Structural and dynamical analysis of an engineered FhuA channel protein embedded into a lipid bilayer or a detergent belt

Journal of Structural Biology 177 (2012) 291–301

Contents lists available at SciVerse ScienceDirect

Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi

Structural and dynamical analysis of an engineered FhuA channel protein embedded into a lipid bilayer or a detergent belt Francisco Rodríguez-Ropero ⇑, Marco Fioroni ⇑ Department of Biotechnology, RWTH Aachen University, Worringer Weg 1, 52074, Aachen, Germany

a r t i c l e

i n f o

Article history: Received 19 September 2011 Received in revised form 6 December 2011 Accepted 7 December 2011 Available online 11 January 2012 Keywords: Molecular dynamics Channel proteins Protein engineering Nanotechnological applications b-Barrel Lipid bilayer

a b s t r a c t Engineered channel proteins are promising nano-components with applications in nanodelivery and nanoreactors technology. Because few of the engineered channel proteins have been crystallized, solution studies based on Neutron Scattering, Circular Dichroism and NMR play a major role. Consequently, the understanding of membrane proteins dynamics in water/detergent solutions or when embedded in a lipid membrane, can clarify how the environment affects protein behavior. In this study, molecular dynamics simulations of the FhuA Escherichia coli outer membrane channel protein and its engineered FhuA D1–159 variant have been performed in two different environments: a DNPC (1,2-dinervonyl-snglycero-3-phosphocholine) lipid bilayer and a water/OES (N-octyl-2-hydroxyethyl sulfoxide) detergent solution. Furthermore the FhuA D1–159 variant has been simulated in the open and closed states, the last induced by the presence of six 3-(2-pyridyldithio)-propionic-acid in the channel inner core. Differences in protein structural and dynamical behavior between the two environments have been found. Considering the FhuA protein characterized by an elliptical–cylindrical symmetry: (a) neither variations on the secondary structure nor axial deformation have been observed in any of the systems; (b) the ellipticity of the channel section (open state) and its fluctuations are enhanced in presence of water/OES, while diminished or suppressed in the DNPC bilayer; (c) the insertion of hydrophobic pyridyl groups into the FhuA D1–159 channel (closed state) induces a higher ellipticity in water/OES solution, while shifting to a circular section in the DNPC membrane; (d) the cork domain represented by the first 159 amino acids does not play a major role for protein stability. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction The aim of nanotechnology is the design of new functional materials and devices through the control of their organization at the atomic and molecular level. One common strategy when designing a new nanodevice is to survey naturally occurring structures with the natural ability to perform the desirable process and use them as scaffold to build the new nanoconstruct (Lowe, 2000). In this sense, the intrinsic nanoscale architecture and rich chemistry of proteins, as well as their catalytic activity in the case of enzymes, may be exploited to build a wide range of specific components in more sophisticated nanosized devices such as nanomotors (Heuvel and Dekker, 2007), nanoreactors (Vriezema et al., 2005) or sensors (Astier et al., 2005).

⇑ Corresponding authors. Present address: Center of Smart Interfaces, TU Darmstadt, Petersenstrasse 32, 64287 Darmstadt, Germany (F. Rodríguez-Ropero), Fraschetta Er Core de Roma, Rennbahn 1, 52062 Aachen, Germany (M. Fioroni). E-mail addresses: [email protected] (F. Rodríguez-Ropero), mfi[email protected] (M. Fioroni). 1047-8477/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jsb.2011.12.021

Following the aforementioned ideas, Meier and coworkers prepared for the first time a nanoreactor in which the outer membrane protein OmpF was inserted into a triblock copolymer PMOXAPDMS-PMOXA (PMOXA: poly-2-methyl-2-oxazoline, PDMS: polydimethyl-siloxane) vesicle containing b-lactamase enzyme (Nardin et al., 2000). Conceptually such nanoreactor or ‘‘synthosome’’ (Onaca et al., 2008), mimics the natural way to permeate specific substrates across cellular outer membranes where the lipid bilayer has been replaced by a polymeric matrix. Thus synthosomes combine the great selectivity of outer membrane proteins with the exceptional technological properties of polymersomes, i.e. mechanical, temperature and pH change stability, substrate and product tolerance (Lalli et al., 2011; Onaca et al., 2006). Synthosomes can be improved by engineering either the protein channel or the polymers constituting the membrane. Concerning the latter aspect, the shape and properties of polymer vesicles can be shifted by modifying the molecular weight of the block copolymer, the mass or volume fraction of each block and the effective interaction energy between monomers in the block (Discher and Eisenberg, 2002). On the other hand up to date OmpF (Ihle et al., 2011; Meier et al., 2000; Nardin et al., 2000, 2001;

292

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

Ranquin et al., 2005), Tsf (Ranquin et al., 2005), Aquaporin Z (Kumar et al., 2007) and FhuA (Nallani et al., 2006; Onaca et al., 2008) Escherichia coli channel proteins have been inserted into different block copolymer membranes. While Aquoporin Z is formed by six transmembrane helices and two half-membrane spanning helices in a right-handed helical bundle forming a 28 Å long channel with a narrow cross-section of 12.5 Å2 (Savage et al., 2003), OmpF, Tsf and FhuA proteins are formed by, respectively, 16-, 12- and 22-stranded b-barrels with diameters spanning from 15– 20 Å in Tsf (Ye and Berg, 2004) to 39–46 Å in FhuA(Guven et al., 2010; Koebnik et al., 2000). The wider diameter of FhuA resulting in a bigger pore area may help to overcome diffusion problems involving synthosomes in biotechnological applications (Nallani et al., 2006). FhuA (Ferric hydroxamate uptake protein component, PDB code: 1BY3) is a member of a family of integral outer membrane proteins. Its biological role is to function as receptor of the ferrichrome-iron peptide in E. coli bacteria, which, coupled with the energy dependent protein TonB, mediates the active transport of ferric siderophores across the outer membrane of Gram bacteria (Ferguson et al., 1998). Its primary sequence is composed by 714 amino acids, where residues 160–714 form a cylindrical b-barrel domain closed in the periplasmic side by the globular N-proximal domain formed by residues 1–159 (Braun et al., 2000; Ferguson et al., 1998, 2000; Locher et al., 1998; Pawelek et al., 2006). The exact mechanism of the TonB-FhuA siderophore-mediated transport machinery is still unknown. However, ferrichromeiron-induced conformational changes and transmembrane signaling are responsible for the iron transport by a possible unplugging/extrusion of the inner cork domain (Galdiero et al., 2007; Gumbart et al., 2007) or by forming a channel, as suggested by cross sectional fluctuations correlated to conductance studies (Ferguson et al., 1998; Locher et al., 1998). However in biotechnology applications, the presence of the cork domain reduces or even precludes the use of FhuA protein as an efficient mass transfer channel (Nallani et al., 2006; Onaca et al., 2006). To reach the goal, two FhuA mutants have been engineered and efficiently expressed in E. coli bacteria (Guven et al., 2010) in which the cork domain has been completely (FhuA D1–159, i.e. deletion of amino acids 1–159) (Braun et al., 1999, 2002; Dworeck et al., 2011) or partially (FhuA D1–129, i.e. deletion of amino acids 1–129) (Nallani et al., 2006) removed. The former mutant has been proven to work as a large passive diffusion channel able to translocate calcein (Onaca et al., 2008), TMB (3,30 ,5,50 -tetramethylbenzidin) (Guven et al., 2010) and single stranded DNA molecules (Onaca et al., 2006) while the latter, in spite of the difficulties to translocate small molecules, displays a channel smaller channel cross-section due to the presence of an additional b-strand inside the b-barrel (Nallani et al., 2006). Insertion of FhuA D1–159 in polyethylene-poly(ethyleneglycol) (Dworeck et al., 2011) and PIB1000–PEG6000–PIB1000 (PIB = polyisobutylene; PEG = polyethyleneglycol) (Muhammad et al., 2011) diblock copolymers has been recently reported. It should be noted that in this last case the protein hydrophobic region was elongated from 3 to 4 nm by ‘‘copy-pasting’’ the last five amino acids of each b-sheet on the periplasmatic side of the b-barrel. More recently a Fhua D1–159 variant with a wider diameter was engineered by doubling the first two b-strands (Krewinkel et al., 2011). To introduce an open/closed switch within the channel, a chemically modified FhuA D1–159 mutant, in which pyridyl- or biotinyl-species were selectively bioconjugated to lysine residues, was successfully used to develop a reduction triggered synthosome loaded with calcein, showing a release strongly modulated by the size of the labeling reagents (Guven et al., 2010; Onaca et al., 2008). In order to engineer new FhuA variants to be used as mass transfer channels in polymersomes a deeper insight into their

