same conditions except with 25 repeats. PCR products were run on a 1% TAE-agarose gel, which was then viewed on a UV transilluminator (Gibco) and imaged using Kodak Molecular Imaging Software v 4.5.1 (Carestream Health, NY, USA). All experiments involving animals were conducted in accordance with the University Animal Ethics Committee (The University of Queensland) guidelines.

Construction of ch-TOG plasmid and [hook-right] MACS Purification of GFP-Tagged Proteins

ch-TOG-GFP was prepared by PCR of human ch-TOG cDNA (in pBSII KS+ vector) using a forward primer (5′-CTCCACCGCGGGATTACAAGGAAAACCTGGA A-3′) to introduce a SacII site and a reverse primer (5′-GCGGGATCCTTTGCGACTGCTCTT TATTC-3′) to introduce a BamH1 site flanking the ch-TOG ORF. The amplified product containing the TOG ORF was digested with SacII and BamH1 and subcloned into pEGFP-C1 (Clontech). The plasmid encoding the TOG-GFP fusion protein (as well as plasmids encoding A2-GFP and A2b-GFP plasmids) were transfected into pEAK Rapid cells (Edge Biosystems, Gaithersburg, MD) using standard Lipofectamine 2000 (Invitrogen) protocol. GFP tagged fusion proteins were purified using the [hook-right] MACS Epitope Tag protocol (Milteyni Biotec). Briefly, 24 h post transfection the medium was removed from the culture dishes and replace with 1 ml pre-cooled (4°C) Lysis buffer (150 mM NaCl, 1% TritonX-100, 50 mM Tris HCL (pH8.0). Cells were scraped, transferred to a 1.5 mL tube and incubated for 30 min on ice with occasional mixing. After centrifugation for 10 m at 10,000xg at 4°C, the supernatant was transferred to a fresh tube and 50 [hook-right] l anti-GFP MicroBeads were added and incubated on ice for 30 m. [hook-right] Columns were prepared by placing in the magnetic field of the[hook-right] MACS Separator and applying 200[hook-right] l Lysis buffer to the column. Reaction mixtures were applied to columns and lysates were allowed to run through. Columns were washed 4×200 [hook-right] l with Wash buffer (150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50mM Tris HCl (pH 8.0) followed by 100[hook-right] l 20 mM Tris HCl (pH7.5). GFP-fusion protein was eluted under native conditions by applying 20[hook-right] l 0.1 M Triethylamine, pH 11.8, 0.1% Triton X-100 to the column, incubating for 5 min at room temperature, followed by another 50 [hook-right] l of the same buffer and collecting the eluate in a tube containing 5 [hook-right] l of 1 M Tris HCl (pH 6.8) for neutralization.

Surface Plasmon Resonance

Surface plasmon resonance (SPR) measurements were performed using a Biacore T100 instrument. Ligand consisted of [hook-right] MACs-purified ch-TOG-GFP immobilized by covalent carbodiimide coupling to a CM5 chip (GE Healthcare). The amount of protein immobilized on the chip was 800 resonance units (RU), corresponding to 0.8 ng/mm2. Mixtures of [hook-right] MACS purified A2-GFP or A2b-GFP (200 nmoles) were incubated with or without 400 nmoles of exon 9 or exon 8,10 antibody in 15[hook-right] l HEPES-buffered saline buffer (10mM HEPES, 3mM EDTA, 0.15 M NaCl, 2mM MgCl2 and 0.05% surfactant p20, pH 7.5) at 4°C overnight. Analyte consisting of the reaction mixtures diluted to a final concentration of 20 nM A2-GFP or A2b-GFP in HEPES-buffered saline buffer and were injected at a flow rate of 30[hook-right] l/min for 180 sec, followed by post injection dissociation in the same buffer for 180 sec at the same flow rate. The surface was regenerated between injections using 2.5M NaCl at a flow rate of 100[hook-right] l/m for 30 s. Each analyte combination was measured in duplicate in at least two separate experiments. Differences between duplicates was less than 2% for each analyte combination.

