Retention of Core Catalytic Functions by a Conserved Minimal Ribonuclease E Peptide That Lacks the Domain Required for Tetramer Formation*

Ribonuclease E (RNase E) is a multifunctional endoribonuclease that has been evolutionarily conserved in both Gram-positive and Gram-negative bacteria. X-ray crystallography and biochemical studies have concluded that the Escherichia coli RNase E protein functions as a homotetramer formed by Zn linkage of dimers within a region extending from amino acid residues 416 through 529 of the 116-kDa protein. Using fragments of RNase E proteins from E. coli and Haemophilus influenzae, we show here that RNase E derivatives that are as short as 395 amino acid residues and that lack the Zn-link region shown previously to be essential for tetramer formation (i.e. amino acid residues 400–415) are catalytically active enzymes that retain the 5′ to 3′ scanning ability and cleavage site specificity characteristic of full-length RNase E and that also confer colony forming ability on rne null mutant bacteria. Further truncation leads to loss of these properties. Our results, which identify a minimal catalytically active RNase E sequence, indicate that contrary to current models, a tetrameric quaternary structure is not required for RNase E to carry out its core enzymatic functions.

In Escherichia coli, endoribonuclease E (RNase E) carries out a wide variety of functions, including the processing of 9S ribosomal RNA (1,2), the degradation of bulk RNA (3)(4)(5)(6)(7), the degradation or processing of a wide variety of messenger and structural RNAs (8 -10) (for recent review, see Refs. 11 and 12), the control of plasmid DNA replication (13), and the removal of poly(A) tails from transcripts (14,15). The 118-kDa E. coli RNase E protein, which is encoded by the rne gene (16), contains three distinct regions: an amino-terminal domain that encodes the catalytic activity (17,18), a centrally located arginine-rich segment that has strong RNA binding activity (17,19), and a carboxyl-terminal region that serves as a scaffold for the binding of other proteins that are assembled into a ribonucleolytic complex known as the degradosome (20 -23). A closely related endoribonuclease, RNase G (previously known as the MreE or CafA protein) (9,24,25), consists almost entirely of sequences showing homology to the catalytic domain of RNase E (Fig. 1A).
The RNase E/G family of proteins has been classified into four subgroups according to the position of the highly conserved catalytic domain (26). Sequence analysis indicates that the genomes of E. coli and many other Proteobacteria encode both a type I RNase E/G enzyme, typically containing 900 -1100 amino acid residues and having the ϳ500-residue catalytic domain at the amino-terminal end, and a type IV family member (i.e. RNase G), consisting almost entirely of sequences conserved between the two enzymes in an ϳ500-amino acid residue protein. However, other bacterial genomes encode only one RNase E/G homolog (26,27). Although all members of the RNase E/G family that have been experimentally studied cleave their substrates in A ϩ U-rich regions, the substrate sites cleaved by RNase G and RNase E are not identical (9,25). Moreover, despite extensive homology between the catalytic domains of all members of the RNase E/G family, biochemical studies indicate that the RNase E of E. coli can degrade substrates quasi-processively in a 3Ј to 5Ј scanning mode, whereas RNase G from the same organism has a distributive mode of action (28).
The evolutionarily conserved region of members of the RNase E/G family of enzymes is ϳ500 amino acids in length, and a 498-residue amino-terminal RNase E fragment of the E. coli RNase E protein retains endoribonucleolytic activity in vitro (17). Bacteria expressing a shorter fragment consisting of residues 1-427 fused with the last 25 residues of the protein are reported to be viable, indicating that the conserved fragment is not required in its entirety for bacterial viability. However, these bacteria are less normal in mRNA degradation than a temperature-sensitive rne mutant at 37°C and are nonviable at 44°C (29). Recently, the crystal structure of the catalytic domain of RNase E was solved using a fragment consisting of amino acid residues 1-529 (30). It was shown that the 529-amino acid segment contains two domains that are joined by a Zn-link (amino acids 400 -415). The Zn-link is required for RNase E to form a homotetrameric quaternary structure, and the formation of such tetramers has been thought to be required for the proper functioning of the enzyme (30,31).
