| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
REPORT |
1 Department of Medicine, Division of Cardiology, University of Colorado Health Sciences Center, Denver, Colorado 80262, USA
2 Department of Pharmacology, University of Colorado Health Sciences Center, Denver, Colorado 80262, USA
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
Keywords: FRET; ARE; HuR; AUF1; protein–protein interaction
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
; Parker and Song 2004; Moore 2005). Each mRNA transcript has several proteins that bind to its 5'-cap structure and to its 3'-untranslated region (3'-UTR), while other proteins bind to specific sequence motifs on the mature mRNA transcript and perform tasks such as delineating exon junctions, facilitating the translocation of the mRNA from the nucleus to the cytoplasm and to specific intracellular sites for translational silencing or for degradation (Kuersten and Goodwin 2005). Of particular interest is the class of A+U-rich (ARE) mRNA binding proteins involved in regulating mRNA turnover. Recent estimates indicate that 5%–8% of all genes (
4000) contain AREs (Bakheet et al. 2006) and are thus subject to rapid decay rates (Chen et al. 1995; Yang et al. 2003).
A number of ARE binding proteins have been identified. Among these is HuR (Ma et al. 1996), a member of the ELAV/Hu family (Robinow and White 1988), which contains the closely related proteins, HuB, HuC, and HuD. Under unstimulated conditions, HuR is primarily localized to the nucleus, but its movement to the cytoplasm has been associated with stabilization of mRNA transcripts (Atasoy et al. 1998; Fan and Steitz 1998a,b). HuR itself has also been shown to be regulated at the protein level (Martinez-Chantar et al. 2006), with changes in expression correlating with a growth phenotype (Blaxall et al. 2000a; Gouble and Morello 2000) and with changes in phosphorylation affecting localization and function (Abdelmohsen et al. 2007; Doller et al. 2007). In contrast to HuR, a number of other proteins, e.g., TTP, KSRP, and BRF1, confer instability to ARE-containing mRNAs (Taylor and Blackshear 1995; Taylor et al. 1996; Carballo et al. 2000; Ming et al. 2001; Stoecklin et al. 2002; Gherzi et al. 2004; Lu and Schneider 2004; Raineri et al. 2004; Schmidlin et al. 2004; Linker et al. 2005; Ogilvie et al. 2005). Interestingly, the family of AUF1/hnRNPD mRNA binding proteins appears to be able to confer either stability or instability to target mRNAs, the effect being cell type and AUF1 isoform-dependent (Laroia and Schneider 2002). The AUF1 gene is expressed as four isoforms designated by their molecular weights as p37, p40, p42, and p45, which are generated by alternative splicing of a single pre-mRNA transcript (Wagner et al. 1998). Common to all four isoforms of AUF1 is an N-terminal, alanine-rich stretch of 28 amino acids putatively required for dimerization of the isoforms in solution and a C-terminal glutamine-rich domain required for high-affinity ARE binding (DeMaria et al. 1997). All four isoforms of AUF1 are predominantly nuclear (Sarkar et al. 2003a) and are able to shuttle between the nucleus and cytoplasm in response to environmental stimuli (Kawamura et al. 2002; Sarkar et al. 2003a; Suzuki et al. 2005).
In this report, we focus on the ARE binding proteins, HuR and AUF1, using a combined cell biological and biophysical approach. Previous reports describing HuR and AUF1 have relied on biochemical techniques that allow for the elucidation of protein–protein and protein–RNA interactions by immunoprecipitation and UV cross-linking studies. However, these biochemical techniques are unable to address the subcellular localization and interactions of ARE binding proteins and RNA in real time. Further, immunoprecipitation colocalization studies are inherently limited in terms of absolute resolution. In contrast, fluorescence microscopy and immunocytochemistry, when combined with fluorescence resonance energy transfer (FRET), allows for the detection of stably associated proteins in live cells in real time (Gordon et al. 1998; Vogel et al. 2006). Using these techniques, we were able to detect protein–protein interactions for p37AUF1 with itself, HuR with itself, and HuR with p37AUF1, in both the nucleus and cytoplasm in real time. We also show that the association of HuR with AUF1 is responsive to the intracellular environment and is subject to conditions of cellular stress such that the association between HuR and AUF1 can be disrupted resulting in the proteins being relocalized to different subcellular compartments within the cytoplasm.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
