Transcript source
April-2021-SeriesTranscript
April Foundation Tech Monthly Seminar Series Version 2
[Image appears of Louise Burton listening on the screen and then the image changes to show Colin Scott talking on the main screen and other participants can be seen in the bar at the top]
Colin Scott: Cool, we’re just hitting the record button now, excellent thank you Louise. I’d like to start the session by acknowledging the traditional owners of the lands on which we meet today. For both of our speakers today that’s the Gadigal people in the Sydney area. And for myself it’s the Ngunnawal people in the Canberra region. We’d like to acknowledge their Elders, past, present and emerging, and acknowledge that we are meeting on the lands for which they have been traditional custodians.
Before we kick off I would like to thank the Ops team, particularly Louise, for making sure that these sessions run as smoothly as possible and for giving me a script to run to. So, thank you very much for that Louise. And just to remind everyone if they want to ask questions please enter the questions into the Chat bar at the side of your screen, and Louise will harvest those questions as we go through and I’ll try and go through as many as we possibly can at the end of the session. If we can’t answer them during the session we’ll farm them out to the speakers and they should be able to answer them by email or otherwise so we’ll make sure that everyone gets their questions answered.
[Image continues to show Colin talking on the main screen while other participants can be seen in the participant bar at the top of the screen]
So, thanks for joining. Today is the Foundation Technologies session, for which I am happy to say, I’ve been given the pleasure of leading, and the Foundation Technologies application domain. And today we’ll be focussing largely on post translational modification of proteins, with our first talk from Nicholas DeBono, who is from, who I’m hoping hasn’t just dropped offline. Looks like you might…
[Image shows Nicholas talking in the participant bar while Colin listens]
Nicholas DeBono: No I’m still here.
Colin Scott: You’re still there, excellent…
[Image briefly shows Nicholas on the main screen and then the image changes to show Colin talking on the main screen and participants listening in the participant bar at the top of the screen]
Macquarie University, and is looking at artificial Golgis. He’s a PhD student who’s been supported by one of the FSP top ups. If you want to take it away Nick.
[Image changes to show Nicholas talking on the main screen while participants can be seen in the participant bar at the top of the screen]
Nicholas DeBono: Awesome, thank you. I’ll just share my screen. Hopefully, it works.
[Image changes to show a new slide on the main screen showing the CSIRO and Macquarie University logos, while Nicholas can be seen talking in the top right and text appears on the slide: A dual glycoengineering system; the prospects of a multi-avenue approach to targeted glycoprotein, Nicholas DeBono]
Everyone can see me? Yep, great OK. So, my name’s Nicholas DeBono. I just wanted to say a quick thank you to Colin and everyone else for giving me the opportunity to speak today. I’ll be presenting, I’m a PhD student, just approaching the end of my first year now. So, I’ve still got a bit to go. I’ll do my best to go through my, part of my PhD project, which is “A dual glycoengineering system and the prospects it has towards a multi-avenue approach towards producing targeted glycoproteins”. So, before I can kind of introduce my glycoengineering system, I feel like I need to introduce glycans. So, what are glycans?
[Image changes to show a new slide on the main screen showing a diagram describing glycans and Nicholas can be seen in the top right talking and text appears: What are glycans?]
Very, very broadly glycans are found as a post-translational modification on glycoconjugates, and they decorate the cell surface and a lot of proteins within the cell. There are a lot of different types of glycan conjugates you can find, from glycosphingolipids to proteoglycans, to glycoproteins themselves.
[Image shows a box appearing around the N-Glycans in the diagram on the slide and Nicholas can be seen talking in the top right corner]
Just for this talk today I’ll be focussing on just N-Glycans here. So, N-Glycans are characterised with their attachment to the asparagine residue through a consensus sequence on a protein. And they normally also have this core structure which you can see in the blue shadow.
[Image changes to show a new slide showing text: Why are glycans important? The presence of glycans on a protein alters how glycoproteins interact within the body, We are unable to modify these glycans the way we modify DNA and proteins]
So, why are our glycans important? Well, the presence of the glycan on a protein alters how that glycan interacts within the body. Additionally, we can’t modify glycans the way we would modify DNA or proteins because they’re not template driven the way DNA and proteins are. They are stochastically added, which means whatever is around is whatever adds on generally. Now, how do these glycans alter the glycoproteins?
[Image shows a diagram appearing below the text on the slide showing a flow chart diagram explaining how glycans can alter glycoproteins]
Well, if we take just one glycoprotein such as IGG, every single little modification of our glycan here – for example if we take off a core Fucose we could increase a certain receptor ligation by up to 100 fold, which promotes cytotoxicity. If we add sialic acid to the terminal end of our N-Glycan, we can change that again. If we add a bisecting GlcNAc residue we can again activate a different receptor. If we take things away we have a Terminal GalNAc different receptors are activated. So, I guess my point for this slide is, is the different glycan present on the different glycoproteins will all have different interactions. But why is this sort of an issue for glycoengineering?
[Image changes to show a new slide showing a diagram explaining why different glycans present on different glycoproteins will have different interactions and text appears above the diagram: Why can glycans be a problem?, Different organisms produce different glycans, making producing therapeutic glycoprotein production in in vivo production systems more difficult]
Well, when we’re trying to produce glycoproteins, or anything like that, what happens is the production system we use will produce a different glycan. So, generally human like glycans will be complex, maybe have a core Fucose on them, maybe will have sialic acids. If we produce our therapeutic proteins in mammalian cells they might have a different sialic acid on them. If we use plant cells we might have the presence of an Alpha 1,3 xylose. If we have insects we normally have paucimannosidic glycose, glycans sorry. And if we’re using these cells, which is a very popular production system for proteins, we have these huge ultra high mannose chains on our glycans. The presence of these will lead to super-quick clearance of our important therapeutic protein, which if we want it to stick around the body and do what it wants to do, can be an issue. So, we try and solve these glycan issues with glycoengineering.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: Current solutions of glycoengineering, glycoengineering is the process by which glycans are modified for a specific purpose or towards a targeted goal]
So, glycoengineering broadly is the process by which we can modify glycans towards our specific purpose or towards our targeted goal. There’s a couple of different ways we can do this glycoengineering.
