Construction, Molecular Modeling, and Simulation of Mycobacterium

Biomacromolecules 2004, 5, 1052-1065

1052

Construction, Molecular Modeling, and Simulation of Mycobacterium tuberculosis Cell Walls Xuan Hong and A. J. Hopfinger* Laboratory of Molecular Modeling and Design (MC 781), College of Pharmacy, The University of Illinois at Chicago, 833 South Wood Street, Chicago, Illinois 60612-7231 Received December 8, 2003; Revised Manuscript Received February 9, 2004

The mycobacterial cell wall is extraordinarily thick and tight consisting mainly of (1) long chain fatty acids, the mycolic acids, and (2) a unique polysaccharide, arabinogalactan (AG). These two chemical constituents are covalently linked through ester bonds. Minnikin (The Biology of the Mycobacteria; Academic: London, 1982) proposed that the mycobacterial cell wall is composed of an asymmetric lipid bilayer. The inner leaflet of the cell wall contains mycolic acids covalently linked to AG. This inner leaflet is believed to have the lowest permeability to organic compounds of the overall cell wall. Conformational search and molecular dynamics simulation were used to explore the conformational profile of AG and the conformations and structural organization of the mycolic acid-AG complex, and overall, an inner leaflet molecular model of the cell wall was constructed. The terminal arabinose residues of AG that serve as linkers between AG and mycolic acids were found to exist in four major chemical configurations. The mycolate hydrocarbon chains were determined to be tightly packed and perpendicular to the “plane” formed by the oxygen atoms of the 5-hydroxyl groups of the terminal arabinose residues. For Mycobacterium tuberculosis, the average packing distance between mycolic acids is estimated to be approximately 7.3 Å. Thus, Minnikin’s model is supported by this computational study. Overall, this modeling and simulation approach provides a way to probe the mechanism of low permeability of the cell wall and the intrinsic drug resistance of M. tuberculosis. In addition, monolayer models were built for both dipalmitoylphosphatidylethanolamine and dimyristoylphosphatidylcholine, two common phospholipids in bacterial and animal membranes, respectively. Structural comparisons of these cell wall phospholipid membrane models were made to the M. tuberculosis cell wall model. Introduction Tuberculosis (TB) is a chronic infectious disease caused mainly by Mycobacterium tuberculosis. Two billion people, one-third of the world’s population, are infected with latent TB. Ten percent of those infected will develop active TB in their lifetimes.2 Even more disconcerting, according to the World Health Organization (WHO), the global incidence of TB is still increasing.3 The increase is attributed to three factors, the breakdown in health services, the co-infection with HIV/AIDS, and the emergence of multidrug-resistant TB.4 Treating mycobacteria infections in general is more difficult than treating infections caused by most Grampositive and Gram-negative bacteria because mycobacteria are resistant to the majority of common antibiotics and chemotherapeutic agents.5 Mycobacteria are also relatively resistant to harsh environments such as drying and the presence of alkali and many chemical disinfectants, which makes prevention of M. tuberculosis transmission difficult. This general resistance and the specific resistance to therapeutic agents are thought to be related to the unusual structure of the mycobacterial cell wall. * Corresponding author. Telephone: 312.996.4816. Fax: 312.413.3479. E-mail: [email protected].

Mycobacteria belong to the Corynebacterium-Mycobacterium-Nocardia branch under the Gram-positive bacteria family. Unlike most Gram-positive bacteria that surround themselves with a porous peptidoglycan cell wall, mycobacteria have an extra cell wall structure that is extraordinarily thick and tight located outside the peptidoglycan layer.6 This cell wall structure contains (1) characteristic long chain fatty acids, mycolic acids, and (2) unique polysaccharides, arabinogalactan (AG).1,7,8 These two constituents are covalently linked together by forming ester bonds. The mycolyl-AG complex, in turn, attaches itself to the peptidoglycan to form the mycolyl-AG peptidoglycan complex. The mycobacterial cell wall also contains many other “free” lipid species, the so-called extractable lipids, that are not covalently linked to the AG-peptidoglycan complex and are solvent-extractable, such as glycolipids, phenolic glycolipids, glycopeptidolipids, and other chemical species.1,7-12 Mycobacterial mycolic acids are distinct from most fatty acids in that (1) they are longer, containing 70-90 total carbon atoms, with a totally saturated R-branch of typically 24 carbons, and mero chains of 40-60 carbons, and (2) in the mero chain, there are normally only two positions that may be occupied by functional groups. The proximal position (nearer the β-hydroxy acid) contains exclusively cis- or transolefin or cyclopropane. However, the distal position may be

10.1021/bm034514c CCC: $27.50 © 2004 American Chemical Society Published on Web 04/06/2004

Mycobacterium tuberculosis Cell Walls

Biomacromolecules, Vol. 5, No. 3, 2004 1053

Figure 1. Structures of mycolic acids found in the M. tuberculosis cell wall. The usual value for k is 23. The sum of n, m, and h is approximately 50. The shorter chain is the R branch, and the longer one is the mero chain. Both R and β carbons have R chirality. (See text for details.)

the same as the proximal position or contain one of a variety of oxygen moieties such as R-methyl ketone, R-methyl methyl ether, methyl-branched ester, or R-methyl epoxide. The exact subclasses of mycolates present and their relative quantities are unique to an individual species or very closely related species, under defined growth conditions (see refs 13 and 14). There are three kinds of mycolates in the M. tuberculosis cell wall, R-mycolates, methoxymycolates, and ketomycolates (Figure 1). Both the R- and methoxymycolates only have the cis-cyclopropyl group at the proximal position, while 17% of the cyclopropyl groups in the keto proximal position are trans.15 R-Mycolates, being the most abundant mycolic acid type, represent about 51% of the total mycolates, whereas the methoxy form represents about 36% and the keto form about 13%.16 All three mycolates contain 24 and 26 carbon R branches in approximately the ratio of 10: 90, with negligible amounts of 22 carbon R branches. Methoxymycolates and ketomycolates have longer mero chains than the R-mycolates. The total carbon numbers for the R, methoxy, and keto forms are 76-82, 83-90, and 8489, respectively.14 The detailed chemical structure of the mycolyl-AG peptidoglycan complex was established through gas chromatography-mass spectrometry and fast atom bombardment mass spectrometry.17-19 Parts A and B of Figure 2 show the structures of an arabinose molecule (Ara) and a galactose molecule (Gal), respectively. Within AG, all arabinose and galactose residues are in the furanose form and the nonreducing termini of arabinan consist of the hexasugar motif [β- D -Araf-(1f2)-R- D -Araf] 2 -3,5-R- D -Araf-(1f5)-R- D Araf, as shown in Figure 2C. The majority of the arabinan chains consist of 5-linked R-D-Araf with branching introduced by 3,5-R-D-Araf. The arabinan chains are attached to C-5 of some of the 6-linked Galf. The galactan consists of

