TcUBP-1, an mRNA Destabilizing Factor from Trypanosomes, Homodimerizes and Interacts with Novel AU-rich Element- and Poly(A)-binding Proteins Forming a Ribonucleoprotein Complex*

Trypanosomes, protozoan parasites causing world-wide infections in human and animals, mostly regulate protein expression through post-transcriptional mechanisms and not at the transcription initiation level. We have previously identified a Trypanosoma cruzi RNA-binding protein named TcUBP-1. This protein is involved in mRNA destabilization in vivo through binding to AU-rich elements in the 3 (cid:1) -untranslated region of SMUG mucin mRNAs (D’Orso, I., and Frasch, A. C. (2001) J. Biol. Chem. 276, 34801–34809). In this work we show that TcUBP-1 is part of an (cid:1) 450-kDa ribonucleoprotein complex with a poly(A)-binding protein and a novel 18-kDa RNA-binding protein, named TcUBP-2. Recombinant TcUBP-1 and TcUBP-2 proteins recognize U-rich RNAs with similar specificity and affinity through the (cid:1) 92-amino acid RNA recognition motif. TcUBPs can ho-mo- and heterodimerize in vitro through the glycine-rich C-terminal region. This interaction was also detected in vivo by co-immunoprecipitation of the ribonucleoprotein complex and using yeast two-hybrid assay. The

invade host cells and differentiate into a replicative amastigote form that differentiates into bloodstream trypomastigotes. The latter stage is able to invade a wide variety of cells, thus propagating the infection. The cycle closes when the hematophagous vector ingests circulating trypomastigotes with its blood meal. Trypanosomes, protozoan parasites of the order Kinetoplastida, have particular features in terms of mechanisms leading to protein expression. RNA polymerase I promoters were identified in rDNA but also in genes encoding the variable surface antigens from African trypanosomes (1). Conversely, only few RNA polymerase II promoters were described (2). As part of the maturation process, a common 39-nucleotide RNA, named spliced leader, is added to all trypanosome mRNAs by a trans-splicing process. This phenomenon was shown to be couple to 3Ј-poly(A) tail addition during polycistronic RNA processing (3). As a consequence, and at variance with higher eukaryotic cells, the control of protein expression in trypanosomatids is mainly post-transcriptional (4). However, little information is available on the relative importance of these processes and how they operate jointly in the parasite.
One of the possible mechanisms to regulate protein expression is through modification of the half-life of mRNAs. Several cis-elements located throughout the mRNA, coding and/or untranslated regions (UTRs), 1 were identified (5)(6)(7). However, only few of them were found to be recognized specifically by trans-acting factors (8). We have found previously (9) two distinct cis-elements located in the 3ЈUTR of the mRNAs of a mucin surface antigen family named SMUG of T. cruzi. A 44-nucleotide AU-rich element (ARE) was shown to destabilize SMUG transcripts in the infective non-replicative trypomastigote stage of the parasite. Conversely, a 27-nucleotide G-rich cis-element stabilizes SMUG in the non-infective replicative epimastigote stage of the parasite (8). These results suggest that both elements, ARE and the G-rich cis-element, act coordinately in a developmentally regulated manner and are recognized by specific RNA-binding proteins (8). Recently, we described that this ARE sequence was recognized by a single RRM-type RNA-binding protein named TcUBP-1, for T. cruzi Uridine-binding protein. In vivo, TcUBP-1 was shown to destabilize SMUG mRNAs in the epimastigote stage of the parasite (10). Furthermore, and at variance with what occurs in yeast, the processes of 3Ј-5Ј and 5Ј-3Ј exonucleolityc cleavage were shown to be active in trypanosomes (11). The identification of an exosome in T. brucei (12) suggests that an early mechanism of mRNA maturation, mediated by 3Ј-5Ј exonucleases involved in 3Ј-end processing, might be a similar although more primitive process, compared with that present in higher eukaryotic cells (13). Therefore, the possibility of regulating the decay of ARE-containing mRNAs exists in trypanosomes.
In this work, we delineate TcUBP-1 involvement in mRNA maturation processes by defining, in part, their protein ligands. We identified a ribonucleoprotein complex containing TcUBP-1 and demonstrated that it interacts specifically with two RRMtype RNA-binding proteins. The first ligand identified was TcPABP1, and its effect in the recognition of U-rich RNA by TcUBP-1 was studied. The second ligand was a novel 18-kDa U-rich RNA-binding protein named TcUBP-2. We provide evidence for the in vitro and in vivo interaction of TcUBP-1 with these ligands, as well as data suggesting that their association modulates TcUBP-1 binding with U-rich RNAs.

