Advertisement
JBC

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M507146200 on August 18, 2005

J. Biol. Chem., Vol. 280, Issue 43, 36397-36408, October 28, 2005
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
280/43/36397    most recent
M507146200v1
Right arrow Submit a Letter to Editor
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Greive, S. J.
Right arrow Articles by von Hippel, P. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Greive, S. J.
Right arrow Articles by von Hippel, P. H.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Assembly of an RNA-Protein Complex

BINDING OF NusB AND NusE (S10) PROTEINS TO boxA RNA NUCLEATES THE FORMATION OF THE ANTITERMINATION COMPLEX INVOLVED IN CONTROLLING rRNA TRANSCRIPTION IN ESCHERICHIA COLI*

Sandra J. Greive, August F. Lins1, and Peter H. von Hippel2

From the Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403

Received for publication, June 30, 2005 , and in revised form, August 17, 2005.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Analytical ultracentrifugation and fluorescence anisotropy methods have been used to measure the equilibrium parameters that control the formation of the core subcomplex of NusB and NusE proteins and boxA RNA. This subcomplex, in turn, nucleates the assembly of the antitermination complex that is involved in controlling the synthesis of ribosomal RNA in Escherichia coli and that also participates in forming the N protein-dependent antitermination complex in lambdoid phage synthesis. In this study we determined the dissociation constants (Kd values) for the individual binary interactions that participate in the assembly of the ternary NusB-NusE-boxA RNA subassembly, and we showed that multiple equilibria, involving both specific and nonspecific binding, are involved in the assembly pathway of this protein-RNA complex. The measured Kd values were used to model the in vitro assembly reaction and combined with in vivo concentration data to simulate the overall control of the assembly of this complex in E. coli at two different cellular growth rates. The results showed that at both growth rates assembly proceeds via the initial formation of a weak but specific NusB-boxA complex, which is then stabilized by NusE binding. We showed that NusE also binds nonspecifically to available single-stranded RNA sequences and that such nonspecific protein binding to RNA can help to regulate crucial interactions in the assembly of the various macromolecular machines of gene expression.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Macromolecular machines involving multiple protein and nucleic acid components drive and regulate many important cellular processes. These include gene expression, protein synthesis and trafficking, as well as signal transduction and many other processes. The assembly of the relevant macromolecular complexes is often controlled by additional regulatory factors that modulate the effective binding affinity of the central components of a self-assembling macromolecular machine. Another regulatory strategy involves controlling the local concentration of a critical component of the complex, which can perturb crucial binding interactions by manipulating mass action effects. In principle this strategy uses changes in the concentration parameter as a "switch" to turn the targeted assembly reaction on or off; a good example of this strategy involves using the control of the synthesis (and thus of the free concentration) of the single-stranded DNA-binding protein of T4 (gp32) to regulate the assembly of the functional T4 DNA replication complex (1). Furthermore, if this component is "tethered" in the vicinity of the assembly to be regulated (e.g. by cis-RNA looping (2, 3)), this switch can be used to confine the regulation of concentration change only to nearby target loci on the genome and not to others.

Characterizing the binding interactions between particular components of a macromolecular complex and elucidating the mechanisms of assembly is a challenging, but vital, part of understanding how cellular processes are regulated. This problem is compounded when RNA is one of the binding partners, because the same RNA molecule can assume different conformations, and proteins that interact with RNA can bind specifically to one particular sequence or conformation and may also bind nonspecifically to other sequences or conformations. RNA-protein complex assembly often begins with one or more binding events that nucleate complex formation by either increasing or decreasing the affinity of other factors for elements of the complex, thus regulating the various competing reaction pathways that function in processes such as RNA transcription, RNA processing, or protein synthesis.

One such regulatory RNA-protein subassembly is the "core" boxA complex that is involved in antitermination control of the transcription of ribosomal RNA genes (rrn) in bacteria (4, 5). RNA transcription can be regulated by specific signal sequences in the DNA, in the nascent RNA, and by protein factors that interact with the RNA polymerase complex. Once initiated, RNA transcription may also be suspended by pause, backtracking, or arrest signals (see Refs. 6 and 7). Termination occurs (generally at pause sites) when the elongation complex dissociates; the RNA transcript is released, and the RNA polymerase is recycled to initiate another round of transcription. In bacteria, the elements that regulate reaction pathways in transcription often involve signals in the nascent RNA and may control the efficiency of either Rho-dependent or intrinsic terminators or both.

The hexameric Rho helicase binds to cytosine-rich sequences and, driven by ATP hydrolysis, translocates 5' -> 3' along the nascent RNA strand to "catch up with" elongation complexes that have been slowed by pause signals. Rho then uses its RNA-DNA helicase activity to remove the RNA from the complex, causing the polymerase also to release the template DNA and thus to terminate transcription (reviewed in Ref. 8). Intrinsic terminators are characterized by a stable terminator hairpin stem-loop structure that forms in the nascent RNA immediately adjacent to the polymerase. This hairpin is followed by a run of uridine residues that, with their complementary deoxyadenine partners, form a particularly weak RNA-DNA hybrid at the terminator. These RNA sequence elements work together to destabilize the elongation complex and bring about transcription termination (9-11). In addition transcript release at these terminators competes with further elongation, is regulated by antitermination factors that involve signals in the RNA that loop back and bind to the RNA polymerase, either directly or through protein co-factors, and modulates the efficiency of vicinal terminators. How this modulation of termination is brought about has been the subject of numerous studies (12, 13).



View larger version (29K):
[in this window]
[in a new window]
  		FIGURE 1.
Assembly of the antitermination complex. A, possible pathway for complex assembly involving a sequential pair of interactions in which NusB and NusE first interact to form a protein heterodimer, which then binds specifically to boxA RNA. B, adapted with permission from Ref. 7; left panel, schematic of the boxA RNA-NusB-NusE core complex that is required for long range {lambda} antitermination of RNA transcription during lytic growth. The final complex also contains NusG and, along with N and NusA bound to boxB RNA, interacts with the elongating RNA polymerase via RNA looping to reduce transcript termination at defined intrinsic and Rho-dependent terminators. Right panel, the ribosomal RNA system requires boxA RNA to be bound by NusB and NusE to recruit NusA, NusG, and ribosomal proteins such as S4 to reduce termination of rRNA transcripts at defined Rho-dependent terminators.

 
Antitermination of Rho-mediated termination of rrn genes is mediated by boxA, which comprises a specific sequence of 11 nucleotide residues located in the leader region of the 16 S RNA and in the spacer region between the 16 S and the 23 S RNA genes on the template (14-16). The boxA sequence is also required for long range N-mediated antitermination during the lytic growth of lambdoid phages (17). It has been proposed that a heterodimer of protein co-factors NusB and NusE (ribosomal protein S10; see Fig. 1A) binds to boxA RNA, along with NusA, NusG, and ribosomal proteins (S4, S2, L4, and L13) in controlling rrn transcription, and with N and boxB RNA in regulating lambdoid phage development (Fig. 1B). These complexes then interact with RNA polymerase to greatly reduce the local efficiency of transcript termination (4, 18-20). Binding of NusB and NusE to boxA RNA has been considered to be a central nucleation event for the assembly of the antitermination core complex, because this interaction is needed to permit the assembly of the other components of both the rrn and the phage {lambda} antitermination systems (21, 22). In this study we characterize the interaction of NusB and NusE with boxA RNA, and we explore the mechanism of assembly of this initial (core) antitermination complex.

Earlier studies (4, 23, 24) of this assembly reaction, carried out using gel shift and surface plasmon resonance assays, have been interpreted in terms of a sequential two-step assembly mechanism for the formation of the NusB-NusE-boxA RNA complex (Fig. 1A). In these studies it was assumed that a NusB-NusE heterodimer forms first, and this heterodimer then binds to the rrn boxA RNA sequence. Based on this model and these data, an equilibrium dissociation constant (Kd)of ~0.1 to 1 µM was estimated for the binding of the protein heterodimer to the boxA RNA sequence. No binding of these proteins to {lambda} nutR boxA RNA was observed in the gel shift experiments (4).

In this study we have used sensitive fluorescence anisotropy and analytical ultracentrifugation methods to characterize the interaction of NusB and NusE with the rrn and {lambda} boxA RNAs, as well as to monitor the two-component assembly of the protein heterodimer itself. Our results show that this assembly process is more subtle than previously appreciated, involving competing sets of binary interactions of the components that are characterized by similar dissociation constants. We have shown that the direct binding of NusE to boxA is nonspecific but results in the stabilization of the specific binding of NusB to this RNA. These findings provide a more detailed view of the assembly pathway that leads to the formation of the boxA-containing core antitermination complex and (as considered under the "Discussion") provide insights into how this pathway might be controlled to regulate antitermination in vivo in both the transcription of the rrn genes and in the lytic phase of phage {lambda} synthesis in E. coli.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Buffers and Reagents—All fluorescence anisotropy titrations and analytical ultracentrifugation analyses were conducted in buffer A (25 mM Hepes, 100 mM potassium acetate, pH 7.5), unless otherwise stated. DNA oligonucleotides used for PCR were synthesized by IDT (Coralville, IA), and RNA oligonucleotides used in fluorescence anisotropy and analytical ultracentrifugation experiments were ordered from Dharmacon (Lafayette CO; TABLE ONE), de-protected according to the manufacturer's instructions, and purified by urea 20% PAGE, followed by further purification over a C18 SepPak column (Waters). RNA concentrations were determined using UV absorbance measurements at 260 nm. NusB and NusE were purified as described below, and protein concentrations were determined via absorbance at 280 nm, on molar extinction coefficients of 14,650 and 1,280 cm-1 based M-1, respectively (25). The concentrations of NusB and NusE used in this study are within 10% of the stated values. Prior to use, RNA and proteins were diluted or dialyzed into buffer A, as stated below. Tobacco etch virus (TEV)3 protease used in the purification of NusB (below) was expressed and purified according the method of Lucast and co-workers (26).


