Membrane binding of Escherichia coli RNase E ... - Semantic Scholar

Download PDF
22 downloads 0 Views 771KB Size Report
Comment
May 1, 2012 - degradosome has been shown to be tethered to the cytoplasmic membrane by the N-terminal region of RNase E (amino acids 1–. 602) (17).
Membrane binding of Escherichia coli RNase E catalytic domain stabilizes protein structure and increases RNA substrate affinity Oleg N. Murashko, Vladimir R. Kaberdin1, and Sue Lin-Chao2 Institute of Molecular Biology, Academia Sinica, Taipei 11529, Taiwan Edited by Joel G. Belasco, New York University School of Medicine, New York, NY, and accepted by the Editorial Board March 22, 2012 (received for review December 8, 2011)

|

bacteria posttranscriptional regulation RNA degradosome

| RNA degradation |

R

Nase E has an essential role in RNA processing and decay in Escherichia coli (1), and it has been postulated that its homologs in other bacteria may have similar functions (2–5). E. coli RNase E, encoded by the rne gene, is a large, multidomain protein of 1,061 amino acids with several functionally distinct regions (1, 6, 7). The first 395 amino acids of E. coli RNase E confer a single-strand–specific endonuclease activity and are highly evolutionarily conserved (8–10). The crystal structure of the truncated RNase E polypeptide [Protein Data Bank (PDB) ID code: 2C4R], determined at 2.9 Å, revealed a structurally defined catalytic domain (designated the “large” domain, amino acids 1–400), followed by a Zn-link spacer (amino acids 400–415) and a “small” domain (amino acids 415–529) that serves as a dimerization interface (11). The large domain consists of several subdomains including the 5′-sensor as well as subdomains structurally similar to protein folds found in S1, DNase I, and RNase H. The C-terminal noncatalytic domain of E. coli RNase E also contains functionally important regions including an arginine-rich region (amino acids 597–684) involved in RNA-binding (12) and a “scaffold” region (amino acids 650–1061) for RNA degradosome assembly with RhlB, enolase, and PNPase (4, 13, 14). In contrast to the N-terminal catalytic domain, the C-terminal half is poorly conserved among RNase E homologs (15). www.pnas.org/cgi/doi/10.1073/pnas.1120181109

The enzymatic activities of RNase E have been detected in membrane fractions of E. coli cell lysates (16), and the RNase E degradosome has been shown to be tethered to the cytoplasmic membrane by the N-terminal region of RNase E (amino acids 1– 602) (17). Thereafter, a specific “segment A” (amino acids 562– 587) has been identified and is required for the membrane binding of the enzyme (18). However, the function of membrane binding and whether other members of the RNase E/G family also are membrane-binding proteins have not been known. In the present study, we found that the E. coli N-terminal fragment of RNase E (NRne, amino acids 1–499), RNase G, and phylogenetically distant RNase E/G homologs from Aquifex aeolicus, Haemophilus influenzae Rd, and Synechocystis sp. interact with the cytoplasmic membrane. We identified four putative membraneattached regions in the most conserved NRne (amino acids 1–400) that form an extremely positive electrostatic potential on the protein surface that enables interaction with the membrane. Liposome-bound NRne and RNA substrates BR13 and GGG-RNAI showed that NRne membrane binding alters its enzymatic activity, and there were no obvious thermotropic structural changes in membrane-bound NRne between 10 and 60 °C. The membrane– protein interaction thus affects the secondary structure of the catalytic domain, stabilizing the folding state of the protein. Membrane binding of the catalytic domain thus increases substrate affinity and enzymatic activities. Our results elucidate the mechanism of the protein–membrane interaction and demonstrate its positive effect on RNase E-mediated RNA decay in bacteria. Results N- and C-Terminal Halves of E. coli RNase E Bind Independently to the Cell Membrane. To determine the localization of E. coli NRne and

C-terminal RNase E (CRne) polypeptides (amino acids 1–499 and 500–1061, respectively), subcellular fractions were prepared using sodium carbonate treatment according to the procedure outlined by Fujiki et al. (19) that prevents unwanted cytoplasmic contaminants from being trapped in the membrane fraction and strips loosely associated proteins from the cell membrane. Cytoplasmic GAPDH and membrane TolA proteins were used as markers to measure the completeness of membrane fraction separation. Endogenous fulllength RNase E (FLRne) and TolA were found exclusively in the membrane fraction (Fig. 1, lane 3), and GAPDH was detected only

Author contributions: O.N.M., V.R.K., and S.L.-C. designed research; O.N.M. performed research; O.N.M. and S.L.-C. contributed new reagents/analytic tools; O.N.M., V.R.K., and S.L.-C. analyzed data; and O.N.M., V.R.K., and S.L.-C. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. J.G.B. is a guest editor invited by the Editorial Board. Freely available online through the PNAS open access option. 1

Present address: Department of Immunology, Microbiology and Parasitology, University of the Basque Country UPV/EHU, Leioa, Spain; IKERBASQUE, Basque Foundation for Science, 48011 Bilbao, Spain.

