Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Asthma is characterized by an infiltration of T cells, neutrophils, and eosinophils into the bronchoalveolar space, associated with airway edema, mucus production, tissue damage, and ultimately fibrosis. The elaboration of cytokines and chemokines by inflammatory cells likely plays a significant role in the pathobiology of asthma (1, 2). In particular, interleukin (IL)-5 and granulocyte macrophage colony-stimulating factor (GM-CSF) induce eosinophilopoiesis and migration from the peripheral blood into the lung (3, 4). These mediators also enhance eosinophil function and survival both in vivo and in vitro (5, 6).

Despite the importance of cytokines in eosinophil biology, little is known about the underlying molecular mechanisms that control their elaboration. The expression of many cytokines, including GM-CSF, is quenched by the extremely rapid decay of their coding messenger RNAs (mRNAs). Upon mitogen treatment, T cells stabilize GM-CSF mRNAs, accounting for most, if not all, of their cytoplasmic accumulation and ensuing secretion (7). Both rapid decay and mitogen-driven stabilization require the presence of 3' untranslated region (3' UTR) adenosine-uridine-rich elements (AREs) (8, 9). Alteration or deletion of the AREs significantly enhanced the stability of mutant GM-CSF mRNAs (10, 11), whereas chimeric globin mRNAs fused to the AREs from GM-CSF decayed at rates similar to wild-type GM-CSF (8).

The recognition of ARE-containing mRNAs likely involves sequence-specific binding proteins. Indeed, several labs have identified such activities in tumor cell lines (12, 13), normal lymphocytes (14), fibroblasts (15), and macrophages (16). Establishing a physiologic role for these proteins has proven more difficult, although antisense-mediated depletion of HuR (an AUUUA-specific RNA binding protein) was associated with destabilization of vascular endothelial growth factor (VEGF) mRNA (17). Thus, it has been proposed that cytokine mRNA stabilization likely occurs when sequence-specific binding proteins interact with, and possibly mask, the AREs from cellular ribonucleases (14).

Therefore, we investigated whether GM-CSF mRNA was regulated by variable decay in an eosinophil-like cell line, AML14.3D10. This cell line displays typical eosinophil morphology and enzymatic activities and thus appears to be an excellent surrogate for normal eosinophils (18). Transfecting full-length GM-CSF mRNA by particle-mediated gene transfer (PMGT), we showed that calcium ionophore inhibits GM-CSF mRNA degradation. Stabilization required intact AREs because a mutant GM-CSF mRNA with AUGUA repeats in place of wild-type AUUUA repeats was nonresponsive to ionophore activation. Interestingly, mutant AUGUA containing GM-CSF mRNA was still quite unstable, suggesting the presence of previously uncharacterized, additional instability elements. Finally, using gel mobility shift analysis, we identified several cytoplasmic mRNA binding proteins specific for the ARE. These binding protein activities were increased in ionophore-treated AML14.3D10 cells concomitant with GM-CSF mRNA stabilization. These observations suggest that several mRNA binding proteins may be involved in stabilizing GM-CSF mRNA in AML14.3D10 cells upon ionophore-mediated activation.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture

AML14.3D10 cells were grown in 150-cm² flasks (Fisher Scientific, Hanover Park, IL) in a 5% CO₂ atmosphere in RPMI 1640 media containing 10% fetal calf serum, supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate, 0.05 mg/ml gentamycin, and 5.5 × 10⁵ M 2-mercaptoethanol. Cells were stimulated with 1 mm ionomycin (Sigma, St. Louis, MO) at a density of 7 × 10⁵ cells per well in 24-well dishes (Corning) for up to 2 h.

