 It's a pleasure to speak here. I've learned a lot already earlier today and yesterday. My presentation will be entirely on medically focused applications of what you could call synthetic biology, but I've tried to relate a little bit to what we heard yesterday as well. So in fact I've heard multiple times that we have 20 amino acids. I mean that is certainly at protein level not entirely true. We have a closer in the neighborhood to 200 different amino acids in proteins because of a process that is called post-translational modification. So 20 amino acids are encoded in the genome but then afterwards after the protein is synthesized or during its folding a lot of modifications are added to those proteins and my lab is focusing on the synthetic biology if you wish of those post-translational modifications. And so I just give you a chart here of the different types of post-translational modifications. I just got this from SwissProt yesterday on experimentally validated post-translational modified sites. You can see there is a whole zoo of modifications that can be added covalently to proteins. And I will talk about this third most abundant one there which is called enlinked glycosylation which is the addition of sugars to the asparagine side chains in eukaryotic proteins in particular. I will spend a little bit of time also on o-linked glycosylation which is the addition of sugars to serine and threonine residues again mostly in eukaryotic cells. So eukaryotic cells do perform these modifications by nature and it's our take on this that we might as well use these natural modifications and re-engineer them to introduce particular functionality into therapeutic proteins. So glycans are not just tiny modifications on a protein. Just to exemplify that this is the crystal structure of influenza virus hemagglutinin. And whatever you see on there which is yellow is enlinked carbohydrate. And as you can see this covers large areas of the protein surface. So this is the dominant antigen in the influenza vaccines that we use today. And every area that's covered by yellow here is basically not available for recognition by antibodies because of steric hindrance. And this is yet a very conserved picture because in crystallography very often the outer branches of carbohydrates are not captured because of large mobility of these structures. So what you see here is effectively the part of the carbohydrate that is rigid enough to yield definable electron density in the crystallography maps. So we're playing with these carbohydrate structures in effect about 55 to 60 percent of all therapeutics we have on the market today. Biopharmaceutical therapeutics are glycoproteins and that includes all monoclonal antibodies or molecules derived from monoclonal antibodies. So it's a very important post translation modification for the biopharmaceutical industry. So we've been talking yesterday about standardization and glycobiology is a fairly recent field in its current generation. And so the standards for nomenclature of carbohydrates have just been agreed upon about less than 10 years ago now. And there's just a few weeks ago there was an update published on this. So just like you have the four letters for the basis in DNA and you have the 20 letters for the amino acids in proteins, we now have symbology for carbohydrates which is universally accepted. And because carbohydrates are branched structures, we cannot suffice with a linear denomination of consisting out of letters. We need to capture stereochemistry in this symbology. And so this can be found it's defined by the consortium for functional glycomics in here. So effectively what this is useful for if you would have to capture the entire chemical detail of a carbohydrate of moderate complexity like this one, I think we could all agree that it's not very convenient to write that out every single time that you would have to talk about such structures. And so the symbology now effectively allows us to reduce that complexity to something more amenable also to automation and to database introduction. So you'll see a lot of this in my talk. Effectively every symbol is a monosaccharide. Squares have glucose topology. So this would be enacetylglucosamine. Sorry, green dots or balls, circles would be mannose. And mannose differs from glucose because of the actual OH hydroxyl functionality on the C2 position. And then this would be galactose which differs from glucose because of an actual hydroxyl functionality on the C4 position of the ring. And so these are the main topologies in the carbohydrates that I would talk about today. So as in other fields we do have models about the biosynthesis of these carbohydrates and models don't always have to be quantitative. In fact we have no idea at all about the quantitative nature of the glycosylation pathway. So this is an entirely qualitative model meaning that we know which enzymes are in which sequence after one another. This is a very simplified depiction of the eukaryotic n-glycosylation pathway. So I'll just walk you through it a little bit. N-glycans, so those that finally end up on the aspergine side chains of proteins, are synthesized as a tetradecaccharide