Kinetics of degradation and oil solubility of ester prodrugs of a model dipeptide (Gly-Phe)

Article abstract of DOI:10.1016/j.ejps.2004.04.013

Oil-based depot formulations may constitute a future delivery method for small peptides. Thus, a requirement is attainment of sufficient oil solubility for such active compounds. A model dipeptide (Gly-Phe) has been converted into lipophilic prodrugs by esterification at the C-terminal carboxylic acid group. The decomposition kinetics of octyl ester of Gly-Phe (IV) has been investigated at pH 7.4 (37°C) and IV was shown to degrade by first-order kinetics via two parallel pathways (1) intramolecular aminolysis resulting in formation of a 2,5-diketopiperazine and (2) hydrolysis of the ester bond producing the dipeptide. The cyclisation reaction was dominating in the decomposition of methyl (II) butyl (III) octyl (IV) decyl (V) and dodecyl (VI) esters of Gly-Phe at pH 7.4. However, this degradation pathway was almost negligible for pH below 6. During degradation of the dipeptide esters in 80% human plasma pH 7.4 (37°C) a minimal amount of cyclo(-Gly-Phe) was formed. A faster degradation of the esters in 80% human plasma pH 7.4 compared to those in aqueous solution pH 7.4 was suggested to be due to fast cleavage of the peptide bond. Low oil solubilities for Gly-Phe and the hydrochlorides of the dipeptide esters III and VI were observed. Although the solubility of Gly-Phe in oil solutions was enhanced by hydrophobic ion pairing with sodium decyl sulfonate the oil solubility was still less than 1 mg Gly-Phe/ml. By addition of a solubiliser, 10% N,N-dimethylacetamide (DMA), to Viscoleo the solubility of the HIP complexes increased significantly. The present study indicates that sufficient oil solubility might only be obtained for relatively small peptides by using the prodrug approach in combination with solubility enhancing organic solvents like DMA.

Full text of DOI:10.1016/j.ejps.2004.04.013

European Journal of Pharmaceutical Sciences 22 (2004) 399–408
Kinetics of degradation and oil solubility of ester prodrugs of a model
dipeptide (Gly-Phe)
Susan Weng Larsena,∗, Michael Ankersenb, Claus Larsena
a
Department of Pharmaceutics, The Danish University of Pharmaceutical Sciences, Universitetsparken 2, DK-2100 Copenhagen, Denmark
b
Novo Nordisk A/S, Novo Nordisk Park, DK-2760 Måløv, Denmark
Received 3 December 2003; received in revised form 6 April 2004; accepted 21 April 2004
Available online 19 June 2004
Abstract
Oil-based depot formulations may constitute a future delivery method for small peptides. Thus, a requirement is attainment of sufficient
oil solubility for such active compounds. A model dipeptide (Gly-Phe) has been converted into lipophilic prodrugs by esterification at the
C-terminal carboxylic acid group. The decomposition kinetics of octyl ester of Gly-Phe (IV) has been investigated at pH 7.4 (37 ◦C) and IV was
shown to degrade by first-order kinetics via two parallel pathways (1) intramolecular aminolysis resulting in formation of a 2,5-diketopiperazine
and (2) hydrolysis of the ester bond producing the dipeptide. The cyclisation reaction was dominating in the decomposition of methyl (II)
butyl (III) octyl (IV) decyl (V) and dodecyl (VI) esters of Gly-Phe at pH 7.4. However, this degradation pathway was almost negligible for pH
below 6. During degradation of the dipeptide esters in 80% human plasma pH 7.4 (37 ◦C) a minimal amount of cyclo(-Gly-Phe) was formed.
A faster degradation of the esters in 80% human plasma pH 7.4 compared to those in aqueous solution pH 7.4 was suggested to be due to
fast cleavage of the peptide bond. Low oil solubilities for Gly-Phe and the hydrochlorides of the dipeptide esters III and VI were observed.
Although the solubility of Gly-Phe in oil solutions was enhanced by hydrophobic ion pairing with sodium decyl sulfonate the oil solubility
was still less than 1 mg Gly-Phe/ml. By addition of a solubiliser, 10% N,N-dimethylacetamide (DMA), to Viscoleo the solubility of the HIP
complexes increased significantly. The present study indicates that sufficient oil solubility might only be obtained for relatively small peptides
by using the prodrug approach in combination with solubility enhancing organic solvents like DMA.
© 2004 Published by Elsevier B.V.
Keywords: Parenteral depots; Dipeptide ester prodrugs; Oil solubility; Hydrophobic ion pair; Intramolecular cyclisation
1. Introduction
4-imidazolidinone prodrugs of the polar local anaesthetic
agent prilocaine, which like many peptides contains an
␣-aminoamide moiety, with the objective to design depot
oil injectables (Larsen et al., 2003). In the present study
attempts to enhance the oil solubility of a dipeptide by
creating lipophilic derivatives were investigated. A model
dipeptide (Gly-Phe) was converted into lipophilic prodrugs
by esterification at the free carboxylic acid at the C-terminal.
A second approach based on formation of hydrophobic ion
pairs has previously been applied to increase the solubility
of peptides and proteins in organic solvents (Meyer and
Manning, 1998). Thus, in order to investigate the effect
of “masking” the positive charge of the ester prodrugs on
the oil solubility hydrophobic ion pairs have been formed
between the positive charged N-terminal of the ester pro-
drugs and sodium decyl sulfonate. The degradation kinetics
including investigation of intramolecular cyclisation of the
dipeptide esters has also been studied.
Parenteral delivery techniques capable of controlling
the release of smaller peptides and peptidomimetics are
of considerable interest. Various approaches in the field
of parenteral administration of peptide and protein com-
pounds have been investigated (Cleland et al., 2001;
Rothen-Weinhold and Gurny, 1997). To this end, the poten-
tial applicability of peptide delivery by parenteral oil-based
depots has not gain much attention. The feasibility of
administering such polar drug substances in the form of
oil solutions is governed by the attainment of sufficient
oil solubility. Increased lipophilicity of peptide-like com-
pounds has been obtained by the use of the prodrug ap-
proach (Bundgaard, 1991). Recently, we have synthesized
∗
Corresponding author. Tel.: +45 3530 6198; fax: +45 3530 6010.
E-mail address: swe@dfuni.dk (S.W. Larsen).
0928-0987/$ – see front matter © 2004 Published by Elsevier B.V.
doi:10.1016/j.ejps.2004.04.013

400
S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
2. Materials and methods
2.2.1.3. Boc-Phe-O-octyl ester. Same procedure as with
Boc-Phe-O-butyl ester. 1H NMR (400 MHz, CDCl3) δ: 0.89
(m, 3H, CH3), 1.27 (b, 10H, (CH2)5), 1.42 (s, 9H, t-butyl),
1.56 (m, 2H, CH2), 3.08 (m, 2H, CH2), 4.08 (m, 2H, CH2),
4.57 (m, 1H, CH), 4.97 (b, 1H, NH), 7.12–7.30 (m, 5H);
LC–MS: Rt = 8.52 min, m/z = 378.4 (m + 1).