The investigations reported here were aimed at defining the minimal fragment of E. coli RNase E that mediates various known functions of this endoribonuclease and elucidating the structural components needed for these functions. Starting with the well studied 498-residue amino-terminal fragment of E. coli RNase E (N-Rne), 4 we constructed a series of rne gene deletions that truncate the protein from either end. Here we show that a peptide that extends 395 or 415 amino acids from the amino-terminal end of the RNase E proteins of E. coli and Haemophilus influenzae is (despite the lack of the Zn-link and the small homodimer-forming domain shown previously to be required for tetramer formation) sufficient for ribonuclease activity, cleavage site specificity, and 3Ј to 5Ј scanning of substrate in vitro, as well as for in vivo complementation of an rne null mutation. Our finding that such functions are mediated by these truncated proteins indicates that, contrary to current models, a tetrameric quaternary structure is not essential for the core enzymatic functions of RNase E.

EXPERIMENTAL PROCEDURES
Plasmids and Bacterial Strains-A plasmid encoding an E. coli N-Rne fragment consisting of amino acid residues 1-498 followed by a hexahistidine affinity tag and a thrombin cleavage site in a pET16b vector (Novagen) has been described (17,32). To construct a plasmid expressing a fragment of H. influenzae RNase E, the coding sequence for residues Ϫ16 to 476 5 was PCR-amplified from genomic DNA of strain KW20 (American Type Culture Collection 51907) and cloned into the inteinbased expression vector pTYB4 (New England Biolabs) by ligation in a manner that added a Pro-Gly dipeptide at the carboxyl terminus of the final expressed protein product. The coding sequence for H. influenzae RNase G protein was introduced into the pTYB4 vector. Coding sequences for shorter fragments of the E. coli and H. influenzae RNase E enzymes were amplified from these plasmids and cloned by ligation into the inteinbased expression plasmid pTYB1 (New England Biolabs) in a manner that left no extraneous residues on the carboxyl termini of the protein products.
For complementation experiments, gene fragments were PCR-amplified using the plasmids described above as templates and were then introduced into the NotI-and XbaI-cleaved Amp r pRNG3 parental plasmid (10) so that the expression of individual gene fragments is controlled by lacUV5 promoter. Specifically, fragments expressing E. coli RNase E residues 1 to 400 and 1 to 395, H. influenzae RNase E residues Ϫ16 to 476 and Ϫ16 to 399, E. coli RNase G residues 1 to 400, and H. influ-enzae RNase G residues 1 to 396 were cloned into the pRNG3 vector for complementation studies.
Protein Expression, Purification, and Crystallization-Expression plasmids were introduced by transformation into E. coli BL21(DE3), and bacteria were grown at 30°C to a cell density corresponding to A 600 ϳ 0.5. Protein expression was induced with 0.7 mM isopropyl-␤-D-thiogalactopyranoside (IPTG), and cells were incubated for an additional 18 h at 25°C. E. coli RNase E (residues 1-498) with an amino-terminal histidine tag was purified by Ni 2ϩ -nitrilotriacetic acid-agarose affinity column purification. Cells were suspended in lysis buffer (500 mM NaCl, 20 mM Tris-HCl, pH 7.9) supplemented with 5 mM imidazole and were disrupted by sonication. The cell lysate was clarified by centrifugation and applied to a Ni 2ϩnitrilotriacetic acid-agarose column previously charged and equilibrated with the same buffer. Supernatant was applied to the column, and the column was washed with ϳ10 column volumes of a buffer of 20 mM Tris-HCl, pH 7.9, 500 mM NaCl, and 60 mM imidazole. Protein was eluted with an elution buffer of 20 mM Tris-HCl, 500 mM NaCl, and 100 mM EDTA. The eluant was then dialyzed into a storage buffer of 20 mM Tris-HCl, pH 7.9, 100 mM NaCl.
