Replacement of the λ boxB RNA–N peptide with heterologous RNA–peptide interactions relaxes the strict spatial requirements for the formation of a transcription anti-termination complex

Authors


*E-mail harada@u-gakugei.ac.jp; Tel. (+81) 42 329 7550; Fax (+81) 42 329 7550.

Summary

In bacteriophage λ, formation of a transcriptional anti-termination complex involving the elongating RNA polymerase is mediated by the interaction of boxB RNA with the RNA-binding domain of the N protein (N peptide). In an attempt to understand the spatial requirements for boxB/N peptide interaction within the anti-termination complex, the effects of changes in the distance between boxA and boxB RNA, the length of the boxB stem, and the distance between the N peptide and remainder of the N protein were examined using a bacterial reporter system. It was found that the requirements for boxB stem length and the distance between N peptide and the remainder of N were optimized and strict. In contrast, replacement of the boxB/N interaction by heterologous RNA–peptide interactions appeared to relax the strict requirement for RNA stem length and the orientation of the RNA-binding peptide, presumably due to the absence of the cooperative interaction between boxB/N and the host factor NusA. In addition, the decrease in activity upon stem lengthening could be partially suppressed by simultaneous lengthening of the RNA spacer. A further understanding of the structural organization of the anti-termination complex may provide insights into how functional ribonucleoprotein complexes may be engineered.

Introduction

In bacteriophage λ, the expression of early operons is controlled by λ N protein through suppression of transcription termination at both Rho-dependent and intrinsic (Rho-independent) termination sites (Das, 1993; Greenblatt et al., 1993; Nudler and Gottesman, 2002; Vieu and Rahmouni, 2004). cis-acting elements termed nut (N-utilization) sites (nutL and nutR) are situated between the promoters pL and pR and the first terminator of the early operons and are necessary for N-mediated anti-termination (Rosenberg et al., 1978; Salstrom and Szybalski, 1978). The nut sites consist of three elements, boxA (Olson et al., 1982; Mogridge et al., 1998a), boxB, which codes for a hairpin RNA (Rosenberg et al., 1978; Somasekhar et al., 1982), and a short intervening spacer. Anti-termination occurs when the NH2-terminal arginine-rich region of the N protein binds to the boxB hairpin within the nut site of the nascent transcript and promotes the assembly of an anti-termination complex (Lazinski et al., 1989; Greenblatt et al., 1993; Chattopadhyay et al., 1995; Tan and Frankel, 1995; Van Gilst et al., 1997). Many protein–protein and RNA–protein interactions are required for this complex to form (Greenblatt et al., 1993). Typically, Escherichia coli host factors NusA, NusG, NusB and S10, and another nut RNA element, boxA, are needed for assembly (Mogridge et al., 1995) (Fig. 1), although anti-termination has been shown to occur in the absence of some factors under certain conditions (Whalen et al., 1988; Rees et al., 1996; Gusarov and Nudler, 2001).

Figure 1.

λ anti-termination complex. Factors in contact and linked by broken lines have been shown to interact with each other by genetic and biochemical analysis (Mogridge et al., 1995).

The interaction of the NH2-terminal arginine-rich region of the λ N protein (the N peptide) with the boxB element has been characterized in considerable detail. The λ boxB RNA consists of 15 nucleotides (nts), and folds into a hairpin structure that contains a pentaloop which forms a structure resembling that of the GNRA-tetraloop (where N is G, A, U or C; R is G or A) upon binding to a bent α-helix conformation of the N peptide (Tan and Frankel, 1995; Su et al., 1997a,b; Legault et al., 1998; Schärpf et al., 2000). The boxB/N structure is well conserved among the lambdoid phages P22 (Tan and Frankel, 1995; Cai et al., 1998) and phi21 (Cilley and Williamson, 2003), while showing somewhat different modes and specificities of binding (Austin et al., 2003; Franklin, 2004). In the case of λ, N peptide-dependent folding of the boxB pentaloop into a GNRA-like structure results in the flipping out of the fourth nucleotide in the loop, and by doing so forms the NusA-binding site (Legault et al., 1998).

A two-plasmid reporter system based on λ anti-termination that can be used to monitor the core interaction between the boxB RNA stem-loop and the N peptide has been developed by Franklin (1993). In this system, N is expressed from a pBR322-based plasmid, and β-galactosidase is expressed from a second pACYC-based reporter plasmid. The reporter plasmid contains four tandem repeats of the strong transcription terminator t1 from the rrnB operon, which are located downstream of the nut site and upstream of lacZ. When the N protein expressed from the first plasmid binds to the boxB hairpin expressed from the reporter plasmid, an anti-termination complex forms (Fig. 1), leading to anti-termination through the terminator sequences and expression of β-galactosidase. It has been shown that the boxB RNA hairpin and the NH2-terminal boxB-binding domain of the N protein (residues 1–20) can be substituted by heterologous pairs of RNAs and polypeptides, thereby enabling the detection of interactions between arginine-rich peptides such as the human immunodeficiency virus (HIV) Rev peptide, a Zif268–Rev fusion, bovine immunodeficiency virus (BIV) Tat peptide, and proteins such as snRNP U1A protein and R17 coat protein, with their counterpart RNAs (Harada et al., 1996; Wilhelm and Vale, 1996; McColl et al., 1999; Peled-Zehavi et al., 2003). Despite a decrease in anti-termination activity upon introduction of heterologous interactions, presumably due to the loss of the interaction between the NusA protein and the boxB/N complex, the two-plasmid system has been shown to be a sensitive method for the analysis of small changes in binding affinity of RNA–polypeptide interactions. This bacterial system has also been shown to be a powerful tool in the screening and selection of novel RNA-binding peptides targeting the HIV Rev-response element (RRE) from combinatorial libraries (Harada et al., 1996; 1997; Peled-Zehavi et al., 2003; Sugaya et al., 2008a), as well as for the selection of novel peptide-binding RNAs (Iwazaki et al., 2005; Sugaya et al., 2008b). However, there appeared to be limitations on the size of RNAs that can be accommodated in the anti-termination complex, with only the interaction of relatively small RNA stem-loops with their cognate polypeptides being detected using the two-plasmid system. While the types of heterologous RNAs and peptides that can be introduced into the anti-termination complex have not been systematically examined, some variability in the distance between the boxA and boxB elements seems to be tolerated (Neely and Friedman, 1998; 2000).

