- Email: [email protected]
Determination of acid dissociation constants of warfarin and hydroxywarfarins by capillary electrophoresis
Accepted Manuscript Title: Determination of acid dissociation constants of warfarin and hydroxywarfarins by capillary electrophoresis Author: Paweł Nowak Paulina Olechowska Mariusz Mitoraj Michał Wo´zniakiewicz Paweł Ko´scielniak PII: DOI: Reference:
S0731-7085(15)00260-5 http://dx.doi.org/doi:10.1016/j.jpba.2015.04.027 PBA 10063
To appear in:
Journal of Pharmaceutical and Biomedical Analysis
Received date: Revised date: Accepted date:
23-12-2014 17-4-2015 18-4-2015
Please cite this article as: P. Nowak, P. Olechowska, M. Mitoraj, M. Wo´zniakiewicz, P. Ko´scielniak, Determination of acid dissociation constants of warfarin and hydroxywarfarins by capillary electrophoresis, Journal of Pharmaceutical and Biomedical Analysis (2015), http://dx.doi.org/10.1016/j.jpba.2015.04.027 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Determination of acid dissociation constants of warfarin and
2
hydroxywarfarins by capillary electrophoresis
3
Paweł Nowaka, Paulina Olechowskaa, Mariusz Mitorajb, Michał Woźniakiewicza*, Paweł
4
Kościelniaka
5
a Jagiellonian
6
Kraków, Poland
7
b
8
Chemistry, Kraków, Poland
9
Corresponding Author:
ip t
1
cr
University in Kraków, Faculty of Chemistry, Department of Analytical Chemistry,
an
us
Jagiellonian University in Kraków, Faculty of Chemistry, Department of Theoretical
*Michał Woźniakiewicz, PhD, Jagiellonian University in Kraków, Faculty of Chemistry,
11
Department of Analytical Chemistry, Kraków, Poland, Ingardena St. 3, 30-060 Kraków,
12
Poland, [email protected], tel./fax: +48 12 663 20 84
M
10
Ac ce p
te
d
13
1 Page 1 of 36
Highlights:
14
o Two acid dissociation equilibria have been described
15
o Thermodynamic pKa values have been determined
16
o CE, CE-DAD and IS-CE methods have been employed
17
o Pharmacological relevance has been presented and discussed
18
o Theoretical explanations of pKa values have been attempted
us
cr
ip t
13
an
19 20 Abstract
22
In this work the acid dissociation constants – pKa of warfarin and its all important
23
oxidative metabolites have been determined by capillary electrophoresis-based
24
methods. It has resulted in a complete description of two acid-base dissociation
25
equilibria, yet not investigated experimentally for phase I metabolites of warfarin.
26
The capillary electrophoresis (CE) method based on the relation between effective
27
electrophoretic mobilities and pH has proven to be a suitable tool for pKa
28
determination, while the spectrophotometric (CE-DAD) and the internal standard
29
methods (IS-CE), have appeared to be promising alternative approaches. The CE-DAD
30
approach based on the change in absorbance spectra between the acidic and basic
31
forms is a combination between capillary electrophoresis and spectrophotometric
32
titration, and yields very consistent values of pKa1 with CE. The IS-CE, in turn, enables
33
an estimation of pKa1 and pKa2 from only two analytical runs, however, less accurate
34
than CE and CE-DAD. The Debye-Hückel model has been confirmed experimentally as
35
a good predictor of pKa values at various ionic strengths. Therefore, it has been used
36
in determination of thermodynamic pKa1 and pKa2, referring to the zero ionic
Ac ce p
te
d
M
21
2 Page 2 of 36
strength. The results are important from the analytical, pharmacological, and
38
theoretical points of view.
39
Keywords: acid dissociation constant, capillary electrophoresis, Debye-Hückel model, drug
40
metabolism, hydroxywarfarins, warfarin
41
Abbreviations:
42
DMSO - dimethyl sulfoxide
43
EOF – electroosmotic flow
44
WAR – warfarin
45
W3 – 3’-hydroxywarfarin
46
W4 – 4’-hydroxywarfarin
47
W6 – 6-hydroxywarfarin
48
W7 – 7-hydroxywarfarin
49
W8 – 8-hydroxywarfarin
50
W10 – 10-hydroxywarfarin
cr us an M d te
Ac ce p
51
ip t
37
3 Page 3 of 36
51 1. Introduction
53
Warfarin (WAR) [3-(α-acetonylbenzyl)-4-hydroxycoumarin] is a widely used drug
54
exhibiting anticoagulant activity, applied in the prevention of thrombosis and
55
thromboembolism, i.e. formation of blood clots and vascular occlusions [1,2]. Its routine
56
application as an effective anticoagulant is however quite problematic owing to complex
57
oxidative metabolism, involving several different isoforms of cytochrome P450 enzyme
58
(CYP). The main oxidative metabolites of WAR are hydroxywarfarins: 3’-hydroxywarfarin
59
(W3), 4’-hydroxywarfarin (W4), 6-hydroxywarfarin (W6), 7-hydroxywarfarin (W7), 8-
60
hydroxywarfarin (W8) and 10-hydroxywarfarin (W10). CYP-mediated metabolism of WAR
61
is highly regio- and enantioselective, the most popular hydroxywarfarins found in vivo are
62
(S)-W7, (R)-W4 and (R)-W10 [1].
63
The acid dissociation constant, usually expressed as its pKa value, is one of the most
64
important parameters describing physicochemical properties of drugs [3,4]. It characterizes
65
acid-base equilibrium of the respective ionizable groups in solution, thus, indicates their
66
dissociation (deprotonation) potential at a given pH. The acidic and basic forms of drugs
67
differ in charge – at least one of them is ionized, therefore, they also may differ in other key
68
properties like water solubility, membrane permeability, affinity of association with
69
proteins, and finally, therapeutic activities. As it was demonstrated in the literature,
70
therapeutic activity of WAR is also dependent on its ionization level [2]. For that reason, the
71
exact value of pKa for WAR is important from the pharmacological point of view, and it was
72
determined many times in the past [5-9]. Another issue is acid-base properties of WAR
73
metabolites, yet not fully recognized. Hydroxywarfarins, contrary to the parent drug, are
74
supposed to be capable of a double dissociation, and this fact complicates predictions of
75
their ionization-related properties. Therefore, experimental determination of their exact
76
pKa1 and pKa2 values is undeniably necessary for the holistic understanding of WAR
77
pharmacokinetics.
78
There are many various analytical techniques that allow for accurate estimation of pKa
79
values in a sound and rapid way [3,10]. Capillary electrophoresis (CE) belongs to the
Ac ce p
te
d
M
an
us
cr
ip t
52
4 Page 4 of 36
subgroup of separation techniques, and in the recent years, has gained particular interest as
81
an efficient tool for pKa determination [11-17]. Its growing popularity originates from the
82
clear analytical benefits: (i) very small consumption of samples and buffers; (ii) relatively
83
high throughput and simple automation; (iii) possibility to separate the sample component
84
from each other, including impurities; (iv) and the accuracy, typically within 0.05 pH units.
85
In a standard approach based on CE, pKa values are calculated from a curve presenting
86
dependence of electrophoretic mobility on pH. For comparative purposes, the use of
87
thermodynamic pKa (pKaT) corresponding to the zero ionic strength and standard
88
temperature 25 ̊C is more suitable than the apparent pKa, determined experimentally in
89
specific conditions of ionic strength and temperature [18,19].
90
In this work, the acid-base equilibria of WAR and six hydroxywarfarins have been
91
investigated and characterized by determination of the respective pKa values. To the best of
92
our knowledge, this is the first experimental study of all pharmacologically important
93
hydroxywarfarins, and in addition, it covers both dissociation equilibria. The standard CE-
94
based method has been used to obtain the most accurate results. Afterwards, usefulness of
95
two alternative approaches utilizing CE have been investigated: the spectrophotometric
96
method (CE-DAD), based on the changes in molecular spectra recorded by DAD detection
97
system; and the internal standard-based method (IS-CE), restricted only to two analytical
98
runs and the known pKa value of an internal standard. The variation of pKa1 with changing
99
ionic strength has been studied experimentally, and compared with the values predicted by
Ac ce p
te
d
M
an
us
cr
ip t
80
100
the Debye-Hückel model. Finally, pKa1T and pKa2T values have been found for each
101
compound, which correspond to the standardized temperature and zero ionic strength. It
102
enables creation of a more complete picture of pharmacologically-important properties of
103
WAR metabolites, theoretical considerations referring to structure-property relationships,
104
and better understanding of their migration profile observed during CE separations.
105
2. Material and methods
106
2.1. Instrumentation
107
WAR (racemic mixture) was supplied by Sigma-Aldrich (St. Louis, MO, USA), W3, W4, W6,
108
W7, W8, and W10 (all racemic mixtures) by LGC Standards (Teddington, UK), while all 5 Page 5 of 36
other chemicals by Avantor Performance Materials Poland. S. A. (Gliwice, Poland). All
110
standard solutions were prepared in the deionized water (MilliQ, Merck-Millipore Billerica,
111
MA, USA) and filtered through the 0.45 μm regenerated cellulose membrane, then degassed
112
by centrifugation. Standard concentration of analytes was 0.1 mg/mL, all analytes were
113
dissolved in water/methanol (1:1 v/v) mixture. Dimethyl sulfoxide (DMSO) was used as the
114
electroosmotic flow (EOF) marker, in 0.2% (v/v) concentration.