PMGT

PMGT of in vitro-transcribed mRNAs into AML14.3D10 cells was performed using the Accell gene gun, as previously described (9). Briefly, capped, adenylated wild-type or mutant (containing AUGUA repeats in place of AUUUA repeats) GM-CSF mRNAs were precipitated at 20°C for 1 h with 1 vol of 2-propanol and 0.10 vol of 5 M ammonium acetate onto gold beads (0.96 µm diameter) at a concentration of 5 µg mRNA/mg of gold beads. A total of 80 to 95% of the inputted mRNA was typically loaded onto the beads. Successive transfections of 2 × 10⁶ cells were pooled and washed twice in culture medium to remove any extracellular mRNAs. Transfected cells were placed in culture at 1 × 10⁷ cells/ml. Some cells were activated with 1 µm ionomycin for up to 2 h before transfection. Whenever activated cells were used, calcium ionophore was maintained at 1 µm throughout the experiment.

mRNA Isolation and Northern Blotting

At the indicated time points, cells were lysed by mixing with 1 ml of TRIreagent (Molecular Research Center, Inc., Cincinnati, OH) and snap-frozen at 80°C. After all time points were taken, total RNA was quantitatively isolated as per the instructions of the manufacturer. RNA samples from each time point were separated by size on denaturing formaldehyde, 1% agarose gels. Ethidium bromide (5 µg) was added to each sample to visualize 28S and 18S ribosomal RNA bands in order to ensure the integrity of the isolated RNA. RNA samples were then transferred to nylon membranes (Magna; Fisher) by vacuum transfer as described by the manufacturer (Pharmacia-LKB Biotechnology, Inc., Piscataway, NJ) and fixed by baking for 30 min at 65°C. Membranes were prehybridized in Quick-hyb (Stratagene) for 1 h at 68°C and then hybridized for 120 min with 2 to 3 × 10⁶ cpm/ml of random-primed cDNA probes (Decaprime kit; Ambion). All probes were labeled to a specific activity of > 10⁹ cpm/µg. Blots were washed twice for 15 min at room temperature with 2× saline sodium citrate (SSC)/0.1% sodium dodecyl sulfate (SDS) and once at 60°C with 0.1× SSC/0.1% SDS for 5 to 15 min before autoradiography or phosphorimaging. Autoradiograms were quantitated by phosphorimaging and GM-CSF-specific signals normalized to signals for glyceraldehyde-3-phosphate dehycrogenase (GADPH) and plotted as a function of time.

Cell Lysis

After incubation with ionophore, AML14.3D10 cells were washed twice with ice-cold phosphate-buffered saline and lysed (25 µl lysis buffer/7 × 10⁵ cells) with a buffer containing 25 mM Tris (pH 8.0), 0.1 mM ethylenediaminetetraacetic acid, 200 µg/ml Pefabloc (Fisher Scientific), and 0.5% Nonidet P-40 (Sigma). Cells were incubated at 4°C for 20 min and occasionally vortexed, followed by centrifugation at 12,000 × g for 15 min at 4°C. The supernatant (crude cytoplasm) was carefully collected, snap-frozen, and stored at 80°C until assayed for binding protein activity.

In Vitro Transcription

Radiolabeled and unlabeled full-length, wild-type, and mutant (containing AUGUA repeats in place of AUUUA repeats) GM-CSF, AUUUA RNA (denoted 2), and AUGUA RNA (denoted 2-AUGUA) were produced by in vitro transcription from the T7 promoter of pGM-CSF, pGM-CSF-AUGUA, pAUUUA, and pAUGUA, respectively, as previously described (11, 12). The 3' UTRs of mutant and wild-type GM-CSF were produced by in vitro transcription from polymerase chain reaction templates, also as previously described (11). [³²P] uridine triphosphate-labeled RNA was quantified by liquid scintillation counting and diluted to 1 × 10⁵ cpm/µl and used as such in band-shift assays. Unlabeled RNA was precipitated with ethanol, resuspended in diethyl pyrocarbonate-treated H₂O, and quantified by absorbance at 260 nm. The integrity of the transcripts was verified on 1% ribonuclease (RNase)-free agarose gels.

