Association of Tumor Suppressor Protein Pdcd4 With Ribosomes Is Mediated by Protein-Protein and Protein-RNA Interactions

Mapping of amino acid residues involved in RNA binding by Pdcd4

We have previously shown that the N-terminal part of Pdcd4 is able to bind to RNA in vitro. ¹⁸ This suggested that this part of the protein might function as an RNA binding domain. To further explore the function of the N-terminal domain of Pdcd4 and to be able to address the role of RNA binding in vivo, we set out to map the amino acid residues that are crucial for the RNA binding activity. As a first step, we expressed different parts of Pdcd4 as glutathione-S-transferase (GST) fusion proteins and analyzed them by an RNA binding assay. As illustrated in Figure 1, the purified proteins were run on a sodium dodecyl sulfate (SDS)–polyacrylamide gel, blotted onto nitrocellulose, and incubated with radiolabeled RNA. Consistent with our previous work, full-length Pdcd4 but not an N-terminally truncated form of Pdcd4 lacking the first 150 amino acids was able to bind RNA. Analysis of the RNA binding activity of subregions of the N-terminal domain revealed that 2 regions located between amino acids 50 to 100 and 100 to 159 showed RNA binding activity, whereas a protein corresponding to amino acids 1 to 50 of Pdcd4 did not bind RNA. We therefore concluded that 2 regions within the N-terminal domain of Pdcd4 are involved in RNA binding.

OPEN IN VIEWER

Figure 1.

RNA binding by Pdcd4 is mediated by 2 subregions within the N-terminal domain of Pdcd4. Glutathione-S-transferase (GST) fusion proteins containing full-length Pdcd4 or different parts of Pdcd4 were displayed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), blotted onto nitrocellulose, and subjected to the RNA binding assay as described in Materials and Methods. The right panel shows an autoradiogram of the blot after incubation with in vitro synthesized, radiolabeled RNA. The left panel shows a Coomassie brilliant blue–stained gel containing the same amounts of the different GST-proteins. The GST fusion proteins are illustrated schematically at the top (numbers indicate Pdcd4 amino acids that are fused to GST in each construct).

Inspection of the amino acid sequence of Pdcd4 revealed 2 clusters of arginine and lysine residues centered around amino acids 63 and 105 (Fig. 2). We refer to these clusters as RBM1 and RBM2, respectively. Both regions show significant amino acid sequence conservation between vertebrate Pdcd4 proteins as well as between Pdcd4 homologs of very divergent species such as Suberites domuncula (a sponge), Drosophila melanogaster, Xenopus laevis, Danio rerio, and the mouse and humans (Fig. 2A). In the case of RBM2, this conservation is especially obvious and extends C-terminally beyond the cluster of basic residues. We reasoned that these clusters, because of their overall positive charge, might be involved in binding RNA. To address this possibility, we mutated several of the basic residues in both clusters to alanine. We then expressed these Pdcd4 mutants as GST fusion proteins and subjected them to the in vitro RNA binding assay. Figure 2B shows that the RNA binding activity of the proteins mutated in RBM1 or RBM2 or in both clusters (RBM1+2) was virtually undetectable, confirming that both clusters of basic amino acid residues are crucial for RNA binding by full-length Pdcd4.

OPEN IN VIEWER

Figure 2.

Identification of 2 conserved amino acid motifs involved in RNA binding by Pdcd4. (A) Schematic structure of Pdcd4 and the location of 2 clusters of basic amino acids (referred to as RBM1 and RBM2). The amino acid sequences of RBM1 and RBM2 in Pdcd4 proteins from the sponge Suberites domuncula (S.d.), Drosophila melanogaster (D.m.), Xenopus laevis (X.l.), Danio rerio (D.r.), Mus musculus (M.m.), and Homo sapiens (H.s.) are shown below. Amino acids identical to human Pdcd4 are marked by black boxes. The asterisks mark the positions of arginine and lysine residues in human Pdcd4. (B) Fusion proteins of glutathione-S-transferase (GST) and full-length Pdcd4 carrying the indicated mutations in RBM1 (mutRBM1), RBM2 (mutRBM2), or both motifs (RBM1+2) were analyzed for RNA binding activity as described in Figure 1. The upper panel shows Coomassie brilliant blue staining of the fusion proteins to confirm equal loading of the proteins. The lower panel shows an autoradiogram of a blot of the same proteins subjected to the RNA binding assay. In both panels, the arrows mark the GST-Pdcd4 fusion proteins. (C) GST-Pdcd4-wt (top) or GST-Pdcd4-mutRBM 1+2 (bottom) proteins were analyzed by chromatography on poly(U)-agarose. Bound proteins were eluted from the column with binding buffer containing increasing amounts of NaCl. Samples of input, flow-through, 2 successive washes, and salt-eluate fractions were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting with Pdcd4 antiserum. Equal aliquots were loaded for input, flow-through, and wash fractions, whereas aliquots of the salt-eluate fractions were 5 times larger.

