Site‐specific cleavage of mRNA has been identified in vivo for the polycistronic colicin E7 operon (ColE7), which occurs between G and A nucleotides located at the Asp52 codon (GAT) of the immunity gene (ceiE7). In vitro, this specific cleavage occurs only in the presence of the ceiE7 gene product (ImmE7). The crystal structure of dimeric ImmE7 has been determined at 1.8 Å resolution by X‐ray crystallographic analysis. We found that several residues located at the interface of dimeric ImmE7 bear surprising resemblance to the active sites of some RNases. These results suggest that dimeric ImmE7 may possess a novel RNase activity that cleaves its own mRNA at a specific site and thus autoregulates translational expression of the downstream celE7 gene as well as degradation of the upstream ceaE7 mRNA.
The plasmid‐encoded E group colicins bind to the vitamin B12 receptor (btuB gene product of Escherichia coli; DiMasi et al., 1973) and have been subdivided by immunity tests into colicins E1–E9 (Watson et al., 1981; Cooper and James, 1984). The bactericidal activities of some E group colicins are well documented: colicin E1 exhibits an ionophore effect on the membrane of target cells (Tokuda and Konisky, 1979); colicin E3 is an RNase (Bowman et al., 1971); colicins E2 and E7–E9 are DNases (Schaller and Nomura, 1976; Toba et al., 1988; Chak et al., 1991).
Three genes, cea, cei and cel, located on the ColE operon are involved in colicin production, protecting colicin‐producing cells against the action of their own colicin and release of the colicin–immunity protein complex respectively. Overexpression of the ColE operon is controlled by an SOS response promoter (Herschman and Helinski, 1967; Little and Mount, 1982). Lysis proteins (the cel gene product) among E group colicins may activate cell envelope phospholipase A and cause extensive damage to the cellular membrane (Pugsley and Schwartz, 1984). Thereby, overproduction of the cel gene can cause cell lysis. This ‘suicidal phenomenon’ arising from dramatic overexpression of the ColE operon has been compared with ‘the programed cell death of bacteria’ (Jensen and Gerdes, 1995; Yarmolinsky, 1995). To stabilize this ‘killer system’, coordinative expression of the ColE operon must be tightly controlled and efficient.
ImmE7 protein can bind to the endonuclease domain of ColE7 and neutralize its toxicity. The interaction between colicin and its cognate immunity protein is not only specific but is also the tightest among all known systems (Wallis et al., 1995a, 1995b). Recently, the three‐dimensional crystal structure of monomeric ImmE7 protein has been determined (Chak et al., 1996). It was suggested that the acidic protruding areas of ImmE7 may be involved in interacting with the basic T2A domain of colicin. From the present study, a specific endonucleolytic cleavage of the transcript has been found in the coding region of ceiE7. Using an in vitro transcription system, we observed that this specific endonucleolytic activity originated from ImmE7 itself. This observation was further supported by crystallographic analysis showing that the interface of dimeric ImmE7 presented possible RNase active sites similar to those of mammalian RNase A (Campbell and Petsko, 1987; Wlodawer et al., 1988) and some microbial RNases, including RNase Rh (Kurihara et al., 1992), barnase (Mauguen et al., 1982), RNase T1 (Koepke et al., 1989), RNase Sa (Sevcik et al., 1991), RNase F1 (Vassylyev et al., 1993) and RNase Ms (Nonaka et al., 1993).
This paper reports the novel ribonucleolytic activty of ImmE7 protein and its role in regulating expression of the ColE7 operon. The biological significance of the ColE7 operon coding a multifunctional ImmE7 protein is discussed.
Measurement of the transcriptional strength across the junctions of the cea–cei–cel polycistronic transcript
Oligonucleotide primer design and the strategy for the reverse transcriptase (retrotherm reverse transcriptase) polymerase chain reaction (RT–PCR) are shown in Figure 1A. The oligonucleotide primer P4 is the common forward primer and primers P5, P1 and P2 are reverse primers. The PCR product (378 bp) amplified by oligonucleotide primers P4 and P5 represents the strength of transcriptional read‐through from the ceaE7 gene to the 5′‐end of ceiE7; the 566 bp fragment indicates the strength of transcriptional read‐through from ceaE7 to the 3′‐end of ceiE7. Similarly, the 757 bp PCR product shows the expression level from ceaE7 to celE7.