115
The P/ACE MDQ Capillary Electrophoresis System (Brea, CA, USA) equipped with a diode
116
array detector was used for all CE-based experiments, using the bare fused-silica capillary
117
of 60 cm total length and 75 μm internal diameter. Separations were conducted in a short-
118
injection mode [20], unless stated otherwise, using a 10 cm long outlet capillary part. Owing
119
to this fact, the total separation time could be vastly reduced. Sample injection was
120
conducted using forward pressure 0.4 psi for 4 s. During separations, 15 kV voltage (anode
121
at injection end) and the additional forward pressure of 0.4 psi were applied. The current
122
measured during separations was between 25 – 125 µA. Capillary was conditioned at 25 ̊C,
123
using the liquid cooling system. Every time, DAD detector collected the whole spectra
124
within 200-600 nm. Signal recorded at 200, 280 and 308 nm was used for the further
125
analysis.
126
Capillary rinsing was conducted between runs applying pressure of 137,9 kPa (20 psi). The
127
particular steps of rinsing were: deionized water for 1 min, 0.1 M NaOH for 2 min, and
128
background electrolyte (BGE) for 2 min. During the first use of the capillary at a working
129
day: methanol for 5 min, 0.1 M HCl for 2 min, deionized water for 2 min, 0.1 M NaOH for 10
130
min, and BGE for 10 min were applied. For the fresh capillary conditioning, the latter
131
sequence was used, but the duration of each individual step was doubled.
cr
us
an
M
d
te
Ac ce p
132
ip t
109
133
2.2. Buffering solutions
134
Composition of the given BGEs was calculated with the use of PHoEBuS 1.3 software by
135
Analis (Namur, Belgium), setting 100 mM as the selected value of ionic strength. The buffers
136
of lower ionic strength were prepared by dilution with deionized water. The obtained
6 Page 6 of 36
recipes for BGEs preparation have been presented in Table 1. The values of pH were
138
verified experimentally prior to CE separations.
139
Table 1
140
2.3. Methods for pKa determination
141
2.3.1. Standard electrophoretic method (CE)
142
In the standard CE approach, pKa values have been estimated by applying the non-linear
143
regression model to the obtained dependency between effective electrophoretic mobility
144
(µeff) and pH. According to the Eq.1, µeff can be calculated from the respective migrations
145
times of analyte and EOF marker:
an
us
cr
ip t
137
(1)
M
146 147
where μeff and μobs are the effective and observed electrophoretic mobilities of analyte (m2 V-
148
1 s-1),
149
total and effective capillary lengths (m), 0.60 and 0.10 m, respectively; V is the separation
150
voltage (V); tobs is the measured migration time of analyte (s), while teof is the time
151
measured for neutral marker of EOF – DMSO (s).
152
The relation between µeff and pH is described by Eq.2 and Eq.3, for the monoprotic and
153
diprotic acid, respectively:
155
d
te
Ac ce p
154
respectively; μeof is the mobility of electroosmotic flow (m2 V-1 s-1); Ltot and Leff are the
(2)
(3)
156
where α values are fitting constants equal to electrophoretic mobility of the deprotonated
157
form of the acid with subscript 1 and 2 equal to the order of dissociation. The pKa values are
158
equal to the pH of inflection points of the fitted sigmoidal-shape curves.
159
2.3.2. Spectrophotometric method (CE-DAD)
7 Page 7 of 36
To enable determination of pKa1 by CE-DAD method [21,22], the absorbance of WAR and its
161
derivatives was measured at the maximum of electrophoretic peaks at 280 and 308 nm.
162
These wavelengths are close to maxima of absorbance of the acidic and basic forms,
163
respectively. Afterwards, the parameter β was calculated according to the three different,
164
arbitrary proposed definitions:
(5)
167
(6)
an
166
cr
(4)
us
165
ip t
160
where β1, β2 and β3 present different calculation methods, while A280 and A308 are the values
169
of absorbance at the given wavelength (nm). As it has been shown in Fig.1, there is a strong
170
shift in absorbance at these wavelength between the acidic and basic forms of WAR. Most
171
importantly, the direction of this shift is opposite for 280 and 308 nm. These wavelength
172
are applicable also for all hydroxywarfarins, whose spectra differ between each other,
173
mostly in the case of W7 (not shown).
174
Figure 1
175
The changing value of β together with changing pH is observed, if the ratio of molar
176
absorptivity coefficients at 280 and 308 nm is considerably different for acidic and basic
177
form. In that case, an every partially ionized state corresponds to the value of β from the
178
range between the two limiting values, related to the protonated and deprotonated forms,
179
respectively. In other words, such parameter is then sensitive on a changing contribution of
180
the acidic and basic forms, characterized by different shape of absorption spectrum. The
181
plot of β against pH gives the curve of similar sigmoidal shape as in the case of the CE
182
method, and analogously, the pKa value is indicated by inflection point located on this curve.
183
Ac ce p
te
d
M
168
2.3.3. Internal-standard method (IS-CE)
8 Page 8 of 36
The IS-CE method was proposed several years ago by Rosés’ scientific group [23-27]. It is
185
based on the use of internal standard of known pKa value, similar to the predicted pKa of
186
analyte, and requires only two runs at two distinct pH values. The first pH should provide
187
total dissociation of both the internal standard and the analyte, while the second one, only
188
partial dissociation. Then, the sought pKa can be calculated according to the following
189
equation:
us
192
given that, (8)
an
191
(7)
cr
190
ip t
184
where pKaIS refers to the internal standard, μA- is the effective electrophoretic mobility
194
measured when both compounds are totally deprotonated, while in the case of μeff,
195
supposed to be partially ionized. To enable the accurate analysis, the analyte and the
196
internal standard are injected together from the same vial. The values of Q are calculated
197
for the analyte and the internal standard independently.
198
Further information on the determination of pKa values has been presented and discussed
199
in the next section of this work.
200
2.3.4. Debye-Hückel model
201
The estimations of thermodynamic pKa were done by using the following equation [18,19]:
d
te
Ac ce p
202
M
193
(9)
203
where γ denotes activity coefficient of the particular species of an analyte: acidic – HA, and
204
basic – A, with a given charge (z). Activity coefficients were calculated based on the
205
extended law of Debye-Hückel [28]:
206
(10)
9 Page 9 of 36
where A and B are the parameters dependent on dielectric constant and temperature, in
208
water at 25°C they are 0.509 and 0.33, respectively, z is the charge of the ion, I is the ionic
209
strength, while a denotes the radius of hydrated ion. The approximate value of a = 5 Å was
210
used for calculations.
211
3. Results and discussion
212
3.1. Application of the CE method
213
At the beginning, pKa1 and pKa2 of WAR and all hydroxywarfarins have been determined by
214
the standard CE method. The measurements of μeff in buffers with gradually increasing pH,
215
but constant ionic strength of 100 mM and temperature of 25 ̊C, have been conducted
216
separately for each compound. Then, the non-linear regression model has been applied,
217
resulting in the fitted curves characterized by one or two inflection points indicating the
218
respective pKa values, according to Eq.2 and Eq.3. All data obtained in this stage have been
219
presented in Fig.S-1 (left part).
220
As it can be seen, both pKa1 and pKa2 have been found. W10 however, in spite of possessing
221
two hydroxyl groups, does not exhibit the second dissociation, similarly as WAR. It
222
originates from alcoholic character of the second hydroxyl group. Interestingly, pKa1 of W10
223
is nearly 1.0 pH unit higher than in the case of the other compounds. A slight difference, in
224
the range of 0.1 – 0.2 pH units, is noted also between W7 and the rest of compounds, except
225
W10. The variation of pKa1 between WAR, W3, W4, W6 and W8 is rather minor, from 4.87
226
to 4.93, generally within the range of method errors. As far as pKa2 is concerned, its
227
fluctuation turns out to be substantially greater. In enables to classify the compounds
228
according to their pKa2 value into three groups. W3 and W4 belong to the first group, pKa2
229
amounts to 11.23 and 11.22, respectively. W6 of the second group exhibits pKa2 10.69,
230
whereas W7 and W8 of the third group show values 9.39 and 9.50, respectively. The
231
uncertainty of pKa2 determination is notably greater than pKa1, nevertheless, qualitative
232
differences between pKa2 of hydroxywarfarins are unambiguously clear.
233
One shall admit that an effective temperature inside capillary could be higher than 25 ̊C due
234
to inefficient cooling, especially in the inlet part. Therefore, the results may be burdened by
235
some intrinsic error related to changes in ionization caused by temperature increase.
Ac ce p
te
d
M
an
us
cr
ip t
207
10 Page 10 of 36
Our results reported herein provide the first experimental and complete picture of the acid-
237
base properties of WAR and all hydroxywarfarins existing in vivo. One must admit that pKa1
238
of WAR was determined oftentimes in the past (4.85 – 5.15) [5-9], while pKa1 of W7 and
239
W10 was estimated by us in our previous work (5.19 and 6.06, respectively) [9]. The
240
experiments described in our previous work however, were conducted in other conditions
241
of temperature and ionic strength, and were devoted mainly to investigation of the
242
supramolecular pKa shifts upon complexation with cyclodextrins. In general, the current
243
results are consistent with the previous reports [5-9]. Further comments and conclusions
244
related to the determined pKa values can be found in Section 3.4.
