Reactions of 4-bis(2-chloroethyl)aminophenylacetic acid (phenylacetic acid mustard) in physiological solutions

Article abstract of DOI:10.1039/a903013a

4-Bis(2-chloroethyl)aminophenylacetic acid (phenylacetic acid mustard, 2) is the major metabolite of the cancer chemotherapeutic agent chlorambucil (1). Although its high antitumor activity and high toxicity to normal tissues have been known for a long time, no detailed chemical data on its reactions in aqueous media have been available. According to the present results 2 is decomposed in aqueous solutions by the same mechanism as other aromatic and aliphatic nitrogen mustards: an intramolecular, rate determining attack of the unprotonated nitrogen to form an aziridinium ion intermediate is followed by attack of an external nucleophile. Species 2 is considerably more stable in whole plasma than in plasma ultrafiltrate (t_1/2 at 37 °C 10 h and 37 min, respectively). Also the product distribution is completely different: while in protein-depleted plasma the only reaction is hydrolysis, in plasma the predominant reactions are non-covalent and covalent binding of 2 to albumin. The present information is important when clinical evaluation of chlorambucil, or in vitro evaluation of phenylacetic acid mustard, is performed in order to determine efficacy and bioavailability of these compounds.

Full text of DOI:10.1039/a903013a

View Article Online / Journal Homepage / Table of Contents for this issue
Reactions of 4-bis(2-chloroethyl)aminophenylacetic acid
(phenylacetic acid mustard) in physiological solutions
Toni Pettersson-Fernholm,a Juhani Vilpo,b Markus Kosonen,a Kristo Hakalaa and
Jari Hovinen*ac
a Department of Chemistry, University of Turku, FIN-20014 Turku, Finland
b Department of Clinical Chemistry, Tampere University Hospital and University of Tampere
Medical School, PO Box 2000, FIN-33521 Tampere, Finland
c Wallac Oy, PO Box 10, FIN-20101 Turku, Finland
Received (in Cambridge, UK) 15th April 1999, Accepted 19th August 1999
4-Bis(2-chloroethyl)aminophenylacetic acid (phenylacetic acid mustard, 2) is the major metabolite of the cancer
chemotherapeutic agent chlorambucil (1). Although its high antitumor activity and high toxicity to normal tissues
have been known for a long time, no detailed chemical data on its reactions in aqueous media have been available.
According to the present results 2 is decomposed in aqueous solutions by the same mechanism as other aromatic and
aliphatic nitrogen mustards: an intramolecular, rate determining attack of the unprotonated nitrogen to form an
aziridinium ion intermediate is followed by attack of an external nucleophile. Species 2 is considerably more stable in
whole plasma than in plasma ultrafiltrate (t1/2 at 37 ЊC 10 h and 37 min, respectively). Also the product distribution
is completely different: while in protein-depleted plasma the only reaction is hydrolysis, in plasma the predominant
reactions are non-covalent and covalent binding of 2 to albumin. The present information is important when clinical
evaluation of chlorambucil, or in vitro evaluation of phenylacetic acid mustard, is performed in order to determine
efficacy and bioavailability of these compounds.
since PAM is known to be considerably more toxic to normal
Introduction
tissues than the parent drug.3 The task of the present work was
4-[4-Bis(2-chloroethyl)aminophenyl]butyric acid (chloram-
to verify the kinetics and mechanism of PAM decomposition in
aqueous solutions, and clarify its reactions in human plasma.