As an alternative RNA binding assay that avoids the denaturation and renaturation of the protein, we subjected GST-Pdcd4-wt and GST-Pdcd4-mutRBM1+2 proteins purified in soluble form to chromatography on poly(U)-agarose. We have demonstrated before that the endogenous Pdcd4 protein binds to poly(U)-agarose and is eluted in buffers of increasing ionic strength. ¹⁸ Figure 2C shows that bacterially expressed wild-type Pdcd4 was able to bind to poly(U)-agarose efficiently, whereas the mutant Pdcd4 protein failed to bind. Most of the wild-type Pdcd4 was eluted in the presence of 200 to 300 mM NaCl, suggesting that it binds to RNA under physiological salt conditions.

RNA-binding activity of Pdcd4 is required for recruiting Pdcd4 to ribosomal complexes

Next, we were interested to know whether the disruption of the in vitro RNA binding activity affects the behavior of Pdcd4 in vivo. A previous report has suggested that Pdcd4 cosediments with ribosomes and/or polysomes in sucrose density gradients ³⁰ ; we therefore set out to compare the ribosomal association of wild-type and mutant Pdcd4 proteins. As a first step, we examined the ribosomal association of endogenous Pdcd4. We prepared cytoplasmic extracts from HeLa cells and analyzed them by sucrose density gradient centrifugation. Figure 3A illustrates the distribution of Pdcd4 and of the 18S and 28S ribosomal RNAs throughout the gradient fractions. A significant amount of Pdcd4 appeared in fractions that contained small ribosomal subunits and monomeric ribosomes with the highest amount of Pdcd4 sedimenting slightly faster than small ribosomal subunits and slower than complete ribosomes. These fractions also contained eIF4A. This suggested that Pdcd4 together with the small 40S ribosomal subunit is part of translation preinitiation complexes and is consistent with the role of Pdcd4 as an eIF4A interacting protein. There was no Pdcd4 in fractions containing polysomes. A second peak of Pdcd4 was detected at the top of the gradient, presumably representing soluble cytoplasmic protein. Similar sedimentation patterns of Pdcd4 were observed when cytoplasmic extracts from other cell lines were analyzed (Fig. 3B and data not shown). To substantiate that Pdcd4 is associated with ribosomes, we treated cytoplasmic extract from HeLa cells with EDTA, which is known to dissociate ribosomes into small and large subunits and to disrupt preinitiation complexes. Under these conditions, all of the Pdcd4 appeared at the top of the gradient (Fig. 3C). Similarly, treatment of cell extract with RNase abolished the peak of Pdcd4 cosedimenting with ribosomal complexes (Fig. 3D). To further corroborate the notion that Pdcd4 is associated with ribosomal complexes, we prepared cytoplasmic extracts from HeLa cells and immunoprecipitated them with Pdcd4-specific antiserum or with preimmune serum. We then analyzed the immunoprecipitates by agarose gel electrophoresis to visualize ribosomal RNAs. We confirmed that the Pdcd4-specific antiserum was able to immunoprecipitate Pdcd4 (see Supplementary Fig. S1). As shown in Figure 3E, 18S and 28S ribosomal RNAs were present in the specific but not in the control immunoprecipitate. The ethidium bromide stainable material was RNA (and not DNA), as demonstrated by its sensitivity toward RNase treatment. Note that the relative amounts of 18S and 28S RNAs were changed in the immunoprecipitate versus the input samples, consistent with the preferential association of Pdcd4 with small ribosomal subunits. Taken together, the data illustrated in Figure 3 demonstrate that Pdcd4 is associated with complexes containing small ribosomal subunits.

OPEN IN VIEWER

Figure 3.

