Regular Articles
Amorphous–Amorphous Phase Separation of Freeze-Concentrated Protein and Amino Acid Excipients for Lyophilized Formulations
2016 Volume 64 Issue 12 Pages 1674-1680
Details
2016 Volume 64 Issue 12 Pages 1674-1680
The objective of this study was to elucidate the mixing state of proteins and amino acid excipients concentrated in the amorphous non-ice region of frozen solutions. Thermal analysis of frozen aqueous solutions was performed in heating scans before and after a heat treatment. Frozen aqueous solutions containing a protein (e.g., recombinant human albumin, gelatin) or a polysaccharide (dextran) and an amino acid excipient (e.g., L-arginine, L-arginine hydrochloride, L-arginine monophosphate, sodium L-glutamate) at varied mass ratios showed single or double Tg′ (glass transition temperature of maximally freeze-concentrated solutes). Some mixture frozen solutions rich in the polymers maintained the single Tg′ of the freeze-concentrated amorphous solute–mixture phase. In contrast, amino acid-rich mixture frozen solutions revealed two Tg′s that suggested transition of concentrated non-crystalline solute–mixture phase and excipient-dominant phase. Post-freeze heat treatment induced splitting of the Tg′ in some intermediate mass ratio mixture solutions. The mixing state of proteins and amino acids varied depending on their structure, salt types, mass ratio, composition of co-solutes (e.g., NaCl) and thermal history. Information on the varied mixing states should be valuable for the rational use of amino acid excipients in lyophilized protein pharmaceuticals.
Freeze-drying is a popular method to formulate therapeutic proteins that are insufficiently stable in aqueous solutions during storage.1) Several freeze-dried therapeutic protein formulations contain multiple excipients, including stabilizers, bulking agents, pH buffer agents, and tonicity modifiers besides active pharmaceutical ingredient (API).2–4) Sucrose and trehalose are popular stabilizers that protect proteins from dehydration-induced irreversible structural changes during the process and from chemical degradation during storage. Certain amino acids and their salts are potent stabilizers in the freeze-drying of proteins, frozen storage of liposomes, and spray drying of vaccines.5–11) Application of the amino acid excipients that protect proteins through particular mechanisms not achievable by saccharides (e.g., aggregation-reducing effect of L-arginine (L-Arg) in aqueous solution) would increase formulation strategies in lyophilization of marginally stable proteins.5,8,12–15)
Physical states (e.g., crystallinity, crystal polymorph) of the components are important factors that determine chemical and conformational stability of proteins during the freeze-drying process and subsequent storage.16–18) The stabilizing effects of disaccharides are attributed to protection of protein conformation by the substitution of surrounding water molecules and by holding protein molecules in lower molecular mobility glass-state amorphous solids. Crystallization of some sugar alcohols (e.g., mannitol) during the freezing segment of lyophilization deprives them of protein-stabilizing interactions. Use of the amino acid excipients in lyophilized formulations also depends on their varied crystallization propensities during the process. Formation of physically stable crystalline cake makes glycine (Gly) a popular bulking agent, while maintenance of amorphous state should be preferable for pH buffering agents (e.g., L-histidine (L-His)) and the stabilizers.19–21)
Mixing state of the protein with excipient molecules in the amorphous freeze-concentrate and in dried solids is another physical factor determining quality of multi-component freeze-dried protein pharmaceuticals.19,22,23) Ice growth during the freezing segment of the lyophilization process significantly concentrates the multiple solutes into a narrow space. Increasing concentration induces phase separation of some solute mixtures, including certain polymer combinations [e.g., polyvinylpyrrolidone (PVP) and dextran], into different amorphous phases rich in one of the components.23,24) Limited protein-stabilizing effects of some glass-forming polysaccharides (e.g., dextran) are explained by insufficient hetero-molecular interactions due to their poor mixing in the frozen solutions and steric hindrance.23)
Information on the mixing states of proteins and low molecular weight stabilizers would be practically more important in the development of lyophilized protein pharmaceuticals. Some spectroscopic and thermal analyses of freeze-dried solids containing proteins and disaccharides demonstrate incomplete mixing and/or multiple amorphous phases that are different in the solute compositions.25–28) Thermal analysis of frozen solutions provides Tg′ (glass transition temperature of maximally freeze-concentrated solutes), which profiles indicate mixing states of non-crystalline solutes.29) Freezing of aqueous solutions concentrates various proteins and up to certain mass ratio of disaccharides into the same amorphous mixture phase. The disaccharide molecules form their dominant concentrated phase above the critical mixing disaccharide/protein mass ratios that vary depending on solute composition (e.g., protein and disaccharide structure, co-solutes) and thermal history.30–32) The proteins and stabilizer molecules should be in the same non-crystalline freeze-concentrated phase to retain protein conformation from the dehydration stresses through the water-substituting molecular interactions (e.g., hydrogen bonding).33–36) The amorphous–amorphous phase separation of freeze-concentrated solutes is considered to affect quality of freeze-dried protein formulations in various ways, including altered physical stability of cake structure and crystallization propensity of certain solutes (e.g., myo-inositol).37,38) Mixing state of freeze-concentrated protein and amino acid molecules is of particular interest.