Cells containing intein-based expression constructs were disrupted by sonication. The cell lysate was clarified by centrifugation, and the supernatant was applied to a chitin affinity column, which was washed with 1-2 liters of lysis buffer. Intein cleavage was then induced by treatment with ϳ50 mM dithiothreitol on the column, and the reaction was allowed to proceed to completion overnight. Protein was eluted and concentrated to an ϳ200-l volume and applied to a Superdex 200 column (GE Healthcare) for gel filtration and buffer exchange into a storage buffer of 20 mM Tris-HCl, pH 7.9, 100 mM NaCl. Crystallization scans were carried out by standard methods (33).
Gel Filtration Chromatography-Purified protein was dialyzed into gel filtration running buffer (20 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 10 mM NaCl, 1 mM dithiothreitol) and adjusted to a concentration of 0.2 mg/ml. A Superdex 75 26/60 column (GE Healthcare) was equilibrated with running buffer and calibrated with blue dextran molecular weight ϳ10 6 (void volume), bovine serum albumin 66 kDa, ovalbumin 44 kDa, and myoglobin 17 kDa. A 1-ml sample was loaded onto a size exclusion column at 4°C with a flow rate at 1 ml/min, and protein elution was monitored by the absorbance at 280 nm.
Activity Assays-BR30M, a 30-mer oligoribonucleotide substrate containing 2Ј-O-methylated nucleotides at positions 16 and 17 was chemically synthesized as described previously (28) and 5Ј-labeled with [␥-32 P]ATP and T4 polynucleotide kinase. Endonuclease activity was assayed in 30-l reaction mixtures containing 20 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 10 mM NaCl, 1 mM dithiothreitol, and 0.1 pmol of 32 P-labeled oligonucleotide. Reactions were incubated at 30°C and stopped at each time point by removing a 6-l aliquot of the reaction mixture and adding it to an equal volume of loading dye containing 80% (v/v) formamide, 5 mM EDTA, 0.05% (w/v) bromphenol blue, and 0.05% (w/v) xylene cyanol FF. The reaction mixtures were then denatured at 85°C for 3 min and analyzed by electrophoresis in 15% urea-acrylamide gels.
In Vivo Complementation Assays-The Ap r pRNG3-based plasmids into which segments coding for fragments of RNase E or RNase G proteins had been inserted were introduced by transformation into E. coli strain KSL2000 to test their ability to complement an rne deletion using methods described previously (10). Briefly, the rne null mutation of KSL2000 was initially complemented by a Km r plasmid carrying the gene for full-length E. coli RNase E under the control of an arabinose promoter. After transformation with an Amp r plasmid encoding an RNase E or G fragment for which transcription was initiated at an IPTG-inducible lacUV5 promoter, cells were cultured on a LB plate containing 0.1% arabinose, 50 g/ml ampicillin, and 30 g/ml kanamycin. The colonies picked from the LB plates were then inoculated into LB medium containing the same concentrations of arabinose, ampicillin, and kanamycin and grown to A 600 ϭ 0.6. At that cell density, 10 l of 1:10 3 to 1:10 6 dilutions of bacteria were spotted onto LB plates containing 50 g/ml ampicillin and amounts of IPTG (1, 10, 100, or 1000 M) sufficient to induce expression of the truncated RNase E/G peptide being tested (12).

Expression and in Vitro Characterization of Stable Fragments
of RNase E and RNase G-N-Rne has been employed previously for investigation of substrate specificity (14,17), mode of action (28), and inhibition by other cellular proteins (34). To identify the minimal domain that retained activity, additional constructs for expression of N-Rne fragments with deletions from either the amino or the carboxyl terminus were made (Fig. 1B) utilizing an intein-based expression and affinity purification methods (see "Experimental Procedures"). RNase E was predicted from its amino acid sequence to have an RNA-binding S1 domain (Pfam PF00575) near the amino terminus, spanning residues 35 to 119 (16,35). Constructs designed to produce a series of RNase E deletion variants lacking amino acid residues through the S1 domain were generated; specifically, residues 1-115, 1-129, and 1-163 were deleted in independent constructs. However, the protein fragments encoded by these constructs showed very low solubility and were found to form inclusion bodies when overexpressed in E. coli.
