 Thank you, Bill. And thanks to the organizers for inviting me. Many of you have seen this video. The Institute where I'm working at belongs to the Leibniz Association and last year was Leibniz year and I happened to be the only female, young, energetic virologist that had time to go to Berlin, so that's why I'm in that video and I think it nicely illustrates what we're all interested in, this little guy, a virus, and it also shows you that viruses can naturally fly, at least some of them, when you blow them in somebody else's face. And this is important if you want to study them with mass spectrometry. The main technique we're using is native mass spectrometry. So just a little bit for you to understand what we're doing. Normally when you run a mass spectrum, you would have your viral protein. In this case, it's a capsic protein of norovirus, VP1. You would have it in organic solvent with a little bit of formic acid. So your nice structure will unfold and since you have a large surface, it will take up a lot of charges. And what you see then in your mass spectrum is along the mass to charge ratio. You see a lot of different peaks, each of the peak having different charge. Here you have the lower charge states, here the higher ones, and from the adjacent peaks we can determine the mass. This is very nice for accurate mass measurements, but it doesn't tell you anything about the complexes you have. If we take the very same protein and put it into buffered solution, which has to be volatile so we can get rid of the salts in the mass spectrometer, then we see that we get a very different mass spectrum. The mass to charge ratio is now an order of magnitude higher and we're looking at the intact virus-like particles, which are in this case T3 particles. Since we here had two species, we couldn't get charge state resolution for the intact capsids, but this is just to show you that it's possible for the hepatitis B capsids. Back then I was able to get mass resolution and resolve the two species. What kind of information can you get from native MS? First of all, of course, since you measured the mass, you can get information about stoichiometry. You can confirm that when you do insolution or gas phase dissociation. That can also tell you something about the topology in your complex where a subunit is located and using i-mobility experiments, we can also say something about the gross shape and look at large conformational changes. And we can do that in a more or less quantitative fashion because the signal intensities usually reflect pretty well the concentrations in solution. That comes in handy if you want to study other processes. For example, ligand binding, that's one of the things I will talk about today. So you can determine the ratio of the unbound protein to the ligand bound protein and determine Kd's and such. You can also study the mass of your complex in a time-resolved fashion to look at, for example, protein complex assembly. And this is just a little video to illustrate how the instrumentation works. We're coming from here with the electrospray. Then very often we have a quadruple mass analyzer where we can select out a single species. Then we can activate it by collision with gas molecules. And what then happens is a subunit unfolds and leaves the complex. Which is a nice thing because now we have two species here. This is an i-mobility cell. And I think Brian will go more into detail about i-mobility. But what happens here is you have a counter gas flow to your ions. So bigger ions experience more friction than smaller ions. It's basically like when you're cycling against the wind. If you're compact, you will be fast. If you do this, you may not move at all. And that's what we measure then. And in the end the main mass analyzer is a time-of-flight which gives us very high mass resolution. And the system we're working on a lot these days is noroviruses. You've all had it and it's not very pleasant as you all know. It's the main cause of viral gastroenteritis and depending on the literature it's about one to twenty particles that's sufficient. You find it in seafood. Oisters are very popular source of noroviruses. It's single-stranded RNA. It's non-enveloped. Which means the capsid protein forming this T3 capsid has to carry out all the function from cell attachment to uncoating protection of the genome and so on. The capsid protein is fairly large with a bit more than 50 kilodalton. It comes in a shell domain which is on its own able to form the capsid and it has a protruding domain which is responsible for the cell attachment and reinforcement of the capsid. And if you just take this P domain you can study binding to the natural ligands which are glycans on the cell surface and for human viruses, Foucault's is the minimum attachment factor which is found in histoplasmic antigens. And until very recently it was thought that each monomer had a single binding site so you would get two binding sites per dimer and that's what we started out with. And we thought this is a nice system to set up our methods and then go to other viruses. That's not quite what happened because it turned out to be more interesting. So the first bit we will look at is the cell attachment of the norovirus. This is my very reductionist view of viral life cycle in bacteria and for an envelope virus which we don't have here but never mind. And this is just to remind you this is the environment that the norovirus is facing. This is an endothelial, endothelial in a blood vessel. Fair enough norovirus is caused most of the problems in the gastrointestinal tract but they can in fact almost any cell in your body and it doesn't look that much different in the gastrointestinal tract. These hairs you can see here, that's all glycans. So the cell surface is entirely covered with that with very high concentrations here and somehow the virus has to get through that. So how do we look at glycane binding with the mass spectrometer? First of all we of course have to look at how our P dimer looks and we also have a reference protein cytochrome C here and when we then put in our histoblactobrub antigen B we get such a spectrum so it's pretty messy. You see a lot of glycans sitting on the P domain but you also see that the cytochrome C which is not supposed to bind this B antigen takes a lot of glycans along. And the reason for this is the electrospray process. We have droplets of a certain size and there is a chance that some free