DNA cleavage and reverse splicing of ribonucleoprotein particles reconstituted in vitro with linear RmInt1 RNA

ABSTRACT

The RmInt1 group II intron is an efficient self-splicing mobile retroelement that catalyzes its own excision as lariat, linear and circular molecules. In vivo, the RmInt1 lariat and the reverse transcriptase (IEP) it encodes form a ribonucleoprotein particle (RNP) that recognizes the DNA target for site-specific full intron insertion via a two-step reverse splicing reaction. RNPs containing linear group II intron RNA are generally thought to be unable to complete the reverse splicing reaction. Here, we show that reconstituted in vitro RNPs containing linear RmInt1 ΔORF RNA can mediate the cleavage of single-stranded DNA substrates in a very precise manner with the attachment of the intron RNA to the 3´exon as the first step of a reverse splicing reaction. Notably, we also observe molecules in which the 5´exon is linked to the RmInt1 RNA, suggesting the completion of the reverse splicing reaction, albeit rather low and inefficiently. That process depends on DNA target recognition and can be successful completed by RmInt1 RNPs with linear RNA displaying 5´ modifications.

Introduction

Group II introns are autocatalytic RNAs that are spliced out of a precursor transcript (pre-RNA) in a two-step mechanism resembling the excision of spliceosomal introns [1,2]. This mechanism involves two sequential transesterification reactions: the 2´-OH group of the bulged adenosine residue in domain VI first attacks the phosphate group at the 5´ splice site, generating the branched intron structure and exposing the 3´-OH group of the 5´ exon. This activated group then reacts with the phosphate group at the 3´ splice site, joining the exons and releasing the intron lariat RNA [3]. For some introns, the first step in the splicing process is hydrolysis, yielding a free linear intron RNA rather than the lariat form [4–8]. Moreover, an excision mechanism generating RNA circles has been observed for some group II introns. This mechanism involves the release of the 3´ exon and a further attack on the first intron residue by a reactive OH group on the terminal nucleotide [9–13].

Some group II introns also act as mobile retroelements, integrating into new genomic locations by a reverse splicing process [14–19]. The functional unit for mobility is a ribonucleoprotein particle (RNP) consisting of the lariat intron RNA and an intron-encoded protein (IEP). Target recognition mostly involves base-pairing interactions between the exon-binding sites (EBSs) in domain I of the intron RNA and the corresponding sequences at the target site (intron-binding sites, IBSs). The intron RNA reverse splices into the top strand and is then reverse-transcribed by the IEP [20]. Host proteins are then responsible for completing the insertion process [21–25]. This reaction has been extensively demonstrated for RNPs formed with lariat intron RNA, but linear group II intron RNAs are generally thought to be unable to complete the reverse splicing reaction [26,27].

The Sinorhizobium meliloti RmInt1 group II intron is an efficient, but unusual mobile retroelement [28] that self-splices in vitro to generate lariat, linear and circular intron molecules together with a wide variety of truncated variants of all of them and aberrant intermediates [10,29,30], but is excised in vivo to form intron lariat and circle RNAs [10]. RmInt1 encodes an IEP without a DNA endonuclease domain (En⁻) [31], but with a C-terminal extension involved in intron function [32]. RmInt1 displays a strong bias for reverse splicing into the lagging strand template during replication, and it has been suggested that this intron recognizes its target principally as a single-stranded (ss)DNA [33]. We recently showed that RmInt1 lariat RNPs reconstituted from self-spliced intron precursor and a maltose binding protein (MBP)-IEP fusion protein could reverse-splice into DNA substrates in vitro [34].

Here, we used the RmInt1 linear intron RNA and the MBP-IEP fusion protein to reconstitute linear RmInt1 RNPs, and to determine their biochemical activities and requirements for DNA cleavage and reverse splicing. We show, for the first time, that RNPs containing linear intron RNA can complete the precise joining of the RNA molecule to both exons of a DNA substrate in vitro. We also show that the efficiency and fidelity of this reaction are compromised by modifications/insertions at the 5´ or 3´ end of the intron sequence or by the proper recognition of the DNA target sequence by the intron RNA. These findings could have important implications for the spread and mobility mechanisms of group II introns that excises as linear molecules in vivo.