View this table:
[in this window]
[in a new window]
  		TABLE ONE
Sequences of the 5'-fluorescein-labeled RNA oligomers used for fluorescence anisotropy experiments

 

Cloning, Expression, and Purification of NusB and NusE—DNA sequences coding for NusB and NusE were amplified by PCR using Pfu DNA polymerase (Stratagene, La Jolla CA) from E. coli K12 genomic DNA. The fragment coding for NusE was inserted into pET11a (Promega, Madison, WI) at BamHI and HindIII restriction sites, whereas the NusB fragment was inserted (at BamHI and XhoI restriction sites) into the pBH expression vector (a kind gift from the Prehoda Lab, Institute of Molecular Biology, University of Oregon), which codes for an N-terminal hexahistidine tag followed by a TEV protease site. Plasmids were transformed into BL21 (DE3) cells, and protein expression was induced with 1 mM isopropyl 1-thio-{beta}-D-galactopyranoside for 4 h when cell growth had reached an A600 of 0.5. Cells containing expressed protein were harvested by centrifugation at 6000 rpm for 10 min in an F6B Sorvall rotor in a Beckman J-6M centrifuge. The resulting cell pellets were stored at -80 °C.

The cell pellets were resuspended in buffer A (25 mM Hepes, pH 7.5, 100 mM potassium acetate) containing EDTA-free protease inhibitor mixture (Roche Applied Science) and 1 mM dithiothreitol (for NusE) and lysed by three passes through a French press at 1000 p.s.i. The resulting lysates were centrifuged at 10,000 x g for 30 min to remove cell debris, and the supernatants were processed as described below. Lysates containing NusE protein were precipitated overnight at 4 °C with an equal volume of 40% ammonium sulfate in buffer A (20% final concentration) added dropwise while stirring. The precipitated protein was pelleted by centrifugation at 10,000 x g for 30 min, washed twice with buffer A, and dissolved in buffer B (6 M urea, 25 mM Hepes, 50 mM potassium acetate, pH 7.5). The protein was purified over a cation exchange column (MonoS; Bio-Rad), washed with 2 column volumes of buffer B, and eluted in 3-ml fractions with a 50-300 mM potassium acetate gradient in buffer B. Fractions containing NusE were identified by SDS-PAGE, pooled, dialyzed against buffer B, and passed over a second MonoS column as described above. Fractions containing purified NusE from the second column were pooled and dialyzed against a 6-0 M urea gradient in buffer A over 24 h, after which precipitated NusE was removed by centrifugation and aliquots of the NusE protein snap-frozen and stored at -80 °C.

NusB lysates were passed over a Talon resin column (BD Biosciences), washed with 10 column volumes of buffer A, followed by 10 column volumes of buffer A containing 20 mM imidazole, and eluted with a 20-250 mM imidazole gradient. Fractions containing NusB were identified by SDS-PAGE analysis, pooled, dialyzed against buffer A, and digested with TEV protease overnight at 16 °C, followed by dialysis against buffer A. The NusB protein was passed over another Talon column to remove uncleaved protein and the protease, dialyzed against buffer A. Aliquots were then snap-frozen and stored at -80 °C.

Analytical Ultracentrifugation—Sedimentation velocity and equilibrium experiments were carried out in a Beckman XL-I analytical ultracentrifuge, and the data were subjected to van Holde-Weischet analysis using Ultrascan (B. Demeler, the University of Texas Health Science Center, San Antonio (27)), Sedfit, and Sedphat (P. Schuck, National Institutes of Health (28-30)) programs for velocity runs and the Winnl106 program (J. Lary, National Ultracentrifuge Facility) for equilibrium runs. Fits using Sedfit and Sedphat were considered satisfactory if the root mean square deviation was less than 0.007, and the residuals were completely random and represented less than 10% of the signal.

The individual proteins, NusB (4.5 µM) and NusE (~20 µM; maximum concentration in phosphate buffer), and their various complexes with one another and with boxA RNA, including NusB-NusE (~5 µM each), NusB-boxA (9 and 2 µM, respectively), and NusB-NusE-boxA (5.4, 6, and 2 µM, respectively), were analyzed by sedimentation velocity methods. Samples were equilibrated by microdialysis with a 5000 molecular weight cut-off membrane with buffer A or buffer C (25 mM potassium phosphate, 200 mM potassium acetate, pH 7) for NusE alone and were inserted into the sample chamber of a two-channel ultracentrifuge cell (buffer A, or buffer C for NusE alone, was inserted into the buffer chamber) using quartz windows and an aluminum centerpiece. Cells were placed in an An60 titanium rotor and centrifuged at 60,000 rpm for 5 h at 20 °C and were scanned continuously using absorbance at 229 and 280 nm for proteins, and 260 nm for RNA. For sedimentation equilibrium analysis, NusB (4.5 µM) in buffer A and buffer A were placed in the sample and buffer chambers, respectively, of a 6-well equilibrium centerpiece, and the cell was assembled with quartz windows. Three scans at 280 nm in an An60 Ti rotor were conducted after equilibrium was reached at each of the following speeds: 10,000, 15,000, 27,000, 31,000, and 40,000 rpm at 20 °C. Data collected at 31,000 rpm were analyzed with the Winnl06 program, assuming a single globular species to determine {sigma} values that were then used to calculate molecular weights.

Mass Spectroscopy—The NusB and NusE proteins were also examined by mass spectroscopy (Matthews Laboratory, Institute of Molecular Biology, University of Oregon), using a Voyager Biospectrometry MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA). Proteins were prepared for analysis in a matrix of sinapinic acid (Sigma), and NusE was desalted before analysis with a ZipTip C18 column (Millipore, Billerica, MA).

Fluorescence Anisotropy—RNA sequences for rrn boxA, {lambda} nutR boxA, and reverse rrn boxA (TABLE ONE), 5'-labeled with 5-carboxy fluorescein, were used for fluorescence anisotropy experiments in buffer A unless otherwise specified. Unlabeled rrn boxA RNA was used in the competition experiments. Proteins (in buffer A) were titrated into a 0.5-cm quartz cuvette (Starna, Atascadero, CA) containing fluorescent RNA and, where indicated, protein as well, and the fluorescence anisotropy at each point was determined at 25 °C using an L-format Jobin-Yvon Horiba Fluoromax fluorimeter.

Data Fitting and Simulation—Anisotropy data were fit to a two-component binding equation (Equation 1) to determine the equilibrium dissociation constant (Kd) by using the Levenberg-Marquardt algorithm and error analysis in the program Profit (Quantumsoft, Uetikon am See, Switzerland) and the following fitting Equation 1,

(Eq. 1)

where A is the anisotropy; Ao is the initial anisotropy, and Po and RNAo are the total protein and RNA concentrations, respectively. Complex equilibria in multiple data sets were fit by global analysis using the Levenberg-Marquardt algorithm in the Dynafit program (Biokin, Pullman, WA (31)), based on the following dissociation constants and equilibrium relationships as shown in Equations 2-7,

(Eq. 2)

(Eq. 3)

(Eq. 4)

(Eq. 5)

(Eq. 6)

(Eq. 7)

where B represents NusB; E represents NusE; (ER)n represents the non-specific binary complex of E and RNA, and R represents the RNA oligomer involved. A range of initial values for each parameter, including those determined experimentally, was used in multiple rounds of fitting to ensure that the final values calculated for K2 and Kber represented the best fit and confidence intervals. The information gained from the global fits were used to simulate boxA assembly kinetics under various conditions using the Berkeley Madonna program (R. Macey and G. Oster, University of California, Berkeley), and data from both analyses of the complex equilibria were plotted using the Profit program.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Association States of NusB and NusE—N-terminal hexahistidine-tagged E. coli NusB protein was purified by chromatography using TALON resin, and the His tag was removed by cleavage with TEV protease (see "Materials and Methods"). This protease leaves two "extra" N-terminal residues (serine and glycine) on purified NusB protein, increasing the predicted molecular mass slightly from 15.67 to 15.8 kDa. However, the function of this protein, which is required (along with NusE and the other components of the {lambda} and rrn antitermination complexes) to increase the transcription rate from long templates (data not shown) and {lambda} N-mediated antitermination activity in vitro,4 was not changed by this modification.