2

To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1120181109/-/DCSupplemental.

PNAS | May 1, 2012 | vol. 109 | no. 18 | 7019–7024

GENETICS

RNase E plays an essential role in RNA processing and decay and tethers to the cytoplasmic membrane in Escherichia coli; however, the function of this membrane–protein interaction has remained unclear. Here, we establish a mechanistic role for the RNase E–membrane interaction. The reconstituted highly conserved N-terminal fragment of RNase E (NRne, residues 1–499) binds specifically to anionic phospholipids through electrostatic interactions. The membrane-binding specificity of NRne was confirmed using circular dichroism difference spectroscopy; the dissociation constant (Kd) for NRne binding to anionic liposomes was 298 nM. E. coli RNase G and RNase E/G homologs from phylogenetically distant Aquifex aeolicus, Haemophilus influenzae Rd, and Synechocystis sp. were found to be membrane-binding proteins. Electrostatic potentials of NRne and its homologs were found to be conserved, highly positive, and spread over a large surface area encompassing four putative membrane-binding regions identified in the “large” domain (amino acids 1–400, consisting of the RNase H, S1, 5′-sensor, and DNase I subdomains) of E. coli NRne. In vitro cleavage assay using liposome-free and liposome-bound NRne and RNA substrates BR13 and GGG-RNAI showed that NRne membrane binding altered its enzymatic activity. Circular dichroism spectroscopy showed no obvious thermotropic structural changes in membrane-bound NRne between 10 and 60 °C, and membrane-bound NRne retained its normal cleavage activity after cooling. Thus, NRne membrane binding induced changes in secondary protein structure and enzymatic activation by stabilizing the protein-folding state and increasing its binding affinity for its substrate. Our results demonstrate that RNase E–membrane interaction enhances the rate of RNA processing and decay.

Fig. 1. The N- and C-terminal halves of E. coli RNase E have independent membrane-binding regions. Subcellular localization of endogenous E. coli FLRne (amino acids 1–1061) and ectopically expressed NRne (amino acids 1– 499) or CRne (amino acids 500–1061). Aliquots containing equal amounts of total protein (T) and equivalent amounts of proteins from cytoplasmic (C) and membrane (M) fractions were separated on 10% SDS/PAGE. Western blotting was conducted as described in SI Materials and Methods. The bands corresponding to endogenous TolA, GAPDH, and FLRne and ectopically expressed NRne or CRne are indicated.

in the cytoplasmic fraction (Fig. 1, lane 2). Ectopically expressed NRne and CRne were found predominantly in the membrane fraction (Fig. 1, lanes 4–9) suggesting that the N- and C-terminal portions of the enzyme contain independent membrane-binding regions. Because interaction of the C-terminal noncatalytic domain of E. coli RNase E has been investigated previously (18), only the evolutionarily conserved N-terminal catalytic domain was used here to study the membrane-binding features. Electrostatic Binding of NRne to Anionic Liposomes. Liposomes have been used extensively to study protein–membrane interactions and their effects on the biochemical properties of membrane proteins (e.g., refs. 20 and 21). To determine the nature of the NRne– membrane interaction, we studied the binding of protein to different types of large unilamellar vesicles (LUVs) prepared from E. coli polar lipids extract (PLE) containing 67.0% phosphoethanolamine, 23.2% phosphatidylglycerol, and 9.8% cardiolipin (for anionic liposomes), or 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) (for neutral liposomes), or DOPE/stearylamine (SA), 1:1 molar ratio (for cationic liposomes). In the control experiment without liposomes, no protein precipitation was found in the pellet (Fig. 2A, lanes 1–3) after spin assay (SI Materials and Methods); however, after incubation of NRne with anionic liposomes (Fig. 2A, lanes 4–6), most of the NRne was found in the pellet, indicating that NRne interacts with the anionic liposomes. In contrast, incubation with cationic liposomes (Fig. 2A, lanes 10–12) did not result in pelleting of NRne. Incubation of NRne with neutral liposomes (Fig. 2A, lanes 7–9) resulted in only a small amount of NRne protein in the pellet, and this protein could be removed by further sodium carbonate treatment (compare Fig. 2B, lanes 4–6, with Fig. 2A, lanes 7–9). These results show that NRne interacts with the anionic liposomes but has no specific interaction with neutral or cationic liposomes. In addition, after incubation with anionic liposomes, NRne was largely retained in the pellet even after sodium carbonate treatment followed by another spin assay (Fig. 2B, lanes 1–3), further suggesting that NRne interacts strongly with anionic liposomes. Finally, NRne showed no association with the anionic membrane in the presence of 1 M NaCl (Fig. 2B, lane 9), indicating that the NRne–membrane interaction is electrostatic. Conformational Changes in NRne upon Binding to Anionic Liposomes.

