VIDEO: Dorothee Dormann on Aberrant Phase Transitions of Neurodegeneration-Linked RNA-Binding Proteins: control mechanisms and functional consequences

Dorothee Dormann (00:02:54):
And you find similar scenario in frontotemporal dementia patients, but just different brain region is affected and therefore patients have different symptoms. So here about 50% of the patients show this TDP aggregation, and in roughly 10% of the patients you have FUS deposited along with some other related hnRNP proteins.

Dorothee Dormann (00:03:16):
And then just for the sake of completeness, I want to mention briefly the other proteins that you find in the subset of patients as aggregating proteins. So in ALS this can also be SOD1, which when mutated can lead to SOD1 pathology. And in frontotemporal dementia you also find cases that have Tau aggregates. So these are the non-RNA-binding protein related pathologies.

Dorothee Dormann (00:03:39):
And interestingly a few years ago it was found that also in up to 50% of Alzheimer’s patients you’ll find a co-deposition of TDP along with, there were more well-known neurofibrillary tangles of Tau and it was shown that also the presence of these Tau aggregates in Alzheimer’s disease correlates with a more severe neurogeneration. So it seems to be also contributing to the pathology of Alzheimer’s disease. So therefore it’s very relevant that we understand how this RNA-binding protein deposition arises and how it actually causes disease. And this is really what we want to find out.

Dorothee Dormann (00:04:23):
So what has fascinated me since I started working on this about, yeah, 15 years ago in the lab of Christian Haass, as a cell biologist, is I found it very puzzling that these proteins are normally predominantly found in the nucleus and I’ll show you in a minute what they do there. However, in the disease cases, in all of these diseases you find them translocated to the cytoplasm. So the nucleus is more or less cleared, you don’t find them anymore in the nucleus and you find them deposited heavily in the cytoplasm. And so I was immediately fascinated by finding out how this translocation and mislocalization happens, and so this is what I’ve been working on since.

Dorothee Dormann (00:05:09):
And so I’ll show you on the next slide, just from a review article, what is known by now about the physiological functions of these proteins. So they basically are known to be involved in all steps of RNA metabolism, so they regulate gene expression. So from the beginning on, so from transcription they regulate various aspects of transcription, also DNA damage repair, but then in particular also splicing regulation, also microRNA biogenesis, long noncoding RNAs. And also in the cytoplasm where they’re less prominently found, they can stabilize certain RNAs, or they can be involved in local translation or RNA transport and local translation.

Dorothee Dormann (00:05:54):
So from these various functions, you can already guess that when you have this translocation from the nucleus into the cytoplasm and the cytosolic deposition that a lot of their normal nuclear functions don’t properly work anymore and you have a nuclear loss of function. And this is exactly what you have in these diseases. So we believe that a large proportion of the downstream mechanism, what goes wrong is the cell, is really due to this nuclear loss of function and abnormal RNA processing which occurs when these proteins are not localized anymore in the nucleus. And also there could be sort of aberrant gain of function mechanisms from having just too much of these proteins in the cytosol or having them deposited in a nonfunctional aggregated form. And so this is believed to drive the neurodegeneration in these diseases.

Dorothee Dormann (00:06:48):
Now what fascinated me, as I said, and what we are now mainly working on in my group is how this is actually happening. So what are the individual molecular events that lead to this mislocalization to the cytoplasm and this cytosolic deposition? We’re also interested in finding out what cellular processes get disturbed when this happens and so how does this actually cause disturbances in cellular homeostasis and neurogeneration. But we are mostly focusing on this aspect. So this is what most of our research projects revolve around, finding this out both for FUS and TDP and related RNA-binding proteins. So yeah, so just to introduce the questions that we are asking in my lab.

Dorothee Dormann (00:07:36):
So I want to step back and then tell you about basic fundamental disease mechanisms I found back in Christian Haass’s lab and that clue about what could be going wrong in these diseases came from genetic mutations that were reported at the time. So at the time, mutations in the FUS gene that cause familial ALS were found and you see that some of them are located in this N-terminal low complexity sequence, but a large number of them was also found in this C-terminal region, which at the time was not known what it was doing. And since it was a very arginine-rich, positively charged signal, I speculated that this could be the protein’s nuclear localization signal and went ahead and tested that hypothesis and this turned out to be the case.

Dorothee Dormann (00:08:32):
So we could show back then that this is indeed the protein’s nuclear localization signal. It’s not a classical one, but a non-classical, so-called proline tyrosine NLS, which is bound by the nuclear input receptor transportin, or TNPO1. But when you have one of these mutations, the transportin binding is weakened and then the nuclear import cannot happen properly because transportin1 is one of those nuclear import receptors which belong to the importin protein family which are known to bind to cargo proteins in the cytoplasm through an NLS sequence and then shuttle cargoes into the nucleus. And so when this affinity is weakened the import process is just much slower and much less efficient. And this is what is happening when you have such a point mutation.

Dorothee Dormann (00:09:24):
And interestingly, there were mutations that had a different degree of severity, there were some disease-linked mutations like these ones, when we quantify this carefully, they only caused a very slight nuclear import defect. So as seen from this simple transfection into cells and then quantifying the nuclear versus cytoplasmic ratio. But other mutations, for example, this P525L mutation caused a very severe nuclear import defect and cytosolic accumulation of the FUS protein.

Dorothee Dormann (00:09:58):
And interestingly there was a correlation about the reported age of onset with the cytosolic mislocalization. So back then that suggested to us that this deficient, defective nuclear import of FUS is really disease causing. And so later on there were even stop mutations or frameshift mutations identified to completely truncate this NLS, the region, and this leads even to juvenile ALS onset. So the more severe the nuclear import defect is, the earlier the patients get the disease.

Dorothee Dormann (00:10:35):
Now what you can also see from this, and that sort of was the next question that we asked ourselves back then, so you can see that when you have such a point mutation, the FUS protein is not properly imported and accumulates in the cytoplasm, but it really sort of stays diffuse; it doesn’t immediately aggregate. And so we wondered what makes it then aggregate in these patients, why do we end up with these large cytosolic inclusions?