Systematic deletions from the carboxyl terminus of N-Rne were made. A fragment consisting of residues 1-494 showed less aggregation than the His 6 -tagged 1-498 N-Rne fragment but also formed hetero-oligomers on gel filtration and failed to form crystals. Using a coarse deletion grid of ϳ20 residues, we found that constructs spanning residues 1 to 400 (Rne 400) or 1 to 300 (Rne 300) produced soluble proteins, whereas constructs expressing fragments of intermediate lengths did not. A finer screening grid identified protein fragments consisting of residues 1-400 (Rne 400) and 1-395 (Rne 395) as being highly soluble. The carboxyl-terminal deletion proteins derived from E. coli N-Rne were assayed for endoribonuclease activity on the 30-nt BR30M substrate ( Fig. 2A) used previously for studies of  Equal amounts of labeled substrate and 50 ng of each protein were used in the assay, and aliquots were taken at the indicated time points. For RNase E derivatives having quasi-processive activity, only the site at the 3Ј end from modified nucleotides can be cleaved, as the enzyme cannot cleave and progress beyond the middle, modified site. This cleavage yields a major product of 25 nt. C, cleavage by fragments of RNase E homologs. The BR30M RNA substrate and 50 ng of protein were used for each assay. Only E. coli RNase G, which cleaves substrate randomly without directionality, can cleave BR30M at the 5Ј end from the modified nucleotides (5 and 6 nt) (25,28).
RNase E cleavage specificity and mode of action (28). BR30M contains three repeats of a sequence that normally is susceptible to RNase E cleavage. However, in BR30M, there are 2Ј-Omethyl-modified nucleotides in the central repeat, preventing cleavage at that site. During quasi-processive RNase E scanning of BR30M from the 3Ј to 5Ј direction, transit of the endoribonuclease past the 2Ј-O-methyl-modified site is blocked, and only the cleavable sequence 3Ј to the modified site is attacked, generating a 25-nt 5Ј-labeled cleavage fragment (28). We found that RNase E fragments as short as 395 residues showed the ability to generate the expected 25-nt fragment, albeit less well than the longer RNase E-derived peptides that we tested (Fig.  2B). A 300-residue peptide was devoid of detectable activity (Fig. 2B), as was one that was only five residues shorter than Rne 395 (data not shown). These findings indicate that truncations that result in RNase E peptides as short as Rne 395 retain the quasi-processive 3Ј to 5Ј scanning mode of action and cleavage site specificity that is intrinsic to E. coli Rne proteins containing the Zn-link region.
As expected from earlier results, (30,36) the Rne 395-truncated E. coli RNase E peptide did not detectably form tetramers, as determined by gel filtration chromatography using the same buffer conditions employed for cleavage assays (Fig. 3). Moreover, rather than showing the dimeric higher order structure reported for larger fragments of the RNase E protein (31,37) Rne 395 was eluted from gel filtration columns predominantly as monomers (Fig. 3).
Fragments of H. influenzae RNase E (residues Ϫ16 to 399; equivalent to E. coli Rne 400 and 85% identical in amino acid sequence in the overlapping region), E. coli RNase G (residues 1-400; E. coli Rng 400), and H. influenzae RNase G (residues 1-396; 36% identical to E. coli RNase E and 71% identical to E. coli RNase G) were expressed, purified, and characterized. All of these truncation fragments lacked the enzyme segment containing the Zn-link region required for tetramer formation by E. coli RNase E. As seen in Fig. 2B, the truncated Ϫ16 to 399 H. influenzae RNase E fragment showed the same cleavage site specificity and mode of action as a longer fragment, Ϫ16 to 476, which contains the Zn-link, and was even more active than E. coli Rne 494, yielding an additional cleavage at the adventitious site found earlier to be cut by highly active longer length RNase E derivatives (28). Equivalent minimal length fragments of H. influenzae Rng (1-396) (Fig. 2B) and E. coli Rng (data not shown) did not show enzymatic activity on this substrate.

