0:49
Just so can I boil it down: 10-15 years ago I was at Horizon Discovery where I started using AAV for cell line engineering.
0:58
And through that we're producing AAV and of course HEK293 cells.
1:05
So I'm glad to share what we've been doing here at Asimov and how we've pushed that in terms of production and manufacture, but also looking at therapeutic application as well.
1:17
So for the presentation today, I'm going to cover the main topics around how you can come to Asimov and use our capabilities around looking at a one stop shop from the GOI, doing the payload design, incorporating promoters or types of optimizations for the CDS.
1:43
And then also how our production system can push those highs titers in terms of reaching up to 1E12 across many different stereotypes.
1:56
So I'll start with an overview of Asimov and then work through and look at payload design and then also the production system for a final summary.
2:07
So why am I at Asimov?
2:08
I'm at Asimov because we bring this hybrid approach in terms of software capabilities and then biology as well and bring that together.
2:16
So there's a melting pot of individuals at our headquarters in Boston which spun out of MIT with synthetic biology capabilities, but then bringing engineers and biologists to solve these issues that we see with the manufacturer of biotherapeutics.
2:36
So this is grounded on the foundation of cell biologists who are creating new GMP whole cell lines.
2:42
We've got synthetic biologists who are looking at building new genetic parts, new genetic systems to help with that expression.
2:50
And then moving towards the computational side, we have data-driven models that have been built by our computational biologists and then we bring all that together in our CAD software.
3:02
So this software allows us to intersect what's going on within the cell and design vectors.
3:09
And I'll talk about that more later.
3:11
And how do you access this technology?
3:14
We deploy that into our customer laboratories and that's part of my role and my team's role to help with that.
3:21
Or we do cell line development services at our base in Boston as well.
3:27
So we set up in 2017, but we're rapidly getting a good strong customer base across the number of innovators, CDMOs as well.
3:38
So where do we operate?
3:40
We operate across therapeutic applications.
3:43
So how can we produce these different modalities that are going into the patient?
3:49
So we've got capabilities around biologics, the CHO edge system with lentivirus production as well.
3:57
And RNA edge system is another offshoot that we're currently exploring.
4:02
Today I'll focus on AAV Edge for production of adeno associated virus.
4:09
If you want to talk to us about any of the other systems that we've built, we're at booth 20 on just the other side of that wall.
4:16
And yeah, please stop by and have a discussion with us.
4:19
But I'll now move forward with our AAV capabilities.
4:26
So it's an integrated approach that we've taken.
4:30
We're not just building an expression system; we’re also considering how we can have an impact on that therapeutic design as well.
4:38
So the earlier that you can come to us, the better so that we can then hard code in capabilities before you need to kind of progress forward to the clinic.
4:49
And the areas that we deal with within therapeutic design is around tissue specific promoters, but also promoters for tunability as well, sequence optimization of the gene of interest.
5:04
And then if the gene of interest is toxic, we have silencing capabilities in the manufacture as well.
5:09
And then the production system is focused around a very robust cell line that's combined with a plasmid system that's delivered like a kit.
5:20
And then we also have model guided process development.
5:24
So delve into each of these individually and going forward.
5:31
So now I'll focus on the payload design.
5:36
So it's about modular design approach of how we can use this CAD software to help with the design of the payload and helping the scientists that need to create these new AAV therapeutics.
5:56
So within kernel shown here on the right hand side, we have capabilities around easy design, construct design, sequence optimization, also modelling of how those different parts can have an impact on expression of your gene of interest.
6:12
So every day I'm using this system to help with the design of expression vectors.
6:18
And then within kernel, there's these applications that are very specific for AAV and I'll delve into each three of these areas in the next few slides.
6:29
So in terms of the sequence optimization to improve AAV expression, we've used machine learning to optimise the therapeutic transgenes here for Luxturna and Zolgensma, the gene of interests are fused to a fluorescent protein to help with the assay to understand the impact of the optimization on the expression of the gene.
6:56
So we optimise these sequences, we generate the AAV in our system and then we transduce HEK293 cells to understand the impact of that optimization.
7:09
And we're using flow cytometry to see the expression.
7:13
So here shown in the middle pane, we have Luxturna and then Zolgensma on the right hand side.
7:20
But key for this is the control in grey non transduced.
7:24
And then the light blue colour is the original sequence for the viral product.
7:30
And then the optimised sequence is shown there in pink.
7:33
So we see this move in the mean for essence given as a confidence that we've also impacted the gene expression for this this gene of interest.
7:46
And here we're showing a 7x improvement.
7:49
Of course this could be different across different gene of interest, but we're very confident in our approach for improving expression of the gene of interest.
