Interaction of Glycolipids with the Macrophage Surface Receptor Mincle – a Systematic Molecular Dynamics Study

Synthetic analogues of mycobacterial trehalose-dimycolate such as trehalose acyl esters have been proposed as novel adjuvants for vaccination. They induce an immune response by binding to the macrophage C-type lectin receptor Mincle. The binding site of trehalose is known, but there is yet only very limited structural information about the binding mode of the acyl esters. Here, we performed a systematic molecular dynamics study of trehalose mono-and diesters with different chain lengths. All acyl chains investigated exhibited a high flexibility and interacted almost exclusively with a hydrophobic groove on Mincle. Despite the limited length of this hydrophobic groove, the distal parts of the longer monoesters can still form additional interactions with this surface region due to their conformational flexibility. In diesters, a certain length of the second acyl chain is required to contact the hydrophobic groove. However, a stable concomitant accommodation of both acyl chains in the groove is hampered by the conformational rigidity of Mincle. Instead, multiple dynamic interaction modes are observed, in which the second acyl chain contributes to binding. This detailed structural information is considered helpful for the future design of more affine ligands that may foster the development of novel adjuvants.


Supplementary Methods -Atom Types and Charges for Trehalose Acyl Esters
The trehalose acyl monoesters and diesters in our MD simulations were assigned atom types and force field parameters of the Glycam06j-1 force field [1]. This force field, which is specifically designed for a description of carbohydrates, has already been applied in previous studies for the simulation of fatty acids [2] and for systems, where glycosids and acyl chains are directly linked together [3,4].
The Glycam06 force field provides predefined residues for some α-D-Glucopyranose isomers with free valences at different positions allowing thereby connectivity to other residues in a modular concept. Examples for already available residues are as shown in Supplementary Fig. S2: Fig. S2. Predefined Glucose residues in the Glycam06 force field and their respective total charges. Free valences are coloured in red.
In order to construct trehalose, they can be combined as (a) 1GA + 0GA, if no acyl chains are present. The partial charges of both residues add up to neutral.
(b) 1GA + 6GA, if a connection to an acyl chain at the o6 atom is desired. This path was chosen for the trehalose-6-monoesters C4, C8, C12 and C18. The partial charges of both residues add up to −0.194 e so that the acyl chain needs a total charge of +0.194 e for electrical neutrality.
However, the residues predefined in Glycam are not sufficient for the construction of trehalose-6,6 -diesters, because no glucose residue exists that has an oxygen, but no hydrogen atom and therefore free valences at both c1 and c6 position. Therefore, we constructed a new residue CGA ( Supplementary Fig. S3), based on the template 1GA using the Amber14 [5] version of the xleap program.
The Glycam06j-1 residues were loaded, and 1GA was opened for editing: source leaprc.GLYCAM_06j-1 edit 1GA In the graphical user interface, the hydrogen atom connected to o6 (leap atom name H6O) was deleted. Then, the modified residue was written to an Amber prep file with the command saveamberprep 1GA CGA.prep In the new prep file, we changed the residue label from 1GA to CGA. In analogy to the conventions of the Glycam06 force field, the atom type of the former hydroxy oxygen atom o6 was changed from Oh to Os. The predefined Glycam residues comprise in total eight different isomers of α-D-Glucopyranose, which have a free valence at the o6 oxygen atom, and another eight, which have a complete hydroxy group at this position. Although these residues differ in the overall number of free valences, a detailed comparison shows that all residues without the hydrogen atom have the same partial charge of −0.458 e at the o6 atom, whereas those with hydrogen atom have a charge of −0.682 e. The charge of the adjacent carbon atoms is not influenced by the existence of the hydrogen. Hence, we changed the charge of the o6 atom to −0.458 e. In consequence, the total charge of the residue had to be modified to −0.388 e.
The new residue CGA was used in the simulations of our trehalose diacylesters 2×C4 and 2×C18 in the combination CGA + 6GA leading to an overall carbohydrate charge of −0.388 e. By adding two fatty acids with a charge of each +0.194 e, electrically neutral diesters could be created.
For the setup of our simulations, we used the trehalose atom coordinates, that were present in the PDB file 4ZRV. Therefore, the residue identifiers of these atoms hat to be corrected to the respective Glycam residue 1GA, 6GA or CGA. Moreover, the atom names were changed to fit the atom names used in the residue templates. Then, we created our topology and the initial Amber coordinate file using tleap. The prep file was loaded using the command loadamberprep. One last necessary step was to define the glycosidic bond between the two residues with the two sugar residues and the ester bonds between sugar and acyl chains by means of the bond command.