245
3.2. Application of CE-DAD method
246
In some cases the CE-DAD instrument gives a possibility to estimate pKa by totally different
247
method than aforementioned one [21,22]. In general, it is the case when the acidic and basic
248
forms of analyte exert different UV-Vis spectra. Then, the advantages of two techniques, CE
249
and spectrophotometric titration, can be combined within one approach. It is worth
250
highlighting that application of voltage is in fact not mandatory, however, it enables
251
separation of sample components from each other and allow to decrease analysis time. As it
252
has been shown in the literature, the only variable parameters used for pKa calculation are
253
absorbance at the given wavelength and analyte concentration. Additionally, a constant
254
value of a molar absorptivity coefficient is required, determined separately for the acidic
255
and basic forms.
256
Our method presented herein for the first time, however, require neither any values of
257
analyte concentration nor absorptivity coefficient. It is instead based on the measurements
258
of absorbance at two distinct wavelengths. Accordingly, the absorbance of WAR or any
259
other compound is measured at the maximum of electrophoretic peaks at 280 and 308 nm,
260
respectively. The outcomes have been shown in Fig.S-1 (see Supplementary Material) and
261
Table 2.
262
Table 2
263
Table 2 demonstrates that the results are the same for β2 and β3, but differ for β1.
264
Remarkably, the excellent accordance is noted between β2, β3, and the reference CE method.
Ac ce p
te
d
M
an
us
cr
ip t
236
11 Page 11 of 36
The less consistent results observed for β1 stem probably from its mathematical definition,
266
since the value of β1 changes in a far broader range than β2 and β3,. Such good outcome
267
proves that the CE-DAD method is fully prospective and gives response consistent with the
268
typical CE approach. The model is based on the stepwise changes in the shape of
269
absorbance spectra, instead of ionization entailing the shifts in migration times in CE. It is
270
worth highlighting that CE-DAD method has its own advantage, since EOF marker is not
271
necessary. Its limitation, in turn, is that the suitable accuracy of measurements can be easily
272
obtained only for relatively high concentration of analyte. Notwithstanding, CE and CE-DAD
273
methods can be applied interchangeably, however in this case, only for pKa1 determination.
274
3.3. Application of the IS-CE method
275
The method based on the use of internal standard of known pKa value was presented as a
276
fully prospective alternative for other methods used in high throughput pKa determination,
277
especially, in drug discovery studies [23-27]. At the moment, however, its true potential is
278
still unknown. In particular, consistency between IS-CE and CE methods applied jointly for
279
the same analytes seems to be an interesting issue to be investigated.
280
To this end, IS-CE has been tested in the numerous variants. Firstly, all the analytes have
281
been used individually as the internal standards for calculation of pKa of the others. As the
282
pKa of internal standards, the values obtained from CE method have been used. In this way,
283
a given pKa1 value has been estimated from six various reference pKa1 values. Secondly, the
284
calculations have been done for three distinct μeff values corresponding to partial ionization,
285
attributed to three distinct pH values. As the value of μeff related to the fully deprotonated
286
species, the value indicated by a bottom horizontal asymptote during function fitting has
287
been applied (see Fig.S-1). In consequence, particular pKa could be determined by IS-CE
288
many times, using different approaches. In addition, pKa2 has been determined for all
289
compounds except WAR and W10. All pKa values have been gathered in Table 3.
290
Table 3
291
In general, as far as pKa1 values are concerned, satisfactory accordance is noticed between
292
the IS-CE and CE methods. In majority, discrepancy is less than 0.1 pH units, and this fact
293
highlights the potential of IS-CE method. The application of W10 as the internal standard
Ac ce p
te
d
M
an
us
cr
ip t
265
12 Page 12 of 36
gives the worst predictions, and also, the values obtained for W10 as the analyte are mostly
295
poorer than the values calculated for other compounds. It seems to be logical that this
296
situation results from the particularly high pKa1 value of W10. Nevertheless, the calculations
297
performed for μeff obtained at pH 5.25 are relatively consistent with CE. It prompts us that a
298
choice of pH from the range between pKa of internal standard and analyte, is actually, the
299
most profitable. It can be also noted that the results obtained for WAR at pH around 6.0 are
300
worse than for hydroxywarfarins. In this case, putatively, possession of the second hydroxyl
301
group might be the main factor affecting accuracy of predictions.
302
In the case of pKa2 the discrepancy between the CE and IS-CE methods is significantly
303
higher, mostly over 0.2 pH units. As it can be easily concluded, the satisfactory results are
304
noted only if pKa2 of analyte is close to pKa2 of internal standard, like for W3 and W4, or W7
305
and W8. Nevertheless, the right qualitative trend between particular compounds is
306
indicated in each column (see Table 3). Owing to this qualitative accordance one can
307
speculate that only two analytical runs and the obtained very approximate pKa2 values
308
should suffice for proper identification of metabolites based on their migration order in CE
309
(compare with Section 3.6).
310
An overall evaluation of the IS-CE method is not straightforward. On the one hand, an
311
accurate pKa value is rather difficult to be obtained when properties of an internal standard
312
and an analyte differ from each other. On the other hand, for compounds of more similar
313
nature the outcomes are appreciably amended. Secondly, a rough pKa value or a direction of
314
its shift with regard to a reference molecule or state, can be obtained very fast. Even if IS-CE
315
is not a perfect tool, its potential of time saving merits attention during development of new
316
high throughput methods for pKa determination.
317
3.4. Thermodynamic pKa
318
Determination of thermodynamic pKa, contrary to apparent values discussed above, enables
319
an easy comparison between different experimental conditions. pKaT refers to the zero ionic
320
strength, and thus, it is defined by the activities of respective species, in this state equal to
321
their concentrations.
Ac ce p
te
d
M
an
us
cr
ip t
294
13 Page 13 of 36
In order to investigate the usefulness of the extended Debye-Hückel model for prediction of
323
pKaT values, the variation of apparent pKa1 with changing ionic strength has been studied
324
experimentally. For that purpose, the CE procedure has been repeated using diluted BGEs,
325
to 25 and 10 mM ionic strength, respectively. It allowed us to compare possible differences
326
between particular ionic strengths obtained experimentally with the values calculated from
327
Debye-Hückel model (see Section 2.3.4). The results have been collected in Table 4.
328
Table 4
329
They suggest that the current version of Debye-Hückel model suitably predicts the variation
330
of pKa1 with respect to changes in ionic strength. The differences between the theoretical
331
model and the values verified experimentally are in the order of only 0.01 pH units.
332
Therefore, the Debye-Hückel correction has been applied for calculation of pKa1T, and also
333
pKa2T. It is worth mentioning that the scale of this correction is greater for pKa2T than for
334
pKa1T, what can be derived from Eq.9 and Eq.10, due to the fact that second dissociation
335
occurs for the singly ionized species.
336
3.5. Theoretical and pharmacological considerations
337
We have further performed preliminary DFT/TZP/BLYP-D3 study using ADF program [29]
338
in order to characterize the structures of all compounds (WAR, W3, W4, W6, W7, W8 and
339
WAR-10) in gas phase. They are presented in Fig.2. It is clearly visible that all compounds
340
contain intramolecular hydrogen bonding of the type O…HO. More importantly, it has been
341
found that the bond lengths within the O…HO bridge are very similar (~1.0 Å for O–H and
342
~1.78 Å for H…O) in all compounds except W10, for which the shortest H…O distance (1.61
343
Å) has been found, see Fig.2. It shows that the O–H proton of W10 is engaged in the
344
strongest intramolecular interaction. It correlates with the trend in the pKa1 values, i.e.
345
markedly larger value of pKa1 is noted for W10 with respect to the remaining regio-isomers.
346
Accordingly, one could suggest that the characteristics of O…HO bonding could directly
347
determine the trend in pKa1 values.
348
Figure 2
Ac ce p
te
d
M
an
us
cr
ip t
322
14 Page 14 of 36
As far as the second pKa2 values are concerned, one shall note that changes in stability when
350
going from monoanion to dianion are of vital importance. We see that W3 and W4 exhibit
351
larger pKa2 values as compared to W6, W7 and W8. It is noticeable that in the former case
352
the hydroxyl is introduced at the single phenyl ring, whereas in the latter one various O–H
353
positions at coumarin moiety are considered. It could be the case that delocalization of
354
negative charge, leading to resonance and induction effects, is stronger for the two aromatic
355
coumarin rings than for the single phenyl ring. Coumarin moiety, in addition, contains the
356
electron withdrawing units, carbonyl and heteroatom. An important role of carbonyl unit in
357
enhancement of anion stability is known, it is reflected e.g. by dissociation of carboxyl acids.
358
Apart from these speculations, the increased electrostatic attraction of the proton caused by
359
the negative charge localized on the oxygen of dissociated OH group should be considered
360
for each hydroxywarfarin. Such effect can account for the higher pKa2 noted for W6, than for
361
W7 and W8.
362
We plan in the future to calculate free energy changes for all compounds, as well as the
363
related theoretical pKa values. It will be performed both in the gas phase and in the water.
364
Furthermore, the electronic structures that could help us to understand factors that
365
determine pKa values on a microscopic level will be also calculated.
