Prodigiosin

Molecular dynamics of the membrane interaction and localisation of prodigiosin

Aarti Ravindran1, Sharmila Anishetty1 and Gautam Pennathur1,2

Abstract

The tripyrrolic antibiotic prodigiosin causes diverse reactions on its targets like energy spilling, membrane leakage, loss of motility and phototoxicity. It has bacteriostatic, bactericidal, anti-fungal, anti-cancer and immunosuppressive properties. Most of the functions suggest the role of prodigiosin in membrane disruption but the exact mechanism remains unknown. A molecular dynamics study was performed to understand the interactions of prodigiosin with the membrane. It was seen that prodigiosin from the solvent enters the membrane immediately either individually or as small clusters. Prodigiosin clusters with more than eight molecules do not appear to enter the membrane. Upon entry, the molecules orient themselves along the membrane-water interface with the pyrrole rings interacting with lipid head groups and with water. This orientation is stabilised by hydrogen bonding and hydrophobic interactions. The presence of prodigiosin molecules in the membrane changes the local lipid architecture and reduces the solvent accessibility of the membrane. The membrane fluidity, thickness or area per lipid head are largely unaffected. This suggests that prodigiosin could cause most damage in the vicinity of a membrane protein and thus could also explain the reason for varied effects on the targets.

Keywords: Interaction, Localisation, Membrane, Molecular Dynamic Simulations, Prodigiosin

1.Introduction

Prodigiosins are a group of tripyrrolic red pigments [1] produced by many strains of Serratia marcescens and certain other bacteria like Hahella chejuensis KCTC 2396 [2], Vibrio gazogenes [3], Pseudoalteromonas rubra [4], Janthinobacterium lividum [5], Actinomadura madurae [6] and Streptomyces coelicolor [7]. Prodigiosin is structurally related to prodigiosin R1 and undecylprodigiosin and these molecules with the characteristic tripyrrolic structure are referred to as prodiginines [8]. Prodigiosin is produced by the condensation of MAP (Methyl Amyl Pyrrole) and MBC (Methoxy Bipyrrole Carbaldehyde) following a multistep pathway which involves fourteen proteins which are produced from genes arranged in a single operon in Serratia species [1]. Prodigiosin is produced in the stationary phase under low glucose conditions by Serratia marcescens and is under the control of the quorum sensing system and other nutrient sensors [9]. The presence of glucose, ATP or phosphate in the medium inhibits the production of prodigiosin. The production is also regulated by the temperature of the environment with the optimum at 25 ti to 30 ti . Prodigiosin is not produced at 37 ti . A small amount of the pigment is secreted into the medium and the bulk remains cell-associated in Serratia species.
Prodigiosin (5-[3-Methoxy-5-(2H-pyrrol-2-ylidene)-1,5-dihydro-2H-pyrrol-2-ylidene]

methyl-2-methyl-3-pentyl-1H-pyrrole) has three pyrrole rings and an alkyl chain whose length determines the type of prodigiosin. The carbon source used for growth of the bacteria appears to influence the type and length of the carbon chain. The molecule has an absorption maxima at 535 nm in ethanol [10] and a log P value of 4.7. It shows poor water solubility but is soluble in methanol, ethanol, DMSO, chloroform and acetonitrile.
The pigment is offers a competitive advantage to the bacterium in the stationary phase. Several molecular mechanisms resulting from the antibiotic effects of prodigiosin have been reported, like membrane potential alteration via anion symport [11], membrane damage [12–14], phototoxicity [15], energy spilling [16] and formation of reactive oxygen species (ROS) [16]. Prodigiosin is localized predominantly in the membrane in Serratia and in the membrane and

nuclear fractions in eukaryotic targets [17]. At sub-inhibitory concentrations, it inhibits the motility of susceptible bacteria [18]. Prodigiosin also has several interesting properties such as immunosuppressive activity (preferentially suppressing polyclonal T cells), anticancer activities toward some types of human cancer cells (inhibiting cell proliferation), as well as bacteriostatic or bactericidal effects against certain bacteria, anti-fungal activity and anti-protozoan activity [19]. It also shows promise as a natural dye. Synthetic prodigiosins like Obatoclax have similar structural and chemical properties.
Though its chemical structure and the mechanism of biosynthesis have been well studied, the mechanism of action of the molecule is poorly understood. Most of the effects of prodigiosin suggest a membrane associated role for the molecule. The mechanism of the interaction of prodigiosin with the membrane is poorly understood. This prompted us to investigate the mechanism by which prodigiosin permeates and/or perturbs the membrane through molecular dynamics simulations.