Ribosomal association of Pdcd4. (A) Cytoplasmic extracts isolated from HeLa cells were analyzed by sucrose density gradient centrifugation. Individual fractions of the gradient were then analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting using antibodies against Pdcd4 (top) and eIF4A (middle) or by agarose gel electrophoresis and staining with ethidium bromide to detect 18S and 28S ribosomal RNAs (bottom). Fractions containing predominantly small ribosomal subunits, complete ribosomes, and polysomes are marked. (B) Cytoplasmic extracts from HEK293 cells were analyzed as described in panel A. (C) HeLa cells were analyzed as described in panel A except that EDTA was used to disrupt ribosomal complexes. (D) HeLa cells were analyzed as described in panel A except that the cell extract was treated with RNAse before centrifugation. (E) Cytoplasmic extracts of HeLa cells, pretreated for 30 min with or without cycloheximide, were immunoprecipitated with Pdcd4-specific antiserum or with preimmune serum. Coprecipitated 18S and 28S ribosomal RNAs were then visualized by agarose gel electrophoresis and staining with ethidium bromide. Aliquots of the samples were treated with DNase or RNase before gel electrophoresis. The bottom panel shows aliquots of the cytoplasmic extracts before immunoprecipitation.

Having established the ribosomal association of wild-type Pdcd4, we next examined the ability of the mutant proteins to associate with ribosomes. We expressed mutant Pdcd4 proteins deficient in RNA binding in HeLa cells and analyzed cytoplasmic extracts by sucrose density gradient centrifugation. In these experiments, coexpressed hemagglutinin (HA) tagged wild-type Pdcd4 served as an internal control for the mutant proteins, all of which were Flag tagged. Figure 4 shows that deletion of the entire N-terminal domain completely abrogated the ribosomal association of Pdcd4, consistent with the notion that RNA binding plays an important role in the binding of Pdcd4 to ribosomal subunits. Mutation of RBM1 only slightly diminished the ribosomal association of the protein, whereas mutation of RBM2 had a much more significant effect on the interaction of Pdcd4 with ribosomes. When RBM1 and RBM2 were mutated simultaneously, the protein was unable to associate with ribosomes.

OPEN IN VIEWER

Figure 4.

Ribosomal association of mutant Pdcd4 proteins. HeLa cells were cotransfected with expression vectors for HA-tagged wild-type Pdcd4 and Flag-tagged mutant Pdcd4 lacking the N-terminal domain (Pdcd4-ΔRBD), carrying mutations in RBM1 (Pdcd4-mutRBM1), RBM2 (Pdcd4-mutRBM2), or both RBM motifs (Pdcd4-mutRBM1+2) or carrying the D418A substitution (Pdcd4-D418A). Cytoplasmic extracts were prepared 20 h after transfection and analyzed by sucrose density gradient sedimentation as described in Figure 3. In each panel, the distribution of Flag-tagged mutant and HA-tagged wild-type proteins is shown at the top and the middle, respectively. The distribution of the ribosomal RNAs is shown at the bottom.

Ribosomal association of Pdcd4 also requires its interaction with eIF4A

Because the interaction of Pdcd4 and eIF4A has been strongly implicated in the translation suppressor activity of Pdcd4, ²² we were interested to know whether the ribosomal association of Pdcd4 was also dependent on the Pdcd4-eIF4A interaction. To address this issue, we made use of the D418A mutant of Pdcd4, which has a strongly reduced affinity for eIF4A. ²³ We transfected HeLa cells with an expression vector for the D418A mutant, followed by sedimentation analysis of cytoplasmic extracts. As illustrated in Figure 4, the mutant protein no longer cosedimented ahead of the small ribosomal subunit but was distributed in the upper part of the gradient. We therefore concluded that the interaction with eIF4A is necessary for the stable ribosomal association of Pdcd4.

Implications for the cytoplasmic function of Pdcd4

Our work leads to several interesting conclusions regarding the role of RNA binding for the cytoplasmic function of Pdcd4. First, our work confirms that Pdcd4 binds RNA via its N-terminal domain and identifies 2 clusters of positively charged amino acids, RBM1 and RBM2, as being essential for the RNA binding activity of Pdcd4. Both clusters are conserved between Pdcd4 homologs from extremely divergent species such as sponges, insects, and vertebrates. This strongly suggests that the ability to bind RNA is an evolutionary conserved feature of Pdcd4. Interestingly, the region located immediately C-terminally to RBM2 is also very highly conserved, suggesting that it plays an important role, either in RNA binding or in another interaction of Pdcd4.

Second, our work demonstrates that a significant fraction of the cytoplasmic Pdcd4 is associated with ribosomal complexes. It has previously been claimed that Pdcd4 cosediments with the polysome fraction in a sucrose density gradient, ³⁰ but our data now clearly show that Pdcd4 is stably associated with a complex sedimenting slightly faster than small ribosomal (40S) subunits and slower than complete (80S) ribosomes. Because of this sedimentation pattern, it appears likely that Pdcd4 is primarily associated with the 48S translation preinitiation complex. This is consistent with the fact that Pdcd4 interacts with eIF4A (which is part of this complex) and is further supported by the observation that the D418A mutant of Pdcd4 disrupts the stable association of Pdcd4 with the ribosomal complex. Asp 418 of Pdcd4 is involved in making direct contacts with eIF4A, and its replacement by alanine has been shown to substantially weaken the Pdcd4-eIF4A interaction. ^23-25 Our data, therefore, indicate that the ribosomal complex with which Pdcd4 interacts contains eIF4A and, furthermore, that this interaction is necessary for the stable association of Pdcd4 with this complex. Taken together, our data demonstrate that Pdcd4 interacts via eIF4A with the translation preinitiation complex.

