0:18
As mentioned in the introduction, part of our activities in Sartorius would be process development.
0:27
We did really a lot of different process developments for different clients.
0:32
They transferred quite a number of them to CMOs or to the sponsors.
0:37
Being in the gene therapy for more than 25 years, developing the chromatographic racing with the purpose to serve gene therapy, meaning no fears in the chromatographic media itself, increasing the recovery, which is basically the key parameter we are looking for when we are trying to build the process.
1:00
So we are talking about very expensive manufacturing costs, every percent which we can gain with the recovery means a lot of money.
1:11
We know that these treatments cost two, three, $4 million each.
1:17
So really very important would be the recovery, which is a bit different than thinking when you are producing proteins where usually you would go for capacity.
1:28
Not saying here is the capacity is not important, but once again, the recovery would be the most important part when you're starting to think about the manufacturing cost.
1:39
So plasma DNA of any kind, bigger, smaller, very big ones, even 30 kilobyte spare plasmids you can nicely purify with great recoveries. Lately we have minicircle, ministering and linear DNA. RNA molecules, single strain, double strain. Adenovirus vectors have been the leading gene therapy vector, 20 - 25 years ago it was regarded it as not safe.
2:10
One of the key issues at that time was 100 capsids, only one gene inside which boosted immune response. Kids were even dying during the treatments.
2:28
One of the reasons why we need to look now for the empty capsids.
2:32
In my opinion, we are going too far in the other direction now trying to get empties down to 10% or less.
2:43
With regard to new response, either it's 10 or 20%, it's not a big deal.
2:47
But the partials within AV, this is of much more concerned because the partials are carrying a lot of unwanted DNA sequences and nobody really knows what they would make in your body and when and what the results would be.
3:04
Vaccines, exosomes, bacteriophages, other particles and nucleic acids.
3:09
We also do some process development within plasma fractionation, but really it's mostly the gene therapy we are dealing with.
3:21
We have heard also some talks in the morning about analytics.
3:26
Without analytics you are blind.
3:28
We are still quite blind when we are dealing with the process development of the upstream which is another talk for another time. Today focuses more towards the downstream.
3:42
So we need more different systems, we need orthogonal methods.
3:47
Sometimes even two orthogonal methods are not enough because they may mislead you.
3:53
We deal with very complex samples, so really important to use orthogonal methods.
4:01
Of course, more analytics, more costs, more time.
4:07
This is why it will be good to have systems which will provide you with more data using just one small amount of sample.
4:14
As we are chromatography people, we always try to do everything also with the chromatography.
4:20
And here is one of the systems we built, it's PATfix.
4:24
You can see it at the Sartorius booth upstairs.
4:31
And what you are aiming for to build this system was first of all to quantify the amount of AV directly in the harvest.
4:39
So what you are doing in this case, we are using a Cation exchange column, we just lower the pH such that we basically would bind any serotype of the AV and we filter the sample.
4:55
Yeah, for five micrometre filter usually we inject 20 microlitres on the monolithic column.
5:05
In this case, that would be the chromatogram you see here. There are many different curves.
5:13
One of the important ones would be this one we see Cation exchanger and here I notify the DNA something probably is wrong.
5:22
It's not really wrong because this DNA is not alone.
5:25
It is in the complex with histones, highly charged, positively charged proteins, which really sticks to most of the chromatographic materials.
5:35
Sometimes you don't really see this material the chromatins, because they may stick inside the column.
5:45
You may see losing the capacity, you may see loss in the resolution.
5:54
So the free DNA of course is in the flow through, but most of the DNA after lysis would be in the histone complexes.
6:03
So big peak is shown in green.
6:07
We know that if you take fluorescent detector, DNA is not fluorescent.
6:12
So we need to find something which sticks to any DNA related impurity.
6:18
And this is exactly what we do.
6:21
So with the fluorescent detector, we can trace on the DNA related impurities.
6:28
And when we see such a big peak, we know that we have a lot of this stuff and we may need to think what to do before we go to the column.
6:38
If we go to the prep, if we enlarge this part so zoom in, that would be the zoom in of this part.
6:46
And now we see better several curves.
6:49
First one of them, the black one represents the multilight scattering detector.
6:55
So this is used as detector.
6:57
We don't use it as a standalone machine to measure the size of the particle.
7:02
We really wanted to have a robust detector which also is not as expensive as the other detectors.
7:12
And we are not going to use this detector to measure the empty and full.
7:17
We are going to measure the empty and full by chromatography.
7:21
So this is now total capsids, this AV capsids would be in this peak and we can quantify the amount of all AV capsids.
7:30
So in the Cation exchange, usually we don't really reach the resolution of the empty and full.
7:38
This method is used in the experiments, which I'm going to show you later on.
7:46
You detect a lot of different peaks, 280 the blue one, 260 the red. This gives you some indication either a peak is more of the DNA or more of the protein. The fluorescent detector is the green one.
8:03
So boosting the signals.
8:05
We see also some smaller amount of the DNA.
