 Good morning everybody and welcome to the Biaxel Student Webinar. This time we have the Winter School 2020 edition and with us there are three presenters. Artemis Benandi from the Italian Institute of Technology, Roshan Shrestan from T-Burban University that is located in Nepal, and Laura John from University of Oxford. I'm Alessandra from the Italian Institute of Technology that I will host together with Julian this Biaxel webinar and with us there is also Marta from ABI Manchester. So today's presenter will be from different countries. So you can see we have Artemis from the Italian Institute of Technology that will speak about modelic electrostatic interaction and salvation in chromatin and it's going from the single nucleosome to the chromatin fiber. Then he will follow for Russian Shrestan that's from T-Burban University and he will speak about molecular dynamics simulation of nanoparticles in model bilayer, lipid phase separation and membrane protein interaction. And the last will be Laura John from the University of Oxford. And she will still speak about membrane, large-scale membrane movement induced by surface charge reversal. Hello everyone, I am Marta Benandi and I'm going to be discussing my PhD project in modeling electrostatic interactions and salvation in chromatin. Going from the single nucleosome to the chromatin fiber. I am a PhD candidate in physics in the University of Genoa and IET Genoa in Italy. Okay, so what is chromatin and why do we study it? Well, if you were to stretch the DNA found in the nucleus of each of your cells, you would get a two meter long fiber which then has to fit in the approximately six micrometers of diameter of the cell nucleus. So how does this work? The DNA binds itself to some proteins named histones to form nucleosomes. Nucleosomes are spools of 147 base pairs of DNA around eight of these histone proteins and nucleosomes are separated by linker DNA. So chromatin compacts and reinforces the DNA and the topology of compaction to use gene expression with repercussions and in various pathologies. So why do we study chromatin? We do because the nucleosome structure is known as atomic resolution, but the determinants of the topology of the second level of folding of DNA are so unknown. So why elixetics and chromatin? Elixetics is a key determinant of chromatin remodeling. I think it is very important, but mainly because the DNA fiber is very highly charged because of the phosphate atoms found in the DNA backbone. This means that self-propulsion in DNA in packing situations needs to be balanced by histone protein, sensor-related ions. As you can see here on the right, we have an exotic map of the solvent-accused surface of 1KX5, the human, not human, let's say industry standard, nucleosome crystal, the one that is mainly used. As you can see, the DNA in red is negative and the proteins are positively charged or mainly neutral. So elixetic stabilization is achieved through a combination of direct interactions and a long range of exotic and solvent screening in chromatin. But this means that modeling inter-nucleosomal interactions with reductionistic potential omits the explicit role of histone tails of counter ions, nucleosome positioning, various factors that need to be taken into account for an accurate depiction of this interaction. So in our study of elixetics in chromatin, we looked at interactions both between nucleosomes and in nucleosomes. In the latter case, we looked at the importance of the histone tails and we did this by developing a methodology to associate conformational changes with elixetic effects in protein DNA systems. So how does it work? First of all, we performed clustering analysis on molecular dynamics trajectories to extract the representative structures and have an indication and have some contribution of the DNA to the system. Then we calculate the exotic potential and the exotic field on the DNA, on the DNA backbone for each structure and infer the elixetic forces acting on the DNA and then compare them to a protein DNA context map that tells us how many protein atoms are found within a certain threshold from the DNA. We can then correlate the structural characteristics of these structures to elixetic effects such as the role of specific residues or the particular role of interstitial disorder proteins, so terminal region truncation among other features. So, as I said, we looked at the role of the histone tails. These are the intrinsically disordered terminal regions of the histone proteins. Each of the eight histones in the nucleosome has an N terminal and two of them have C terminals. And we were able to look at the dependence on transcription or histone tail positioning. We were able to separate the radial and axolubic forces acting on the two DNA gyres on the nucleosome. And among us, we also looked at the effect of histone tail truncation. When it comes to interactions between nucleosomes, we use Delphi, which is a post involvement solver interface with nanoshaper on structures on nucleosome pairs generated through Python scripts for intermediate distances, or using Hadock for very close distances in order to estimate the interaction energy of these pairs. In this process, we also developed some contributions to the Biopython project. We developed a parser for PQR files and an input-output module for PQR files, which have been integrated in Biopython as of one year