"Hello, everyone, and welcome to our webinar today with the title Speed up cell line development at Sartorius using our 4CellChow platform.

Before we start, I would like to first introduce myself. My name is Lisa Blaschke, and I am actually product manager for our 4CellChow platform.

And I'm very happy to guide you through the webinar today.

So let me introduce our speaker today. It's Juliana Bitchoff. She's a product specialist for cell line development at Sartoris. She holds a Master of Science in Pharmaceutical Biotechnology.

And she began her career at Sartorius in the product development department, where she was actively developing three generations of Sartorius proprietary CLD platform.

So she has all the experience and currently she works with prospective clients to identify the customers' needs and to outline their COD proof scope.

So with that, I will hand over to Juliana. We can get started and enjoy the webinar everyone.

So thank you, Lisa, for the kind introduction. And before I start, I just shortly wanted to say hello. Thank you all for joining.

And I hope you will enjoy the webinar. So as Lisa said, I'm Juliana Bishop. I'm working as product specialist for cell line development at Sartorius and I am based in Ulm, South Germany.

So today the topic of the webinar will be that I will talk about the 4CellChow platform.

And I will explain to you how we work for cell line development or how it is performed the cell line development at Sartorius and especially with new technologies we have implemented to speed up the whole process. At the agenda of today, I would like to start to give you a short introduction into Sartorius Stedim CellCat.

I will explain the general workflow of the CLD platform. And my major topic will be that I will explain in a bit more detail and with a bit of a scientific background, which new technologies we have implemented to find to improve the high producer selection and also to shorten the timeline. And to make it a bit of a full picture for you with the complete CELD service, I will shortly touch a bit on other platform features.

And the last topic will be that I will give you a short overview on recently launched other packages that are a bit in the context of CRD. So this will be intensified processing and key knockout services.

Then at last, I will give a short summary and then we have time for the questions.

So Sartorius is a global company. We have around sixty locations worldwide and roughly fourteen thousand employees are working for Sartorius. The focus of Sartorius is really strongly on the biopharmaceutical market. So we have a portfolio with a lot of lab products such as, for example, you might have heard of the Optech device or the IQ screener, a lot of tools and consumables.

And all these things cover the whole process for the customer coming from molecular development, through cell line and process development, and then up to upstream and downstream processing. So with the stainless steel bioreactors and filters. And we will be focusing today on this part with the webinar.

So yeah, the topic of today is cell line development. And I would like to show you with this history timeline a bit where cell line development at Sartorius is coming from, where it all started. So it started back in two thousand and five with the when CELLCA was founded by Doctor. Asus Kahili. At this time, we had a very strong collaboration with Benshla Biopharm and we were also placed in a building on their ground.

And then the upcoming years afterwards, a small team of lab experts invested a lot of time in developing the first platform version. So in this case, that means they focused on generating a robust whole cell line combined with an effective expression vector and also then the media system.

So the platform fall behind it is that it works for basically every product type that might come along. One big milestone in this timeline is then in twenty fifteen when CELLINK acquired by Sartorius, so since that we are called Sartorius Steel and CELLINK.

And then it was followed by another big milestone for us, that a new building, a new facility was built.

So now we are based in Ulm since twenty nineteen.

And today we are here, we have done a lot of CELLINK projects already. We have two market approvals. And we just recently, in the end of last year, we launched the newest platform version for the fifth version, and this will be the major topic I will be talking about today.

So a bit more into the CRD service we offer, I would like to start to introduce or to highlight to you the key features of the platform.

So, for example, speed is always a topic in biopharmaceutical industry. Since the pandemic situation, it got even more important. So we can now with the Bot CellCode platform, go from DNA to RCB in nine weeks. And during cell line development, you have to screen a lot of clones.

So in this platform version, we strongly focused on improving and implementing new technologies to improve the high producer selection, and to do this also as early as possible in the process. With the new platform version, we can reach tight product titles up to ten gram per liter. And also you have seen this in the timeframe as well. We have done CELD for quite a while now.