2.Methods

2.1.Simulation setup

Molecular dynamics simulations were carried out to understand the localisations and the interactions of prodigiosin and a dipalmitoylphosphatidylcholine (DPPC) bilayer using GROMACS 5.1.2 [20,21] and the GROMOS54a7 force field [22]. The force field parameters were modified to include lipid parameters. The lipid parameters used (Berger lipids) were as described in the literature [23]. The DPPC layer was made up of 128 DPPC molecules (64 in each layer) with the necessary water molecules in each simulation. The box size and hence the number of solvent molecules was different for each simulation as the number of prodigiosin molecules in the solvent region was varied. In the force field, Ryckaert-Bellemans dihedral potential was used, and a scaling factor of 0.125 was applied to Lennard-Jones 1-4 interactions. Particle Mesh Ewald (PME) method was used for Coulombic interactions. The structure of prodigiosin (C20H25N3O) was downloaded

from NCBI PubChem (ID: 5351169) (National Center for Biotechnology Information. PubChem Compound Database; CID=5351169, https://pubchem.ncbi.nlm.nih.gov/compound/5351169) and the United atom topology was generated using the ATB server [24,25]. The molecule is called 74EL in certain figures and in the raw data files (which are not included in this manuscript). The SPC model was used to represent water. Periodic boundaries were applied in all directions.
The membrane with prodigiosin was solvated and energy minimisation was performed with the steepest descent minimisation and 1.2 nm as the short-range Van der Waals cut-off. The energy minimised structure was equilibrated under NVT (particle number, cell volume and temperature) for 100 ps and NPT (isothermal-isobaric) with position restraints for 1 ns. The LINCS algorithm was used for the constraints and the Nose-Hoover thermostat for the temperature coupling. The system reached a temperature of 323.15 K and an average density of 1014.23 kg/m3 with an RMSD of 3.53 Å. The pressure progression was also checked. The energy was analysed to check if the box vectors (X and Y) had stabilised. The box vectors and the density profile of the membrane components is shown in the Supplementary file 1 (Figure 1-4). The position restraints were then removed and MD (Molecular Dynamics) simulations were carried out. The simulation was performed in triplicates.
First, a single molecule of prodigiosin was placed at the centre of the membrane and simulations were carried out until 100 ns. In the next simulation, a single molecule of prodigiosin was placed at the surface of the solvent layer and the simulations were carried out until the molecule entered the membrane. There were three groups of molecules: prodigiosin, DPPC and water. The Centre of Mass (COM) motion of the molecules was removed separately as compared to the solvent and DPPC. Simulations were also carried out with multiple molecules of prodigiosin placed randomly in the solvent region.
2.2.Analyses

The RMSD (Root Mean Square Deviation), Density profiles, RDF (Radial Distribution Function), solvent accessibility and MSD (Mean Square Deviation) studies were found out using

GROMACS utility tools. Hydrogen bonding and visualisation of the simulation and rendering of time snap images were performed with VMD (Visual Molecular Dynamics) [26]. The hydrogen bond length was kept at 0.35 nm and the bond angle cut off was taken as 30°. The graphs were rendered using Grace (http://plasma-gate.weizmann.ac.il/Grace/). The area per lipid head and thickness of the membrane were calculated using FATSLiM [27].

3.Results and Discussion

3.1.Localisation of prodigiosin

Prodigiosin targets the membrane of bacteria, fungi, protozoa, plants and animals as mentioned in the introduction. Biological membranes usually have a mixture of various lipids. Since prodigiosin acts on both bacterial and eukaryotic cells, it was suspected to act on the phosphoryl groups or the glycerol esters in the lipid membrane. DPPC was chosen, in this study, to represent a simple biomembrane. It has been used to analyse the interaction of antibacterial peptides with a simulated biomembrane [28–30] and has also been used to mimic the plasma membrane of Serratia marcescens NIMA [31] in experimental studies. It represents microbial membranes [32]
and mammalian membranes [33,34] in other studies.

