0:11
OK, so today I'm going to talk about how I utilise the Prometheus Panta to explore the physical stability of biologics and the impact of excipients.
0:25
So I'm going to talk about Co-formulation, which is what I actually did.
0:29
And I'm also going to talk about the Prometheus Panta, which was the instrument I did everything on.
0:35
And then I'm going to go through some of the data from both of the individual component screens and the Co-formulation screen.
0:44
And then I'm going to touch on how I utilise the Panta to do some viscosity screening on the Co-formulated samples.
1:02
So Co-formulation is the act of mixing 2 biologic drug products together into a single formulation.
1:10
And the two biologics I co-formulated were NISTmab, which is just monoclonal antibody reference material and a synthetic GLP 1.
1:25
So this is a complex challenge because both molecules are significantly different in structure.
1:33
So the antibody has 22 buried tryptophan within its tertiary structure.
1:41
So upon the introduction of thermal heat, the only change we should be able to see is the unfolding of the antibody and the tryptophan's becoming exposed to the solvent.
1:55
Conversely, the peptide in its monomeric form has a single tryptophan that is exposed to the solvent.
2:05
So upon heat we should only really see aggregation of this GLP-1 and because of this we will see the tryptophan becoming buried.
2:15
So we will see a change in the fluorescent emissions of the tryptophan.
2:21
So although they are vastly different in molecular weight, when we look at the size of both molecules, there isn't really much difference.
2:30
So classically we would not maybe expect to see the two molecules resolved by DLS, but the fact I'm mentioning that will become clearer in a bit.
2:42
So because this is a complex challenge, we have to approach it in a systematic way.
2:48
So first I performed individual screens on each components with a number of different excipients to try and probe the homogeneous interactions within each molecule.
3:02
After this I co-formulated the two biologics and with the same excipients with varying biologic compositions I screened the co-formulated mixtures.
3:14
So I did this by using multiplexing optical methods to deconvolute this overall quite complex challenge.
3:23
The instrument that I used was the Prometheus Panta with automated robot auto sampler.
3:31
So this allows for combination DLS, nano DSF and turbidity.
3:37
You only need 10 micro litres of sample volume and specifically I did a thermal ramp from 20°C to 95°C at a rate of 1°C per minute and the samples were loaded automatically for me from 384 well plates and I was able to do 24 samples at once.
3:58
So just to stress how important the low sample volume is for my antibody screen, I did 4 antibody concentrations, 3 excipients, multiple excipient concentrations and triplicate repeats and all of this used less than 16 milligrammes of antibody.
4:16
So now on to the data.
4:19
So prior to performing a thermal ramp, the Panta allows you to look at the colloidal stability through dynamic light scattering or DLS.
4:29
So initially the hydrodynamic radius which looks at the size including the solvated shell.
4:36
The hydrodynamic radius of NISTmAb in my solution was 5.25 +- 0.02.
4:43
When you look at the COA supplied by NIST you can see that when you include the error these match up pretty well.
4:50
And secondly, I can tell that this is a highly monodispersed sample because of the very low polydispersity.
4:58
So anything below a 0.2 polydispersity index tells you have a highly monodispersed sample.
5:05
So from this I was able to confirm that in 25 millimolar histidine, NISTmAb is colloidally stable, which is no surprise because it's quite similar to the formulation buffer.
5:17
So then I looked at the effect of each of the three excipients.
5:21
So I looked at the effect of arginine, proline and salt and you can actually see across all three excipients there is no true significant change in the hydrodynamic radius.
5:34
Although there is a change, you can see that this never really goes above 5.5 nanometers hydrodynamic radius.
5:40
So we can confirm that NISTmAb does not appear to be affected and we're just looking at the hydrodynamic radius.
5:50
So before I talk about my own thermal ramp data, I just want to show what you would classically expect to see from NISTmAb on a thermal ramp.
6:00
So we are looking at the ratio of the fluorescent emission at 350/330 nanometres.
6:09
So we are looking at the changes of the Tryptophans highlighted in black.
6:15
So when we do a thermal ramp, we can see that we are able to look at the unfolding of each of the three domains.
6:24
So we can see the CH2 domain unfolding and then we can see although it's not highlighted here at this point, the Fab domain unfolds and we also at higher temperatures we see the CH3 domain unfolding.
6:40
So we do not see aggregation classically, but we do see a size change and the size change is only unfolding induced.
6:48
So the change just becomes freer and you can see a size change instead of aggregation.
6:53
Stressing this because I when I looked at my own data did not see that.
7:00
So the black line you can see here is with just 25 millimolar histidine and with increasing salt concentrations.