[Image shows a diagram appearing on the bottom right showing how a glycan can be modified and text appears on the left: Genetic change, N-glycosite addition, deletion or move to POI, Addition of non-native GT’s or similar enzymes (e.g. sugar transporters), KO of endogenous GT’s or related enzymes, Environmental change in vivo, Growth conditions, substrate change (azidic sugars), Enzymatically modifying proteins post production in vitro, Addition of GT’s plus required components to a solution containing POI, EndoS glycosylation system]
First we can do it in vivo, inside the host cell through a genetic change. We could say, change at a specific site where glycan is added. So, no glycan site, change our amino acid sequence, and we have a glycan appearing. We can add non-native glycosyl transporters of similar enzymes to modify the types of glycans being added, or we could just knock some out and that will also change our glyco profile so to speak.
[Image changes to show a new diagram on the slide on the right showing Azide reporter generation]
We can also change the environment that we grow our cells in. If we change growth conditions for example, if we have human cells and we change the amount of oxygen present in the chamber that we grow those human cells in we can alter the glycol profile there, and we can also introduce non-natural sugars to give us reported glycans so that we can see where they’re going.
[Image shows a new diagram appearing on the slide on the right hand side showing a diagram explaining enzymatic modification using the EndoS glycosylation system]
Another modification is to do enzymatic modification. So, once we’ve produced our glycoproteins we can then modify them post production in vitro. And this is a really cool method called the EndoS glycosylation system. I won’t be talking into this a little, I won’t be going into this one but it is a very neat way to glycoengineer proteins. So, my point for this slide is that there are a lot of different solutions to glycoengineering and all of them have their advantages and different disadvantages, and a thought I had towards the start of my PhD was, “Well, what if I can combine multiple glycoengineering solutions together?”.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: A dual glycoengineering solution In vivo glycoengineering, Potential initial loss of fitness, Delay in results, Lower protein production without strain engineering, Cheaper long run average cost (if it works), High throughput capacity, Generally produces a heterogenous mixture of glycans]
So, if I was to take a glycoengineering solution that was mostly in vivo, well my potential downsides are that when I’m initially doing my genetic modification of that host I could have an initial loss of fitness, so my host won’t survive as well. Because it’s a life science and cells things take a while to grow, they take a while to get results from. In some host organisms the act of glycoengineering reduces the fitness of the cell which can lead to lower protein production overall. But the upside is averaged out over the long run, it’s generally cheaper if you can get it to work because once you’ve, you’re producing your glycoprotein you can scale up, and that leads to high throughput. However, this generally produces a heterogenous mixture of glycans. So, if we think back to that slide at the start where one specific glycoform can lead to a very specific effect, if I have a whole mix then that effect gets diluted.
[New text appears on the right of the slide: Traditional In vitro glycoengineering, Does not rely on fitness, Dependent on enzymatic availability or chemical synthesis pathways, Faster immediate result, Lower throughput, Higher long run average cost, More homogeneous mix of glycans]
If I was to go down the in vitro glycoengineering route, so not in life, in the test tube so to speak, of course this doesn’t rely on cell fitness anymore because I’ve taken it out of the cell but it is dependent on enzymatic availability or chemical synthesis pathways. So, essentially what enzymes are available for me to use to do my glycoengineering. I can get a faster result straight away because I don’t need to wait for cells to grow but generally my throughput is down a little bit, and enzymes can get expensive. So, over the long run that could be a high cost. But because I’m controlling what enzymes I’m adding in, I can generally get a more homogenous mix. My thought for this at the start of my PhD was, “Well what if I can, you know, combine the two together?”
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: In vivo glycoengineering, Minimise initial loss of fitness, Less delay in results, Higher protein production without strain engineering, More controllability, Cheaper long run average cost (if it works), in vitro glycoengineering, Does not rely on fitness, Less enzymes required for synthesis of product, Faster immediate result, Higher final throughput, Better controllability, Offset long run average cost]
That would lead to a general minimise of initial loss of fitness. I could theoretically get quicker results because I’m doing half in vivo, half in vitro. I could get higher protein production because I’m doing less strain engineering so that I get less loss of fitness. If I can just produce a starting glycan and then finish the modification in vitro I could have more controllability and that cheaper overall cost is there.
So, what… the solution I came up with for my dual glycoengineering solution was to combine a well-known glycoengineered protein production strain such as Pichia with a in vitro system that myself and a, my PhD supervisor came up with, Dr Edward Moh, called the artificial Golgi column, which is what I’ll be going into.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a diagram Pichia pastoris in a bio reactor then through a Golgi column to achieve a targeted glycan structure and text appears: A dual glycoengineering solution, Create a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation]
So, the way this system works is I’m planning on producing a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation. So, if we were to map it out we would start with our Pichia pastoris strain in a bio-reactor – I know it’s the best picture I could make – we start with our ultra high in mannose glycans, we do some strain engineering and we finally come up with a glycoprotein that has a much more truncated glycan structure. We can then take this glycan structure and put it inside my artificial Golgi column – which I’ll explain soon – modify that and come out with a final glycan, targeted glycan structure which is what we’re aiming for.
Now, that doesn’t have to be this structure here with the hybrid mannose and the terminal sialic acid. The beauty of the artificial Golgi column is that it’s tuneable and if I was to just swap out the enzymes in this column I could create a different glycan structure theoretically. So, I’ll be mostly talking about the AGC today but I’ll quickly duck into the Pichia pumic because it is a dual solution.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a photo of Pichia under a microscope and text appears: Chosen in vivo production host, Pichia pastoris, Grows to a high biodensity, Frequently used yeast production system, Glycoengineering tools exist, Already used for glycoengineering efforts]
So, the reason why I chose Pichia is it grows to a higher bio-density, it’s a frequently used yeast production system, it’s already been used for glycoengineering.