The results of PCR amplification of the DNA template from the ColE7 plasmid are shown in Figure 1B (lane 2). This positive control experiment indicated that the amplification intensity of the 757, 566 and 378 bp fragments were derived from the same copy number of DNA templates. Using this RT–PCR method to amplify the mRNA extracted from cells containing the ColE7 operon at different time intervals, we found that the intensities of these three fragments at the 30 min time point were similar (Figure 1B, lanes 3–5). Comparing the relative intensities of the RT–PCR products (Figure 1B, time intervals 30–90 min), it was found that the copy number of the target mRNA amplified by oligonucleotide primers P4 and P5 was far more than that of mRNA amplified by the other two sets of oligonucleotide primers. In contrast to the increase in promoter activity during induction (oligonucleotide primer PB as extension primer; data not shown), the quantities of mRNA produced from the RT–PCR products gradually decreased. Moreover, production for these three fragments appeared to be different. It was noticed that the amounts of the 566 and 757 bp fragments decreased faster than that of the 378 bp fragment. These unexpected patterns of amplification indicated that some unknown factor(s) may be involved in the regulation of copy number of mRNAs during induction. This phenomenon may most probably be due to post‐transcriptional processing or decay of the polycistronic ColE7 mRNA.
Localization of site‐specific cleavage of the ceiE7 transcript
To investigate the cause of the differential degradation rates of mRNA during induction, we performed primer extension with P1 as the extension oligonucleotide primer and cei–cel transcript as the template. Surprisingly, a 5′‐end truncated transcript which terminated at a specific site between the G and A nucleotides of the Asp52 codon of ceiE7 (Figure 2) was detected. This result was further confirmed by using oligonucleotide primer P2 for primer extension (data not shown). Neither primer P1 nor P2 could be used to detect the transcription initiation site for the cei gene (Figure 2). However, two transcription initiation sites (Soong et al., 1994) for the cei gene were detected when primer P5 was used. It seems most likely that breakage of the transcript at the Asp52 codon of the ceiE7 transcript is indeed a specific cleavage site created after transcription. This specific cleavage was found under both non‐induced and induced conditions (Figure 2). Furthermore, the accumulation of a larger amount of truncated cei–cel transcript coincided with an increase in transcriptional initiation at the SOS response promoter of the ColE7 operon (Soong et al., 1994). Because of this specific cleavage, intact cea–cei–cel transcripts should be rare in vivo and can only be detected by a more sensitive RT–PCR amplification method (Figure 1, the 757 bp fragment).
Site‐specific cleavage of the ceiE7 transcript is mediated by ImmE7 protein
RNase E and RNase III are the two most common endonucleases for degradation of RNA (for reviews see Belasco and Higgins, 1988; Higgins et al., 1992). However, the ceiE7 mRNA extracted from RNase‐deficient mutants (Escherichia coli N3433, N3431, RS6521 and SW001) carrying the ColE7 operon could not prevent the transcripts from being cleaved at the specific cleavage site (data not shown).
In vitro transcription was employed to study whether the specific cleavage occurred spontaneously or conditionally. When under the control of the SP6 promoter (Melton et al., 1984), a full‐length ceiE7 run‐off transcript (528 nt) initiated at the SP6 promoter was obtained (Figure 3). Specific cleavage was not observed under these in vitro conditions (Figure 3, lane 1). Specific cleavage was reproduced only if either purified ImmE7 or a crude extract from E.coli harboring a ceiE7 gene was present in the assay system (Figure 3, lanes 2, 3, 6 and 7). These experimental results thus imply that the ImmE7 protein is required for this specific cleavage. Probably, ImmE7 exhibits a novel RNase activity to cleave its own transcripts at a specific site.