2.1. Materials
In the synthesis of Gly-Phe esters the following chemicals
were used: 4-dimethylaminopyridine 99% and 1-(3-dimeth-
ylaminopropyl)-3-ethylcarbodimide hydrochloride 98%
from Aldrich (Germany); N,N-diisopropylethylamine and
Phe-O-methyl ester, HCl from Fluka Chemie (Switzer-land);
1-hydroxybenzotriazole, Boc-Phe-OH, Boc-Phe-OH from
Novo Nordisk A/S (Denmark); 1-butanol from Rieldel
de Häen (Germany); 1-decanol and from Lancaster (Eng-
land); 1-octanol and n-dodecanol from Merck (Ger-
many). Cyclo(-Gly-Phe) was obtained from Bachem
(Switzerland) and Gly-Phe and castor oil were from
Sigma (USA). Fractionated coconut oil (Viscoleo®) and
N,N-dimethylacetamide 99% (DMA) were purchased from
P. Broeste A/S (Denmark) and Acros Organics (USA), re-
spectively. Sodium decyl sulfonate (1-decanesulfonic acid
sodium salt) was obtained from Buchem bv (Holland).
Other chemicals and solvents were of analytical or chro-
matographic grade. Demineralized water was used through-
out.
2.2.1.4. Phe-O-octyl ester, HCl. Same procedure as with
Phe-O-butyl ester, HCl. 1H NMR (400 MHz, DMSO) δ: 0.86
(m, 3H, CH3), 1.22 (m, 10H, (CH2)5), 1.42 (m, 2H, CH2),
3.02–3.25 (m, 2H, CH2), 4.02 (m, 2H, CH2), 4.24 (b, 1H,
CH), 7.24–7.35 (m, 5H), 8.69 (b, 3H, NH3); LC–MS: Rt
= 5.23 min, m/z = 278.4 (m + 1).
2.2.1.5. Boc-Phe-O-decyl ester. Same procedure as with
Boc-Phe-O-butyl ester. 1H NMR (400 MHz, CDCl3) δ: 0.88
(m, 3H, CH3), 1.27 (b, 14H, (CH2)7), 1.42 (s, 9H, t-butyl),
1.57 (m, 2H, CH2), 3.08 (m, 2H, CH2), 4.08 (m, 2H, CH2),
4.57 (m, 1H, CH), 4.97 (b, 1H, NH), 7.12–7.30 (m, 5H);
LC–MS: Rt = 9.12 min, m/z = 406.4 (m + 1).
2.2.1.6. Phe-O-decyl ester, HCl. Same procedure as with
Phe-O-butyl ester, HCl. 1H NMR (400 MHz, DMSO) δ: 0.86
(m, 3H, CH3), 1.24 (m, 14H, (CH2)7), 1.43 (m, 2H, CH2),
3.02–3.23 (m, 2H, CH2), 4.02 (m, 2H, CH2), 4.24 (b, 1H,
CH), 7.23–7.35 (m, 5H), 8.62 (b, 3H, NH3); LC–MS: Rt
= 5.35 min, m/z = 306.4 (m + 1).
2.2. Synthesis of hydrochlorides of Gly-Phe esters
2.2.1. Esterification of Phe
2.2.1.1. Boc-Phe-O-butyl ester. To an ice-cold stirred
solution of Boc-Phe-OH (1.91 g, 7.2 mmol), 1-butanol
(0.66 ml, 7.2 mmol) and 4-dimethylaminopyridine (0.44 g,
3.6 mmol) in 50 ml of dichloromethane was added
1-(3-dimethylaminopropyl)-3-ethylcarbodimide hydrochlo-
ride (1.38 g, 7.2 mmol). The reaction mixture was stirred
at 0 ◦C for 3 h, and overnight at ambient temperature. The
reaction mixture was transferred to a separating funnel with
100 ml dichloromethane. The dichloromethane layer was
washed with 10% NaHSO4 (50 ml), H2O (50 ml), satu-
rated aqueous NaHCO3 (50 ml), H2O (50 ml) and saturated
aqueous NaCl (50 ml). The organic layer was dried with
anhydrous MgSO4 and concentrated under reduced pres-
sure to an oil. 1H NMR (400 MHz, CDCl3) δ: 0.91 (m, 3H,
CH3), 1.32 (b, 2H, CH2), 1.42 (s, 9H, t-butyl), 1.56 (m,
2H, CH2), 3.08 (m, 2H, CH2), 4.09 (m, 2H, CH2), 4.57 (m,
1H, CH), 4.97 (b, 1H, NH), 7.13–7.31 (m, 5H); LC–MS:
Rt = 7.00 min, m/z = 322.4 (m + 1).
2.2.1.7. Boc-Phe-O-dodecyl ester. Same procedure as
with Boc-Phe-O-butyl ester. H NMR (400 MHz, CDCl3)
δ: 0.88 (m, 3H, CH3), 1.26 (b, 18H, (CH2)9), 1.42 (s, 9H,
t-butyl), 1.58 (m, 2H, CH2), 3.08 (m, 2H, CH2), 4.08 (m,
2H, CH2), 4.57 (m, 1H, CH), 4.98 (b, 1H, NH), 7.12–7.31
(m, 5H); LC–MS: Rt =9.65 min, m/z = 434.4 (m + 1).
1
2.2.1.8. Phe-O-dodecyl ester, HCl. Same procedure as
1
with Phe-O-butyl ester, HCl. H NMR (400 MHz, DMSO)
δ: 0.86 (m, 3H, CH3), 1.24 (m, 18H, (CH2)9), 1.42 (m,
2H, CH2), 3.01–3.23 (m, 2H, CH2), 4.02 (m, 2H, CH2),
4.25 (m, 1H, CH), 7.23–7.35 (m, 5H), 8.65 (b, 3H, NH3);
LC–MS: Rt =6.53 min, m/z = 334.2 (m + 1).
2.2.2. Synthesis of Gly-Phe esters (Fig. 1)
2.2.2.1. Boc-Gly-Phe-O-methyl ester.
A
solution of
Boc-Gly-OH (1.49 g, 8.5 mmol), 1-hydroxybenzotriazole
(1.30 g, 8.5 mmol) and 1-(3-dimethylaminopropyl)-3-ethyl-
carbodimide hydrochloride (1.63 g, 8.5 mmol) in 30 ml of
dimethylformamide was stirred at ambient temperature for
30 min. To the solution N,N-diisopropylethylamine (2.20 g,
17 mmol) and Phe-O-methyl ester, HCl (1.52 g, 8.5 mmol)
were added and the reaction mixture was stirred at ambient
temperature overnight. The reaction mixture was transferred
to a separating funnel with 100 ml ethyl acetate. The ethyl
acetate layer was washed with 10% NaHSO4 (50 ml), H2O
(50 ml), saturated aqueous NaHCO3 (50 ml), H2O (50 ml)
2.2.1.2. Phe-O-butyl ester, HCl. Boc-Phe-O-butyl ester
(2.1 g, 6.6 mmol) was dissolved in ethyl acetate followed
by 3.8 M HCl in ethyl acetate. After 30 min the reaction
mixture was rotary evaporated under reduced pressure to
obtain a white powder. 1H NMR (400 MHz, DMSO) δ:
0.82 (m, 3H, CH3), 1.18 (m, 2H, CH2), 1.42 (m, 2H, CH2),
3.02–3.24 (m, 2H, CH2), 4.03 (m, 2H, CH2), 4.24 (m, 1H,
CH), 7.24–7.36 (m, 5H), 8.66 (b, 3H, NH3); LC–MS: Rt
= 3.63 min, m/z = 222.4 (m + 1).

