 So, good afternoon everyone, thank you for coming and hello to everybody on Zoom. So it is a great honour for CJ and myself to welcome Dr. Shor Sharris who kindly flew all the way over the pond from the MRC LMB in Cambridge UK to give the international steamboat lectures today and we're going to get another lecture tomorrow. So Shor is and has been really a pioneer in the field of cryo-EM with a particular focus on the development of algorithms for image processing and really pushing forward what is sort of the limit of the technique and what is currently possible. And so I can give you a little bit about his academic background. So I apologise for my Dutch pronunciation here. So Shor stood his PhD in Utrecht with Piet Cross, in extra crystallography actually, right? And then I guess he saw the light and moved to Jose Maria Carazzo's lab in Spain to study cryo-EM and develop algorithms for cryo-EM. And in 2010 he was made a group leader at the MRC LMB in Cambridge, sort of the Premier Molecular Biology Research Institute in the UK. And really he's been at the forefront of the field for all those years. He has many awards and I'm not going to go through them but I did see that in 2014 you were one of nature's 10 people that mattered, which that's kind of cool, right? You matter, at least for a year. So that was good. In 2021 he was elected a Fellow of the Royal Society and has many other awards as well. So Shor is very well known for the development of the programme Relyon, which is the most used cryo-EM processing package. I googled it quickly before I came. There are 20,000 cryo-EM structures in the EMDB and Relyon is listed as being the processing package on a little under half of them and probably it's not listed on some that it was used on. And we're going to use, we're going to hear more about that tomorrow I believe and see some atomic resolution cryo-EM maybe. But today we're going to hear about Shor's exciting work in the amyloid fibrils and how he's used cryo-EM to really uncover some really interesting stuff about them. So thank you very much for joining us Shor's and please take it away. Thank you Tim. Thank you to you and CJ for the very kind invitation. This is my first post-pandemic international trip for work. So I didn't realise how much I'd missed all these interactions and just chatting with people about what they do and had a wonderful lunch with the students. I just wanted to say you all matter, no matter what nature says. OK, so my lab, does that not work? OK, I'll stay here. That did work, you know. All right, don't worry. It changed out. Oh, it did, yeah, excellent. My lab, as Tim already alluded to, we have kind of two aspects of our work. I originally started out after my post-doc with Xosumeria echaratha doing image processing developments also at the LMB and we write the software package called Reliom. So I'll speak more about Reliom tomorrow and today focus my lecture on the other part on the amyloid structures where we've looked mostly at tau and alpha sinuclein but also a few other proteins which I'll show you in a minute. So also this work was kind of made possible first through image processing developments and that's why we're kind of keen to have both an experimental side of the lab and an image processing side of the lab because they're highly synergistic and you develop methods that make new experiments possible and your experiments inform on what kind of new methods you need and we like that cycle to go on with all our colleagues at the LMB but also specifically in our own group. So Shouda he a tantra PhD student in the lab, he implemented helical reconstruction algorithms inside Relan which then made it possible for us to look at these helical filaments. So all the work I'll be talking to you about today is done in very close collaboration with Michelle Goodard here who's a group leader in the neurobiology division of the LMB. So I just walked down the corridor to the other division and have almost daily chats with Michelle. We kind of share the supervision of all the students and the postdocs that work on these projects and kind of LMB works slightly different than many universities in that we don't have individual group budgets. We all grab from one very big pot until the end comes inside and the director tells us could you grab a bit less. We don't have budgets and so on and it makes for a very collaborative atmosphere and it also then doesn't really matter whether it's your student or my student and so it just makes for a very very nice way to work together. So with Michelle we've been looking at this protein called tau since 2015, 2016 or so and the abnormal aggregation of tau into filamentous aggregates characterizes multiple different neurodegenerative disease. In total there's more than 20 different diseases and they are all characterized by accumulation of tau filaments. But how they do that and what types of neuronal cells and what type of aggregates you get kind of already at the neuropathological level and this is just from four different diseases, Alzheimer's disease, PICS disease, progressive sukinookia, palsy and cortical basal degeneration. You already these are all stained for tau in black or in purple here. Already at the kind of cellular level these diseases look very different even though this is always the same protein that aggregates. Now tau in its normal happy healthy life is a mostly disordered protein and then it binds to microtubules and is thought to stabilize microtubules. That's kind of as far as people have gotten with understanding what it does. Liz Kellogg in Ivano Gala's lab did a beautiful structure of tau constructs bound to microtubules which of course structures of that if I was a big expert in and so there's now some idea of how it binds but that's not what today's talk is about because what happens is this mostly monomeric unstructured tau perhaps partially bound to microtubules somehow in disease self-assembles into forming these very long filamentous aggregates and if you put these in an x-ray beam then you get this typical what people have called cross beta diffraction patterns. So in one direction you get a very strong signal at 4.7 angstroms which is the distance between beta strands and beta sheets and in the perpendicular direction at about 10 to 12 angstroms or so you get a another peak and that's more or less the distance that two beta sheets can pack against each other and that kind of forms this x-shaped cross beta it's called a diffraction pattern. That arises from these filaments being amyloids in that you have multiple copies of the protein which stack on top of each other to form very long beta sheets which then because those structures tend to twist and you get these helical type of filaments so there's basically very long beta sheet in this direction and then in each run of the beta sheets you have one or more of these protein molecules but the beta sheets themselves kind of fold over and pack against themselves with this typical cross beta packing which gets