Acid-base profiling of imatinib (Gleevec) and its fragments

Article abstract of DOI:10.1021/jm049546c

The site-specific basicities of imatinib (Gleevec, a new signal transduction inhibitor drug of chronic myeloid leukemia) and two of its fragment compounds were quantitated in terms of protonation macroconstants, microconstants, and group constants by NMR-pH and pH-potentiometric titrations. Sequential protonation of imatinib follows the N³⁴, N¹¹, N³¹, N¹³ order, in which N¹¹ and N³¹ show commensurable basicity, but negligible intramolecular interaction. Fragment compounds include two "halves" of imatinib, and their moiety-specific basicities confirm the NMR-based protonation sequence of the parent compound. NMR-pH profiles, macro- and/or microscopic protonation schemes, and species-specific distribution diagrams are presented. On the basis of these data, imatinib is shown to be predominantly neutral, monocationic, and tricationic at intestinal, blood, and gastric pH, respectively. The molecular hypotheses on imatinib binding to the Bcr-Abl oncogene fusion protein are interpreted at the site-specific level in view of the moiety basicities of imatinib.

Full text of DOI:10.1021/jm049546c

J. Med. Chem. 2005, 48, 249-255
249
Acid-Base Profiling of Imatinib (Gleevec) and Its Fragments
Zolta´n Szaka´cs,† Szabolcs Be´ni,‡ Zolta´n Varga,§ La´szlo´ O¨ rfi,‡ Gyo¨rgy Ke´ri,§,| and Be´la Nosza´l*,‡
Department of Inorganic and Analytical Chemistry, Lora´nd Eo¨tvo¨s University, Pa´zma´ny Pe´ter se´ta´ny 1/a,
H-1117 Budapest, Hungary, Department of Pharmaceutical Chemistry, Semmelweis University, Ho¨gyes E.u. 9,
H-1092 Budapest, Hungary, Cooperative Research Center, Semmelweis University, POB 131,
H-1367 Budapest, Hungary, and Peptide Research Group of the Hungarian Academy of Sciences,
Department of Medical Chemistry, Semmelweis University, Puskin u. 9, H-1088 Budapest, Hungary
Received June 9, 2004
The site-specific basicities of imatinib (Gleevec, a new signal transduction inhibitor drug of
chronic myeloid leukemia) and two of its fragment compounds were quantitated in terms of
protonation macroconstants, microconstants, and group constants by NMR-pH and pH-
potentiometric titrations. Sequential protonation of imatinib follows the N34, N11, N31, N13 order,
in which N11 and N31 show commensurable basicity, but negligible intramolecular interaction.
Fragment compounds include two “halves” of imatinib, and their moiety-specific basicities
confirm the NMR-based protonation sequence of the parent compound. NMR-pH profiles,
macro- and/or microscopic protonation schemes, and species-specific distribution diagrams are
presented. On the basis of these data, imatinib is shown to be predominantly neutral,
monocationic, and tricationic at intestinal, blood, and gastric pH, respectively. The molecular
hypotheses on imatinib binding to the Bcr-Abl oncogene fusion protein are interpreted at the
site-specific level in view of the moiety basicities of imatinib.
Introduction
imatinib, we combined pH-potentiometry with NMR-
pH titrations. The same methodology plus UV-pH
titrations were applied to characterize two fragment
compounds, 4-methyl-N-3-(4-pyridin-3-yl-pyrimidin-2-
yl)-benzene-1,3-diamine 2 and 4-(4-methyl-piperazin-1-
ylmethyl)-benzamide 3 (for structures and numbering,
see Figure 1). 2 is also known to be a pharmaceutical
impurity of Gleevec.7 Potentiometric titrations provided
the protonation constants in the range 2 < log Ki < 12.
To define the loci for the potentiometric results and to
determine the protonation constants in highly acidic pH,
1H NMR-pH titrations were carried out for all ligands.
Analysis of the NMR-pH profiles allowed a site-specific
insight into the alternative protonation pathways in 1
and 2. In particular, the intrinsic basicity of the pyridyl
and piperazinyl nitrogens of 2 have been characterized
in terms of microscopic protonation constants, while for
imatinib 1 the main route of protonation could be
identified.