Results

DNA cleavage and reverse splicing of RNPs containing linear RmInt1 RNA

We previously showed high levels of in vitro cleavage and partial reverse splicing into ssDNA substrates for RNPs containing the RmInt1 lariat, whereas full reverse splicing was barely detectable [34]. To test the ability of RNPs reconstituted in vitro with the linear RmInt1 (ΔORF) RNA to perform these activities, we carried out DNA cleavage assays with 5´ and 3´ end-³²P-labeled 70 nt ssDNA substrates encompassing the RmInt1 target site. The products obtained were separated by electrophoresis in a denaturing polyacrylamide gel (Figure 1A). We found that the RNPs containing linear intron RNA cleaved the DNA substrate as efficiently as RNPs containing lariat RNA (lanes 3–4 versus lanes 5–6). Notably, low-mobility bands were detected at the top of the gel regardless of the end-labeled substrate used. Both types of RNPs yielded mostly partial reverse splicing products (Figure 1A, top of the gel, lanes 4 and 6), but the RNPs containing linear intron RNA also generated products corresponding to intron RNA molecules attached to the 5´ exon, similar to the RNPs containing lariat intron RNA (5´ end-labeled substrate, lanes 3 and 5). As expected, DNA cleavage/reverse splicing activity was not detected in the absence of the RNP complex (-RNPs, lanes 1 and 2) or for RNPs reconstituted with a catalytically inactive linear intron RNA (Linear dV mutant RNPs, lanes 7 and 8). To confirm the nature of the low-mobility bands, we subjected DNA cleavage reaction mixtures to RNase digestion before gel electrophoresis (Figure 1B). The DNA cleavage bands remained visible at the bottom of the gel, while the bands migrating with a low mobility, possibly corresponding to the linear intron RNA attached to the 5´ exon (lane 3) or to the 3´ exon (lane 4), completely disappeared on RNase treatment (lanes 1 and 2). Similar results were obtained by subjecting reaction mixtures to alkaline digestion at 70°C for 10 minutes (data not shown). These results confirm that the bands at the top of the gel are RNA-DNA hybrid molecules susceptible to RNA degradation.

Figure 1.

Biochemical activities of RNPs reconstituted with linear RmInt1 intron RNA. (A) Comparison of the DNA cleavage activities of linear and lariat intron RNPs with DNA substrates radioactively labeled at their 5´ (5´*) and 3´ (3´*) ends. (B) Verification of hybrid RNA/DNA molecules by RNase digestion. The bands are identified on the right and their sizes are indicated on the left. 5´ DNA exon, black box; 3´ DNA exon, white box; RNA, thin line. Detailed panels correspond to the same area of the gel overexposed.

Time-courses of DNA cleavage/reverse splicing of RNPs containing linear intron RNA (Figure 2) revealed that almost immediately after the mixing of the reaction components, products derived from DNA cleavage and low migrating bands were detected with both 5´ end- and 3´ end-labeled substrates. We detected 3´ end-labeled low-mobility products barely the two minutes after the addition of RNPs to the reaction tube, and the amounts of these products reached the saturation point 30 minutes into the reaction (left panel). However, even though the 5´ end-labeled product resulting from DNA cleavage also appeared two minutes after initiation of the reaction, the band at the top of the gel corresponding to a product in which the DNA substrate is attached to the intron RNA was of a much weaker intensity than that for the partial reverse splicing product (see the upper plot for quantification). Similar behavior was observed in assays with RNPs reconstituted with lariat RNA (Figure S1). A strong accumulation of free exons was observed after two hours of reaction (Figs. 1A and 2), whereas the levels of low-mobility products stabilized after about 30 minutes (Figure 2, lanes 9–10, both panels). Spliced exon reopening (SER) has been described for other group II introns and is usually observed during forward-splicing assays, consistent with the hydrolytic cleavage of the joined exons at the splice site catalyzed by group II introns, but without intron insertion [4,35]. The accumulation of free ³²P-labeled 3´ exon may therefore indicate the occurrence of a parallel reaction contributing to exon reopening. Taken together, our data indicate that the first step of RmInt1 linear intron reverse splicing is rapid, specific and precise, but that the completion of intron integration is rate-limiting.

Figure 2.