In this study, in order to understand the spatial requirements within the anti-termination complex for detection of RNA–polypeptide interactions, and to gain insight into the structural organization of the anti-termination complex, we examined the effect of the spatial context of the boxB or other RNA sites with respect to the boxA element, and of the distance between the N-peptide (residue 1–20) or other RNA-binding peptides and the remainder of the N protein (residue 21–107). We utilized the two-plasmid assay described above (Franklin, 1993) and its modified systems in which the boxB RNA–N peptide interaction (boxB/N) was replaced with the heterologous RNA–protein complexes of the RRE RNA and the Rev peptide (RRE/Rev), the BIV trans-acting responsive element (TAR) RNA and the BIV Tat peptide (TAR/Tat), U1 hairpin II (hpII) RNA and the U1A protein (hpII/U1A), and the complexes of two selected RRE-binding peptides, RSG-1.2 and K1, with the RRE (RRE/RSG-1.2 and RRE/K1 respectively) (Harada et al., 1996; 1997; Harada and Frankel, 1998; Peled-Zehavi et al., 2003). The use of heterologous interactions was expected to be informative because the structures of the RNA–peptide complexes are diverse, and may be expected to impose different spatial requirements on the anti-termination complex. The Rev and K1 peptides bind to an internal loop of RRE IIB in α-helical conformations (Tan et al., 1993; Battiste et al., 1996; Sugaya et al., 2008a), the RSG-1.2 peptide binds to the same site in an extended-turn-helix conformation (Gosser et al., 2001; Zhang et al., 2001), the BIV Tat peptide binds to the BIV TAR bulge in a β-hairpin conformation (Puglisi et al., 1995), and the U1A protein binds to the apical loop of the U1 hpII as an RNP motif (Oubridge et al., 1994).

We found that lengthening the boxB stem and insertion of a peptide spacer between residues 20 and 21 of N dramatically reduced anti-termination activity, presumably due to the strict spatial requirements for the formation of the boxB/N/NusA ternary complex, while increasing the distance between the boxA and boxB elements led to only a gradual decrease in activity. In contrast, replacement of the boxB/N complex with the heterologous RNA–peptide interactions led to a considerable relaxation in the stringency of the length of the peptide linker in particular, and also stringency of the RNA stem length, although to a lesser extent. In addition, simultaneous lengthening of the nut spacers was shown to partially compensate for the loss of function by the stem elongation in some cases. These results suggest that a further understanding of the spatial requirements of the anti-termination complex may enable the incorporation of a wider range of RNA–polypeptide complexes in the place of the boxB–N peptide complex, thereby demonstrating how a functional ribonucleoprotein complex may be engineered in a predictable manner.

Results

Lengthening the stem of boxB and heterologous RNA sites

The effect of the insertion of RNA duplexes into the base of the boxB and U1 hpII stem, and the lower stem of the RRE and TAR stem-loop was examined (Fig. 2A and B, Table S1). LacZ reporter plasmids where 5 or 11 base pair (bp) duplexes were inserted into the base of the λnutR boxB or a nut site in which boxB was replaced with HIV RRE, BIV TAR or U1 hpII (designated st+5 or st+11, Fig. 2A) were constructed. The secondary structures of the RNAs were consistent with those obtained using the MFOLD RNA folding algorithm (Jaeger et al., 1989; 1990; Zuker, 1989) (data not shown). Anti-termination activities in the presence of the cognate or non-cognate binding peptide or protein were assayed for β-galactosidase activity by colony colour assays on plates containing 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (Xgal), and in solution using o-nitrophenyl-β-D-galactopyranoside (ONPG). We have already shown that anti-termination activities reported using β-galactosidase and scored as the intensity of colony colour correlates well with the strength of the RNA–peptide interaction replacing the boxB RNA and N peptide, and should be an accurate measure for the stability of the anti-termination complex. Colony colour was scored from 0 to 8 plusses (8+) by comparison with a standardized set of controls (see Fig. 3 for representative colony colours). For example, the interaction between the RRE RNA and the DLA peptide (Kd = 0.5 nM), which is related to the K1 peptide used in this study, reported an anti-termination activity of 6+ (Sugaya et al., 2008a), while that of the RRE and the RSG-1.2 peptide (Kd = 6 nM) was 5+ (Harada et al., 1997), and that of the RRE and the Rev peptide (Kd = 40 nM) was 3+ (Harada et al., 1996), where the number of plusses corresponds to the intensity of blue colony colour. One colony colour unit has been shown to correspond to roughly a fivefold difference in RNA–peptide binding affinity (M. Sugaya and K. Harada, unpublished data).