Microinjection in Live Cells

Mature rat oligodendrocytes were injected with a mixture of labeled MBP mRNA (which contains the A2RE sequence) and anti-exon 9 or anti-exon 8/10, or water as a control. Transcription of labeled mRNA was performed according to manufacturer’s specifications (Epicentre, Ampliscribe T7 Flash, WI, USA). The final concentration of Alexa488-UTP labeled full length MBP mRNA in the mix was 3.5 μg/μl solution. Thirty minutes after injection, the cells were imaged by confocal microscopy in a microscope incubation chamber pre-equilibrated to 37°C, 5% CO2. Images were collected by confocal laser scanning microscopy. The proportions of cells with granular or diffuse labeling under each injection condition were analyzed by the Fisher exact test.

Supplementary Material

Supp Fig s1

Supplementary Figure 1. Exon 9 staining is restricted to nuclei:

Cells were immunostained with anti-exon 9, which recognizes hnRNPs A2 and B1. Staining with Hoechst 32258 dye was performed to define nuclear boundaries. Bars represent 20 μm.

Supp Fig s2

Supplementary Figure 2. A2b-GFP is present in the cytoplasm of rat but not human cells:

Cells were transfected with A2b-GFP, fixed, and immunostained with anti-GFP to enhance the fluorescence signal. Staining with Hoechst 32258 dye was performed to define nuclear boundaries. In rat cells (hippocampal and B104), A2b-GFP was expressed in the cell soma and processes, whereas in human cells (SH-SY5Y and HeLa), A2b-GFP was restricted to nuclei. Bars represent 20 μm.

Supp Fig s3

Supplementary Figure 3. Specificity of antibodies:

A) Neuronal cells were immunostained with anti-A2/B1, anti-exon 8/10 and anti-exon 9 antibodies in the presence of no peptide, specific peptide (the peptide against which the antibody was raised) or an unrelated peptide. Specific peptides, but not unrelated peptides, successfully reduced the level of signal of each antibody. B) Western blot of mouse whole brain lysate immunostained with anti-exon 8/10 and 9.

Supp Fig s4

Supplementary Figure 4. Surface Plasmon resonance analysis of function-blocking antibodies:

Full length TOG GFP protein was covalently immobilized as ligand on the surface of a Biacore chip. Analyte (hnRNP A2-GFP (Panel A) or hnRNP A2b-GFP (Panel B) mixed with either ex9 antibody or ex8/10 antibody) was injected at 75 sec and replaced with buffer at 250 sec. Spectrograms are shown for analyte with no antibody (red), ex9 antibody (blue) or ex8/10 antibody (green).


This work was supported by grants from the Australian National Health and Medical Research Council and the Cancer Council of Queensland to RS and JAR and grants to JHC (NS15190) and EB (NS19943) from the National Institutes of Health. HSP was supported by a University of Queensland Research Scholarship. We would like to thank Yuan Li for assistance in data collection and analysis, Adam Skarshewski for making the mGFP construct, Dr. Rachel de las Heras for isolating the hippocampal progenitor cells, Dr. Pam Silver for providing the anti-GFP antibody, Dr. Markus Kerr for providing the SH-SY5Y cells and Dr. Peter Dodd for providing the human brain cDNA.