8:11
This one probably in the protein complex because we see 260 signal high this one as well.
8:20
But we see this one is not really overlapping with the protein signals.
8:25
There may be another complex we didn't really investigate it in. If you are now looking for DNA removal, we have an orthogonal method.
8:34
Its important to say that it is more sensitive than the PCR for the whole cell DNA.
8:42
If the whole cell DNA is packed inside the histones, it may not be able to replicate it by the primer because the static exclusion is really limiting this possibility. In some cases we put dual channels first and detector with the second channel we are going to trace residual proteins.
9:06
So we have orthogonal method.
9:08
Now for the capsids, which basically would use orthogonal to ELISA/PCR, we have the method to look for the residual DNA and we have also the method to look for residual proteins.
9:27
So its a nice system to have and it can basically be used in every step.
9:33
So you have more process control, you can develop your processes faster and more important, you can have better traceability when something goes wrong in the manufacturing.
9:46
And believe me, in almost every manufacturing process something goes wrong even after a few years of manufacturing stuff for the market.
9:56
Now going more to the process itself, before discussing the process, keep in mind that AV is not as stable as you may see it in the literature, because the ability of the AV usually would be represented by PCR.
10:13
But you don't really know if your genome is still inside the particle or if it's somewhere between the two particles which are already partially destroyed.
10:29
So not all the genomes are properly packed.
10:32
Especially if you are using lower pH, you are starting to destroy the capsids and you are losing the recovery.
10:41
So measuring the infectious titre or potency, you may see that already from left to right, from seven, you're already starting to lose the potency.
10:57
So keep in mind the AV is not as stable as usually seen in the literature.
11:06
The process, the platform and process, it would be nice to have it.
11:11
But believe me, it’s taken a decade or more to get to such platforms.
11:21
Every AV is different; every media used, every transaction involves scaling up many different things.
11:32
So that’s what we are aiming for but it will take time.
11:38
So most of the AVS are produced intracellularly.
11:41
So we need to do lysis.
11:43
There are few which are extracellular which are much nicer, much easier to purify, much easier to manufacture.
11:51
Unfortunately, still most require lysis.
11:56
So we have detergent.
11:57
Usually we use detergent lysis.
12:00
If you'd really like to study the lysis, do not freeze before you’ve already make some lysis.
12:06
Detergent is fine.
12:07
But when you do the scale up, when you go to the 500 litre reactor, 1000 litre reactor and you do the lysis with detergent, you have a headache with the filters because detergent make micelles and micelles are very difficult to control, so they'll be clogging your filters.
12:27
So salt will be better in most cases.
12:30
Salt would be OK.
12:31
Sometimes you need to add some detergents, not 0.5, maybe 0.1%. Salt also helps to prevent at least partially the chromatic structures being made.
12:43
Then usually we go with the DNA treatments.
12:46
Not a cheap step.
12:48
So the aim would be not to use the DNA.
12:52
So there are some other approaches which may help you not use the DNAse treatment.
12:59
One of them would be flocculation.
13:02
Recently we introduced solid phase extraction, which allows you to basically reduce the amount of the chromatic structures which produce a lot of headaches or also avoid the use of DNA in some cases.
13:18
The traditional way would be TFF/DNAase treatment which is also fine.
13:22
In some cases, if your expression is very good, you can go directly to the Cation exchange column.
13:31
After you prepare the material, you can go to Cation exchange, which we prefer. There are many different reasons for it including affinity.
13:37
And then the last step, empty/full/partial separation.
13:41
So we did the comparison between the affinity approach and strong ion exchange approach, in this case TFF/DNAse.
13:49
With the cation exchange, you need to reduce the pH down to 4 or even 3.5.
13:57
The lower the pH, the better the purity.
14:00
But of course we are out of the comfort zone of the stability, so we are destroying a bit of the product.
14:07
With affinity, we don't have this issue.
14:08
We go to the column and from the Cation exchange, you can already elute at neutral pH or even higher pH.
14:17
And then you go to the empty/ full/partial separation with affinity.
14:21
You need to elute with the lower pH 2.5, even 2.
14:27
Then sometimes citric buffer.
14:29
There may be some headache with the citric buffer.
14:34
If I have time, I'll raise this later on.
14:37
In principle, two different processes.
14:39
And if you measure what you get from this column, this column immediately after the column, you will not see the key difference.
14:45
The key difference will only be seen here.
14:54
If you have good expression, you really get quite nice purified peak.
14:57
If your expression is not that good, you will start to see some impurities here and maybe here.
15:03
But most of the junk flows through her and strongly binds to the column.
15:11
And then we loaded the material to the same column to the same QA column, and we can see that the material coming from the affinity a little sooner.
15:21
It's not because the column is not working properly, but because the first layer of cations and the anions around the AV is different.
15:30
The buffers used here are different than the buffers used here.
15:34
Keep in mind not only for the column but also for the transfection to the cell, the first layer of cations and anions is important.
15:44
It may change the property of binding; it may change also the other properties of the capsid.
