Excipient effects on the stuctural and colloidal stability of proteins A rational approach to the formulation of protein pharmaceutics?. Agenda: Protein stability, reversible and irreversible transitions Preferential interactions and reversible stability Aggregation and colloidal stability
Bratislava, May 20th 2008
Peter Westh ([email protected])
Typical stability for a small globular protein at “normal” conditions
DG° ~ 20-50 kJ/mol
Hen lysozyme has ~200 intramolecular hydrogen bonds – the bond energy for each of these is ~25kJ/mol
Evolutionary aspects –
Is it so hard to design a stable protein conformation?
Functional aspects –
An enzyme is a nanomaschine often doing 100-100,000 cycles per sec
Technological aspects –
Product stability may be a limiting factor
Hen lysozyme with 185 bound water molecules
Wilson et al 1992.
K=[D] / [D] and hence DG°=-RTlnK
Far from Equil. mI << mD
The irreversible transition is directly relevant to formulation protocols. The reversible transition is experimentally and theoretically accessible and empirically related to the kinetic stability
Does not happen since
DG°blue+red > DG°blue + DG°red
The difference is the red-blue domain interaction free energy
Calorimetry is a simple experimental principle (Lavoisier had nice calorimeters in 1780) which has been developed to extreme sensitivity (10-100nW or ~ 10-7°C)
Constant P: qp=DH=CpDT
Proteins denature both upon heating and cooling
The binding of a ligand to the native state brings abour stabilization – The dicplacement of the peak along with the change in transition enthalpy quantifies the binding strength
Preferential exclusion : of proteinsinhibits the formation of interface
Præferential binding : promotes larger interfaces
Destabilize SolvateProtein stability – solute (excipient) effects
Preferentially excl. excipient
Preferentially bound excipient
Relies on surface charge density and polarizability
Anionic kosmotropes (e.g. F- and SO42-) bind 10-20 water molecules strongly but leaves the bulk rather unperturbed.
Anionic chaotropes affect bulk properties
Koga et al. 2004, Westh et al. 2006
Naturally occurring inorganic osmolytes are kosmotropes
Timasheff et al.
Small alcohols do not, however, universally destabilize the protein
Td vs. [alcohol]
At 10C ethanol stabilizes lysozyme up to ~4M
At 40C ethanol is neutral up to ~1M
Propanol strongly reduces Td for Lysozyme
Velichelebi and Sturtevant 1981
Calorimetric and neutron scattering studies
provide a ”state diagram” for cutinase in SDS
Effetcs of a ”traditional” stabilizer: Trehalose
Babtista et al Biopolymers 89 538 (2008)
Nielsen et al 2006, Nielsen J. Phys Chem B 111 2941 (2007)
Preferential Interaction parameter3 (m3/m2)T,P,3
Gm3 is the number of excipient molecules, which has to be added to re-establish its chemical potential (activity) upon the addition of one protein molecule to the system.
3 > 0 : preferential binding
3 < 0 : preferential exclusion
For non-specific interactions:
The effect of changing the logarithm of the activity (or roughly, the logarithm of the concentration) of any excipient on the standard free energy of protein unfolding is proportional to the adsorption of the excipient on the protein’s surface (the preferential interaction parameter) times the surface area change caused by unfolding.
Wyman & Gill, Binding and Linkage, 1990
Gibbs adsorption equation (The surface excess)
N D I
As seen by DSC
Jahn & Radford 2004
Separate mechanisms or a
universal propensity of polypeptides to
form intermolecular b-structure?
Kinetically stable dispersion of two phase system – emulsion, foam, suspension, aerosol etc.
The simplest case: a plane surface (negatively charged)
Interactions between two particles
For a particle (e.g. a protein)
Salts may strongly modulate the stability of liquid protein formulations – but not always in a bad way !
For ”non-specific” interactions, G scales with the surface area.
N D Ag
Do stabilizers promote aggregation ?
Do denaturants promote shelf-life?
Effect of typical stabilizer such as Trehalose or SO42-
Are these (equilibrium) considerations relevant to a kinetically controlled process ?