structural and dynamical behavior appears to be necessary. In fact the engineered FhuA embedded into polymeric membranes opens a further set of problems related to the hydrophobic mismatch (Andersen and Koeppe, 2007; Xu et al., 2008) or mechanical stress experienced by the protein in the new artificial environment. Furthermore because none of the engineered channel proteins have been crystallized, structural information is obtained mainly in solution studies, i.e. Neutron Scattering, Circular Dichroism or NMR. As a consequence, the understanding and comparison of the membrane proteins dynamical behavior in water/detergent solutions (Gall et al., 2002) or when embedded into a lipid membrane, can clarify how the environment modifies the protein dynamics especially if solution behavior is ‘‘extrapolated’’ to studies in lipid or polymer membranes. Given the intrinsic time scale and the atomistic nature of such phenomena, molecular dynamics (MD) simulations provide an appropriate theoretical framework to accomplish such objectives (Adcock and McCammon, 2006). Here, we report a comprehensive structural analysis of the FhuA, FhuA D1–159 (open channel) and FhuA D1–159 pyridyl-labeled (closed channel, where the inner channel Lys-NH2 are labeled by hydrophobic pyridyl groups (Onaca et al., 2008)) based on five independent 50 ns long MD simulations in aqueous solution/OES detergent (N-octyl-2-hydroxyethyl sulfoxide) or embedded into a DNPC (1,2-dinervonyl-sn-glycero3-phosphocholine) lipid bilayer. A detailed analysis of the protein dynamics is conducted to understand whether studies in solution (Gall et al., 2002) can be compared to those within lipid bilayers. The general aim of this study is a step toward the understanding, at the atomistic level, of the behavior of the FhuA and FhuA D1–159 protein variant (D) for further engineering and optimization in applications such as nanochannels in nanodelivery or nanoreactors technology.

2. Methods 2.1. System description FhuA protein is formed by 22 b-strands of variable length connected by short periplasmic turns and long extracellular loops, plus a globular cork domain embedded into a cylindrical b-barrel. The whole FhuA sequence is displayed in Table 1. Though the first 19 residues are not visible in the electron density map, their existence has been proven after six cycles of sequential Edman degradation (Locher et al., 1998). The FhuA D1–159 engineered protein is obtained by deleting the whole plug domain (first 159 aminoacids) in the FhuA wild type (Onaca et al., 2008). Based on previous works (Faraldo-Gómez et al., 2003; Locher et al., 1998), FhuA D1–159 has been divided in three topological domains, i.e. b-strands (b), loops (L) and turns (T), which are defined in Fig. 1 and Table 2. The b-barrel domain is characterized by an elliptic cylinder shape with a cross section of 46  39 Å in the part delimited by the b-turns (T region), compared to the loop side (L region) with a smaller cross section of 40  35 Å (Fig. 1, data deduced from crystal structure). The b-strands and b-turns in the L and T regions are extremely flexible while the hydrophobic b-barrel domain maintains a rigid geometry (Faraldo-Gómez et al., 2003) opening the possibility to perform simple geometric analyses (Fig. 1). In Fig. 2 geometrical parameters of the hydrophobic b-barrel domain are shown, where DL and DT refer to the longest axis in the extracellular and periplasmatic sides respectively, dL and dT refer to their shorter counterparts and h is the height of the hydrophobic b-barrel domain. Centers of mass of the hydrophobic bbarrel domain (CM), mass centers of the two amino acid rings placed at the border of the rigid b-strands domain (CML and

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301 Table 1 FhuA sequence. Amino acids belonging to b-strands and loop domains are highlighted with underline and italics respectively. Pyridyl-labeled lysine residues are highlighted in bold. Amino acids belonging to the hydrophobic b-barrel domain are doubly underlined. Plug domain SAWGPAATIAARQSATGTKTDTPIQKVPQSISVVTAEEMALHQPKSVKEALSYTPGV SVGTRGASNTYDHLIIRGFAAEGQSQNNYLNGLKLQGNFYNDAVIDPYMLERAEIMR GPVSVLYGKSSPGGLLNMVSKRPTTE b-Barrel domain PLKEVQFKAGTDSLFQTGFDFSDSLDDDGVYSYRLTGLARSANAQQKGSEEQRYAIA PAFTWRPDDKTNFTFLSYFQNEPETGYYGWLPKEGTVEPLPNGKRLPTDFNEGAKNN TYSRNEKMVGYSFDHEFNDTFTVRQNLRFAENKTSQNSVYGYGVCSDPANAYSKQC AALAPADKGHYLARKYVVDDEKLQNFSVDTQLQSKFATGDIDHTLLTGVDFMRMR NDINAWFGYDDSVPLLNLYNPVNTDFDFNAKDPANSGPYRILNKQKQTGVYVQDQA QWDKVLVTLGGRYDWADQESLNRVAGTTDKRDDKQFTWRGGVNYLFDNGVTPYF SYSESFEPSSQVGKDGNIFAPSKGKQYEVGVKYVPEDRPIVVTGAVYNLTKTNNLMA DPEGSFFSVEGGEIRARGVEIEAKRPLSASVNVVGSYTYTDAEYTTDTTYKGNTPAQV PKHMASLWADYTFFDGPLSGLTLGTGGRYTGSSYGDPANSFKVGSYTVVDALVRYD LARVGMAGSNVALHVNNLFDREYVASCFNTYGCFWGAERQVVATATFRF

