 about the protein and all over structure of protein and protein unfolding and now we are moving into a greater detail and we have the pleasure to welcome Esko Oxanen who is the instrument scientist on the beam line for a neutron diffraction at the ASS and he's also very well recognized in the field of protein crystallization and using Newton crystallography to unravel the internal secrets of proteins so please Esko. Thank you Tommy and so thanks for giving me the the opportunity to give you give a talk on this course and I was asked to talk about protein crystallization but then after some discussion we decided that it would be nice to also talk a little bit about protein crystallography with neutrons in general because there doesn't seem to be really another talk on that so okay I don't seem to be able to advance my slides now again okay I'm sorry can you hear me actually but now because my zoom is somehow frozen completely so that I'm not I'm sorry a technical problem here we can see you but not can you see the slide yeah I can see the slide and and you are changing because it seems like I can't do anything so do you want to log in again or yes I may have to restart my computer because it seems that I can't do anything I'm really sorry about that this is the whole thing is completely frozen yeah so so so the people attending the course maybe you can think of a clever question for Esko something that you really want to know about the internal structures of protein and if you can unravel that with the neutrons so I think you get impression that most computers today are tired of zoom they are worn out but all those zoom activities in the in the past year now okay yes now I'm back let's see what we get sorry for the technical issues I tried to check everything beforehand and of course that's that works only up to a point so do you now see the slide and no extras okay I still have a big box saying the meeting is being recorded in front of me so I can't see my own slides but that's fine yeah so so I thought I would start with a bit of motivation why we want to use neutrons for crystallography and then talk a little bit about how it is done in practice with the focus on the crystallization parts so first a little bit of general introduction to how we grow protein crystals to begin with and Trevor will today talk about how to make the protein so I will not touch on that too much and then specifically for neutron crystallography how do we make these crystals very big and then I will talk a little bit about the instruments that we use for protein crystallography and and what we do with the data eventually and I would encourage you to ask questions in between because if you wait until the end you will probably have forgotten them by the time so please interrupt please use the raise hand function in zoom so it should pop up in front of the top of my screen don't hesitate to do that so why do we want to use neutrons for protein crystallography basically the simple answer is to see hydrogen atoms and hydrogens are about half of the atoms in the typical protein structure and they are often the ones that are interesting for function as we will see and in this example we have a simple example of a glutamine residue so from this rather high resolution x-ray map you couldn't really see which one is the nitrogen with two hydrogens and which one is the oxygen with none however if we use neutrons instead of x-rays as the radiation even at relatively modest resolution and with a rather bad quality map you could say you can still see very clearly where the nitrogen with the two hydrogens is so the visibility of hydrogen is very similar to heavier atoms you may note that the aliphatic hydrogens in this map are not shown this is because they're actually negative so they even cancel out some of the density because these have not been exchanged to deuterium we will talk about later but firstly a couple of examples of one what kind of systems hydrogen is actually of interest because we don't want to do this just because we can we want to do it in cases where we're really interested in where the hydrogens are and this is an example from an enzyme that has a different natural substrate but it is actually used to detoxify nerve agents it's called isopropyl fluorophosphatase and there were basically two mechanistic proposals one of which has a deprotonated aspartate attacking the phosphorous and another one has a hydroxide ion and the neutron structure clearly shows an H2O or D2O actually bound to the catalytic calcium ion so it is clearly the mechanism A that is consistent with the experimental data and not B so the the objective often is to look at very few hydrogen atoms in a rather large structure so a few out of several thousand another example is from drug development basically so this is from my colleague Zoe Fisher when she was still in the States and she was studying an enzyme called carbonic anhydrase 2 which is a drug target for a drug called acetase of amide which you see on the left and the problem here was that the drug was quite well binding but not very specific because there are many different carbonic anhydrases and to to develop it further it was necessary to know what protonation state it actually has to start sort of computational drug design and there were three different potential protonation states the crystallization pH which were around eight and then the the neutron structure clearly showed that it is number three that you actually see in the in the crystal and of course you see in the crystal in the x-ray crystal structures then that you know from your in your x-ray crystal structures you know what protonation state you have so you're going to start doing drug design and so that we have hydrogen here uh i can't point can i you can't see my i can't see any