Imatinib mesylate (Gleevec, 4-(4-methyl-piperazin-1-
ylmethyl)-N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-
ylamino)-phenyl]-benzamide methanesulfonate) is in-
dicated for the treatment of chronic myeloid leukemia
(CML) and unresectable and/or metastatic malignant
gastrointestinal stromal tumors (GIST).1 CML is a
clonal hematopoietic stem cell disorder characterized by
the presence of the Philadelphia chromosome. A hall-
mark of this leukemia is a reciprocal chromosomal
translocation involving the bcr gene from chromosome
9 and the c-abl gene from chromosome 22. The resulting
Bcr-Abl fusion proteins have elevated nonreceptor ty-
rosine kinase catalytic activity, causing cell transforma-
tions and the concomitant malignancy.2 The most
potent, selective inhibitor of Bcr-Abl is imatinib, signal
transduction inhibitor 571 (STI 571). This ATP mimic
drug has a high affinity for Abl kinase, while being
essentially inactive against Ser/Thr-kinases and most
of the tyrosine kinases.3,4
In spite of its enormous therapeutic importance, the
main physicochemical properties of imatinib have not
been reported. The only data on its acid-base properties
are based on pH-dependent capillary zone electrophore-
sis,5 claiming that protonation of the neutral imatinib
1 starts below pH 5, but providing no quantitative
parameters. Considering the protonation constants of
N-methylpiperazine6 (log K ) 9.85), the lack of proton
binding of 1 above pH 5 is highly unlikely.
Results and Discussion
Potentiometric Titrations. Potentiometric titration
curves were transformed into Bjerrum plots (Figure 2),
which show the mean number of protons associated with
one ligand molecule at any arbitrary pH. In the pH
interval 2 through 12, three protonation steps are
observed for imatinib 1, while both 2 and 3 bind two
protons. The resulting protonation constants are sum-
marized in Table 1. The outstanding uncertainty of log
K1 ) 7.7 of 1 is due to ligand precipitation above pH
6.5 even at 1 mM concentration. Aqueous solvent
mixtures such as 65% (w/w) DMSO or 40% (w/w)
methanol eliminate the solubility problem, but they
heavily modify the equilibria. Similar precipitation
occurred for 2 at concentrations >3 mM above pH 7.
The fact that a log K1 value about 8 has been determined
for both 1 and 3 but not for 2 confirms that the N-methyl
Here we report the site-specific acid-base properties
of imatinib. In order to quantitate the basicity of
* Corresponding author. Phone: (36) 1 2170891. Fax: (36) 1
2170891. E-mail: nosbel@hogyes.sote.hu.
† Lora´nd Eo¨tvo¨s University.
‡ Department of Pharmaceutical Chemistry, Semmelweis Univer-
sity.
§ Cooperative Research Center, Semmelweis University.
| Department of Medical Chemistry, Semmelweis University.
10.1021/jm049546c CCC: $30.25 © 2005 American Chemical Society
Published on Web 12/13/2004

250 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1
Szaka´cs et al.
Figure 1. Structure and numbering of the compounds studied (for systematic names, see text). For comparison, the numbering
of imatinib 1 is retained for the corresponding atoms of 2 and 3.
Figure 3. 1H NMR-pH titration curves of 3, with computer
fits in solid lines.
Figure 2. Bjerrum plots from the potentiometric titrations
of 1, 2, and 3. Computer fits are shown in solid lines.
The pH-dependent chemical shifts of all carbon-bound
Table 1. Macroscopic Protonation Constants in H2O at 25 °C, I
) 0.15 M (NaCl), Determined by Potentiometric (and in Part by
1H NMR) Titrations
protons are depicted in Figure 3. Since protonation
processes are rapid on the NMR time scale, the observed
chemical shifts (δobs) are weighted averages8 of those of
compd
log K1
log K2
log K3
log K4
+
2+
the distinct protonation forms (δL, δHL , and δH L ):
2
1
2
3
7.7 ( 0.1a 3.88 ( 0.03
3.10 ( 0.01 1.71b ( 0.02
obs ) δLxL + δHL+xHL+ + δ
x
(1)
4.75 ( 0.01 3.72 ( 0.01 1.16b ( 0.03
7.95 ( 0.01 3.37 ( 0.02
δ
H2L2+ H2L2+
a Uncertainties are estimates of standard deviation (1σ values).
Expressing the x mole fractions in terms of pH and
logarithms of K protonation constants, the master equa-
tion to fit the NMR-pH titration curves is obtained:
b Determined by 1H NMR-pH titration.
piperazinyl nitrogen is the binding site in the weakly
basic pH range.
δL + δHL+(10logK -pH) + δ
(10logK +logK -2pH
)
1
1
2
The protonation at pH < 2 cannot be accurately
studied by potentiometric titrations due to experimental
difficulties involving the large buffering capacity of
solvent water and the non-Nernstian response of the
glass electrode. However, NMR titrations do not suffer
from these limitations and reveal an additional proto-
nation step at pH < 2 for both 1 and 2 (see below).
NMR-pH Titration of 3. The aromatic part of the
1H NMR spectrum of 3 contains the characteristic
AA′BB′ multiplet of the para-disubstituted phenyl ring.
The nonequivalent amide protons give rise to two broad
singlets near 8.1 and 7.3 ppm at pH < 7. These signals
disappear in alkaline solutions as the rate of base-
catalyzed amide proton exchange increases. In the
aliphatic region, the piperazinyl methylenes show two
exchange-broadened peaks due to the conformational
dynamics of the piperazine ring, which collapse to a
single, broad resonance at pH extremes.