EXPERIMENTAL PROCEDURES
Parasite Cultures-T. cruzi CL-Brener cloned stock (14) was used. Different stages of the parasite were obtained as described previously (15).
Recombinant TcUBPs Protein Expression and Purification-Tcubp-1 and Tcubp-2 cDNAs and partial deletions of both were amplified by PCR (see Table I). They were cloned into the BamHI and EcoRI restriction endonuclease sites of the pGEX-2T vector (Amersham Biosciences), generating a glutathione S-transferase (GST) fusion and transformed in Escherichia coli strain BL21 DE3 pLysS (Novagen). Cultures were induced with isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. Recombinant proteins were purified using GST-agarose columns (Sigma). GST fusion proteins were cleaved with thrombin (Sigma). The reactions were performed with 1 unit of thrombin/100 g of protein in a buffer containing 20 mM Tris-HCl, pH 7.6, 150 mM ClNa, and 2.5 mM Cl 2 Ca during 2 h at 25°C.
Antibody Production and Western Blot Analysis-An antibody against TcUBP-1 RRM was prepared as described (10). Two specific peptides of TcUBP-1, VSQYDPYGQTAC, and TcUBP-2, RNRNGVST-FGAC, were synthesized (Sigma GENOSYS). The C-terminal cysteine residues were added to conjugate the peptides with maleimide-activated keyhole limpet hemocyanin according to the manufacturer's instructions (Pierce). Keyhole limpet hemocyanin-conjugated peptides were injected into rabbits and mice with Freund's adjuvant three times at 2-week intervals. For Western blot analysis, samples were fractionated on SDS-PAGE gels transferred to HybondC nitrocellulose (Amersham Biosciences), probed with primary antibodies, and developed using horseradish peroxidase-conjugated anti-rabbit and anti-mouse antibodies and the Supersignal West Pico chemiluminescent substrate (Pierce), according to manufacturer's instructions.
Protein Extract Preparation-Cytosolic protein extracts and subcellular fractionation were done as described (8).
Immunoprecipitations-A cytosolic extract corresponding to 10 8 parasites or gel filtration chromatography column fractions were incubated with mouse pre-immune serum, anti-TcUBP-1, anti-TcUBP-2, or anti-TcPABP1 polyclonal antibodies for 2 h at 4°C with gentle mixing. A 50% slurry of protein A-Sepharose was added to the mixture and incubated for 2 h more. The mixture was centrifuged at 1000 ϫ g for 1 min, and the resin was washed four times with 300 l of Tris-buffered saline-Tween 0.05%. Proteins were eluted with 2ϫ Laemmli buffer.
Gel Filtration Chromatography-Gel filtration was carried out with 300 l of an epimastigote protein extract (10 mg/ml) on a Bio-Sil SEC 250 column (Bio-Rad). A Superdex 75 HR 10/30 (Amersham Biosciences) was used for the determination of apparent molecular masses of the recombinant proteins. Columns were equilibrated in 20 mM Tris-HCl, pH 6.8, and 150 mM NaCl. Fractions of 500 l were collected and analyzed by Western blot. Where indicated, the extract was pretreated with RNase A (USB) at a concentration 0.1 g/l during 30 min.
Cross-linking Studies-TcUBP-1, TcUBP-2, and the deletion mutant proteins (Table I) were treated at room temperature with glutaraldehyde at a final concentration of 0.01% as indicated (17). The products were run on a SDS-PAGE and stained with Coomassie Blue.
Dihydrazide-agarose RNA Cross-linking-Homoribopolymers (rA, rC, rG, rU) were oxidized with NaIO 4 and cross-linked to adipic acid dihydrazide-agarose beads (Sigma) as indicated previously (18). Purified proteins or trypanosome extracts were incubated for 1 h with RNA cross-linked beads and washed. Protein elution was done with 2ϫ Laemmli buffer.
In Vitro Protein Binding Experiments-GST pull-down was carried out with 5 g of GST or GST fusion proteins immobilized on glutathione beads (Sigma) and 5 g of TcUBP-1 or TcUBP-2, without GST tag. Columns were equilibrated with Tris-buffered saline, and after washing with Tris-buffered saline-Triton 0.1%, eluted proteins were loaded onto a SDS-PAGE and stained with Coomassie Blue.
In Vitro Transcription-RNA production was done according to pre- a All constructions were cloned in the pGEX-2T vector (Amersham Biosciences). b The restriction sites used for the cloning procedure are underlined.