point i'm a little afraid to click no uh yeah anyway so then the added benefit of doing neutron crystallography is also that we get a full picture of the hydrogen bonding network and it's uh interesting to note that uh usually the hydrogen bonding network is a little bit different from what we would imagine from just taking all the donor acceptor distances that you have so not all the hydrogen bonds that you might imagine are usually formed in a protein so you always tend to have surprises when you do some neutron crystal what is important to understand as well though is that uh neutron crystallography is still something experimentally very difficult so uh when i started doing this back in 2003 my supervisor told me that it's conceptually the same as x-ray crystallography just experimentally about a thousand times more difficult and if you look at the number of pdb depositions now this this should say 2021 sorry i did update the numbers but that didn't update the year apparently uh so uh the growth of uh pdb depositions in the last seven or 18 years or so uh has followed a similar trend so even if neutron crystallography has remained quite difficult the number of structures has grown significantly so uh so it is becoming more possible it is still not something uh you would do unless you absolutely have to so the uh the important thing is that you do neutron crystallography when you have exhausted the possibilities of x-ray crystallography basically and you always do an x-ray structure as well as we will see so why is it so difficult uh well the main reason is of course that we have so weak neutron sources uh i don't think you had a specific lecture about neutron sources so far but i think most of you are familiar with the type of neutron sources that we have but uh they in practice uh to do protein crystallography we are still at a level that is below laboratory x-ray sources in practice so we need very large crystals uh and to make it a little bit quicker so to do things in weeks instead of months uh in some cases uh we also use uh so called lower diffraction polychromatic diffraction to have many wavelengths at once um and then we have an added problem which is incoherence scattering from the light isotope of hydrogen uh and uh this uh uh we can deal with partially by exchanging everything every whole hydrogens in the sample to deuterium um this we will discuss a bit later just to remind you i'm not sure how many of you are familiar with the concept of incoherence scattering uh but um the um what happens in incoherence scattering is that the phase of the scattered radiation is randomized so if you think about bragg's law where uh i guess most of you are familiar with bragg's law which basically states that you will have constructive interference between two diffracted rays uh if there is a specific relation between the angle and the the repeat distance uh but this is only true if the phase is conserved or changed in the same way for every ray so if it is randomized we basically only ever can produce background and not signal for bragg's scattering which is what we do in crystallography um and uh the important thing here is that the incoherence scattering cross section for light hydrogen one H is about 40 times higher than the scattering cross section for coherence scattering as you see in the table here right and uh for deuterium the situation is somewhat better uh where the scattering cross sections are comparable it still means we have to deal with that with a high background technique um so this is really another important challenge uh in addition to the weak neutral sources any questions at this point yes i think sorry you said that the um the typical kind of neutral experiment takes uh crystallography experiment takes a very long time what would be kind of a typical um measurement time for these crystals so um usually something from a couple of days at the shortest to two weeks it's hard to get allocated much more than two weeks beam time usually uh but in in principle you can collect as long as you want it but uh usually there is a there's a diminishing return in having longer exposure times with the detectors that we use so uh using exposure times longer than 24 hours is not usually that useful so it's a question of how desperate you are but i mean i would say from three days to three weeks that's right thanks good so then um as i was asked to talk about protein crystallization so um i will i will not talk about how we make uh pure protein actually crystallizing proteins was first developed to purify them and crystallization is a very good purification technique sometimes sometimes we even actually reuse uh more recrystallized protein to purify it's not very easy but uh it's possible uh and it's interesting that that protein crystals have been known for much longer than protein crystallography has existed so uh um it there was in the 20s and 30s a big discussion about whether proteins have a structure at all and the fact that there are crystals of proteins and that uh they they give rise to diffraction patterns was basically proof that proteins do have a structure and then of course there was the uh the big breakthrough of having the first protein crystal structure but uh the important thing about why we need crystals is that uh it is a repeating object uh and uh because either with x-rays or with neutrons we don't have enough scattering from a single molecule a single object uh to get signal get enough signal so we have to amplify the signal by repeating the uh object the molecule and if they are randomly oriented as they would be in solution we also average all the atomic level positional information so if you want atomic level information using x-rays or neutrons or basically any other technique you need some way of amplifying the signal