RNA Mobility Shift Assay

Mobility shift assays were performed essentially as previously described (11, 12). Briefly, 5 to 15 µg of AML14.3D10 cytoplasmic protein was incubated with radiolabeled RNA (10⁵ cpm) with 5 U RNasin (Promega, Madison, WI), 1× binding buffer (containing 15 mM Hepes [pH 8], 10 mM KCL, 10% glycerol, and 1 mM dithiothreitol), and 1 µg/µl yeast transfer RNA. Where indicated, unlabeled competitor RNAs were added to the reaction mix 15 min before the addition of radiolabeled probe. The entire reaction mixture was incubated at either 37°C or 4°C for 10 min, then incubated further with 40 U of RNase T1 at 37°C or 4°C for 45 min. Samples were ultraviolet (UV) crosslinked for 5 min on ice in a Stratalinker (Stratagene, La Jolla, CA) on automatic setting before the addition of 3.5 µl 4× SDS loading buffer, boiled for 3 min, and loaded onto 12% polyacrylamide gels. After electrophoresis, gels were dried and exposed overnight to Kodak X-AR film with two intensifying screens or to a PhosphorImager screen (Molecular Dynamics).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Decay Kinetics of Transfected GM-CSF mRNA in AML14.3D10 Cells

We sought to transfect AML14.3D10 cells with GM-CSF RNA precipitated onto gold beads using PMGT. This would allow determination of the decay kinetics of GM-CSF RNAs in the absence of transcriptional poisons such as actinomycin D (Act D). These drugs have recently been shown to interfere with cytokine mRNA decay (9, 11), although the mechanisms underlying this effect are unknown. In addition, determination of the half-life (t_1/2) of transfected, mutant GM-CSF RNAs would establish the contribution of the 3' UTR AREs to decay. Figure 1A shows typical autoradiograms of northern blots of AML14.3D10 cells after PMGT with blank beads or those with precipitated, wild-type, capped, and adenylated (90 adenosine residues) GM-CSF RNA. GM-CSF-specific hybridization was not detected in cells transfected with blank beads (Figure 1A, column Blank), demonstrating that endogenous GM-CSF gene expression was not induced by PMGT. In unstimulated AML14.3D10 cells, wild-type GM-CSF RNA decayed exceptionally quickly with a t_1/2 of about 5 min (Figure 1C). After 1 h of ionomycin treatment, the degradation of transfected GM-CSF RNA was significantly slowed to a t_1/2 of 14 min, which further increased to 22 min after 2 h of ionophore treatment (Figure 1B). Mutant GM-CSF RNA, lacking the 3' UTR AREs, displayed a constant t_1/2 of 19 to 22 min, which was completely unaffected by ionophore activation (Figures 1B and 1D). Thus, these data demonstrate that ionophore can stabilize GM-CSF RNA but that this effect requires intact AREs. Further, the kinetics of stabilization are rapid, with the half-maximal response achieved in less than 1 h.

View larger version (18K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   View larger version (31K): [in this window] [in a new window]   View larger version (29K): [in this window] [in a new window]   Figure 1.   Ionomycin activation stabilizes transfected GM-CSF mRNA in AML14.3D10 cells. Cells were cultured and stimulated with ionomycin before transfection with wild-type (A) or mutant (B) GM-CSF mRNA by PMGT as described (see MATERIALS AND METHODS). After rapid washing, the cells were harvested at the times shown along the top and total RNA was isolated and Northern blotted. Blots were hybridized with radiolabeled GM-CSF or GAPDH complimentary DNA before washing with 2× SSC/0.1% SDS at room temperature followed by 0.1× SSC/0.1% SDS at 60°C for 5 to 15 min. "Blank" refers to AML14.3D10 cells transfected with naked gold beads lacking GM-CSF RNA. Northern blots shown in A and B were quantified by PhosporImaging, plotted as percent of GM-CSF RNA signal strength remaining versus time, and are shown in C and D, respectively. Signal intensity at time zero was defined as 100%. The intersection of the dotted vertical lines with the x-axis equals t1/2.

We also examined the translational competence of transfected wild-type and mutant RNAs. In this way, the effects of alterations in decay rate on GM-CSF production can be evaluated. Equal numbers of cells were transfected with either wild-type or mutant RNAs and cells cultured for 2 additional hours before measurement of secreted GM-CSF by enzyme-linked immunosorbent assay (ELISA). Using an ELISA with a sensitivity of 4 pg/ml, we were repeatedly unable to measure any GM-CSF from supernatants over cells transfected with wild-type RNA. However, 30 pg/ml/10⁷ cells was typically detected in the supernatant from cells transfected with mutant GM-CSF RNA. Thus, a 4-fold increase in t_1/2 led to a > 7.5-fold increase in GM-CSF secretion. As the transfected GM-CSF RNAs share identical 5' untranslated and coding regions, differential translational efficiency is unlikely to account for these data.

