The low permeability of the outer cell wall of Gram-negative bacteria has been identified as one of the key problems in antibiotic research (see IMI call devoted to “New drugs for bad bugs”). Within the above mentioned call IMI will support about 50 teams investigating broad aspects of antibiotic permeability ranging from basic science to large pharmaceutical companies with their antibiotic research departments. The lMI project involves mainly postdocs and senior scientists. For additional 12 young researchers, the present ITN project will provide a unique possibility to participate in this particular research environment. As all teams are contributing to the larger goal, the individual young researcher will have a substantially increased number of possibilities for training and networking.

Moreover, science is advancing on an international level and for a researcher it is of key importance  to understand the organization of R&D in different countries. Due to the high level of internationalization, Europe needs a large pool of scientists having a European view and understanding of research. This will help to overcome national differences and to promote a global view. Furthermore, an intensive exchange of personnel is a very efficient tool to connect the expertise of the participating partners. Thus, it is seen as an essential target of this research proposal to have a pool of junior researchers involved in joined projects based on a shared supervision of the work

The discovery of new agents to treat resistant Gram-negative infections generally relies on developing agents that can penetrate through at least one, often both of the membranes of the Gram-negative cellular envelope.  This envelope, comprised of an outer and inner membrane, which have significantly different properties, provides Gram negative bacteria with an excellent physical barrier to antibacterial agents.  Furthermore, if an agent is able to penetrate one or both of these membranes, it is often subject to rapid efflux out of the cell by numerous broadlyacting efflux pumps—rendering the agent ineffective.  This combination of an intrinsic penetration barrier, combined with potential for rapid efflux often leads to where a compound with good intrinsic activity at an intracellular or periplasmic target displays poor whole cell antibacterial activity.  This represents a real, unsolved problem for drug discovery efforts.   If these processes were better understood, it would provide the industrial, biotech and academic community renewed optimism and tool for the discovery of new antibiotics to treat the severe, multi-resistant Gram-negative infections that are becoming more and more common in Europe and across the globe.

In the following sections detailed descriptions of the teams and their tasks are presented:

ESR1 (Bremen-JUB-MW)

Satya Prathyusha Bhamidimarri

Supervisors: Mathias Winterhalter, Ulrich Kleinekathöfer, Jacobs University Bremen

Cosupervisor: Jean-Marie Pagès (AMU), Niels Fertig (NAN), Bert van den Berg (NEW)


Characterization of the influx of antibiotics or efflux pump blockers through porins using electrophysiology and complementary techniques. Here we could finish a first investigation ahead of time. For Cmpylobacter we are finalizing a ms. Whereas for Providencia we have a ms. Just accepted and one under review. We identified a further bacterial channel with the unusual property to allow passive diffusion of large carbohydrates like alpha-cyclodextrin. The objective during the first period was to elucidate the underlying mechanism of transport of such a bulky molecule through the nanopore and to exploit this to further understand the biophysics behind this which will aid in understand the transport of molecules across a porin in general


The CymA porin from Klebsiella oxytoca is responsible for passive cyclodextrin uptake. Passive uptake of such large molecules by a porin channel was very much surprising and we subsequently investigated transport properties at single molecular level. Based on the high resolution crystal structure obtained in Newcastle the theory group at JUB-UK could perform all-atom modelling to obtain a complete picture: Cyclodextrin binds to the extracellular side of CymA with its plane parallel to the membrane, tilts vertically and glides through the channel. At the periplasmic side there is a flexible loop which acts as a gate preventing free flow of molecules but allows the transport of cyclodextrin through the channel. In addition to a general understanding CymA might be used as a potent new type of biosensor. Cyclodextrin may be bound covalently at the entrance which allows in developing a specific cyclodextrin based sensor for small molecules. Overall this will provide a sound understanding on molecular transport across channels.

In Gram negative bacteria the outer membrane acts as a selective uptake barrier. It contains protein channels (porins) which provide an entry pathway for hydrophilic molecules like small nutrient molecules and β-lactam antibiotics. Understanding uptake of these molecules via porins is vital to comprehend the transport mechanism across the cell membrane. Electrophysiology forms a promising approach to study the permeation of molecules across outer membrane and thereby understanding molecular interactions with the channel.  However effects due to external applied voltage on the molecular permeation in porins remain unclear in this technique. Voltage effects are even prominent on neutral molecules which are related to the presence of electro osmosis flow. Here we present biophysical characterization of CymA from K.oxytoca which has the ability to take up cyclodextrin (neutral molecules). Detailed single channel analysis revealed inherent asymmetric gating characteristics of the channel. And also voltage dependent interaction of neutral molecule (cyclodextrin) with CymA presumed due to Electro osmosis. To further elucidate these effects, substrate interaction studies are performed under various external conditions. To obtain an atomistic view we complement our studies with all-atom molecular dynamics simulation to study the flow of water and also its directionality in the channel in the absence of substrate and in the presence as well. For the first time we have explicitly shown the existence of such an effect in the outer membrane uptake channel.

Figure 1: Left Panel: Crystal Structure of CymA (A-C) without substrate and with substrate (D-F) . Right Panel: Trace depicting substrate interaction of CymA with alpha Cyclodextrin at different applied voltages.


B.van den Berg, S.P. Bhamidimarri et al. Outer membrane translocation of bulky small molecules by passive diffusion. PNAS (Manuscript accepted)


ESR2 (Munich-NAN)

Jiajun Wang

Supervisor: Niels Fertig, Nanion Technologies Munich, 

Co-Supervisor: Mathias Winterhalter (JUB)


Screening of antibiotic permeation through porins towards low throughput <100recordings/day. Nanion developed a unique planar patch clamp chip used for the electrophysiological analysis of mammalian cells. Planar bilayer recordings are attractive for investigations of membrane proteins not accessible to patch clamp analysis, like e.g. proteins from organelles or bacteria. This technique offers substantial advantages as compared to traditional patch clamp and BLM recording, in terms of facile handling and improved sensitivity. Particular the enhanced sensitivity will improve the time resolution of the measurements.