Empirically corrected CFP and YFP spectral bleed-through values were used to calculate FRET intensity using a tandem CFP:YFP construct (Galperin and Sorkin 2003), which served as a positive control. By definition, a one-to-one molar ratio of CFP to YFP yields the desired 1:10 signal intensity ratio of CFP:YFP. Figure 1 shows a representative image of a DDT1-MF2 smooth muscle cell transfected with the tandem CFP:YFP construct. Figure 1A depicts the individual channel signals as well as the merged CFP/YFP channel and FRETc signal. As expected, the tandem CFP:YFP protein is expressed uniformly throughout the cell. Due to the spherical nature of DDT1-MF2 cells, an architecture particularly evident in live cells, the FRETc image is of higher signal intensity at the center and of lower intensity at the cell's periphery. Mean CFP, YFP, and raw FRET intensity values collected from identically sized masks (1.7 µm in diameter) across the entire cell were used to calculate two different FRET indices, FRETc and FRETn, and the theoretical FRET efficiency, Et, (Fig. 1B–D). When FRETc is plotted versus CFP, a linear relationship independent of CFP intensity is observed. A linear relationship is also obtained if FRETc is plotted versus the mean YFP intensity (data not shown). Although FRETn can be a highly useful means of FRET correction, because of the nonlinear relationship between the abundance of donors or acceptors and FRETn (Vogel et al. 2006), we chose not to use this method of correction. Whereas FRETc and FRETn values are specific to each experimental setup, the FRET efficiency, E, allows direct comparison between experimental setups for the same cell type and fluorescent protein FRET pairs (Vogel et al. 2006). The theoretical FRET efficiency (percent) for the tandem CFP:YFP construct, which is the maximal FRET efficiency obtainable for this fluorescent protein pair in DDT1-MF2 cells, is 0.22 (Table 1). The FRET efficiencies for HuR/HuR, p37AUF1/p37AUF1, and HuR/p37AUF1, by definition, will be lower than the theoretical maximum obtained for the tandem CFP:YFP protein. Of the three interactions measured (see below), the HuR homodimer exhibited the highest FRET efficiency with E=0.11. The efficiency of the p37AUF1 homodimer was somewhat less with E=0.07. Finally, the HuR/p37AUF1 heterodimer exhibited the lowest efficiency with E=0.04 (Table 1).
|
|
). This leads to the possibility of HuR and AUF1 binding to target mRNAs simultaneously and the possibility of direct protein–protein interaction for HuR and AUF1. Therefore, we wished to examine the potential for protein–protein interaction of HuR and p37AUF1 using FRET in both fixed cells and in live cells in real time.
In previous studies, AUF1 has been shown to form homomultimers (DeMaria and Brewer 1996; DeMaria et al. 1997; Wilson et al. 1999). However, the multimeric state of AUF1 in the nucleus and cytoplasm has not been determined in vivo. Figure 2A shows the intracellular localization of p37AUF1 proteins labeled with either eYFP or eCFP and cotransfected into DDT1-MF2 cells. In unstimulated cells, and at times of <4 h, p37AUF1 protein was present almost exclusively in the nucleus. The formation of higher-order stable complexes between the two fluorescently labeled monomers of p37AUF1 is indicated by the strongly positive FRETc signal.
|
Given that both HuR and AUF1 may be part of a larger RNA mRNA binding protein complex, we wished to examine the possibility that there was direct protein–protein interaction between the two proteins in both nuclear and cytoplasmic compartments. Since HuR is typically found associated with the ribosomal and higher molecular weight polysome fractions, compartments that typically lack AUF1 (Pende et al. 1996; Lal et al. 2004), the possibility existed that the association of HuR and AUF1 might be location specific. Figure 2C shows a cell transiently cotransfected with both HuR-eYFP and p37AUF1-eCFP. In unstimulated cells, HuR and p37AUF1 colocalize to the nucleus. As indicated by the robust FRETc signal, HuR and p37AUF1 appear to form stable heterodimers.
To verify the specificity of these protein–protein interactions, additional FRET studies were performed with proteins not expected to interact with HuR. To this end, eCFP versions of the cytoplasmically expressed protein, PK-A R11A, and the nuclear and cytoplasmically localized protein, PH-PL-C, were analyzed. In each case, no evidence of FRET pairing with HuR-eYFP was evident (data not shown).
We have previously shown that anisomycin, a MAPK (p38, JNK) activator, increases the half-life of the 
2-adrenergic receptor (AR) mRNA, an ARE-containing mRNA endogenously expressed in DDT1-MF2 cells (Headley et al. 2004). Thus, the 2-AR is similar to other ARE-containing mRNAs in that MAPK activation prolongs their half-lives (Mahtani et al. 2001; Dean et al. 2003). Further, the movement of HuR from nucleus to cytoplasm has been correlated with its ability to stabilize ARE-containing mRNAs (Atasoy et al. 1998). The anisomycin-dependent increase in the half-life of ARE-containing mRNAs could be due to increased binding of stabilizing factors, such as HuR, a decrease in the binding of destabilizing factors, such as AUF1, or a combination of both. Therefore, we wished to determine the effect of anisomycin on the intracellular localization of HuR and AUF1 in live cells.
To explore HuR shuttling at the level of higher-order associations, FRET was used to assess the interaction of HuR with itself once it has transited from nucleus to cytoplasm. As demonstrated in Figure 2B (bottom row), movement of HuR from the nucleus to the cytoplasm does not alter its ability to form homodimers. In contrast to the rapid shuttling observed for HuR, anisomycin-induced shuttling of p37AUF1 is a prolonged process that can take in excess of 24 h (Fig. 2A, bottom row). Like HuR, and as indicated by the FRETc signal, p37AUF1 maintains its ability to form homodimers in the cytoplasm. Similarly, anisomycin stimulated shuttling of AUF1 and HuR does not alter their heterodimeric association (Fig. 2C, bottom row).
