0:00
Ross Walton is a senior application scientist for biologics at Unchained Labs, and he has published work on single particle protein analysis by transmission electron microscopy and the also on the direction of the innate immune system and viruses in cancer.
0:22
So he did his undergraduate work at UC Berkeley and then earned his PhD in molecular genetics in microbiology from at Duke University.
0:32
And yeah, so we are very much looking forward to Doctor Walton's talk if we can pull it up.
2:02
So, but I'll at least get started with my introduction.
2:04
So my name is Ross.
2:06
I am working for Unchained Labs.
2:08
We sell a variety of products that are useful for looking at kind of all kinds of aspects of biologics research and gene therapy research.
2:17
So if you've got a problem in either of those spaces, there's a distinct possibility that we could help you find a solution for that issue.
2:25
We are headquartered in Pleasanton, CA, which is in the Bay Area.
2:30
And yeah, we try to have fun.
2:32
So hopefully that will come across and I'll be talking to you today about how to fast track formulation screens of biologics.
2:41
So one of the things that I did this summer was actually watch the Olympics because it was on and there was an event that I really enjoyed watching, which was the track cycling events.
2:48
So I don't know if you guys have ever watched track cycling, but it's really, you know, it's looks like this is you've got a bunch of cyclists up on a banked track like this.
2:57
And the races are really interesting because they start with the all these cyclists sort of jockeying for position and very carefully and methodically trying to navigate around each other and find the best line on the track.
3:10
Then when the starting goes off, starting gun goes off, they Sprint like crazy as fast as they possibly can to get around this track.
3:17
And in my mind, that's kind of like how I approach formulation streaming.
3:21
So you can very methodically and very carefully plan out your experiments.
3:25
But once you're ready to execute those experiments, you want them to be finished as quickly as possible so you can get to the data and actually start making decisions about which formulations are optimal for your particular biologic or your LNP or whatever sample you happen to be looking at.
3:40
Now.
3:40
To that end, Unchain Labs offers four different products which form the Biologics Formulations Dream Team.
3:46
So we have Stunner, which is a plate based system for doing protein quantification as well as looking at the size and polydispersity of protein molecules.
3:55
We have Unagi which is a benchtop buffer exchange platform which can handle anywhere from 0.5 up to 8 millilitres of sample and has eight different sort of sample holders.
4:06
So we can actually exchange quite a bit of sample in that system.
4:09
Then we've got Uncle, which is a the 1st thermal stability platform for looking at biologics as well as looking at viral vectors and lipid nanoparticles.
4:19
And then last, well, but not but certainly not least, the most recent addition of this group is Honeybun, which is a viscometer and rheometer that is specifically designed for protein solutions.
4:30
And so for the case study that I'm going to talk to you about today, I took four different monoclonal antibodies, 2 of which are commercially available, adalimumab and trastuzumab.
4:39
I'm sure you're familiar with those ones.
4:41
And then buffer exchanged them into the same formulation, which was 10 millimolar histidine at pH 6, but with several different excipients.
4:49
So all of them had polysorbate 80 at the same concentration.
4:52
But in one of the formulations, we had sodium chloride, one of them we had sucrose and one of them we had arginine, and one of them we had a combination of sucrose and arginine.
5:01
A lot of you guys who are probably familiar with formulations are not in your head.
5:04
You've seen all of these before.
5:06
You know, sodium chloride and sucrose is commonly added as for a lyophilization stability.
5:11
You know sodium chloride often necessary for a lot of lipid and nanoparticles.
5:15
Arginine typically gets added to protein solutions that are of high viscosity in order to try and decrease their viscosity.
5:21
And that's particularly important if you're trying starting to consider subcutaneous injections rather than intravenous injections.
5:28
Subcutaneous injections tend to be at much higher concentrations and are much more viscous and adding arginine can in some cases help reduce that viscosity.
5:35
And then the last one combination of sucrose and arginine.
5:40
So we'll start off with trying to get our proteins are four different monoclonal antibodies into the proper formulations.