rise to this perpendicular signal here. Now then for many amyloids only part of the protein sequence forms the ordered core with the beta sheets and the rest at the end and possibly the C-termini then hangs over and forms what Tony Crowter already back in the 80s when he looked at tau filaments extracted from the brains of patients who died with Alzheimer's disease he kind of saw what he called the fuzzy coat so these are disordered parts either on the end on C-termini part of the of the of the stable core that just adopt unstructured confirmations. Cool. Now in human brain you then get six different isoforms of tau and that's due to alternative splicing of map T the tau gene and axons two and three can be spliced out or not an axon 10 here which encodes for one of the four microtubule binding repeats and these are more or less homologous repeats that there are four of in the in the tau sequence that the protein uses to bind to tau and that's the kind of constructs that that that Liz used in her structure with the microtubules so you can get alternative splicing of axon 10 you can splice out the second repeat called R2 and then you get depending on whether these these two and terminal axons are spliced out you get three four repeat tau isoforms and three three repeat tau isoforms now being a structural biology and so normally splicing wouldn't really be the first thing on my mind if you think of proteins but in disease there's this interesting observation if you now look at the biochemistry you can extract these filaments from from brain tissue of patients which which have died with these different diseases and it turns out if you've died of Alzheimer's disease or chronic traumatic ansophilopathy the disease that you get after that you may get if you have repeated brain trauma for example from boxing or perhaps american football so on i shouldn't say that because i understand the stadium is right next door but then all six isoforms are accumulated in these filaments that's not true if you if you have pix disease because then only the three repeat tau isoforms are present in the filaments and if you have some of the other diseases like like cvd or psp then you get only four repeat tau isoforms in the filaments so that's kind of interesting already in itself from a biochemical point of view and then tony kraut as i said back in the 80s in the 90s already was working with michelle using the electron microscopy techniques available back then negative stain and and software that he wrote himself and to kind of look at some of these tau filaments and already just from low resolution negative stain images he could see that the morphology of the filaments how why they are and how much they twist was different among the different diseases now much more recent work by by this japanese group they then do limited proteolysis to chip away the unstructured fuzzy code and then look at with masspexy which which constructs are protected because they would be part of the ordered core and they could see that the the extent of the ordered course would be different in some of these different diseases now that kind of together with observations that during the progression of disease you have what appears to be a spreading so the different diseases start in different brain regions and will lead to different early symptoms also disease but most of these diseases have in common that you get what appears to be a spreading throughout disease as disease progresses so all those observations together then has kind of led to this hypothesis that tau may behave in a way that the prion would do in these neurodegenerative diseases where you have templated seeding you have a filament which which acts the ends of which acts as kind of little magnets for for otherwise happily disorder protein to to join in and form and grow this filament and these filaments might then break and you have more rounds and you get kind of exponential growth and spread of tau throughout the disease and then perhaps the the different isoforms and might might correspond to then forming different structures and then the different structures that would then be like like the different tau strains that people have that's kind of part of the of the prion hypothesis with different confirmations of these filaments then are responsible for different diseases how that would work was completely unclear when we started all this work and after having had when our resolution revolution and more about that tomorrow the new software developed by shout out the new microscopes with the new detectors we thought it would be a good time in 2016 to go back look at the same type of filaments that Tony already isolated in the 80s and 90s and now do modern-day cryo electron microscopy on them so we started off with the brain of a 74 year old lady here in the US who died of Alzheimer's disease back down we would use pretty big chunks of it cortex region here and then you can use sarcastal solubilization and then you can do different steps of centrifugation to to separate out your filaments from a lot of other stuff that is in brain and the good thing about imaging is you can recognize these filaments quite easily in the images so biochemical purification you could you can kind of be relatively loose because as long as you can see the images and there's not too much muck around it you can you can still do pretty good structures already Tony in the 80s had identified there is two different types of filaments in Alzheimer's disease he called these ones paired helical filaments they are the characteristic kind of wide narrow wide narrow and these are he called the straight filaments because they lack this alternation of width so we saw the same this is a negative stain image from this brain and you can then do you can make cryo electron microscopy grids where you also see here with blue arrows the paired helical filaments and here is a small piece of a straight filament with the green arrow so using shouda software using our back down relatively new cryos with the falcon two detector i think or k2 k2 i think it was we could then calculate structures to about three and a half angstroms resolution for both the paired helical and the straight filaments and this is just kind of to give you an idea of how this looks like so this is false collet here of the paired helical and the straight now this is already a projection of the 3d reconstruction but if there were no noise in the images and then and you would zoom in on this you would be able to kind of see this is the 4.7 angstrom different layers that that you would get from the beta sheets here so you can this is kind of a view side view of the three year reconstruction where if you then zoom in if i now turn this over 90 degree i'm going to look at the filament head on and then i see these two c-shaped proto filaments as we call them and each proto filament on each layer of this this beta sheet structure on each beta round as we call it is one copy of the tau molecule and and from the end and the