In the first simulation, a single prodigiosin molecule was placed on the surface of the water. The molecule entered the membrane within 2 ns of the start of the simulation with the alkyl chain entering first, followed by the pyrrole rings. This permeation of prodigiosin into the membrane, immediately after the start of the simulation is justified as prodigiosin is sparingly soluble in water.
In a separate simulation, a single molecule of prodigiosin was placed at the centre of the membrane in its hydrophobic core and simulation was carried out for 100 ns. Prodigiosin moved to the interface of the membrane and the solvent layer within 10 ns. After this period there was lateral movement of the molecule diffusing along the membrane. The molecule remained within the membrane till the end of the simulation at the membrane – water interface (data not shown).

3.2.Orientation of the molecule

The prodigiosin molecule orients itself with the pyrrole rings aligned towards the solvent- membrane interface and the alkyl chain aligned with the acyl groups of the DPPC. This could be due to electrostatic interaction between the hydrogens attached to the pyrrole rings of prodigiosin or the oxygen atom in the prodigiosin molecule with the lipid head groups or the water molecules outside the bilayer.
To confirm this, the radial distribution function was calculated for various groups on the prodigiosin molecule with water (Figure 1A) and with polar groups on the DPPC (Figure 1B). The RDF was calculated using the data generated during the simulation of a single molecule of prodigiosin placed at the centre of the membrane over the course of 100 ns. The molecule remained in the membrane the entire time and hence this simulation data was used to determine the RDF values.
The peak at 0.25 in Figure 1A indicates hydrogen bonding of the hydrogen attached to the N atom in the pyrrole ring to the oxygen in the water molecule. The peak at 0.25 in Figure 1B indicates hydrogen bonding of the hydrogen of attached to the pyrrole ring of the prodigiosin to the oxygen atom in the DPPC molecule. This indicates that the interaction of the prodigiosin is with both water and the lipid head groups in the membrane. Some of the water interactions were below the bulk density but the data obtained was consistent with those obtained by Chen et at, 2016 [35]. The numbering scheme of the atoms of the prodigiosin molecule and the DPPC molecule are shown in the Supplementary data file 2 (Figure S1 and S2 respectively).
Hydrophobic interactions of the prodigiosin alkyl chain with the acyl chains of the DPPC are also possible. The orientation of the molecule during entry into the membrane suggests that hydrophobic interactions also play a role in the membrane orientation of the molecule. The minimum distance between any two given residues can be used to determine the existence of hydrophobic interactions. Data represented in the Supplementary file 2 (Figures S3, S4 and S5) shows that hydrophobic interactions play a role in molecule stability. As observed in the figures, the

minimum distance between H12 of prodigiosin (a residue which is involved in forming hydrogen bonds) and O9 and P8 of DPPC (residues involved in electrostatic interactions) is 0.25 nm in certain time frames of the simulation (X axis of the figures). The minimum distance between C1 of prodigiosin (the terminal alkyl carbon) and C21 of DPPC (the carbon atom in the acyl chain most likely to lie in the vicinity of C1 of prodigiosin due to its length) shows less variation throughout the simulation than the data obtained from the bonds of H12 with O9 and P8 mentioned above. This suggests that hydrophobic interactions between the alkyl chain of prodigiosin and the acyl groups of DPPC contribute significantly to the localisation of the prodigiosin molecule in the membrane.

3.3.Effect of increasing the concentration

Since prodigiosin is not soluble in water, it can be assumed that when multiple molecules are placed in the solvent their behaviour could be different as compared to when a single molecule is placed in the solvent. To corroborate this, two separate simulations were carried out with eight and sixteen molecules respectively of prodigiosin inserted using the gmx insert-molecules command in the solvent region. Each simulation was performed for at least 100 ns and extended as required until all the molecules entered the membrane.
When eight prodigiosin molecules were placed in the solvent, the molecules started clumping within 2 ns and forms small clusters. This continued until they formed a single cluster (Figure 2) and the cluster diffused randomly in the solvent until it came in contact with the membrane (at around 25 ns). Then, it entered as a cluster. The cluster entered with the alkyl chain of one or two prodigiosin molecules first interacting with and then entering the membrane. This was followed by the entry of the rest of the molecules, each entering one by one or in groups by orienting alkyl groups with the membrane. They then stayed at the interface between the membrane and water and the cluster was eventually dispersed across the membrane. The time of entry of the cluster is possibly entirely random, as increasing the box size and hence the number of solvent molecules would mean that the cluster could take longer to come into contact with the membrane.