Dorothee Dormann (00:10:58):
And we noticed back then that whenever we stressed cells that had such a cytosolic FUS accumulation due to an NLS mutation, that when we stressed them for example with heat shock or oxidative stresses that then the cytosolically mislocalized proteins started heavily accumulating in these clusters. And at the time, initially, we thought they are like irreversible solid aggregates, but this wasn’t the case, they were actually, this was reversible when we took away the stress.

Dorothee Dormann (00:11:27):
And so then we, by reading more about RNP granules and so on, we figured out that these could be stress granules, which at the time were obviously already widely studied. And so there were nice marker proteins known for those stress granules and by co-staining with various of those stress granule markers, we could show that these accumulations of FUS that we see up after stress are stress granules.

Dorothee Dormann (00:11:59):
And so that led us to the hypothesis that when you have too much of the FUS protein in the cytoplasm due to an NLS mutation, then cells experience cellular stress and forms stress granules that then FUS heavily accumulates in stress granules and then you have a very high local concentration of the FUS protein in these granules and that those granules could really be the precursors to the pathological aggregates.

Dorothee Dormann (00:12:25):
And in support of that model, we collaborated with neuropathologists, with Manuela Neumann and Ian Mackenzie and they stained brain sections of patients, FTD and ALS patients, with markers of these stress granules like the Poly(A)-binding protein shown here. And they could find that a significant proportion of the FUS and TDP inclusions actually colocalized with such stress granule markers. And then Wolozin’s group had similar findings around the same time, that there is a colocalization with TIA1, for example. And so this supported the idea that stress granules could be possibly precursors to these aggregates. Now, so this was what we knew basically more than 10 years ago.

Dorothee Dormann (00:13:11):
And then at the time phase separation or condensation was not so widely known yet in biology, but it just emerged at the time and a few years later basically the model emerged that stress granules could be forming through phase separation. And so then we had this working model and then by reading about membraneless organelles and that they could be forming through phase separation, we got interested in looking into this more.

Dorothee Dormann (00:13:42):
And so this is, obviously to this audience, not a new thing. You all are familiar with this concept. But at the time it was a new concept that became clear through work by Tony and then Cliff Brangwynne and others that stress granules are forming through phase separation and there were observations that these physiological liquid granules can transition into a solid state. And so this led us to speculate that these aggregates that we thought were forming through stress granules could actually be forming through such aberrant phase transitions.

Dorothee Dormann (00:14:27):
And so then this paper, this seminal paper from 2015 from the labs of Simon Alberti and Tony Hyman was published and they used FUS as a model protein to show that this protein can form physiological phase-separated droplets at physiological concentration. So this is roughly the physiological concentration of FUS in the nucleus. So it’s in the low micromolar range. And they observed in this nice in vitro reconstitution assay, as you probably all know, that over time FUS ages into these solid fibrils.

Dorothee Dormann (00:15:04):
And they showed that this was concentration dependent. So this is accelerated at higher concentrations. And so we figured that this could be probably what could be going on in these stress granules, where we had shown that FUS heavily accumulates when you have too much FUS in the cytoplasm and that this is exactly what could be going on in this process leading to RNA-binding protein aggregation. So that was really a very nice working model at the time.

Dorothee Dormann (00:15:35):
And so since that time we have expanded our working model, how these RNA-binding protein aggregates may be forming. So we now believe that the first step is this nuclear import defect bringing the proteins into the cytosol, having elevated cytosolic concentrations and that can then promote their phase separation And also obviously external stress could help to trigger that and that then aberrant phase transitions could be going on in these granules where you have high concentrations of these proteins.

Dorothee Dormann (00:16:09):
And so since the time my lab is very interested understanding the phase separation of FUS and also TDP and related proteins better. And we are trying to figure out what are the factors that suppress this phase separation process or what are the factors that promote it in disease. And more recently we’ve also gotten very interested in thinking about what is the physiological relevance of phase separation of these proteins, why have they evolved to do this so efficiently? Could this have also functions in their normal job as RNA-processing regulators? Because presumably they have adopted this behavior for a reason and for a functional reason. And so that’s also something we are trying to look into now.

Dorothee Dormann (00:17:00):
And now I want to show you some of the major findings we had in the last years and some of the factors we have found to suppress or promote phase separation. And how we into came across one important regulator is by looking more closely at how FUS actually phase separates. So at the time it was mainly known that this long intrinsically disordered LCD region was driving FUS phase separation, but the contribution of this C-terminal RNA-binding portion wasn’t looked at very much.

Dorothee Dormann (00:17:36):
And so we speculated because these were also, this region apart from these folded regions is also largely disordered and has a lot of disordered RGG motifs. We speculated that this region could also be important for driving phase separation. So a PhD student in my lab at the time, Mario, he made mutations in FUS where he either deleted the C-terminal region, this NLS region which I previously shown you was the binding site for transportin. And then he also mutated all the arginines to lysine to test whether the arginines are important. And then I think by now this is also quite well known, that by doing such mutations you actually severely impair FUS phase separation.

Dorothee Dormann (00:18:25):
So, both is in this C-terminal, RGG-rich region, and is important for FUS phase separation and also the arginines, which Simon and Tony and other people at the same time had had similar findings. So then after seeing this, we wondered now if the C-terminal NLS region is so important for phase separation of FUS, could the nuclear import receptor that binds to this region be modulating phase separation?

Dorothee Dormann (00:19:05):
And so we tested this by adding recombinant transportin to FUS and testing how it influences phase separation. And so we did this in very simple phase separation assays. So we always purify FUS from bacteria with an N-terminal MBP tag and then by liberating the MBP tag with protease as we formed these FUS droplets or we can also by aging, get these more solid aggregate-like structures structures.