glycans are in that very same droplet and just right down onto our protein. That's why we have the reference protein to correct for that. But you can see that here we have a nice ladder and here we have much more on the P dimer so there is specific binding underneath this unspecific clustering. So we can correct for that and what we then get are these nice bar graphs where we see just the specific binding and what you can see here is that we have up to four B antigens bound to a dimer instead of two that were anticipated originally. And this is not an artifact of our measurements. This is a study we did on a sarcopy dimer and our colleagues used the very same strain and the saturation transfer NMR where you can measure how much glycans you've bound and what they see if they titrate in the B saccharide, they see four discrete steps. It's not necessarily the binding pockets but it's intriguing that they see four steps when we see four binding events and here's just a zoom in. And collaborators from Heidelberg did a crystal structure with a different strain though and just Foucault's but they see four Foucault's molecules in the binding cleft so there's pretty good indications that they are indeed four instead of two binding sites. We also went to another strain MI001 which belongs to the same genome group like the one we've looked at before which usually infect humans which means they bind to Foucault's but this one can also infect mice and usually if you have murine noroviruses they bind Cialic acid. So if this one can infect both the question is obvious can bind both. So we first looked whether it retained the binding to Foucault's and we tied it in the B saccharide and indeed we see it can bind and we also see the four binding sites again. Most of the times I will only show you single point measurements but we do these titrations to get better estimates of the KD. So if we then take Cialic acid containing glycan like GM3 we see that we do get binding and we tested at two different ionic strains. Initially we had started with high ionic strength but it turns out that lower ionic strength is better so we have to put in less glycan, we get less clustering so it's easier to analyze the data. We used the A antigen which is just different in this moiety I think from the B and we see again binding and for the B we see binding but a lot more than for GM3 and A. We don't see the four binding sites with GM3 and A that could easily mean that affinity is just lower as is evident from the amount of binding we see but it could also be that the additional binding sites are not opening up that something is not happening with these two. So just to summarize that again GM3 and A behave similar and B has a much higher affinity and definitely has these additional binding sites. So if we just look at individual sugars which we can't do in mass spectrometry but SCD and MR can do that single galactose that you can see here this is the reference spectrum this is the difference spectrum of sample with them without protein you see no signals which means no binding single galactose cannot bind as well as single cellic acids can't bind FUGUS can. So we thought okay let's just go to a larger galactose sugar so we have a proper negative control. So we went to this GB4 antigen which contains of anacetyl galactosamine galactoses and the glucose and shouldn't bind. Surprisingly enough it has a very similar KD for the first binding event as has the B antigen around 100 micromolar. So this is obviously not a negative control and we have tested other things it seems like almost anything that has at least three sugars in a row or branched like the B combined. So then the question of course arises are all these binders functional that can't be there must be some specificity to decide what cell to enter and to have the species trochism. So when we take a closer look at the B antigen binding so this is now again for each charge state separate before I had the sum charge state what you can see is that for the two strains we've looked at so far we see very little binding for the low charge states and much more for the high charge states and for this one it's even more evident. What is so interesting about that? Well the higher the charge state usually the higher the surface if you remember from the beginning the denatured protein which took up a lot of charges. So this could be an indication that there is a preference for a slightly more open structure for the B antigen to bind. So this could be an indication for a structural change that is ongoing and that would hopefully not be observed for these binders that are not supposed to have an effect and we're now testing that with negative effectors which are human milk, oligosaccharides and I can tell you already that if you plot this nicely you get a certain trend for the B antigen but the GB4 which we hope to have as a negative control gives a flat line we don't know whether the milk, oligosaccharides do the opposite thing yet. That would be nice of course. Okay so much about binding. Let's go to capsid assembly and disassembly because ultimately we want to look at the glycan binding on the VLPs which means we have to characterize them first. So back then when I was still doing my PhD in Utrecht we did already a study on the Norwalk virus capsid and we received from a different lab a prep and we tried to reproduce our old data here and while it looks similar we get a T equals three but the prep is not as nice as the one we had back then. So we also have a bit of T equals one in there and it's known that Norovirus can form both sizes. Maybe the preparation here has one purification step less. We don't know maybe there is a mutation if you ask the collaborators they always say we do it the very same way. We also receive different strains and we had to look how they look at neutral pH and this is a G217 Kawasaki very recent strain which caused an epidemic in Japan and also here we see T3 and T1 and you can also see in the EM some smaller particles here but most of them are T3. Then there is another Japanese strain which didn't cause epidemics but is relatively new and what we see is just a little bit of T3 some malformed particles and T equals one and that's also evident in the EM where you have lots of small particles. So then what we wanted to look at is how stable are these particles as we change the conditions. In Norovirus it was the following if we went to alkaline pH but kept high