Formulation and Delivery of Proteins and Peptides - American

Creighton, T.E., Ed.; IRL Press: New York, NY, 1988; pp. 331-345. 6. Lee ... Ashwood-Smith, M.J.; Warby, C. Cryobiology 1972, 9, 137-140. 29. Prestels...
0 downloads 0 Views 2MB Size
Chapter 9 Interactions of Stabilizers with Proteins During Freezing and Drying 1

2

1

John F. Carpenter , Steven J. Prestrelski , Thomas J. Anchordoguy, and Tsutomu Arakawa 2

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

1

School of Pharmacy, University of Colorado Health Sciences Center, Denver, CO 80262 Amgen, Inc., Amgen Center, Thousand Oaks, CA 91320 2

During the course of processing, shipping and storage, proteins are often subjected to either freezing or drying stress, or both. The preservation of proteins during freezing and drying is fundamentally different for each process. Protection of labile proteins during freezing appears to be due to the same thermodynamic mechanism that accounts for solute-induced stabilization in nonfrozen aqueous solution. Namely, preferential exclusion of the solute leads to stabilization of the native state of the protein. Protection of proteins during drying by sugars is due to the sugar hydrogen bonding to the dried protein and serving as a water replacement. Recovery of enzyme activity after freeze-drying and rehydration correlates directly with the maintenance of the native protein structure in the dried state. Sugars protect labile proteins by inhibiting lyophilization-induced unfolding. Finally, certain polymers (e.g., polyvinylpyrrolidone) protect freeze-dried multimeric enzymes during freeze-drying and rehydration. These compounds cannot serve as "water replacements" in the dried state. Rather they inhibit the freezing- and drying-induced dissociation. The fully polymerized oligomers are more resistant to these stresses. With increasing numbers of applications for recombinant and naturally-derived proteins, the need for stable protein formulations is growing. When the inherent stability of the protein, and/or the logistics of product shipping and use, preclude storage as an aqueous solution, the protein is usually freeze-dried. Often, the focus during formulation development is on the terminal stress of dehydration and on the effects of additives on the stability of the protein in the dried solid. However, freezing stress also occurs during lyophilization, as well as at other points during processing, shipping and storage. Freezing can occur either by design or by accident. For example, a protein may be very stable in a liquid formulation at 4°C, but if this same protein accidently freezes during shipping, then the product can be 0097-6156/94/0567-0134$08.00/0 © 1994 American Chemical Society In Formulation and Delivery of Proteins and Peptides; Cleland, J., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

9.

CARPENTER ET AL.

Stabilizer-Protein Interactions in Freezing & Drying 135

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

destroyed. Planned freezing includes situations where the protein is stored frozen and thawed by the end-user, lyophilized proteins are rehydrated by the end-user and then aliquoted and frozen, and frozen storage is used as an interim measure for holding batches of proteins, prior to the lyophilization of a large lot. Thus, for the development of successful lyophilized formulations, both freezing and drying stresses must be taken into consideration. This review will focus on stress-specific preservation of proteins and the mechanisms by which additives protect the structure and function of labile proteins during freezing and drying. That is, we are concerned primarily with proteins that are acutely sensitive to the freezing and/or drying stresses and that can be irreversibly damaged in the absence of a stabilizing additive. Another related category of proteins are those that unfold during freeze-drying, but resume their native conformation upon rehydration (see chapter by Prestrelski et al., for further discussion of dehydration-induced structural changes). Our view is that, in this case, stabilizing additives can be used to prevent the reversible denaturation and that the native protein may be more resistant to degradation during long-term storage of the dried product. This suggestion is mostly speculative at this time, because only recently has this type of acute drying-induced structural transition been identified. Finally, certain proteins are intrinsically resistant to the acute effects of lyophilization and retain their native conformation in the dried state. However, without the proper additives, these proteins may degrade chemically during long-term storage. We will not address this level of stabilization since it will be reviewed rigorously by others in this volume. However, we wish to emphasize, that for more labile proteins, the initial preservation of the native conformation in the dried formulation appears to be a fundamental requirement for subsequent longterm storage stability. Mechanism for Protein Preservation During Freezing A wide variety of compounds will protect labile proteins during freeze-thawing (reviewed in ref. 1-4). These include sugars, amino acids, polyols, methylamines, synthetic polymers (e.g., polyethylene glycol, PEG), other proteins (e.g., bovine serum albumin, BSA) and even inorganic salts (e.g., potassium phosphate and ammonium sulfate). For most proteins, the cryoprotectant must be at a concentration of several-hundred millimolar to confer maximum protection. Exceptions are polymers such as PEG, BSA and polyvinylpyrrolidone (PVP), which at concentrations of even less than 1% (wt/vol) ftilly protect sensitive enzymes such as lactate dehydrogenase (LDH) or phosphofructokinase (PFK). Based on the results of freeze-thawing experiments with L D H and P F K and a review of the literature on protein freezing, we have determined (1) that protein cryopreservation can be explained by the same universal mechanism that Timasheff and Arakawa have defined for solute-induced protein stabilization in nonfrozen, aqueous solution (3-5). Detailed explanations of this mechanism can be found elsewhere (3-5). For the purpose of the current review a brief summary will suffice.