Dorothee Dormann (00:19:36):
And then when we did this in the presence of equimolar amounts of transportin, the phase separation of FUS was completely abolished. And this was not the case for other control proteins that we added. So PRMT1 is the major arginine methyltransferase which also binds to FUS RGG regions and methylates the protein, but that didn’t have that effect. And an importin that does not interact with FUS also didn’t have that effect.

Dorothee Dormann (00:20:06):
So we also then wanted to show this mechanism is happening in cells and for that we came up with a few assays and I’m just showing for the time the sake of time one assay that we also now like to use a lot in my lab and this was developed by Saskia Hutten. She adopted this assay that we usually use to study nuclear import of proteins. So here we took HeLa cells that in this case they were stressed and had stress granules and then she semi-permeabilizes them with Digitonin and then you can wash out soluble factors and add back recombinant proteins. So you can do this to reconstitute nuclear import, but in this case we block the nuclear pores with wheat germ agglutinin so that this import cannot happen.

Dorothee Dormann (00:20:55):
And when we add FUS it actually does partition into the preformed stress granules or it associates with these preformed stress granules. And then we asked, well if we do this now in the presence or absence of transportin does it affect the stress granule partitioning of FUS? And indeed, so we see here in the top panel that FUS added alone, it nicely sticks to these preformed stress granules, which here we are colabeling with G3BP. But if we did this in the presence of transportin, this stress granule association does not happen.

Dorothee Dormann (00:21:33):
And we also did some as essays in intact cells, which I’m not showing for the time reason because also this is all published a few years ago and so you may anyway be familiar with it. So we could show also in intact cells that this chaperoning mechanism was happening and that transportin in the cytoplasm has the capacity to suppress stress granule recruitment of FUS.

Dorothee Dormann (00:21:59):
And now when you have what now was sort of the immediate next thought that well now if you have an ALS-linked mutation in this signal that prevents or weakens transportin binding as I showed you in the very beginning that now you would predict that now obviously also this chaperoning mechanism, this shielding mechanism is not working properly anymore. And this is what we could show exactly to be the case.

Dorothee Dormann (00:22:24):
So we showed this by purifying either the wild-type protein or a protein with such an ALS-associated mutation, P525L, and then tested whether this shielding or chaperoning was happening. And you see it obviously happens very beautifully for the wild-type protein but the ALS mutant protein is sort of resistant to the chaperoning. So it cannot be properly chaperoned by transportin because the binding cannot occur so efficiently.

Dorothee Dormann (00:22:55):
And this is also the case in cells. Now this is an assay where we cytosolically anchor FUS so that it cannot be imported. And basically now you can investigate its stress granule association or partitioning independent of its transport into the nucleus. And also here in this assay you see that the mutant protein has a much higher tendency to partition into stress granules.

Dorothee Dormann (00:23:26):
What these ALS-associated mutations in this NLS region are doing, they’re actually doing two bad things by disrupting this transportin binding mechanism: they not only interfere with proper nuclear import of FUS but they also disable this shielding or chaperoning mechanism in the cytoplasm that makes them, the FUS protein, more prominently go into stress granules and accumulate there at even higher levels. And so that’s how we think why these mutant proteins so efficiently end up in stress granules and then can probably aggregate very efficiently there.

Dorothee Dormann (00:24:03):
Now this mechanism, this shielding or chaperone mechanism, is not only specific to FUS and transportin. Around the same time other labs showed that it was also the same for TDP-43 and it’s cognate import receptors. And we went on to expand this concept to other positively charged repeat proteins that arise in ALS and FTD patients. So this is a gene mutation that you find frequently in familial ALS and FTD cases. It’s a long repeat expansion in an intron of this C9orf72 gene which was shown to give rise to a repeat RNA, which is then noncanonically translated and gives rise to these different dipeptide repeat proteins.

Dorothee Dormann (00:24:58):
So if you read this repeat sequence in the three different reading frames, both in the sense and antisense direction, that’s the protein sequences that you end up with. And so these are not made in any normal sort of healthy individual because they don’t have this long repeat expansion. But in these patients, these proteins, these aberrant dipeptide repeat proteins are made.

Dorothee Dormann (00:25:21):
And so they were discovered by Dieter Edbauer and Petrucelli around, I guess, 10 years ago or so. And when these were known to be arising in these patients, we speculated that because they’re so arginine rich, they’re looking a little bit like NLS sequences that they could also be directly bound by these importins. And Saskia tested this hypothesis and could show that this was the case.

Dorothee Dormann (00:25:53):
And then we and others had noticed that these very positively charged DPRs actually have the property to very efficiently interact with negatively charged molecules like DNA or RNA and also with low complexity sequences, for example aromatic-rich IDRs that you find in many RNA-binding proteins, also in TDP. And so when you mix them together you actually get co-condensates of this GR repeat or the PR repeat in these proteins or the nucleic acids.

Dorothee Dormann (00:26:35):
And then, so I’m just cutting a very long story short and then sort of showing you that what now happens when you now add import receptors that directly bind to these dipeptide repeat proteins, you can do that and by that you can actually sequester them away and sort of shield them from aberrantly interacting with these proteins and nucleic acids. And so we showed this for TDP-43.

Dorothee Dormann (00:27:01):
So that’s in this paper and here I’m showing as a proof of principle, showing you just an experiment with the with RNA. So here we used fluorescently-labeled poly-GR peptides, that together with RNA from these RNA-GR co-condensates and this we can actually prevent efficiently by adding this importin beta type import receptors but not the, sorry, I have to move away the videos, but not the export receptors, export receptor CRM1 for example.

Dorothee Dormann (00:27:37):
And so this shows that these import receptors can be quite broadly protective and then sort of shield such positively charged phase separation-prone proteins. And this concept now really has emerged from multiple other studies. As I mentioned before, Jim Shorter back to back with us found this was also the case for TDP-43 and also other disease-linked RNA-binding proteins like the hnRNP-A family or the whole FET family. So they are, cognate import receptors can very efficiently sort of act as chaperones or even disaggregators of aggregates.