RNA-Protein
Interactions-Native mobility gel shift assays were done with the indicated protein amounts and RNA as previously described (10). All gel shift experiments were performed at least twice. Bands were quantified using Image Analysis software (Eastman Kodak Co.). K d values were calculated by plotting the ratio of bound RNA/free RNA against protein concentration.
Yeast Two-hybrid Assay-Saccharomyces cerevisiae L40 cells were transformed sequentially with pBTM116 plasmid expressing a LexA-DNA-binding domain (DBD) fusion protein and pVP16 plasmid expressing a VP16 transcription activation domain fusion protein and were grown at 28°C in synthetic medium Yc-TULLH (Trp Ϫ , Ura Ϫ , Leu Ϫ , Lys Ϫ , His Ϫ ) (19). An interaction between the DBD and the activation domain of the fusion proteins results in the production of the HIS3 gene product and enables the cells to grow on minimal medium lacking histidine. ␤-Galactosidase assays were carried out as described previously (19). Plasmids pBTM116 and pVP16 were a kind gift from Dr. Tellez-Inon (Instituto de Investigaciones en Ingeniería Genética y Biología Molecular).
TcUBP-2 Binds poly(U)-RNA and Is Expressed Differentially during Parasite Development-A Western blot analysis was performed with an anti-RRM polyclonal antibody using a total parasite extract of the epimastigote stage. The anti-RRM antibody recognized three protein bands of 18, 27, and 45 kDa ( Fig.  2A). Both 27-and 45-kDa protein bands were suggested to be TcUBP-1 isoforms, but only the 27-kDa band has the expected size for Tcubp-1 coding region. Previously, we observed that the 45-kDa isoform increases after parasite transformation with Tcubp-1. This result suggested that the 45-kDa isoform might correspond to a post-translationally modified form of TcUBP-1 (10). To simplify the nomenclature, TcUBP-1 will refer to the 27-kDa protein and TcUBP-1m, m from modified, to the 45-kDa protein.
To confirm that they are indeed codified by Tcubp-1, we raised antibodies against the N-terminal region of TcUBP-1 and named it anti-TcUBP-1 antibody. This antiserum identified both the 27-and 45-kDa bands, signals that were undetectable when the antiserum was pre-adsorbed with GST-TcUBP-1 protein (Fig. 2B). On the other hand, the 18-kDa protein band corresponds to the expected size derived for Tcubp-2 coding region (165 amino acids, 18.4 kDa). To confirm whether this protein is indeed TcUBP-2, we raised antibodies against a peptide of its central region from the amino acids 111 to 122 (Fig. 1A) and named it anti-TcUBP-2. This antibody recognized GST-TcUBP-2 and not GST-TcUBP-1 (Fig. 2C). Additionally, the specificity of the anti-TcUBP-2 antibody was demonstrated. Pre-adsorption of the antibody with GST-TcUBP-2 abolished its interaction with the 18-kDa polypeptide (Fig. 2C). This result demonstrated the specificity for the anti-TcUBP-2 antibody and also confirmed that the 18-kDa protein is indeed TcUBP-2. TcUBP-2 expression levels were analyzed by Western blot, using total parasite extracts from the different stages of the parasite. TcUBP-2 was detected preferentially in epimastigotes (Fig. 2D), showing that it is expressed differentially during parasite development. As control of this experiment, the same

mRNA-binding Protein Interactions in Trypanosomes
filter was reacted with an antibody made against the poly(A)binding protein (16), showing that it is detected in all stages of the parasite. To verify whether TcUBP-2 presents functional RNA-binding activity, an in vitro binding assay was performed. The homoribopolymers (rA, rC, rG, and rU) were cross-linked to dihydrazide-agarose beads. A trypanosome total protein extract from the epimastigote stage was prepared and passed through the four columns, and the eluted proteins were run in SDS-PAGE, followed by Western blot using an anti-RRM antibody. TcUBP-2, as TcUBP-1 isoforms, was recognized only in the fraction bound to poly(U) column and not to other homoribopolymers (Fig. 2E). The cellular localization of TcUBPs proteins was performed by a biochemical fractionation, demonstrating that they are located preferentially in the polysomal fraction (not shown).