Crystal structure of dimeric ImmE7
The ImmE7 protein maintains an equilibrium between monomeric and dimeric forms that can be detected by chromatography (data not shown). This protein crystallizes in two distinct forms with the vapor diffusion method (Ku et al., 1995). The monomeric ImmE7 structure was solved by the multiple isomorphous replacement method using orthorhombic crystals (Chak et al., 1996). Dimeric ImmE7 crystallizes in monoclinic space group P21 with cell dimensions a = 28.94 Å, b = 101.67 Å, c = 52.62 Å and β = 92.6°, indicating that four ImmE7 molecules are arranged in an asymmetric unit. The Matthew's coefficient (Matthews, 1968) would be 2.04 Å3/Da and the estimated solvent content is 40%. The structure was solved by the molecular replacement method using the monomeric ImmE7 structure (Chak et al., 1996) as searching model. Details of structure determination are described in Materials and methods.
The final model contains two dimeric pairs (M1/M3 and M2/M4) of ImmE7 with 2792 non‐hydrogen atoms and 127 water molecules. The structure of the four individual molecules, M1, M2, M3 and M4, in the asymmetric unit contains all of the 87 residues, which are numbered 101–187, 201–287, 301–387 and 401–487 respectively. Details of crystallographic statistics are listed in Table I. The Ramachandran plot (Ramakrishnan and Ramachandran, 1965) shows that all but three residues have φ, ψ angles out of the allowed range. The geometry was further checked by the program PROCHECK (Laskowski et al., 1993), in which 303 non‐glycine residues (93.5%) are in the most favored area and 21 (6.5%) in allowed regions.
The backbone secondary structure of the dimer as assigned by PROCHECK is found to be identical to monomeric ImmE7 (Chak et al., 1996) in that they all consist of four anti‐parallel α‐helices: Helix 1 (12–25), Helix 2 (32–45), Helix 3 (52–55) and Helix 4 (65–78). In the monomeric structure, the N‐terminal strand is parallel to Helix 2, however, in the dimeric structure, it is almost perpendicular to Helix 2. The ‘open’ conformation of the N‐terminal strand helps to interlock the counter subunit in the dimeric structure (see Figure 4). The major interactions between the two ImmE7 molecules in a dimer involve Helix 2. The two helices propagate in opposite directions, with the nearest backbone atoms separated by 6.5 Å and an angle between the two Helices 2 of ∼75°. Three residues, Val33, Val36 and Leu37, from each monomer are involved in hydrophobic contacts. These side chains form a hydrophobic core structure at the dimer interface. Surface areas of individual ImmE7 monomers range from 5538 to 5711 Å2. The buried solvent‐accessible surfaces for each molecule are 882 Å2 for the dimer M1/M3 and 859 Å2 for the dimer M2/M4 (calculated by Areamol in CCP4). A stable dimer of comparable molecular weight covers ∼750 Å2 (Janin et al., 1988), therefore, this dimeric ImmE7 structure should be stable in solution. In addition to the central hydrophobic core, residues Leu34, Leu38 and/or Val42 form a hydrophobic patch which may be responsible for docking the N‐terminal strand of the counter monomer. Several acidic amino acid residues, Asp32, Asp35 and Glu39, located in the protruding area of Helix 1, Helix 2 and Variable Loop 2 of monomeric ImmE7 (Chak et al., 1996) were predicted to be involved in interaction with ColE7. In contrast, these residues are buried in dimeric ImmE7, indicating that interaction of dimeric ImmE7 with colicin seems impossible. This result suggests that dimeric ImmE7 could have some distinct biological functions other than inhibition of its cognate colicin.