Omp35 and Omp36 are compared to their homologues (OmpF, OmpC) by using antibiotics with different biophysical properties on single molecular level. In my studies, enrofloxacin, cefepime, imipenem gave different association rate between OmpF homologues. Magnesium ions also gave different modification to the interaction of norfloxacin with OmpC and Omp36. Combined with structural information (St. Andrews) our data will be used in all atom modelling (Cagliari) to predict the pathway of antibiotic penetration.

Screening of antibiotic permeation through porins towards low throughput <100recordings/day. First we did  complementary measurements to ESR1 but on the chip. Reconstitution of MOMP/OMP50 and testing for a larger compound library.

The investigation of the penetration strategies of molecules through the bacteria cell membrane at the basis of influx and efflux is the aim of the whole project. In this study, the interactions of outer membrane proteins (OMPs) with antibiotic molecules were analyzed on the single molecular level. Omp35 and Omp36 from Enterobacter aerogenes are the porin homologs of OmpF and OmpC found in Escherichia coli and shares high similarities in gene sequence. In my studies, enrofloxacin, cefepime, imipenem gave different association rate between OmpF and its homologue Omp35. Magnesium ions also gave different modification to the interaction of norfloxacin with OmpC and its homologue Omp36. With the aid of crystallization and computerization, the function of residue groups and antibiotic interaction strategies would be addressed.

Instrument for bilayer study with improved sensitivity and time resolution designed by Nanion Technologies GmbH such as port-a-patch, Orbit 16 and Orbit mini has been used. The interaction between purified protein (OmpF) and antibiotics are recorded by port-a-patch as well as Orbit 16 and compared to BLM techniques. The results are published as Antibiotic translocation through porins studied in planar lipid bilayers using parallel platforms. The purified proteins are reconstituted with the planar lipid bilayer. The cell free expression protein tackled with the expression problems for the proteins but the relatively low reconstitution rate to bilayer remains a problem. Different instruments such as Orbit 16 were used to meet this requirement. The cell free expressed protein is proved to be functionalized and published as Cell-free expression of a functional pore-only sodium channel.


C. Weichbrodt, H. Bajaj, G. Baaken, J. Wang, S. Guinot, M. Kreir, J. C. Behrends, M. Winterhalter and N. Fertig, Analyst, 2015, DOI: 10.1039/C4AN02335H.


ESR3 (Cagliari-UCA-MC)

Silvia Acosta Gutierrez

Supervisor: Matteo Ceccarelli, Department of Physics, University of Cagliari

Co-supervisors: Ulrich Kleinekathöfer (JUB)


The group combined MD simulations with an acceleration scheme to follow the translocation of antibiotics through porins at atomic scale. Multi-scale algorithms extend the simulation time to the range to milliseconds, making it possible to obtain the reactive pathway that antibiotics follow during passive diffusion. Accelerated MD simulations have revealed a putative translocation pathway for penicillins and fluoroquinolone  through OmpF and porins extracted from resistant strains. Through the associated free energy surface (FES) of this process, affinity sites and activation barriers can be identified.


Our activity focuse on the permeability of antibiotics through general channels, namely OmpF/OmpC from E.coli and MOMP from C.jejuni (collaboration with AMU, JUB and STA). For the former we have several data available and in particular X-ray co-complex structures with penicillins that we aim to compare with our modelling results. We started characterizing the internal electrostatic of pores looking at the ordering of water molecules, being the latter a natural probe for quantifying an internal electric field because of their dipole.

Starting from the X-ray structures obtained by USTAN, we simulated MOMP, both the momomeric and trimeric form, embedded in a phospholipid bilayer. Also from our modelling the two forms do not differ, therefore we used the small monomeric form to investigate the diffusion of ciprofloxacin.

We developed an analysis method to evaluate the internal electric field of pores. Using spheres of water molecules and the Clausius-Mossotti formula, we were able to correlate the ordering of water dipole moment to the internal electric field inside OmpF/OmpC and mutants from clinical starins. These channels show a very strong transversal component of electric field that can change orientation along the axis of the pore.

We quantified the interactions of a few penicillins within OmpF and we compared our free energy minima structure with the X-ray soaked complex structures solved by B. Roux, obtaining a very good agreement. From the one-dimensional free energy surface knowledge we are developing a method to evaluate the flux of molecules traversing the channel upon application of a concentration gradient.


S. Acosta-Gutierrez, M. A. Scorciapino, I. Bodrenko and M. Ceccarelli, The Journal of Physical Chemistry Letters, 2015, DOI: 10.1021/acs.jpclett.5b00612, 1807-1812.


ESR4 (Newcastle-NEW)

Monisha Pathania

Supervisor: Prof. Dr. Bert van den Berg, Newcastle University

Co-supervisors: M. Ceccarelli (UCA), U. Kleinekathöfer (JUB)


Our goal is to clone, express and purify OmpU and OmpT, the two key OM uptake channels of Vibrio cholerae to give a better overall picture of the structure of a diverse set of porin proteins. In addition we will attempt to co-crystallize these OM channels with substrates, giving a rich body of structural information to help inform computation simulations in this and other WPs

Newcastle will work to solve structures of Vibrio cholerae OM channels by X-ray crystallography. In addition the substrate specificity of the porins will be investigated by using in vitro and in vivo transport assays.


The main aim of the project at NEW is to study and understand the antibiotic permeation in the OmpF/C-like outer membrane proteins OmpU and OmpT of Vibrio cholera. For this, we will solve the OmpU/T X-ray crystal structures along with structural studies on other porins of Gram-negative bacteria for comparison purposes, like Enterobacter cloacae (OmpE35 and OmpE36) and Klebsiella pneumonia (OmpK36). In addition to solving structures we will test uptake of different antibiotics via these outer membrane proteins via antibiotic permeation assays (eg. liposome swelling experiments and disk diffusion assays). The interaction of OmpU and OmpT with antibiotics will also studied by single-channel electrophysiology. The results will be combined and used to obtain co-crystal structures of selected antibiotics with OmpU and/or OmpT. Together the project should give new insights into how antibiotics enter V. cholerae and other pathogenic vibrios.