Differential circular dichroism (CD) spectroscopy was used to detect conformational changes that may occur upon binding of NRne to liposomes and to calculate binding constants (22). To record CD spectra of free and membrane-bound NRne under precisely the same conditions, we used a specifically designed tandem cuvettete (Fig. S1A) with two separate chambers that are optically in tandem, as described previously (23). CD spectra were recorded for samples of NRne and liposomes placed in separate chambers of a tandem cuvettete and then were recorded again after the samples were mixed in the same cuvettete. Among the four spectra analyzed (Fig. 2C), the far UV CD spectra (200–250 nm) of lipid-free NRne (line 1) and NRne mixed with cationic liposomes 7020 | www.pnas.org/cgi/doi/10.1073/pnas.1120181109

Fig. 2. Binding of NRne to anionic liposomes occurs through electrostatic interactions and affects the conformation of the protein. (A) NRne–liposome binding spin assays using different types of phospholipids. NRne (16 μM) was mixed with neutral, anionic, or cationic liposomes (1.35 mg/mL) as described in SI Materials and Methods. The total reaction mixture (T), supernatant (S), and pellet (P) were separated by 10% SDS/PAGE and analyzed by Coomassie Blue staining. (B) NRne interacts specifically with anionic liposomes through electrostatic interaction. Individual pellets (P), as shown in A, were divided and suspended in 100 mM sodium carbonate or 1 M NaCl, incubated for 30 min on ice, and repelleted using the procedure described in SI Materials and Methods. The repellet and supernatant are denoted P1 and S1, respectively. (C and D) Detection of conformational changes in NRne bound to the different types of liposomes by CD difference spectroscopy using a specifically designed tandem cuvettete (see Fig. S1A and Materials and Methods for details). The mixture conditions of protein and individual types of liposomes are identical to those used in the spin assays shown in A. CD difference spectra corresponding to individual types of NRne– liposome mixtures are shown. The affinity of NRne and liposomes was calculated by monitoring conformational changes upon binding as observed by CD difference spectroscopy (Materials and Methods). NRne (1 μM) was mixed with different concentrations of liposomes (0–20 μM). Data were fitted to a one-site binding hyperbolic curve by OriginPro software and are presented as mean values with SDs calculated from three independent experiments. NRne binds to the anionic liposomes with a high affinity (solid line, Kd = 298 ± 29 nM). NRne mixed with cationic liposomes shows no binding (dashed line).

(line 4) showed a similar shape and magnitude, suggesting the absence of a specific interaction between NRne and cationic liposomes. This result is in agreement with results from the spin assay (Fig. 2A, lanes 10–12). In contrast, the CD spectra of NRne mixed with anionic liposomes (Fig. 2C, line 2) revealed a substantial bathochromic shift, indicating conformational reorganization of the secondary structure of NRne. This bathochromic shift was reversed by increasing the NaCl concentration, as shown in Fig. S1B. A small bathochromic shift in the CD spectra occurred for NRne mixed with neutral liposomes (Fig. 2C, line 3). These results agree with the spin assay results, which showed some nonspecific interaction between NRne and this type of liposome (Fig. 2A, lanes 7–9) that could be disrupted by sodium carbonate (Fig. 2B, lanes 4–6). Calculation of percentages of secondary structure components of NRne in the presence and absence of anionic liposomes revealed a decrease in α-helix content and an increase in β-sheet and random coil content upon interaction with anionic liposomes (Tables S1 and S2), suggesting that membrane binding induces a structural change in NRne that may lead to a flexible protein conformation. A similar effect was observed by Raman spectroscopy (Fig. S1C and Tables S1 and S2). Using CD difference spectroscopy as described above and a fixed amount (1 μM) of the NRne titrated with various concentrations (0–20 μM) of liposomes, we determined the binding affinity Murashko et al.

Fig. 3. Calculation and mapping of the membrane-binding surface of NRne. (A) Electrostatic potentials of NRne. We used the 3D structures of the catalytic domain of E. coli RNase E (PDB ID codes 2VRT and 2C4R for the open and closed form of the enzyme, respectively) to calculate electrostatic potentials of NRne as described in SI Materials and Methods. Blue and red surface areas represent positive and negative equipotential contours, respectively. The positively charged surface (the region above the dashed line) likely approximates the extent of the envelope of the polar head group region of the membrane surface. NRne subdomains are indicated. The figure was produced using PyMol. (B) Subcellular localization of GAPDHtagged E. coli RNase E polypeptides. Names of RNase E variants are indicated above each gel. Subscript numbers indicate the position of NRne amino acids that were fused with GAPDH. Cellular fractions were prepared and analyzed as described in the legend of Fig.1. The protein bands corresponding to plasmid-expressed GAPDH fusion polypeptides ( ) and endogenous GAPDH (*) are indicated. (C) The NRne fragment (amino acids 1–499) is shown divided into subdomains according to Callaghan et al. (11). The four putative NRne membranebinding regions identified in our study are shown as gray rectangles. (D) Spin assay analyses of the binding of the DNase I subdomain to anionic liposomes. DNase I subdomain (16 μM) was mixed with anionic liposomes (1.35 mg/mL) as described in SI Materials and Methods. (Upper) The total reaction mixture (T), supernatant (S), and pellet (P) were separated by 15% SDS/ PAGE and analyzed by Coomassie Blue staining. (Lower) The pellet (P) obtained in lane 6 (Upper) was divided and suspended in 100 mM sodium carbonate or 1 M NaCl, incubated for 30 min on ice, and repelleted as described in SI Materials and Methods. The repellet and supernatant are denoted as P1 and S1, respectively. (E) The DNase I subdomain binds specifically with anionic liposomes. Kd (solid line, Kd = 387 ± 34 nM) was calculated as described in the legend of Fig. 2. The DNase I subdomain shows no interaction with cationic liposomes DOPE/SA (dashed line).