linear alternating 5- and 6-linked β-D-Galf, and the galactan region of AG is linked to a peptidoglycan via the linker disaccharidephosphate. The mycolic acids are located in clusters of four on the terminal hexa-arabinofuranoside through 1,5 linkages, but only two-thirds of these are mycolated. Figure 2D provides an overview of the mycolylAG peptidoglycan complex. A special class of proteins, the porins, are also present in the mycobacterial cell wall, though in very small amounts.20 Porins produce nonspecific and narrow aqueous diffusion channels across the cell wall to facilitate small molecule transport. In the early 1980s, Minnikin1 proposed the currently accepted physical organization model for the mycobacterial cell wall (Figure 3). In this model, the mycobacterial cell wall is composed of an asymmetric lipid bilayer while the inner leaflet contains mycolic acids covalently linked to AG and the outer leaflet contains the extractable lipids. The bulk of the mycolic acid hydrocarbon chains are tightly packed in a parallel fashion and predominantly oriented in a direction perpendicular to the cell wall surface. The extractable lipids accommodate the uneven surface of the inner leaflet caused by the two branches of mycolic acid that are not equal in length. Porins reside in the middle of mycolic acids and extractable lipids. The Minnikin model has received strong experimental support from biophysical structural characterization studies such as X-ray diffraction.21 Spin label studies conducted by Nikaido also suggests that the mycobacterial cell wall forms an asymmetric bilayer with the outer leaflet being moderately fluid and the inner leaflet having very low fluidity.22 This low fluidity of the inner leaflet may be the result of the unique chemical structures of mycolic acids and their tight packing, which may also account for the low permeability of the overall cell wall. It is important to point

1054

Biomacromolecules, Vol. 5, No. 3, 2004

Hong and Hopfinger

Figure 2. Structures of the chemical components of the mycolyl-AG peptidoglycan complex of mycobacteria and an overview of the organization of the complex (parts C and D are with permission from the Annual Review of Biochemistry, Volume 64, copyright 1995 by Annual Reviews www.annualreviews.org; minor modifications made).

out that although a porin produces a channel for the transport of some types of hydrophilic compounds, this hydrophilic pathway is inefficient owing to the small numbers of porins distributed over the mycobacterial cell wall.6,23 Most evidence suggests that a large number of compounds, even some hydrophilic anti-TB drugs such as isoniazid, across the

mycobacterial cell wall through the lipid bilayer without “help” from the porins.23-25 The cell wall permeability barrier prevents drugs from entering the cell, which attributes, at least in part, to the intrinsic drug resistance of mycobacteria.5 Even though experimental evidence favors a lipid bilayer model, the Minnikin model still faces a major criticism that


Figure 3. Mycobacterial cell wall model proposed by Minnikin. (The picture is taken from www.niaid.nih.gov/newsroom/focuson/tb02/ target.htm with modifications; reprinted with permission from Patrick Brennan. Copyright 2001 NIAID). The funnel-shape structure in the center of the cell wall represents a porin.

close packing of mycolic acid hydrocarbon chains is difficult because they are covalently linked to AG, a macromolecular polysaccharide.17,19,26 In this study, we have investigated the validity of the Minnikin model, sought additional insight regarding the M. tuberculosis cell wall structure, and explored the cell wall permeability barrier mechanism, using molecular modeling methodologies. First, the conformational profiles of AG were explored by conformational search methods. Next, the conformation and physical organization of the mycolyl-AG complex were studied using molecular dynamics simulation (MDS). Then, focusing on the inner leaflet, we constructed a M. tuberculosis cell wall model. Finally, to study the difference of physical organization between the M. tuberculosis cell wall and common biological phospholipid membranes, monolayer models composed of dimyristoylphosphatidylcholine (DMPC) and dipalmitoylphosphatidylethanolamine (DPPE) were constructed, respectively, and comparisons were made to the M. tuberculosis cell wall model. Conformational search methods have been widely used to study the conformation behavior of proteins and polymers.27-30 MDS methods have proven useful in probing time, temperature, and spatial profiles of phospholipid membranes.31-33 However, combining these two modeling methods and applying them to explore mycobacterial cell wall structure has not been reported. Methods 1. Construction of a M. tuberculosis Cell Wall Model. (1) Construction of the Arabinan Chain. It is not realistic to apply a conformational search to the entire AG complex of three arabinans and one galactan. The AG complex involves nearly 100 sugar residues and hundreds of torsion angles, and a complete conformational search is very time-consuming. Thus, the 17 arabinose residues that are close to the nonreducing termini of the arabinan chain and covalently linked to the mycolic acids were chosen to represent the structural characteristics of the AG complex (see Figure 4). In Figure 4 the eight arabinoses that are covalently linked to mycolic acids are represented as Ara_a to Ara_h. The rest


Figure 4. Chemical structures and reference coding of the 17 arabinose residues. The legends used in this figure are the same as those in Figure 2D. Inside the ellipse are 17 arabinose residues that have been studied by the conformational search method. The blue letters, a-h, represent the eight arabinose residues that bind with mycolic acids, and the blue italic numbers represent the remaining nine sugar residues.

of the sugars are indexed by numbers, such as Ara_1. The 17-arabinose chain was built in the neutral form using the Sugar Builder module of the HyperChem 6.01 software.34 The MM+ molecular mechanics force field also implemented in the HyperChem software package was utilized. The arabinose chain was subjected to conjugate gradient minimization with the Polak-Ribiere first derivative method35 until a derivative convergence criterion of 0.1 kcal/(mol Å) was reached or until a maximum of 4380 iterations were performed. The HyperChem software sets the maximum number of iterations to be 15 times the number of atoms in the system. The energy-minimization routine of the HyperChem software package, as just described, was applied throughout this study unless stated otherwise. The resulting energy-optimized arabinose chain structure was used as the initial structure in the conformational search. (2) Conformational Search of the Arabinan Chain. The random sampling conformational search module in the Cerius2 package was used to perform the conformational search.36 The Pcff_300_1.01 force field implemented in the package was applied, and partial atomic charges were calculated using Gast_original 1.0 method also from the Cerius2 program. New conformations generated after randomly altering torsion angles37 (also referred to as dihedrals) in the search model were subjected to conjugate gradient minimization using the Polak-Ribiere first derivative method until a derivative convergence criterion of 0.1 kcal/(mol Å) was reached or until a maximum of 2000 iterations had been performed. Ring torsion angles were not considered in this study. Only spatially unique conformers, defined as having an average corresponding atom positional root-mean-square difference greater than 0.5 Å, were retained in the analysis. (3) Construction of the Pseudo-Mycolic Acid (PMA). As mentioned in the Introduction, mycolic acids have two aliphatic chains of unequal length. The extra surface area of the longer chain, the mero chain, is covered by the extractable lipids (Figure 3). However, the precise identities, locations, and orientations of these extractable lipids remain unclear.