S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
401
5.03 (b, 1H, NH), 6.46 (b, 1H, NH), 7.09–7.30 (m, 5H);
I
R = H
LC–MS: Rt = 7.72 min, m/z = 435.4 (m + 1).
II
R = CH3
O
III R = (CH2)3CH3
IV R = (CH2)7CH3
2.2.2.6. Gly-Phe-O-octyl ester, HCl (compound IV). Same
procedure as with Gly-Phe-O-methyl ester, HCl. How-
ever, after 30 min the reaction mixture was rotary evapo-
rated under reduced pressure to obtain a brown powder.
The product was further purified by recrystallization from
ethanol/tert-butylmethylether. 1H NMR (400 MHz, DMSO)
δ: 0.86 (m, 3H, CH3), 1.23 (b, 10H, (CH2)5), 1.48 (m, 2H,
CH2), 2.99 (m, 2H, CH2), 3.54 (m, 2H, CH2), 3.98 (m,
2H, CH2), 4.53 (m, 1H, CH), 7.21–7.31 (m, 5H), 8.19 (b,
3H, NH3), 9.02 (b, 1H, NH); LC–MS: Rt = 5.26 min, m/z
= 335.4 (m + 1); Anal. Calcd. (C19H30N2O3, HCl): C,
61.53; H, 8.42; N, 7.55; Found: C, 61.64; H, 8.76; N, 7.56;
HPLC: Rt = 39.7 min (A1), Rt = 42.6 min (B1).
H2N
OR
NH
O
V
R = (CH2)9CH3
VI R = (CH2)11CH3
Fig. 1. Chemical structures of Gly-Phe (I) and ester prodrugs thereof
(II–VI).
and saturated aqueous NaCl (50 ml). The organic layer was
dried with anhydrous MgSO4 and concentrated under re-
duced pressure to an oil. H NMR (400 MHz, CDCl3) δ:
1.42 (s, 9H, t-butyl), 1.42 (s, 3H, CH3), 1.45 (s, 3H, CH3),
3.13 (m, 2H, CH2), 3.70 (m, 3H, CH3), 4.85 (m, 1H, CH),
4.85 (b, 1H, NH), 6.86 (b, 1H, NH) 7.11–7.30 (m, 5H);
LC–MS: Rt = 4.75 min, m/z = 337.4 (m + 1).
1
2.2.2.7. Boc-Gly-Phe-O-decyl ester. Same procedure as
with Boc-Gly-Phe-O-methyl ester. 1H NMR (400 MHz,
CDCl3) δ: 0.88 (m, 3H, CH3), 1.27 (b, 14H, (CH2)7), 1.44
(s, 9H, t-butyl), 1.59 (b, 2H, CH2), 3.12 (m, 2H, CH2),
3.79 (m, 2H, CH2), 4.09 (m, 2H, CH2), 4.87 (m, 1H, CH),
5.04 (b, 1H, NH), 6.47 (b, 1H, NH), 7.09–7.30 (m, 5H);
LC–MS: Rt = 8.67 min, m/z = 463.6 (m + 1).
2.2.2.2. Gly-Phe-O-methyl ester, HCl (compound II).
Boc-Gly-Phe-O-methyl ester (2.1 g, 6.3 mmol) was dis-
solved in ethyl acetate followed by 3.8 M HCl in ethyl
acetate. After 30 min a white powder precipitates. The pow-
der was isolated and dried. 1H NMR (400 MHz, DMSO) δ:
3.00 (m, 2H, CH2), 3.53 (m, 2H, CH2), 3.62 (s, 3H, CH3),
4.56 (m, 1H, CH), 7.21–7.32 (m, 5H), 8,13 (s, 3H, NH3),
8.99 (b, 1H, NH); LC–MS: Rt = 2.35 min, m/z = 237.4 (m
+ 1); Anal. Calcd. (C12H16N2O3, HCl): C, 52.85; H, 6.28;
N, 10.27; Found: C, 52.50; H, 6.32; N, 10.04; HPLC: Rt
= 16.1 min (A1), Rt = 18.2 min (B1).
2.2.2.8. Gly-Phe-O-decyl ester, HCl (compound V). Same
procedure as with Gly-Phe-O-methyl ester, HCl. However,
after 30 min the reaction mixture was rotary evaporated un-
der reduced pressure to obtain a brown powder. The product
was dissolved in water and applied to a C-18 Sep-Pak car-
tridge (10 g), which had been prewashed with isopropanol
and water. Then a gradient of an eluent consisting of wa-
ter/acetonitril (+0.1% HCl) (10, 20, 30, 40, 50 and 60%
acetonitrile in water) was run through the Sep-Pak. The rel-
evant fractions were combined and freeze-dried to obtain a
white powder. 1H NMR (400 MHz, DMSO) δ: 0.86 (m, 3H,
CH3), 1.24 (b, 14H, (CH2)7), 1.48 (m, 2H, CH2), 2.99 (m,
2H, CH2), 3.54 (m, 2H, CH2), 3.99 (m, 2H, CH2), 4.54 (m,
1H, CH), 7.22–7.31 (m, 5H), 8.08 (b, 3H, NH3), 8.95 (b, 1H,
NH); LC–MS: Rt = 6.15 min, m/z = 363.2 (m + 1); Anal.
Calcd. (C21H34N2O3, HCl): C, 63.22; H, 8.84; N, 7.02;
Found: C, 63.41; H, 8.83; N, 6.99; HPLC: Rt = 44.9 min
(A1), Rt = 48.6 min (B1).
2.2.2.3. Boc-Gly-Phe-O-butyl ester. Same procedure as
with Boc-Gly-Phe-O-methyl ester. LC–MS: Rt = 5.95 min,
m/z = 379.4 (m + 1).
2.2.2.4. Gly-Phe-O-butyl ester, HCl (compound III). Same
procedure as with Gly-Phe-O-methyl ester, HCl. However,
after 30 min the reaction mixture was rotary evaporated un-
der reduced pressure to obtain a brown powder. The prod-
uct was further purified by recrystallization from ethanol.
1H NMR (400 MHz, DMSO) δ: 0.85 (m, 3H, CH3), 1.24
(m, 2H, CH2), 1.48 (m, 2H, CH2), 2.92–3.06 (m, 2H, CH2),
3.53 (m, 2H, CH2), 4.01 (m, 2H, CH2), 4.54 (m, 1H, CH),
7.21–7.32 (m, 5H), 8,09 (b, 3H, NH3), 8.96 (b, 1H, NH);
LC–MS: Rt = 4.08 min, m/z = 279.4 (m + 1); Anal. Calcd.
(C15H22N2O3, HCl): C, 57.23; H, 7.36; N, 8.90; Found:
C, 57.61; H, 7.47; N, 8.87; HPLC: Rt = 27.2 min (A1), Rt
= 29.1 min (B1).