The objective of this study was to elucidate the mixing state of proteins and amorphous amino acids or their salts (amino acid excipients) concentrated in frozen solutions. Aqueous solutions containing L-Arg, its salt, fibrous protein, (gelatin) and a highly purified globular (recombinant human globulin) proteins were used as model systems.39,40) Better understanding of the mixing state in pharmaceutically relevant protein and amino acid systems should assist rational design of the formulations and the processes.
All chemicals and proteins employed in this study were obtained from the following sources: recombinant human albumin expressed in rice (rHA, Cellastim™), gelatin from porcine skin (gel strength 300, type A) (Sigma-Aldrich Co., St. Louis, MO); L-Arg, L-arginine monohydrochloride (L-Arg·HCl), monosodium L-glutamate monohydrate (Na·L-Glu, MSG), and sodium chloride (Wako Pure Chemical Industries, Ltd., Osaka, Japan); purified lower molecular weight porcine gelatin (RM-50, 40–60 kDa; Jellice Co., Sendai, Japan)13); dextran FP40 (SERVA Electrophoresis GmbH, Heidelberg, Germany).
The rHA was dialyzed overnight against 5 mM sodium phosphate buffer (pH 7.0) before measuring concentration by absorbance at 280 nm. Gelatin was dissolved in a buffer solution and filtered (0.22 µm, polyvinylidene difluoride (PVDF), Millipore, U.S.A.) before adjusting the concentration of the protein as specified in the product data sheet. The gelatin solutions and their mixtures were prepared and maintained at 40°C before thermal analysis. The amino acid excipients were dissolved in sodium phosphate buffer (5 mM, pH 7.0) and used without pH adjustment. Aqueous L-arginine monophosphate (L-Arg·H3PO4) solutions were prepared by mixing aqueous L-Arg and phosphoric acid solutions.
Thermal AnalysisA differential scanning calorimeter (DSC Q10; TA Instruments, New Castle, DE, U.S.A.) and software (Universal Analysis 2000, TA Instruments) were used for thermal analysis of frozen solutions. Aliquots (10 µL) of aqueous solutions in hermetic aluminum cells were cooled from room temperature to −70°C at 10°C/min before heating scans at 5°C/min. Some scans were paused at −25 to −5°C (−10°C, otherwise mentioned), kept at the temperature for 1–480 min (30 min, otherwise mentioned), then cooled again to −70°C before second heating scans at 5°C/min. The post-freeze heat treatment of gelatin-containing solutions was performed at −5°C. The Tg′ of frozen solutions was obtained from the temperature of the peaks in derivative thermograms.
Derivative DSC curves of frozen solutions containing various polymers or amino acid excipients (100 mg/mL, 5 mM sodium phosphate buffer) obtained in the first heating scans are shown in Fig. 1. The frozen solutions showed a Tg′ transition peak at −46.9°C (L-Arg·HCl), −50.0°C (Na·L-Glu), −42.7°C (L-Arg), −31.4°C (L-Arg·H3PO4), −12.0°C (gelatin type A), −12.7°C (gelatin RM-50), and −13.9°C (dextran 40k). The transition of frozen gelatin RM-50 solution was more clear than that of gelatin type A.12) In contrast, the frozen rHA solution showed only a gradual shift of the DSC curve from approximately −23°C, suggesting a broader Tg′ starting from the low temperature. Second scans of frozen solutions after heat treatment did not alter the thermal profiles (data not shown). Observation of the Tg′ transitions, absence of solute crystallization peaks, and gradual baseline shift between the Tg′ and the ice melting endotherm are typical for frozen solutions containing amorphous concentrated solutes.
Figure 2 shows Tg′s of frozen solutions containing varied mass ratios of rHA and L-Arg·HCl (total 50, 100, 150 mg/mL) obtained before and after heat treatment. Frozen solutions containing a higher mass ratio of rHA (0.4–0.8, total: 100 mg/mL) showed single Tg′ transition before and after heat treatment, suggesting that the solutes remain mixed in the amorphous freeze-concentrated phase. Some mixture frozen solutions predominant in L-Arg·HCl showed two Tg′ transitions in the initial scan. The heat treatment at −10°C induced splitting of the Tg′ at an intermediate mass ratio range, suggesting phase separation of these concentrated solutes. Lower temperature Tg′, appearing at temperatures closer to that of the single-solute L-Arg·HCl solution, may indicate physical changes of the phase dominant in L-Arg·HCl. The higher temperature transitions may be Tg′ of the solute–mixture phase containing rHA and up to certain mass ratios of L-Arg·HCl.
