Central Region of the Human Splicing Factor Hprp3p Interacts with Hprp4p* 210

Human splicing factors Hprp3p and Hprp4p are associated with the U4/U6 small nuclear ribonucleoprotein particle, which is essential for the assembly of an active spliceosome. Currently, little is known about the specific roles of these factors in splicing. In this study, we characterized the molecular interaction between Hprp3p and Hprp4p. Constructs were created for expression of Hprp3p or its mutants in bacterial or mammalian cells. We showed that antibodies against either Hprp3p or Hprp4p were able to pull-down the Hprp3p-Hprp4p complex formed in Escherichia coli lysates. By co-immunoprecipitation and isothermal titration calorimetry, we demonstrated that purified Hprp3p and its mutants containing the central region, but lacking either the N-terminal 194 amino acids or the C-terminal 240 amino acids, were able to interact with Hprp4p. Conversely, Hprp3p mutants containing only the N- or C-terminal region did not interact with Hprp4p. In addition, by co-immunoprecipitation, we showed that intact Hprp3p and its mutants containing the central region interacted with Hprp4p in HeLa cell nuclear extracts. Primer extension analysis illustrated that the central region of Hprp3p is required to maintain the association of Hprp3p-Hprp4p with U4/U6 small nuclear RNAs, suggesting that this Hprp3p/Hprp4p interaction allows the recruitment of Hprp4p, and perhaps other protein(s), to the U4/U6 small nuclear ribonucleoprotein particle.

(1-3). Each snRNP particle consists of one (U1, U2, and U5) or two (U4/U6) snRNAs complexed with a set of Sm or Sm-like proteins and several particle-specific proteins (4 -6). These snRNPs recognize conserved sequences of the pre-mRNA and assemble into a catalytically active spliceosome that catalyzes the two cleavage-ligation reactions of pre-mRNA splicing (1,2,7,8). Spliceosome assembly follows an ordered pathway through the formation of several intermediate complexes (2). First, a pre-spliceosome (A complex) is formed once U1 and U2 snRNAs (as part of the U1 and U2 snRNPs) associate with the conserved 5Ј-splice site and the branch point of the intron, respectively (9,10). Then, U4/U6 snRNP, in which U4 and U6 snRNAs base pair over an extended complementary region, forming a Y-shaped junction, is recruited together with U5 snRNP, presumably via protein/protein interactions, to the pre-spliceosome to form the mature spliceosome (B complex) (11)(12)(13).
Prior to the first catalytic step of splicing, important conformational rearrangements occur to create a catalytically active spliceosome. For example, U1 snRNA dissociates from the 5Јsplice site, and the U4/U6 snRNA association is disrupted (14 -16), leaving the U6 snRNA free to base pair with both the U2 snRNA and the 5Ј-splice site (17). The U2 and U6 snRNAs, together with the pre-mRNA, may form the catalytic core of the spliceosome, whereas the U5 snRNP interacts with exonic sequences adjacent to both splice sites, possibly aligning the exons during splicing (18). After a round of splicing, the snRNPs are recycled to participate in further splicing, whereas the intron RNA is degraded (19 -21).
The assembly of U4/U6 snRNP, its recruitment into the pre-spliceosome, and numerous conformational changes of its snRNA components are poorly understood processes. Proteins associated with U4 and U6 snRNAs are of particular interest because they could mediate the conformational changes that occur between U4 and U6 snRNAs (4,22) and the recruitment of the U4/U6 snRNP to the pre-spliceosome (23). U4/U6 snRNP contains at least five specific proteins, viz. the 15.5-kDa protein (Snu13p in yeast) that binds the 5Ј-stem-loop of U4 snRNA (24 -27), the 61-kDa protein (Prp31p in yeast) (4), and the heterotrimer cyclophilin H-Hprp4p-Hprp3p complex (26 -29). We previously identified, along with others, Hprp3p and Hprp4p as homologs of the yeast U4/U6 snRNP-specific factors Prp3p and Prp4p, respectively (28,30,31). Hprp3p and Hprp4p can be isolated from HeLa cells together with the 20-kDa cyclophilin H as a heterotrimer protein complex (32). Within this complex, cyclophilin H associates with Hprp4p, but not with Hprp3p (28,29), suggesting that Hprp3p may interact directly with Hprp4p in the complex.
On the other hand, it has been shown that human Hprp3p and Hprp4p failed to interact with each other in the yeast two-hybrid system (30). Therefore, whether these two human splicing factors directly interact with each other and how they interact with the U4/U6 snRNP remain unclear. Understanding these interactions within the spliceosome is of particular importance because it will help to elucidate the molecular mechanisms of an essential genetic process, splicing. Here, we show that Hprp3p directly interacts with Hprp4p. Furthermore, we have identified the domain of Hprp3p involved in this interaction as being within the central region of this molecule. Our results suggest a possible role for Hprp3p in the recruitment of Hprp4p for the U4/U6 snRNP assembly.