Time-course of the reaction between the 5´-triphosphorylated linear intron and the DNA substrates labeled at the 3´end (left) or 5´end (right gel). Samples were taken at 0, 2, 5, 10, 15, 20, 30, 60, 120, and 300 minutes. The progress of the reaction for band quantification was plotted over time, with reactions performed with 3´end-labeled DNA substrates shown in gray and reactions carried out with 5´end-³²P-labeled DNA substrates shown in black. The inset graph in the reverse splicing plot focus closely on the reverse splicing levels of RNPs reconstituted with linear RNA on 5´-labeled DNA substrates. For product identification, see the caption to Figure 1; *, labeled molecules.

Identification of low migrating intron products generated with RNPs containing linear RmInt1 RNA

To test whether the linear intron was linked to the 5´and 3´exons, the low-mobility products of the reaction of linear intron RNPs with labeled ssDNA substrates were gel-purified and sequenced after RT-PCR (Figure 3). Different combinations of primer pairs were used to detect the putative fully reverse-spliced product (primer 1 and 4) and to identify sequences at the 5´and 3´- integration junctions from the linear intron RNA (5´ junction, primers 1/2; and 3´ junction, primers 3/4). PCR products were cloned and sequenced as individual clones. The sequences of low-mobility products from different independent reactions were determined (Figure 3). Amplification with a primer pair binding to the 3´ intron-exon junction mostly generated a product corresponding to the 3´ end of the RmInt1-ΔORF intron attached to the 3´ exon. By contrast, an analysis of the 5´-integration junction revealed that most of the products corresponded to the 5´end of the intron attached to the 5´exon, but that smaller products were also generated, consistent with an intron truncated at nucleotide position 31, close to a loop region in subdomain IA. The analysis of the sequence at that position did not shed light about the appearance of this product because we could not identify any sequence resembling the intron EBSs. Finally, the sequencing of products amplified from a cDNA obtained with random oligonucleotides and primers binding to both exons identified the full-length RmInt1 ΔORF intron, inserted into the target site. Thus, our results indicate that RNPs containing the linear RmInt1 RNA can promote the full reverse splicing of the intron into ssDNA substrates at the correct insertion site in vitro. This is the first time, to our knowledge, that full reverse splicing has been detected in vitro for group II intron RNPs reconstituted with linear intron RNA.

Figure 3.

Identification of the reverse splicing products obtained with RNPs reconstituted with linear RmInt1 intron RNA. (A) The sites of primer binding for PCR amplification are shown. 5´ DNA exon, black box; 3´ DNA exon, white box; RNA, thin line. (B) A representative gel of the amplification products corresponding to the 5´ junction, 3´ junction and the full reverse splicing molecule is presented. The template for the negative control (−) was the −35/ΔORF/+75 RNA precursor transcript. The template for the positive control (+) was the cDNA obtained from a run-off transcript of the precursor intron containing −35/+75 exon sequences. Lanes −/+ RT identify the templates derived from the −/+ reverse transcription of the low-mobility bands extracted from polyacrylamide gels. (C) Sequences of the 5´ and 3´ integration junction resulting from the reverse splicing of linear RNPs, obtained by RT-PCR. Numbers on the right side correspond to the number of individual clones presenting the same sequence for the 5´ junction (upper sequences) and for the 3´ junction (lower sequences). The sequences derived from the full reverse splicing product (Full Rv Sp) are excluded in these surveys. The position numbers at the beginning and end of the intron sequences are indicated as for the ΔORF intron sequence. The expected sequence is shown in bold and the underlined sequences represent the intron binding sites (IBS1, 2 and 3). The experiment was performed at least four times, with different RNPs and substrate preparations.

DNA cleavage and full insertion of linear RmInt1 RNA depend on DNA target recognition