approximately. And yeah, what we do is that we begin from the human-nucleosome crystal structure, which is actually, again, it is used for the study of chromatin in humans, but it is actually not human histones. I wanted to clarify that. But yeah, it is the industry standard, as I said before. We use this to sample the possible rotations and translations and therefore forming different nucleosome pairs and record the energy of each possible combination. We also studied the influence of the presence of linker DNA in these conformations and record the energy dependence of different linker DNA configurations. And we are interested in nucleosome relative orientation and not just the different distances between nucleosomes, because different orientations are very important for fiber compaction locally and have global consequences, because, for example, of different fiber-style patterns in chromatin. So how did we choose which structures to study? We just chose them based on the CERC class criterion using nanoshaper to calculate the volume of the structures and then to discard the nucleosome pairs that presented CERC clashes. This, of course, led to less statistics in small distances, which we were able to compensate using the HADUK DOCT poses. But this data is very relevant for type compaction situation, because it doesn't just give us a measure of the interaction energy, but it also tells us which nucleosome positioning configurations are feasible in the fiber, in the chromatin fiber. So then one thing we can see from these calculations is that we can see where the threshold in internucleosome distance is after which we can just reduce nucleosomes to the monopole approximation. And our calculations enable us to create a map of interaction energy between nucleosomes, the data from which can be used to parameterize a functional form to describe nucleosome interaction energies. Here in this plot, you can see some points that look like outliers. They don't follow the general trend of other structures around them. What this is actually the contribution of the linker DNA that I was discussing before, looking at these structures, we saw that the linker DNA adopted particular configurations, which, as we can see from the energetics, are favorable. They lead to favorable nucleosome positions. So what we want to do is to be able to describe electronic interactions in the chromatin fiber in scales that are relevant for chromatin compaction. What we do is that for large distances, we just use the monopole approximation by describing nucleosomes as dielectric spheres immersed in solution. For intermediate distances, we also place a dipole inside the nucleosome, or introducing the orientation dependence. And this has the added benefit that in large distances it falls naturally into the previous case, into the monopole approximation. We can gauge how well we're doing this by comparing with the numerical Poisson-Boltzmann data that I showed before. And finally, to tie everything together, what we want to do is to expand in five interacting centers, one placed in the center of mass of the nucleosome, two more centers in the linker DNA entry and exit sites, which then again provide orientation and on the linker DNA in order to be able to describe electrostatics in chromatin relative orders of magnitude. Besides all this, we also looked at solvation in chromatin. We observed the fact that consistent with previous literature, the nucleosome core particle presents high solvent and dissolved ion accessibility, meaning that as you can see here, we built the solvent to the surface of 1KX5, the industry standard crystal that I mentioned before. And we were able to see the channel traversing the histone core in blue here and the adjacent open cavity. This channel is important because it makes the histone core accessible to solvent and therefore to dissolved ions, enhancing the screening of the charges of the DNA. Furthermore, on the, on the salvation, let's say point of view, we, we measured the set of potential on nucleosomes. The potential is the mean value of electrokinetic potential measured by light scattering techniques. It is widely used in colloid chemistry and in other fields in biology and to our knowledge we are introducing it to the study of chromatin. And we measured the set of potential on single nucleosomes under varying ionic conditions. What we saw, of course, was what we were expecting to see the fact that these other potential becomes less negative as the ionic concentration becomes higher. And we also, we were able to calculate to reproduce these results. Computationally, we were able to calculate the values of the data potential on the estimated position of the slipping plane using the full nonlinear Poisson-Balsman equation. Again, we were able to observe the a similar trend here. So finally I'm just going to leave you with some conclusions. Chromatin is a multi-scale system in both space and time and therefore it's study requires the synergy of multiple modeling paradigms and experimental data. Chromatin electrostatics is very important in chromatin folding because of the high charge on the DNA backbone. And we looked at chromatin electrostatics both through inter-nucleosome interactions and inter-nucleosome interactions. Through interactions in nucleosomes, we developed a methodology to investigate the exotic determinants of protein nucleic acid interactions and