Dual energy landscape: The functional state of the β-barrel outer membrane protein G molds its unfolding energy landscape

  • Published on

  • View

  • Download

Embed Size (px)


<ul><li><p>RESEARCH ARTICLE</p><p>Dual energy landscape: The functional state of the</p><p>b-barrel outer membrane protein G molds its unfoldingenergy landscape</p><p>Mehdi Damaghi1,2, K. Tanuj Sapra2,3, Stefan Koster4, Ozkan Yildiz4, Werner K .uhlbrandt4</p><p>and Daniel J. Muller1,2</p><p>1 ETH Z .urich, Department of Biosystems Science and Engineering, Basel, Switzerland2 Biotechnology Center, University of Technology, Tatzberg, Dresden, Germany3 Chemistry Research Laboratory, University of Oxford, Oxford, UK4 Max-Planck-Institute of Biophysics, Department of Structural Biology, Frankfurt am Main, Germany</p><p>Received: April 13, 2010</p><p>Revised: May 16, 2010</p><p>Accepted: May 17, 2010</p><p>We applied dynamic single-molecule force spectroscopy to quantify the parameters (free energy</p><p>of activation and distance of the transition state from the folded state) characterizing the energy</p><p>barriers in the unfolding energy landscape of the outer membrane protein G (OmpG) from</p><p>Escherichia coli. The pH-dependent functional switching of OmpG directs the protein alongdifferent regions on the unfolding energy landscape. The two functional states of OmpG take</p><p>the same unfolding pathway during the sequential unfolding of b-hairpins IIV. After theinitial unfolding events, the unfolding pathways diverge. In the open state, the unfolding of</p><p>b-hairpin V in one step precedes the unfolding of b-hairpin VI. In the closed state, b-hairpin Vand b-strand S11 with a part of extracellular loop L6 unfold cooperatively, and subsequentlyb-strand S12 unfolds with the remaining loop L6. These two unfolding pathways in the openand closed states join again in the last unfolding step of b-hairpin VII. Also, the conformationalchange from the open to the closed state witnesses a rigidified extracellular gating loop L6.</p><p>Thus, a change in the conformational state of OmpG not only bifurcates its unfolding pathways</p><p>but also tunes its mechanical properties for optimum function.</p><p>Keywords:</p><p>Atomic force microscopy / Interactions / Mechanical properties / Nanoproteomics / pH</p><p>gating / Single-molecule force spectroscopy</p><p>1 Introduction</p><p>Outer membrane proteins (Omps) are found in the outer</p><p>membranes of Gram-negative bacteria, mitochondria, and</p><p>chloroplasts. These b-barrel-forming transmembrane proteinsare imperative to a cells survival owing to their function in</p><p>controlling the transport of solutes in and out of a cell.</p><p>Whereas some Omps, like OmpC, OmpF, and OmpG, are</p><p>non-selective in what they transport through their pores, others</p><p>like LamB, FhuA, BtuB, are solute specific [16]. The gating</p><p>mechanisms of the b-barrel forming Omps attract continuousinterest and remain to be investigated in detail [1]. Several of</p><p>the transmembrane pores formed by Omps of Escherichia coliare pH gated. Low pH induces the closing of the pores of, for</p><p>example, OmpC, OmpF, OmpG, LamB, and PhoE [26]. What</p><p>conformational changes drive pore closure has long been</p><p>debated [1, 7]. In 1999, high-resolution atomic force micro-</p><p>scopy (AFM) imaging, for the first time, showed that at low pH</p><p>the large extracellular loops of the OmpF collapsed onto theAbbreviations: aa, amino acids; AFM, atomic force microscopy;</p><p>DFS, dynamic single-molecule force spectroscopy; F-D curve,</p><p>single-molecule force-distance curve; Omp, outer membrane</p><p>protein; SMFS, single-molecule force spectroscopy; WLC, worm-</p><p>like-chain</p><p>These authors contributed equally to this work.</p><p>Colour Online: See the article online to view fig. 3 in colour.