The Arabidopsis ortholog of the 77kDa subunit of the cleavage stimulatory factor (AtCstF-77) involved in mRNA polyadenylation is an RNA-binding protein

1. Introduction 

The processing of pre-mRNAs in eukaryotic cells is a complicated, yet critical process involving numerous players [1], [2], [3]. Briefly, the events involved in RNA processing include addition of the 5′ methyl guanosine cap, intron/exon splicing, and addition of a 3′ polyadenosine [poly(A)] tail. These processes are highly coordinated, co-transcriptional nuclear events that show no true separation spatially or temporally from other events in gene expression [3], [4], [5]. The formation of the 3′ end of the pre-mRNA has been well studied in several eukaryotes spanning multiple kingdoms [1], [2]. This process involves several multisubunit complexes that can be separated, purified, and assayed for activities important for the complete process.

One of the complexes involved in polyadenylation in mammals is the so-called cleavage stimulation factor, or CstF [6]. In mammals, this complex is comprised of 50kDa, 64kDa, and 77kDa subunits [6]. CstF64 possesses an RNA recognition motif (RRM)-type RNA-binding domain near the N-terminus [7], [8], [9], [10], [11], [12], [13], while the C-terminal region interacts with other members of the polyadenylation complex [14], [15], [16]. The sequence preference of the CstF64 RRM has been determined to be G/U-rich sequences downstream of the poly(A) site (AAUAAA) in mammals and A-rich sequences in yeast [9], [10], [11], [12]. CstF50 is a WD repeat protein that interacts with CstF77 and also with the BRCA-associated ubiquitin ligase subunit BARD [16], [17].

CstF77 possesses a series of so-called HAT (for the half-a-TPR) domains [18]; this feature of CstF77 is conserved in all eukaryotes. The HAT domain is flanked at the C-terminus of the protein by a C-terminal domain that exhibits very lower evolutionary conservation. Functionally, CstF77 acts as a scaffold and a bridge to other polyadenylation factors, as it has multiple interactions with the other CstF subunits and with the cleavage and polyadenylation specificity factor (CPSF) subunits CPSF160 and symplekin [16], [19], [20], [21]. CstF77 exists as a dimer [20], [22] that is held together by subunit–subunit interactions involving HAT motifs in the central third of the protein. This dimer may be modeled as a concave shape, with the C-terminus residing within the cavity formed by the rest of the dimer (see Fig. 1A).

View Large Image Download to PowerPoint

Fig. 1. Structures of the AtCstF77 derivatives used in this study. (A) Illustrations of the 3-dimensional structures of the monomeric (left) and dimeric (right) HAT domains of the mouse CstF77. These illustrations were made using Cn3D 4.1 and are based on the PDB accessions 2OOE and 2OND, respectively. The portion corresponding to the HAT-N domain is colored in yellow, and the HAT-C domain in purple or blue. The C-terminal domain of CstF77 was not in the crystal structure; the position of this domain, based on the location of the C-termini of the chains shown, is noted with black arrows. (B) Amino acid sequence alignment of the Arabidopsis and mouse CstF77 proteins. The HAT-N, HAT-C, and C-terminal domains are set off with boxes with differently-colored outlines.

As with its mammalian counterpart, the Arabidopsis CstF77 ortholog (AtCstF77) interacts with CstF64 and CPSF160 [23], [24]. Additionally, AtCstF77 also binds the Arabidopsis ortholog of the 30kD subunit of CPSF, AtCPSF30 [24]. In this study, we describe an analysis of the interaction between AtCstF77 and AtCPSF30. We also report an unexpected finding arising from these studies, that AtCstF77 can bind RNA. Together, these results corroborate those of previous studies purporting an interaction between AtCstF77 and AtCPSF30. Additionally, they add a new RNA-binding capability to the plant polyadenylation complex, and with it an added complexity to this aspect of polyadenylation in plants.

2. Results 

Previously, it was reported that the C-terminal half (approximately) of AtCstF77 interacted with AtCPSF30 in a yeast two-hybrid assay [24]. To confirm this interaction, in vitro pull-down assays were performed. Using the crystal structure of murine CstF77 [22] as a guide, it was determined (Fig. 1A) that AtCstF77 could be subdivided into three domains, with a C-terminal region that, while not apparent in the structures, could be placed within a conceptual cavity formed by the dimeric HAT part of the protein (see the right panel of Fig. 1A). The C-terminal domain is not well-conserved evolutionarily (Fig. 1B). For AtCstF77, the CTD consisted of residues 500–734, (Fig. 1B).