The crystal structure of dimeric ImmE7 presents possible RNase active sites
The structures of RNase A and its complexes with various substrate analogs have established a mechanism for the hydrolysis reaction (Wodak et al., 1977; Nachman et al., 1990; Zegers et al., 1994). Three residues in the active site, His12, Lys41 and His119, of RNase A participate in the hydrolysis reaction. His12 and His119 are located on either side of the cleaved phosphate group and the amino group (NZ) of Lys41 lies in the vicinity of the ribose phosphate portion of the inhibitor. His12 behaves as a general base and substracts a proton from the 2′‐OH group, whereas His119 acts as a general acid and furnishes a proton to the 5′‐oxygen as it is being displaced from the phosphorus by nucleophilic attack of the 2′‐oxygen atom. Lys41 forms a salt bridge with the negatively charged oxygen bound to the phosphorus and stabilizes the transition state.
By examining the structure of dimeric ImmE7, we found a pair of histidine residues (His40) which are located in the interface of the two subunits (Figure 4). His140 and His340 refer to the same His40 residue from different subunits (M1 and M3 respectively). The distance between the α‐carbons of the two histidine residues, His140 and His340, is ∼7 Å, compared with 9 Å for RNase A (His12 and His119). A lysine residue (Lys343) which is located close to the histidine residues is also found in ImmE7. Superimposition of the three residues (His12, His119 and Lys41 in RNase A and His340, His140 and Lys343 in ImmE7) by least squares fitting of the corresponding atoms gave an average r.m.s. deviation of only 1.20 Å. As shown in Figure 5A, when the active site of RNase A is superimposed on the proposed one of ImmE7, the imidazole rings of the histidines in the two enzymes can be overlapped in the same region by rotating the dihedral angle (Cα–Cβ) of His140 by 120° in the ImmE7 structure. This rotation is allowed because it would not cause additional unfavorable Van der Waal's contacts. It has been shown that the general acid His119 in RNase A exhibits conformational mobility. By rotating about its Cα–Cβ bond, the imidazole side chain can occupy two distinct conformations (Zegers et al., 1994). It is possible that the conformation of His40 in ImmE7 may change after binding to RNA.
The striking resemblance between RNase A and ImmE7 is not only found in the proteins, but also in their water structures. In the structure of phosphate‐free RNase A (Wlodawer et al., 1988), two water molecules were found in the phosphate binding pocket. When compared with the structure of substrate‐bound RNase A, one of the water molecules overlapped with the phosphate oxygen atom (near position O2). In dimeric ImmE7, there are also two water molecules located between the two histidine residues (see Figure 5A), indicating that His140 and His340 in dimeric ImmE7 could act as a general acid and base for RNase activity.
In addition to mammalian RNases, the structure of one enzyme, RNase Rh, in the larger microbial family has also been solved. The active site of RNase Rh (Kurihara et al., 1992) is similar to that of RNase A. Furthermore, study of a smaller family of microbial RNases suggested that a histidine and a glutamate are the general acid and base. These two residues in RNase Ms (Nonaka et al., 1993) nearly coincide with the His12 and His119 residues in RNase A. Despite the differences in their primary amino acid sequences and tertiary structural folding, there are similarities between mammalian RNase A and microbial RNases in the arrangement of the catalytic residues (Nonaka et al., 1993). These enzymes may have similar chemical mechanisms for phosphodiester hydrolysis.
In the vicinity of the active pocket we also find a nearby glutamate (Glu39) that could act as a general base, as observed in the smaller family of microbial RNases. It seems possible that the general acid and base of ImmE7 are a histidine and a glutamate from either monomer. As shown in Figure 5B, the residues His240, Glu439 and Lys243 of dimer M2/M4 overlap nicely with the corresponding residues of barnase. With distinct ImmE7 dimer pairs (Figure 5A and B) for comparison, it was demonstrated that the structures of the active sites in dimers M1/M3 and M2/M4 are identical. The average r.m.s. deviation between the corresponding atoms of the residues involved in the active site in barnase and ImmE7 is 1.43 Å.