A Ribonucleoprotein Complex Containing TcUBP-2 and the Destabilizing Factor TcUBP-1-Because TcUBP-1 was shown to be involved in post-transcriptional regulation (10), it might exert its effect through protein-protein interactions. Thus, we asked whether TcUBP-1 forms a protein complex in vivo in the epimastigote stage. Gel filtration experiments were performed with cell-free extracts pre-treated with Buffer A (150 mM NaCl) and Buffer B (150 mM NaCl, ϳ0.1 g/l RNase A). After pretreatment of the extracts for 30 min, they were applied on a Bio-Sil SEC 250 column, and the eluates were analyzed by Western blot using anti-RRM antibody. In the treatment of the extract with Buffer A, we observed that TcUBP-1 isoforms and TcUBP-2 were detected in a high molecular mass RNA-protein complex of ϳ450 kDa (Fig. 3A, lane 9). TcUBP-1 was also detected in fraction 17 of an apparent molecular mass of ϳ27 kDa. This band might correspond to the monomer form. Conversely, TcUBP-2 was detected between fractions corresponding to 17 and 45 kDa (Fig. 3A, lanes 16 -19) and also in fraction 12 of ϳ250 kDa. In the presence of Buffer B, all protein bands from the fraction corresponding to the ϳ450-kDa complex dis-appeared (Fig. 3B, lane 9), whereas the position of TcUBP-1 and TcUBP-2 in between fractions 16 and 19 did not change significantly. Thus, because Buffer B contains RNase A, these results suggest that an RNA component might be important for the RNA-protein interactions and complex formation. The amount of a trypomastigote stage protein extract used in two different gel chromatography experiments did not allow the detection of bands in Western blot analysis (not shown).
The RNA-Protein Complex Contains TcUBP-1/TcUBP-2 Heterodimer and TcPABP1-To determine whether TcUBP-1 and TcUBP-2 proteins interact in vivo, as suspected from the gel chromatography profiles from Fig. 3A, immunoprecipitation experiments were performed. Fraction 9 in Fig. 3A was pretreated or not for 30 min with RNase A and immunoprecipitated with pre-immune, anti-TcUBP-1, or anti-TcUBP-2 polyclonal antibodies followed by a Western blot using the anti-RRM antibody (Fig. 4A). Both TcUBP-1 and TcUBP-2 were detected independently of the treatment of the extract with RNase, suggesting that they were associated in vivo by protein-protein interaction. However, it was difficult to detect TcUBP-1m under these experimental conditions (Fig. 4A). TcUBP-1 and TcUBP-2 homo-and heterodimerization was corroborated through a two-hybrid assay, performed in L40 yeast strain and using LexA-DNA binding and VP16 activation domain hybrid proteins (see below).
In higher eukaryotic cells, PABP1 protein is part of a multiprotein complex formed in the 3ЈUTR of ARE-containing

mRNA-binding Protein Interactions in Trypanosomes
mRNAs (21). Because TcUBP-1 binds to SMUG 3ЈUTR and regulates its mRNA stability levels, we asked whether TcPABP1 might be present in this complex. An immunoprecipitation was carried out with anti-TcUBP-1, anti-TcUBP-2, or anti-TcPABP1 antibodies followed by a Western blot using the anti-TcPABP1 antibody. We found evidence that the ribonucleoprotein complex containing the heterodimer also contains TcPABP1 (Fig. 4B). Although two different bands of 55 and 66 kDa were detected, the 66-kDa band was described as being TcPABP1 (16).
TcUBP-1 and TcUBP-2 Interact in Vivo in a Two-hybrid System-A two-hybrid assay was performed to corroborate TcUBP-1 homo-and heterodimerization with TcUBP-2 in a heterologous system. The coding sequence of TcUBP-1 was fused to the LexA-DBD in the plasmid pBTM116 to serve as bait (19). Alone, LexA-DBD-TcUBP-1 did not give any detectable levels of GAL4-dependent synthesis of lacZ in the L40 yeast strain used (not shown). Therefore, TcUBP-1 and TcUBP-2 cDNAs were cloned into the pVP16 fused to VP16 activation domain. We test them in binding assay with LexA-DBD-TcUBP-1. The co-transformation of L40 yeast strain with LexA-TcUBP-1 and VP16-TcUBP-1 or VP16-TcUBP-2 allowed the growth of transformants on medium without histidine and also displays lacZ activity (not shown). Negative results were obtained with control yeast transformed with LexA-DBD-TcUBP-1 and VP16 alone. Thus, the interaction between TcUBP-1/TcUBP-1 and TcUBP-1/TcUBP-2 was specific. These results demonstrate that TcUBP-1 and TcUBP-2 proteins have the potential to interact in vivo in a heterologous system, corroborating the immunoprecipitation results (Fig. 4A).