so the crystal is a way of amplifying your scattering signal basically and how do crystals actually grow so it turns out that uh there is an activation energy to forming a crystal an ordered aggregate and uh and there is there are many theories about how this actually happens in practice but but you have basically two phases you have nucleation forming the critical nucleus that can then grow and then the the you don't have the activation energy of growing of of getting enough molecules together to actually form a repeating structure and then you have a growth phase where you just add molecules to each surface so uh so this is important that you have nucleation which is uh not very well understood step and then grows that is better understood and uh and we can actually represent this in the form of a phase diagram which is quite important so we have on the the y-axis protein concentration and then on the x-axis we have uh been here is precipitant concentration but it could be any other parameter like temperature that we vary and there could be multiple precipitants so it could be a multi-dimensional diagram actually and there we have different zones so we have the under saturated zone below the the solubility limit so this means that existing crystals will dissolve uh then above the solubility limit where existing crystals will actually grow we will not form any new crystals because we're not above the uh the critical concentration of forming nuclei and above the metastable zone so existing crystals will grow but no new ones will form there is a fuzzy line a super saturation limit uh after which nuclei start appearing and the nucleation is a kind of a stochastic process so you can easily make two drops two crystallization drops with exactly the same conditions you get crystals in one you don't get crystals in the other and then if you both weigh too much then you will start forming on uh disordered aggregates which we call precipitate but uh but this is basically the basis of rational understanding of how to grow protein crystals however uh we don't a priori know in what conditions a given protein will crystallize so uh so we have to uh usually test this with um uh with an empirical approach and uh therefore it is usually useful to have a setup or it is very useful to have a setup where we can explore a large area of this this face diagram and this we do typically by vapor diffusion which is quite easy to automate so what you do is you make a on this picture is really bad for some reason um sorry for that but you basically uh you set up your protein solution um in a drop that is in contact with a reservoir solution in a sealed volume and uh then uh water vapor evaporates from your drop and eventually you reach equilibrium through the vapor phase with your reservoir solution so your uh your protein concentration increases and your precipitate concentration increases at the same time um and this we can do in either very small volumes um something like uh 100-200 nanoliter volumes using a pipetting robot which you see here in the middle um to screen many conditions or we can scale it up to some microliters or we can even scale it up further for the neutron crystallization which you see at the the bottom um where you can fit up to 200 microliters of protein so um so once we when we embark on uh solving a neutron crystal structure we typically already know the crystallization conditions from from the x-ray structure so we have somewhere to start from so I'm not going to talk too much about the screening of uh crystallization conditions there are there's a lot to talk about that as well uh but but I will skip that however um when we start thinking about growing big crystals for neutron crystallography the first question is to ask how big crystals are actually big enough how big big crystals do we need and I'm generally against showing equations in lectures because it tends to be not very efficient and in this horrible equation which is basically an equation for the the average signal that we get there is a term that has the number of unit cells which is basically the volume of the crystal divided by the volume of the unit cell so if we substitute that in what we get is uh the volume of the unit cell in the square in the denominator uh what this means is that the volume of the unit cell is very critical for how big crystals do we need because if we have uh twice as big unit cell we need uh four times as big crystals so uh the unit cell volume varies a lot depending on the size of the protein itself and also how it ends up being packed so you can't directly relate it to the size of the protein so this is why it's important to try and choose crystal forms with small unit cells and this also explains why in some cases you get away with relatively small crystals and in other cases even quite large crystals don't deflect neutral as well so uh in deciding how many how big crystals do you need you first need to know what is your unit cell like and of course we try to grow as big the crystals can't really be too big so uh it's more a question of what what would be sufficient now I think there is something as somebody or something in the chat I don't see the yes so um there was a question about whether defect defects can form uh yes they can but usually if you form large enough defects that is what stops the crystal from growing eventually um so it is not actually very clear always what controls the quality of protein crystal because you have to remember also that protein crystals are um they contain up to 70 percent solvents so they have solvent contents ranging from 30 to 70 percent so it is more like an ordered jelly really than a salt crystal and the theoretically the slower you grow your crystal the less defects you will have but what you have to keep in mind as well