between NRne and anionic liposomes, as described previously (23). The far UV CD spectra were collected and normalized to the ellipticity (θ) at 222 nm. Data fitted to a one-site binding hyperbolic curve using OriginPro software (OriginLab) yielded an apparent dissociation constant (Kd) of 298 ± 29 nM (Fig. 2D), demonstrating high-affinity binding of NRne to anionic liposomes. Therefore, our Murashko et al.

data suggest the existence of a positively charged surface that mediates interaction of NRne with the inner membrane. Four Putative Membrane-Binding Sites Mapped to the Positively Charged Surface of NRne Calculated in Silico. We used the PDB2PQR (24) and

APBS (25) programs to calculate the electrostatic properties of PNAS | May 1, 2012 | vol. 109 | no. 18 | 7021

GENETICS

◀

published NRne structures [PDB ID codes 2VRT and 2C4R for the “open” and “closed” form of the enzyme, respectively (11, 26)]. Calculations revealed the existence of several positively charged clusters in both forms of the enzyme. These clusters form a continuous surface (shown in blue in Fig. 3A) within the RNase H, 5′sensor, and DNase I subdomains that is likely to be in contact, via electrostatic interactions, with the negatively charged phospholipid head groups of the membrane. To provide further experimental evidence for the existence of the membrane-binding surface of NRne calculated in silico, we determined the subcellular localization of NRne polypeptides corresponding to different subdomains. Because the coordinates of each subdomain can be inferred from the crystal structure of the NRne (11), we generated individual NRne-derived polypeptides that do not disrupt intact α-helices and/or β-sheets (Fig. S2). However, these protein segments are unlikely to have folded into the native conformation determined from the crystal structure of NRne unbound to membrane. Because both endogenous and ectopically expressed GAPDH were present exclusively in the cytoplasmic fraction (Fig. 3Ba, lanes 1–3), and the subcellular localization of NRne-GAPDH and GAPDH-CRne fusion proteins (Fig. 3Ba, lanes 4–9) mimicked the localization of untagged NRne and CRne (Fig. 1, lanes 4–9), as detected by monoclonal anti-GAPDH antibodies or polyclonal antibodies against RNase E, respectively, we next analyzed the presence of NRne polypeptides in the cytoplasmic and membrane fractions by immunodetection of their GAPDH-tagged variants with the anti-GAPDH antibodies. Using this GAPDH-tagged approach, we identified four putative membrane-binding sites: amino acids 20–40 (Fig. 3Bb, lanes 1–6), amino acids 111–160 (Fig. 3Bc, lanes 1–9), amino acids 216–279 (Fig. 3Bd, lanes 1–3), and amino acids 280–400 (Fig. 3Bd, lanes 4–12). In contrast, the fusion polypeptides Rne41–110 (Fig. 3Bc, lanes 1–3) and Rne401–499 (Fig. 3Bd, lanes 10–12), containing the S1 domain and the Zn-link spacer plus the small domain, respectively, were present exclusively in the cytoplasmic fraction, indicating that these domains are exposed to the cytoplasm. Collectively, these results provide evidence consistent with in silico data suggesting that the positively charged surface within RNase H, 5′-sensor and DNase I subdomains mediate the interaction of the enzyme with the membrane. DNase I Subdomain Contributes to High-Affinity Binding of NRne to Anionic Liposomes. We found that the positively charged residues

(Fig. S3) and their distribution within the membrane-binding interface are highly conserved among RNase E/G homologs, particularly in the DNase I subdomain (Fig. S4). To evaluate the membrane-binding properties of the DNase I subdomain of E. coli RNase E, we analyzed its interaction with anionic liposomes by spin assay. The DNase I subdomain interacted strongly with these liposomes (Fig. 3D, lanes 4–6), because it remained predominantly in the pellet even after sodium carbonate treatment (Fig. 3D, lanes 7–9); the DNase I subdomain was released from the membrane only after treatment with 1 M NaCl (Fig. 3D, lanes 10–12). Determination of the binding affinity of the subdomain with anionic liposomes yielded an apparent Kd value of 387 ± 34 nM (Fig. 3E), a value only slightly lower than that obtained for NRne and anionic liposomes. Membrane Binding is a Common Property among RNase E/G Homologs.

