1. Advanced Characterization for Proteins and Polymers
Dr. Dierk Roessner
Multidetector Approach for different sample analysis

2. Table of content Polymer examples Online Viscosity Protein examples
Take home message
Why using multi detector approach

3. Multi Detector Approach DLS Online
Gum Arabic
Multi Detector Approach
DLS Online

4. Application
Gum Arabic is a stabilizer for lemonade and other beverages, food and drug (E414)
Coca Cola, Gummy Bears, pills etc.
It allows to keep dyes, vitamins and other additives in a stable solution
It contains glycoproteins and polysaccharides
SEC MALS is used as incoming inspection the natural gum Arabic they buy from different vendors
Average molar mass is the most important parameter
SEC MALS used for quality control of incoming GA

5. Gum Arabic – Molar Mass
Sample A
Sample B
dRI
Molar mass
blue GA has larger content of large species
results in delayed elution and is detected in MALS
The molar mass show differences to the expected SEC mode. Log Molar mass not linear to the elution volume
Plot of molar mass versus elution volume. The dRI signal is plotted as an overlay.
Molar mass of both sample not related to elution volume.

6. DLS Detector online
DLS Detector

7. Gum Arabic – Separation
Sample A
Sample B
Difference in Rh indicates column interaction
Smaller peak elutes perfectly to SEC mode so separation by hydrodynamic radius.
Online dynamic light scattering (DLS) measures the hydrodynamic radius.
Higher hydrodynamic radii in DLS for second species indicate a column interaction.

8. Gum Arabic – Separation
Sample A
Sample B
Difference in rms radius is strong. That implements a very different structure. To show the structure a Burchard-Stockmeyer Plot can be done.
Plot of rms radius versus elution time. measures the hydrodynamic radius.
The radius of the two sample is very different to the elution volume.

9. Gum Arabic – Structure information
1.4
1.5
1.1
0.9
Results show the difference in structure very clear. Sample B is more elongated as sample A.
If one want to look to the structure more into details a conformation plot can be done.
Burchard-Stockmayer Plot, ratio of RMS radius/Rh (Q) vs. elution volume.
0.78 corresponds to compact sphere
1.00 corresponds to hollow sphere or random coil structure
>1.00 show more and more rod like shape

10. Gum Arabic – Structure information
0.66
0.59
Sample 3
Sample 5
0.58
0.46
Plot of rms radius versus molar mass show the structure of the sample. The different species show also in this plot different structure and it also confirms the result we had already with DLS.
Notes:
The experiment was open in ASTRA® twice in order to determine the different slopes.
Different slopes indicates the different structures of the different peaks.
Two slopes for the different fractions of the blue GA sample, indicating different degrees of branching
rms conformation plot
1 is a stiff rod
0.6 random coil
0.5 to 0.4 indicates branching
0.33 is a sphere

11. Single Chain Nano-Particle (SCNP)
Multi Detector Approach
Viscosity Online

12. SCNP SCNP are new developed drug delivery system.
A synthetic polymer is is internaly crosslingt so that a more compact structure is generated
In SEC the internal crosslinked molecule should elute later, but the molar mass should be the same. D A
SCNP: crosslinking to create a sphere-like particle

13. SEC Chromatogramm of Polymer and SCNP
Molar Mass
dRI
Different elution due to different Rh caused by the crosslinking of the linear polymer to create SCNP, but the molar mass is still comparable -> SEC with column calibration estimates a wrong molar mass due ti teh compact structure of hte SCNP
Plot of the molar mass of a polymer (red) and the SCNP (blue) vs. retention volume. The dRI signal is plotted as an overlay.
Due to the different hydrodynamic radius the elution time is very different.

14. Molar Mass distribution
Polymer
SCNP
The molar mass distribution shows that the molar masses are quite similar and most of the branched material is really internally crosslinked.
But 80 % of the sample is too small to do branching calculation from light scattering.