Interaction between TcUBP Proteins and TcPABP1 in Different Stages of Differentiation-We next analyzed the interaction of TcUBPs and TcPABP1 throughout parasite development. Total trypanosome extracts were prepared from the different parasite stages, incubated with anti-TcUBP-1 antibody for immunoprecipitation, followed by Western blot with anti-TcPABP1. This experiment demonstrated that TcUBP-1 interacts with TcPABP1 in the epimastigote and trypomastigote stages (Fig. 4C). Conversely, TcUBP-2 co-precipitates with TcPABP1 in the epimastigote stage (Fig. 4B) and not in other parasite stages (not shown). Because the complex containing TcUBPs in epimastigotes was disrupted by RNase A treatment (Fig. 3B), we asked whether the interaction between TcUBP-1 and TcPABP1 in this stage was dependent on RNA. Immunoprecipitations were done with anti-TcUBP-1 antibody using cell-free extracts pre-treated or not for 30 min with RNase A. When the extract was RNase-treated there was no detectable TcPABP1 in the immunoprecipitate (Fig. 4D). Conversely, in the trypomastigote stage the interaction was not dependent on RNA, showing that TcUBP-1 and TcPABP1 make direct protein-protein contact.
TcUBP-1 and TcUBP-2 Present the Same Binding Affinity and Specificity to Homoribopolymers and U-rich RNA-The binding specificity of TcUBP-2 was tested and compared with TcUBP-1. An electrophoresis mobility shift assay (EMSA) was performed with a labeled U-rich element named P1, in the presence of different amounts of each of the four homoribopolymers (Fig. 5B). We found that TcUBP-1 and TcUBP-2 RNAbinding activity was competed in a concentration-dependent manner with poly(U) and not by poly(A) or poly(C). In addition, it can be competed at high concentrations (500ϫ) of poly(G) (Fig. 5B). These results confirmed that recombinant proteins recognize poly(U) RNA, as do the native ones (see Fig. 2E).
Different U-rich RNAs were used to determine by EMSA the apparent dissociation constants (K d ) of the ribonucleoprotein complexes formed by TcUBP-2 and TcUBP-1 (Fig. 5). The apparent K d for each reaction was calculated by determining the protein concentration at which 50% of the RNA was bound. TcUBP-2 recognized P1 RNA with an apparent K d of ϳ100 nM, and TcUBP-1 recognized it with a K d of ϳ150 nM. A new RNA, which has three uridine to guanosine changes in the last U-rich stretch of P1, was made and named P1-GGG (see Fig. 5A). These changes had a slight effect on TcUBP-2 K d (ϳ200 nM) and TcUBP-1 K d (ϳ250 nM). As it was described with TcUBP-1 (10), the S RNA (see Fig. 5) is bound by TcUBP-2 with a slightly high K d (ϳ250 nM). Conversely, P2 and P3 RNAs (Fig. 5), which  5. TcUBP-1 and TcUBP-2 bind U-rich RNAs with similar specificity and affinity. A, sequences of the RNAs used in the EMSA experiments. B, an EMSA with ϳ1 M TcUBP-1 or TcUBP-2 and P1 RNA was performed and competed with increasing amounts of homoribopolymers (1, 10, 50, and 500ϫ). RNA, binding reaction without protein; (Ϫ), binding reaction without unlabelled competitor. C, an EMSA was performed to determine the apparent dissociation constants between TcUBP-2 and the indicated RNAs. D, EMSA to determine the apparent dissociation constants between TcUBP-1 and the indicated RNAs. The amount of proteins used in panels C and D were 0, 10, 50, 250, 500, and 1000 nM as indicated.

mRNA-binding Protein Interactions in Trypanosomes
have shorter U-rich stretches, were recognized by TcUBP-2 but with lower affinities (K d values of ϳ750 and ϳ850 nM, respectively) (Table II). These results suggest that U-rich stretches are the most important component of the RNA recognized by TcUBPs proteins. Also, GU-rich RNAs are better recognized than AU-rich sequences, and the RNA-binding activity of TcUBP-1 was clearly similar to TcUBP-2 (Fig. 5, C and D). A summary of TcUBPs RNA-binding activities is shown in Table II. This comparison reflects that TcUBP-1 and TcUBP-2 have similar K d values to the U-rich RNAs tested in this work.