[Image shows an article appearing at the bottom of the screen showing a text heading: Glycoengineering of antibody (Herceptin) through yeast expression and in vitro enzymatic glycosylation]
And there are some really nice papers that have come out such as this paper here with really good examples of a dual system already through yeast expression and in vitro enzymatic glycosylation. Critically the different here is this in vitro enzymatic glycosylation was relatively slow whereas my artificial Golgi column solution is a little bit faster.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a photo of Dr Carol Hartley and text appears: Chosen in vivo production host, Pichia pastoris, Grows to high biodensity, Frequently used yeast production system, Glycoengineering tools exist, Already used for glycoengineering efforts, Dr Carol J. Hartley, Expert in flow biocatalysis and P. Pastoris, To be started late 2021]
And helping me explore this part of the project, Dr Carol Hartley from CSIRO, who is an expert in flow biocatalysis and in Pichia down in Canberra. She’s going to be helping me out. I haven’t had a chance to start the Pichia glycoengineering just yet but I’m planning on starting in late this year.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing glycoengineered Pichia and text appears: My solution (MRes and PhD continuation), Create a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation]
So, that’s our Pichia side, and then if we move, so we’ve, say we’ve glycoengineered our Pichia and we’ve produced a truncated glycan structure such as this from our high mannose and then we want to modify this even further to get our final product. How do we do that? Well, my solution was to put it through my artificial Golgi column. Now, I’ve called it an artificial Golgi column because I’m trying to replicate what happens in the back half of the Golgi inside cells, which is a lot of post translation modification, and additional protein folding.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing the Artificial Golgi Column and text appears: the Artificial Golgi Column – an in vitro post-production glycoengineering system]
And I’ve taken some of those glycosyltransferase enzymes and I have bound them to the inside of a column. The system I’ve used for this is just Ni-NTA, you know, HisTrap histidine column binding with my [14:24] glycosyltransferase enzyme, down to the beads inside this column here. All you add in at the start is your starting glycan structure, and your other essential factors such as your metal co-factor and your starting sugars. It goes through the column, you get your enzymatic reaction, and you get a hopefully final glycan product. The beauty of this system here is that I’ve, all the reactions I’ve done, happened in about one to two minutes, as opposed to a usual enzymatic reaction which you’d do in a test tube which can happen over hours.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing the B4GalT1 diagram and text appears: Immobilised enzyme used, B4GalT1, Found in trans-Golgi, Adds Beta, 1-4 galactose to terminal GlcNAc]
So, the first enzyme that I immobilised onto my column to test, and I’ve got some preliminary results to show you, is B4GalT1 which we normally find in the trans-Golgi, and it adds a galactose residue in a Beta, 1-4 form position to a terminal GlcNAc sugar. All these sugars mean slightly different types but if you just want to follow the colours that’s fine, I’ve got the key down here. But critically what I’m looking for in my results of my experiments that I’m about to show you is, pardon me, an addition of this galactose residue and a higher finishing concentration on my intact glycans.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of preliminary results using the Trastuzumab and the 3mm I.D on the left of the screen and a diagram of the three main types of glycans on the right can be seen and text appears: Preliminary Results]
So, what I did was the first test I used, I bought some commercial Trastuzumab sample. This is an IGG and I analysed its glycol profile at the start and I found that there were three main types of glycans. There was the FA2 glycan which is this black bar here and I’m going to call that my starting glycan. And then FA2G1 – G1 because it has one glycosyl transferase on it – and then FA2G2 obviously because it has two.
We can see from the start to after the reaction there has been a large decrease in the starting glycan, and a huge increase in the final dye glycosylated glycan here which told me that that’s great, that’s excellent. My column that adds galactose residues to glycan is working. Now, ideally I would like this to be, you know, 100% FA2G2, but these are preliminary results and the fact that this is working is really great for me. So, it means I can take it further and optimise it down the track.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of preliminary results using Fetuin and on the right structures can be seen of A3, A3G1, A3G2 and A3G3]
I did some more tests. I changed proteins. I was wondering if the protein might have an effect and now I’m using Fetuin which has a slightly different dominant glycan. We call this one A3 and that’s it here in the black bars. And using the same column again what’s really interesting with this result here is that I can actually see the appearance of the tri-antennary, tri-galactose sialic glycan here, where it’s not present at all in the starting sample, and then present at the end, albeit in lower concentrations but definitely there. So, I tested two different proteins and then I thought to myself, “OK, well what if I change the enzyme that I’ve immobilised on?”.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram of the structure of B4GalT1 and then the structure of ST6Gal1, and text appears: Changing the immobilised enzyme, B4GalT1 Found in trans-Golgi, Adds Beta, 1-4 galactose to the terminal GlcNAc, ST6Gal1, Found in trans-Golgi, Adds Alpha, 2-6 sialic acid to Galactose]
So, I go from B4GalT1 which I was using previously that adds that galactose residue to the terminal GlcNAc, and then I thought, “OK, well what if I can take something that adds to this terminal galactose residue?”. So, I grabbed human ST6Gal1, which is also found in the trans-Golgi, and that adds a sialic acid on to that galactose residue that I’ve just added on previously.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph showing preliminary results of tests using Fetuin on the left and structures of A3G3, A3G3S1, A3G3S2, and A3G3S3 on the right and text appears: Preliminary results – St6Gal1 – gfp]
So, I did some tests here. I used Fetuin and I found that yes there was an increase in the sialic acids present on my Fetuin sample but it clearly wasn’t as effective as the final, as the B4GalT1 column that I was using. The results were still there though, so I said to myself, “Alright, well I know both columns are working, yes I need optimisation but I know those columns are working, so what if I stick them together and combine and start with say this product here, minus those yellow sugars and finish all the way over here?”. Let’s see what happens.
[Image changes to show Nicholas talking in the top right and a new slide showing a photograph of the machine used to perform the tests and text appears: Sequentially connecting the two reactors]
So, here’s a picture of me and my columns. Your sample comes in this way, goes across, goes through the first addition, and once those glycans have been added on, it’ll go through to the second addition, and then I’ll collect at the other end. And these are the results I got from that slide there.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of test results using Fetuin-S-G on the left and the structures of A3, A3G3, and A3G3S1 on the right]
Now, this is a bit of a confusing slide so the important thing is to look at is, is this bar here, this one, and this one, and I’ve shown those down the bottom. And what we’re seeing happen is we, we start, this is all starting, compound, we start with the majority of tri-antennary glycans but without any galactose residues at all and after they pass through both columns I see a large increase in A3G3 which is the tri-galactose sialic glycan. And I also see an increase, which is this bar here, which is the A3G3 but with one sialic acid on it. So, my B4Gal column worked great, was excellent, but my sialic acid column left a little bit to be desired. And that was OK, this is all my preliminary testing, so it gave me a lot to go back on and a lot to look at.