Identification of Proteins Forming Complexes with the 3' UTR of GM-CSF mRNA

A number of ARE-specific mRNA binding proteins have recently been identified (12, 19). Several, including AUF-1 and HuR/HuD, have been cloned (13, 20). Although the physiologic role of these proteins remains poorly understood, they may interact with the ARE and target or block appropriate RNases, leading to mRNA decay or stabilization, respectively. To date, however, the existence of such proteins in eosinophils or eosinophil-like cell lines has yet to be demonstrated. Thus, we isolated cytoplasmic lysates from cycling AML14.3D10 cells and performed RNA mobility shift assays with radiolabeled, full-length, in vitro- transcribed wild-type (containing the 3' UTR AUUUA motifs) and mutant GM-CSF RNAs (3' UTR lacking the ARE). Cytoplasmic lysates and radiolabeled probe were incubated for 10 min at 37°C, followed by 40 U of RNase T1 to digest any unprotected radiolabeled RNA. Protein- RNA complexes were resolved by 12% SDS-polyacrylamide gel electrophoresis after UV treatment and visualized by autoradiography. As shown in Figure 2, GM-CSF RNA was completely digested by RNase T1 in the absence of cytoplasmic lysate. However, radiolabeled complexes of 55, 60, 85, 100, and 125 kD were detected when the probe was incubated with AML14.3D10 lysate followed by RNase T1. No complexes were observed when mutant GM-CSF RNA was used in place of wild-type RNA (Figure 2). To assess whether the resolved bands were in fact RNA-protein complexes, we incubated a constant amount of radiolabeled probe with increasing amounts of cytoplasmic lysate. As shown in Figure 2, the intensity of the complexes increased as the concentration of lysate increased, suggesting a saturable event. In addition, proteinase K completely abolished the complexes. Therefore, GM-CSF RNA clearly interacts with at least one and possibly five cytoplasmic proteins present in cycling but unstimulated AML14.3D10 cells.

View larger version (44K): [in this window] [in a new window]   Figure 2.   AML14.3D10 cell cytoplasm contains multiple GM-CSF RNA binding proteins. RNA mobility shift assays were performed as described in MATERIALS AND METHODS with 10 or 15 µg of cytoplasmic protein from cycling AML14.3D10 cells and radiolabeled wild-type or mutant full-length GM-CSF mRNAs. Molecular-weight markers are shown along the left.

The RNA binding specificity was determined by competition experiments using wild-type and mutant, ARE-negative GM-CSF RNAs. Unlabeled RNA was added to the lysate in increasing concentrations 15 min before radiolabeled GM-CSF RNA. As shown in Figure 3, increasing molar amounts of unlabeled, wild-type GM-CSF RNA successfully inhibited RNA-protein complexes. ARE-negative, mutant GM-CSF RNA, however, failed to complete substantially even at 50-fold molar excess (10× is the maximum concentration shown in Figure 3). Therefore, AML14.3D10 cytosol contains proteins that specifically interact with the 3' UTR AREs of GM-CSF RNA.

View larger version (52K): [in this window] [in a new window]   Figure 3.   GM-CSF RNA-protein complexes are specific for the 3' UTR AREs. Increasing amounts of unlabeled (concentrations in molar [M] excess over probe), competitor RNAs (as shown along the top) were added to 5 µg of AML14.3D10 cytoplasmic lysate before radiolabeled GM-CSF RNA probe and mobility shift assay. Molecular-weight markers are shown along the left. The molecular weights (in kD) of the protein-RNA complexes are presented along the right.