15. MALS – Viscometer - System

16. Molecular Structure from Mark-Houwink Plot
MHS plot: allows comparison of the instrinsic viscosity of the samples -> gives information of the structure / branching,
If the SCNP would be a sphere -> slope would be = 0
Mark-Houwink plots of epoxy resin, linear polystyrene, linear poly(methyl methacrylate, linear poly(benzyl methacrylate), linear poly(iBuPOSSMA) and star-branched poly(isobutyl methacrylate).

17. Online Determination of Intrinsic Viscosity
𝜂 𝑠𝑝 = 4𝐷𝑃 𝐼𝑃−2𝐷𝑃
DP imbalance pressure across the bridge
IP pressure drop from inlet to outlet
sp specific viscosity
c concentration (g/mL)
α RI detector calibration constant (RIU/V)
dn/dc specific refractive index increment
Erroneous dn/dc makes the same error in molar mass as in the case of MALS!!!
𝑐=𝛼 𝑅𝐼 𝑠𝑎𝑚𝑝𝑙𝑒 − 𝑅𝐼 𝑏𝑎𝑠𝑒𝑙𝑖𝑛𝑒 𝑑𝑛/𝑑𝑐
Specific viscosity extrapolated to concentration 0 -> intrinsic viscosity
Do not explain the special features of ViscoStar : Autobalacing, bridge structure etc. Takes too much time and is not in focus.

18. Mark-Houwink-Sakurade Plot
Polymer
SCNP
a = 0.59
a = 0.74
a < 0.27
a = 0.87
Random coil 0.7
SNCP / internally crosslinked forming a sphere at molar masses above 200kDa -> indicated by slope 0.27
Slope in the Mark-Houwink-Sakurade plot (intrinsic viscosity vs. molar mass) shows the structure.
SCNP of 200 kDa and higher more compact.

19. Hydrodynamic Radius Determination
DLS and Viscosity Online

20. Hydrodynamic Radius from QELS
BSA
Rh nm
Uncertainty
25 µg
2.9
8.0%
50 µg
3.2
5.4%
100 µg
3.3
3.4%
200 µg
3.5
2.7%
3.2 nm +/- 8 %
Plot of the hydrodynamic radius (QELS) versus retention volume. The UV 280 nm is plotted as an overlay.
The dimer has too low concentration to measure it.

21. Hydrodynamic Radius: DLS versus Viscometry
ViscoStar can be used to determine the Rh
Comparison of DLS and ViscoStar data
Conformation plots of polystyrene based on hydrodynamic radius determined by online viscometer and dynamic light scattering.

22. Hydrodynamic Radius
mAb1&2
Bispecific mAb
Globular protein
ViscoStar : higher sensitivity than online DLS, even at low sample concentrations
Hydrodynamic radius of different antibodies determined by ViscoStar 3.
The antibody 1 and 2 show normal size as well as peak shape. The bispecific antibody elutes too late for its size.

23. Protein/Protein-Complex
Multi Detector Approach
UV, MALS, RI Online

24. Conjugate Analysis: ADCs
Protein conjugate analysis uses MALS + UV + RI and measures:
Amount component A - Amount component B
Molar mass component A - Amount component B
Molar mass of the complex
Protein conjugate analysis is used for:
Glycosylated proteins
Pegylated proteins
Membrane protein
Protein-protein-complex