[Image changes to show Nicholas talking in the top right and a new slide appears showing a diagram on the right and text appears on the left: Moving forward, Test enzymes from different hosts for effectiveness – bacterial vs human, Test different physical column parameters – residence time, amount of enzyme etc., Alternative enzymes for addition of glycans, Investigate alternative glycoproteins which also have great therapeutic and monetary value, Begin work on in vivo glycoengineering in P. pastoris]
And moving forward for some more of my, more part of my PhD, I really want to be testing enzymes from different hosts for their effectiveness. I really want to see, you know, what’s more effective in a column format, bacterial enzymes, human enzymes, even enzymes from other sources that aren’t bacterial or human. I would love to be able to test different physical parameters of the column itself. So residence time, the amount of enzyme present, any metal co-factors that I may use. Just looking at alternative enzymes in general. So, not just glycosyltransferases, other enzymes that may also be able to do that function. And I really want to investigate other glycoproteins on top of IGGs and monoclonal antibodies which also have a really good therapeutic end value. And of course later in this year I’ll start on the first half of that dual glycosylation system which is the in vivo glycoengineering.
[Image changes to show Nicholas talking in the top right and a new slide showing photos of Nicki Packer and Edward Moh and text appears: Distinguished Professor Nicki Packer, Dr Edward Moh, Thank you]
So, I just wanted to say a quick thank you to everyone for listening in today. I think my talk might have gone a little bit quick and a thank you to Nicki Packer and Edward Moh who have both really helped me out. These are my Macquarie University supervisors for my PhD. I’m really looking forward to continuing my PhD and I’m really thankful to the CSIRO for helping me achieve my goals through the FSP Scholarship. So, thank you.
[Image changes to show Colin Scott in the top right talking to the camera]
Colin Scott: Thanks Nicholas. We’ve got time now for one or two questions before moving on to the second speaker. The first question is from Tom P, and it’s actually a question I’d like to follow up with a few of my own thoughts as well.
[Image changes to show a photo of bushland and then the image changes to show Colin on the main screen talking and other participants can be seen in the bar at the top of the screen]
Tom asks, “Are the glycotransferase enzymes that you use affected by metal argons and if so is the HisTrap column that you were using, which would be full of nickel ions going to be a problem?”. And I guess I’d ask a similar question which is about contamination by nickel of the final product which is, you know, eventually going to be a human therapeutic, is nickel going to be a problem for you in those systems?
[Image changes to show Nicholas talking on the main screen and the participants can be seen inset at the top of the screen listening]
Nicholas DeBono: Yeah, so great questions. Yes, the glycosyltransferases are affected by metal ions. That’s what uses them, normally it’s Ng2 plus which is an essential co-factor in those glycosyltransferase enzymes. My plan is for the future to actually change the binding system that’s used inside the column to move away from a nickel NTA. I’m still investigating different binding systems. I was looking for one that like you said wouldn’t have, wouldn’t have the potential to loop anything with itself such as nickel ions into the downstream product.
[Image changes to show Colin talking on the main screen and Nicholas and other participants can be seen listening in the participant bar at the top of the screen]
Colin Scott: Thanks, so we’ll have a conversation about that in the near future because we’re looking at similar systems and we’ve got really nice silicon binding system that Raquel Walker is working with in the lab at the moment. So, we can have that discussion later.
[Image changes to show Nicholas talking to the camera and then listening and the other participants can be seen in the participant bar at the top of the screen]
Nicholas DeBono: Yeah, I’m really excited to have a chat with everyone about that.
[Image changes to show Colin talking to the camera on the main screen and the participants can be seen inset in the participant bar at the top listening]
Colin Scott: Great stuff. So, second question is from Daniel Winter, and he’s asking, “How the reaction rate of glycosylation is controlled as it passes through the column, do you have to make sure that there’s a very slow flow rate and so a long residence time, or do you have a semi-continuous process, where you have pauses in the flow rate to ensure that you get full glycosylation?”. And I guess you could ask even, you know, with a very complex glycosylation patterns, how would you control how they come out, and how do you control the homogeneity of the product that you get at the end?
[Image changes to show Nicholas talking on the main screen and other participants can be seen listening in the participant bar at the top of the screen]
Nicholas DeBono: Mm, yeah great question. So, the… first of all the more enzymes you add into the sequence, the higher you need your reaction completion to be at every single step. You know, if you have 90% completion at two steps, well you’ve only got 81% of your final product. So, that’s a big part of what I’m trying to do moving forward, which is optimise every single individual reactor to try and get as close to 100% completion as possible.
I haven’t had a chance to test how I can increase that glycosylation efficiency. At the moment I’m thinking that I’ll just be increasing residence time through the column, or potentially increasing the length of the column itself so that I can keep the same flow rate. But these are all factors that I haven’t had a chance to test yet. So, yeah I’ll probably be in touch with you to see what thoughts you have.
[Image changes to show Colin talking on the main screen and the other participants can be seen listening in the participant bar at the top of the screen]
Colin Scott: OK, that sounds like a really good plan… left to me now to wrap up and say hopefully everyone will be back for the next one of these seminars in the series. So, thank you very much everyone.
[Image shows Louise clapping inset in the participant bar at the top and Daniel talking]
Daniel Winter: Thank you all.
[Image changes to show Colin talking on the main screen and then the image changes to show Daniel listening on the main screen]
Colin Scott: And lots of clapping icons coming through. That’s very good. Thank you.