These data suggest that protein binding required intact AUUUA motifs. The 3' UTR of GM-CSF contains two clusters of AREs composed of three and five AUUUA motifs. Both clusters were deleted from the mutant GM-CSF RNA that we used. However, to further evaluate the sequence constraints on protein interactions, we asked whether a short (80-base) RNA with four consecutive AUUUA pentamers (denoted 2) could compete for binding of wild-type GM-CSF RNA. A control of identical size and composition, but containing tandem AUGUA repeats (2-AUGUA), was also included to ensure that competitor length did not influence binding. Figure 4A shows that 2 RNA prevented GM-CSF-protein interactions in a concentration-dependent fashion, whereas 2-AUGUA had no effect (not shown). The concentration of 2 required to compete was 2- to 4-fold greater than that seen with wild-type GM-CSF RNA (see previous discussion of Figure 3), suggesting a lower affinity interaction than seen for full-length GM-CSF RNA.

View larger version (63K): [in this window] [in a new window]   View larger version (68K): [in this window] [in a new window]   Figure 4.   An 80-base RNA containing four AUUUA repeats (2) competes with GM-CSF RNA or itself for binding to cytoplasmic proteins. Increasing concentrations of 2 RNA were added 15 min before radiolabeled RNA and performance of the mobility shift assay at 37°C. (A) Radiolabeled GM-CSF RNA was used as probe. (B) Radiolabeled 2 was used as probe. Molecular-weight markers are shown along the left.

We also performed the converse experiment, namely mobility shift experiments with radiolabeled 2 and AML14.3D10 cytoplasm. As shown in Figure 4B, under these conditions a dominant 47-kD RNA-protein complex was detected along with weaker bands at 53 to 55, 65, and approximately 100 kD. All complexes were specific, as they could be entirely competed with 25-fold M excess of 2 RNA (Figure 4B, columns marked Competition) but not by equivalent concentrations of 2-AUGUA (not shown). These data cumulatively suggest that ARE binding proteins have higher affinity for ligands with multiple clusters of AUUUA motifs, as found in wild-type GM-CSF RNA. Alternatively, an up- or downstream sequence present only in full-length GM-CSF RNA facilitated RNA-protein interactions. Inasmuch as 2 RNA was able to compete with labeled GM-CSF RNA, these RNAs must recognize identical proteins, albeit with different affinities. The variations in size of the RNA-protein complexes generated by 2 and GM-CSF RNA thus most likely reflect differences in protected (and presumably protein-associated) RNA after RNAse T1 digestion.

Calcium Ionophore Upregulates AUUUA Binding Protein Activity

The previous data demonstrate that GM-CSF RNA can be stabilized by ionophore in AML14.3D10 cells. A series of ARE-specific binding proteins are present in the cytoplasm of these cells and capable of binding RNA ligands in vitro. Thus, we asked whether binding activity could also be modulated by ionophore activation of AML14.3D10 cells. If these proteins were involved in GM-CSF RNA destabilization, their activity would be decreased by ionophore. Conversely, if these proteins were involved in ionophore-mediated GM-CSF RNA stabilization, their activity would likely be increased by activation. Thus, we stimulated AML14.3D10 cells with ionomycin and temporally assessed binding protein activity using mobility shift assays. To simplify data interpretation we employed radiolabeled 2 as the probe, which, as shown in Figure 4B, dominantly generates a 47-kD RNA-protein complex. This complex was preferentially enhanced by PhosphorImager analysis, permitting direct quantitation of its intensity over time. A typical RNA mobility shift is shown in Figure 5A, using cytoplasmic lysates from ionomycin-treated AML14. 3D10 cells. As expected, basal activity was observed in cycling but untreated cells that remained quite constant over time (2 h). Upon ionophore treatment, the 47-kD 2 RNA- protein complex increased by 1.5-fold within 1 h and by nearly 3-fold within 2 h. Similar data were obtained for the entire set of five RNA-protein complexes when full-length GM-CSF RNA was used as the probe (not shown). Thus, binding activity increased contemporaneously with enhanced GM-CSF RNA stability (see Figures 1A and 1C). These data show that enhanced binding protein activity is tightly associated with decreased degradation of GM-CSF RNA and imply that the in vivo interaction of binding protein and GM-CSF RNA may attenuate its decay.