366
Another important issue is an equilibrium between distinct tautomers of WAR and its
367
derivatives which can coexists in solution: the open form and the two diastereomeric
368
hemiketals, depicted in Fig.3. It is remarkable that different compounds may exhibit
369
different partition between particular tautomers, depending also on the solvent
370
composition [30]. Furthermore, amount and spatial distribution of the potential hydrogen
371
donors and acceptors varies between the open and the hemiketal forms. Therefore, acid-
372
base equilibrium defined by proximity of given molecule regions can be associated with
373
equilibrium between the tautomers. This aspect cannot be omitted during elucidation of
374
structure-related properties of WAR and its metabolites.
375
Figure 3
376
A pharmacological relevance of our results is in fact noticeable. The values of pKa1T and
377
pKa2T should be discussed in two different areas of interest. The first one is elucidation of
Ac ce p
te
d
M
an
us
cr
ip t
349
15 Page 15 of 36
WAR pharmacokinetics. It is commonly known that a role of the phase I oxidative
379
metabolism of drugs mediated by CYP enzymes is to increase a drug hydrophilicity and
380
solubility. This increase may stem from the two reasons: appearance of novel ionizable
381
groups yielding charge upon dissociation, or formation of groups which do not dissociate
382
but change an overall character of given molecule sites. One of the decisive factors for the
383
prevalence of one effect over other one is the relation between pKa of this group and pH of
384
environment. Formation of the second OH group during phase I metabolism of WAR, as can
385
be concluded from pKa2T values, does not trigger an increase in overall charge of molecule at
386
physiological pH. Yet more interesting conclusions can be drawn when referring to the
387
pharmacological relevance of pKa1T. Regarding the range of pH of natural intracellular
388
environment or body fluids like blood, urine or bile – generally from 5.5 to 8.0, one can
389
conclude that both parent compound and every hydroxylated metabolite occur rather in the
390
ionized form. The intriguing exception can be however W10 with pKa1≈ 6. In this case the
391
pKa1value may suffice for existence of significant fraction of the non-ionized form, far more
392
hydrophobic, especially in slightly acidic matrices like e.g. urine (pH ≈ 5.5). The conclusion
393
could be that phase I metabolism leading to W10 may cause in fact the adverse effects on
394
solubility in some matrices. This is of special relevance because (R)-W10 is one of the main
395
metabolites found in biological material collected from patients.
396
The second pharmacologically-relevant area of interest is, still enigmatic, the phase II
397
conjugative metabolism of WAR. It is quite likely that the reactions of glucuronidation
398
catalyzed by UDP-glucuronosyltransferases depend on hydroxylation site, thus, exert high
399
regioselectivity [31-33]. This may be in consequence one of the reasons why a diverse
400
profile of WAR phase II metabolites found in vivo is correlated with the expression of given
401
CYP enzymes. The last issue is a hypothetic modulation of WAR metabolism directly by
402
hydroxywarfarins. The two different phenomena should be considered, either a
403
transcriptional modulation of genes expression or a direct inhibition of CYP enzymes e.g. by
404
blocking of the active sites of enzymes. One may predict that these phenomena are
405
somehow related to structure and conformation of the particular metabolites and
406
respective stabilities of O–H bonds. In consequence, their pKa values would be of greater
407
biochemical relevance.
Ac ce p
te
d
M
an
us
cr
ip t
378
16 Page 16 of 36
3.6. pKa-related migration profile
409
Apart from the theoretical structure-activity relationships and the pharmacological
410
importance of the data obtained in this work, another interesting issue is a separation of all
411
compounds from each other mixed in a single sample. To this end, the CE method based on
412
differences in already known pKa values seems to be the simplest opportunity. To attempt
413
this approach, CE-based separation has been conducted at pH 10.4. This pH has been
414
selected as optimal for obtaining different ionization levels of the particular compounds,
415
thus also high selectivity. In Fig.4 the migration profile has been depicted.
416
Figure 4
417
It has occurred that the individual migration times properly reflect the order of pKa2 values.
418
In particular, almost total but still slightly different ionization can be predicted from the
419
position of W8 and W7 peaks, the ionization level below 50% from W6 peak, while yet
420
lower but still visible ionization from W3 and W4 peaks. The lack of separation between W3
421
and W4 proves that their pKa2 is very similar. Interestingly, W10 has been separated from
422
WAR, and this is probably caused not by different ionization, but rather by a different
423
hydrodynamic resistance during electromigration. Hypothetically, W10 as capable of
424
forming stronger intramolecular hydrogen bonds may migrate in a more occlusive way, and
425
thus, actually a more mobile form than WAR. This hypothesis is supported by different
426
shape of the W10 molecule presented in Fig.2. Further investigation of these analytically
427
important conundrums will yet be the aim of our future work.
428
4. Conclusions
429
The classical CE method has allowed us to determine thermodynamic pKa values with the
430
aid of extended Debye-Hückel model, upon its experimental verification. The CE-DAD
431
method has occurred to be very useful for determination of pKa1, and provides totally
432
different physicochemical background than CE. Its additional advantage is that information
433
about concentration of analyte and absorptivity coefficient values is not needed. Both
434
methods are complementary and can be applied interchangeably. Accuracy of the IS-CE is
435
strictly dependent on degree of similarity between an internal standard and an analyte, and
436
this method appears to be especially useful for qualitative predictions. Due to vast
Ac ce p
te
d
M
an
us
cr
ip t
408
17 Page 17 of 36
economization of time and materials, however, its potential is still worth to be appreciated
438
and further tested.
439
One may conclude that the values of pKa1T and pKa2T obtained in the current work are
440
important from the analytical, pharmacological and theoretical points of view. The acid-
441
base properties depend on hydroxylation site. In particular, relatively large variations are
442
observed between pKa2T values. W10 exhibits specific properties, namely it undergoes only
443
one dissociation contrary to other hydroxywarfarins, and its pKa1T surpasses the values
444
noted for any other compounds. These features are related to the specific molecular
445
structure and alcoholic character of the second OH group. It turns out that phase I
446
metabolism leads to creation of more hydrophilic molecules, however still only singly
447
ionized at physiological pH, similarly as the parent drug. The specific conformations and
448
hydrogen binding energies, as well as the actual proximity of the given hydrogen donors
449
and acceptors, may be very important factors for explanation of regioselective phase II
450
metabolism and modulation of CYP-mediated metabolism, not studied in depth so far.
451
Selection of proper pH enables the efficient separation by CE of almost all tested
452
compounds mixed together, based on their pKa2 values.
453
The authors declare no competing financial interest.
454
Acknowledgments
455
Author Paweł Nowak has received the financial support from Krakowskie Konsorcjum
456
“Materia-Energia-Przyszłość” within the subsidy KNOW.
457
The research was carried out with equipment purchased with financial support from the
458
European Regional Development Fund within the framework of the Polish Innovation
459
Economy Operational Programme (contract no. POIG.0 2.01.00-12-0 23/08).
Ac ce p
te
d
M
an
us
cr
ip t
437
460
18 Page 18 of 36
References
461
[1] L.S. Kaminsky, Z-Y. Zhang, Human P450 Metabolism of Warfarin, Pharmacol. Ther. 73
462
(1997) 67–74
463
[2] M. Gebauer, Synthesis and structure-activity relationships of novel warfarin derivatives,
464
Bioorg. Med. Chem. 15 (2007) 2414–2420.
465
[3] S. Babić, A.J.M. Horvat, D. Mutavdžić-Pavlović, M. Kaštelan-Macan, Determination of pKa
466
values of active pharmaceutical ingredients, Trac-Trend. Anal. Chem. 26, (2007) 1043–
467
1061.
468
[4] I. Ghosh, W.M. Nau, The strategic use of supramolecular pKa shifts to enhance the
469
bioavailability of drugs, Adv. Drug Deliver. Rev. 64 (2012) 764–783.
470
[5] H. Wan, A. Holmen, M. Någård, W. Lindberg, Rapid screening of pKa values of
471
pharmaceuticals by pressure-aassisted capillary electrophoresis combined with short-end
472
injection, J. Chromatogr. A 979 (2002) 369–377.
473
[6] J.M. Cabot, E. Fuguet, C. Ràfols, M. Rosés, Fast high-throughput method for the
474
determination of acidity constants by capillary electrophoresis. II. Acidic internal standards,
475
J. Chromatogr. A 1217 (2010) 8340–8345.
476
[7] M. Shalaeva, J. Kenseth, F. Lombardo, A. Bastin, Measurement of Dissociation Constants
477
(pKa Values) of Organic Compounds by Multiplexed Capillary Electrophoresis Using
478
Aqueous and Cosolvent Buffers, J. Pharm. Sci. 97 (2008) 2581–2606.
479
[8] Y. Ishihama, M. Nakamura, T. Miwa, T. Kajima, N. Asakawa, A Rapid Method for pKa
480
Determination of Drugs Using Pressure-Assisted Capillary Electrophoresis with Photodiode
481
Array Detection in Drug Discovery, J. Pharm. Sci. 91 (2002) 933–942.
482
[9] P. Nowak, M. Garnysz, M.P. Mitoraj, F. Sagan, M. Woźniakiewicz, P. Kościelniak,
483
Analytical aspects of achiral and cyclodextrin-mediated capillary electrophoresis of
484
warfarin and its two main derivatives assisted by theoretical modelling, J. Chromatogr. A
485
1377 (2015) 106–113.