precursor, so 14 monosaccharides. There is an entire sequence which I have not depicted here of enzymes that build up this lipid linked, dolicol linked precursor oligosaccharide. And then as a protein that is in eukaryotic cells designed for the plasma membrane or other intracellular membranes or for secretion outside of the cell, when such proteins all contain single peptides that's recognized by single recognition particle in the cytoplasm until the ribosome SRP particle docks to the translocon, the sex 61 translocon in the endoplasmic reticulum membrane, then the protein translation resumes and the proteins kind of pumped inside the lumen of the endoplasmic reticulum. And at that stage, so prior to folding these n-glycan precursors are added to the protein. And then this entire process in the endoplasmic reticulum of eukaryotic cells is universally conserved in the whole eukaryotic world. And the reason for that most likely is that these glycans play a very important role in catalyzing the protein folding in the endoplasmic reticulum. So, some of these carbohydrates, and I will not go in detail today, some of these carbohydrates recruit proteins like calnexin and calreticulin to the folding protein. And with calnexin and calreticulin you get protein disulfide isomerases, which assists in resolving disulfide bonds that have formed inappropriately in the protein. And so after that has happened, so in other words, you can't touch this very much. If you remove glycans by simply mutating the sequence to which they tend to be attached, many, many proteins will simply not fall anymore in eukaryotic cells. This is also one of the main reasons why so many proteins do not fold in E. coli, because E. coli does not have this machinery at all. So that's why the biotech industry needs eukaryotic expression systems on top of E. coli for manufacturing purposes. This is the very single most important reason why that is the case. Then after this folding catalyzing functions have occurred, the protein is then processed to this mannus-8 click-nag-2 structure, and that's recognized for export from the endoplasmic reticulum to the Golgi apparatus. And in the Golgi apparatus, depending on which eukaryotic cell you're looking at, there will be an entire battery of glycosyl transferases, which I've just simplified here by this swarm of arrows, that will process that mannus-8 structure into rather complex end-glycan structures that are then secreted from the cell. And this is just depicting what a pathway in a very simplified way would look like in a mammalian cell, say a liver cell that produces the majority of protein in our serum. If you'd be looking at a yeast cell, this battery of glycosyl transferases would be entirely different. So the protein is secreted, modified with an entirely different set of glycans, and that causes a whole set of problems in the biopharmaceutical industry. And this is the reason why still today mammalian cells are the main expression hosts on the eukaryotic site in that industry. So this is the knowledge that has been gathered by many, many colleagues and giants on whose shoulders we stand for then engineering this pathway. Even in eukaryotic cells, in mammalian cells, such pathway will not convert with 100% efficiency every intermediate to every product. And so to build this, we need about 16 enzymatic steps starting from this mannus-8 structure. And all of that has to happen in the about 10 to 15 minutes that a protein has between the time point when it enters the Golgi and the time point when it leaves the Golgi for secretion from the cell. So in 15 minutes, you need 16 enzymes to work on that substrate in a sequential way to convert all of that to the final product. I think you can easily understand that this is not going to happen with 100% efficiency at every step. And the consequence is that then you get the synthesis of a large array of these glycan structures, some of which would be similar to the final product, but most of which would be biosynthetic intermediates along that path. And this makes it very difficult to assign structure-function relations to glycoproteins, to glycoforms. And so a goal that we've set for ourselves more than 10 years ago is to design a set of eukaryotic expression systems, expression hosts, that would produce particular glycoforms, single structures with a level of homogeneity that will be sufficient to be able to purify it from that mixture. And so that's what I will be talking about today. So as every synthetic biologist, one needs analytics, high throughput and highly reliable analytics to quantify what we are doing. And so what I did back in the days when I was still a graduate student is to use DNA sequencers, capillary DNA sequencers as the analytical platform to very quickly profile carbohydrate mixtures using capillary electrophoresis. And so that gave us the throughput that has been used to sequence the human genome to now be able to do clonal analysis of all these different clones that we generate along this engineering path. So we label, we basically take off the sugars from the protein, if that would be the blue line, it would be the protein. We take it off with an enzyme, then we