applied it to the histone tails where we studied their role in the stabilization of nucleosomes and the tuning of chromatin transcription. And then we studied exotic interactions between nucleosomes and developed a map of interaction energies using energy calculations in nucleosome pairs at different relative distances and rotations. And this gave us estimates of the interaction intensity and the basis for the parameterization of an analytical model. Finally we looked at salvation in chromatin both experimental and computationally and placed all of this in the wider context of custom cross-prand approaches for chromatin. So to conclude, I would like to thank my supervisors in IIT, Prof. Alberto Diaspro, Dr. Vaterokia, Dr. Silvia Dante, and the Anoscopy and Concept Labs in IIT. And of course the organizers of the Biaxel Winter School and this webinar who allowed me to present here and all of you for your attention. Thank you. Thank you, Attemi. And now we move to our second speaker. Thank you, Alessandra. Hi, all. My name is Rosen Schrester. Today I'll be presenting our work on the molecular dynamic simulations of nanoparticles, where I'll be briefly talking about the earlier work being done by our group in terms of lipid phase separation in mixed values due to the presence of nanoparticles and then work on the interaction between membrane protein and nanoparticle. I'll be presenting this work on behalf of Dr. Sangyong Noh from the University of Warwick and Dr. Antoninast from the University of Oxford. I'll be dividing my presentation basically two parts. In the first part, I'll give you a preview on why nanoparticle is important, some work being done within our group in terms of the interaction of nanoparticles with mixed values. And then in the later part, I'll shift the emphasis on our current work where we build on this earlier work in between the striped and hydrophobic nanoparticle with the transmembrane protein. So the aim of our research is to understand whether nanoparticles have any impact on the conformational arrangement of integral transmembrane proteins or not. And to achieve the same, we are particularly looking at the tips. The first one is, will hydrophobic nanoparticle have an impact on the, have any impact on the helical packing of a modal integral transmembrane protein? And will striped nanoparticle have any effect or impact on the helical packing of modal integral transmembrane protein? So as you can see that there are enormous varieties of nanoparticles having lots of functionalities because are being extensively used in different fields. And the field that's being used mostly these days are in the field of biomedical research is a different types of the nanoparticles that have been developed and designed to deliver the drug. For our work, we have used Lian functionalized gold nanoparticles. Basically, we have used gold nanoparticle because of its versatility. We can easily functionalize it in different ways and it can be prepared in range of sizes. And also because of its biocompatibility. Now I will discuss in brief this particular work on the aggregation of striped nanoparticles in mixed phospholipid bilayers. We've found the basis and work on incorporating protein in our models, which I'll talk about shortly. This work is a course where we use martini force field to build a system to look at the interaction of striped nanoparticles with these striped nanoparticles consist of hydrophobic liens on the inter-region and just liens in the exterior region. So, if you can actually look at the course snapshot, then you can see it is lipid x-in state nanoparticle and it is in the unsaturated women against saturated women. There is an increased liquid like disordered state near the nanoparticles. So, basically in this diagram, what you can actually see that this DPC is represented by rate. And this DPC is represented by green and this DPC is represented by blue. So, I want to say that we performed for the multi-nanoparticle system. So, we can see that there is aggregation of nanoparticles within the insaturated women of the bilayers, which I saw during this start animation here. So, in the control simulation with just the DPC, we can see some aggregation behavior of nanoparticles. Now, as we add more complex constituents like when we add cholesterol, there is enhanced aggregation. Whereas, when we further add DPC and DPC, there is local liquid x-ins around the nanoparticles. So, we can actually see the formation of women's around the nanoparticles. Now, the main intuition for us to do our current work is that membranes are complex and cloud-environment. The work that I actually showed you earlier that can actually clearly see that the nanoparticles are not just some static objects, nothing to the membrane environment. Now, nanoparticles are not just an impact on the bilayer when these nanoparticles are embedded on the bilayer. Now, one of the major components of the membrane is membrane proteins. And in our work, we want to see what impact or effect these nanoparticles have on the membrane protein when they are embedded in the bilayers. So, now the reason why we're interested in membrane proteins. The reason why we're interested in membrane proteins is approximately one-third of genes in the human genome encode transplants, and they form more than half of all drug targets. However, it's very difficult to... Getting of these proteins, membrane