1056


Hong and Hopfinger

Figure 5. Chemical structure of a PMA molecule.

Moreover, the innermost part of the cell wall consisting solely of mycolic acids is believed to be the most tightly packed region and has the lowest permeability for the composite cell wall. Therefore, construction of this part of the M. tuberculosis cell wall is the focus of the work reported in this paper. PMAs were designed to model this cell wall region (Figure 5). There are 24 carbons in the R branch of a PMA molecule. The mero chain was shortened to the same length as the R branch. A cis-cyclopropyl group was chosen to be the functional group at the proximal position. The advantage of selecting this structural design is that a PMA molecule is able to represent most of the major types of mycolic acids present in M. tuberculosis because (1) they have similar structures except for the distal moieties (Figure 1) and (2) a cis-cyclopropyl group is the dominant functional group at the proximal position. PMAs were built using the HyperChem software. Each structure was energy-minimized, first using the MM+ molecular mechanics force field and then by the AM1 semiempirical method.38 Partial atomic charges were estimated by performing the single point calculation for the minimized structure. (4) Construction of the Mycolyl-Arabinan Complex. The conformations of the arabinan chain realized in step 2 were ranked with respect to their conformational potential energies. The low-energy conformations were chosen to form complexes with PMAs. For each of these low-energy conformations, the carboxyl groups of eight PMA molecules were linked to the 5-OH groups of the low-energy conformer to form the ester bonds (i.e., mycolic acids were transformed to mycolates). The resulting complexes were then energyminimized in HyperChem using the MM+ molecular mechanics force field. The complex with the lowest overall conformational energy was used as the starting structure in step 5. (5) MDSs of Mycolyl-Arabinan Complexes. The mycolyl-arabinan complex from step 4 has only eight mycolate chains, which is not adequate to study the properties of the whole cell wall. Therefore, “twin” complexes (i.e., two mycolyl-arabinan complexes of identical structures) were aligned relative to one another using the Crystal Builder module in the HyperChem package with the distance between each “twin” set at 6, 7, 8, and 9 Å. Overall, four sets of “twin” complexes were generated using the Crystal Builder. MDS was then employed to sample the conformational states available to each “twin” complex. The MDSs were performed using the MOLSIM package39 with an extended MM2 force field.40,41 Terminal carbons of mycolate mero chains were assigned heavy mass values of 1000 amu to simulate the restrictions in movement that would occur if the mero chains had not been shortened. The 1-OH of Ara_5 (Figure 4) was etherified by methanol, and a mass of 1000 amu was also

assigned to the methyl carbon. The reason for assigning this heavy mass is to take the influence of further binding into consideration. The periodic boundary conditions were altered in a manner corresponding to the different distances selected between the “twin” complexes. The temperature for the MDS was set at 311 K and held constant during the MDS by coupling the system to an external fixed-temperature bath.42 The simulation sampling time ranged from 90 to 130 ps with intervals of 0.001 ps for a total sampling of 90 000-130 000 conformations for the complexes. The atomic coordinates of conformations during the MDS were recorded every 0.1 ps. The distances among the ester oxygen atoms of the complexes were measured over the MDS trajectories, and the average value was calculated and used in step 6. (6) Construction and MDS of the Inner Leaflet Model. The carboxyl group of a PMA molecule was esterified with methanol, and the molecule generated was referred to as a pseudo-mycolate (PM). A (5 × 5 × 1) monolayer of PMs was built using the MI-QSAR software.43 The distance between a pair of PM molecules was set as the average value calculated from step 5 using the ester oxygens. The surface orientation angle γ was set as 90° because mycolates are perpendicular to the cell surface. The MOLSIM package with an extended MM2 force field was used to perform the MDSs in the same manner as in step 5 at a constant simulation temperature of 311 K. Methyl ester carbons and the terminal carbons of PM mero chains were each assigned heavy masses of 1000 amu, for the same reason as given in step 5. Various periodic boundary conditions were tested to find the optimal conditions corresponding to the most stable assemblies. The simulation sampling time was 210 ps with intervals of 0.001 ps. 2. Construction and MDS of a DPPE Monolayer. A single DPPE molecule was built using the HyperChem package. The structure was energy-minimized first under the MM+ molecular mechanics force field and then by the AM1 semiempirical method. Partial atomic charges were estimated from the AM1 calculation. The structure of a DPPE molecule is given in Figure 6. However, there is no consistent data reported regarding the distance between a pair of DPPE molecules in a biological membrane. The surface area of a membrane DPPE, measured from a surface pressure/area isotherm generated by the Langmuir-Blodgett (LB) method, is 38 Å2.44 This average area yields a distance between two DPPEs of about 6.16 Å, a value significantly smaller than the common distance among phospholipids in biological membranes. DLPE (dilauroylphosphatidylethanolamine), an analogue of DPPE with four less carbons for each aliphatic chain, was cocrystalized with acetic acids. The surface orientation angle γ and the distance between two DLPEs were reported as 92° and 7.773 Å, respectively.45 However,



Figure 6. Chemical structure of a DPPE molecule.

Figure 7. Chemical structure of a DMPC molecule.

the study done by Philips and Chapman indicates that DPPE and DLPE have quite different molecular areas.46 Hence, in this study the distance between a pair of DPPE molecules was varied from 6.5 to 8.0 Å with intervals of 0.5 Å, and four (5 × 5 × 1) DPPE monolayers were built accordingly. The MI-QSAR software was employed in the construction of DPPE monolayers. The surface orientation angle, γ, was defined as 92°. The MDSs were performed using the MOLSIM package under the same force field and temperature used for the inner leaflet model. Four sets of periodic boundary conditions were employed with the common c and γ values (i.e., 80 Å and 92°, respectively) and different values for a and b: [1] a ) 32.5 Å, b ) 32.5 Å; [2] a ) 35.0 Å, b ) 35.0 Å; [3] a ) 37.5 Å, b ) 37.5 Å; and [4] a ) 40.0 Å, b ) 40.0 Å. The monolayers were simulated for 70 ps with intervals of 0.001 ps. 3. Construction and MDS of a DMPC Monolayer. A DMPC molecule was built and energy-minimized in the same manner as for a DPPE molecule (see Figure 7 for the structure of a DMPC molecule). A (5 × 5 × 1) DMPC monolayer was constructed using the MI-QSAR package. The distance between a pair of DMPC molecules was set as 8 Å, and γ was defined as 96°, according to the available experimental data.47 Periodic boundary conditions were employed (a ) 40 Å, b ) 40 Å, c ) 80 Å, and γ ) 96°). The simulation sampling time was 70 ps with intervals of 0.001 ps. Additional information regarding the construction of a DMPC monolayer using the MI-QSAR software is given in refs 31-33. Results 1. M. tuberculosis Cell Wall Model. Figure 4 is a schematic representation of the 17 arabinose residues considered in the conformational search and simulation studies. All arabinose residues are in R configuration, except for Ara_a, c, e, and h, which are in the β configuration. 1,5 linkages connect the β arabinose residues to their adjacent R sugar residues. Ara_4 is linked to Ara_5 by a 1,5 linkage, whereas a 1,3 linkage connects Ara_5 to Ara_6. Hence, the two chains originating from Ara_5 experience different