View larger version (43K): [in this window] [in a new window]   View larger version (28K): [in this window] [in a new window]   Figure 5.   A 47-kD protein-2 RNA complex was upregulated in AML14.3D10 cells after ionomycin activation. (A) RNA mobility shift assays were performed with 5 µg cytoplasmic protein from AML14.3D10 cells treated for the indicated times with ionomycin. Molecular-weight markers are shown along the left. (B) Three independent mobility shift experiments were quantified by PhosphorImaging, normalized to cytoplasmic protein, and analyzed by one-sided t test. Significant P value (< 0.05) is shown at 2 h after activation.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Eosinophils are well established effector cells contributing to the pathology of asthma (21). After allergen challenge, peripheral blood eosinophils migrate into the lung where they exacerbate pulmonary dysfunction and likely contribute to damage and ultimately fibrosis. GM-CSF accumulates to high levels in the airways after sustained release by activated immune cells, fibroblasts, and eosinophils themselves (1, 5, 22). This cytokine has profound effects on eosinophil biology (6, 23), survival (24), and differentiation (25) both in vitro and in vivo. As such, understanding the underlying molecular mechanisms that control GM-CSF elaboration from activated eosinophils is of critical importance. In T lymphocytes, most cytokines, including GM-CSF, interferon-, and IL-2 are regulated at the post-transcriptional level through variable mRNA stability (7). In this way, cells can respond quite rapidly to changes in the environment with changes in cytokine production and release. Given the importance of GM-CSF to eosinophil biology and the prior data in T cells, we first investigated the rate at which GM-CSF mRNA decayed in AML14.3D10 cells. To our knowledge, there are no data in the literature related to this question. Our second question was: If GM-CSF mRNA stability varied in activated eosinophilic cells, by what mechanism might this be affected?

Typically, mRNA decay is measured after transcriptional arrest with inhibitors of RNA polymerase II such as Act D (26). Unfortunately, we have recently shown that these drugs themselves can profoundly stabilize GM-CSF mRNAs (9, 11). Therefore, we employed the alternative approach of transfecting AML14.3D10 cells with GM-CSF RNA precipitated onto gold beads by PMGT. These mRNAs contain a 5' guanosine cap and a 90-base polyadenylate tail, permitting efficient translation. In lymphocytes, we have shown that exogenous mRNAs were rapidly mobilized onto polysomes with newly translated GM-CSF protein detectable in as little as 15 min (11). Thus, it is likely that transfected mRNAs are appropriately localized for physiologic decay. In those cells, GM-CSF mRNA decayed with a t_1/2 of approximately 8 min, or quite comparable to that observed here. Transfected, ARE-negative GM-CSF mRNA was substantially more stable than wild-type, with a t_1/2 of approximately 22 min. Thus, the ARE of GM-CSFand, by inference, the AREs of other cytokine and proto-oncogene mRNAstargets them for extremely rapid decay in AML14.3D10 cells. Because the ARE-negative mutant still decayed quite quickly, these data suggest that the remaining 5' UTR, coding region, or 3' UTR contains additional destabilizing elements. Indeed, Iwai and colleagues have demonstrated that regions 5' to the AREs contributed to phorbol ester- and concanavalin A-mediated regulation of GM-CSF mRNA decay (27). However, these elements need not be in the UTR, as both c-myc and c-fos mRNAs contain destabilizing regions in the coding region (28).

Activation with calcium ionophore markedly enhanced GM-CSF RNA t_1/2. Similar data have been shown for GM-CSF and IL-3 mRNA in activated mast cells (29). It is notable that stabilization through ionophore-mediated signaling pathways required the 3' UTR AREs, inasmuch as a mutant GM-CSF lacking these elements was not responsive. The kinetics of stabilization were prompt, with over 50% of the ultimate increase occurring within 1 h. These data suggest that preformed effectors, most likely proteins, were modified, leading to GM-CSF mRNA stabilization. Because the decay of most mRNAs is not affected by ionophore, it seems unlikely that inhibition of generic RNase activity could account for the observed specificity. As such, we hypothesized that protein(s) may bind the ARE of GM-CSF mRNA in AML14.3D10 cells. In the simplest model, RNase access to the AREs would be prevented or delayed by ARE-binding proteins. Under such conditions, GM-CSF mRNA would be specifically stabilized.