CMT) as well as the tilt of the strands toward the barrel axis (a) are included together with the reference direction of the backbone (C) and hydrogen bond direction (H). Ellipses eccentricities of both domain sides (eL and eT) have been calculated by using Eqs. (1) and (2):

h

eL ¼ ðDL =2Þ2  ðdL =2Þ2

i1=2

=ðDL =2Þ

ð1Þ

eT ¼ ½ðDT =2Þ2  ðdT =2Þ2 1=2 =ðDT =2Þ

293

ð2Þ

Used throughout the text, we will refer to the analyzed quantities to be ‘‘axial-latitudinal’’ if parallel or ‘‘equatorial-longitudinal’’ if perpendicular to the cylinder main axis. 2.2. MD simulations All the simulations were performed using GROMACS 4.0.7 (Berendsen et al., 1995) molecular dynamics simulation package (www.gromacs.org (Lindahl et al., 2001)). Each simulated system was placed in the center of a dodecahedron for the protein/ detergent/water simulations or triclinic box (a = b – c and a = b = c = 90°) for the lipid bilayer. Dodecahedron boxes are the smallest and most regular space-filling unit cells being each of the 12 periodic images at the same distance and allows saving about 29% of CPU-time when simulating spherical macromolecules (Lindahl et al., 2001), as it is in the case of the protein embedded in the detergent belt. Explicit water molecules, represented by the SPC model (Spoel et al., 2005) was used. Due to the different characteristics between the systems, refer to ‘‘System assembling and equilibration’’ section for further details. Periodic boundary conditions were applied and a time step of 2 fs was used for the numerical integration of the equations of motion. Atomic coordinates were saved every 5 ps. Simulations were conducted at a constant temperature of 300 K and a constant pressure of 1 bar (Berendsen et al., 1984). Solvent (i.e. 100 mM salt solution plus counterions), OES (N-octyl-2-hydroxyethyl sulfoxide) or DNPC (1,2-dinervonyl-sn-glycero-3-phosphocholine) and protein were independently coupled to a temperature bath, with a coupling constant of sT = 0.02 ps, by a V-rescale thermostat (Bussi et al., 2007). An isotropic pressure coupling for the water solution simulations was used, with a coupling constant of sP = 1.0 ps and a compressibility of 4.5  105 bar1 by a Berendsen barostat (Berendsen et al., 1984). All DNPC lipid bilayer simulations were performed using a semi-isotropic coupling. The GROMACS Force Field (ffG53a6) (Oostenbrink et al., 2004), in which aliphatic carbons are treated using the united atom representation, was used. Energy minimizations were performed using a steepest descent algorithm followed by a constrained molecular dynamics. Constraints on the protein backbone atoms were applied by a harmonic potential with a force constant of 10 kJ mol1 Å2 and slowly diminished to 0 kJ mol1 Å2 within the equilibration time. Bond distances were constrained using the LINCS algorithm (Hess et al., 1997) while the van der Waals interactions were modeled using a 6–12 Lennard–Jones potential with a cutoff at 1 nm. The electrostatic interactions were calculated by using the Particle Mesh Ewald algorithm (PME) (Darden et al., 1993) with a cutoff of 1 nm for the direct space calculation. The reciprocal space calculation was performed using a fast Fourier transform algorithm. Details for each of the simulated systems are provided in Table 3. 2.3. Parameterization of OES detergent, pyridyl label and DNPC lipid 2.3.1. OES detergent: N-octyl-2-hydroxyethyl sulfoxide OES bonding and LJ interactions were parameterized using the GROMOS96 G53a6(Oostenbrink et al., 2004) parameters, while charges were obtained by an ab initio estimation using the CHelpG (Breneman and Wiberg, 1990) procedure under the Gaussian09 program (Frisch et al., 2009) on an optimized OES geometry at the HF/6–31 + G⁄ level of theory. Topology of the OES can be found in the ESI.

Fig. 1. Upper panel. Axial projection of the FhuA D1–159. b-Strands (b) are connected by short turns (T) and extended loops (L). OES detergent molecules are represented by solid spheres. Lower panel. FhuA D1–159 embedded into the DNPC bilayer. Snapshot after 50 ns of simulation.

2.3.2. Pyridyl label: 3-(2-pyridyldithio)-propionic-acid In our group (Guven et al., 2010), to control the flux through the FhuA D1–159 channel a 3-(2-pyridyldithio)-propionic-acid

294

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

Table 2 Definition of the b-strands (b), loops (L) and turns (T) in the b-barrel domain. b1 L1 b2 T1 b3 L2 b4 T2 b5 L3 b6

161–169 170–172 173–183 184–189 190–202 203–208 209–223 224–226 227–242 243–273 274–290

T3 b7 L4 b8 T4 b9 L5 b10 T5 b11 L6

291–293 294–317 318–339 340–367 368–369 370–393 394–419 420–442 443–444 445–462 463–466

b12 T6 b13 L7 b14 T7 b15 L8 b16 T8 b17

467–485 486–489 490–501 502–515 516–527 528–534 535–551 552–558 559–579 580–581 582–597

L9 b18 T9 b19 L10 b20 T10 b21 L11 b22

598–611 612–622 623–629 630–639 640–654 655–666 667–674 675–681 682–704 705–714

To check the validity of the model, the area/lipid ratio was calculated finding a value of 0.70 ± 0.033 nm2 (309 K) in satisfactory agreement with the experimental value of 0.67 ± 0.019 nm2 (309 K) (Lewis and Engelman, 1983). The transition temperature is Tt = 297 K (Lewis and Engelman, 1983). 2.3.4. System assembling and equilibration Starting conformations of both FhuA and FhuA D1–159 were based on the crystal structure (PDB code: 1BY3). The FhuA D1– 159 engineered protein was obtained by deletion of the whole plug domain in the FhuA wild type, while pyridyl-labeled FhuA D1–159 was constructed by coupling the 3-(2-pyridyldithio)-propionicacid to the six lysines in the inner part of the channel: Lys167, Lys344, Lys364, Lys526, Lys545 and Lys575 (Guven et al., 2010; Onaca et al., 2008). 2.3.4.1. Detergent equilibration. OES (N-octyl-2-hydroxyethyl sulfoxide) detergent molecules (Table 3) were placed around the external protein hydrophobic ring followed by a minimization in vacuum maintaining the protein constrained. Water molecules were added together with Na+ counterions to balance the overall protein charge at pH 7.0 (Table 3). One ns of equilibration with constrains on the protein backbone at 309 K was followed by further 2 ns of molecular dynamics by releasing the protein constrains. 50 ns of production run were performed for each of the FhuA and FhuA D1–159 in the open and closed states.