But as soon as it comes into contact with the membrane, it enters and then does not leave the membrane.
When sixteen molecules are placed in the solvent, they form a cluster but the cluster does not split or enter the membrane during the 100 ns of simulation. This could be because the cluster is so large that it cannot overcome the force required to permeate the membrane or that the solvent region is so large that it does not come into contact with the membrane as explained above. This could also explain why prodigiosin production in Serratia marcescens is accompanied with a secretion of serrawettin, a surfactant [36]. The presence of the surfactant could break up the clusters of prodigiosin leading to better membrane permeation.

3.4.Dynamics of the cluster of prodigiosin molecules

The prodigiosin cluster formed by placing eight molecules of prodigiosin in the solvent was analysed. The minimum COM distance of prodigiosin from the COM of DPPC was calculated and is shown in the figure (Supplementary data file 2: Figure S6).
The mass density profile of the prodigiosin molecules over the course of the simulation (Figure 3) indicates that the molecules stabilise at the water-membrane interface. The movements are mostly lateral (along the membrane rather than across) and the molecules continue to be at the interface of the membrane and the water. They do not leave the membrane or even move to the other leaflet of the bilayer. They continue to remain at the same side of the bilayer where they entered the membrane. This could mean that individual molecules show more affinity for the lipid environment of the membrane than for water. Hence it is a reflection of their high log P value (4.7) leading to better lipid solubility. The fact that they disperse across the membrane could mean that they could now affect a larger area of the membrane.
Selected time frame images of the duplicate and the triplicate run of the simulation with eight molecules of prodigiosin or four molecules placed in the solvent are shown in Supplementary data file 2: Figure S7 and S8 respectively. In the second run of the simulation, a single molecule of

prodigiosin enters the membrane at 50 ns separating from the rest of the cluster. The other molecules entered the membrane between 75 ns and 100 ns. By the end of the simulation run (200 ns) the molecules dispersed across the membrane. The time frames shown in Supplementary data file 2: Figure S7 shows the prodigiosin molecules starting to come together at 1 ns. In the third run of the simulation only four molecules of prodigiosin were placed in the solvent. The smaller size of the cluster formed resulted in delayed entry of the molecules into the membrane. In this run of the simulation, the molecules of prodigiosin entered the membrane as individual units rather than as clusters.
One important fact remains that the permeation of prodigiosin into the membrane, the formation of a cluster and the entry of the molecule either individually or as a small cluster occur at random. If the simulation is repeated the events do occur but most likely at different time points.

3.5.Interactions of prodigiosin with the membrane

An analysis of the Radial Distribution Function (RDF) values indicated that hydrogen bonding could be responsible for the orientation of prodigiosin along the membrane-water interface. Hence hydrogen bonding of the eight prodigiosin placed in the solvent was observed over the course of 200 ns. The results (Figure 4) show that there is a shift in the hydrogen bonding pattern of the prodigiosin from bonding with water initially when it is in the solvent layer, to then interacting with DPPC upon entering the membrane, then forming bonds with other prodigiosin molecules while the cluster is still intact and then again with DPPC and water as the cluster disperses to finally orient itself along the membrane-water interface either as single molecules or in smaller groups much like in the first simulation of just a single molecule.
Hydrogen bonds are formed between the hydrogen attached to the pyrrole rings and the oxygen in the water. Bonds are also formed between the oxygen atoms of the DPPC molecule (specifically the phosphoryl oxygen atoms, the carbonyl oxygen atoms of the glycerol ester and the ester oxygen atoms) with the hydrogen attached to the pyrrole rings, consistent with the RDF

values. Selected residues of the simulation showing the hydrogen bonds with the residues involved are shown in the Supplementary data file 2: Figure S9.

3.6.Effect of prodigiosin on the membrane

The area per lipid head and the thickness of the membrane are important parameters for assessing membrane stability and function. They were calculated for the data obtained from the simulation of eight prodigiosin molecules placed in the solvent. The area per lipid head changes at the time of entry of the prodigiosin molecules, but is restored to the initial values as simulation progresses (Supplementary data file 2: Figure S10). The MSD values of the phosphate group (P8) of the DPPC molecules is an indicator of the membrane fluidity [22]. The mean square deviation of the P8 group of the DPPC residues was calculated over the course of the entire simulation. The MSD values obtained in this simulation (Supplementary data file 2: Figure S11a) indicate that the DPPC membrane lipids are in the fluid state. The thickness of the membrane does not seem to be affected by the entry of prodigiosin into the membrane (Supplementary data file 2: Figure S11b). This could be due to the fluidity of the membrane.
One effect of the presence of prodigiosin at the membrane-water interface is the possible change in the solvent accessibility of the membrane. In this case, there are three possible groups for which solvent accessibility could be calculated: DPPC, prodigiosin and DPPC with prodigiosin. It was seen that the overall solvent accessibility of the membrane (DPPC with prodigiosin) reduced after the cluster of prodigiosin entered the membrane (Supplementary data file 2: Figure S12) and also after the molecules dispersed along the membrane. As expected individual solvent accessibility of prodigiosin alone or DPPC alone was more after prodigiosin entered the membrane (data not shown). This could be explained by the increased interactions of prodigiosin with the lipid polar groups, thereby preventing their access to water. Solvent accessibility is also a measure of the hydrophobicity of the system. There have been reports that prodigiosin in the membrane leads to