label these sugars with a fluorophore called APTS, which imparts negative charge and fluorescence to the carbohydrates. Then there's a bit of cleanup steps, and then we go on the sequencer. And so the anode, the positive pole will be here, the negative one will be there, and we just separate them out using electrophoresis. And just like you would do for DNA or for proteins, we include size standards, which are depicted there. That's just a hydrolyzate of a carbohydrate polymer called starch. And so in this way we can very quickly at high throughput analyze these carbohydrates. Okay, sorry. So the platform that we now have built over the last 10 years allows us to make every single of the structures that you see here with a homogeneity of higher than 85%. And that has taken us many PhD thesis and many postdocs work to arrive at this particular stage. What it allows us to do, amongst other things, is to target glycoproteins. So this would be the protein that's the sugar, of course not in proportion here. But what it allows us to do is to target biofarmaceuticals to particular tissues, because different tissues in the human body express a different complement of sugar binding proteins called lectins. So by injecting proteins modified with particular carbohydrates, we can target therapeutics to particular organs. And so I will talk about one particular application that we've been working on, which was published a few years ago, which is the modification of carbohydrates with Manosix phosphate residues, which allows obviously to target Manosix phosphate receptors. And this is important in treating a category of human disease, which is called lysosomal storage diseases. And so I'll come back to that as we go. Our most recent addition to this platform is this. And that's I think what looks most like synthetic biology. We were at a certain point not happy anymore with what nature had to offer. And we basically reduced complexity of this pathway to its bare minimum, effectively allowing to produce just these very tiny dye and trisaccharides. And I'll also explain to you why that is important, at least in our mind, for therapeutic applications. Okay, so the first stage of this work involved the building, the construction, the transfer of the human angloic oscillation Golgi pathway, module, if you wish, to the yeast Golgi apparatus. And the reason why we wanted to do that is that yeast is tremendously more scalable in manufacturing than mammalian cell cultivation is. It's also much cheaper to do that. And so this is the scale at which yeast fermentation occurs in the industry. This is a factory for human serum albumin, which is built in the north of Japan, where human serum albumin is produced in literally metric ton quantities for use as a serum expander blood expander during surgery, where people just if you don't have enough blood as a surgeon, you can take just take the blood you have and expand it by diluting it with a solution of serum albumin. And so the doses that you have to inject there are in the hundreds of grams. And so you can make that in yeast and you can make it in yeast only at the for the time being. And so being able to construct things like monoclonal antibodies in yeast would change the economics of antibody production quite dramatically. But you need to get the glycosylation right. And so if this would be, in fact it is, the crystal structure of the FC part of human IgG. So for those of you who know for the mathematicians, antibodies are the molecules in our serum that help us combat infectious disease amongst other things. And they have a typically Y shape. Maybe there's some chalk here. So human IgG has this kind of shape. Sorry for that. A little bit like this. So there's a linker region here and what you see there is the bottom part of this. That's called the FC region. And this would be the FAP region of which you have two. These binds for example things on the surface of pathogens. This is where then the so-called effector functions are localized. This binds to other proteins on immune cells and that then triggers reactions like killing activity for the pathogen that sits here. Okay. So what you see here is the bottom part and you can see the blue squares there. Those are sugars and that's conserved in all antibodies that we have in our body. There's a universally conserved carbohydrate there and an n-link glycan of which the structures are normally like this. And what's been found about 10 years ago that if you remove this little triangle there, just one monosaccharide, which is called fucose, the antibodies become approximately 100-fold more potent in their triggering of killing activity for tumor cells. So this is called antibody-dependent cellular cytotoxicity. So if you remove that, you get much better anti-tumor antibodies. And so this is now being commercialized by companies like Roche. So their newest generation of anti-tumor antibodies are manufactured in mammalian cell lines that do not make such fucose anymore. Now yeast doesn't make it in the first place. So you don't have to engineer it out, which makes this somewhat attractive. Also 60% of all biopharmaceuticals today are antibodies contained in this part