These FRETc images indicate that HuR and AUF1 are in sufficient proximity to functionally associate in both the nucleus and the cytoplasm. Interestingly, the length of time required for anisomycin-induced shuttling of p37AUF1 appears greatly different when p37AUF1 is present as a complex with HuR (cf. Fig. 2C, bottom row, and Fig. 2A, bottom row). Although speculative, it is possible that the formation of a stable complex of p37AUF1 with HuR permits AUF1 to shuttle out of the nucleus on a shorter "HuR" time scale, when stimulated by anisomycin. This remains to be demonstrated.
To verify that the images obtained for AUF1 and HuR in live cells could be recapitulated in fixed cells, a parallel set of experiments was performed. Figure 3, A and B, shows the intracellular localization of HuR-eYFP and p37AUF1-eCFP in fixed DDT-MF2 cells. Under control conditions, HuR and p37AUF1 remain colocalized within the nucleus, producing a high-intensity FRET signal. A graph depicting the intensity of each fluorescent protein and their associated FRET signals is presented to the right of each panel of images. The line intensity graph demonstrates clearly that in the control, untreated cells, there is no movement of HuR or p37AUF1 from the nucleus to the cytoplasm. However, following 24 h of anisomycin treatment, a substantial proportion of p37AUF1-eCFP and HuR-eYFP shuttled from nucleus to cytoplasm. A FRETc signal is present in the nucleus and throughout the entire cytoplasm with the highest signal intensity being found closest to the nucleus. Whether the decrease in FRET signal intensity further away from the nucleus is due to dissociation of HuR from p37AUF1 or due merely to lower concentrations of HuR/p37AUF1 complexes is not known. As has been noted previously, both HuR and p37AUF1 are excluded from the nucleoli of the nucleus (Fan and Steitz 1998a; Fawal et al. 2006). This can be seen clearly from the line graphs, which indicate that the concentration of both HuR and p37AUF1 in the nucleus is nonhomogeneous. The concordance of the results presented in Figures 3 and 2 demonstrates that data obtained from fixed cells can be an accurate representation of the live cell conditions in DDT1-MF2 cells.
|
|
|
). However, data are emerging indicating that ARE mRNA binding protein complexes are both large and dynamic (Tenenbaum et al. 2002; Mazan-Mamczarz et al. 2003; Lal et al. 2004; Keene and Lager 2005). Under conditions of stress, discrete cytoplasmic entities composed of high concentrations of proteins and RNA have been identified (Abraham et al. 2002; Kedersha and Anderson 2002; Cougot et al. 2004; Brengues et al. 2005; Kedersha et al. 2005; Anderson and Kedersha 2006). These "cytoplasmic" or "stress" granules have been shown to be composed of proteins involved in mRNA translation initiation, translation control, and mRNA decay (Newbury et al. 2006).
We were interested in determining the localization and interaction of HuR with AUF1 during times of cellular stress. Figure 6 shows immunocytochemical detection of several mRNA binding proteins in DDT1-MF2 cells in response to sodium arsenite. TIA-1 was used as a marker of stress granules (Kedersha et al. 1999, 2000, 2005). The left-hand column of Figure 6 shows the individual channel data at time zero prior to arsenite treatment. As noted previously, endogenous AUF1 and HuR are present mainly in the nucleus, whereas TIA-1 is present in both the nucleus and to some degree in the cytoplasm. In the absence of arsenite treatment, no stress granules are detected (left column). However, 30 min after arsenite treatment, TIA-1-positive stress granules are readily detected (Fig. 6, center column). Arsenite-induced stress also causes HuR to move into stress granules, and the colocalization of HuR and TIA-1 to the stress granules is readily apparent. The observed stress-induced movement of HuR to cytoplasmic granules is consistent with previous results using other types of stressors (Kedersha et al. 2000, 2005; Kedersha and Anderson 2002). In contrast, arsenite treatment had no obvious effect on the intracellular localization of AUF1.
|
mRNA (Fig. 7, top), and to the more complex ARE 231 nucleotides from the 2-AR mRNA (Fig. 7, bottom). As has been shown previously, the banding pattern of the gel shift assays indicates that the size and composition of the protein–mRNA complex formed is dependent on the amount of protein present (Wilson et al. 1999), as both HuR and p37AUF1 are able to form multiple complexes with a single ARE. Competition of p37AUF1 with a preformed HuR–mRNA complex had the potential to yield several possible results. If p37AUF1 displaced HuR, then no substantial change in the relative migration of bands should be observed, as HuR and p37AUF1 are relatively close in molecular weight. However, if p37AUF1 is able to bind to a preformed HuR–mRNA complex, then larger, more slowly migrating bands should form. Figure 7 shows that for both TNF and the 2-AR AREs, the addition of GST-p37AUF1 (or His6-p37AUF1, data not shown) leads to larger protein–mRNA complexes, indicating that both HuR and p37AUF1 are complexing with a single RNA. Although the data do not directly support the conclusion that the two proteins are competing for individual binding sites on the ARE, the resolution of a gel shift assay with a large complex ARE is likely insufficient to eliminate this possibility.
|
demonstrates that p37AUF1 is able to bind to the ARE simultaneously with HuR and form a single complex. Due to the size of this mRNA molecule, 53 nt, the possibility exists for one stable nucleoprotein complex. This result, however, is not able to distinguish between the possible order and location of binding of HuR and p37AUF1 to the ARE. For the 2-AR ARE, a greater number of combinations of protein and mRNA are possible. However, only two ribonuclear protein complexes are formed at the highest protein concentrations. These results clearly demonstrate that HuR and AUF1 can bind additively to the AREs tested and presumably have the potential to do so on other ARE containing RNAs.