5:47
And the way to do that is through buffer exchange.
5:49
Unagi can handle up to 8 samples at a time with volumes anywhere from 450 microliters up to 8 millilitres and takes only about 15 minutes’ worth of hands on time in order to set up.
5:59
So extremely flexible system bench top buffer exchange.
6:03
And it's also great because like you said, everything's handled automatically.
6:07
You just set up the software, load your samples and can walk away from it and it'll even give you text messages when it's done, which is something I've really appreciated.
6:15
So the way that Unagi works is through positive pressure, UFDF, ultra-filtration, diafiltration.
6:22
Basically you load your samples, Unagi measures the sample volume, pressurises the sample and when the sample gets pressurised that actually forces the buffer that's currently in the sample out through the bottom of this little consumable, what we call the Una.
6:36
So that bottom of the consumable is actually a molecular weight, and a molecular weight cut off filter membrane.
6:41
So you know, when we have a variety of different pore sizes available for those Unas, then after pressurisation cycle or after pressurisation, the Unagi measures the sample volume again and refills it up to whatever setting you get.
6:57
You told it to with the new buffer.
6:59
It repeats that cycle until it reaches whatever percent exchange you defined at the beginning of the experiment.
7:04
Typically for my experiments, I usually go to 96, but you go to 99, 99.9% exchange.
7:10
Then as a last step, it will also concentrate that sample.
7:14
So reduce that sample volume one more time in case you know, had to do exchange, you wanted to do exchanges that say 1 mg per ml, but you want to do your testing at 10 mgs per ml.
7:24
Yeah, the cycle time as well as the cycle pressure is actually adaptive.
7:28
So whatever is going on, whatever unique thing is going on with a particular sample, Unagi is tracking it and responding to it.
7:36
The other nice thing that about Unagi is that during that pressurisation cycle, it is doing orbital mixing the entire time.
7:43
Now that is great because that prevents this dead infiltration that tends to happen with certain filter based buffer exchange methodologies.
7:51
You know you'll get membrane fouling at which point you're you know slows down the entire buffer exchange process.
7:57
But with the orbital mixing that doesn't happen.
7:59
So it goes faster and keep your solutions a lot more homogeneous.
8:05
After buffer exchange we then have to do quantification which we do with Stunner based off of just UV-Vis quantification and this also has ADLS module in it, dynamic light scattering.
8:14
So that tells us the size and polydispersity of our protein samples, which gives us a clue as to whether or not these samples are actually happy in the formulations we've added into.
8:24
And as an added bonus, Stunner can also do B22 and KD, which is a way of measuring the aggregation propensity of an antibody in a particular formulation.
8:35
Does all that from just two microliters of sample.
8:37
And since it's a plate based system, we can run 96 samples at a time.
8:42
So when you're looking thinking about having done buffer exchange and that sort of first thing that you need to look at when you are running a formulation screen, you need to figure out first how much of the antibody do I actually have?
8:54
Is it the amount that I expected to have?
8:56
What was my percent recoveries coming off of the buffer exchange system?
8:59
And then also do I have any aggregates, have those formed during the course of buffer exchange?
9:04
Is the antibody actually happy being inside of the formulation that we're tested in and that's where UV-Vis answers that protein quantification method or question and then DLS dynamic light scattering answers that is it aggregated question.
9:20
So here we actually have the results from that screen that I talked about right at the beginning of our four different monoclonal antibodies in our with our four different excipients.
9:30
And generally, I'd say, you know, the antibodies were all pretty happy in the formulations we tested with 1 notable exception.
9:37
Adalimumab really did not like being exchanged into the sucrose solution.
9:40
It aggregated like crazy, which you can tell because the hydrodynamic diameter of the adalimumab solution is a lot higher than the others.
9:48
Typically monoclonal antibodies have hydrodynamic diameter of a round 10 nanometres and typically their polydispersity index or PDI is about 0.1 when they are happy in a formulation, when they are mono dispersed.