7:10
So here and here we can see that the fluorescent emission ratio changes.
7:20
This becomes far clearer when we look at the first derivative of this ratio and we can see that there are peaks and the formation of negative peaks.
7:29
So just like the classical thermal ramp that we would expect, we see the CH2 domain unfolding.
7:37
We see the fab domain unfolding.
7:41
And then actually with increasing salt, we see the formation of a negative peak.
7:47
So we can tell that this is bearing of tryptophan due to aggregation because of the increase in cumulant radius.
7:54
So cumulant radius basically tells us the average size of the molecules present in the sample.
8:01
And we can see at very high salt concentrations and at very high temperatures we are seeing the formation of large aggregated species.
8:11
So because this happens just after the Fab unfolding domain, we can say that this aggregation is driven by the unfolding of the Fab domain and by the increasing salt concentration.
8:26
So rather than just looking at these graphs, we can actually plot the data.
8:31
So here you are looking at the CH2 domain unfolding.
8:36
So we are looking at the Inflexion point of that first very large peak.
8:41
So with increasing arginine and increasing salt, we can see that there is a decrease in the onset of the unfolding.
8:50
We can tell that this is decreasing the stability of NISTmAb because less thermal energy is required to unfold NISTmAb.
9:00
So arginine and salt are having quite a significant effect on the thermal stability of NISTmAb.
9:08
So the side chain of arginine is charged and salt is charged.
9:13
So we can probably say this is due to electrostatic screening.
9:18
Proline on the other hand, although is a zwitterion, it does not have a charged R group.
9:26
So we can see very clearly that upon increasing the Proline concentration, there is not really a change in the onset of this unfolding.
9:38
We can say that Proline has minimal effect and both arginine and salt have quite a significant effect on the unfolding of the CH2 domain.
9:49
So now we are looking at the formation of the negative peak that we previously saw.
9:57
So arginine and salt and proline all have three quite different effects.
10:04
So arginine above 50 millimolar has an increase in the temperature required for the antibody to aggregate.
10:13
Therefore, we can say that there is increasing stability of the unfolded form of NISTmAb because if you remember this occurred after the Fab domain unfolding.
10:25
So we can say here that arginine is stabilising the unfolded form of NISTmAb.
10:31
So when we consult literature, we can see that arginine is capable of forming cation π interactions.
10:40
So it's probably that these cation pi interactions are stabilising the unfolded NISTmAb.
10:46
We can tell that this is most likely not due to electrostatics because of the effect of salt.
10:52
So we can see that at the high antibody concentrations there is a decrease and then a plateau.
10:59
And then this effect is seen pretty much through all the excipient, all the antibody concentrations.
11:07
So we can say the electrostatic screening is also driving the aggregation propensity of NISTmAb because less thermal energy is required to aggregate NISTmAb.
11:18
Again, Proline doesn't have as much of a clear effect.
11:23
So you can see that really there may be a decrease at higher antibody concentrations, but at low antibody concentrations, there's not much of an effect compared to arginine and salt.
11:39
So as I said, we are looking at aggregation and we want to confirm that we are seeing aggregation, and the Panta allows us to do this by looking at the turbidity and cumulant radius.
11:52
So this is the effect on salt.
11:54
So as you can see that we get a response from the turbidity at high salt concentrations.
12:03
So the turbidity allows us to look at the formation of species bigger than 12.5 nanometers in size.
12:10
You can see that this is the case when we then consult the cumulant radius.
12:16
So we can confirm that salt concentration is really important when we're looking at the aggregation of NISTmAb.
12:26
So as before, we can plot all this data rather than just looking at graphs.
12:30
So instead we can look at different graphs and we can look at the onset of the increase in cumulant radius.
12:38
So this is specifically looking at the formation of species bigger than 15 nanometers.
12:45
So as you can see, arginine and salt both form larger species at lower temperatures.
12:53
So again, as I previously mentioned, both are electrostatic in nature.
13:00
So this is most likely the effect.
13:05
We can confirm this. We can confirm these trends by consulting the turbidity onsets.
13:12
So we can see that the trends match up pretty well.
13:18
So as you may have noticed, the proline graphs were not present on those last few slides.
13:23
This is because when we look at the thermal ramp and we look at the fluorescent emission ratio and first derivative of this, we can see that as classically we would expect we have the CH2 domain unfolding, the Fab domain unfolding and not as much of a negative peak because of this.
13:43
And when we look at the cumulant radius, we can see that we are not actually seeing any significant aggregation.
13:49
So we are actually just seeing size change due to the unfolding of the Fab domain like we would actually expect.