Dorothee Dormann (00:28:22):
And then already a long time ago Dirk Gorlich had a similar finding for histones and ribosomal proteins that he showed that in vitron these also tend to aggregate with nucleic acids with RNA and that this can be suppressed with import receptors. And the same is true for FG-rich nucleoporins as shown by my colleague here in Mainz, Edward Lemke a few years ago.

Dorothee Dormann (00:28:48):
So there is obviously a few other studies that I didn’t have space to fit on the slide. So there is a whole bunch of literature in the meantime that supports this concept that nuclear import receptors are broadly protective against cytoplasmic protein aggregation, especially against such positively-charged DNA- and RNA-binding proteins that otherwise would tend to heavily aggregate in the cytoplasm.

Dorothee Dormann (00:29:18):
And so therefore this concept has been proposed by us and Jim and others that possibly this could be used in new therapies if you would manage to somehow enhance the interaction of the import receptors with the pathological proteins or elevate their levels. That could be a strategy to relatively broadly defend against such protein pathologies. Now obviously putting that into a therapy and into practice is obviously easier said than done. But just conceptually I think this has emerged from this work.

Dorothee Dormann (00:29:56):
Now obviously these nuclear import receptors are not the only sort of chaperone-like or holdase-type of molecules that are acting against or on these RNA-binding proteins. So classical chaperones or small heat-shock proteins have also been shown by many different labs to be effective and probably there could be other interacting proteins that have a similar activity.

Dorothee Dormann (00:30:24):
So with this I want to leave the topic of import receptors and for the last 10 minutes or so talk about the topic of post-translation modifications, which my lab is also very interested in. So here I’m just showing you a bunch of post-translation modifications that are commonly found on RNA-binding proteins including FUS and TDP. And yeah these can be very complex chain-like PTMS like Ubiquitination or PARylation. But what has been shown to be altered in the disease state, and that’s why we have studied them, is arginine methylation of FUS and phosphorylation of TDP. And this is what I’ll want to briefly talk about.

Dorothee Dormann (00:31:08):
So I first have a few slides on our work on FUS because there we found already a number of years ago still was with Christian Haass that FUS arginine methylation is actually altered and deficient in frontotemporal dementia. So this cartoon shows that normally FUS is arginine methylated on all these arginines in its three RGG domains. So this is known from a lot of methyl proteomic studies and because we at the time found that the methylation of this signal here alters transportin binding and nuclear import of FUS, we were interested at looking into the methylation status of FUS in ALS and FTD patients and therefore we raised methyl specific antibodies against this RGG-3 region and we were lucky that we got nice monoclonal antibodies that were either specific to the unmethylated RGG-3 region or monomethylated or asymmetrically dimethylated.

Dorothee Dormann (00:32:10):
And then we could use those highly methyl-specific FUS antibodies and investigate the methylation status of FUS in ALS and FTD patients. And that showed that, so I’m not only showing you frontotemporal dementia because in ALS didn’t find any changes but in FTD there was a striking change in the methylation pattern. So normally in healthy tissue, and this is not only true in cortex but generally it’s pretty much most tissues that you look at, FUS is asymmetrically dimethylated, and you get staining with this antibody and not really any prominent staining with these other forms of these other antibodies.

Dorothee Dormann (00:32:49):
However, in the FTD patient cortex where you have FUS deposited in insoluble aggregates, you find a different pattern. You don’t find asymmetrically dimethylated FUS, but you find mainly unmethylated and monomethylated FUS. And so now we still don’t really know exactly where the switch or this deficiency is coming from. I mean, we speculate that it’s probably on the level of PRMT1 and that FUS cannot be properly methylated by its major arginine methyl transferase. But there is so far just more a speculation and we don’t have really proof for that. So it’s not really known how that arises. But what we then looked into what are the consequences of this when FUS loses these methyl groups.

Dorothee Dormann (00:33:37):
And so having shown that these arginines in these RGG regions are so important for FUS phase separation, we obviously speculate that this could affect phase separation or these phase transitions of FUS and the aggregation behavior of FUS. And so we tested this by in vitro methylating FUS. So purified recombinant FUS with PRMT1 and SAM as a methyl donor. So this worked very efficiently and we can, just showing you here with our specific antibodies, we can fully convert the unmethylated bacterial protein into asymmetrically dimethylated FUS. And then we could compare these different proteins, the methylated and unmethylated, in phase separation assays or also look at the differences in dynamics in FRAP assays and so on. And that showed that the methylated protein phase separates less efficiently is more dynamic and also tends to accumulate less in stress granules. So that sort of supported, that led us to the conclusion that loss of these methyl groups makes the protein phase separate more and probably makes it more aggregation prone.

Dorothee Dormann (00:34:56):
Now we are still following up what additional consequences loss of this methylation could have. And here I’m just showing you some so far unpublished data that we sort of still have to work through mechanistically, but what we also use these, the unmethylated and the methylated protein for is in vitro pulldowns from just brain lysates where we just then after thorough washing checked which proteins preferentially bind to unmethylated versus methylated FUS And there was indeed a very, quite large spectrum of proteins that showed differential binding. So there were the nice positive controls coming up. For example, transportin binds much more to the unmethylated FUS as expected, as we have previously shown.

Dorothee Dormann (00:35:49):
Interestingly we found a lot of large number of ribosomal proteins and SRSF proteins, so splicing factors as well as translational regulators in this sort of cloud of proteins that tends to accumulate, to bind more strongly to the unmethylated protein. So this is something we’re still following up mechanistically what that could mean.

Dorothee Dormann (00:36:14):
And then a postdoc in my lab, Erin Sternburg, is currently looking at the consequences of FUS methylation on RNA-binding and I’m just showing you one piece of in vitro data that suggests that the unmethylated protein also has a deficiency in its interaction with RNA. So we are currently still doing sort of dynamic and affinity measurements for this. And Erin has now also taken that into cells and there obviously it’s much more difficult, or it’s actually not possible to really specifically manipulate or change the methylation status of FUS specifically without changing the methylation status of other proteins.