Figure 2.

The effect of the lengthening of the nut site linker and the stem of the boxB or heterologous RNA site on anti-termination activity.
A. The secondary structures of the nut site are shown with RNA sequence insertions indicated using one-sided arrows, and RNA sequence replacements indicated using two-sided arrows. The broken arrows under the sequence of sp+32/18nt SL and sp+32/26 nt SL indicate formation of an RNA stem.
B. The effect of lengthening the RNA stem of the boxB and heterologous RNA sites on the anti-termination activities of boxB/N and heterologous RNA–peptide interactions. The numbers of plusses above bars indicate the β-galactosidase activities assayed by the colony colour on Xgal plates.
C. The effect of lengthening the spacer length between the boxA element and the boxB or heterologous RNA sites on the anti-termination activities of boxB/N and heterologous RNA–peptide interactions.

Figure 3.

Representative colony colours of standardized controls used in the β-galactosidase colony colour assays after incubation for 30 h (24 h at 37°C followed by 6 h at 24°C) or 72 h (an additional 42 h at 24°C) on Xgal/tryptone plates. The anti-termination activity was represented as the number of plusses; boxB/N: 8+ RRE/K1: 6+ RRE/RSG-1.2: 5+ RRE/Rev: 3+ RRE/Rev(A20): 2+ boxB/N(A20): 1+ boxB/Tat: 0.

When 5 bp and 11 bp stems were inserted into the bottom of the boxB RNA stem-loop, a complete loss in anti-termination activity (scored as 0) from a colony colour of 8+ in the wild-type context (designated st+0) was observed (Fig. 2B). However, when the boxB/N interaction was replaced by the heterologous interactions, with the exception of RRE/RSG-1.2, residual activity of 0.5+ to 1.5+ was observed upon insertion of a 5 bp stem (st+5), showing that the lengthening of the stem was less disrupting with the heterologous RNA–polypeptide interactions. This difference in the effect of stem lengthening between the wild-type boxB/N and the heterologous interactions is presumably related to the presence of the boxB/N/NusA ternary complex in the wild-type context, which is absent in the case of the heterologous interactions.

In the case of the hpII/U1A, RRE/K1 and RRE/Rev interactions, further lengthening of the stem (st+11) did not reduce activity, and in the case of RRE/K1, a slight increase in activity from 1+ to 1.5+ (Fig. 2B), which was consistent with the results of the solution assay (Table S1), was observed. The copy number of the reporter plasmids was estimated by quantification of the amount of plasmid DNA per OD unit of bacterial culture, and were found to be similar (Fig. S1 and Table S2), showing that the differences in anti-termination activity were not due to changes in the amount of reporter plasmid. This suggested that the reduction in anti-termination activity was not simply the result of lengthening the RNA stem, but may also be due to the rotation of the RNA site around the helix axis. Indeed, lengthening of the upper stem region of the RRE had only a small effect on anti-termination activity (Fig. S2). Interestingly, when comparing the effect of the lengthening of the RRE stem in the presence of the K1 and Rev peptides, a somewhat similar behaviour was observed in that activity was retained for both st+5 and st+11, suggesting that the orientation of these two RNA–peptide complexes may be similar.

Intervening spacer between boxA and boxB

Next, the effect of the change in the length of the flexible single-stranded spacer connecting the boxA element and boxB or the heterologous RNA site on anti-termination activity was analysed. We constructed reporters that have a deletion of 2 or 4 nts (designated sp-2 or sp-4), and an insertion of 8, 16 or 32 nts (designated sp+8, sp+16 or sp+32), respectively, with respect to the wild-type nut spacer (Fig. 2A), and analysed their anti-termination activities on plates containing Xgal (Fig. 2C) and in solution using ONPG (Table S3). As a result, shortening of the linker by two bases (sp-2) in the boxB/N context led to a dramatic reduction in anti-termination activity to 2.5+ compared with 8+ for the wild-type context (sp+0), consistent with a previously report (Doelling and Franklin, 1989). A similar reduction in activity was observed in the hpII/U1A context upon shortening the linker by 2 nts. In contrast, shortening the linker by 2 nts (sp-2) in the RRE/Rev, RRE/RSG-1.2, RRE/K1 or TAR/Tat context resulted in anti-termination activities higher than those in the wild-type linker context (sp+0). However, further shortening of the linker by 2 nts (sp-4) in the case of the RRE led to a dramatic loss of activity. These results showed that the optimal length of the linker differs for the different RNA–peptide complex, and suggest that the wild-type linker has been evolutionarily optimized for the boxB/N interaction.

On the other hand, the elongation of the RNA linker led to a gradual decrease in anti-termination activity. In the case of the boxB/N and hpII/U1A interactions, a considerable amount of residual activity was observed even after insertion of a 32 nt linker, which is in contrast to the case of the RRE/K1, RRE/RSG-1.2, RRE/Rev and TAR/Tat interactions. This difference may reflect differences in the role that the boxA element plays in the formation of the wild-type and modified anti-termination complexes, and in the case of the hpII/U1A interaction, due to the high expression level of the U1A protein and a high background of anti-termination (Fig. S3A, lower).