Table 1. Relative Conformational Energies of the 16 Lowest Energies and Unique Conformations Generated from the Random Sampling Conformational Searcha rank

coding

∆E (kcal/mol)

rank

coding

∆E (kcal/mol)

1 2 3 4 5 6 7 8

conformer 1 conformer 2 conformer 3 conformer 4 conformer 5 conformer 6 conformer 7 conformer 8

0.0 0.4 3.7 4.0 4.2 4.4 5.7 6.0

9 10 11 12 13 14 15 16

conformer 9 conformer 10 conformer 11 conformer 12 conformer 13 conformer 14 conformer 15 conformer 16

6.1 7.3 7.8 7.9 7.9 8.9 9.9 10.9

a Relative conformational energies are calculated as the energy difference between the corresponding conformer and conformer 1.

chemical environments. This is also the case for Ara_a and Ara_b, Ara_c and Ara_d, Ara_e and Ara_f, and Ara_g and Ara_h pairs of chains. Therefore, each of the eight sugar residues that are linked to mycolic acids is distinct. The 1,5 and 1,3 linkages of the arabinan backbone also provide every arabinose residue freedom to move, which is also characteristic to arabinan chains. After an exhaustive conformational search, 4969 unique conformations were identified for the arabinan chain. From this set of unique conformations, the 16 conformations having conformational potential energies within 10 kcal/mol of the lowest potential energy were chosen as most plausible and their potential energies are given in Table 1. Among these 16 conformations, four chemical configuration patterns can be identified for the eight arabinose residues that bind mycolic acids. These chemical configurations are both shown and defined in Figure 8. The first distinct pattern is shown in Figure 8A and includes conformers 1-3, 7, 8, 10, 12, and 14. A representative geometry of conformer 1 is given for illustration purposes as a part of Figure 8A. This conformer is characterized by the oxygen atoms of the four 5-OH groups from Ara_a, b, c, and d being nearly coplanar with a small torsion angle of -6.4° (see ref 37 for the definition of this torsion angle). Moreover, the locations of these oxygens form an approximate rectangle with Ara_c

1058


Hong and Hopfinger

Figure 8. Chemical configurations of the 17 arabinose residues after random sampling conformational searching. The coding of arabinose residues is given next to their 5-hydroxyl groups. The square-like shape formed by straight lines between a, b, c, and d illustrate the chemical configuration of these four residues. The dashed lines indicate that the spatial locations of residues involved form an irregular shape. See text for details. (The carbon atoms are shown in gray and the oxygen atoms in red. The hydrogen atoms are not shown for clarity.)

and Ara_d close to Ara_b and Ara_a, respectively. In addition, Ara_b is close to Ara_e, while Ara_a and Ara_f are adjacent. The second pattern found for the eight arabinose residues comes from conformers 4 and 5 where Ara_b, a, and e are adjacent in space, while Ara_d, c, and f are close as can be seen in Figure 8B. The rectangular spatial relationship of the 5-hydroxyl oxygens of Ara_a, b, c, and d seen in the first pattern still holds for this second pattern. However, the torsion angle of these oxygens is relatively large. Conformer 4 has an angle of -25.0°, and conformer 5 has an angle of 23.6°. Pattern 3, shown in Figure 8C, includes conformers 6, 9, 13, 15, and 16. This pattern is similar to the first pattern. Ara_a is adjacent to Ara_c and Ara_f, whereas Ara_b is close to Ara_d and Ara_e. The torsion angle formed by the 5-hydroxyl oxygens of Ara_a, b, c, and d is small and the locations of the four oxygen atoms still form a rectangular shape in space. For example, the angle is 8.5° for conformer 6. The fourth pattern only includes conformer 11 (see Figure 8D). In this pattern Ara_a is adjacent to Ara_b, which is close to Ara_f. Ara_d is adjacent to Ara_c and Ara_e. The four oxygen atoms from Ara_a, b, c, and d, forming a rectangle, are on the same plane with a torsion angle of 0.9°. In the four chemical configuration patterns found among Ara_a, b, c, and d, these four sugar residues are closer to Ara_e and f than to Ara_g and h. However, the configuration pattern among Ara_e, f, g, and h is not clear. The 5-OH oxygen atoms of these four residues are not coplanar with each other or with the 5-OH oxygen atoms from Ara_a, b, c, and d. Their spatial locations do not form a regular and a

Table 2. Relative Conformational Energy Values for the Four Mycolyl-AG Complexes after Energy Minimizationa ∆E (kcal/mol) PMA-conformer 1 PMA-conformer 4 PMA-conformer 6 PMA-conformer 11

0.0 -90.7 +156.0 +77.4

a Relative conformational energies are calculated as the energy difference between the corresponding complex and the PMA-conformer 1 complex.

common shape across the 16 conformers. The energy differences among the 16 conformations are relatively small when one considers their different chemical configurations. These small differences in conformational energy are due to the great flexibility of the arabinan backbone. At this point, it is quite difficult to determine which chemical configuration is preferred. However, after energy minimization of the four mycolyl-arabinan complexes generated by attaching PMAs to the conformers that have the lowest energy from each of the four patterns, the PMAconformer 4 complex was found to have a significantly lower (more stable) energy as compared to the other three (see Table 2). The hydrocarbon chains of the eight mycolates of this complex are quite ordered and oriented perpendicular to the plane formed by the ester oxygen atoms as is shown in Figure 9. From the side view, Figure 9B, it appears that the cis-cyclopropyl groups bend the hydrocarbon chains. However, in general the long chains are tightly packed and in parallel alignment despite the presence of cis-cyclopropyl groups. The torsion angle for the 5-OH oxygen atoms of



Figure 9. Chemical structure of the PMA-conformer 4 complex after energy minimization.