The molecular surface of ImmE7 calculated using GRASP (Nicholls and Honig, 1991) shows a groove at the interface of dimeric ImmE7 between two adjacent Helices 2 (Figure 6A). A deep hole in the middle of the groove could be the binding site for the phosphate group and the two histidines are located exactly above and beneath the hole. Figure 6B shows a close‐up of the proposed RNase active site in dimeric ImmE7, with the three residues His140, His340 and Lys343 located on the molecular surface, reachable by mRNA. There are several exposed basic residues located along the groove (Lys43, Lys81 and Arg76) (Figure 6A). Thus, this long groove in the dimeric interface could be the binding site for ribonucleic acids.
Site‐specific cleavage of ceiE7 mRNA can regulate translational expression of the downstream celE7 transcript
celE7 is the third gene of the ColE7 operon and is promoterless (Chak and James, 1985; Chak et al., 1991). The regulation of expression of celE7 is still unclear. It has been noted that not only is the translational stop codon of ceiE7 located within the T1 stem–loop (Figure 7A), but also that the ribosome binding site of celE7 is situated near the 3′‐end of the T1 stem–loop structure (Figure 7A).Therefore, we speculate that cleavage of ceiE7 mRNA may play an important role in the regulation of translational expression of celE7 mRNA.
In order to investigate the regulation of celE7 expression, three translational fusion mutants were generated. The recombinant plasmids, pHK001‐13R, pHK001‐14R and pSE4R, containing lacZ promoter–ceiE7–celE7 fusions (Figure 7B), were constructed. pHK14R was an out‐of‐frame fusion mutant that created a new translational stop codon, UAG, 88 nt upstream of the original stop codon (Figure 7A). pSE4R produced a translational stop site (UGA) located 29 nt downstream of the ceiE7 stop codon. With isopropyl β‐d‐thiogalactoside (IPTG) induction, cells containing either pHK001‐13R (in‐frame fusion; Figure 7A) or pSE4R started to lyse 30 min after induction (Figure 7B), whereas cell lysis was not observed in cells carrying pHK001‐14R upon IPTG induction. These results suggest that ribosome pausing at the translational stop codon of the ceiE7 cistron may hinder formation of the T1 hairpin structure in such a way that binding of the ribosome to the Shine–Dalgano sequence of celE7 is not interfered with and as a result translation of celE7 is initiated (Figure 7B). In contrast, as in the case of pHK001‐14R (Figure 7B), ribosome pausing occurs 88 nt upstream of the original termination codon and thus formation of the stem–loop structure in the T1 region is favored. As a result, this hairpin structure may block binding of the ribosome to the Shine–Dalgano sequence of celE7 and thus translation of the celE7 gene is impossible. Obviously, cleavage of the ceiE7 transcript may have a similar effect in favoring formation of the T1 stem–loop structure. Consequently, translational coupling of ceiE7 and celE7 is disrupted.
Both in vivo and in vitro evidence support the idea that ImmE7 may autoregulate the cleavage of its own mRNA. The crystal structure of dimeric ImmE7 shows that the possible RNase active site is located at the dimer interface. The cleavage event may induce decay of upstream mRNA and at the same time turn off translational expression of downstream celE7. The sequence specificity of this putative enzyme is unclear, but the specific cleavage site of ceiE7 mRNA has been identified and located in the proposed secondary structure (Figure 7A). Furthermore, ImmE7 protein is a multifunctional protein. It is well documented that this protein protects colicin‐producing cells by inactivating colicin toxicity. The monomeric crystal structure of ImmE7 suggests that the highly negatively charged Helix‐1–Loop–Helix‐2 region of the protein may bind to the T2A domain of colicin (Ku et al., 1995; Chak et al., 1996) and as a result the DNase domain of colicin is inactivated. Translocation of colicins is coupled with that of its cognate immunity protein, suggesting that Imm protein may act as a chaperone‐like substance to assist exportation of colicins through the membrane. Interestingly, cells containing the ceiE7 gene are more resistant to low temperatures than host cells containing no cei gene (data not shown), suggesting that the functional targets of ImmE7 may be far broader than we previously expected.