Preparation of HA-tagged Hprp3p Mutant Constructs for Expression in Mammalian and Escherichia coli Systems-
The pcDNA3-3xHA-HPRP3 construct containing the 2.2-kb coding region of Hprp3p (31) was used as a template to create Hprp3p deletion mutants. The coding region of the mutants was amplified by PCR using the following primers: N1, 5Ј-aaaggtaccagatctaccatgggtgggcggatcttttac-3Ј; N2, 5Ј-aaaagcggccgcttcatcatgaatgatgccattgag-3Ј; N3, 5Ј-aaaagcggccgcaagaaggaacagaaaaaacttc-3Ј; C1, 5Ј-aaactcgagtcaagtggcagcctgggagggctg-3Ј; C2, 5Јaaactcgagtcacttggtaagatatactcccag-3Ј; and C3, 5Ј-agtcgaggctgatcagccag-3Ј. All PCR amplifications were performed with the Pfu DNA polymerase (Stratagene, La Jolla, CA). The design of the Hprp3p mutant constructs is illustrated in Fig. 2B. The 5Ј-primer N1 and either the 3Ј-primer C1 or C2 were used to generate deletion mutant fragments I and II, respectively. Either the 5Ј-primer N2 or N3 in combination with the 3Ј-primer C3 were used to generate deletion mutant fragments III and IV, respectively. To build mammalian expression constructs, all four deletion fragments were subcloned at the NotI and XhoI restriction sites of a modified pcDNA3 vector (Invitrogen, Carlsbad, CA) containing an HA epitope tag just upstream of the HPRP3 start codon. For construction of the corresponding set of E. coli expression plasmids, the HA epitope-containing BamHI-XhoI fragment from pcDNA3-HPRP3 and those from pcDNA3-HPRP3 deletion mutants were subcloned individually into pET28a (Novagen, Madison, WI) bearing an N-terminal His tag for expression in the E. coli system. pET28a-HPRP4 was previously constructed (31). DNA sequencing with an automated DNA sequencing system (Li-Cor Model 4000L) confirmed the final constructs.
Protein Overexpression and Purification-E. coli BL21(DE3) cells harboring pET28a carrying HPRP4, HPRP3, or its mutants were cultured at 37°C in LB medium containing 50 g/ml kanamycin and 0.4% glucose. When the culture reached A 600 ϳ 0.6, isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 0.1 mM, and the cells were allowed to grow for an additional 7 h at room temperature to induce the target proteins. Cells were harvested by centrifugation at 8000 ϫ g for 10 min at 4°C. The cell pellet was resuspended in 0.04 volume of the original culture of sonication buffer (100 mM sodium phosphate, 500 mM KCl, 10% glycerol, 1ϫ protease inhibitor mixture (Roche Molecular Biochemicals), 8 mM ␤-mercaptoethanol, and 0.1% Tween 20, pH 7.5). The cell suspension was sonicated and centrifuged at 8000 ϫ g for 30 min. The supernatant (cell lysate) was used for immunoprecipitation assays and for further affinity purification of the His-tagged proteins. The relative content of induced protein was examined by 8% SDS-PAGE. The Bradford assay (33) was used to determine the total protein content in the soluble fraction.
Immunoprecipitation Assays-Fifty l (2 g/l total protein) of Hprp3p or Hprp4p soluble cell lysates alone (for negative controls) or Hprp4p and HA-Hprp3p soluble cell lysates together were incubated with rabbit preimmune serum (negative controls) or antiserum (against either Hprp3p or Hprp4p) in 500 l of buffer A (100 mM potassium phosphate, 500 mM KCl, 1ϫ protease inhibitor mixture, and 0.1% (v/v) Tween 20, pH 7.5) for 1 h at 4°C with constant mixing. For the experiment shown in Fig. 1C, buffer A contained 150, 250, or 500 mM KCl. Ten l of preimmune serum or antiserum and 10 l of protein Aor protein G-agarose beads were used in each experiment. After the incubation, the complex was washed five times with 1 ml of buffer A and twice with 1 ml of buffer B (100 mM potassium phosphate, 50 mM KCl, 1ϫ protease inhibitor mixture, and 0.1% (v/v) Tween 20, pH 7.5). The bound proteins were resolved by 8 or 10% SDS-PAGE and immunoblotted with the enhanced chemiluminescence system (ECL, Amersham Biosciences). Similar experiments were performed using purified Hprp4p and HA-Hprp3p or HA-Hprp3p mutants (10 g of protein). Immunoprecipitation assays were also carried out using purified proteins (10 g) and HeLa nuclear extracts (150 g of total protein). In this experiment, the HA-Hprp3p constructs were incubated with HeLa nuclear extracts, and the mixture was then added to protein G-agarose beads pre-bound with anti-HA antibody. Hprp4p was detected in the pulled-down complexes by Western blotting. Nuclear extracts from HeLa cells transiently transfected with HA-Hprp3p mutants were also used for immunoprecipitation experiments in a similar way. Rabbit antisera against Hprp3p or Hprp4p were generated in our animal facility. The mouse monoclonal anti-HA antibody was purchased from BABCO.