We have expressed and purified OmpU from V. cholerae as well as OmpK36 from Klebsiella pneumonia and OmpE35/36 from Enterobacter cloacae. All proteins are available in milligram quantities. We have solved the X-ray crystal structures of OmpU and OmpE36 to high resolution.

Proteins of Vibrio cholerae

The main aim of the project at NEW is to study and understand the antibiotic permeation in the OmpF/C-like outer membrane proteins OmpU and OmpT of Vibrio cholera.  For this, we will solve the OmpU/T X-ray crystal structures along with structural studies on other porins of Gram-negative bacteria for comparison purposes, like Enterobacter cloacae (OmpE35 and OmpE36) and Klebsiella pneumonia (OmpK36). In addition to solving structures we will test uptake of different antibiotics via these outer membrane proteins via antibiotic permeation assays. (eg. liposome swelling experiments and disk diffusion assays). The interaction of OmpU and OmpT with antibiotics will also studied by single-channel electrophysiology. The results will be combined and used to obtain co-crystal structures of selected antibiotics with OmpU and/or OmpT. Together the project should give new insights into how antibiotics enter V. cholerae and other pathogenic vibrios.

We have expressed and purified OmpU from V. cholerae as well as OmpK36 from Klebsiella pneumonia and OmpE35/36 from Enterobacter cloacae. All proteins are available in milligram quantities. We have solved the X-ray crystal structures of OmpU and OmpE36 to high resolution.

Structure of OmpU- The crystals of OmpU showed one trimer in the asymmetric unit. Each monomer is a 16-stranded β-barrel with longer loops on the extracellular side and short turns on the periplasmic side. Thus, the OmpU structure is very similar to that of other porins. The most interesting part of the OmpU structure is the N-terminus, which does not form a salt bridge with the C-terminus as observed in all porins solved so far. Instead, the N-terminus of OmpU extends into the pore from the periplasmic side and constricts the pore. This is a novel feature.

Structure of OmpE36- Thecrystal of OmpE36 diffracted to ~ 1.5 Å resolution with two trimers in the asymmetric unit. As expected from the sequence similarity, the overall structure of OmpE36 is very similar to other OmpC proteins from Enterobacteriaciae.

LPS attached to OmpE36 protein- Surprisingly, bound lipopolysaccharide (LPS) molecules are present for each monomer of the OmpE36 trimer. One OmpE36 monomer has two bound LPS molecules. So far only one other structure of an OMP (the TonB dependent transporter FhuA) has been reported with bound LPS.


Figure 1: (A) Cartoon diagram of OmpU trimer, side view. (B) Rainbow colour cartoon of single monomer.


Figure 2: (A) Cartoon diagram of OmpE36 trimer, side view. (B)Rainbow colour cartoon of OmpE36 single monomer (bottom view). (C) Full OmpE36 trimer shown with four bound LPS molecules.


ESR5 (St. Andrews-STA)

Luana Ferrara

Supervisor: James H Naismith, School of Chemistry, University of St Andrews

Co-supervisors: M. Ceccarelli (UCA), J.-M. Pagès (AMU)


The group uses x-ray crystallography to study the structures of membrane proteins. We are particularly interested in channels which conduct ions and polar molecules. For example, the crystal structure of OmpC mutants revealed that changes in antibiotic transport were more likely due to changes in the transverse electrostatic field. The group is also pioneering the use of PELDOR spectroscopy to monitor conformational change in integral membrane proteins.


My goals are to determine the structure of MOMP and Omp50 porins from the pathogen Campylobacter jejuni and Omp35, Omp36 and Omp37 porins from the pathogen Enterobacter Aerogenes. To do so, all the proteins have been expressed in a suitable system, a purification protocol have been optimized and the purified proteins have been crystallized. I have solved the structure of MOMP and Omp36. The structure will serve as entry for all-atom modeling of the antibiotic entry.

The increasing of multidrug resistance (MDR) amongst bacteria is a global concern for public health.  MDR is most serious in Gram negative bacteria due to their additional outer membrane, whose low permeability makes the influx of antibiotics more difficult. Embedded in the outer membrane, water filled channels, known as porins, represent a “gate” through which small hydrophilic molecules such as sugar, drugs and chemicals can enter the cell (1). As a resistance mechanism, porins can be down-regulated and/or mutated when bacteria are pressured by antibiotics. (2). A better understanding of porins structure will help us to clarify how they interact with antibiotics and hence accelerate the design of new drugs.

MOMP has been purified from the native Campylobacter jejuni. Culture preparation and purification were performed in Prof Pages‘s lab (secondments). The purified protein was then successfully crystallized and the structure solved at 2.89Å resolution. The structure shows that MOMP is a trimeric 18-stranded porin with the typical β-barrel character (Fig.1). MOMP structure was also solved using the protein overexpressed in a E. coli system. Comparison of the two structures did not show any significant difference. Also, conductance profiles in planar membrane of both native and recombinant MOMP are the same (native MOMP G= 2.3±0.3 nS; recombinant MOMP G=2.2±0.2) (secondment JUB).

Figure 1. The 2.89Å resolution structure of MOMP trimer from top (left) and monomer from the side (right).

Purification of Omp50 from the native campylobacter has been challenging. Omp35, Omp36 and Omp37 have been cloned and purified. All of the three porins have been already crystallized and Omp36 structure solved at 2.4Å resolution.  The structure shows that Omp36 is a trimeric 16-stranded porin with the typical β-barrel character (refinement is in progress) (Fig2).