We have worked by today on two forty projects already.

Some others are still ongoing. And so we have a quite a big track record and we have worked with a lot of different product molecules. So our platform is quite versatile, that is proven in the track record. So we've worked not only with monoclonal antibodies, but also with bi and multi specifics.

We have done some projects where we generated satellites for enzymes and hormones. We had also a lot of sticky molecules, which gets more and more trendy now. And we have also worked with biosimilars. And what you have seen also in the timeline is that we constantly also optimize our platform.

So if there are new technology trends or new demands from the market, we attach this, really on, yeah, we work constantly on that.

So, line development is only one part of the Flow platform. So if a customer comes to us for cell line development, we have of course some throughput analytics that we can do in parallel to the cell line development, but we also have the capability to do protein characterization. So for example, to really check the product quality and the functionality, and there we have a lot of ready to use assays that are already developed, or we can also start assay development from scratch on the customer specific needs. Furthermore, we can also do then with the identified lead clone, EMP mast and working cell banks in a special dedicated facility, and the cell banks can be released by a qualified person. And these bios from the cell bank, can also move them to the biosafety testing lab team, so that we can check that the cell line is free of contaminants such as microplasma or viruses.

And all these parts are supported by our own four cell media system, the Smart Shopper system, Smart Shopper media system. So this is a medium that is chemically defined and it got and was developed for optimal performance with our CLT platform.

In our opinion, there are three pillars that make up a good CLD platform. So one is of course the whole cell line and the expression vector. We at Sartorius, are working with the whole DG34 whole cell line. That cell grows in suspension and is fast dividing. We have a full cell line history report that one needs to hand into the authorities.

And also in the recent years, the development team has worked a lot on optimizing the expression vectors. Basically every element there was optimized. The vector is freedom to operate.

The second pillar is the media system. This media system got especially optimized to work best with our cell line in the platform. Actually there was no media screening required.

The medium is of course chemically defined, it does not have any animal components, and it really supports a good cell growth and a good productivity.

And together with that, the last, the third pillar is that we have a predefined scale up process. So at the end of a CELLINK development project, we perform a five liter bioreactor and this process protocol can be then easily taken for scaling up, to at least two thousand liters. And we also transfer this protocol to the later production facility, if it's either the client itself or to a CMO.

Now to the main part of the presentation, the CIG workflow and all the improvements we have done in a bit more detail. I would like to explain this what changes we have implemented to this platform version by comparing it to the old platform workflow. So this is shown here in the upper part. So we can check the cells with our expression vector. We generated an expanded tool, and these tools went into single cell sorting with a Pax device. So fluorescent activated cell sorting, the clones were expanded and selected, and a certain amount of clones, standard it's the top twenty four clones, they were chosen based on fake flask batch titer, and then move into a deeper clone evaluation in the Ember fifteen system.

And then out of that you get the top four clones where an RCB is generated. With the old platform, it was around fourteen weeks.

You take the top four clones into a stability study and then at the end you have your final clone.

So with the new platform version, we completely got rid of the pool phase. So already on day ten after transaction, we move into single fact learning, and here comes, and here we implemented to use another device, it's called the Cell Selector. I will explain this in the next slide, how it works in more detail.

And the Cell Selector offers very mild conditions, so the cells after single cell cloning recover faster. So we could also shorten the timeline of clone expansion.

And then we introduced also another layer of clone selection. So we are not choosing the top four clones for the Embrane system based on the batch titers, but we implemented a ninety six well small scale fed batch. So it's based on batch titers. I will also explain the system later on. And then out of the and then it's the clone evaluation with the MBR fifteen, you get again your top four clones and RCB is generated. But here we save compared to the old version five weeks. So now the RCB the chartered timeline is ready after nine weeks, and then it follows the same procedure as in the old platform version.

So as I said, in a new platform version, we are using the Cell Selector to perform the single sec loading. And actually this device has many technical highlights. So if you look on this picture that shows the device, it has a robotic arm that does the automatic cell picking from one plate to the other.