10:01
In the case of adalimumab with sucrose, both the hydrodynamic diameter and the PDI were a lot higher than what we would have wanted.
10:10
So this was just a bad formulation for this particular antibody, and we excluded it from the rest of the rest of our studies.
10:17
But you also probably seen that a lot of these formulations have PDIs that are above that 0.1 threshold.
10:24
Now why?
10:24
But their size is in the size range that you'd expect.
10:29
You also may be noticing that in every case of that's happening, it is in a sucrose containing formulation.
10:37
And in order to understand what's going on with that, we actually have to look a little bit deeper at the DLS results rather than the sort of top line summary.
10:46
And what we can see is the reason for it pretty much is because sucrose actually shows up in DLS results.
10:52
So here we have intensity distributions, which is another type of data that comes off of DLS.
10:57
When you've got a nice, you know, single population of particle sizes in solution, you get this nice single peak right here.
11:05
When you have multiple sizes, you'll get multiple peaks.
11:07
And in the case of monoclonal antibody 1 plus sucrose, you can see a peak showing up right here at about 0.8 nanometres and that is the sucrose peak that shows up whenever you're using sucrose in a formulation and looking at it in DLS, it's not in it, you know, it's an indicator.
11:21
There's nothing wrong, it just shows up in the data.
11:24
And but by looking at these intensity distributions, you actually get a little bit more information about what's going on with your solution.
11:29
And this happened in every one of the solutions, every one of the formulations that contains sucrose.
11:33
So can still conclude that the antibodies were not aggregated.
11:37
So they were all happy in this formulation at room temperature.
11:40
And we can move on to our actual stability studies.
11:44
And the first of those sort of, I'd say, stability studies is using what we call is using the second virial coefficient or B22 and the diffusion interaction parameter or KD.
11:55
What we're basically looking for here is positive slopes.
11:59
If the slope of either the Debye plot or the diffusion coefficient plot is positive, that's an indicator that the antibody has a very low propensity to aggregate.
12:10
If it's negative, it means that the antibody is much more likely to aggregate in those solutions, especially as concentrations get higher.
12:17
That's an important thing to realise or notice when you're sort of doing low concentration tests of an antibody that you're intending to use for high concentration delivery, for example in subcutaneous.
12:29
So positive is good, negative is bad.
12:32
Down here we've got mAb1 really liked being in with that sucrose solutions.
12:35
Pretty much all of the other ones, we had negative values for B22 and for KD, indicating that this antibody really prefers to be in sucrose relative to all of the other excipients.
12:45
And we did these same tests for all of our other antibodies and all of our other formulations highlighted here which gave the best results, which gave the highest values of B22 and KD.
12:57
And the one that I do want to turn our attention to is adalimumab before I move on.
13:00
And that had really none of these excipients had positive values of B22 or of KD.
13:06
So really none of these excipients are ideal for adalimumab, but they're not bad enough necessarily for me to exclude them from the studies with the exception of that sucrose containing solution.
13:15
But generally most of these antibodies really liked being with sucrose.
13:18
mAb2 actually really liked that combination of sucrose and arginine, although arginine by itself was also had stabilising effects on this antibody.
13:28
So we've got our sort of room temperature baseline stability tests.
13:31
We know a little bit about what's likely to happen as we go into higher concentrations.
13:36
But for more deep analysis of the stability of these antibodies in these formulations, we're going to turn to Uncle, which was the first all in one biologic stability platform.
13:45
And it contains 3 different detection methods for looking at the stability of proteins.
13:50
So it has full spectrum fluorescence, which we use to look at the unfolding behaviours of proteins or the conformational stability.
13:56
We have static light scattering, which we use for looking at aggregation of a protein in solution.
14:01
And then we also have dynamic light scattering also used for looking at aggregation and for checking the size of the proteins at various temperatures.
14:10
Uncle only requires 9 microlitres of sample in order to get all of these measurements and has temperature controls anywhere from 15 to 95°C, which covers most of the relevant range for formulation screens.