H2L2+
δobs
)
1 + 10logK -pH + 10logK +logK -2pH
1
1
2
(2)
Equation 2 holds separately for each nucleus. The log
K values obtained by potentiometry are used to deter-
+
2+
mine the individual chemical shifts δL, δHL , and δH L
2
by parameter estimation, and the results are collected
in Table S1 of the Supporting Information. The proto-
nation shifts ∆1δ ) δHL - δL and ∆2δ ) δH L - δHL
+
2+
+
2
0
1
hold information on the protonation sequence of the
molecule.
All NMR-pH titration curves exhibit inflections near
the log K values. The protonation shifts in 1H NMR
spectroscopy are greatest for nuclei in the vicinity of the
basic site.9 Upon coordination of the first proton to N34
with log K1 ) 7.95, the greatest effects are observed at
the piperazinyl H33/H35 nuclei (0.76 ppm) and the

Acid-Base Profiling of Imatinib and Its Fragments
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1 251
Figure 5. Macroscopic and microscopic protonation scheme
of 2. P, G, and N represent the pyridyl N11, the guanidine N13,
and the amine N21 nitrogens, while K and k denote macroscopic
and microscopic protonation constants, respectively.
Figure 4. 1H NMR-pH titration curves of 2, with computer
fits in solid lines.
N-methyl H37 (∆01δ ) 0.63 ppm). Protonation of N31
with log K2 ) 3.37 is monitored most sensitively by the
nearby H30 and H32/H36 methylenes, with downfield
shifts of ∆21δ ) 0.81 and 0.87 ppm, respectively.
NMR-pH Titration of 2. The multiplets in the
aromatic part of the 1H NMR spectrum of 2 exhibit
mostly first-order pattern even at 250 MHz. Figure 4
shows the chemical shifts of the carbon-bound protons
as functions of pH. The NMR-pH datasets were fitted
by the following, triprotic extension of eq 2:
protonation fraction of the pyridyl nitrogen can be
obtained, which enables the calculation of kP, one of the
unknown microconstants, by fitting the following func-
tion:
δoHb9s - δL,H9
P
10logk -pH + 10logK +logK -2pH
1
2
fP )
)
1 + 10logK -pH + 10logK +logK -2pH
1
1
2
δ
- δL,H9
H2L,H9
(4)
Thus, a log kP ) 4.07 value was obtained. Since the
K macroconstants are known from potentiometry, the
knowledge of log kP allows the calculation of the other
three microconstants (Table 2), using the relationships
between micro- and macroconstants.9,10 Microconstants
differing by 0.01 unit were obtained when the evaluation
was repeated for nuclei H8 or H10.
As a further check of the consistency of the micro-
constant set, similar calculations were carried out to
obtain kN from the NMR titration curve of each “amine
reporter” nuclei H16, H17, and H19, resulting in log kN
values of 4.64, 4.65, and 4.71, respectively.
1
1
2
δ
obs ) [δL + δHL+(10logK -pH) + δH L (10logK +logK -2pH) +
2 2+
δH L3+(10logK +logK +logK -3pH)]/[1 + 10logK -pH
+
1
2
3
1
3
10logK +logK -2pH + 10logK +logK +logK -3pH] (3)
1
2
1
2
3
This calculation yielded the chemical shifts of the
individual HiLi+ species (Table S2, Supporting Informa-
tion) and the value of the third protonation constant log
K3 ) 1.16, which could not be obtained by potentiometry.
All NMR-pH profiles in Figure 4 show two steps. The
large downfield shift occurs in the pH interval 6 to 2.3
and reflects the coordination of the first two protons.
The ∆δ protonation shifts in Table S2 hold useful
information to identify these two protonation sites. The
first protonation occurs mainly at the N21 amine group,
as H19, H17, and H16 of the phenyl ring are most affected.
The largest ∆21δ values are observed at H8, H9, H10, and
H12, indicating the protonation of the pyridyl nitrogen
Interaction between sites N and P can be quantified
in terms of log kP - log kNP ) log kN - log kNP, the
interactivity parameter. Its 0.25 value in log units
indicates that these sites are in weak interactions in 2,
which is, however, irrelevant in the parent compound
where N21 exists in amide bond, bearing thus no basicity
in this pH range.
1
NMR-pH Titration of Imatinib 1. The H NMR
N11. Although the main protonation pathway is N21
f
titration of 3 mM imatinib 1 was restricted to the pH <
6 range to avoid precipitation. The piperazine methyl-
enes appear in broad signals between 3 and 4 ppm, so
these peaks were omitted from the analysis. The chemi-
cal shifts of other carbon-bound hydrogens are plotted
as functions of pH in Figure 6. By evaluation of these
curves, the first protonation step of 1 was disregarded
and the coordination of three protons to HL+ was
considered. Using K2 and K3 from potentiometry, eq 3
was fitted to all NMR-pH datasets simultaneously to
obtain log K4 ) 1.71 and the chemical shifts of the
individual HiL species (Table S3, Supporting Informa-
tion). The ∆δ protonation shifts help to identify the
protonation pattern of 1.