RNA dependence of the ARE binding protein interaction
To address whether the stable, cytoplasmic interaction of HuR and p37AUF1 that we have shown in Figures 2 and 3 is RNA dependent, DDT1-MF2 cells were transiently transfected with HuR-eYFP and p37AUF1-eCFP, incubated for 4 h with anisomycin, and treated with ribonuclease A prior to fixation (Maison et al. 2002). Figure 8 demonstrates that there is no qualitative difference in the HuR/p37AUF1 interaction, as indicated by the FRETc signal, between mock treated and cells treated with ribonuclease A. To further verify this result, the FRET efficiency value, E, was calculated for the HuR/p37AUF1 interaction in mock- versus ribonuclease A treated cells. The E value, calculated from a sampling of >100 individual cells, was 0.04, a value identical to that found for the HuR/p37AUF1 heterodimer in live and fixed cells. This result supports the conclusion that at least in the whole cell context, RNA binding is not an absolute requirement for the AUF1/HuR protein–protein interaction.
|
support the ability of AUF1 and HuR to directly compete for binding but only at select AREs. For both the TNF 53mer and the 2-AR 231mer AREs, the data also do not permit us to distinguish between scenarios A and C (Fig. 9). What is apparent, however, is that a small amount of a higher molecular weight complex is formed in both competition experiments. Thus, scenario D may occur if, after the binding sites on the ARE are saturated, proteins are then able to bind to themselves to form even larger mRNA–protein–protein complexes. This result has previously been shown for AUF1 alone, forming a dimer in solution but a hexamer when bound to an ARE (DeMaria and Brewer 1996; DeMaria et al. 1997; Lal et al. 2004).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
FRET is a powerful biophysical technique that is increasingly being used to visualize cellular processes in vivo (Galperin et al. 2004; Vogel et al. 2006). Intrinsically, FRET is dependent on both the absolute distance between, and the correct orientation of, two fluorophores that constitute defined FRET pairs (Forster 1948). The strength of the FRET signal obtained from a pair of interacting, fluorescently labeled proteins is also dependent on the FRET efficiency, E, which is the maximum energy transfer from the FRET donor to acceptor.
To attempt to quantify the relative efficiency of FRET between the ARE binding proteins p37AUF1 and HuR, the maximum theoretical FRET efficiency, Et, was first obtained by using a tandem CFP:YFP protein construct (Galperin and Sorkin 2003). Under these conditions, the maximum efficiency obtained in our live cell system was 22%. In the same system, HuR homodimers, p37AUF1 homodimers, and p37AUF1/HuR heterodimers yielded Ed values of 11%, 7%, and 4% respectively, Table 1. These Ed values are in the range measured for other protein pairs in other cellular systems (Galperin et al. 2004). The calculated Ed values for HuR and p37AUF1 show that both HuR and p37AUF1 have the potential to form strong homodimers in the cell (Figs. 2, 3). The lesser Ed for the HuR/AUF1 heterodimer may be due to any number of factors. One possibility is that the formation of HuR and p37AUF1 homodimers sequesters HuR and p37AUF1, thereby preventing the formation of p37AUF1/HuR heterodimers. If so, this would result in an artifactual reduction in the FRET signal. It is also possible that mRNA-dependent binding of HuR and p37AUF1 imposes steric factors onto the respective proteins such that the protein–RNA interaction(s) precludes or diminishes the FRET detectable interactions. Alternatively, and perhaps most simply, the possibility exists that the heteromeric interactions between p37AUF1 and HuR are weaker than the respective homomeric interactions. In support of this later interpretation is the fact that Ed is independent of the concentration(s) of fluors (and thus is likely not an artifact of differential expression), and the fact that FRET efficiency can under certain circumstances be correlated directly with the affinity of protein–protein interaction (Chilibeck et al. 2006).