1. Sossin WS, DesGroseillers L. Intracellular trafficking of RNA in neurons. Traffic. 2006;7(12):1581–1589. [PubMed]
2. Martin KC, Zukin RS. RNA trafficking and local protein synthesis in dendrites: An overview. J Neurosci. 2006;26(27):7131–7134. [PubMed]
3. Shan JG, Munro TP, Barbarese E, Carson JH, Smith R. A molecular mechanism for mRNA trafficking in neuronal dendrites. J Neurosci. 2003;23(26):8859–8866. [PubMed]
4. Gao Y, Tatavarty V, Korza G, Levin MK, Carson JH. Multiplexed Dendritic Targeting of {alpha} Calcium Calmodulin-dependent Protein Kinase II, Neurogranin, and Activity-regulated Cytoskeleton-associated Protein RNAs by the A2 Pathway. Mol Biol Cell. 2008;19(5):2311–2327. [PMC free article] [PubMed]
5. Ainger K, Avossa D, Morgan F, Hill SJ, Barry C, Barbarese E, Carson JH. Transport and localisation of exogenous myelin basic protein messenger-RNA microinjected into oligodendrocytes. J Cell Biol. 1993;123(2):431–441. [PMC free article] [PubMed]
6. Munro TP, Magee RJ, Kidd GJ, Carson JH, Barbarese E, Smith LM, Smith R. Mutational analysis of a heterogeneous nuclear ribonucleoprotein A2 response element for RNA trafficking. J Biol Chem. 1999;274(48):34389–34395. [PubMed]
7. He Y, Smith R. Nuclear functions of heterogeneous nuclear ribonucleoproteins A/B. Cell Mol Life Sci. 2008;66:1239–1256. [PubMed]
8. Kwon S, Barbarese E, Carson JH. The cis-acting RNA trafficking signal from myelin basic protein mRNA and its cognate trans-acting ligand hnRNP A2 enhance cap-dependent translation. J Cell Biol. 1999;147(2):247–256. [PMC free article] [PubMed]
9. Hatfield JT, Rothnagel JA, Smith R. Characterization of the mouse hnRNP A2/B1/B0 gene and identification of processed pseudogenes. Gene. 2002;295(1):33–42. [PubMed]
10. Kamma H, Horiguchi H, Wan LL, Matsui M, Fujiwara M, Fujimoto M, Yazawa T, Dreyfuss G. Molecular characterization of the hnRNP A2/B1 proteins: Tissue-specific expression and novel isoforms. Exp Cell Res. 1999;246(2):399–411. [PubMed]
11. He YW, Brown MA, Rothnagel JA, Saunders NA, Smith R. Roles of heterogeneous nuclear ribonucleoproteins A and B in cell proliferation. J Cell Sci. 2005;118(14):3173–3183. [PubMed]
12. Kamma H, Satoh H, Matusi M, Wu WW, Fujiwara M, Horiguchi H. Characterization of hnRNP A2 and B1 using monoclonal antibodies: intracellular distribution and metabolism through cell cycle. Immunol Lett. 2001;76(1):49–54. [PubMed]
13. Mizukami K, Ishikawa M, Iwakiri M, Ikonomovic MD, Dekosky ST, Kamma H, Asada T. Immunohistochemical study of the hnRNP A2 and B1 in the hippocampal formations of brains with Alzheimer’s disease. Neurosci Lett. 2005;386(2):111–115. [PubMed]
14. Sueoka E, Goto Y, Sueoka N, Kai Y, Kozu T, Fujiki H. Heterogeneous nuclear ribonucleoprotein B1 as a new marker of early detection for human lung cancers. Cancer Res. 1999;59(7):1404–1407. [PubMed]
15. Sueoka E, Sueoka N, Iwanaga K, Sato A, Suga K, Hayashi S, Nagasawa K, Nakachi K. Detection of plasma hnRNP B1 mRNA, a new cancer biomarker, in lung cancer patients by quantitative real-time polymerase chain reaction. Lung Cancer. 2005;48(1):77–83. [PubMed]
16. Wu SL, Sato M, Endo C, Sakurada A, Dong BM, Aikawa H, Chen Y, Okada Y, Matsumura Y, Sueoka E, Kondo T. hnRNP B1 protein may be a possible prognostic factor in squamous cell carcinoma of the lung. Lung Cancer. 2003;41(2):179–186. [PubMed]
17. Pollard VW, Michael WM, Nakielny S, Siomi MC, Wang F, Dreyfuss G. A novel receptor-mediated nuclear protein import pathway. Cell. 1996;86(6):985–994. [PubMed]