Isothermal Titration Microcalorimetry-The ITC data were generated using a Microcal VP-ITC instrument at 25°C. All solutions were carefully degassed before titration. Each titration experiment consisted of 28 injections of 10 l of Hprp3p (20 M) or Hprp3p mutant (50 M) into a cell containing 1.8 ml of 16 M Hprp4p. Titrations were conducted in 100 mM potassium phosphate, pH 6.0, 500 mM KCl, and 10% glycerol. Phosphate buffer was chosen by virtue of its small ionization enthalpy change. To correct for dilution and mixing effects, a series of control injections was carried out in which the heat of dilution was measured in blank titrations by injecting the protein into the buffer and then subtracted from the binding heat.
Transfection and Immunostaining-HeLa cells (ATCC CCL2) were cultured in ␣-minimal essential medium with 10% (v/v) fetal bovine serum at 37°C. Cells for immunostaining were seeded in six-well plates with glass coverslips and transfected at 40 -60% confluency with 1 g of DNA and 12 g of LipofectAMINE (Invitrogen) for each well under serum-free conditions as recommended by the manufacturer. Cells for nuclear extract preparation were seeded in 10-cm dishes and transfected at 40 -60% confluency with 6 g of DNA and 72 g of Lipo-fectAMINE for each dish under serum-free conditions. Immunostaining was performed at room temperature 24 h post-transfection. The transfected cells grown on coverslips were washed three times with PBS and fixed for 20 min with 4% (w/v) paraformaldehyde in PBS. The plasma membrane was permeabilized with 0.2% (v/v) Triton X-100 in PBS for 10 min. The cells were incubated in blocking solution (0.5% bovine serum albumin in PBS) for 1 h and then with mouse monoclonal anti-HA antibody in blocking solution for 1 h. The cells were washed three times with PBS and incubated with FITC-labeled goat anti-mouse IgG (H ϩ L) antibody (Invitrogen). Following three washes with PBS, the coverslips were mounted in Vectashield mounting medium with 4,6-diamidino-2-phenylindole. Fluorescence micrographs were recorded using a 100ϫ objective. All micrographs are representative of experiments repeated at least three times.
Nuclear Extract Preparation-The HeLa nuclear extracts were prepared as described previously (34,35). HeLa cells in suspension culture were harvested in early mid-log phase, and the pellet were washed with 5ϫ packed cell volume and resuspended in 2ϫ packed cell volume of buffer C (10 mM HEPES, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM dithiothreitol, pH 7.9). The cells were lysed by 30 strokes with a Dounce homogenizer (A-type pestle). The nuclei were pelleted by centrifugation at 750 ϫ g for 5 min at 4°C. The nuclear pellet was resuspended in buffer D (20 mM HEPES, 10% glycerol, 1.5 mM MgCl 2 , 0.42 M KCl, 0.5 mM dithiothreitol, 0.2 mM EDTA, and 0.5 mM phenylmethylsulfonyl fluoride, pH 7.9). The nuclei were lysed by 30 strokes with a B-type pestle. The resulting lysate was incubated at 4°C for 30 min with gentle agitation and then centrifuged at 25,100 ϫ g for 30 min at 4°C. The resulting supernatant was dialyzed against 3 liters (1 liter ϫ 3 changes) of dialysis buffer (20 mM HEPES, 10% glycerol, 1.5 mM MgCl 2 , 0.1 M KCl, 0.5 mM dithiothreitol, 0.2 mM EDTA, and 0.5 mM phenylmethylsulfonyl fluoride, pH 7.9) for 3.5 h. The nuclear extracts from HeLa cells transfected with pcDNA-HA-HPRP3 or its deletion mutants were prepared in a similar manner. The nuclear extracts were quick-frozen in liquid nitrogen and stored at -80°C.
Primer Extension-The presence of U4 and U6 snRNAs in immunocomplexes precipitated with anti-HA antibodies was determined by primer extension. RNA extractions from immunocomplexes and primer extension were performed as described (31,36). Two primers (U4-82, 5Ј-ggtattgggaaaagttttcaattag-3Ј; and U6-91, 5Ј-tcacgaatttgcgtgtcatcctggc-3Ј) that yield extension products specific to human U4 (82 nucleotides) and U6 (91 nucleotides) snRNAs, respectively, were used in the experiment. Primers were labeled at the 5Ј-end with [␥-32 P]ATP using T4 polynucleotide kinase (MBI Fermentas Inc., Burlington, Ontario, Canada). To anneal the end-labeled primers to their target snRNAs, the U4 and U6 primers (ϳ5 pmol each or ϳ15,000 cpm) were added, individually or together as indicated in the legend to Fig. 5C, to RNA samples in a solution containing 50 mM Tris-HCl, pH 8.3, and 25 mM KCl. The mixtures were incubated at 90°C for 2 min and then slowly cooled to 30°C. Extension reactions (20 l) were carried out at 37°C for 1 h in buffer containing 50 mM Tris-HCl, pH 8.3, 50 mM KCl, 10 mM dithiothreitol, 10 mM MgCl 2 , and 200 M each dNTP plus 1 unit of OmniScript reverse transcriptase (QIAGEN Inc.). The extended products were phenol-extracted, ethanol-precipitated, and resuspended in 5 l of loading buffer (85% formamide, 20 mM EDTA, 0.1% bromphenol blue, and 0.1% xylene cyanol). They were then resolved on 5% acrylamide gels containing 8 M urea and Tris borate/EDTA, and their migration positions were visualized by autoradiography. For primer extension controls, total nuclear RNAs (0.1 g) from HeLa cells were used.