Results obtained from nondenaturing electrophoresis experiments (Fig. 7) indicate that biochemical associations between HuR and p37AUF1 are indeed observable. For both p37AUF1 and HuR, binding either to the TNF 53mer ARE or the 2-AR 231mer ARE, homomeric higher-order structures are observed. These results are concordant with our previous observations and with those of others (Wilson et al. 1999; Blaxall et al. 2000b; Blaxall et al. 2002). In competitive binding experiments where p37AUF1 is added to mRNA pre-bound to HuR, rather than displacing HuR from either ARE, p37AUF1 appears to further shift the mRNA–protein complex. These data provide strong evidence that for both a simple ARE, such as that of TNF 53mer, or for a long, complex ARE, that of the 2-AR 231mer, co-residence of both proteins on an ARE can occur. What nondenaturing gel shift assays cannot reveal is the RNA dependence of these protein–protein interactions. In this regard, it is well established that AUF1 can form RNA-independent homomultimers (DeMaria et al. 1997; Wilson et al. 1999). HuR also appears to homomultimerize (Soller and White 2005), the RNA dependence of that interaction being unknown. Conversely, coexistence of HuR and AUF1 on AREs appears to be dependent on the specific ARE. What remained unclear was the RNA dependence of these interactions. As demonstrated by Lal et al. (2004), coimmunoprecipitation of HuR and AUF1 does not occur when the protein–RNA complex is subjected to RNase pretreatment. In contrast, in the current manuscript, we demonstrate that HuR/AUF1 interaction (Fig. 8) can be detected by FRET even after RNase treatment. These findings are not incongruous as the preparations and conditions are entirely different, one being in the whole cell context and the other being in a purified biochemical preparation. Regardless, the dependence or independence of protein–protein interaction on RNA binding in no way diminishes or precludes the possibility of functionally relevant interactions between the two proteins, both direct or indirect, e.g., by RNA remodeling affecting the binding of one or the other protein (Wilson et al. 2001a,b).
As detailed in Figure 9, there are several ways in which AUF1 and HuR might compete for binding to, or co-reside on, a single ARE. Each of the AREs examined herein, as well as other AREs we have examined previously (Blaxall et al. 2000b, 2002), are capable of binding both AUF1 and HuR with high affinity. However, certain AREs appear to bind only AUF1, others bind only HuR, whereas still others bind either AUF1 or HuR (Lal et al. 2004). Other studies have clearly demonstrated that binding of HuR to the AREs of a broad spectrum of mRNAs is associated with their stabilization (Fan and Steitz 1998b; Peng et al. 1998; Dean et al. 2001; Figueroa et al. 2003; Raineri et al. 2004). Although somewhat less consistent, and perhaps splice-form dependent, there is a general association between AUF1 binding and mRNA destabilization (Buzby et al. 1996; Sirenko et al. 1997; Laroia et al. 1999; Sarkar et al. 2003b). What remains to be examined, especially in the case of an mRNA containing a complex ARE, is the functional significance of more than one protein simultaneously occupying the same ARE or the same mRNA on sequential AREs.
Regulation of the life cycle of an mRNA molecule is a complex process dependent on a number of variables. One level of regulation is that of the abundance of the mRNA binding proteins themselves. To this end, the functional, intracellular concentrations of HuR and AUF1 are regulated by multiple factors including external stimuli, the cell cycle, and pathophysiological processes such as cellular hypertrophy or hyperplasia (Pende et al. 1996; Lafon et al. 1998; Blaxall et al. 2000a; Gouble and Morello 2000; Wang et al. 2000a). A second level of regulation is that of the intracellular localization of the mRNA binding protein and/or the mRNA. There is well-documented evidence both in the literature and presented herein that AUF1 and HuR undergo nuclear/cytoplasmic shuttling.
One major regulatory pathway for both the binding and function of AUF1 is phosphorylation. As demonstrated by Wilson and colleagues (2003a), phosphorylation of AUF1 can markedly affect its affinity for an ARE. Phosphorylation of AUF1 can also induce structural remodeling of the ARE (Wilson et al. 2003a,b; Fialcowitz et al. 2005). Further, the peptidyl–prolyl isomerase, Pin1, binds to AUF1 (Shen et al. 2005). Isomerization by Pin1 could potentially alter the binding affinity of AUF1 by inducing conformational changes in phosphorylated forms of the protein. Like AUF1, and other mRNA-destabilizing proteins including TTP, BRF1, and KSRP, there is recent evidence to date that HuR is also a phosphoprotein. In the case of HuR, induction of cytoplasmic shuttling and increased binding to COX-2 mRNA has recently been linked to phosphorylation by PK-C (Doller et al. 2007). HuR has also been shown to be phosphorylated by the cell cycle checkpoint kinase Chk2 (Abdelmohsen et al. 2007). However, phosphorylation by Chk2 does not appear to effect HuR shuttling. Given the impact of phosphorylation state on the binding of AUF1 to an mRNA, it is reasonable to conclude that this mode of regulation might impact the interaction of AUF1 and HuR.
The persistent association of HuR and/or AUF1 with a target mRNA throughout the life of an mRNA in the cell has yet to be definitively tested. HuR has been associated with the stabilization of mRNA transcripts in the cytoplasm, and studies have been conducted on the movement of HuR from the nucleus to the cytoplasm (Atasoy et al. 1998; Fan and Steitz 1998a,b; Seko et al. 2004). However, it is not clear whether HuR binds to a target mRNA and remains bound to a particular mRNA transcript, not only as it is translocated from the nucleus to the cytoplasm, but also through its lifespan in the cytoplasm. This idea can be extended to include whether or not AUF1 also binds to one mRNA transcript for its lifespan and whether or not combinations of other destabilizing and stabilizing RBPs do so as well.