Sedimentation velocity experiments with purified NusB protein, using the van Holde-Weischet analysis, revealed that the protein exists as a monomer in solution and is characterized by a corrected sedimentation coefficient (s20,w) of 1.7 S and a relatively spherical shape (Fig. 2A, note the absence of nonideality in these data). Analysis with the Sedfit program also yielded an s20,w value of 1.7 S and an estimated molecular mass of 16 kDa, as well as a frictional ratio of 1.35, indicative of a globular shape (Fig. 2, B and C). The monomer molecular mass of NusB was confirmed by sedimentation equilibrium runs at 31,000 rpm, using absorbance optics at 280 nm. Winnonlin software was used to analyze the equilibrium data. Assuming a single globular species yielded a {sigma} factor of 2.179 for NusB, corresponding to a molecular mass of 16.2 kDa (Fig. 2D). This value was also confirmed by MALDI-TOF mass spectrometry (data not shown) and is consistent with previous studies and with a monomer of 15.8 kDa based on the amino acid residue sequence (24, 32).



View larger version (35K):
[in this window]
[in a new window]
  		FIGURE 2.
Sedimentation velocity and equilibrium experiments with free NusB. A, van Holde-Weischet analysis of sedimentation velocity data at 280 nm and 60,000 rpm yields a sedimentation coefficient (s20,w value) of 1.7 S and shows that the NusB protein sediments as a globular monomer species. B and C, further analysis of these data with Sedfit (28) also yields a plot of c(s) versus s20,w and shows that the main component in this sedimentation profile sediments with an s20,w of 1.7 S, which is equivalent to a molecular mass of 16 kDa for a (assumed globular) NusB monomer in a c(M) versus s20,w plot. D, a sedimentation equilibrium plot of NusB protein concentration versus radial position (R) in the ultracentrifuge cell. Sedimentation was at 31,000 rpm, and the data (circles) from scans at 280 nm were fit to a single species model using the Winnonlin program. A {sigma} value of 2.18 and a mass of 16.2 kDa were obtained.

 
E. coli NusE was purified as described above. Despite previous reports that this protein is insoluble in its native form (24, 33), we note that concentrations of NusE up to a concentration of ~20 µM could be obtained by slow dialysis through a urea gradient (from 6 M to native buffer). Previous data from proteolytic digests and chemical degradation experiments had suggested that free E. coli NusE might exist in a relatively unstructured form (34), and because NusE from Mycobacterium tuberculosis has also been shown (by CD spectroscopy and NMR) to be relatively unstructured (even when bound to NusB (35)), it seems likely that E. coli NusE is also largely unstructured. It also seems likely that NusE functions within the core complex in a manner similar to that of {lambda} N protein in the N-dependent antitermination system. N has also been shown to be largely unstructured in free solution and to assume a partially helical structure (in its N-terminal domain) on binding to its specific (boxB) RNA target. N also binds to its various other reaction partners (including NusA and RNA polymerase) in a modular fashion, using its other (initially unstructured) sequence domains (36, 37).

The sedimentation properties of free NusE in phosphate buffer (see "Materials and Methods") were determined from sedimentation velocity experiments analyzed with Sedfit. The data were noisy, due both to the strong tendency of the free protein to aggregate and to its paucity of aromatic amino acid residues and small size. However, despite these limitations, the data collected at 229 nm could be reasonably well fit (with a root mean square deviation of 0.03 and residual values somewhat greater than 10% of the signal to obtain an s20,w of 0.5 S (Fig. 3A) and a frictional ratio of 2.6 (consistent with a largely random coil structure)). A molecular mass of 10.5 kDa (Fig. 3B) was estimated from these data, consistent with the existence of this protein as a monomer in free solution. This estimated molecular mass was confirmed by mass spectrometry, which yielded a mass of 11.8 kDa for the NusE protein (data not shown), consistent with expectations from the amino acid residue composition. Further analysis of NusE alone, using native PAGE or CD spectroscopy, was rendered difficult by the low solubility of the free protein.

Association States of the Ternary NusB-NusE-boxA RNA Complex and Its Binary Components—Because it appears that NusB interacts individually with both NusE and boxA RNA, as well as participating in a ternary core antitermination complex with both components (4, 23, 24), we have explored the association states of the constituent binary complexes and the final ternary complex by sedimentation velocity techniques. Analysis (with the Sedfit Program) of experiments conducted at 280 nm with an equimolar mixture of NusB and NusE yielded two peaks with corrected sedimentation coefficients (s20,w) of 1.7 S and 2.2 S for NusB and the binary complex, respectively (Fig. 4A). An average frictional ratio of 1.29 was obtained for the binary protein complex, suggesting that this complex (the 2.2 S peak) is, like NusB alone, roughly globular in shape. Converting the c(s) profile into c(M) data revealed an estimated molecular mass of ~27 kDa for the 2.2 S species (Fig. 4B), suggesting that the NusB and NusE proteins do interact to form a heterodimer with a stoichiometry of 1:1 (24).



View larger version (18K):
[in this window]
[in a new window]
  		FIGURE 3.
Analysis of free NusE by sedimentation velocity analytical ultracentrifugation. A, free NusE was centrifuged at 60,000 rpm at 20 °C and scanned at 229 nm continuously during the experiment. Analysis of the data with Sedfit using the c(s) model yielded a s20,w of 0.5 S. B, conversion of the c(s) profile into c(M) using Sedfit (and assuming a globular protein) resulted in an estimated mass of 10.5 kDa.

 
A binary mixture of NusB protein and boxA RNA was also analyzed by sedimentation velocity at 60,000 rpm and 20 °C, using absorbance measurements at 260 nm and a mixture of fluorescently labeled rrn boxA RNA and a 5-fold excess of NusB (Fig. 4C). Three peaks were observed in these runs, with s20,w values of 1.49 S and 2.33 S (the major peak) likely corresponding to free RNA and to a (1:1) NusB-RNA dimer, respectively. Statistical analysis with the fitting program revealed that the 3.03 S peak was not real. Conversion of these c(s) data into estimated molecular weights requires the use of the correct value of partial specific volume () for each complex. RNA molecules in sodium salt have been reported previously to have values ranging from 0.5 to 0.55 ml/g (38), with 0.55 ml/g usually being taken as the "standard" value. A of 0.59 ml/g has been reported for RNA in potassium salts (39), presumably reflecting the contribution of cation condensation to the measured values of this parameter. We used this value in our analyses (with the Sedphat program) of complexes containing RNA. The so-called "hybrid continuous and discrete species" version of this program allows the frictional coefficients for each species to be fit individually, with the value for the second species then estimated from the fit (30). The frictional coefficient for the species with an s20,w value of 2.3 S was found to be 1.55, consistent with a somewhat elongated shape. The molecular mass for this species was estimated to be 23 kDa, corresponding to the predicted mass of 22.6 kDa for a NusB-boxA RNA complex with 1:1 stoichiometry.

The stoichiometry of the ternary NusB-NusE-boxA RNA complex was explored using a 3-fold excess of an equimolar mixture of NusB and NusE relative to the limiting concentration of fluorescently labeled rrn boxA RNA. Analysis of sedimentation velocity data measured at 260 nm revealed peaks with s20,w values of 1.2 S, 2.1 S (the major peak), and 2.51 S. We suggest that these peaks correspond to free boxA RNA, a binary NusB-boxA RNA complex, and a ternary (1:1:1) NusB-NusE-boxA RNA complex, respectively (Fig. 5). The shift in the apparent s20,w of the RNA species monitored at 260 nm, relative to its s20,w value in the previous experiment, indicates that the RNA is in rapid equilibrium with the protein components in this and in the preceding experiment. Fitting these data with the hybrid continuous and discrete species version of Sedphat yielded a mass of 23 kDa for the 2.1 S species, consistent with that determined in the experiment shown in Fig. 4C and also confirming a 1:1 stoichiometry for the NusB-boxA RNA complex.

This 2.1 S peak is somewhat broader in the c(s) distribution plot (Fig. 5) than the 2.3 S peak obtained in the previous experiment (Fig. 4C), suggesting that this intermediate species was also engaged in rapid (relative to the speed of sedimentation) dissociation and re-association during the ultracentrifuge run. The same Sedphat analysis described above estimated a mass of 33.4 kDa for the 2.51 S peak (Fig. 5), which is comparable with the expected 34.56 kDa for the NusB-NusE-boxA RNA ternary complex and confirms that the stoichiometry of the NusB-NusE-boxA RNA complex is indeed 1:1:1.