[Image shows Louise waving in the participant bar and then the image changes to show Louise’s photo on the main screen]
[Image appears of Louise Burton listening on the screen and then the image changes to show Colin Scott talking on the main screen and other participants can be seen in the bar at the top]
Colin Scott: Cool, we’re just hitting the record button now, excellent thank you Louise. I’d like to start the session by acknowledging the traditional owners of the lands on which we meet today. For both of our speakers today that’s the Gadigal people in the Sydney area. And for myself it’s the Ngunnawal people in the Canberra region. We’d like to acknowledge their Elders, past, present and emerging, and acknowledge that we are meeting on the lands for which they have been traditional custodians.
Before we kick off I would like to thank the Ops team, particularly Louise, for making sure that these sessions run as smoothly as possible and for giving me a script to run to. So, thank you very much for that Louise. And just to remind everyone if they want to ask questions please enter the questions into the Chat bar at the side of your screen, and Louise will harvest those questions as we go through and I’ll try and go through as many as we possibly can at the end of the session. If we can’t answer them during the session we’ll farm them out to the speakers and they should be able to answer them by email or otherwise so we’ll make sure that everyone gets their questions answered.
[Image continues to show Colin talking on the main screen while other participants can be seen in the participant bar at the top of the screen]
So, thanks for joining. Today is the Foundation Technologies session, for which I am happy to say, I’ve been given the pleasure of leading, and the Foundation Technologies application domain. And today we’ll be focussing largely on post translational modification of proteins, with our first talk from Nicholas DeBono, who is from, who I’m hoping hasn’t just dropped offline. Looks like you might…
[Image shows Nicholas talking in the participant bar while Colin listens]
Nicholas DeBono: No I’m still here.
Colin Scott: You’re still there, excellent…
[Image briefly shows Nicholas on the main screen and then the image changes to show Colin talking on the main screen and participants listening in the participant bar at the top of the screen]
Macquarie University, and is looking at artificial Golgis. He’s a PhD student who’s been supported by one of the FSP top ups. If you want to take it away Nick.
[Image changes to show Nicholas talking on the main screen while participants can be seen in the participant bar at the top of the screen]
Nicholas DeBono: Awesome, thank you. I’ll just share my screen. Hopefully, it works.
[Image changes to show a new slide on the main screen showing the CSIRO and Macquarie University logos, while Nicholas can be seen talking in the top right and text appears on the slide: A dual glycoengineering system; the prospects of a multi-avenue approach to targeted glycoprotein, Nicholas DeBono]
Everyone can see me? Yep, great OK. So, my name’s Nicholas DeBono. I just wanted to say a quick thank you to Colin and everyone else for giving me the opportunity to speak today. I’ll be presenting, I’m a PhD student, just approaching the end of my first year now. So, I’ve still got a bit to go. I’ll do my best to go through my, part of my PhD project, which is “A dual glycoengineering system and the prospects it has towards a multi-avenue approach towards producing targeted glycoproteins”. So, before I can kind of introduce my glycoengineering system, I feel like I need to introduce glycans. So, what are glycans?
[Image changes to show a new slide on the main screen showing a diagram describing glycans and Nicholas can be seen in the top right talking and text appears: What are glycans?]
Very, very broadly glycans are found as a post-translational modification on glycoconjugates, and they decorate the cell surface and a lot of proteins within the cell. There are a lot of different types of glycan conjugates you can find, from glycosphingolipids to proteoglycans, to glycoproteins themselves.
[Image shows a box appearing around the N-Glycans in the diagram on the slide and Nicholas can be seen talking in the top right corner]
Just for this talk today I’ll be focussing on just N-Glycans here. So, N-Glycans are characterised with their attachment to the asparagine residue through a consensus sequence on a protein. And they normally also have this core structure which you can see in the blue shadow.
[Image changes to show a new slide showing text: Why are glycans important? The presence of glycans on a protein alters how glycoproteins interact within the body, We are unable to modify these glycans the way we modify DNA and proteins]
So, why are our glycans important? Well, the presence of the glycan on a protein alters how that glycan interacts within the body. Additionally, we can’t modify glycans the way we would modify DNA or proteins because they’re not template driven the way DNA and proteins are. They are stochastically added, which means whatever is around is whatever adds on generally. Now, how do these glycans alter the glycoproteins?
[Image shows a diagram appearing below the text on the slide showing a flow chart diagram explaining how glycans can alter glycoproteins]
Well, if we take just one glycoprotein such as IGG, every single little modification of our glycan here – for example if we take off a core Fucose we could increase a certain receptor ligation by up to 100 fold, which promotes cytotoxicity. If we add sialic acid to the terminal end of our N-Glycan, we can change that again. If we add a bisecting GlcNAc residue we can again activate a different receptor. If we take things away we have a Terminal GalNAc different receptors are activated. So, I guess my point for this slide is, is the different glycan present on the different glycoproteins will all have different interactions. But why is this sort of an issue for glycoengineering?
[Image changes to show a new slide showing a diagram explaining why different glycans present on different glycoproteins will have different interactions and text appears above the diagram: Why can glycans be a problem?, Different organisms produce different glycans, making producing therapeutic glycoprotein production in in vivo production systems more difficult]
Well, when we’re trying to produce glycoproteins, or anything like that, what happens is the production system we use will produce a different glycan. So, generally human like glycans will be complex, maybe have a core Fucose on them, maybe will have sialic acids. If we produce our therapeutic proteins in mammalian cells they might have a different sialic acid on them. If we use plant cells we might have the presence of an Alpha 1,3 xylose. If we have insects we normally have paucimannosidic glycose, glycans sorry. And if we’re using these cells, which is a very popular production system for proteins, we have these huge ultra high mannose chains on our glycans. The presence of these will lead to super-quick clearance of our important therapeutic protein, which if we want it to stick around the body and do what it wants to do, can be an issue. So, we try and solve these glycan issues with glycoengineering.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: Current solutions of glycoengineering, glycoengineering is the process by which glycans are modified for a specific purpose or towards a targeted goal]
So, glycoengineering broadly is the process by which we can modify glycans towards our specific purpose or towards our targeted goal. There’s a couple of different ways we can do this glycoengineering.