increased hydrophobicity [37], but the studies were not conclusive about the mechanism due to which this occurs. This study could lend further strength to this argument.
Inspection of the membrane structure just after the entry of the prodigiosin cluster (eight molecules) and also at the end of the simulation, when the cluster of eight molecules has broken up indicate this (Figure 5). The lipid architecture in the vicinity of the prodigiosin molecule was different (Figure 5A). The lipid heads appeared to orient themselves with the prodigiosin and not with the water. This could lead to local loss of membrane integrity and would be in concordance with reports on prodigiosin-induced leakage of intracellular substance from bacteria [12]. It can be seen, however, that this does not lead to diffusion of water across the membrane. They could aid the entry of certain molecules into the membrane or if they are in the vicinity of a membrane protein,they could lead to loss of the protein or its function. In fact, this could be the reason for the varied effects of prodigiosin on its targets. It is possible that the localisation of prodigiosin in close proximity to membrane proteins like the flagellar motor, the F0-F1 ATPase and the ion transporters could be crucial to its anti-bacterial activity. The fact that the prodigiosin residues disperse across the membrane could mean that they could now affect a larger area of the membrane (Figure 5B).
Further studies of the interaction of the prodigiosin in the membrane with membrane proteins like the flagellar motor, proton pump or other important membrane proteins could shed light on its effects on membrane proteins. It would also be interesting to study the dynamics of other prodiginines with the membrane. That could give us more insight into the role of the alkyl chains in the membrane interactions. Further studies to study the interactions of prodigiosin could involve simulations of prodigiosin with POPE membranes for bacteria with/without relevant membrane proteins.

4.Conclusion

The membrane localisation and interactions of prodigiosin have been investigated using a molecular dynamics approach. Individual prodigiosin molecules are insoluble in water and enter the

membrane immediately. They orient themselves along the membrane water interface, with the pyrrole rings interacting with the lipid head groups in the bilayer and with the water molecules. Hydrogen bonding and hydrophobic interactions are responsible for its localisation in the membrane and its interactions. When the concentration of prodigiosin increases in the solvent, the molecules clump together and small clusters readily enter the membrane. Bigger clusters do not enter the membrane. The small clusters of prodigiosin disperse laterally upon entering the membrane indicating better solubility in lipids as compared to water. The molecules cause changes in the local lipid architecture in the membrane and also decrease the solvent accessibility of the membrane. They do affect the lipid head area, membrane thickness and membrane fluidity. This indicates that they could cause most damage in the proximity of a membrane protein.

Acknowledgements

This work was supported by a grant from Department of Biotechnology, Government of India, (BT/PR5106/BRB/10/1073/2012). We also thank the Department of Science and Technology, Government of India for their funding. AR thanks Council of Scientific and Industrial Research, Government of India for her Fellowship (CSIR-NET). We thank the Biotechnology Information Services programme (BTIS, DIC, AU, BT/BI/03/013/2002), Department of Biotechnology, Government of India for computational facilities. We also thank the DBT – BUILDER project (BT/PR12153/INF/22/200/2014) for computational facilities and for providing the necessary infrastructure.

List of References
[1]N.R. Williamson, P.C. Fineran, F.J. Leeper, G.P.C. Salmond, The biosynthesis and regulation of bacterial prodiginines, Nat. Rev. Microbiol. 4 (2006) 887–899. doi:10.1038/nrmicro1531.
[2]H.K. Lee, Red to Red – the Marine Bacterium Hahella chejuensis and its Product Prodigiosin for Mitigation of Harmful Algal Blooms, J. Microbiol. Biotechnol. 18 (2008) 1621–1629.
[3]P.M. Gummadidala, M.E. Holder, J.L. O’Brien, N.J. Ajami, J.F. Petrosino, C. Mitra, Y.P.