of the antibody molecule. So what we've been working on is to, you could call it B-cell eyes, the yeast. So you try and make a yeast which looks like a B-cell in terms of glycosylation at least. And so this is what yeast normally does. It takes man-8, and then it converts this man-8 into these hugely complicated, so-called hypermanus-related structures. Whereas this is what you need on the antibody. So at first I couldn't be more different, but it all comes from the same precursor structure. So by knocking out the first steps of this pathway and then, like we do here, and then building in the human pathway in a very optimized way, you could hope of arriving at least at this structure. Mind that this would also require you to have all the sugar nucleotides to build these monosaccharides on top of one another, which is also, certainly when we started, it was not obvious at all. Certainly for galactose, it was not at all clear that yeast made UDP galactose. In fact, it turned out it does, and we still don't know what it makes it for in nature here. So to do this, to build this pathway into yeast, we need to get all of these glycosyl transphases in the appropriate location in the Golgi apparatus, at appropriate expression levels and activities for this pathway to work with high efficiency. And so we do that by targeting the catalytic domains of these glycosyl transphases to the Golgi, by fusing them to transmembrane signals from yeast glycosyl transphases. So enzymes that, by nature, go to the Golgi apparatus of yeast. And exactly how to make that fusion, that's part of the part of the optimization work that needs to be done when you do this engineering. So this is the current results that we are getting from this after about 10 years of work, I must say. So this is what the yeast glycosylation pathway would look like, by nature, very complicated, and not all the structures you want. And then by gradually building in one glycosyl transphase after the other, we build up a glycosylation profile that looks like this, which is in fact better than what you would get from Choké one cells, which are the mammalian cells that are used today in biomanufacturing. So Bram Locans is working on this, finishing his PhD as we speak. So we introduced also a new trick, which unfortunately I can't tell much about yet, to remove all the yeast endogenous glycans, because if you build that pathway, there will be remaining yeast endogenous glycans remaining, about 10-15% of the total glycan pool. And that's really complicated for the pharmaceutical industry to get rid of. So you have to engineer it out, and we will be publishing very soon how we have done that now, effectively allowing us to make pretty homogeneous glycosylation profiles. Okay, we were talking yesterday about modularity and orthogonality, and so I just thought this is not published yet. But when you engineer these pathways, very often what we observe is that the new intermediates that we synthesize along that path, in this case, this Manos V structure, this has never been seen in evolution by the yeast glycosyl transphases, by the yeast Choké glycosyl transphases. So in other words, there also has not been selective pressure for the yeast glycosyl transphases, not to recognize these human intermediates. And sometimes they do, as you can see here. This is a glycosylation site, one site on a human protein called interleukin-22. And what we expected in this train was to get Manos V here, as we do on almost all other proteins that we've tested. With all of a sudden, there was like 30% or more of this new carbohydrate, which we'd never seen before. And so we did all the structural analysis, which quite a lot of work to do, including an NMR analysis of this, to find out that it was this Manos V structure, but now with a tetrasaccharide on top of it involving three, sorry, four different glycosyl transphases to build this. So all of a sudden now you make an intermediate from the human pathway, which yeast has never seen before, and the yeast-choké glycosyl transphases start building on top of that. They're just illustrating the kind of complexity you can get and the kind of non-orthogonality of these modules, these glycosylation modules. Fortunately, we could solve that problem now, because if you get this, this is likely to be very, very immunogenic. It's an entirely new structure. And so you certainly wouldn't want to have that in a therapeutic preparation. Okay, so we can build the human pathway into yeast after a lot of work that works, allowing us to make antibodies in yeast at like gram per liter quantities in a very short space of time. I mean mammalian cells, you need about three to four weeks of fat batch cultivation to get up to the gram per liter levels that people talk about. In yeast this takes three days. Okay, so you have a space time yield, which is at least in order of magnitude better than with mammalian cells. Okay, and then the second target that I was referring to is to try and re-engineer the glycosylation pathway to deliver therapeutic glycoproteins to the lysosomes of patient cells. Now whether the lysosomes, lysosomes are vesicles inside mammalian cells of which the main function