14:22
And combining those three different detection methods with that 15 to the 95° temperature control, we actually have over a dozen different applications built into the software and can cover a wide range of different types of modalities.
14:34
So proteins and monoclonal antibodies, which is what I'm discussing here, but we've also done experiments with viruses and with the nanoparticles as well and gotten useful data from those particular products also.
14:47
So for proteins though, this is how basically how Uncle works.
14:51
So you saw that animation happening as a protein heats up, it starts to unfold.
14:56
As it unfolds, the hydrophilic amino or the hydrophobic, excuse me, the hydrophobic amino acids, tryptophan, tyrosine mainly inside of the protein gets exposed to the aqueous environment that causes the fluorescence be emission of the protein to decrease over time or decrease during the thermal wave and also shift over to the right.
15:17
So that is what is known as a, let me make sure I'm doing this right in my head, red shift.
15:23
I get that right.
15:24
OK, thank you.
15:25
I get my right and left backwards sometimes.
15:27
So it's useful to have confirmation.
15:29
So that is what is known as a red shift.
15:31
And by monitoring that red shift in the protein's fluorescence, we can actually track is how much has the protein unfolded at different temperatures.
15:39
Uncle, since we're measuring this entire spectrum of fluorescence, we can also take a look at the thermal stability of proteins that don't redshift, because not all of them do.
15:47
Some of them blue shift and some of them actually don't show any shift at all.
15:50
It's just they show the decrease.
15:51
All of those are things that Uncle can track and use to determine TMs.
15:57
We're also using static light scatter.
15:59
We use this fluorescence detection also to look at the static light scattering of our two different lasers, our 266 nanometre laser and our 473 nanometre laser, which is what we're using to look at aggregation behaviour during the thermal ramp.
16:11
It's real simple as a protein starts to aggregate, as it gets larger, the SLD static light scattering increases and that shows up in those two different peaks there at 266 and 473 nanometres.
16:24
So when we take a look at the change in intrinsic fluorescence of monoclonal antibody 1 with our four different formulations as graphed by this, what we call a BCM or barycentric mean graph, we can actually assign TMs right here at the transition points.
16:39
So from this you can actually see that addition of either sucrose or the combination of sucrose and arginine caused higher TMs than the sodium chloride did.
16:49
So yeah, monoclonal antibody 1 doesn't much care for sodium chloride.
16:54
The addition of sucrose or arginine does improve its stability.
16:57
What's interesting though, is that the combination of sucrose and arginine is nearly the same as the TM for just arginine or just sucrose.
17:05
So there's really no combinatorial effect.
17:07
It's just, you know, slight improvement from one or the other, but there's really no combinatorial effect between them.
17:15
When we add in the SLS data.
17:17
And again, we're looking for the increase in SLS as an indicator of aggregation, we can see that the addition of sodium chloride actually has some really bad effects for this antibody.
17:28
Massive amounts of static light scattering.
17:30
This thing aggregates like crazy when you add sodium chloride.
17:33
Other thing to note is that if you add just arguing by itself, that does give you the best Tagg or aggregation temperature.
17:41
That's the onset of aggregation.
17:42
We want that number to be as high as possible, but the trade-off for that is we also get a very high SOS signal, which is usually an indicator that the protein has started to aggregate by a lot.
17:54
So it may aggregate late, but it aggregates a lot, which can be quite bad for if it's something you're intending to deliver to patients.
18:03
So balancing out Tagg versus the extent of aggregation is part of the science or maybe even part of the art of looking at formulation screening.
18:12
So yeah, addition of excipients really has a massive influence over how both unfolding and aggregation proceeds during the course of a thermal ramp.
18:23
Now the nice, the other nice thing we've got is that since we have dynamic light scattering as well as SLS, we can actually confirm that behaviour that we saw before and actually quantify it to some extent.
18:34
So we saw that this antibody mAb 1 with the combination of sucrose and arginine seemed to have runaway, extremely extensive aggregation with just sucrose by itself.
18:45
It did aggregate slightly earlier, a few degrees earlier, but that didn't seem to be as bad based off the SLS data.