Ac ce p
te
d
M
an
us
cr
ip t
460
19 Page 19 of 36
[10] J. Reijenga, H. van Arno, L. van Antonie, B. Teunissen, Development of Methods for the
487
Determination of pKa Values. Anal. Chem. Insights 8 (2013) 53−71.
488
[11] P. Nowak, M. Woźniakiewicz, P. Kościelniak, Application of capillary electrophoresis in
489
determination of acid dissociation constant (pKa) values, J. Chromatogr. A 1377 (2015) 1–
490
12.
491
[12] S.K. Poole, S. Patel, K. Dehring, H. Workman, C.F. Poole, Determination of acid
492
dissociation constants by capillary electrophoresis, J. Chromatogr. A 1037 (2004) 445–454.
493
[13] M. Riesová, J. Svobodová, K. Ušelová, Z. Tošner, I. Zusková, B. Gaš, Determination of
494
thermodynamic values of acidic dissociation constants and complexation constants of
495
profens and their utilization for optimization of separation conditions by Simul 5 Complex,
496
J. Chromatogr. A 1364 (2014) 276–288.
497
[14] E.C. Demiralay, Z. Ustun, Y.D. Daldal, Estimation of thermodynamic acidity constants of
498
some penicillinase-resistant penicillins, J. Pharmaceut. Biomed. 91 (2014) 7–11.
499
[15] J.L. Wang, X.J. Xu, D.Y. Chen, Determination of pK(a) values of a triptolide derivative
500
and its impurities by pressure-assisted capillary electrophoresis, J. Pharmaceut. Biomed. 88
501
(2014) 22–26.
502
[16] V. Solínová, V. Kašička, Determination of acidity constants and ionic mobilities of
503
polyprotic peptide hormones by CE, Electrophoresis 34 (2013) 2655–2665.
504
[17] H.X. Ren, L.C. Wang, X.S. Wang, X. Liu, S.X. Jiang, Measurement of acid dissociation
505
constants and ionic mobilities of 3-nitro-tyrosine and 3-chloro-tyrosine by capillary zone
506
electrophoresis, J. Pharmaceut. Biomed. 77 (2013) 83–87.
507
[18] K. Vceláková, I. Zusková, E. Kenndler, B. Gaš. Determination of cationic mobilities and
508
pK(a) values of 22 amino acids by capillary zone electrophoresis, Electrophoresis 25 (2004)
509
309–317.
510
[19] V. Solínová, V. Kašička, D. Koval, M. Cesnek, A. Holý. Determination of acid-base
511
dissociation constants of amino- and guanidinopurine nucleotide analogs and related
512
compounds by capillary zone electrophoresis, Electrophoresis 27 (2006) 1006–1019.
Ac ce p
te
d
M
an
us
cr
ip t
486
20 Page 20 of 36
[20] Z. Glatz, Application of short-end injection procedure in CE, Electrophoresis 34 (2013)
514
631–642.
515
[21] X. Wu, S. Gong, T. Bo, Y. Liao, H. Liu, Determination of dissociation constants of
516
pharmacologically active xanthones by capillary zone electrophoresis with diode array
517
detection, J. Chromatogr. A 1061 (2004) 217–223.
518
[22] S.P. Ozkorucuklu, J.L. Beltrán, G. Fonrodona, D. Barrón, G. Alsancak, J. Barbosa,
519
Determination of Dissociation Constants of Some Hydroxylated Benzoic and Cinnamic Acids
520
in Water from Mobility and Spectroscopic Data Obtained by CE-DAD, J. Chem. Eng. Data 54
521
(2009) 807–811.
522
[23] E. Fuguet, C. Ràfols, M. Rosés, A fast high throughput method for the determination of
523
acidity constants by capillary electrophoresis. 3. Basic internal standards, J.
524
Chromatography A 1218 (2011) 3928–3934.
525
[24] J.M. Cabot, E. Fuguet, C. Ràfols, M. Rosés, Determination of acidity constants by the
526
capillary electrophoresis internal standard method. IV. Polyprotic compounds, J.
527
Chromatogr. A 1279 (2013) 108–116.
528
[25] R.S. Gravador, E. Fuguet, C. Ràfols, M. Rosés, Temperature variation effects on the
529
determination of acidity constants through the internal standard–capillary electrophoresis
530
method, Electrophoresis 34 (2013) 1203–1211.
531
[26] J.M. Cabot, E. Fuguet, M. Rosés, Internal Standard Capillary Electrophoresis as a High
532
Throughput Method for pKa Determination in Drug Discovery and Development, ACS Comb.
533
16 (2014) 518–525.
534
[27] J.M. Cabot, E. Fuguet, M. Rosés, Determination of acidity constants of sparingly soluble
535
drugs in aqueous solution by the IS-CE method, Electrophoresis 35 (2014) 3564–3569.
536
[28] P. Debye, P. Huckel, On the theory of electrolytes, Phyz. Z. 24 (1923) 305–325.
537
[29] G. teVelde, F.M. Bickelhaupt, E.J. Baerends, C. Fonseca Guerra, S.J.A. van Gisbergen, J.G.
538
Snijders, T. Ziegler, Chemistry with ADF, J. Comput. Chem. 22 (2001) 931–967.
Ac ce p
te
d
M
an
us
cr
ip t
513
21 Page 21 of 36
[30] B. Castleberry, E.J. Valente, D.S. Eggleston, Open-cyclic warfarin isomerism: 5-
540
hydroxywarfarin, J. Cryst. Spectrosc. 20 (1990) 583–593.
541
[31] C.P. Pugh, D.L. Pouncey, J.H. Hartman, R. Nshimiyimana, L.P. Desrochers, T.E. Goodwin,
542
G. Boysen, G.P. Miller, Multiple UDP-glucuronosyltransferases in human liver microsomes
543
glucuronidate both R- and S-7-hydroxywarfarin into two metabolites, Arch. Biochem.
544
Biophys. 564 (2014) 244–253.
545
[32] D.R. Jones, G.P. Miller, Assays and applications in warfarin metabolism: What we know,
546
how we know it and what we need to know, Expert Opin. Drug Met. 7 (2011) 857–874.
547
[33] D.R. Jones, J.H. Moran, G.P. Miller, Warfarin and UDP-glucuronosyltransferases: Writing
548
a new chapter of metabolism, Drug Met. Rev. 42 (2010) 53–59.
an
us
cr
ip t
539
549
Ac ce p
te
d
M
550
22 Page 22 of 36
550 Fig.1. Absorbance spectra obtained during CE-DAD separation of WAR at pH 2.5 and 8.0,
552
respectively. Under these conditions only one form: the acidic (protonated) or the basic
553
(deprotonated) exists in solution. Vertical dashed lines indicate the wavelengths (280 and
554
308 nm) for which the absorbance values were recorded and further processed.
555
Fig.2. DFT/TZP/BLYP-D3 optimized structures of WAR and hydroxywarfarins together with
556
the selected bond distances (in Å).
557
Fig.3. Tautomeric forms of WAR, potential hydrogen donors and acceptors have been
558
pointed by red and blue colors, respectively.
559
Fig.4. Electropherogram presenting the separation of all analytes mixed together,
560
conducted at pH 10.4 in 25 mM ionic strength (in the
an
us
cr
ip t
551
Ac ce p
te
d
M
561
23 Page 23 of 36
561
Table 1. Composition and predicted pH of all buffering solutions prepared for experiments,
562
calculated for 50 mL total volume and 100 mM ionic strength.
