 Excellent. Well, thank you very much indeed. Thank you for the introduction Erica and thanks to the Organising Committee for inviting me here today. So I'm going to talk to you about recombinant membrane protein production in microbial hosts. And what I'm going to do is the talk is in two parts. And because it's a workshop I've tried to make it quite practical. So first of all, I'm going to talk to you about a review that I wrote with Bruno, who I see is in the audience here today, where we surveyed membrane protein production in microbial hosts and by that we mean E. Coli and two species of yeast. So saccharomyces cereviscii and pycopastorus, and I'll just survey our findings in terms of the most commonly used routes to success. And then in the second part, I'm just going to give you a slightly more in-depth analysis of an example of one membrane protein that we've recently expressed in pycopastorus in my lab and how we also then extracted it with styrene mulletic acid. And for those of you who aren't so familiar with that, that's a polymer based extraction technique that's rather different from traditional detergents. All right. So just to make sure we're all on the same page then. What we're talking about is recombinant protein production. And as I'm sure you all know, we effectively take the DNA encoding the protein of interest. We feed that into a host cell. And then we ask our cell to behave like a cell factory, and we culture the cells so that they produce the protein of interest to us. So those of you who have done this before know that it never works first time, and often you need iterative rounds because what you need to ensure is that the protein you've produced is functional. So you will, of course, need to spend some time on protein characterization. And often you may find that you haven't got enough protein, or the protein isn't sufficiently stable, or if you're going to embark on a structural biology campaign that maybe your protein won't crystallize. What you do then is you can start to think about maybe mutating the DNA to make a mutant protein. So you reintroduce your mutant form of the DNA back into your host cell, and then adapting the culture conditions hopefully you'll end up with what you need. And I would like to stress that actually spending some time on culturing the cells can be a very fruitful way to make sure that you get enough functional protein of interest. And I'll come back to that a little bit in a second part of the talk. We all know how cells are just factories, they're complicated. And so of course, what we need to think about it are all the various things that can go wrong. And we need to make sure that the cell, which we're using as our cell factory here, we try and manage the stress that it experiences, when we're asking it to do two things for us just to be a cell, but also to be our protein production factory. So of course what we need to do is introduce our DNA. We typically have its transcription under the control of a promoter it's often a strong promoter. And then of course there are lots of options therefore for problems to arise you may find there are degradation pathways that are up regulated. We also need to worry about the structure of the mRNA. And of course then this is a template for the energy dependent translation process that provides us with our recombinant protein and when it's a membrane protein of course it's folded and enters the secretory pathway and becomes inserted into the membrane. And of course again things can go wrong, and the protein may end up being degraded and broken down so we need to manage all of this through the KF4 management of our culture and regime that we're using to produce the protein of interest. So let's start to think about the components of a protein production experiment. The first thing of course is the gene sequence so that's the sequence encoding the target protein of interest. And of course we can think about all sorts of things here about code and optimization. We almost certainly will be adding a purification on detection tag to the gene sequence. And we also need to think quite carefully about signal sequences. You can see my PhD student Thanos in the audience he's thinking a lot about this at the moment, and I'm not going to spend much time on this but this is really something that there's a lot of literature out there upon. And we certainly need to think about carefully designing the gene sequence in order to get the best outcome of our recombinant protein production experiment. So having identified the sequence that we're going to use we then want to put that into a plasmid. We can either have a plasmid that sits in the site of plasm so an epizomal plasmid, or in some cases and particularly when you're working with the East pick your pastoralist, you'll integrate your gene sequence into the genome to make a stable strain. So typically when we're talking about microbes, bacteria and saccharomyces rely on epizomal plasmids and pick your pastoralist tends to rely on genomic based integrated expression. So this is really important, inducible versus constitutive so do you grow the cells and then turn on the expression, or is the expression occurring throughout the growth phase. And then having chosen your host cell, whether it's E. Coli or saccharomyces or pick your pastoralist, you may also think about choosing a specific strain. For example, you may want to work in a protease deficient background, or you may consider doing something