In order to show that the reduction of anti-termination activity upon lengthening the spacer was due to an entropic effect, and not due to the disruption of anti-termination complex formation by steric clash, two additional constructs based on λ boxB (sp+32) were prepared. Both constructs contain 32 extra nts in the spacer, but one was capable of forming a 18 nt stem-loop [designated (sp+32/18 nt SL)], while the other can form a 26 nt stem-loop ([sp+32/26 nt SL]), resulting in an effective linker length corresponding to that of a 16 base linker (sp+16) and 8 base linker (sp+8) respectively (Fig. 2A). As might be expected, as the spacer region in H-19B nut site has been shown to contain similar stem-loop structures (Neely and Friedman, 1998; 2000), the activities of the two spacer constructs capable of forming stem-loops were dependent on the length of single-stranded region within the elongated spacer, and reported higher activities corresponding to the shortening of the effective length of the spacer (Fig. 2C, boxB/N, open circle and diamond).

Combination of boxB stem elongation and the spacer length of nut

As the dramatic loss of activity upon lengthening the stem region of the RNA site may be due in part to the disruption of interactions with factors linked to the boxA element (Fig. 1), it seemed possible that lengthening the spacer between boxA and the RNA site may suppress this loss of activity. In order to test this hypothesis, reporter constructs with combinations of both the elongated stems and spacers described above were prepared, and anti-termination activities were assayed on plates containing Xgal and in solution using ONPG (Fig. 4 and Table S3). As a result, while lengthening the spacer did not completely suppress the loss of function by stem lengthening, a significant recovery of activity was observed in some cases. In the case of the boxB/N interaction, simultaneous lengthening of the stem and spacer by 5 bp and 16 nts [boxB (sp+16; st+5)], respectively, resulted in a particularly high level of recovery to 4+ for the colony colour assay compared with an activity of 0 when only the stem of boxB was lengthened [boxB (sp+0; st+5)]. In contrast, when the stem of boxB was lengthened by 11 bp, the highest recovery was observed when the RNA spacer was lengthened by 8 nts [boxB (sp+8; st+11)], possibly reflecting the different spatial requirements imposed by lengthening the stem by 5 and 11 bp. Similarly, a moderate suppression of the loss of activity was observed in the case of the hpII/U1A, the RRE/Rev, and the TAR(st+11)/Tat interaction. In the case of the RRE/K1 and TAR(st+5)/Tat interactions, while suppression of the loss of activity was not observed by simultaneous lengthening of both the stem and linker, lengthening the stem appeared to result in a reduction of the decrease in activity upon lengthening the RNA linker. Taken together, the above results suggest that the length of the RNA stem and linker are somehow linked in the formation of a functional anti-termination complex.

Figure 4.

The effect of the combination of lengthening the RNA stem of the boxB and heterologous RNA sites and the spacer length between the boxA element and the boxB or heterologous RNA sites on the anti-termination activities of boxB/N and heterologous RNA–peptide interactions. The numbers at the bottom of the bars indicate the nucleotide length inserted to the spacers. The percentage of wild-type (WT) is normalized to the average β-galactosidase Miller units of the wild-type reporter (sp+0; st+0) assayed the same day. The numbers above bars indicate the β-galactosidase activities assayed by the colony colour on Xgal plates.

Insertion of peptide linkers following the NH2-terminal RNA-binding domain of N

As the lengthening of the RNA spacer within the nut site did not completely suppress the loss of function by the lengthening of the stem of boxB or the other RNA sites as shown above, we next examined the effect of lengthening the space between the RNA-binding peptide (N1-20 or heterologous peptides) and the activation domain (N21-107) of the N protein which is involved in NusA and RNA polymerase binding (Fig. 5 and Table S4). A tandem repeat of one, two or four linker peptides consisting of five alanine residues (AAAAA, or A5), or a combination of three glycines and two serines (GSGSG, or GS5), designated A5, A10, A20 and GS5, GS10 GS20, respectively, in addition to a C-terminal alanine residue in the linkers for cloning purposes, was inserted into the BsmI site of the N expressor plasmid, resulting in the insertion of 6, 11 and 21 residues respectively (Fig. 5A). These polyalanine or glycine/serine linkers were expected to possess somewhat different structural features in that oligoalanines have a high α-helix propensity, while glycine/serine repeats are flexible, thereby possibly influencing the stability of the RNA–peptide complex as well as that of the anti-termination complex. We have previously shown that four alanines linked to the C-terminus of the α-helical Rev peptide led to a substantial increase in anti-termination activity towards the RRE, while a triglycine linker reduced activity (Harada and Frankel, 1998). On the contrary, when the triglycine linker was attached to the BIV Tat, which forms a β-hairpin, an increase in activity was observed (Harada and Frankel, 1998).

Figure 5.

The effect of the lengthening of the linker connecting the N peptide (N1-20) and the remainder of N (N21-107) on anti-termination activity.
A. Schematic representation of N and N-fusion protein and their NH2-terminal sequences expressed by the N-expressor plasmids. The Lys19 of the N protein was changed to Asp19 to generate a BsmI site with a negligible effect on N function (Franklin, 1993). Two sided arrows indicate the substitution of the N peptide sequence with heterologous RNA-binding peptides, and one-sided arrows indicate the insertion of the peptide linker.
B and C. The effect of peptide linker insertion on the anti-termination activity of boxB/N and heterologous RNA–peptide interactions. Closed triangles indicate AAAAA linkers and open circles indicate GSGSG linkers. β-Galactosidase activities were measured on Xgal plates (B) and in solutions using ONPG (C). Fold activation (ONPG) is the ratio of β-galactosidase in Miller units of the cognate reporter in the presence of N or N-fusion divided by the β-galactosidase in Miller units of non-cognate reporter (TAR for N, K1, RSG-1.2 or Rev series and boxB for Tat series).
D. North Western blotting of N and RSG-1.2–N fusion protein with peptide linker insertions. N and RSG-1.2–N fusion proteins with peptide linker insertions were expressed in E. coli N567 with (+) or without (−) IPTG induction and subjected to 15% SDS-PAGE followed by North Western blotting with biotinylated RNA probes. Each lane contains the extract from 0.08 OD600 units of original culture (1 ml). An arrowhead and white asterisks in the left panel indicate the position of non-specific bands and the bands of N protein with or without peptide linkers respectively.