Thus, to test the interaction between the C-terminus of AtCstF77 and AtCPSF30, the maltose binding protein (MBP) tag was fused with amino acids 500–734 of AtCstF77; the MBP tag was used for pull-down assays and for purification of the proteins. Subsequently, the co-purification of biotinylated AtCPSF30 with MBP-AtCstF77 CTD was assayed. The results of these experiments showed that AtCPSF30 co-purified with the MBP-AtCstF77 CTD (“MBP-77 CTD” in Fig. 2B). In contrast, AtCPSF30 did not bind to MBP (Fig. 2B). This binding of the AtCstF77 CTD was specific for AtCPSF30 (and not for the biotin tag), as the major biotinylated Escherichia coli protein (BCCP in Fig. 2B; [25]) present in the extracts did not bind to any of the MBP proteins. These results corroborate previous results [24] and place the binding site of AtCPSF30 apart from the conserved HAT domains, at the C-terminus of AtCstF77.

View Large Image Download to PowerPoint

Fig. 2. The C-terminal domain of AtCstF77 binds to AtCPSF30. (A) Illustration of the proteins used in this study. BCCP=E. coli biotin carboxyl carrier protein. Not drawn to scale. (B) Results of pull-down assays. The top row shows the co-purification of AtCPSF30 with the MBP proteins (labeled at the top), the middle row shows the co-purification of the BCCP, and the bottom row shows the recovery of MBP fusion proteins on the amylose beads. AtCPSF30 and BCCP were detected using alkaline phosphatase-conjugated stretpavidin, and the MBP proteins by staining with Coomassie Brilliant Blue. The “input” column shows the amount of biotinylated AtCPSF30 and BCCP that was added to each reaction.

The activities of AtCPSF30 are affected by several interacting partners [26], [27]. Thus, it was hypothesized that AtCstF77 may also affect the activity of AtCPSF30. Accordingly, RNA binding by AtCPSF30 was assayed in the presence of the C-terminal domain of AtCstF77. For these assays, the AtCstF77 CTD was cleaved from the MBP tag. In addition, an MPB fusion protein containing an AtCPSF30 mutant lacking one of the three CCCH zinc finger motifs [26] was assayed. It was necessary to use the MBP fusion protein for these assays since it was not possible to purify enough biotinylated AtCPSF30 to perform the RNA-binding assays. Also, this mutant was chosen because it interacts with AtCstF77 (B Addepalli, unpublished results), and because it lacks the inherent endonuclease activity of the wild-type AtCPSF30 [26] and thus is easier to assay for RNA binding. The RNA for these assays was derived from the cauliflower mosaic virus polyadenylation signal [28], [29] and possessed all of the elements needed for efficient polyadenylation in vivo (Fig. 4A).

View Large Image Download to PowerPoint

Fig. 4. Binding of the C-terminal domain of AtCstF77 to different RNAs. (A) Illustration of the three RNAs used in this study. At the top is shown the structure of the CaMV 35S RNA polyadenylation signal [28], [29], showing the locations of the FUE (black bar within the gray box), NUE (vertical black bar), and poly(A) site (vertical tic beneath the “An” designation). Beneath this are shown the compositions of the three RNAs used: “wt” consists of the unmodified polyadenylation signal, “dPA” is the wt RNA in which the NUE has been deleted (represented as a space), and “NUE” is an RNA that carries just the NUE and sequences extending to the poly(A) site. The numbers (“−181” and “+80”) next to the depiction of the RNA indicate the nucleotides upstream and downstream from the poly(A) site that are contained in this RNA. (B) Results of the binding assays. Varying concentrations of the purified and cleaved MBP-AtCstF77 C-terminal domain were incubated with the indicated RNAs and analyzed as described in Section 4. Binding activity was normalized such that the maximal activity for each curve was set to a value of 1.0.

As expected, AtCPSF30 possessed readily-detectable RNA-binding activity (Fig. 3A, lane 2). Considerable RNA-binding activity was also seen when a mixture of AtCPSF30 and the AtCstF77 C-terminal domain fusion protein was assayed (Fig. 3A, lane 3); this indicates that RNA binding activity remains when AtCPSF30 and the C-terminus of AtCstF77 are present in the same reaction. Surprisingly, RNA binding was also seen when AtCPSF30 was omitted from these assays (Fig. 3A, lane 1), suggesting that the C-terminal domain of AtCstF77 itself possesses RNA-binding activity.