Chen, A.W. Decho, A. Chanda, Complete Genome Sequence of Vibrio gazogenes ATCC 43942, Genome Announc. 5 (2017) e00733-17. doi:10.1128/genomeA.00733-17.
[4]D. Fehér, R.S. Barlow, P.S. Lorenzo, T.K. Hemscheidt, A 2-Substituted Prodiginine, 2-(p- Hydroxybenzyl)prodigiosin, from Pseudoalteromonas rubra, J. Nat. Prod. 71 (2008) 1970– 1972. doi:10.1021/np800493p.
[5]P.D. Schloss, H.K. Allen, A.K. Klimowicz, C. Mlot, J.A. Gross, S. Savengsuksa, J. McEllin, J. Clardy, R.W. Ruess, J. Handelsman, Psychrotrophic Strain of Janthinobacterium lividum from a Cold Alaskan Soil Produces Prodigiosin, DNA Cell Biol. 29 (2010) 533–541. doi:10.1089/dna.2010.1020.
[6]N.N. Gerber, Prodigiosin-like pigments from Actinomadura (Nocardia) pelletieri and Actinomadura madurae., Appl. Microbiol. 18 (1969) 1–3. doi:10.7164/antibiotics.24.636.
[7]A.M. Cerden, M.J. Bibb, G.L. Challis, Analysis of the prodiginine biosynthesis gene cluster of Streptomyces coelicolor A3 ( 2 ): new mechanisms for chain initiation and termination in modular multienzymes, Chem. Biol. 8 (2001) 817–829.
[8]D.X. Hu, D.M. Withall, G.L. Challis, R.J. Thomson, Structure, Chemical Synthesis, and Biosynthesis of Prodiginine Natural Products, Chem. Rev. 116 (2016) 7818–7853. doi:10.1021/acs.chemrev.6b00024.
[9]F. Grimont, P.A.D. Grimont, The Genus Serratia, Prokaryotes Vol. 6 Proteobacteria Gamma Subclass. (2006) 219–244. doi:10.1007/0-387-30746-X_11.
[10]A.H. Faraag, A.I. El-Batal, H.H. El-Hendawy, Characterization of prodigiosin produced by Serratia marcescens strain isolated from irrigation water in Egypt, Nat. Sci. 15 (2017) 55–68. doi:10.7537/marsnsj150517.08.Keywords.
[11]N. Busschaert, P.A. Gale, Small-molecule lipid-bilayer anion transporters for biological applications, Angew. Chemie – Int. Ed. 52 (2013) 1374–1382. doi:10.1002/anie.201207535.
[12]Suryawanshi, Rahul K , C.D. Patil, S.H. Koli, J.E. Hallsworth, S. V Patil, Antimicrobial activity of prodigiosin is attributable to plasma-membrane damage, Nat. Prod. Res. 31 (2017) 572–577. doi:10.1080/14786419.2016.1195380.
[13]Q. Yan, L.D. Lopes, B.T. Shaffer, T.A. Kidarsa, O. Vining, B. Philmus, C. Song, V.O. Stockwell, J.M. Raaijmakers, K.L. McPhail, F.D. Andreote, J.H. Chang, J.E. Loper, C.Q. Yan, E.E. Steven Lindow, Secondary Metabolism and Interspecific Competition Affect Accumulation of Spontaneous Mutants in the GacS-GacA Regulatory System in Pseudomonas protegens, MBio. 9 (2018) e01845-17. doi:10.1128/mBio.01845-17.
[14]T. Danevcic, M.B. Vezjak, M. Tabor, M. Zorec, D. Stopar, Prodigiosin induces autolysins in actively grown Bacillus subtilis cells, Front. Microbiol. 7 (2016) 1–10. doi:10.3389/fmicb.2016.00027.