Anisomycin, by virtue of its ability to activate MAPKs, stabilizes ARE-containing mRNAs (Winzen et al. 1999; Mahtani et al. 2001; Neininger et al. 2002; Dean et al. 2003; Stoecklin et al. 2004; Brook et al. 2006; Martinez-Chantar et al. 2006), and causes rapid shuttling of HuR from nucleus to cytoplasm. Figure 5A demonstrates that the time course of shuttling of HuR (30 min) coincides with previous data on the stabilization of ARE-containing mRNAs by MAPK activation (Headley et al. 2004). Conversely, short-term exposure to anisomycin has a more modest effect on the localization of AUF1. These observations lead to the question of how RNPs move from the nucleus to the cytoplasm. The nuclear pore complex (NPC) allows passive diffusion of protein complexes under 60 kDa; larger molecules require assistance for transport. Monomeric HuR and AUF1 are between 30 and 45 kDa. However, when they are bound to an ARE as multimers, the size of an ABP-bound RNA complex necessitates assistance for translocation from the nucleus. Studies detailing the nucleocytoplasmic shuttling of HuR and AUF1 demonstrate that two different mechanisms appear to be utilized for the shuttling of each of these proteins. HuR has been shown to use the CRM1 assisted method of shuttling (Gallouzi et al. 2001; Kimura et al. 2004; Higashino et al. 2005; Cherradi et al. 2006; Fries et al. 2006; Prechtel et al. 2006). The exact mechanism of AUF1 nucleocytoplasmic shuttling has yet to be elucidated; however, AUF1 shuttling appears to be carrier-mediated but not CRM1 associated (Sarkar et al. 2003a).
One condition under which the protein composition bound to a particular mRNA can be altered is that of cellular stress. Upon stress, such as that engendered by treating cells with arsenite, specific, well-defined granules appear. These so-called stress granules (SGs) are known to be protein- and mRNA-dense foci. The mRNA transcripts present in these SGs are those present in the cytoplasm at the onset of the stress conditions and by one mechanism or another are transported into these structures. An example of this would be the TNF mRNA, which appears to be localized to SGs when associated with the ARE binding protein, tristetraprolin (TTP) (Kedersha et al. 2005; Lian et al. 2006). Other proteins commonly found in stress granules are those involved in RNA deadenylation and in translational silencing (Kedersha et al. 2005). In our model system, we have demonstrated that HuR is present and colocalizes with TIA-1 in stress granules. In contrast, AUF1 does not appear to colocalize to SGs, at least not under conditions of arsenite induction. This result is fully consistent with results obtained in other cell types (Newbury et al. 2006; Eulalio et al. 2007). As stress granules are an accumulation of cytoplasmic mRNA putatively not required during the stress response and which are stabilized and held in translational arrest (Kedersha et al. 2000, 2005; Kedersha and Anderson 2002), the presence of a stabilizing ARE binding protein, such as HuR, would be fully expected. The lack of AUF1 in stress granules implies either that (1) a cellular mechanism exists for removing AUF1 from an mRNA transcript prior to localization to a stress granule, (2) that mRNA transcripts with AUF1 bound to them are excluded from stress granules and remain present in the cytoplasm at times of stress, or (3) AUF1-bound mRNA would be more rapidly degraded under these conditions with AUF1 returning to the nucleus. The results presented here show that sodium arsenite-induced stress does not alter the cellular AUF1 localization. However, the data presented here do not address the fate of cellular AUF1-bound mRNA transcripts at the onset of a stress response.
In conclusion, the results presented herein demonstrate that two ARE binding proteins with potentially opposite functional effects on mRNA stability can co-associate in mRNA–protein complexes. These interactions occur in both the nucleus and cytoplasmic compartments. What remain to be determined are the functional consequences of this interaction. An extension of the current findings is that the use of FRET to detect these interactions opens up distinct new possibilities in the ability to detect the interactions of multiple individual mRNA binding proteins in live cells in real time.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
Cell culture
DDT1-MF2 hamster smooth muscle cells were grown in 100 mm dishes at 37°C, 5% CO2 in Dubelcco's Modified Eagle Medium (Invitrogen) supplemented with 5% Fetal Bovine Serum (Hyclone) and 0.1% penicillin/streptomycin solution (Invitrogen). For transient transfections, cells were grown on either 18-mm or 25-mm glass coverslips to 80% confluence. For immunocytochemistry, cells were grown on 18-mm glass coverslips to 60% confluence. Where indicated, cells were treated with anisomycin (75 nM in dimethyl sulphoxide) or sodium arsenite (0.5 mM).
Transient transfection
Vectors were transfected into DDT1-MF2 cells using FuGene 6 (Roche) following the manufacturer's protocol. The amount of plasmid DNA transfected was adjusted empirically to yield the desired 1:1 molar ratio of the CFP protein construct to the YFP protein construct.