Mapping of the Minimal Region Required for RNA Binding in TcUBPs Proteins-We demonstrated that TcUBP-1 and TcUBP-2 recognized the P1 RNA with similar apparent K d values (Fig. 5). These results suggest that the differences in the primary sequence between both proteins might not contribute to RNA recognition and/or modulation of the RNA-binding activity. In addition, TcUBP-1 and TcUBP-2 dimerize when binding to RNA containing U-rich regions, and therefore, we hypothesize that the auxiliary domains might be crucial for their function in protein-protein interaction. To verify this hypothesis, several deletion mutants were constructed by PCR (Table  I). The different proteins were named TcUBP-1⌬N, lacking the N-terminal region; TcUBP-1⌬Q, lacking the glutamine-rich region; TcUBP-1⌬NQ, lacking its N-and glutamine-rich regions; and two other constructs named TcUBP-1⌬QG1 and TcUBP-1⌬QG2 (Fig. 6A). The difference between these two last deletion mutants resides in their VR. The first one has this portion, composed by 14 extra amino acids (PGIAGAVGDGNGYL), after the predicted ␤4 sheet (Fig. 6A). However, both lack the motif GAYGGYGAY within the glycine-rich region. Similarly, TcUBP-2 deletions were TcUBP-2⌬N, lacking the N-terminal region; TcUBP-2⌬C, which lacks the C-terminal region; and TcUBP-2⌬CG1 and TcUBP-2⌬CG2 (Fig. 6C). The difference between these two last proteins is that the first one presents the VR region, composed by nine extra amino acids (NRNGVSTGF), in the C-terminal region.
The predicted secondary structure and heteronuclear-nuclear Overhauser efffect NMR spectra of TcUBPs 2 suggest that the auxiliary domains behave as random-coil, so none of these deletions should have any effect on TcUBP-1 and TcUBP-2 final folding. The apparent K d values of all TcUBPs constructions were determined by EMSA. The majority of them present similar K d values of ϳ100 -200 nM (Fig. 6, B and D). However, TcUBP-1⌬QG2 showed a K d Ͼ1000 nM (Fig. 6B), showing that the RNA affinity of this mutant is decreased more than 5-fold. Conversely, TcUBP-1⌬QG1 that contains the VR region at the C-terminal of the RRM motif showed the same K d as the complete TcUBP-1 protein. Similarly, all TcUBP-2 mutants, except for TcUBP-2⌬CG2, interact with RNA with the same affinity as the wild type TcUBP-2 protein (Fig. 6D). These experiments demonstrate that the ϳ92-amino acid RRM motif is the minimal region required for RNA binding in TcUBPs proteins. The function of C-terminal VR extension in the RRM motif require further investigation. It might be homologous to that of the RRM C-terminal extension present in RNA-binding proteins from other cell types (see "Discussion").
The Glycine-rich Region of the Auxiliary Domain Is Critical for Dimerization-A decrease in the mobility of the ribonucleoprotein complex formed in the presence of protein amounts greater than 500 nM was observed with TcUBP-1 and all mutant proteins except for TcUBP-1⌬QG1 and TcUBP-1⌬QG2 (Fig. 6). Both proteins lack the glycine-rich region. Thus, these results suggest that this region might be important for dimerization. To study the formation of dimers in vitro, we performed cross-linking reactions with glutaraldehyde at a final concentration of 0.01% (Fig. 7). This is a zero-length cross-linker reagent that was described to form covalent bonds between proteins that are in contact (17). For this purpose, different TcUBP-1 mutant proteins listed in Fig. 7A were analyzed. Dimers were detected at concentrations greater than 500 nM (not shown). TcUBP-1⌬Q protein, which presents the VR-and glycine-rich regions within the C-terminal, was cross-linked efficiently (Fig. 7B, lane 4). The percentage of cross-linking in this protein was about 10 -15%. Similar values of cross-linking were obtained with proteins from other cell types (17). Conversely, TcUBP-1⌬QG1 mutant protein formed dimers much less efficiently than TcUBP-1⌬Q (Fig. 7B, compare lanes 4 and  6), and TcUBP-1⌬QG2 did not react with the cross-linker (Fig.  7B, lane 8). The formation of dimers in TcUBP-1 was hardly detectable with the amount of protein used in this assay (Fig.  7B, lane 2). It might be possible that its C-terminal glutamine- rich region regulates protein contacts negatively because of its presumed random-coil nature (see "Discussion"). Finally, the same cross-linking results were obtained with TcUBP-2 and its mutant proteins (not shown), suggesting that the glycine-rich region is the principal domain of both proteins required for dimer formation. A chromatographic analysis was performed to corroborate dimer formation on TcUBPs proteins. A similar elution profile was detected with TcUBP-1⌬Q and TcUBP-2⌬C. Both elute at a position corresponding to dimer and monomer forms (not shown), thus confirming the dimerization results.