The RmInt1 RNP complex containing the lariat intron recognizes the DNA target site via both its RNA and protein components [36,37]. In particular, the RNA component base-pairs via its EBS sequences to nucleotide residues in the 5´ and 3´ exons (mostly at positions −13 to +1 relative to the intron insertion site), whereas the IEP interacts with distal positions in the 5´ exon (T −15 nucleotide). We assessed the relevance of target-site recognition for the efficiency of DNA cleavage and full reverse splicing in vitro for RmInt1 linear RNA, by incubating RNPs reconstituted with linear RmInt1 introns with 5´ end-labeled ssDNA substrates modified at different positions in the DNA target site (Figure 4). The substitutions made were chosen on the basis of retrohoming results obtained in vivo [36]: in the 5´ exon, the sequence was independently modified at T-15 (−15G), T-12 (−12G), and G-2 (−2C); and in the 3´ exon, the sequence was modified at C + 1 (+1A) and G + 4 (+4T). All mutant target DNAs affected DNA cleavage and/or full reverse splicing efficiency (Figure 4). The modifications to the 3´ exon slightly increased the DNA cleavage activity of the linear RNA RNPs, but did not proportionally increase the amounts of linear intron RNA attached to the 5´exon (lanes 11 and 12). Moreover, the intron RNA yielded a double cleavage band when incubated with the +1A mutant DNA substrate (lane 11), probably due to the misrecognition by EBS3 (G). Nucleotide modifications to the 5´ exon had different effects on DNA cleavage and full reverse splicing activities. The −2C substitution affecting IBS1-EBS1 pairing increased DNA cleavage and full intron insertion into the target site (lane 10). By contrast, nucleotide changes in the IBS2 (−12G) or distal 5´ exon (−15G) regions decreased DNA cleavage. However, only modifications affecting the distal region of the 5´ exon, potentially recognized by the IEP, resulted in lower levels of the low migrating product (lane 8). As expected, neither dV-mutant intron RNA (lanes 1 to 6) nor RNPs mixed with a substrate with an alien sequence (data not shown) generated any specific DNA cleavage-derived products. Similar results were obtained for DNA cleavage with RNPs containing lariat RNA, although the reverse splicing rates might indicate that the RNPs containing lariat RNA discriminate better the different mutated DNA substrates than those containing linear RNA (Figure S2, compare lanes 2 and 8, or 3 and 9). Then, the insertion of the linear intron into the target DNA therefore seems to be limited mostly by recognition of the distal 5´exon region enabling EBS-IBS pairing to occur, as with RNPs containing lariat intron RNA. Nevertheless, the lariat conformation in the RNP complexes seems to constrict the recognition the DNA substrate for reverse splicing.

Figure 4.

RNP activity is dependent on DNA target recognition. DNA cleavage and reverse splicing products observed when 5´ end-labeled point-mutated DNA substrates react with wild-type or domain V-inactive linear intron RNP complexes. The detailed upper panel corresponds to the same area of the gel overexposed. Quantification graphs show the mean values for three independent assays with different RNP preparations. Error bars correspond to the standard errors.

Linear RmInt1 RNA modified at the 5´ and 3´ ends can still insert into the target DNA site

We assessed whether modifications at the 5´ and 3´ ends of the linear intron RNA influence its insertion into the DNA target. RNPs prepared with various linear RmInt1 RNAs modified at their boundaries were incubated with 70 nt 5´ and 3´end-³²P-labeled ssDNA substrates (Figure 5). Modifications to the 5´ end of the intron had little, if any, effect on intron activity (lanes 3 to 8), whereas changes at the 3´ end affected intron functionality (lanes 9 to 16). DNA cleavage and reverse splicing levels were similar to those of the wild-type RmInt1 RNA when intron transcripts extended by 5 nucleotides at the 5´-end were used for RNP reconstitution, indicating that intron insertion may involve the incorporation of extra residues at the 5´ end of the intron (compare lanes 1 and 2 with 3 and 4, or 7 and 8), which was further confirmed by amplification and sequencing the corresponding low-mobility products. In addition, we found that the first position (G) of the intron RNA seemed to be dispensable for DNA cleavage and intron insertion in vitro (lanes 5 and 6), but a C-to-A point mutation (C740A) at the 3´ end of the intron RNA decreased intron functionality (lanes 15 and 16), whereas the same substitution at the penultimate position in the intron (C739A) had no effect on intron activity (lanes 13 and 14). The presence of these modifications in the linear intron molecules inserted between the two DNA exons was checked by sequencing. Finally, the addition of 5 nt at the 3´end of the intron greatly decreased reverse splicing reactions regardless of whether the last nucleotide was an A or the natural C (lanes 9 to 12). Together, these data suggest that the linear RmInt1 RNA can be modified at its 5´end with no significant effect on the activity of the intron RNP.

Figure 5.