like choosing an engineered strain to minimize for example the bioconstellation pathway, and that can be important in the East. And I've already mentioned really thinking hard about the culture conditions and optimizing them in order to maximize functional yield and manage the stress experienced by the cell is a really important component for designing a good protein production experiment. The most recombinant mammalian membrane protein structures used recombinant proteins produced in yeast. And the reason to think about protein production experiments that results in structures is this is a really good way to define high yielding production experiments that have given really high quality protein. So in the in the article in the review article I'm going to summarize for you in a minute, we use that as the basis of our findings because we figured that if we were focusing on proteins that have had had their structures sold. They must have been produced through high quality protein production experiments. And there are two examples here, these were published in 2005. So this protein here was expressed in saccharomyces or an SCI, and this protein here was expressed in picking a pastoralist. And I've also got a note for you here as well at the bottom of the slide. And that's concerning the fact that we know that yeast can sometimes be problematic in terms of post translational modifications. And there's always a concern that that might hinder crystallization and I just wanted to point out to you that there is a high resolution structure of a glycosylated worm, P glycoprotein from C elegans. And that was synthesizing P pastoralists, and it gives some comfort that yeast glycosylation doesn't necessarily hinder crystal formation. So we go to the first part of the talk now. So this is an article that I published a few years ago now in 2018, together with Peter Henderson in Leeds, who uses E. Coli, and also with Bruno Marou, who's sitting in the audience and was responsible for the discovery and exploitation of the C 41 and C 43 strains that you'll be very familiar with. So together with the co-authors here on this paper, we surveyed the microbial expression of membrane proteins using this approach of looking at proteins that have had their structures solved and deposited in the protein data bank. So what let's start with a little bit of context here. What we noticed back then, and this is more or less still true, is that 31% of all membrane protein coordinate files deposited in the protein data bank were derived from the component protein so that's quite a large proportion and it's growing. We also noted that 71% of all unique structures were derived from microbial sources so you can see that microbes themselves as host cell factories are making a really important contribution to the structural biology of membrane proteins. And that's going to lead to the generation of high quality protein in high yields. So E. Coli is by far the most important as you can see here in terms of numbers 64% and then the rest in these two species. And you also notice that if you look into the protein data bank into the protein data bank, you will see that there are other less commonly used microbes available as well and these tend to be used on a case by case basis. So if you now is to summarize some of the findings that we published in that paper, but I would encourage you if you're interested in you have more questions to look into the paper itself there's a lot of data and now tables, a lot of information and references to the wider literature. So if you are interested, it's a good resource to have a look at microbial expression of membrane proteins. The first thing to note then is the target. And what you'll see in the next few graphs is that the gray bars represent proteins that have been expressed in yeast and black bars that are proteins that have been expressed in E. Coli. You can see the majority of the proteins expressed in microbial hosts are alpha helical in nature, a few are monotopic. And then, in terms of beta barrels, they are only and are exclusively expressed in E. Coli which is perhaps not surprising. If you also then look at the size of protein that has been produced, what you can see across the size range is that E. Coli tends to be used at the smaller end of the size range, and yeast is making a contribution across the full range of sizes of the proteins that were produced for structural biology. The next thing to think about is tags. So if we look again at proteins produced in E. Coli, in black, and proteins produced in yeast and gray, what you can see in both cases is that the polyhystidine tag predominates as the tag of choice in both cases. It's by far the most widely used of the tags. There are other tags out there, of course, GFP, the maltose binding protein, Flagstrap and GST are used. It's the yeast tag that dominates. And some proteins are multiply tagged. And that's certainly a technique that we use in my group and is used widely by others. And there may be reasons why you want to be able to collect and detect your protein using more than one tag. So that's something to bear in mind. If you think about where the tag is, as we surveyed the database, what we found was there was a more or less equal distribution between putting the tag at the C terminus and putting the tag at the N terminus with a slight preference to put the tag at the C terminus. Now, of course, this is a determination that you really need to make based on your