bucil, CLB, 1) is an alkylating drug routinely used in the chemo-
therapy of chronic lymphocytic leukemia.1 Other indications
include Hodgkin’s lymphoma, non-Hodgkin’s lymphoma,
Results
Waldenström’s macroglobulinaemia, ovarian and breast cancer,
some other tumors and certain autoimmune diseases. CLB is an
orally administrated drug that is readily absorbed from the gas-
trointestinal tract giving peak plasma concentrations 1 h after
ingestion.2 CLB is metabolized predominantly by β-oxidation
of the butyric acid side chain,3 giving an intermediate (E)-4-[4-
bis(2-chloroethyl)aminophenyl]but-3-enoic acid, that is further
converted into the final metabolite 4-bis(2-chloroethyl)amino-
phenylacetic acid (phenylacetic acid mustard, PAM, 2; Scheme
1). Despite its known high anti-tumor activity and broad area
It is widely accepted that aromatic and aliphatic nitrogen mus-
tards decompose in aqueous media by a mechanism involving
an intramolecular, rate determining attack of the unprotonated
nitrogen to form an aziridinium ion intermediate followed
by an attack of an external nucleophile.7–11 To verify if this is
true also in the present case, PAM was allowed to react under
various strictly controlled conditions, and the reactions were
followed via HPLC techniques at 267 nm.
pH–rate profile
The effect of pH on the rate of PAM decomposition was
investigated at 37 ЊC at constant ionic strength (1 M, adjusted
with NaClO4) in the absence of chloride ion (Fig. 1). PAM
disintegration followed cleanly first order kinetics over the
whole pH range studied. In the absence of nucleophiles (i.e. in
cacodylic acid buffers and in solutions of perchloric acid) the
only significant reaction observed was PAM hydrolysis yielding
4-(N-chloroethyl-N-hydroxyethylamino)phenylacetic acid (3)
as the reaction intermediate and 4-bis(2-hydroxyethyl)amino-
phenylacetic acid (4) as the stable end product. When the reac-
tions were performed in formic acid and acetic acid buffers,
Scheme 1
under curve (AUC),4–6 no detailed chemical data on the reac-
tions of 2 in human fluid matrixes have been available. In order
to understand the therapeutic index of CLB it is essential to
know its metabolism and protein binding in detail, especially
J. Chem. Soc., Perkin Trans. 2, 1999, 2183–2187
This journal is © The Royal Society of Chemistry 1999
2183

View Article Online
Table 1 HPLC/MS analyses of the reactions of PAM in the presence
of 5 mM nucleophiles analyzed at t = 2 t1/2 [T = 37 ЊC, pH = 6.8, I = 1 M
(NaClO4)]. For chromatographic conditions, see Experimental
Nucleophile
SCNϪ
tR/mina
[M Ϫ 1]
Compound
18.7
22.8
23.6
28.4
30.1
32.0
18.3
18.7
21.2
23.6
32.0
238
279
256, 258
320
297, 299
274, 276
527
4
7
3
6
5
2
9
4
8
3
2
GSH
Fig. 3 End product distribution of PAM decomposition as a function
238
of thiocyanate ion concentration [T = 37 ЊC, pH = 6.8, I = 1.0 M
(NaClO4)]. Compound 4 (filled circles), compound 6 (open squares),
compound 7 (open circles).
545, 547
256, 258
274, 276
a Refers to retention time on mass detector.
The effect of glutathione and thiocyanate ion
The effect of two biologically significant nucleophiles, gluta-
thione (GSH) and thiocyanate ion, on the rate and product dis-
tribution of PAM decomposition was investigated (pH 6.8,
I = 1.0 M). In the presence of 5 mM SCNϪ, PAM was converted
into two reaction intermediates (3,5) and three stable end
products (4,6,7). The identity of 2–7 was easily confirmed on
HPLC/MS analysis (Table 1). By contrast, HPLC/MS analysis
of the reactions of PAM with GSH was more complicated. The
products of PAM hydrolysis (3,4) could be characterized as
above. Analogously, the molecular ions of the reaction inter-
mediate 8 and the stable glutathione monoadduct 10 could be
analyzed, although their molecular ions were considerably
weaker than those of the corresponding thiocyanate derivatives
(5,7). By contrast, the expected molecular ion of the gluta-
thione bis-adduct, 9, could not be detected. This is not sur-
prising, since the loss of glutamic acid residues of glutathionyl
(GS) derivatives of chlorambucil on thermospray LC/MS
analysis is well documented.12 In all likelihood, a similar
problem is also present in electron spray MS analysis.