View larger version (23K):
[in this window]
[in a new window]
  		FIGURE 4.
NusB binds to NusE and to boxA RNA with a 1:1 stoichiometry. A and B, an equimolar mixture of NusB and NusE was analyzed by sedimentation velocity (using absorption optics at 280 nm at 60,000 rpm) with the Sedfit program. The c(s) distribution plot (A) shows a peak representing a species with an s20,w value of 2.2 S, and the c(M) distribution (B) reveals a peak at 27 kDa, equivalent to the expected molecular mass of a NusB-NusE heterodimer. C, sedimentation velocity experiment with a mixture of fluorescently labeled rrn boxA RNA and a 10-fold excess of NusB was monitored at 260 nm, and the data were analyzed with Sedfit. The c(s) distribution reveals two major peaks of 1.49 S and 2.33 S. Conversion of these data into molecular mass using Sedphat and partial specific volume () of 0.59 ml/g for RNA results in an estimated mass of 21.5 kDa for the 2.3 S peak. This mass is equivalent to that expected for a binary NusB-RNA complex and demonstrates that this interaction has a 1:1 stoichiometry.

 
NusB Binds Specifically to boxA RNA—RNA binding interactions were characterized by fluorescence anisotropy using three RNA oligomers labeled at their 5'-ends with a fluorescein adduct (see TABLE ONE). These oligomers correspond to the rrnG boxA sequence (rrn boxA) that was used previously by Squires and co-workers (20) as a competitor in in vitro antitermination assays, a similar sequence containing the sequence of the {lambda} nutR boxA ({lambda} boxA), and a sequence corresponding to the exact reverse of the rrn boxA (reverse boxA). The reverse boxA sequence has been shown previously not to function in rrn antitermination experiments (20) nor to bind protein in gel shift assays (4).



View larger version (8K):
[in this window]
[in a new window]
  		FIGURE 5.
The core complex shows a 1:1:1 stoichiometry. rrn boxA RNA was added to a 3-fold excess of an equimolar mixture of NusB and NusE and analyzed by sedimentation velocity analytical ultracentrifugation at 60,000 rpm and 260 nm. The data were analyzed with Sedfit. A c(s) distribution showing three peaks with s20,w values of 1.2 S, 2.1 S, and 2.51 S. Conversion of the c(s) data into molecular mass data with the hybrid continuous and discrete model of the Sedphat program, using a value of 0.59 ml/g for free RNA determined by using molecular mass values of 23 and 33.4 kDa for the 2.1 S and 2.51 S species, respectively. The latter is consistent with a NusB-NusE-RNA complex with a stoichiometry of 1:1:1.

 
No change in anisotropy was observed when 50 nM of reverse boxA RNA were titrated with NusB (Fig. 6), indicating that no interaction occurs between these two molecules under our conditions. However, titration of 50 nM rrn or {lambda} boxA RNA with NusB resulted in an increase in fluorescence anisotropy for both RNAs (Fig. 6), and when fit to a two-component binding isotherm ("Materials and Methods," see Equation 1) yielded dissociation equilibrium constants (Kd) of ~850 and ~600 nM for the NusB interaction with rrn RNA and with {lambda} boxA RNA, respectively. We note that these values are 10-fold smaller (i.e. tighter affinity) than the Kd values estimated by Luttgen et al. (24) from surface plasmon resonance experiments. Our results show that NusB binds specifically to both the rrn and the {lambda} boxA RNA sequences and shows no nonspecific RNA binding in the concentration range used in this study.

NusE Binds RNA Nonspecifically—Preliminary experiments in which changes in the intensity of the Rayleigh scattering peak were observed in the fluorimeter on adding NusE to various solutions indicated that light scattering by aggregates of NusE protein is significant at free NusE concentrations exceeding ~1 µM (data not shown). Because scattered light is known to artificially increase anisotropy values (40), data at free NusE concentrations in excess of 1 µM must be interpreted with caution. Fluorescence anisotropy titrations with 50 nM concentrations of rrn boxA and reverse RNA were performed with NusE protein. The resulting binary RNA-NusE complexes were soluble, and these data show that NusE binds equally well to reverse RNA and to rrn boxA RNA (Fig. 7A) under these conditions. If there is a specific component in the binding of NusE to boxA RNA in these experiments, it was too small to detect, or it was masked by the nonspecific interaction. We note that this binary (nonspecific) affinity of NusE for boxA RNA is relatively weak and was also not detected by earlier workers using gel shift assays (4).



View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 6.
NusB binds to boxA RNA specifically. Purified NusB was titrated with 50 nM reverse boxA RNA (squares), rrn boxA RNA (circles), or {lambda} boxA RNA (diamonds). All the RNA oligomers were 5'-labeled with fluorescein. The anisotropy data were fit using a two-component binding equation ("Materials and Methods," Equation 1), and Kd values of 850 and 600 nM were determined for the rrn (solid line) and {lambda} (dashed line) RNA sequences, respectively.

 



View larger version (29K):
[in this window]
[in a new window]
  		FIGURE 7.
NusE binds to RNA nonspecifically. A, fluorescence anisotropy was used to monitor the titration of NusE with 50 nM of 5'-fluorescein-labeled rrn boxA RNA (circles) or reverse boxA RNA (squares). B, the titrations for both rrn (circles) and reverse boxA (squares) RNA were continued until a total of 8 µM NusE was added. The data for the reverse RNA were fit by using a two-component binding isotherm (dashed curve). A Kd of ~3 µM was measured for this nonspecific binding interaction. The anisotropy values of NusE bound to boxA RNA continue to rise and may be a result of a both specific and nonspecific interactions with RNA creating a network of protein-RNA molecules (see text).

 
We have observed that when the titration of NusE with either reverse boxA RNA or with rrn boxA RNA was continued up to a concentration of ~8 µM, the trends observed for the two sets of data appear to diverge (Fig. 7B). The anisotropy for the reverse boxA sequence reaches a plateau and can be fitted with a two-component binding isotherm (Equation 1) and a Kd value of ~3 µM, assuming that each RNA molecule was bound nonspecifically by a single NusE protein. This assumption, if borne out, means that NusE binds nonspecifically to these RNA oligomers with a site size between 11 and 20 nucleotides. The data for NusE binding to rrn boxA RNA does not reach a plateau but rather continues to increase linearly. This may reflect a combination of specific and nonspecific binding that results in more than one protein bound to a single RNA molecule and/or in more than one RNA molecule bound to a single NusE protein. Such a network of interactions, if present, would create large complexes with increased apparent anisotropy values as a result of scattered light. These data were not analyzed further because of protein insolubility problems.



View larger version (26K):
[in this window]
[in a new window]
  		FIGURE 8.
NusB binding to boxA RNA in the presence of NusE. A, fluorescence anisotropy data for the titration of 50 nM 5'-fluorescein-labeled rrn (circles), {lambda} (diamonds), or reverse (squares) boxA RNA with a premixed solution containing equimolar concentrations of NusB and NusE. Curves were fit to the {lambda} boxA data (solid line) and the rrn boxA data (dashed line) using a two-component binding isotherm (Equation 1) to yield apparent Kd values of ~100 and ~200 nM for the {lambda} and rrn boxA RNA complexes, respectively. B, solutions containing 50 nM 5'-fluorescein-labeled rrn boxA RNA (circles) or {lambda} boxA RNA (diamonds), together with 400 nM NusE, were titrated with the NusB protein, and the resulting fluorescence anisotropy data were fit with a two-component binding curve (rrn; solid line and {lambda}; dashed line). C, unlabeled boxA RNA was titrated into a 50 nM solution of equimolar NusE-NusB and 5'-fluorescein-labeled rrn boxA RNA. The data were fit using a two-component binding isotherm.

 
NusE Increases the Affinity of NusB for boxA RNA—The assembly of the core boxA complex was further analyzed by examining the binding of both proteins to boxA RNA. In one series of experiments a 50 nM concentration of rrn RNA, {lambda} boxA RNA, or reverse RNA was titrated with an equimolar mixture of NusB and NusE (Fig. 8A). The resulting data indicated that the NusB-NusE complex binds specifically to boxA RNA. When the data were fit with a two-component binding isotherm (Equation 1), again based on assuming a two-component interaction between a NusB-NusE heterodimer and the boxA RNA, the apparent Kd values obtained were ~100 and ~200 nM for the interactions with rrn and {lambda} boxA RNA, respectively. These data can be interpreted to show that the presence of NusE increases the affinity of NusB for boxA RNA ~10-fold.

This latter point is confirmed by the data of Fig. 8B, which shows an experiment in which NusB was titrated into a 50 nM concentration of either labeled rrn boxA RNA or labeled {lambda} boxA RNA, in the presence of a large excess (400 nM) of NusE (Fig. 8B). Because the concentrations of both NusE (400 nM) and RNA (50 nM) in the cuvette were low relative to the Kd for nonspecific binding of NusE, it is likely that there was not much interaction between NusE and RNA prior to the addition of NusB. This conclusion is consistent with the relatively small difference between the starting anisotropy value of this experiment (Fig. 8B) and that of the titrations of boxA RNA with NusE alone (Fig. 7A). We note that both curves reach their respective plateaus at or below ~200 nM concentrations of added NusB for these titrations containing excess NusE, indicating that a ternary complex is formed and that the binding interactions within this ternary complex are significantly tighter than those for either binary protein-RNA complex alone.