Figure 2. The 2.4Å resolution structure of Omp36 trimer from the top (left) and monomer from the side (right)


ESR6 (Frankfurt-GUF)

Jenifer Cuesta Bernal

Supervisor: Martin K. Pos, Biochemistry, Goethe Universität Frankfurt

Co-supervisor: Paolo Ruggerone, Ulrich Kleinekathöfer, Jacobs University Bremen


The RND component AcrB has been intensively studied using biochemical and structural methods. High-resolution structures of wild-type AcrB and several of its single-site variants have been determined via X-ray crystallography. Latest structural data revealed multiple binding sites for drugs simultaneously within the loose and tight promoters.  Insights into the drug binding sites are key to further studies on how to develop inhibitors of the RND component and are important sources for forthcoming computational analysis on the drug efflux mechanisms.


We selected 6 RND multidrug efflux transporter genes from S. typhimurium and C. jejuni for proof-of-principle of a streamlined, time-saving and economical approach to identify stable protein candidates to enter the larger production and purification. Twelve constructs were synthesized in E. coli as a GFP fusion protein using the versatile FX-cloning procedure. Expression yield and stability within different detergents were directly visualized in crude E. coli detergent extracts. From the initial 6 cloned genes, we obtained 3 pure and stable RND efflux transporters, of which two produced crystals in the initial crystallization screen. These results are very promising towards the elucidation of RND efflux transporter structure, which will yield valuable data for the design of new antibiotics/inhibitors in the fight against multiple drug resistance.  The data will serve as input for all atom MD in Cagliari.

Identification of stable RND protein candidates for crystallization studies

We focused initially on RND proteins from C. jejuni and S. typhimurium (CmeB, CmeF, AcrB, AcrD, AcrF and MdsB), in order to establish a strategy based on Membrane Protein - Green Fluorescent Protein (MP-GFP) fusions1. Initially 12 RND-GFP constructs (pT7 and pBAD based) were generated by FX cloning 2. Screening and optimization of expression conditions in E. coli was followed by whole cell fluorescence measurements and in-gel fluorescence (Figure 1).

Figure 1: Detection of RND-GFP fusion proteins by in gel fluorescence and Western Blot.

For S.typhimurium AcrB and C. jejuni CmeB, first crystal hits were obtained (Figure 2). S. typhimurium AcrB crystals diffracted to 8 Å resolution. C. jejuni CmeB crystals have not been exposed to X-rays yet.


Figure 2: Crystals of S. typhimurium AcrB (left) and C. jejuni CmeB (right)

The results obtained are in-line with the objectives. We invested time to establish a stream-lined approach, which in the next months will be extended to RND multidrug efflux transporters from E. aerogenes and K. pneumonia. We also start with the inhibitor co-crystallization of E. coli AcrB and in due time, with RND proteins from S. typhimurium and C. jejuni.


ESR7 (Bremen-JUB-UK)

Fabio Grassi

Supervisors: Ulrich Kleinekathöfer, Jacobs University Bremen

Co-supervisor: Paolo Ruggerone, Martin K. Pos, Biochemistry, Goethe Universität Frankfurt


The computational group has expertise in simulating the ion and substrate transport through porins but more importantly in modelling TolC and AcrB (in cooperation with the groups from UCA and GUF). It is still an open question how and why TolC and its homologoues assume open conformations upon assembling of the tripartite complex. Our group will complement ongoing experimental work using all-atom MD studies. Molecular-level hypotheses by the experimental partners can be tested and new experiments can be suggested. Attention will be devoted to the assembly of the different efflux pump components beyond simple static docking models and learning from the recent CusAB structure. 


The work performed by Fabio Grassi has been focused on the study of the E. Coli TolC by means of molecular dynamics simulations. TolC is a major component of bacterial efflux systems which pumps out the toxic compounds from the cell interior that inhibits the bacteria growth. Results indicate that thus far unreported residues play a key role in conformational stability and identify ion selective sites. These findings can provide valuable insight into the opening mechanism of TolC, thus improving upon the current understanding of efflux pumps and, therefore, aiding in the design of new antibiotics.

In a further collaboration between the European Screening Port IME Fraunhofer institute and both ESRs at Jacobs University we are working on a joint project with electrophysiology, all-atom MD simulation and docking studies on TolC.

Molecular dynamics simulations of wild type TolC, as well as of nine mutants, were performed with neutral systems and at [KCl] = 0.1 M and 1 M. The effect of ions was studied both by applying electric fields to the system and by restraining the ions outside the protein. Ions were restrained by applying an energy penalty to movement in one dimension, thus restraining them on a plane perpendicular to the main axis of TolC. The criterion for selecting mutants was based upon existing experimental literature and on collaboration with Dr. Vassiliy Bavro.

Findings based on the analysis of hydrogen bonds and salt bridges confirm the importance of certain key residues, namely R367, D371 and D374, located at the periplasmic end, while others, particularly D162, are shown to be of less significance than hitherto believed. Two residues close to the equatorial domain, R328 and R18, are shown to be consistently involved in two interchain salt bridges, while closer to the tip, the salt bridge E359 – R135 is shown to feature prominently throughout the whole simulation: to our knowledge, this is the first time that the effect of these residue pairs is observed. Lastly, analysis of the trajectories of potassium ions has revealed three ion pockets in the beta barrel close to the junction with the periplasmic domain: these sites could potentially provide a target for a properly designed blocking molecule. Work on the AcrA-TolC complex is yet be commenced: the investigation of the mutants which exhibited unexpected behaviour proved to be more prolonged than anticipated, resulting in a delay of the second part of the project.

Figure 1. Left: TolC periplasmic domain intramonomer links. Center: TolC periplasmic domain intermonomer links. Right: ion pocket sites.