Fig. 1 pH–rate profile of PAM decomposition at 37 ЊC and 1.0 M
ionic strength (adjusted with NaClO4). The solid line is the theoretical
curve derived from the rate law [eqn. (4)] developed in the Experimental
section.
As expected11 SCNϪ concentration had no effect on the rate
of PAM decomposition: no rate enhancement was observed
when [SCNϪ] was increased from 0 to 0.1 M, which is con-
siderably larger than the biologically relevant thiocyanate
concentration.13 By contrast, the effect of SCNϪ on the product
distribution was significant: when SCNϪ was over 0.02 M,
practically no PAM hydrolysis could be detected (Fig. 3). In
contrast to SCNϪ, GSH retarded slightly the rate of PAM
decomposition (t1/2 was 7.5 and 30 min in the presence of 0
and 0.1 M GSH, respectively). This is due to the decrease of the
pH of the reaction medium caused by the acidity of GSH;
the buffer capacity of the reaction mixture was not high enough
to keep the pH constant at high [GSH].
Reactions of PAM in physiological media
The reactions of PAM in whole human plasma and its ultrafil-
trate were investigated at 37 ЊC. In protein-depleted plasma, t1/2
of PAM decomposition was 37 min, and the only reaction
detected was hydrolysis, yielding 3 as the reaction intermediate,
and 4 as the stable end product (Fig. 4A). When the same reac-
tion was performed in whole plasma, the rate of PAM decom-
position was considerably reduced (t1/2 was 10 h). Furthermore,
there was practically no 3 and 4 in the reaction mixtures when
analyzed via HPLC techniques after protein precipitation
(Fig. 4B). Similar rate retardation and product distribution
was detected when the reactions of PAM were followed in a
cacodylic acid buffer in the presence of purified HSA (Table 2).
Fig. 2 The effect of chloride ion on the rate of PAM decomposition
[T = 37 ЊC, pH = 6.8, I = 1.0 M (NaClO4)].
additional reaction intermediates and stable end products were
formed, which were tentatively assigned as resulting from the
reactions of acetate and formate ions with the aziridinium ions
derived from 2. No catalysis by acetate or formate ions was
observed.
The chloride effect
Addition of chloride ion is known to retard the rate of decom-
position of aromatic nitrogen mustards significantly.7–11 Also
in the present case the rate of PAM decomposition was con-
siderably slower in the presence of ClϪ. The rate retardation
was dependent on chloride ion concentration in the solution
(Fig. 2).
Discussion
The proposed mechanism of PAM decomposition in aqueous
solutions is shown in Scheme 2. As in the case of other nitro-
gen mustards, the mechanism is supported by the following
2184
J. Chem. Soc., Perkin Trans. 2, 1999, 2183–2187

View Article Online
Table 2 The effect of HSA and ClϪ on the rate of PAM disintegration at 37 ЊC
Albumin/
[ClϪ]/
mM
Matrix
pH
g dmϪ3
kobs/10Ϫ5 sϪ1
Reaction observed
Protein-depleted plasma
Cacodylic acid buffer
Plasma
HSA in cacodylic acid
buffer
7.5
7.0
7.9
7.3
—
—
48.3
55.1
114
100
105
—
31.5 0.4
38.4 0.2
1.94 0.05
0.69 0.03
hydrolysis
hydrolysis
binding to proteins
binding to HSA
Fig. 4 Reversed phase HPLC traces of the reaction of PAM in human plasma ultrafiltrate (A) and human plasma (B) analyzed at t ≈ 2 t1/2. For
chromatographic conditions, see Experimental.
of the carboxylate ion of the acetic acid side chain: protonation
of the acetate ion decreases inductively the electron density at
the anilinic nitrogen and diminishes its capacity to act as a
nucleophile. Furthermore, the electrostatic stabilization of the
positively charged transition state by the carboxylate ion is lost.