Bovine seminal RNase, which belongs to the same superfamily as RNase A, is the only dimeric RNase isolated thus far (Piccoli et al., 1992). Two disulfides reinforce the non‐covalent links between the two subunits. In the crystal structure of naturally dimerized seminal RNase, the two subunits swap their N‐terminal segments and the composite active sites are made by residues from the two different subunits. The nature of these shared active sites is similar to that proposed for dimeric ImmE7. Rop protein, encoded by the ColE1 plasmid of E.coli, is another example of a dimeric RNA‐binding protein (Predki et al., 1995). It binds hairpin pairs formed between RNA II and RNA I and regulates plasmid copy number. X‐ray crystallographic studies (Banner et al., 1987) have shown that dimeric Rop forms a four helix bundle structure and a recent genetic analysis demonstrated that RNA binding determinants reside on one face of the protein formed by two helices (1 and 1′), one from each monomer. The residues form a thin stripe down the center of this face and it is suggested that the protein and RNA interact in a ‘parallel’ fashion. The central core of dimeric ImmE7 is similar to a four helix bundle structure and we suspect that RNA binds between the two Helices 2. The predicted secondary structure of the mRNA around the cleavage site is a hairpin structure. It is also possible that ImmE7 specifically recognizes this kind of RNA structure, similarly to Rop, the binding of which has been shown to be structure specific.
The proposed autoregulatory pathway of expression of the ColE7 operon controlled by ImmE7 is shown in Figure 8. The central theme of this autoregulation is the dynamic equilibrium between the monomeric and dimeric form of ImmE7. We propose that the function of monomeric ImmE7 is to inactivate colicin by binding to the T2A domain. In addition, dimeric ImmE7 exhibits a novel RNase activity and can be expected to regulate translational expression of the cel as well as the cea gene by specific cleavage of it own transcripts. ceiE7 contains two constitutive promoters (Soong et al., 1994). Therefore, we hypothesize that expression of the ceiE7 gene is independent of the SOS response promoter of the ColE7 operon. This speculation has been confirmed by Western blot analysis, which demonstrated that ImmE7 is at least 20 times in excess of colicin E7 (our unpublished data). Therefore, the amount of housekeeping ImmE7 protein is more than enough to protect cells from the toxic effects of either exogenous or endogenous colicin produced by basal level expression of the operon. If the amount of ColE7 polycistronic transcript increases substantially by overexpression from the inducible promoter, then a new state of dynamic equilibrium will be created, whereby more dimeric ImmE7 from the ImmE7 pool will be formed to cleave its own transcript. As a result, autoregulation of translational expression of the ColE7 operon will be achieved.
Delayed expression of the cel gene in response to induction of the SOS promoter (Lu and Chak, 1996) has been observed. This phenomenon can be explained by the presently proposed autoregulatory pathway. Excessive amounts of housekeeping dimeric ImmE7 may cleave the cea–cei–cel polycistronic transcripts at a specific site located in the cei gene, leading to uncoupling of translational expression (Figure 7B) of the cel gene. However, if induction of the SOS response promoter is sustained for 100 min, as in the case of the ColE7 operon (Lu and Chak, 1996), the amount of housekeeping dimeric ImmE7 is no longer sufficient to cope with the vastly increased numbers of cea–cei–cel polycistronic transcripts and, in this situation, overexpression of the cel gene would cause cell lysis.
It is known that translation influences the accessibility of endonuclease cleavage sites (Belasco and Higgins, 1988; Higgins et al., 1992). Pauses in polypeptide chain elongation have been reported to result from unusual codon usage for colicins (Varenne et al., 1982; Morlon et al., 1983). We have also observed that many non‐optimal codons are used in ceiE7 mRNA. This observation indicates that translational pausing may occur during translation of cei mRNA, thereby the formation of a specific mRNA secondary structure due to this translational pause may be susceptible to ribonucleolytic attack by dimeric ImmE7.
