|Home | About | Journals | Submit | Contact Us | Français|
U2 small nuclear ribonucleoprotein auxiliary factor (U2AF) is an essential component of the splicing machinery that is composed of two protein subunits, the 35 kD U2AF35 (U2AF1) and the 65 kD U2AF65 (U2AF2). U2AF interacts with various splicing factors within this machinery. Here we expand the list of mammalian splicing factors that are known to interact with U2AF65 as well as the list of nuclear proteins not known to participate in splicing that interact with U2AF65. Using a yeast two-hybrid system, we found fourteen U2AF65-interacting proteins. The validity of the screen was confirmed by identification of five known U2AF65-interacting proteins, including its heterodimeric partner, U2AF35. In addition to binding these known partners, we found previously unrecognized U2AF65 interactions with four splicing related proteins (DDX39, SFRS3, SFRS18, SNRPA), two zinc finger proteins (ZFP809 and ZC3H11A), a U2AF65 homolog (RBM39), and two other regulatory proteins (DAXX and SERBP1). We report which regions of U2AF65 each of these proteins interacts with and we and discuss their potential roles in regulation of pre-mRNA splicing, 3’-end mRNA processing, and U2AF65 sub-nuclear localization. These findings suggest expanded roles for U2AF65 in both splicing and non-splicing functions.
U2AF65 (U2AF2) is an essential component of the U2 small nuclear ribonucleoprotein auxiliary factor (U2AF) . In multicellular animals and plants, the carboxyl- (C-) terminal two-thirds of U2AF65 is comprised of three distinct domains, two of which are canonical RNA recognition motifs (RRM) that directly contact the polypyrimidine tract of the pre-mRNA at a point adjacent to the 3’ splice site [2, 3]. The third C-terminal domain is a protein-protein interaction domain, now commonly defined as a U2AF homology motif (UHM) . The UHM binds to UHM ligand motifs (ULM), which contain an invariant core tryptophan (W) residue. A consensus ULM [(K/R)4-6X0-1W(D/E/N/Q)1-2)] is found in many proteins, although only the ULM regions within a few splicing factor proteins have been directly shown to function in protein-protein interactions [4-6]. Indeed, U2AF65 also has a ULM of its own, which interacts with UHMs on other proteins, including that in U2AF35 (U2AF1), its heterodimeric interaction partner. The U2AF35-binding domain is within amino acids (a.a.) 85-112 of U2AF65 . Crystal structures showed that U2AF35 contacts tryptophan 92 (W 92) within the U2AF65 ULM and, in a reciprocal fashion, U2AF35-W 134 is enveloped by a proline repeat on U2AF65 near the ULM [4, 8].
The amino- (N-) terminal third of U2AF65 is predominated by an intrinsically unstructured arginine/serine-rich (RS) domain that promotes pre-spliceosome assembly by contacting the pre-mRNA branchpoint sequence and modulating interaction of the U2 small nuclear ribonucleoprotein (snRNP) complex with the branchpoint [9, 10]. RS domains are found in many proteins of the splicing machinery and often mediate protein-protein interactions . However, the RS domain of U2AF65 has rarely been reported to bind other proteins and, instead, is thought to participate in splicing primarily through its RNA-binding capacity [10, 12]. This domain was also shown to target U2AF65 to the nucleus and was sufficient to target a heterologous protein to nuclear speckles , which are sub-nuclear structures enriched for splicing-, transcription-, and 3’ end RNA-processing-factors . Interestingly, U2AF65 is cleaved near aspartate 128 (D 128) by caspases during early apoptosis and the liberated RS-containing N-terminal fragment translocates from nuclear speckles to nucleolus-like structures . Thus, regions outside of this N-terminal fragment might also participate in sub-nuclear localization .