is that your protein will not necessarily stay stable in time so if you would let your if you're if your protein at the same time aggregates for example when your crystals grow then what might easily happen is that uh if you try to go through too slowly then your protein just precipitates because it goes old so I think the short answer to this is yes but it mostly doesn't matter terribly much right so when we try to grow uh large crystals um we have a number of tricks that uh we start trying on the first one which is by far the easiest is to just add more protein to our vapor diffusion drops which we call feeding uh and this means that in the phase diagram uh we move the uh the state up again to the the metastable zone without going above the the nucleation limit and this is really difficult because we don't know where the nucleation limit is so uh so this is something we only really know by trying in some cases works really well you just add one microliter of protein into your five microliter drop and crystals continue growing uh and then you can you can let it equilibrate for a week and then add more but what it usually what how it usually fails is that instead of growing the existing crystals you get a lot more new crystals um and uh what is often the case as well is that even before we have started the vapor diffusion process so we have when we set up the drop we're already in the metastable zone so then nucleation occurs at the same time as the equilibration occurs so then it's really difficult to know uh what has happened so vapor diffusion is a good technique for screening but it's not very easy to control the conditions there the second thing uh that we would try after feeding if that doesn't work uh is uh simply to scale up the volume now you might think it's easier to scale up the volume than to start feeding drops which is true if you have an unlimited amount of protein but usually you don't so the the effort of of scaling up is really the effort of producing more protein and you will as you will hear from from trevor this afternoon especially if you were dealing with produce rated protein this is not necessarily cheap either so so in scaling up we we basically then go to a different type of plate such as this nine well sandwich plate and um related to the previous question uh it is uh it is actually good that uh this the larger the drops the slower the equilibration so often just having a bigger drop favors larger crystals because the surface volume ratio is different then uh the often does happen that vapor diffusion is simply too difficult to to control to really grow large crystals so the the main problems are that because the the path along the phase diagram is is complex and uh is not reversible so uh so then uh basically we don't know what is happening when and whatever we try it is difficult to to make crystals grow bigger and at that point we typically move to uh techniques that are more complicated to set up experimentally but uh can give us better control and reversibility and one of the mist dialysis where we basically put the protein solution into a little chamber which we seal off with a semi-permeable membrane which lets through the the buffer and of course water and the precipitant in some most cases but not the protein and and this means that by just changing the reservoir solution we can go back and forth in the in the phase diagram and as you can see in the in the the diagram in the left we have a different motion because basically the protein concentration remains the same but we are changing just the precipitant concentration so we have a different trajectory along the phase diagram which is not as good for screening crystallization so it's not as easy to make nuclei but when we try to grow large crystals this is actually a good thing so we can suppress residual nucleation more easily and then we can even use temperature as a variable when we know the conditions very well so because the the play you have with temperature is quite small so you basically already need to have existing crystals but by then changing the temperature either lowering or raising it depending on what type of solubility the protein has we can actually grow existing crystals very nicely reversibly and we've even built automated devices for combining dialysis with with temperature control to to combine the two to do this in basically three dimensions but this requires a well understood and well characterized crystallization system so so you have to first have reproducibility in order for this to work and also significant amounts of protein so on so then if we do have large crystals we still have the issue of the high background from the incoherence capturing so we can't take our fully hydrogenated crystal in H2O because we would have a horrible background on that and we we try to replace all the hydrogen atoms in the sample by by deuterium and the simple thing of course is to exchange the the mother liquor usually by by vapor diffusion or by just growing the crystals under D2O and that takes care of something like 95% of the protein of the hydrogens in the beam because you have to remember that a lot of the protein crystal is water and it is also surrounded by a drop of solvent for it to not dry out so the actual protein is a fairly small part of the sample in the beam eventually however it is helpful to also replace all the aliphatic and aromatic hydrogens in the protein and in this case we have to grow the we have to produce the the protein hydrologically in an organism that can grow in D2O and this is what Trevor will talk about later this afternoon but I want to stress that this is not a requirement for doing protein crystallography it is a way of getting away with smaller crystals but it can be very helpful so so we try to do it whenever it's