In Formulation and Delivery of Proteins and Peptides; Cleland, J., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

136

FORMULATION AND DELIVERY O F PROTEINS AND PEPTIDES

Timasheff, Arakawa and their colleagues have observed experimentally that there is a deficiency of stabilizing solutes (e.g., sugars and polyols) in the immediate vicinity of the protein, relative to the bulk solution, and that the protein is preferentially hydrated. That is, the solutes are preferentially excluded from contact with the surface of the protein. The presence of the preferentially excluded solutes in a protein solution creates a thermodynamically unfavorable situation, because the chemical potentials of both the protein and the additive are increased. Consequently the native structure of monomers or the fully polymerized form of oligomers are stabilized because denaturation or dissociation, respectively, would lead to a greater surface area of contact between the protein and the solvent, and therefore, exacerbate this thermodynamically unfavorable effect. An overview of the key thermodynamic aspects of this mechanism follows (reviewed in 5-5). Setting component 1 = principal solvent (here water), component 2 = protein, and component 3 = solute (e.g., sucrose or PEG), the preferential interaction of component 3 with a protein is expressed, within close approximation, by the parameter, (bm^lbm^^^, at constant temperature and pressure, where μ and m are the chemical potential and molal concentration of component i , respectively. A positive value of this interaction parameter indicates an excess of component 3 in the vicinity of the protein over the bulk concentration (i.e., preferential binding of the solute). A negative value for this parameter indicates a deficiency of component 3 in the protein domain. Component 3 (the solute) is preferentially excluded and component 1 (water) is in excess in the protein domain. χ

x

The preferential interaction parameter is a direct expression of changes in the free energy of the system induced by component 3, and has the relation: (ô/x /ÔW3) = - (bmjbmi)^ 2

m2

^lbm^)

(1)

m2

Equation 1 indicates that those compounds that are excluded (i.e., (bm^lbm^^ < 0) from the surface of the protein will have positive values of (b^bm^'y they will increase the chemical potential of the protein, rendering the system more thermodynamically unfavorable. In the presence of excluded solutes, the exclusion will be greater for the denatured form of the protein than for the native form because the former has a larger surface area, as indicated by: (bn^/bm^ < (bm /bm ) < 0. Consequently, the increase in chemical potential is greater for the denatured form than for the native form in the presence of a preferentially excluded solute, as indicated by: (fy* /ôm ) > (ô/x /ôw ) > 0. N

3

D

2

3

m2

2

N

2

3

m2

This effect of the solute leads to an increase in the free energy difference between the native and denatured forms, thus stabilizing the native state. Ultimately, it is this difference between the effects of the solute on chemical potential of the native versus the denatured state that determines i f a compound will serve as a protein stabilizer. As long as the degree of exclusion is greater for the denatured state, the solute will be a protein stabilizer. The opposite is seen for potent protein dénaturants such urea and guanidine HC1.

In Formulation and Delivery of Proteins and Peptides; Cleland, J., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

9.

CARPENTER E T A L .

Stabilizer-Protein Interactions in Freezing & Drying 137

These solutes bind preferentially to both the native and the denatured form of the protein (3-5) and hence decrease the chemical potential of the protein. Since the number of available binding sites is increased upon unfolding of the protein, an increase in preferential solute binding occurs as indicated by: (bn^lbrn^f > (bm^lbm^f > 0. There is a concomitant decrease in protein chemical potential, which is greater for the denatured state: ( W ^ ) , ^ < ( δ μ ^ & η ^ " < 0. This serves to lower the free energy difference between the two states, and when the native state becomes the higher energy state, protein denaturation should result.