The results of the peptide linker insertions assayed in the wild-type nut context (sp+0; st+0) are shown in Fig. 5B and C and Table S4. Insertion of even a relatively short A5 or GS5 linker in the boxB/N context led to a dramatic decrease in activity (8+ to 0.5+), similar to when the boxB stem was lengthened. On the other hand, in the case of the heterologous RNA–peptide interactions, lengthening the peptide linker led to only a gradual decrease in anti-termination activity. Insertion of the peptide linkers between residues 20 and 21 of the N protein result in changes in the C-terminal region of what has been defined as the N peptide (residues 1–22). Although residues 20–22 of the N protein have been shown to be dispensable for specific binding to boxB in vitro (Cilley and Williamson, 1997; Su et al., 1997a), in order to confirm that insertion of peptide linkers does not affect boxB binding or expression levels of N, Western and North Western blotting analyses of wild-type N as well as Rev, RSG-1.2 and K1 peptide N-fusion proteins without and with peptide linkers were carried out. Western blotting analysis showed that while protein levels were somewhat different depending on the peptide fused to N21-107, and on the type of linker, these differences did not correlate with the differences observed in anti-termination activity (Fig. S3). In addition, North Western analysis of the N and the RSG-1.2–N fusion proteins, indicated that the insertion of peptides linkers did not affect boxB and RRE binding respectively (Fig. 5D).

In the case of the heterologous RNA–peptide interactions, no significant difference in the linker type was observed in the RRE/K1 and TAR/Tat context, while the flexible glycine linker was favoured in the RRE/RSG-1.2 interaction, and the alanine linkers with high α-helical propensity seemed to be preferred in the RRE/Rev context. These differences in the types of peptide linkers preferred may reflect the different constraints that the RNA–peptide complexes impose on the anti-termination complex, because it has been shown by nuclear magnetic resonance spectroscopy that the Rev and RSG-1.2 peptides bind to the RRE in completely different orientations (Battiste et al., 1996; Gosser et al., 2001; Zhang et al., 2001), so that the remainder of the N protein (residues 21–107) would be positioned in a different direction.

As in the case of boxB RNA stem lengthening, the dramatic decrease in activity for the boxB/N interaction upon insertion of peptide linkers into N may reflect strict spatial requirements for the orientation of the boxB/N/NusA ternary complex (Mogridge et al., 1995; Legault et al., 1998). In the case of the heterologous RNA–peptide pairs, a significant relaxation in the stringency towards the length of the peptide linker was observed, presumably due to the loss of the interaction with NusA. As the loss of activity upon lengthening the boxB RNA stem could be partially suppressed by lengthening the nut linker RNA, we investigated the possibility that loss of activity upon lengthening the RNA stem could be suppressed by peptide linker insertion. However, unlike in the case of simultaneous lengthening of the RNA stem and nut linker (Fig. 4), no significant compensation of the decreased activity by stem lengthening was observed upon lengthening the peptide linker (Fig. S4 and Table S4). However, as in the case of the simultaneous RNA stem/nut spacer elongation, the decrease in activity observed upon lengthening the peptide linker in the st+0 context appeared to be more gradual when the RNA stem was lengthened by 5 or 11 bp (st+5 and st+11). For example, in the case of RRE/K1 context, the activity of (st+0)/(GS0) was down from 6+ to 3+ for (st+0)/(GS20), while the activity of (st+5)/(GS0) was as same as (st+5)/(GS20). This suggested that the RNA stem and peptide linker length are somewhat correlated even though the decrease in activity of stem elongation could not be recovered by peptide linker elongation.

Discussion

The spatial requirement for the wild-type nut site RNA and N peptide within the anti-termination complex is optimized and strict

It has been shown that high concentrations of N can induce anti-termination in vitro, even in the absence of boxB, demonstrating the central role of N in this process (Rees et al., 1996). However, as the interaction of N with the transcription complex has been estimated to be relatively weak, the formation of a boxB/N/NusA complex with RNAP has been shown to be necessary for anti-termination in vivo, thereby allowing transcription through terminators located just downstream of the nut site (Whalen and Das, 1990; Mason et al., 1992). Processive anti-termination through distant terminators requires NusB, NusG and S10 along with boxA RNA, suggesting that these proteins can be considered as processivity factors. The formation of the core boxB/N/NusA complex has been shown to occur first by the binding of the amino terminal region of N as a bent α-helix with the 5′-strand of the boxB stem and the first three residues of the loop (Legault et al., 1998). This is accompanied by the extrusion of the fourth nucleotide in the loop so that the remaining nucleotides form a GNRA fold. NusA interacts with this newly created boxB/N binding surface, while also forming a protein–protein contact with amino acids 34–47 of N (Mogridge et al., 1998b), and by doing so further strengthening the interaction of N with RNAP (Fig. 1). Although NusA has been shown to bind fairly tightly to the free N protein with Kds as high as 70 nM (Van Gilst et al., 1997), mutations in the boxB loop such as the deletion of the extruding fourth nucleotide of boxB lead to a loss of NusA-binding in vitro, while still maintaining the boxB/N interaction (Chattopadhyay et al., 1995; Legault et al., 1998).