Because the N-terminal catalytic domain of RNase E is evolutionarily conserved (15), we investigated the membrane-binding properties of RNase E/G homologs (Fig. S5). A significant fraction of their GAPDH-tagged variants was found in the membrane fractions for all the E. coli RNase E/G homologs tested: E. coli RNase G and A. aeolicus, H. influenzae Rd, and Synechocystis sp. homologs ectopically expressed in DH5α E. coli (Fig. 4, lanes 1– 12). This result indicates that membrane association is a common feature of RNase E/G homologs and that one or more membrane-binding regions of these proteins are located within their evolutionarily conserved N-terminal catalytic domains (Fig. S5B). 7022 | www.pnas.org/cgi/doi/10.1073/pnas.1120181109

Fig. 4. RNase E/G homologs bind to the cytoplasmic membrane via the evolutionarily conserved catalytic domain of the enzyme. Subcellular distributions of GAPDH-fused derivatives including E. coli RNase G (Rng), RNase E/G homologs from A. aeolicus (AaeRne), and the putative catalytic domains of H. influenzae Rd (HinNRne, amino acids 1–418) and Synechocystis sp. (SynNRne, amino acids 1–396). Cellular fractions were prepared and analyzed as described in the legend of Fig.1. The bands corresponding to endogenous GAPDH (*) and ectopically expressed RNase E/G protein derivatives fused with GAPDH ( ) are indicated.

◀

Specific Interaction of NRne with Anionic Liposomes Increases Its Substrate Affinity and Endoribonucleolytic Cleavage Activity. To

understand the role of membrane binding in RNase E-mediated decay, we investigated whether NRne attachment to the membrane could alter its enzymatic activity by in vitro cleavage assay using liposome-free and liposome-bound NRne. Cleavage assays were performed with GGG-RNAI and its truncated variant BR13, which have been used as model substrates of RNase E (27–29). The binding of anionic liposomes to NRne increased the enzymatic cleavage rate (Fig. 5 A and B; compare the rates of product formation in lanes 8–12 and 2–6). In agreement with the liposomebinding data shown in Fig. 2, anionic but not cationic or neutral liposomes increased the rate of NRne cleavage (Fig. 5 A and B, lanes 8–12, and Fig. S6 A and B, lanes 2–6 and 8–12). Membrane binding increased the enzymatic activity of NRne by approximately twofold (P < 0.05, Student’s t test) (Fig. S6 C and D; compare the slopes of the black and the red lines). Furthermore, Michaelis–Menten enzyme kinetics of BR13 cleavage by NRne showed a 2.3-fold (P < 0.001, Student’s t test) decrease in the Km value upon NRne binding to anionic liposomes (Fig. 5C), revealing that membrane binding increases the RNA substrate affinity of NRne. NRne–Liposome Binding Stabilizes Protein Folding and Prevents Protein Structure Changes During Thermal Denaturation. We fur-

ther characterized the thermotropic structural change in NRne between 10 and 60 °C in the absence or presence of anionic liposomes by CD spectroscopy. The thermally induced unfolding of the soluble NRne started at 44 °C (Fig. 5D, filled circles; melting temperature of ∼49 °C). In contrast, under the same conditions, the membrane-bound NRne showed no obvious thermotropic structural changes and retained a thermodynamic intermediate state (Fig. 5D, filled triangles). Furthermore, far UV CD spectra of NRne (amino acids 1–499) in the absence (Fig. 5E, line 1) and presence (Fig. 5E, line 3) of anionic liposomes and spectra of anionic liposomes alone (Fig. 5E, line 5) were recorded at 25 °C. The samples were heated gradually to 60 °C; then the samples were cooled, and the CD spectra were recorded again at 25 °C (Fig. 5E, lines 2, 4, and 6, respectively). As assessed by CD spectral analysis, the membrane-bound NRne retained its protein-folding structure after cooling, whereas the soluble NRne remained in an irreversibly unordered state that could not be refolded (Fig. 5E, lines 3 and 4 vs. lines 1 and 2). In addition, the membrane-bound NRne consistently retained its normal cleavage activity after cooling (Fig. 5F, lanes 3 and 4), whereas the membrane-free NRne lost its enzymatic activity completely (Fig. 5F, lanes 1 and 2). These data indicate that binding of NRne to the membrane stabilizes the protein structure and prevents thermally induced unfolding. Discussion Like many other peripheral membrane proteins, such as cytochrome c and apolipoprotein H (30), native NRne exhibits both water-soluble and membrane-bound states. Membrane binding of a protein can affect protein structure, increasing or decreasing Murashko et al.

the enzymatic activity of the protein (31, 32). Because the physicochemical features of the bulk aqueous phase differ from those adjacent to the membrane surface (33), the stability of the open and closed protein structures of RNase E (11, 26) can depend on the enzyme’s localization in vivo. In this study, we detected membrane-induced conformational changes in NRne bound to anionic liposomes but not in NRne bound to other types of liposomes, and the membrane-bound structures gave rise to higher enzymatic activity by increasing the substrate affinity. The membrane-bound NRne appears to be retained in a thermodynamic membrane-bound state with a native-like structure or with an intermediate structure that differs from both the native and the denatured structures (Fig. 5E, lines 3, 1, and 2, respectively). As a result, the binding protects the NRne folded structure and its enzymatic activity against thermally induced denaturation. Thus, the conformational changes induced in NRne by anionic lipids may sustain the membrane-bound state of the protein and facilitate the proper 3D orientation of NRne for highaffinity binding and faster degradation of the RNA substrate. In contrast, without membrane binding, the soluble NRne became an irreversibly disordered structure upon thermally induced transition, indicating that lack of the membrane-induced conformationally changed state may result in aggregation of RNase E in the cytoplasm upon its synthesis at higher temperatures. Murashko et al.