The Formation of TcUBP-1, but Not TcUBP-2, Homodimers Is Disrupted by TcPABP1 Interaction in Vitro-We have shown previously that TcPABP1 interacts in vivo with TcUBPs proteins in both epimastigote and trypomastigote stages (Fig. 4). Thus, we asked whether this interaction is also detected in vitro. For these propose, Tcpabp1 was cloned and expressed as a GST fusion protein, rendering GST-TcPABP1. To determine whether TcUBP-1 and TcUBP-2, without GST tag, interact in vitro with GST-TcPABP1, a GST pull-down assay was done. Under the same assay conditions, GST-TcPABP1 interacts with TcUBP-1 but failed to interact with TcUBP-2 (Fig. 8A), suggesting that the binding of TcUBP-1 to TcPABP1 is mediated by protein-protein interactions and not mediated by RNA (Fig. 8A). We also analyzed by EMSA the effect of adding increasing concentrations of TcPABP1 in TcUBP-1 and TcUBP-2 U-rich RNA-binding assays. To ensure complete RNA recognition, ϳ2 M TcUBP-1 and TcUBP-2 proteins were used. Although no observable effect was detected on TcUBP-2 RNAbinding, TcUBP-1-binding activity was modified by increasing concentrations of TcPABP1 (Fig. 8B). TcPABP1 did not displace TcUBP-1 protein from RNA, because no free RNA is detected at high amounts. Conversely, it disrupted or knocked out dimer formation on U-rich RNA, rendering a monomeric RNA recognition pattern in solution (Fig. 8B). TcUBP-1 interaction with TcPABP1 might involve N-or C-terminal regions of TcUBP-1, because these regions are different between TcUBP-1 and TcUBP-2.

DISCUSSION
In this work we have provided evidence for the existence of two very closely related RNA-binding proteins, TcUBP-1 and TcUBP-2, that present an identical RRM motif with distinct Nand C-terminal regions. The C-terminal region of TcUBP-1 can be divided into two parts, a short glycine-rich region and a large glutamine-rich random-coil domain. Conversely, TcUBP-2 presents a glycine-rich region larger and with three YGG motives (Fig. 1). TcUBP-1 presents affinity for U-rich elements in vitro and in vivo (10; and this work), and both proteins recognize with high affinity AU-and GU-rich elements. Our ongoing NMR structure of TcUBP-1 and TcUBP-2 suggest that they present a similar ␤␣␤␤␣␤ topology. Thus, this might explain why both proteins present similar K d values with the RNAs used for the EMSA experiments (see Table II). Although the RRM is almost identical, the C-terminal extension of this motif, named VR region, is different, and it might have a regulatory role. Deletion of the VR region produces an increase in the apparent K d in both TcUBPs proteins (Fig. 6). Similarly, an NMR and biochemical study of U1A RRM motif but lacking its C-terminal extension showed that it has the same folding topology but does not bind RNA (22). Moreover, these residues were involved in structure stability of the free protein (23). In heterogeneous nuclear RNP-D, the C-terminal of RRM2 was important in conferring high affinity for ARE binding (24). This phenomenon suggests that those residues FIG. 8. Protein-protein interaction between TcUBP-1 and TcPABP1. A, GST pull-down assay between GST or GST-TcPABP1 and TcUBP-1 or TcUBP-2 proteins, in the absence of RNA. Input, 10% of the protein added to the pull-down assay. The integrity of the unbound proteins was monitored by SDS-PAGE after reaction was completed (not shown). B, an EMSA with TcUBP-1 or TcUBP-2 proteins and P1 RNA was performed. The effect of increasing amounts of TcPABP1 (100, 250, 500, and 1000 nM) was tested. The position of the monomer and dimer forms is indicated.

mRNA-binding Protein Interactions in Trypanosomes
localized in the VR region might be important in conferring structure stability by a possible interaction with residues in the ␤1␤3 sheets that make contact with the RNA.