N11, the alternative N11 f N21 protonation pathway
cannot be neglected either. Because of overlap in the
two protonation steps, log K1 and log K2 are not direct
measures of the pyridyl (P) and amine (N) basicity. The
exact, site-specific characterization takes the micro-
scopic protonation scheme (Figure 5). This treatment
distinguishes between the protonation isomers (mi-
crospecies) denoted by P and N that have the same
overall composition HL+. The four microscopic proto-
nation constants, kN, kP, kPN, kPN, characterize the basic-
ity of the P and N sites directly. In order to determine
these constants, “reporter” nuclei were chosen to selec-
tively monitor the protonation of each group through
their chemical shifts. The aromatic protons of the
pyridyl ring are separated by more than 9 bonds from
the N21 amine group, so any of them can well be
assumed to be a selective indicator of the N11 protona-
tion. Using the chemical shift-pH profile of H9, the fP
Figure 7 shows the protonation scheme of 1. Once the
first proton to the piperazinyl N34 (labeled as group A)
is attached, two alternative pathways occur, leading
either to the N11-protonated microspecies PA or to the
N31-protonated one, labeled here as AB. To determine

252 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1
Szaka´cs et al.
Table 2. Microscopic Protonation Constants of 1 and 2, Determined by 1H NMR-pH Titration (T ) 25 °C, I ) 0.15 M with NaCl,
Solvent H2O)a
microconstants
N11
N31 in 1 and N21 in 2
N13
compd
log kPA
log kPAB
log kP
log kPN
log kBA
log kBPA
log kN
log kNP
log kGPAB
log kGPN
1
3.78 ( 0.02
3.78 ( 0.02
3.20 ( 0.02
1.71 ( 0.02
1.16 ( 0.03
3.20 ( 0.02
2
4.07 ( 0.04
3.82 ( 0.02
4.65 ( 0.01
4.40 ( 0.04
a See Figures 5 and 6 to assign microconstants.
Figure 6. 1H NMR-pH titration curves of imatinib 1, with computer fits in solid lines.
Figure 7. Macroscopic and microscopic protonation scheme of imatinib 1. P, G, A, and B represent the pyridyl N11, the guanidine
N13, and the piperazine N34 and N31 nitrogens, while K and k denote macroscopic and microscopic protonation constants, respectively.
the microconstant kAP, H10 was chosen as protonation
sensor for the pyridyl N11 (labeled as P). With the
calculation outlined above, the microconstants in Table
2 were obtained and similar results are gained from the
NMR-pH titration curve of H12. The microconstants
indicate that the pyridyl nitrogen is 3.8 times more basic
than N31. Two pairs of microconstants, referring to the
same group, are equal. This fact indicates that the
protonation of one site does not diminish appreciably
the basicity of the other one, which is, in principle, an
example of the group constant approximation.10 This
independence of groups indicates not only the long
covalent distance between N11 and N31 but also the
rigidity of imatinib, since flexible molecules of similar
intermoiety distance show 0.2-0.3 interactivity in log
k units.12
The last protonation of 1 occurs at N13. Accordingly,
the corresponding microconstant log kGPAB practically
equals log K4 ) 1.71. The value of this constant clearly
shows that this moiety is a 2-amino-pyrimidine struc-
ture rather than formally resembling a guanidino unit.
Figure 8. Distribution curves of the macrospecies (HiL) and
microspecies (PA and AB) of imatinib 1.
The macroscopic and microscopic protonation con-
stants enable the calculation of the species distribution
of 1 (Figure 8) in compartments of various pH of the
human body. In the blood (pH 7.4), 33% of imatinib is
monocationic (HL+), holding the single proton on the
piperazine A site. At the duodenal pH of 4.6, 80% of

Acid-Base Profiling of Imatinib and Its Fragments
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1 253
) 0.1 (chloroform:methanol ) 10:1); retention times: 2.54 min
(HPLC1) and 2.45 min (HPLC2). MS: (M + H)+ ) 234. 1H
NMR (DMSO-d6, 300 MHz) δ (ppm): 7.91 (bs, 1H, NH2R), 7.81
imatinib is monocationic, while the remaining fraction
is present in the form of two dicationic (H2L2+) proto-
nation isomers. In the stomach (pH 2), the tricationic
H3L3+ (63%) and the fully protonated H4L4+ (33%) are
the most abundant species.
3
3
(d, 2H, J ) 8.2 Hz), 7.35 (d, 2H, J ) 8.1 Hz), 7.30 (bs, 1H,
NH2â), 3.48 (s, 2H, CH2), 2.34 (bs, 8H), 2.14 (s, 3H, CH3).