c-termines as we'll see in a minute they'll be sticking out this ordered parts which disappear in our cryo and reconstruction because there's no structure there now as we'll see near the end there's still potentially a lot of interesting biology but cryo cannot help you there and i was discussing the chat this morning you know perhaps NMR could be more useful to study those parts of the filaments so the straight filament is also made of these the same two c-shapes proto filaments but whereas there is a symmetry axis here a helical symmetry axis right down the middle there is no symmetry here they pack against each other in an asymmetrical way but and because this structure now is more or less as high as it is wide it kind of looks straight in projection but this structure is much wider than it's high and then it depends on whether you look at it from the top here or from the side here whether you get either a wide view or a narrow view so that's kind of that to kind of put you back into the 2d image so you can zoom in a bit and this is then the in blue the reconstructed density the three and a half angstrom that's then of enough detail to be able to recognize the protein sequence and build a unique model so there's only one model that could explain these structures it starts at valine 306 and ends at finnualine 378 and then the 300 up to 305 residues on the end terminal part of this are all floppy and the 70 odd at the c-terminal part are also all floppy in the in the disordered core and if you look at the straight filament it's it's the same two protofilaments as i said and they pack against each other in this asymmetric manner where these four lysines kind of coordinate to this blob of density which is probably not how and as of today we still don't really know what it is it's probably some negatively charged molecule or we don't really know but to coordinate all the positive charges of the lysines it could be perhaps post-translational modifications although we don't see great connections towards them but it's probably important to keep the straight filament together you can superimpose them they really did more or less the same c-shaped protofilament structures and then the kind of structural biologists call it ultra structural polymorphs so you have the same protofilament that comes together and to form two different types of dimers if you like dimers is a strange word but now then after doing that then we got people saying okay but this was one lady right with Alzheimer's disease what happens with the with the next person with Alzheimer's disease so in total we looked at 20 different cases of Alzheimer's disease in negative stain and we did three more cryoam structures from from three more cases and they're all the same since then we've looked at even more cases and and if you have Alzheimer's disease there's always phs and straight filament sometimes the ratios vary a bit etc but they tend to be usually much more phs than than straights but they're basically always the same so we think this fault is characteristic of Alzheimer's disease now the extent of the ordered chord then explains why you got incorporation of both three repeat and four repeat isoforms in these structures and that is because valine 306 is the first residue in the third microtubule binding repeat which means that the entirety of the second repeat lies in the fuzzy code that lies in that n-terminal parts which is which is this orders so whether the second repeat is present or not it doesn't matter it's part of the of the fuzzy code so you can get indiscriminate incorporation of both three and four repeat isoforms and that kind of illustrated on the next slide whether you have the orange bit of R2 or not kind of just doesn't matter and they all get incorporated now what happens then with some of the other diseases so if the next one I'd like to speak to you is chronic traumatic and syphilopathy because that one as in Alzheimer's disease had both three and four repeat isoforms and people thought you know that probably the filaments from CTE would be possibly the same as the ones from Alzheimer's disease for that reason biochemically also you know this limited proteolysis you wouldn't be able to distinguish them probably so for this we then used some more advances in reliance in a particular algorithms developed by Jacenco Zivanov to correct for errors that you may have made when you aligned the scope or in this case this was data collected at Diamond at the national facility in the UK actually the thermal fissure had left certain three treffle in the column and they came back in they aligned the column etc but Jacenco wrote software to detect this and to correct for it a little bit more about that tomorrow but it allowed us to go from 2.7 without correction to 2.3 angstroms so we looked of three different cases at three different cases of CTE one ex-professional American football player and two ex-professional boxers and as is with Alzheimer's disease if you have CTE different individuals with CTE all had the same structures and again we saw two different types of structures who recall this is CTE type one structure and this CTE type two which we only detected in case one and two not in case three but as you can see the structures are somewhat different from the Alzheimer's disease structures so actually the extent of the ordered core is very similar so this one has what perhaps one more residue at both the end and the C terminus but otherwise more or less the same also the morphology is similar and and I here compare now only the the protofilament fold so colored by the same residues you get that more or less a similar topology but whereas the it was truly C shape structure in Alzheimer's disease in CTE it becomes much more of a straight structure and also if you if I go back and look at the density especially here in the 2.3 angstrom map and this is done where resolution starts to matter you can see there's there's a density inside the kind of of the cavity of the loop that fold where the whole fault folds back on itself again and even at 2.3 angstroms that does not connect at all to the tau the to the density of the side chains for tau so we don't think this is a post translational modification and then looking at the cavity itself it's all rather either hydrophobic residues or there's two serins in the cavity so we thought you know perhaps there's some sort of hydrophobic cofactor that coassembles with CTE filaments that might be important to to kind of determine the difference between the AD fold and the CTE fold we still don't know to date what this is either although we're getting more insights and I'll come back to that near the end this pointer is a bit slow now then what happens to tauopathies where the actual biochemical composition of the filaments are different so the first other disease we done looked at was a pig's disease where I told you you only have three repeat tau isophoens which assemble into filaments so this is a picture of the pig brain that we used and