proteins into their function form, as well as how they misfold into a disease-resistant form. Now, very difficult areas of study, and they are remaining so. In the... You can see that there are different types of membrane proteins. The membrane protein that you all may already know is the integral membrane protein where the subunits are represented by different colors. C is equal to the trans membrane protein and D is equal to the beta-barrel trans membrane protein. Well, the structural stability of the trans membrane domain of membrane protein depends on the inter-helical backing of conserved amino acid motifs, which are basically the specific arrangement of amino acids. Now, these special arrangements, or let's see, specific arrangement of amino acids such as GG4, leucine-zipper, and alanine-zipper, which you can see on this figure on the left, are a crystal role, are understanding for this trans membrane protein. The model trans membrane protein domain that we are using belongs to the glycophorin-8 integral trans membrane protein. Now, it is a basically homotimer consisting of two identical peptides. Now, their stability as a homotimer is facilitated by these GG4 motifs. Now, basically, in this GG4 motif, you can actually see that this glycine residue repeats after the fourth position. Now, what does this GG4 motif does is that they can maximize the interfacial, interfacial-pander-walls interaction and or hydrogen bonding by allowing the inter-helices to be in proximity, that is, in close with each other. Now, what we want to see is if there is any disruption to the interaction between these GG4 motifs in the presence of nanoparticle or not. We want to know that this GG4 motif is a crucial role in the folding of the membrane proteins. Now, in our work, we are looking at basically two different types of nanoparticles. One is hydrophobic nanoparticle, another one is strap nanoparticle. Now, and we are actually looking at the interaction between these nanoparticles with the trans membrane domain of glycophorin-8. We thought about how we can look for changes in the structure of protein, and that's why we are performing these case and control simulations. Now, in control, we are not expecting to see any changes, where we have the intermolecular distance between the nanoparticle and protein separated by more than 9 nanometers. Whereas, in the case simulation, the nanoparticle is placed next to the protein, and we expect the interaction between these nanoparticle and the membrane protein. Now, I'll be talking about the model later as well. Now, we have actually used the DTPPC forcefully by leading here, so we can see on the figure on the right that the trans membrane proteins and the nanoparticle are either placed next to each other for both the stripe and the hydrophobic nanoparticle for the hydrophobic nanoparticle, and they are placed for the distance around the 9 nanometer when we are doing control simulations. Now, if you look closely at the nanoparticle structure, then the nanoparticle core consists of inner and outer gull atoms, and the thylator gull atoms, which are the base of the ligands, and the structures come from actually derived functionalized nanoparticles. These actually consist of aromatic functional groups, while the polar group consists of hydroxyl anomide groups. Now, the ligand sites themselves are attached onto AUS motifs, each with slightly different angles and connectivity parameters depending on the shape of the EU core itself. Now, the ligand themselves can be through PDB formats of the desired ligands. Now, if you look at the structure of the glycophore, then we build the system in the ember force field where we used the TIPTP water model, and the DDPC lipids were used, and again, I have already told you that we used that for the trans-membrane to a homodimer, a domain that is basically a homodimer consisting of two identical peptides. Now, the GZ4 motifs is colored in yellow. Now, it mediates this trans-membrane to a man dimmer. Now, the GZ is basically a right-handed dimmer with a crossing angle of approximately 30 degrees. Now, the analysis that we actually did, there are different analysis that we actually did, but we do principal component analysis can be used to reduce or let's say simplify large and complicated sets of data. In the PC analysis, the greatest degree of variance can be explained by the initial few principal components, here the projection of the distribution onto the space defined by these principal components result in a low-dimensional representation of the data set. Now, basically, we can actually understand or find looking for some structural variance of our system, of our protein for now. So, we are using PC to discover do the atomic coordinates cluster into unit stress or not. Now, in the plots on the left, in the plots on the left, the red one shows the PC1 as a function of time and the black one shows the PC2 as a function of time. Now, clearly, there are conformational states indicated by small amount of... So, sorry. So, basically, there are, clearly, there are conformational states being indicated by very small amount of structural variances, which are visited the most amount of time during the explanation. We can actually see it through the contour plots on the right. Now, here, these arrow indicates the confirmation sampling as time passes. Now, before we look at any structural work, we