Effect of pH on the rate of asparagine deamidation in polymeric formulations: pH -rate profile

Article abstract of DOI:10.1002/1520-6017(200102)90:2<141::AID-JPS5>3.0.CO;2-Y

The rate of Asn deamidation of a model hexapeptide (L-Val-L-Tyr-L-Pro-L-Asn-Gly-L-Ala) was measured as a function of effective pH ('pH') in glassy and rubbery polymeric solids containing poly(vinyl pyrrolidone) (PVP) and in solution controls at 70°C. The reaction exhibited pseudo-first-order kinetics in all samples over a wide 'pH' range (0.5 <'pH'<12); the formation of similar products suggests that the reaction mechanism is unaffected by matrix type. Rates of deamidation were comparable for the polymeric and solution samples in the acidic range ('pH'<4). Solution-state rates were faster than those in polymeric solids at neutral 'pH' (6 <'pH'<8), increasing to a >10,000-fold difference in the basic range ('pH'>8). Specific base catalysis was observed in solution and in the polymeric solids under neutral conditions (6 <'pH'<8). In solution, the reaction exhibited general base catalysis for 'pH'>8, whereas the reaction was 'pH'-independent in the polymeric solids in this range. The 'pH' - rate profile and supporting buffer catalysis data are consistent with a change in the rate-determining step in the basic range from 'pH'-dependent attack of the deprotonated backbone amide nitrogen on the Asn side chain in solution to 'pH'-independent ammonia expulsion in the polymeric solids. The results suggest that polymer matrix incorporation not only affects the magnitude of the deamidation rate constant but also the 'pH' dependency of the reaction and the rate-determining step in the basic 'pH' range.

Full text of DOI:10.1002/1520-6017(200102)90:2<141::AID-JPS5>3.0.CO;2-Y

Effect of `pH' on the Rate of Asparagine Deamidation
in Polymeric Formulations: `pH'±Rate Pro®le
YUAN SONG, RICHARD L. SCHOWEN, RONALD T. BORCHARDT, ELIZABETH M. TOPP
Department of Pharmaceutical Chemistry, The University of Kansas, Lawrence, Kansas 66047
Received 3 July 2000; revised 28 August 2000; accepted 29 August 2000
ABSTRACT: The rate of Asn deamidation of a model hexapeptide (L-Val-L-Tyr-L-Pro-L-
Asn-Gly-L-Ala) was measured as a function of effective pH (`pH') in glassy and rubbery
polymeric solids containing poly(vinyl pyrrolidone) (PVP) and in solution controls at
ꢀ
7
0 C. The reaction exhibited pseudo-®rst-order kinetics in all samples over a wide `pH'
range (0.5<`pH'<12); the formation of similar products suggests that the reaction
mechanism is unaffected by matrix type. Rates of deamidation were comparable for the
polymeric and solution samples in the acidic range (`pH'<4). Solution-state rates were
faster than those in polymeric solids at neutral `pH' (6<`pH'<8), increasing to a
>
10,000-fold difference in the basic range (`pH' > 8). Speci®c base catalysis was
observed in solution and in the polymeric solids under neutral conditions (6<`pH'<8).
In solution, the reaction exhibited general base catalysis for `pH' > 8, whereas the
reaction was `pH'-independent in the polymeric solids in this range. The `pH'±rate
pro®le and supporting buffer catalysis data are consistent with a change in the rate-
determining step in the basic range from `pH'-dependent attack of the deprotonated
backbone amide nitrogen on the Asn side chain in solution to `pH'-independent ammonia
expulsion in the polymeric solids. The results suggest that polymer matrix incorporation
not only affects the magnitude of the deamidation rate constant but also the `pH'
dependency of the reaction and the rate-determining step in the basic `pH' range. ß 2001
Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 90:141±156, 2001
Keywords: peptide stability; asparagine deamidation; solid state; polymer; poly(vinyl
pyrrolidone); effective `pH'
INTRODUCTION
the rational formulation of these promising new
drugs.
A number of physicochemical factors have been
Therapeutic peptides and proteins are often
formulated in the solid state to prolong shelf life
or to provide a platform for controlled release.
Although many protein degradation reactions are
appreciably slower in the solid state than in solu-
tion, reactivity is often suf®cient to compromise
drug potency. An understanding of the mechan-
isms of peptide and protein degradation reactions
in the solid state and of the physical and chemical
factors affecting their rate is therefore central to
shown to in¯uence protein degradation reactions
in solids. These include temperature, pressure,
the mobility of reactant or product species in the
solid matrix, moisture content, the concentrations
1,2
of various excipients, and the effective pH. The
studies presented here address the role of effec-
tive pH in determining the rate and mechanism of
asparagine (Asn) deamidation for a model peptide
in solid polymeric matrices. Because this reaction
3
±6
is known to be sensitive to pH in solution, it is
reasonable to expect a dependence on hydrogen
ion activity in the solid state as well.
Correspondence to: E.M. Topp (Telephone: 785-864-3644;
Fax: 785-864-5736; E-mail: topp@ukans.edu)
Assessing the role of effective pH in solid-state
reactions is complicated by the fact that pH
Journal of Pharmaceutical Sciences, Vol. 90, 141±156 (2001)
ß 2001 Wiley-Liss, Inc. and the American Pharmaceutical Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001
141

142
SONG ET AL.
technically is de®ned only in aqueous solution.
Although there have been numerous attempts to
provide a comparable measure of hydrogen ion
activity in solids, to date there is no single,
generally accepted method. Methods that have
been proposed for the measurement of hydrogen
ion activity in solids include electron paramag-
The studies reported here address the effect
of `pH' on the rate of Asn-deamidation in the
hexapeptide
L-Val-L-Tyr-L-Pro-L-Asn-Gly-L-Ala
incorporated into lyophilized powders containing
poly(vinyl pyrrolidone) (PVP). This peptide was
selected because deamidation is its sole degrada-
tion reaction both in solution and in the solid
7
4,15
netic resonance (EPR), confocal ¯uorescence
state,
and because the kinetics and mechan-
8
9
microscopy, Fourier transform infrared (FTIR)
isms of the reaction have been thoroughly char-
10
4
or diffuse re¯ectance FTIR spectroscopy, and
nuclear magnetic resonance (NMR) spectros-
copy. These methods generally employ a pH-
sensitive probe molecule (e.g., a pH-sensitive spin
probe or ¯uorescent dye ) or isotopically labeled
pH-sensitive functional group (e.g.,
acterized in solution, as shown in Figure 1. In
addition, the absence of higher-order structure
common to larger proteins permits a focus on
chemical rather than physical degradation. PVP,
a linear vinyl polymer with a single pendant
functional group, has been selected as a polymeric
excipient and serves as a relatively inert amor-
phous matrix in which the peptide is embedded.
Previous studies in our laboratories have estab-
lished that lyophilized PVP can be plasticized
with glycerol and/or sorbed water so that both
11
7
8
15
N-histi-
11
dine ). An alternative approach adopted in many
stability studies de®nes the ``effective pH'' of the
solid as the pH measured in aqueous solution
prior to lyophilization or following reconstitution
1
2±14
of the solid.
The effective pH measured in
this way is often termed the `pH', where the
quotation marks indicate an effective value.
glassy and rubbery solids can be produced in the
12
19,20
temperature range of interest.