understanding of the protein that you're working on. And if it so happens, there's an important functional group or a functional domain in the carboxy terminus, perhaps you don't want to put your tag there, but that's not necessarily always the case. And there are some examples where the tag has been put in another part of the protein, maybe one of the extra cellular loops, for example. If you look at the poly-histodine tag as the most widely used tag as well, what you see is hexahistodine is the most commonly used. And here's an example of a hexahistodine tag here. But larger and longer tags are increasingly becoming more important. You can see here is a decahistodine tag. And that's particularly particularly the case if you're going to use something like the SMA code polymer, which is quite bulky. And so then there's a feeling that having a longer tag can certainly help in purification. And if we think about then how we're going to attach our tag to our protein, what you can see is the use of protease cleavage sites as being a really useful way of having an optionally cleavable tag. So one very commonly used protease site is the TEV protease. And that's because the TEV protease itself is active in the presence of most commonly used detergents, but also the thrombone protease is widely used. So that's something about the construct itself and where you might be putting the tag. And then if we think about a really important component of a protein production experiment, that's the promoter. So what we find when we surveyed the promoters that we use is neither the bar colouring has a different meaning. So here in black are proteins that are heterologous to E. coli. So these are non E. coli membrane proteins being expressed in E. coli. And gray means E. coli proteins expressed in E. coli. And what you can see is, again, there's a real predominance of one choice here and that's the T7 promoter system, the IPTG induced support T7 promoter system. Other promoters are used as well, but by far the most common choice is to use T7. What you can see as well for homologous membrane proteins, perhaps this isn't quite so surprising, is that they often use their own native promoter in order to be produced in E. coli. Let's have a look at pycopastorus, which is one of the E species we use in my group and that is used quite widely for the production of a range of different membrane proteins. Here I've also included information on the foremost commonly used strains. So this is X33, one that we use quite commonly. And then here we have a protease deficient strain, which is used very commonly for proteins that have been produced for structural biology. And what you can see here is we've really only got two choices in in pycopastorus. You have this inducible alcohol oxidase promoter AOX1. And you also have the option of a constitutive PMO1 promoter. And what you can see in the survey that we did was that by far the most common combination was this inducible alcohol oxidase promoter with a protease deficient strain. And you can see there was one example where we didn't know what the promoter was in the literature. And of note here is that typically we use integrative plasmids for expression in pycopastorus. What that means is we integrate the gene sequence with any of the tags that we've added into the genome and make a stable strain. And then the usual thing is to grow up the cells to a very high cell density. This is something that's very specific to pycopastorus. We can get very, very high cell densities. You can see here greater than 100 grams per litre of dry cell weight, for example, or greater than 500 OD units per mil. And the idea then is that you grow the cells up to these very high cell densities. And then you turn on the induction of your gene expression with the alcohol oxidase one promoter, which is induced using methanol. And then hopefully because you've got lots of membranes, you also get lots of protein. The Saccharomyces cerevisia promoters, you've got a much wider range of strains in which to integrate, to use your plasmid. But again, a limited number of promoters. So here our inducible system is the gal promoter induced by galactose addition. And again, this constitutive PMA1 promoter here. And now using Saccharomyces cerevisia, as I mentioned before, the tradition is to use epizonal plasmids used for expression in Saccharomyces rather than integrating. And you can see here there's just a range of different strains that are being used, typically with this galactose inducible promoter. So what I wanted to do now is to just give you an idea of some of the different approaches we've taken to expressing different types of proteins. I'm going to briefly tell you about a GPCR and an aquapoint that we've expressed in my group. And then I'm going to spend a bit more time on the third membrane protein called CD81, which is a tetra-spanin. So let's start with the GPCR. This is an expression of a human, the human adenosine 2A receptor in Saccharomyces cerevisiae. And what you have here are some fluorescence microscopy images of the wild type adenosine A2A receptor. And what you can see here is that it's been tagged with green fluorescent protein. And you can see the protein throughout the cell and you get some nice expression of the plasma membrane. So that's all looking good. We then added a tag. We were doing a series of experiments where we had a specific tag we wanted to use for a variety of reasons. And what happened then when we added that