Fig. 2. Hydrophobic b-barrel domain geometrical parameters. DL and DT: the longest axis in the extracellular and periplasmatic sides; dL and dT: their shorter counterparts; h: height of the hydrophobic b-barrel domain; CM: center of mass of the hydrophobic b-barrel domain; CML and CMT: center of masses of the b-barrel domain at the verges; a: tilt of the strands toward the barrel axis with the reference direction of the backbone (C) and hydrogen bond direction (H).

coupled to a Lys-NH2 was used (pyridyl label). The corresponding MD model was parameterized using the GROMOS96 G53a6 (Oostenbrink et al., 2004) bonded and LJ interactions, while the electrostatic contribution was derived by an ab initio estimation using the CHelpG (Breneman and Wiberg, 1990) procedure under the Gaussian09 program (Frisch et al., 2009) on an optimized 3-(2-pyridyldithio)-propionic-acid geometry at the HF/6–31 + G⁄ level of theory. Topology of the pyridyl label can be found in the ESI. 2.3.3. DNPC: 1,2-dinervonyl-sn-glycero3-phosphocholine The 1,2-dinervonyl-sn-glycero3-phosphocholine (DNPC) has been parameterized following the work of Kukol (Kukol, 2009) based on the GROMOS96 53a6 Force Field (Oostenbrink et al., 2004). Both monounsaturated acyl chains in the DNPC, 24:1, are represented by the nervonyl acyl radical, –[CH2]7CH@CH[CH2]13cis-15-Tetracoseno. Topology of the DNPC is reported in the ESI. Table 3 Details of the simulated systems studied in this work: Total number of particles (Natoms), number of OES molecules (NOES), number of solvent molecules (Nwaters), number of sodium atoms (NNa+) and number of DNPC molecules (NDNPC). System

Natoms

NOES

Nwaters

NNa+

NDNPC

Wild type FhuA (OES) D1–159 FhuA (OES) Pyridyl labeled D1–159 FhuA (OES) D1–159 FhuA (DNPC) Pyridyl labeled D1–159 FhuA (DNPC)

70225 66844 65555 82068 84036

100 130 100 – –

20584 19760 19437 21390 22016

12 14 20 14 20

– – – 184 184

2.3.4.2. DNPC bilayer equilibration. Initial coordinates of the 288 DNPC molecules (144 for each leaflet) were created by using a grid of 12(x)  12(y)  2(z), distributed in a triclinic box. The center of mass of each single lipid molecule was placed on a point grid, considering a spacing of VdW +0.2 nm in order to avoid malignant superpositions. The DNPC germinal conformer was selected in the ‘‘extended’’ conformation, where both alkyl chains are stretched in parallel. A further random rotation around the main molecular axis followed. The obtained double bilayer of 288 DNPC molecules was then slowly shrunk until VdW overlay resulted in a high potential energy. Water was finally added. An equilibration time of 50 ns was performed checking the convergence of the area/lipid ratio vs. time. 2.3.4.3. FhuA and D variants embedding into DNPC bilayer. Embedding of the FhuA and D variants into the previous equilibrated DNPC bilayer was performed by applying the ‘‘shrinking’’ method reported by Tieleman (Kandt et al., 2007). The method involves a spreading of the equilibrated lipids on a large spaced grid to be gradually shrunk until the density of the lipid bilayer reaches a plateau, ensuring a densely packed lipid environment around the protein. An equilibration time of 10 ns followed, monitoring the energy components of the lipid-lipid, lipid-protein, lipid-water and protein-water till a constant value was reached (Kandt et al., 2007). No deformation of the system was reported with the number of DNPC molecules equally distributed between the two leaflets. 3. Results and discussion 3.1. RMSD, RMSF and secondary structure comparison of the FhuA wild type and engineered FhuA D1–159 variants in OES and DNPC The conformational stability was measured by calculating the backbone Root-Mean-Square Deviation (RMSD) (Table 4) and the Root-Mean-Square Fluctuation (RMSF) of the individual residues averaged over the last 25 ns in OES detergent, (Fig. 3a–c) and embedded in the DNPC bilayer, (Fig. 3d and e). DNPC was selected

295

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

due to its long mono-unsaturated acyl chain (24:1), representative of the model bilayer used in liposome experiments (Guven et al., 2010). The RMSD of the wild type in OES (Fig. 3a–c and Table 4), shows a higher stability compared to the D variant in both open and closed states. Furthermore the plug domain, embedded in the internal b-barrel core, shows an averaged RMSD of 1.11 ± 0.12 Å underlining a very low mobility. The obtained results referred to the wild type protein are in qualitative agreement with those previously reported by Sansom and coworkers (Faraldo-Gómez et al., 2003) based on 10 ns long MD of the FhuA and FhuA-ferrichrome complex embedded in a phospholipid bilayer. In case of the D variant, the lack of the cork domain induces obvious higher fluctuations in all the T, L and b-strand domains. However a comparison between the values of the D variant in the open and closed states does not show a clear trend. In fact the L region in the closed state reports a lower deviation as consequentially shown also by the total protein backbone. Such behavior will be later explained when geometrical parameters defined on the FhuA geometric shape, i.e. elliptical cylinder, will be introduced. When embedded in the DNPC bilayer (Fig. 3d and e), the overall behavior of the D variant in both open and closed states is comparable to the FhuA wild type in detergent, though showing higher deviations. Interestingly the T region has a lower deviation if compared to all the simulations in OES, with no main differences

between the open and closed states (Table 4). This finding can be correlated to the two different environments experienced from the turns characterizing the T domain. In fact, while in the DNPC bilayer, the turns are buried into the DNPC colin heads layer, in presence of OES turns are more exposed to water inducing higher deviations (Fig. 1b). One further consideration must be taken into account. In the current simulations a protein/OES molar ratio of 1:100 has been used, a value typically employed to solubilize the FhuA and D variants (Guven et al., 2010). However at higher concentrations of OES over the CMC (Critical Micelle Concentration), FhuA is expected to be embedded in a detergent layer that mimics a biological lipid bilayer, predicting a similar behavior for both cases (Bond et al., 2004). For all three analyzed OES simulations, the shape of the detergent layer surrounding FhuA is comparable to an oblate micelle as previously reported and suggested (Boeckmann and Caflisch, 2005; le Maire et al., 2000). The secondary structure time evolution of each of the five systems under study was determined by using the DSSP algorithm (Kabsch and Sander, 1983) (see ESI) showing no major variations in the secondary structure within the simulation time. Minimal differences might be due to the different starting conditions instead of a ‘‘real’’ different evolution between the systems, furthermore underlining that 50 ns are far to the b-sheets refolding time (lsms scale) (Colombo et al., 2002).

Table 4 RMSD values calculated on the last 25 ns for the wild type FhuA, open and closed FhuA D1–159 variants in OES detergent and DNPC bilayer.