Hence the formation of the aziridinium ion is less favorable
and a rate retardation is observed. A similar effect has been
reported also for CLB8 but the effect is naturally less significant
due to the longer distance of the carboxylic acid group from the
reaction center. The solid line in Fig. 1 is the best theoretical
fit of eqn. (4), which gives the dissociation constants Ka1
=
(9.46 1.67) × 10Ϫ3 M (i.e. pKa1 2.02), and Ka2 = (4.90 1.78) ×
10Ϫ5 M (i.e. pKa2 4.31), and the rate constants kdiss1 = (3.73
0.24) × 10Ϫ4 sϪ1 and kdiss2 = (6.73 0.21) × 10Ϫ4 Ϫ1. (ii) Chlor-
s
ide ion retards the rate of decomposition by attacking the
aziridinium ion derived from 2 and reforming the unchanged
PAM. As shown in Fig. 2 and eqn. (5) the rate retardation is
inversely proportional to chloride ion concentration. (iii) The
rate of decomposition is of zero order with respect to external
nucleophiles.
It is of common knowledge that approximately 98–99% of
CLB is bound to plasma proteins.14 We have shown that the
non-covalent binding is practically immediate, and albumin is
the main protein responsible for the binding in human blood.15
The binding to albumin also has a stabilizing effect on CLB.
Indeed, PAM was also found to be considerably more stable in
human plasma and in the presence of purified human serum
albumin (HSA) than in the plasma ultrafiltrate.
Albumin is the most abundant plasma protein; its concentra-
tion is normally 0.6 mM.16 It its known to provide a very high
drug binding capacity. It has different binding sites for cationic
and anionic drugs. Also van der Waals interactions between
albumin and hydrophobic drugs are important: the albumin
binding correlates with the drug hydrophobicity. Hence, the
large rate retardation of the PAM disintegration in plasma is
Scheme 2
observations. (i) The rate of decomposition is proportional to
the mole fraction of unprotonated PAM in solution. Since only
deprotonated nitrogen can form the aziridinium ion intermedi-
ate, the reactivity of the protonated PAM is insignificant: the
rate of decomposition drops sharply below the pKa of the
nitrogen atom (Fig. 1, for development of the rate law, see
Experimental). Furthermore, the rate levels off when the mole
fraction of the deprotonated PAM reaches 1. The additional
inflection at ca. pH 4 can be shown to result from the protolysis
J. Chem. Soc., Perkin Trans. 2, 1999, 2183–2187
2185

View Article Online
easily explained by its non-covalent binding to the hydro-
phobic pockets of albumin: in the absence of bulk water the
hydrolysis is naturally less prominent. Furthermore, the hydro-
phobic environment makes the formation of the aziridinium
ion intermediates less favorable. Similar effects, though less
significant, have been reported for CLB stabilization in the
presence of cyclodextrins17 and micelles.10
Drug binding to albumin is normally instantaneous and
reversible. In accordance with our previous findings with
CLB,15 covalent drug binding is also observed. These inter-
actions between PAM and albumin may explain the inter- and
intra-individual differences in the pharmacokinetics and bio-
availability of orally administered CLB, as shown previously.18
Molecular details of these interactions are currently being
explored in our laboratories.
d[Itot]
dt
(1A)
= 0
⇒ kdiss1[SH] ϩ kdiss2[SϪ] Ϫ kass1[IHϩ][ClϪ] Ϫ
kass2[I][ClϪ] Ϫ k1[IHϩ][Nu] Ϫ k2[I][Nu] = 0 (1B)
where
[Itot] = [IHϩ] ϩ [I]
[Itot]
[Itot]
[Hϩ]
[IHϩ] =
; [I] =
Ka3
1 ϩ
1 ϩ
Hϩ
[Ka3]
[Stot] = [SH2ϩ] ϩ [SH] ϩ [SϪ]
[Stot]
[Stot]
[Hϩ]
Experimental
General
[SϪ] =
; [SH] =
[Hϩ]
[Hϩ]2
Ka2
[Hϩ]
1 ϩ
ϩ
1 ϩ
ϩ
Ka2
Ka1Ka2
Ka1
4-Bis(2-chloroethyl)aminophenylacetic acid (2) was synthesized
as described previously.19 All inorganic reagents were of
analytical grade. GSH was purchased from Fluka and human
serum albumin (99%; globulin and fatty acid free) from Sigma.