This titration also shows that the ternary complex is stabilized by the presence of excess NusE. Apparent Kd values of 50 nM were determined using a two-component binding isotherm (see "Material and Methods," Equation 1) for the interaction of the putative pre-formed NusB-NusE heterodimer with both the rrn and the {lambda} boxA RNA oligomers. Not only is the affinity of NusB for boxA RNA increased significantly in the presence of NusE, but the presence of excess NusE effectively masks any minor differences in binding affinity between rrn and {lambda} boxA RNA with the putative NusB-NusE heterodimer.

The difference in plateau values observed for rrn boxA and for {lambda} boxA RNA sequences titrated with both NusB and NusE (Fig. 8, A and B) was not seen in the experiments using NusB alone (Fig. 6). This difference in maximal anisotropy may be because of different conformations of the ternary complex, potentially resulting from a weak but specific interaction of NusE with rrn boxA RNA when complexed with NusB, as compared with a nonspecific interaction with {lambda} boxA RNA in the presence of NusB.

In the previous experiment the plateaus were reached at significantly higher concentrations of protein than required in an excess of NusE, consistent also with the notion (see Fig. 8A) that the binary interaction affinities are not very strong and that the ternary reaction can be significantly driven toward completion by the presence of excess NusE protein. This confirms our previous finding that the presence of NusE increases the affinity of NusB for boxA RNA and suggests that the putative heterodimer binds somewhat more tightly to rrn boxA RNA than to {lambda} boxA RNA, as also suggested by the earlier surface plasmon resonance measurements and gel shift assays (4, 24). The observed binding is not because of, or perturbed by, the labeling fluorophore, because unlabeled rrn boxA RNA competes equally well with the fluorescent boxA sequence for binding to NusB and NusE (Fig. 8C).

Assembly of the Core boxA RNA Complex with NusB and NusE Involves Multiple Three-component Reaction Pathways—A simple two-component interaction (i.e. between a "tight" protein heterodimer and an RNA oligomer) should maintain a constant apparent Kd value for all input component concentrations. However, further analysis of the boxA complex assembly using different concentrations (50, 100, 200, 300, 400, and 500 nM) of fluorescently labeled rrn boxA RNA for each titration with premixed equimolar concentrations of NusB and NusE showed that a fit using the two-component binding isotherm (Equation 1) yields Kd values that change with increasing RNA concentration, with the apparent Kd value increasing as the RNA concentration was raised (TABLE TWO). This indicates that the interaction of NusB and NusE with boxA RNA is not a simple two-component interaction and that assembly does not progress exclusively by initial formation of the NusB-NusE heterodimer followed by RNA binding (Fig. 1A).


View this table:
[in this window]
[in a new window]
  		TABLE TWO
Summary of equilibrium dissociation constants (Kd) fit using a two-component binding model for each concentration of rrn boxA RNA

 
Because, in addition to participating in heterodimer formation, NusE binds to RNA nonspecifically, and NusB alone is able to bind boxA RNA specifically, it is possible that the net change in anisotropy observed in these experiments may be due to NusB and NusE binding to boxA RNA separately to form either an intermediate or a competing binary complex, rather than binding as a pre-formed heterodimer prior to final formation of the ternary complex. As the RNA concentration increases, the number of RNA molecules bound by single protein monomers must also increase, contributing to the observed anisotropy change and thus leading to the observed change in the apparent (two-component) Kd value. However, these data demonstrate that NusB binds boxA RNA much more tightly in the presence of NusE than in its absence, and also confirm that the interaction between NusB-NusE and rrn boxA RNA is tighter than the binding of these proteins to {lambda} boxA RNA. This suggests that in the presence of NusB, NusE may bind rrn boxA RNA with a different conformation than when bound to {lambda} boxA RNA. These data also suggest that the mechanism of complex assembly does not proceed only through initial NusB-NusE heterodimer formation, followed by binding to boxA. Rather these results show that the Kd values for the formation of the various possible binary complexes are all relatively similar.

Although the boxA core complex consists of only three components, each is able to interact with multiple partners (Fig. 9 and Equations 2-7). NusE binds to NusB (Fig. 9A, K1), binds specifically (but weakly) to boxA RNA (Fig. 9B, K3), and binds tightly to a binary complex of NusB and boxA RNA (Fig. 9C, K6). NusE also binds nonspecifically to RNA (Fig. 9B, Kns), whereas NusB binds specifically to boxA RNA (Fig. 9C, K5) and to NusE (Fig. 9A, K1), and thus must play a central role in the nucleation of the core complex. Thus boxA RNA is bound nonspecifically by NusE (Fig. 9B, Kns) and specifically by NusB (Fig. 9C, K5) or by a heterodimer of NusB and NusE (Fig. 9A, K2). These multiple interactions make this a difficult assembly system to solve analytically, particularly because the Kd values for the NusB-NusE interaction and any specific NusE-RNA interaction that might be present could not be determined directly in our analyses.



View larger version (8K):
[in this window]
[in a new window]
  		FIGURE 9.
Potential assembly pathways of the NusB-NusE-boxA RNA complex. A, as described in Fig. 1, assembly may proceed through NusB binding to NusE, characterized by dissociation constant K1, followed by binding to boxA RNA (K2). B, NusE is able to bind RNA in a nonspecific manner (Kns) and may also bind boxA RNA specifically (K3), after which NusB may bind (K4) to form the final complex. C, NusB binds specifically to boxA RNA (K5), and this association is tightened in the presence of NusE (K6) to form the core ternary NusB-NusE-boxA RNA complex.

 
The data obtained in the experiment performed with varying boxA RNA concentrations described above (see TABLE TWO) were fit globally to the complex equilibria shown in Fig. 9 using the Dynafit program (Fig. 10) (31). The Kd value for the NusB-NusE complex was estimated to be ~100-200 nM, and the macroscopic dissociation constant Kber was found to be ~5000 nM2. Despite displaying some systematic behavior, the residuals for this fit represent less than 10% of the signal (Fig. 10). However, although the {chi}2 surface areas for these parameters are very broad (data not shown), constraining the parameters (within limits) away from the best-fit value reduced the quality of the fit significantly. The various Kd values determined experimentally and by modeling methods are summarized in TABLE THREE.


View this table:
[in this window]
[in a new window]
  		TABLE THREE
Summary of the equilibrium dissociation constants (Kd) determined by fluorescence anisotropy in this study

 



View larger version (35K):
[in this window]
[in a new window]
  		FIGURE 10.
Global fit of data from titrations with increasing boxA RNA concentrations. Premixed equimolar concentrations of NusB and NusE protein were titrated with 50 (closed circles), 100 (open squares), 200 (closed triangles), 300 (open circles), 400 (closed squares), and 500 nM (open triangles) concentrations of 5'-fluorescein-labeled rrn boxA RNA. The resulting anisotropy data were globally fit to a three-component model, which included the competing nonspecific binding of RNA by NusE, using the Dynafit program (31). The residuals of this global fit, representing less than 10% of the signal, are depicted in the lower panel. A Kd of ~100 nM was estimated for the formation of the NusB-NusE heterodimer.

 
The Assembly Pathways of the Core boxA Complex—The information derived experimentally and theoretically from the global fits (above) was used to simulate the dynamics of boxA complex assembly under various conditions. When NusE was present in initial excess relative to boxA RNA as the concentration of NusB was increased (in experiments such as that described in Fig. 8B), the measured equilibrium constants suggest that assembly of the NusB-NusE complex probably does occur first, followed by binding of the protein heterodimer to boxA RNA (see Fig. 9A and Fig. 11A). However, when the concentration of NusE is limiting, assembly is more likely to proceed via the NusB-boxA RNA pathway (see Fig. 9C and Fig. 11B).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
In Vitro Assembly of the Central Components of the boxA-dependent Antitermination Complex—In this study we have used equilibrium measurements to explore the assembly of the NusB, NusE, and boxA RNA components of the core boxA subassembly. This subassembly serves to nucleate the formation of both the complex that is required for antitermination during transcription of rRNA genes in bacteria and the complex that is involved in the N protein-dependent antitermination events that regulate the lytic growth of lambdoid phages.



View larger version (27K):
[in this window]
[in a new window]
  		FIGURE 11.
Simulation of the concentrations of the boxA complex and assembly intermediates during titrations with NusB. A, the Kd values for the three-component equilibria determined experimentally and theoretically in previous experiments were used to simulate the concentrations (nM) of the boxA complex (BER, solid line) and subassemblies (NusB-boxA RNA, BR, dashed line; and NusB-NusE, BE, dashed/dotted line) during the titration of 50 nM rrn boxA RNA and 400 nM NusE with NusB (Fig. 8B). The concentrations of complexes containing RNA exceeds that of the NusB-NusE subcomplex at all NusB concentrations, consistent with the notion that assembly here proceeds primarily through pre-formation of the NusB-NusE heterodimer. B, the above simulation was repeated for a situation where 100 nM boxA RNA and 10 nM NusE were titrated with NusB. The concentrations (nM) for the core boxA complex (BER, solid line) and subassemblies (NusB-boxA RNA, BR, dashed line; and NusB-NusE, BE, dashed/dotted line) are shown during the course of the titration. In this case, the concentration of NusB-boxA RNA complexes exceeds that of NusB-NusE complexes, suggesting that assembly under these conditions occurs predominantly through specific NusB binding to boxA RNA, followed by NusE binding to stabilize the complex.