possible there's a couple of points about when you if you would do crystallize under deuterated conditions because then the protein solubility changes a little bit and most importantly pH is not the same as pd there are formula for for that difference but you need to you need to take that into account and once you do that it is usually quite easy to reproduce the the crystallization conditions under deuterated conditions but the crystals are not necessarily quite as big as under hydrogenated conditions because the protein solubility is actually a bit lower so we can't get as much protein into the solution now if we have a fully hydrogenated crystal grown under hydrogenated conditions what we usually do is we actually exchange the mother liquor with deuterium in the capillary mount where we where we actually collect the data so before we collect the data we mount the crystal in a in a capillary which you see some pictures over the bottom and then because we need to anyway add some liquid some well solution to keep the crystal happy and hydrated so instead of using the reservoir the original reservoir solution would make a deuterated version of the reservoir solution which then through the vapor phase exchanges the hydrogens in the crystal so it is always bad idea to soak crystals in d2o that were grown in h2o simply because the osmotic shock between h2o and d2o is so large that you're basically quite sure to to crack your crystals so so we want to exchange due to deuterium but we need to do that gently so once we have grown a crystal mounted it exchanged it to deuterium we would have to go to a beam line a neutron instrument to collect some data and there are basically three different types of neutron crystallography instruments around depending on the neutron source that we use and the technique we use so on reactor sources which are continuous sources we can either use the monochromatic oscillation method like we use in x-ray crystallography the disadvantage there is that we get less neutrons because we throw most of them away in the monochromatization step but we do have a lower background because we only have background from the wavelength that we use and not all the other wavelengths and we have actually slightly easier to use software for processing the data so i will show later then if we want to make want to collect the data as fast as possible and here the limitation is really usually the time that it would take that if it would take a month to collect data in a monochromatic beam line and if we can do it in a week collecting many wavelengths at once using the lower technique then the lower technique makes sense so this is why we can we can use smaller crystals in practice then the disadvantage is that the data is more difficult to process and we have a higher background because we have background from all the wavelengths and signal for a given reflection only at that particular wavelength and then using sphalation sources we get in principle the best of both worlds using time of flight long way so we use all wavelengths at once but resolve wavelengths using the neutron time of flight i think you will hear about this later i will not go into that too much but in practice the sphalation sources the existing sphalation sources still have relatively weak flux compared to the reactor facilities so this is still not a huge advantage it's rather on par with the best reactor instruments but also the data processing is a little bit more complicated so so all of these have their advantages and disadvantages depending on what kind of crystal system you have then one thing about which instrument or the the neutron instruments that that is important is the wavelength range that is available and it is also important to understand what wavelengths do we want to use and the answer is in again in this horrible equation where we have a lumped up to the fourth term and again this is an equation for our average signal so to speak so this is a very strong driver to use as long wavelengths as we possibly can the limiting factor there is that is bragg's law because if we use very long wavelengths then all our data is going to be at very high scattering angles so so this means that we have to strike a compromise and a reasonable compromise is usually somewhere around two and a half two point six angstrom which nicely corresponds with the the peak of a typical cold moderator spectrum this is an an early spectrum calculated spectrum from an ESS moderator but we can't go to ridiculously wavelengths because then we would get wouldn't get any data but if we use sort of two and a half angstrom wavelengths we can still collect data if we have a detector that goes all around the sample and we have much stronger scattering so where can we practically do this so just start with the monochromatic reactor instruments there are a couple of diffractometers at the Japan research reactor in Tokai which unfortunately is still not operational or is closed waiting to be open so to speak so the the main instrument that that we use and this is the bio diff instrument the f2 which you see on the right so it is a very simple setup with monochromator and then velocity selector for cleaning up the beam and then a cylindrical detector around the sample so I can't this is where I would have liked to point but on the diagram on the right you can see the the blue circle is the detector and the sample sits in the middle so it's a it's a cylindrical neutral image plate there is also an instrument called D19 at the ILL that is sometimes used for a very high resolution protein crystallography work but because it has a thermal beam so much shorter wavelengths it only really works for quite small cells and quite large crystals so then moving on to to laue instruments just to repeat that with laue we