Interaction of Hprp3p and Hprp4p
Expressed in E. coli-Hprp3p and Hprp4p are known to be present in the same  Fig. 1 in the presence of 500 mM KCl, and immunoblotted (I.B.) with monoclonal anti-HA antibodies. As a control, the same mixture of HA-Hprp3p and Hprp4p was added to protein Aagarose beads pre-blocked with rabbit preimmune serum (PIS) (lane 6). Lower panel, the same blot as in the upper panel was analyzed with polyclonal anti-Hprp4p antibody (diluted 1:4000). The position of the band corresponding to Hprp4p is indicated. The asterisk indicates background due to the polyclonal antibody used for the co-immunoprecipitation B: shown is a schematic diagram depicting the size and origin of full-length Hprp3p and its mutants produced from their corresponding mammalian or bacterial expression vectors. For mammalian expression, the corresponding coding sequences for these proteins were cloned into pcDNA3 with the HA epitope added to the N terminus. For bacterial expression, the same set of HA-tagged DNA sequences were cloned into pET28a. The origin of each mutant protein was defined by the amino acid positions indicated by the numbers. ⌬, Hprp3p mutant. U4/U6 snRNP particle (28,30,31). However, their role in U4/U6 snRNP assembly and the nature of their interaction are not clear. Because the splicing machinery is absent in bacterial cells, we first studied the interaction of Hprp3p and Hprp4p using soluble cell lysates of E. coli cells expressing these proteins. The cDNAs encoding Hprp3p and Hprp4p were cloned into pET28a and expressed in E. coli. The expression conditions were optimized for the yield of soluble proteins. More than 60% of Hprp3p and Hprp4p are soluble when cells are induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside at room temperature for 6 -8 h. After cell lysis under native conditions (see "Materials and Methods"), the soluble cell lysates containing Hprp3p or Hprp4p were used for co-immunoprecipitation analysis with polyclonal antibodies against Hprp3p or Hprp4p in buffer containing 150 mM KCl. As shown in Fig. 1A, Hprp3p was detected in the complex immunoprecipitated with anti-Hprp4p antibodies (lane 1), but was not present in the sample immunoprecipitated with preimmune sera (lane 4). As expected, anti-Hprp4p antibodies failed to pull-down Hprp3p in the absence of Hprp4p (Fig. 1A, lane 2), indicating that there is no cross-reaction with Hprp3p and antibodies against Hprp4p. Likewise, anti-HA antibodies did not recognize Hprp4p (Fig.  1A, lane 3), suggesting there is no cross-reaction between Hprp4p and the anti-HA antibodies used in this experiment. Reciprocal experiments using anti-Hprp3p antibodies to precipitate the complex and anti-Hprp4p antibodies to detect the presence of Hprp4p yielded the same conclusion, i.e. Hprp3p and Hprp4p are present in the same protein complex (Fig. 1B,  lane 1).
Hprp3p and Hprp4p together with the 20-kDa cyclophilin H can be purified from HeLa cells as a heterotrimer protein complex in the presence of Ͼ500 mM NaCl (28). To examine whether the Hprp3p-Hprp4p complex formed in bacterial cell lysates can withstand similar conditions, co-immunoprecipitation experiments were performed with increasing salt concentrations (up to 500 mM KCl). As shown in Fig. 1C (lanes 1-3), the Hprp3p-Hprp4p complex was not disrupted even at 500 mM KCl. These results support the notion that Hprp3p and Hprp4p interact with each other. However, because these experiments were performed with soluble cell lysates, the possibility of a third partner in the system nonspecifically mediating the interaction of the two proteins cannot be discarded. For example, the interaction may proceed through another protein or even RNA that contacts both Hprp3p and Hprp4p. To exclude this possibility, His-tagged HA-Hprp3p and Hprp4p were produced and purified to Ͼ90% homogeneity using Ni 2ϩ -nitrilotriacetic acid affinity chromatography (Supplemental Fig. 1). Equal amounts of the two highly purified proteins were incubated and added to protein A-agarose beads pre-absorbed with antisera against Hprp4p. The presence of Hprp3p in the complex precipitated by antisera against Hprp4p was confirmed by immunoblotting using antibodies against the HA tag of Hprp3p ( Fig.  2A, upper panel, lane 5). Because highly purified proteins were used in this experiment, it is unlikely that molecules in E. coli cell lysates would play a role in mediating the Hprp3p/Hprp4p interaction.