In Vivo Complementation of an rne Null Mutation in E. coli by RNase E and RNase G Fragments-The retention of in vitro
endoribonuclease activity of truncated RNase E and G fragments prompted us to test their in vivo activity as assayed by their ability to confer colony forming ability on an E. coli strain carrying an rne null mutation. Using methods described previously (10), plasmid-encoded ribonuclease proteins were expressed under control of a lacUV5 promoter regulated by IPTG added to the culture medium. The rne null mutation was complemented by E. coli Rne 395 in the presence of 10 M IPTG, although cell growth was slower (Fig. 4)   The viability of the cells in absence of arabinose and kanamycin represents the ability of the RNase E or G protein fragments expressed from the pRNG3-derived plasmids to complement the rne null mutation. Left frame, control in which full-length E. coli RNase E expressed from a plasmid complements the null mutation; subsequent frames, increasing levels of IPTG-induced expression of RNase E/G fragments indicated on the left after loss of the plasmid for full-length RNase E (see "Experimental Procedures"). 10-l serial dilutions of bacteria were spotted onto LB plates, and pictures were taken 24 h after incubation. Plates were then incubated for an additional 24 h to verify the inability of cells expressing the truncated E. coli and H. influenzae RNase G fragments to yield colonies. enzae RNase G fragment crystallized readily from 3.5 M sodium acetate, pH 4.0, but the crystals diffracted poorly. The H. influenzae RNase E fragment crystallized under several different conditions; crystals grown from 1.8 M (NH 4 ) 2 SO 4 , 20 mM CoCl 2 , and 100 mM MES, pH 6.5, diffract to ϳ3.4 Å (space group P6 2 or 4 22; a ϭ 105.80 Å, c ϭ 456.54 Å; native data have been collected to 3.4-Å resolution with R sym ϭ 0.076). 6 The H. influenzae RNase E fragment 1-399, from which the putatively extraneous amino-terminal residues were deleted, was also expressed and purified, and it crystallized with unit cell parameters and diffraction limits similar to those of the Ϫ16 to 399 fragment.

DISCUSSION
In E. coli, RNase E is essential to cell viability (3,38,39) and is required for the processing or degradation of multiple RNAs. Homologs of RNase E have been conserved widely during bacterial evolution, and certain of these homologs also have been shown experimentally to have an important biological role in their native species (26,27,40) (also see review in Ref. 41). Earlier work has shown that the 498-residue amino-terminal segment of E. coli RNase E (i.e. N-Rne), which encompasses a fragment in which the amino acid sequence is highly conserved across all members of RNase E/G families, retains the structural features needed to enable the enzyme to cleave at specific RNA sites (17), confers a quasi-processive 3Ј to 5Ј scanning mode of enzymatic action (28,42), and supports bacterial viability. However, an Rne protein fragment containing only the first 321 amino acid residues was inactive (17). The carboxyl-terminal ("scaffold") region enhances the activity of the RNase E protein (29,43) and is required for assembly of degradosomes containing polynucleotide phosphorylase, enolase, and the RhlB helicase, as well as RNase E. Inhibitors of RNase E activity also interact with this region (34). Global analysis of mRNA decay in E. coli has shown that deletion of the rne gene sequence encoding the carboxyl-terminal region of RNase E significantly affects the half-lives of some, but not all, mRNAs in E. coli (44). These data have suggested a physical separation of functions along the length of the RNase E/G proteins in which the highly conserved ϳ500-residue fragment represented by E. coli N-Rne is necessary and sufficient for endoribonuclease activity and the C-terminal extension of RNase E participates in quaternary interactions with other proteins.
The first indication that the N-Rne fragment might not be required in its entirety for activity was provided by Ow et al. (29) who found that a fusion protein consisting of 427 aminoterminal residues plus 25 carboxyl-terminal residues, when overexpressed ϳ20-fold in vivo, could rescue an rne deletion. However, the cells were described as "viable but very defective in mRNA decay at 37°C and nonviable at 44°C," and the enzymatic properties of the truncated protein were not examined (29).