tag was that the protein ended up in the vacuole, which is not helpful at all. I don't know whether you can see there's no protein at all around the surface of the cell where the plasma membrane is. You can see that very pale purple ring around the edge. And what you can see instead is that the protein is in the vacuole. So the reason I'm showing you all of this is that this is an example where choosing the right strain really helps. So what we did was when we expressed our protein in this deletion strain, SBG3 delta, which is a Saccharomyces cerevisiae strain, we could restore expression, functional expression to the cell surface. And this is a confocal image where you can really nicely see that the protein has ended up in the plasma membrane of our Saccharomyces cerevisiae cells. And we were able then to do some binding studies to show that the protein was indeed functional. So the take a message from this slide for you is that a choice of strain can make a really big difference in terms of whether an expression experiment is successful or not. If we take the same adenosine A2O receptor, and we express that now in the second species of yeast that we've been talking about P pastores, then I wanted to show you an example of how we've done that and how we've also extracted that protein using styrene Malay acid. So for those of you who are not so familiar with the SMA technology, I'm sure most of you are. So this is the SMA 2000 copolymer in a two to one ratio and that's solved by Cray. And what happens is that instead of using a traditional detergent like DDM or octal glucoside, you add this polymer to your membrane preparation and the polymer wraps around both the protein and lipid. This is the so called styrene Malay acid lycoparticles. So you end up with your protein embedded in the surrounding lipid and encased in the SMA polymer. And this gives us lots of really interesting ways to examine the importance of protein lipid interactions when you're looking at the protein of interest. So here we have an example where we've expressed our human A2O recombinantly. In this case we've got an N terminal, his 10 tag, because we were aware we would be working with SMA and we wanted to make sure we had a long enough tag for the purification. So we're using the P pick Z alpha a expression plasmid. So this has an alpha secretion signal in front of the A2A gene. And then we did the expression in the P pastoral strain x 33 under the control of the alcohol oxidase one gene. And in this particular set of experiments we also took a we introduced a mutation into this gene sequence in order to preclude hyperglycosylation of our target protein. And what you can see here is when we express and then extract using the SMA polymer, we end up with a single band when we do our illusion from nickel NTA column using 250 millimolar imidazole. And we get a nice Western blot here, nice thing about here showing that we've been able to produce a high quality, very pure protein. And we also want to be sure that our protein is active and what I wanted to show you that was something that was quite interesting to us is when we did the same experiment, but we extracted our protein in the DDM. Then we didn't get any specific binding when we looked at binding of this radio ligands here to the density native a receptor that we've expressed in P pastores. And this is a dope in cholesterol hemisuccinate in order to see any activity any binding activity, and this is consistent with the known dependence of a to a on cholesterol in order to get full binding. If you notice when we extract an SMA. Now we've got both protein and lipid within our lipoparticle, we see binding, and this was really interesting to us, because of course there's no cholesterol and yeast, we're working with these membranes. And we were still finding that we were getting activity here. So it's possible, either that the agostral, which is the equivalent sterile in yeast membranes to cholesterol is doing the job of the cholesterol, when we have to dope it in, in a detergent based case, or it might be that somehow the way that the protein is held within that liquid environment is allowing it to be in a functionally folded form, and you can see when we put cholesterol into the estimate make no difference at all. So, as well as GPCRs, we're very fond of aquaporants in my group, and we published right a year ago now a really nice study with Susanna who's also in the audience, and she did some work for us using a recombinant expression of the Q and aquaporin-4 and P pastores, and this is the experiment we were using that protein for. So Susanna was able to produce and purify using traditional detergent, wild type aquaporin-4, and you can see what we're looking at here in this orange is the binding of aquaporin-4 to calm modulin, and we were able to do a series of experiments using the purified aquaporin-4 that Susanna had produced, and the calm modulin that also was produced in her lab, and to examine the binding of aquaporin-4 to calm modulin. And using a series of different experiments we were able to identify that we could inhibit that by adding this molecule called trifluoropericine, which is a calm modulin inhibitor, and you can see the knockout binding there. We could also remove part of the C-terminus, which is in these blue triangles, and you can see we knocked out the binding if the binding site wasn't there. We could also knock it out by adding EDTA because we have a calcium dependent mechanism, and