Third, our results demonstrate that mutations of the RNA binding motifs affect the ribosomal association of Pdcd4 and thus provide the first evidence that RNA binding by Pdcd4 plays a crucial role in vivo. Mutation of RBM1 had only a minor effect in vivo, whereas replacement of only 2 arginine residues within RBM2 had a significantly stronger effect on the ribosomal association of Pdcd4. The combination of both RBM mutants eventually led to complete disruption of the ribosomal association of Pdcd4. Although both single RBM mutants had virtually completely lost RNA binding activity in vitro, they showed residual association with ribosomes. Presumably, the ability of both mutant proteins to still bind to eIF4A and to RNA with the intact RBM explains why they are still associated with the ribosomal complexes, at least to some extent. In conclusion, therefore, our data demonstrate that the association of Pdcd4 with translation preinitiation complexes is mediated by protein-protein interactions between Pdcd4 and eIF4A as well as protein-RNA interactions between Pdcd4 and, possibly, the mRNAs present in these complexes. Whether the Pdcd4-RNA interaction is specific for certain mRNAs and whether RNA binding mediates the recruitment of Pdcd4 to preinitiation complexes containing specific mRNAs are very interesting questions that can be addressed once specific target mRNAs for Pdcd4 have been identified.

Eukaryotic expression vectors

A human Pdcd4 expression vector was generated by amplifying the coding region of human Pdcd4 from total cDNA of T47D breast cancer cells using 5′-CTGGATCCGCCACCATGGATGTAGAAAATGAGCAG-3′ and 5′-GACTCGAGTTTAGTAGCTCTCTGGTTTA-3′ as forward and reverse primers and inserting the PCR product between the BamHI and XhoI sites of pCDNA4. A C-terminally HA-tagged expression vector for human Pdcd4 was generated by a subsequent PCR using the reverse primer 5′-GACTCGAGTTAGGCGTAGTCGGG CACGTCGTAGGGGTAGTAGCTCTCTGGTTTA-3′ and cloning the PCR product between the BamHI and XhoI sites of pCDNA4. An N-terminally Flag-tagged version was constructed by replacing pCDNA4 polylinker sequences between the KpnI and BamHI sites immediately upstream of the Pdcd4 coding region by the sequence GGTACCGAATTCGCCACCATGGATTACAA GGACGACGATGACAAGGGATCC. pCDNA4-Flag-hPdcd4- D418A is a point-mutated derivative in which Asp 418 was changed to Ala using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and appropriate primers. In plasmid pCDNA4-Flag-hPdcd4-RBD-Mut1, the human Pdcd4 coding sequence was changed from AAGGCAAAAAGGCGACTAAGGAAAAACTCA to AAGGCAGCGGCCGCACTAGCGGCAAACTCA, resulting in mutation of the RBM1 from KAKRRLRKN to KAAAALAAN. In plasmid pCDNA4-Flag-hPdcd4-RBD-Mut2, the human Pdcd4 coding sequence was changed from GCTGGATAGGCGATCCAGAT to GCTGGATGCGGCAT CCAGAT, resulting in mutation of the RBM2 from LDR RSRSG to LDAASRSG. Plasmid pCDNA4-Flag-hPdcd4-RBD-DM contains a combination of both mutations. pCD NA4-hPdcd4-ΔRBD encodes an internally deleted human Pdcd4 protein lacking amino acids 16 to 150. All mutants were verified by sequencing.

HeLa cells were transfected with the desired expression vectors using the calcium-phosphate coprecipitation method, as described before. ³¹ Cells were harvested 16 to 20 h after transfection.

Bacterial expression vectors

Expression vectors coding for different GST-Pdcd4 fusion proteins were generated by cloning the full-length or different parts of the mouse Pdcd4 coding region between the BamHI and NotI sites of the expression vector pGex-6P-2. The respective parts of the coding region were amplified by PCR using appropriately designed primers, and the final constructs were verified by sequencing. Point mutants of the RNA binding motifs RBM1, RBM2, or both were generated as fusion proteins of GST and full-length mouse Pdcd4. In the case of RBM1, the amino acid sequence was changed from KAKRRLRKN to KAAAALAAN; in the case of RBM2, the sequence was changed from LDRRSRSG to LDAASRSG. GST fusion proteins were expressed and purified as described. ³² To detect GST-Pdcd4 by Western blotting, a Pdcd4 antiserum reacting with the C-terminus of human Pdcd4 (Acris, Herford, Germany) was used.