Dorothee Dormann (00:36:52):
But what she has done and is currently investigating is proteins where we have mutated these RRG motifs. So where we have mutated the arginine to lysine, so KGG mutant FUS, if you will. And she’s now looking globally at changes to RNA-binding of FUS and its RNA processing functions. And so through that we are trying to get a handle at addressing how FUS that shows altered methylation would show deficiencies in its RNA interactions and RNA processing functions.

Dorothee Dormann (00:37:31):
And so for the last few minutes, I’m leaving FUS and I want to talk a bit about TDP-43 and particularly it’s phosphorylation because that has been shown already a long time ago to be altered in ALS and FTD patients. So it’s always been observed from the beginning on that in disease tissue where you have insoluble TDP, that TDP is present, is a higher migrating form and this was shown to be hyperphosphorylated TDP. And there are also some phospho-specific antibodies, particularly these five sites have been raised and they pretty much always stain the deposited insoluble protein in patients, but not the physiological protein in the physiological conditions.

Dorothee Dormann (00:38:21):
And from mass spectrometry analysis of ALS brains, so there was only a couple of patients were analyzed in this study, but nevertheless they showed that the majority of the phosphorylated sites were located in the C-terminal intrinsically disordered region of TDP. And surprisingly, I wasn’t really clear how this phosphorylation affects the aggregation status of FUS or its phase separation propensity.

Dorothee Dormann (00:38:52):
So there were only studies where people had globally overexpressed or inhibited various kinases that were shown to phosphorylate TDP. And this from these studies which were done in cell culture models, or C. elegans, for example, or drosopohila, those had all come to the conclusion that phosphorylation of TDP actually promotes TDP aggregation or is sort of pathological and drives TDP pathology.

Dorothee Dormann (00:39:24):
And then there were a couple of studies where people had introduced phosphomimetic mutations into the C-terminal fragment of TDP and they actually showed that this reduces aggregation in cells. And so that was really contradictory literature and so we decided that we need some in vitro studies that clearly answer this question, how phosphorylation of TDP now actually changes the phase separation propensity of TDP.

Dorothee Dormann (00:39:50):
And so my former PhD student Lara tackled that question initially by in vitro phosphorylating TDP. So we found that casein kinase 1 delta, which we got from a collaborating lab in Frankfurt, can very efficiently in vitro phosphorylate our bacterial TDP proteins. So you see a nice upshift of the protein. And then in simple sedimentation assays that Lara started out with, she could show that in vitro phosphorylating TDP with this kinase actually makes it shifted more to the soluble fraction, as you can also see from this quantification. And this was not the case by just this ATP alone or the kinase alone. So that was surprising, because actually we hadn’t expected this finding in the beginning. So indeed phosphorylating TDP seems to make it more soluble.

Dorothee Dormann (00:40:45):
And then we decided to look into this more deeply and we made a series of phosphomimetic mutations actually in the sites that were found to be phosphorylated in patients. So these were either just these two aspartates, sort of serine to aspartate as phosphomimetic mutations, or these five ones or 12 different ones. So all of the ones that had been found by mass spectrometry in patients. And so then various in vitro phase separation assays on these phosphomimetic proteins or control proteins where we changed the same serines to an alanine showed that having increased numbers of phosphomimetic mutations really suppresses phase separation of FUS.

Dorothee Dormann (00:41:32):
And when we looked at the morphology of the condensates that you get with the increasing numbers of phosphomimetic mutations, you see that the condensates become really much more round and large and appear more liquid-like than the relatively amorphous, not very liquid like looking wild-type TDP condensates.

Dorothee Dormann (00:41:56):
And so then we wanted to confirm this by live imaging of the different proteins where we were looking for fusion events and we found that pretty much the wild-type protein you almost never see fusion events happening over the timecourse of a minute or so. But for the phosphomimetic 12D protein we could readily see that. And also in FRAP experiment we found that the phosphomimetic TDP was more dynamic, and more readily exchanged after half bleaches.

Dorothee Dormann (00:42:34):
Then Francesca Simonetti, another PhD student in my lab set up nice aggregation assays for TDP and an SDD-AGE assay that allows us to visualize these higher molecular form SDS-insoluble aggregates or also we can visualize sort of insoluble TDP aggregates by confocal microscopy if we use fluorescently labeled TDP. And that also showed that having these increasing numbers of phosphomimetic substitutions clearly suppressed this aggregate formation or higher molecular weight SDS-insoluble species.

Dorothee Dormann (00:43:12):
And in cells, when we looked at the behavior of these proteins in cells, this really tracked with the in vitro behavior that, while the wild-type TDP after oxidative stress or also the 12A control protein, they readily partitioned into stress granules. But this was also the case for arsenite treatment where you induce these nuclear stress bodies where TDP goes to. But in the phosphomimetic protein shown here, you really see that it’s stays dispersed and doesn’t go into stress granules or into these nuclear stress bodies.

Dorothee Dormann (00:43:51):
And Hendrik from Dieter Edbauer’s lab also looked at this in primary neurons. He looked at the same panel of mutations either like nuclear proteins or cytosolic proteins and he did nice filter prep assays to look at the protein solubility or also the pattern you see in neurons. And that confirmed exactly what we saw in HeLa cells, that phosphomimetic protein is actually much more soluble and tends to aggregate less as SDS-stable aggregates and granules basically in the neuronal cytoplasm.

Dorothee Dormann (00:44:30):
And so this led us to, this is just a summary model from all of these in vitro and cellular studies, that surprisingly led us to find that phosphorylation of TDP, which you do find in a ALS and FTD patients actually doesn’t make the protein aggregate more, but rather the opposite. It makes it phase separate and aggregate less. You end up with this more liquid-like dynamic behavior and really a suppression of phase separation or insolubility. And so we can clearly conclude from this, I believe, that TDP phosphorylation is definitely not a direct driver of TDP aggregation.

