0:00
So good morning everyone.
0:03
Today I would like to talk about our new engineered show host cell lines for high title production for biopharmaceuticals.
0:11
Just a couple of slides to give introduction of who we are.
0:15
So Sartorius Cell and Development Centre is located in Ulm in Southern Germany.
0:21
We are actually a big facility with over 100 employees working on site with us.
0:28
We have a capacity to handle 50 cell and development projects every year.
0:35
And just to give you some more insights, in the last 15 years we have completed over 300 different customer projects across different protein modalities and very often achieving titers up to 10 grammes per litre.
0:50
Currently, we have over 80 molecules in clinic and we have received market approvals for additional 8 molecules.
1:00
The agenda for today would be I would first give a short introduction about our Joe cell and development platform.
1:09
Next, I will tell you more about how we engineered our chose cells.
1:13
Here I will mainly focus on how we identified our molecular target and how we performed all the validations.
1:19
And finally, I will tell you more about our CLD campaigns where we evaluated our current wild type post and compared to our novel engineered host across different protein modalities.
1:34
So coming to the CLD technology, the CLD process begins when we received the sequence for the antibody from our customers that we synthesise and clone into our proprietary CLD vector.
1:49
The vector is then transfected into our host chose cells following by metabolic selection and generation of large pools.
1:59
From the large pools, we perform single cell cloning and the clones are further evaluated on growth parameters and productivity.
2:08
And finally, we select the top 4 clones.
2:12
I will not go into details of each and every step but I would like to emphasise on the single step which is the single cell cloning.
2:22
We use a device called Cell selector to to perform our single cell clonings.
2:28
It's a semi automated system with robotic arms and inbuilt microscope in it.
2:35
And to perform a single cell cloning using this device, you start with seeding your cells at low densities using the nano well plates.
2:43
The nano well plates are basically 24 well plate.
2:46
But each well of this 24 well plate is further subdivided into 4000 nano wells.
2:52
The concept here is that each clone will occupy the nano wells.
2:57
Although they are separated, they will still share the same media.
3:01
So by using this strategy we overcome lot of problems that one faces when you use the traditional fax sorting, flow cytometer based sorting approaches.
3:13
So we have a really good outgrowth of cells and they're also in general very happy and growing very fast.
3:22
The advantage of having microscope are, are twofolds.
3:25
First, it takes photos that each and every step of the cell, single cell cloning base.
3:33
This allows to trace back our monoclonality which is an important factor when it comes to regulatory authorities.
3:40
And secondly, we take a lot of the the microscope while taking photos, it records lot of different parameters for each of the clones with respect to the growth, viability, productivity, etcetera.
3:54
So all this data that we have collected now we put it into our our machine learning algorithms, which allows us to predict a high performing clones already within the four days of single cell cloning.
4:08
So this is just an example how we can use now the machine learning tools along with our single cell cloning device to identify high producers very early on in the cell and development as a as a risk mitigation strategy.
4:22
I'm happy to talk about more on a lot of different experts of it in the break, but then I will move about the main topic for today's talk.
4:30
So here I would like to focus more on how we identified our molecular targets and how we performed validations in the last several years.
4:40
We spend a lot of time to to improve our in house gene editing capabilities.
4:47
The next step was for us to identify right molecular targets which when we knock out would improve our CLT performance.
4:56
There are a lot of different ways of one how one could do the screening, but we decided to go for more targeted approach.
5:03
So our goal was to identify a molecular targets which are energy intensive for the cells to make and to secrete in the media.
5:14
And our hypothesis was that if you knockout energy intensive endogenous host proteins then we might be able to divert all the energy that is saved to make this proteins to our gene of interest.
5:31
So keeping this in mind, we decided to analyse 4 different producer clones.
5:37
We started Ember to fed batch process and took samples at different time points throughout the the Fed batch process.
5:46
The samples were then further analysed using mass spectrometry.
5:50
The image in the left is just to show you how.
5:53
It's an example of how the analysis looks like.
5:57
So we have a Venn diagram showing different molecular targets that are secreted by different clones during the Fed fetch process.
6:05
So we were only interested in the ones that is common for all the clones during the process, not only in one time point but throughout the different time points.
6:17
So just to summarise what we found.
6:18
So we identified 1254 molecular targets.
6:24
These are basically all the proteins that are secreted by the clones in the supernatant during the Fed patch process.
6:29
But of this, only 351 proteins were common for all the clones and for all the different time points.
6:37
Of this 351 clones, only 67 were abundantly expressed, while most of them were below the quantification limit.
6:46
So we decided to focus on this 67 molecular targets that we identified in the screen.
6:52
Just a site.
6:53
Note that although the 65 targets that we identified only represents 5% of the total proteins that is secreted by the clones, it represents almost 63% of the total protein mass that was secreted during the Fed patch process.
7:11
So then the next step would be to individually validate this molecular targets.
7:16
So we started with 67 targets that we identified in the screen and we added further 10 targets by data mining in the literature.
7:24
So in total we have 77 different targets.
7:27
So we the way we do it is we take our wild type host cells, we individually knockout each and every gene that we identified in the screen and we performed large pools.
7:40
The pools are then further transpected with the model antibody followed by selection and standard shape flask fat batch process, just to give you an example how the performance evaluations after the fat batch process looks like.
7:54
So we are mainly interested in how the cells were growing.
7:57
So the peak VCC, the productivity parameters like specific productivity and final title and then the overall process duration.
8:07
So 0 here would indicate that there was no difference in the in any of the parameters compared to the controls or the Y type.
8:14
Negative or minus would indicate a negative impact on the growth parameters and productivity while plus would indicate a positive impact.
8:24
So just to summarise the screening.
8:26
So we screened 77 targets out of which 47 targets had no impact on either growth parameters or titers.