[Image shows a diagram appearing on the bottom right showing how a glycan can be modified and text appears on the left: Genetic change, N-glycosite addition, deletion or move to POI, Addition of non-native GT’s or similar enzymes (e.g. sugar transporters), KO of endogenous GT’s or related enzymes, Environmental change in vivo, Growth conditions, substrate change (azidic sugars), Enzymatically modifying proteins post production in vitro, Addition of GT’s plus required components to a solution containing POI, EndoS glycosylation system]
First we can do it in vivo, inside the host cell through a genetic change. We could say, change at a specific site where glycan is added. So, no glycan site, change our amino acid sequence, and we have a glycan appearing. We can add non-native glycosyl transporters of similar enzymes to modify the types of glycans being added, or we could just knock some out and that will also change our glyco profile so to speak.
[Image changes to show a new diagram on the slide on the right showing Azide reporter generation]
We can also change the environment that we grow our cells in. If we change growth conditions for example, if we have human cells and we change the amount of oxygen present in the chamber that we grow those human cells in we can alter the glycol profile there, and we can also introduce non-natural sugars to give us reported glycans so that we can see where they’re going.
[Image shows a new diagram appearing on the slide on the right hand side showing a diagram explaining enzymatic modification using the EndoS glycosylation system]
Another modification is to do enzymatic modification. So, once we’ve produced our glycoproteins we can then modify them post production in vitro. And this is a really cool method called the EndoS glycosylation system. I won’t be talking into this a little, I won’t be going into this one but it is a very neat way to glycoengineer proteins. So, my point for this slide is that there are a lot of different solutions to glycoengineering and all of them have their advantages and different disadvantages, and a thought I had towards the start of my PhD was, “Well, what if I can combine multiple glycoengineering solutions together?”.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: A dual glycoengineering solution In vivo glycoengineering, Potential initial loss of fitness, Delay in results, Lower protein production without strain engineering, Cheaper long run average cost (if it works), High throughput capacity, Generally produces a heterogenous mixture of glycans]
So, if I was to take a glycoengineering solution that was mostly in vivo, well my potential downsides are that when I’m initially doing my genetic modification of that host I could have an initial loss of fitness, so my host won’t survive as well. Because it’s a life science and cells things take a while to grow, they take a while to get results from. In some host organisms the act of glycoengineering reduces the fitness of the cell which can lead to lower protein production overall. But the upside is averaged out over the long run, it’s generally cheaper if you can get it to work because once you’ve, you’re producing your glycoprotein you can scale up, and that leads to high throughput. However, this generally produces a heterogenous mixture of glycans. So, if we think back to that slide at the start where one specific glycoform can lead to a very specific effect, if I have a whole mix then that effect gets diluted.
[New text appears on the right of the slide: Traditional In vitro glycoengineering, Does not rely on fitness, Dependent on enzymatic availability or chemical synthesis pathways, Faster immediate result, Lower throughput, Higher long run average cost, More homogeneous mix of glycans]
If I was to go down the in vitro glycoengineering route, so not in life, in the test tube so to speak, of course this doesn’t rely on cell fitness anymore because I’ve taken it out of the cell but it is dependent on enzymatic availability or chemical synthesis pathways. So, essentially what enzymes are available for me to use to do my glycoengineering. I can get a faster result straight away because I don’t need to wait for cells to grow but generally my throughput is down a little bit, and enzymes can get expensive. So, over the long run that could be a high cost. But because I’m controlling what enzymes I’m adding in, I can generally get a more homogenous mix. My thought for this at the start of my PhD was, “Well what if I can, you know, combine the two together?”
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing text: In vivo glycoengineering, Minimise initial loss of fitness, Less delay in results, Higher protein production without strain engineering, More controllability, Cheaper long run average cost (if it works), in vitro glycoengineering, Does not rely on fitness, Less enzymes required for synthesis of product, Faster immediate result, Higher final throughput, Better controllability, Offset long run average cost]
That would lead to a general minimise of initial loss of fitness. I could theoretically get quicker results because I’m doing half in vivo, half in vitro. I could get higher protein production because I’m doing less strain engineering so that I get less loss of fitness. If I can just produce a starting glycan and then finish the modification in vitro I could have more controllability and that cheaper overall cost is there.
So, what… the solution I came up with for my dual glycoengineering solution was to combine a well-known glycoengineered protein production strain such as Pichia with a in vitro system that myself and a, my PhD supervisor came up with, Dr Edward Moh, called the artificial Golgi column, which is what I’ll be going into.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a diagram Pichia pastoris in a bio reactor then through a Golgi column to achieve a targeted glycan structure and text appears: A dual glycoengineering solution, Create a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation]
So, the way this system works is I’m planning on producing a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation. So, if we were to map it out we would start with our Pichia pastoris strain in a bio-reactor – I know it’s the best picture I could make – we start with our ultra high in mannose glycans, we do some strain engineering and we finally come up with a glycoprotein that has a much more truncated glycan structure. We can then take this glycan structure and put it inside my artificial Golgi column – which I’ll explain soon – modify that and come out with a final glycan, targeted glycan structure which is what we’re aiming for.
Now, that doesn’t have to be this structure here with the hybrid mannose and the terminal sialic acid. The beauty of the artificial Golgi column is that it’s tuneable and if I was to just swap out the enzymes in this column I could create a different glycan structure theoretically. So, I’ll be mostly talking about the AGC today but I’ll quickly duck into the Pichia pumic because it is a dual solution.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a photo of Pichia under a microscope and text appears: Chosen in vivo production host, Pichia pastoris, Grows to a high biodensity, Frequently used yeast production system, Glycoengineering tools exist, Already used for glycoengineering efforts]
So, the reason why I chose Pichia is it grows to a higher bio-density, it’s a frequently used yeast production system, it’s already been used for glycoengineering.
[Image shows an article appearing at the bottom of the screen showing a text heading: Glycoengineering of antibody (Herceptin) through yeast expression and in vitro enzymatic glycosylation]
And there are some really nice papers that have come out such as this paper here with really good examples of a dual system already through yeast expression and in vitro enzymatic glycosylation. Critically the different here is this in vitro enzymatic glycosylation was relatively slow whereas my artificial Golgi column solution is a little bit faster.
