`
`RESEARCH ARTICLE
`
`Formulation of Proteins in Vacuum-Dried
`Glasses. II. Process and Storage Stability in
`Sugar-Free Amino Acid Systems
`
`Markus Mattern,1,2 Gerhard Winter,2 Ulrich Kohnert,3
`and Geoffrey Lee1
`1DepartmentofPharmaceuticalTechnology,Friedrich-AlexanderUniversity,
`Erlangen,Germany
`2BoehringerMannheimGmbH,GalenicalDevelopment,Mannheim,Germany
`3BoehringerMannheimGmbH,BiochemicalResearch,Penzberg,Germany
`
`Received February 18, 1998; Accepted August 6, 1998
`
`ABSTRACT
`
`The purpose of this research was to investigate the freeze- and vacuum-drying behavior of (cid:108)-amino
`acids of current/potential use as adjuvants for formulating proteins. The analytical methods used
`were wide-angle x-ray diffraction, differential scanning calorimetry, and scanning electron micros-
`copy. Protein analysis was performed either as an activity assay (lactate dehydrogenase [LDH]) or
`by size-exclusion chromatography (granulocyte colony-stimulating factor [rhG-CSF]). After samples
`were freeze-dried, only the four basic amino acids (arginine, lysine, histidine, and citrulline) formed
`amorphous solids, which, however, were partially crystalline. The remaining amino acids all formed
`fully crystalline solids. After samples were vacuum-dried, (20(cid:176) C, 0.1 mbar, 1 ml fill volume in 2-ml
`vials) fully crystalline solids were formed by all of the amino acids. For arginine, the addition of
`either HCl, H3PO4, or H2SO4 sufficient to form the respective salt produced amorphous solids after
`vacuum-drying, but they had high residual water contents and low glass transition temperatures (Tg).
`Addition of phenylalanine to arginine base inhibited crystallization of the latter at low concentrations
`during vacuum-drying procedure, leading to formation of a pure rubbery solid. At higher concentra-
`tions the phenylalanine crystallized, producing dry products with glass transition temperatures of
`.60(cid:176) C. The process and storage stability of LDH and rhG-CSF in the vacuum-dried phenylalanine/
`arginine glasses was greatly improved at temperatures up to 40(cid:176) C compared with the unprotected
`proteins. Uptake of moisture during storage was, however, a complicating factor, reducing Tg, promot-
`ing crystallization, and leading to decreased protein stability. The PO4 salt of arginine produced
`especially high glass transition temperatures after it was vacuum-dried. These sugar-free amino acid
`formulations thus are potential stabilizes for proteins.
`KEY WORDS: Amino acid; Freeze-drying; Phosphoric acid; Protein stabilization; Vacuum-drying.
`
`Address correspondence to Geoffrey Lee, Lehrstuhl fu¨r Pharmazeutische Technologie, Cauerstr. 4, 91058 Erlangen, Germany. Fax:
`09131/85 95 45. E-mail: g.lee@pharmtech.uni-erlangen.de
`
`Copyright (cid:211)
`
`1999 by Marcel Dekker, Inc.
`
`www.dekker.com
`
`199
`
`-199-
`
`
`
`
`MAIA Exhibit 1039
`MAIA V. BRACCO
`IPR PETITION
`
`
`
`200
`
`Mattern et al.
`
`INTRODUCTION
`
`A number of amino acids are frequently cited as being
`suitable bulking agents for freeze-dried formulations.
`There exists some evidence that certain amino acids also
`act as cryoprotectants. These protect sensitive active
`agents, especially peptides and proteins, against various
`destabilizing effects occurring in aqueous solution during
`the initial freezing phase, e.g., freeze concentration and
`pH shifts (1). Carbohydrates, amino acids, glycerol, and
`polyethylene glycols (2) are all thought to act by their
`so-called preferential exclusion from the protein/water
`interface (3). This maintains native protein structure, pro-
`vided the cryoprotectants are present in sufficient quanti-
`ties (4). There is no evidence that amino acids alone (i.e.,
`not combined with carbohydrates) can also act as lyopro-
`tectants, which protect proteins against physical and
`chemical instabilities in the dried state after they are
`freeze-dried. Indeed, it is carbohydrates, especially disac-
`charides such as sucrose, maltose, and trehalose, which
`are the most effective lyoprotectants (5).