[15]F. Wang, Prodigiosin found in Serratia marcescens y2 initiates phototoxicity in the cytomembrane, Electron. J. Biotechnol. ISSN. 16 (2013). doi:10.2225/vol16-issue4-fulltext- 7.
[16]P.L. Haddix, S. Jones, P. Patel, S. Burnham, K. Knights, J.N. Powell, A. LaForm, Kinetic analysis of growth rate, ATP, and pigmentation suggests an energy-spilling function for the pigment prodigiosin of Serratia marcescens, J. Bacteriol. 190 (2008) 7453–7463. doi:10.1128/JB.00909-08.
[17]J.C. Lapenda Lins, C.C.S. Maciel, H.S. Xavier, C.A. Alves da Silva, G.M. Campos-Takaki, Production and toxicological evaluation of prodigiosin from Serratia marcescens UCP/WFCC1549 on Mannitol Solid Medium, Int. J. Appl. Res. Nat. Prod. 7 (2014) 32–38.
[18]N. Darshan, H.K. Manonmani, Prodigiosin inhibits motility and activates bacterial cell death revealing molecular biomarkers of programmed cell death, AMB Express. 6 (2016). doi:10.1186/s13568-016-0222-z.
[19]N.R. Williamson, H.T. Simonsen, R.A.A. Ahmed, G. Goldet, H. Slater, L. Woodley, F.J. Leeper, G.P.C. Salmond, Biosynthesis of the red antibiotic, prodigiosin, in Serratia: Identification of a novel 2-methyl-3-n-amyl-pyrroie (MAP) assembly pathway, definition of the terminal condensing enzyme, and implications for undecylprodigiosin biosynthesis in Streptomyces, Mol. Microbiol. 56 (2005) 971–989. doi:10.1111/j.1365-2958.2005.04602.x.
[20]H.J.C. Berendsen, D. van der Spoel, R. van Drunen, GROMACS: A message-passing parallel molecular dynamics implementation, Comput. Phys. Commun. 91 (1995) 43–56. doi:10.1016/0010-4655(95)00042-E.
[21]M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindah, Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers, SoftwareX. 1–2 (2015) 19–25. doi:10.1016/j.softx.2015.06.001.
[22]J.A. Lemkul, From Proteins to Perturbed Hamiltonians: A Suite of Tutorials for the GROMACS-2018 Molecular Simulation Package, v1.0, Living J. Comput. Mol. Sci. In Press (2018) 1–52. doi:10.33011/livecoms.1.1.5068.
[23]O. Berger, O. Edholm, F. Jähnig, Molecular dynamics simulations of a fluid bilayer of dipalmitoylphosphatidylcholine at full hydration, constant pressure, and constant temperature, Biophys. J. 72 (1997) 2002–2013. doi:10.1016/S0006-3495(97)78845-3.
[24]K.B. Koziara, M. Stroet, A.K. Malde, A.E. Mark, Testing and validation of the Automated Topology Builder (ATB) version 2.0: Prediction of hydration free enthalpies, J. Comput. Aided. Mol. Des. 28 (2014) 221–233. doi:10.1007/s10822-014-9713-7.
[25]A.K. Malde, L. Zuo, M. Breeze, M. Stroet, D. Poger, P.C. Nair, C. Oostenbrink, A.E. Mark, An Automated force field Topology Builder (ATB) and repository: Version 1.0, J. Chem.