RMSD RMSD RMSD RMSD

[Å] [Å] [Å] [Å]

backbone b-strand loops turns

Wild type FhuA (OES)

D1–159 FhuA open (OES)

D1–159 FhuA closed (OES)

D1–159 FhuA open (DNPC)

D1–159 FhuA closed (DNPC)

2.02 ± 0.37 1.44 ± 0.23 1.70 ± 0.26 4.37 ± 1.05

3.92 ± 0.14 2.21 ± 0.30 4.14 ± 0.12 4.81 ± 0.41

3.37 ± 0.19 3.38 ± 0.35 2.75 ± 0.10 5.39 ± 0.51

2.48 ± 0.19 2.13 ± 0.22 2.40 ± 0.26 3.27 ± 0.40

3.20 ± 0.18 2.17 ± 0.23 3.43 ± 0.26 3.57 ± 0.30

Fig. 3. Evolution of the backbone RMSD relative to the initial configuration (upper part) and the RMSF averaged over the last 25 ns (lower part) for the FhuA wild type (a), FhuA D1–159 (b) and pyridyl-labeled FhuA (c) proteins in OES and for the FhuA D1–159 (d) and pyridyl-labeled FhuA (e) embedded in DNPC lipid bilayer. Horizontal lines in the RMSF plots represent each of the b-strands present in the b-barrel domain.

296

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

Fig. 3 (continued)

As previously reported, the cork domain does not play a crucial role in the stabilization of the FhuA protein secondary structure which is mainly correlated to the consecutive b-strands hydrogen bond network. These findings are in qualitative agreement with

the CD analysis performed on the D variants showing a clear b-sheet structure (Guven et al., 2010; Onaca et al., 2008). Figs. 4 and 5 depict for each simulated system the initial structure and two snapshots obtained after 25 and 50 ns of MD

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

297

Fig. 5. MD snapshots of the equatorial and axial projections of the FhuA D1–159 (a) and FhuA D1–159 pyridyl-labeled FhuA (b)at t = 0, 25 and 50 ns embedded in the DNPC bilayer. Bilayer has been removed for clarity purpose.

Fig. 4. MD snapshots of the equatorial and axial projections of the FhuA wild type (a), FhuA D1–159 (b) and pyridyl-labeled FhuA (c) proteins in OES detergent at t = 0, 25 and 50 ns. OES has been removed for clarity purpose.

simulation. By a fast glimpse it is clear the D variant embedded in the DNPC bilayer tends to be circular, while in the OES detergent, especially for the pyridyl labeled variant, the protein tends to maintain or even increase the ellipticity. To better analyze the aforementioned behavior relevant geometric parameters are discussed and defined in the next section. 3.2. Geometrical analysis comparison of the FhuA wild type and engineered FhuA D1–159 variants in OES and DNPC The rigid b-strand domain of the FhuA protein can be represented by a cylinder characterized by two elliptical bases, with slight different semi-axial dimensions. Following the geometrical parameters definition as shown in Fig. 2 and introducing the ellipticity parameter (Eq. (1) and (2)), calculated results for all the five simulations are reported in Table 5. Moreover the time evolution of

the height of the b-strands domain as well as for the elliptical axes DL, dL, DT and dT are shown in Fig. 6. In OES simulations, the height h of the b-strands domain is unperturbed neither by the plug domain deletion nor by the introduction of highly hydrophobic species such as the pyridyl labeling agent into the channel. In all cases the height of the b-barrel domain surrounded by detergent molecules remains 21 Å. Moreover the CML–CM–CMT angles (CM = center of mass of the hydrophobic b-barrel domain, CML and CMT = center of mass of the two amino acid rings placed at the border of the rigid b-strands domain, see Fig. 2) are very similar behaving, as expected, as a rigid structure. However major differences are observed in the elliptical axes. The presence of the pyridyl groups in the hydrophilic inner core induces an elongated shape in the elliptical channel leading to higher eccentricities eL and eT in comparison to the FhuA wild type and unlabeled FhuA D1–159. Interestingly the detergent belt, which behaves as an oblate micelle, is covering the protein surface during the whole trajectory being its average eccentricity comparable with those reported in Table 5 for each FhuA variant in OES (see ESI for the time evolution of the eccentricity of the detergent belt). Analyzing the D variant simulations embedded in the DNPC bilayer, the main elliptical axis and, consequentially, the ellipticity changes considerably, with a tendency to gain more circular shape as it can be seen in Fig. 6 b where DT  dT. This fact is especially remarkable in the labeled D variant. Furthermore the height h together with the CML–CM–CMT angle tends to higher values, with a Dh  1 Å and a D(CML–CM–CMT) 3° with an expected lower value of a defined as the tilt of the strands toward the barrel axis (Fig. 2). Interestingly between the methods used to analyze the FhuA and D variant stability, a different trend is found if geometrical parameters such as backbone RMSD, RMSF (Fig. 3), domain center of masses and channel ellipticity (Table 4) or if methods such as

298

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

Table 5 Averaged geometric parameters for the wild type FhuA, open and closed FhuA D1–159 proteins, calculated over the last 25 ns MD trajectories in OES detergent and DNPC bilayer.

h [Å] CML–CM–CMT angle [°] a [°] DT [Å] dT [Å]

eT DL [Å] dL [Å]

eL

Wild type FhuA (OES)

D1–159 FhuA open (OES)

D1–159 FhuA closed (OES)

D1–159 FhuA open (DNPC)

D1–159 FhuA closed (DNPC)

21.16 ± 0.24 171.60 ± 0.86 39.55 ± 4.30 51.16 ± 1.17 41.21 ± 1.25 0.589 ± 0.047 48.81 ± 0.51 36.71 ± 0.57 0.659 ± 0.021

21.43 ± 0.21 171.76 ± 0.92 37.80 ± 5.21 51.43 ± 1.95 38.77 ± 2.38 0.648 ± 0.075 46.45 ± 0.77 39.03 ± 1.19 0.537 ± 0.057

21.71 ± 0.18 171.65 ± 0.81 38.39 ± 4.65 52.19 ± 3.27 30.58 ± 3.92 0.802 ± 0.064 50.08 ± 0.66 34.38 ± 0.97 0.726 ± 0.025

22.42 ± 0.13 175.51 ± 0.88 37.66 ± 3.94 48.37 ± 1.33 42.98 ± 1.24 0.45 ± 0.09 46.86 ± 0.70 37.38 ± 0.98 0.60 ± 0.03

22.39 ± 0.15 174.38 ± 0.91 37.35 ± 0.35 47.21 ± 1.66 43.50 ± 1.62 0.57 ± 0.04 46.07 ± 0.81 37.45 ± 1.13 0.579 ± 0.049

Ramachandran dihedrals and hydrogen bond connections defining the secondary structures through DSSP (see ESI) are considered. In fact, while the ellipticity or the RMSD and RMSF show quite important differences between FhuA and D variants in both OES

and DNPC simulations, the secondary structure behaves similarly reporting no main differences. Therefore when channel proteins with a cylindrical symmetry similar to the FhuA family, i.e. OmpF, are considered, the ellipticity,

Fig. 6. Time evolution of the height of the b-strands domain (a), and the semiaxis DT and dT (b) and DL and dL (c) for the FhuA, open and closed FhuA D1–159 proteins. In (b) it should be noted that DT > dT at time 0.