4-Methyl-N-3-(4-pyridin-3-yl-pyrimidin-2-yl)-benzene-
1,3-diamine (2). 3-Acetylpyridine (64.1 cm3, 583 mmol) and
94.0 cm3 (700 mmol) of N,N-dimethyl-formamide dimethyl-
acetal were refluxed in 250 cm3 of ethyl alcohol overnight. The
reaction mixture was evaporated under reduced pressure, 100
cm3 of diethyl ether was added to the residue, and the reaction
mixture was cooled to 0 °C. 3-Dimethylamino-1-pyridin-3-yl-
propenone (71.2 g) was filtered off as yellow crystals. (Yield:
69%.) This material was used in the subsequent steps without
further purification.
2-Methyl-5-nitro aniline (22.8 g, 150 mmol) was dissolved
in 60 cm3 of ethyl alcohol, and 10.5 cm3 of concentrated HNO3
was added to the solution dropwise followed by 25.2 cm3 (300
mmol) of a 50% aqueous solution of cyanamide. The reaction
mixture was refluxed overnight and then cooled to 0 °C. N-(2-
Methyl-5-nitro-phenyl)-guanidinium nitrate (20.6 g) was fil-
tered off and washed well with ethyl acetate and diethyl ether.
(Yield: 53%.)
To a suspension of 26.4 g (150 mmol) of 3-dimethylamino-
1-pyridin-3-yl-propenone and 40.8 g (157.5 mmol) of N-(2-
methyl-5-nitro-phenyl)-guanidinium nitrate in 220 cm3 of
2-propanol was added 6.95 g (175 mmol) of NaOH. The
reaction mixture was refluxed for 24 h. Then it was cooled to
0 °C. The precipitate was filtered off, suspended in water,
filtered off again, and finally washed with 2-propanol and
diethyl ether. This reaction gave 30.1 g of (2-methyl-5-nitro-
phenyl)-(4-pyridin-3-yl-pyrimidin-2-yl)-amine. (Yield: 65%.)
Conclusion
Several hypotheses were reported on the possible
electrostatic and H-bonding contact points between
imatinib and the complementary Bcr-Abl oncogene.2-4
In the model of Corbin et al.,3 a hydrogen bond is
assumed between N34 of imatinib and Glu258 in the ATP
binding pocket of the active form of the Abl kinase
domain. This interaction is supported by the protonation
constant of N34 (log K ) 7.7), which makes N34 the most
basic site of imatinib and facilitates its proton-donor role
in such bindings. In the binding pattern, the role of the
pyridyl N11 atom is barely understood. In the crystal
structure of the Abl catalytic domain with a variant of
imatinib, N11 was described2 as an acceptor in its
unprotonated form for an H-bond with the Met318
peptide group in the unprotonated form. Similarly, the
guanidino N13 atom is assumed2,4 to be a proton acceptor
in an H-bond from Thr315. The condition of this H-bond,
the unprotonated state of N13, is certainly met in view
of our log K ) 1.71 data, indicating that N13 is basic,
but largely unprotonated under all physiological cir-
cumstances, accommodating thus proton donor moieties
of counter molecules.
A solution of 42 g (186 mmol) of SnCl2‚2 H2O in 115 cm3 of
concentrated hydrochloric acid was added to 12.3 g (40 mmol)
of (2-methyl-5-nitro-phenyl)-(4-pyridin-3-yl-pyrimidin-2-yl)-
amine while the suspension was vigorously stirred. After 30
min of stirring the mixture was poured onto crushed ice, made
alkaline with K2CO3, and extracted three times with 100 cm3
of ethyl acetate. Organic phases were combined, dried over
MgSO4, and evaporated to dryness. Recrystallization from
dichloromethane resulted in 8.9 g of 2 (yield: 81%) as an off-
white solid. Anal. (C15H15N5) C; N calcd, 5.45; found, 6.16; N
calcd, 25.25; found, 19.48. Mp: 134-136 °C. Rf ) 0.5 (chloro-
form:methanol ) 10:1), retention times: 3.64 min (HPLC1)
and 4.62 min (HPLC2). MS: (M + H)+ ) 278. 1H NMR (DMSO-
d6, 300 MHz) δ (ppm): 9.22 (s, 1H, NH), 8.67 (d, 1H, 3J ) 4.4
Experimental Section
Materials. The free bases of 1, 2, and 3 were synthesized
as detailed in the following section. Other base chemicals of
analytical grade were purchased from commercial suppliers
and used without further purification. All solutions were
prepared from bidistilled Millipore water (conductivity: 1.1
µS cm-1).
Synthetic Chemistry. General Experimental Proce-
dures. All reagents and solvents were obtained from Sigma-
Aldrich Co. and used without purification and drying. Melting
points were determined with a MEL-TEMP capillary melting
point apparatus and are uncorrected. Synthesis-supporting 1H
NMR spectra were recorded using a Bruker Avance DRX 300
MHz spectrometer. Chemical shifts are given in parts per
million relative to TMS. Electrospray ionization mass spectra
were recorded on a Waters 2795 HPLC, equipped with a
Waters 996 photodiode array detector and a Micromass ZMD
2000 LC-MS system. Retention factors were determined on
silica gel GF254 TLC plates eluting with mixture of chloroform/
methanol 10/1. The purity of the compounds was also checked
in two RP-HPLC systems, using an acetonitrile/water gradient
(HPLC1) and a methanol/water gradient (HPLC2); other
relevant parameters are listed in Table S5 of the Supporting
Information).