Ara_a, b, c, and d is -14.4°, smaller than the value for conformer 4 found in the conformational search. The rectangular shape remains for these four oxygen atoms. In fact, the lengths of each of the four sides of the rectangle are close to one another, having the values of 7.4 Å for Ara_a-Ara_b, 7.0 Å for Ara_b-Ara_d, 6.3 Å for Ara_dAra_c, and 6.1 Å for Ara_c-Ara_a. Therefore, the locations of these oxygens form an approximate square. For Ara_e, f, g, and h, the torsion angle of their 5-OH oxygen atoms is 177.0°, indicating these four oxygen atoms are quite coplanar. However, their spatial locations form an approximate line, with the sequence of Ara_e-Ara_f-Ara_g-Ara_h, where the bending angle for Ara_e-Ara_f-Ara_g is 5.2° and the bending angle for Ara_f-Ara_g-Ara_h is 140.1° (see ref 48 for the definition of these bending angles). Overall, a clear and consistent chemical configuration for Ara_e, f, g, and h remains difficult to identify. Extensive MDS leads to the “gap” between the “twin” PMA-conformer 4 complexes decreasing, regardless of the different initial distances between the “twin” complexes, and the mycolates “merge” into a uniform chemical environment (see Figure 10). The mycolate hydrocarbon chains remain tightly packed and perpendicular to the cell surface at the end of the MDS. The cis-cyclopropyl groups cause the PMA assembly to slightly disorganize at the tails of R branches and mero chains. The ester oxygen atoms from Ara_a, b, c, and d remain coplanar and approximately square in shape. However, unambiguous identification of a distinct chemical configuration pattern for Ara_e, f, g, and h remains prob-

lematic. The ester oxygen atoms of these four residues are not coplanar, having torsion angles ranging from 50 to 150° for the “twin” PMA-conformer 4 complexes having different distances between the “twins”. The relative orientations also change with the different distances between the “twin” complexes. No regular and consistent configuration is found. Overall, only the four sugar residues that are consistently well-organized, that is, Ara_a, b, c, and d, were further studied from this point forward in the study. Three types of average distances between a pair of the ester oxygen atoms from these residues were calculated on the basis of the MDS trajectories after the complexes had reached equilibrium. The three distances are absolute distance, horizontal distance, and perpendicular distance and were calculated from the Cartesian coordinates of a pair of the ester oxygen atoms adjacent to one another vertically or horizontally, but not diagonally. The absolute distance stands for the distance between two ester oxygen atoms in three dimensions, while the other two distances are on the XY plane and along Z axis, respectively. The distances were computed as given in ref 49. One PMAconformer 4 complex contains four ester oxygen atoms from Ara_a, b, c, and d. Thus, for the “twin” complexes the total number of ester oxygen atoms is eight (see Figure 11). The perpendicular and horizontal distances were calculated because of the fact that the ester oxygen atoms are not perfectly coplanar. The results of these distance calculations are given in Table 3. The value of the average packing distance is estimated to be between the values of the average absolute and horizontal distances, that is, between 7.2 and

1060


Hong and Hopfinger

Figure 10. Chemical structure of the “twin” mycolyl-AG complexes before and after MDS.

Figure 11. Ester oxygen atoms from Ara_a, b, c, and d of the “twin” PMA-conformer 4 complexes used in the distance calculation. Each bar in the graph links one pair of the ester oxygen atoms together and represents one value of the distance. There is a total of 10 values for the distance for one “twin” PMA-conformer 4 complex. See text for details.

7.6 Å. Because the horizontal distance represents the packing distance in a more precise manner than does the absolute distance, the average packing distance is predicted as 7.3 Å, a value close to the average horizontal distance. The average surface area of packed R-mycolic acids is about 0.48 nm2, as measured by the LB technique, which yields an average packing distance of 7 Å.50 The surface area generated from a surface pressure/area isotherm using the LB technique is

for the hypothetical state of an uncompressed, close-packed layer that resembles a two-dimensional solid.51 The average packing distance calculated from this study is very close to the experimental value from the LB study suggesting that mycolic acids in the inner leaflet of the M. tuberculosis cell wall are, indeed, tightly packed and exist in a nonliquid state. The (5 × 5 × 1) PM monolayer after the MDS is shown in Figure 12. The monolayer is highly ordered. The energy



Table 3. Absolute, Horizontal, and Perpendicular Distances between a Pair of PMs after MDSa

d (Å)

average absolute distance (Å)

average horizontal distance (Å)

average perpendicular distance (Å)

6 7 8 9 average

7.6 7.8 6.9 8.2 7.6

7.1 7.3 6.7 7.4 7.2

1.9 2.0 1.3 2.8 2.0

a

See text for the definitions.

Figure 12. Side view of the (5 × 5 × 1) PM monolayer after the MDS.

Figure 13. Energy profile of the PM monolayer from the MDS trajectory. (1) E stretching, stretching energy; (2) E bending, bending energy; (3) E torsion, torsion energy; (4) E 1,4, the interaction energy between two atoms separated by three bonds; (5) E vdw, the van der Waals energy; (6) E electrostatic, electrostatic energy including hydrogen bond interactions; and (7) E potential, total potential energy.

profile from the MDS trajectory of the monolayer is given in Figure 13. The van der Waals interaction is a significant stabilizing force for the system as implied by the negative (stabilizing) values of the van der Waals energy. Hydrogen bond interactions are also important in stabilizing the system and are discussed in detail in the following. The stretching

Figure 14. Conformation of conformer 4. Only five residues, that is, Ara_1, a, b, c, and d, are shown for clarity. The only hydrogen atoms shown are those of the hydroxyl groups.

and bending energies appear to be the major destabilization contributions to the total potential energy of the system. However, these Valence geometry components of a MM force field have a somewhat arbitrary zero energy value and remain essentially constant over the MDS trajectory. Hydrogen bond interactions play a very important role in stabilizing the conformations of the arabinan chain. Figure 14 shows the hydrogen bonds formed in conformer 4. To keep the structures clear to view, only the Ara_1, a, b, c, and d residues are shown. Intra-residue hydrogen bonds are formed between the 3-OH and 5-OH for both Ara_b and Ara_d. (For the substituent coding, see Figure 2A.) These two residues belong to the majority of sugars that are in the R configuration. Thus, we would expect similar intra-residue hydrogen bond arrangements to exist for other R arabinose residues. Inter-residue hydrogen bonds are present between the 2-OH of Ara_1 and 2-OH of Ara_c and the 3-OH of Ara_c and the 5-OH of Ara_a. The distances between the atoms of the inter-residue hydrogen bonds are shorter than those of the intra-residue hydrogen bonds. Hence, the interresidue hydrogen bonds are likely stronger than the intra-

1062


Hong and Hopfinger

Figure 15. Side views of the (5 × 5 × 1) DPPE monolayers with different packing distances after the MDS. The same color coding as in Figure 8 is used here with the addition that the phosphorus atoms are colored orange and the nitrogen atoms blue.