instead of these kind of raindrop like tangles tau tangles in Alzheimer's disease you have these more like spherical pig bodies in the in the in the in the in the neuropathology the histology here and again we saw from the filaments there's two different types what we call the pig narrow filaments and the white filaments and the white filaments we never got beyond kind of eight or nine angstroms resolution there weren't too many of them and they may flop a bit but for the narrow filament we got a 3.2 angstrom resolution structure and I hope I can convince you that we probably think that the white filament is made up of kind of head to head dimers of the of the narrow filaments although we never could build an atomic model for that we could build an atomic model for the narrow filament and Benzy was quite good if we look at the assignments now what is part of the ordered call it starts here at lysine 254 which is part of r1 and then remember only three repeat tau isophoens have incorporated in the tau filament so all the filaments then go straight from lysine 274 to valine 306 so the numbering is a bit confusing here perhaps but and that happens then right here but so what you can see now because part of repeat one is present in the ordered call you can now compare the the homology between r1 and r2 there's there's these three different residues and you if you would try and place a four repeat tau molecule in this filament you would get clashes and things wouldn't work out so again the the extent of the ordered core it now explains the biochemical composition of the filament so the four repeat tau molecules would just not not be able to adopt this type of structure but of course overall this structure now looks this termingly different from the ad and the cte structures right it's it's it's a it's a very elongated we go with kind of j-shaped protofilament and all the interactions are very different you know in between ad and cte there was still some commonality but that's all out of the window with the pig's disease now since then we've gone on and we did structures of corticobasal degeneration and then and then we kind of kept doing new diseases and then last year we just decided to do one big paper and kind of put them all together and that then led to this kind of structure-based classification of tautopathies so cbd agd psp ggt these are all diseases which have four repeat tau isoforms only in their inclusions and that is because repeat two which is always colored in in in light blue is now an intrinsic part of the ordered core and then you can do the opposite as what we did with pig so if you now put three repeat tau isoforms where the r1 which is colored in purple there would be residues that just wouldn't fit and then you couldn't you couldn't form these structures but you can see there's there's multiple different structures they have some things in common right so there's there's especially the the cyan and the green bit they kind of look similar topology wise but the details always different so what we found is that distinct tau faults always define distinct diseases so it's not the opposite so if you have different diseases they could have still the same fault for example Alzheimer's disease and familiar brain ish or British or Danish dementia all have the Alzheimer fault and the same is true for this fault which we call the agd fault if you have r tag or some some splicing mutants of tau they also adopt this fault but distinct structures do always define different diseases and we now have an explanation of their biochemical composition and we can do this sort of classification where we have a second level of of kind of we call this four-layer faults and these three-layer faults so PSP and CBD which based on neuropathology and based on clinical symptoms sometimes had been grouped together actually much more different from each other at the molecular level than for example agd and CBD are and so these are new findings from from this kind of classifications now also for these diseases you know you can look at multiple individuals and they all have the same structure so the multiple individuals with the same disease all have identical structures and if you now look I will I will kind of point out some of the densities which which are inside or outside of these filaments which are not explained by the by the tau molecule main chain or side chains so these could be other factors like like this hydrophobic potentially hydrophobic co-factor that we saw in the CTE structure we see a potentially hydrophilic co-factor because it's surrounded by lysines inside the CBD fold but also on the outside of the Alzheimer's disease I mentioned the one that holds the the straight filaments together there's all these densities which are relatively fuzzy they never really connect to tau we're not entirely sure they are they would be post translational modifications some people have tried to claim that based on some mass spec data but we're not we're not convinced at all yet but it could be that that either post translational modifications and or other molecules are important in defining what happens in the different diseases why do you have the same protein with the same sequence adopting all these different structures in these different diseases something specific must be going on oh sorry so these are this is this was just a I went by the co-factor of the CBD now for one of the cases that was was diagnosed as progressive sucronucleopold we did in total seven cases of PSP and six of them all had the same what we now call PSP fold but this case had a different structure and it had it there's also a co-factor involved you can see some positively charged residues pointing towards it but it was different from the PSP structure if you look at it in details for example you can see the c-terminal bit in red is pointing upwards in the in in this structure where it's pointing more or less downwards in the PSP structure so based on that observation and on the observation that that until then all always it helped that different folds identified different characterized different diseases we think this meant that we actually identified a new disease entity so we then went back to the neuropathologist and the look back at the clinical profiles and they then thought oh perhaps this case wasn't actually typical PSP and there were some atypical observations especially in the neuropathology and so now they're kind of going back and trying to find more cases of what they suspect now is a new disease so you can use cryoam to identify new diseases isn't isn't that cool okay so similar observations we've now made for for all kind of for different proteins that and so if you have a given disease because tau characterizes all these tauopathies but of course alpha sinuclein is characteristic the aggregation of alpha sinuclein is characteristic for Parkinson's disease dementia lewis bodies multiple system atrophy and then for example Alzheimer's disease you also besides the accumulation of tau