Choosing optimal strain for photoswitchable antibiotic
As can be read in the photoswitchable antibiotics section, the idea with the photoswitchable antibiotic was that it would kill the decoy B. subtilis but would not cause any harm to the engineered strain containing the DNA sequence.
When not treated properly with the right wavelength of UV radiation (see figure 1), spirofloxacin remains in its inactive state. Because the sample sent to the receiver contains hundreds to thousands of times more decoy spores, it cannot be sequenced directly by the recipient (see also Decoding Fidelity). Instead, the receiver must activate the antibiotic. Once done, the spirofloxacin-resistant message-containing bacteria would outnumber the decoys.
We explored the activity of spirofloxacin on different strains of E. coli, since it was a species of bacteria whose susceptibility to spirofloxacin had been previously well measured [7] (see figure 2).
At this point of the lab work, experiments trying to measure the MIC of spirofloxacin “on”/”off” against B. subtilis had been inconsistent. We thought of Molecular Dynamics (MD) studies as a good alternative to measure how well the antibiotic performs on Bacillus, thus allowing us to continue faster with the engineering of the resistant strain.
It is important for our system that the ratio of resistance to susceptibility in engineered and wild type is optimal. That means that in its inactive state, spirofloxacin must not have a high bactericidal activity and when activated, it must be potent enough to kill the wild-type cells while the engineered strain will survive.
First, we needed to find a suitable crystallographic structure of a type-II topoisomerase (a protein-DNA complex) bound to a fluoroquinolone. Topoisomerase IV is the main fluoroquinolone’s target in Gram-positive bacteria, while gyrase (the other type of topoisomerase II) is the main target in Gram-negatives [1]. The crystallographic structure recently reported by Veselkov et. al. (2016) offered a good alternative as it has the two fluoroquinolone molecules bound to the protein-DNA complex.
Secondly, we needed to create and adapt an appropriate force-field that would reproduce the behavior of the binding process at a reasonable computational cost. A force-field is a set of parameters that tells the software (we used GROMACS v5.0.4) how each atom behaves and interacts with others. As reported elsewhere [1][2] the binding process involves cation- and hydrogen-bonding interactions that can only be reproduced in the atomistic level. However, the computational cost of simulating a 150 kDa protein (embedded in a solvated box) is high enough to consider using lower-resolution scales. In addition, the crystallographic structure already is a bound topoisomerase – antibiotic complex, so there is no need for that high level of resolution. Coarse-grained models are refined enough to offer insights into the affinity of protein-ligand interactions (see i.e. [4]). Some of the molecular parameters for the spiropyran part of the molecule (the part that gives it is photoswitchable behavior) had been studied elsewhere [9]; nevertheless, its parameterization into atomistic and further coarse-grained requires some extra work.
While the MARTINI force-field has been adapted and optimized for both proteins and DNA [3][6] it cannot reproduce cation- and hydrogen-mediated binding of ligands. So our best method is to use umbrella sampling in which basically the protein and the ligand are artificially placed in its “correct” orientation and dragged away from each other, measuring the Potential of Mean Force (PMF) [5].
Before actually starting the umbrella simulations and due to the lack of consistent experimental results, we decided to change the photoswitchable antibiotic approach in our project. However, for the sake of clarity in the next figure we show what we expected to obtain from the simulations.
References
- [1] Aldred, K., Kerns, R. and Osheroff, N. (2014). Mechanism of Quinolone Action and Resitance. Biochemistry, 53, 1565-1574
- [2] Lupala, C., Gomez-Gutierrez, P. and Perez, J. (2013). Molecular Determinants of the Bacterial Resistance to Fluoroquinolones: A Computational Study. Current Computer-Aided Drug Design, 9: 281-288
- [3] Monticelli, L., Kandasamy, S., Periole, X., Larson, R., Tieleman, P., and Marrink, S. (2008). The MARTINI Coarse-Grained Force Field: Extension to Proteins. J. Chem. Theory and Comput., 4(5): 819-834
- [4] Naughton, F., Kalli, A. and Sansom, M. (2016). Association of Peripheral Membrane Proteins with Membranes: Free Energy of Bingind of GRP1 PH Domain with Phosphatidylinositol Phosphate-Containing Model Membranes. J. Phys. Chem. Lett., 7(7): 1219-1224
- [5] Torrie, G. and Valleau, J. (1977). Nonphysical sampling distributions in Monte Carlo free-energy estimation: Umbrella sampling. J. Comput. Phys., 23(2): 187-199
- [6] Uusitalo, J., Ingólfsson, H., Akhshi, P., Tieleman, P., Marrink, S. (2015). Martini Coarse-Grained Force Field: Extension to DNA. J. Chem. Theory Comput., 11(8): 3932-3945
- [7] Velema, W., Hansen, M., Lerch, M., Driessen, A., Szymanski, W. and Feringa, B. (2015). Ciprofloxacin-Photoswitch Conjugates: A Facile Strategy for Photopharmacology. Bioconjugate Chemistry, 26: 2592-2597
- [8] Veselkov, D., Lapanogov, I., Pan, X., Selvarajah, J., Skamrova, G., Branstrom, A., Prasad, L., Fisher, M. and Sanderson, M. (2016). Structure of a quinolone-stabilized cleavage complex of topoisomerase IV from Klebsiella pneumoniae and comparison with a related Streptococcus pneumoniae complex. Acta Cryst. D72: 488-496
- [9] Zhai, GH., Yang, P., Wu, SM., Lei, YB. and Dou, YS. (2014). A semiclassical molecular dynamics of the photochromic ring-opening reaction of spiropyran. Chinese Chemical Letters, 25: 727-731