2.2.2.9. Boc-Gly-Phe-O-dodecyl ester. Same procedure
1
as with Boc-Gly-Phe-O-methyl ester. H NMR (400 MHz,
CDCl3) δ: 0.89 (m, 3H, CH3), 1.26 (b, 18H, (CH2)9), 1.44
(s, 8H, t-butyl), 1.58 (m, 2H, CH2), 3.12 (m, 2H, CH2),
3.79 (m, 2H, CH2), 4.08 (m, 2H, CH2), 4.86 (m, 1H, CH),
5.05 (b, 1H, NH), 6.48 (b, 1H, NH), 7.09–7.30 (m, 5H);
LC–MS: Rt = 8.92 min, m/z = 491.4 (m + 1).
2.2.2.5. Boc-Gly-Phe-O-octyl ester. Same procedure as
with Boc-Gly-Phe-O-methyl ester. 1H NMR (400 MHz,
CDCl3) δ: 0.89 (m, 3H, CH3), 1.28 (b, 10H, (CH2)5), 1.44
(s, 8H, t-butyl), 1.59 (b, 2.7H, CH2), 3.12 (m, 2H, CH2),
3.80 (m, 2H, CH2), 3.90 (m, 2H, CH2), 4.88 (m, 1H, CH),
2.2.2.10. Gly-Phe-O-dodecyl ester, HCl (compound VI).
Same procedure as with Gly-Phe-O-methyl ester, HCl. How-
ever, after 30 min the reaction mixture was rotary evaporated

402
S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
under reduced pressure to obtain a white powder. The prod-
uct was further purified by recrystallization from ethanol.
1H NMR (400 MHz, DMSO) δ: 0.85 (m, 3H, CH3), 1.24 (b,
18H, (CH2)9), 1.48 (m, 2H, CH2), 2.91–3.05 (m, 2H, CH2),
3.54 (m, 2H, CH2), 3.99 (m, 2H, CH2), 4.54 (m, 1H, CH),
7.22–7.31 (m, 5H), 8.04 (b, 3H, NH3), 8.92 (b, 1H, NH);
LC–MS: Rt = 7.03 min, m/z = 391.2 (m + 1); Anal. Calcd.
(C23H38N2O3, HCl): C, 64.69; H, 9.21; N, 6.56; Found: C,
64.88; H, 9.56; N, 6.54; HPLC: Rt = 39.79 min (B6).
1973). These reactions were followed for a time period cor-
responding to 1–2% degradation of compound IV.
2.4.2. Degradation in human plasma
The stability of the compounds I–VI, cyclo(-Gly-Phe)
and Phe-O-methyl ester was investigated in human plasma:
0.1 M phosphate pH 7.4 4:1 (v/v). The reactions were initi-
ated by adding 50 ␮l of a stock solution in methanol (in the
case of compound I and cyclo(-Gly-Phe) an aqueous stock
solution was used) to 5 ml of preheated plasma to give an
initial concentration of 1×10−4 to 4×10−4 M. The reaction
mixture was kept in a water bath at 37 0.5 ◦C and at appro-
priate time intervals samples were taken and deproteinised
with either acetonitrile or methanol in the ratio of 1:2 (v/v)
and 1:3 (v/v), respectively. The mixture was immediately
vortexed and after centrifugation for 3 min at 13 000 rpm the
supernatant was analysed by HPLC for remaining ester. For
the compounds II–VI the amount of cyclo(-Gly-Phe) formed
was determined at the end of the study.
2.3. Hydrophobic ion pairing between Gly-Phe esters and
sodium decyl sulfonate
Hydrophobic ion pair (HIP) complexes between sodium
decyl sulfonate and compounds III and VI were formed by
the following procedure: To an aqueous solution of the hy-
drochloride of the Gly-Phe ester an equimolar amount of
sodium decyl sulfonate dissolved in demineralised water
was added. The HIP complex precipitated from the solution
and after stirring for 5 min the complex was isolated and
freeze-dried. The concentration of sodium decyl sulfonate
was kept below the critical micelle concentration (CMC:
44 mM (Bromberg and Klibanov, 1994)).
2.5. Solubility measurements
The solubility of compound III, VI, Gly-Phe and the de-
cyl sulfonate salts of compound III and VI was measured in
Viscoleo, Viscoleo/castor oil 2:1 (v/v) and Viscoleo/DMA
9:1 (v/v). To 1 ml oil solution in a screw capped test tube
an amount of the investigated compound was added and the
tube was rotated at ambient temperature. Addition of com-
pound was repeated until solid material was remaining af-
ter rotating for at least 3 days. The supernatant was trans-
ferred to an Eppendorf tube and centrifuged at 13 000 rpm
for 20 min. Hundred microliters of the supernatant was ade-
quately diluted with ethanol prior to HPLC analysis. A solu-
tion containing less than 5% (v/v) oil in ethanol was applied
to the column. The amount of Gly-Phe in the oil solution
was quantified by peak area measurements in comparison to
those of standards chromatographed under the same condi-
tions.
2.4. Kinetic measurements
2.4.1. Degradation in aqueous buffer solutions
The degradation of compound IV was investigated in the
pH interval 1.14–9.67 whereas the stability of compounds
II, III, V and VI only were investigated at pH 7.38. All
rate studies were carried out in aqueous buffer solution
at 37 0.5 ◦C and the buffers used were phosphate (pH
6.02–7.38) and borate (pH 8.22–9,67). At pH below 2.5 hy-
drochloric acid was used. In the pH range 8.22–9.67, the
influence of the borate buffer on the degradation rate was in-
vestigated using three different buffer concentrations, 0.02,
0.05 and 0.1 M. At pH 6.02–7.38, the degradation rates were
determined at three different phosphate buffer concentra-
tions, 0.005, 0.01 and 0.015 M. A constant ionic strength of
0.5 was maintained for each buffer by adding a calculated
amount of potassium chloride. The reactions were in most
cases initiated by adding 100 ␮l of a 1 × 10−2 M stock so-
lution of the compounds in methanol (in the case of com-
pound I and cyclo(-Gly-Phe) an aqueous stock solution was
used) to 10.0 ml of preheated buffer solution resulting in
a final concentration of approximately 1 × 10−4 M. At ap-
propriate time intervals samples were withdrawn and anal-
ysed immediately by HPLC. In case of fast reactions (at pH
10), samples were taken and hydrochloric acid was added
to pH 5 prior to analysis in order to stop the degradation
process.
2.6. Determination of pKa value of compound III
The pKa value of compound III was determined by three
independent potentiometric titrations running from pH 3 to
10 at 37 0.5 ◦C and at an ionic strength of µ = 0.5 under
a slow argon flow. A Sirius GLpKa instrument with Sirius
pKa LOGPTM revision 5.2a software from Sirius Analytical
Instruments (Forest Row, UK) was used.
2.7. Analyses
2.7.1. 1H NMR, LC/MS and elemental analyses
1H NMR were recorded on a Bruker 400 MHz using
CDCl3 or d6-DMSO as solvent, and chemical shifts (δ) are
in ppm relative to tetramethylsilane.
Elemental analyses were obtained from Novo Nordisk
Microlab.