One possible explanation for the dramatic decrease in anti-termination activity upon lengthening the boxB stem and the N peptide linker (Figs 2B and 5B and C) may be the loss of cooperativity in the formation of the boxB/N/NusA complex, indicating that the spatial requirement of the individual components is important for stable ternary complex formation. However, considering that the stability of the boxB/N and RRE/Rev peptide have been shown to be fairly similar by in vitro gel shift experiments under similar conditions (20 nM and 40 nM respectively) (Tan and Frankel, 1995; Harada et al., 1996), loss of the NusA interaction, for example by peptide linker lengthening, would be expected to result in residual activity similar to that of the RRE/Rev interaction (3+). Therefore, a more likely reason for the dramatic loss of activity in the wild-type context may be that the orientation of the boxB/N/NusA complex within the anti-termination complex is restricted, and that slight changes in the orientation of this ternary complex may cause strain to other interactions such as those involving RNAP, boxA, and other host factors, and destabilize the anti-termination complex. The partial suppression of the complete loss of activity upon lengthening the boxB stem by 5 bp (Figs 2B and 4, boxB/N, st+5) by the insertion of a 16 base linker to 4+ (Fig. 4, boxB/N, st+5/sp+16) may be regarded as a partial restoration of the cooperativity of the boxB/N/NusA ternary complex within the anti-termination complex. On the other hand, the introduction of mutations in the boxB loop that disrupt NusA-binding, while leading to an overall decrease in anti-termination activity, may be expected to result in a relaxation of the strict spatial requirement for the boxB/N complex, as in the case of the heterologous RNA/peptide interactions. In order to test this hypothesis, mutant boxB reporter plasmids where the fourth base (A) in the loop was substituted to an G, C and U, or deleted were constructed, and anti-termination activities in the presence of wild-type N and mutant N's with peptide linkers were analysed (Fig. S5 and Table S4). However, the anticipated relaxation of the stringent spatial requirement was not observed for the single nucleotide mutations in the loop, presumably because these mutations were not sufficient to abolish NusA binding in vivo. The deletion of the fourth nucleotide in the loop presumably led to disruption of both NusA- and N-binding, and diminished anti-termination activity.

The lengthening of the RNA spacer connecting boxB and the boxA element, which is mainly involved in positioning the processivity factors NusB, NusG and S10, resulted in only a gradual decrease in activity (Fig. 2C). As we also showed that relatively large stem-loops could be inserted into the nut RNA linker (Fig. 2C), the gradual decrease in activity for the nut RNA linker is most likely due to the increased entropic costs involved with an increase in flexibility as observed in the case of artificially engineered DNA-binding or RNA-binding proteins (Ribas de Pouplana et al., 1996; Kim and Pabo, 1998; Campisi et al., 2001; Moore et al., 2001). In addition, for the boxB/N complex, optimal anti-termination activities were observed in the wild-type context, indicating that the spatial context of this interaction has been evolutionarily optimized.

Relaxed spatial requirements for the heterologous RNA–peptide modified anti-termination complex and implications for the engineering of functional ribonucleoprotein complexes

Replacement of the boxB/N complex with heterologous RNA–peptide interactions was shown to lead to a considerable relaxation of the strict spatial requirements for the RNA stem and N peptide linker observed in the formation of the wild-type anti-termination complex (Figs 2B and 5B and C), presumably due to the absence of the interaction of boxB/N with NusA in the wild-type complex. However, as a consequence of the replacement of the boxB/N interaction by heterologous RNA–peptide interactions, a considerable decrease in anti-termination activity was observed. For example, replacement of the boxB/N interaction with RRE/Rev resulted in a decrease in colony colour of 8+ to 3+, and a 40-fold reduction in β-galactosidase units. As the Kd of the boxB/N interaction (20 nM) has been shown to be similar to that of the RRE/Rev interaction (40 nM) (Tan and Frankel, 1995; Harada et al., 1996), the difference in anti-termination activity of 8+ and 3+, respectively, may account for the contribution of the interaction of boxB/N and NusA on the stability of the anti-termination complex. On the other hand, replacement of the RRE/Rev interaction with those of RRE/RSG-1.2 and RRE/DLA with Kds of 6 nM and 0.5 nM, resulted in increases in anti-termination activities to 5+ and 6+ respectively (Harada et al., 1997; Sugaya et al., 2008a). This shows that the decrease in anti-termination activity due to the loss of the NusA interaction can be partially recovered to about 15% that of the boxB/N interaction in β-galactosidase units by increasing the affinity of the heterologous RNA–peptide interaction, even though the RNA–protein interaction network within the anti-termination complex has been considerably modified.

While it is not clear whether full anti-termination activity as seen in the wild-type context can be achieved for the heterologous RNA–peptide interactions (Xia et al., 2003), a number of strategies are conceivable. First, optimization of the interaction of the individual components may further increase the stability of the anti-termination complex. In fact, we have found that optimization of the NusA-binding domain of N (residues 34–47) by randomization and selection leads to N variants with increased anti-termination activity (H. Suzuki and K. Harada, unpublished), demonstrating the potential of such an approach. A second strategy for the improvement of anti-termination activity is the simultaneous manipulation of the orientation of multiple components of the anti-termination complex as attempted in this study. While dramatic increases in activity were not observed in this study, a further understanding of the structural organization of the anti-termination complex may lead to the identification of alternative parameters for optimizing the spatial orientation of individual domains.