you also have aggregation of amyloid beta peptides so what we've seen now is that different individuals with one disease always have the same structures but possibly different diseases are characterized by different structures so for alpha sinuclein we looked at five different cases of multiple system atrophy msa and their their structures and we saw there's two different types of filaments in msa as well and some individuals only have one and some have only the other and in some it's a mixture but msa is characterized by these type of structures of alpha sinuclein when we try to do the same with dementia with lewis bodies which has alpha sinuclein filaments as well we saw the filaments were mostly untwisted and probably different from the msa although that still needs to be confirmed because we've never been able to solve the structures and then down here very recently this january young young in the lab she was able to isolate with new extraction procedures bonafide a a beta 42 peptides filaments from a range of different disease brains and if you have sporadic Alzheimer's disease you always you tend to not always have these and if you have familiar forms of Alzheimer's disease or other diseases you can have a beta accumulation together with alpha sinucleinopathy or so but then the the filaments tend to look like this i mean why this all is we don't understand at all so i'm we're kind of just gathering data and images and and trying to to to understand this now in all these efforts we then also found some filaments and this started out in in some cases of a rare familiar tauopathy where most of the filaments were not tau at all and we solved their structures for a few months in the lab we called them the mystery filaments because you know you have a three and a half angstrom structure it obviously wasn't tau but we had no clue what it was it wasn't until alexa mursin who is our walking protein structure encyclopedia in the lab alexa developed the scott database for a classification of protein structures perhaps some of you know but he's seen many protein structures and he read this paper about this transmembrane protein called team m106b and he recognized there was four glycosylation sites and he kind of recognized he had recognized already previously that there were there were four different glycosylation sites on these filaments just from kind of looking at the density and then it all clicked together and he could build the the atomic model of team m106b just straight bum into the map so and then we had we hadn't really understood because normally these these densities are linear you know you have an n-terminal part of the ordered core and then it can do whatever complicated things and then it comes out at the c-term this this structure appeared to be had a cyclic kind of density in it so we thought oh perhaps you know our reconstruction wasn't really right or so but we hadn't kind of really found out but then it turns out there's actually an intramolecular disulfide bond in the team m106b which beautifully fit into place so that was one of these familiar hypotheses then we look then we suddenly started seeing these filaments in all kind of different diseases and that kind of made us worried that this would be an artifact we then ordered some control brains from the brain bank or varying age and what we saw already in western both with with with antibodies against team m106b was that all the younger controls did not have this but some of the older controls and this includes a 101 year old person who was completely cognitively healthy when he when he died of pneumonia or something or perhaps just of old age i don't know but you know they also have team m106b filaments in their brain so and then we even with the antibody that we'd raised against the c-terminal domain of team m106b we could do even immunohistology we could solve the same structures from older brains but not in younger brains and that kind of coincides younger brains don't have this pathology whereas the older brains in the many of the disease cases but they're all old as well and the older controls we do get these kind of i think the neuropathologist tell me this is kind of astrocytic type of of of inclusions and we have one kind of odd disease case which is a tragic case of a 15 year old girl who died of early onset dementia with lewie bodies and in that case we could not find any team m106b filaments so the current we don't really know what team m assembly means whether it if you have a disease whether it interacts with disease and makes it worse or not you know there has been some genetic risk factors for some of the tdp43 opethes but what we do think is that team m106b assembles as you age so how it then affects disease we don't know but we don't think it's the primary cause of disease anyway that's that's a little sidetrack so we've had this strange observation that each disease is characterized by a different fault and this is very specific and reproducible for the different diseases so something specific must go on at the biochemical level for these different structures to form and we would really like to understand so what's next one thing we can now do is we can you know people are quite keen to develop a positron emission tomography ligands to to to image the accumulation of tau in living patients all this you can only do from post mortem brains so and people are quite keen to develop ligands that are specific for the different diseases so you could hope to do sort of structure based drug design now and kind of to show that this is possible yang shi did a compound from aprenoya with and without ad filaments and you can calculate difference maps and then at quite high sigma levels you get different density so we now kind of know where these molecules bind and and the computational chemist can have a have a field day with this personally we don't really want to go into that direction we're much more interested in the molecular mechanisms that would underlie disease so what leads to these different filaments in diseases these cofactors or post translation modifications etc we really like to find out and in order to study this there's only so much you can do with post mortem brains it's kind of experimentally that system as well because you can't perturb the experiment of filament formation so we thought it would be important to go back and try to do in vitro assembly with recombinant tau to be able to gain understanding of what is important for generating these different filaments and then ultimately what we hope is to use knowledge gained in this to then develop better cellular and and and possibly even animal systems for the different diseases we've looked at a tau mouse that that michel has in the lab it structures from that it's bit preliminary but again the structures look very different from any of the human diseases so there's a problem with model systems in general in the amyloid field that because these proteins are so structurally versatile you you don't know whether the