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

CML–CM–CMT and a angles (Murzin et al., 1994) can be important parameters to gain further information on protein dynamics. In synthesis, based on the previous analysis it can be concluded: (a) both FhuA wild type and D variant show a differential behavior if parallel (axial-latitudinal) or perpendicular (equatorial–longitudinal) variations toward the main cylindrical axis are compared; (b) when embedded into a DNPC bilayer, the D variant tends to adopt a circular section, while in OES, enhances the original ellipticity. The first finding (a) is quite straightforward analyzing the hydrogen bond scheme of the b-barrel skeleton. A higher energy penalty is expected if perturbations parallel to the main axis are applied. In fact an axial-latitudinal deformation would involve a mutual sliding and stretching of the anti-parallel b-sheets characterizing the b-barrel domain, causing an ‘‘out of phase’’ of the hydrogen bonds. Such stretching applied to single a-helices (Ireta et al., 2005) or b-sheets (Rossmeisl et al., 2004) are known to induce structural transitions, i.e. a?p-helix or a?310-helix involving high energy penalties due to the re-organization of numerous hydrogen bonds. The axial-latitudinal rigidity is definitively correlated to the hydrophobic mismatch phenomenon, defined as the difference between the hydrophobic length of membrane proteins and the hydrophobic thickness of the membrane they span (Andersen and Koeppe, 2007; McIntosh and Simon, 2006). In fact the hydrophobic mismatch will determine if and how the protein will insert into the bilayer (Andersen and Koeppe, 2007; McIntosh and Simon, 2006), by a mutual re-organization of the lipid bilayer geometry and, though less, the geometry of the protein (Xu et al., 2008). In this context our simulations indicate that FhuA protein is very resistant to forces parallel to the membrane normal behaving as a rigid cylinder. This behavior is similar to that of Aquoporin-0 channel protein, whose structure remains unaffected independently of the lipid environment (dimyristoyl phosphatidylcholine bilayer or in E. coli polar lipids) inducing a reorganization of the lipid bilayer (Hite et al., 2010). The second finding (b) has been correlated to the different local density fluctuations (Dq) between the two media. Dq has been calculated considering the number of OES and DNPC molecules within a cave cylinder of 5 Å of thickness surrounding the b-strand domain, i.e. the only involved in the hydrophobic interactions with the alcane chains of DNPC and OES, and averaged in time. The Dq number gives the ‘‘feeling’’ on how the protein experiences changes in density in the nearby environment. In the OES case, the local fluctuations have been found 5% higher compared to the DNPC counterpart. This behavior is also intuitively expected being a slab of DNPC more isotropic compared to an OES micelle. Furthermore in OES the T domain is more exposed to the water environment, introducing a further noise to the protein-environment interactions, while in DNPC the same domain floats in a quite uniform colin sea. As a consequence the less isotropic medium made by OES compared to DNPC, introduces random perturbations to the FhuA D1– 159 protein, which enhances instead to suppress the elliptical shape. However why the elliptical shape in the FhuA wild type? In fact due to the 22 b-sheets equivalency, the FhuA protein would be expected to show a circular shape instead of an elliptical one. A glimpse to FhuA wild type underlines the presence of the central cork domain, characterized by an elliptical shape (43  35 Å, see ESI) with the two main axis aligned with the protein elliptical axis (minor vs. minor and major vs. major). By this observation, it can be proposed that the cork domain is the main responsible of the elliptical distortion found in the FhuA wild type b-barrel domain.

299

This finding can also explain why the D variant embedded in an isotropic medium like a DNPC bilayer show a clear tendency toward a circular shape, loosing the ‘‘memory’’ of the elliptical symmetry. Regarding the pyridyl labeled FhuA D1–159 (closed state), the water density inside the channel is in average 20% lower (997.0 and 797.6 kg/m3) than in its open state, due to the presence of pyridyl hydrophobic groups. This phenomenon involves a differential pressure on the protein walls inducing the geometrical changes and fluctuations observed during the MD simulations, particularly amplified in OES. We are aware that the previous findings are on the limit of the MD simulation time. For example, as previously reported, due to the FhuA D 1–159 symmetry, all the 22 b-sheets characterizing the rigid b-barrel core are equivalent. As a consequence the protein is expected to change ellipticity, inducing main axis ‘‘regression’’ around the ellipses perimeters, a phenomenon not noticed in the reported simulations. However within the same time frame, the FhuA and D variant behavior in OES and DNPC shows clear differences that underline a differential geometric behavior between the two media. Consequentially experimental data obtained in solution should be cautiously extrapolated to the lipid environment. 4. Conclusions The perspective gained in this work from our simulations of the FhuA and D variants in water/OES detergent and water/DNPC lipid bilayer can be summarized in four main conclusions: (a) All five systems under study, within the 50 ns of simulation time, are able to retain their secondary/tertiary structure with the FhuA D1–159 variant showing a remarkable stability independently from the presence of the cork domain. These findings are in agreement with kinetic and dichroic data analysis obtained in solution (Guven et al., 2010) or in a liposome (Guven et al., 2010) and polymersome (Onaca et al., 2008). The stability of FhuA protein and its D variant mainly derives from its intrinsic b-barrel architecture based on an inter-strand hydrogen bonding network. (b) Though the secondary structure of FhuA D1–159 (open and closed states) shows no main differences, the channel ellipticity is function of the environment and presence of the pyridyl moiety. In water/OES detergent the presence of hydrophobic pyridyl groups into the FhuA D1–159 channel (closed state), induces strong fluctuations in the main elliptical axes increasing the ellipticity, less perturbed if the open state or the FhuA wild type are considered. However when embedded in a DNPC bilayer, FhuA D1–159 in both open and closed states reduces its ellipticity toward a nearly circular one. Such behavior has been correlated considering the difference in local density fluctuations between the two environments, with the water/OES more anisotropic compared to the isotropic DNPC bilayer. Especially in the case of the (hydrophobic) pyridyl labeled FhuA D1–159, higher fluctuations and ellipticity are linked to the inner channel reduced water density. Previous results suggest that the FhuA crystal structure obtained by water/OES solutions (PDB ID: 1BY3) report a ‘‘memory’’ of the crystal showing an ellipticity lost when inserted into a lipid bilayer. Furthermore the ellipticity in the non-engineered FhuA seems to be induced by the ellipsoidal cork domain, through its volume and hydrogen bonds with the b-sheets characterizing the b-barrel structure. (c) A glimpse to the secondary structure analysis performed by Ramachandran plot and DSSP algorithm (Determination of Secondary Structure of Proteins), no main changes are