Pseudo-first-order degradation rate constants (kobs) were
determined from the slopes of linear plots of the logarithm of
intact ester prodrug against time. For the slowly proceeding
reaction at pH 2.14 and 6.02 the pseudo-first-order rate con-
stant was derived by using the initial rate method (Connors,

S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
403
The LC–MS method consisting of a HP1100 MSD
3. Results and discussion
equipped with binary pump, column compartment, diode
array detector and a single quadrupole mass spectrometer
detector. Electrospray ionisation was performed in posi-
tive ion made mode. The analysis was performed at 40 ◦C
on 3 mm × 50 mm YMC-pack ODS-A column. The mo-
bile phase consisted of mobile phase A: 0.1% TFA in 5%
acetonitrile, and mobile phase B: 0.1% TFA in 90% ace-
tonitrile. A linear gradient from 0 to 100% B over 15 min
was performed using a flow rate of 1 ml/min.
3.1. Synthesis of hydrochlorides of Gly-Phe esters
Structures of the synthesised ester derivatives of Gly-Phe
are shown in Fig. 1. The compounds were prepared by the
following steps: Esterification of the Boc protected phenyl-
alanine using the appropriate alcohol, removal of the protect-
ing group with hydrochloric acid in ethyl acetate, coupling
between the hydrochloride of the phenylalanine ester and
the Boc protected glycine, removal of the protecting group
with hydrochloric acid in ethyl acetate and if necessary pu-
rification by recystallization or by using Sep-Pak cartridge
followed by freeze-drying.
2.7.2. HPLC methods
Solutions of compounds II–V dissolved in methanol were
analysed using two HPLC systems. HPLC (method A1): The
HPLC analysis was performed using a Waters 2690 systems
fitted with a Waters 996 diode array detector. UV detections
were collected at 214, 254, 276, and 301 nm on a 218TP54
4.6 mm × 250 mm × 5 ␮m C-18 silica column (The Seper-
ations Group, Hesperia), which was eluted at 1 ml/min at
42 ◦C. The column was equilibrated with 5% acetonitrile in
a buffer consisting of 0.1 M ammonium sulphate adjusted
to pH 2.5 with 4 M sulphuric acid. After injection the sam-
ple was eluted by a gradient of 5–60% acetonitrile in the
same aqueous buffer during 50 min. HPLC (method B1): The
HPLC-analysis was performed on same system described in
method A1 using another gradient system. The column was
equilibrated with mobile phase consisting of 0.1% TFA in
5% acetonitrile. After injection the sample was eluted by a
gradient of 5–60% acetonitrile (with 0.1% TFA added) dur-
ing 50 min. The HPLC-analysis of compound VI was per-
formed as described in method B1 however a gradient of
5–90% acetonitrile was used (method B6).
Samples from the stability studies were analysed by a
HPLC system consisted of a Merck Hitachi L-6000 pump
connected to a Merck Hitachi L-4000 UV detector. Re-
versed phase chromatography was carried out using an In-
ertsil ODS-2 (250 mm × 4.6 mm, 5 ␮m particles) column
equipped with an Inertsil ODS-2 precolumn (Chrompack).
The mobile phases consisted acetonitrile and 0.1% phos-
phoric acid in demineralised water which were made 1 mM
with respect to triethylamine. The content of acetonitrile in
the mobile phase was adjusted to 5–55% (v/v) to give ap-
propriate retention time for the actual compound. The flow
rate was 1 ml/min and the column effluent was monitored at
220 nm. Some of the stability studies were performed using
a HPLC-system consisting of a Merck Hitachi L-7100 pump
connected to a Merck-Hitachi L-7400 UV detector and with
an automatically sampling procedure using a Merck Hitachi
L-7200 autosampler. These reaction mixtures were kept at
constant temperature by using a Merck L-7200 Peltier Sam-
ple Cooler.
The synthesis processes were not optimized thus prod-
1
uct yields of the ester derivatives were 40–50%. H NMR,
LS/MS and elemental analyses confirm the structures of the
compounds and the compounds were of high purity (HPLC
system A1, B1 and B6 > 99%). The retention times (Rt) of
the compounds in the gradient HPLC systems A1 and B1
increased from compound II (16 and 18 min) to compound
V (45 and 49 min) and with Rt of compound VI exceeding
the analysis time of 50 min. By using another HPLC gradi-
ent system (method B6) the lipophilic compound VI could
be detected.
3.2. Stability studies
The overall degradation of all ester prodrugs followed
apparent first-order kinetics at fixed pH, temperature, ionic
strength and buffer concentration. Product analysis (decom-
position in 0.1 M phosphate buffer pH 7.38 at 37 0.5 ◦C)
revealed that the disappearance of compound IV was accom-
panied by parallel formation of Gly-Phe and cyclo(-Gly-Phe)
(Fig. 2).
8E-05
7E-05
2E-07
6E-05
1,5E-07
5E-05
1E-07
4E-05
5E-08
3E-05
0
0
5
10
15
20
25
30
2E-05
1E-05
0
Time, hour
0
5
10
15
20
25
30
Samples from the solubility measurements were analysed
by a HPLC system consisting of a Shimadzu LC-6A pump
and a Shimadzu SPD-6A UV detector operating at 260 nm,
otherwise the HPLC analyses were performed at the same
settings and conditions as described for the stability studies.
Time, hour
Fig. 2. Time course for compound IV (᭢), Gly-Phe (᭹) and
cyclo(-Gly-Phe) (᭿) during the degradation of compound IV in 0.1 M
phosphate buffer at pH 7.38 at 37 0.5 ◦C. The full lines have been
derived using Eqs. (2) and (3).

404
S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
O
H2N
OR
NH
E
O
k1
k2
O
O
O
+
ROH
N
H2N
OH
+
ROH
NH
N
O
I
C
Scheme 1.
Through the experiment the sum of the concentration of
compound IV and the two products was 100 3% and
cyclo(-Gly-Phe) was the major degradation product (99.8%)
formed. Compound IV is suggested to degrade via two
parallel pathways as depicted in Scheme 1; intramolecular
aminolysis resulting in formation of a 2,5-diketopiperazine
(cyclo(-Gly-Phe)) and direct hydrolysis of the ester bond
producing the dipeptide (Gly-Phe). Thus, the overall rate
constant kobs can be expressed:
k2
obst
[I] =
[E]0(1 − e−k
)
(5)
kobs
with rate constants k1 (3.8×10−3 min−1) and k2 (7.5 ×
10−6 min−1) being in accordance with those obtained by
the former method. In conclusion, the results confirm a par-
allel degradation mechanism of compound IV at the exper-
imental conditions employed. Similar degradation profiles
were seen for dipeptide derivatives by Gu and Strickley
(1987, 1988) and formation of 2,5-piperazinedione as a
result of intramolecular aminolysis has been described for
other esters of dipeptides (Jensen and Bundgaard, 1991;
Meresaar and Ågren, 1968; Purdie and Benoiton, 1973)
and for p-nitroaniline amides of a dipeptide (Goolcharran
and Borchardt, 1998). In the decomposition of com-
pound IV cyclo(-Gly-Phe) was only formed in significant
amounts for pH > 6. Similarly, cyclo(-Gly-Phe) was the
major degradation product in the decomposition of com-
pounds II–VI at pH 7.4. These observations suggest that
cyclo(-Gly-Phe) formation results from a nucleophilic attack
of the free amino group on the carbonyl group of the ester
bond.
kobs = k1 + k2
(1)
where k1 and k2 are the first-order rate constants for the cy-
clisation process and the hydrolysis of the ester bond, re-
spectively. The rate constants were obtained from the fol-
lowing expressions:
ꢀ
ꢁ
k1
ln([C]∞ − [C]) = ln
[E]0 − kobst
(2)
(3)
kobs
ꢀ
ꢁ
k2
ln([I]∞ − [I]) = ln
[E]0 − kobst
kobs
where [I] and [C] are the concentration of Gly-Phe and
cyclo(-Gly-Phe), respectively. Calculation of k1 and k2 was
performed from the intercepts using the obtained kobs and
the initial concentration of compound IV expressed as [E]0.