Our data show that protein–membrane association likely involves a continuous, positively charged surface formed by four membrane-binding regions distributed within the RNase H, 5′sensor, and DNase I subdomains of NRne and that the protein– membrane binding is electrostatic. Furthermore, in silico analyses of charge distributions of NRne and RNase E/G homologs based on the linearized Poisson–Boltzmann equation revealed a highly conserved pattern within these subdomains, particularly in the DNase I subdomain (56% amino acid identity) (Figs. S3 and S4). These findings suggest that the characteristics of the interaction of NRne with the membrane are common among RNase E/G homologs in bacteria. The cellular membrane plays an important role in compartmentalization of many biological processes, such as metabolism, transport, and respiration, in bacteria and other organisms. Here, our studies suggest a role of the membrane in RNase E-mediated RNA processing and decay in bacteria. Finally, an RNase E region (segment A; amino acids 562–587) that was proposed to have the propensity to form an amphipathic α–helix was found by Khemici et al. (18) to be required for RNase E binding to membrane. However, the endoribonuclease activity of the renatured his-RNase E was not affected by vesicle binding in vitro. Additional experiments showed that enzyme tagged at its C-terminal end with yellow fluorescence protein was delocalized from the plasma membrane and formed aggregates at the poles PNAS | May 1, 2012 | vol. 109 | no. 18 | 7023

GENETICS

Fig. 5. Binding to liposomes stabilizes the folding state of NRne and increases the enzymatic activity by increasing substrate affinity. (A and B) In vitro cleavage assays. As described in SI Materials and Methods, 20 pmol of BR13 or GGG-RNAI was incubated with 1 μg of liposome-free NRne (control) or liposome-bound (anionic liposomes) NRne at 37 °C. Aliquots taken after 2, 4, 6, 8, and 10 min (BR13) or 10, 20, 30, 40, and 50 min (GGG-RNAI) of incubation were separated on 20% or 8% sequencing gels, respectively. The position of bands corresponding to the substrates and products of cleavage are indicated at the left of each panel. (C) Michaelis–Menten analysis of BR13 cleavage by NRne in the presence (NRne+anionic liposomes) or absence (NRne) of anionic liposomes. The concentration of NRne was 2.5 nM in each sample; the concentration of substrate varied between 2.5 nM and 50 nM. The vo values reflecting the accumulation of cleavage product were fitted to a one-site binding hyperbolic curve using OriginPro software and are presented as mean values with SDs calculated from three independent experiments. (D) Thermally induced transition curves of NRne obtained by recording θ at 222 nm. The curves represented by filled and open circles correspond to the signal for NRne in the absence of anionic liposomes upon gradual temperature increase (from 10–60 °C, filled circles) or decrease (from 60–10 °C, open circles), respectively. The same measurements were performed for NRne in the presence of anionic liposomes (filled and open triangles, respectively). (E) Effect of temperature on CD spectra of NRne. Far UV CD spectra of NRne (amino acids 1–499) in the absence (line 1) and presence (line 3) of anionic liposomes as well as of anionic liposomes alone (line 5) were recorded at 25 °C. The samples were heated gradually to 60 °C and then were cooled; the CD spectra were recorded again at 25 °C (lines 2, 4, and 6, respectively). The compositions of the analyzed samples were the same as in the experiments presented in Fig. 2. (F) Thermal inactivation of NRne. Cleavage assays of BR13 were performed with NRne before (lane 1) and after (lane 2) thermal inactivation or with the liposome-bound variant before (lane 4) and after (lane 4) thermal inactivation using the NRne samples that were used for recording the CD spectra shown in D. The mixture conditions of protein and anionic liposomes are identical to those shown in A. Aliquots withdrawn after 4 min incubation were analyzed on a 20% sequencing gel.

of cells when segment A was absent from the otherwise fulllength protein. Our results indicate that segment A is not required for interaction of native NRne (amino acids 1–499), which was untagged and lacks the C-terminal half of the protein, with anionic liposomes. Thus, segment A is necessary for membrane binding of full-length RNase E but is dispensable for membrane interaction of the catalytically active segment, although addition of segment A may aid such binding. We hypothesize that segment A may mediate membrane binding of the full-length protein by assisting proper assembly of degradosome complexes on the intrinsically unstructured RNase E C-terminal scaffolding region (amino acids 499–1,061) (34). In contrast, the membrane-binding of NRne and of natural RNase E/G homologs that consist almost entirely of this region may induce conformational changes that are independent of the C-terminal region and which alter the catalytic functions and thermostability of the protein. Materials and Methods