We undertook to systematically define the minimal catalytic fragment necessary for the enzyme to function properly in vitro and in vivo and to elucidate the structural components required for these functions. Additionally, current models suggest that proper functioning of RNase E requires that the enzyme form a tetrameric quaternary structure, and we wished to learn the biochemical effects of removal of the Zn-link region recently found to be necessary for tetramer formation (36). Our results show that Rne protein fragments that contain ϳ400 amino acid residues of the ϳ1000-residue RNase E enzymes of E. coli and H. influenzae and lack the Zn-link retain the structural determinants required for site-specific endonuclease activity and the unique quasi-processive 3Ј to 5Ј scanning mode of action characteristic of RNase E in vitro.
Recently, Callaghan et al. (30) solved the crystal structure of the catalytic domain of E. coli RNase E by co-crystallizing the peptide containing amino acids 1-529 and RNA oligomeric substrate; they found that the 529-amino acid residue segment of the protein consists of two separate domains, a small one and a large one. The two domains of the 529-amino acid peptide analyzed by Callaghan et al. (30) are joined by a Zn-link, and their analysis of the crystal structure suggested that the Zn-link positions small domains of individual RNase E molecules to form a homotetrameric quaternary structure rather than link two dimers as proposed previously (36). Our computational analysis of the published crystal structure using a protein interaction server (Protein-Protein Interaction Server, version 1.5) found that the region likely to provide the most stable interface for interaction between the RNase E catalytic domains also spans the Zn-link and the small domain. Our biochemical evidence that a peptide virtually congruent with the large domain of Callaghan et al. (30) (i.e. the first 400 amino acid residues) sufficient for RNase E catalytic activity, site-specific cleavage of substrates, and a 3Ј to 5Ј scanning mode of action is consistent with the observation that the catalytically functional minimal RNase E peptide region exists predominantly as a monomer under the conditions we employed for our analyses.
It has been shown that RNase E homologs, including E. coli RNase G (12) and Streptomyces coelicolor RNase ES (26), confer colony forming ability on an rne null mutant strain when overexpressed in E. coli, suggesting that RNase E-like proteins that differ significantly outside the conserved N-Rne fragment have similar biological abilities. Although in some cases, significantly greater induction of gene expression was required for functional complementation by the truncated peptides we studied, our results indicate that peptides that cleaved the RNA substrates used in these studies in vitro also had the ability, when expressed in vivo, to complement the rne null mutation. Conversely, truncated fragments of RNase G that showed no detectable enzymatic activity on a substrate containing specific RNase E cleavage sites failed to complement the rne null mutation at any level of induction tested. The RNase E/G fragments that we have characterized provide both positive and negative examples that establish a close correlation between in vitro endonuclease activity on an RNase E substrate and the capability to complement an rne null mutation in vivo.
The structural integrity of the minimal peptide fragments was inferred from the observed enzymatic and biological properties of the fragments and was further confirmed by their crystallization. Shorter, structurally stable fragments ϳ300 residues in length were isolated, but they did not retain enzymatic activ-ity. The requirement for a minimum of 395 amino acids for endonuclease activity, site-specific cleavage, and a quasi-processive mode of action of RNase E suggests that both the S1 domain near the amino-terminal end of the protein and the small arginine-rich region spanning amino acid residues 307 to 390 are essential to substrate binding and cleavage. However, the ϳ100 amino acids that follow this fragment, although highly conserved among RNase E/G homologs, are not essential. Amino-terminal (ϳ400 residues) fragments of the RNase G proteins of E. coli and H. influenzae also displayed structural integrity, as demonstrated by crystallization of the fragment of the H. influenzae protein. However, they did not show endoribonuclease activity on the RNase E substrates used in these studies, and correspondingly, they did not complement an rne mutation in vivo.
Our results indicating that a peptide that lacks the Zn-link segment required for tetramer formation can nevertheless carry out the core catalytic functions of the enzyme argue strongly that tetramer formation is unnecessary for the enzyme to bind to the phosphate at the 5Ј termini of RNA substrates (45) and then scan for cleavage sites from the 3Ј end (28). The Zn-link region is also not necessary to confer colony-forming ability on an rne null mutant. We postulate that tetramer formation by RNase E instead may assist RNase E in maintaining RNA substrates in close proximity to the enzyme to facilitate efficient cleavage.