Plasma and protein-depleted plasma were obtained as described
previously.15 Albumin concentrations were measured immuno-
turbinometrically on a Cobas Integra 700 instrument.
and Ka1, Ka2 and Ka3 are the equilibrium constants.
[SHϩ][Hϩ]
[SϪ][Hϩ]
[SHϩ]
[I][Hϩ]
[IHϩ]
Ka1
=
; Ka2
=
; Ka3 =
ϩ
[SH2
]
Substituting [SϪ], [SH], [IHϩ] and [I] into eqn. (1B), the total
concentration of the reaction intermediates is given by eqn. (2).
HPLC analyses
kdiss1
[Hϩ]
Ka1
kdiss2
[Hϩ]
ϩ
[Hϩ]2
These were performed on a Waters 2000 instrument consisting
of a UV-detector (λ = 267 nm) and a reversed phase column
(Hypersil C18, 4.6 × 240 mm, particle size 6 µm). Mobile
phase: Buffer A = 0.1 M ammonium acetate, Buffer B = 0.1 M
ammonium acetate in 50% (v/v) aq. acetonitrile. The following
solvent systems were used as the eluents. System 1: isocratic
elution of 50% A (pH rate profile, the effect of ClϪ, SCNϪ and
GSH on the rate of PAM decomposition). System 2: Gradient:
from 0 to 10 min, 100% A; from 10 to 40 min, a linear gradi-
ent from 100% A to 100% B (product analyses, reactions of
PAM in human fluid matrixes and in the presence of purified
Ka2
1 ϩ
ϩ
1 ϩ
ϩ
[Hϩ]
Ka2
Ka1Ka2
[Itot] =
[Stot] (2)
k1[Nu] ϩ kass1[ClϪ] k2[Nu] ϩ kass2[ClϪ]
ϩ
Ka3
[Hϩ]
[Hϩ]
1 ϩ
1 ϩ
Ka3
The observed rate constant is then given by eqn. (3), where [Itot]
dP
ͫ ͬ= (k1[IHϩ] ϩ k2[I])[Nu] =
dt
HSA). In all the cases, flow rate was 1.0 ml minϪ1
.
k1[Nu]
k2[Nu]
[Hϩ]




ϩ
[Itot] (3)
Ka
3
Kinetic measurements
1 ϩ

1 ϩ
[Hϩ]
K
a3
These were performed as described previously,13 except in the
case of reactions in the presence of proteins, which were pre-
cipitated with a known amount of ethanol prior to HPLC
analyses.
is given in eqn. (2). If [ClϪ] = 0 (or in general k1,2[Nu] ӷ kass1,2
[ClϪ]), then eqn. (3) is simplified to eqn. (4) at constant pH and
d[P]
dt
kdiss1
[Hϩ]
kdiss2
[Hϩ]
Qualitative product analyses
=
ϩ
(4)
Ka2
[Hϩ]
[Hϩ]2
1 ϩ
ϩ
1 ϩ
ϩ
These were performed on a Perkin-Elmer Sciex API 365LC/
MS/MS triple quadrupole mass spectrometer in negative detec-
tion mode (HPLC System 2; 0.01 M buffer concentration).
Reaction mixtures containing 0.005 M GSH or SCNϪ and 0.1
mM PAM were analyzed at t = 2t1/2 (pH 6.8; T = 37 ЊC).
Ka1
Ka2
Ka1Ka2
[Nu] eqn. (3) is simplified to eqn. (5), i.e. the rate constant is
1
linearly dependent on
.