Aliquots (10 µL) of solutions were scanned from −70°C at 5°C/min prior to or after heat treatment at −10°C (30 min) (n=3, average±standard deviation (S.D.)).
Frozen solutions containing lower (50 mg/mL) and higher (150 mg/mL) total concentration rHA and L-Arg·HCl showed similar Tg′ profiles, including a double Tg′ at the higher L-Arg·HCl ratios. This indicated the relevance of the solute mass ratio, rather than the absolute concentrations, on the phase behavior in the frozen solutions. Unclear transitions in derivative DSC curves, particularly in lower concentration frozen solutions, would partially explain the differences between the Tg′ profiles.
Effect of heat treatment temperature (A) and time (B) on derivative DSC curves of a frozen solution containing 30 mg/mL rHA, 70 mg/mL L-Arg·HCl, and 5 mM sodium phosphate buffer are shown in Fig. 3. Post-freeze heat treatment at high temperature and longer time induced splitting of Tg′ transition, suggest a contribution of increased solute molecular mobility in the non-ice region for spatial reordering to thermodynamically more stable states. The frozen solution maintained the two Tg′s during the longer heat treatment at −10°C (480 min). Contrarily, some protein-rich frozen solutions (e.g., 60 mg/mL rHA and 40 mg/mL L-Arg·HCl, C) kept the single Tg′ during the long heat treatment. The post-freeze heat treatment (annealing) is a popular method to induce crystallization of certain small molecule APIs or bulking agents (e.g., mannitol), as well as to obtain faster-subliming larger ice crystals by Ostwald ripening.1,14)
Aliquots (10 µL) of solutions were scanned from −70°C at 5°C/min prior to or after a heat treatment.
Mixing states of the freeze-concentrated polymers and amino acid excipients were further studied through thermal analysis. Tg′ profiles of frozen solutions containing various mass ratios of gelatin RM-50 or dextran 40k and L-Arg·HCl obtained in scans prior to and after a heat treatment are shown in Fig. 4. The gelatin- or dextran-rich frozen mixture solutions showed single Tg′ that shift cotinuously depending on the mass ratios. Some mixture frozen solutions dominant in L-Arg·HCl showed two Tg′ transitions both before and after heat treatment. Heat treatment of some frozen solutions induced splitting of the Tg′ transition. The clear Tg′ transition, makes purified gelatins a good model to study mixing states of proteins and excipients in frozen solutions, although their physical and chemical characteristics are much different from several globular therapeutic proteins. It is possible that the gelation and associated solution compartmentalization before freezing affect mixing state of freeze-concentrated solutes.41) Dextran and L-Arg·HCl mixture frozen solutions showed two transitions at a much narrow mass ratio than proteins and L-Arg·HCl, suggesting a high critical mixing L-Arg·HCl/polymer mass ratio in the freeze-concentrate.
Heating DSC scans were performed from −70°C (5°C/min) prior to or after a heat treatment at −5°C (gelatin RM-50) or −10°C (dextran 40k) (30 min) (n=3, average±S.D.).
The transition temperature profiles of frozen solutions containing rHA and L-Arg, or L-Arg·H3PO4 are shown in Fig. 5. These frozen solutions showed single Tg′ peaks in derivative DSC curves at all mass ratios studied. Broader transitions, as well as the discontinuities in Tg′ profiles observed in some mixture frozen solutions rich in L-Arg or L-Arg·H3PO4, suggested incomplete mixing of the freeze-concentrated solutes. The different Tg′ profiles in the systems containing L-Arg·HCl (Fig. 2) and L-Arg·H3PO4 (Fig. 5) suggested contribution of salt types on the varied mixing states in frozen solutions. The high Tg′ of frozen L-Arg·H3PO4 solutions may be attributed to their multiple hydrogen bonding network.5,42) While L-Arg tends to crystallize during the heating process of single-solute frozen aqueous solutions, it remained amorphous in the presence of low concentration sodium phosphate buffer.5,42)
Heating DSC scans were performed from −70°C (5°C/min) prior to or after heat treatment at −10°C (30 min) (n=3, average±S.D.).
The effect of NaCl on derivative DSC curves of frozen solutions containing rHA (20 mg/mL) and L-Arg (80 mg/mL) obtained before and after a post-freeze heat treatment are shown in Fig. 6. The mixture showed single Tg′ transition both before and after heat treatment in the absence of NaCl. Addition of NaCl shifted the single transition to a lower temperature in the first scans. Heat treatment of the NaCl-containing frozen solutions at −10°C induced splitting of the Tg′ transition, suggesting separation of freeze-concentrated rHA and L-Arg into the two phases. The effect of NaCl would be explained by shielding of the electrostatic interaction between rHA and L-Arg.