ESR8 (Cagliari-UCA-PR)

Venkata Krishnan Ramaswamy

Supervisor: Paolo Ruggerone, Department of Physics, University of Cagliari

Co-supervisors: M. Ceccarelli (UCA), J.-M. Pagès (AMU)


UCA together with JUB and GUF has performed MD and state-of-the-art docking studies on AcrB addressing also the impact of a point mutation on AcrB activity. Additionally, UCA, in collaboration with BPI, has investigated the recognition mechanism of imipenem and meropenem by MexB. Still unclear are several issues, such as to what extent the functional rotation, i.e., the specific series of sequential conformational changes, is essential for the drug extrusion and whether cooperativity effects are also involved. Thus, molecular details of the mechanism, recognition and uptake for AcrB and MexB require further investigations that will be performed by UCA in collaboration with JUB and GUF. Inhibitors, used in combination with antibiotics, expand the spectrum of antibacterial activity, reverse resistance and dramatically reduce the rates of resistance development, but the molecular details of their action are still elusive. The ‘non-specificity’ of the transporters asks for the role of the pump’s putative affinity in resistance and inhibition. Additionally, insights on possible allosteric sites in the efflux pumps will be gained by extended MD simulations and indicate sites to be targeted


Our computational research is focused on the efflux pumps of the resistance-nodulation-division (RND) family, among the major responsible for the appearance of multidrug resistance. We are performing a comparative study of transporters of the RND family showing different substrate specificity, i.e., AcrD and AcrB of E. coli and MexY and MexB of P. aeruginosa. For instance, AcrD and MexY transport aminoglycosides but AcrB and MexB do not. The results of this comparative investigation will offer insights into the microscopic details of the functioning of the efflux systems and, when mapped onto the chemical structures of the compounds considered in the present study, will possibly help the design of molecules able to escape and/or to inhibit the efflux systems.

Venkata is carrying out a very thorough computational study of the RND transporters of E. coli and P. aeruginosa. Comparison of the interaction pattern of compounds that are affected differently by different efflux systems; (1) Comparison of interaction pattern of compounds that are poor and good substrates of a specific efflux system;

The research requires a series of steps:

  1. For two transporters, namely AcrD and MexY, no crystal structures are to date available, although the purification of AcrD has already been completed and crystallization is ongoing within the ITN Translocation Consortium (K.M. Pos, GUF). Structures of AcrD and MexY have to be built by homology using the crystal structures of AcrB and MexB as template. Different models have been built and validated by bioinformatics tools. ACHIEVED
  2. MD simulations of the selected models. Equilibration of the models is performed by MD simulations with the transporters inserted in the membrane, in the presence of the solvent and at physiological temperature. The resulting trajectories will be used to characterize the key features of the dynamics of the systems, i.e., interaction with solvents, relevant interactions stabilizing the proteins, structural features of key regions and domains. The comparison with results obtained by Cagliari’s group on AcrB and MexB will be helpful to select putative relevant regions in AcrD and MexY. ONGOING
  3. Molecular Docking. Docking runs of ceftobiprole and kanamycin, two compounds known to be extruded by MexY, have already been performed using the homology model of the transporter. Furthermore, the extensive trajectories of Point b will be exploited to identify the relevant conformations explored by the systems during the simulation time. These representatives will be used for the docking runs to analyse the binding of selected compounds to dynamical structures. Similar procedure will be adopted for AcrD. ONGOING
  4. Interaction pattern. The poses extracted by the docking runs of Point c will undergo validation and equilibration by MD simulations. The trajectories will be analysed to extract key interaction patterns of the different compounds with the transporters. The interaction with the solvent in terms of residence time of the water molecules in specific regions will be also examined. Finally, the affinity of the compounds will be dissected in residue contributions to highlight hotspots of the interaction.
  5. MD simulations of AcrB. Simulations of AcrB already performed in Cagliari will be extended and analysed in tune with the strategy highlighted in Point b. Behaviour of waters, relevant motions of domains and regions will be identified. This will offer insights into possible coupling between key parts of the transporter. Additionally, the interactions with AcrB of compounds that are known as poor substrate of AcrB but are affected by AcrD will be studied. Analogous research will be done for MexB.


ESR9 (Hamburg, ESP, now Fraunhofer IME-SP)

Alessia Gilardi

Supervisor: Dr. Philip Gribbon, Frauhofer IME

Cosupervisor: Mathias Winterhalter (JUB)


Our principle approach is to improve the availability of antibiotics at their site of action by selectively blocking porin function. We will quantify isolated porin efflux systems using the Iongate (Surf2er) and Ionovation (Compact) in- vitro cell free electrophysiology technologies, by characterizing electrochemical activation via capacitance-based readouts. A screening process, to monitor the ability of compounds to modulate the function of porins, will be defined and validated. The aim is to develop industrial quality primary electrochemical assays with the capacity to process 100’s of compounds per week. Hit compound selection will be supported by high content secondary assays and classical fluorescence efflux measurements in cell based systems. We will work with molecular modelling teams to understand the putative mechanism of interaction of Hit compounds with key candidate porins. The ESP has a comprehensive compound logistics, compound screening and profiling infrastructure (in-vitro and in-silico) at its disposal and access to a range of drug and lead-like compound libraries for assay validation and screen prosecution.

The young researcher will test reconstitution systems and establish the assay robustness in terms of classical HTS quality criteria including: Z’ factor (quantifies variation in High and Low controls relative to the assay window); pharmacological response to standard compounds (to be defined and provided by industry partners); and the temporal and spatial stability of the signal. This state of the art equipment at the ESP will allow throughput of 200 compounds per week which permits for screening of  > 2000 compounds over the project period. The optimal screening collection will be carefully selected from the available library of 200,000 compounds ESP using pharmacophore and compound docking approaches available in-house (Gold, Flex-x). High Content based secondary assays will be developed on the Opera confocal scanning microscope to confirm compound activity translates to cellular system. A suite of image analysis tools (Acapella, Harmony and Columbus – Perkin Elmer) are available for primary data processing. These software packages are supported by the industry grade data reduction tools (ActivityBase-XE, IDBS) for plate based data analysis and end-point quality control assessments. Compound activity will be confirmed in whole bacteria based on fluorescence efflux assays. In-silico assessments of Hit compounds will be made in conjunction with WP1 and WP2. The predicted physico-chemical properties of key Hit compounds (cLog P, TPSA etc) will be determined. To confirm that the effects observed are not the result of promiscuous or non-specific inhibitors, a preliminary assessment of off-target effects will be made using readily-available assays against representative targets from therapeutically validated classes  (proteases, kinases and HDACs). To prioritize hits, an initial measure of possible general compound cyto-toxicity properties will be performed in engineered (HEK) and primary human cell lines (PBMC’s).  