View Large Image Download to PowerPoint

Fig. 3. The C-terminal domain of AtCstF77 binds RNA. (A) Assay of RNA binding by AtCPSF30 and the C-terminal domain of AtCstF77. The identities of the proteins used in the RNA binding reactions were: lane 1 – AtCstF77-C; lane 2 –AtCPSF30; lane 3 – AtCstF77-C+AtCPSF30; lane 4 – MBP; lane 5 – no protein. The positions of the RNA–protein complexes and free RNA are noted. The “∗” denotes the top of the gel, where some insoluble precipitate sometimes accumulated (even in the control – see lane 5). (B) Binding of full-length AtCstF77-MBP fusion proteins to poly(A) sepharose. Lane 1 – pre-stained size standards (sizes noted on the left). Lane 2 – purified MBP-AtCstF77 before purification on poly(A) sepharose (the quantity loaded is approximately equal to that used for the sample in lane 3). Lane 3 – MBP fusion proteins bound to poly(A) sepharose. The arrow on the right denotes the full-length MBP-AtCstF77 fusion protein.

To confirm this result, the binding of full-length MBP-AtCstF77 to RNA was assayed. Since the preparations of the full-length protein contained significant quantities of truncated protein (Fig. 3B, lane 2), RNA binding was assayed by testing the ability of the fusion protein to bind immobilized poly(A). Because the detection used focuses on the MBP tag that is fused at the N-terminus of AtCstF77, this approach has the added facet that a deletion analysis of a sort is obtained, with the demarcation between bound and unbound polypeptides defining the amino acid sequences needed for binding to poly(A). When this was done, the full-sized MBP-AtCstF77 was found to be bound efficiently to the immobilized poly(A) (Fig. 3B, lane 3). In contrast, forms of the MBP-AtCstF77 that had been truncated at their C-termini by as few as about 70 amino acids did not bind to poly(A). This result indicates that the full-sized AtCstF77 binds RNA, and it corroborates the finding in Fig. 3A by localizing sequences needed for binding to the C-terminal 70 amino acids (approximately) of the protein.

The finding that AtCstF77 binds RNA raises the possibility that this binding might be specific for one of the sub-elements of a plant polyadenylation signal. To test this, RNA binding by the isolated C-terminal domain was assayed with CaMV-derived RNAs lacking either the far-upstream (FUE) or near-upstream (NUE) elements that comprise a plant poly(A) signal [1], [28], [29], [30]. The design of these RNAs (Fig. 4A) was guided by earlier studies [28], [29], [30]; thus, the NUE mutant carried a deletion of the AAUAAA motif that is required for efficient functioning of this poly(A) signal, and the FUE mutant consisted of a deletion lacking the UGUAA elements that are required as well for efficient polyadenylation. As shown in Fig. 4B, RNAs that lacked either the NUE motif or the FUE were bound by AtCstF77 much as was the intact CaMV polyadenylation signal. With all three RNAs, half-maximal binding was seen at protein concentrations of about 320nM. Thus, AtCstF77 does not show a noticeable preference for an intact plant polyadenylation signal or its sub-elements.

3. Discussion 

The results reported here add another RNA-binding protein to the plant polyadenylation apparatus and pose some interesting questions as to how pre-mRNAs are recognized and subsequently handled during mRNA 3′ end formation in plants. There are indications that the CstF complex varies slightly between plants and mammals, mainly in the protein–protein interactions within the hypothetical complex [23], [24]. Specifically, in contrast to the case in mammals [16], CstF50 does not appear to interact with CstF77 in plants. RNA binding has not been reported for either the mammalian CstF77 or its yeast counterpart, Rna14. However, the part of AtCstF77 that binds RNA is not well-conserved in eukaryotes (see Fig. 1B), and in fact is missing from Rna14 [31]. Thus, it would seem as if the RNA-binding activity of AtCstF77 is a novel, plant-specific property.

The fact that the C-terminal domain of AtCstF77 binds RNA raises a number of interesting questions. If it is assumed that AtCstF77 assumes the same dimeric structure as the animal protein (illustrated in Fig. 1B), then this places the two RNA binding sites of AtCstF77 within the cavity formed by the CstF77 dimer (see Fig. 1B). The C-terminus of AtCstF77 also binds AtCstF64 [24], itself an RNA-binding protein [23]. The C-terminus of AtCstF77 also interacts with an Arabidopsis Fip1 ortholog (AtFIPS5; [31]), yet another RNA-binding protein [32]. Finally, as shown in this report, the C-terminus of AtCstF77 binds AtCPSF30, another RNA-binding polyadenylation factor subunit. This places as many as eight RNA binding sites within the cavity of the hypothetical CstF77 dimer. These eight sites have at least four possible specificities (reflecting that four different proteins contribute to the RNA-binding potential of the hypothetical complex). This number of RNA binding sites and specificities exceeds the number of known cis elements that together constitute a plant polyadenylation signal [1], thereby raising questions as to the roles of these RNA binding sites in the polyadenylation reaction. Whatever these roles may be, the results shown in Fig. 4 argue against the possibility that the C-terminus of AtCstF77 is solely responsible for recognition of one of these cis elements. These questions remain to be addressed experimentally.

To summarize, the results reported here indicate that the plant-specific C-terminal region of AtCstF77 interacts with AtCPSF30 and that it possesses a distinctive RNA-binding activity. This adds another RNA-binding domain to the suite of such activities in the plant polyadenylation apparatus and raises interesting questions regarding the functioning of AtCstF77 in the polyadenylation reaction.

4. Materials and methods 

The coding region for the AtCstF77 CTD was PCR amplified from Arabidopsis thaliana Columbia cDNA [33] and cloned into the pMAL-c4G vector (New England Biolabs); to facilitate this, XhoI and BglII sites were incorporated into the PCR primers (see Table 1). AtCPSF30 (At1g30460) was cloned into the pT7-MAT-FLAG expression vector (Sigma–Aldrich) as an XhoI-BglII fragment. In addition, a C-terminal biotinylation site (GLNDIFEAQKIEWHE) was incorporated into the construct using suitably-designed oligonucleotides (Table 1). Briefly, the T7-Mat-FLAG-AtCPSF30 construct was digested with the BglII site at the 3′ end of the AtCPSF30 coding region and the annealed oligonucleotides encoding the biotinylation motif ligated into the digested plasmid. Recombinants were identified by PCR and restriction digestion. The clone encoding the full-length AtCstF77-MBP fusion protein has been described before [31].

Table 1. Oligonucleotides used in this research.

	Primers	Sequence (5′→3′)	Used for
	C77Ca	CCCGAATTCAACGACCTTGATCATTTAGCCAGAC	Cloning AtCstF77 C-terminal domain
	C77Cb	CCCTCTAGATTAGCCAGTGCTACCAGAAAGCTCG	Cloning AtCstF77 C-terminal domain
	C30MATFLAG5′a	TTTCTCGAGTGGAGGATGCTGATGGACTTAGCTTCGAT	Cloning AtCPSF30
	C30MATFLAG3′	GGGAGATCTTCCAGGAAGCTTTGCATGCCTGTACCGAC	Cloning AtCPSF30
	TEVBio-1	GATCGGAGAACCTGTACTTCCAAGGTGGCGGTCTGAACGACATCTTCGAGGCTCAGAAAATCGAATGGCACGAATAA	Adding a biotinylation site to AtCPSF30
	TEV-Bio2	GATCTTATTCGTGCCATTCGATTTTCTGAGCCTCGAAGATGTCGTTCAGACCGCCACCTTGGAAGTACAGGTTCTCC	Adding a biotinylation site to AtCPSF30

The growth, induction, and purification of the MBP-containing proteins was done as described elsewhere [33]. BL21(DE) that had been transformed with the pBirAcm plasmid (http://www.genecopoeia.com/product/avitag/) was used for the production of biotin-tagged AtCPSF30; in this case, cells were grown, induced, and extracts prepared as described elsewhere [33]. Crude extracts were used for in vitro pull-down assays. MBP and the MBP-AtCstF77 CTD fusion proteins were purified using affinity chromatography as described [33]. For the assay shown in Fig. 4B, the MBP-AtCstF77CTD protein was cleaved with Genenase (New England Biolabs) prior to assay. The MBP-ZF3 AtCPSF30 protein was a gift of Dr. Balasubrahmanyam Addepalli.

The protocols for in vitro pull-down assays have been detailed elsewhere [31], [32]. RNA-binding assays were conducted as described previously [26], [31], [33], using 12pmol of purified protein and 2pmol of RNA per 10μl reaction except where noted (e.g., Fig. 4B). The binding of purified MBP-AtCstF77 to poly(A) Sepharose was performed using the protocol described by Forbes et al. [32], except that poly(A) Sepharose was used in place of poly(G) Sepharose, 12μg of purified protein was used, and assays and washes were performed in 50mM Tris HCl, pH 7.5+0.15M NaCl+1mM EDTA. For this assay, the input and bound proteins were analyzed by immunoblotting with anti-MBP monoclonal antibodies (New England Biolabs).