Biochemical and Molecular Biology Techniques. Cell fractionation and RNase E cleavage assays were performed as described (19, 29) (SI Material and Methods). Preparation of Liposomes. LUVs with an approximate diameter of 0.1 μm (Fig. S7) were formed by extrusion with a miniextruder (Avanti Polar Lipids, Inc.) (SI Material and Methods). CD Difference Spectroscopy. CD spectra were recorded using an Aviv CD spectrometer (Model 400; AVIV Biomedical) with a 0.875-cm specifically designed tandem cuvettete (Hellma) (Fig. S1A). To study possible conformational changes upon NRne interaction with artificial membranes, liposomes (1.35 mg/mL) and NRne (16 μM) were diluted in 50 mM NaCl, 25 mM Na+/K+ phosphate buffer (pH 7.5), placed into separate chambers of the tandem cuvettete (Fig. S1A), and allowed to equilibrate to solvent conditions for 30 min at 25 °C. Partners in two chambers of the cuvettete then were mixed by inversion and allowed to equilibrate for 30 min at 25 °C. Three spectra for each condition were collected before and after mixing in the wavelength range of 200–250 nm using a bandwidth of 1 nm and 1-nm resolution. Data are expressed as mean residue ellipticity: [θ] (degree·cm2/dmol). The fractional percentage of the secondary structure was calculated by the DichroWeb (35). Thermotropic transition experiments were performed at 222 nm a speed of 1 °C/min from 10–60 °C. For the measurement of binding affinity, see SI Material and Methods.

Bacterial Strains, Plasmids, and Protein Expression. E. coli strain DH5α (Invitrogen) was used for the protein localization study. The plasmid pBAD-EBFP2 (Addgene) was used for plasmid constructions and protein expression for subcellular localization experiments as described in SI Material and Methods. E. coli strain BL21(DE3) (Invitrogen) was used for protein isolations. For overexpression of GST-tagged NRne (amino acids 1–499) or the DNase I subdomain (amino acids 280–400), the plasmids were created in one step by in-frame insertion of the corresponding DNA fragments into the EcoRI and XhoI sites of pGEX-4T-1. The primers used in this study for PCR are listed in Table S3. The GST-tags were removed from isolated proteins by thrombin, and the resulting polypeptides (i.e., untagged NRne and the DNase I subdomain) were purified further by gel filtration and then were used for spin-down assays, cleavage assays, and Raman and CD spectroscopy (SI Material and Methods).

ACKNOWLEDGMENTS. We thank Drs. H. Kuhn, A. Peña, and M. Loney for help in editing the manuscript; Dr. Chak (National Yang-Ming University) for providing the specific antibody for TolA; S.-P. Tsai (Institute of Molecular Biology, Academia Sinica) and Y.-C. Chao (National Taiwan University) for technical support; and the reviewers and editor for their critical comments. This work was supported by National Science Council Frontier Research Grants NSC 97- to 100-2321-B-001 from the National Science Council, Taiwan, and by Academia Sinica Grant AS 23-23 (to S.L.-C.). V.R.K. and O.N.M. were supported by Academia Sinica Grant AS 23-23.

1. Cohen SN, McDowall KJ (1997) RNase E: Still a wonderfully mysterious enzyme. Mol Microbiol 23:1099–1106. 2. Hagège JM, Cohen SN (1997) A developmentally regulated Streptomyces endoribonuclease resembles ribonuclease E of Escherichia coli. Mol Microbiol 25:1077–1090. 3. Jäger S, et al. (2001) An mRNA degrading complex in Rhodobacter capsulatus. Nucleic Acids Res 29:4581–4588. 4. Kaberdin VR, et al. (1998) The endoribonucleolytic N-terminal half of Escherichia coli RNase E is evolutionarily conserved in Synechocystis sp. and other bacteria but not the C-terminal half, which is sufficient for degradosome assembly. Proc Natl Acad Sci USA 95:11637–11642. 5. Zeller ME, et al. (2007) Quaternary structure and biochemical properties of mycobacterial RNase E/G. Biochem J 403:207–215. 6. Casarégola S, et al. (1992) Cloning and analysis of the entire Escherichia coli ams gene. ams is identical to hmp1 and encodes a 114 kDa protein that migrates as a 180 kDa protein. J Mol Biol 228:30–40. 7. Cormack RS, Genereaux JL, Mackie GA (1993) RNase E activity is conferred by a single polypeptide: Overexpression, purification, and properties of the ams/rne/hmp1 gene product. Proc Natl Acad Sci USA 90:9006–9010. 8. Caruthers JM, Feng Y, McKay DB, Cohen SN (2006) Retention of core catalytic functions by a conserved minimal ribonuclease E peptide that lacks the domain required for tetramer formation. J Biol Chem 281:27046–27051. 9. Lin-Chao S, Wong TT, McDowall KJ, Cohen SN (1994) Effects of nucleotide sequence on the specificity of rne-dependent and RNase E-mediated cleavages of RNA I encoded by the pBR322 plasmid. J Biol Chem 269:10797–10803. 10. McDowall KJ, Lin-Chao S, Cohen SN (1994) A+U content rather than a particular nucleotide order determines the specificity of RNase E cleavage. J Biol Chem 269: 10790–10796. 11. Callaghan AJ, et al. (2005) Structure of Escherichia coli RNase E catalytic domain and implications for RNA turnover. Nature 437:1187–1191. 12. McDowall KJ, Cohen SN (1996) The N-terminal domain of the rne gene product has RNase E activity and is non-overlapping with the arginine-rich RNA-binding site. J Mol Biol 255:349–355. 13. Miczak A, Kaberdin VR, Wei CL, Lin-Chao S (1996) Proteins associated with RNase E in a multicomponent ribonucleolytic complex. Proc Natl Acad Sci USA 93:3865–3869. 14. Py B, Higgins CF, Krisch HM, Carpousis AJ (1996) A DEAD-box RNA helicase in the Escherichia coli RNA degradosome. Nature 381:169–172. 15. Kaberdin VR, Singh D, Lin-Chao S (2011) Composition and conservation of the mRNAdegrading machinery in bacteria. J Biomed Sci 18:23. 16. Miczak A, Srivastava RA, Apirion D (1991) Location of the RNA-processing enzymes RNase III, RNase E and RNase P in the Escherichia coli cell. Mol Microbiol 5:1801–1810. 17. Liou GG, Jane WN, Cohen SN, Lin NS, Lin-Chao S (2001) RNA degradosomes exist in vivo in Escherichia coli as multicomponent complexes associated with the