residue hydrogen bonds. The composite shape of this 5-arabinose residue complex can be described as a four-pedal flower with a stem. Ara_a, b, c, and d are each a pedal and Ara_1 is the stem. Inter-residue hydrogen bonds help to hold the pedals together and in alignment to one another. 2. DPPE Monolayer Model. Figure 15 shows models of the DPPE monolayer for different packing distances after MDS. Apparently, a 6.5-Å packing distance is too small for the DPPE molecules because the whole monolayer is remarkably tilted (kanting has occurred), whereas an 8.0-Å packing distance is too large because the monolayer forms an irregular-shaped pseudo “micelle” assembly. When the packing distance is 7.0 Å, the head groups of the DPPE molecules that are in the center of the monolayer model significantly “bulge out”, and the slightly tilted monolayer forms a bow-like shape. These “bulging” and “bowing” distortions imply that a packing distance of 7.0 Å is still somewhat too small for the DPPE monolayer. When the packing distance increases to 7.5 Å, the monolayer appears to adopt a reasonable and regular shape. As mentioned in the Methods section, the packing distance measured for DPPE molecules using the LB technique is about 6.2 Å, a value that is too small for the DPPE monolayer on the basis

Figure 16. Energy profile of the DPPE monolayer, with an initial packing distance of 7.5 Å, from the MDS trajectory. The legend used in this plot is the same as in Figure 13.

of MDS. This result suggests that, unlike the PM monolayer, the monolayer of DPPE is not a compressed, close-packed layer. The energy profile of the DPPE monolayer with the average packing distance of 7.5 Å is given in Figure 16. The composite set of electrostatic interactions that includes hydrogen bonding plays a very important role in aligning the DPPE molecules and stabilizing the monolayer assembly.


Figure 17. Side view of the (5 × 5 × 1) DMPC monolayer after the MDS. The same color coding as in Figure 15 is used in this figure.

Figure 18. The energy profile of the DMPC monolayer from the MDS trajectory. The legend used in this plot is the same as in Figure 13.

3. DMPC Monolayer Model. The DMPC monolayer model after MDS is shown in Figure 17. Under the packing distance of 8.0 Å, the monolayer is much more organized than the DPPE monolayer under the same packing distance. The surface area of the packed DMPC molecules is approximately 47 Å2, larger than that of the DPPE molecules.46 The difference can be explained by their different head groups. The nitrogen of the head group of the DMPC molecule is fully methylated, while the amine residue of the DPPE head group is fully protonated. The energy profile of the DMPC monolayer is given in Figure 18, where it can be seen that the electrostatic interaction stabilizes the monolayer, however, in a less significant way than that for the DPPE monolayer. The reason might be that the extent of hydrogen bonding in the DMPC monolayer is much less, almost nonexistent, as compared to that in DPPE. Discussion The major findings regarding the conformations of the arabinan chain are that the 5-OH groups of Ara_a, b, c, and d, where the mycolic acids bind, are approximately coplanar and, secondly, tightly packed at an average separation distance of 7.3 Å. This intermolecular packing distance is shorter than that found among phospholipids of common biological membranes, such as DPPE and DMPC. Moreover, the packing distance of the mycolates found in the MDS is very close to the packing distance indirectly measured using the LB technique. However, the MDS packing distances are


greater than those found by the LB method for both DPPE and DMPC monolayers. Considering that the LB technique generally produces highly ordered films, with densely packed structures, we postulate that the mycolic acids are tightly packed and the inner leaflet of the M. tuberculosis cell wall exists in a relatively highly ordered state. This ordering is reflected by the high phase transition temperature, usually in the 60-70 °C range, of the bulk of the mycolates in the mycobacterial cell wall.15 Consequently, the criticism17,19,26 placed on the lipid bilayer model1 that states mycolic acids cannot form a tightly packed assembly because of their covalent binding with the polysaccharide AG complex is likely not valid. In fact, the structure of the arabinan chain is the source of how this seemingly improbable tight-packed arrangement can be achieved. The backbone of the arabinan chain is made up of arabinofuranose units joined as 1,5 linkages permitting great conformational flexibility. This conformational flexibility is demonstrated by the rather small energy differences among a range of distinct conformers generated from the conformational search done in this study. Thus, the arabinan “head group” has an exceptionally flexible structure that likely facilitates the lateral packing of mycolic acid chains. For the entire AG complex, conformational flexibility might be even greater than that of only the arabinan chains. The backbone of AG is composed of galactofuranose units connected by 1,6 linkages, with the 1,5-linked arabinofuranose units composing its side chains. A 1,6 linkage allows more conformational flexibility than a 1,5 linkage because an additional methylene group is part of the linkage. The flexible AG backbones allow, or at least give the opportunity for, the hydroxyl groups of arabinoses and galactoses to form extensive inter-residue and intra-residue hydrogen bonds. These hydrogen bond interactions, in turn, hold the arabinose residues, which bind with mycolic acids, close together. Hence, the AG complex, instead of playing a negative role in the tight packing of the inner leaflet, is, in fact, crucial for the formation and stabilization of a compact layer. The mycolate chains in the inner leaflet of the M. tuberculosis cell wall have been determined to be approximately perpendicular to the “plane” formed by the oxygen atoms of the 5-hydroxyl groups of Ara_a, b, c, and d. This finding supports one of the postulated features of the lipid bilayer model of Minnikin in which mycolates forming the inner leaflet are assumed to be perpendicular to the cell surface. Intermolecular hydrogen bonding is much stronger in the DPPE monolayer than in the PM monolayer, whereas intermolecular hydrogen bonding is negligible for the DMPC monolayer. This is not surprising if one considers that each DPPE molecule has two amine hydrogen atoms, one amine nitrogen atom, and eight oxygen atoms, all of which are adjacent to one another and available to form both interresidue and intra-residue hydrogen bonds. In contrast, a PM molecule has only one hydroxyl hydrogen atom and three oxygen atoms available for hydrogen bonding, and there are no hydrogen bond donor atoms in a DMPC molecule. van der Waals forces stabilize all three monolayers in the following order: PM > DPPE ≈ DMPC. The different