The overall rate constant kobs (3.6×10−3 min−1) calculated
by the sum of the obtained k1 (3.6 × 10−3 min−1) and k2
(6.9 × 10−6 min−1) was in good agreement with the kobs
(3.9 × 10−3 min−1) determined from the slope of the lin-
ear plot of the logarithm of intact ester against time demon-
strating the adequacy of the suggested degradation mech-
anism. Alternatively k1 and k2 were obtained from linear
The degradation of compound IV was subjected to gen-
eral acid–base catalysis by borate buffers and the estimated
rate constants at zero buffer concentration were obtained
from the intercepts of the linear plots of kobs versus total
buffer concentration (0.02–0.1 M) as described by Larsen
and Bundgaard (1977). General acid–base catalysis was
also observed by phosphate buffer solutions in the de-
composition of all the investigated ester prodrugs at pH
7.4 and of compound IV at pH 6.02 and 6.90. Non-linear
relationship between kobs and the total phosphate buffer
concentration, which was most pronounced at pH 7.4 are
illustrated in Fig. 3. The observed non-linear phosphate
buffer-catalytic effect is in accordance with results reported
by Jensen and Bundgaard (1991) and the buffer-independent
pseudo-first-order rate constants were estimated from these
plots of [C]/[E]0 and [I]/[E]0 against (1 − e−k t) (Larsen
obs
and Bundgaard, 1979) according to:
k1
obst
[C] =
[E]0(1 − e−k
)
(4)
kobs

S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
405
4.0E-03
3.0E-03
2.0E-03
1.0E-03
0.0E+00
0
-1
-2
-3
-4
-5
-6
-7
0
2
4
6
8
10
12
0
0.02
0.04
0.06
0.08
0.1
0.12
pH
Conc.,M
Fig. 5. The pH-rate profile for the degradation of compound IV at
37 0.5 ◦C and µ = 0.5 (kobs: observed pseudo-first-order rate constant
(min−1)). The full lines have been calculated from Eq. (6).
Fig. 3. The effect of phosphate buffer concentration on the
pseudo-first-order rate constant kobs for the degradation of compound IV
at pH 6.02 (᭿), 6.90 (᭹) and 7.38 (᭡) (37 0.5 ◦C and µ = 0.5).
buffer-independent pseudo-first-order rate constants (kobs
)
are plotted against pH.
The shape of the pH-rate profile of compound IV can be
accounted for in terms of a specific acid (kH ) and specific
plots by using the approximately linear relationship at buffer
concentrations of 0.005, 0.01 and 0.015 M.
+
Although enhanced catalytic effect of phosphate buffer
was seen as the pH changed from 6.02 to 7.38 (Fig. 3)
a plot of the catalytic rate constants kcat as a function of
−
base (kOH ) catalysed reaction of the protonated form and
specific base (kOꢀ H− ) catalysed reaction of the free base. A
spontaneous (or water-catalysed) hydrolysis of the free base
kinetically equivalent to the specific base-catalysed reaction
of the protonated form may take place. The presence of a
spontaneous or water-catalysed hydrolysis of the protonated
form has not been investigated. A similar pH-rate profile has
been reported for the degradation of glycylglycine benzyl
ester (Jensen and Bundgaard, 1991). The full lines in Fig. 5
were obtained from the expression:
2−
the fraction of HPO4 did not result in a linear relation-
ship. Likewise, non-linear relationship between kcat and
2−
the fraction of HPO4
was observed for degradation of
Phe-Pro-p-nitroaniline (Goolcharran and Borchardt, 1998)
and dipeptide model prodrugs (Lepist et al., 2000). More-
over, observations in the present study have shown that
an increase in the catalytic effect of phosphate buffer was
followed by an increase in the amount of cyclo(-Gly-Phe)
formed. Consequently, the phosphate buffer is suggested
to catalyse primarily the cyclo(-Gly-Phe) formation. Com-
pared to the other ester prodrugs the methyl ester prodrug
(compound II) was most sensitive to phosphate buffer
catalysis (Fig. 4).
+
+
aH
aH
+
+
−
−
kobs = kH aH
+ kOH aOH
Ka
+
+
aH + Ka
aH + Ka
+ kOꢀ H− aOH
(6)
−
+
aH + Ka
+
−
where aH and aOH refer to the hydrogen and hydroxide
The influence of pH on the rates of degradation of com-
pound IV is shown in Fig. 5 where the logarithms of the
ion activity, respectively. The best fit was obtained by using
kH = 3.2 × 10−3 M−1 min−1, kOH = 1950 M−1 min−1
,
+
−
kOꢀ H− = 170 M−1 min−1 and Ka = 10−7.6. The ionisation
constant, Ka, for compound IV could not determined by po-
tentiometric titration due to low alkaline solubility. However,
the pKa value for compound III was determined to 7.60
0.002 at 37 ◦C and the magnitude is in accordance with the
reported pKa values of 8.00 (25 ◦C), 7.94 (25 ◦C), 7.8 (60 ◦C)
for methyl (Purdie and Benoiton, 1973), ethyl (Meresaar and
Ågren, 1968) and benzyl (Jensen and Bundgaard, 1991) es-
ters of Gly-Gly, respectively.
1.0E-02
8.0E-03
6.0E-03
4.0E-03
2.0E-03
0.0E+00
The minor deviation between the observed and calculated
kobs values at pH 6 may arise from the complexity of the
phosphate buffer catalysis and the uncertainty related to the
determination of the buffer-independent pseudo-first-order
rate constant by the initial rate method. Complexity in
the degradation of dipeptide derivatives using phosphate
buffer solutions have also been observed by Goolcharran
and Borchardt (1998) and Lepist et al. (2000). In compar-
0
0.005
0.01
0.015
0.02
Conc., M
Fig. 4. The effect of phosphate buffer concentration on the
pseudo-first-order rate constant kobs at pH 7.4 for the degradation of com-
pound II (᭹), III (᭺), IV (᭿), V (ᮀ) and VI (᭡) (37 0.5 ◦C and µ
= 0.5).