Dorothee Dormann (00:45:13):
And so we are currently trying to get more evidence for the hypothesis that this could be a protective cellular response that really is coming up maybe when TDP starts becoming insoluble. And then you maybe have, so this is what we currently trying to test, through again in vitro and cellular assays, we’re testing what happens when we make TDP more solid-like versus more liquid-like. Do we actually then see changes in its levels, in its phosphorylation levels, and we are now trying to set up cellular models of phosphorylation where we can actually study under what conditions phosphorylation of TDP actually is coming up and where we can test this hypothesis that it’s really, it could potentially be a cellular response to make TDP more more soluble and counteract the aggregation.

Dorothee Dormann (00:46:10):
Now what I want to just briefly mention the last minute is that something that we already started back then, but what is still a possibility and something we are also testing at the moment, that obviously we cannot exclude that having a more liquid like and condensation deficient TDP, what we are seeing with the C-terminal phosphorylation that this could actually be detrimental because maybe it could be interfering with TDP’s physiological functions. And that got us now very interested in looking into the functional relevance of TDP phase separation and whether that changes its functions in RNA processing, for example.

Dorothee Dormann (00:46:58):
So at the time we looked a little bit into that. So this is the data that is in the paper. So we could exclude at the time that the phosphorylation or the phosphomimetic residues directly interfere with RNA-binding of TDP and also at least in a few splice targets or like autoregulation of TDP that we looked at this was not disturbed for the phase separation deficient 12D mutant protein.

Dorothee Dormann (00:47:26):
But we are now broadening this at the moment and have made now a large panel of either more phase separation prone or phase separation deficient versions of TDP with mutations in different regions and now are looking into how this actually disturbs TDP’s ability to regulate certain physiological functions in RNA metabolism. So we want to look here at, for example, transcriptional regulation or splicing or translational regulation.

Dorothee Dormann (00:47:56):
So this will give us hopefully more insights into the physiological relevance of TDP phase separation and obviously this could also be extended to other proteins like FUS. Yeah, so sorry, I forgot to advance to this slide. So this is more or less what we want to test: if we modulate TDP’s ability to phase separate, how do we disturb its physiological functions that I showed you in the very beginning.

Dorothee Dormann (00:48:24):
Okay, so with this I’m at the end. Thanks for sticking around. So I hope I could show you some of the current view, how we are believing these pathological RNA-binding protein aggregates in neurogenerative diseases, how they’re formed. So this is a current working model and we believe that phase separation, aberrant phase transitions for example in stress granules, could be one pathway how this could be happening. And I hope I could show you some of the regulatory important factors that we have found like import receptors and PTMs.

Dorothee Dormann (00:49:01):
Obviously there’s much more to discover and I’m happy to go into a nice discussion with you to answer all your questions. But before I do this I want to acknowledge my fantastic team. A lot of them moved with me to Mainz and we settled here now very well at Johannes Gutenberg University and the IMB. I mentioned some of the group members that did some of the studies I showed to you, and these are some of our collaborators. And I want to point out that we are still recruiting, especially postdocs and I also have a technician position open. So if you’re interested, contact me. Now I’m at the end I’ll exit my presentation and I look forward to answering questions.

Edgar Boczek (00:49:54):
Great talk, Dorothee. Thank you very much. This was amazing and I think we have a question from Paul Kaufman, if you want to go ahead and ask a question.

Paul Kaufman (00:50:07):
Yes, hello.

Dorothee Dormann (00:50:08):
Hi.

Paul Kaufman (00:50:09):
Great talk. I really enjoyed that. I was wondering at the very end where you showed the unaltered splicing regulation by the different mutant TDPs, have you looked at that at a genome-wide scale or just looked at-

Dorothee Dormann (00:50:22):
Yeah, no, this is sort of what we are planning, yeah? And what we are now planning to do with even more mutants, like not only this 12D versus 12A, but yeah, so therefore I think this question is still not fully answered because in our previous study we only looked at a few splice events. So obviously that could be, if you look globally, there could be changes.

Paul Kaufman (00:50:47):
Great, thank you.

Dorothee Dormann (00:50:48):
And also beyond splicing, it could be in other functions, there could be deficiencies. So that’s what we want to look into.

Paul Kaufman (00:50:53):
Great.

Dorothee Dormann (00:50:54):
But I can’t answer it yet.

Edgar Boczek (00:50:59):
Great, thank you. So I actually also have a question, so it’s more of a fundamental question maybe. So do you actually believe that neurodegeneration is caused by a loss of function in nucleus or a gain of function in the cytoplasm?

Dorothee Dormann (00:51:13):
Yeah, I guess it looks, I would say from all the data that’s out there from various, for example, mouse models or neuronal cell models, I would say there is evidence that for TDP it’s really a large component of what yeah drives neurogeneration is the nuclear loss of function because for example, you see very prominent splicing aberrations in patients as well. And so this indicates that there is a nuclear loss of function at least happening in the disease.

Dorothee Dormann (00:51:47):
For TDP, sorry, for FUS, there has been nice mouse model studies, for example, from Luc Dupuis’s group and others that support the idea that it’s actually both, that there is also a cytosolic gain of function toxicity from having too much FUS in the cytoplasm. It has been shown that this causes, for example, translational alterations and all the interactions of the cytosolic protein. So there it looked, and that also they showed that having having nuclear, basically FUS knockout is not toxic and does not drive most neurogeneration, but having NLS mutations and having too much FUS in the cytoplasm does drive neurogeneration. So for FUS it looks like it’s an additional component of the cytosolic gain of function.

Edgar Boczek (00:52:41):
Okay, thank you.

Dorothee Dormann (00:52:42):
In principle both are still possibilities, I think.

Edgar Boczek (00:52:47):
Yeah, yeah. Okay. We have a question from Kamran. Do you want to ask a question yourself, Kamran?

Kamran Rizzolo (00:52:56):
Hello? Yep. Thank you for the great talk Dorothee. I was just curious, do you find that there is any effect on the binding of the transporters when there’s these different patterns of PTMs on the C-terminus of these proteins and so you see right, the more diffuse signal in the cytoplasm and in your summary you also show the nucleus, but I wonder is transport affected?