This study has also highlighted the importance of the length of RNA stem regions in the structural manipulation of the anti-termination complex, and possibly also for the engineering of any functional ribonucleoprotein complex. This is because even small changes in the length of RNA stem regions may result in dramatic changes in the orientation of associated factors. As illustrated in Fig. 6, insertion of a 5 bp stem would result in a completely different orientation for the remainder of the N protein (19–107) that contains the NusA and RNA polymerase binding domain, while insertion of an 11 bp stem would result in an orientation similar to the RNA site without an insertion (st+0). This may explain why a 5 bp insertion leads to a dramatic decrease in activity, while further decrease in activity is suppressed, or in the case of RRE/K1 a slight recovery of activity is observed, compared with st+5 upon further lengthening to st+11 (Fig. 2B and Table S1). Similarly, this may also explain why in the boxB/N interaction, the loss of activity in the st+5 stem was optimally suppressed by the sp+16 linker, while the shorter sp+8 linker was optimum for the st+11 stem (Fig. 4).

Figure 6.

A schematic illustration of the topological change of the boxB RNA stem-loop and the N protein upon elongation of the bottom stem in RRE/K1.

Taken together, the above results lead us to believe that a further understanding of the spatial and structural requirements for anti-termination complex formation may enable the manipulation of the anti-termination complex stability in a predictable manner, and as a result, allow the accommodation of a wider range of RNA–peptide interactions. Such studies are also expected to provide useful insights into general principles for the engineering of functional ribonucleoprotein complexes.

Experimental procedures

Plasmids

Oligonucleotides used to construct the pAC and pBR constructs described in this paper were synthesized at Hokkaido System Science or Espec Oligo Service. Reporter pAC plasmids containing a wild-type spacer (sp+0) or shortened spacer (sp-4 or sp-2) at the nut site, and a part of reporters containing spacer with XhoI site were constructed by cloning synthetic oligonucleotides into the PstI and BamHI sites of the previously described pAC plasmid (referred to as pAC-TAT13 in Franklin, 1993). Based on the generated pAC plasmids containing the XhoI site, synthetic oligonucleotide cassettes were cloned into the PstI and XhoI site to construct spacer elongated constructs, and XhoI and BamHI for the replacement of the boxB site with appropriate sequences. These newly generated plasmids were further digested with the combination of PstI and XhoI, or XhoI and BamHI, and replaced with synthetic oligonucleotide cassettes to yield appropriate constructs. N or N-fusion protein expressors with peptide linkers were generated by insertion of synthetic oligonucleotides cassettes into the BsmI sites of the previously described pBR-based plasmids encoding N (Franklin, 1993), HIV Rev-N and BIV Tat-N (Harada et al., 1996), RSG-1.2–N (Harada et al., 1997) or U1A-N and K1-N (Peled-Zehavi et al., 2003). The sequence of the sense strands of inserted linkers is as follows, A5: 5′-GCCGCTGCGGCCGCGGCA-3′; A10: 5′-GCCGCTGCGGCCGCGGCAGCCGCCGCGGCTGCA-3′; A20: 5′-GCCGCTGCGGCCGCGGCAGCCGCCGCGGCTGCAGCTGCGGCGGCGGCCGCTGCGGCTGCCGCAGCA-3′; GS5: 5′-GGCAGCGGTAGCGGCGCA-3′; GS10: 5′-GGCAGCGGTAGCGGCGGAAGTGGCAGCGGTGCA-3′; GS20: 5′-GGCAGCGGTAGCGGCGGAAGTGGCAGCGGTGGAAGCGGTAGTGGCGGCAGCGGCAGCGGTGCA-3′. The pBR N- plasmid which does not express N protein was generated by insertion of synthetic oligonucleotides cassettes, whose sense strand is 5′-CATGGCCTGACTGACTGACTGACGAATGCA-3′, into the NcoI and BsmI sites of the pBR plasmid described above.

Monitoring β-galactosidase expression by colony colour and solution assays

Competent E. coli N567 cells were prepared by standard CaCl2 methods and transformed with the appropriate plasmids by heat shock as described previously (Harada and Frankel, 1999). Each transformant was spread onto tryptone plates containing ampicillin (200 mg l−1), chloramphenicol (20 mg l−1), isopropyl β-D-thiogalactoside (IPTG) (0.05 mM) and Xgal (80 mg l−1). When plates were incubated at 37°C for 18–24 h and further at 24°C for 16–24 h, blue intensity of these colonies was monitored and represented in the number of plusses from 0 to 8+ (see Fig. 3 for representative colours). Solution assays using ONPG were carried out as described previously (Cocozaki et al., 2008) except for differences in antibiotic concentration for overnight culturing (100 mg l−1 ampicillin and 20 mg l−1 chloramphenicol).

Comparison of the copy number of reporter plasmids

Escherichia coli N567 cells transformed with reporter pAC plasmids were spread onto tryptone plates containing chloramphenicol (20 mg l−1) and tetracycline (6 mg l−1) as antibiotics and incubated at 37°C overnight. Three independent colonies were picked and cultured in tryptone medium containing both antibiotics with aeration at 37°C overnight. DNA was prepared from each overnight culture (1 ml) using QIAprep® Spin Miniprep Kit (Qiagen). Each DNA was eluted from the QIAprep spin column with 150 μl of Buffer EB (10 mM Tris-Cl, pH 8.5).