model system you have is actually the relevant one and i'll briefly show you some examples of that so back already in 2019 when john in the lab she did heparin induced so you can make tau filaments recombinant in ecoli they will be staying solution forever it's very soluble protein but if you add a bit of heparin it will readily come out of solution to form filaments yet the structures again looked very different from any of the ones that we've seen in disease so sofia leuvestam who started in the lab as a phd student a bit over two years ago now set out to to try and replicate the structures observed in alzheimer's disease from recombinantly overexpressed protein in ecoli and the key to this was to cut off part of the and terminal and the c-terminal fuzzy code so again the the full length protein will stay in solution and this was already known if you could if you start cutting off and in c-terminus then certain constructs grow readily from from filaments people have used constructs which by the names of k18 and k19 which were cut right in the middle of the ordered core of ad filament so we already knew those would never be able to form ad filaments but if she cut at 297 to 391 a construct also investigated already before by louis serpo in the uk then adapting the protocols that louise had developed which did not give ad but after some tinkering around sofia was able to make ad filaments first at relatively low purity but again by further optimization she she now can reliably from recombinant tau make 95 pure phd filaments in the test tube and the structures she did then to confirm by cryo that these are indeed exactly the same structures that you see in disease even some of the you know these fuzzy densities that some people have said are post translational modification at least in this prep i know for sure they're not post translational modification because this is an ecoli ecoli prep they're probably just the accumulation of negatively charged phosphate ions from the buffer in front of the positively charged lysines which which point towards these densities that doesn't mean necessarily in disease it's not post translational modification but it might very well not be now interestingly if she would add certain salts to the buffer then she could get different structures and the addition of sodium chloride in specific was very interesting because then suddenly she saw type 2 filaments as were exactly the same as in cte including the the little density at the head of the of the protofilament fault so then by changing the ion the positive ion lithium potassium she could change the the the shape of the structure so you it would be either the c shaped without from Alzheimer's disease without any of these monovalent cations but if you added the monovalent cations you would get opening up just like you you saw in cte and and the different cations lead to different degrees of opening up and different densities there we then used a prototype microscope with a falcon for detector and as electric sex energy filter and a cold fag in in in microscope in eindhoven in the Netherlands a microscope i'll tell you more about tomorrow with the atomic resolution reconstructions that we did but we did that now to do a 1.9 angstrom structure of these amyloids in the presence of potassium chloride and at 1.9 you can see beautiful density for what we believe are now potassium and chloride atoms and if you look at densities kind of all makes more or less sense so that all together then makes us think that probably probably in the sorry in the cte like structure that we get after addition of sodium chloride there's additional density here which isn't resolved to such high resolution but we now think are also pairs of of sodium and chloride atoms now that's not proven yet but there's not much else in the buffer present it's basically a phosphate anti-phosphate buffer with every sodium chloride so does that mean that in cte it's sodium chloride atoms in inside that cavity that we don't know and we've tried some experiments you know positron induced x-ray emission experiments with alspeth garmin in oxford to see you know is there more sodium or more chloride in cte extracted filaments than in ad extracted filaments but brain samples are up to some extent always dirty and we haven't been able to come to any conclusion yet but it's kind of an interesting observation that in vitro just the addition of sodium chloride can change the fault from an ad fault to a cte fault so sofia did a extensive search of which constructs to to use for this you know is 297 and 391 are they kind of special no on the end terminus you you have quite a lot of flexibility on the c terminus a bit less so and then interestingly if you if you either leave on the entire fuzzy code at the c terminus or the entire fuzzy code at the end terminus then you don't get any filaments as you would have with full length tau but if you then in the c terminus she'd looked at the paper by Judith Steen who'd use mass spec to identify post-translation modifications that occur on tau in Alzheimer's disease and she said oh there's these four that I would like to look at so she made phosphor mimetic mutations of her tau construct on the c terminus and then she could get filaments from the from the full c terminus construct just by including these four mutations and they don't form phs but they do they do adopt the c-shaped protofilament fault from Alzheimer's disease but they come together in different ways or they stay single protofilament so not yet beautiful phs but we now think that post-translational modifications in the fuzzy code can actually affect what type of structures you form in the ordered core and this is something I mentioned to you this morning chat where I think you know it would be great to look with with nuclear nuclear magnetic resonance you know what what do these kind of alterations of the of the sequence mean in terms of what happens to the fuzzy code now in doing all this Sofia managed to solve 76 different structures in the span of one year and these are some of the structures that are all different from any of the disease related structures we had observed before so this is all tau all assembled in vitro different construct different phs and salts etc just to kind of bring home to you the the observation that proteins have evolved with an amino acid sequence to adopt one fault or perhaps no fault with intrinsically disorder proteins like tau but the same is not true for amyloids you know there is not a this is the amyloid structure of this protein I think is the wrong way of thinking about it because the same protein can adopt many many different confirmations so that took her quite a while to submit that all out to the EMDB and she's provided some feedback on how that process could be improved but it kind of shows that what we now call high throughput cryo-am and I don't want to put numbers on how many you can do per time unit but you know we've now reached a stage where just solving a cryo-am structure