Oligonucleotide‐directed site‐specific mutagenesis of the cei gene has been carried out to investigate the actual binding site between ImmE7 and ColE7 protein, the interaction face for dimeric ImmE7 and the actual location of the putative RNase active site of ImmE7. We have also designed a set of experiments to locate the mRNA binding site of dimeric ImmE7. Moreover, co‐crystallization of ImmE7 with ColE7 or RNA analogs is under investigation.
Materials and methods
Bacterial strains, plasmids and growth conditions
Escherichia coli W3110 (ColE7‐K317) from the collection of Dr R.James (Chak et al., 1991) and E.coli W3110 and JM109 (Yanisch‐Perron et al., 1985) were the hosts of the plasmids used in this study. Escherichia coli K12 isogenic strains N3433 (lacZ43, relA, spoT1, thi1), N3431 (rne‐3071ts), RS6521 (rnc::Tn10/Tetr rnc−) and SW001 (rnc::Tn10/ts (rne‐3071) Tetr rnc−) were kindly provided by Dr Sue Lin‐Chao. Plasmids pHK001‐13R and pHK001‐14R were constructed from derivatives of ExoIII‐digested pHK001 (Soong et al., 1992). pHK001‐13R contains the DNA fragment of cei–cel from nt −110 to the EcoRI site of pColE7 (the first nucleotide of the ceiE7 start codon is defined as +1), while pHK001‐14R and pSE4R include the sequence from +141 to the EcoRI site and +252 to the EcoRI site respectively. These three DNA fragments were cloned into pUC19. pYAN13R was as pHK001‐13R except that the cloning vector was pGEM3Z(f). Bacteria were cultured in L broth. For selection or induction, ampicillin, IPTG and mitomycin C (MMC) were added to final concentrations of 100, 20 and 0.5 μg/ml respectively.
DNA and RNA techniques
Restriction enzymes, SP6 RNA polymerase, RNase‐free DNase I, T4 polynucleotide kinase, M‐MLV and retrotherm reverse transcriptases, DNA sequenase and DNA ligase were purchased from BRL, Promega, Boehringer Mannhein or Epicentre Co. and were used according to the manufacturer's instructions. Plasmid isolation, restriction, ligation and transformation were carried out as described by Sambrook et al. (1989).
Isolation of total RNA, primer extension and RT–PCR
Total RNA was extracted from lysed cells by the hot phenol method as described by von Gabain et al. (1983). For induction, bacteria were grown to an OD600 of 0.4 in L broth before addition of MMC (0.5 μg/ml) or IPTG (20 μg/ml). Samples were withdrawn at different time points and total RNA extracted and analyzed by primer extension (Sambrook et al., 1989) or used as templates for RT–PCR amplification.
Oligonucleotide primers used in the experiment were as follows: P1, 5′‐CTGGCTTACCGTTAGCAGC; P2, 5′‐GACGACGGTGATACTGTCC; P4, 5′‐GACTTAGGTTCTCCTGTTCC; P5, 5′‐CTAACACATCATCAGT; PB, 5′‐CCACCTGTGTTATGTGCGCC.
RT–PCR was performed at 94°C for 1 mim, 40°C for 5 min and 70°C for 10 mim for the first cycle and at 94°C for 1 min, 45°C for 75 s and 70°C for 2 min for another 30 cycles.
Preparation of crude extracts and purification of ColE7 and ImmE7
Assay of RNase activity and in vitro transcription
RNase activity assays were performed as described by Mackie (1991). In vitro transcription of ceiE7 was carried out using SP6 RNA polymerase with a linear DNA template containing the cei gene. The detailed procedure was as described previously by Melton et al. (1984).
X‐ray data collection and processing
Single monoclinic crystals were grown at room temperature by the sitting‐drop vapor diffusion method from ammonium phosphate solution as described earlier (Ku et al., 1995). X‐ray diffraction data were collected using an R‐AXIS II image plate detector equipped with a Rigaku RU‐300 generator operating at 50 kV and 80 mA. The data collected on a single monoclinic crystal at room temperature were processed using RAXIS software. Detailed statistics for the native data are listed in Table I.