[ClϪ]
Quantitative product analyses
dP
constant1
These were performed via HPLC at λ = 267 nm. This wave-
length was chosen since it is the isosbestic point of the reactions
of PAM with non-chromophoric nucleophiles, as judged on UV
spectrophotometry. At this wavelength all reaction intermedi-
ates and end products have the same ε, and the mole fractions
of each reaction component are proportional to the integrals of
their HPLC signals. For details, see ref. 11.
=
(5)
dt constant2 ϩ constant3[ClϪ]
1
⇒
∝ [ClϪ]
kdiss
Acknowledgements
Financial support from the Turku University Foundation, the
Medical Research Fund of Tampere University Hospital and
the Finnish Cancer Organization is gratefully acknowledged.
The authors wish to thank Leiras Oy (Turku, Finland) for the
generous gift of the HPLC instrument.
Effect of pH and nucleophiles on the rate of PAM hydrolysis
If the intermediates I and IHϩ are assumed to be very unstable
and their concentrations reach steady state, then at steady state
eqn. (1) applies,
2186
J. Chem. Soc., Perkin Trans. 2, 1999, 2183–2187

View Article Online
10 P. M. Cullis, R. E. Green and M. E. Malone, J. Chem. Soc., Perkin
References
Trans. 2, 1995, 1503.
1 C. B. Faguet, J. Clin. Oncol., 1994, 12, 974.
2 W. B. Pratt, R. W. Ruddon, W. D. Ensminger and J. Maybaum, The
Anticancer Drugs, 2nd edn., Oxford University Press, New York,
1994, ch. 6.
3 A. McLean, A. D. Newell, G. Baker and T. Connors, Biochem.
Pharmacol., 1980, 29, 2039.
4 N. H. Greig, E. M. Daly, D. J. Sweeney and S. I. Rapaport, Cancer
Chemother. Pharmacol., 1990, 25, 320.
5 M. M. Oppitz, E. Musch, H. P Malek, H. Rüb, G. E. von Unruh,
U. Loos and B. Mühlenbruch, Cancer Chemother. Pharmacol., 1989,
23, 208.
6 L. Bastholt, C.-J. Johansson and P. A. Pfeiffer, Cancer Chemother.
Pharmacol., 1991, 28, 205.
7 W. R. Owen and P. J. Stewart, J. Pharm. Sci., 1979, 68, 992.
8 D. C. Chatterji, R. L. Yager and J. F. Gallelli, J. Pharm. Sci., 1982,
71, 50.
11 J. Hovinen, T. Pettersson-Fernholm, M. Lahti and J. Vilpo, Chem.
Res. Toxicol., 1998, 11, 1377.
12 D. M. Dulik, O. M. Colvin and C. Fenselau, Biomed. Environ. Mass
Spectrom., 1990, 19, 248.
13 J. Hovinen, R. Silvennoinen and J. Vilpo, Chem. Res. Toxicol., 1998,
11, 91.
14 Goodman & Gilman’s The Pharmacological Basis of Therapeutics,
9th edn., ed. J. G. Hardman and L. E. Limbird, McGraw-Hill,
New York, 1995.
15 T. Peters, Jr., All About Albumin, Academic Press, San Diego, 1996.
16 K. Löf, J. Hovinen, P. Reinikainen, L. M. Vilpo, E. Seppälä and
J. A. Vilpo, Chem.-Biol. Interact., 1997, 103, 187.
17 A. R. Green and J. K. Guillory, J. Pharm. Sci., 1989, 78, 427.
18 M. J. Lind and C. Ardiet, Cancer Surv., 1993, 17, 157.
19 J. Hovinen, Acta Chem. Scand., 1996, 50, 1174.
9 G. C. Kundu, J. R. Schullek and I. R. Wilson, Pharmacol. Biochem.
Behav., 1994, 49, 621.
Paper 9/03013A
J. Chem. Soc., Perkin Trans. 2, 1999, 2183–2187
2187