564 565
cr
ip t
NaH2PO4 (100 mM) 4.80 CH3COONa (500 mM) 10.00 10.00 Na2HPO4 (100 mM) 4.70 13.29 16.25 NaOH (1 M) 1.02 2.13 2.50 KCl (1 M) 4.36 0.45
an
M
d
Ac ce p
563
H3PO4 (100 mM) 18.52 CH3COOH (500 mM) 14.16 2.52 NaH2PO4 (100 mM) 3.59 1.01 0.12 Na2B4O7·10H2O (50 mM) 39.77 28.67 25.05 NaOH (1 M) 0.64 4.55
te
Phosphate buffer I 2.50 Acetic buffer 4.50 5.25 Phosphate buffer II 6.00 7.00 8.00 Borate buffer 9.40 10.00 11.00 NaOH / KCl 12.00 13.00
Buffer composition [mL]
us
pH
24 Page 24 of 36
565
Table 2. Comparison of CE and CE-DAD methods used for pKa1 determination
CE-DAD
CE-DAD
CE-DAD
4.87 ± 0.02
4.71 ± 0.03
4.86 ± 0.03
4.86 ± 0.03
W3
4.87 ± 0.04
4.77 ± 0.02
4.89 ± 0.02
4.88 ± 0.02
W4
4.93 ± 0.04
4.77 ± 0.03
4.91 ± 0.03
4.91 ± 0.03
W6
4.92 ± 0.07
4.62 ± 0.02
4.92 ± 0.02
4.92 ± 0.02
W7
5.09 ± 0.07
5.08 ± 0.04
5.10 ± 0.03
5.10 ± 0.04
W8
4.88 ± 0.06
4.64 ± 0.03
4.84 ± 0.03
4.84 ± 0.03
W10
5.80 ± 0.03
5.68 ± 0.03
5.79 ± 0.02
5.79 ± 0.02
an
566
us
WAR
cr
CE (ref)
ip t
pKa1
M
567
Ac ce p
te
d
568
25 Page 25 of 36
Table 3. Apparent pKa values (in 100 mM ionic strength) obtained by the IS-CE method for
569
all possible combinations between potential internal standards and analytes, based on the
570
selected data obtained by the CE method. Internal standard pH of partial
WAR
W3
W4
W6
W7
4.88
4.91
4.90
4.94
4.85
4.84
4.85
4.71
4.71
4.67
ionization 4.87
W6
W7
W8
W10
4.92
4.67
4.66
4.90
4.89
4.93
4.88
5.11
4.87
4.90
4.88
4.95
4.83
4.82
5.06
4.93
4.97
4.92
5.14
4.94
4.94
5.01
4.89
4.89
4.88
5.12
4.96
4.91
5.14
4.94
4.92
4.99
4.93
4.92
5.15
5.04
5.27
5.07
5.14
5.08
5.32
4.86
pH ≈ 5.25
4.89
pH ≈ 6.00
5.03
4.83
pH ≈ 4.50
4.89
4.91
pH ≈ 5.25
4.96
4.93
4.93
4.96
pH ≈ 6.00
5.09
4.93
pH ≈ 4.50
4.89
4.90
4.92
pH ≈ 5.25
4.94
4.92
4.91
pH ≈ 6.00
5.12
4.96
4.96
pH ≈ 4.50
5.02
5.03
5.05
5.05
pH ≈ 5.25
5.09
5.06
5.06
5.07
pH ≈ 6.00
5.29
5.13
5.13
5.08
pH ≈ 4.50
4.86
4.87
4.89
4.89
4.93
pH ≈ 5.25
4.90
4.87
4.87
4.88
4.90
pH ≈ 6.00
5.09
4.93
4.93
4.88
4.89
pH ≈ 4.50
5.55
5.56
5.59
5.58
5.62
5.57
pH ≈ 5.25
5.75
5.72
5.72
5.73
5.75
5.73
pH ≈ 6.00
5.77
5.61
5.61
5.57
5.57
5.56
4.87
4.87 4.87
te
W4
4.85
pH ≈ 4.50
Ac ce p
W3
4.89
4.87
M
pH ≈ 6.00
5.12
us
pH ≈ 5.25
d
WAR
W10
4.89
an
pH ≈ 4.50
W8
cr
Analyte
ip t
568
4.92
5.09
5.11 4.88
4.95 5.12 5.80
26 Page 26 of 36
pH ≈ 10.50
11.23
pH ≈ 11.00
W7
10.90 10.36 10.37
11.23
10.93
pH ≈ 9.90
11.22
10.82 10.23 10.22
pH ≈ 10.50
11.20 11.22 10.87 10.34 10.35
pH ≈ 11.00
11.22
pH ≈ 9.90
11.09
11.09
pH ≈ 10.50
11.02
11.04 10.69 10.15 10.16
pH ≈ 11.00
10.99
11.00
pH ≈ 9.90
10.38
10.38
pH ≈ 10.50
10.26
10.27
pH ≈ 9.90
10.49
10.50
10.10
9.51
pH ≈ 10.50
10.36
10.37
10.03
9.49
10.91
pH ≈ 11.00
9.98 9.93
9.39
9.38 9.40
9.50
d
W8
10.10 10.09
M
pH ≈ 11.00
cr
W6
11.25
us
W4
10.83 10.24 10.24
an
W3
11.23
ip t
pH ≈ 9.90
As pKa of the internal standard, the values calculated from the CE method (in 100 mM ionic
572
strength) have been used.
574
Ac ce p
573
te
571
27 Page 27 of 36
Table 4. Apparent (valid at given ionic strength) and thermodynamic (standardized to zero ionic strength) pKa
pKa2 100 mM
pKa2T
4.87 ± 0.02
4.89 ± 0.04
4.99 ± 0.03
4.99
4.87 ± 0.04
4.89 ± 0.04
4.93 ± 0.03
4.97
11.23 ± 0.06
11.55
W4
4.93 ± 0.04
4.98 ± 0.05
4.97 ± 0.04
5.03
11.22 ± 0.06
11.54
W6
4.92 ± 0.07
4.94 ± 0.06
4.97 ± 0.04
5.01
10.69 ± 0.08
11.01
W7
5.09 ± 0.07
5.09 ± 0.04
5.10 ± 0.04
5.16
W8
4.88 ± 0.06
4.88 ± 0.03
4.93 ± 0.04
4.97
W10
5.80 ± 0.03
5.93 ± 0.03
5.92 ± 0.05
W3
Δcalculated
0.03
Δpredicted
0.04 Δcalculated
581 582
9.71
9.50 ± 0.12
9.82
5.95
0.03
Δpredicted 0.02 has been calculated as an average from three values obtained for three distinct ionic strength by
Ac ce p
576 577 578 579 580
pKa1T
9.39 ± 0.16
d
WAR
cr
pKa1T
us
pKa1 10 mM
an
pKa1 25 mM
M
pKa1 100 mM
ip t
values.
te
574 575
applying respective Debye-Hückel corrections; Δcalculated denotes an average difference between apparent pKa for the same compound and obtained experimentally for two distinct ionic strengths; Δpredicted denotes a difference between apparent pKa for the same compound and predicted for two distinct ionic strengths from Debye-Hückel model.
long-end injection mode).
28 Page 28 of 36
Ac
ce
pt
ed
M
an
us
cr
i
*Graphical Abstract
Page 29 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1 bw print only
Page 30 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1 color web only
Page 31 of 36
Ac ce p
te
d
M
an
us
cr
ip t
Figure 2 revised color- web only
Page 32 of 36
Ac ce p
te
d
M
an
us
cr
ip t
Figure 2 revised bw- print only
Page 33 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3 bw print only
Page 34 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3 color web only
Page 35 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4
Page 36 of 36
S0731-7085(15)00260-5 http://dx.doi.org/doi:10.1016/j.jpba.2015.04.027 PBA 10063
To appear in:
Journal of Pharmaceutical and Biomedical Analysis
Received date: Revised date: Accepted date:
23-12-2014 17-4-2015 18-4-2015
Please cite this article as: P. Nowak, P. Olechowska, M. Mitoraj, M. Wo´zniakiewicz, P. Ko´scielniak, Determination of acid dissociation constants of warfarin and hydroxywarfarins by capillary electrophoresis, Journal of Pharmaceutical and Biomedical Analysis (2015), http://dx.doi.org/10.1016/j.jpba.2015.04.027 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Determination of acid dissociation constants of warfarin and
2
hydroxywarfarins by capillary electrophoresis
3
Paweł Nowaka, Paulina Olechowskaa, Mariusz Mitorajb, Michał Woźniakiewicza*, Paweł
4
Kościelniaka
5
a Jagiellonian
6
Kraków, Poland
7
b
8
Chemistry, Kraków, Poland
9
Corresponding Author:
ip t
1
cr
University in Kraków, Faculty of Chemistry, Department of Analytical Chemistry,
an
us
Jagiellonian University in Kraków, Faculty of Chemistry, Department of Theoretical
*Michał Woźniakiewicz, PhD, Jagiellonian University in Kraków, Faculty of Chemistry,
11
Department of Analytical Chemistry, Kraków, Poland, Ingardena St. 3, 30-060 Kraków,
12
Poland, [email protected], tel./fax: +48 12 663 20 84
M
10
Ac ce p
te
d
13
1 Page 1 of 36
Highlights:
14
o Two acid dissociation equilibria have been described
15
o Thermodynamic pKa values have been determined
16
o CE, CE-DAD and IS-CE methods have been employed
17
o Pharmacological relevance has been presented and discussed
18
o Theoretical explanations of pKa values have been attempted
us
cr
ip t
13
an
19 20 Abstract
22
In this work the acid dissociation constants – pKa of warfarin and its all important
23
oxidative metabolites have been determined by capillary electrophoresis-based
24
methods. It has resulted in a complete description of two acid-base dissociation
25
equilibria, yet not investigated experimentally for phase I metabolites of warfarin.
26
The capillary electrophoresis (CE) method based on the relation between effective
27
electrophoretic mobilities and pH has proven to be a suitable tool for pKa
28
determination, while the spectrophotometric (CE-DAD) and the internal standard
29
methods (IS-CE), have appeared to be promising alternative approaches. The CE-DAD
30
approach based on the change in absorbance spectra between the acidic and basic
31
forms is a combination between capillary electrophoresis and spectrophotometric
32
titration, and yields very consistent values of pKa1 with CE. The IS-CE, in turn, enables
33
an estimation of pKa1 and pKa2 from only two analytical runs, however, less accurate
34
than CE and CE-DAD. The Debye-Hückel model has been confirmed experimentally as
35
a good predictor of pKa values at various ionic strengths. Therefore, it has been used
36
in determination of thermodynamic pKa1 and pKa2, referring to the zero ionic
Ac ce p
te
d
M
21
2 Page 2 of 36
strength. The results are important from the analytical, pharmacological, and
38
theoretical points of view.