Dorothee Dormann (00:53:27):
So when you have different patterns of post-translational modifications you mean?

Kamran Rizzolo (00:53:32):
Yeah, exactly.

Dorothee Dormann (00:53:33):
Yeah. So I didn’t talk about this, but for FUS we have shown that altered methylation changes, so methylation in the RGG motifs changes the transportin binding affinity. So, because the transportin directly binds to RGG motifs and then methylation status can change that. For TDP, this is still not known. And it’s also not known whether the NLS region actually is abnormally post-translationally modified in the disease, although that’s a possibility.

Kamran Rizzolo (00:54:10):
I see.

Dorothee Dormann (00:54:12):
So definitely yeah, the binding in the NLS region or around it, or if that’s modified that this can change the affinity to the import receptors.

Kamran Rizzolo (00:54:25):
Thank you.

Dorothee Dormann (00:54:26):
Yeah, and there could be other PTMs that we haven’t discovered or studied that could also do that. That’s definitely a possibility.

Edgar Boczek (00:54:34):
Okay, cool. So we have a question from Emily Spaulding. You want to go ahead?

Emily Spaulding (00:54:39):
Sure. Thank you. I wondered if you know whether the nuclear import receptor TNPO1, I think it was called, also regulates phase separation of wild-type FUS when it’s in the nucleus where it might be fairly concentrated?

Dorothee Dormann (00:54:55):
Yeah, so in the nucleus we believe that, or it’s known that the RanGTPase displaces the binding of import receptors to the cargoes. And so in the nucleus this complex by the presence of RanGTPase gets dissociated and therefore we believe that in the nucleus the chaperone mechanism is not happening. Or maybe then other chaperones or the binding to the nucleic acid sort of take over that shielding or sort of folding function. So we think this nuclear import receptor chaperoning is more relevant in the cytoplasm where you have a high affinity binding between the import receptors and the cargoes.

Emily Spaulding (00:55:42):
Okay.

Edgar Boczek (00:55:43):
Then Shruti from Dewpoint, you want to ask your question?

Shruti Jha (00:55:47):
Hi, thank you very much for a great talk. I just have a technical question. What was the difference between the phosphomimetic and control construct that you used for experiment?

Dorothee Dormann (00:55:59):
Yeah, so as a control we just mutated the serine to an alanine. So to a non-charged, yeah, neutral amino acid.

Shruti Jha (00:56:13):
Okay.

Dorothee Dormann (00:56:14):
So this more or less, yeah, it behaved largely similar to wild-type or it, in the contexts of mythology it tends to make TDP rather a bit more amorphous or a more aggregate-like looking.

Shruti Jha (00:56:26):
Do you know why you chose alanine? It could be any amino acid?

Dorothee Dormann (00:56:30):
Yeah, we just wanted to have something neutral. But yeah, I mean in principle that’s why we chose alanine and that is sort of what is typically used as a control for these phosphomimetics for, yeah, if you changed something to a charged amino acid, then you choose something non-charged basically.

Shruti Jha (00:56:51):
Thank you.

Edgar Boczek (00:56:54):
Okay. Then we have a question from Katie Copley. Katie, you want to go? You’re mute Katie.

Katie Copley (00:57:03):
Oh, can you hear me now, or?

Dorothee Dormann (00:57:04):
Yeah, mm-hmm (affirmative).

Edgar Boczek (00:57:04):
Yes.

Emily Spaulding (00:57:06):
Okay. I was wondering whether you think to get at the idea of if the reason we see just hyperphosphorylated TDP-43 in patients is if its like effect of being in the surviving neurons, if it would be a good idea or if you thought about doing live cell imaging of neurons and trying to track TDP-43 phosphorylation and see if unphosphorylated TDP-43 in a neuron like that neuron dies earlier, whereas the neurons with phosphorylated TDP-43 kind of survive longer? Or do you have any other ideas to try and get out-

Dorothee Dormann (00:57:39):
Yeah, I mean, yeah, mean that would be a great experiment. It’s just that, so we still have to find ways to actually manipulate TDP phosphorylation in a cell model. So it’s not so trivial because as I said, if you just globally inhibit or overexpress kinases that phosphorylate TDP, that will often affect a lot of other proteins. And then it’s sometimes difficult to interpret where the effect is coming from if you see then whatever in neuro toxicity or a rescue. And so that’s why we so far have only looked at psychosomimetics.

Dorothee Dormann (00:58:17):
And now we are looking at signals or stimuli that it can actually trigger TDP phosphorylation in a cell model. So because this is actually not really known. So this is something that we are, so yeah, it looks like we have found different stressors that can do it. So I mean also I guess stressors have the disadvantage that you’re also changing a lot of other things in the cell and not just TDP phosphorylation. So in principle, yeah, it would be a very great experiment to somehow trigger, well one neuron phosphorylated TDP and another neuron have it non phosphorylated and directly compare those, but it is technically not trivial, so we haven’t really found a way to do that.

Edgar Boczek (00:59:01):
Cool. Along those lines, I have a question as well. So do you actually know the kinase or phosphotase targeting this TDP-43? Are you thinking about doing CRISPR screens or kinase, like using kinase inhibition libraries or?

Dorothee Dormann (00:59:13):
Yeah, so I mean, yeah, there’s a few kinase report in literature, actually maybe a handful. This casein kinase 1 delta, 1 epsilon, casein kinase 2, and Tau-tubulin kinase and a few others. So yeah, we actually are considering, at least with a small a small kinase inhibitor library which we have, looking once we have a good cell model for TDP phosphorylation just comparing which one seems maybe to be the most efficiently inhibiting TDP phosphorylation. But obviously this could also all be cell type specific and could really be that there is a redundancy in that multiple kineses can do it.

(00:59:59):
So it’s definitely has been shown, if you go through literature, you find a handful or so of kinases that can both in vitro and in cells phosphorylate TDP. And it’s not necessarily clear which one is the major, or if there is a major one in the disease setting in these patients.

Edgar Boczek (01:00:21):
Yeah, I see.