is to degrade biopolymers that the cell doesn't need anymore. Okay, this is in a very simplified way what lysosomes do, at least one of their functions. And so if for this degradation of biomacromolecules to the monomer building blocks, you need depolymerizing enzymes and there's about 40 of them in the human genome. So genes that code for such acid pH optimum degradative enzymes and just one disease is illustrated here of what happens when you don't have such degradative enzyme. So this is in the case of Pompey's disease. There you don't have the alpha glucosidase in the lysosomes and this leads to the accumulation of glycogen in the muscle cells of these patients and gradually this becomes toxic to the cells and the cells lose their functionality. So Genzyme, one of the main biotech companies, one of the oldest biotech companies in the world, contributed tremendously to the outlook for these patients by realizing that the targeting system for these lysosomal hydrolysis involved carbohydrates. Normally when they come from the goji to the lysosome there is a receptor that recognizes Manus 6-phosphate on the carbohydrates. Now this receptor that normally targets these enzymes to the lysosomes is also expressed at the cell surface of almost all of our cells. So if you inject a glycoprotein, such a missing enzyme from the lysosomes, if you inject that in the human bloodstream, if it's modified with Manus 6-phosphate these receptors will capture it and endocytose it and transport that protein into the lysosomes of patient cells. So these molecules are amongst the most expensive therapeutics we have today. So the alpha glycosidase for pompous disease cells at about 300,000 euro per year, life long, for these patients. And in fact because these molecules are made by mammalian cells today, their levels of Manus 6-phosphate are very very low. On this particular protein is less than 5% on the currently used formulation. So that means more than 80% is cleared by delivering the first few minutes after injection and this is not a very efficient therapeutic. Although at the high doses that is used today it has changed the life of patients but you know there's clearly room for improvement on the manufacturing side. So what we've done is to take the yeast glycosylation pathway and yeast by nature builds Manus 6-phosphate like structures into its cell wall. The only difference being that there is another Manus residue sitting in the way, substituting the phosphate there and that blocks these sugars from recognizing the human Manus 6-phosphate receptor. So what we've done is to build the to pump up the levels of Manusylphosphorylation in yeast cells by overexpressing the genes that are the rate limiting catalytic rate limiting steps of this pathway. And then the question became can we now convert these kind of sugars into the kind of sugars you need for lysosomal delivery. We have no enzymes to do that you could do that chemically without any problems whatsoever but there is a problem there is a protein here very labile protein that you have to keep intact during that conversion. So we figured we needed new enzymes to do that. And then so this is just I'll skip this so this became the question how are we going to do this right. So as always as biotechnologists I think our first reflex is to go out in nature and think about where in nature there might have been evolutionary pressure selective pressure for such enzymes to originate. And what we realized is that actually fungal cell walls are very very abundant carbon sources in the biosphere. So we thought there must be bacteria that have evolved to grow on fungal cell walls as their sole carbon source. And in fact just going through the biological literature you can easily find a number of bacteria like that. One is this particular species cellulose microbium cellulose. And so when you grow these bacteria on yeast cell walls as their only carbon source you get the induction of a whole array of carbohydrate hydrolytic enzymes including some enzymes as we found that can indeed catalyze these reactions that we were looking at. So this has been done by these two very very good scientists in the lab. There was a project that took quite some time as you can imagine. So this is just showing you that we indeed could find such enzymes. And so I'll just walk you through this. So we have here are the carbohydrates that are made by this yeast cell that we've engineered to express very high levels of mannose 6 phosphate mannose. And those two peaks here would be structures that contain that. So when we take one of these enzymes that we found from this bacterium which we have coined with this name here. When you treat that they all of a sudden in capillary electrophoresis move faster which means that either they are smaller or they have more charge. And then when we treat those with calve intestinal phosphatase an enzyme that removes end standing phosphates now they run slower again showing that indeed these phosphates are now terminal and you can remove them. And we found a second enzyme that then also can remove these mannoses there and you also need to get rid of those for these ligands to be recognized by the