Immunocytochemistry
DDT1-MF2 cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences), permeabilized with 0.2% Triton X-100, treated with 0.1% sodium borohydride (Sigma), and blocked with 10% horse serum in 1% BSA. Cells were incubated overnight at 4°C with 1°Ab and with 2°Ab for 1 h at room temperature in the dark. Coverslips were mounted on glass slides using a 12.5% solution of Mowiol (Calbiochem) in 50% glycerol. Detection of HuR was via a 1:400 dilution of monoclonal 1°Ab (Santa Cruz Biotechnology) and a 1:2000 dilution of 2°Ab labeled with Alexa Fluor 488 (Molecular Probes); detection of AUF1 was via a 1:500 dilution of a polyclonal 1°Ab (Upstate), which recognizes all four isoforms of AUF1, and a 1:800 dilution of 2°Ab labeled with Cy3 (Jackson ImmunoResearch); detection of TIA-1 was via a 1:500 dilution of polyclonal 1°Ab (Santa Cruz Biotechnology) and a 1:400 dilution of 2°Ab labeled with Alexa Fluor 647 (Molecular Probes).
Subcellular fractionation
Cytoplasmic and nuclear fractions were produced as described by Wang et al. (2000b). Briefly, cells were rinsed with cold phosphate buffered saline followed by incubation on ice in hypotonic buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl2) containing protease inhibitors (leupeptine [1 µg/mL], aprotinin [1 µg/mL], and 0.5 mM PMSF). Cells were lysed by the addition of buffer A with protease inhibitors containing 2.5% Nonidet P-40. Lysed cells were centrifuged (3500 rpm for 4 min) to pellet nuclei. The resultant supernatant was isolated, subjected to five freeze–thaw cycles, and recentrifuged. Cytosolic fractions were prepared by centrifugation of the second supernatant (14,000 rpm for 60 min at 4°C). Nuclear pellets were incubated on ice in extraction buffer C (20 mM HEPES [pH 7.9], 0.45 M NaCl, 1 mM EDTA) with protease inhibitors followed by centrifugation (14,000 rpm for 10 min at 4°C).
Western blotting
Endogenous HuR and AUF1 proteins were detected as described previously (Blaxall et al. 2000a). Briefly, AUF1 and HuR proteins were detected using polyclonal 1° antibodies (1:1000, Upstate) and detected by a GAR 2° antibody (1:2000, Jackson ImmunoReserearch). For normalization, blots were stripped and reprobed with a mouse monoclonal anti-actin antibody (1:1000) with detection by a GAM 2° antibody (1:4000, Jackson ImmunoResearch).
Cell fixation
DDT1-MF2 cells were fixed with 3.4% formaldehyde. The coverslips were mounted on glass slides using a 12.5% solution of Mowiol (Calbiochem) in 50% glycerol.
Microscopy
Images of live and fixed cells were captured at 100x magnification under oil immersion, in 2x2 binning mode using a Nikon Eclipse TE3000 fluorescence microscope running Slidebook software (3I, version 4.0.1.43), or a Zeiss Axiovert 200M fluorescence microscope, fitted with an atmosphere controlled chamber, running Slidebook software, (3I, version 4.1.0.12). All images were collected using the following criteria: integration time >25 msec, channel intensity <4000. All images shown are background corrected by subtracting the intensity of a cell devoid area from each individual image. For images containing CFP- and YFP-labeled proteins, integration time was between 25 and 250 msec. Cells containing CFP- and YFP-labeled proteins were viewed via the CFP channel to prevent photobleaching of YFP. All images presented have a signal intensity >100 after background correction. FRETc images were produced by Slidebook using spectral bleed-through numbers determined experimentally. FRETc data was calculated using the raw intensity values obtained from Slidebook.
Spectral bleed-through number calculations
The following constructs were used to determine the average CFP spectral bleed-through number: eCFP vector, p37AUF1-eCFP, HuR-eCFP, TTP-eCFP and KSRP-eCFP. To determine the YFP spectral bleed-through number, the following vector constructs were used: eYFP vector, HuR-eYFP, and BRF1-eYFP. As a further proof of the bleed-through numbers, transient transfection of non-FRET pairs was used for empirical correction of the spectral bleed-through number; eCFP vector and eYFP vector; HuR-eYFP and eCFP vector; p37AUF1-eCFP and eYFP vector; HuR-eYFP with both an unrelated cytoplasmically localized protein (PKA R11A) and an unrelated nuclear and cytoplasmically localized protein (PH-PLC), labeled with CFP. Both unrelated protein vector constructs were kindly provided by Dr. Mark Dell'Acqua, University of Colorado.
Calculation of FRETc, FRETn, and FRET efficiency E:
To calculate FRETc, FRETn, and the FRET efficiency, E, the following equations were used:
|
| (1) |
FRETc is the corrected FRET value, raw FRET is the uncorrected FRET intensity, D is the donor fluorophore intensity, X is the donor fluorophore spectral bleed-through number, A is the acceptor fluorophore intensity, and Y is the acceptor fluorophore spectral bleed-through number.
|
| (2) |
FRETn is the normalized FRET value, FRETc is the corrected FRET value, D is the donor fluorophore intensity, and A is the acceptor fluorophore intensity.
|
| (3) |
E is the FRET efficiency, FRETc is the corrected FRET value, D is the donor fluorophore intensity, and G is a constant determined for the specific filter set used (Gordon et al. 1998; Galperin et al. 2004). To calculate the FRETc, FRETn, and E calculations, the following criteria were used: YFP:CFP ratio between 9 and 11, CFP intensity >100.