406
S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
Table 1
Table 2
Solubilities (mean
Half-lives of the compounds at pH 7.4 and 37 0.5 ◦C in aqueous solution
at zero buffer concentration and 80% human plasma (n = 2)
standard deviation (n = 4)) in Viscoleo (SV), Vis-
coleo/castor oil 2:1 (v/v) (SV/C) and Viscoleo/DMA 9:1(v/v) (SV/DMA) at
ambient temperature
Compound
t1/2 (min)
Compound
SV
SV/C
SV/DMA
pH 7.4
80% Human plasma
Gly-Phe
III
HIP complex of III
VI
HIP complex of VI
<0.004
<0.02
0.15
<0.01
0.19
<0.004
0.32
0.80
0.07
0.65
<0.004
0.93
12.52
0.18
I
II
III
IV
Stablea
162
555
819
717
70–73
8–12
6–17
42–43
0.01
0.02
0.01
0.02
0.01
0.81
0.03
0.14
0.02
0.03
3.79
V
61–62
The hydrophobic ion pair (HIP) complexes is obtained from sodium decyl
sulfonate. The solubility of the compounds is expressed as mg Gly-Phe/ml
solution.
VI
648
108–116
No degradation after 7 h
Cyclo(-Gly-Phe)
Stablea
a
No degradation after 10 days in 0.02 M phosphate buffer pH 7.4.
ison to the other Gly-Phe esters, the methyl ester deriva-
tive (compound II) was degraded much faster at pH 7.4
(Table 1).
lipophilic prodrugs with sufficient oil solubility. Optimal sol-
ubility of these ester prodrugs in oil is expected when the
amino group at the N-terminal is uncharged. However, for
the investigated dipeptide esters it is not possible to obtain
the free base form due to intramolecular cyclisation. Thus,
in order to enhance the oil solubility ester derivatisation in
combination with hydrophobic ion pairing (HIP) may be
used. Previous studies have shown promising results in the
area of hydrophobic ion pairing for enhancement of solubil-
ity of proteins in apolar media (Meyer et al., 1995; Meyer
et al., 1996; Matsuura et al., 1993). The solubility of insulin
in octanol was for example increased from below 0.03 to
about 3 mg/ml by ion pairing with sodium dodecyl sulfonate
(Matsuura et al., 1993).
Measurements of oil solubilities of compounds III and VI,
as representatives for the dipeptide esters, were performed
for both the hydrochlorides and the HIP complexes with
sodium decyl sulfonate. The solubility studies were done in
duplicate. From each experiments two samples were with-
drawn and analysed and the results are presented in Table 2,
where the solubilities are expressed as mg Gly-Phe per ml
oil solution. Three oily solutions were used (i) Viscoleo,
(ii) Viscoleo/castor oil 2:1 (v/v) and (iii) Viscoleo/DMA 9:1
(v/v).
As to be expected the solubility of Gly-Phe in the oil so-
lutions was very low (below the limit of detection) and like-
wise low oil solubilities of the hydrochlorides of Gly-Phe
butyl and dodecyl ester were observed. The solubility of
Gly-Phe in oil solutions was enhanced by at least a factor 3
by replacement of the chloride anion with decyl sulfonate.
The solubilising enhancing effect of hydrophobic ion pair-
ing seems to be more pronounced for the long chain ester
derivative (dodecyl) compared to the more short chain ester
(butyl) of Gly-Phe.
Previously, a significant solubilising effect of castor oil
relative to that of other triglycerides including Viscoleo has
been observed (Jumaa and Müller, 2001; Larsen et al., 2002;
Riffkin et al., 1964). The presence of hydroxy groups in ri-
cinoleic acid, the dominant fatty acid component in castor
oil, might account for the solubilising effect due to their hy-
drogen bond donor capabilities (Larsen et al., 2002). Castor
oil, per se, might be less applicable due to the high viscos-
The in vitro stability data, per se, may indicate that the
esters of Gly-Phe are not to be considered as bioreversible
derivatives due to significant intramolecular cyclisation with
formation of cyclo(-Gly-Phe) in aqueous buffer solutions
at pH > 6. However, in vivo the ester derivatives might
function as prodrugs in case cleavage of the ester bond
by esterases proceeds much faster than formation of the
2,5-piperazinedione. Compared to aqueous solution pH 7.4
at zero buffer concentration decomposition studies of the
compounds in 80% human plasma pH 7.4 revealed much
faster disappearance of the ester derivatives (Table 1). Ex-
cept from compound II less than 1% cyclo(-Gly-Phe) was
formed during degradation of the dipeptide esters in 80%
human plasma reflecting fast esterase cleavage of the ester
bond or fast cleavage of the peptide bond by aminopepti-
dases since significant amount Gly-Phe was not formed. In
order to elucidate the degradation mechanism of the dipep-
tide esters in human plasma the formation of the degradation
products was followed after incubation of compound II in
80% human plasma. In comparison to the other dipeptide es-
ters the degradation of compound II resulted in formation of
15% cyclo(-Gly-Phe). The major degradation product was
Phe-O-methyl ester (approximately 78%) with less than 1%
Gly-Phe formed. Since the degradation rate of Phe-O-methyl
ester in 80% human plasma was relative slow (t1/2 = 82 min)
compared to the overall degradation rate of compound II
(t1/2 = 10 min) it is suggested that compound II in 80% hu-
man plasma is degraded by cleavage of the peptide bond and
to minor extent by the cyclisation process. Thus, most likely
the half-lives for the degradation of the esters in 80% human
plasma shown in Table 1 do not reflect the rate of esterase
cleavage. A comparison between the rate of esterase cleav-
age and the cyclisation process can, however be performed
using aminopeptidase stabilised dipeptides.
3.3. Oil solubility
Esterification of the free carboxylic acid at the C-terminal
of small active peptides may constitute a means to obtain

S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
407
Bundgaard, H., 1991. Novel chemical approaches in prodrug design.
Drugs Future 16, 443–458.
Cleland, J.L., Daugherty, A., Mrsny, R., 2001. Emerging protein delivery
methods. Curr. Opin. Biotechnol. 12, 212–219.
Connors, K.A., 1973. Reaction Mechanisms in Organic Analytical Chem-
istry. Wiley, New York, pp. 41–110.
ity (681 mPa s (Larsen et al., 2001)) resulting in a less con-
venient syringeability. It has previously been reported that
addition of castor oil to Viscoleo results in enhanced oil sol-
ubility (Jumaa and Müller, 2001; Larsen et al., 2002) and
this solubilising effect of castor oil was also observed in the
present study. However, less than 1 mg Gly-Phe was dis-
solved per ml oil composed of Viscoleo/castor oil 2:1 (v/v).
Thus, in order to obtain feasible oil solubility addition of sol-
ubilisers might be necessary. By adding 10% DMA, an oil
miscible cosolvent used in parenteral product (Spiegel and
Noseworthy, 1963; Strickley, 1999), to Viscoleo the solubil-
ity of compounds III, VI and their HIP complexes increased
by a factor of 18–84. Interestingly, the solubilising effect
of DMA was most pronounced for compound III. In the
present study the highest oil solubility of 12.5 mg Gly-Phe-
per ml was obtained for the butyl ester of Gly-Phe ion paired
with sodium decyl sulfonate dissolved in Viscoleo contain-
ing 10% DMA (Table 2).