RNA-binding assays

RNA binding assays of GST-Pdcd4 fusion proteins were performed as described before. ¹⁸ GST fusion proteins were fractionated by SDS-PAGE and blotted onto nitrocellulose. Radiolabeled RNA was synthesized in vitro using T7 RNA polymerase, [³²P]-UTP, and the control template (pGEM Express) provided in the Riboprobe System T7 kit (Promega, Madison, WI), as described previously. ¹⁸ Blots were incubated first with 6 M urea, 0.2% NP40 for 10 min; washed 4 times 15 min each with RNA binding buffer (10 mM Tris-HCl, pH 7.5; 5 mM KCl; 2 mM MgCl₂) lacking RNA; and then incubated for 15 min in RNA binding buffer containing radiolabeled RNA. Subsequently, blots were washed 3 times with binding buffer lacking RNA (10 min each wash) and analyzed by autoradiography. All incubations were performed at room temperature. For chromatography on poly(U)-agarose (all steps at 4°C), GST-Pdcd4 fusion proteins were purified by binding to glutathione-sepharose and eluting them overnight with 15 mM glutathione in 10 mM Tris-HCl, pH 7.5. Eluted proteins were then bound to poly(U)-agarose (Sigma, P8563) equilibrated in RNA binding buffer. Proteins not bound (flow-through), eluted with 2 successive washes with binding buffer (wash fractions 1 and 2), and eluted with binding buffer supplemented with increasing concentrations of NaCl were analyzed by SDS-PAGE and Western blotting.

Sucrose density gradient centrifugation of cytoplasmic extracts

Cells were treated with 50 µg/mL cycloheximide for 30 min before harvesting. Cells were then washed with ice-cold phosphate-buffered saline (containing 50 µg/mL cycloheximide) and lysed in hypotonic buffer (10 mM HEPES, pH 7.5; 5 mM KCl; 2 mM MgCl2; 1 mM dithiothreitol [DTT]; 0.5% NP40; 1 mM phenylmethanesulfonyl fluoride [PMSF]; supplemented with 40 U/mL RNaseOUT [Invitrogen, Carlsbad, CA] and a protease inhibitor cocktail containing 5 ng/mL pepstatin A, 1 ng/mL leupeptin hemisulfate, and 5 ng/mL aprotinin) for 20 min on ice. The lysed cells were pelleted at 14,000 rpm for 15 min at 4°C. The supernatant was collected as cytoplasmic fraction and layered on top of a 10% to 50% sucrose density gradient made in hypotonic buffer. Gradients were centrifuged in a SW41 rotor at 36,000 rpm for 2 h at 4°C. The gradients were fractionated, and aliquots of the fractions were analyzed by 10% SDS-PAGE and Western blotting to determine the distribution of Pdcd4 and eIF4A or treated with Proteinase K and 1% SDS and analyzed by agarose gel electrophoresis to determine the distribution of ribosomal RNAs. Rabbit anti-Pdcd4 serum was described before ¹⁶ ; anti-eIF4A, HA, and Flag antibodies were from Abcam (Cambridge, UK), Hiss Diagnostics (Freiburg, Germany), and Sigma (Munich, Germany), respectively. For EDTA treatment of ribosomal complexes, 25 mM EDTA (final concentration) was added to the cytoplasmic extracts. In this case, the sucrose gradient was made in hypotonic buffer lacking MgCl₂ and supplemented with 25 mM EDTA. For RNase treatment of ribosomal complexes, the extract was digested with 100 µg/mL RNaseA for 1 h at 37°C.

Immunoprecipitation

Cells were pretreated with or without cycloheximide, and cytoplasmic extracts were prepared as described above. Aliquots of the extracts were immunoprecipitated with rabbit anti-Pdcd4 serum ¹⁶ or with preimmune serum for 3 h on ice. Immunoprecipitates were collected on protein A sepharose, washed with hypotonic buffer, and finally eluted from the sepharose with 1% SDS. Samples of the immunoprecipitates or the input material were treated with Proteinase K before they were analyzed by agarose gel electrophoresis followed by ethidium bromide staining. Samples were optionally treated with RNase-free DNase or with RNase before gel electrophoresis.