8:02
The other area is around now we've got a catalogue of promoters, but also, we spend a lot of time using our machine learning capabilities and artificial intelligence to delve into potential promoters that would have a greater on target expression versus off target expression for AAV applications.
8:24
So we've generated a library of tissue specific promoters.
8:28
I'm here showing the ones for cardiac expression.
8:33
I went dwell on this too much longer.
8:35
But yeah, please see us at the booth to understand this in more detail.
8:40
But showing the in vivo study here, it's very clear with the expression highlighted there for this cardiac promoter that's very centralised, low off target, all on target for this.
8:53
So be happy to talk about this more later on as well.
8:59
Then in terms of, we've heard some areas where people are having difficulty with toxic genes of interest and that's where we come with our concepts and ideas to try and overcome that.
9:12
So using a typical triple plasmid system for AAV expression, if that payload plasmid has a transgene that creates a toxic protein for HEK293 cells that are used for the manufacturer of AAV, you may see an impact on that overall titer being produced from that cell line.
9:33
That cell line might, have poor viability and the AAV is not being produced at the maximum capacity that it should be.
9:42
So we've come up with a genetic module to overcome that.
9:48
So we're able to then within the manufacturing HEK293 cell line to lower knock down the expression of the transgene.
10:00
So then that's not expressed within that manufacturing cell line.
10:04
So then you get the cells growing at their preferred viability capacity to produce the AAV.
10:12
So here's shown on the right hand side, we've got the transgene GFP with no gene of interest silence in clear GFP expression.
10:23
But then incorporating our science in plasmid a genetic module, then we have completely no expression of that GFP.
10:31
And then in terms of the functional titer and taking that virus and then transducing a target cell line, we still have good comparable titer production and also, we still have the expression of the transgene and the target cell.
10:54
So now I'll move on to the AAV Edge production system.
11:01
So it's comprised mainly of these components where we have the robust HEK293 cell line and transient plasmid system and then model guided PD.
11:18
And I'll delve into these again into some more details, but these are the three pillars of what we offer.
11:25
Our scientists are building this as well for future manufacturing scale up.
11:34
So in terms of the cell line, the cell line was established from a collection ATC adherent HEK293 cell line.
11:45
We suspension adapted that into a chemically defined media and then generated a number of different clones that were then interrogated for ways that they could improve AAV production and growth characteristics as well.
12:03
And then once we've selected that high performing clone that's being GMP banked to ICHQ standards, that clone is now ready for GMP manufacturer in the future for transient production of AAV.
12:18
We have here looked at some of the details around the cell line.
12:22
The cell line recovers very fast from the revival from the cryopreserve stock.
12:28
It's allowing you to kind of get into that seed train as quick as possible.
12:34
It has a very good 25 hour doubling time.
12:38
And then also one of the key benefits that we see is that as we're selecting that particular clone, we're looking for clones that were not very clumpy. 10-15 years ago, when I was working with the 293-F cells from Thermo.
12:52
You know, they're not a great cell line to work with, but having that, that non clumpy cell line allows us to count those cells very reliably, robustly.
13:03
So that then when we're coming to doing the GMP scale production, we can be very consistent with lots of lots and the transfection ability as well as that cell line.
13:21
Now moving on to the model guided process development.
13:24
So no, we're not just a company of cell biologists.
13:29
We've also got capabilities around machine learning and when we're building these expression systems for whatever therapeutic we're targeting, we're always considering how that's going to be scaled up at the end of the day.
13:45
So we always ensure that our cell line meets that scalability and the expression capabilities.
13:55
And key to that is using machine learning to kind of understand what's happening within that process and then apply future changes that provide benefits to the overall process.
14:08
So for the AAV system that we're talking about today, AAV Edge, we've done 500 plus runs looking at 12 different factors, parameters that can impact overall AAV production, but also the quality of the AAV being produced and tested in different scales as well to ensure that we've got that scalability and then transferability of the process as well.
14:36
So looking at different facilities, but then also different bioreactor types as well to give that robustness and a system that's ready to transfer from our lab to your lab or from our lab to a CDMO as well.
14:53
But in terms of a case study where we applied our machine learning approach here shown on the right hand side there, we started with the classical triple plasmid system, but then quickly identified moving towards a two plasmid system.
15:11
And I'll show you more details about that later, what that comprises of.
15:18
But then looking at that two plasmid system in the experiment two, we found other parameters that we could improve and adjust to get that real boost in percentage full and a slight increase in titer.
15:32
And then another step change that we made was then looking at genetic boosters as well that we can incorporate within this production system to then go up to experiment five where we get this well capsulated AAV high 85% full capsid and then over the 1E12 titer as well.