mannose 6 phosphate receptor. So this organism effectively makes two of these enzymes. We also crystallized one of them this one here to explain why it worked and it's all in the paper if you're interested in that. So with two E. coli produced enzymes you could basically convert the yeast produced mannose 6 phosphate containing carbohydrate into what we really needed for therapeutic delivery. And so this was the effect then if we exposed patient cells so we took cells from pompous disease patients fibroblasts in this case because that's the only cells you can easily get from patients. And then we exposed these cells to increase in concentrations of our recombinant enzymes and we looked at how much enzyme got into the cells. The current therapeutic drug had this behavior here and then our yeast made drugs we needed about 15 times less of that enzyme to reach similar levels of activity intracellularly. And this is just because we have about 15 times more mannose 6 phosphates on these on these drugs. This is also translated in a mouse model and unpublished yet in a non-human primate model in increased clearance of glycogen from the muscles of of these experimental animals and it's now moving into phase one slash two clinical trials because this isn't genetic disease you can actually do phase one and phase two at the same time because a precursor drug I mean the current drug is the protein part is exactly the same. So there's less issues with suspected toxicity of these kind of products. So this is being commercialized in this company called Oxyrane and the R&D division is in in Ghent. Okay how much more time do I have right now? Okay perfect. So everything I've said so far is basically taking existing pathways or more or less existing pathways from other organisms and putting them into another one which in our case was yeast. And the second story involved the discovery of new enzymes that we needed which were not present in the human pathway but which we needed to convert the yeast pathway into the human pathway. So the next part and this is something that started about five six years ago in the lab is when talking to biopharmaceutical industry people very often they they'll tell you that they have to live with carbohydrates on their molecules because that's that happens to be how nature modifies these proteins and then you have to engineer them to such an extent that they don't harm your function but if you actually could get rid of carbohydrates altogether in eukaryotic cells for a large number of applications that's what they would prefer but they can because it introduces so much complexity to the manufacturing processes and it introduces heterogeneity in the pharmacodynamic behavior of these molecules but you can't get rid of them so far because you need them for protein folding so that there has been this catch-22 for a very long time and and so we figured perhaps it is useful if we could find a solution for that and so I'll skip that if you have questions on what applications would be useful we can come to that I'll show you a few. So this is what wild type cells mammalian cells do they'll make n-glycans of the very complicated type that they've already shown you they also make oglycans which can also introduce quite a bit of complexity there's about in the range of about 200 different structures known in humans here there's about 500 actually known on the oglyc hospitalation site so there's tremendous complexity that can be introduced so what we made is something we call glycol delete where we re-engineered this pathway short-cutted it basically so that we now construct n-glycans of this trisaccharide shape that is identical to the outer branches of what you normally get in mammalian cells and then we've taken it one step further and that's as yet unpublished where we then completely remove heterogeneity only keeping one glucic residue at the n-glycol oscillation sites and entirely destroying oglycol oscillation because oglycol oscillation is not important for folding with very few exceptions so you can't get rid of it actually and from mammalian cells and keep the cells alive this is rather drastic and it was not known at the moment when we started whether this would be compatible with cellular life so this is one of the I think a small example of how sometimes this kind of engineering can teach you more about what is really necessary for cells and for from unicellular life okay so I think the concept was simple you recognize this from the ER we get this manus-8 structure and then we wanted to kind of put a pair of scissors in the in the goji an enzyme that would take off the glycans entirely but for the last residue here and then we were hoping that the endogenous glycosyl transphases galactosyl transphase and cialyl transphase would take that as a substrate and build this trisaccharide which is identical in structure to what you normally get on the outer branches so this was done by Leander in the lab who is still with me right now so then we started playing and you need selective pressure like everything if you were doing engineering you need some kind of selective pressure to select for what you want and for sugars such tools