Generation of 3D structures
We created 3D structures and Amber prep files for acyl chains with n ∈ {2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22} carbon atoms, which we call F02, F04, . . . , F22. Although not all these chain length were necessary for the present MD simulations, we generated a complete set covering a large variety of chain lengths in order to facilitate future studies. In order to make them easily combinable with the glucose residues of the trehalose, our acyl chain residues begin at the carbonyl carbon atom c1 and do not contain the acidic hydroxy group. The ester oxygen atom is in our setup still part of the sugar moiety ( Supplementary  Fig. S4). However, to allow for a RESP/ESP fit of charges, we created also PDB files of complete alkyl acids including the acidic hydroxy group. All initial PDB files were generated with Avogadro 1.1 [6] by drawing the necessary atoms and adjusting inter-atomic distances, angles and torsions. Supplementary Fig. S4. Subdivision of sugar and acyl chain residues in our setup.

Calculation of partial charges
In order to set up partial charges for the acyl chains, the RESP ESP Charge Derive Server 3.0 [7] was used. Quantum mechanical calculations were performed with Gaussian09 [8]. In analogy to the derivation of charges for the predefined Glycam residues, the protocol RESP-C2 (HF//6-31G*//HF/6-31G*) was selected.
The charge fit was performed for every acyl chain residue with the respective fatty acid methyl ester. However, the charges for the carbonyl carbon atom, the carbonyl oxygen atom and the methyl moiety were calculated only once for methyl acetate and then set as a constraint for all further systems.
The methyl acetate molecule (structural formula with atom names in Supplementary Fig. S5) was defined to be electrically neutral. Then, a charge constraint of q = 0 was posed upon the methyl group connected to the carbonyl carbon atom, and a charge constraint of q = −0.194 e was set for the entity of ester oxygen atom and the methyl group connected to the ester oxygen. This approach lead to the implicit assignment of the charge q = +0.194 e for the carbonyl group (carbonyl carbon and oxygen atom). After an energy optimization of the initial structure, the RESP charge fit lead to the results summarized in Supplementary Table S1. In a second step, a RESP fit was calculated for the different acyl chain lengths (F02, F04, . . . , F22). Therefore, the structure of the respective fatty acid methyl ester was submitted, and defined to be in total electrically neutral. The partial charges as listed in Supplementary Table S1 were chosen as a constraint. Furthermore, all aliphatic hydrogen atom charges of the alkyl chain were set to zero according to the Glycam force field conventions.

Supplementary
Thereby, it was possible to assign a charge of q = +0.194 e to our acyl chain residues (carbonyl group + c2 to cn) as required for electrically neutral trehalose acyl esters. Simultaneously, the alkyl chain (c2 to cn) as an entity remained uncharged, but not its individual carbon atoms. The Pearson correlation coefficients determined for the fit of the different fatty acid methyl esters are summarized in Supplementary Table S2.

Generation of prep files and assignment of atom types
The Amber tool antechamber was used to convert the PDB files into the prep file format. Then, all prep files were combined into one single file. In order to assign the correct atom types and the partial charges, this file was edited manually. The atom types were assigned following the description in the Glycam parameter file $AMBERHOME/dat/leap/parm/GLYCAM_06j.dat and in analogy to the residue myr (myristic acid), which is predefined in the Glycam force field (Supplementary Table S3).  The residues are colored according to their average number of contacts with the acyl chain (gray) during the MD trajectory: red > orange > yellow > green > blue. (b) Sequence alignment for the carbohydrate recognition domains of bovine Mincle (sequence from PDB 4ZRV) and human Mincle (sequence from Uniprot). The alignment was generated using Clustal Omega with default settings. The residues are shaded according to the color code from panel (a).