[Image changes to show Nicholas talking in the top right and a new slide can be seen showing a photo of Dr Carol Hartley and text appears: Chosen in vivo production host, Pichia pastoris, Grows to high biodensity, Frequently used yeast production system, Glycoengineering tools exist, Already used for glycoengineering efforts, Dr Carol J. Hartley, Expert in flow biocatalysis and P. Pastoris, To be started late 2021]
And helping me explore this part of the project, Dr Carol Hartley from CSIRO, who is an expert in flow biocatalysis and in Pichia down in Canberra. She’s going to be helping me out. I haven’t had a chance to start the Pichia glycoengineering just yet but I’m planning on starting in late this year.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing glycoengineered Pichia and text appears: My solution (MRes and PhD continuation), Create a dual glycosylation system that can produce specific glycoproteins with controllable glycosylation]
So, that’s our Pichia side, and then if we move, so we’ve, say we’ve glycoengineered our Pichia and we’ve produced a truncated glycan structure such as this from our high mannose and then we want to modify this even further to get our final product. How do we do that? Well, my solution was to put it through my artificial Golgi column. Now, I’ve called it an artificial Golgi column because I’m trying to replicate what happens in the back half of the Golgi inside cells, which is a lot of post translation modification, and additional protein folding.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing the Artificial Golgi Column and text appears: the Artificial Golgi Column – an in vitro post-production glycoengineering system]
And I’ve taken some of those glycosyltransferase enzymes and I have bound them to the inside of a column. The system I’ve used for this is just Ni-NTA, you know, HisTrap histidine column binding with my [14:24] glycosyltransferase enzyme, down to the beads inside this column here. All you add in at the start is your starting glycan structure, and your other essential factors such as your metal co-factor and your starting sugars. It goes through the column, you get your enzymatic reaction, and you get a hopefully final glycan product. The beauty of this system here is that I’ve, all the reactions I’ve done, happened in about one to two minutes, as opposed to a usual enzymatic reaction which you’d do in a test tube which can happen over hours.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram showing the B4GalT1 diagram and text appears: Immobilised enzyme used, B4GalT1, Found in trans-Golgi, Adds Beta, 1-4 galactose to terminal GlcNAc]
So, the first enzyme that I immobilised onto my column to test, and I’ve got some preliminary results to show you, is B4GalT1 which we normally find in the trans-Golgi, and it adds a galactose residue in a Beta, 1-4 form position to a terminal GlcNAc sugar. All these sugars mean slightly different types but if you just want to follow the colours that’s fine, I’ve got the key down here. But critically what I’m looking for in my results of my experiments that I’m about to show you is, pardon me, an addition of this galactose residue and a higher finishing concentration on my intact glycans.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of preliminary results using the Trastuzumab and the 3mm I.D on the left of the screen and a diagram of the three main types of glycans on the right can be seen and text appears: Preliminary Results]
So, what I did was the first test I used, I bought some commercial Trastuzumab sample. This is an IGG and I analysed its glycol profile at the start and I found that there were three main types of glycans. There was the FA2 glycan which is this black bar here and I’m going to call that my starting glycan. And then FA2G1 – G1 because it has one glycosyl transferase on it – and then FA2G2 obviously because it has two.
We can see from the start to after the reaction there has been a large decrease in the starting glycan, and a huge increase in the final dye glycosylated glycan here which told me that that’s great, that’s excellent. My column that adds galactose residues to glycan is working. Now, ideally I would like this to be, you know, 100% FA2G2, but these are preliminary results and the fact that this is working is really great for me. So, it means I can take it further and optimise it down the track.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of preliminary results using Fetuin and on the right structures can be seen of A3, A3G1, A3G2 and A3G3]
I did some more tests. I changed proteins. I was wondering if the protein might have an effect and now I’m using Fetuin which has a slightly different dominant glycan. We call this one A3 and that’s it here in the black bars. And using the same column again what’s really interesting with this result here is that I can actually see the appearance of the tri-antennary, tri-galactose sialic glycan here, where it’s not present at all in the starting sample, and then present at the end, albeit in lower concentrations but definitely there. So, I tested two different proteins and then I thought to myself, “OK, well what if I change the enzyme that I’ve immobilised on?”.
[Image changes to show Nicholas talking in the top right and a new slide showing a diagram of the structure of B4GalT1 and then the structure of ST6Gal1, and text appears: Changing the immobilised enzyme, B4GalT1 Found in trans-Golgi, Adds Beta, 1-4 galactose to the terminal GlcNAc, ST6Gal1, Found in trans-Golgi, Adds Alpha, 2-6 sialic acid to Galactose]
So, I go from B4GalT1 which I was using previously that adds that galactose residue to the terminal GlcNAc, and then I thought, “OK, well what if I can take something that adds to this terminal galactose residue?”. So, I grabbed human ST6Gal1, which is also found in the trans-Golgi, and that adds a sialic acid on to that galactose residue that I’ve just added on previously.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph showing preliminary results of tests using Fetuin on the left and structures of A3G3, A3G3S1, A3G3S2, and A3G3S3 on the right and text appears: Preliminary results – St6Gal1 – gfp]
So, I did some tests here. I used Fetuin and I found that yes there was an increase in the sialic acids present on my Fetuin sample but it clearly wasn’t as effective as the final, as the B4GalT1 column that I was using. The results were still there though, so I said to myself, “Alright, well I know both columns are working, yes I need optimisation but I know those columns are working, so what if I stick them together and combine and start with say this product here, minus those yellow sugars and finish all the way over here?”. Let’s see what happens.
[Image changes to show Nicholas talking in the top right and a new slide showing a photograph of the machine used to perform the tests and text appears: Sequentially connecting the two reactors]
So, here’s a picture of me and my columns. Your sample comes in this way, goes across, goes through the first addition, and once those glycans have been added on, it’ll go through to the second addition, and then I’ll collect at the other end. And these are the results I got from that slide there.