Central Region of Hprp3p Mediates the Hprp3p/Hprp4p Interaction-Currently, the domain structure and function of Hprp3p are virtually unknown. Because we demonstrated above that Hprp4p is likely to interact with Hprp3p, the first step toward unraveling Hprp3p domain structure and function is to determine the regions responsible for such an interaction. Therefore, we generated four HA-tagged Hprp3p deletion mutants (Hprp3p mutants I-IV) (Fig. 2B). Each mutant was cloned into pET28a for the expression of His-tagged HA-Hprp3p in E. coli. As with the wild-type protein, each purified Hprp3p mutant was incubated with purified Hprp4p, and the complex was co-immunoprecipitated with anti-Hprp4p antibodies. Similar to full-length Hprp3p, mutants II and III formed a complex with Hprp4p ( Fig. 2A, upper panel, lanes 2  and 3, respectively), but mutants I and IV did not (lanes 1 and  4, respectively). This result indicates the formation of an Hprp3p-Hprp4p complex, in particular through the central region of Hprp3p, and suggests that neither the N-nor C-terminal region of Hprp3p is necessary for the interaction.
Characterization of Hprp3p/Hprp4p Interaction by ITC-ITC is an increasingly used technique for the detection of macromolecular interactions. It enables unmodified native forms of proteins to be characterized in solution phase (37,38). The ITC experiment usually consists of injections of one binding partner into the other, with both solutions contained in exactly the same buffer. The instrument measures the amount of energy as heat required to maintain a constant temperature with respect to a reference cell. As the binding sites are saturated, each injection peak becomes smaller until the heat reflects only the dilution of the titrant molecule. This feedback power is the base-line level in the absence of any reaction, which is provided in an independent experiment with only buffer in the cells.
In this experiment, the interactions of Hprp3p and its mutants with Hprp4p were measured by ITC. Before analyzing these interactions, blank titrations of the titrators (i.e. Hprp3p and its mutants as well as Hprp4p) into a buffer were performed (see "Materials and Methods"). Surprisingly, a large endothermic enthalpy signal was observed in the blank titration of Hprp3p (9 mg/ml) (Supplemental Fig. 2). The enthalpy signal dropped as the Hprp3p concentration increased in the buffer, thus suggesting a possible dissociation reaction of oligomeric assemblies of Hprp3p in the syringe. The enthalpy signal was dependent on the concentration of Hprp3p. When a lower concentration of Hprp3p (1.8 mg/ml) was titrated into the buffer, the heat of dilution was negligible (Supplemental Fig.  2), further supporting the notion that Hprp3p, at a high concentration, may form oligomers. Interestingly, titration of Hprp3p mutant I at a high concentration (10 mg/ml) into buffer showed a much smaller signal (ϳ1 cal/s) (data not shown). The heat of dilution of other mutants (II-IV) was slightly higher than that of mutant I (data not shown).
To avoid the problem of oligomerization, Hprp3p and its mutants at relatively low concentrations (as indicated in the legend to Fig. 3) were chosen for the binding experiments with Hprp4p (1.2 mg/ml), and the signals from blank titrations were subtracted from the binding heat after integration of each injection peak. As shown by the representative data in Fig. 3, full-length Hprp3p as well as mutants II and III interacted with Hprp4p, whereas neither mutant I nor IV showed measurable interactions with Hprp4p, consistent with results from the co-immunoprecipitation experiment. We were unable to obtain dissociation constants (K d ) and binding stoichiometry with high confidence from these data for the following reasons. (i) For Hprp3p, we had to use a low concentration (1.8 mg/ml) for the binding experiment to avoid the problem of oligomerization; thus, the binding reaction is not saturable (Fig. 3A). (ii) For mutant II or III, the titration curves were irregular, which is not uncommon for a binding reaction involving two large proteins and may reflect the complexity of the nature of the interaction (39). Nevertheless, the large enthalpy signals obtained when Hprp3p or mutant II or III was titrated into Hprp4p and the decrease of this signal as the concentration of injected protein increased clearly demonstrated a direct interaction. The fact that the ITC results are in agreement with data obtained from the co-immunoprecipitation experiment strongly supports the conclusion that Hprp3p interacts with Hprp4p directly and that the central domain (amino acids 195-443) of Hprp3p is responsible for this interaction.
Hprp3p/Hprp4p Interaction in HeLa Nuclear Extracts-So far, we have provided in vitro evidence that the human splicing factor Hprp3p interacts with Hprp4p, and we have further shown that the central domain of Hprp3p is responsible for this interaction. To examine whether such an interaction can be observed in human cells, full-length Hprp3p and mutants I-IV were cloned into the pcDNA3 expression vector and expressed in HeLa cells. Hprp3p is normally localized in the nucleus, but whether deletion at either end affects its cellular distribution or function is unknown. Before examining their interaction with Hprp4p, we performed immunohistochemistry analysis to examine the expression and cellular localization of the mutant  , c, e, g, i, and k). The localization of the HA-tagged proteins is visualized as green immunofluorescence (b, d, f, h, j, and l). Mocktransfected cells did not display any immunostaining signal under the FITC channel. Fluorescence micrographs were recorded using a 100ϫ objective.