13:56
So what we can say here is that Proline actually has a minor effect on NISTmAb size and also probably NISTmAb stability.
14:04
It's not too much of a stretch to say that.
14:09
So also utilising the DLS data the Panta gives us, we can actually look at beginning to look at the effect of all the excipients at once.
14:23
So we can do this by using the KD.
14:26
So the KD here is the diffusion interaction parameter and it's plotted by looking at the diffusion coefficient against the molecule in question, concentration.
14:40
And then you do some linear regression and you get a value.
14:44
So the size of this value and whether it's positive or negative tells you about what's happening within the solution if the value is positive.
14:54
So you just have to imagine this dotted line was actually at 0.
14:58
If it's positive, you can see that there are repulsive interactions between NISTmAb molecules in solution.
15:05
When this value is negative, you can say that there are attractive interactions present between NISTmAb molecules.
15:13
And then when we look at the effect of each excipient, you can see that both salt and arginine to a certain concentration make the interactions between NISTmAb molecules more attractive.
15:25
So they are driving the interactions.
15:29
Conversely, with proline, they stay positive.
15:33
So when we're saying proline prevents NISTmAb interactions, this is relative to arginine and salt, but it's just a nice way that you can begin to estimate a value of the effects of all the excipients.
15:50
So due to the amount of time I have, I'm only going to show you the DLS data of the GLP-1.
15:57
So here I looked at the colloidal stability of GLP-1 in 25 millimolar histidine, which was what the antibody screen was done in.
16:07
You can see here we have very large aggregates forming.
16:13
So it wouldn't be a good idea for me to formulate or co-formulate in 25 millimolar histidine.
16:21
So instead I looked at 25 millimolar sodium acetate pH 5 and you can see that we have a singular peak with a hydrodynamic radius of 2.1 which is what we would expect from monomeric GLP-1 and we have a very low polydispersity.
16:38
So we can say that GLP-1 is colloidally stable at pH 5 and 25 millimolar sodium acetate.
16:48
Because of this I then decided to move forward with my co-formulation screen.
16:52
In this buffer you don't see the data, but NISTmAb is also stable in 25 millimolar sodium acetate.
17:01
So when we look at the initial DLS of the co-formulated mixtures, we can see that in some solution conditions we were able to separate each peak.
17:15
So this was surprising because classically DLS is only capable of resolving species that are three to five times magnitude in different in size and the antibody is 5 nanometers and the peptide is 2 nanometers.
17:29
So you wouldn't usually expect to see two separate peaks, but actually at a one to four concentration ratio and a two to four concentration ratio, we can actually see both species.
17:42
So the red line here in all three cases corresponds to 25 millimolar sodium acetate and then the coloured lines here correspond to the addition of 100 millimolar proline.
17:58
So this is just a highlight that in that when we begin to change the solution conditions, we actually could not see the two peaks in the end.
18:07
But it was really nice that we were able to with the DLS resolve the two peaks.
18:15
So now when we look at the thermal ramp data, you can see that we have a negative peak again here.
18:23
And we can definitely confirm this is aggregation due to the response from the turbidity and actually the size distribution.
18:31
So with the size distribution we can see that we are actually very clearly forming a second species at a larger size and high salt concentration is driving this.
18:43
Proline on the other hand we do not see a negative peak and we do not see any change in the size distribution.
18:52
So Proline is not affecting the co-formulated mixture whereas salt as we saw with the antibody screen is still driving the instability.
19:03
So finally I did some viscosity measurements.
19:06
Obviously viscosity is very important and I wanted to see the effect of all of the three excipients and the solution viscosity.
19:15
So all the coloured icons you are seeing here are different ratios of co-formulation.
19:21
So there is a lot to look at here.
19:24
But the main message from this is Arginine shows the largest increase in viscosity.
19:32
Salt and proline show less of an increase in viscosity, which was interesting for me because obviously salt had the largest effect on these instability of the mixture.
19:44
So I was expecting salt to have the largest effect on the viscosity.
19:48
But clearly this is not the case.
19:51
So it was interesting that Proline, although affects the viscosity, does not actually affect the instability of the drug products and solution.
20:01
So just to conclude, I used multi-parametric high-throughput stability screening using the Prometheus Panta to deconvolute the physical stability of co-formulated mixtures using certain excipients.
20:18
From this I was able to determine that electrostatic screening destabilised NISTmAb and the Co formulated mixtures.
20:26
Surprisingly, the individual components in certain solution conditions were able to be resolved.
20:33
And finally, I used combination high throughput viscosity measurements to understand the effect of recipients.
20:41
Here are some people I would just like to acknowledge.
20:43
Thank you very much for listening.