The monoclinic crystal form of ImmE7 belongs to space group P21. Calculations of self‐rotation maps at different resolution ranges using XPLOR (Brunger, 1992) yielded prominent peaks for two non‐crystallographic dyads in the directions of the a‐ and c‐axes, corresponding to spherical angles of ψ = 90° and φ = 0 and 90° respectively. The refined structure from the orthorhombic I222 crystal was employed as a search model in a cross‐rotational search. Several different resolution limits, ranging from 5 to 2.5 Å, were used in the rotation function calculations. The rotation function searches yielded 150–200 highest peaks, which were then subjected to filtering by PC refinement procedures in XPLOR. No straightforward solution was obtained from a single search, but among the peak positions with the highest PC values, four sets of angles were extracted by comparing results from different resolution ranges. These angles persistently appeared at the top of the lists and represented four possible orientations of ImmE7 molecules in the monoclinic unit cell. They are (θ1, θ2, θ3) = (65.5°, 66.5°, 97.5°), (66.5°, 63.0°, 273.5°), (228.0°, 73.0°, 159.5°) and (238.5°, 77.0°, 333.5°). We noticed that the first orientation (M1) is related to the second (M2) and the third (M3) is related to the fourth (M4) by a non‐crystallographic 2‐fold symmetry that is almost parallel to the c‐axis.
The translational function search was divided into two parts: (i) a search in the a–c plane for each of the four ImmE7 molecules; (ii) a search along the b‐axis for relative positions between the molecules. To carry out (i), different resolution ranges were tested to produce a clearly interpretable map for each molecule. Four major peaks at (x, z) = (0.188, 0.094) for M1, (0.281, 0.094) for M2, (0.483, 0.047) for M3 and (0.031, 0.063) for M4 were found. For the translational search (ii), diffraction data between 12 and 3 Å resolution with F/σ > 3 were used in the calculation of intermolecular translation functions in which M1 was fixed at the origin while M2, M3 or M4 was moved along the b–axis. A unique peak was observed for M2 at y = 0.789, with the addition of the half‐lattice vector (0, 0, 1/2); for M3 it was at y = 0.031 with vector (1/2, 0, 1/2); for M4, y = 0.750 with vector (1/2, 0, 0). Highly consistent results were obtained with one of the other three molecules fixed and these reconfirmed the disposition of ImmE7 molecules in the monoclinic unit cell.
The N‐terminus of M1 was in poor contact with Helix 2 of M3 near residue 36 in the starting model. The N‐terminus of M3 was also in contact with Helix 2 of M1, again near residue 36. Similar relationships were observed between M2 and M4. In spite of the N‐terminus and Helix 2 being in poor contact, no other parts of the protein molecules interacted unfavorably with its neighbors. Rigid‐body refinement of the four molecules against 19 710 reflections between 20 and 2 Å resolution with F/σ > 2 yielded an R value of 0.42. The N‐terminal residue Leu3 was displaced by 5 Å, because there were densities corresponding to this new position, and relieved the poor contact with Helix 2 of its neighbor. There were also densities in the adjacent regions for possible extension towards the two terminal residues Met1 and Glu2, which were not observed in the orthorhombic crystal. The conformations of the N‐terminal strands were similar but not identical in all four cases. The final model consists of four polypeptide chains, in which the residues were numbered 101–187, 201–287, 301–387 and 401–487, and 127 water molecules. With 8% of data reserved for Rfree, the R factor is 18.4% and Rfree is 26.8 %, using a total of 24 783 reflections with F > 2 σ(F) in the 6.0–1.8 Å resolution shell.
This work was support by the Academia Sinica and National Science Council (NSC) of the Republic of China with grants to K.‐F.C. (NSC85‐2331‐B010‐105 and NSC86‐2314‐B010‐011) and H.S.Y. (NSC86‐2313‐B001‐005).
↵† S.‐Y.Hsieh and T.‐P.Ko contributed equally to this work
- Copyright © 1997 European Molecular Biology Organization