Also attached is a fluorescent microscope that does the imaging of the wells.

And you can screen or image a lot of cells simultaneously with this device.

Also with the CellCelector comes an imaging software, where we can analyze the images the microscope takes and we do the high producer identification.

The cells then get picked with a glass capillary. The glass capillary is single use, so it will only be used for one product to prevent cross contamination.

And the actual generation of single cells happens quite easily in these special twenty four well, nano well plates. So what you do is you take a pipette, you distribute equally the cell suspension over the wells and the cells singly go into the nano wells by gravitation.

So to explain this a bit more, it's the twenty four well plate, but each of these wells has approximately more than four thousand nano wells. One nano well holds a volume of four nanoliters, so the wells are really small and the cells can't hide in there.

Also the device has proven compliance with regulatory requirements.

And yeah, another very nice advantage of these using these plates is that the cells sit single as singles in the nano wells, but they share a medium layer which lays on top of it.

They share this medium layer and through this shared medium layer, they can cross, they can have a crosstalk in a molecular way. But cells do not jump from one cell to each other, but cells are a bit like human, they like the proximity of each other. So all this offers really mild condition to the cells.

Therefore that the cells find these mild conditions, it's possible to do the single cell cloning already that early in the process.

These mild conditions with a Cell Selector also have a positive effect on the clone out growth after the single cell cloning. So if you compare the clone outgrowth after single cell cloning with the fact sorter and with the cell selector, you see in this graph where we did this outgrowth for six test products, that in black the percentage of outgrown clones with a Cell Selector is a lot higher, so up to fifty percent higher compared to the fax device.

For the pre selection of clones, we developed also a novel productivity assay. How this assay works is shown in this small schematic.

So after some days after the single cells are laid out in a nanowell, we add special beads to the cells. These beads are coated with protein A and the cell in these few days secretes the products or the antibody into the outer well part and the secreted antibody can bind to protein A on the beads. And then on top of that, we add a fluorescently labeled antibody. So basically based on the level of fluorescence that gets detected by the microscope, we can say, is this clone likely to be a high producer in the end or not?

And you see in these pictures, in these microscope pictures down here, how the process during CLD works, sorry, how the process of CLD works. So on day zero and day one, we take pictures to detect the monophonality of the cells. So you have to be sure that really only one cell at the beginning of single cell cloning was present in the well. Then on day four, we detect the outgrowth of the clone.

So it has to be at least eight cells in the well. So we can say the clone has outgrown. Then you see already here these black dots. These black dots are the beads that we add also on day four.

And then we detect the fluorescence of the clone. So basically you see here in the middle, this clone has a higher fluorescence intensity than the wells that are around it. So this seems to be a very nice one because it's monoclonal, it has outgrown and it shows a high fluorescence. So the device will take this clone automatically and transfer it to the next well stage.

So during the development phase, of course we wanted to evaluate just that this principle works. So we were taking some clones that had a high fluorescence, and also we were taking some clones that had a low fluorescence and checked them in a fat batch condition. So we inoculated small scale fat batches and indeed we saw that the high fluorescent clones also had higher titers later on in the fat batch compared to the low ones. What you can also do is you can also select cells without the productivity assays, so you can select them randomly.

Randomly does not mean we pick blindly the cells, but we take them based on all the other parameters the Cell Selector provides to us. So up to eighty parameters are collected with the Cell Selector and a special combination out of that. We take also into consideration to select a clone. But however, the chance to find a high produce we see as increase if you use the productivity assay.

Next, I will like to shortly touch upon a very interesting topic, what the development team has done.

They were trying or they are working on developing a predictive data platform. So working with data modeling for cell line development in order to predict.

So the goal of this was to use the data of the CellCelector and with applying a data model, being able to predict the performance of the clone later in the ENDRA fifteen system.

But this here shortly explains how it was done. So at first, the first step is that you collect the data. In this case, it came from five test products. The data source is the CellCelector.