</p><p>Correspondence: Professor Daniel J. Muller, ETH Z .urich,</p><p>Department of Biosystems Science and Engineering, Matten-</p><p>strasse 26, 4058 Basel, Switzerland</p><p>E-mail: daniel.mueller@bsse.ethz.ch</p><p>Fax: 141-61-387-39-94</p><p>&amp; 2010 WILEY-VCH Verlag GmbH &amp; Co. KGaA, Weinheim www.proteomics-journal.com</p><p>Proteomics 2010, 10, 41514162 4151DOI 10.1002/pmic.201000241</p></li><li><p>pore entrance [8]. This supported the hypothesis that the Omp</p><p>pores are gated by the conformational changes of the flexible</p><p>extracellular loops. Experiments on the maltoporin LamB from</p><p>E. coli, which is specific for malto-oligosaccharides, corrobo-rated the gating model. When lacking the major extracellular</p><p>loops L4 and L6, LamB failed to close at lower pH [9].</p><p>Among the pH-gated Omps, the structure and function</p><p>relationship of the OmpG from E. coli represents possibly thebest-studied example. The OmpG structure solved by X-ray</p><p>crystallography [10, 11] and NMR [12] comprises 14 b-strands(S1S14) that form a transmembrane b-barrel. On the peri-plasmic side the b-strands are connected by six short poly-peptide turns (T1T6). On the extracellular side the b-strandsare connected by seven longer loops (L1L7) that exhibit</p><p>enhanced intrinsic flexibility [11, 12]. A pH-dependent gating</p><p>controls the flux of small molecules through the OmpG pore</p><p>[6]. X-ray structures obtained from three-dimensional OmpG</p><p>crystals grown at neutral (pH 7.5) and acidic (pH 5.6) pH</p><p>provide insight into the conformational changes that may</p><p>guide the gating mechanism [10]. At low pH, the largest</p><p>extracellular loop L6 folds into the pore, thereby constricting</p><p>its entrance. However, the three-dimensional crystals of</p><p>solubilized OmpG grown at different pHs showed different</p><p>packing arrangements, and some extracellular loops formed</p><p>crystal contacts with adjacent OmpG molecules. Thus, it may</p><p>be assumed that the conformations observed may not</p><p>represent those that naturally occur in the gating mechanism</p><p>of OmpG. To test this hypothesis, OmpG was reconstituted</p><p>in native E. coli lipids and imaged by high-resolution AFM inbuffer solution at room temperature [13]. The AFM topo-</p><p>graphs confirmed that the pH-dependent gating mechanism</p><p>suggested from the X-ray structures indeed occurred in</p><p>physiological conditions.</p><p>Besides the gating mechanisms of the Omps, the</p><p>mechanisms that guide their folding and unfolding are of</p><p>pertinent interest [1420]. So far most experiments investi-</p><p>gating the folding of Omps have first denatured Omps in</p><p>detergent and/or urea and then characterized the refolding</p><p>into a lipid bilayer or in a detergent [14, 2124]. Such bulk</p><p>unfolding experiments suggest that OmpG unfolds and</p><p>refolds reversibly [22]. The folding process of Omps is</p><p>described as being coupled with membrane insertion [17]. A</p><p>folding and insertion process has been recently described for</p><p>the b-barrel transmembrane protein PagP from E. coli [25].PagP solubilized and denatured in 10 M urea is found to</p><p>adsorb to the lipid headgroups of the bilayer, where it forms a</p><p>transition state that tilts and inserts into the lipid membrane</p><p>to complete the folding process. Apparently, these models</p><p>contrast with the results from single-molecule force spectro-</p><p>scopy (SMFS) in which single OmpG molecules have been</p><p>mechanically stressed to induce their unfolding from the</p><p>native lipid membrane in a buffer solution [26]. These single-</p><p>molecule experiments clearly show that OmpG molecules</p><p>unfold via many sequential unfolding intermediatesdescribing a detailed unfolding pathway. The unfolding step</p><p>of a single b-hairpin characterizes the transition from one</p><p>unfolding intermediate to the next one. However, the differ-</p><p>ences between chemical denaturation and refolding and</p><p>mechanical unfolding experiments may have different</p><p>origins. First, one may assume that the refolding mechanism</p><p>in the absence of any external force does not reflect unfolding</p><p>under an applied force. Second, it may be that the experi-</p><p>mental conditions alter the unfolding and folding pathways</p><p>chosen by b-barrel membrane proteins. In case of a-helicaltransmembrane proteins it has been shown that alterations in</p><p>the temperature and buffer solution within the physiological</p><p>relevant range can considerably modify their unfolding</p><p>pathways [27, 28]. Therefore, it is not surprising that the</p><p>exposure of membrane proteins to urea, detergent, and</p><p>mechanical stress may force them along very different</p><p>unfolding and folding pathways.