The main target of our work is TolC, outer membrane protein of the efflux system AcrAB-TolC present in E. coli. Addressing TolC, our goal is to identify small molecules that can modulate the transport through the channel, interfering with the functionality of the complex. For that, we developed a strategy based on in silico approach followed by in vitro confirmation. After a structure-based characterization of the protein, we selected the periplasmic site of TolC as a target for virtual screening campaigns and compound libraries for database setup. The outcome will be analyzed and the selected hits will be tested in cell-free based in vitro assays (using label free electrophysiological and biochemical technologies, such as BLM and SPR) in order to confirm and characterize compounds activity. A constant collaboration has been setup with the ESRs from the Jacobs University (M.Winterhalter and U.Kleinerkathoefer’s groups) for a complete understanding of TolC function and role in antibiotic resistance.

In the frame of the ITN Translocation project the main goal is a better understanding of the mechanisms involved in gram-negative bacteria antibiotic resistance and the finding of reliable solutions to overcome this problem. Down-regulation of porin expression and increasing of translocation of molecules through efflux systems are the key mechanisms used by bacteria to reduce the presence of toxic compounds, such as antibiotics, in their intracellular and periplasmic space. Within the project developed at the Fraunhofer IME ScreeningPort, the efflux system AcrAB-TolC present in E. coli has been selected as a target, focusing major attention on the outer membrane protein TolC of this tripartite complex.

TolC is a homotrimer of ~53 KDa, presenting an alpha/beta-barrel structure, that spanning from the periplasm to the outer membrane, creates a channel through which the molecules can be transported to the extracellular space. In its structure 3 main domains can be identified: an outer membrane domain, consisting in a rigid beta-barrel inserted in the outer membrane; an equatorial domain, where TolC can essentially get in contact with the periplasmic adaptor protein AcrA; and a periplasmic domain, site of interaction with AcrB, catalytic protomer of the complex. The periplasmic site seems to have an important role in the control of TolC functionality, in fact it presents two Aspartates ring (two residues for each monomer) are interacting together to keep a close conformation of the protein channel. So far only few compounds have been identified showing inhibitorial activity against this protein (all toxic) and a molecule of hexaamminecobalt has been co-crystalized with TolC, interacting with the aspartate rings present in the periplasmic site.

Based on these studies a strategy was developed for the identification of small molecules that could act on the periplasmic site of TolC and specifically with aspartate rings, keeping the protein structure in its closing position. A virtual screening has been set up using different software and tools to prepare the protein structure as well as a deep analysis of the target site, virtual compound library preparation and docking parameters. At the same time an assay was established to characterize functional and electrophysiological features of TolC, confirming previous literature data in order to have a robust system where screen compound hits resulting from the in silico screening. Alternative approaches have been planned to target the AcrAB-TolC complex. Recent publications have shown how the interaction between AcrA and TolC could be fundamental for TolC to switch to its opening state and complete its function. For this reason it would be interesting to investigate which regions of the proteins are involved in the cited interaction and conduct an in silico and in vitro campaign to identify compounds able to interact in that area.  

Virtual Screening

The X-ray structure of the wild type TolC (PDB ID: 1EK9) and the co-crystallized structure with the hexaamminecobalt bound (PDB ID: 1TQQ) have been deeply analyzed and prepared for the virtual screening campaign. In first place, the structures have been protonated using the “protonate 3D” function of the MOE suite, while the energy minimization step was done using the SYBYL tool, after a testing process of different parameters for the identification of the best conditions. The structures were then analyzed using validation tools such as Procheck and Prosa. A database has been prepared starting from the CleanLeads subset (ZINC - is not commercial – library): from the original group of about 6 million compounds, a first selection has been done discarding the molecules without any positive charged amino group and following the accordance with druglikeness (Lipinski’s rule of 5).

300000 compounds have been docked using GOLD docking software (parameters selected after a test on a set of benchmark proteins, data not shown), based on a Lamarckian Genetic Algorithm to predict the position of interaction between protein and ligand. The best ranked poses have been selected for a post-docking analysis. Almost 3000 compounds have been re-docked to obtain more precise information about protein-ligand interaction and Ligand Efficiency (LE). Interesting its will be selected for in vitro screening.


Figure 1: Structure TolC (PDB ID: 1EK9) with highlighted site targeted for the virtual screening (left). Interactions between a docked compound with aminoacid residues of the target site (right).


ESR10 (Basel-BAS)

Pauline Maturana

Supervisor: Dirk Bumann (BAS) 

Co-supervisor: Jean-Marie Pagès, Aix Marseille University


Current whole-cell assays for screening antimicrobials rely on standardized, well-accepted in vitro conditions. Although useful, such conditions may not fully reproduce relevant conditions that pathogens encounter in infected host tissues.  Gram-negative bacteria such as Pseudomonas readily adapt to different conditions by comprehensively remodelling their cell envelope properties such as differential expression of one or more of the some 30 porins, induction of one or more of their ~20 efflux pumps, or modifications to the lipopolysaccharide. This envelope remodelling can substantially affect envelope penetration of antimicrobials. As a consequence, antimicrobials with promising activity under standard in vitro conditions might fail under relevant in vivo conditions because of insufficient penetration or increased expulsion/degradation in that environment.