Production of ARE-containing mRNAs
PCR amplification of a 231 nt section of the 3' -UTR of the hamster 2-AR gene using the following primers (5'-CCGCTCGAGCAGGCTTT; 3'-GATCTCTAGATTTAAGTT, both shown in the 5'-3' direction) was cloned into the pCDNA3.1 zeo+ vector. This construct was used to make the 231 nt ARE containing mRNA fragment utilized for the nondenaturing gel shift assays. The production of the TNF 53 nt mRNA was as described (Stoecklin et al. 2001).
Production of recombinant GST-p37AUF1 protein
The coding region of p37AUF1 (Pende et al. 1996) was inserted into a pGEX-2T plasmid (Amersham Biosciences). The resulting pGEX-2T-AUF1 plasmid was transformed into Escherichia coli BL21 cells, and protein expression was induced with 1 mM IPTG at 37°C for 3–4 h. Cells were centrifuged at 4000 rpm for 10 min, 4°C, and resuspended in 5 mL of Buffer A (50 mM Tris pH 8.0, 200 mM NaCl, 1 mM EDTA). When fully resuspended, 500 µL of 10 mg/mL lysozyme was added. The cell suspension was incubated on ice for 20 min followed by centrifugation at 13,000 rpm, 30 min, 4°C. The supernatant was incubated with 1 mL of glutathione sepharose beads slurry (50% slurry) for 1 h at 40°C with agitation. The beads were washed twice with Buffer B (50 mM Tris-HCl pH 8, 200 mM NaCl, 1 mM EDTA, and 1% Triton X-100), and fractions were eluted with Elution Buffer (50 mM Tris-HCl [pH 8], 5 mM Glutathione). Human alpha-lactalbumin (Sigma) was added to a final concentration of 100 µg/mL prior to storage at –80°C. Production of recombinant GST-HuR and His6-p37AUF1 has been described previously (Blaxall et al. 2002).
Nondenaturing gel shift assays
Briefly, 1 fmol of 32P-labeled, in vitro transcribed RNA (5000 CPM) (Pende et al. 1996) was incubated with increasing concentrations of purified recombinant protein in a final volume of 10 µL of binding buffer (10 mM Tris-HCl [pH 7.5], 5 mM magnesium acetate, 100 mM potassium acetate, 2 mM DTT, 0.1 µg/µL of BSA, 10 units of RNasin [Promega], 0.2 mg/mL yeast tRNA, 5 µg/µL heparin, and 0.1 µg/µL poly[C]) for 10 min on ice. Complexes were resolved by electrophoresis through nondenaturing, 5%–6%, polyacrylamide gels and visualized by autoradiography.
In vivo RNA digestion
DDT1-MF2 cells grown on 18 mm glass coverslips were transiently transfected as above. Cells were permeabilized with 0.01% Triton X-100, treated with 1 mg/mL ribonuclease A (Sigma) for 10 min at room temperature, and fixed with 4% paraformaldehyde (Maison et al. 2002). Coverslips were mounted on glass slides using a 12.5% solution of Mowiol in 50% glycerol.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
probe template and Dr. Mark Dell'Acqua for the PKA(R11A) and PH-PLC CFP labeled constructs. We would also like to thank Drs. Alexander Sorkin, Emilia Galperin, and Mark Dell'Acqua for many helpful discussions pertaining to FRET microscopy. |
Footnotes |
|---|
Article published online ahead of print. Article and publication date are at http://www.rnajournal.org/cgi/doi/10.1261/rna.501707.
Received February 8, 2007; accepted May 29, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSACKNOWLEDGMENTSREFERENCES |
|---|
Abdelmohsen, K., Pullmann Jr, R., Lal, A., Kim, H.H., Galban, S., Yang, X., Blethrow, J.D., Walker, M., Shubert, J., Gillespie, D.A., et al. 2007. Phosphorylation of HuR by Chk2 regulates SIRT1 expression. Mol. Cell 25: 543–557.[CrossRef][Medline]
Abraham, W.T., Gilbert, E.M., Lowes, B.D., Minobe, W.A., Larrabee, P., Roden, R.L., Dutcher, D., Sederberg, J., Lindenfeld, J.A., Wolfel, E.E., et al. 2002. Coordinate changes in Myosin heavy chain isoform gene expression are selectively associated with alterations in dilated cardiomyopathy phenotype. Mol. Med. 8: 750–760.[Medline]
Anderson, P. and Kedersha, N. 2006. RNA granules. J. Cell Biol. 172: 803–808.
Atasoy, U., Watson, J., Patel, D., and Keene, J.D. 1998. ELAV protein HuA