again we knocked that binding out. And then we were able to anticipate how the binding was likely to be happening on the C-terminus, so we'd identified on the C-terminus the potential for this calm modulin binding site to be formed. So for the C-terminus to go from a rather unstructured peptide into something that looked more like this nice alpha helix here, revealing this hydrophobic binding site here, and if we knocked that out in this mutant here, as you can see, and again we would knock out the binding. And Susanna also did some experiments where she made a phosphor-nometic mutant here, and again she increased the binding in that case. So this is a really nice use of a recombinant aquaporin for this particular set of experiments. And what we're doing now, this is some work done by one of my PhD students, Lucas, I think is also here with us. We're now trying to produce human aquaporin form and then to extract it using SMA. And what we want to do now is we've got a C-terminally HIST6 tagged version of aquaporin form in the Ppix ZB expression plasmid using X33, again under the control of the alcohol oxidase promoter. And what you'll notice here is that we're even though we know we've got a functionally important component of the binding site in the C-terminus, we can have this HIST tag and that doesn't make any difference to the activity of the protein. So in the last part of my talk, I just wanted to give you a little bit more detail around expression of another protein because you're going to hear a lot more about aquaporin form, Christina Headphile in a bit. So I'm going to choose to talk about a different protein, and I'm going to talk about human CD81, its expression in yeast and some work on its SMA extraction. So this is work that's a combination of work from a PhD student with Broadbent in my lab, and also some work that we published at the end of last year. So tetraspinins are four transmembrane domain proteins as you can see this is a nice diagram that Lucas drawn so you can see the four transmembrane domains, and then you have two extracellular loops, a small extracellular loop here and this large extracellular loop. So tetraspinins are a family of proteins that have a wide range of functions in human and other cells. What's of interest to us around CD81 is it's involved in the infectivity of a range of human pathogens, including hepatitis C virus, but also malaria and influenza. So we know CD81 binds the hepatitis C virus to like a protein and that's how hepatitis C viral infection occurs. And we also know that CD81 has had its structure solved. So it has a so called waffle cone structure like this. And this is rather an unusual structure that was that was solved a few years ago now. And then the top of the aquaporin where the loops are for what you might think of as being the ice cream part of the ice cream cone. And it's thought that this part at the top can open and close. There are lots of open questions around both the oligomerization status of this protein, which is thought to be diameric, but also what the physiological relevance are of these open and closed states. So many of these questions around CD81 are really about protein lipid interactions. And so what we wanted to do was to express to start with our protein in yeast, and ultimately move into a more native mammalian cells. And also to make sure we could express and then purify our protein using SMA in order to preserve those lipid interactions. So here we have our expression in ppxd, encoding the C terminally his six times human CD81 in strain x 33 again under the control of the So if we think a little bit now about CD81, what we can see is we can express it in yeast, and then we can use both SMA polymers or conventional detergents to extract it. So what we have in this top part of the graph here is just some of the membranes from yeast. And I wanted to show you the visual representation of what happens when you add SMA 2000 this polymer. And you can see it's much clearer. You can see this as represented here on this graph, if we're measuring just OD 600, you can see that the membrane suspension is here. And then when you add either two versions of the SMA polymer, or in DDM, then you find that the C drops because you've cleared the membrane solubilised membrane. We can also see very clearly that the solubilisation efficiency of CD81 from yeast membrane is very similar, whether you use a conventional detergent like DDM or these SMA polymers. And one thing I did want to point out to you is that if you do use SMA polymers, and we've used two here, we've used the two to one Cray Valley polymer, but also a three to one polyscope polymer, which give very similar results. Sometimes excess SMA can mask signals and Western blocks, so it's actually quite important when you're looking at solubilisation efficiencies that you look at the pellet because that pellet's been washed away. So we look at the P here, the pellet, in order to see how the solubilisation is progressing. You can see here in DDM, you've got this nice strong band in a solubilised fraction, and that's not mirrored at all here because of the masking effect of the excess SMA. So something else that's really interesting, especially with notice when using yeast as our system is that when you add SMA to the yeast membranes, they clear really quickly, and you can see that here in this graph. So the OD drops very quickly and is maintained in that state over several hours, you can see within just a few