proteins (pcDNA3-3xHA-HPRP3 mutants I-IV) in mammalian cells. The expression plasmids were introduced into HeLa cells through transient transfection and immunostained with anti-HA antibody 48 h post-transfection. Full-length Hprp3p and the empty vector were employed as positive and negative controls, respectively. An FITC-conjugated secondary antibody was used to visualize the location of protein expression, and 4,6-diamidino-2-phenylindole staining was used to mark the cell nuclei by fluorescence microscopy (Fig. 4). Full-length Hprp3p, as expected, and its mutants II and III were localized in the nucleus (Fig. 4, j, d, and f, respectively). On the other hand, Hprp3p mutants I and IV were distributed both in the cytoplasm and in the nucleus, as shown by immunostaining (Fig. 4, h and b, respectively) and immunoblot analysis (data not shown). It is possible that the deleted regions of these mutant proteins are necessary for Hprp3p nuclear translocation. Also, because mutants I and IV are small, ϳ26 and 35 kDa in size, respectively, they could enter and leave the nucleus by diffusion if lacking the ability to bind to Hprp4p, which is exclusively localized in the nucleus. 2 To examine whether the Hprp3p mutants interact with Hprp4p in HeLa cells, nuclear extracts from the transfected cells were used for immunoprecipitation with anti-HA antibodies. The immunocomplexes of different mutants were analyzed for the presence of Hprp4p by immunoblotting using anti-Hprp4p antibodies. As shown in Fig. 5A (upper panel), mutants II and III (lanes 2 and 3, respectively), but not mutants I and IV (lanes 1 and 4, respectively), were able to interact with endogenous Hprp4p, suggesting that the central region of Hprp3p is indeed required for the interaction. Under these experimental 2 C. Ushida and Y. Tomabechi, unpublished data.  1-4). Nuclear extracts from non-transfected cells (HeLaNE) and from cells transfected with the empty vector were used as controls (lanes 5 and 6, respectively). The migration positions of molecular mass markers (in kilodaltons) run in parallel are given. Middle panel, the nuclear extracts from these cells were analyzed by immunoblotting with monoclonal anti-HA antibodies (diluted 1:1000 as shown in lanes 1-4. Nuclear extracts from non-transfected cells and from cells transfected with the empty vector were used as controls (lanes 5 and 6, respectively). Proteins were fractionated by 10% SDS-PAGE. Lower panel, 5 l of nuclear extracts (10 g of total protein) from these cells were also analyzed by immunoblotting with polyclonal antiantibodies against Hprp4p (diluted 1:4000) as shown in lanes 1-6. The position of the band corresponding to Hprp4p is indicated. B: upper panel, HA-Hprp3p mutants I-IV and HA-Hprp3p were incubated with HeLa nuclear extracts and co-immunoprecipitated with protein G-agarose beads pre-blocked with anti-HA antibody, and the presence of Hprp4p in the complex was detected by Western blotting using anti-Hprp4p antibodies (lanes 1-5, respectively). As a control, purified Hprp4p (lane 6) was incubated with the protein G-agarose beads preblocked with anti-HA antibody. Lane 7 shows the presence of Hprp4p in the HeLa nuclear extract. Lower panel, the blot was stripped and reprobed with anti-HA antibody. The presence of both HA-Hprp3p mutants and Hprp3p is shown. C: detection of U4 and U6 snRNAs by primer extension. Total nuclear RNA (0.1 g) (lanes 2-4) or RNAs extracted from the immunoprecipitates prepared as described for A (lanes 6 -10) were subjected to primer extension analysis. As a control, RNA extraction and primer extension were performed with immunoprecipitates from nuclear extracts of HeLa cells transfected with the empty vector, pcDNA3 (lane 5). Individual primers for U4 (lane 2) and U6 (lane 3) and a combination of U4 and U6 (lanes 4 -10) were used in the analysis. The migration positions of U4 (82 nucleotides) and U6 (91 nucleotides) are indicated on the left. The asterisk indicates a partial extension product from U4 snRNA. The extended products were resolved on a 5% polyacrylamide gel under denaturing conditions. conditions, both Hprp3p (mutants) and Hprp4p were present in similar levels as found in the cell and likely in their natural state of post-translational modification.
We then tested the ability of Hprp3p to affinity-select endogenous Hprp4p from HeLa nuclear extracts (Fig. 5B, upper  panel). For this experiment, highly purified Hprp3p or its mutants were added to HeLa nuclear extracts, and the mixture was incubated with protein G-agarose beads pre-blocked with anti-HA antibodies. In agreement with the previous experiment, Hprp3p and mutants II and III were able to coprecipitate Hprp4p present in the HeLa nuclear extracts (Fig. 5B, upper  panel, lanes 5, 2, and 3, respectively). Mutants I and IV did not show any affinity for Hprp4p (Fig. 5B, upper panel, lanes 1 and  4, respectively). These results are consistent with data obtained with proteins of Hprp3p and Hprp4p expressed in E. coli and further suggest that the interaction Hprp3p and Hprp4p is likely to occur in vivo.