Goolcharran, C., Borchardt, R.T., 1998. Kinetics of diketopiperazine for-
mation using model peptides. J. Pharm. Sci. 87, 283–288.
Gu, L., Strickley, R.G., 1987. Diketopiperazine formation, hydrolysis, and
epimerization of the new dipeptide angiotensin-converting enzyme
inhibitor RS-10085. Pharm. Res. 4, 392–397.
Gu, L., Strickley, R.G., 1988. Preformulation stability studies of the new
dipeptide angiotensin-converting enzyme inhibitor RS-10029. Pharm.
Res. 5, 765–771.
Jensen, E., Bundgaard, H., 1991. Peptide esters as water-soluble prodrugs
for hydroxyl containing agents: chemical stability and enzymatic hy-
drolysis of benzyl esters of glycine, diglycine and triglycine. Int. J.
Pharm. 71, 117–125.
Jumaa, M., Müller, B.W., 2001. Development of a novel parenteral formu-
lation for tetrazepam using a lipid emulsion. Drug Dev. Ind. Pharm.
27, 1115–1121.
Klixbüll, U., Bundgaard, H., 1984. Prodrugs as drug delivery systems.
XXX: 4-imidazolidinones as potential bioreversible derivatives for the
␣-aminoamide moiety in peptides. Int. J. Pharm. 20, 273–284.
Larsen, C., Bundgaard, H., 1977. Kinetics and mechanism of degradation
of mecillinam in aqueous solution. Arch. Pharm. Chem. Sci. Ed. 5,
66–86.
Larsen, C., Bundgaard, H., 1979. Hydrolysis of phensuximide. Isolation
of degradation products and development of a stability-specific HPLC
assay. Arch. Pharm. Chem. Sci. Ed. 7, 119–127.
Larsen, D.B., Fredholt, K., Larsen, C., 2001. Addition of hydrogen bond
donating excipients to oil solution: effect on in vitro drug release rate
and viscosity. Eur. J. Pharm. Sci. 13, 403–410.
Larsen, D.B., Parshad, H., Fredholt, K., Larsen, C., 2002. Characteristics
of drug substances in oily solutions. Drug release rate, partitioning
and solubility. Int. J. Pharm. 232, 107–117.
Larsen, S.W., Sidenius, M., Ankersen, M., Larsen, C., 2003. Kinetics of
degradation of 4-imidazolidinone prodrug types obtained from reacting
prilocaine with formaldehyde and acetaldehyde. Eur. J. Pharm. Sci.
20, 233–240.
Lepist, E.-I., Kusk, T., Larsen, D.H., Andersen, D., Frokjaer, S., Taub,
M.E., Veski, P., Lennernäs, H., Friedrichsen, G., Steffansen, B., 2000.
Stability and in vitro metabolism of dipeptide model prodrugs with
affinity for the oligopeptide transporter. Eur. J. Pharm. Sci. 11, 43–50.
Matsuura, J., Powers, M.E., Manning, M.C., Shefter, E., 1993. Structure
and stability of insulin dissolved in 1-octanol. J. Am. Chem. Soc.
115, 1261–1264.
Although enhanced oil solubility in this study was ob-
served for the HIP complexes between ester prodrugs of
Gly-Phe and decyl sulfonate compared to the hydrochlo-
rides, very low oil solubilities were obtained indicating that
for peptides the use of hydrophobic ion pairing may consti-
tute a less promising approach to obtain sufficient oil solu-
bilities. To evaluate the relative effect of a homologous series
of ester derivatives with increasing chain length on the oil
solubility other model peptides which are not susceptible to
intramolecular cyclisation such as tripeptides must be used.
For active dipeptides other types of bioreversible derivati-
sation, e.g. 4-imidazolidinones (Rasmussen and Bundgaard,
1991; Klixbüll and Bundgaard, 1984) and cyclisation by
linking the C- and N-terminal functionalities (Borchardt,
1999) might be used in order to increase the lipophilicity
and enhance the oil solubility. It appears likely from the pre-
liminary data of this study that in case of relatively small
peptides sufficient oil solubility requires involvement of pro-
drug derivatisation in combination with the use of solubilis-
ers like DMA.
Meresaar, U., Ågren, A., 1968. Hydrolysis of protolytic esters. VI. Al-
kaline hydrolysis and ring closure of glycylglycine ethyl ester. Acta
Pharm. Suec. 5, 85–94.
Meyer, J.D., Manning, M.C., 1998. Hydrophobic ion pairing: altering the
solubility properties of biomolecules. Pharm. Res. 15, 188–193.
Meyer, J.D., Matsuura, J.E., Kendrick, B.S., Evans, E.S., Evans, G.J.,
Manning, M.C., 1995. Solution behavior of ␣-chymotrypsin dissolved
in nonpolar organic solvents via hydrophobic ion pairing. Biopolymers
35, 451–456.
Meyer, J.D., Kendrick, B.S., Matsuura, J.E., Ruth, J.A., Bryan, P.N.,
Manning, M.C., 1996. Generation of soluble and active subtilisin and
␣-chymotrypsin in organic solvents via hydrophobic ion pairing. Int.
J. Peptide Protein Res. 47, 177–181.
Acknowledgements
We would like to thank Peter Andersen (synthesis)
and Nina Støber Vejlemand (kinetics) for excellent tech-
nical assistance. This study has been supported by The
PharmaBiotec Research Center, Copenhagen Denmark and
by The Center for Drug Design and Transport, Copenhagen
Denmark.
References
Purdie, J.E., Benoiton, N.L., 1973. Piperazinedione formation from esters
of dipeptides containing glycine, alanine, and sarcosine: the kinetics
in aqueous solution. J. Chem. Soc., Perkin Trans. II, 1845–1852.
Rasmussen, G.J., Bundgaard, H., 1991. Prodrugs of peptides. 10. Pro-
tection of di- and tripeptides against aminopeptidase by formation of
bioreversible 4-imidazolidinone derivatives. Int. J. Pharm. 71, 45–53.
Borchardt, R.T., 1999. Optimizing oral absorption of peptides using pro-
drug strategies. J. Control. Release 62, 231–238.
Bromberg, L.E., Klibanov, A.M., 1994. Detergent-enabled transport of
proteins and nucleic acids through hydrophobic solvents. Proc. Natl.
Acad. Sci. 91, 143–147.

408
S.W. Larsen et al. / European Journal of Pharmaceutical Sciences 22 (2004) 399–408
Riffkin, C., Huber, R., Keysser, C.H., 1964. Castor oil as a vehicle for
parenteral administration of steroid hormones. J. Pharm. Sci. 53, 891–
895.
Rothen-Weinhold, A., Gurny, R., 1997. Controlled and/or prolonged par-
enteral delivery of peptides from the hypothalmic pituitary axis. Eur.
J. Pharm. Biopharm. 43, 115–131.
Spiegel, A.J., Noseworthy, M.M., 1963. Use of nonaqueous solvents in
parenteral products. J. Pharm. Sci. 52, 917–927.
Strickley, R.G., 1999. Parenteral formulations of small molecules thera-
peutics marketed in the Unites States (1999)—Part I. PDA. J. Pharm.
Sci. Technol. 53, 324–349.