will see that there is variance in the protein, atomic coordinates, and we can see that the protein is adopting different confirmations. Right? So, basically, basically, the protein can be, like, in this state, in most of the time, in this state, in this state, and in this state. We can see similar things in other kind of dissimilar work as well. Now, we did the distance measurement where you basically calculate the distance between the center of geometry of the nanoparticle and protein when they are placed next to each other. For both the hydrophobic and strapped nanoparticle. Right? So, here, in the case strapped nanoparticle, what you can actually see that this strapped nanoparticle moved away from the protein. So, here in the strapped nanoparticle, we are good nanoparticle, you can actually see that this strapped nanoparticle moved away from the transmittable protein at around 5.8 nanometer, and then it's just to this average distance of 5 nanometer. However, the hydrophobic nanoparticle remains within this interaction distance of the interaction distance with transmittable protein, and we can actually see the increase in the crossing angle for the hydrophobic nanoparticle. Now, another work that we did was what we're trying to understand is like, is there any change in the secondary structure of the transmittant dimer or not due to the presence of the nanoparticle or not. Now, we looked at the helical secondary structure of the protein as a function of time, and we started to compare the hydrophobic to either versus hydrophobic apart. That is our case and control. And we did it for same for the strapped nanoparticle as well. So we basically calculated the helical content in the protein for both the hydrophobic as well as the strapped in both cases. That means when they are apart and when they are together. Now, from the plot on the right, we found that we could not account for any change in the helical structure of the protein due to the presence of nanoparticle. So basically, there is no change in the helical content, right? But it's not just the alpha helical content that we looked after. We actually looked after other structures as well. So we can actually see from these structures, the second structures that we analyzed as we did from the VMD. From the VMD. So the helical content was extracted from the VMD secondary structure and like calculation. And basically we can see that there is not much of a change in the helical content for the cases when they were either apart or when they were together because they had for big nanoparticle and for the strapped nanoparticle as well. Now, now the thing that we actually did was well, Russian sorry, there is one minutes left. Thank you. Thank you. So well, the structure stability of the transfer domain of the main protein depends on the inter-helical backing of conserved amino acid motifs. Now, what we need to understand is this that we also calculated the distance between the center of mass of the protein and the of these two peptides for both the when the transfer domain and the nanoparticle were together, that is for the hydrophobic as well as for the strapped when they were together and also for when they were apart. Now, we can actually see that there is that that the effect the GC4 packing interaction at all, right? So, so we can't account for any change in the change in the GC4 packing interaction due to the presence of the nanoparticle. So, so basically what we did that we have performed initial to me all the time simulation of of these of the look complexity. We'll nanoparticle in the presence of model transfer and protein diameter model by here. Now, now we know now we can actually say that this four is known to stabilize this glycophorinate interactions. That is neither one nanoparticle effects GC4 packing. Now, however, from the distance calculation that we did that the strapped nanoparticle will nanoparticle moves away from protein, where is the hydrophobic will nanoparticle continues to interact with the protein and the protein's pressing angle is adjusted. However, we want to do is what the more structural analysis modifications want to see if the modification of will nanoparticle legals will account for changes in the protein structure or not that we actually did from the CLS. So, this wonderful opportunity to present our work and thank you all. Thank you. Thank you. And now we move to our to Laura. That is our last talk of today. Yeah, so hi, my name is Laura. I'm a PhD student in the lab of Professor Mark Sansom. And in my project, I'm looking at the interactions between membranes and proteins. And for this, I use a combination of Newton Reflectrometry measurements and molecular dynamic simulations. And today I want to present you some very interesting effects I saw in my experimental model membrane system and how I use simulations to explain these effects. So that's how my experimental system looks like. You have a surface and on top of that surface, you have a gold layer. And on top of that gold layer, you self assemble a self assembling mono layer. Then you add your lipids in form of vesicles. And via a procedure we developed, you gain a lipid by layer shown in blue here. And crucial is now that the lipid by layer and the self assembling mono layer are not in contact. There's a water layer in between. So the lipid by layer is basically free floating. That's why I call the system also free floating membrane system. Then you can add, for example, protein to your system and measure everything with nutrients. Why nutrients? Well, nutrients have a few advantages. They are deeply penetrating the system without destroying anything. So you get information about your membrane and your protein without destroying your protein. Then you can use physiological conditions. And you can change the environment during your measuring. So you can first measure the by layer, then you can add protein and measuring the whole time. You measure basically the affected beam. And the information you get is something similar to a density profile from an MB simulation along the X axis. So you can get information, for example, about how deeply the protein penetrates the membrane. How far the membrane is away from the surface. What's the state of the membrane? So how rough or how curved is the membrane? But you can also get information about, for example, lipid distribution within the membrane or how dense the protein is bound to the membrane surface. And you can use this information in form of constraints for your simulations. And in turn, you can use your simulations to refine your fitting analysis process from the neutron data. So both techniques can feed each other for gaining structure and functional information of your system. So in my case, before I could actually add a protein to my system, I wanted to see how different soil concentrations and buffer concentrations influence the membrane behavior in my sample system. So I started measuring the system in the presence of two millimolar calcium, because that's also the soil condition you deposit your membrane. And I had a distance between the same surface and the bilayer of around 12 angstrom. When I removed the calcium, the membrane went basically crazy. It went far away and it became a very rough state. We are talking about a distance from around 200 to 600 angstrom from the surface. When I measured the system in the presence of 200 millimolar sodium, we ended up at an intermediate state. So the membrane came partially back to the surface, but the distance is still around 4200 angstrom. When I re-added calcium, the system, the membrane came fully back to its initial position. This process is repeatable and reversible. And it's looking like this in your neutron data. What you see here is a so-called scattering length density profile you gain when you fit your reflectivity data. On the x-axis, you see the distance from the surface. In my case, from the gold surface. In yellow here, that's basically where the gold sits. And the brown layer here is where your SAM is. In the SLD profile, in my experiments, the first tip is basically always belonging to the SAM. The second tip is always belonging to the membrane. And in the different colors, you see the different salt conditions I used, and you see it for different bilayers. So you can see here the different lipid compositions. In case of calcium present in the system, the membrane was always close to the SAM. As you can see here in orange, when I remove the cut ions, the membrane went far away and was very rough. Now you can see really the roughness on this really broad distribution. And when you measure the system in the presence of sodium, you had this intermediate state shown in green here for all the lipid bilayers. So that was very interesting. And this effect you saw like this movement was really, really large compared to a membrane movement. And I just wanted to know, obviously, what's going on in there. So where do the cut ions interact? What's the difference between sodium and calcium? And what's the driving force behind these effects? And to solve these questions, I use molecular simulations. I started with a coarse-grained setup of my SAM. So just the SAM in a simulation box. And you can see here the mapping between coarse-grained and all atomistic. And that's monomolecule from SAM. The SAM is also called OEG-SAM because it has this oligoethylene glycol groups here shown in orange. At the bottom, you see the hexagonal grid along which I aligned my molecules of the SAM before and after the simulation. I use the coarse-grained system to validate my SAM. So I looked if it behaved as it should be compared to experimental results in terms of thickness, for example, distance between molecules and this angle, for example. And then I converted this coarse-grained structure to an all-atomistic. The force field here was Martini 3. And now I moved on to CHAM 36. I simulated my SAM in all-atomistic in presence of sodium and in presence of calcium to gain the energy profile. You can see it for calcium in red and for sodium in black. In case you didn't see such energy profiles before, what it basically says when it's getting deeper here, you're well. It means something binds stronger. And you can see that the depth is very similar. So both ions seem to bind similarly strong. However, the sodium seems to bind more diffused because you see the distribution is much broader compared to calcium. Well, now I wanted to compare these energy, binding energies, to sodium and calcium when it binds to different lipid bilayers. So I simulated all-atomistic lipid bilayers in presence of calcium at the top and in presence of sodium at the bottom. And here you see the different lipids I use basically. When we start with calcium and compare it to the red curve for the SAM, we can see that calcium binds definitely stronger to the SAM, but it also binds to the different bilayers because we always have here a well basically below zero. In case of sodium, it's different. We see only ion binding to the membrane when the highly negative discharge clip three lipid was included in the membrane. With POPC and POPC-POS, we didn't see sodium binding to the membrane. So again here, sodium seems to prefer to intact the SAM rather than with bilayer. And similar to the SAM, we saw also here that the sodium binding was, when there was binding, more diffuse. The distribution is here much broader compared to the calcium distribution. So what can we conclude from this? Well, both ions seem to interact more with the SAM. However, calcium also interacts with the bilayer. And it will be the calcium interaction with the SAM will switch the negative recharge from the SAM to the positive and will attract the membrane. Sodium instead will also interact with the SAM. However, it has lower charge and it has more diffuse binding to the SAM and it has, in most of the case, no binding to bilayer. So there will be less attraction and less bridging between these two layers and leading to this intermediate position. I should mention that both layers, the membranes and the SAM are negatively charged. So there's actually like a repulsion force between them. So if you have this ions in between, which change charge of the SAM, you change this repulsion force. So why, when we say, okay, there's now an attraction, why are they not in contact these two layers in the presence of calcium? Well, we can really nicely see this when we go to the Corspan multi-layer system I simulated. So now I went back to Corspan and simulated SAM and bilayer in the same simulation box. Hereby, the bilayer had all the time a hole throughout which the ions and water could equilibrate during the simulation. And then I just determined basically after one microsecond the density profiles of water shown in blue and of the hot ions shown in black for calcium on the right side and for sodium on the left side. And when you look at these blue curves, you can see that there seems to be water layers. And that's known about this SAM, OEG SAM, these oligoethylene glyphosams, that they form the structured water layers on top. So it was really nice to actually see these water layers in the simulations as well. And we saw it in the all atomistic as well. And this causes a repulsion hydration repulsion force. So although there's an attraction, the water prevents from the prevents the bilayer to come in too close contact to the SAM. So you have always this water layer in between. With this, I want to conclude and answer my previously asked questions. So where do the captions interact? They will preferably interact with SAM. Although the calcium will also interact with bilayer. What's the difference between calcium and SAM? Calcium binds stronger, especially to the membrane and will probably bridge between these layers. Sodium has a weaker binding, especially to the membrane and will have a less bridging effect, especially due to its more diffuse binding. What's the driving force behind these effects? The driving force is that you reduce the repulsion between the two layers because they are not fully negatively charged anymore. And the SAM surface will change its charge basically due to cation binding. Yeah, that's a nice summary. But what I really want you to take home from this short talk is that you have heard from this free-floating bilayer model membrane system. Because it's really nice, this system actually, because it has on both sides of your membrane biologically relevant water layer, which you don't have in the other planar model membrane systems so far. And you can actually tune this water layer by just changing the salt concentration in your system at physiological conditions. And the other point is what I want you to remember is that neutronary spectrometry and molecular dynamic simulations are working well together and can feed each other by gaining new insights in membrane behavior, especially with the system and also in protein membrane interactions. With this, I want to thank BioXL for this really nice winter school we had and also for giving me the opportunity here to talk. And I want to thank my two supervisors, Mark and Luke, my research group, and my funding bodies, BDB, SSC, Oxford, and BioScience, DTP, and ISIS, Neutron, and Moonsource. And last but not least, I also want to thank you for listening and I'm happy to answer questions. The first question is, please could you describe in more detail the plot of Coulombic versus reaction field energy? And please could you give some more details on how the dimers or pairs of nucleosomes were picked? Okay, so, yeah, basically what we did is that we took the... we took two identical structures, two identical 1KX5 plus linker DNA, and we also did this with 3AFA, which is another nucleosome crystal which doesn't have histone tails because we wanted to check out how histone tails change the interactions and those results are in my thesis. But yeah, basically what we did is that we picked the nucleosome crystal that we were interested in, be that 1KX5, 1KX5 plus linker DNA, 3FA, whatever, then we kept one nucleosome fixed, first of all, we superimposed them, then we kept one nucleosome fixed and rotated the other one, and then we translated