ionic strengths we would shift the equilibrium to T equals one some weird A-teamer and lots of smaller intermediates at low ionic strengths but high pH we abolished assembly and if we put this back into neutral pH high ionic strengths we would recover the T equals three. With our new prep it again looks slightly different. We do lose the T equals three signal at high ionic strengths high pH. No, it's not from patients. It's VLPs, virus-like particles. So they are producing insect cells. No patient samples. I would have six students all the time. Yeah, highly contagious. No way we would work on the real thing. So no genome. Sorry that I didn't make that clear. But yeah, we do see some intermediates. Unfortunately if we go to the ionic strengths some of the T equals one is very persistent doesn't want to go away but we do see less of these other species and if we go back from this condition to neutral pH we do regain some of the T equals three. At least qualitatively it's similar. Not quite happy yet but at least it disassembles and it reforms. If we now go to the Kawasaki strain and test different pH what we see is that the T3 doesn't care and NISA does the T1, not a lot is happening. There seems to be some smaller species here which actually do disassemble so this is the VP1 dimer, this big peak here. And we also did EM to confirm that the particles are stable and you can see more or less nice T equals three in each of the graphs. So nothing is happening to those. Not quite what we expected but there seems to be quite a difference from strain to strain but what we will now do on this one is we will add glycans and see whether our effectors that we think induce structural change actually help us disassemble these guys under multi alkaline conditions probably pH eight where we still know that the glycans can bind. This would be a proof that there is indeed something happening. With this I wanted to switch gears and I've shown you that just two minutes ago or so. Back then in Utrecht we looked at all these intermediates with our mobility to get some information about their structure and come up with an assembly model which was very nice but our resolution was very low. What we would like to see is really this atomic resolution but of course these intermediates can't be purified and that's what drove me to the European XFL and I will bother you now again with my idea what I want to do there. The European XFL is an X-ray free electron laser. Currently under construction in the Hamburg area. We start out at the Dezis Synchrotron which some of you might know was the electron gun and then we run 3.4 kilometers to the next federal state where we have the experimental hall. You can see the experimental hall down here. This is the main building and underneath the main building is the experimental hall. It's nearing completion. We had first light in the experimental hall and first user experiments are starting in September and the single particle consortium that I'm part of will have beam time in November and we're all very excited so stay tuned. We have user labs so you can prepare your biological samples or also other samples on campus. There are six instruments for biology, for meteorology, material science, femtosecond chemistry and so on. So the XFL delivers femtosecond long X-ray pulses of very high intensity. The peak brilliance is about a billion times higher than a third generation synchrotrons so we're dealing with a lot more photon flux in a short time. It's mostly coherent radiation. That's why we call it the laser and I have a little video on the next slide to explain a little bit why it is coherent and for Bogdan because he was mentioning anisotropic radiation the other day. Since it's superconducting linear accelerator, we can run at high pulse rate so in total we will get 27,000 pulses per second coming in this weird time structure so we have a 10 hertz pulse rate and then within 600 microseconds we have 2,700 pulses in a row. And with this repetition rate and the peak brilliance it's superior to other free electron lasers that are available around the world for hard X-rays which you need for structural studies that's LCLS in Stanford and Sackler in Japan and there are also soft X-ray lasers available. The first two were flash which is conveniently also in Hamburg and works with the very same time structure and found me in Italy. There's one now also in Switzerland nearing completion and there's another hard X-ray FEL in Korea almost ready. So this is a video that illustrates how the light is produced. This is our electron packet which is wiggling close to the speed of light along a magnetic structure called undulator. And now we see a zoom in into our electron cloud and if you send charged particles at the speed of light along a curved pass they don't very much like that they start to mid-light and I guess most of you being physicists understand that much better than I do. But since they travel for a certain distance aside each other they do interfere. So we get a microstructure here in the electrons which leads them to be in the same phase basically and then a certain phase within the light is amplified and you get this spontaneous self-amplified emission and very bright light. And that's also one of the reasons why we need this long structure because they have to travel in parallel for a long time to really get these bright pulses. So why would you like to do biology or look at structure at a free electron laser? Why not go to a synchrotron? So most high resolution structures usually come from X-ray crystallography where you have your proteins nicely in order to amplify your signal from the repetitive structure. But in my view proteins are more like this. They all take up different conformations and they don't like to be packed in crystals and that especially holds for my viral intermediates. So in order to image those you need very high intensities like we have at the free electron laser but unfortunately your particle will explode and the plasma is not very informative of your original protein structure. That's where the short femtosecond pulses come in handy. They will have scattered at the particle and left the particle before the explosion actually can take place and our diffraction pattern still comes from the original particle structure. But of course you can't reimage that plasma cloud so you have to send in more particles, take a lot of diffraction patterns very much like in