[Image changes to show Nicholas talking in the top right and a new slide showing a bar graph of test results using Fetuin-S-G on the left and the structures of A3, A3G3, and A3G3S1 on the right]
Now, this is a bit of a confusing slide so the important thing is to look at is, is this bar here, this one, and this one, and I’ve shown those down the bottom. And what we’re seeing happen is we, we start, this is all starting, compound, we start with the majority of tri-antennary glycans but without any galactose residues at all and after they pass through both columns I see a large increase in A3G3 which is the tri-galactose sialic glycan. And I also see an increase, which is this bar here, which is the A3G3 but with one sialic acid on it. So, my B4Gal column worked great, was excellent, but my sialic acid column left a little bit to be desired. And that was OK, this is all my preliminary testing, so it gave me a lot to go back on and a lot to look at.
[Image changes to show Nicholas talking in the top right and a new slide appears showing a diagram on the right and text appears on the left: Moving forward, Test enzymes from different hosts for effectiveness – bacterial vs human, Test different physical column parameters – residence time, amount of enzyme etc., Alternative enzymes for addition of glycans, Investigate alternative glycoproteins which also have great therapeutic and monetary value, Begin work on in vivo glycoengineering in P. pastoris]
And moving forward for some more of my, more part of my PhD, I really want to be testing enzymes from different hosts for their effectiveness. I really want to see, you know, what’s more effective in a column format, bacterial enzymes, human enzymes, even enzymes from other sources that aren’t bacterial or human. I would love to be able to test different physical parameters of the column itself. So residence time, the amount of enzyme present, any metal co-factors that I may use. Just looking at alternative enzymes in general. So, not just glycosyltransferases, other enzymes that may also be able to do that function. And I really want to investigate other glycoproteins on top of IGGs and monoclonal antibodies which also have a really good therapeutic end value. And of course later in this year I’ll start on the first half of that dual glycosylation system which is the in vivo glycoengineering.
[Image changes to show Nicholas talking in the top right and a new slide showing photos of Nicki Packer and Edward Moh and text appears: Distinguished Professor Nicki Packer, Dr Edward Moh, Thank you]
So, I just wanted to say a quick thank you to everyone for listening in today. I think my talk might have gone a little bit quick and a thank you to Nicki Packer and Edward Moh who have both really helped me out. These are my Macquarie University supervisors for my PhD. I’m really looking forward to continuing my PhD and I’m really thankful to the CSIRO for helping me achieve my goals through the FSP Scholarship. So, thank you.
[Image changes to show Colin Scott in the top right talking to the camera]
Colin Scott: Thanks Nicholas. We’ve got time now for one or two questions before moving on to the second speaker. The first question is from Tom P, and it’s actually a question I’d like to follow up with a few of my own thoughts as well.
[Image changes to show a photo of bushland and then the image changes to show Colin on the main screen talking and other participants can be seen in the bar at the top of the screen]
Tom asks, “Are the glycotransferase enzymes that you use affected by metal argons and if so is the HisTrap column that you were using, which would be full of nickel ions going to be a problem?”. And I guess I’d ask a similar question which is about contamination by nickel of the final product which is, you know, eventually going to be a human therapeutic, is nickel going to be a problem for you in those systems?
[Image changes to show Nicholas talking on the main screen and the participants can be seen inset at the top of the screen listening]
Nicholas DeBono: Yeah, so great questions. Yes, the glycosyltransferases are affected by metal ions. That’s what uses them, normally it’s Ng2 plus which is an essential co-factor in those glycosyltransferase enzymes. My plan is for the future to actually change the binding system that’s used inside the column to move away from a nickel NTA. I’m still investigating different binding systems. I was looking for one that like you said wouldn’t have, wouldn’t have the potential to loop anything with itself such as nickel ions into the downstream product.
[Image changes to show Colin talking on the main screen and Nicholas and other participants can be seen listening in the participant bar at the top of the screen]
Colin Scott: Thanks, so we’ll have a conversation about that in the near future because we’re looking at similar systems and we’ve got really nice silicon binding system that Raquel Walker is working with in the lab at the moment. So, we can have that discussion later.
[Image changes to show Nicholas talking to the camera and then listening and the other participants can be seen in the participant bar at the top of the screen]
Nicholas DeBono: Yeah, I’m really excited to have a chat with everyone about that.
[Image changes to show Colin talking to the camera on the main screen and the participants can be seen inset in the participant bar at the top listening]
Colin Scott: Great stuff. So, second question is from Daniel Winter, and he’s asking, “How the reaction rate of glycosylation is controlled as it passes through the column, do you have to make sure that there’s a very slow flow rate and so a long residence time, or do you have a semi-continuous process, where you have pauses in the flow rate to ensure that you get full glycosylation?”. And I guess you could ask even, you know, with a very complex glycosylation patterns, how would you control how they come out, and how do you control the homogeneity of the product that you get at the end?
[Image changes to show Nicholas talking on the main screen and other participants can be seen listening in the participant bar at the top of the screen]
Nicholas DeBono: Mm, yeah great question. So, the… first of all the more enzymes you add into the sequence, the higher you need your reaction completion to be at every single step. You know, if you have 90% completion at two steps, well you’ve only got 81% of your final product. So, that’s a big part of what I’m trying to do moving forward, which is optimise every single individual reactor to try and get as close to 100% completion as possible.
I haven’t had a chance to test how I can increase that glycosylation efficiency. At the moment I’m thinking that I’ll just be increasing residence time through the column, or potentially increasing the length of the column itself so that I can keep the same flow rate. But these are all factors that I haven’t had a chance to test yet. So, yeah I’ll probably be in touch with you to see what thoughts you have.
[Image changes to show Colin talking on the main screen and the other participants can be seen listening in the participant bar at the top of the screen]
Colin Scott: OK, that sounds like a really good plan… left to me now to wrap up and say hopefully everyone will be back for the next one of these seminars in the series. So, thank you very much everyone.
[Image shows Louise clapping inset in the participant bar at the top and Daniel talking]
Daniel Winter: Thank you all.
[Image changes to show Colin talking on the main screen and then the image changes to show Daniel listening on the main screen]
Colin Scott: And lots of clapping icons coming through. That’s very good. Thank you.
[Image shows Louise waving in the participant bar and then the image changes to show Louise’s photo on the main screen]