Of great importance is the modular RRM organization of RNA-binding proteins in higher eukaryote cell types. They present multiple RRM motifs, and its composition is what determines the final affinity for the RNA. The first RRM (RRM1) is the most important in terms of RNA affinity and specificity. However, the other RRM motifs have an accessory function, such as oligomerization and localization (25), structure stability in solution (23), and protein-protein interaction (26). Analysis of HuR deletion mutants allowed Park et al. (27) to suggest that RRM1 is critical for ARE binding but that at least one additional RRM is required to achieve a K d in the nanomolar range. Both RRM1 and RRM2 are important for ARE affinity, even though they are not sufficient (24). It is important to conclude that certain RNA-binding proteins require only one domain to achieve RNA-binding specificity and affinity, such as U1A (28). Conversely, other proteins have been shown to require combinations of at least two (27) or occasionally more domains (29). Several common features have emerged from the comparison of two tandem RRM structures such as those in Sxl (30) and PABP1 (31). Each of these proteins contains short interdomain linkers of about 8 -11 residues. These regions are highly mobile in the free protein making the two domains structurally independent (32).
Homodimerization, the common feature of prokaryotic transcription factors, is extended in eukaryotes by heterodimerization. This possibility increases the range of DNA sequences that can be recognized and the degree of binding specificity. In evolutionary terms, this increase in range is necessary when genomes become larger and more complex (33). We have provided here evidence demonstrating that the glycine-rich region of TcUBPs is important for dimerization (see Figs. 6 and 7). However, the fact that no dimers can be detected by crosslinking with TcUBP-1, at difference with TcUBP-1⌬Q (Fig. 7), suggests that its C-terminal region might regulate the accessibility and interaction with itself and/or other proteins. It was described that shorter coiled-coils of some transcription factors might either promote or prevent formation of homo-and/or heterodimers (33). Thus, it is likely that in the parasite both homo-and heterodimerization might be regulated processes. In this work we showed that TcUBPs, having a single RRM motif, bind RNA with high affinity and can dimerize by the glycinerich flexible region. Thus, an even higher affinity might be achieved in vivo through homo-and/or heterodimerization of TcUBPs, allowing the simultaneous binding of two RRMs to the RNA. The glycine-rich region might contribute to the mobility and flexibility of the RRM motif in the dimer form, as deduced from our results and comparison with RRM-type proteins in other cell types. In general terms, the glycine-rich domains found in several RNA-binding proteins do not have any structural features beyond the likelihood that stretches of residues such as Gly or Pro result in flexible coils. This region was demonstrated to be involved in protein-protein interactions and to contribute to the RNA-binding free energy (34,35).
TcUBP-1 and TcUBP-2 form part of a ribonucleoprotein complex containing TcPABP1. TcUBP-2 co-precipitates in the epimastigote stage, whereas TcUBP-1 precipitates with TcPABP1 in epimastigote and trypomastigote stages (Fig. 4). In vitro, TcUBP-1 interacts with TcPABP1, in the absence of RNA, and the effect of this interaction was to disrupt the formation of TcUBP-1, but not TcUBP-2, homodimers (Fig. 8). The domains involved in these interactions have not been identified. However, we speculate that the C terminus of TcUBP-1 might be involved, because this is the most dissimilar region when comparing with TcUBP-2. Although TcUBP-1 lacks the PABC consensus sequence described for PABP1 partners in human (36), this possibility must be tested further. In higher eukaryotes, when complexed to the poly(A)-tail, PABP1 circularizes mRNA molecules via its interaction with translational initiation factors, which results in mRNA stabilization (21). Recently, it has been shown that binding of some AU-rich-binding proteins to ARE may alter the interaction between PABP1 and the poly(A)-tail, thereby providing access to a poly(A)-ribonuclease (37). Rapid mRNA degradation involves the recognition of AREs by AU-rich-binding proteins, several of which have been shown to recruit the exosome and cause the reduction of PABP1 affinity for the poly(A)-tail (37). In our model (Fig. 9), the interaction of TcUBP-1 and TcPABP1 in vitro disrupts the formation of TcUBP-1 homodimers and might also change the affinity of TcPABP1 for the poly(A)-tail. The possibility that in trypomastigotes, the presence of TcUBP-1 protein without TcUBP-2 on SMUG 3ЈUTR may recruit the exosome or a poly(A)-ribonuclease activity (Fig. 9), is unlikely (21). We suggest that, in the epimastigote stage, the presence of TcUBP-2 and other yet unidentified factors might stabilize the interaction between TcUBP-1 and TcPABP1 within the ribonucleoprotein complex. This can prevent, at least in part, the disruption of TcUBP-1 homodimers (Fig. 8). Although our results provide evidence for the role of these ARE-binding proteins in the regulation of mRNA turnover in trypanosomes, establishing the mechanism by which TcUBPs stimulate mRNA stabilization/destabilization will require additional experiments.