8:34
19 targets showed a negative impact on titers by 5 lethal targets, meaning when we knock this genes out the this results in cell death.
8:45
But we were able to identify one target that significantly improved protein titers and this is what we further evaluated.
8:53
So coming to the next step here, I want to also focus on how we generate our new host cell lines.
9:00
So we again started with our our wild type host cells.
9:04
We, we performed genetic engineering, meaning we knocked out the one target that we identified and generated the host pools.
9:13
But this time we also went ahead and we did a single cell cloning to identify a lot of different clones.
9:20
From this hundreds of clones that we identified, we narrowed narrowed it down to top 35 engineered host cell clones.
9:28
This screening was based on strict parameters about monoclonality to ensure that this clone is indeed coming from one single clone on day zero.
9:37
This is possible using the cell selected device that I mentioned before.
9:40
And also we did a thorough genetopic characterizations to ensure that they were really knockout.
9:46
And no, none of the proteins is produced from any of the genetic alleles, for example.
9:53
Then the 35 engineered clones were further transfected with our model antibody.
9:58
Following selection, we started with the shape classified batch process and thorough evaluation of all these 35 clones.
10:06
What you see here in the bottom panel is the performance of the top 4 clones that we identified.
10:12
So what you see it in the black is the wild type of host cells.
10:17
What is startling is once the wild type cells reach the peak BCC, it shows the sudden decline in the BCC.
10:26
But compared to that of all the knockout clones that we analysed, show decreased peak BCC and elongated viability.
10:35
And and if you look at the titles, they have significantly higher titles compared to the wild type host cells.
10:42
What it basically suggests is the lower peak BCC means that even with the low amount of cells, it is able to produce a lot of proteins.
10:51
So from this we selected the best performing clone and then this is what we used now for all our future CLD campaigns.
11:00
So all the data that I've showed here so far was done by using one model antibody.
11:06
So we wanted to make sure that it's not product specific, but it's true across different protein modalities.
11:12
So we then took the the best performing clone and then we transvected the pools using different types of antibodies, so fusion molecules bispecific and IgG 1 and we were happy to see that it was quite reproducible.
11:28
The the data that we showed showed in the earlier slide that we still still see a decrease in the peak cell density.
11:36
We show an overall average increase of one point fold in their titles.
11:42
And because the peak BCC is low, we see a huge improvement in the cell specific productivity.
11:49
So we were so far very happy with all the data that we that we found.
11:53
And then next we decided to do a proper CLD campaign where we do a side by side comparison of our wild type host versus our new engineered host using our CLD platform.
12:05
We used three different products.
12:07
This time we used the IgG 1, IgG 4 and FC fusion molecule.
12:12
And here we did not stop at the clone generation, but we followed up with stability studies as well as we did scale up studies to ensure that you know, all the observations that we made in in the small scale or scaled down models are also true when we scale up the process.
12:30
Here I will not show you the data for all, each and every step, but I will show you the data from the top 12 prones using Ember 15 system, then how we do the step 80 studies.
12:41
And then I will show you the data from the bio reactors.
12:46
Now coming to the Ember 15 data for the top 12 clones for each product and for Wildtap as well As for the new engineered clone, we see similar to what I showed earlier that we see a reduction in the peak PCC increase in specific productivity and overall titles.
13:06
And here we did stability studies.
13:09
So basically we cultivate the cells over the period of eight weeks and we have the cells go through 70 generations at time point of two weeks and time point of eight weeks.
13:21
We do shake phosphate patch to make sure that the stability is preserved and there is no drop in the protein titers as well as at the same time point we look at the copy number analysis to ensure that there is no instability in the copies in the in the process of eight weeks or over 70 generations.
13:40
So as you can see in the left side, so this is the data from from the Fed patches normalised to the controls.
13:50
So here you see that all the titers that remain overall consistent apart from clone 9 which shows a slight reduction in titers.
13:58
And if you now look at the copy numbers, I would like to emphasise that even with comparatively low copies of the of the product, we are able to get high titers.
14:08
And also the copy numbers does not change between the Week 2 and week 8, establishing that the stability is quite preserved over the period of 70 generations.
14:20
Just just as a side note.
14:21
So here like during all the processes for the new engineered host cells, we see 90% stability compared to what we have for the current wild type host which is around 80% stability.
14:36
And finally, this is scale up using the universal 5 feet of bioreactor system.
14:41
So here we selected the top clones from the for the IgG 4 and IgG 1.
14:47
And here just wanted to make sure that all the observations that we made about the peak PCC, about the cell growth and the titers are also grew when we scale up the systems.
14:57
And here you can nicely see if you look at the yellow dots, the one with the dashed line is the wild type.
15:03
And you see that it reaches very High Peak PCC.
15:06
But the engineered ones, they have still very low peak PCC.
15:11
And now if you look at the titers, which is on the right Y axis, you see a huge improvement in the final titles for both the different molecules.
15:24
So with this, I would like to already summarise what I talked so far.
15:31
So we successfully implemented targeted proteomic strategy to identify key engineering targets.
15:39
We also generated engineered knockout choke host cells and show that you know it.
15:45
It has a robust performance across different modalities, that is, IgG execution and bispecific molecules.
15:53
And finally, we did also 3 CLD campaigns and further demonstrated the superior performance of our knockout loans compared to our current wild type host across different parameters, including stability, which is one of the most important, important parameters.
16:11
With this, I would like to acknowledge a lot of different teams.
16:15
So on our side, we have a product development team, which were mainly involved in doing all the experiments in the lab.
16:21
We have the operations team, we have the quality and regulatory affairs team and the corporate research team.
16:28
So with this, I've already end my talk, and I'm open to any questions you have.
16:32
Thanks.