Hprp3p/Hprp4p Interaction Mediating the Association of Hprp4p with U4/U6 snRNAs-Genetic analyses of yeast Prp3p and Prp4p mutants indicated that their interaction is critical for the assembly of U4/U6 snRNP (40); however, it is not clear how these proteins associate with U4/U6 snRNAs. Using E. coli produced proteins, we have shown that Hprp3p, but not Hprp4p, has RNA-binding activity in vitro (data not shown). Therefore, we speculated that Hprp4p associates with U4/U6 snRNAs through its interaction with Hprp3p. To test this hypothesis, we examined, by primer extension analysis, the presence of U4/U6 snRNAs in Hprp3p-Hprp4p complexes precipitated with anti-HA antibodies. As shown in Fig. 5C, both U4 and U6 snRNAs were present in the complexes of HA-tagged wild-type Hprp3p (lane 6) and mutant III (lane 9), but their levels were undetectable in immunocomplexes of Hprp3p mutant I (lane 7) and mutant IV (lane 10), which does not contain the Hprp3p central region. The absence of U4 and U6 snRNAs in the immunocomplexes of Hprp3p mutant II (Fig. 5C, lane 8), which retains the central region, was due to the lack of the Hprp3p C-terminal region required for RNA binding (data not shown). As a negative control, immunoprecipitation was performed with extracts from HeLa cells transfected with the pcDNA3 cloning vector, and the amount of U4 and U6 snRNAs present in the immunoprecipitates was determined by the same primer extension analysis. There was no nonspecific precipitation of U4 or U6 snRNA by the anti-HA antibodies (Fig.  5C, lane 5). These results suggest the importance of the Hprp3p central region for maintaining the association of Hprp3p-Hprp4p with U4/U6 snRNAs. Because the Hprp3p central region was demonstrated to be essential for interaction with Hprp4p and because Hprp4p does not bind to RNA (data not shown), we interpret that this interaction is important for the recruitment of Hprp4p to the U4/U6 snRNP.

DISCUSSION
Hprp3p and Hprp4p, two integral components of the human U4/U6 snRNP, are required for the assembly and activation of the spliceosome (41)(42)(43)(44), although their specific involvement in this process is still not clear. Nevertheless, a large body of indirect evidence suggests that these proteins play a major role during spliceosome assembly through their interactions with each other and other components of the splicing machinery. Overexpression of yeast Prp3p suppresses temperature-sensitive alleles of the yeast PRP4 gene (40,41) and the prp4-1 prp3-1 double mutant exhibits synthetic lethality (45). These observations indicate a genetic interaction between the two proteins. In addition, it has been reported that the WD domain of yeast Prp4p interacts with Prp3p in the yeast two-hybrid system (46). Hprp3p and Hprp4p can be copurified with cyclophilin H in a stable complex from HeLa cells (28), also suggesting an Hprp3p/Hprp4p interaction. On the other hand, the observation that there is no interaction between human Hprp3p and Hprp4p in the yeast two-hybrid system (30) has been confirmed by our group (data not shown). This result may simply reflect the limitation of the yeast two-hybrid system in the analysis of protein/protein interactions. Proteins might not be able to fold into the correct conformation and retain activity as fusion proteins in yeast. False negatives can also be caused by the failure of binding domain X (X, Hprp3p or Hprp4p) and/or activation domain Y (Y, Hprp4p or Hprp3p) to localize to the yeast nucleus. It has been estimated that the number of false negatives in the two-hybrid system is ϳ45% (47,48).