`It was found previously that vacuum-drying could be
`improved by the judicious combination of disaccharides
`with the crystallizing amino acid phenylalanine (Phe), to
`produce partially amorphous phases having residual
`moisture of ,1.5% (6). Proteins could be rendered stor-
`age stable at room temperature in these vacuum-dried
`sugar/amino acid glasses. The question then arose
`whether pure amino acids also form amorphous glasses
`of sufficiently low residual moisture and high glass tran-
`sition temperature when vacuum- or freeze-drying proce-
`dures are used. If so, do the amino acids act as lyoprotec-
`tants and can proteins be stabilized in such sugar-free
`systems? The literature indicates that glycine, for exam-
`ple, crystallizes during freeze-drying (7), its behavior be-
`ing dependent on pH and salt form. The literature did
`not, however, yield a comparative study of a wider spec-
`trum of amino acids regarding their behavior during
`freeze-drying. In this paper we identify those amino acids
`which form amorphous glasses during either freeze- or
`vacuum-drying procedures. The importance of salt forms
`for the glassy amino acids is a feature of this work, as is
`the use of mixed amino acid systems to reduce residual
`water content. The lyoprotective ability of the vacuum-
`dried amino acid glasses was investigated using lactate
`dehydrogenase (LDH) and recombinant human granulo-
`cyte colony-stimulating factor (rHG-CSF).
`
`MATERIALS AND METHODS
`
`All of the (cid:108)-amino acids used were of analysis grade
`and obtained from either Merck (Darmstadt, Germany),
`
`Sigma (Munich, Germany), Clintec Salvia (Frankfurt,
`Germany), or Degussa (Frankfurt, Germany). Hydrochlo-
`ric and phosphoric acids were obtained from Merck. Wa-
`ter was double-distilled from an all-glass apparatus. LDH
`and rhG-CSF were used as described previously (6).
`
`Preparation and Physicochemical
`Characterization
`
`All solutions of the amino acids investigated were first
`filtered through a 0.22-m m membrane filter and filled into
`2 ml vials. These were then either freeze- or vacuum-
`dried using a freeze-dryer with 0.6 m2 shelf area
`(Scha¨fer & Hof, Lohra, Germany), as described previ-
`ously (6). Vacuum-drying was conducted for 24 hr at
`20(cid:176) C and down to 0.1 mbar. Each 2 ml vial contained
`1 ml of solution, the rubber stoppers being pressed into
`place at the end of the drying cycle after the chamber
`was flooded with dry air. The freeze-drying conditions
`were as follows: shelf temperature 225(cid:176) C, primary dry-
`ing at 0.1 mbar, secondary drying at shelf temperature
`rising to 10(cid:176) C and at 0.01 mbar.
`Karl Fischer (Mettler, Greifensee, Switzerland) titra-
`tion was used to determine the residual water content of
`each dried sample. Glass transition and crystallization
`temperatures were determined by differential scanning
`calorimetry (DSC 7, Perkin-Elmer, Munich, Germany).
`Samples (10–15 mg) in sealed Al pans were repeatedly
`cooled and heated at 10(cid:176) C/min in the range 250–150(cid:176) C.
`The crystallinity of the dried samples was examined us-
`ing wide-angle x-ray diffraction (PW 1720, 40 kW, Cu-
`Ka
`, l 5 0.15418 nm; Phillips, Kassel, Germany). Pow-
`dered samples (300 mg) were examined at 25 6 1(cid:176) C.
`The dried samples were also examined using scanning
`electron microscopy (SEM, DSM 902, Zeiss, Germany)
`after gold sputtering was performed in an Al sample
`holder.
`
`Stability of LDH and rhG-CSF in Dried
`Amino Acids
`
`Solutions of the amino acids of interest were prepared
`containing either 165 units LDH/ml or 0.35 mg rhG-CSF
`per 1 ml of solution. After vacuum-drying was per-
`formed, these samples were stored at various tempera-
`tures and examined periodically during up to 9 months
`storage. The stability of the rhG-CSF in the dried prod-
`ucts was assessed by determining of the amounts of mo-
`nomer and aggregated dimer using size exclusion chro-
`matography (SEC-HPLC). Each dry sample was first
`reconstituted in water and 20 m
`l was injected into an
`HPLC system (Shimadzu, Munich, Germany) with a re-
`
`-200-
`
`
`
`
`
`
`Formulation of Proteins in Vacuum-Dried Glasses. II
`
`201
`
`frigerated autosampler (TM 717, Waters, Frankfurt, Ger-
`many). A TSK-Gel G 2000 SW (7.5 3 300 mm) column
`was used (Toso Haas) with detection at 214 nm (LC-GA,
`Shimadzu). The mobile phase was 0.1 M Na/K phos-
`phate buffer of pH 6.2, with a flow rate of 0.6 ml/min
`at room temperature. Under these conditions the retention
`time of the monomer was 14.5 min against an rhG-CSF
`external standard. With LDH a direct determination of
`enzymatic activity was possible. Potassium phosphate
`buffer (2.5 ml 0.1 M, pH 7.0), pyruvate (0.1 ml 0.02 M),
`and reduced nicotinamide adenine dinucleotide (NADH)
`(0.05 ml 0.011 M) were mixed at 25(cid:176) C in a plastic cu-
`vette. To this 0.05 ml of the test LDH solution was added,
`and after the solution was mixed, the extinction was mea-
`sured at l 5 365 nm over a period of 5 min (552 UV/
`VIS photometer, Perkin-Elmer). The enzymatic activity
`(U/ml) was then calculated from the rate of change in
`extinction pro min, D E/D
`t, using
`U/ml 5 (D E/D
`t (cid:215) volume of solution)/
`(extinction coefficient (cid:215) pathlength
`(cid:215) volume of LDH solution)
`
`(1)
`
`RESULTS AND DISCUSSION
`
`Freeze-Drying of Amino Acids
`
`Only the four basic amino acids (lysine, arginine
`[Arg], histidine, and citrulline) showed conspicuous glass
`transitions after they were freeze-dried. For the example
`of (cid:108)-arginine, we found 1.2% w/w residual water and a
`glass transition at Tg 5 42(cid:176) C, followed by an exothermic
`crystallization peak at Tc 5 61(cid:176) C. This behavior is typi-
`cally seen with amorphous phases of, for example, carbo-
`hydrates (8). All of the other amino acids examined (Ta-
`ble 1) showed no detectable Tg or Tc and were apparently
`fully crystalline after being freeze-dried under the condi-
`tions used here. This finding underscores the frequently
`cited use of amino acids as crystalline bulking agents (2)
`[for example, glycine (6)]. Wide-angle x-ray diffraction
`reveals that the four basic amino acids were not, however,
`fully amorphous directly after being freeze-dried. The
`x-ray diffractogram of the freeze-dried (cid:108)-arginine [Fig.
`1(a)], for example, shows several crystalline reflections
`superimposed on the characteristic amorphous halo. The
`positions of these peaks are identical to those of anhy-
`drous crystalline (cid:108)-arginine (not shown). The degree of
`crystallinity in Fig. 1(a) is approximately 10%, which lies
`within the range quantifiable using standard methods of
`x-ray diffraction (8). Heating to above a sample’s Tc pro-
`duced spontaneous conversion to the fully crystalline
`form of the respective basic amino acid, as could be con-
`
`firmed from its x-ray diffractogram. Thus, under the stan-
`dard freeze-drying conditions used here, the four basic
`amino acids form amorphous solids, but with some crys-
`talline content.
`The addition of mineral acids to the (cid:108)-arginine base
`solution before freeze-drying was performed greatly al-
`tered the properties of the dried product. The amount of
`acid added was chosen to give the correct ratio of Arg
`base to acid required to form the corresponding salt. An
`equimolar solution of (cid:108)-arginine and HCl is thus assumed
`to form Arg Cl salt, and resulted in a more than doubled
`residual water content, and Tg was accordingly sharply
`reduced from 42 to 18(cid:176) C (Table 2). The use of a third
`molar part of H3PO4 in the (cid:108)-arginine solution (Arg PO4)
`also resulted in a higher residual water content than with
`the base, but there was a dramatic increase in Tg from 42
`to 93(cid:176) C (Table 2). In addition, the dried product of Arg
`PO4 was fully amorphous [Fig. 1(b)], as there were no
`signs of the crystalline reflections seen with Arg base
`[Fig. 1(a)]. The large increase in Tg is, therefore, indepen-
`dent of the change in residual water content, indicating
`clearly that Arg base and Arg PO4 have quite different
`abilities as glass-formers. Presumably, the higher water
`solubility of a salt compared with the base reduces nucle-
`ation tendencies during freezing. Indeed, salt forms of
`glycine were also found to crystallize more slowly than
`the base (7).
`
`Vacuum-Drying of Amino Acids
`
`In contrast to the primary drying stage of freeze-dry-
`ing, water loss during vacuum-drying is by evaporation.
`The properties of the dried product (state and residual
`water content) are determined essentially by the drying
`pressure, temperature, and time (9), as well as the volume
`to be dried in a specific container (6). Carbohydrates were
`found to form amorphous structures after being vacuum-
`dried under the same conditions used here, although they
`remained as rubbers, owing to high residual water content
`(6). The amino acids that crystallized here during freeze-
`drying also crystallized during vacuum-drying. The re-
`sidual water contents for the two processes (Table 1) do
`not differ within the experimental error expected using
`Karl Fischer titration of these small amounts of water
`(2% w/w residual water < 1 mg water). The vacuum-
`drying conditions fortuitously were good enough to give
`the same results as the freeze-drying conditions. In each
`case the x-ray diffractogram was identical to that of the
`fully crystalline amino acid, as with freeze-drying, and
`no Tg or Tc was discernable on the DSC scan (result not
`shown). Differences between freeze- and vacuum-drying
`only emerge with the four basic amino acids, which
`
`-201-
`
`
`
`
`
`
`202
`
`Mattern et al.
`
`Table 1
`
`Results of Freeze-Drying or Vacuum-Drying Aqueous Amino Acid Solutions
`
`After Freeze-Drying
`
`After Vacuum-Drying
`
`Residual Water
`Content (% w/w)
`
`1.1 6 0.1
`0.9 6 0.1
`0.7 6 0.2
`1.3 6 0.2
`0.8 6 0.3
`0.6 6 0.3
`1.1 6 0.2
`7.8 6 0.1
`0.9 6 0.1
`1.3 6 0.2
`2.8 6 0.1
`1.8 6 0.1
`0.7 6 0.1
`1.9 6 0.1
`0.9 6 0.2
`0.8 6 0.2
`
`DSC Behavior
`
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Tg 5 68 6 2.1
`Tg 5 42 6 2
`Tg 5 37 6 5.6
`Tg 5 64 6 0.7
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`
`Residual Water
`Content (% w/w)
`
`DSC Behavior
`
`0.8 6 0.2
`0.8 6 0.2
`0.6 6 0.2
`1.6 6 0.3
`1.1 6 0.3
`0.4 60.1
`0.5 6 0.2
`2.6 6 0.8
`0.8 6 0.2
`0.5 6 0.1
`0.8 6 0.2
`4.9 6 0.1
`1.9 6 0.4
`1.5 6 0.1
`1.6 6 0.4
`3.0 6 2.6
`
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`Crystalline
`
`Amino Acid
`
`Glycine
`(cid:108)-Alanine
`(cid:108)-Valine
`(cid:108)-Leucine
`(cid:108)-Isoleucine
`(cid:108)-Serine
`(cid:108)-Threonine
`(cid:108)-Cysteine
`(cid:108)-Lysine
`(cid:108)-Arginine
`(cid:108)-Histidine
`(cid:108)-Citrulline
`(cid:108)-Proline
`(cid:108)-Phenylalanine
`(cid:108)-Methionine
`g -Aminobutyric acid
`
`In all cases a 0.24 M solution was used, with the exception of (cid:108)-histidine, (cid:108)-isoleucine, (cid:108)-leucine, and (cid:108)-methionine, which were prepared as 0.12
`M solutions because of their low solubility. During freeze-drying at a shelf temperature of 225(cid:176) C the (cid:108)-arginine and (cid:108)-lysine collapsed during
`primary drying, and were experiments repeated at 260(cid:176) C.
`
`formed freeze-dried amorphous solids, but fully crystal-
`line products after they were vacuum-dried (Table 1).
`The residual water contents, therefore, tend to be lower
`than those observed after the material is freeze-dried.
`Only (cid:108)-citrulline showed a higher residual water content
`in the crystalline state. The lower rate of water removal
`from the amino acid solution during vacuum-drying al-
`lowed sufficient time for complete nucleation and crystal
`growth.
`The addition of either HCl, H3PO4, or H2SO4 to the
`basic amino acid solutions to form the respective salts
`causes crystallization to be completely suppressed during
`vacuum-drying. The results, given in Table 2 for the ex-
`ample of (cid:108)-arginine, show fully amorphous solids that are
`rubbers at room temperature, owing to their high residual
`water contents. The x-ray diffractograms (not shown) are
`identical to that of the fully amorphous freeze-dried Arg
`PO4 shown in Fig. 1(b). As seen with freeze-drying, salt-
`formation reduces the tendency to nucleation; in this
`case, in the supersaturating amino acid solution during
`evaporative water loss. This effect is, however, salt spe-
`cific; neither nitrate nor acetate counter ions inhibited
`crystallization during vacuum-drying and the dried prod-
`
`ucts were fully crystalline with high residual water con-
`tents (Table 2).
`
`Vacuum-Drying of Phe/Arg Base Mixtures
`
`The addition of HCl, H2SO4, or H3PO4 inhibits crystal-
`lization of Arg during vacuum-drying. The dried prod-
`ucts were, however, all still in the rubbery state at room
`temperature, and as such, would certainly be unsuitable
`for stabilizing peptides or proteins (2). The reason for the
`rubbery products is as follows. Under the vacuum-drying
`conditions used here (0.1 mbar, 20(cid:176) C, 1 ml fill volume
`in 2 ml vial), the time required for the Arg/acid solution
`to reach the glass transition exceeds the 24-hr drying
`time. Of those factors that determine the time to the glass
`transition during vacuum-drying [temperature, pressure,
`nature and concentration of glass former, gradient of Tg
`(w), and surface area of evaporation (10)], it is the last-
`named that is responsible for this poor vacuum-drying
`behavior. As already found with carbohydrates, the small
`surface-to-volume ratio of the solution contained in the
`vial prevents sufficient water being lost after 24 hr at
`20(cid:176) C and 0.1 mbar to achieve the glassy state (6).
`
`-202-
`
`
`
`
`
`
`Formulation of Proteins in Vacuum-Dried Glasses. II
`
`203
`
`A combination of the Arg base with Phe, both of
`which form fully crystalline products when vacuum-dried
`separately without acid (see Table 1), can, however,
`greatly reduce residual water content. Figure 2 shows the
`residual water content (wg) and Tg of the dried Phe/Arg
`mixtures as a function of increasing molar proportion of
`Phe in the Arg. Recall that vacuum-dried Arg base is fully
`crystalline with a residual water content of ,1% (Table
`1). The addition of 1 mole-part Phe to 7 mole-parts Arg
`yields a clearly discernable glass transition at approxi-
`mately 2(cid:176) C (Fig. 2), indicating a change away from fully
`crystalline toward amorphous character. This is con-
`firmed by the x-ray diffractogram for this mixture in Fig.
`3(a), where crystalline peaks, identifiable from their an-
`gular positions as Arg and not Phe, lie superimposed on
`an amorphous halo. This small proportion of added Phe
`has, therefore, partially suppressed the crystallization of
`the Arg during vacuum-drying. Consequently, wg jumps
`from ,1% for pure fully crystalline Arg to approxi-
`mately 12% for this partially amorphous product (Fig.
`2), which as explained above cannot then be effectively
`vacuum-dried under the conditions used. Further increase
`in the mole fraction of Phe in the mixture produces, how-
`ever, remarkable changes in wg and Tg. Figure 2 shows
`how wg then decreases, at first accompanied by only a
`marginal increase in Tg. Thus, 1 mole-part Phe to 5 mole-
`parts Arg is still a rubbery solid at room temperature (Tg
`5 2(cid:176) C), but is fully amorphous [Fig. 3(b)]. The Phe com-
`pletely suppresses crystallization of the Arg at and above
`this mole fraction. With 1 mole-part Phe to 2 mole-parts
`Arg, however, new crystallization peaks emerge from the
`amorphous halo [Fig. 3(c)], identifiable as being those of
`crystalline Phe. This incipient crystallization of Phe
`within the amorphous Arg leads to a change in behavior
`of the vacuum-dried Phe/Arg product. Thus, further in-
`
`Figure 1. Wide-angle x-ray diffractograms of freeze-dried
`amino acids taken at 25(cid:176) C. (a) (cid:108)-Arg; (b) (cid:108)-Arg (0.24 M) 1
`H3PO4 (0.12 M).
`
`Table 2
`
`Influence of Counter Ions on the Freeze-Drying or Vacuum-Drying Behavior of (cid:108)-Arginine
`
`After Freeze-Drying
`
`After Vacuum-Drying
`
`Arginine/Counter Ion
`
`(cid:108)-Arg base (0.24 M)
`(cid:108)-Arg (0.24 M) 1 HCl (0.24 M)
`(cid:108)-Arg (0.24 M) 1 H3PO4 (0.12 M)
`(cid:108)-Arg (0.24 M) 1 H2SO4 (0.12 M)
`(cid:108)-Arg (0.24 M) 1 HNO3 (0.24 M)
`(cid:108)-Arg (0.24 M) 1 Ch3COOH (0.24 M)
`
`Drying conditions were as in Table 1, n 5 3.
`
`Residual Water
`Content (% w/w)
`
`1.3 6 0.2
`3.5 6 0.18
`2.2 6 0.07
`
`Tg
`((cid:176) C)
`
`42 6 2
`18 6 0.2
`93 6 1
`
`Residual Water
`Content (% w/w)
`
`0.5 6 0.1
`6.5 6 0.1
`3.3 6 0.2
`3.2 6 0.3
`2.7 6 0.1
`11.1 6 0.8
`
`Tg
`((cid:176) C)
`
`Crystalline
`3.5 6 0.3
`5.2 6 0.6
`6.7 6 0.3
`Crystalline
`Crystalline
`
`-203-
`
`
`
`
`
`
`204
`
`Mattern et al.
`
`Figure 2. Residual water content (s) and glass transition temperature (n) of vacuum-dried mixtures of Phe and Arg. In all cases,
`a 0.24 M solution of amino acid in water was vacuum-dried for 24 hr at 20(cid:176) C and down to 0.1 mbar. Fill volume was 1 ml in a
`2-ml vial.
`
`Figure 3. Wide-angle x-ray diffractograms of vacuum-dried Phe/Arg mixtures taken at 25(cid:176) C. (a) Phe/Arg (1:7); (b) Phe/Arg (1:
`5); (c) Phe/Arg (1:2); (d) Phe/Arg (1:1). The drying conditions are given in Fig. 2.
`
`-204-
`
`
`
`
`
`
`Formulation of Proteins in Vacuum-Dried Glasses. II
`
`205
`
`crease up to equimolar parts (1:1) Phe and Arg produces
`a precipitous fall in wg from 10% to approximately 2%,
`with a corresponding jump in Tg from 2(cid:176) C up to 60(cid:176) C
`(Fig. 2). Simultaneously, the degree of Phe crystallinity
`of the system increases [Fig. 3(d)]. Further increase in
`the molar proportion of Phe tends to increase wg slightly,
`accompanied by a slight decrease in Tg (Fig. 2). The crys-
`talline peaks in the x-ray diffractogram become larger
`and the pattern resembles more that of pure Phe (result
`not shown).
`It is already known that Phe can form a crystalline
`network within vacuum-dried carbohydrates (6). The
`same occurs with the Phe/Arg mixtures. In this case,
`however, the crystallization of Arg is suppressed by the
`addition of small amounts of Phe, which in increasing
`concentrations, itself crystallizes out within the amor-
`phous Arg. This is clearly seen in the SEM of the vac-
`uum-dried Phe/Arg (2:1) mixture shown in Fig. 4. At
`low magnification only a monolithic slab of dried product
`is evident, some 20 m m thick. Higher magnification re-
`veals that the cake is, however, highly porous and is com-
`posed of a network of crystallized Phe, which is coated
`with a thin pellicle of the amorphous Arg. The small layer
`thickness of the Arg pellicle (,2 m m) and its large sur-
`face-to-volume ratio result in sufficient loss of water un-
`der the vacuum-drying conditions used to reach and ex-
`ceed the glass transition after 24 hr of drying time. The
`result is a mixed amorphous/crystalline product of low
`wg and high Tg. The latter clearly characterizes unambig-
`uously the glassy state of the Arg. More care is needed
`when considering wg, however, because the moisture may
`not be homogeneously dispersed in the mixed amor-
`phous/crystalline products. The state of these systems
`will be expected to be only metastable. The presence of
`a counter ion to the arginine base is still important. HCl
`has no effect (Table 3), but with H3PO4, Tg is increased
`to .110(cid:176) C without change in the residual water content
`of approximately 1%. The particular ability of H3PO4 to
`increase the Tg of the basic amino acids is currently the
`subject of further, more detailed experiments.
`
`Stability of LDH in Phe/Arg/H3PO4
`
`Because the binary amino acid mixtures are partly
`amorphous and partly crystalline, it is uncertain if a sub-
`stantial protein lyoprotection can be expected. The initial
`loss in LDH activity occurring during the vacuum-drying
`process (process stability) of the unprotected bulk en-
`zyme is approximately 40% (Fig. 5). This is reduced to
`approximately 20% loss with the Phe/Arg/H3PO4 (1:1:
`1) formulation (Fig. 5), showing that the binary amino
`
`Figure 4. SEMs of vacuum-dried Phe/Arg (2:1). (a) Magni-
`fication 5003; (b) magnification 50003. The vacuum-drying
`conditions are given in Fig. 2. At this high molar proportion of
`Phe, the crystallization of the Phe is clearly evident.
`
`-205-
`
`
`
`
`
`
`206
`
`Mattern et al.
`
`Vacuum-Drying of Phe/Arg Mixtures at Molar Proportion of (1:1) with Mineral Acids
`
`Table 3
`
`Formulation
`
`Arg (0.24 M) 1 Phe (0.24 M)
`Arg (0.24 M) 1 Phe (0.24 M) 1 HCl (0.24 M)
`Arg (0.24 M) 1 Phe (0.24 M) 1 H3PO4 (0.12 M)
`
`Drying conditions were as in Table 1, n 5 3.
`
`Residual Water
`Content (% w/w)
`
`1.1 6 0.1
`1.0 6 0.1
`1.1 6 0.2
`
`Tg ((cid:176) C)
`63 6 0.4
`70 6 1.2
`113 6 3
`
`acid formulation can partly protect the enzyme during
`vacuum-drying at 20(cid:176) C. Franks reported that LDH could
`be vacuum-dried with a sucrose copolymer without any
`loss of activity (10). The higher temperature used for vac-
`uum-drying (up to 40(cid:176) C) and the smaller volumes used
`(250 m
`l) allowed, however, more rapid attainment of the
`glass transition than under the conditions used here, and
`hence less activity loss was observed during passage
`through the rubbery state. The unprotected LDH also
`shows rapid, temperature-dependent loss of activity dur-
`ing storage (Fig. 5).
`The storage stability of LDH in the binary amino acid
`mixture under normal storage conditions (3(cid:176) C) as well
`as at elevated temperatures (40 and 60(cid:176) C) was examined.
`Clearly, the modes or rates of degradation (in this case,
`aggregation of the LDH) may be quite different at these
`higher temperatures. Nonetheless, we were interested in
`the ability of the formulations to protect proteins even
`under stress conditions. The rate of activity loss of LDH
`is lower in the Phe/Arg/H3PO4 (1:1:1) than with the
`bulk enzyme, although activity still decreases by some
`20% over 26 weeks for the example of 40(cid:176) C. The loss
`in enzymatic activity is temperature-dependent in the
`range 3–60(cid:176) C examined here. Thus, despite a Tg of
`.110(cid:176) C for this binary amino acid formulation (Table
`3), the LDH shows substantial loss of activity on storage
`at temperatures more than 50(cid:176) C below Tg. At the Tg of
`vacuum-dried material, a transition in macroscopic prop-
`erties occurs (it becomes rubbery), as well as a change to
`greater molecular mobility. The latter has been associated
`with a shift in degradation kinetics from a slower model
`(Arrhenius) to a faster model (Williams–Landel–Ferry
`[WLF]) (11,12). At temperatures below Tg, therefore,
`degradation rates will be a function of temperature and
`should follow the slower kinetic model. This temperature
`dependence is, however, made more complicated by an
`increase in residual water content of the formulation on
`storage. After 26 weeks of storage the water contents in-
`
`creased from initially 1.1% (see Table 3) to 1.7% at 3(cid:176) C,
`2.9% at 40(cid:176) C, and 3.2% at 60(cid:176) C storage temperatures.
`Correspondingly, Tg decreased from initially 113 to 92(cid:176) C
`at 3(cid:176) C, 80(cid:176) C at 40(cid:176) C, and 78(cid:176) C at 60(cid:176) C storage tempera-
`tures. These changes are a result of sorbed moisture re-
`leased from the stoppers of the vials, which is taken up
`by the dried product. Clearly, the difference (Tg 2 T) is
`reduced on storage, but always remains positive. Because
`the conditions in the vials were continually changing dur-
`ing storage, a closer analysis of the kinetics and its rela-
`tion to (Tg 2 T) is prohibited. The storage stability of
`LDH in the sugar-free amino acid formulation is, how-
`ever, quantitatively identical to that in a vacuum-dried
`maltose/Phe (5:1) formulation (6).
`
`Figure 5. Process and storage stability of LDH, determined
`by enzymatic activity assay: bulk LDH, vacuum-dried, stored
`at 3(cid:176) C (.), 21(cid:176) C (r), and 40(cid:176) C (1); Phe/Arg/H3PO4 (1:1:
`0.5) LDH formulation, vacuum-dried, stored at 3(cid:176) C (n), 40(cid:176) C
`(s), and 60(cid:176) C (n).
`
`-206-
`
`
`
`
`
`
`Formulation of Proteins in Vacuum-Dried Glasses. II
`
`207
`
`Stability of rhG-CSF in Phe/Arg HCl
`
`To obtain more information about the effect of mois-
`ture uptake on protein stability in the region of the formu-
`lation’s Tg, we vacuum-dried rhG-CSF in the Phe/Arg/
`HCl (1:1:1) mixture. This mixture has a lower Tg than
`that with H3PO4 of just 70(cid:176) C at 1% residual water (Table
`3). Unprotected rhG-CSF bulk shows 99.6% intact mono-
`mer immediately after the vacuum-drying process. This
`value is maintained with the Phe/Arg/HCl (1:1:1) for-
`mulation [Fig. 6(a)]. Subsequent storage stability is
`strongly dependent on temperature. Although no more
`than 20% monomer is lost after 39 weeks of storage at
`up to 40(cid:176) C, at 60(cid:176) C and above more than 80% loss in
`
`monomer is evident during the first 12 weeks. As men-
`tioned above, the modes and rates of degradation may be
`quite changed at these high stress temperatures. As seen
`with LDH in the Phe/Arg/H3PO4 glass, this protein insta-
`bility at storage temperatures below Tg is complicated by
`uptake of moisture into the glass from the vial stopper.
`The already low Tg of the Phe/Arg/HCl glass of 70(cid:176) C
`causes, however, more dramatic temperature-dependent
`changes in the state of the glass than seen with Phe/Arg/
`H3PO4. Thus, after 39 weeks of storage at 3(cid:176) C the initial
`residual water content of 1.0% (Table 3) increases to
`2.4%; although Tg consequently decreases to 66(cid:176) C, there
`is no loss of rhG-CSF monomer up to 39 weeks [Fig.
`6(a)]. The higher the storage temperature, the greater is
`
`Figure 6.
`(a) Process and storage stability of rhG-CSF in Phe/Arg/HCl (1:1:1), determined by size exclusion chromatography,
`after vacuum-drying and subsequent storage at 3(cid:176) C (h), 21(cid:176) C (d), 30(cid:176) C (n), 40(cid:176) C (.), 60(cid:176) C (r), and 80(cid:176) C (1). (b) Comparison
`of stabilizing action of Phe/Arg plus HCl or H3PO4 formulations at 3 and 80(cid:176) C.
`
`-207-
`
`
`
`
`
`
`208
`
`Mattern et al.
`
`the increase in water content during storage, until at 40(cid:176) C
`we find 3.9% water, a Tg of only 40(cid:176) C, and a loss of 20%
`rhG-CSF over 39 weeks. It is, therefore, not possible to
`distinguish between the destabilizing effects of higher
`storage temperature and accelerated moisture uptake.
`The x-ray diffractograms of all systems stored at up
`to 40(cid:176) C remained unchanged. At the higher storage tem-
`peratures (60 and 80(cid:176) C) the behavior is, however, quite
`different; after 39 weeks of storage, both systems had
`changed to fully crystalline (determined by x-ray diffrac-
`tion) and no Tg was detectable using DSC, although the
`water content was unchanged. This amorphous to crystal-
`line change is a typical phenomenon (see ref. 13 for su-
`crose and lactose) and runs parallel to the drastic decrease
`in protein stability between 40 and 60(cid:176) C in Fig. 6(a). Wa-
`ter uptake depresses Tg below the storage temperature,
`initiating full crystallization of the amino acids. The fully
`crystalline formulation loses its ability to stabilize pro-
`teins, as evident in Fig. 6(a).
`The general importance of a high Tg for storage stabil-
`ity of the rhG-CSF is clearly illustrated by the compari-
`son of the Phe/Arg/HCl and Phe/Arg/H3PO4 formula-
`tions in Fig. 6(b). Recall that both formulations have
`approximately 1% residual water directly after being vac-
`uum-dried, but greatly different Tg values of 70 and
`113(cid:176) C, respectively. With both formulations there is no
`loss of rhG-CSF monomer after 39 weeks of storage at
`3(cid:176) C. At 80(cid:176) C, however, rhG-CSF aggregation is much
`less in Phe/Arg/H3PO4 than in Phe/Arg/HCl, the latter
`being completely inactivated after just 4 weeks. Although
`protein stability at such high temperatures cannot be com-
`pared with that at ambient temperatures, the improved
`stabilizing effect of the PO4 system is clearly evident.
`
`CONCLUSIONS
`
`This study shows that most amino acids crystallize
`during freeze-drying, and therefore are in principle suit-
`able as bulking agents. It is possible, however, to obtain
`amorphous products by either freeze- or vacuum-drying
`the phosphate or chloride salts of the basic amino acids.
`With freeze-drying, optimization of the process condi-
`tions is necessary to reduce the somewhat high residual
`moisture contents obtained here. This procedure appears
`to be particularly promising for phosphate salts, which
`give glasses of notably high Tg. The vacuum-drying pro-
`cedure could only be improved by using a binary amino
`acid mixture, yielding products of high Tg and low wg.
`Because the products are mixed amorphous/crystalline
`
`structures, the question of their metastability needs to be
`considered. Despite this, these sugar-free amino acid for-
`mulations act as lyoprotectants for both LDH and rhG-
`CSF during storage. Despite the low pressure used during
`secondary drying (0.01 mbar), the stoppers retained re-
`sidu