4-(4-Methyl-piperazin-1-ylmethyl)-benzamide (3). 4-
Chloromethyl-benzoyl chloride (378 mg, 2 mmol) was dissolved
in 50 cm3 of tetrahydrofuran and cooled to 0 °C. Concentrated
ammonia solution (0.27 cm3, 4 mmol) was added slowly to the
cold solution. The reaction mixture was vigorously stirred for
10 min at 0 °C, and then the mixture was poured on 50 cm3 of
ice water. The aqueous layer was extracted three times with
50 cm3 of ethyl acetate. The combined organic layers were
dried over magnesium sulfate and evaporated to dryness.
Acetonitrile (20 cm3) and 1.11 cm3 (10 mmol) of N-methylpip-
erazine were added to the residue, and the reaction mixture
was refluxed for 4 h. Then the reaction mixture was cooled to
0 °C. Crystalline 3 (292 mg, 63%) was filtered off after 1 h.
Anal. (C13H19N3O) C calcd, 66.92; found, 64.84; H calcd, 8.21;
found, 7.55; N calcd, 18.01; found, 17.34. Mp: 156-159 °C. Rf
3
3
Hz), 8.63 (s, 1H), 8.44 (d, 1H, J ) 5.1 Hz), 8.39 (d, 1H, J )
3
4
8.0 Hz), 7.51 (dd, 1H, J ) 7.9 Hz, J ) 4.8 Hz), 6.85 (d, 1H,
3J ) 8.0 Hz), 6.77 (d, 1H, 4J ) 1.4 Hz), 6.32 (dd, 1H, 3J ) 8.0
4
Hz, J ) 2.0 Hz), 4.80 (s, 2H, NH2), 2.04 (s, 3H, CH3).
4-(4-Methyl-piperazin-1-ylmethyl)-N-[4-methyl-3-(4-py-
ridin-3-yl-pyrimidin-2-ylamino)-phenyl]-benzamide (1).
4-Methyl-N-3-(4-pyridin-3-yl-pyrimidin-2-yl)-benzene-1,3-di-
amine (3.3 g, 12 mmol) was dissolved in 150 cm3 of N,N-
dimethylformamide and cooled to 0 °C. Then 3.4 g (18 mmol)
of 4-chloromethylbenzoyl chloride was added. The reaction
mixture was stirred for 2 h before it was poured onto 500 g of
crushed ice. This mixture was neutralized with NaHCO3 and
stirred for an hour. The precipitate was filtered off and re-
crystallized from ethyl alcohol, providing 3.1 g of 4-chloro-
methyl-N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-ylamino)-phenyl]-
benzamide (yield: 61%) as a yellowish solid. 4-Chloromethyl-
N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-yl-amino)-phenyl]-
benzamide (1.6 g, 3.8 mmol) and 4.3 mL (38 mmol) of N-methyl
piperazine were added to 150 cm3 of acetonitrile. The reaction
mixture was heated under reflux for 6 h, then evaporated to
50 cm3, and cooled to 0 °C. The precipitate was filtered off and
recrystallized from acetonitrile to provide 1.25 g (yield: 68%)
of 1 as creamy white crystals. Anal. (C29H31N7O) C calcd, 70.56;
found, 69.03; H calcd, 6.33; found, 5.51; N calcd, 19.86; found,
25.14. Mp: 206-209 °C. Rf ) 0.15 (chloroform:methanol ) 10:
1), retention times: 3.24 min (HPLC1) and 4.48 min (HPLC2).
1
MS: (M + H)+ ) 494, (M - H)- ) 492. H NMR (DMSO-d6,
4
300 MHz) δ (ppm): 10.14 (s, 1H, NH), 9.27 (d, 1H, J ) 1.8

254 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1
Hz), 8.95 (s, 1H, NH), 8.68 (dd, 1H, 3J ) 4.7 Hz, 4J ) 1.5 Hz),
8.50 (d, 1H, J ) 6.9 Hz), 8.49 (m, 1H), 8.08 (d, 1H, J ) 1.9
Hz), 7.90 (d, 2H, 3J ) 8.2 Hz), 7.54-7.41 (m, 5H), 7.20 (d, 1H,
3J ) 8.34 Hz), 3.52 (s, 2H), 2.35 (bs, 8H), 2.22 (s, 3H, CH3),
2.15 (s, 3H, CH3).