minutes, the membrane suspension clears, and that gives the impression that solubilisation itself, operating of interest is also very quick. And that's actually not necessarily true. So what you find is the membrane itself breaks up very quickly, and so the optical density drops. But if you actually look at the protein, and this is a now a Western blot of our CD81, what you can see is the solubilisation is much slower and that's represented by this gray line here in soluble CD81. And dropping with time, you can see it takes several hours for the protein to be solubilised effectively. And we also see that if we increase the amount of SMA that we're using as well. We wondered whether we might be able to make the solubilisation more efficient by increasing the amount of SMA that we use, but that doesn't actually end up being the case for CD81. So the rate of solubilisation is both protein and expression system specific in our experience. And also you really do need to think about measuring the protein specifically and simply monitoring the optical density is not sufficient. So having expressed our CD81 and encapsulated it in SMA to make a small, this is then our purification gel you can see here over here we've used nickel NTA and we've eluted the protein using 300 millimolar imidazole here. This is a reasonably pure protein, as you can see in these lanes here. And it was important to know whether this protein was functionally folded or not. So we were able to use three different confirmation of the active antibodies you can see here that they bind to our CD81 and our control does not. And importantly as well, that this E2 glycoprotein binds to our small protein, but our control does not. So we know that our protein is functionally folded. So we've been minimised by physically our protein. And what you can see here is a comparison of CD81 that's being expressed in DDM and CD81 that's been, sorry, CD81 has been expressed in yeast and purified in DDM and CD81 that's been purified using SMA. And you can see that the SMALT, these SMA-like particles are larger, about 10 nanomolar, much larger than the DDM encapsulated CD81 micelles. You can also see that the CD spectrum is rather different between the DDM and the SMALT purified materials. And particularly that the amount of alpha helical content is affected between the SMA purified and the DDM purified material. And we wonder whether that might be a function of the way that the protein is packed within the SMA-like particle. We can use these CDs in order to look at thermostability of our proteins. One of the things that has been observed in previously is that SMA can sometimes extend the thermostability of a protein. So what we did was we followed with temperature this peak here at 207 nanometers. So first of all, we noticed was that what we find is that there is a secondary structure of CD81 appears to be more or less equally thermostable in a SMALT compared to the DDM purified material as we've got here. We also noticed that the aggregation of CD81 appears to be much less in DDM here. You can see that compared to the SMALT, you get much higher aggregates in the SMALT. But interestingly to us, it was the CD81 in SMALT that was more thermostable with respect to this important extracellular loop. So the large extracellular loop, the one that goes over the top and forms the ice cream in the waffle cone, is where the important E2 binding occurs. And that binding activity appears to be maintained for longer and at higher temperatures in the SMA compared to the DDM purified material. So when we did the size exclusion primatography, we saw something really interesting. We've got two peaks that look rather similar, as you can see here. But we found that only the second peak, the small one here, was the one that was binding E2. So that was the one that was functional. So we really were interested in trying to get more of this material. So you can see this is our typical peak for us. We're getting lots of peak one, which is the one that doesn't bind E2, and not very much of peak two, which does bind E2. And you can see here on this chart, this is present in lower quantities. So I've mentioned earlier on it was important to think about your culture conditions. And so I wanted to bring that back up here because what we were able to do was increase the amount of the second peak, as you can see here. By inducing it to a lower cell density, an OD of about one rather than an OD of greater than five, which is what we normally do. And also doing the induction for a very long time, 22 hours in order to get some protein that was then able to be identified as peak two, as you can see here. We also optimized our purification buffer, and rather than using HEPI, we use HEPIs rather than Tris, and we added some glycerol and some sodium chloride. You can see when we do a biophysical characterization of these two peaks that they overlay exactly. And so peak one seems to have the same secondary folded secondary structure as peak two. The main difference is their size, so peak one particles are approximately twice the size of peak two particles. And maybe that's because there's a dimer formation, or maybe there's some kind of conformational change. We don't know the answer to that yet. And finally, I just wanted to very quickly show you how we were able to analyze the liquid content of the material that we had purified. So we were able to do some mass spectrometry on the crude membrane, on the material that had been