39
Keywords: acid dissociation constant, capillary electrophoresis, Debye-Hückel model, drug
40
metabolism, hydroxywarfarins, warfarin
41
Abbreviations:
42
DMSO - dimethyl sulfoxide
43
EOF – electroosmotic flow
44
WAR – warfarin
45
W3 – 3’-hydroxywarfarin
46
W4 – 4’-hydroxywarfarin
47
W6 – 6-hydroxywarfarin
48
W7 – 7-hydroxywarfarin
49
W8 – 8-hydroxywarfarin
50
W10 – 10-hydroxywarfarin
cr us an M d te
Ac ce p
51
ip t
37
3 Page 3 of 36
51 1. Introduction
53
Warfarin (WAR) [3-(α-acetonylbenzyl)-4-hydroxycoumarin] is a widely used drug
54
exhibiting anticoagulant activity, applied in the prevention of thrombosis and
55
thromboembolism, i.e. formation of blood clots and vascular occlusions [1,2]. Its routine
56
application as an effective anticoagulant is however quite problematic owing to complex
57
oxidative metabolism, involving several different isoforms of cytochrome P450 enzyme
58
(CYP). The main oxidative metabolites of WAR are hydroxywarfarins: 3’-hydroxywarfarin
59
(W3), 4’-hydroxywarfarin (W4), 6-hydroxywarfarin (W6), 7-hydroxywarfarin (W7), 8-
60
hydroxywarfarin (W8) and 10-hydroxywarfarin (W10). CYP-mediated metabolism of WAR
61
is highly regio- and enantioselective, the most popular hydroxywarfarins found in vivo are
62
(S)-W7, (R)-W4 and (R)-W10 [1].
63
The acid dissociation constant, usually expressed as its pKa value, is one of the most
64
important parameters describing physicochemical properties of drugs [3,4]. It characterizes
65
acid-base equilibrium of the respective ionizable groups in solution, thus, indicates their
66
dissociation (deprotonation) potential at a given pH. The acidic and basic forms of drugs
67
differ in charge – at least one of them is ionized, therefore, they also may differ in other key
68
properties like water solubility, membrane permeability, affinity of association with
69
proteins, and finally, therapeutic activities. As it was demonstrated in the literature,
70
therapeutic activity of WAR is also dependent on its ionization level [2]. For that reason, the
71
exact value of pKa for WAR is important from the pharmacological point of view, and it was
72
determined many times in the past [5-9]. Another issue is acid-base properties of WAR
73
metabolites, yet not fully recognized. Hydroxywarfarins, contrary to the parent drug, are
74
supposed to be capable of a double dissociation, and this fact complicates predictions of
75
their ionization-related properties. Therefore, experimental determination of their exact
76
pKa1 and pKa2 values is undeniably necessary for the holistic understanding of WAR
77
pharmacokinetics.
78
There are many various analytical techniques that allow for accurate estimation of pKa
79
values in a sound and rapid way [3,10]. Capillary electrophoresis (CE) belongs to the
Ac ce p
te
d
M
an
us
cr
ip t
52
4 Page 4 of 36
subgroup of separation techniques, and in the recent years, has gained particular interest as
81
an efficient tool for pKa determination [11-17]. Its growing popularity originates from the
82
clear analytical benefits: (i) very small consumption of samples and buffers; (ii) relatively
83
high throughput and simple automation; (iii) possibility to separate the sample component
84
from each other, including impurities; (iv) and the accuracy, typically within 0.05 pH units.
85
In a standard approach based on CE, pKa values are calculated from a curve presenting
86
dependence of electrophoretic mobility on pH. For comparative purposes, the use of
87
thermodynamic pKa (pKaT) corresponding to the zero ionic strength and standard
88
temperature 25 ̊C is more suitable than the apparent pKa, determined experimentally in
89
specific conditions of ionic strength and temperature [18,19].
90
In this work, the acid-base equilibria of WAR and six hydroxywarfarins have been
91
investigated and characterized by determination of the respective pKa values. To the best of
92
our knowledge, this is the first experimental study of all pharmacologically important
93
hydroxywarfarins, and in addition, it covers both dissociation equilibria. The standard CE-
94
based method has been used to obtain the most accurate results. Afterwards, usefulness of
95
two alternative approaches utilizing CE have been investigated: the spectrophotometric
96
method (CE-DAD), based on the changes in molecular spectra recorded by DAD detection
97
system; and the internal standard-based method (IS-CE), restricted only to two analytical
98
runs and the known pKa value of an internal standard. The variation of pKa1 with changing
99
ionic strength has been studied experimentally, and compared with the values predicted by
Ac ce p
te
d
M
an
us
cr
ip t
80
100
the Debye-Hückel model. Finally, pKa1T and pKa2T values have been found for each
101
compound, which correspond to the standardized temperature and zero ionic strength. It
102
enables creation of a more complete picture of pharmacologically-important properties of
103
WAR metabolites, theoretical considerations referring to structure-property relationships,
104
and better understanding of their migration profile observed during CE separations.
105
2. Material and methods
106
2.1. Instrumentation
107
WAR (racemic mixture) was supplied by Sigma-Aldrich (St. Louis, MO, USA), W3, W4, W6,
108
W7, W8, and W10 (all racemic mixtures) by LGC Standards (Teddington, UK), while all 5 Page 5 of 36
other chemicals by Avantor Performance Materials Poland. S. A. (Gliwice, Poland). All
110
standard solutions were prepared in the deionized water (MilliQ, Merck-Millipore Billerica,
111
MA, USA) and filtered through the 0.45 μm regenerated cellulose membrane, then degassed
112
by centrifugation. Standard concentration of analytes was 0.1 mg/mL, all analytes were
113
dissolved in water/methanol (1:1 v/v) mixture. Dimethyl sulfoxide (DMSO) was used as the
114
electroosmotic flow (EOF) marker, in 0.2% (v/v) concentration.
115
The P/ACE MDQ Capillary Electrophoresis System (Brea, CA, USA) equipped with a diode
116
array detector was used for all CE-based experiments, using the bare fused-silica capillary
117
of 60 cm total length and 75 μm internal diameter. Separations were conducted in a short-
118
injection mode [20], unless stated otherwise, using a 10 cm long outlet capillary part. Owing
119
to this fact, the total separation time could be vastly reduced. Sample injection was
120
conducted using forward pressure 0.4 psi for 4 s. During separations, 15 kV voltage (anode
121
at injection end) and the additional forward pressure of 0.4 psi were applied. The current
122
measured during separations was between 25 – 125 µA. Capillary was conditioned at 25 ̊C,
123
using the liquid cooling system. Every time, DAD detector collected the whole spectra
124
within 200-600 nm. Signal recorded at 200, 280 and 308 nm was used for the further
125
analysis.
126
Capillary rinsing was conducted between runs applying pressure of 137,9 kPa (20 psi). The
127
particular steps of rinsing were: deionized water for 1 min, 0.1 M NaOH for 2 min, and
128
background electrolyte (BGE) for 2 min. During the first use of the capillary at a working
129
day: methanol for 5 min, 0.1 M HCl for 2 min, deionized water for 2 min, 0.1 M NaOH for 10
130
min, and BGE for 10 min were applied. For the fresh capillary conditioning, the latter
131
sequence was used, but the duration of each individual step was doubled.
cr
us
an
M
d
te
Ac ce p
132
ip t
109
133
2.2. Buffering solutions
134
Composition of the given BGEs was calculated with the use of PHoEBuS 1.3 software by
135
Analis (Namur, Belgium), setting 100 mM as the selected value of ionic strength. The buffers
136
of lower ionic strength were prepared by dilution with deionized water. The obtained
6 Page 6 of 36
recipes for BGEs preparation have been presented in Table 1. The values of pH were
138
verified experimentally prior to CE separations.
139
Table 1
140
2.3. Methods for pKa determination
141
2.3.1. Standard electrophoretic method (CE)
142
In the standard CE approach, pKa values have been estimated by applying the non-linear
143
regression model to the obtained dependency between effective electrophoretic mobility
144
(µeff) and pH. According to the Eq.1, µeff can be calculated from the respective migrations
145
times of analyte and EOF marker:
an
us
cr
ip t
137
(1)
M
146 147
where μeff and μobs are the effective and observed electrophoretic mobilities of analyte (m2 V-
148
1 s-1),
149
total and effective capillary lengths (m), 0.60 and 0.10 m, respectively; V is the separation
150
voltage (V); tobs is the measured migration time of analyte (s), while teof is the time
151
measured for neutral marker of EOF – DMSO (s).
152
The relation between µeff and pH is described by Eq.2 and Eq.3, for the monoprotic and
153
diprotic acid, respectively:
155
d
te
Ac ce p
154
respectively; μeof is the mobility of electroosmotic flow (m2 V-1 s-1); Ltot and Leff are the
(2)
(3)
156
where α values are fitting constants equal to electrophoretic mobility of the deprotonated
157
form of the acid with subscript 1 and 2 equal to the order of dissociation. The pKa values are
158
equal to the pH of inflection points of the fitted sigmoidal-shape curves.
159
2.3.2. Spectrophotometric method (CE-DAD)
7 Page 7 of 36
To enable determination of pKa1 by CE-DAD method [21,22], the absorbance of WAR and its
161
derivatives was measured at the maximum of electrophoretic peaks at 280 and 308 nm.