RNA probes for North Western analysis

RNAs (boxB and RRE) for the preparation of biotinylated probes were transcribed in vitro using T7 RNA polymerase. The RNAs were transcribed at 37°C for 2 h in the reaction mixture containing 80 mM HEPES-KOH (pH 8.1), 5 mM DTT, 1 mM spermidine, 0.001% Triton X-100, 80 mg ml−1 polyethylene glycol-8000, 42 mM MgCl2, 8 mM GTP, 8 mM UTP, 4 mM ATP, 4 mM CTP, 0.4 U μl−1 RNasin® Plus RNase Inhibitor (Promega), 1.6 U μl−1 T7 RNA polymerase (Promega) and 500 nM synthetic DNA template (boxB, 5′-GGCCCTTTTTCAGGGCCTTCCTATAGTGAGTCGTATTAC-3′; RRE, 5′-GGCCTGTACCGTCAGCTTGCGCTGCGCCCAGACCTATAGTGAGTCGTATTAC-3′) annealed with T7 promoter sequence (5′-GTAATACGACTCACTATA-3′). RNAs were treated with Turbo™ DNase (0.02 U μl−1) (Ambion) at 37°C for 15 min, purified on 12% polyacrylamide/8 M urea gels, eluted from the gels in 0.4 M sodium acetate (pH 5.4), isopropyl alcohol precipitated, and washed with 70% ethanol. Purified RNAs were labelled with biotin at their 5′-ends using 5′-EndTag™ Nucleic Acid Labeling System (Vector Laboratories) and treatment with biotin maleimide (Vector Laboratories). The RNAs were dissolved to 10 μM in renaturing buffer containing 200 mM Tris-HCl, pH 7.5 and 1 M NaCl, then heated at 70°C for 5 min and slowly cooled to room temperature for annealing.

Western blotting and North Western blotting

Escherichia coli N567 cells transformed with pBR plasmids were spread onto tryptone plates containing ampicillin (100 mg l−1) and tetracycline (6 mg l−1) as antibiotics and incubated at 37°C overnight. Generated colonies were inoculated to tryptone medium containing both antibiotics and incubated with aeration at 37°C overnight. The overnight cultures were diluted 1:50 in A medium (4 or 5 ml) containing 10.5 g l−1 K2HPO4, 4.5 g l−1 KH2PO4, 1.0 g l−1 (NH4)2SO4 and 0.5 g l−1 trisodium citrate dihydrate supplemented with 0.4% (w/v) glucose, 1 μg ml−1 vitamin B1, 1 mM MgSO4 and both antibiotics, and incubated with aeration for 6 h at 37°C. Then, each culture was divided into two aliquots and further incubated with or without addition of 0.5 mM IPTG for 30 min. The cells were harvested from each culture (1 or 2 ml) and washed with ice-cold acetone (1 ml). Each pellet was resuspended in 100 or 150 μl of 1× sodium dodecyl sulphate (SDS) sample buffer containing 62.5 mM Tris-HCl (pH 6.8), 2% (w/v) SDS, 0.002% (w/v) bromophenol blue, 10% (v/v) glycerol and 5% (v/v) 2-mercaptoethanol, and heated at 100°C for 5 min. These solutions were separated by 15% SDS-PAGE and then transferred to polyvinylidene difluoride membrane (Millipore) in transfer buffer (25 mM Tris, 192 mM glycine, 20% (v/v) methanol). The amounts of the solutions loaded on the gels are otherwise noted. For Western blotting, the membrane was blocked with 2% (w/v) Membrane Blocking Agent (GE Healthcare) in TBST (20 mM Tris-HCl, pH 7.5; 154 mM NaCl; 0.05% Tween® 20). The blocked membrane was treated with antisera against N protein (a gift from Franklin, N. C.) and ECL Rabbit IgG, HRP (horse radish peroxidase)-Linked Whole Ab (from donkey) (GE Healthcare), and visualized with Amersham ECL Plus Western blotting Detection Reagents (GE Healthcare). Chemiluminescent signals were detected with Cooled CCD Camera System Light-Capture II (ATTO). For North Western blotting, the membranes after transferring were blocked in wash buffer (10 mM Tris-HCl, pH 7.5; 100 mM NaCl; 1 mM EDTA, pH 8.0; 0.01% (v/v) NP-40) containing 2% (w/v) Membrane Blocking Agent (GE Healthcare) and 5 mg ml−1 ribonucleic acid from torula yeast type VI (Sigma) for 1–2 h at room temperature. The blocked membrane was washed by several changes of wash buffer. Then, the membrane was incubated with 50 nM biotinylated RNA probes in wash buffer for 1 h at 4°C. After several washings, the membrane was incubated in streptavidin-HRP (Invitrogen) diluted to 1:5000 with wash buffer for 1–5 h at room temperature or 4°C, washed by several changes of wash buffer, visualized and detected as for Western blotting.

Acknowledgements

The authors thank Naomi Franklin for the kind gift of N antisera, Akihiro Oguro and Koichi Ito for advice on experimental techniques, and Colin Smith for comments on the manuscript. This work was supported by a Grant-in-Aid for Scientific Research in Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (K.H. and S.M.).

Ancillary