25. Conjugate Analysis: How it Works
ASTRA measures dRI, dUV and LS to solve:
UV signal
dRI signal
dUVADC =
cmAb
εmAb
cdrug
εdrug +
Contribution
from mAb
dRIADC =
cmAb dn
dcmAb
cdrug +
dcdrug
LSADC =
cADC
MADC dn
dcADC ) ( 2
So how do we determine the Drug antibody ratio from our light scattering data. First we should look at the equation down here in black. The LS-traces from our ADC are directly proportional to the molar mass of the adc and the concentration of our ADC and also the dn/dc of our conjugate which is the refractive index increment. So we can already see the complication here. The protein component has a specific dn/dc while the drug also has a specific dn/dc. So if we combine both we have some kind of a weighted average that is not known to us and that prevents the determination of an accurate molar mass.
So in conjugate analysis we utilize a triple detection system.
Dn/dc of the drug is usually determined beforehand in DMSO
The UV signal that we get is the combined Uv signal from the antibody x ext.coeff and the drug times ext.coeff.
Likewise our RI-signal is going to be the sum of the responses from our antibody and the drug.
Lets assume the easiest case where our drug is not UV active. We get the UV response, can integrate the peak area and we know how much protein we have. We also know our proteins dn/dc and therefore what the response should be in the RI detector. And that means the excess dRI response comes from the linked drug.
So when we know the mass contributions of both compounds we know their relative abundances and can apply that information to the light scattering data.
And that is the basis of our conjugate analysis in ASTRA

26. Task
Two protein complex – protein A 17.6 (=0.653) and protein B 18.3 kDa (=0.970)
Is the complex 1:1, 1:2 or 2:1
Complex 1:1 -> 35.9 kDa could also be the dimer from one of the proteins
Complex 2:1 -> 53.5 kDa ( = 0.758)
Complex 1:2 -> 54.2 kDa ( = 0.864) Mw too close together to decide with MALS
How can the multi detector approach help here?

27. Molar Mass determination
Protein A kDa (=0.653)
Protein B kDa (=0.970)
Complex 1:1 -> 35.9 kDa
Complex 2:1 -> 53.5 kDa ( = 0.758)
Complex 1:2 -> 54.2 kDa ( = 0.864)
34.2 kDa dimer
52.5 kDa hetero trimer?
Mp of main peak hints towards 2:1 Complex due to closer molar mass, but the difference between both complexes is too small to be sure
Plot of the molar mass versus retention volume.
The molar mass in the peak of the complex is not constant.

28. Protein Conjugate Approach
Protein A kDa (=0.653)
Protein B kDa (=0.970)
Complex 1:1 -> 35.9 kDa
Complex 2:1 -> 53.5 kDa ( = 0.758)
Complex 1:2 -> 54.2 kDa ( = 0.864)
35 kDa
2*18.3 kDa protein B
19 kDa
17.6 kDa protein A
Prot conjugate reveals a 1:2 complex due to the analysis of the UV exticntion coefficient
Plot of the protein fraction versus retention volume
The main peak show the stochiometry of 1:2

29. Protein Conjugate – Protein Fraction
Protein A kDa (=0.653)
Protein B kDa (=0.970)
Complex 1:1 -> 35.9 kDa
Complex 2:1 -> 53.5 kDa ( = 0.758)
Complex 1:2 -> 54.2 kDa ( = 0.864)
0.35
Plot of the protein fraction versus retention volume.
The main peak show the stochiometry of 2.1
Left and right of the the fraction is different indicating a contamination

30. Extinction Coefficient from RI
Protein A kDa (=0.653)
Protein B kDa (=0.970)
Complex 1:1 -> 35.9 kDa
Complex 2:1 -> 53.5 kDa ( = 0.758)
Complex 1:2 -> 54.2 kDa ( = 0.864)
 = 0.807
 = 0.751
 = 0.842
 = 0.690
Using the measured exticntion coefficient the composition of the complex in the main peak and the shoulder can be resolved
Plot of the molar mass versus retention volume.
The molar mass in the peak of the complex is not constant.
Expected extinction coefficient is 0.864

31. Take Home Message

32. Summary Multi detectors allow multiple views
Molar mass
RMS radius
Hydrodynamic radius
Intrinsic viscosity
MHS-plots for structure and branching information
Extinction coefficient
Copolymer analysis
Wyatt support to find the right solution
Especially complex samples