Theory Comput. 7 (2011) 4026–4037. doi:10.1021/ct200196m.
[26]W. Humphrey, A. Dalke, K. Schulten, VMD — Visual Molecular Dynamics, J. Mol. Graph. 14 (1996) 33–38.
[27]S. Buchoux, FATSLiM: A fast and robust software to analyze MD simulations of membranes, Bioinformatics. 33 (2017) 133–134. doi:10.1093/bioinformatics/btw563.
[28]Z. Sajjadiyan, N. Cheraghi, S. Mohammadinejad, L. Hassani, Interaction of aurein 1.2 and its analogue with DPPC lipid bilayer, J. Biol. Phys. 43 (2017) 127–137. doi:10.1007/s10867- 016-9438-z.
[29]N. Joondan, S. Jhaumeer-Laulloo, P. Caumul, A study of the antibacterial activity of l- Phenylalanine and l-Tyrosine esters in relation to their CMCs and their interactions with 1,2- dipalmitoyl-sn-glycero-3-phosphocholine, DPPC as model membrane, Microbiol. Res. 169 (2014) 675–685. doi:10.1016/j.micres.2014.02.010.
[30]D. Lopes, C. Nunes, P. Fontaine, B. Sarmento, S. Reis, Proof of pore formation and biophysical perturbations through a 2D amoxicillin-lipid membrane interaction approach, Biochim. Biophys. Acta – Biomembr. 1859 (2017) 803–812. doi:10.1016/j.bbamem.2017.01.031.
[31]S. Merino, J.L. Vázquez, Ò. Domènech, M. Berlanga, M. Viñas, M.T. Montero, J. Hernández-Borrell, Fluoroquinolone-biomembrane interaction at the DPPC/PG lipid-bilayer interface, Langmuir. 18 (2002) 3288–3292. doi:10.1021/la015627p.
[32]A.L. and Y.N. Kaznessis, Protocols for Studying Inhibition and Eradication of Bacterial Biofilms by Antimicrobial Peptides, Methods Mol. Biol. 618 (2017) 15–30. doi:10.1007/978- 1-60761-594-1.
[33]T. Malaspina, G. Colherinhas, F. De Oliveira Outi, E.E. Fileti, Assessing the interaction between surfactant-like peptides and lipid membranes, RSC Adv. 7 (2017) 35973–35981. doi:10.1039/c7ra04537a.
[34]S. Jing, S. Yu, C. Liu, J. Sun, H. Zhu, The dependence of quorum sensing in Serratia marcescens JG on the transcription of luxS gene, (2015). doi:10.1007/s00203-015-1099-y.
[35]L. Chen, J. Chen, G. Zhou, Y. Wang, C. Xu, X. Wang, Molecular Dynamics Simulations of the Permeation of Bisphenol A and Pore Formation in a Lipid Membrane, Sci. Rep. 6 (2016) 1–7. doi:10.1038/srep33399.
[36]J. Hage-Hülsmann, A. Grünberger, S. Thies, B. Santiago-Schübel, A.S. Klein, J. Pietruszka, D. Binder, F. Hilgers, A. Domröse, T. Drepper, D. Kohlheyer, K.E. Jaeger, A. Loeschcke, Natural biocide cocktails: Combinatorial antibiotic effects of prodigiosin and biosurfactants, PLoS One. 13 (2018) 1–23. doi:10.1371/journal.pone.0200940.
[37]S.R. Burger, J.W. Bennett, Droplet enrichment factors of pigmented and nonpigmented

Serratia marcescens: Possible selective function for prodigiosin, Appl. Environ. Microbiol. 50 (1985) 487–490.

List of Figures

Figure 1: Radial Distribution Function of selected atoms of prodigiosin. HW and OW represent the atoms in a water molecule, the other atoms shown are present in prodigiosin (74EL) and DPPC. (Please see supplementary data file 2: Figure S1 and S2 for naming scheme of the molecules).(A) Interaction of prodigiosin with water. O1-HW (black), OW-H12 (red) and OW-H19 (blue). (B) Interaction of prodigiosin with DPPC. O9_DPPC-H12_prodigiosin (black), O9_DPPC- H19_prodigiosin (red), O1_prodigiosin-P8_DPPC (blue).
Figure 2: Images of selected time frames of the simulation of eight prodigiosin molecules. The water molecules are shown in red above and below the membrane. The DPPC residues are shown using light blue lines and the eight prodigiosin residues are coloured by residues ID (i.e. a difference colour for each residue) and are depicted using the Van der Waals radius representation in all the images.
Figure 3: Density profile of the membrane after 100 ns simulation of a single molecule of prodigiosin showing the preferred localisation of prodigiosin. (A) Partial density of DPPC and water (B) Partial density of eight molecules of prodigiosin over the course of 200 ns of simulation. 74EL represents the localisation of the molecule at the start of the simulation.
Figure 4: Hydrogen bonding of the various residues during the simulation of eight molecules of prodigiosin dispersed randomly in the water over the course of 200 ns (A) DPPC – water (B) DPPC – prodigiosin (C) Prodigiosin – prodigiosin (D) Prodigiosin – water. Increased hydrogen bonding of prodigiosin with the DPPC (B) is accompanied with a corresponding decrease of its hydrogen bonding with the water molecules (D).
Figure 5: Depiction of the membrane architecture with eight prodigiosin molecules. Side view (A) showing the change in lipid orientation and top view (B) showing localisation on the membrane.

Molecular dynamics of the membrane interaction and localisation of prodigiosin
Highlights:
•Molecular dynamics simulations were carried out to understand the membrane
localisation and interactions of prodigiosin in a DPPC bilayer
•It was found that prodigiosin entered the membrane upon simulation either as individual molecules or as small clusters due to its sparing solubility in water
•The molecules, upon entry, had interactions with the DPPC head groups and with the water molecules.
•The localisation of the molecules resulted in a local lipid architecture in the membrane and in a decrease in the solvent accessibility of the system

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: