 The next coming in our case study lecture is named selectivity of binding of vector molecules to the fallout receptor alpha observed by molecular dynamics, and it will be presented from professor Anelay Ivanova from the physical chemistry department, faculty of chemistry, pharmacy, University. So I'll ask once again for your attention this time to professor Ivanova. Let me first start by saying that this is the team from our lab, the colleagues together with whom we have been working on these projects during the last five years already almost. And I would like to start the talk by saying that we did this computational study in the framework of drug delivery. As we have chosen, as our model drug, the same actually chemical therapeutic, as we was talking about through within. The reason to select a chemical therapeutic for the study was that, as we all know, chemotherapy is the most usual treatment of cancer, but together with the positive effects it has also some very unpleasant I would say severe side effects, which are listed here. One of the main problems with chemo therapeutics is that these aggressive substances, aggressive to our body, they kill the cancer cells, but they also affect in an undesired way the health issues. One of the strategies to alleviate this undesired side effect is to create their delivery systems which are based on the so called active targeting. Active targeting in principle means that there are some additional, let's say component together with the pharmaceutical that are able to steer it primarily to the tumor cells to the neoplastic cells. And in this way, reduce the side effects by not affecting the healthy issues. There are different strategies to achieve this. They are listed here, but I will focus on the implementation of small organic molecules, which are able to if bound to the drug, they are able to steer it towards a particular part of the cell membranes. And then targeting usually a membrane embedded receptor, and then sometimes in our case together with part of the membrane using endocytosis, the drug is internalized into the side role of the neoplastic cell cells. This is the pair of proteins that we selected to test whether we are able to describe computationally this targeting affinity of the protein. The ligand is the well known folic acid, which in the human body and physiological conditions is doubly charged, and the anion is usually called folate. The naked receptor of this ligand in the human organism is the so called well it has several maybe receptors but one of them is the so called folate receptor alpha which is the crystal structure which is depicted here on the screen. This is the primary transporter of this ligand across the cell membrane. And we decided that it's a good pair to study not only for several reasons, one of them is that this receptor is over expressed mostly on the surface of cancer cells. Of course, this present also on the normal tissues but in very, very much lower quantity than on the cancer cells. Also, it is available, not only on certain type of cancer cells but on many types of cancer cells. And it has been reported that the over expression can reach from 100 to even one million times compared to the healthy cells. Another advantage of this pair is the high affinity of the ligand for this receptor. Here you can see that the association constant is small. Also, the sector is rather small which from computational perspective is an advantage. It has well defined crystal structure which was published several years ago, and which we used as the basis for our models. And also folate the natural ligand it's simple. It's organic compound with, let's say standard functional groups in it, and it's simple and abundant also experimentally. This was the reason to choose this pair. And what were our goals. First, we decided to describe the molecular structure of a series of folate like vector ligands in Salin, only the folate but also some of its derivatives which could have the potential of steering some drugs to the folate receptor alpha. The second goal of the study was to construct a model cancer cell membrane with an embedded receptor and to simulate its dynamics so these were the two preparatory steps for the actual part of the study. Then we monitored by molecular dynamics, I think of these set of ligands to the folate receptor alpha and characterize their interactions and try to rationalize the selectivity of binding of these ligands to the particular protein receptor. And finally, we tried to assess the effect. We tried to connect carbon to the targeting ability of these vector molecules. And all the stages of the study were carried classical or to me stick molecular dynamics in relations. So let's discuss about the models. Here is the set of ligands that we tried to, that we decided to investigate. First, of course, is the folate, folic acid in the W ionized state. So, the other four ligands from this set, they are all the anions so negatively charged. This is an abrogation to a Routy traxate and metal traxate. They are available synthetically and they are used as chemo therapeutic on their own. This ligand here, which we abbreviated MTHF is another form of folic acid, which is slightly different but actually it is more bio available inside the human body than the folate which we intake with the food. This ligand is a bit different. Okay, in red, you can see everywhere on the molecular structure. What are the differences compared to the initial folic acid molecule. So this bond here, which is a therial ornithin molecule is different than the others in the sense the only three therionic molecule in the set. It's also synthetically available and was suggested by a group of researchers as an inhibitor of one of the other folate receptors. So you see the compounds are somehow related. But they also have differences which might affect their behavior toward their receptor. The model membrane just a couple of words about it. We tried to construct this model, mostly based on experimental data or I would say as much as possible based on experimental data. That's why our model consisted of 370 molecules distributed asymmetrically both in terms in terms of number and in terms of charge, which is typical for the human cell and race. And the bilayer consisted of five different types of lipids, which were obtained by combining these lipid head groups, the standard for sort of your committees and so on we also included higher amount of cholesterol, which is known for cancer cells and so I'm not going to go into detail here just I would like to mention that we have several kinds of lipids that are negatively charged, and all of them are situated in the inner liquid of this model membrane. And then we have here the protein receptor which is embedded into the outer leaflet of the model membrane by because you also deal in a total anchor by a GP anchor that is what happens also in experiment according to the literature sources. And you see that. Yeah, something else that might be important for the membrane model. The parts of the lipids were not only saturated alkane residues, but they, some of them contained also more non saturated. It means one double bond or with multiple bottom double bonds along the carbon chain. The composition is practically taken from these two experimental studies. And finally the protein. This is the crystal structure again. This is the BB ID which we used for this protein molecule it consists of 204 amino acids and the overall charge is plus four. Again, the construction of the protein bound to the GP anchor was based on experimental data. And this GP anchor serves to immobilize the receptor on the surface of the cell membrane that's what we did also in the model where this receptor is usually found in the so called liquid crafts which are rich in transport proteins and different proteins. In this case, there were experimental measurements that it is usually found either as a monomer or as dimer so we used a single receptor molecule anchored into the model membrane. And the overall charge of the GP anchor is minus one so you see we have a lot of charges in addition to that. And this is how the four different types of models so that we study to look like the first one is contains one legumes on each of the series of six immersed in saline and we maintained their physiological concentration of sodium chloride also in all of the models. The second is immersed in a cubic periodic box and simulated with order on the boundary conditions in the three dimensions of space. This is the first type of models, the second type contains a membrane with one molecule of unembedded protein receptor immersed in water and sodium chloride. And of course in a larger box here we imposed hexagonal periodic boundary conditions because this is the normal packing type of the little thing in the membrane. In the third model into this we also put a ligand in the system, three ligands on bounds and let it move the model system and observed whether and how it will interact with the protein receptor or with the membrane or with the components model. And finally, we also test the performance simulations, almost the same model, with the exception of the three ligand was not any more free, but we attached a peptide complex, a drug peptide complex to it to test how to attach the drug carbon. Just in short, the simulation protocol of mystic molecular simulations, they are all followed by some additional DFT calculations to assess the binding energy between the protein and the ligand. We used the charm 36 to describe the protein, the sugars and the lipids. And we use them and for the ions and we use CGN effects for the ligand molecules and for the drug peptide for the drug and for the peptide of course charm 36 again. And water was deep, very standard protocol for the MD simulations, with the exception that we maintain some constant surface pressure on the models with the membrane corresponding to the surface pressure that is measured for human living cells and the temperature was physiological and the pressure was ambient. Here are some details about the MD simulation, I will not go into detail here if you have questions we can discuss later. Just the trajectory lens did 150 nanoseconds for the three ligands, one microsecond for the receptor embedded membrane, and from 20 to 400 or 500 nanoseconds for the systems containing those ligands. I would like to mention that we also performed four independent trajectories for the three ligands and two independent trajectories for the drug peptide complexes models. Final note on the protocol we saved snapshots into the trajectories every 10 seconds. Okay. So the preparatory steps we wanted to learn more about the ligands and about the membrane themselves without when they are not together. I will tell you here that the whole series of ligands turned out to be quite flexible. The state had definitely judging the square deviation of the atoms. For late during the last 15 nanoseconds of the MD simulation, but we can see that it definitely preferred conformations and interchanges between them quite fast. The main thing is true about PMPTX about bone and the other three ligands they have more or less one preferred conformation, but nevertheless, all of them are very mobile. Here is some summary of the cluster analysis, and they have at least two preferred structures which are populated in different share throughout the trajectories. So that this flexibility might be important for the binding to the receptor. Concerning the membrane and the lipid bilayer as part of this membrane model. It is more or less different to try whether we have obtained the correct structure of the lipid bilayer, it seems to be the case of the thickness of the bilayer experimental bounds. It is also known that in the increased presence of cholesterol, as is our case, the lipid bilayers are usually in the so-called liquid ordered state, and this analysis corresponds exactly to such a liquid ordered phase state of the membrane. We can see also for most of the lipids, we have areas per lipid molecule, occupied on the surface of the bilayer that are very, very close to the experimental data. The order of the lipid tails is an essential determinant of the phase state of the lipid bilayer, and as we can see from these order parameters of the lipid tails. They such tails are the most ordered one. The more unsaturated are the least ordered one, so the closer to zero these values are the disorder, so the closer to a liquid state the membrane is. And finally the poorly unsaturated fatty acids, of course, they are known to be the most disordered ones, once which is reproduced by our results. So the overall picture of this lipid tail order parameters corresponds exactly to a liquid ordered state. So we assume from this analysis that our model is correct. The results of the protein, how do we behave during the MD simulations, some extracts of the results about the protein structure at half of the trajectory at about 550 nanoseconds, this is the X-ray structure. So even visually you can see that the secondary structure is quite well preserved. Here is the numerical comparison here. The structure from the molecular simulations is very close to the experimental one. There is some unfolding of the helical regions and some slight changes in the secondary structure. But when we did a more careful analysis of the data, it turned out that this partial, let's say unfolding of the protein is caused by interactions of its Germany with the molecules from the membrane. Since the experiment was done in the absence of a membrane for the protein as it is without a membrane. So we believe that this is okay as a structure of the protein. And here just to have an impression about the flexibility of the bite, which is in this part of the protein of the ligand binding site. We have estimated the fluctuations of the volume of this binding pocket during a piece of the trajectory. And this is the experimental value, this here. So you can see that the pocket fluctuates, it's relatively mobile but the volume predicted by the simulation is very close to the experimental value from the X-ray structure. So we were relatively sure that our model was fine and then we studied how the ligands behave in the presence of this protein, just a second. I think I need the other way to be able to start. So I'm going to show you how each of the six ligands bind or interacts with the receptor. Before showing the movie, I would like to underline once more that there was no bias in our end situations. So we just monitored the natural dynamics of the... Here is the folate and here in this movie we have tried to capture the moment when it binded to the active site territory. So these residues here in gray, they represent the ligand binding site of the protein and folate did not have doubts where to bind. It went directly to the pocket and bound there. And as you can see from the minimum distance and the protein, the ligands remain there till the end of the simulation. Here is a... happened in the four independent trajectories well for folate the same in four out of four cases it went to the pocket and bound. And in most of the trajectories in three of them, it stayed there just in the fourth trajectory it was more mobile and also... I'm close to the base of the protein somewhere here. Here is an illustration of what RTX did. Okay, let me play the movie. Well, the same. It found the active site of the protein and bound there. Pay attention now it will rearrange for some time and then assume it's just even going away for a while. And this is the right post of the correct post of binding into the active site of the protein, which was not the same by the way for the... And here is the plot showing the binding time of the ligand and that it remains in the active site until the end of the simulation. This is a summary of the behavior of RTX. Yeah, in two out of four times it was able to identify it and stay there, bind there. But in the other two, so in 50% of our fourth trajectories, it found an allosteric site on the... This allosteric site at least to our knowledge was not reported for the folate receptor alpha and it is here just opposite the binding. This one here in dark gray and RTX is immobilized there by the amount of clasts of the protein and it's also very stable binding poles for this ligand. Okay, oh, sorry. Sorry, sorry, sorry. I'm going to share it again for this interaction. But yeah, even from this small number of trajectories, we were able, I hope you see the screen again. Yes, it's visible again. Okay. We had a good explanation of the relative affinities of the two ligands measured experimentally. It is known that RTX has about 60% for this receptor compared to folate. Of course we cannot be quantitative but qualitatively we were able to explain the distance due to the binding. Okay, so what made these ligands bind at this spot of the protein? We did analysis of the ligand protein ligand interaction in molecular dynamics trajectories and they showed that it's a complex of interactions which were reported also for the crystal structure. So we were able to detect some by stacking of the aromatic parts of the ligands and some kind of gen bonds. Just be aware that these are very persistent kind of gen bonds because we were extracting snapshots at very long intervals of 10 seconds from our trajectories and for RTX the difference is that the most intensive stacking is between the so-called tearing part of the ligand and some aromatic amino acids from the pocket. While folate, the most intensive stacking is between the middle part of the ligand, the glutamate, and also the charged part of the ligand, and the middle part with some other amino acid just from the pocket. So they really have two different binding modes, these two ligands. To rationalize this difference in the binding, we calculated with the DFT with dispersion corrections presented to snapshots from the molecular dynamics trajectories taking the structures from there and extracting just the ligand together with its very closely lying amino acids. So we calculated binding energies, so which are quite strong with you can see, of course, this is due to the very intensive electrostatic system. Again, when having a set of binding energies, we use the components analysis to correlate these binding energy values, here they are scaled down for the purpose of the statistics. So we saw from the molecular dynamics trajectories that might be important for the binding of the ligand to the number and length of hydrogen bonds, number of contacts, so as a measure of the van der Waalsen corrections, number of stacking partners, the number of electrostatic partners, number of closely lying amino acids. And here I have highlighted the most important coefficients of the first three principle components which were able to explain more than 80% of the binding energy that the two ligands have different, slightly different mechanisms of binding to the protein. While the whole eight relies on van der Waalsen correction on piloting bonds and on analytics, but they can be somehow compensated for each other because they are found in the three different principle components. These are RTX, almost all of the important interactions are concentrated in the first principle component. These are again the number of piloting bonds, dispersion, but acting together at the same time to keep this RTX bound to the pocket. And they are complemented taking and by the closely lying amino acids to get a more visual impression on the, on the mode of binding here I have extracted some sensitive interaction maps, and you see the amino acids from the pocket which contribute to the interaction with the ligand by van der Waalsen mostly. And those are other amino acids which contribute with quite strong some of them. So it's a really collective interaction of the ligand with several amino acids at the same time to accomplish this to the active site of the protein. And by the way, most of these amino acids are the same, which were outlined in the crystallographic of the for later protein complex. So we were able, however, to identify a new amino acid this article in 61, not listed in the crystal structure interaction map, but we saw in the molecular dynamics that it is very important for initially attracting the ligand to the binding pocket. Okay, what happens with the other ligand here is ptx. Contrary to what we might expect the similar chemical structure did not behave in the same way it near the pocket but it never went into it and didn't bind in the cavity. So in summary, it was very much attractive overall to the surface of the protein. As you can see from the distance it's really sticky to the protein, but it goes all over the place. Either to the other side, the same, which was registered for a text or to the surface, whatever it likes and scan virtually the surface of the project. Some hydrated form of the folate. It examples the folate itself to some extent, because it was stable in three out of the four to check to restore identify the point there. As you can see also here that it's coming close to the pocket, but as you can see also from the fluctuations of the minimum distance, the binder as stable as that of the folate. So we attribute this by analyzing the data to the larger volume of distance, which led to the inability of this reduced ligand to fit well into the charity of the protein will be part of, or to take part in all of the necessary interactions. The key is to ligands were still able to identify the protein and beats. Unlike. Okay, what were the respective functions. Yeah, much weaker by stacking also for empty chiefly registered some surface activity. It was attracted from time to time by the membrane, which also prevented stable binding to them to the pocket of the protein. And as you hear, wonderful interactions are much more expressed there, but that's most of it. It is not able to form such strong hydrogen bonds as for late in its non reduced form. So for PTX here in Brown, you can see all the amino acids that interact in the four different trajectories that we generated, and you can see also on interaction map and on the PC a insult that was stabilized by a non specific of underval center interactions. So this ligand behaving somewhat different way than the first two but okay they still bind to the protein, irrespective of the nature of the binding. So this is to the final two and metal tech state and aerial or anything. As you will see from the movie for mpx he doesn't like the protein. It works much more to bear into the membrane or to stay in the solution. So this is also evidence from the table. Why this is happening because the mpx has the smallest structural difference than for late but it turned out that this is enough to make it a special. And finally the last thing that the period or meeting well it did bind to the protein only in the first trajectory for a very short or relatively short time at the end, and you see very far from the pocket. None of the other three trajectories. The state goes to the protein for a measurable amount of time. Well, of course this is really explainable. This was the only Twitter I am in the series so this was an evidence that the double negative charge for these ligands to be able to bind to this receptor. Okay to summarize the results from. Yeah. Both a positive and a negative charge and the protein surface is really highly charged. It just so randomly to some of these charged residues. Okay, but, yeah, this is proof. I just explained to you that the mpx very much likes to stay on the surface of the membrane. This is the interface of the outer one, and that it is, whenever it binds to the protein very rarely stays here at the allosteric site and stabilized by a high number of hydrogen bonds. And one has you see very high number of charged amino acid as the pre 14 production partners on the other side of the protein. Okay, so to summarize the results from the unbounded they are able to detect the protein at least some of them. And if I have to to sort them in terms of capability to be or potential to use this vector ligands for cargo, maybe our text and for late will be applied first and then followed by PTX and MTHF so we decided to keep these four ligands and they are targeting the ability by attaching bioactive cargo. Again, we took from experimental data. Biodegradable linker that has been tested experimentally to be suitable touch to it. The two valence complex of the same chemotherapeutic dogs will be seen that was talking about absorb those are especially tailored to binding that so this is our bioactive cargo and tested by molecular dynamics whether the two ligands will be able to carry this system to the protein as they did as they interacted in the free state and here are some short movies. This is complex, as you can see, it goes to the protein but unfortunately they're okay in much the complex way to the active site and in none of the cases was the ligand able to bind in the same way to the active site of the protein as it did when there was no cargo and only the four days. Sorry for the interruption just to mention approaching half an hour. One slide away from you. Thank you. Thank you. Okay. So yeah, as we suspected, PTX and MTHF didn't do so well. They were not able even to stay behind the bioactive cargo to the protein so here is a summary of what we learned from these studies. We found out that the different ligands have specific structure in shape and which is important in a way for their binding but mostly the chemical composition turns out to be the important one for the selectivity of binding. We maintained the liquid order state in all simulations. The structure of the receptor locally solution close to the membrane and only locally when the ligands were bound to it. And I'm not going to read this in detail because I told you about the behavior of the different ligands. We discovered another state center which attracts most of the four leg derivatives but not itself. And this leads to diminished specificity of the binding of these derivatives. And then we found that the cargo, the drug conflicts and fears quite significantly with the steering affinity of the vector molecules and that's why this carbon needs to be tuned carefully to enable efficient functioning of the target in ligands. But apart from that, we can outline RTX and as the most promising candidates for this targeted drug delivery. For fully this is known but for RTX at least we couldn't find our reports so far. And finally, I would like to acknowledge the support both in terms of funding for the research and the computational resource and thank you for your attention. If you have some questions, of course, I will be happy to answer them. Thank you, Professor for the wonderful and interesting presentation. Are there any questions from the audience. I think it's visible till now in the chat. Yes, but to be able to see them. Okay. Sure. Okay, I think this is not a question for me but yes she's here with me in the room. Okay, thank you. I have a question that is not exactly in the focus of your expose but I was wondering, you mentioned that the receptor for the for late was overexpressed in some types of cancer cells. What kinds of cells. Muscle cells. Thank you. Many types of cancer cells over expressed this receptor breast cancer for sure your cancer bladder cancer. Most of the solid tumors over expressed this receptor. The primary reason is that the cancer cells need to feed the vote and for the case. It is required for the growth of the cells and for feeding this division process. So that's why it is there on many types of cancer. Even on bank cancer cells, they register over expression of this receptor. Okay. Thank you. Thanks for the answer.