The student will address this issue by determining in vivo envelope properties with a focus on porins and efflux systems of Salmonella in a mouse typhoid fever model during disease progression as well as during treatment with β lactam and fluoroquinolone antibiotics. We have previously developed a powerful approach for pathogen purification from infected host tissues, and have shown that this approach coupled to mass spectrometry-based proteomics reveals comprehensive quantitative data on pathogen properties in vivo. Current data suggest substantial downregulation of porins OmpC, OmpD, and OmpF during infection, but upregulation of Tsx and FadL as well as upregulation of the efflux pump AcrAB. These proteome changes are expected to dramatically alter differential cell envelope permeability for a wide range of antibiotics. Using a systems biology approach, we will incorporate these accurate quantitative data in a quantitative envelope model that also integrates kinetic data on individual porins/efflux pumps as determined in WP1 and WP2. Through iterative improvements we expect to generate an envelope model that accurately predicts antibiotic penetration under in vivo conditions and identifies potential targets for perturbation and increased antibiotic efficacy.


We set up the highly sensitive and accurate Selective Reaction Monitoring technique by using heavy isotope labeled peptides in order to obtain in vivo absolute quantification of the outer membrane proteome of Salmonella i.e. the copy number per cell. This targeted proteomics technique enables the envelope characterization of different subsets of Salmonella mutants in various environments and will be used as a guide to optimize in vitro conditions that mimic the in vivo Salmonella envelope properties.

The bacterial envelope is an interface to the host that is crucial for virulence and a potential target structure for vaccination. The outer membrane of Salmonella is a dynamic compartment that modulates its composition according to various stresses. However, our knowledge of in vivo quantitative envelope composition remains incomplete as most of the outer membrane proteins are expressed below the quantification threshold of standard shotgun proteomics approaches. To obtain absolute quantification of the outer membrane proteome of Salmonella in vivo, we set up the highly sensitive and accurate Selective Reaction Monitoring proteomics technique. As an example, we can identify a peptide from OmpF with high sensitivity and quantify it accurately using a mixed-in heavy isotope labeled synthetic peptide with the same sequence (Fig. 1A).  This targeted approach enabled the characterization of 68 outer membrane proteins of different Salmonella strains and compare it to WT outer membrane proteins under various in vitro and in vivo conditions. Fig. 1B shows a comparison of WT Salmonella with an rpoE mutant that has a defect in responses to envelope stress. Most proteins are similarly abundant but the simple porin OmpD and the gated porin FhuE are more abundant in the mutant, while the efflux-associated outer membrane channel YohG is less abundant. We currently obtain large-scale data sets for Salmonella subpopulations with different growth rates and stress level from infected mouse tissues.


Fig. 1: Proteome analysis of Salmonella outer membrane proteins. A) Identification and quantification of OmpF using an internal heavy isotope-labeled reference peptide. B) Comparison of outer membrane protein abundance in WT and rpoE Salmonella.

Outlook for second year

We will optimize the proteomics approach to cover all 85 relevant outer membrane proteins. We will enlarge our in vivo data set by including Salmonella samples from antibiotics-treated mice. Finally, the accumulating in vivo envelope composition data will be used as a guide to optimize in vitro conditions that mimic the in vivo Salmonella envelope properties. 


ESR11 (Marseille-AMU)

Sushovan Dam

Supervisor: Jean-Marie Pagès, Aix Marseille University

Co-supervisor : James naismith (STA) 


The second part of this WP focuses on genetic regulation of drug transport. Several specific regulators play a key role in modulating the membrane permeability via the porin/efflux pump expression and contribute to MDR. In addition, various compounds such as salicylate, imipenem or chloramphenicol are able to induce or select the MDR response. This phenomenon has been observed in vitro by adding drugs to bacterial cultures as well as in clinical settings during antibiotic treatment of infected patients. Regulation of membrane permeability directly affects the intracellular accumulation of antibiotics.

In E. coli, the two main general porins OmpF and OmpC have been shown to be post transcriptionally regulated by small interfering RNAs (sRNAs) micF and micC, respectively. In this work, we outline results showing micC induction mechanism(s) — including environmental conditions (growth conditions, chemicals) and regulatory pathways — as well as the effect of micC on porin expression in E. coli. First, we used transcriptional lacZ fusion assays to screen through numerous stress conditions and genetic backgrounds. Then, the porin expression profile was evaluated in selected growth conditions and in various mutants by Western blot analysis. Another aspect of our study is elucidating the common regulatory mechanism between micC induction and the adjacent porin gene ompN. Although OmpN is a quiescent porin, we hypothesize that it may play a key role during bacterial adaptation to stress. AMU currently performs an antibiotic susceptibilities analysis on their collections of clinically important Enterobacteriaceae. Moreover, AMU has recently developed a platform allowing the high throughput determination of the activity of large number of antibacterial agents e.g. last generation ß-lactams, fluoroquinolones, etc on several strains at the same time. We will identify which porins are expressed in resistant and susceptible strains, respectively. The relationships between porins and antibiotics efficiencies will be further explored by a rate-killing approach.29 In such experiments, the activity of antibiotics is measured as a function of incubation-time on an E. coli strain expressing a selected porin. AMU has developed a gene-fusion assay (omp-lacZ) to follow the expression of outer membrane porins in the presence of various chemicals. This genetic approach will be used to investigate the expression of porins and regulators (micC, etc). This monitors the kinetics of porin regulation during external stresses. While previous studies prove the control of membrane permeability in large populations of bacterial cells, they miss the response kinetic and the link in the regulation. Here we characterize the dynamics of gene expression. The analysis of regulators under external stresses allowed us to select appropriate regulators and determine its structure and the effect of mutations or chemical effector on their functional structure. 


Infectious diseases caused by multidrug resistant bacteria are a major concern worldwide. In clinically important Enterobacteriaceae, such as Escherichia coli and Enterobacter aerogenes, the major mechanisms of resistance to β-lactam antibiotics involve (i) enzymatic degradation due to overexpressed β-lactamases and (ii) a decrease in outer membrane permeability. In this regard, alterations of outer membrane porins, conferred by downregulation of porin synthesis and/or porin modifications, restrict the access of antibiotics to their periplasmic targets. In order to design new drugs with enhanced translocation property across the outer membrane, it is important to decipher the molecular mechanisms underlying regulation of porins. In our study we are looking into the post transcriptional and translational regulation of porin expression. Small uncoding regulatory RNAs play a major role in modulating outer membrane porins OmpC and OmpF (and OmpN) and we are interested in elucidating more about their mode of regulation by using genetic tools and outer membrane proteins analyses.