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

0

As noted above, the same principles apply to the influence of solutes on the degree of assembly of multimeric proteins. Preferentially excluded solutes tend to induce polymerization and stabilize oligomers since the formation of contact sites between constituent monomers serves to reduce the surface area of the protein exposed to the solvent. Polymerization reduces the thermodynamically unfavorable effect of preferential solute exclusion. Conversely, preferential binding of solute induces depolymerization since there is greater solute binding to monomers than to polymers. Since it is not possible to measure preferential interactions between solutes and proteins in frozen samples, we cannot state that cryoprotectants are actually preferentially excluded from the frozen proteins. However, preferential interactions of solutes with proteins in nonfrozen, aqueous solutions can be used to predict the cryoprotective capacity of solutes. For example, the degree of perturbation of protein chemical potential arising in the presence P E G and sucrose, respectively, can be used to explain why PEG is a much more potent cryoprotectant than sucrose. The data for one case, which are shown in Table I, will serve to illustrate this point. The increase in chymotrypsinogen chemical potential ,^ /bm ) , in the presence of either of two different molecular weights of P E G (e.g., M r = 400 or 6000) is greater than that noted in the presence of the sucrose; even though the P E G is excluded to a lesser degree, on a per mole of solute basis. Comparing the two P E G molecules indicates that the larger the PEG the less it is excluded on a mole basis, but the more that it increases protein chemical potential. 2

3

m2

The basis for these observations can be explained by examining Equation 1. The other major component in determining the effect of solute on protein chemical potential is the self-interaction parameter for the solute, (δμ /δ^ ) . The value for this parameter is several-fold greater for P E G 400 and almost three orders of magnitude greater for P E G 6000, than that for sucrose. The self-interaction parameter is given by: 3

3

/η2

= [(RT/m ) + RTiffli^/aw^wJ 3

(2)

where Jf is the activity coefficient of the solute and R is the universal gas constant (5-7). The molal concentrations needed for preferential exclusion of P E G are very small and the activity coefficient of P E G is quite large, relative to values for sucrose. Therefore, the self-interaction parameter for P E G is very large compared 3

In Formulation and Delivery of Proteins and Peptides; Cleland, J., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

138

FORMULATION AND DELIVERY O F PROTEINS A N D PEPTIDES

to that for sucrose. In addition, as the size of P E G increases there is a great increase in such nonideality (Table I).

Table I. Parameters for Solute Interactions with Chymotrypsinogen Solute

Cone.

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

Sucrose

0

P E G 400

1.27 m d

PEG6000

a

b

c

d

(δ/τ^/δτηΛ,α^

10% w/v (0.27 m) d

l%w/v (0.0017 m)

-0.62

480.00

1

kcal (mol of solute) (mol of solute in 1000 kg H20) kcal (mol of protein)" (mol of solute)" Data taken from ref. 6 Data taken and calculated from ref. 7 1

297.6

1

1

When preferential exclusion is considered on a mass specific basis, (àg /ôg ), the degree of exclusion increases with increasing PEG molecular weight. The exclusion is due to steric hindrance of P E G interaction with the protein (7). Also, the nonideality for P E G and its perturbation of protein chemical potential increase with molecular weight. Based on these values and the above discussion, one would predict that the larger the PEG the more effective it should be at protecting labile enzymes during freeze-thawing. The data presented in Figure 1 support this hypothesis. L D H is completely protected during freeze-thawing by P E G M W 8000 at concentrations of >_ 0.01 % (wt/vol). In contrast, full protection in the presence of P E G M W 400 is not realized until the concentration is at least 2.5% (wt/vol). On a weight percentage basis, P E G M W 8000 is 250-fold more potent as a cryoprotectant. On a molar basis, the higher molecular weight P E G is 5000-fold more potent. Thus, we conclude that cryoprotection of proteins by P E G is due to preferential exclusion, which is in turn due to steric hindrance of the interaction of P E G with the protein. 3

2

Since degree of stabilization correlates directly with the increase in protein chemical potential in the presence of a solute, it is not surprising that P E G is much more effective than sugars at protecting labile enzymes during freezing. Interestingly, this correlation does not hold for high temperature denaturation experiments. Sugars increase the melting temperature for proteins, but P E G decreases protein stability at high temperature (6,8). This effect has been ascribed to increased hydrophobic interaction of P E G with proteins as temperature is increased, which leads to preferential binding to the denatured form (reviewed in ref. 9).

In Formulation and Delivery of Proteins and Peptides; Cleland, J., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

Downloaded by PENNSYLVANIA STATE UNIV on June 7, 2012 | http://pubs.acs.org Publication Date: August 19, 1994 | doi: 10.1021/bk-1994-0567.ch009

9.

CARPENTER E T A L .

Stabilizer-Protein Interactions in Freezing & Drying 139

100

"D Φ k_ V > Ο

υ ω

m/%

80

o

PEG 8000

60

CE

m

- m m

Q Ψ

PEG 400

m

m

V o ^ ^

A A

/

Δ

40 "ο