Dorothee Dormann (01:00:22):
Because in cell models we can find out, but yeah, I cannot really answer which is the major one. And it’s also the same for phosphatase. The canonical protein phosphatase 1 and 2 can dephosphorylate it, and I think a few others. But yeah, this is definitely interesting to look into, but they probably broadly affect a lot of other proteins as well. Targeting those enzymes will probably not help.

Edgar Boczek (01:00:55):
That’s my related question, actually. So how would you envision drugging RNA-binding proteins, right? So are you specifically looking into such mechanisms and direct binders or enhancers of import activity or modulation of PTM modifiers, these kinds of things, are you-

Dorothee Dormann (01:01:13):
Yeah, so I have to say that with my lab we are not doing that. I’m really basically academically just interested in basic mechanisms. So we’re not directly doing that here, but obviously I would be very thrilled if other labs would maybe take up on our findings and then pursue these strategies.

Dorothee Dormann (01:01:42):
So yeah, I guess one could, I guess for the nuclear import receptor targeting, it may be maybe even easier to maybe do a chemical screen for finding drugs that maybe enhanced the binding of input receptors to these aggregating proteins or something. But it’s not something that we are doing here.

Edgar Boczek (01:02:14):
Yeah. Yeah. Thank you. So Jarelle Joseph, you have a question as well? You want to go ahead and ask yourself?

Jerelle Joseph (01:02:17):
Hi, Dorothee.

Dorothee Dormann (01:02:18):
Hi.

Jerelle Joseph (01:02:18):
I loved the talk. I have a question as it relates to the interplay between heat shock and the PTMs. Have you looked at that in the context of FUS and TDP-43 or? Yeah, just wanted to hear your thoughts on that.

Dorothee Dormann (01:02:34):
You mean whether PTMs are changing upon heat shock? Is that what?

Jerelle Joseph (01:02:39):
In terms of if you have TDP-43 or FUS that have these PTMs and you expose them to heat shock conditions, does that change the mechanisms that you see there in the absence of those stresses?

Dorothee Dormann (01:03:00):
Yeah, no. So as I mentioned, we are now looking at different stressors, not only heat but also others that can potentially change the PTM profile of these proteins. And so yeah, this is definitely something that I think a lot of stressors are causing changes to PTMs. I mean although more globally, not only to those proteins specifically. But this is definitely something we are following.

Dorothee Dormann (01:03:25):
So I cannot really answer it yet, how exactly the … But that’s where we would love to arrive at in a few years, maybe have a more global overview of how the whole PTM profile of these proteins changes specifically also with different stressors. And it’s definitely a possibility that this then also changes their condensation properties or general interactions in the cell a lot.

Jerelle Joseph (01:03:52):
Thank you very much. Beautiful talk.

Edgar Boczek (01:03:57):
I have another question related to what we are doing at Dewpoint, right? So when you have a drug discovery program, you of course you want to look at the function of the protein at some point then. Are there assays that you would actually recommend setting up when targeting those kinds of complicated proteins?

Dorothee Dormann (01:04:19):
Yeah, good question. Yeah, I mean, I guess the best readout for the function is to look at some of their downstream RNA processing functions like with some splicing reporters that have been developed or just whatever sequencing or checking for certain splicing regulatory events, which are known from large scale transcriptomic studies, or so.So I guess that is a possibility.

Dorothee Dormann (01:04:56):
Yeah, it would be great if one, I guess I’m also very interested in establishing or something we’re working on is also establishing functional in vitro assays, but we can reconstitute some of these functions in vitro so that we can, for example, directly with recombinant proteins, then check if we now change whatever, the PTMs or the sequences, how that affects the functions, but.

Edgar Boczek (01:05:23):
Okay. And what do you think about the RNA transport granule function, do you consider?

Dorothee Dormann (01:05:28):
Yeah, I guess, I mean. Yeah. I have to say we are working a lot in simple cell models like whatever, HeLa cells, and not so much in neurons. But definitely in neurons it looks like these proteins have an important transport function and they’re present in these neuritic transport granules and that seems to be one of their important functions.

Dorothee Dormann (01:05:51):
So they are looking for RNA transport. I mean, there’s a few known targets that they are transporting, for example, along axons or dendrites. So that could, in a neuron, that could definitely also be a readout, although also obviously a bit more complicated readout.

Edgar Boczek (01:06:15):
Okay. Yeah, absolutely. Well this, was very-

Dorothee Dormann (01:06:16):
They’re very multifunctional, right? So in principle, you can lead a lot of different functions with different assays. So yeah, I think the simplest one would be a simple, there’s for TDP for example, there’s some nice splicing reporters that you can just bring into your cells. But I mean, it’s obviously just tracking one particular splicing change, but at least it’s a start for looking at complete protein deficiency, or, yeah.

Edgar Boczek (01:06:47):
Yeah. Thank you. Great. So Dorothee, this was great. Thank you very much.

Dorothee Dormann (01:06:52):
Thank you.

Edgar Boczek (01:06:52):
We really enjoyed it. There were a lot of questions, was a nice discussion, absolutely, I would say.

Dorothee Dormann (01:07:00):
Yeah, I also enjoyed it.

Edgar Boczek (01:07:01):
And yeah, it’s great to have you here. Jill, you want to say a few words?

Dorothee Dormann (01:07:08):
Oh, we can’t hear you, Jill.

Jill Bouchard (01:07:22):
Yeah. All right, I can hear me now. Thank you. I don’t know what happened. Yes, I’d like to echo the thank you. This was a fantastic discussion and a fantastic talk. So I think we all learned a lot and had a good time today. So thanks, Dorothee. Thanks, everybody, for joining on Zoom at the Kitchen Table, at the centrosone in our other office. And we will have more of these discussions down the line, probably in the spring. So please feel free to come back and thanks again, Dorothee. Everybody have a good day and thanks for coming.

Dorothee Dormann (01:07:56):
Thank you.

Edgar Boczek (01:07:56):
Thank you.

Dorothee Dormann (01:07:56):
Bye.