Potentiometric Titrations. pH-metric titrations were
carried out at 25.0 ( 0.1 °C. A Metrohm 716 DMS Titrino
autoburet (Metrohm, Switzerland) equipped with a Metrohm
6.0234.110 combined glass electrode was used. The electrode
system was calibrated in terms of hydrogen ion concentration11
by titrating 2 mL of 0.05 M HCl with standardized 0.05 M
NaOH. Both stock solutions contained calculated amounts of
NaCl to ensure a constant ionic strength of 0.15 M during
titration. For protonation constant determinations, a weighed
amount of the ligand was dissolved in the HCl stock solution
and titrated with NaOH. Measurements were performed at
various ligand concentrations, 1-5 mM for 1 and 2 and 2-10
mM for 3.
Szaka´cs et al.
of sufficiently basic or acidic solutions, respectively (Table S4).
A special approach was necessary for dichloroacetate, where
complete protonation could not be reached even at pH 0. In
this case, the indicator parameters log K ) 1.06 ( 0.01 and
δHInd ) 6.328 ppm were determined in a separate 1H NMR
titration at 1 M ionic strength.16 The protonation constant was
converted to 0.15 M ionic strength by the Davies equation17
to yield log K ) 1.14, while δHInd was converted empirically to
0.15 M ionic strength.
An important criterion of using in situ pH indicators is the
lack of their interaction with other molecules in the sample.
Since the δInd and δHInd values of the same indicator vary from
one NMR titration to another within the experimental error
(ca. 0.002 ppm), such disturbing interferences are assumed to
be negligible.
3
4
The NMR-pH datasets were evaluated with the OPIUM
computer program.18 For microconstant calculations, in-house
least-squares C++ programs12 were used.
All titration curves were evaluated by our nonlinear least-
squares regression program PROTC.12 The acid-base chem-
istry of imatinib, a tetravalent base, is characterized in terms
of log K protonation constants, although the term pKa, the well-
known acid dissociation parameter, is certainly widespread in
medicinal and pharmaceutical chemistry. In order to comply
with data in Critical Stability Constants,13 “stoichiometric”
protonation constants containing hydrogen ion concentration
were calculated. To get “practical” protonation constants
involving hydrogen ion activity, the log K values reported here
should be increased by 0.12 unit.
Acknowledgment. This work was supported by
Grants OTKA T043579 and ETT 535/03. The authors
are grateful to Istva´n Ko¨vesdi (EGIS Pharmaceutical
Works, Budapest) and Miklo´s Boros (Semmelweis Uni-
versity, Budapest) for access to the 500 MHz spectrom-
eter.
Supporting Information Available: Tables S1, S2, and
S3 displaying 1H NMR chemical shifts [ppm] of individual HiL
species of 3, 2, and 1, respectively, and their changes upon
protonation. Table S4 displaying protonation constants, pH
ranges, and limiting 1H NMR chemical shifts [ppm] of the
NMR-pH indicator molecules. Table S5 displaying summary
of experimental data confirming the identity and purity of
compounds 1-3. This material is available free of charge via
the Internet at http://pubs.acs.org.
1H NMR Titrations with in Situ pH Monitoring. To
acquire a pH-dependent series of 1H NMR spectra, the recently
developed14 electrodeless single tube NMR titration was ap-
plied. In this method, titration of the ligand is carried out in
a single NMR tube, which contains also pH-monitoring indica-
tor molecules (dichloroacetic acid, chloroacetic acid, acetic acid,
and tris(hydroxymethyl)aminomethane [TRIS]), the chemical
shift of which show the actual pH after each titrant addition.
A single NMR sample solution of 0.7 mL was prepared,
which contained 3 mM imatinib 1, the indicators listed above
(2 mM each), 0.05 M HCl to set the starting pH to 1.7, 1 mM
sodium 3-(trimethylsilyl)-1-propanesulfonate (DSS) as internal
chemical shift reference, and 0.096 M NaCl to give a total ionic
References
(1) Levitzky, A. Signal Transduction Therapy. In Molecular Patho-
mechanism and New Trends in Drug Research, 1st ed.; Ke´ri,
Gy., To´th, I., Eds.; Taylor & Francis: New York, 2003; pp 55-
58.
(2) Schindler, T.; Bornmann, W.; Pellicena, P.; Miller, W. T.;
Clarkson, B.; Kuriyan, J. Structural Mechanism for STI-571
Inhibition of Abelson Tyrosine Kinase. Science 2000, 289, 1938-
1942.
(3) Corbin, A. S.; Buchdunger, E.; Pascal, F.; Druker, B. J. Analysis
of the Structural Basis of Specificity Inhibition of the Abl Kinase
by STI571. J. Biol. Chem. 2002, 277, 32214-32219.
(4) Noble, M. E. M.; Endicott, J. A.; Johnson, L. N. Protein Kinase
Inhibitors: Insights into Drug Design from Structure. Science
2004, 303, 1800-1805.