solubilized with SMA, and then the material that we had purified and contained only CD81 in slabs. And what you can see is that for the P-pastor's membranes, the mass spec are dominated by PC in green, and relatively long polyunsaturated chains, and there are several different PE species in blue. When we do a solubilization, so we remove the unsolubilized material, but this is crude solubilized membranes. We've lost the triosal glycerols here, we've lost the single myelin here, and we keep the complex PC, but some different PE species. That's my 35 minutes. And then when we purify the CD81, we have an almost complete loss of PE, and then we get PI and PA present as well. So let me conclude then by saying that microbial hosts dominate in the production of recombinant membrane creatines for structural studies. And there are lots of things to think about when you're designing a good experiment, including the gene, the tags, the promoter, the strain, and the culture conditions. And then I've shown you a little bit more detail about expression of CD81 and P-pastor's and its solubilization and purification using SMA, that we can generate an SMA encapsulated CD81 that retains its native-folded structure, that we can optimize our expression and buffer conditions to improve protein quality, and that we can interrogate the lipid environment of CD81 within our smarts, which are enriched with negatively charged lipids. So finally, thank you to everybody who's been involved. So along with me here is my colleague Alice Rossney, who together we supervise this PhD student of ours, Paul Ayo, who is the first author on the work that I've just shown you on CD81, and together with Michelle Claire, who was the postdoc at the time. I've mentioned Luke, Lukas and Thanos already. I'm with Phil Kitchenworks with Luke on the Aquaporin project, and my colleagues here at Aston have been involved in the CD81 project, together with colleagues at Warwick, and my colleagues at the German Center for Infection Research. We're supported by BBSRC, and I'd like to thank all of my colleagues at Aston's member in protein and lipids group, and I'd like to very much thank you for your attention. Thanks so much, Rossney. It was a wonderful presentation. We have quite many questions, so I'm sure that we will get back also in the general discussion. The first question was from Ariane. He's asked that, do you think that the use of different harsh expressions will evolve as a result of increased use of cryoEM as structure determination method of choice. Do you think the use of different harsh expressions will evolve as a result of increased use of cryoEM as structure? Yes, so yes I do. Hey Ariane, nice to see you. Of course, I mean one of the things about when we realized more heavily on crystallography was we really needed to have huge amounts of protein and of course microbes are really good for that because you can grow them in large quantities and they're relatively cheap. And of course now we've seen these strides in cryoEM. I'm sure we will see a wider range of hosts being used for that because you don't need as much protein. In fact, I think that's a primary driver there. And if you see, do you purify the SMA polymer in any way before you use it? What we have is we have quite a defined protocol that we use. So we buy in the raw materials and then we have to reflux it because we have to, we buy it in as an anhydride and then we have to reflux it to make the acid form. And then we do a series of dialysis steps. So I'll be happy to share that protocol with you. My colleague Alice Rossney who I showed you this photograph of in the last slide with me is an expert on this. So I'm very happy to introduce you to her and she can give you her protocols. We have also from Philip Pamula, he asked, did you check how many lipids per membrane protein you are able to extract with the SMA? We'd really like to do that. So unfortunately our mass effect wasn't quantitative so we haven't been able to do that for that particular experiment but it is possible to do that. Yes, one of the really interesting things about using SMA is that in principle there seems to be a limit to how many lipids you could get into the nano disk because it's 10 nanometers dimension. So there are some really big proteins that have been expressed and purified using SMA and you quite wonder how many lipids really are in there but we haven't been able to calculate that with ours. And the answer, yes, can you purify SMA polymer and use it for bacterial proteins? So I guess, yes you can. And if you see there is a question from Julie Tucker, maybe it's easier if you. I'm interested now, how does the speed of salivation test make a matter of that soluble detergent and whether this impacts on the function and stability of the extracted protein? Yes, so what's really interesting about SMA is that the speed of solubilisation seems to be really protein and host dependent. So it can actually be quite slow. In some cases we found it to be quite slow for yeast, for example, so the example I showed you with CD81 we were still seeing solubilisation after four hours, whether this impacts on the function and stability of the extracted protein. So for us, it seemed to be fine. It's almost once it's been encapsulated in the SMA. The SMA can hang around for quite a long time at four degrees and there doesn't seem to be too much degradation of function. Thank you. Thank you so much. And also thank you very much for the presentation.