15:58
Just want to mention as well that these titers that I'm sharing today are all based off pulling the harvest and directly doing the title at that point we're not we're not purifying or any additional purification.
16:14
These are cruise supernatant titers that we're sharing today.
16:21
Then in terms of the two plasmid system, so as I mentioned removed from the triple transaction classic to the two plasmid system where we have the rep and the cap with the gene of interest on one plasmid.
16:39
And then the helper plasmid has the genetic booster, then the E2A, E4, and VA helper genes as well.
16:49
And we don't just want to supply you a vial of plasmid.
16:52
This is about helping with the design of the payload and those plasmids that are being incorporated into the expression system.
17:00
So we have a streamlined workflow that is incorporated within the kernel that then helps with that design of the initial gene of interest plasmid helps with the sequence optimization.
17:15
We've got capabilities now that we’ll be deploying later around the automated compiling of those plasmids.
17:23
So, that will help with enhanced design, but then also reduce any errors that you get, cut and paste errors that I used to have building plasmids.
17:34
And then within kernel you can track that order as well.
17:37
So we can do the product the creation of that plasmid at our Boston site in the US, track the status.
17:44
And then also within kernel, we have the protocols housed and the electronic lab notebook as well.
17:51
The electronic lab notebook has got an additional function which I quite like, but it's also that functions part of the genetic design where we've got a AI system where you can ask it questions and it will provide you details by searching through our repositories and searching through the electronic lab notebooks to find… can I ask you a question about what titers reached this level in these particular experiments and it will drop back that answer to you.
18:24
I find it very useful to sort of delve into all of those lab experiments that I've got coming through.
18:33
Then in terms of just bringing it all together, how does that two plasmid system and the cell line come into one and this is showing you know across different serotypes.
18:47
Again, this is cruise supernatant titers not purified that we can with clinically relevant serotypes from 2, 5, 6, 8, and 9, we're getting up to that E12 vg/ml based on DDPCR assay.
19:08
And then looking at the assessment of empty full again, we've shown already the AAV9 going up to 85%.
19:18
But for the other clinically relevant serotypes as well, we're getting very good full capsid, so we're getting full plasmid from 30% and above.
19:35
AAV2 is still one of those areas that we'd need to do more work on.
19:40
So yeah, watch this space as we try and improve that as well.
19:47
So I've got 5 minutes left, so I'll just go through a benchmark case study.
19:53
So taking an off the shelf system that's out there and comparing that to the capabilities that we have.
20:01
So using VPC 2.0 and off the shelf free plasmid system, we then also compared that using the same VPC 2.0 cell line but with our plasmid system and then our full package which is our cell line with our two plasmid system.
20:19
And then we're doing key analytics around the quality, the performance, the safety, the last one there, the replication competent, we haven't got that data to share today, but we have seen early indications that we could do sequencing of the virus that we don't see rep and cap genes present.
20:38
So watch this space, it'll be coming out in the future.
20:42
But to delve into this a little bit deeper, so taking that crude harvest and running it on an SDS page, we get this improved desired ratio 1:1:10 in terms of VP1, VP2, and VP3. With our full platform system, we're getting very close to that.
21:05
You know, this is a semi quantitative analysis from SDS page.
21:08
We'll be looking to take this again and delve into this further, but there's definitely an improvement over the benchmark system.
21:20
Then delving into the percent full.
21:25
So benchmarking against the current standard percent full is very comparable.
21:33
But again, we get this titer increase and some of this is based around our plasmid system being incorporated.
21:38
Remember the hybrid has the same cell line as the benchmark, but then it's the two plasmid system that we've changed.
21:45
And then with AAV edge, it has the full cell line and plasmid.
21:53
Then looking at impurities, so for residual host cell protein and host cell DNA, again, we get this improvement with the hybrid and AAV edge.
22:02
Interestingly, how that could be associated with changing the plasmid, we want to delve into a little bit deeper.
22:08
But yeah, there's definitely an improvement over the benchmark system that we used.
22:14
And then the last but not final piece is the analytical ultra centrifugation.
22:20
So the more gold standard for looking at full capsid versus empty capsid because also we can now capture partial capsid.
22:29
So remember this is expressing the GFP.
22:32
So it's not fully utilising the payload in AAV, but we'll be exploring this more and more with different payload sizes as we go forward.
22:42
But again, we've got that improvement with the full capsid for AAV edge.
22:48
So now I'll just summarise, bring it all together.
22:51
So we have therapeutic design production system and then the high performing platform that I've shared with you today.
22:57
So please stop by our booth 20 just in the main hall and we'd love to discuss more with you today.
23:05
And on that.
23:06
Thank you for your attention and I'll take any questions that you may have.
