Kasper Huus, Principal Scientist Peptide and Protein Formulation Injectable Drug Product Global Research Novo Nordisk A/S

Transcription

1 Insulin thermostability Paris, 26th September Insulin thermostability, ligand interactions, and correlation of DSC data with other stability indicating methods 2nd European Microcal Symposium Institut Pasteur, Paris 26th-27th September, 2016 Kasper Huus, Principal Scientist Peptide and Protein Formulation Injectable Drug Product Global Research Novo Nordisk A/S

2 Insulin thermostability Paris, 26th September Outline Insulin structure and selfassociation forms DSC interpretation of insulin thermal unfolding Correlation between DSC results and other methods for chemical and physical stability Thermostability of insulin analogues

3 Insulin thermostability Paris, 26th September Human Insulin structure Primary structure of human insulin Secondary and tertiary structure A21 B1 A1 B30 Insulin monomer from T 6 crystal structure Brange et al., Diabetes Care, 1990

4 Insulin thermostability Paris, 26th September Human Insulin self-association forms Monomer Dimer T 6 hexamer 2 Zn 2+ /6 insulin R 6 hexamer 2 Zn 2+ /6 insulin 6 phenol/6 insulin B1-B8 convert to α-helix A21 B1 A1 B30 Circulation in blood stream Receptor binding Building block for hexamer Crystalline storage in β- cells granula in vivo Pharmaceutical formulation (with additional excipients)

5 Insulin thermostability Paris, 26th September Stability of insulin formulations Physical stability issues: Primarily fibrillation, precipitation, crystallization Chemical stability Main degradation pathways in neutral, soluble insulin formulations: Asn B3 deamidation Dimerization Polymerization

6 Insulin thermostability Paris, 26th September Biphasic thermal denaturation of human insulin DSC thermogram obtained under formulation relevant conditions Phenolic ligand Buffer Zn 2+ /hexamer 3.0 Anionic ligand ph 7.4 Human insulin 0.6 mm Phenol 20 mm Phosphate 7 mm Sodium chloride 50 mm

7 Insulin thermostability Paris, 26th September Biphasic thermal denaturation of human insulin Known origins of biphasic thermal denaturation Unfolding of separate domains at different temperatures Larger proteins and antibodies Oligomeric proteins: Dissociation of oligomer followed by unfolding of monomer Ligand induced stabilization of native protein when the ligand binding is not saturated

8 Thermal Denaturation of Human Insulin Parameters effecting the biphasic properties Insulin thermostability Paris, 26th September Zinc concentration effect Insulin concentration effect Monomeric analogues zinc-free Fixed ratio Zn 2+ ratio (2 Zn 2+ /hexamer). Insulin concentrations 0.1, 0.3, 0.6 mm (left to right) 2 Zn 2+ /hex Fixed insulin concentration at 0.6 mm. Zn 2+ concentrations are indicated in the figure. Huus et al.,biochemistry, 2005 Human insulin Insulin Asp B28 Insulin Asp B9,Glu B27

9 Thermal Denaturation of Human Insulin NUV-CD studies Insulin thermostability Paris, 26th September NUV-CD thermal unfolding NUV-CD. From top to bottom: Insulin Asp B9,Glu B27, Insulin Asp B28, and HI - all zinc-free Bottom: HI 3 Zn 2+ /hexamer 22 C -400 CD 275 nm Zn/hexamer Biphasic denaturation is caused by zinc redistribution during heating Huus et al.,biochemistry, 2005

10 Insulin thermostability Paris, 26th September Formation of amyloid fibrils Amyloid fibrils Insoluble, supramolecular with cross β-sheet structure formed of peptide/protein monomers Physical change no change of covalent bonds (but hydrogen bonds, van der Waals interactions, hydrophobic effect) Stochastic process Requires build-up (lag time) of critical mass (seed) May occur in bulk solution (no physical stress) or dependent on surface/interface interaction Physical stress (e.g. shear forces) required Interaction with primary packaging crucial Unacceptable in a pharmaceutical drug product Cryo Electron Microscopy (Top) Model (Below) [Jimeniz et al., 2001] Partially unfolded structure Denatured peptide Nucleus

11 Insulin thermostability Paris, 26th September Insulin formulations: An unexpected effect of phenol ThT assay:45h, 30 C, 960 rpm shake 8 replica, 1 µm ThT Phenol binds to the T 6 insulin hexamer Conformational change to R 6 insulin hexamer with improved chemical and physical stability Pulls the equilibrium further away from the monomeric state Lag time (h) Lagtime (h) Recovery (% ) Insulin (no zinc) Insulin + 2 Zn/6 ins Recovery (%) mm phenol + 10 mm phenol + 15 mm phenol + 30 mm phenol mm phenol + 10 mm phenol + 15 mm phenol + 30 mm phenol 6x phenol No effect of phenol on insulin lag time Shorter lag times of insulin+zn with increasing phenol! Morten Schlein, unpublished data