Modifications to the ends of the intron RNA used for reconstitution alter RNP activity. (A) Diagram showing the wild-type intron RNA and the intron RNAs with modifications to the 5´ end and the 3´ end used for the preparation of RNPs. Sequence insertions and substitutions are shown in bold. (B) DNA cleavage and reverse splicing products observed when substrates radioactively labeled at their 5´ end (5´*) or 3´ end (3´*) were incubated with the different linear intron RNP preparations. The lower panel corresponds to the bottom part of the gel, whereas the upper panel shows the top of the gel but after a longer exposure. For product identification, see the legend to Figure 1.

Discussion

The reconstitution of RNPs in vitro has been a cornerstone of the development of potential biotechnological applications for group II introns since the late 1990s [34,38–45]. We have analyzed here the DNA cleavage and reverse splicing activities of RNPs reconstituted with linear RmInt1 RNA and an MBP-IEP fusion protein and show that they can catalyze the attachment of the intron RNA to both exons of single-stranded DNA substrates in vitro, unlike other described group II introns [43,46–48].

RNPs containing linear RmInt1 RNA catalyzed the covalent binding of the intron RNA to DNA exons but also the hydrolytic cleavage of DNA substrates

In vitro endonuclease reactions on double-stranded DNA substrates have shown that Ll.ltrB group II intron RNPs generate a double-strand break leading either to complete intron insertion if the lariat RNA is used for reconstitution [38,39,43], or to partial insertion of the intron into the top strand if linear RNA is used for RNP reconstitution [43]. Other linear group II introns lacking a protein partner can bind to the 3´ exon (DNA or RNA) but cannot reverse the first step of forward splicing [46,47], even if a potential leaving group is placed at the 5´ end [48]. Like these introns, the linear RmInt1 RNA attaches to a precise position in vitro, in the 3´ exon, due to the reversal of the second step of the splicing reaction. Furthermore, RmInt1 RNPs containing linear RNA catalyze the binding of the linear ribozyme to the last residue of the 5´ exon. As we were able to sequence the low-mobility bands isolated from our DNA cleavage reactions, it therefore seems likely that the 5´ exon is linked to the 5´triphosphate linear intron molecules by a 3´-5´phosphodiester bond releasing pyrophosphate. Nevertheless, preliminary results indicate that the attachment of the 5´ exon to a 5´ monophosphate linear molecule is also produced with similar efficiency (data not shown). Previous studies on other group II introns have yielded conflicting results. Experiments using a linear ai5γ intron modified with a suitable leaving group (a triphosphate or an m⁷G-cap) at the 5´ end generated no products corresponding to reversal of the first step in splicing [48], whereas a linear bI1 intron extended 12 nt at its 5´end and attached to the 3´exon seemed to use the breaking energy of an α-β bond in a triphosphate group for transesterification reconstituting a 5´ exon-12 nt-extended intron RNA-3´ exon molecule [47]. The dephosphorylated version of this 12 nt-extended bI1 intron also supported transfer of the 5´ exon, albeit less efficiently and generating molecules truncated at the 5´ exon-intron junction [47]. Future research should determine the mechanism by which the linear RmInt1 RNA is bound to the 5´exon.

It is plausible that molecules of 5´ exon covalently bound to the linear RmInt1 RNA but with no 3´ exon could be produced by the reaction of a hydroxyl group at the end of free 5´ exon generated by SER with the 5´ end of the triphosphorylated linear intron [47]. However, reverse splicing assays including only the 5´ exon and RmInt1 linear RNA RNPs have shown that these products did not react (Figure S4). Furthermore, we detected full intron insertion at the target site, demonstrating the occurrence of complete reverse splicing.

The differences in band intensity between experiments performed with 5´ and 3´ end-labeled substrates suggest that the second step in the reverse splicing of linear RmInt1 intron RNA may be rate-limiting. This is not particularly surprising given the low levels of full reverse splicing observed for RmInt1 lariat RNPs with ssDNA substrates [34] or for the lariat form of other group II introns in vitro with their own RNA substrates [46–48]. Indeed, the reverse branching reaction is generally considered to be inefficient in wild-type introns due to structural constraints on the catalytic core due to the change in conformation occurring between the two steps of reverse splicing [41,49–53]. On the other hand, our time-course experiments showed a rapid first-phase reaction, followed by a slow second-phase reaction. These observations are similar to the kinetic findings for the reverse splicing of other lariat (full reverse splicing) and linear (partial reverse splicing) group II introns in vitro. In these situations, the biphasic kinetics can be explained by three concomitant reactions: reverse splicing, forward splicing and SER [48,51,54]. We observed the accumulation of free exons over long reaction times. We therefore assumed that SER occurred and that this irreversible reaction competed for the substrate. The observed slower rate of hydrolytic cleavage of the substrate (SER) by linear RmInt1 RNPs than by other group II introns [48] may be related to the presence of the IEP, which not only facilitates correct intron RNA folding, but may also restrict the access of water or metal-coordinated hydroxyl groups to the catalytic core.