1064


values of the van der Waals packing energies are inversely related to the different packing distances found for the three monolayers. The conformational search and MDS methods applied in this study seemingly provide reasonable conformations and assemblies of the component molecules in the construction of the M. tuberculosis cell wall model. The general features, found by experiment, of the mycobacterial cell wall structure provide both a basis for, and the constraints of, the computational model. Therefore, the computational model developed here should also be applicable to other mycobacterial cell walls composed of mycolic acids with structures similar to those of the M. tuberculosis cell wall. However, the current computational model could benefit from (1) larger conformational sampling of the 17-arabinose residue segment; (2) more sugar residues included in the conformational search; (3) four, or even more, units of the PMA-conformer 4 complex included in the MDS; and (4) structural analysis of a monolayer composed of the mycolyl-AG structure rather than only mycolates. It is also very important to be able to identify the chemical configuration pattern of Ara_e, f, g, and h. X-ray diffraction studies21 done on the purified M. chelonae cell wall reveal that the tightly packed and largely parallel hydrocarbon chains in the cell wall exist in a quasi-crystalline array,52 a special polymorphic form consisting of multiple packing patterns that displays long-range orientational order but lacks periodic translational order. Considering that the tightpacking state in which mycolic acids in the M. tuberculosis cell wall exist resembles a crystalline phase, the results from the X-ray diffraction experiments would suggest that the mycolic acids likely exhibit more than one packing pattern. The difficulty encountered in this study in identifying a definitive chemical configuration pattern for Ara_e, f, g, and h might be an indication of the polymorphic packing nature of the mycolates. Also, additional experimental data regarding the mycobacterial cell wall structure is needed to refine the current cell wall model. In particular, knowledge of the structure of the AG peptidoglycan complex, a very important structural unit located between the inner leaflet of the cell wall and the cytoplasmic membrane, is needed for a full picture of the cell wall. The cell wall model constructed in this study permits visualization of the physical organization and molecular geometry of the M. tuberculosis permeability barrier and, thereby, provides a way to study the mechanism of low permeability of the cell wall and the intrinsic drug resistance of M. tuberculosis. A MDS study of a series of molecules, including drugs for the treatment of TB, was next performed to determine their diffusion and permeation properties through the inner leaflet model of the M. tuberculosis cell wall constructed in the work reported in this paper. The MDS transport study is reported in the following companion paper.53 Acknowledgment. Resources of the Laboratory of Molecular Modeling and Design were used in performing this study. We gratefully appreciate financial support from The

Hong and Hopfinger

Chem21 Group, Inc., and X.H. acknowledges a University Fellowship from UIC. We also thank Mr. Jianzhong Liu of our research group and Professor Scott G. Franzblau of the Institute for Tuberculosis Research at UIC for helpful comments and discussion. References and Notes (1) Minnikin, D. E. In The biology of the Mycobacteria; Ratledge, C., Standford, J. L., Eds.; Academic: London, 1982; Vol. 1, pp 95184. (2) Reichman, L. B.; Tanne, J. H. Timebomb: the global epidemic of multi-drug resistant tuberculosis; McGraw-Hill: New York: 2002. (3) WHO report 2003. http://www.who.int/gtb/publications/globrep/ index.html (accessed June 2003). (4) WHO Tuberculosis Fact Sheet. http://www.who.int/mediacentre/ factsheets/who104/en/index.html (accessed May 2003). (5) Jarlier, V.; Nikaido, H. Mycobacterial cell wall: structure and role in natural resistance to antibiotics. FEMS Microbiol. Lett. 1994, 123, 11-18. (6) Nikaido, H. Prevention of drug access to bacterial targets: permeability barriers and active efflux. Science 1994, 264, 382-388. (7) Minnikin, D. E.; Goodfellow, M. In Microbiological classification and identification; Goodfellow, M., Board, R. G., Eds.; Academic: London, 1980; pp 189-256. (8) Dobson, G.; Minnikin, D. E.; Minnikin, S. M.; Parlett, J. H.; Goodfellow, M. In Chemical methods in bacterial systematics; Goodfellow, M., Minnikin, D. E., Eds.; Academic: London, 1985; pp 237-265. (9) Brennan, P. J. In Microbial lipids; Ratledge, C., Wilkinson, S. G., Eds.; Academic: London, 1988; Vol. 1, pp 203-298. (10) Brennan, P. J. Structure of mycobacteria: recent developments in defining cell wall carbohydrates and proteins. ReV. Infect. Dis. 1989, 11 (Suppl.), 420-430. (11) Hunter, S. W.; Murphy, R. C.; Clay, K.; Goren, M. B.; Brennan, P. J. Trehalose-containing lipooligosaccharides. A new class of speciesspecific antigens from mycobacterium. J. Biol. Chem. 1983, 258, 10481-10487. (12) Hunter, S. W.; Gaylord, H.; Brennan, P. J. Structure and antigenicity of the phosphorylated lipopolysaccharide antigens from the leprosy and tubercle bacilli. J. Biol. Chem. 1986, 261, 12345-12351. (13) Brennan, P. J.; Nikaido, H. The envelope of mycobacteria. Annu. ReV. Biochem. 1995, 64, 29-63. (14) Barry, C. E., III; Lee, R. E.; Mdluli, K.; Sampson, A. E.; Schroeder, B. G.; Slayden, R. A.; Yuan, Y. Mycolic acids: structure, biosynthesis and physiological functions. Prog. Lipid Res. 1998, 37, 143-179. (15) Liu, J.; Barry, C. E., III; Besra, G. S.; Nikaido, H. Mycolic acid structure determines the fluidity of the mycobacterial cell wall. J. Biol. Chem. 1996, 271 (47), 29545-29551. (16) Yuan, Y.; Crane, D. C.; Musser, J. M.; Sreevatsan, S.; Barry, C. E., III. MMAS-1, the branch point between cis- and trans-cyclopropanecontaining oxygenated mycolates in Mycobacterium tuberculosis. J. Biol. Chem. 1997, 272 (15), 10041-10049. (17) Daffe, M.; Brennan, P. J.; McNeil, M. Predominant structural features of the cell wall arabinogalactan of Mycobacterium tuberculosis as revealed through characterization of oligoglycosyl alditol fragments by gas chromatography/mass spectrometry and by 1H and 13C NMR analysis. J. Biol. Chem. 1990, 265, 6734-6743. (18) McNeil, M.; Daffe, M.; Brennan, P. J. Evidence for the nature of the link between the arabinogalactan and peptidoglycan of mycobacterial cell walls. J. Biol. Chem. 1991, 265, 18200-18206. (19) McNeil, M.; Daffe, M.; Brennan, P. J. Location of the mycolyl ester substituents in the cell walls of mycobacteria. J. Biol. Chem. 1991, 266, 13217-13223. (20) Kartmann, B.; Stengler, S.; Niederweis, M. Porins in the cell wall of Mycobacterium tuberculosis. J. Bacteriol. 1999, 181 (20), 65436546. (21) Nikaido, H.; Kim, S.-H.; Rosenberg, E. Y. Physical organization of lipids in the cell wall of Mycobacterium chelonae. Mol. Microbiol. 1993, 8, 1025-1030. (22) Liu, J.; Rosenberg, E. Y.; Nikaido, H. Fluidity of the lipid domain of the cell wall from Mycobacterium chelonae. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 11254-11258. (23) Nikaido, H.; Jarlier, V. Permeability of the mycobacterial cell wall. Res. Microbiol. 1991, 142 (4), 437-443.