15:53
So working with particles, not only with AV, also with LNPs, always take into account that your particle is charged from the surface and that some cations or anions may make quite strong bond with the surface.
16:17
So magnesium or calcium bind to the AV capsids very strongly.
16:22
This was presented by Fraser Wright 25 years ago in his paper.
16:28
To remove residual magnesium from the capsid you need to use EDTA which you don't want to use in the process.
16:35
Magnesium comes from the DNAse treatment but may also come from the impurities of your buffers.
16:42
Its really important to take into account when you are trying to develop a robust process.
16:47
Just to compare the two processes of orthogonal methods: PCR and PATfix.
16:58
So the total amount of capsids.
17:01
Of course each of them gives different data, because in one case you are using it just for the genome, in the other case you look for all the capsids.
17:15
Looking at the two columns, you don't really see such a big discrepancy in the recoveries yet.
17:20
The main difference is on the last column.
17:23
This is where a lot of mistakes are made because you compare columns after immediately after the and the column not taking into account what kind of product you still have, is it still intact to other particle or it's already partially degraded.
17:42
So the difference is really in the last column because the last column shows the issue or the damage which the column before or maybe preparation before made.
17:52
So in this case with both methods we ended up with abou30% more of the product which is a lot of money.
18:01
If you compare the impurities basically quite the same host cell DNA removal, pDNA removal, HCP and ETX impurity removal. It is comparable, although Cation exchanger does not show affinity but there is still removal.
18:16
OK, now we are coming to another issue, low recovery and also due to the damage capsids the affinity of material is not as good.
18:34
So again orthogonal methods, mass photometry or the PATfix system, the chromatography, in both cases we can see that the purity of the material obtained by affinity QA is not as good as by cation exchanger and anion exchanger.
18:53
So 55% in comparison to 70%.
18:57
So not only the recovery but the purity is an issue.
19:05
Comparing the capacities, usually both columns would have comparable capacity.
19:12
Take into account if you are using the monolith you can pump your samples very fast.
19:18
You can work with 1-2 column volume per minute, meaning you can go directly to the column.
19:25
If you have very good expression, you still basically get your purity.
19:31
And also important, the Cation exchange approach has been already well proven for several years in the manufacturing.
19:43
So the first manufacturing batches for use were prepared already in 2018. Now they’ve been used for five years for treatments.
19:53
Over 4000 kids have been successfully treated for SMA1, surviving very well.
20:01
And now we go to the empty and full traditional methods including CAR methods would not really provide you with a good resolution.
20:10
And now we are coming to the surface of the capsid.
20:17
So, Just playing with buffers, different cations and anions, we are able to improve the resolution by 5, even 10 times.
20:26
So from such kind of resolutions, so the same column, the same material, everything the same, just addition to the buffer A and buffer B, some cations and anions and a lot of work we don't yet have a mechanism behind to understand.
20:43
So the four months of the experiments can improve the resolution.
20:52
So empty, partial, full and damaged capsids need the fluorescent multi light scattering detector.
21:01
It is possible to scale up.
21:04
And if you are really picky, you can even improve this resolution by introducing the cell buffer.
21:12
We get the resolution empty, all these partials, and the fulls.
21:16
So better seen on the prep empty partials very heterogeneous and the fulls are now a bit separated.
21:25
We could get in this case with AV8 more than 99 pure full capsid with the fluorescent detector.
21:34
So fluorescent is much more sensitive for the impurities than UV detector.
21:41
Would it be possible to do this with AAV9?
21:43
No, not yet, not so far.
21:47
But in some cases you get such really nice peaks, from 60 to 80.
21:55
But if you put this on the mass photometry, yeah.
21:59
What is now your product?
22:00
Is this one or is this one?
22:02
If you look for the PCR, you have the genome all over.
22:06
So keep in mind if you are using the genome which is not feeling cure capsids fully like here, you will get such product.
22:15
And this is impossible to separate from all other partials you're full of is something which is completely problematic because you will have maybe 20, 30 or 40 different NGS structures.
22:29
And my advice here is go back and prolong your genome.
22:33
Just add 1 kilobase pair, 1.5 kilobase pair of nothing, just a single base that filled the genome, fill the particle with the genome of 4.6 or 4.7.
22:51
So there is just no AV platform yet available and it will still take a lot of years to get to this.
22:58
So my long experience in this, you need a lot of analytics.
23:05
We developed a system which we believe can be very helpful for everybody developing the process and then understanding the process.
23:15
In many cases replacing affinity with Cation exchange would result in better recovery, meaning much cheaper manufacturing.
23:24
30% of recovery increase is much cheaper manufacturing at the end of the day. Also better full/empty ratios and better purity.
23:37
If you take into consideration the surface of the capsid and start to play with cations and anions. you will be able to improve the resolution between the empty and more importantly partial and full capsids.
23:54
Take into account the manufacturing process early enough in order not to go back and try to redevelop your product.
24:08
So develop the product in a way that you know that you are going to be able to manufacture it properly later on.