The RmInt1 IEP may control the rate of reverse splicing of the RmInt1 RNA

For group II introns, insertion site selection has been shown to be associated not only with the precise structure of the catalytic core, but mostly, to the correct recognition of the IBS-EBS helical structure [55,56]. As expected, DNA cleavage and reverse splicing requirements in vitro of linear RmInt1 RNPs are similar to those described for in vivo retrohoming of the RmInt1 intron [36,57]. However, conditions in vitro seem to be more permissive because single mutations of the DNA target that completely abolish retrohoming in vivo do not abolish reverse splicing activity in vitro. The residues directly involved in the base pairing with the intron RNA seem to be important for cleavage of the DNA substrate. However, the mutant target −15G (5´exon distal region) did not only produces a ladder of aberrant cleavage products, but also it allows even lower reverse splicing levels than the other mutated DNA substrates. It has been suggested that this residue is recognized by the RmInt1 IEP [36], which would suggest that the rate of reverse splicing is also dependent on the correct positioning of the protein component of the RNP complex. Indeed, an analysis of the cryo-EM structure of the L. lactis L1.LtrB group II intron in complex with its IEP suggested a direct role of the thumb domain of the LtrA protein in the stabilization of EBS-IBS interactions [58]. Moreover, it has been considered that the RNPs bind to its DNA target nonspecifically, scan the DNA, and eventually a rate-limiting conformational event of the complex occurs at suitable sequences, when the specific recognition of the DNA target by the protein component take place [41]. We cannot rule out that the architecture of the linear intron and/or the large MBP-IEP fusion protein could impair the structural reorganization of the RNP complex to recognize specifically the DNA target, generating an abundance of unproductive reaction complexes for integration. Thus, it is plausible that the forward splicing reactions are favored in our in vitro assays, resulting in a very low detection of reverse splicing products.

Modifications of nucleotides at RNA-intron boundaries may alter catalytic core architecture

The nucleotides at the intron boundaries of group II introns are strongly conserved, with consensus sequences of GUGYG at the 5′ end of the intron and AY (where Y is C or U) at the 3′ end [59], although the RmInt1 intron has two C residues at its 3´ end. The effect of these terminal residues has been studied in detail only for forward splicing efficiency. Lariat formation seems to require a non-Watson–Crick interaction between the first and penultimate nucleotides of the intron [60,61]. It is therefore feasible that the absence of the G + 1 nucleotide or the replacement of the penultimate nucleotide of the linear RmInt1 ΔORF RNA had little effect on reverse splicing efficiency. Furthermore, the last residue of the intron (γ′) is engaged in a Watson–Crick interaction with a purine (γ) residue in the segment separating domain 2 from domain 3 (J2/3), which is required for the positioning of the 3′ splice site for exon ligation [62–64] and stabilization of the catalytic core [52,58]. Despite discrepancies concerning the location of the γ position in structural studies, either in [52,58] or outside the active center [53,65], the γ´ residue seems to be located in the catalytic core. It therefore seems possible that substitution of the 3´terminal nucleotide of the intron decreases the activity of the linear intron RNP.

Finally, our detection of 5´ exon molecules linked to linear intron RNAs carrying a five-nucleotide extension at their 5´ end is a remarkable finding. Such a situation has been described for 10 organelle group II introns lacking the bulging A of domain VI, in which the end of the IBS1 sequence and the GUGYG consensus sequence are separated by one to 33 intervening nucleotides [8]. However, the catalytic core is subject to structural constraints [52,53,58,65], and the presence of additional nucleotides at intron boundaries (5´and 3´ends), and, consequently, in the active site of the ribozyme, may distort the conformation of the reactive groups, decreasing reverse splicing activity.