cytoplasmic membrane via the N-terminal region of ribonuclease E. Proc Natl Acad Sci USA 98:63–68. Khemici V, Poljak L, Luisi BF, Carpousis AJ (2008) The RNase E of Escherichia coli is a membrane-binding protein. Mol Microbiol 70:799–813. Fujiki Y, Hubbard AL, Fowler S, Lazarow PB (1982) Isolation of intracellular membranes by means of sodium carbonate treatment: Application to endoplasmic reticulum. J Cell Biol 93:97–102. Gerlach H, et al. (2010) HIV-1 Nef membrane association depends on charge, curvature, composition and sequence. Nat Chem Biol 6:46–53. Wang SX, Sun YT, Sui SF (2000) Membrane-induced conformational change in human apolipoprotein H. Biochem J 348:103–106. Fasman GD, ed (1996) Circular Dichroism and the Conformational Analysis of Biomolecules (Plenum Press, New York). Kentsis A, et al. (2001) The RING domains of the promyelocytic leukemia protein PML and the arenaviral protein Z repress translation by directly inhibiting translation initiation factor eIF4E. J Mol Biol 312:609–623. Dolinsky TJ, et al. (2007) PDB2PQR: Expanding and upgrading automated preparation of biomolecular structures for molecular simulations. Nucleic Acids Res 35(Web Server issue):W522–525. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA (2001) Electrostatics of nanosystems: Application to microtubules and the ribosome. Proc Natl Acad Sci USA 98: 10037–10041. Koslover DJ, et al. (2008) The crystal structure of the Escherichia coli RNase E apoprotein and a mechanism for RNA degradation. Structure 16:1238–1244. Huang H, Liao J, Cohen SN (1998) Poly(A)- and poly(U)-specific RNA 3′ tail shortening by E. coli ribonuclease E. Nature 391:99–102. Lin-Chao S, Cohen SN (1991) The rate of processing and degradation of antisense RNAI regulates the replication of ColE1-type plasmids in vivo. Cell 65:1233–1242. McDowall KJ, Kaberdin VR, Wu SW, Cohen SN, Lin-Chao S (1995) Site-specific RNase E cleavage of oligonucleotides and inhibition by stem-loops. Nature 374:287–290. Sui SF (2000) Membrane-induced conformational change of proteins. Adv Colloid Interface Sci 85:257–267. Nemat-Gorgani M, Dodd G (1977) The interaction of phospholipid membranes and detergents with glutamate dehydrogenase. Eur J Biochem 74:139–147. Vécsey-Semjén B, Lesieur C, Möllby R, van der Goot FG (1997) Conformational changes due to membrane binding and channel formation by staphylococcal alphatoxin. J Biol Chem 272:5709–5717. McLaughlin S (1989) The electrostatic properties of membranes. Annu Rev Biophys Biophys Chem 18:113–136. Callaghan AJ, et al. (2004) Studies of the RNA degradosome-organizing domain of the Escherichia coli ribonuclease RNase E. J Mol Biol 340:965–979. Whitmore L, Wallace BA (2008) Protein secondary structure analyses from circular dichroism spectroscopy: Methods and reference databases. Biopolymers 89:392–400.

7024 | www.pnas.org/cgi/doi/10.1073/pnas.1120181109

18. 19.

20. 21. 22. 23.

24.

25.

26. 27. 28. 29. 30. 31. 32.

33. 34. 35.

Murashko et al.