 
Rigorous thermodynamic binding assays using fluorescence anisotropy and analytical ultracentrifugation involve fewer assumptions and have greater sensitivity than the gel shift and surface plasmon resonance assays used to measure binding in previous studies (4, 24), and have permitted us to analyze the assembly process as a total system of three separate components involving several constituent binary interactions. In addition, the fact that we have been able to obtain and examine soluble free E. coli NusE, although over a limited concentration range, has been crucial to obtaining estimates of binding parameters for the various equilibria listed in Equations 2-7. These quantitative measurements of the interactions of the individual constituents have allowed us to develop a model for the assembly of the ternary core complex.

In summary, we have confirmed that E. coli NusB exists in free solution as a monomer with a molecular mass of 16 kDa, as determined previously (32), and that NusE is also a monomer and (as expected from amino acid composition data) has a molecular mass of 11.8 kDa. NusB and NusE have been shown to form a 1:1 heterodimer, and stoichiometries of 1:1 and 1:1:1 have been demonstrated for the NusB-boxA RNA and NusB-NusE-boxA RNA complexes, respectively. This study also showed that NusB binds specifically, and with relatively equal affinity, to both rrn RNA and {lambda} boxA RNA (Fig. 6). Furthermore, having purified soluble NusE, and despite the potential complication introduced by scattered light, this study has shown that NusE binds to RNA nonspecifically and with a Kd of ~3 µM and that this protein increases the affinity of NusB for boxA RNA by ~10-fold. The presence of NusE also significantly increases the affinity of NusB for both rrn and {lambda} boxA RNA, yielding apparent Kd values of ~100 and ~200 nM, respectively, for these two RNA ligands.

Finally, this study has shown that the assembly of the boxA-containing core antitermination complex does not proceed simply via a pair of discrete and sequential binding interactions, but rather that a network of alternative interactions is involved that is regulated by the local concentrations of the components and the relative dissociation constants of the binary reaction intermediates. Global fitting of these data to a multiple equilibria model is consistent with the experimentally determined values of these Kd parameters and yields a Kd of ~100 nM for the NusB-NusE interaction, which is consistent with the apparent Kd values observed in the experiments with excess or equimolar NusE. As described below, this model contributes to our understanding of how the assembly of this complex might be regulated in the cell and may also be useful as a paradigm for the quantitative analysis of other RNA-protein complexes.

Physiological Considerations for E. coli and Bacteriophage Development—It is, of course, relevant and interesting to attempt to apply these results to the in vivo situation in E. coli. In this organism the concentrations of host factors required for cell survival, such as ribosomal components and transcription factors, vary greatly with changes in bacterial growth rate (TABLE FOUR) (41, 42)). In simulating the mechanism of ternary complex assembly, we conservatively estimate that only 10% of the total nonspecific mRNA-binding sites not covered by ribosomes is bound by NusE, and that 50% of NusB is unable to participate in the boxA complex. At low growth rates (0.5 doublings/h) the concentration of NusE is limiting (TABLE FOUR), and assembly is likely to occur via pre-formation of the NusB-boxA RNA subcomplex (Fig. 12A).


View this table:
[in this window]
[in a new window]
  		TABLE FOUR
Concentrations of cellular (E. coli) components at different growth rates

 
Although the concentration of free NusE increases at higher growth rates (2.5 doublings/h; see TABLE FOUR), it is still limiting and thus the assembly of the boxA core complex probably continues via pre-formation of the NusB-boxA interaction under these conditions (see Fig. 12A). The concentration of boxA RNA undergoes the largest change in moving from low to high growth rates. We therefore simulated the concentration of each binary and the ternary complex relative to this change. Assuming that the concentration of all the components of this complex varies by the same ratio across all growth rates, we can simulate the concentration of the different binary and ternary complexes during growth by using the Kd values determined in this work (Fig. 12B). The amount of the NusB-boxA complex present is in excess at all concentrations of boxA RNA, suggesting that assembly of the ternary complex occurs through this pre-formed complex, followed (and stabilized) by the binding of NusE (Fig. 9C). Thus this model suggests, at the growth rates for which protein concentration data are available, that the ternary complex probably does assemble primarily via an initial NusB-boxA RNA subcomplex.

Binding of NusE to the NusB-boxA RNA complex would then localize the ribosomal antitermination complex specifically to the rRNA transcript and could serve to nucleate the formation of the entire rrn antitermination complex required for the synthesis of full-length rRNA. For {lambda} antitermination, boxA binds NusB and NusE, followed by NusG binding, and this complex presumably interacts with the {lambda}N-boxB-NusA subassembly to form an antitermination complex that is sufficiently stable to support long range antitermination effects in this system (17)4.



View larger version (21K):
[in this window]
[in a new window]
  		FIGURE 12.
Simulation of the relative in vivo concentrations of the core boxA complex and subassemblies at low and high growth rates. A, data on the concentration of transcription and translation machinery at two growth rates (µ = 0.5 and 2.4 doublings/h; see TABLE FOUR) (41, 42) were used to simulate the concentrations (nM) of the core boxA complex and subcomplexes at each growth rate. The concentration of the ternary core complex (BER, black bars) and subassemblies (NusB-boxA RNA, BR, gray bars; and NusB-NusE, BE, white bars) were determined for each growth rate. At both growth rates, the concentration of the NusB-NusE heterodimer exceeds that of the other components, suggesting that this may be the preferred pathway of core complex assembly in vivo under most conditions. B, simulation of the concentration of the ternary core complex (BER, dashed/dotted line) and the binary complexes (NusB-boxA RNA, BR, solid line; and NusB-NusE BE, dashed line). The concentration of total boxA RNA varies the most between low and high growth rates.

 
Understanding how the assembly of this complex takes place and is regulated may contribute to a broader understanding of how transcription is regulated generally. Because in this work NusE alone was shown not to bind specifically to boxA RNA, at least not at detectable levels (Fig. 7A), we suggest that NusB acts as a specificity factor to nucleate complex assembly on the boxA RNA sequence (Fig. 13). This is consistent with our simulations of in vivo assembly reactions in E. coli that suggest that the NusB-boxA complex is in excess at the bacterial growth rates for which information is available. NusE can thus be considered to be an "affinity regulator," with some preference for the rrn boxA sequence over the {lambda} boxA sequence, as it significantly increases the affinity of NusB for boxA RNA (Fig. 8), presumably by creating a network of interactions that supplement the binding surfaces of the boxA RNA to stabilize the specific interaction with NusB. The other members of the complex (see Fig. 1B) are then able to bind to the core boxA complex and also interact with RNA polymerase (see below) by means of cis looping interactions with the nascent RNA (2, 3).

The unstructured nature of NusE (S10) allows this small protein of 103 amino acid residues to participate in multiple interactions, including rRNA antitermination and integration into the ribosome. It is unlikely that much NusE exists free in solution in the cell, because it is insoluble as the free protein, even at "moderate" concentrations in "native" buffers. However, this protein is obviously soluble in the cell lysate during the purification process (33), suggesting that NusE may be involved in multiple solubilizing reactions in vivo, including nonspecific complex formation with available RNA, as a specific complex with NusB or NusB-boxA RNA and as an integral component of the ribosome. These interactions are all consistent with our experimental findings in this paper and with our calculations based on the component concentration data available for E. coli (TABLE FOUR) (42). The observed nonspecific binding of NusE with RNA increases the complexity of the binding analysis by requiring a binding model with multiple equilibria to describe accurately the overall assembly reaction. This nonspecific binding interaction may also have perturbed the apparent dissociation constants obtained in experiments in which a tRNA carrier was used as a component of the binding buffer (4).



View larger version (38K):
[in this window]
[in a new window]
  		FIGURE 13.
Model for the assembly of the rrn antitermination complex during the transcription of rRNA. NusB binds to boxA RNA as it is synthesized (A), followed by NusE association with the NusB-boxA RNA binary complex (B). C, the other Nus factors (A and G), along with various ribosomal proteins (such as S4 and L4; see Ref. 20), also participate in the formation of the antitermination complex bound to polymerase. D, once the ribosomal RNA has been transcribed and the binding sites for individual ribosomal proteins that participate in antitermination are available, these proteins bind preferentially to the ribosomal subassembly. E, the complex is destabilized and NusA and NusG are released, followed by the release of NusE, which is then assembled into the ribosome. This leaves NusB bound weakly to boxA, after which it is likely to be recycled to a boxA site located on a new transcript, and the original boxA sequence is degraded.