have much more reflections to deal with and it is also more sensitive to the the crystal mosaicity which often increases when the crystals grow bigger so the picture on the right is just to show you what we're dealing with so this is a simulated laue pattern with the wavelength color coded so we're dealing with a lot of reflections and we also I'm going to talk about this too much but in practice we're not doing full laue in the sense that we are actually limiting the wavelength that we use so that we know what is the maximum and minimum wavelength so this is why it is called quasi laue and this limits significantly the the overlap of this reflection that makes the person easier and this is what a typical laue image would look like so the if you look at the top of the image the that's where you see a lot of the the small the weak reflections that's where the the real interesting data is if you will so the you have quite a lot of reflections to deal with and the instruments where to do this you have the the real workhorse of neutron bergain crystallography has been the ladi 3 instrument at the isle and they're commissioning another similar instrument now and it is very similar to bio diff it has actually the the same detector the main difference is that instead of a monochromator it has a filter for selecting a wavelength range so so we get much more of the spectrum out and on the wavelengths here are can be adjusted by by moving the filter and there is a very similar instrument called imagine at the high flux isotope reactor in tennessee ridge the main difference is basically the selection of wavelengths is done with a set of mirrors instead of a filter but otherwise it's actually the the same the very same sector and readout system so then in the interest of time fits well with the time of flight i will not talk about very much about spallation sources how we resolve the the wavelengths using time of flight but basically we can this means we can spread the background over many time bins and we can resolve eventual spot overlaps using an extra dimension and the existing instruments at both sns and j park are quite similar it looks like a big hedgehog so this is something like a three meters high what you see on the right the detector array of the mandi instrument and you actually have to lower the sample down in an elevator it goes two meters down in the setup so it's quite a quite a thing and they have similar geometry different kinds of detectors but otherwise quite quite similar instruments both in the us and japan what we are building here in lund is going to be a little bit different so because we the ess source is a little bit different it's a long pulse source so the instrument here will be quite a bit longer 158 meters from the moderator sample and we will not be putting the detectors all around the sample permanently but rather mounting the detectors on robots so that we can actually work in a much more open geometry so this is just a rendering of what the experimental area will look like and when we have ess operating at reasonable power we should get significantly more flux than existing instruments so we we expect to be able to collect data much faster so the objective is to be able to collect the dates in a day and just to show you how it looks like actually no today we've already built control hutch here but it's still we have some buildings where the experimental area is here but there's no components in it yet so so this is still several years away from being operational just a few words on the data processing and refinement so the main the sort of basic workflow of crystallography be it x-ray or neutron is that we have to first find where the diffraction spots are we have to index them we have to give them the have to assign miller indices to them and find the unit cell and then we have to integrate their intensities put those intensities on a common scale determine the phases which is a significant hurdle in x-ray crystallography and then we computational refine atomic coordinates to the structure and then we investigate the maps now there are some challenges compared to x-ray crystallography here the main ones actually being in integration because we have high backgrounds so that is the main hurdle also if we do a law where we need to scale together multiple wavelengths which is another step one thing we don't need to worry is determining the phases because we always have an x-ray structure of the same thing that we put in the neutron beam and the refinement is also tricky because we basically have the data to parameter ratio by increasing the number of atoms by about double and and we don't actually have any more data we have less data than than we would have in x-ray crystallography and we there we also typically because of the long wavelengths we have low completeness of the maps so they're not as pretty I will not talk more about that but an important thing that we do to to get around the data to parameter ratio issue is that we actually do joint neutron x-ray refinement and this means that we never do a neutron structure without doing an x-ray structure because for this refinement we ideally want to do it from the same crystal sometimes this is a crystal from the same drop and importantly at the same temperature because we refine temperature factors atomic displacement factors that are temperature dependent so the two dates just need to be at the same temperature so whenever we collect a neutron data set we try to collect an x-ray data set from the same crystal after the neutron data collection or if not that's not possible then we collect from a crystal from the same batch and in addition to the the joint refinement we can also use information from high resolution cryogenic temperature x-rays for example in this example of an