Potential biological relevance

Natural group II introns excising exclusively as linear molecules appear to have lost their mobility [6,8]. Interestingly, some bacterial introns generating only trace amounts of branched products during in vitro splicing are present as multiple copies in their host genomes [66–68]. Moreover, Lambowitz and coworkers demonstrated the retrohoming of a linear group II intron (Ll.ltrB) in eukaryotes by partial reverse splicing followed by reverse transcription of the intron RNA. The cDNA generated was then attached to the 5´ exon through a variant of the non-homologous end-joining (NHEJ) mechanism, a process dependent on host enzymes (Lig4 and PolQ) that frequently leads to imprecise intron insertion, with sequence losses and gains at the 5´ intron-exon junction [43–45]. Our observations that the full reverse splicing reaction of linear RmInt1 RNA occurs in vitro, even though is low and inefficient, open the possibility that some group II introns which excises as linear molecules in vivo might spread and impact the evolution of the host bacterial genome.

Materials and methods

In vitro transcription for linear RmInt1 RNA preparations

A 740 nt RmInt1-ΔORF linear intron was obtained by run-off transcription with the T7 RNA polymerase from a PCR product containing the T7 promoter at the 5´ end. The template for in vitro transcription was amplified from pKGEMA4 [69] with the Phusion DNA polymerase (Thermo Scientific) and the sense primer LMS94 and the antisense primer Oligo1, an oligonucleotide 2´-OMe-modified at the last two nucleotides at the 5´ end to prevent the addition of untemplated nucleotides to the 3´end of the transcripts [70,71] (see Table S1). We performed DNA cleavage assays on 5´-labeled 70-mer ssDNA substrates, with RNPs reconstituted from homogeneous intron transcripts with identical 3´ ends (Li 3´Homo) or with additional nucleotides at the 3´-end (Li 3´Hetero) (Figure S4), obtaining similar levels of DNA cleavage activity for both types of reconstituted RNPs, although reverse splicing levels were slightly higher when homogeneous intron RNA molecules with identical 3´-termini were used to prepare the RmInt1 RNPs [70,71]. The RmInt1 intron RNA precursor was synthesized as described by Molina-Sánchez et al. (2016) [34]. For modifications introduced into the 5´end of the RNA, the Oligo1 primer was used as the 3´ primer; conversely, if the RNA was modified at its 3´end, the corresponding primer in 3´ was LMS94. Transcripts were generated with 15 µg of the PCR product and 70 units of T7 RNA polymerase in 1 ml reaction buffer containing 1× transcription buffer [10× transcription buffer: 150 mM MgCl₂, 400 mM Tris–HCl, pH 7.5, 20 mM spermidine, and 50 mM DTT], 0.96 mM NTPs, 10 mM DTT and 400 units of RNaseOUT (Invitrogen). RNA products were precipitated after 5 h of incubation at 37°C, treated with DNase I and gel-purified before use [30,34]. RNA molecules of the correct size were excised from the gel and eluted by simple diffusion during overnight incubation in elution buffer (0.3 M NaCl, 10 mM MOPS pH 6, 1 mM EDTA). RNA was precipitated, dried and stored in 10 mM MOPS pH 6 and 1 mM EDTA.

RmInt1 IEP expression and purification

The MBP-IEP fusion protein was produced and purified essentially as described by Molina-Sánchez et al. (2016) [34]. Freshly transformed E. coli Rosetta-gami (DE3):pLysS cells harboring pMALFlagIEP were cultured in 50 ml of LB medium supplemented with 0.2% glucose, 100 µg·ml⁻¹ ampicillin and 50 µg·ml⁻¹ chloramphenicol for 4 h at 37°C with vigorous shaking, until the OD₆₀₀ reached 0.5 units. They were then induced with 0.3 mM IPTG at 20°C for 16–22 h. Cells were lysed by freeze-thaw cycles followed by sonication. Cleared lysates were passed through chromatography columns (BioRad) containing amylose beads (New England Biolabs). MBP-IEP was eluted by adding 10 mM maltose (Sigma-Aldrich), dialyzed and concentrated with YM-30 centrifugal filters (Amicron Ultra, Millipore). Protein concentrations were determined by the Bradford method, with the Bio-Rad protein assay reagent and BSA as the standard (Bio-Rad). Protein preparations were analyzed by electrophoresis in Coomassie blue-stained 0.1% SDS-10% polyacrylamide gels and by immunoblot analysis with antibodies against the Flag epitope (Sigma-Aldrich).