</p><p>SMFS has been particularly successful in characterizing</p><p>the unfolding pathways of membrane proteins and to</p><p>quantify the interactions and energies of the intermediates</p><p>in the unfolding pathways [28, 29]. SMFS provides detailed</p><p>insights into the nature of molecular interactions and most</p><p>importantly allows to locate and quantify these interactions</p><p>structurally with an accuracy of E26 amino acids (aa). Inthis work we have performed dynamic SMFS (DFS) to probe</p><p>the strength of the interactions that stabilize the unfolding</p><p>intermediates of OmpG at different loading rates (applied</p><p>force over time). The dependence of these interaction</p><p>strengths on the loading rate allows quantifying the</p><p>unfolding energy barriers of the intermediates [30, 31].</p><p>These measurements provide the position of the transition</p><p>state, the transition rate of the intermediate from the folded</p><p>to the unfolded state, and the energy of activation to cross</p><p>the transition barrier. Because the sensitivity of SMFS</p><p>permits to directly determine the sequence at which the</p><p>unfolding barriers are located along an unfolding pathway,</p><p>we can chart the unfolding energy landscape of OmpG in</p><p>the two pH-dependent conformational and functional states.</p><p>The energy landscapes reveal detailed insights into how</p><p>interactions can change the unfolding pathways and the</p><p>gating mechanism of OmpG. We show that the molecular</p><p>interactions associated with a change in pH not only drive a</p><p>conformational change but also influence the mechanical</p><p>properties of the region responsible for the conformational</p><p>change. We propose that the functional properties of the</p><p>protein are related to its mechanical properties.</p><p>2 Materials and methods</p><p>2.1 SMFS and DFS</p><p>OmpG was purified from inclusion bodies, refolded in</p><p>detergent, and reconstituted into native E. coli lipids [10].Membranes showed the OmpG molecules being densely</p><p>packed and assembled into two-dimensional crystals. These</p><p>OmpG membranes were adsorbed onto freshly cleaved mica</p><p>(E30 min) in buffer solution (pH 7.0, 25 mM Tris-HCl,</p><p>4152 M. Damaghi et al. Proteomics 2010, 10, 41514162</p><p>&amp; 2010 WILEY-VCH Verlag GmbH &amp; Co. KGaA, Weinheim www.proteomics-journal.com</p></li><li><p>25 mM MgCl2, 300 mM NaCl or pH 5.0, 25 mM Na-acetate,</p><p>25 mM MgCl2, 300 mM NaCl). The adsorbed membranes</p><p>were localized by AFM in the same buffer solution at room</p><p>temperature [32]. For SMFS, the AFM cantilever tip (60 mmlong Biolever, Olympus) was pushed onto the OmpG</p><p>membrane applying forces E500750 pN for E500 ms. InE0.1% of all cases the OmpG terminus attached to theAFM tip. Then the AFM tip was retracted at a specific</p><p>pulling velocity to induce unfolding. A force-distance (F-D)</p><p>curve recorded the forces required to overcome the interac-</p><p>tion strengths that stabilized the unfolding intermediates of</p><p>the membrane protein. F-D spectra recorded from OmpG,</p><p>which was either densely packed or crystallized two-</p><p>dimensionally showed no difference in the force pattern.</p><p>DFS experiments were performed at seven pulling velocities</p><p>(100, 300, 600, 900, 1200, 2500, and 5000 nm/s). Before</p><p>and after each experiment the spring constant of each</p><p>cantilever (E0.03 N/m) was estimated from its thermalnoise using the equipartition theorem [33]. To minimize</p><p>errors that may occur due to uncertainties in the</p><p>cantilever spring constant calibration, OmpG was unfolded</p><p>using at least three different cantilevers for each pulling</p><p>velocity.