By RNA mobility shift analysis, we were able to resolve RNA-protein complexes of 55, 60, 85, 100, and 125 kD from AML14.3D10 cells using full-length, radiolabeled GM-CSF RNA. Nakamaki and associates showed that human embryonic lung fibroblasts contained five major ARE- binding protein complexes of 70, 45, 40, 38, and 32.5 kD (15). The observed differences in complex mass may reflect their use of RNase A rather than RNase T1, as used here. Because RNase T1 cuts single-stranded RNA 3' to guanosine, whereas RNase A cuts 3' to pyrimidines, the latter can cleave GM-CSF RNA within the AREs and the former cannot. Thus, larger RNase-resistant fragments would be anticipated after T1 digestion.

On the basis of competition assays, all RNA binding proteins were highly specific for the AUUUA repeats. At this time we do not know how many distinct proteins generate the observed number of RNA-protein complexes. For example, it is possible that multimers of the 55- or 60-kD activities produce the 100- and 125-kD complexes, respectively. We are currently performing protease mapping to determine the relatedness of the protein components of these complexes.

When an 80-base RNA containing four consecutive AUUUA repeats (2) was used for mobility shift assays at 4°C, a profile of RNA-protein complexes was detected similar to that seen with full-length GM-CSF RNA (not shown). Inasmuch as unlabeled 2 was also capable of competing with GM-CSF RNA whereas a mutant 2 containing AUGUA motifs was not, similar or identical proteins likely interact with these two RNAs through their shared AREs. However, the competition data also revealed significant differences in affinity for full-length versus short ARE-containing ligands, with most binding proteins preferring the full-length ligand. When mobility shift assays were performed at 37°C with 2, the dominant complex had a molecular mass of approximately 47 kD whereas assays performed at 4°C were indistinguishable from those employing full-length GM-CSF RNA. We attribute this to differential affinity, which becomes far more apparent under conditions of greater stringency (i.e., 37°C). The affinity may be lower because 2 lacks additional GM-CSF RNA sequence, which contributes to protein recognition.

During ionophore-mediated activation, the activity of ARE-specific binding proteins increased substantially. Within 2 h the overall activity was increased by 2.5-fold, which was highly significant. The timing of this event coincided with GM-CSF mRNA stabilization, suggesting a cause-and-effect relationship. The simplest mechanism to account for this phenotype would be for binding proteins to mask the AREs from RNases. Recently, overexpression of HuR, an AUUUA-specific RNA binding protein, stabilized c-fos mRNA in transfected mouse L939 cells (30). Heterogeneous nuclear ribonuclear proteins (hnRNP) C and L have been shown to stabilize amyloid protein precursor (31) and VEGF (32) mRNAs, respectively. Thus, there is substantial precedent for mRNA stabilization after binding protein interactions in mammalian cells.

Given the rapidity of binding-protein upregulation in AML14.3D10 cells, it is unlikely that increased transcription or protein synthesis can account for this effect. The adenosine-uridine binding factor was phosphorylated in response to phorbol ester-mediated or ionophore-mediated activation of peripheral blood mononuclear cells (33). This modification was associated with increased binding activity toward AUUUA RNA ligands. Conversely, hnRNP C protein binding activity in activated cells was enhanced by dephosphorylation (34). Thus, post-translational modifications involving the addition or removal of phosphate possibly mediate the changes in AUUUA binding ativity identified here in AML14.3D10 cells.

The biologic role of mRNA stabilization in GM-CSF elaboration by activated eosinophils requires additional study. Even modest increases in stability often have profound effects on protein synthesis by increasing steady-state mRNA levels. Protein production can be increased up to 20-fold through mRNA stabilization (9, 11). The data shown here suggest GM-CSF secretion increased by at least 7.5-fold when the more stable GM-CSF mutant mRNA was transfected. Thus, it is likely that mRNA stabilization may play an important role in enhancing GM-CSF production by activated eosinophils.