 
Assembly of the core boxA RNA complex may take place through either NusB-NusE heterodimer formation followed by boxA binding (Fig. 9A) or by NusB binding to boxA RNA, followed by the sequestration of NusE to the boxA complex (Fig. 9C). The actual pathway of assembly depends upon the relative concentrations of free NusE and naked mRNA, and because the concentration of NusB is in excess at all growth rates, it is likely that the in vivo assembly pathway for this core antitermination complex also proceeds via an initial binding of NusB to boxA RNA, followed by NusE binding to form the stable and specific ternary complex (Fig. 9C and Fig. 12).

Ribosomal RNA transcription is controlled during cell growth by factors that regulate the initiation of transcription, as well as by antitermination of transcription by the boxA RNA-containing complex. The relative concentrations of ribosomal proteins and ribosomal RNA are modulated by a feedback mechanism that incorporates the ribosomal proteins as components in the antitermination system (13, 20). NusE (ribosomal S10 protein) is a small protein that binds specifically to 16 S RNA in the final phase of 30 S subunit assembly (33, 43, 44), as well as participating as an antitermination factor in rRNA transcription. Therefore, it seems reasonable that NusB determines the specificity of the assembly of the core antitermination complex assembly on boxA RNA sequences (Figs. 12 and 13). However, NusE and other ribosomal proteins are also required to form part of the antitermination complex in order to maintain the balance of ribosomal RNA and protein components at normal growth rates (45-47) and hence the increase in affinity of NusB for boxA RNA in the presence of NusE.

The nonspecific RNA binding activity of NusE, in addition to serving to enhance the specificity of the NusB interaction with boxA (perhaps somewhat differentially as suggested by the slightly increased affinity of the NusB and NusE proteins for rrn boxA RNA as compared with the {lambda} sequence), contributes to the stability of the core boxA complex by binding to the RNA transcript and to RNA polymerase (21), as well as to NusB (23), and is directed specifically to boxA through the NusB-boxA RNA interaction. This "nucleation" complex is extended to bring together a network of RNA-protein interactions that includes NusG, NusA, S4, and other ribosomal proteins that also contact RNA polymerase and, for some proteins, the nascent RNA (Fig. 13) (20, 48-50). Once synthesis of the rRNA transcript is completed and assembly of the 30 S ribosomal subunit reaches the stage of NusE incorporation, NusE probably binds to the partially assembled ribosome specifically and with high affinity. This interaction competes successfully with the NusB-NusE-boxA complex and sequesters NusE away from the antitermination complex. NusB is left bound specifically, but weakly, to the boxA RNA, and this might assist in recycling NusB to subsequent nascent transcripts that contain the boxA sequence (Fig. 13).

Lambdoid phages, including {lambda}, {phi}21, and {phi}22, require the boxA complex to facilitate assembly of the entire antitermination system (including NusG, along with N and NusA bound independently to boxB) onto the phage mRNA. This network of interactions between N, NusA, and boxB, followed by the interactions between NusE, NusB, and the boxA RNA sequence and then the interactions of N, NusA, NusE, and NusG with RNA polymerase (22, 48, 49, 51), creates a complex network that is stable and active during transcription over thousands of nucleotide residues and enables expression of the phage late genes during the switch from lysogeny to lysis. The {phi}21 and {phi}22 boxA sequences are identical to the rrn boxA sequence, and so NusE is likely to bind to this NusB-RNA complex at lower concentrations than the free protein concentration at which it binds to the {lambda} boxA sequences. The {lambda} boxA RNA sequence renders {lambda} phage sensitive to the NusE71 mutation, which is complemented by the NusB101 mutation (52), and suggests that these amino acid residues and the nucleotide residues that differ between the rrn and the {lambda} boxA sequences are involved in forming the NusB-NusE-boxA RNA interface. These changes in the {lambda} boxA RNA sequence may also be required for some other function, such as to regulate RNA transcription and/or for phage-phage competition reactions.

Extrapolation to Other and More Complex Regulatory Systems—The assembly of the core boxA complex discussed here represents a relatively simple model system that encounters some of the same issues faced by larger and more complicated RNA-protein complexes. It is possible that nonspecific binding to RNA may reflect a requirement for electrostatic interaction between protein and RNA that is needed to support the specific binding of a particular protein to a defined RNA sequence or structure, perhaps by increasing the local concentration of binding partners until the correct binding structure can form. Alternatively, such nonspecific binding may increase the variety of interactions in which such a small protein can participate. A basic mechanism of complex assembly appears to involve nucleation by one or more specificity factors that bind to a specific RNA sequence or structure, followed by the binding, through specific protein-protein interactions, of other cofactors or subassemblies that also bind to the RNA nonspecifically or with weak specificity, finally resulting in the creation of a network of protein-protein and protein-RNA interactions that stabilize the complex.

This general process applies also to the assembly of RNA virions and potentially in directing mRNA splicing, as well as controlling RNA transcription. For instance, the nucleocapsid of human immunodeficiency virus, type 1, binds to RNA nonspecifically and forms an RNA-protein network, mimicking the process that must occur during virion assembly after nucleation by the binding of the packaging signal (53). In another example, during assembly of the spliceosome, Sm proteins bind non-specifically to any RNA that is relatively uridine-rich, and this interaction is directed to uridine-rich small nuclear RNAs by the SMN complex (54). Thus nonspecific binding of proteins to RNA may be a vital intermediate step in complex assembly processes, allowing cofactors to participate in more than one macromolecule machine by binding to specificity factors while also contributing to the overall stability of the final complex, and thus serving as a switch that specifically activates or deactivates a regulatory function. Finally, the additive stability of each layer of these interaction networks also allows for sequential disassembly, "switching" a process on or off when required to permit the finely balanced and integrated regulation of cellular processes. Modular interaction of components could also significantly expand the repertoire of activities that a finite number of proteins can control and might contribute to limiting the number of components that the cell must contain to support the large number of processes of gene expression. Hence, despite the complexities of the necessary data collection and analysis, the role of nonspecific RNA binding in complex assembly should be considered more thoroughly, and greater importance than has previously been the case should probably be attributed to this aspect of RNA-protein interactions in considering the assembly and regulation of the various macromolecular complexes involved in gene expression.


   FOOTNOTES
 
* This work was supported by National Institutes of Health Grants GM-15792 and GM-29158 (to P. H. v. H.), by American Heart Association Postdoctoral Fellowship 0425713Z (to S. J. G.), and by an American Cancer Society Research Professorship (to P. H. v. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Back

1 Present address: Dept. of Cell and Molecular Biology, University of Hawaii, Manoa, 1960 East-West Rd., Biomed. T-514, Honolulu, HI 96822. Back

2 To whom correspondence should be addressed: Institute of Molecular Biology and Dept. of Chemistry, University of Oregon, Eugene, OR 97403. E-mail: petevh{at}molbio.uoregon.edu.

3 The abbreviations used are: TEV, tobacco etch virus; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight. Back

4 C. Conant, unpublished data. Back


   ACKNOWLEDGMENTS
 
We are grateful to our laboratory colleagues, particularly Neil Johnson and Jim Goodarzi, for numerous useful discussions and much helpful advice.