In this study we have standardized the β-galactosidase assay for checking the expression of selected genes. Initially, we created transcriptional lacZ fusions for micC and ompN in the promoterless pFus2K vector (Masi et al., J. Bacteriol., 2005). Further, we evaluated the effect of various growth and stress conditions on the micC and ompN expression in E. coli MC4100. micF-lacZ was obtained from (Mizuno et al., PNAS, 1984) and used as a control. Here, we are representing the gene expression as fold expression by dividing the Miller units obtained at the stressed condition to the unstressed condition (Fig. 1).

Figure 1: Activation of the micC-lacZ, ompN-lacZ and micF-lacZ fusions under various growth conditions in E. coli MC4100 cells.

The conditions that showed multifold induction of both micC and ompN were further taken for OmpF and OmpC expression analysis. To do this, cells were grown in the appropriate conditions and outer membrane fractions were isolated. Subsequently, OmpF and OmpC expression levels were quantified by Western blot analysis with specific antibodies prepared against the denatured form of OmpF and OmpC (Fig. 2) 

As expected we observed a decrease in the OmpC expression in all the conditions where an increase in micC induction was observed (Fig. 2).


Figure. 2: Effect of MicC inducing conditions on OmpC and OmpF expression. Cells were grown in Luria Bertani (LB), Mueller Hinton 2 (MH2), Nutrient broth (NB), NB-Sorbitol 20% (NBS)or in LB supplemented with  1/4th of sub-inhibitory concentration of imipenem (Imip) or biapenem (Biap). Outer membrane fractions were prepared and outer membrane proteins were analyzed by Western blots. (a)  Western blots analysis of OmpC, OmpA and OmpF. Each blot is specified by the respective porin expression and the antibody used for identifying the porins. The d in subscript to the antibody specifies its identification for the denatured form of the porin. (b)(c) OmpF and OmpC expression levels after quantification and normalization with OmpA.


ESR12 (Basel-BVS)

Vincent Trebosc

Supervisor: Dr. Christian Kemmer, BioVersys AG 

Co-supervisor: Dirk Bumann (BAS)


BioVersys follows the approach of inhibiting global or local transcriptional regulators of resistance gene expression. Our TRIC (Transcriptional Regulator Inhibiting Compounds) technology platform allows for the identification of target specific, non-cytotoxic and non-antibacterial small molecules that potentiate the activity of antibiotics. The combinatorial application of BioVersys` TRIC adjuvant compounds with existing antibiotics has been shown to allow for killing of even extensively resistant pathogens at clinically relevant doses of the antibiotic. One of our most advanced projects is specifically focusing on inhibition of efflux-mediated resistance via targeting the transcriptional regulator of the respective efflux-gene resistance cluster.


We designed a novel gene knockout platform, which does not require any antibiotic resistance marker, to be able to do gene deletions in gram negative multidrug resistant clinical isolates of the ESKAPE pathogens. We applied this system and validated the transcriptional regulator AdeR, suggested to be essential in the tigecycline resistance pathway of Acinetobacter baumannii, as a putative drug target. We confirmed the important role of AdeR in conferring tigecycline resistance by upregulation of the efflux pump AdeABC. However, we demonstrated that targeting AdeR is not sufficient to switch-off tigecycline resistance in clinical isolates to rejuvenate the activity of this antibiotic. Our work clearly devalidated AdeR as a potential drug target. We currently use the developed knockout technology to validate several transcriptional regulators involved in diverse antibiotic resistance pathways as putative drug targets.

In order to construct scarless adeR knockout mutants in multidrug resistant clinical isolates of the ESKAPE pathogen Acinetobacter baumannii we developed a novel two-step KO strategy that requires no antibiotic selection markers. State-of-the art gene knockout technics require antibiotic selection markers that may not be applied to clinical MDR-strains, because these pathogens carry a multitude endogenous and acquired resistance pathways. By using this technology we produced ten adeR knockout mutants in A. baumannii MDR isolates of diverse origins and determined the antibiotic profiles of the mutants and their parental strains. The deletion of adeR reduced the MICs of highly tigecycline resistant strains to MICs ≤ 3.1 µg/ml. To understand the underlying cellular antibiotic resistance mechanism we evaluated the expression level of the adeABC efflux pump by quantitative real time PCR (qRT-PCR) and demonstrated that AdeR is required for the expression of the AdeABC efflux pump. We also sequenced the adeRS TCS in highly tigecycline resistant strains and identified non-synonymous mutations that probably conferred the overexpression of the AdeABC efflux pump. 

Our data suggest that overexpression of AdeABC mediated by AdeR is the main tigecycline resistance pathway in A. baumannii clinical isolates. However, we found that there is an alternative AdeR unrelated tigecycline resistance mechanism that confers moderate tigecycline resistance (MIC = 3.1 µg/ml). This alternative pathway was present in 50% of the tested clinical isolates (5 out of 10). Our data lead to the devalidation of AdeR as a target for TRICs in A. baumannii because a deactivation of AdeR by small molecule modulators will not be sufficient to overcome tigecycline resistance in A. baumannii isolates.

In parallel with the knockout technology development we cloned various expression vectors and successfully purified the A. baumannii derived transcription factor AdeR after heterologous expression in Escherichia coli. This protein was used to develop a fluorescence polarization based DNA-binding assay and an enzyme linked immunosorbent assay (ELISA) as part of the TRIC technology platform. We also identified the corresponding DNA binding sequence and proofed specific DNA-interaction with the purified protein.

Our developed gene knockout technology is now routinely applied in MDR isolates to evaluate a selection of transcriptional regulators that are involved in the regulation of diverse antibiotic resistance pathways in different pathogens of the ESKAPE-group. This approach results in a rapid evaluation of putative drug targets that will be, after successful validation, incorporated into the TRIC technology platform of BioVersys.