Reconstitution of RNP particles in vitro

RmInt1-ΔORF RNP particles were reconstituted from in vitro-synthesized intron RNA and purified MBP-IEP, as described by Molina-Sánchez et al. (2016) [34]. Linear intron RNA (2.5 µM) in 40 mM Tris-HCl pH 7.5 was denatured by heating at 90°C for 1 minute and then renatured by successive incubations at 50°C for 1 minute and slow cooling at room temperature, before the addition of 500 mM NH₄Cl, 5 mM MgCl₂ and 5 µM purified MBP-IEP protein. The RNA/protein mixture was transferred to a water bath at 30°C for 1 h, and the reconstituted RNPs were then kept on ice until use.

Substrate labeling and DNA cleavage assays

DNA cleavage assays were carried out principally on a 70-mer single-stranded DNA oligonucleotide substrate containing the intron insertion site at position 35 (see Table S2). The substrate was labelled at its 5´ end with [γ-³²P]ATP (6,000 Ci/mmol; Perkin-Elmer) and T4 polynucleotide kinase (New England Biolabs), or at its 3´ end with terminal transferase (New England Biolabs) and cordycepin 5´-triphosphate [α-³²P] 3´-deoxyadenosine (6,000 Ci/mmol; Perkin Elmer). The sequences of the other mutated 70-mer single-stranded DNA substrates used for the experiment described in Figure 4 are included in Table S2.

Reconstituted RNP particles (2.5 µl; ~900 nM) were incubated with the substrate (300,000 cpm) at 37°C for 2 h in reaction buffer containing 50 mM Tris-HCl pH 7.5, 10 mM KCl, 25 mM MgCl₂ and 5 mM DTT. For time-course experiments, samples were taken at the indicated times and placed on ice. The reaction mixtures were cleaned up by extraction with phenol-chloroform-isoamyl alcohol (25:24:1) followed by ethanol precipitation. Products were analyzed by electrophoresis in denaturing 7 M urea-6% (w/v) polyacrylamide gels and quantified with Quantity One software (Bio-Rad).

Identification of low-migrating products

Low-migrating products were gel-purified and eluted by simple diffusion overnight in elution buffer (0.3 M NaCl, 10 mM MOPS pH 6, 1 mM EDTA). The DNA/RNA hybrid molecules were precipitated, dried and stored in RNase-free water. cDNA was synthesized with 2 µg N₆ random oligonucleotides or 10 pmol S70ds/DN (complementary to the 3´ end of the substrate) and 200 units SuperScript II reverse transcriptase (Invitrogen) in the presence of 1.5 mM dNTPs, 15 units RNase OUT (Invitrogen), 50 mM Tris-HCl pH 8.3, 75 mM KCl, 2 mM MgCl₂ and 20 mM DTT for 2 h at 42°C. Integration junctions were identified by PCR with 10 pmol of each primer and 2 units Taq polymerase. For the 5´ exon-intron junction, S70ds/UP (complementary to the 5´ end of the substrate) and R1 (complementary to positions 348–365 of intron domain I) were used, whereas for the intron-3´ exon junction, H3 (located at position 520–539 of the intron domain IV) was used with oligonucleotide +9 (binding between positions 9 and 28 in the 3´ exon). The PCR conditions were as follows: denaturation at 94°C for 3 min, followed by 25 cycles of 20 s at 94°C, 20 s at 60°C, 30 s at 72°C, and a final extension at 72°C for 10 min. The full reverse splicing product was amplified with S70ds/UP and S70ds/DN, by increasing the extension time to 60 s and the number of cycles to 35. The RT-PCR products were visualized on agarose gels and inserted into the pGEM-T easy vector (Promega) for the sequencing of individual molecules. The primer sequences are indicated in Table S3.

Funding Statement

This work was supported by the Ministerio de Ciencia e Innovación [BIO2017-8244-P]; Ministerio de Ciencia e Innovación [BIO2014-51953-P].

Acknowledgments

We thank Jose María del Arco and Ascensión Martos for technical assistance. This work was supported by research grants [BIO2014-51953-P] and [BIO2017-8244-P] from the Plan Nacional de I+D+i, biotechnology program of the Spanish Ministerio de Ciencia, Innovación y Universidades including European Regional Development Funds (ERDF).

Disclosure statement

The authors have no competing interests to declare.

Supplementary material

Supplemental data for this article can be accessed here.