   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. von Hippel, P. H., Kowalczykowski, S. C., Lonberg, N., Newport, J. W., Paul, L. S., Stormo, G. D., and Gold, L. (1982) J. Mol. Biol. 162, 795-818[CrossRef][Medline] [Order article via Infotrieve]
  2. Whalen, W. A., and Das, A. (1990) New Biol. 2, 975-991[Medline] [Order article via Infotrieve]
  3. Van Gilst, M. R., Rees, W. A., Das, A., and von Hippel, P. H. (1997) Biochemistry 36, 1514-1524[CrossRef][Medline] [Order article via Infotrieve]
  4. Nodwell, J. R., and Greenblatt, J. (1993) Cell 72, 261-268[CrossRef][Medline] [Order article via Infotrieve]
  5. Sharrock, R. A., Gourse, R. L., and Nomura, M. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 5275-5279[Abstract/Free Full Text]
  6. Weisberg, R. A., and Gottesman, M. E. (1999) J. Bacteriol. 181, 359-367[Free Full Text]
  7. Greive, S. J., and von Hippel, P. H. (2005) Nat. Rev. Mol. Cell. Biol. 6, 221-232[CrossRef][Medline] [Order article via Infotrieve]
  8. Richardson, J. P. (2003) Cell 114, 157-159[CrossRef][Medline] [Order article via Infotrieve]
  9. Wilson, K. S., and von Hippel, P. H. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8793-8797[Abstract/Free Full Text]
  10. Rees, W., Weitzel, S., Das, A., and von Hippel, P. (1997) J. Mol. Biol. 273, 797-813[CrossRef][Medline] [Order article via Infotrieve]
  11. Komissarova, N., Becker, J., Solter, S., Kireeva, M., and Kashlev, M. (2002) Mol. Cell 10, 1151-1162[CrossRef][Medline] [Order article via Infotrieve]
  12. Van Gilst, M. R., and von Hippel, P. H. (2000) Methods Enzymol. 323, 1-31[Medline] [Order article via Infotrieve]
  13. Condon, C., Squires, C., and Squires, C. L. (1995) Microbiol. Rev. 59, 623-645[Abstract/Free Full Text]
  14. Berg, K. L., Squires, C., and Squires, C. L. (1989) J. Mol. Biol. 209, 345-358[CrossRef][Medline] [Order article via Infotrieve]
  15. Heinrich, T., Condon, C., Pfeiffer, T., and Hartmann, R. K. (1995) J. Bacteriol. 177, 3793-3800[Abstract/Free Full Text]
  16. Pfeiffer, T., and Hartmann, R. K. (1997) J. Mol. Biol. 265, 385-393[CrossRef][Medline] [Order article via Infotrieve]
  17. DeVito, J., and Das, A. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 8660-8664[Abstract/Free Full Text]
  18. Nodwell, J. R., and Greenblatt, J. (1991) Genes Dev. 5, 2141-2151[Abstract/Free Full Text]
  19. Van Gilst, M., and von Hippel, P. (1997) J. Mol. Biol. 274, 160-173[CrossRef][Medline] [Order article via Infotrieve]
  20. Torres, M., Condon, C., Balada, J. M., Squires, C., and Squires, C. L. (2001) EMBO J. 20, 3811-3820[CrossRef][Medline] [Order article via Infotrieve]
  21. Mason, S. W., and Greenblatt, J. (1991) Genes Dev. 5, 1504-1512[Abstract/Free Full Text]
  22. Mogridge, J., Mah, T. F., and Greenblatt, J. (1998) J. Biol. Chem. 273, 4143-4148[Abstract/Free Full Text]
  23. Mason, S. W., Li, J., and Greenblatt, J. (1992) J. Mol. Biol. 223, 55-66[CrossRef][Medline] [Order article via Infotrieve]
  24. Luttgen, H., Robelek, R., Muhlberger, R., Diercks, T., Schuster, S. C., Kohler, P., Kessler, H., Bacher, A., and Richter, G. (2002) J. Mol. Biol. 316, 875-885[CrossRef][Medline] [Order article via Infotrieve]
  25. Gill, S. C., and von Hippel, P. H. (1989) Anal. Biochem. 182, 319-326[CrossRef][Medline] [Order article via Infotrieve]
  26. Lucast, L. J., Batey, R. T., and Doudna, J. A. (2001) BioTechniques 30, 544-550[Medline] [Order article via Infotrieve]
  27. Demeler, B., and van Holde, K. E. (2004) Anal. Biochem. 335, 279-288[CrossRef][Medline] [Order article via Infotrieve]
  28. Schuck, P. (2000) Biophys. J. 78, 1606-1619[Medline] [Order article via Infotrieve]
  29. Dam, J., and Schuck, P. (2005) Biophys. J. 89, 651-666[CrossRef][Medline] [Order article via Infotrieve]
  30. Dam, J., Velikovsky, C. A., Mariuzza, R. A., Urbanke, C., and Schuck, P. (2005) Biophys. J. 89, 619-634[CrossRef][Medline] [Order article via Infotrieve]
  31. Kuzmic, P. (1996) Anal. Biochem. 237, 260-273[CrossRef][Medline] [Order article via Infotrieve]
  32. Berglechner, F., Richter, G., Fischer, M., Bacher, A., Gschwind, R. M., Huenges, M., Gemmecker, G., and Kessler, H. (1997) Eur. J. Biochem. 248, 338-346[Medline] [Order article via Infotrieve]
  33. Culver, G. M., and Noller, H. F. (1999) RNA (N. Y.) 5, 832-843
  34. Yaguchi, M., Roy, C., and Wittmann, H. G. (1980) FEBS Lett. 121, 113-116[CrossRef][Medline] [Order article via Infotrieve]
  35. Gopal, B., Papavinasasundaram, K. G., Dodson, G., Colston, M. J., Major, S. A., and Lane, A. N. (2001) Biochemistry 40, 920-928[CrossRef][Medline] [Order article via Infotrieve]
  36. Greenblatt, J., and Li, J. (1981) J. Mol. Biol. 147, 11-23[CrossRef][Medline] [Order article via Infotrieve]
  37. Van Gilst, M., Rees, W. A., and von Hippel, P. H. (1995) Nucleic Acids Symp. Ser. 33, 145-147
  38. Cohen, G., and Eisenberg, H. (1968) Biopolymers 6, 1077-1100[CrossRef][Medline] [Order article via Infotrieve]
  39. Bonifacio, G. F., Brown, T., Conn, G. L., and Lane, A. N. (1997) Biophys. J. 73, 1532-1538[Medline] [Order article via Infotrieve]
  40. Lakowicz, J. R. (1999) Principles of Fluorescence Spectroscopy, 2nd Ed., pp. 302-303, Kluwer Academic/Plenum Publishing Corp., New York
  41. Swindle, J., Zylicz, M., Georgopoulos, C., Li, J., and Greenblatt, J. (1988) J. Biol. Chem. 263, 10229-10235[Abstract/Free Full Text]
  42. Bremmer, H., and Dennis, P. P. (1996) in Escherichia coli and Salmonella Cellular and Molecular Biology (Neidhardt, F. C., ed) 2nd Ed., pp. 1553-1569, American Society for Microbiology, Washington, D. C.
  43. Mizushima, S., and Nomura, M. (1970) Nature 226, 1214-1218[CrossRef][Medline] [Order article via Infotrieve]
  44. Held, W. A., Ballou, B., Mizushima, S., and Nomura, M. (1974) J. Biol. Chem. 249, 3103-3111[Abstract/Free Full Text]
  45. Koch, A. L. (1970) J. Theor. Biol. 28, 201-231[Medline] [Order article via Infotrieve]
  46. Gausing, K. (1977) J. Mol. Biol. 115, 335-354[CrossRef][Medline] [Order article via Infotrieve]
  47. Dodd, J., Kolb, J. M., and Nomura, M. (1991) Biochimie (Paris) 73, 757-767
  48. Gill, S. C., Weitzel, S. E., and von Hippel, P. H. (1991) J. Mol. Biol. 220, 307-324[CrossRef][Medline] [Order article via Infotrieve]
  49. Li, J., Horwitz, R., McCracken, S., and Greenblatt, J. (1992) J. Biol. Chem. 267, 6012-6019[Abstract/Free Full Text]
  50. Van Gilst, M. R., and von Hippel, P. H. (1997) J. Mol. Biol. 274, 160-173[CrossRef][Medline] [Order article via Infotrieve]
  51. Mah, T. F., Li, J., Davidson, A. R., and Greenblatt, J. (1999) Mol. Microbiol. 34, 523-537[CrossRef][Medline] [Order article via Infotrieve]
  52. Court, D. L., Patterson, T. A., Baker, T., Constantino, N., Mao, X., and Friedman, D. I. (1995) J. Bacteriol. 177, 2589-2591[Abstract/Free Full Text]
  53. Stoylov, S. P., Vuilleumier, C., Stoylova, E., De Rocquigny, H., Roques, B. P., Gerard, D., and Mely, Y. (1997) Biopolymers 41, 301-312[CrossRef][Medline] [Order article via Infotrieve]
  54. Pellizzoni, L., Yong, J., and Dreyfuss, G. (2002) Science 298, 1775-1779[Abstract/Free Full Text]

Add to CiteULike CiteULike   Add to Complore Complore   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us   Add to Digg Digg   Add to Reddit Reddit   Add to Technorati Technorati    What's this?


This article has been cited by other articles:


Home page
 Nucleic Acids ResHome page
S. Prasch, M. Jurk, R. S. Washburn, M. E. Gottesman, B. M. Wohrl, and P. Rosch
RNA-binding specificity of E. coli NusA
Nucleic Acids Res., June 10, 2009; (2009) gkp452v1.
[Abstract] [Full Text] [PDF]

Home page
 J. Bacteriol.Home page
K. B. Arnvig, S. Zeng, S. Quan, A. Papageorge, N. Zhang, A. C. Villapakkam, and C. L. Squires
Evolutionary Comparison of Ribosomal Operon Antitermination Function
J. Bacteriol., November 1, 2008; 190(21): 7251 - 7257.
[Abstract] [Full Text] [PDF]

Home page
 MicrobiologyHome page
P. J. Lewis, G. P. Doherty, and J. Clarke
Transcription factor dynamics
Microbiology, July 1, 2008; 154(7): 1837 - 1844.
[Abstract] [Full Text] [PDF]

Home page
 J. Bacteriol.Home page
G. P. Doherty, D. H. Meredith, and P. J. Lewis
Subcellular Partitioning of Transcription Factors in Bacillus subtilis
J. Bacteriol., June 1, 2006; 188(11): 4101 - 4110.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
280/43/36397    most recent
M507146200v1
Right arrow Submit a Letter to Editor
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Greive, S. J.
Right arrow Articles by von Hippel, P. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Greive, S. J.
Right arrow Articles by von Hippel, P. H.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 All ASBMB Journals   Molecular and Cellular Proteomics 
 Journal of Lipid Research   ASBMB Today 
Copyright © 2005 by the American Society for Biochemistry and Molecular Biology.
Advertisement
spacer
Advertisement
Advertisement