the rotated nucleosome. This was done with biopython through custom scripts that I wrote and yeah, then basically we fed those structures to Delphi and in that process, we first used nanoshaper to discard the structures, the pairs that presented steric clashes. Basically what we did is that we calculated the volume of each pair and if it was equal or smaller than the volume of two separate nucleosomes, we discarded that structure because that meant that we had a steric clash there and yeah, that's the criterion we used and then we just measured using Delphi, the columbic interaction energy and the reactant fiend energy of the pairs. So that's the plot you saw, the scatter plot that you saw in which we had the columbic interaction energy that as I said from a point on started following the monopole approximation which means that rotations are no longer relevant and each nucleosome sees the other as a point charge and in the other case we had the reaction field energy which is basically the interaction energy because of the fact that the nucleosomes are dielectrics and yeah, I don't know if that's too detailed or not detailed enough, please let me know, that's pretty much it. Great, thank you very much for the answer. The next question we have is for Laura and the question is what kind of softwares can be used for these sorts of MD simulations? The MD simulations were done with Gromax, just standard Gromax version and the analysis was either with Gromax or VMD. I hope this answers the question, I'm not sure. I think it does, thank you very much. The next question we have is for Artemis again and the question is asking for more information about the zeta potential data. Namely how difficult is it to do these measurements and how reproducible are the simulations? And the question ends with by asking is this data published anywhere? Is there a place where people could go to read up about the work you've done? Yeah, thanks for that question. So okay, full disclosure about the experiment, the zeta potential experiment, it was my experimental supervisor, Dr. Silvia Dante in IIT who actually did the experiment. I can't do it, I just did the simulations and I used the data. Okay, let's start from the easy part of the question. This data is not yet published because it's in a manuscript that we are going to, that we're going through right now. So basically the first part of my talk has already been published, the part on the histone tails and protein DNA interactions. And then the second part of my talk is going to be published. So it's not yet public. I can send you a draft of my thesis where you can find more information and you can contact me privately if you need that. And yeah, okay. So about the reproducibility of the experiments, yeah, the reproducible, they were not particularly challenging to do as measurements, at least from what I know, but then I was not the person actually in the room doing measurement. But yeah, we used, we bought nucleosomes and we just performed the measurements on those. Yeah, and then the calculated values, they are absolutely reproducible. The only slight issue, let's say, is that the position of the slipping plane is estimated. It's estimated as a function of the ionic strength taking as the reference the physiological ionic strength and the bilength. So that's why if you look at the plot that I showed in my presentation, you can see that there's some discrepancies. Basically, we were a bit underestimating the value of this potential, but in that we believe that this is due to the position of the estimate of the position of the slipping plane. But yeah, the results are reproducible and the experiment is not particularly, let's say, costly from the complication point of view. Let's put it that. If you need any more information, please don't hesitate to contact me. So yeah, thanks. Great. Thank you very much for that in-depth answer. The final question that we have for today is from Davide. The question is for Zora and the question is, is neutral reflectometry enough, sensitive enough to study transient protein-medal interactions? It is. So I'm not entirely sure what you mean by transient, but I guess you mean when the protein goes away again from the membrane. So I just measure proteins which bind to the membrane and even like I did just experiments with pH domains, so not very large proteins but it's sensitive enough to see the binding of these protein domains basically. And there are a lot of different tricks to make it more sensitive, so you can, for example, deuterate your protein, which gives you then in neutral reflectometry a different contrast and to see even smaller things. So it's very sensitive, but sometimes it's quite a lot of work to make it more sensitive. But like with a standard neutron experiment, I was able to see a pH domain binding to my bilayer. But what you see is an average, so you can't see like just, I don't know, you can't observe one protein, you add like several pH domains, right? And then you see an average, so you can't observe like a transient binding event like in simulations. Thank you very much for that answer and I'd like to take this opportunity to thank all three of our speakers again for the great presentations that they gave. There's a number of people saying in the questions chat, thank you for the answers, thank you for the wonderful talks, et cetera, et cetera. They were very good.