So the data is a lot of different cell parameters and fluorescence parameters. And the raw data itself from all these different products needed to be pre processed. So for the pre processing, we have done a round of data normalization. The data normalization is very important to be done, because if you have data from different products, clones tend to in the plot tend to cluster to form clusters per product.

And this data clustering, it really, yeah, can have a negative influence of the learning of the data model. So if you perform data modeling, then you get a more representative data set.

And then you can actually start with moving forward to the data modeling.

So the challenges for the data modeling were that of course you have to generate a data model that works for a lot of different products. And also it is really almost impossible or hard to achieve to predict a really precise titer value, and also to consider all the biological complexities that different molecules can bring.

So as a first step in data modeling, the first model was a PCA classification. So the classification gave us the best prediction. So that means that we were not saying a clone will have a certain titer, but we said that we can classify the clone. So saying a clone will have a certain titer threshold. So for example, everything above four gram per liter will be claimed as a high producer and everything below that as a low producer. So the two classes high and low producer.

To be a bit more confident with the model and also to improve it, We did also a machine learning method. So these are called random forest classifier and artificial network classifier. So basically that means if three models will give you the same result, you can be quite sure that correct what it's saying.

And these data models will be integrated into a one button solution in Simka, just that the lab teams on a daily basis will work more easily with the system. And then the output is a ranking of clones that says, that tells you, okay, these are the high producers.

We have run this with some test products and we got already really nice results.

So we evaluated this under fed batch conditions and compared to different selection strategies. So one part of the clones were chosen randomly, so randomly with using the productivity assay and all the other parameters from the Cell Selector. The other selection strategy was using on top of that the data model. And you see that in the fed bed condition, the titer of the top four clones in pink is increased. But also more important is that the system tries to get away from only finding a lucky hit, but to increase the mean titer of all your clones, to shift it upwards. But of course, you can understand that a data model is profiting from a big data set. So our aim is to constantly improve the system with feeding it with more data in the future.

As I said before, as a second layer, and we see this a bit more as a fine tuning tool to increase the chance to find high producers, we implemented a ninety six well small scale fat batch, and these small scale fat batches work as a scaled down model to the ember. The ember has fifteen milliliters as a working volume, and the small scale fat batch only has three hundred microliters.

And the cells are cultivated in ninety six well plate and for this we are using the tooth system. The tooth system is a special platelet and the platelet consists out of different filter layers and these filter layers prevent the very small working volume to evaporate, but they also still allow oxygen or gas transfer from the outside to the wells. So cells will be still getting oxygen.

These plates can be installed basically in every shaking incubator, and the daily handling of the small scale fat batches we have done, we are doing with a liquid handler system.

During the development phase, we of course wanted to check if the scale down principle of small scale fed batch to M15 works. So this is a small snapshot out of the data.

We have inoculated, we have taken eight clones and inoculated a small scale package with them. And we also in parallel run an EMBER fifteen run. And what you see is that the data is quite nicely correlating. So viable cell concentration profile looks almost the same.

The viability is quite comparable and more important is of course that the product titers, not only with the height of the titer, but also with the ranking of the clones is very comparable. So the orange clone we saw in the small scale fat batch is quite a nice one. Also this has been seen again in the AMBER system. So the principle of the scale down model of the AMBER fifteen system works.

And the other small scale fat batch titer. And if you check the normalized MBO titers, you see that with the small scale fat batch, you can get rid of these really low and non producers. So the range of the titer is not so big anymore. And you can also shift the mean titer of clones upwards. So you increase the likelihood that it or you improve the high producer selection.

Another benefit is that with a small scale fat batch, you can predict also the molecular performance. So for example, if you have a different product in a small scale fat batch and in an amber, you see for example for product A, which was a difficult to express product.

We saw already in the small scale fed batch that it's more on the load titer level. And this was again the same seen in the EMBER fifteen system. And yeah, load titer compared to other products.