enzyme called urate oxidase you can see that the neutron map on the the right is not very it's not very clear where the heavy atoms are but once you know where the heavy atoms are from the x-ray map you can see the hydrogens rather clearly and even knowing that you have for example water molecules so on the the left hand side you can see there is a chain of water molecules that goes from from one side of the substrate to the other on the neutron map you see this extension of the density you wouldn't model a model of water there if you didn't know that there is a water there but this actually makes sense when we do quantum chemistry calculations but there is approach and relay so this type of combining information from x-rays and neutrons is really key for getting the most out of neutrons so it is not something you ever do on its own basically so with that I would like to thank you for your attention sorry for going four minutes over time and I'd be happy to answer any questions you still have I have a question you say that you want to use deutron to see the crystal structure and did you on can we use the deutrated monomer and then crystal as a tendency to so you mean deuterated protein yes yes yes so we can but we still have to crystallize that deuter or we have to have the solvent around it deuterate it that is more important than having deuterated protein because the the non-exchangeable hydrogens in a protein are a rather small fraction of the total amount of hydrogen in the sample because we have if we have a one cubic millimeter crystal which has 50 solvents then we have maybe five times or ten times more solvent around it just to keep the the crystal in the capillary then it is more important to exchange all the other components than the the protein aliphatic hydrogens and I have a basic question can you give a definition what is protein crystal so a protein crystal is an ordered array of protein molecules so it is a crystal that is held together by by intermolecular interactions but what you have to understand is that you have the the number of interactions holding together a typical protein crystal is similar to what is holding together a typical sugar crystal but you have about a thousand times more atoms in a protein than you have in a sucrose molecule for example then what is the difference between protein crystal and protein amyloid like they are also repeated aggregation of the protein yes but if you have like for example a fiber it has it is repeating in one dimension it is not repeating in three dimensions okay so ordered it is not ordered in three dimensions so you can have you can have i mean you can have two-dimensional crystals for example that is also possible you can have layer crystals that are ordered in two dimensions and not in the third one that that sometimes happens as well but here we are using 3d crystals and we have a question in the chat there's a neutron scattering better for membrane protein so that membrane proteins are really difficult to to work with so there are about i mean even membrane protein x-ray structures it is less than i think five percent of the existing protein structures so so it is really difficult to to produce them it is really difficult to to grow crystals at all and there are no examples yet of of neutron structure the membrane proteins there is one potassium channel where they've collected some data but not not good enough to to really determine the structure anything else so can i ask a question i ask or even if i know far from past expiry for being a student so how much faster will it be to do a diffractogram of a protein with ASS compared to say mlz so that that is a very tricky question because it depends very much on the on the protein and we don't really have a lot of comparative data from of the same crystal from from different instruments and because the the problem is also that the crystals are fragile enough that when you transport them they sometimes stop diffracting as well so it's not even i mean you can't even be sure that you have exactly the same thing but the the the rough calculation is that ESS operating at 5 megawatts we should get something like 17 times more neutrons time averaged than Ladi which is currently the best instrument i would say and there are different opinions on what is the real gain of using the time of light some say it's a factor of hundreds some say it's a factor of five essentially but my take is it's the basically the ratio of the pulse length to the the bandwidth so it's about a factor of 20 so if we take a factor of 20 then my guess is that 5 megawatts given again there is another thing about the the detector efficiency so which is also and the detector not necessarily just efficiency but the the stability of the detector how long you have to count to get the background flat so my guess is that we should be able to collect a dataset in a day that from a crystal that is 10 to 100 times smaller than what we would usually today need for collecting a dataset in five or six days but that is really difficult to know and it's it's difficult to know how it scales with the ESS power because it's going to be some time before we operate at 5 megawatts okay very good there is one in the chat another in the chat so yeah about secondary structure so the secondary structure to be honest we see the secondary structure very well from x-ray crystallography so there is no reason to this is not usually something you need neutrons for so so we do see of course the I mean you see the hydrogens I mean the the the hydrogens in beta sheets for example are pretty much the first ones you see because they are the best ordered they are also I mean in high resolution x-ray structures these hydrogens are often visible as well so you don't actually need neutrons to study that to be honest