</p><p>2.2 Data selection and analysis</p><p>We analyzed only F-D curves that corresponded to the</p><p>length (475 nm) of a fully stretched and unfolded OmpGpolypeptide (281 aa). This selection criterion ensured that</p><p>OmpG was mechanically unfolded by stretching one of its</p><p>termini [28]. Previously we have shown that OmpG predo-</p><p>minantly attaches with its N-terminal end to the AFM tip</p><p>[26]. All F-D curves having a length of 475 nm showed an F-D pattern similar to those published previously [26, 34].</p><p>Thus, we could conclude that the F-D curves were recorded</p><p>upon mechanically unfolding of OmpG from the N-terminal</p><p>end. For analysis each force peak of each F-D curve was</p><p>fitted using the worm-like-chain (WLC) model to reveal the</p><p>contour lengths of the unfolded polypeptides [28]. Deter-</p><p>mining the unfolded polypeptide stretches allowed assign-</p><p>ing the structural regions that form stable unfolding</p><p>intermediates [26, 28].</p><p>2.3 Calculating xu and k0 from DFS data</p><p>According to the BellEvans theory [35, 36], the</p><p>most probable unfolding force F plotted versus ln rf describes the most prominent unfolding energy barriers</p><p>that have been crossed along the force-driven reaction</p><p>coordinate [30]. The relation between F and rf can bedescribed by:</p><p>F kBTxu</p><p>lnrurfkBTk0</p><p> 1</p><p>where kB is the Boltzmann constant, T the absolutetemperature, rf the loading rate, xu the distance between thefree energy minimum of the folded intermediate state and</p><p>transition state barrier, and k0 the unfolding rate of theintermediate at zero force. rf is the product of pullingvelocity, v, and the slope of the WLC fit at each force peak.Experimental loading rate and force histograms (Supporting</p><p>Information Figs. S1 and S2) were fitted with Gaussian</p><p>distributions. The resulting F was semi-logarithmicallyplotted versus rf . xu and k0 were obtained by fitting Eq. (1)using a non-linear least-squares algorithm. Only unfolding</p><p>forces and loading rates corresponding to the main force</p><p>peaks were considered for analysis.</p><p>2.4 Calculating transition barrier height and rigidity</p><p>The height of the free energy barrier, DGz, separating thefolded and the unfolded states was assessed using an</p><p>Arrhenius-like expression:</p><p>DGz kBT lntDk0 2where tD denotes the diffusive relaxation time. Typicalvalues for tD found for proteins are in the order of107109 s [37, 38]. Therefore, assuming tD5 10</p><p>9 s seemsto be reasonable for determining the free energy barrier</p><p>heights. This value has also been used for molecular</p><p>dynamics simulations of protein folding [39, 40]. We have</p><p>used tD5 109 s in all our calculations. Varying tD within</p><p>the above-mentioned range changes the free energy of</p><p>activation by o15%. Moreover, even if tD was wrong byorders of magnitude, the influence on the error of tD wouldbe the same for all conditions and values and, hence, would</p><p>not affect the qualitative results. Errors in DGz were esti-mated by propagation of the errors of k0.</p><p>Without any information on the energy potential shape,</p><p>we assumed the shape of the energy potential to be a simple</p><p>parabola. Hence, the spring constant k of the folded struc-ture was calculated using DGz and xu [41, 42]:</p><p>k 2DGz</p><p>x2u3</p><p>Errors in DGz and xu were propagated for estimation oferrors in k.</p><p>3 Results and discussion</p><p>3.1 The functional state of OmpG directs its</p><p>unfolding route</p><p>In previous experiments we have established SMFS-based</p><p>unfolding of single OmpG molecules that were recon-</p><p>stituted into membranes of native E. coli lipids [26]. In thoseexperiments we first localized OmpG membranes by AFM</p><p>imaging [13, 32]. Then the AFM tip was pressed onto OmpG</p><p>Proteomics 2010, 10, 41514162 4153</p><p>&amp; 2010 WILEY-VCH Verlag GmbH &amp; Co. KGaA, Weinheim www.proteomics-journal.com</p></li><li><p>to facilitate the non-specific attachment of the terminal end</p><p>(Fig. 1A). The non...</p></li></ul>