So in conclusion to this topic, we demonstrated the performance of the Faucet platform in the EMBER fifteen system. As I said before, the EMBER fifteen system is a scaled down model to a bioreactor. So here with these three test products, you see that we really with this platform can achieve products or product titles up to ten gram per liter. So here is one, the other clone reads seven gram per liter, we see a nice viability curve and also nice growth. And also it can address the challenges that you face with a typical expressed protein, which you see in this case where the title level is also quite nice.

So to make the picture of the cell line development service complete, I will now shortly touch a bit upon other key platform features.

So as I said before, the three pillars of a good platform is also one of them is the medium. So we invested also time on improving our media system. So we launched recently an upgraded version.

The new medium is called the Smart Show. We increased there the filter ability, also the feed stability got improved and the harvest ability. And also we saw that cell specific productivities are systematically higher. So it means that with the amount of cells you have, you get more product. And the performance is also supporting high titer package processes.

And there is also a variant of the Smart Show, it's called Smart Show PE, that works for intensified process formats. And also the product quality was not negatively impacted by the optimizations we have done.

So once you have your top four clones defined after the AMBER run, and you have an RCB prepared, we in the cell line development workflow jump right into a stability study. The stability study is eight weeks long and above and covers around seventy generations. And this will show the phenotypical stability of the clone. So what we are doing is we run head patches with cells from RCB stage and with cells from the time after eight weeks. And these head batch performances should not differ more than twenty percent.

So and here in this graph, if you take all the clones, we generated in this case from seventy four projects, and more than eighty percent of the clones are stable. And we never had a case where a customer where one of the top, where none of the top four clones was stable.

Also as I said before, our process is quite nicely stable and scalable. So if we compare here the performance, the viable cell concentration, the viability titre and lactate level in a five liter scale compared with a two hundred and two thousand liter scale, you see that all data sets are really nicely comparable and scalable. Also, the scalability is proven in product quality. So if you take here the glycosylation patterns in the different stages or different scales, they are all really the same.

So the last topic of this webinar will be shortly about the packages we have launched concerning intensified processing and G knockout services. So service we offer for intensified processing is called high inoculation batch. So it addresses the topic to have a protocol for intensified process format in place. So with the fine ocalation fed batch, the volumetric productivity per production process is increased.

So it means we can achieve higher titers, or the principle behind intensified process is you get more out of your production run. And one important feature of this is that you can shift the biomass production. So generating the cells, you shift it from this large and very or more expensive production process to smaller pre stage and then and that saves you a lot of cost of goods.

So how it works is that for the biomass production for generating cells, use an n minus one fusion process in a rock emulsion system to generate the cells. This takes about a week and then we can inoculate the high inoculation fat batch in a bioreactor, in a stainless steel bioreactor. Both systems will be run with the medium type I mentioned before. So that's Smart Sho PE, the perfusion medium, which is a base on medium that in general supports the intensified process format.

So this is a small snapshot out of the development phase. So for the development of the process design, ten different clones were taken and a lot of different conditions were checked. So the inoculation density, different feeding strategies and so on. And then in five liter scale, we did proof of concept runs with six different clones. And what you see, when you especially look graphs, see that at the time the process is finished, the titer is increased for the high inoculation, a bit in yellow, or one could also say, I shortened the very, very expensive production process and get the same height level as I had before with a normal fed batch process.

And the other topic that gets more and more relevant in the biopharmaceutical industry is gene engineering. So gene engineering can be used to improve the product quality or also the efficiency of a product. And as a first product, we have launched the FUT8 knockout. So this is more an example target to be seen. FUT8 is the gene that is responsible or that code for the Fukushil Transferase. The Fukushilotransferase is responsible to transfer a Fukose molecule to a monoclonal antibody.

However, the Fukushilation of the antibody has a negative impact or can decrease the efficiency of the product. So especially concerning the ADCC function, the antibody dependent cellular cytotoxicity, So meaning the product is less efficient in it use in using cell death to the target cell. But if you knock out this gene, you will get an afrocosylated antibody. And this is much more efficient in doing its mode of action. But as I said, the Food Aid Knockout service is more to be seen as a first product.