U2AF65 also interacts with other splicing factors. Splicing factor-1 (SF1) binds between a.a. 367 and 475 of U2AF65 . The SF1:U2AF65 interaction is mediated specifically via ULM:UHM contacts between these proteins, respectively, which cooperatively strengthen the SF1:U2AF65 interaction with the pre-mRNA in a phosphorylation-dependent manner [17, 18]. The ULM of splicing factor-3B1 (SF3B1/SAP155), a component of the U2 snRNP complex, also interacts with the UHM region of U2AF65 [19-21]. BAT1A (UAP56/BAT1 in humans), a splicing factor and DExD/H-box protein, interacts with a.a. 138-183 of human U2AF65 . Interaction of BAT1A with U2AF65 in the pre-spliceosome is dependent on the ATP-binding and ATPase activities of BAT1A . Recently, U2AF65 was shown to directly bind RBM5, which regulates alternative splicing of apoptotic genes. Removal of the N-terminal RS domain (a.a. 1-92) of U2AF65 disrupts the interaction between U2AF65 and RBM5 . RNA binding motif protein 17 (RBM17/SPF45) and poly(U)-binding splicing factor-60KDa (PUF60, a U2AF65 homolog) each contain UHM domains that can bind the U2AF65 ULM [5, 6, 25]. More complete lists of putative U2AF65-associated proteins, including both known primary interactors and other components of U2AF65-containing complexes, can be found at http://mint.bio.uniroma2.it , www.thebiogrid.org  and www.hprd.org .
In addition to its role in pre-spliceosome assembly, U2AF65 has been linked to several other cellular processes. These include pre-mRNA 3’ end formation and processing [12, 29, 30], transcription , and apoptosis [5, 15, 24]. U2AF65 is also reported to continuously shuttle between the cytoplasm and nucleus , and it remains associated with mRNAs after they have been spliced, suggesting that U2AF65 may be more than just a pre-mRNA splicing factor .
Here we present an expanded analysis of the U2AF65 interactome. In addition to finding previously reported U2AF65-interacting proteins, which validated the conditions of our screens, we found nine previously unreported nuclear U2AF65-interacting proteins. Five of these were either known splicing-related proteins that were not known to directly contact U2AF65 or are homologues of splicing-related proteins that had been shown to interact with U2AF65. The other four proteins were putative regulatory factors, including RNA-binding proteins, two poorly characterized zinc-finger family members, and an apoptosis regulatory protein. The expanded interactome presented here suggests that regulators of numerous cellular processes interact with U2AF65, which may participate in tailoring splicing to specific activities or states of the cell.
All two-hybrid screens and confirmatory interactions used the pGBKT7+1 bait plasmid for expression of U2AF65-derived proteins ; bait proteins are diagrammed in Fig. 1A. Amplification and cloning of bait constructs (Table 1) was as follows: mouse U2AF65 (1-468) (clone encodes all except the last three a.a. of U2AF65; GenBank ID: NM_133671) was amplified from random-primed mouse embryonic day 10.5 placental first-strand cDNA using U2af65-N-forward and U2af65-C-reverse (Table 1), digested with Sal I/Not I, and inserted into pGBKT7+1; U2AF65 (1-308) (a.a. 1-308 of U2af65) was PCR-amplified from pGBKT7+1-U2af65-FL using U2af65-N-forward and U2af65-N1-reverse, digested with Sal I/Not I, and inserted into pGBKT7+1; U2AF65 (1-97) was PCR-amplified from pGBKT7+1-U2AF65-FL using U2af65-N-forward and U2af65-N2-reverse, digested with Sal I/Not I, and inserted into pGBKT7+1; U2AF65 (88-468) was PCR-amplified from pGBKT7+1-U2af65-FL using U2af65-C-forward and U2af65-C-reverse, digested with Sal I/Not I, and inserted into pGBKT7+1. All plasmids were verified by sequencing. All baits were tested for auto-activity.
A C57Bl/6J mouse embryonic day 10.5 whole pregnant uteri cDNA prey library was constructed and inserted into the pGADT7SN vector as previously described . Assembly of the liver prey library was as follows. Total RNA was extracted and CsCl-purified from F1 hybrid C57Bl/6J x BALB/cJ hybrid adult male livers as described previously . Poly(A+) mRNA from these samples was purified using Oligo-(dT)25 Dynabeads (Dynal Biotech ASA, Oslo, Norway) following the manufacturer's protocols. The liver library was constructed using 3.4 μg of poly(A+) mRNA and the Superscript plasmid system for cDNA synthesis and cloning (Invitrogen, Carlsbad, CA), which yields cDNAs containing 5’ Sal I and 3’ Not I overhangs. cDNAs were ligated into Sal I/Not I-digested pGADT7SN. The liver library contained ~3.2 × 106 independent recombinants with 83% bearing inserts. The average insert size in this library exceeded 1 kb (data not shown). All renewable resources used in this study, including two-hybrid libraries and plasmids, are freely available for unrestricted non-profit use upon request unless specifically restricted by another party.
All screens, confirmatory/auto-activation analyses, and sub-domain interaction tests were completed in Saccharomyces cerevisiae strain AH109 (BD Bioscience). Library- and small-scale yeast transformations, and yeast “spot” dilution plate experiments, were carried out as previously described . Two-hybrid screens were performed on synthetic complete medium (SC) agar plates lacking leucine, tryptophan, and histidine (SC-L-W-H) (screen1), SC-L-W-Adenine (SC-L-W-A) (screen 2), or SC-L-W-H + 0.5 mM 3-aminotriazole (3-AT), a competitive inhibitor of the His3 gene product (screen 3). Primary colonies that grew robustly on the screen plates were transferred to SC-L-W-H media that contained increasing amounts of 3-AT or SC-L-W-A agar plates. Prey plasmids from clones that grew under more stringent conditions (i.e. higher concentrations of 3-AT) were isolated by glass bead lysis and transformed into DH5α bacteria; inserts were sequenced to determine the cDNA identity. All prey plasmids that encoded a U2AF65-interacting prey protein were re-transformed into AH109 with the bait and grown on selective media to verify the interaction.
Two-hybrid screens were performed to identify U2AF65-interacting proteins. Specific conditions of each screen are summarized in Table 2. Screen 1 used the pregnant uteri cDNA prey library and the U2AF65 (1-468) bait (Fig. 1A), and evaluated ~3.0 × 106 primary transformants. Screen 2 used the liver library and the U2AF65 (1-468) bait. This screen evaluated ~1.2 × 106 primary transformants. Screen 3 used a mixture of liver and pregnant uteri libraries (Table 2) and the C-terminally truncated U2AF65 (1-308) bait (Fig. 1A), which lacked the C-terminal β-strand of RRM2 and the UHM [8, 34]. It is likely that this results in an unfolded and therefore inactive RRM2 domain. This screen evaluated ~0.6 × 106 primary transformants. U2AF65-interacting prey clones were isolated from all three screens; the identities and protein regions encoded by each clone are summarized in Table 3. In total, we isolated clones encoding fourteen different U2AF65-interacting proteins. Validating the conditions of the screen, seventeen isolated clones encoded proteins previously reported to interact with U2AF65. In addition, nine proteins that were not previously recognized as U2AF65-interactors were identified.
Most of the prey proteins that were identified in these screens were either known components of the splicing machinery or were homologous to known components of the splicing machinery. Six clones encoding U2AF35, the heterodimeric partner of U2AF65 in U2AF, were isolated verifying the selection conditions used in the screens. Also, for example, one clone we isolated, RNA binding motif protein 39 (RBM39/CAPERα), is homologous to U2AF65, itself. RBM39, like U2AF65, contains an N-terminal RS domain followed by two central RRM domains and a C-terminal UHM [35, 36]. Another member of this family, PUF60, is similar to U2AF65 and RBM39, except that it lacks the RS domain . PUF60 has previously been shown to interact with U2AF65 and regulates both splicing and transcriptional processes [6, 25]; however, to our knowledge RBM39 was not previously identified as a U2AF65-interactor. In addition to proposed splicing roles, RBM39 has been shown to regulate transcription . The interaction of U2AF65 and RBM39 supports an expanded role for U2AF65 in coupling of splicing and transcription.
Previously reported U2AF65-interacting proteins that came out of the screens were all splicing factors. In addition to U2AF35, these included SF1 (1 clone), SFRS2IP (1 clone), SF3B1 (4 clones), and BAT1A (5 clones; Table 3)(for example references [20, 22, 36]). Other splicing components identified in these screens that were not previously known to bind U2AF65 included DEAD box polypeptide 39 (DDX39), splicing factor arginine/serine-rich 3 (SFRS3), and small nuclear ribonucleoprotein polypeptide A (SNRPA/U1A) [37, 38]. The identification of DDX39 as a U2AF65-interacting protein was not surprising. DDX39 is 89% identical to BAT1A (see above), which is known to interact with U2AF65. BAT1A and DDX39 are DEAD-box RNA helicases with roles in spliceosome assembly [37, 39]. Like U2AF65, BAT1A and DDX39 interact with proteins that are integral not only to splicing, but also to transcriptional regulation and mRNA transport. For example, both BAT1A and DDX39 interact with CIP29/HCC-1 (Fig. 2), a heterogeneous ribonucleoprotein that enhances DDX39 RNA helicase activity [38-40]. BAT1A and DDX39 were also previously shown to bind THOC4/ALY, an mRNA export factor [41, 42] (Fig. 2). The association of U2AF65 with the DDX39/BAT1A interaction network provides additional evidence linking U2AF65 with cellular processes that lie beyond its role in constitutive splicing.
Several of the proteins that interacted with U2AF65 contained RS domains. For many SR proteins, the phosphorylation state of this domain affects protein-protein interactions . Because Saccharomyces cerevisae lack SR proteins [11, 44], they may not properly phosphorylate mammalian SR proteins. The current manuscript catalogs interactions of some SR proteins with U2AF65. Further studies will be required to determine which of these occur in mammalian cells and to identify SR proteins whose interaction with U2AF65 are phosphorylation-dependent.
To more precisely identify the regions of U2AF65 that were required for each interaction, we tested whether each prey could bind U2AF65 (1-468) or three different truncated U2AF65 baits (Fig. 1A). The three truncated baits were designed based on domain architecture of the U2AF65 protein such that U2AF65 proteins specifically lacking or containing the N-terminal RS domain or the C-terminal UHM region could be assessed for interactions with each prey protein. As expected, all preys recapitulated the interaction with the original bait used in each screen (Fig. 1B, Tables 2, ,3).3). In previous studies, both SF1 and SF3B1 bound the C-terminal UHM of U2AF65 [18-20], and this was verified by their binding activities with the truncated U2AF65 bait proteins used in this study (Figs 1A, B). The ULM region of SF1 that binds U2AF65 is located in the extreme N terminus of SF1 (a.a. 15-24) [5, 16], which was present in the SF1 prey clone we isolated. Like SF1, SF3B1 and zinc finger CCCH-type containing 11A (ZC3H11A) also interacted with U2AF65 (1-468) and U2AF65 (88-468), but did not interact with U2AF65 (1-97) or U2AF65 (1-308) (Fig. 1B). Four clones of SF3B1 were isolated in these screens, the shortest clone encoded a.a. 298-1304 of the protein (Table 3). Importantly, this clone included only the most C-terminal of the ‘RWDETP’ repeats, entitled either the W7 or ULM-5 motif (a.a. 333-340) [5, 20]. Our results suggest that W7/ULM-5, alone, is sufficient to mediate an interaction with U2AF65.
Four of the U2AF65-interacting proteins, U2AF35, RBM39, SFRS3, and Fas death domain-associated protein (DAXX) interacted with all U2AF65 baits (Figs 1A, B). Of these, only U2AF35 was previously known to bind U2AF65 [7, 45]. Studies on U2AF65 in Drosophila demonstrated that a 28 a.a. region, corresponding to a.a. 85-112 of human U2AF65, was sufficient for U2AF65 binding to U2AF35 . The results of our mapping studies may refine the interaction domain more precisely. Since only ten residues (a.a. 88-97, Fig. 1A) were shared by all bait vectors in the current study, it suggests this short motif, VRKYWDVPPP, was sufficient to mediate U2AF35 binding. This result agrees with previous studies demonstrating that the central WDV residues are integral to the U2AF35-U2AF65 interaction  and X-ray structures showing the W 92 and the nearby proline stretch on U2AF65 directly contact U2AF35 . The mapping results for DAXX, RBM39, and SFRS3 (Fig. 1B) suggested that each of these proteins either binds within the same ten a.a. region on U2AF65 or binds to multiple sites on the U2AF65 protein.
Three of the prey clones, those encoding BAT1A, DDX39, and SNRPA, interacted with both U2AF65 (1-308) and U2AF65 (88-468), but not with U2AF65 (1-97) (Fig. 1B). This suggested that each protein required a motif lying within a.a. 88-308 of U2AF65 for binding. The interaction of BAT1A with U2AF65 (1-308) and U2AF65 (88-468) was consistent with published results showing that a.a. 138-183 of human U2AF65 is sufficient for binding of human BAT1 .
Two of the U2AF65-interacting proteins identified in the screens, splicing factor arginine/serine-rich 18 (SFRS18) and zinc finger protein 809 (ZFP809), interacted with U2AF65 (1-97) and U2AF65 (1-308) (Fig. 1B). This suggested that each of these proteins binds within the first 97 a.a. of U2AF65, which includes the RS domain (Fig. 1A). SFRS18, like SFRS3 and RBM39 (see above), also contains RS-rich regions in its C-terminus [46, 47], and a truncated clone of SFRS18, containing only the extreme C-terminus (a.a. 721-814) of this protein interacted with the RS domain-containing a.a. 1-97 of U2AF65 (Table 3, Fig. 1B). The interaction of ZFP809 with U2AF65 is interesting in that it bound to U2AF65 proteins containing only the N-terminal region of the protein, but it did not interact with the nearly full-length U2AF65 (1-468) bait (Fig. 1B, Table 3). The clone encoding ZFP809 was isolated in screen 3, which used the U2AF65 (1-308) bait lacking the U2AF65 C-terminal region. The lack of interaction with U2AF65 (1-468) suggested that the C-terminal region of U2AF65 inhibited ZFP809 binding. The failure to interact with full-length U2AF65 but ability to interact with the N-terminal portion of the U2AF65 is intriguing in light of a recent study of U2AF65 during apoptosis. Very early following Fas-dependent induction of apoptosis, caspase-mediated cleavage of U2AF65 near D 128 separates the N- and C-terminal portions of the protein . The N-terminal fragment of this cleavage redistributes from nuclear speckles into nucleolus-like bodies, and accumulation of this fragment is associated with alternative splicing of Fas pre-mRNA; the C-terminal portion translocates to the cytoplasm . This may be relevant to the interactions we detected with ZFP809. It remains to be determined whether the domain-restricted interaction of ZFP809 with U2AF65 participates in the nuclear redistribution of the N-terminal fragment, or if this interaction is required for alternative splicing of Fas pre-mRNA during apoptosis.
Many of the U2AF65-binding proteins identified in our screens are also known to interact with other U2AF65-binding proteins, suggesting a possible network of protein-protein interactions. We constructed a mammalian U2AF65 interactome map including proteins identified in the current screens (Fig. 2, blue lines) and other previously reported interactions (Fig. 2, red lines). For simplicity, secondary protein interactions (proteins that do not directly interact with U2AF65; gray lines) were included only if they could be connected back to a known U2AF65-interacting protein via a single step.
Identification of protein interaction networks and the functions of proteins within these networks have led to an increased understanding of how cellular processes are interconnected . In this study, we identified previously unrecognized protein interactions with U2AF65, an auxiliary factor that participates in pre-mRNA splicing [1, 49] and may also participate in other aspects of mRNA maturation [12, 29, 30], transcriptional regulation , and apoptosis [5, 15, 24]. Consistent with the proposed roles for U2AF65, the screens reported here identified known protein components of the spliceosomal machinery. Moreover, we report previously unrecognized protein-protein interactions with U2AF65, including interactions with proteins having suspected roles in splicing, transcription, apoptosis, or sub-nuclear localization, all of which are in agreement with U2AF65 acting as an information hub that coordinates splicing with numerous processes in the cell.
The authors thank M. Rollins, O. Lucas, and S. Ramsey for technical assistance. This work was supported by The NSF , The NIH [AI055739 to EES; RR020185 to VMB], an award from the Montana State University Undergraduate Scholars Program to AMS, and an appointment from the Montana Agriculture Experiment Station to EES.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.