is no longer the main goal of a project it becomes a tool that you can repeatedly use in in a project to find out certain questions which I think as my techniques mature is where we really want to get to so that brings me to my conclusion slides amyloids are structurally versatile if there's only one thing you want to remember please remember that but despite that observation in disease something highly specific happens and I think it's very important that we find out why and how that this happens if we are to kind of understand the molecular mechanisms of disease hopefully in the end being able to intervene in in the process of disease so in vitro assembly is starting to yield insights there I think Sophia's work is an exciting start but there's a lot more to be done and high throughput cryo-am will be continue to be a very important tool we're also starting to use NMR ourselves now and I've alluded to some of these things already so where this is then lead to what I hope is that will from these in vitro studies will be able to gain knowledge you know it's these type of post-translational modifications or these types of cells which have these types of molecules where you get done these types of filaments that this type of knowledge ultimately will lead us to better cellular model systems and ultimately better animal model systems that we can have model systems that reflect what happens in the different diseases better then we can start to do you know screen for ways of of how to intervene in those cool that brings me to my acknowledgement slide all this as I said is done in close collaboration with Michelle he's partially covered by Akis here so all these people we worked with together on a daily basis and Ape Kutetscha at Thermofficial Scientific in Eindhoven uses kind of this case as for their company to prove you know the high throughput cryo-am that's what the pharma companies all want them want to invest in so he uses our samples I think Sophia this week is sending Ape and another 100 samples and he'll put it on his scope and go all the way automatically through so we collaborate with a lot of people who provide us with these brain tissues so a lot of neuropathologists or samples already purified from brain tissues and without those people we couldn't do any of this work so I thank you very much for your attention and we'd be happy to take any questions. Thank you Joth that was great and thank you to the audience for not writing at the phrase American football. So has anyone got any questions they would like to ask? Beautiful lecture thank you very much one of the features that to me is hard to understand is that between protofilaments you have several examples where the surface area of contact is extremely small right yet these appear to be stable two filament structures are there any insights into what is stabilizing that and how I mean because it's a very small steric zipper or maybe a few hydrogen bonds how can that hold together two large filaments do you understand that? Not really no and it's very specific as well so for example the Alzheimer's fold we've now if you looked in detail some of the structures that Sophia did there was quite a few which have something akin to the Alzheimer's fold but many of them pack against each other very differently for example this KCL structure had like a trimer rather than a dimer and we have no clue why that is and so the interactions between them you're right some of them are very seem very weak you know PHF is it's just backbone kind of interactions possibly a bit of hydrophobic interactions they're quite tightly packed again the distance between them is very small because there's through some glycines but why that is so strong that you always see these PHFs because we don't see any single protofilaments in of tau in Alzheimer's brain for example. And so you don't have sufficient resolution to see if the main chain exchanges between the two filaments it doesn't it doesn't okay we do have the resolution to see that down. So I'm curious about the variability among tau amino acid sequences across the human population and if you've explored these this variation with respect to variation in structure. Yeah so there are some mutations in tau they're quite rare but you have familiar forms of neurogeneration which are caused by mutations in the tau gene so there are some of these intramutations which lead to different splicing and probably they just lead to more four repeat tau isoforms relative to three repeat that seems to lead to disease and some of the structures we did were from some of those cases I think the familiar British and Danish are also mutations that are at least not in the ordered core but and that but then there is one residue P301 so that's in in repeat 2 that's kind of a famous one so you can have that mutate to s or to l or to t and then you get neurogeneration which is familiar. Now the Michelle's mouse model which I mentioned is actually P301 s mouse model because of tau itself being quite soluble not easily to form filaments in vitro but if you mutate P301 to s l or t you get a spontaneous formation of recombinant tau in vitro and a lot of model systems use this you know mark diamond has these biosensors cells which you can put some filaments at the outside and then they get taken up and you get massive amplification of amyloids inside the cells it's kind of used as a readout for infectivity and so on so those are based all on the P301 s system but Michelle's P301 s mouse we did the structure now of we haven't published it yet we're writing it up but the structure is again different from any of the other diseases now we don't have human brains of P301 s because I don't think there's any available anymore but we do now have P301 l and P301 t brains and again that's not published yet but they're they're probably different again from the P301 s mouse and different from the things we've seen before although it's not completely finished yet that works so yes so you see all these different forms you've seen all these different forms now that you haven't been able to link to disease do you think that there are those diseases exist somewhere or there's some reason why those cannot form and form disease is that a yeah I got that question a lot so I think you know people do experiments in vitro and we all wish they were relevant but I think just some experiments we do in vitro are not relevant so there is not that many topopathies left to do I can't exclude that some of the structures we will find we have stumbled on in our in vitro work but I suspect that most of the structures I showed on one of my last slides by Sophia are all artifacts so uh you showed uh two structure two different isoforms of tau fibrils in CTE were those uh in two different isoforms within the same no no two different proto filaments oh two two different filament types and the proto filament structure is always the same so that's that's something that not in msa but in all the topopathies we often see at least two different