are less elaborate than for proteins and nucleic acids but what we do have is cytotoxic lectins sugar binding proteins that have a toxin domain like rice and toxin there rice and toxin binds terminal galactos right and that if that binds it will kill you as you know rice and is very toxic now if you re-engineer such pathway and you make a mutant that does not make galactosides anymore then the cells become resistant right and in this way you can select for a glycosylation phenotype in your cells that you want and we wanted to have these mannose 5 because the scissors that are available and only recognize such structures they wouldn't recognize the wild type carbohydrates that mammalian cells produce so then we got the scissors from this beast which is hypochrye jack arena that's a fungus that's used for making cellulases nowadays in the biofuels industry and when we studied the secretome of this fungus we found that it actually secretes its glycoproteins modified with just a single glutinac residue indicating that it must secrete an endoglycosidase of the type that we were after and so we clone that gene and it has this gene structure if you wish open reading frame coding sequence structure so it has the intermal pre-prosequence for secretion and the catalytic core and it has a c-terminal prosequence a peptide that as long as this protein is in the endoplasmic reticulum falls back onto the catalytic core and with the crystal structure in the meantime falls back on the catalytic core and inhibits the enzyme basically protecting the glycosylation pathway of these eukaryotic cells from this degradative activity and this enzyme this propeptide only gets removed in the trans-goji apparatus then fully activating the enzyme at the goji which means after the stage where glycans are important in the cell for folding catalysis which is in the ER so it kind of activates in the trans-goji and then removes the carbohydrates when they're not needed anymore for folding catalysis so again we used the lectin trick to select for the cells we wanted in this case we used concanavalin A which binds terminal mannocytes and there again when we remove these terminal mannocytes with this n-dot as we call it this enzyme pair of scissors we get the gluconate we don't get any terminal mannose anymore and that makes the cells resistant against concanavalin A selection so we analyzed that using this particular protein which is an important pharmaceutical human granulocyte macrophage colonist stimulating factor GMCSF and so it has two and glycosylation sites which you can separate from one another using cleavage with trypsin protease so that allows you to do simple mass spectrometry on these glycopeptides and so on the parental cell line where we got the high mannose sugars you get these of course then when we looked at the same glycopeptide from glyco-delete cells all of that is gone literally everything there's nothing detectable anymore and we get new carbohydrate structures which are the intermediates towards what we want to build here and now moldy mass spectrometry is now very quantitative I'll spare you the details on how we quantitated this but we have about 70% of this about 20% of that about 10% of this just without any further engineering so we're now engineering this further to pump it towards this structure and so we have good ideas on how to do that right now so this reduces heterogeneity quite tremendously on glycoproteins so this is gmcsf you see the mass here from about 14 and a half kilodalton to about 20 kilodalton all of that as heterogeneity introduced by the carbohydrates so it took me half a day to actually get any spectrum from that because there's so many different forms sure so many different forms here that it all tends to go into the baseline now in glyco-delete you have remaining heterogeneity but most of that is because of the O-glycans which are also present on the protein this sequence here is O-glycosylated there so Francis who's in the audience in the back took this one step further well he first looked at an application of this again here this is the antibodies which have this conserved glycan so if we produce a human monoclonal antibody in these cells we were looking at would that be helpful in any way for pharmaceutical applications what we found is this if you inject these antibodies in mice so this is in mice that must admit we haven't done it yet in larger animals but at least in mice when you inject these proteins from the wild app cells or from the glyco-delete cells what you'll see is that immediately after injection normally you lose about and that's true for all monoclonal antibodies you lose half to two-thirds of your antibody in the first half an hour due to so-called bio-distribution but if you actually look in the literature about what the mechanism might be for that there's absolutely nothing known there's no rigid scientific experimentation on that and so you lose it but we don't know how at the moment and so exactly at that stage we lose our glyco-delete antibody much less than the wild app-produced antibody resulting in about double circulating levels of the antibody with the same dose and now this is huge