Heating DSC scans (5°C/min) were performed from −70°C prior to or after a heat treatment at −10°C (30 min).
Mixing states of the freeze-concentrated proteins (rHA, gelatin RM-50) and Na·L-Glu) were also studied (Fig. 7). Sodium glutamate is a potent stabilizer used in freeze-dried vaccine formulations.43) Some protein-rich mixture frozen solutions showed sharp single Tg′ transition peak that shifted depending on the solute mass ratio. Contrarily, frozen solutions containing a higher mass ratio of Na·L-Glu showed broad transitions that have a peak at temperatures close to Tg′ of the single-solute frozen Na·L-Glu solution in the derivative DSC curves. Two Tg′ transitions were observed in the intermediate mass ratio ranges. The mixture frozen solutions with gelatin RM-50 showed the lower-Tg′ of the possible Na·L-Glu-dominant phase at wider mass ratio ranges. Heat treatment of the frozen solutions did not alter the Tg′ profiles. It is possible that broader transition makes Tg′ of the protein and Na·L-Glu mixture phase unclear in the mixture frozen solutions rich in Na·L-Glu.
Heating DSC scans were performed from −70°C (5°C/min) prior to or after a heat treatment at −10°C (rHA) or −5°C (gelatin RM-50) (30 min) (n=3, average±S.D.).
The results indicate varied mixing states of proteins and amino acid excipients in frozen solutions. Various frozen solutions containing combinations of structurally different proteins or polysaccharide and salts of acidic (L-Glu) or basic (L-Arg) amino acids showed amorphous–amorphous phase separation at certain mass ratios. These solutes may form the multiple amorphous phases basically in the same mechanism with previously reported phase separation in some polymer and disaccharide mixture frozen solutions.30,31) Freezing of the protein-rich solutions should concentrate all the solutes into practically stable highly viscous amorphous non-ice mixture phase. In contrast, solutions rich in amino acid excipients would form solute–mixture and amino acid-dominant phases during the initial freeze-concentration and/or upon post-freeze heat treatment. Observation of two Tg′ transitions in the first heating scans of frozen solutions (e.g., rHA and L-Arg·HCl) indicated higher propensity of the solutes for phase separation during the ice-growth process.
The transition temperature profiles of proteins and amino acid excipients indicated a large variation in the critical mixing amino acid/protein mass ratio depending on combinations and co-solute compositions. Apparent variation in the chemical and physical properties of amino acid excipients (e.g., pH), as well as their electrostatic interaction with protein molecules, may explain the larger variation in their mixing states than those with sugars and sugar alcohols.21) In addition, the size and structure of the proteins may also affect thermodynamic and kinetic factors determining the mixing states. Further clarification of factors determining the phase behavior may contribute to formulation developments.
The large effect of post-freeze heat treatment on the mixing states indicated relevance of thermal history and associated process factors on component mixing states. Careful characterization of the candidate formulations by thermal analysis and freeze-drying microscopy (FDM) or small-scale lyophilization studies may assist appropriate product temperature control to avoid physical collapse during primary drying and/or solid shrinkage during secondary drying segment of the multi-phase systems.44,45) In addition, understanding the physical properties of particular formulations would enable rational application of new technologies employed to reduce the lyophilization process time (e.g., controlled nucleation).
The mixing state of protein and amino acid excipients would affect stability of proteins in frozen solutions and freeze-dried solids in several ways. Phase separation at the amino acid-rich frozen solutions should not directly damage conformation of the protein distributed to the solute–mixture phase. The amino acid molecules concentrated with the proteins in the amorphous mixture phase would provide intermolecular hydrogen bonding that protects the protein conformation from low temperatures and dehydration stresses.5–9) However, the formation of an amino acid-dominant phase would alter local environment surrounding the protein molecules (e.g., pH, salt concentration) through uneven contribution of third components (e.g., NaCl, phosphate buffer).
This study showed that mixing states of amorphous proteins and amino acid excipients in frozen solutions varies depending on their combinations, mass ratios, co-solute compositions, and thermal history. Further clarification of factors determining mixing states in pharmaceutically relevant amorphous systems may assist development of complex formulations. Some amino acid excipients have high propensity to crystallize in frozen solutions.46) Mixing state of proteins and those amino acid excipients, and their effect on the excipient crystallization is the topic for our next study.
This work was partly supported by the Health and Labour Sciences Research Grants.
The authors declare no conflict of interest.