The `pH'
The recent literature provides several exam-
ples of `pH'-dependent solid-state degradation
reactions of peptides and proteins. Oliyai et al.
have demonstrated that the rate of deamidation of
lyophilized formulations of the hexapeptide L-Val-
L-Tyr-L-Pro-L-Asn-Gly-L-Ala increases with `pH' in
dependence of the deamidation reaction in glassy
and rubbery solids and in a PVP-containing
solution control are examined here. The results
demonstrate that the `pH'±rate pro®le is signi®-
cantly altered in the polymeric media relative to
the solution control and provide evidence for a
matrix effect on the rate-determining step in basic
media.
15
the range 5±8. This group also observed `pH'-
dependent isomerization of the lyophilized Asp-
containing analog L-Val-L-Tyr-L-Pro-L-Asp-Gly-L-
14
Ala.
Several groups have reported a `pH'
dependence for the degradation of lyophilized
insulin, with degradation reactions involving
covalent dimerization, cyclic anhydride forma-
MATERIALS AND METHODS
Materials
12,13
16
tion, and aggregation.
Pikal et al. have
reported `pH'-dependent aggregate formation in
lyophilized human growth hormone, and Town-
L-Val-L-Tyr-L-Pro-L-Asn-Gly-L-Ala (Asn-hexapep-
tide) was synthesized by Dr. Madhup Dhaon
(Abbott Laboratories, North Chicago, IL) and
17
send and DeLuca have observed a nonlinear
pH'-dependent loss of RNase activity. These
`
L-Val-L-Tyr-L-Pro-L-Asp-Gly-L-Ala
(Asp-hexa-
studies establish the importance of `pH' in a
variety of solid-state protein degradation reac-
tions, but are limited in that the number
of `pH' values studied is often small and in a
fairly narrow range. This limitation prevents
effective analysis of the `pH'±rate pro®le, which
in solution-state reactions has provided valu-
peptide) was synthesized by Gemini Biotech
(Woodlands, TX). The cyclic-imide-containing
hexapeptide (Asu-hexapeptide) was isolated and
puri®ed from degraded Asn-hexapeptide by pre-
parative high-performance liquid chromatogra-
phy (HPLC). Poly(vinyl pyrrolidone) (PVP), under
the trade name Kollidon K17 (MW ˆ 10, 000), was
purchased from BASF Corporation (Parsippany,
NJ). Glycerol was purchased from Aldrich Che-
mical Company, Inc. (Milwaukee, WI). Tri¯uor-
oacetic acid (TFA) was purchased from Pierce
(Rockford, IL). The salts and organic solvents
used to make buffer solution and mobile phase,
respectively, were obtained from Fisher Scienti®c
4,18
able information on reaction mechanisms.
Similarly, although the use of large therapeutic
proteins is of practical interest, these complex
molecules are often subject to multiple simulta-
neous degradation reactions, which again compli-
cates the mechanistic interpretation of `pH'
effects.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
143
Figure 1. Pathways of L-Val-L-Tyr-L-Pro-L-Asn-Gly-L-Ala deamidation in aqueous
solution. At `pH' > 5, the reaction proceeds via the formation of a cyclic imide, which is
subsequently hydrolyzed to form the L-Asp and L-iso-Asp hexapeptides. Racemization to
the D-cyclic imide also occurs, which is hydrolyzed to the D-Asp and D-iso-Asp
hexapeptides (not shown). At `pH'<4, the L-Asp hexapeptide is formed from the L-Asn
hexapeptide by direct hydrolysis. The L-Asp hexapeptide (`pH'<4) is then subject to
cleavage of the Asp±Gly peptide bond, producing a tetrapeptide (L-Val-L-Tyr-L-Pro-L-
Asp) and a dipeptide (Gly-L-Ala). Adapted from Ref. 4.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

144
SONG ET AL.
(
Fair Lawn, NJ). All chemicals were analytical
Therefore, this buffer was also judged to be
unsuitable.
grade or better and used as received. All water
used throughout the study was deionized with a
Millipore MILLI-Q water system.
As a result of these screening studies, the
following salts were selected to maintain `pH' in
various regions of interest. In the very acidic
region (`pH' 0.5±1.5), H SO was used. Monobasic
TM
2
4
PVP Puri®cation
and dibasic potassium phosphates (KH PO /
2 4
K HPO ) in varying ratios were used to maintain
2
4
Prior to its use in formulations, PVP was
dissolved in water and dialyzed for 48 h using
Spectra/Pro molecular weight cut-off (MWCO)
cellulose membranes (MWCO ˆ 3500 Da; Spec-
trum Medical Industries, Houston, TX) to remove
low MW impurities that may interact with the
model hexapeptide. After dialysis, the polymer
was freeze-dried for 3 days with a VirTis Freeze-
mobile 5SL (Gardner, NY).
`
pH' in the slightly acidic to neutral region (`pH'
.0±7.0). Similarly, in the slightly basic region
`pH' 7.0±8.0), dibasic and tribasic potassium
phosphates (K HPO /K PO ) were employed.
3
(
2
4
3
4
For `pH' values between 9.0 and 10.5, mixtures
of monobasic and dibasic sodium carbonate salts
(
NaHCO /Na CO ) were used. Finally, for `pH'
3 2 3
values > 10.5, sodium hydroxide was added. With
the exceptions of samples at extreme `pH' (i.e.,
containing H SO or NaOH) and those used in
2
4
buffer catalysis studies, all buffer solutions were
prepared at a concentration of 0.15 M. The ionic
strength of all buffer solutions was maintained
constant at 0.5 by the addition of sodium chloride.
Buffer pH was measured with a pH-meter (Orion
model 410A) equipped with a Ross combination
electrode.
Preparation of Buffer Solutions
Since control of `pH' in the solid state is crucial in
a study of the effects of hydrogen ion activity on
rate, preliminary studies were conducted to
evaluate and select appropriate buffers. An
effective buffer for these accelerated stability
studies: (i) must be stable at the storage tempera-
ꢀ
ture (70 C); (ii) must not be subject to differential
Preparation of Formulations in
Solution and Solid State
precipitation of salt species during lyophilization,
as has been observed for sodium phosphate
21
buffers; (iii) must not contain volatile species
e.g., acetate, HCl), which may be lost during
Peptide stability was determined in glassy and
rubbery polymeric solids at various `pH' values,
and in polymer-containing solutions. To prepare
the formulations, a 5% (w/v) solution of dialyzed
PVP was ®rst prepared in one of the buffer
solutions described above. The solution was then
divided and glycerol added to half the solution
(30% w/v); glycerol was used as a plasticizer to
produce rubbery polymeric solids. The Asn-hex-
apeptide was then added to each solution to a
concentration of ꢁ0.5 mg/mL. After addition of
the peptide, aliquots of each solution were placed
(
lyophilization and/or storage at elevated tempera-
ture; (iv) should show minimal interference with
the HPLC assay for the peptide and its degra-
dation products; and (v) should not interact
with the peptide or polymer components of the
formulation.
Formulations were prepared using a variety of
salt species and evaluated for their ability to meet
the criteria just outlined. Preliminary studies
indicated that p-toluene sulfonic acid has a strong
ultraviolet (UV) absorbance at 214 nm and co-
elutes with the iso-Asp degradation product
during HPLC analysis, and so is unsuitable for
use in these studies. In a previous study, we
noted that Tris buffer [tris-(hydroxy methyl)
aminomethane] degrades at 70 C both in solution
and in PVP-containing solids to produce formal-
dehyde, which then forms a covalent adduct with
the Tyr residue in this peptide. Tris buffer was
omitted for this reason. Solution and solid
formulations prepared with trichloroacetic acid
showed a dramatic shift in `pH' during processing
and initial storage, from `pH' 3.1 to `pH' 10.
ꢀ
in a 70 C oven to serve as solution-state controls
in the stability studies. The remaining amounts of
the glycerol-containing and glycerol-free solutions
were placed into a series of 2-mL glass lyophiliza-
tion vials, with 100 mL in each vial. The samples
were then lyophilized using a VirTis BenchTop
lyophilizer (Gardner, NY). The following lyophili-
zation cycle was applied: (i) freezing at a shelf
ꢀ
2
2
ꢀ
temperature of � 35 C for 2 h; (ii) primary drying
ꢀ
at a shelf temperature of � 35 C for 10 h; and (iii)
secondary drying at shelf temperatures of � 15,
ꢀ
� 5, 5, 15, 20, and 25 C, for 10, 8, 6, 6, 12, and 10 h,
respectively.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
145
Accelerated Stability Studies
tide, L-Val-L-Tyr-L-Pro-L-Asp. Approximate elu-
tion times of the hexapeptides were iso-Asp-
hexapeptide, 8 min; D-Asp-hexapeptide, 10 min;
L-Asp-hexapeptide, 10.5 min; L-Asn-hexapeptide,
Accelerated stability studies were conducted in a
ꢀ
7
0 C oven. Solution-state samples were placed in
closed containers. Solid samples were placed in
open vials in a chamber containing a saturated
solution of NaBr to maintain a constant relative
13 min; and Asu-hexapeptide, 19 min. The
tetrapeptide was eluted at ꢁ6 min. The products
of some side reactions were also detected and re-
solved with this method. The limit of detection of
this assay is ꢁ0.1 mg/mL for the Asn-hexapeptide.
On repeated injection of samples containing
PVP, the retention times of the known compounds
tended to decrease, compromising the separation
of the iso-Asp- and Asp- hexapeptide degradation
products. To regenerate the column, a high
organic content mobile phase (50:50 methanol:-
water) was ¯ushed through the column after
every 50 to 100 injections. Retention times and
peak resolution were indistinguishable from those
in a new C18 column following this procedure.
Previous studies employing this assay assumed
that the concentrations of the Asn-, Asp-, and iso-
Asp- hexapeptides were proportional to peak area,
a reasonable assumption given that the principal
contribution to UV absorbance at 214 nm is the
peptide bond. Because peptide recovery and mass
balance were of interest in the present studies, we
determined the UV sensitivities of known com-
pounds involved so that their concentrations
could be determined with greater accuracy.
Standard calibration curves were constructed by
injecting different concentrations of authentic
samples of the Asn-, Asp-, and Asu-hexapeptides
and of the tetrapeptide; the respective UV
absorbances were then determined. Because a
synthetic standard for the iso-Asp-hexapeptide
was not available, its theoretical sensitivity was
determined by mass balance. The results are
summarized in Table 1, and were used in
calculating peptide concentrations from chroma-
tographic peak areas.
23
humidity of ꢁ50%. Additional, supporting stu-
dies were performed at 70 and 95% RH, using
saturated NaCl and K SO solutions, respec-
2
4
23
tively. At appropriate time intervals, triplicate
samples of each solid formulation or a 100-mL
aliquot of the solution formulations were removed
from the oven for analysis. Because water sorp-
tion is very rapid, with 95% of the equilibrium
19
mass reached within 20 min, sampling intervals
of <20 min were considered impractical. All
samples were either diluted (solutions) or recon-
stituted (solids) using 1 mL of 0.05 M ammonium
acetate buffer solution (pH 4) because the Asn-
hexapeptide is maximally stable between pH 3
4
and 5. The samples were then analyzed by
HPLC-UV, using the method described later.
The pH values of the solutions and the reconsti-
tuted lyophilized solids were measured for each
sample to verify that pH did not change during
the studies. Stability studies were terminated
when the concentration of Asn-hexapeptide was
no longer detectable and the concentrations of the
major degradation products showed no further
change with time.
Sample Analysis
Analysis of the Asn-hexapeptide and its deamida-
tion products was performed by isocratic
reversed-phase HPLC with UV detection. Details
4
of the method have been reported previously,
although there were minor modi®cations. Brie¯y,
the method employed an Econosphere C18
reversed-phase column (4.6 Â 250 mm, 5-mm par-
ticle size) from Alltech Associates, Inc. (Deer®eld,
IL) for separation. The chromatographic system
consisted of a Shimadzu SCL-10Avp system
controller, a SPD-10Avp UV-vis detector, a SIL-
Kinetic Analysis
In liquid and solid formulations, the observed rate
constants (kobs) for the disappearance of the Asn-
hexapeptide were determined from ®rst-order
plots of the fraction of the initial Asn-hexapeptide
concentration remaining versus time, according
to eq. 1:
1
0ADvp autoinjector, and a LC-10ATvp pump.
The mobile phase contained 8% (v/v) acetonitrile
and 0.1% (v/v) tri¯uoroacetic acid in 0.050 M
ammonium acetate at pH 4.6. The ¯ow rate was
0
.8 mL/min and detection was at 214 nm. The
method provided for separation and quantitation
of the Asn-hexapeptide and its major deamidation
products, including iso-Asp, D-Asp, L-Asp, the
cyclic imide (Asu-hexapeptide), and the tetrapep-
A ˆ Aoexpꢂ� kobst†
ꢂ1†
where A is the amount of Asn-hexapeptide at time
t, and Ao is the initial peptide concentration.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

146
SONG ET AL.
Table 1. Assay Sensitivities for Asn-Hexapeptide and Its Major Degradation Products
Parameter
Asn
Asp
Tetra
Asu
iso-Aspa
Sensitivity Æ SD
7:16 Æ 0:13
7:03 Æ 0:11
5:94 Æ 0:09
8:70 Æ 0:20
5:62 Æ 0:10
peak area/(mg/mL) Â10 6)
�
(
a
Determined by calculation.
Previous studies have shown that the deamida-
tion of the Asn-hexapeptide exhibits pseudo-®rst-
order kinetics both in solution and in polymeric
tion > rubbery solid > glassy solid. This ordering
of reactivity was observed throughout the neutral
to basic `pH' range, as discussed in detail later.
4,19,20
solids.
Values of the observed rate constants
(
kobs) were determined by nonlinear regression
Products Formed
using the software package Origin (Microcal Soft-
ware, Inc., MA). In solution studies at pH ꢃ 9.5,
the reaction was too rapid to permit the frequent
sampling and temperature re-equilibration neces-
sary to determine kobs values by this method. For
these conditions, experiments were carried out at
room temperature (22 C) and k values at 70 C
were estimated using the reported Arrhenius
activation energy (Ea) of 22 kcal/mol for the
reaction in solution.
This section presents information on the products
formed during deamidation, with the intent of
establishing that the same reaction pathways are
operative in solution and in polymeric solids. A
complete comparative analysis of the kinetics
of product formation and of the corresponding
microscopic rate constants in solution and poly-
meric solids will be presented in a subsequent
report.
ꢀ
ꢀ
obs
4
In constructing the `pH'±rate pro®le, kobs
values were measured at several buffer concen-
trations and linearly extrapolated to zero buffer
concentration to obtain the reported ko value.
Exceptions to this procedure occurred in the very
acidic (`pH'<2) or very basic (`pH' > 10.9) regions
in which buffer catalysis studies could not be
Under acidic conditions (`pH'<3), the deamida-
tion products observed in the solution controls
were those previously observed in acidic solution,
4
with the Asp-hexapeptide as the major initial
degradation product, and the tetrapeptide (L-Val-
L-Tyr-L-Pro-L-Asp) dominating at later times.
Although the Asu- and iso-Asp-hexapeptides were
also detected in these samples, they were present
in smaller amounts, a ®nding consistent with
direct hydrolysis as the dominant pathway of
conducted because strong acids or bases (H SO4
2
or NaOH) were used to control `pH'. In these
cases, the reported k value is equal to the k
o
obs
3,4,6
value measured at a single H SO or NaOH
deamidation.
In polymeric solids at acidic `pH',
2
4
concentration.
the Asu-hexapeptide was the major initial degra-
dation product, whereas the tetrapeptide and
iso-Asp-hexapeptides were the major products
at later times. The Asp-hexapeptide was also
detected in the acidic polymeric solids throughout
the study, but in lesser amounts. These results
suggest that the dominant pathway for deamida-
tion at acidic `pH' may shift from direct hydrolysis
in solution to a combination of direct hydrolysis
and cyclic imide formation in polymeric media.
The ratio of the maximum concentration of Asu-
hexapeptide to that of the Asp-hexapeptide at
acidic `pH' is shown in Figure 3A; higher ratios in
the polymeric solids than in solution support the
increased importance of the cyclic imide pathway.
Possible alternative explanations for these obser-
vations include: (i) a difference between the ``true''
hydrogen ion activity in the polymeric solids and
the effective `pH' measured by reconstitution,
with the ``true'' hydrogen ion activity being more
RESULTS AND DISCUSSION
Deamidation Kinetics
Figure 2 shows representative kinetic curves for
the deamidation of the Asn-hexapeptide in aqu-
eous solution, and in rubbery and glassy PVP
ꢀ
solids at `pH' 10.4 and 70 C. The lines repre-
sent nonlinear regression ®ts of the data to the
pseudo-®rst-order model given in eq. 1. The data
demonstrate apparent-®rst-order kinetics for the
disappearance of the Asn-hexapeptide for all
three formulations under these experimental
conditions. Similar results were observed for all
formulations and `pH' values studied (data not
shown). The data also demonstrate that the
reaction rates (kobs) decrease in the order solu-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
147
Figure 2. Representative kinetic profiles for Asn-hexapeptide loss in aqueous solu-
tion (&, Panel A), rubbery PVP solids (~, Panel B) and glassy PVP solids (*, Panel C) at
ꢀ
`
pH' 10.4 and 70 C. Apparent-first-order rate constants (kobs) obtained by fitting the
data to eq. 1 are shown in each panel. The line in each panel represents this regression.
Error bars represent standard deviations; n ˆ 3.
nearly neutral, and (ii) formation of the Asu-
hexapeptide from the Asp-hexapeptide via back-
reaction (Figure 1).
At acidic `pH' in glassy polymeric solids, all the
chromatographic peaks observed corresponded to
known solution-state deamidation products of the
Asn-hexapeptide, suggesting that no new reaction
pathways contribute to deamidation in glassy
solids. In rubbery polymeric solids at acidic `pH',
two additional peaks were observed, suggesting
the formation of unknown products (data not
shown). The cumulative areas of these new peaks
corresponded to a maximum of 20±25% of the
initial peak area for the Asn-hexapeptide, and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

1
48
SONG ET AL.
samples. No correction for these unknown pro-
ducts was made.
At neutral to basic conditions (`pH' > 5), the
dominant degradation products were the iso-Asp-
and Asp-hexapeptides both in solution and in the
polymeric solids. The ®nal ratio of iso-Asp- to Asp-
hexapeptide was in the range of 3.0 to 5.0 for all
samples, as shown in Figure 3B, a value similar to
that previously reported for the reaction in
3,4,6
aqueous solution.
This result suggests that
deamidation at neutral to basic `pH' occurs via
formation of the Asu-hexapeptide in the poly-
meric solids, as previously established for the
3,4,6
reaction in solution.
No unknown chromato-
graphic peaks were detected in the polymeric
solids in this `pH' range, indicating the absence of
competing side reactions.
Because the degradation products detected in
the PVP-containing solution controls and in the
polymeric solids are comparable to those pre-
viously reported in polymer-free aqueous solu-
tions, it is reasonable to assume that the
dominant degradation mechanisms are similar
Figure 3. Product formation ratios for Asn-hexapep-
tide deamidation in solution (&), rubbery PVP solids
ꢀ
(
~), and glassy PVP solids (*) at 70 C. Panel A: Ratio
of the maximum concentrations of Asu- to Asp-hex-
apeptides at acidic `pH'. Panel B: Ratio of the ®nal
concentrations of iso-Asp- to Asp-hexapeptides at
neutral to basic `pH'. The lines in Panel A are included
to clarify trends, and do not indicate regression. Error
bars represent standard deviations; n ˆ 3.
Figure 4. `pH'±rate prole for Asn-hexapeptide loss in
solution (&), rubbery PVP solids (~), and glassy PVP
solids (*) at 70 C. Lines represent regression of the
data to eq. 2. Error bars represent standard deviations;
n ˆ 3.
were greatest toward the end of the stability
studies. It is suspected that these products are the
result of side reactions between the peptide and
the glycerol plasticizer used in the rubbery
ꢀ
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
149
Table 2. Rate Constants (ko) for Deamidation of the Asn-Hexapeptide in Solution and in Polymeric PVP Solidsa
ko ( Æ SD), h�
1
`
pH'
Buffer Salt
Solution
Glassy PVP Solid
Rubbery PVP Solid
1
2
2
3
3
3
3
3
2
1
1
0
0
0
1
1
2
4
5
5
6
7
7
9
9
.17
H
H
H SO4
H
H
2
SO
SO
4
4
1:34ꢂÆ 0:11†  10�
�
� 2
� 1
� 2
� 2
� 3
.65
.99
.45
.89
.97
.05
.08
.97
.6
.02
.96
.03
.5
2
3:24ꢂÆ 0:22†  10
2:04ꢂÆ 0:27†  10
8:51ꢂÆ 0:86†  10
2:51ꢂÆ 0:16†  10
8:13ꢂÆ 0:66†  10
3:80ꢂÆ 0:15†  10
3:24ꢂÆ 0:19†  10
8:91ꢂÆ 0:26†  10
9:77ꢂÆ 0:42†  10
1:28ꢂÆ 0:05†  10
1:74ꢂÆ 0:04†  10
2:82ꢂÆ 0:03†  10
1:62ꢂÆ 0:10†  10
1:48ꢂÆ 0:09†  10
1:26ꢂÆ 0:11†  10
1:48ꢂÆ 0:12†  10
1:78ꢂÆ 0:20†  10
1:66ꢂÆ 0:15†  10
1:41ꢂÆ 0:06†  10
2:40ꢂÆ 0:11†  10
�
�
�
�
�
�
�
�
� 3
� 3
� 4
� 4
� 4
� 4
� 4
� 3
� 3
� 3
1:38ꢂÆ 0:06†  10
7:76ꢂÆ 0:49†  10
2
2
SO
SO
4
4
6:92ꢂÆ 0:19†  10
2:13ꢂÆ 0:14†  10
2
2:18ꢂÆ 0:06†  10
7:24ꢂÆ 0:71†  10
KH PO /K HPO
1:29ꢂÆ 0:05†  10
2
2
2
2
4
4
4
4
2
2
2
2
4
4
4
4
4
KH
KH
KH
PO
PO
PO
/K
/K
/K
HPO
HPO
HPO
1:55ꢂÆ 0:10†  10
� 3
� 3
4:57ꢂÆ 0:41†  10
2:04ꢂÆ 0:08†  10
2:34ꢂÆ 0:12†  10
3:63ꢂÆ 0:10†  10
KH PO /K HPO
1:45ꢂÆ 0:12†  10
2
4
4
4
2
3
3
�
0
K
K
2
HPO
HPO
/K
/K
PO
PO
4
4
2:82ꢂÆ 0:15†  10
� 2
2
1:14ꢂÆ 0:06†  10
4:07ꢂÆ 0:59†  10
�
2
� 1
NaHCO /Na CO
NaHCO
NaHCO
2:88ꢂÆ 0:011†  10
3:24ꢂÆ 0:33†  10� 1
3
3
3
2
2
2
3
3
3
3
3
1
b
� 2
� 2
� 2
� 2
� 2
� 2
/Na
/Na
CO
CO
1:83ꢂÆ 0:072†  10
4:90ꢂÆ 0:087†  10
1
1
1
1
1
1
0.03
0.43
0.4
0.91
2.03
1
NaHCO /Na CO
5:25ꢂÆ 0:21†  10
3
2
1
1
2
b
� 1
� 1
� 1
NaHCO
NaOH
NaOH
3
/Na
2
CO
3:02ꢂÆ 0:22†  10
9:55ꢂÆ 0:62†  10
1:86ꢂÆ 0:20†  10
4:17ꢂÆ 1:21†  10
b
3:80ꢂÆ 0:18†  10
b
4:07ꢂÆ 0:42†  10
a
ꢀ
n ˆ 3; 70 C
b
ꢀ
Calculated from the kobs at 22 C and Ea ˆ 22:2 kcal/mol.
to those de®ned for the reaction in solution. In the
remaining sections of this paper, we analyze the
ni®cant buffer catalysis was observed. Several
features of the `pH'±rate pro®le are noteworthy.
Overall, the curve is U-shaped in solution and in
the polymeric solids, suggesting a degree of
catalysis by both acidic and basic species as
`pH'-dependence of the pseudo-®rst-order disap-
pearance of the Asn-hexapeptide adopting this
assumption.
4
reported previously for the reaction in solution.
The `pH' of maximum stability occurs between 2
and 5 for all samples, and is relatively unaffected
by the type of formulation.
In the acidic region (`pH'<2.0), ko values are
comparable in all three types of formulation. The
curves are linear, with slopes approximately
equal to � 1 (Table 3). This result is consistent
with speci®c acid catalysis, as reported previously
`
pH'±Rate Pro®le
The relationship between the deamidation rate
constants and `pH' is shown in Figure 4; the
corresponding rate constants are given in Table 2.
The rate constants are designated ``k '' to indicate
that the values have been extrapolated to zero
buffer concentration for samples in which sig-
o
Table 3. Approximate Values of the Slopes of the `pH'±rate Pro®le (Figure 4) in
Selected Regions
�
1
a
Slopes of the `pH'±Rate Pro®le, log (h ) Æ SE
Formulation Type
`pH' < 2.0
5 < `pH' < 8
`pH' > 9
Solution
Rubbery solid
Glassy solid
� 1:00 Æ 0:06
� 1:22 Æ 0:05
� 1:13 Æ 0:006
0:86 Æ 0:06
0:46 Æ 0:06
0:18 Æ 0:03
0:39 Æ 0:07
0:05 Æ 0:02
� 0:005 Æ 0:02
aCalculated by linear regression of the data in the `pH' region indicated.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

150
SONG ET AL.
4
for the reaction in aqueous solution. Note that
catalysis. Accordingly, nonlinear regression was
used to ®t the data in Figure 4 to the following
equation:
both the k values and the absolute values of the
o
slopes are slightly greater for the rubbery poly-
meric solids than for the other samples. These
minor differences may re¯ect the formation of
unknown products, as already noted.
‡
�
�
k ˆ k ‰H Š ‡ k ‡ k ‰OH Š=fK ‡ ‰OH Šg ꢂ2†
o
H
W
B
In the neutral region (5<`pH'<8), k values
where k , k , and k are the rate constants for
H W B
o
decrease in the order: solution > rubbery > glassy.
The values in solution are ꢁ3±10 times greater
than in the rubbery solids; values in the rubbery
and glassy solids differ by a similar proportion.
The `pH'±rate pro®le is approximately linear in
this region, with slopes that decrease in the order
the acid, water, and base catalysis of the reaction,
respectively, and K is a constant related to the
charge of the transition state in the rate-deter-
mining step. The curves in Figure 4 represent the
best ®t for each type of sample; the excellent
agreement between the data and the regression
curves suggests that the model adequately
describes the data for all three types of samples.
Values of the regression coef®cients are given in
Table 4. Particularly noteworthy is the > 3000-
fold variation in the value of kB in the three
observed for the k values (Table 3). The smaller
o
slopes in the polymeric solids suggest a decreased
sensitivity of the reaction rate to `pH' changes in
these samples relative to the solution control.
Whereas the slope for the solution sample ( ‡ 0.86
�
1
Æ 0.06 log (h )) approaches the value of ‡ 1
expected for speci®c base catalysis of the reaction,
the values in the polymeric solids are signi®cantly
smaller, suggesting that such a mechanism
cannot be invoked.
samples. In solution at pH 10, the k term in eq. 2
B
is the dominant contribution to k , and is more
o
than four orders of magnitude greater than the kW
term. However, in glassy solids, the kB term is
only 30-times greater than the kW term, suggest-
ing a decrease in the importance of base catalysis
relative to water catalysis at basic `pH'. The value
In the basic region (`pH' > 9), k values differ
o
dramatically in the three types of samples. The ko
values in solution are as much as 1000-fold
greater than that in the rubbery polymeric solid,
which in turn is ꢁ30-fold greater than that in the
glassy solid. The `pH'±rate pro®le reaches a
plateau for the polymeric solids for `pH' > 9
�
5
of K is on the order of 0.1±1.0 Â 10
M for all
three samples, and is consistent with the devel-
opment of a plateau in the `pH'±rate pro®le
between `pH' 9 and 10, with larger values of K
corresponding to initiation of the plateau at
higher `pH' values. Values of kW and kH are
comparable in the three types of samples.
(
Figure 4), suggesting that the reaction rate is
`pH' independent in this region. Calculated values
of the slopes of the curves in the basic region
con®rm a decreasing sensitivity to `pH' for all
three types of samples. For the solution sample,
the slope is ꢁ0.40 log(h ), roughly half the value
in the neutral region (Table 3). The slopes
calculated for the rubbery and glassy solids are
approximately equal to zero (Table 3), as expected
given the plateaus shown in Figure 4.
Buffer Catalysis Studies
�
1
To further explore the effects of `pH' on deamida-
tion in the basic range, buffer catalysis was
investigated in the range 6.5<`pH'<10.5 by
measuring kobs at varying buffer concentrations
in both solution and solid formulations while
maintaining constant `pH' and ionic strength. The
slope of the relationship between kobs and `pH' is
The shape of the `pH'±rate pro®le suggests that
the reaction is subject to both acid and base
Table 4. Contributions of Acid (kH), Water (kW), and Base (kB) Catalysis to the `pH'±
a
Rate Pro®le of Figure 4
2
4
2
5
kH Â 10
kW Â 10
kB Â 10
K Â 10
�
1
� 1
� 1
� 1
Formulation Type
(M
h
)
(h
)
(h
)
(M)
Solution
Rubbery solid
Glassy solid
15 Æ 2
70 Æ 10
6 Æ 2
11 Æ 2
22 Æ 5
6 Æ 1
7000 Æ 600
40 Æ 5
3:0 Æ 0:9
0:7 Æ 0:3
0:2 Æ 0:1
2 Æ 0:2
aValues determined by nonlinear regression of the data in Figure 4 to eq. 2.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
151
de®ned as the catalytic rate constant, kcat. The
results are summarized in Figure 5 and Table 5.
In solution at `pH' 6.6, there is no apparent
buffer catalytic effect. The kobs value is unaffected
by buffer concentration (Figure 5A) and the kcat
value (Table 5) is near zero. Because the `pH'±rate
pro®le shows a dependence of kobs on `pH' in this
range, this result suggests that the solution-state
reaction is subject to speci®c base catalysis (i.e.,
�
by OH ) in this region. In contrast, there is a
distinct buffer catalytic effect at higher solution
pH (9.5, 10.5; Figure 5A), with the value of kcat
increasing with increasing pH. When combined
with the pH dependence demonstrated in the
`pH'±rate pro®le, this result suggests that the
reaction is subject to general base catalysis (i.e.,
by the buffer anion) in this region, as reported
4,6
previously.
Like the solution-state samples, glassy poly-
meric solids at `pH' 6.6 show no apparent buffer
catalytic effect (Figure 5B). Again, because the
`pH' rate pro®le shows a `pH'-dependence in this
region, this result is consistent with speci®c base
catalysis. At higher `pH' values (9.5, 10.5), how-
ever, no buffer catalysis was observed and the kcat
value was not signi®cantly different from zero. In
addition, the `pH'±rate pro®le is independent of
`pH' for the glassy solids in this region. The
absence of both `pH' and buffer catalytic effects
suggests that the rate-determining step is `pH'
independent in glassy polymeric solids at basic
`pH'. This suggestion is in contrast to the general
base catalysis of the solution samples, and is
consistent with a change in the rate-determining
step in the polymeric solids relative to the solution
control.
Effect of Relative Humidity
Previous studies in our laboratories have demon-
strated that the deamidation of the Asn-hexapep-
tide in PVP matrices is sensitive to the relative
19,20
humidity (RH) of the surrounding gas phase,
an experimentally controllable parameter that
determines the equilibrium moisture content of
the polymer matrix. These studies also demon-
strated that water serves as both a reaction
Figure 5. Effect of buffer concentration on kobs
values in solution (Panel A) and in glassy PVP solids
(
Panel B) at basic `pH'. Symbols correspond to `pH'
values: &, `pH' 6.6; *, `pH' 9.5; ~, `pH' 10.4 in both
ꢀ
19,20
panels. Studies were performed at 70 C, with the
medium and a plasticizer in these systems.
exception of solution samples at `pH' 9.5 and 10.4 (*,
To determine whether the observed plateaus in
the `pH'±rate pro®le in basic polymeric solids
were in¯uenced by moisture content, additional
stability studies were conducted at 70 and 95%
RH in rubbery PVP matrices. The results are
shown in Figure 6, and are compared with the
~
, Panel A), for which rapid reaction prevented
effective sampling; these studies were performed at
ꢀ ꢀ
2 C, and the kobs values were extrapolated to 70 C.
2
Catalytic rate constants (kcat) determined from the
slopes of the lines are listed in Table 5. Error bars
represent standard deviations; n ˆ 3.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

1
52
SONG ET AL.
a
Table 5. Catalytic Rate Constants ꢂkcat† for the Buffer Catalysis Studies of Figure 5
2
� 1 � 1
kcat  10 (M
h )
Formulation Type
pH 6.6
pH 9.5
pH 10.4
b
b
Solution
Glassy solid
5:8 Æ 1:4
41 Æ 14
88 Æ 14
0:010 Æ 0:002
0:23 Æ 0:02
0:010 Æ 0:002
a
Determined by linear regression of the data in Figure 5.
These studies at 22 C; others at 70 C.
b
ꢀ
ꢀ
observed for both glassy and rubbery polymeric
solids and at varying matrix moisture content.
Reaction Mechanism
The appearance of identical degradation products
in solution and in polymeric solids and the
similarities in the overall shape of the `pH'±rate
pro®le suggest that the mechanism of Asn-
deamidation is fundamentally unaltered in the
polymeric media. However, the notable differ-
ences in the `pH'±rate pro®le in the neutral to
basic `pH' region are consistent with a shift in the
rate-determining step of the reaction in the
polymeric media. Speci®cally, in the polymeric
solids: (i) the slope of the `pH'±rate pro®le in the
neutral to basic region (i.e., 5<`pH'<8) decreases
relative to that for the solution control (Figure 4);
(
ii) the corresponding values of k decrease (Table
B
3); and (iii) a distinct plateau appears in the `pH'±
rate pro®le for `pH' > 9 (Figure 4). These observa-
tions suggest the lack of involvement of either the
hydroxide ion or the basic form of the buffer
species (Figure 5B) in the rate-determining step
in basic polymeric solids, a feature present in the
proposed mechanisms for the reaction in basic
Figure 6. Effect of relative humidity (RH) on deami-
dation rate in solid PVP systems at basic `pH'. Symbols
indicate varying RH and/or glycerol content: &, 70%
RH, 0% glycerol; *, 95% RH, 0% glycerol; ~, 50% RH,
3,4,6
solution.
We propose that the polymeric media
0
% glycerol; !, 50% RH, 30% (w/w) glycerol. Rate
constants are k values (i.e., extrapolated to zero buffer
o
exert a strong and differential effect on the rates
of the individual steps in the reaction, and that, in
the basic region, the rate-determining step shifts
from base-catalyzed ring formation to `pH'-inde-
pendent loss of ammonia.
concentration) for 50% RH data, but are kobs values (i.e.,
determined at a single buffer concentration of 0.15 M)
for 70% and 95% RH data. Error bars represent
standard deviations; n ˆ 3.
Figure 7A shows the proposed mechanism of
Asn-hexapeptide deamidation at neutral to basic
`pH'. Initially, a small amount of a highly reactive
corresponding 50% RH data from the `pH'±rate
pro®le for both glassy (0% glycerol) and rubbery
species is generated through the deprotonation of
the backbone NH center. Further attack by this
species on the side-chain amide functional group
of Asn leads through the ®rst transition state,
labeled TSattack, to a tetrahedral intermediate,
(
30% glycerol) samples. The k values increase
o
19,20
with increasing RH, as observed previously,
but `pH'-independent behavior is observed in all
these basic polymer matrices. Thus, the plateau
in the `pH'±rate pro®le in the basic region is
T � . Because progress through TS
requires
attack
the anionic form of the backbone amide nitrogen,
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
153
Figure 7. Proposed mechanism and energetics of Asn-hexapeptide deamidation in
solution and in polymeric solids. Panel A: Proposed mechanism. TSattack represents a
transition state formed by attack of the amide nitrogen on the Asn-carbonyl. TSexpulsion
represents a second transition state for the expulsion of ammonia. Panel B: Proposed
reaction energetics, summarizing `pH' and matrix effects on activation energies (Ea) for
TSattack and TSexpulsion. As shown, the diagram represents the relationship for the `pH'-
dependent reaction observed in basic aqueous solutions, in which (Ea, TSattack) > (Ea,
TSexpulsion). For the `pH'-independent reaction in polymeric solids, (Ea, TSexpulsion) >
(
Ea, TSattack) (not shown).
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

154
SONG ET AL.
this step is hydroxide ion dependent. The tetra-
hedral intermediate T � does not accumulate to
appreciable levels, but either decomposes to
regenerate the reactants or accepts a proton to
form a second tetrahedral intermediate, T Æ . The
intermediate T Æ may lose a proton to regenerate
T � , or progress through a second transition
between the two values. However, the high ratios
of Asu- to Asp-hexapeptide (Figure 3) in polymeric
samples at acidic `pH' may suggest that the true
hydrogen ion activity is less than the effective
value (i.e., the ``true `pH' '' is higher) because the
Asu-hexapeptide is not formed in appreciable
amounts in acidic aqueous solution (Figures 1
3,4,6
state, labeled TS
, forming the cyclic imide
expulsion
and 3).
Of course, the high ratios of Asu- to
intermediate and releasing ammonia into the
surrounding medium. This process is expected to
be independent of medium `pH'.
Asp-hexapeptide may also suggest that the reac-
tion mechanism is altered in acidic polymeric
media; discriminating between these explana-
tions is not possible unless the ``true'' hydrogen
ion activity is known. Similarly, the plateau in the
`pH'±rate pro®le in basic polymeric solids may be
interpreted mechanistically as a shift in the rate-
determining step, as already described, but may
also indicate that the `pH' value overestimates the
``true'' hydrogen ion activity (i.e., underestimates
the ``true pH''). Thus, the possible lack of corre-
spondence between ``effective'' and ``true'' pH
values should be borne in mind when considering
our results. This dif®culty also suggests that
further development of methods for the direct
measurement of hydrogen ion activity in solids is
warranted.
A second important limitation involves the
recovery of the Asn-hexapeptide and its degrada-
tion products from the polymeric solids. In the
studies reported here, the observed degradation
products were those recovered and detected
following the sample preparation and HPLC
assay methods already described. For the solution
samples, the cumulative amount of the Asn-
hexapeptide and its known degradation products
detected was equal to the initial Asn-hexapeptide
load at all times, so that the peptide recovery was
100%. In contrast, in the polymeric solids, peptide
recovery decreased with storage time, with <80%
of the initial peptide load recovered following 50 h
of storage. The causes of this incomplete recovery
are unknown, but may involve a change in
polymer morphology during storage that affects
peptide recovery, or an interaction between the
polymer and the peptide and/or its degradation
products that is not disrupted during sample
preparation and analysis. In analyzing the data,
we have assumed that when peptide recovery is
incomplete it is not preferential, so that the
compounds recovered are representative of the
relative amounts present in the solid matrix.
Finally, by labeling the samples as ``solution'',
``rubbery'', and ``glassy'', we have implied that
molecular mobility is of primary importance in
controlling reactivity, but polymer matrix incor-
This mechanism provides a framework for
interpreting the `pH'±rate pro®les in the neutral
to basic region. If the activation energy for TSattack
is greater than for TSexpulsion (Figure 7B), ``attack''
will be the rate-determining step and the overall
reaction will be observed to be `pH' dependent. If
the converse is true, ``expulsion'' will be rate
determining and the reaction as a whole will
appear to be `pH' independent. Because the
formation of TSattack is dependent on the concen-
tration of base, the activation energy for this step
is expected to decrease with increasing `pH'. With
regard to TSexpulsion, the loss of ammonia is
expected to occur readily in solution because of
the high mobility of both reactants and products,
but may be restricted in the more rigid polymeric
solids. The activation energy for TSexpulsion is then
expected to be less in solution than in the
polymeric media. The plateaus observed in the
`pH'±rate pro®le in the polymeric media can then
be understood to represent a shift in the rate-
determining step, from amide anion attack (via
TSattack) to expulsion of ammonia (via TSexpulsion).
This shift occurs at lower `pH' in the polymeric
samples, perhaps because mobility restrictions in
the solid state increase the activation energy for
TSexpulsion, so that a smaller hydroxide ion con-
centration is required to produce the condition
TSattack<TSexpulsion
.
Limitations of the Method
The experimental data and mechanistic interpre-
tations just presented should be viewed with some
caution in light of the limitations in the methods.
First, it is important to note that although the Asn
deamidation rate in polymeric solids has been
related to the `pH' (effective pH), the relationship
between this value and the ``true'' hydrogen ion
activity in the matrix is unknown. Similarities in
the general shape of the `pH'±rate pro®les of
solution and polymeric samples and in the `pH'
of maximum stability suggest a correspondence
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

EFFECT OF EFFECTIVE `pH' ON ASPARAGINE DEAMIDATION RATE
155
poration may in¯uence the reaction in other ways.
For example, inclusion in a polymer matrix may
affect the conformation and/or higher order
structure of peptides and proteins. Secondary
structure has been shown to have signi®cant
effects on peptide deamidation rates in solu-
polymeric solids is less sensitive to `pH' in the
neutral range (5<`pH'<8) than the solution
control, as indicated by the slopes of the `pH'±
rate pro®le. In basic polymeric solids (`pH' > 10), a
region in which the solution-state reaction shows
general base catalysis, the reaction is `pH'
independent. These results, together with sup-
porting studies of buffer catalytic effects, are
consistent with a change in the rate-determining
step of the reaction in the basic region, from `pH'-
dependent attack of the deprotonated backbone
amide nitrogen on the Asn side chain in solution
to `pH'-independent expulsion of ammonia in the
polymeric solids. As a whole, the results pre-
sented here challenge the notion that the `pH'
dependence of protein degradation reactions in
24±26
tion.
Although the small hexapeptide studied
here is unlikely to be structured in solution, a
degree of folding may be imposed by the polymer
matrix. This possibility has not been investigated.
In addition, the polarity of the polymer matrix
may differ from that in solution. Deamidation
occurs more rapidly in solutions of higher polarity
(
as measured by the solvent dielectric constant),
a ®nding attributed to greater stabilization of
the charged transition states (Figure 7A) by more
27,28
12
polar solvents.
We have assumed that the
the solid state simply parallels that in solution,
solution and polymer samples are similar in this
regard, but no attempt has been made to de®ne
and measure the ``effective polarity'' of the polymer
matrices. Lastly, speci®c interactions between the
peptide and the polymer may in¯uence the
deamidation rate in the polymer matrices. PVP
has been shown to interact with compounds
and provide strong evidence for a matrix effect on
the relative rates of the mechanistic steps of the
deamidation reaction.
ACKNOWLEDGMENTS
29,30
containing aromatic functional groups,
which
This work was supported by grant GM-54195
from the National Institutes of Health.
suggests the possibility of an interaction between
the polymer and the peptide tyrosine residue. The
degree to which such an interaction occurs in our
polymer samples and its effect on the rate of
deamidation are unknown.
REFERENCES
1
. Topp EM, Song Y, Wilson A, Li R, Hageman MJ,
Schowen RL. 2000. Solid state chemical stability of
peptides and proteins: Application to controlled
release formulations. In: Wise DL, editor. Hand-
book of pharmaceutical controlled release technol-
ogy. New York: Marcel Dekker, Inc., pp. 693±724.
. Lai MC, Topp EM. 1999. The solid state chemical
instability of peptides and proteins. J Pharm Sci
CONCLUSIONS
The rate of deamidation of the Asn-hexapeptide is
sensitive to `pH' both in polymer-containing
aqueous solutions and in glassy and rubbery
polymeric solids. The reaction follows pseudo-
2
3
88:489±500.
®
rst-order kinetics over a broad `pH' range (0.5±
. Brennan TV, Clarke S. 1995. Deamidation and iso-
Aspartate formation in model synthetic peptides:
the effects of sequence and solution environment.
In: Aswad DW, editor. Deamidation and iso-
Aspartate formation in peptides and proteins.
Ann Arbor, MI: CRC Press, pp 65±90.
1
2) in all three systems; similarities in the
products formed and their ratios suggest that
the mechanism of reaction is unaffected by
incorporation in polymeric media. In the acidic
range, rates of deamidation are comparable in
solution and in polymeric solids. The slope of
the `pH'±rate pro®le is approximately � 1 for all
three systems in acidic media, a result consistent
with speci®c acid catalysis. In the neutral to basic
4
. Patel K, Borchardt RT. 1990. Chemical pathways of
peptide degradation. II. Kinetics of deamidation of
an asparaginyl residue in a model hexapeptide.
Pharm Res 7:703±711.
5
. Xie M, Vander Velde D, Morton M, Borchardt RT,
Schowen RL. 1996. pH-induced change in the rate-
determining step for the hydrolysis of the Asp/
Asn-derived cyclic-imide intermediate in protein
degradation. J Am Chem Soc 118:8955±8956.
`pH' range, the deamidation rate is strongly
dependent on the sample type, with as much as
a 10,000-fold difference in rate between solution
and glassy polymeric solids. The shape of the
`
pH'±rate pro®le is also affected by matrix type in
6. Capasso S, Mazzarell L, Sica F, Zagari A, Salvadori
S. 1993. Kinetics and mechanism of succinimide
the neutral to basic range. The reaction in
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001

1
56
SONG ET AL.
nd
for pharmacists, 2 edition. New York: John Wiley
& Sons, pp. 43±59.
ring formation in the deamidation process of
asparagine residues. J Chem Soc Perkin Trans
2
:679±682.
19. Lai MC, Hageman MJ, Schowen RL, Borchardt RT,
Topp EM. 1999. Chemical stability of peptides in
polymers. 1. Effect of water on peptide deamidation
in poly(vinyl alcohol) and poly(vinyl pyrrolidone)
matrices. J Pharm Sci 88:1073±1080.
20. Lai MC, Hageman MJ, Schowen RL, Borchardt RT,
Laird BB, Topp EM. 1999. Chemical stability of
peptides in polymers. 2. Discriminating between
solvent and plasticizing effects of water on peptide
deamidation in poly(vinyl pyrrolidone). J Pharm
Sci 88:1081±1089.
21. Murase N, Franks F. 1989. Salt precipitation
during the freeze-concentration of phosphate buffer
solutions. Biophys Chem 34:293±300.
22. Song Y, Schowen RL, Borchardt RT, Topp EM.
Formaldehyde production by Tris buffer in peptide
formulations at elevated temperature (Note), sub-
mitted for publication in J Pharm Sci.
7
. Brunner A, M aÈ der K, G oÈ pferich A. 1999. pH and
osmotic pressure inside biodegradable micro-
spheres during erosion. Pharm Res16:847±853.
. Fu K, Pack DW, Klibanov AM, Langer R. 2000.
Visual evidence of acidic environment within
degrading poly(lactic-co-glycolic acid) (PLGA)
microspheres. Pharm Res 17:100±106.
. Costantino HR, Greibenow K, Langer L, Klibanov
QM. 1997. On the pH memory of lyophilized
compounds containing protein functional groups.
Biotechnol Bioeng 53:345±348.
8
9
1
0. Juang RW, Storey DE. 1998. Quantitative determi-
nation of the extent of neutralization of carboxylic
1
acid functionality in Carbopol 974P NF by diffuse
re¯ectance Fourier transform infrared spectrome-
try using Kubelka±Munk function. Pharm Res
1
5:1714±1720.
1
1
1. Munowitz M, Bachovchin WW, Herzfeld J, Dobson
CM, Grif®n RG. 1982. Acid-base and tautomeric
equilibria in the solid state: 15N NMR spectroscopy
of histidine and imidazole. J Am Chem Soc
23. Nyquist H. 1983. Saturated salt solutions for
maintaining speci®ed relative humidities. Technol
Prod Manuf 4:47±48.
24. Xie M, Schowen RL. 1999. Secondary structure and
protein deamidation. J Pharm Sci 88:8±13.
25. Stevenson CL, Friedman AR, Kubiak TM, Donlan
ME, Borchardt RT. 1993. Effect of secondary
structure on the rate of deamidation of several
growth hormone releasing factor analogs. Int J
Peptide Protein Res 42:497±503.
1
04:1192±1196.
2. Strickley RG, Anderson BD. 1996. Solid state
stability of human insulin. I. Mechanism and the
effect of water on the kinetics of degradation in
lyophiles from pH 2-5 solutions. Pharm Res
1
3:1142±1153.
1
1
3. Costantino HR, Langer R, Klibanov AM. 1994.
Moisture-induced aggregation of lyophilized insu-
lin. Pharm Res 11:21±29.
4. Oliyai C, Patel JP, Carr L, Borchardt RT. 1994.
Chemical pathways of peptide degradation. VII.
Solid state chemical instability of an aspartyl
26. Kosky AA, Razzaq UO, Treuheit MJ, Brems DN.
1999. The effects of alpha-helix on the stability of
Asn residues: Deamidation rates in peptides of
varying helicity. Protein Sci 8:2519±2523.
27. Brennan TV, Clarke S. 1993. Spontaneous degra-
dation of polypeptides at aspartyl and asparaginyl
residues: Effects of the solvent dielectric. Protein
Sci 2:331±338.
residue in
1:901±908.
a model hexapeptide. Pharm Res
1
1
5. Oliyai C, Patel J, Carr L. Borchardt RT. 1994. Solid
state stability of lyophilized formulations of an
asparaginyl residue in a model hexapeptide. J
Parent Sci Technol 48:167±173.
28. Li R, D'Souza AJ, Laird BB, Schowen RL, Borch-
ardt RT, Topp EM. 2000. Effects of solution polarity
and viscosity on peptide deamidation. J Peptide Res
56:326±334.
1
6. Pikal MJ, Dellerman KM, Roy ML, Riggin RM.
29. Plaizier-Vercammen JA. 1983. Interaction of povi-
done with aromatic compounds IV: Effects of
macromolecule molecular weight, solvent dielectric
constant, and ligand solubility on complex forma-
tion. J Pharm Sci 72:1042±1044.
30. Otagiri M, Imai T, Koinuma H, Matsumoto U.
1989. Spectroscopic study of the interaction of
coumarin anticoagulant drugs with polyvinylpyr-
rolidone. J Pharm Biomed Anal 7:929±935.
1
991. The effects of formulation variables on the
stability of freeze-dried human growth hormone.
Pharm Res 8:427±436.
1
7. Townsend MW, DeLuca PP. 1990. Stability of
ribonuclease A in solution and in the freeze-dried
state. J Pharm Sci 79:1083±1086.
1
8. Connors KA, Amidon GL, Stella VJ. 1986. In:
Chemical stability of pharmaceuticals: A handbook
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 2, FEBRUARY 2001