300

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301

observed between all five systems, while geometrical parameters like ellipticity, center of mass of the different domains and backbone RMSD or RMSF show quite interesting different evolutions. Based on our results, channel ellipticity and center of mass distance–angle of the different protein domains are important parameters that should be considered when cylindrical channel proteins are studied. (d) As expected, geometrical variations parallel (axial–latitudinal) or perpendicular (equatorial–longitudinal) to the main cylindrical axis of the FhuA D1–159 variant, show a different sensitivity, with the axial-latitudinal reporting lower changes/fluctuations in values compared to the equatorial– longitudinal one. Such behavior is a natural consequence of the hydrophobic mismatch governing the protein response to the external lipid bilayer. In fact to avoid unfavorable exposure of hydrophobic surfaces to a hydrophilic environment, the FhuA hydrophobic length must be similar to the hydrophobic bilayer thickness. However if such favorable condition is not satisfied, the protein and the lipid bilayer react by re-organizing the lipid bilayer and protein geometry. As a consequence, FhuA D1–159 can be well approximated to a semi-rigid cylinder in the latitudinal direction while showing a flexible behavior in the longitudinal one (i.e. anisotropic compressibility). In order to get a deeper insight onto the FhuA D1–159 variant behavior, further theoretical research when embedded into lipid bilayers with different chain length or block copolymers is required. Such computational studies, actually in progress, will better clarify the hydrophobic mismatch response, stability and performance of the channel protein, proposing a detailed and targeted strategy for a directed evolution approach. Acknowledgments The authors thank the BioNoCo (Biocatalyis using non-conventional media, Deutsche Forschungsgemeinschaft), RWTH Aachen University and the computational resources offered by the RZ (Rechen Zentrum, Super Computing Center) RWTH Aachen University. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jsb.2011.12.021. References Adcock, S.A., McCammon, J.A., 2006. Molecular Dynamics: Survey of methods for simulating the activity of proteins. Chem. Rev. 106, 1589–1615. Andersen, O.S., Koeppe II, R.E., 2007. Bilayer thickness and membrane protein function: an energetic perspective. Annu. Rev. Biophys. Biomol. Struct. 36, 107– 130. Astier, Y., Bayley, H., Howorka, S., 2005. Protein components for nanodevices. Curr. Opin. Chem. Biol. 9, 576–584. Berendsen, H.J.C., Spoel, D.v.d., Drunen, R.v., 1995. GROMACS: a message-passing parallel molecular dynamics implementation. Comput. Phys. Commun. 91, 43– 56. Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., Haak, J.R., 1984. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81, 3684– 3690. Boeckmann, R.A., Caflisch, A., 2005. Spontaneous formation of detergent micelles around the outer membrane protein OmpX. Biophysical J. 88, 3191–3204. Bond, P.J., Cuthbertson, J.M., Deol, S.S., Sansom, M.S.P., 2004. MD simulations of spontaneous membrane protein/detergent micelle formation. J. Am. Chem. Soc. 126, 15948–15949. Braun, M., Killmann, H., Braun, V., 1999. The beta-barrel domain of FhuD5–160 is suffcient for TonB-dependent FhuA activities of Escherichia coli. Mol. Microbiol. 33, 1037–1049. Braun, M., Killmann, H., Maier, E., Benz, R., Braun, V., 2002. Diffusion through channel derivatives of the Escherichia coli FhuA transport protein. Eur. J. Biochem. 269, 4948–4959.

Braun, V., Braun, M., Killmann, H., 2000. Iron transport in Escherichia coli – crystal structure of FhuA, an outer membrane iron and antibiotic transporter. Adv. Exp. Med. Biol. 485, 33–43. Breneman, C.M., Wiberg, K.B., 1990. CHelpG. J. Comput. Chem. 11, 361–373. Bussi, G., Donadio, D., Parrinello, M., 2007. Canonical sampling through velocityrescaling. J. Chem. Phys. 126, 014101. Colombo, G., Roccatano, D., Mark, A.E., 2002. Folding and stability of the threestranded b-sheet peptide Betanova: insights from Molecular Dynamics simulations. Proteins: Struct., Funct., Genet. 46, 380–392. Darden, T., York, D., Pedersen, L., 1993. Particle mesh Ewald: an N log(N) method for Ewald sums in large systems. J. Chem. Phys. 98, 10089–10092. Discher, D.E., Eisenberg, A., 2002. Polymer vesicles. Science 297, 967–973. Dworeck, T., Petri, A.-K., Muhammad, N., Fioroni, M., Schwaneberg, U., 2011. FhuA deletion variant D1–159 overexpression in inclusion bodies and refolding with polyethylene-poly(ethylene glycol) diblock copolymer. Protein Expres. Purif. 77, 75–79. Faraldo-Gómez, J.D., Smith, G.R., Sansom, M.S.P., 2003. Molecular Dynamics simulations of the bacterial outer membrane protein FhuA: a comparative study of the ferrichrome-free and bound states. Biophys. J. 85, 1406–1420. Ferguson, A.D., Hofmann, E., Coulton, J.W., Diederichs, K., Welte, W., 1998. Siderophore-mediated iron transport: crystal structure of FhuA with bound lipopolysaccharide. Science 282, 2215–2220. Ferguson, A.D., Braun, V., Fiedler, H.P., Coulton, J.W., Diederichs, K., Welte, W., 2000. Crystal structure of the antibiotic albomycin in complex with the outer membrane transporter FhuA. Protein Sci. 9, 956–963. Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., et al. 2009. Gaussian 09 Gaussian 09, Revision A.1 ed. Gaussian, Inc., Wallingford, CT. Galdiero, S., Galdiero, M., Pedone, C., 2007. Beta-barrel membrane bacterial proteins: structure, function, assembly and interaction with lipids. Curr. Prot. Pept. Sci. 8, 63–82. Gall, A., Seguin, J., Robert, B., Bellissent-Funel, M.-C., 2002. Membrane proteins in bulk solution can be used for quasi-elastic neutron scattering studies: the case for the photochemical reaction center. J. Phys. Chem. B 106, 6303– 6309. Gumbart, J., Wiener, M.C., Tajkhorshidz, E., 2007. Mechanics of force propagation in TonB-dependent outer membrane transport. Biophysical J. 93, 496–504. Guven, A., Hauer, B., Fioroni, M., Schwaneberg, U., 2010. Molecular understanding of sterically controlled compound release through an engineered channel protein (FhuA). J. Nanobiotechnol. 8, 14. Hess, B., Bekker, H., Berendsen, H.J.C., Fraaije, J.G.E.M., 1997. LINCS: a linear constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472. Heuvel, M.G.L.v.d., Dekker, C., 2007. Motor proteins at work for nanotechnology. Science 317, 333–336. Hite, R.K., Li, Z., Walz, T., 2010. Principles of membrane protein interactions with annular lipids deduced from aquaporin-0 2D crystals. EMBO J. 29, 1652. Ihle, S., Onaca, O., Rigler, P., Hauer, B., Rodríguez-Ropero, F., Fioroni, M., Schwaneberg, U., 2011. Nanocompartments with a pH release system based on an engineered OmpF channel protein. Soft Matter 7, 532–539. Ireta, J., Neugebauer, J., Schleffer, M., Rojo, A., Galván, M., 2005. Structural transitions in the polyalanine alfa helix under uniaxial strain. J. Am. Chem. Soc. 127, 17241–17244. Kabsch, W., Sander, C., 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22, 2577–2637. Kandt, C., Ash, W.L., Tieleman, D.P., 2007. Setting up and running molecular dynamics simulations of membrane proteins. Methods 41, 475–488. Koebnik, R., Locher, K.P., Gelder, P.V., 2000. Structure and function of bacterial outer membrane proteins: barrels in a nutshell. Mol. Microbiol. 37, 239–253. Krewinkel, M., Dworeck, T., Fioroni, M., 2011. Engineering of an E. coli outer membrane protein FhuA with increased channel diameter. J. Nanobiotechnol. 9, 33. Kukol, A., 2009. Lipid models for united-atom molecular dynamics. J. Chem. Theory Comput. 5, 615–626. Kumar, M., Grzelakowski, M., Zilles, J., Clark, M., Meier, W., 2007. Highly permeable polymeric membranes based on the incorportation of the functional water channel protein Aquaporin Z. Proc. Natl. Acad. Sci. U S A 104, 20719–20724. Lalli, M., Facey, S.J., Hauer, B., 2011. Protein containers—promising tools for the future. ChemPhysChem 12, 1519–1521. le Maire, M., Champeil, P., Moller, J.V., 2000. Interaction of membrane proteins and lipids with solubilizing detergents. Biochim. Biophys. Acta 1508, 86–111. Lewis, B.A., Engelman, D.M., 1983. Lipid bilayer thickness varies linearly with acyl chain length in fluid phosphatidylcholine vesicles. J. Mol. Biol. 166, 211–217. Lindahl, E., Hess, B., Spoel, D.v.d., 2001. GROMACS 3.0: a package for molecular simulation and trajectory analysis. J. Mol. Model. 7, 306–317. Locher, K.P., Rees, B., Koebnik, R., Mitschler, A., Moulinier, L., Rosenbusch, J.P., Moras, D., 1998. Transmembrane signaling across the ligand-gated FhuA receptor: crystal structures of free and ferrichrome-bound states reveal allosteric changes. Cell 95, 771–778. Lowe, C.R., 2000. Nanobiotechnology: the fabrication and applications of chemical and biological nanostructures. Curr. Opin. Chem. Biol. 10, 428–434. McIntosh, T.J., Simon, S.A., 2006. Roles of bilayer material properties in function and distribution of membrane proteins. Annu. Rev. Biophys. Biomol. Struct. 35, 177– 198. Meier, W., Nardin, C., Winterhalter, M., 2000. Reconstitution of channel proteins in (polymerized) ABA triblock copolymer membranes. Angew. Chem. Int. Ed. 39, 4599–4602.