18:51
And the DLS data confirms exactly that behaviour.
18:54
At 20°C, at the beginning of the experiment, both antibodies were mono dispersed.
19:00
At 90°C, the antibodies in the combination of sucrose and arginine had a much larger average size and a much wider distribution, a much higher PDI.
19:08
So we can tell that this sample is much more heterogeneous and these proteins have gotten a lot larger when in combination of sucrose and arginine.
19:16
Probably not the optimal formulation for these.
19:21
I'm not going to go through every single one of those curves for every antibody in front of you.
19:24
I did that already.
19:25
So I'm just going to summarise the result, which is identifying the optimal act, the optimal excipients for each one of our antibodies both at 1 mg per mill and at 10 milligrammes per millilitre.
19:35
Since I was able to do all of that basically in a single run because Uncle can handle up to 48 samples at a time.
19:41
So I was able to do all of this in triplicates just in a single run.
19:44
When we do that for mAb 1, trastuzumab, sucrose was the clear favourite at both concentrations.
19:50
But in the case of mAb 2, we actually had different results for the two different antibodies.
19:54
So we had a combination of sucrose and arginine and then just arginine by itself was the preferred in the case of the higher concentrations.
20:02
How am I doing on time?
20:03
3 minutes perfect I can do this so?
20:09
So now we've got a set of formulations that at low concentration have a green flag.
20:12
We know which ones are optimal at those low concentrations.
20:16
So the question becomes are those same concentrations, are same formulations optimal when we go to higher concentrations go to say 100 milligrammes per millilitre, not necessarily the case.
20:27
So we have to check do those formulations still have that green flag and we can actually use Uncle to still look at the thermal stability of our formulations even at 100 mgs per mil and actually beyond I've done tests of over 300 mgs per mil and still gotten the stability data.
20:43
So when we just still look at the Tagg results for each one of our for three of our antibodies, we can still see sucrose is preferred for mAb 1.
20:50
Combination of sucrose and arginine is preferred for mAb 2, and for adalimumab.
20:55
And really none of these antibodies at high concentration liked being with sodium chloride.
21:00
It aggregated extremely readily with the addition of sodium chloride.
21:05
Last step is to check the viscosity of our high concentration protein samples at 100 mgs per mil.
21:09
We're using Honeybun for that, which could run 10 samples at a time and takes only 35 microlitres of sample and finishes experiments in a few minutes actually in order to get those viscosity measurements.
21:20
And actually that's my, it can, it actually can also cool as well.
21:24
So we can measure the viscosity at 4° up to 45°C.
21:29
If you're curious about how Honeybun works, I'll be happy to discuss that with you at a later time because here's the results.
21:34
Basically, sucrose formulations tend to have a slightly higher viscosity than the sodium chloride formulations for antibodies, but they were also stabilising.
21:42
So I think that's a worthwhile trade off.
21:43
And the other thing that I think is worth noting is that in the case of mAb 1, adding arginine actually increased the viscosity relative to sucrose by itself.
21:54
So the combination increased the viscosity.
21:56
But if you look at mAb 2, the addition of arginine decreased the viscosity by quite a bit.
22:00
So it really the addition of arginine is not a magic bullet for viscosity.
22:04
Sometimes it works, sometimes it doesn't.
22:06
And it's worth checking with a specific antibody if that's actually going to be appropriate.
22:11
So now we're able to crown our winning formulations for mAb 1.
22:14
Sucrose was the winner for trastuzumab, also sucrose from mAb 2, that combination of sucrose and arginine, when we consider all of the data together, B22, KD, TM, Tagg and viscosity.
22:24
And unfortunately, really none of these were optimal for adalimumab.
22:28
So it'll have to try again later.
22:32
So excuse me.
22:33
So I hopefully you'll agree that, you know, no matter with the right equipment, anyone can go fast.
22:37
With the right tools, anyone can get a formulation screen done quickly.
22:41
All right, thanks for your attention.
22:42
I'll happy to take any questions.