We have now the systems analytics established, and we can basically do also customized knockout. So if a customer has another target, we can also perform the knockout here and work on other different product quality types.

And the service can be integrated in parallel into the COD process and does not have too much time addition on top.

How the principle works compared to the standard platform is shown here. So we do not only transacted gene of interest, we also transact the knockout enzyme. And on the stage of single cell cloning, perform knockout screening. So this knockout screening is to genetically prove and to select the clones that have the knockout on a genetic basis. And these clones are then taken also into the clone evaluation in the EMBER fifteen system. But afterwards, we have to do a proof of the afoxylation in a functional way. So we do glyco analytical assays to really prove, okay, the product is afoxylated.

Then we generate also the cell bank, the research cell banks, but this will be approximately one week later than the standard process.

This data set gives proof that for seven test clones that were performed in the development phase, the knockout of the gene does not influence the PET batch titer in the end. So you see that compared to the wild type clones, the knockout clones show comparable ranges of titers in a fed batch process. Also the cell specific productivity is not influenced. And more important is here that we prove that the knockout clones have one hundred percent hundred percent are focused related. So no focused relation was measured here compared to the wild type where levels reached up to eighty percent. And also what is not shown here on this slide, the growth performance in a fed batch titers is not influenced.

So now I'm coming to the end of the webinar with a small summary. So I have been spoken about the four CellChow platform, where we can go in the shortest timeline from DNA to RCB in nine weeks, and we can reach product titers up to ten grams per liters. In the new CHO platform version, we have implemented a lot of new technologies that address especially the selection of high producers. These were the CELL Selector for single cell cloning, we also implemented a predictive data model, and also implemented the ninety six small scale fat batch as a second layer for high producer selection. Also, have optimized the medium for the cell line development. So we improved the durability, harvestability, productivity and stability.

And we in general for our platform have a proven scalability of the process and clones are stable. And other parts, we have recently launched address intensified processing with the high inoculation set batch and we also are able to perform the knockout service for customers.

So with that, I would like to thank you for listening and I'm looking forward questions you have.

Very good. Thank you very much Juliana for this nice presentation and the interesting information about all the updates.

The audience, if you haven't done so yet, you can paste your questions into the chat now.

Yeah, we already have some questions, but before we go to the Q and A sessions, we want to make this little announcement.

If you want to learn more about our COD services and what you can gain through it, you can join our free event in March or on March sixteenth, the Sartorius Summit twenty three, which is called Shaping COD Solutions through Expert Insight.

And you also have the link down here.

So let me check.

Me review the questions. We already have a few.

So the first question, Juliana, for you, how can you achieve that you don't need the pool phase anymore by using the CELL Selector?

So what I have explained before is that the CELL Selector offers really mild conditions to the cells during single cell cloning.

So one part of that is what I haven't mentioned before, that the Cell Selector device does sit in an incubator flow box. So that means it's, of course, a sterile environment. But also, as it functions as an incubator, the cells will be provided with temperature, and also with co2 gassing and humidity. So they during the time the cells are handled with a device, they will find the same conditions as in the incubator, which is really good for their health, let's say.

And the other topic I mentioned before is that in the plate and the twenty four nanowell plate, they share this medium layer which lays on top on the nanowells and the cells can crosstalk. So crosstalk means, as I said, that they can share on a molecular basis.

They can cross talk on a molecular way. So this means more signal proteins are shared. Node cells are jumping around. But all these things offer very mild conditions would, which means also the cells grow out better. And also, makes it able that we can really after ten days after transaction do the single site cloning. And if you want to shorten the process, then for sure, this pool phase that takes several times, yeah, needs to be cut out. Otherwise, it's not working to shorten the timeline, at least in our case.

Okay, thank you very much for this nice explanation.

There is another question on the cell selector. I think you might need to explain that again a little bit with the GFP signal or the fluorescence signal. So"