We have demonstrated the Hprp3p/Hprp4p interaction without the presence of cyclophilin H by using co-immunoprecipitation assays with proteins expressed in bacteria. We have also shown that these proteins interact with each other in HeLa nuclear extracts, where Hprp3p and Hprp4p are present in a similar concentration range as found in the cell. Similarly, because the proteins are in their natural state of post-translational modification, interactions that require phosphorylation or dephosphorylation are more realistically assessed. Considering that other factors could be present in the immunoprecipitation system and that the proteins in question might be part of a larger complex, it cannot be concluded using only these assays that Hprp3p and Hprp4p directly interact. Consequently, we used another approach (ITC) to further investigate whether the Hprp3p/Hprp4p interaction is direct. The ITC experiments combined with the study of these proteins expressed in HeLa cells suggest that direct interaction between these splicing factors may also occur in vivo. Furthermore, our analyses of the interactions between Hprp3p mutants and Hprp4p indicate that the first 194 amino acids of Hprp3p (mutant I) are not required for the interaction. By amino acid sequence comparison, it has been noted that this region is poorly conserved in Prp3p, the Saccharomyces cerevisiae homolog (31). This difference may reflect the complexity of the splicing system of higher eukaryotes in comparison with yeast. We have also shown that the last 240 amino acids of Hprp3p (mutant IV) are not required for its interaction with Hprp4p. Because this region is highly conserved from human to yeast, it is likely that the C terminus plays an important role in RNA splicing through its interaction with other conserved components of the splicing machinery. This assumption is consistent with the finding that this region of Hprp3p (amino acids 554 -626) shares a 40% similarity with the double-stranded binding domain of RNase III (amino acids 142-222), pointing to the possibility that Hprp3p could be a U4/U6 snRNA-binding protein (30,49). Results from our group suggest that Hprp3p (and, in particular, its C-terminal region (mutant IV)) is involved in RNA binding. 3 As demonstrated in our primer extension analysis (Fig. 5C), both U4 and U6 snRNAs could be coprecipitated with Hprp3p or its mutant III, but not mutants I, II, and IV. Mutant II, which lacks the C-terminal region of Hprp3p, failed to yield U4 and U6 primer extension products, suggesting that it is required for RNA binding. Because this mutant could be coimmunoprecipitated with Hprp4p (Figs. 2 and 5), the association of Hprp4p with U4 and U6 snRNAs (31) is likely mediated through Hprp3p. U4 or U6 snRNA could not be co-immunoprecipitated with mutant IV (Fig. 5C), indicating that the Cterminal region alone is not sufficient to form a stable RNAprotein complex. Our data indicate that the Hprp3p middle region (amino acids 195-442), present in mutants II and III, contains the domain required for interaction with Hprp4p because deletion of this region from either end abolishes its interaction with Hprp4p. This interaction is critical for the recruitment of Hprp4p to the U4/U6 complex because Hprp4p does not directly interact with RNA (data not shown).
Based on our results and those of other groups (23,28,30,31,40,46,50,51), we propose the following model to explain the role of Hprp3p in spliceosome assembly (Fig. 6). In our model, Hprp3p interacts with both proteins and U4/U6 snRNAs. The Hprp3p central region directly contacts Hprp4p, whereas its C-terminal region interacts with U4/U6 snRNAs. Because yeast genetic analyses suggest that the WD domain (the seven ␤-transducin repeats) of Prp4p is involved in the association with Prp3p, it is likely that the central region of human Hprp3p interacts with the WD domain of Hprp4p (46). In addition, we suggest that Hprp3p directly interacts with the stem II region of U4/U6 snRNAs; and thereby, the other proteins associated with Hprp3p interact with the U4/U6 snRNAs. Hprp3p does not have any known RNA-binding motif; however, it is the most positively charged (pI 9.90) protein in a heterotrimer protein complex that associates with U4/U6 snRNAs, and its C-terminal region exhibits homology to RNase III, consistent with its role as an RNA-binding protein. Moreover, several lines of evidence from studies of human and yeast U4/U6 snRNP components support the above proposition. (i) Both Hprp3p and Hprp4p have been shown to be associated with U4/U6 snRNP (31). (ii) Biochemical analysis of yeast U4/U6 snRNP demonstrated that yeast Prp4p is associated with the 5Ј-region of yeast U4 snRNA (52). (iii) Saturation mutational analysis of yeast U4 snRNA revealed that U4 stem II is highly sensitive to mutations, indicating the potential for molecular interactions (36). (iv) The steady-state level of U6 snRNA is dramatically reduced in temperature-sensitive PRP3 and PRP4 yeast mu-tant strains at nonpermissive temperature (40). Further supporting our model is the finding that SPF30, a U2 snRNP-associated protein, may dock the U4/U6/U5 tri-snRNP to the pre-spliceosome by preferentially binding Hprp3p (23,51). Because neither the N-terminal region of Hprp3p nor SPF30 is conserved in S. cerevisiae, we speculate that both SPF30 and the Hprp3p N terminus have evolved to serve higher eukaryotic splicing functions. In our model, the N-terminal region of Hprp3p is proposed to interact with SPF30 to dock the whole U4/U6/U5 tri-snRNP to the prespliceosome, a crucial step that triggers conformational rearrangements to activate the spliceosome. Recently, a protein domain (the PWI motif) has been identified in Hprp3p (50). The PWI motif is present in the mammalian (mouse and human) splicing factor Prp3p, but not in Caenorhabditis elegans and yeast Prp3p. The function of the PWI motif is not known, but its presence in two splicing factors (SRm160 and mammalian Prp3p) and in proteins related to splicing (MAL3P4.20 and W04D2) suggests that it may be important for pre-mRNA splicing (50). Furthermore, because the N terminus of Hprp3p contains the PWI motif, a likely protein/protein-binding domain (50), it can be speculated that SPF30 might interact with Hprp3p throughout this motif. Interestingly, the biological significance of Hprp3p has been highlighted by a very recent report indicating that mutations in HPRP3 lead to autosomal dominant retinitis pigmentosa (53). The authors speculated that the U4/U6/U5 tri-snRNP recruitment to the pre-mRNA could be a rate-limiting step in splicing and that its deficiency could be detrimental to a highly metabolically active tissue such as the retina. In conclusion, we propose that Hprp3p plays an important role in recruiting other splicing factors such as Hprp4p to the U4/U6 snRNP and docking the U4/U6/U5 tri-snRNP to the pre-spliceosome.