18. Burd CG, Swanson MS, Gorlach M, Dreyfuss G. Primary structures of the heterogeneous nuclear ribonucleoprotein A2-protein, and C2-protein – A diversity of RNA-binding proteins is generated by small peptide inserts. Proc Natl Acad Sci USA. 1989;86(24):9788–9792. [PMC free article] [PubMed]
19. Weighardt F, Biamonti G, Riva S. Nucleocytoplasmic distribution of human hnRNP proteins A – search for the targeting domains in hnRNP A1. J Cell Sci. 1995;108:545–555. [PubMed]
20. Steinert PM, Mack JW, Korge BP, Gan SQ, Haynes SR, Steven AC. Glycine loops in proteins - Their occurrence in certain intermediate filament chains, loricrins and single-stranded RNA-binding proteins. Int J Biol Macromol. 1991;13(3):130–139. [PubMed]
21. Friend LR, Han SP, Rothnagel JA, Smith R. Differential subnuclear localisation of hnRNPs A/B is dependent on transcription and cell cycle stage. Biochim Biophys Acta. 2008;1783(10):1972–1980. [PubMed]
22. Krecic AM, Swanson MS. hnRNP complexes: composition, structure, and function. Curr Opin Cell Biol. 1999;11(3):363–371. [PubMed]
23. Stamm S, Ben-Ari S, Rafalska I, Tang Y, Zhang Z, Toiber D, Thanaraj TA, Soreq H. Function of alternative splicing. Gene. 2005;344:1–20. [PubMed]
24. Li Q, Lee JA, Black DL. Neuronal regulation of alternative pre-mRNA splicing. Nat Rev Neurosci. 2007;8:819–831. [PubMed]
25. Maggipinto M, Rabiner C, Kidd GJ, Hawkins AJ, Smith R, Barbarese E. Increased expression of the MBP mRNA binding protein HnRNP A2 during oligodendrocyte differentiation. J Neurosci Res. 2004;75(5):614–623. [PubMed]
26. Brumwell C, Antolik C, Carson JH, Barbarese E. Intracellular trafficking of HnRNP A2 in oligodendrocytes. Exp Cell Res. 2002;279(2):310–320. [PubMed]
27. Gama-Carvalho M, Carmo-Fonseca M. The rules and roles of nucleocytoplasmic shuttling proteins. FEBS Lett. 2001;498(2–3):157–163. [PubMed]
28. Kanai Y, Dohmae N, Hirokawa N. Kinesin transports RNA: Isolation and characterization of an RNA-transporting granule. Neuron. 2004;43(4):513–525. [PubMed]
29. Shan JG, Moran-Jones K, Munro TP, Kidd GJ, Winzor DJ, Hoek KS, Smith B. Binding of an RNA trafficking response element to heterogeneous nuclear ribonucleoproteins A1 and A2. J Biol Chem. 2000;275(49):38286–38295. [PubMed]
30. Moran-Jones K, Wayman L, Kennedy DD, Reddel RR, Sara S, Snee MJ, Smith R. hnRNP A2, a potential ssDNA/RNA molecular adapter at the telomere. Nucleic Acids Res. 2005;33(2):486–496. [PMC free article] [PubMed]
31. Dreyfuss G, Kim VN, Kataoka N. Messenger-RNA-binding proteins and the messages they carry. Nat Rev Mol Cell Biol. 2002;3(3):195–205. [PubMed]
32. Kosturko LD, Maggipinto MJ, D’Sa C, Carson JH, Barbarese E. The microtubule- associated protein tumor overexpressed gene binds to the RNA trafficking protein heterogeneous nuclear ribonucleoprotein A2. Mol Biol Cell. 2005;16:1938–1947. [PMC free article] [PubMed]
33. Huynh JR, Munro TP, Smith-Litière K, Lepesant JA, St Johnston D. The Drosophila hnRNPA/B homolog, Hrp48, is specifically required for a distinct step in osk mRNA localization. Dev Cell. 2004;6(5):625–35. [PubMed]
34. Ma ASW, Moran-Jones K, Shan J, Munro TP, Snee MJ, Hoek KS, Smith R. hnRNP A3, a novel RNA trafficking response element binding protein. J Biol Chem. 2002;277:18010–18020. [PubMed]
35. Winer BJ, Brown DR, Michels KM. Statistical Principles in Experimental Design. 3. USA: Mc-Graw-Hill; 1991.
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...