cryo-AM classify then do 3D averaging and eventually reconstruct. Since it's possible to do single particles it's obvious that everything that can't be crystallized is of interest. So my former postdoc supervisor is mostly interested in non-reproducible stuff like intact cells. You can also look at membrane complexes but also dynamic systems, disordered proteins and most importantly transient species but this poses another problem. They only make up a sub-fraction of your data sets so it's very difficult to find them and means you have to acquire a lot of data which takes a lot of time. So it would be nice if you somehow could filter them out which brings us to the point how we get sample in. The most popular way are liquid jets like the ones you can see here and they provide atomic resolution from small crystals. This is one of the first examples published in 2013 on a protein from terpranosoma so that's a sleeping disease. It's very nice for the cereal crystallography stuff but single particles you can only image in the water window because you have a lot of water around in the liquid jet which creates a background and this is a 10 nanometer protein roughly the size of RNA polymerase so it's not a small complex but there's obviously a lot more water. So the only way you can look at that is if you go to the water window and that limits your resolution to something like four nanometers and here's an example from bacteriophage P22 and the resolution doesn't go very high out so you have to go to the gas phase somehow. There are aerosol injectors I use which look a bit like a gun and it's kind of similar but also there you have a high background for small particles because the initial droplets you create are 100 times larger in volume than your 10 nanometer object plus you have a high particle consumption and they are very difficult to pulse so you lose material while you have no beam and all the sorting has to be done online so we still face a problem that the analysis time is much, much longer than the acquisition time but nevertheless it works nicely for large things like memoviruses, half a micron across or caboxysomes. So in my view the ideal sample delivery system has a low sample consumption so time particle release we should work from a natural environment but as I pointed out no background is very important so we have to forget about the natural environment rather have a diffraction pattern and no natural environment and we would like to select species from a mixture and do some pre-sorting and speed up the data analysis and this brings us back to native mass spectrometry where we can do all this we have the nano-electro spray source up front which gives us low background or controllable background I should say and a low sample consumption this of course comes with a drawback we have little particles to hit but since we're dealing with ions we can trap them and time the particle release with the free electron lasers and make up for the low ion density then we have the quadrupole I showed you before where we can single out the species that we want to look at and thereby purify low abundant species we can go one step further and behind the quadrupole introduce ion mobility separation to pick out certain conformations and here's an old example where we first selected was a quadrupole single species and this is just a fragmentation experiment to prove that it is what it is and on a neighboring peak we separated to species with ion mobility so this is perfectly possible yeah Robert they do have slightly elevated temperature because they still experience gas collisions and take up energy but it's not straightforward to calculate what you have what you can see is if you activate too much your profiles change and your proteins unfold and if you activate too little your water will be stuck to it so we have an indirect readout I didn't I never dared to calculate it because it's really complicated but you could also additionally cool or heat if you wish to and we will also attempt to align the molecules according to their dipole this would again be helpful for the data analysis but I won't go into detail because it's a challenge on its own and we will keep the time of flight for online diagnostics so we know all the system is working we won't see the particles that we have imaged but those that we missed and we have pretty well defined now how this region should look with the trap and all the mass and conformation filters we've determined the ion flux and it's much higher than we originally expected so we should get the decent hit rate with the laser and we performed the first experiments at flash here you can see our setup so this is basically a commercial mass spectrometer into which we drilled a hole to connect it to the vacuum system of the beam line at flash we were not interested in structure because we have soft x-rays we can't get to meaningful resolution but we have a very similar pile structure so we can test a lot of things and the idea was that we would get multi-photon absorption and depending on the intensity we would go from plasma to peptide or single protein ejection sub-complex formation so that's very interesting on its own for proteomics applications and this is the kind of data we got without flash we got a normal mass spectrum and with flash we got a lot of new peaks which are hopefully peptides my postdoc had to do other things but he's now back on this data trying to assign the masses resolution was not as good as we were hoping but next time yeah I'm almost done that's perfect so with this I'd like to sum up so I hope I've convinced you that mass spectrometry is a versatile tool to study viruses for example strain specific dynamics and assembly but also ligand binding and that in the future you want to run all your samples where you're interested in structure at the XFLMS system we call it vis-a-vis that's the name of the funding last year we finally got some funds and now we're taking up momentum and finally progress is happening with this I'd like to thank my group especially Ellen who's working on the XFL project how Julia who is Ronja who work on the neurovirus and the microscopy platform who are responsible for the AM data the collaborators that provide sample and additional data the colleagues at XFL and a few companies that helped me with designing the instruments and of course all the funding sources and you for your attention I'm happy to take any first questions