Mycobacterium tuberculosis Cell Walls (24) Jackson, M.; Raynaud, C.; Laneelle, M. A.; Guilhot, C.; LaurentWinter, C.; Ensergueix, D. Inactivation of the antigen 85C gene profoundly affects the mycolate content and alters the permeability of the Mycobacterium tuberculosis cell envelope. Mol. Microbiol. 1999, 31, 1573-1587. (25) Raynaud, C.; Laneelle, M. A.; Senaratne, R. H.; Draper, P.; Laneelle, G.; Daffe, M. Mechanisms of pyrazinamide resistance in mycobacteria: importance of lack of uptake in addition to lack of pyrazinamidase activity. Microbiology 1999, 145, 1359-1367. (26) McNeil, M.; Brennan, P. J. Structure, function and biogenesis of the cell envelope of mycobacteria in relation to bacterial physiology, pathogenesis and drug resistance; some thoughts and possibilities arising from recent structural information. Res. Microbiol. 1991, 142, 451-463. (27) Hunt, N. G.; Gregoret, L. M.; Cohen, F. E. The origins of protein secondary structure. Effect of packing density and hydrogen bonding studied by a fast conformation search. J. Mol. Biol. 1994, 241 (2), 214-225. (28) Kopple, K. D.; Bean, J. W.; Bhandary, K. K.; Briand, J.; D’Ambrosio, C. A.; Peishoff, C. E. Conformational mobility in cyclic oligopeptides. Biopolymers 1993, 33 (7), 1093-1099. (29) Bassolino-Klimas, D.; Bruccoleri, R. E. Application of a directed conformational search for generating 3-D coordinates for protein structures from R-carbon coordinates. Proteins: Struct., Funct., Genet. 1992, 14 (4), 465-474. (30) Potenzone, R.; Hopfinger, A. J. Conformational analysis of glycosaminoglycans. III. Conformational properties of hyaluronic acid and sodium hyaluronate. Polymer J. 1978, 10 (2), 181-199. (31) Kulkarni, A.; Hopfinger, A. J. Membrane-interaction QSAR analysis: application to the estimation of eye irritation by organic compounds. Pharm. Res. 1999, 16 (8), 1245-1253. (32) Kulkarni, A.; Han, Y.; Hopfinger, A. J. Predicting caco-2 cell permeation coefficients of organic molecules using membraneinteraction QSAR analysis. J. Chem. Inf. Comput. Sci. 2002, 42, 331342. (33) Iyer, M.; Mishru, R.; Han, Y.; Hopfinger, A. J. Predicting bloodbrain barrier partitioning of organic molecules using membraneinteraction QSAR analysis. Pharm. Res. 2002, 19 (11), 1611-1621. (34) HyperChem Program Release 6.01 for Windows; Hypercube, Inc.: Gainesville, FL, 2000. (35) Polak, E. Computational Methods in Optimization; Academic Press: New York, 1971. (36) Cerius2, version 4.0; Molecular Simulations, Inc.: San Diego, CA, 1999. (37) The torsion angle of four atoms A, B, C, and D is defined as the angle between the two planes, one containing atoms A, B, and C and the other containing atoms B, C, and D. Thus, if the torsion angle is 90°, it means that the two planes are perpendicular to each other, whereas the torsion angle of a value close to 0, 180, or -180° means the four atoms are approximately coplanar. (38) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. AM1: a new general purpose quantum mechanical model. J. Am. Chem. Soc. 1985, 107, 3902-3909.

Biomacromolecules, Vol. 5, No. 3, 2004 1065 (39) Doherty, D. C. MOLSIM User’s Guide, The Chem21 Group, Inc.: Lake Forest, IL, 1997. (40) Allinger, N. L. Conformational analysis. 130. MM2. A hydrocarbon force field utilizing V1 and V2 torsional terms. J. Am. Chem. Soc. 1977, 99, 8127-8134. (41) Hopfinger, A. J.; Pearlstein, R. A. Molecular mechanics force-field parametrization precedures. J. Comput. Chem. 1984, 5, 486-492. (42) Berendsen, H. J. C.; Postman, J. P. M.; Gunsteren, W. F. V.; Nola, A. D.; Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 1984, 81, 3684-3690. (43) MI-QSAR User’s Manual, version 2.0; The Chem21 Group, Inc.: Lake Forest, IL, 1997. (44) Alexandre, S.; De´rue, V.; Garah, S.; Monnier, C.; Norris, V.; Valleton, J. Submolecular structures in dipalmytoylphosphatidylethanolamine Langmuir-Blodgett films observed by scanning force microscopy. J. Colloid Interface Sci. 2000, 227, 585-587. (45) Hitchcock, P. B.; Mason, R.; Thomas, K. M.; Shipley, G. G. Structural chemistry of 1,2-dilauroyl-DL-phosphatidylethanolamine: molecular conformation and intermolecular packing of phospholipids. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 3036. (46) Phillips, M. C.; Chapman, D. Monolayer characteristics of saturated 1,2,-diacylphosphatidylcholines (lecithins) and phosphatidylethanolamines at the air-water interface. Biochim. Biophys. Acta 1968, 163, 301-313. (47) Stouch, T. R. Lipid membrane structure and dynamics studies by all atom molecular dynamics simulations of hydrated phospholipid bilayer. Mol. Simul. 1993, 1, 335-362. (48) The bending angle used in this study is the angle formed by three 5-OH oxygen atoms of three arabinose residues. (49) [1] The absolute distance was calculated as the mean of [(Xi - Xi+1)2 + (Yi - Yi+1)2 + (Zi - Zi+1)2]1/2, where Xi, Yi, and Zi and Xi+1, Yi+1, and Zi+1 represent Cartesian coordinates of a pair of the ester oxygen atoms adjacent to each other vertically or horizontally but not diagonally. [2] The horizontal distance, calculated as the mean of [(Xi - Xi+1)2 + (Yi - Yi+1)2]1/2. [3] The perpendicular distance, calculated as the mean of [(the absolute distance)2 - (the horizontal distance)2]1/2, that is, the mean of [(Zi - Zi+1)2]1/2. The index “i” is between 1 and 9. See text and Figure 11 for details. (50) Hasegawa, T.; Nishijo, J.; Watanabe, M.; Funayama, K.; Imae, T. Conformational characterization of R-mycolic acid in a monolayer film by the Langmuir-Blodgett technique and atomic force microscopy. Langmuir 2000, 16, 7325-7330. (51) Hann, R. A. In Langmuir-Blodgett Films; Roberts, G., Ed.; Plenum Press: New York, 1990, p 21. (52) Shechtman, D.; Blech, I.; Gratias, D.; Cahn, J. W. Metallic phase with long-range orientational order and no translational symmetry. Phys. ReV. Lett. 1984, 53, 1951-1953. (53) Hong, X.; Hopfinger, A. J. Molecular Modeling and Simulation of Mycobacterium tuberculosis Cell Wall Permeability. Biomacromolecules, 2004, 5, 1066-1077.

BM034514C

Construction, Molecular Modeling, and Simulation of Mycobacterium

Recommend Documents