types of filaments but they usually arise from different inter proto filament packings with the common fold of the proto filament so were those both found in the same the same brain or among okay yes so we looked at three different brains and uh two of them had both types and then in one of them we only could see type one okay do you have any idea of why that might be the case no okay no so different types of filaments ratios do vary for msa for example we have two different types and we have some individuals with only one and other individuals with only the other and that's kind of interesting we we could use that for in vitro seeded assembly and you know you've probably seen the the paper where we could then get different seeded reactions of the different types and it's the only way we now have now to be able to separate them was the luck to have found an individual with only type one or with only type two so have you been able to get any information on um structural heterogeneity a long one filament I guess I'm particularly thinking at the ends and if the the ends kind of look like the middle yeah so the type of averaging that we do kind of precludes us from much to looking into that much because the averaging we do assumes that each of these layers is exactly the same so if you want to study the structure at the end you would have to let go of helical symmetry and the number of particles you would have would tumble down quite dramatically because then each end is is unique right so we haven't really tried to do that but yeah one could in principle try I can get around this edge I always get good exercise daily hello um so this feels like kind of a dumb question but towards the end you're talking about the role of potentially post translational modifications um promoting that assembly of aggregation and filamentation but I was wondering if would it be possible that the presence of a filament themselves could be promoting more filaments to be formed and then the post translational modifications are just being signaled to tend to that yeah so the presence of filament definitely promotes more filaments to be formed so that's the whole kind of underlying principle of this prion like seated aggregation right and for example like I already alluded to it so for alpha sinuclein we did that experiment where we took different cases with msa and took a little bit of the brain the purified alpha sinuclein filaments from brain to add it to recombinant alpha sinuclein and under the experimental conditions that we chose the recombinant protein on its own would not assemble but you add a bit then that acts as a seed of which you get growth of the filaments and then that's the kind of thing what I said you know they could break and you get exponential growth so you if you add tiny amounts of seeds you then get very rapid assembly of the recombinant alpha sinuclein but Sofia did the structure of the products of those and if you use the case which only had these type 2 msa filaments she could kind of replicate half of the structure so we think that also here there is a cofactor in the middle that may have been missing from our in vitro assembly reaction because we don't know what it is which could be an explanation why we only get half of the structure but for other cases which have mixtures of the type 1 and the type 2 we then get all these structures here shown in green and and blue which are very different from the actual structures of the seeds that we put in so yes the addition of filaments accelerates growth but it's not necessarily along the type of mechanisms that people have thought that you get templated seeding of the same structure that then faithfully replicates what you get in vitro that doesn't mean that if others don't repeat the experiment on the different conditions they also don't succeed i'm just saying we didn't succeed so it could happen that you don't succeed that's as fast as i move i was interested in this heparin catalyzed filament formation in vitro so is that also just a heparin also required in the presence of the different salts are those still heparin catalyzed in the in vitro system so all of the structure sofia did there was no heparin there was no heparin no heparin because if you add heparin heparin is beautiful it's very long kind of polymeric like molecules with lots of negative charges right we think is all the lysines on the on the tau sequence of the ordered course kind of you kind of get an entropic effect where you it acts as a glue along the helical axis kind of keeping everything together so there is again fuzzy densities on the outsides of the cores of the heparin structures which suggests that heparin binds kind of and forms like a ruler on which to build so it becomes a structural component of heparin to anything it will assemble like crazy so it's like a no heparin in the brain right but there is but heparin is is basically a nucleic acid analog in its structure so there's other molecules in the brain which could can has anybody tried just adding nucleic acid instead of heparin you can add RNA as well yeah so we did the structure of that I don't think we it's published yet but it's it's again a different structure no I think one of the structures and what RNA was it oh I don't remember yeah some probably something but potentially that could be acting as a seed in the in cells yeah so people have hypothesized that binding of RNA to these filaments could could promote growth etc but yeah and it does so in vitro what role plays in disease I don't I don't know yeah okay one more and then I think we've had a stop thanks yeah I wanted to follow up on Dave's questions about the in vitro work I'm curious how stable those structures are or can they be disassembled and can you add a different salt to then reassemble them in a different structure have you done anything all these amyloids are quite some of them are very stable in vitro you can't reflect the in vitro ones most of them are also quite we haven't done extensive studies on all of them because at some point you know there's two life is too short and there's too many experience you could do but the ones in disease for example you you can boil if you boil them you can take them apart and you boil them in SDS you can get monomers on but you can add salt to them and just adding salt they wouldn't come apart so I think what Sophia did try was you know she has the structures with salt formed with salt and without salt so she used those as seeds in the conditions of the contrary experiments so she used the salt formed structures to seed aggregation without salt and she used the non-salt containing structures in seeded aggregation with salt and I think one of those experiments did replicate the same structures but the other experiment did not replicate the structure of the seed but the replicated the structure of the condition but I I'm sorry I forgot which which way around it was yes early days yes okay thank you very much let's all so that there is now a reception outside as well so