12 Insulin thermostability Paris, 26th September Phenolic ligands influence on insulin thermostability 20 mm phenolic ligand 80 mm phenol 20 mm phenol Without phenol 100 mm phenolic ligand Destabilization of zinc-free insulin Stabilization of zinc hexamers Huus et al.,biochemistry, 2006

13 Insulin thermostability Paris, 26th September DSC vs physical stability of formulations in vials DSC data Sample Tm,onset ( C) Tm,hexamer ( C) A B Ref Sample A Start 28 days Huus, unpublished data

14 Covalently linked insulin dimer - Extremly stable Insulin thermostability Paris, 26th September Disulfide linked monomers (B25C) Dimer and hexamer structure equivalent to self-association forms of human insulin Does not bind to the insulin receptor DSC Thioflavin T fibrillation assay Vinther et al, PLoS One, 2012

15 Insulin thermostability Paris, 26th September Insulin analog with additional disulfide bond CysA10, CysB4, desb30 insulin Self-association similar to human insulin Increased affinity to the insulin receptor Vinther et al, Protein Science, 2013 Does not adopt R-confirmation, stays in T-conformation DSC in zinc-free condition 4SS-insulin monomers are more stable than HI zinc hexamers Also strongly reduced propensity to fibrillate

16 Insulin thermostability Paris, 26th September Ligands and insulin chemical stability Effect of zinc on AsnB3 deamidation and covalent dimer formation Without phenol 20 mm phenol Zinc-hexamer formation stabilizes slightly against deamidation. Contradictory to the effect of zinc on thermostability, additional zinc increases deamidation in T 6 hexamer In R 6 hexameric insulin this effect is not seen. Possibly a specific interaction in the C-terminal B-chain Huus et al., Pharm. Res., 2006

17 Insulin thermostability Paris, 26th September Ligands and insulin chemical stability Effect of zinc on other chemical degradation Increased stability against fragmentation and formation of other related substances with increased zinc levels: Correlates to thermostability Without phenol 20 mm phenol Huus et al., Pharm. Res., 2006

18 Insulin thermostability Paris, 26th September Ligands and insulin chemical stability Effect of phenolic and anionic ligands (with 3 Zn 2+ /hexamer) Effect on the biphasic insulin denaturation Effect on Asn B3 deamidation correlates with thermostability 3 Zn 2+ /hexamer 3 Zn 2+ /hexamer + 20 mm phenol 3 Zn 2+ /hexamer + 20 mm phenol + 50 mm NaCl Huus et al., Pharm. Res., 2006

19 Insulin thermostability Paris, 26th September DSC for stability screening of new insulin analogues Different insulin analogues formulated similarly Data from DSC screening: T m,onset T m, hexamer unfolding Huus, unpublished data

20 Insulin thermostability Paris, 26th September HMWP formation of various insulin analogues Good correlation with T m,onset data HMWP formation rate (%/month) HMWP formation rate (%/month) T m, hexamer ( C) T m, onset ( C) Huus, unpublished data

21 Insulin thermostability Paris, 26th September Purity loss of various insulin analogues Slight correlation with T m,onset data Purity loss rel. to reference Purity loss rel. to reference T m, hexamer ( C) T m, onset ( C) Huus, unpublished data

22 Insulin thermostability Paris, 26th September Conclusions and applications - I Zinc-free vs. zinchexamers Biphasic denaturation at low concentrations of zinc Understanding of the mechanism provides more rational use of DSC in insulin formulation development

23 Insulin thermostability Paris, 26th September Conclusions and applications - II DSC as a method to study ligand binding to the insulin hexamer Correlation between thermal and chemical stability studied at formulation-relevant zinc concentrations Can be used for studying both phenolic and anionic ligands Useful in examination of zinc binding and self-association properties of insulin analogues Clear correlation for B3 deamidation but not for dimerization/other Low zinc results (biphasic denaturation) correlate with high-zinc results DSC is useful for studying insulin formulations and identifying new stabilizing ligands Limitations: Binding is studied at high temperatures may not correlate to lower temperatures Interactions with the unfolded state Limitations: DSC may give false positive or false negative results (examples: KSCN and resorcinol) DSC results are exclusively related to the structural stability of the protein

24 Insulin thermostability Paris, 26th September Acknowledgements University of Copenhagen Sven Frøkjær Marco van de Weert Bent W. Sigurskjold Novo Nordisk Helle B. Olsen Svend Havelund Morten Schlein Tine Vinther Ida Boldsen and many more