Typical destabilizer such as Urea or ClO4-
Urea and the solubility of apo-myoglobin
N(aq) D Ag
Conceptual scheme (if D is reasonably populated):
N D D Ag
Rate governed by the concentration [D]
Equlibrium constant: K= [D]/[D]
Solute effects on aggregation kinetics depends on relative size of DG and DG
TS theory and protein aggregation
Chi et al. 2003
Baines & Trout 2004
Thermal stability of amylase and apo-amylase
Nielsen et al 2003a
o of proteins
Heat flow (
tTitration calorimetry and ion-stripping
Isothermal titration calorimetry detects the heat of reaction when a small amount of titrand is added to a calorimetric (stirred) cell
Addition of EDTA to amylase at 60°C
Aggregation of a-amylase at 60C
”Additives which stabilize the N-state of glucagon phosphorylase also promotes the protein’s aggregation”
N D Ag
D form accumulates in Urea solutions
Urea promotes accumulation of the D-form in accordance with its lyotropic properties
Buffer, glycerol or
[D]~0 in buffer or solutions of stabilizers
Almost negligible !!!
BSA forms “-aggregates” in which a moderate part of the native helices are converted into intermolecular -sheets.Militello et al. 2004
Time course of aggregation at Tm
Bagger et al, 2007
MALLS RI of proteins
MALLS RISEC-MALLS-RI analysis of quenched samples
MALLS-Signal proportional to
RI and UV Signals proportional to Cm
30 min at Tm (buffer and 0.5 M sucrose)
Ratio of MALLS:RI signals is a measure of Mw
SEC-MALLS-RI analysis of BSA thermal aggregation
EFFECTS OF ADDITIVES ON BSA AGGREGATION
Many small aggregate particles – no net effect on ”life-time”.
Few and small particles – promotes kinetic stability.
No ”Hofmeister-effects” in either case.
Hofmeister effect – preferential interactions – are not important for the rate of aggregation
Salts promote stability ????
Electrolyte efffects generally relate to lower concentrations !
Preferential interactions ~0.2-2 M
Electric double layer ~ 0.01 – 100 mM
SEC-MALS measurements of a-Amylase aggregation in 5mM
HEPES buffer, pH 8, 60C. LS signal (@90) is shown in panel A,
UV signal panel B.
Colloidal stability (pH 8.0, 60C) of a-amylase (net charge -10)
5mM HEPES + 10 mM NaCl
5mM HEPES + 5 mM NaCl
Aggregate Concentration (nM)
CONSTANT: adsorption to existing particles.
Colloidal stability (pH 9.0, 60C) of a-amylase (net charge -16)
Monovalent counterion (Na+)
5 mM buffer (Na-Borate)
buffer + 10 mM NaCl
buffer + 20 mM NaCl and buffer + 40 mM
Trivalent counterion (Co3+)
buffer + 10 mM Co[(NH3)6]Cl3
20 and 30 mM Co[(NH3)6]Cl3
60 and 120 mM Co[(NH3)6]Cl3
1st or 2nd order kinetics ?
40 mM NaCl
120 mM Co[(NH3)6]Cl3
Evans & Wennerstrøm, The Colloidal Domain 1999
At CCC the potential barrier is negligible – further addition of salt will not accelerate the aggregation.
Note counter-ion valens to the 6th !!
Hence 700 x increase from +1 to +3 – We see ca 400 X
The time course suggested 1st order when salt was added and 2nd order in pure buffer.
This is confirmed in rate vs. protein concentration measurements
2nd derivative FTIR spectrum of native and aggregated BHA with and without added salt
2nd order (UU complex)
1st order (N is rate limiting)
High salt: 1st order
Production of U from N is rate limiting
Experimentallyobservedlevel in 5 mMBorate, pH 9.0
Low salt : 2nd order
§Record et al. 1991, 1998
Weakly bipolar proteins (even charge distribution):
Coulombic protein-protein interactions promotes colloidal stability – conventional colloid stability (salts promote aggregation).
Strongly bipolar proteins:
Coulombic protein-protein interactions drives proteins together – salts weaken this initial step of aggregation.