(5) Rodr´ıguez Flores, J.; Berzas, J. J.; Castaneda, G.; Rodr´ıguez,
N. Direct and Fast Capillary Zone Electrophoretic Method for
the Determination of Gleevec and its Main Metabolite in Human
Urine. J. Chromatogr., B 2003, 794, 381-388.
(6) Frenna, V.; Vivona, N.; Consiglio, G.; Spinelli, D. Amine Basici-
ties in Benzene and in Water. J. Chem. Soc., Perkin Trans. 2
1985, 12, 1865-1868.
1
strength of 0.15 M. After recording the H NMR spectrum of
this sample on a 500 MHz Varian Unity spectrometer, the
sample tube was ejected and a small (3-20 µL) portion of an
alkaline titrant solution (0.11 M NaOH, 0.12 M NaCl) was
added from a medical syringe. After the contents of the NMR
tube were homogenized, a new spectrum was acquired. Since
1
the solution contained no D2O, all H spectra were measured
without spectrometer lock at 25 °C and the resonance of
solvent H2O was diminished by a selective presaturation pulse
before the observation pulse.19 The NMR titrations of 2 and 3
were performed analogously on a 250 MHz Bruker Avance
DRX spectrometer.
The freeware program Mestre-C15 was used for spectral
processing, followed by first-order multiplet analysis. 1H
chemical shifts are referenced to internal DSS, which does not
alter its peak position even under strongly acidic circum-
stances.16,20
(7) Vivekanand, V. V.; Rao, D. S.; Vaidynathan, G.; Sekhar, N. M.;
Kelkar, S. A.; Puranik, P. R. Validated LC Method for Imatinib
Mesylate. J. Pharm. Biomed. Anal. 2003, 33, 879-889.
(8) Jaroszewski, J. W.; Matzen, L.; Frolund, B.; Krogsgaard-Larsen,
P. Neuroactive Polyamine Wasp Toxins: Nuclear Magnetic
Resonance Spectroscopic Analysis of the Protolytic Properties
of Philanthotoxin-343. J. Med. Chem. 1996, 39, 515-521.
(9) Szaka´cs, Z.; Kraszni, M.; Nosza´l, B. Determination of Microscopic
Acid-Base Parameters from NMR-pH Titrations. Anal. Bioanal.
Chem. 2004, 378, 1428-1448.
The pH belonging to each spectrum was determined from
obs
Ind
the actual chemical shift of the appropriate indicator δ
,
using a rearranged form of the Henderson-Hasselbach equa-
tion:
(10) Nosza´l, B. Group Constant: A Measure of Submolecular Basicity.
J. Phys. Chem. 1986, 90, 4104-4110.
obs
δ
- δHInd
[Ind]
[HInd]
Ind
pH ) log KInd + log
) log KInd + log
(11) Irving, H. M.; Miles, M. G.; Pettit, L. D. A Study of Some
Problems in Determining the Stoichiometric Proton Dissociation
Constants of Complexes by Potentiometric Titrations Using a
Glass Electrode. Anal. Chim. Acta 1967, 38, 475-488.
(12) Nosza´l, B.; Szaka´cs, Z. Microscopic Protonation Equilibria of
Oxidized Glutathione. J. Phys. Chem. B 2003, 107, 5074-5080.
(13) Martell, A. E., Smith, R. M., Eds. Critical Stability Constants;
Plenum Press: New York, 1974-1989.
obs
Ind
δInd - δ
(5)
The log KInd protonation constants were determined in
separate potentiometric titrations, and the results in Table
S4 (Supporting Information) show a good agreement with
literature values. The limiting chemical shifts δInd and δHInd
refer to the unprotonated and protonated forms of the indica-
tor, respectively. These values are obtained from NMR spectra
(14) Szaka´cs, Z.; Ha¨gele, G.; Tyka, R. 1H/31P NMR pH Indicator Series
to Eliminate the Glass Electrode in NMR Spectroscopic pKa
Determinations. Anal. Chim. Acta 2004, 522, 247-258.

Acid-Base Profiling of Imatinib and Its Fragments
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 1 255
(15) MestRe-C computer program, Version 3.5.1Beta, http://
www.mestrec.com.
(19) Hoult, D. I. Solvent Peak Saturation with Single-Phase and
Quadrature Fourier Transformation. J. Magn. Reson. 1976, 21,
337-347.
(20) De Marco, A. pH Dependence of Internal References. J. Magn.
Reson. 1977, 26, 527-528.
(16) Szaka´cs, Z.; Ha¨gele, G. Accurate determination of low pK values
by 1H NMR titration. Talanta 2004, 62, 819-825.
(17) Davies, C. W. The Extent of Dissociation of Salts in Water. Part
X. Dissociation Minima. J. Chem. Soc. 1945, 460-464.
(18) OPIUM computer program, M. Kyvala, I. Lukes, Chemometrics
’95, Pardubice, 1995, Book of Abstracts, p 63; http://www.
natur.cuni.cz/∼kyvala/opium.html.
JM049546C