F. Rodríguez-Ropero, M. Fioroni / Journal of Structural Biology 177 (2012) 291–301 Muhammad, N., Dworeck, T., Fioroni, M., Schwaneberg, U., 2011. Engineering of the E. coli outer membrane protein FhuA to overcome the hydrophobic mismatch in thick polymeric membranes. J. Nanobiotechnol. 9, 8. Murzin, A.G., Lesk, A.M., Chothia, C., 1994. Principles determining the structure of beta barrels in proteins. J. Mol. Biol. 236, 1369–1381. Nallani, M., Benito, S., Onaca, O., Graff, A., Lindemann, M., Winterhalter, M., Meier, W., Schwaneberg, U., 2006. A nanocompartment system (synthosome) designed for biotechnological applications. J. Biotechnol. 123, 50–59. Nardin, C., Widmer, J., Winterhalter, M., Meier, W., 2001. Amphiphilic block copolymer nanocontainers as bioreactors. Eur. Phys. J. E 4, 403–410. Nardin, C., Thoeni, S., Widmer, J., Winterhalter, M., Meier, W., 2000. Nanoreactors based on (polymerized) ABA-triblock copolymer vesicles. Chem. Commun., 1433–1434. Onaca, O., Nallani, M., Ihle, S., Schenk, A., Schwaneberg, U., 2006. Functionalized nanocompartments (synthosomes): limitations and prospective applications in industrial biotechnology. Biotechnol. J. 1, 795–805. Onaca, O., Sarkar, P., Roccatano, D., Friedrich, T., Hauer, B., Grzelakowski, M., Güven, A., Fioroni, M., Schwaneberg, U., 2008. Functionalized nanocompartments (synthosomes) with a reduction-triggered release system. Angew. Chem. Int. Ed. 47, 7029–7031. Oostenbrink, C., Villa, A., Mark, A.E., Van Gusteren, W.F., 2004. A biomolecular force field based on the free enthalpy of hydration and solvation: the

301

GROMOS force-field parameter sets 53A5 and 53A6. J. Comput. Chem. 25, 1656–1676. Pawelek, P.D., Croteau, N., Ng-Thow-Hing, C., Khursigara, C.M., Moiseeva, N., Allaire, M., Coulton, J.W., 2006. Structure of TonB in complex with FhuA, E. coli outer membrane receptor. Science 312, 1399–1402. Ranquin, A., Versées, W., Meier, W., Steyaert, J., Gelder, P.V., 2005. Therapeutic nanoreactors: combining chemistry and biology in a novel triblock copolymer drug delivery system. Nano Lett. 5, 2220–2224. Rossmeisl, J., Norskov, J.K., Jacobsen, K.W., 2004. Elastic effects behind cooperative bonding in beta sheets. J. Am. Chem. Soc. 126, 1314013143. Savage, D.F., Egea, P.F., Robles-Colmenares, Y., O’Connell, Stroud, R.M., 2003. Architecture and selectivity in aquaporins: 2.5 angstrom X-ray structure of aquaporin Z. PLoS Biol. 1, 334–340. Spoel, D.V.d., Lindahl, E., Hess, B., Groenhof, G., Mark, A.E., Berendsen, H.J.C., 2005. GROMACS: fast, flexible, and free. J. Comput. Chem. 26, 1701–1718. Vriezema, D.M., Comellas-Aragonès, M., Elemans, J.A.A.W., Cornelissen, J.J.L.M., Rowan, A.E., Nolte, R.J.M., 2005. Self-assembled nanoreactors. Chem. Rev. 105, 1445–1489. Xu, Q., Kim, M., David Ho, K.W., Lachowicz, P., Fanucci, G.E., Cafiso, D.S., 2008. Membrane hydrocarbon thickness modulates the dynamics of a membrane transport protein. Biophysical J. 95, 2849–2858. Ye, J., Berg, B.v.d., 2004. Crystal structure of the bacterial nucleoside transporter Tsx. EMBO J. 23, 3187–3195.