extrapolation okay if you'd be able to do this in patients what it would allow you to do if this were your therapeutic threshold you know your trough level that you need to have for getting therapeutic efficacy it would allow you to reduce dosing with a factor of two and like for antibodies with anti-TN where you treat against rheumatoid arthritis against TNF alpha these patients have to go to the hospital every four to six weeks for injections now if you could double that to every eight to twelve weeks this means quite a bit for for patients and we're certainly not at the end of what we can achieve there so i think we could get this even much further reduced and as we go forward in human medicine monoclonal antibody therapy will be much more driven by patient comfort rather than efficacy and so this might turn monoclonal antibody treatments into something that you can inject maybe three or four times a year all right so i'm almost done so this is the one step further that francis took it so francis also got rid of oglycosylation entirely in these cells and that's yet to be published but i can show you the result of that what we did is basically tackle sugar nucleotide metabolism for this so this is a western blood which is a method to analyze protein heterogeneity molecular weight heterogeneity and you can see here from wild lab cells this gmcsf is hugely spread out as i already showed you in the mass spectrum as well in glycol delete it's quite a bit reduced but still has remaining heterogeneity due to the oglycosylation and some level of the n-glycans which are not entirely the trisaccharide structure and now in glycol delete cells it's effectively becoming one band and the remaining heterogen glycol double deletes and the remaining heterogeneity we have is due to site occupancy of the n-glycan sites so n-glycosylation sites can be can or cannot be occupied with an n-glycan so if you take off the n-glycans entirely what you'll find is not modified and modified with a single glutinac residue so i think for the first time now we can actually make glycoproteins in mammalian cells with the same level of homogeneity as we could make non-glycoproteins in E. coli for the longest time and what we still have remaining is a single glutinac residue and so that sugar residue which sits on the n-glycans is actually chemo orthogonal to the protein backbone so we can do chemistry on that carbohydrate on that monosaccharide residue to couple whatever you wanted to do it there so that's hence my question of last of yesterday you know what could you do with more than two or three i mean this is another one that we can add to it and it's made by nature in case you're wondering this doesn't do much to the physiology of the cell glycol deletes adapts completely it's just behaves like wild type cells glycol double delete suffers a bit more but as you know eukaryotic cells and living cells go you can evolve them to adapt to that new glycosylation phenotype and we are still doing that so we're you know passaging it through normal cultivation conditions and it's improving still almost back to normal levels of cell growth okay so i had one more story but i'll skip that i guess i just will tell you that we have implemented glycol delete technology now also in plant seeds and this has been done together with and the picker who is in the department of mark from muntagu so mark as you may have known developed agrobacterium tumor fascians transformation of plants in the very early days of plant biotechnology and we were struck i think as all of you by the the dreadful situation of the Ebola epidemic which happened last year where we had antibodies that could prevent the disease but they couldn't be manufactured fast enough because they were manufactured in a transient expression system in tobacco and so to do that in tobacco you have to grow the tobacco plants which takes months right until you have the doses so what we figured is that if we could implement antibody production in legume seeds in this case soybeans if we could do that you could decouple the production stage from the purification stage of pharmaceutical manufacturing so these seeds as you all know i mean legume seeds you can store them at room temperature for years and they will still keep their germinating power which means that all the proteins in there are intact so we can actually store monoclonal antibodies at room temperature for many years in huge quantities using legume seeds and then having them available when such an emergency happens to purify the proteins and done the one problem was that plant glycans or allergenic to a large section of the human population if you inject them in the bloodstream of course we eat plants all the time and that's safe but that doesn't mean if you inject them in the bloodstream that it's safe and so we engineered out this allergenicity from the plants in a very simple pathway which is implementable in crops okay i'm done this is the funding there is a poster from our group that i want to point your attention to by morgan and then we have a similar meeting to this one in january so if engend is not far away from here if you want to join feel free all right thank you