162
These wavelengths are close to maxima of absorbance of the acidic and basic forms,
163
respectively. Afterwards, the parameter β was calculated according to the three different,
164
arbitrary proposed definitions:
(5)
167
(6)
an
166
cr
(4)
us
165
ip t
160
where β1, β2 and β3 present different calculation methods, while A280 and A308 are the values
169
of absorbance at the given wavelength (nm). As it has been shown in Fig.1, there is a strong
170
shift in absorbance at these wavelength between the acidic and basic forms of WAR. Most
171
importantly, the direction of this shift is opposite for 280 and 308 nm. These wavelength
172
are applicable also for all hydroxywarfarins, whose spectra differ between each other,
173
mostly in the case of W7 (not shown).
174
Figure 1
175
The changing value of β together with changing pH is observed, if the ratio of molar
176
absorptivity coefficients at 280 and 308 nm is considerably different for acidic and basic
177
form. In that case, an every partially ionized state corresponds to the value of β from the
178
range between the two limiting values, related to the protonated and deprotonated forms,
179
respectively. In other words, such parameter is then sensitive on a changing contribution of
180
the acidic and basic forms, characterized by different shape of absorption spectrum. The
181
plot of β against pH gives the curve of similar sigmoidal shape as in the case of the CE
182
method, and analogously, the pKa value is indicated by inflection point located on this curve.
183
Ac ce p
te
d
M
168
2.3.3. Internal-standard method (IS-CE)
8 Page 8 of 36
The IS-CE method was proposed several years ago by Rosés’ scientific group [23-27]. It is
185
based on the use of internal standard of known pKa value, similar to the predicted pKa of
186
analyte, and requires only two runs at two distinct pH values. The first pH should provide
187
total dissociation of both the internal standard and the analyte, while the second one, only
188
partial dissociation. Then, the sought pKa can be calculated according to the following
189
equation:
us
192
given that, (8)
an
191
(7)
cr
190
ip t
184
where pKaIS refers to the internal standard, μA- is the effective electrophoretic mobility
194
measured when both compounds are totally deprotonated, while in the case of μeff,
195
supposed to be partially ionized. To enable the accurate analysis, the analyte and the
196
internal standard are injected together from the same vial. The values of Q are calculated
197
for the analyte and the internal standard independently.
198
Further information on the determination of pKa values has been presented and discussed
199
in the next section of this work.
200
2.3.4. Debye-Hückel model
201
The estimations of thermodynamic pKa were done by using the following equation [18,19]:
d
te
Ac ce p
202
M
193
(9)
203
where γ denotes activity coefficient of the particular species of an analyte: acidic – HA, and
204
basic – A, with a given charge (z). Activity coefficients were calculated based on the
205
extended law of Debye-Hückel [28]:
206
(10)
9 Page 9 of 36
where A and B are the parameters dependent on dielectric constant and temperature, in
208
water at 25°C they are 0.509 and 0.33, respectively, z is the charge of the ion, I is the ionic
209
strength, while a denotes the radius of hydrated ion. The approximate value of a = 5 Å was
210
used for calculations.
211
3. Results and discussion
212
3.1. Application of the CE method
213
At the beginning, pKa1 and pKa2 of WAR and all hydroxywarfarins have been determined by
214
the standard CE method. The measurements of μeff in buffers with gradually increasing pH,
215
but constant ionic strength of 100 mM and temperature of 25 ̊C, have been conducted
216
separately for each compound. Then, the non-linear regression model has been applied,
217
resulting in the fitted curves characterized by one or two inflection points indicating the
218
respective pKa values, according to Eq.2 and Eq.3. All data obtained in this stage have been
219
presented in Fig.S-1 (left part).
220
As it can be seen, both pKa1 and pKa2 have been found. W10 however, in spite of possessing
221
two hydroxyl groups, does not exhibit the second dissociation, similarly as WAR. It
222
originates from alcoholic character of the second hydroxyl group. Interestingly, pKa1 of W10
223
is nearly 1.0 pH unit higher than in the case of the other compounds. A slight difference, in
224
the range of 0.1 – 0.2 pH units, is noted also between W7 and the rest of compounds, except
225
W10. The variation of pKa1 between WAR, W3, W4, W6 and W8 is rather minor, from 4.87
226
to 4.93, generally within the range of method errors. As far as pKa2 is concerned, its
227
fluctuation turns out to be substantially greater. In enables to classify the compounds
228
according to their pKa2 value into three groups. W3 and W4 belong to the first group, pKa2
229
amounts to 11.23 and 11.22, respectively. W6 of the second group exhibits pKa2 10.69,
230
whereas W7 and W8 of the third group show values 9.39 and 9.50, respectively. The
231
uncertainty of pKa2 determination is notably greater than pKa1, nevertheless, qualitative
232
differences between pKa2 of hydroxywarfarins are unambiguously clear.
233
One shall admit that an effective temperature inside capillary could be higher than 25 ̊C due
234
to inefficient cooling, especially in the inlet part. Therefore, the results may be burdened by
235
some intrinsic error related to changes in ionization caused by temperature increase.
Ac ce p
te
d
M
an
us
cr
ip t
207
10 Page 10 of 36
Our results reported herein provide the first experimental and complete picture of the acid-
237
base properties of WAR and all hydroxywarfarins existing in vivo. One must admit that pKa1
238
of WAR was determined oftentimes in the past (4.85 – 5.15) [5-9], while pKa1 of W7 and
239
W10 was estimated by us in our previous work (5.19 and 6.06, respectively) [9]. The
240
experiments described in our previous work however, were conducted in other conditions
241
of temperature and ionic strength, and were devoted mainly to investigation of the
242
supramolecular pKa shifts upon complexation with cyclodextrins. In general, the current
243
results are consistent with the previous reports [5-9]. Further comments and conclusions
244
related to the determined pKa values can be found in Section 3.4.
245
3.2. Application of CE-DAD method
246
In some cases the CE-DAD instrument gives a possibility to estimate pKa by totally different
247
method than aforementioned one [21,22]. In general, it is the case when the acidic and basic
248
forms of analyte exert different UV-Vis spectra. Then, the advantages of two techniques, CE
249
and spectrophotometric titration, can be combined within one approach. It is worth
250
highlighting that application of voltage is in fact not mandatory, however, it enables
251
separation of sample components from each other and allow to decrease analysis time. As it
252
has been shown in the literature, the only variable parameters used for pKa calculation are
253
absorbance at the given wavelength and analyte concentration. Additionally, a constant
254
value of a molar absorptivity coefficient is required, determined separately for the acidic
255
and basic forms.
256
Our method presented herein for the first time, however, require neither any values of
257
analyte concentration nor absorptivity coefficient. It is instead based on the measurements
258
of absorbance at two distinct wavelengths. Accordingly, the absorbance of WAR or any
259
other compound is measured at the maximum of electrophoretic peaks at 280 and 308 nm,
260
respectively. The outcomes have been shown in Fig.S-1 (see Supplementary Material) and
261
Table 2.
262
Table 2
263
Table 2 demonstrates that the results are the same for β2 and β3, but differ for β1.
264
Remarkably, the excellent accordance is noted between β2, β3, and the reference CE method.
Ac ce p
te
d
M
an
us
cr
ip t
236
11 Page 11 of 36
The less consistent results observed for β1 stem probably from its mathematical definition,
266
since the value of β1 changes in a far broader range than β2 and β3,. Such good outcome
267
proves that the CE-DAD method is fully prospective and gives response consistent with the
268
typical CE approach. The model is based on the stepwise changes in the shape of
269
absorbance spectra, instead of ionization entailing the shifts in migration times in CE. It is
270
worth highlighting that CE-DAD method has its own advantage, since EOF marker is not
271
necessary. Its limitation, in turn, is that the suitable accuracy of measurements can be easily
272
obtained only for relatively high concentration of analyte. Notwithstanding, CE and CE-DAD
273
methods can be applied interchangeably, however in this case, only for pKa1 determination.
274
3.3. Application of the IS-CE method
275
The method based on the use of internal standard of known pKa value was presented as a
276
fully prospective alternative for other methods used in high throughput pKa determination,
277
especially, in drug discovery studies [23-27]. At the moment, however, its true potential is
278
still unknown. In particular, consistency between IS-CE and CE methods applied jointly for
279
the same analytes seems to be an interesting issue to be investigated.
280
To this end, IS-CE has been tested in the numerous variants. Firstly, all the analytes have
281
been used individually as the internal standards for calculation of pKa of the others. As the
282
pKa of internal standards, the values obtained from CE method have been used. In this way,
283
a given pKa1 value has been estimated from six various reference pKa1 values. Secondly, the
284
calculations have been done for three distinct μeff values corresponding to partial ionization,
285
attributed to three distinct pH values. As the value of μeff related to the fully deprotonated
286
species, the value indicated by a bottom horizontal asymptote during function fitting has
287
been applied (see Fig.S-1). In consequence, particular pKa could be determined by IS-CE
288
many times, using different approaches. In addition, pKa2 has been determined for all
289
compounds except WAR and W10. All pKa values have been gathered in Table 3.
290
Table 3
291
In general, as far as pKa1 values are concerned, satisfactory accordance is noticed between
292
the IS-CE and CE methods. In majority, discrepancy is less than 0.1 pH units, and this fact
293
highlights the potential of IS-CE method. The application of W10 as the internal standard
Ac ce p
te
d
M
an
us
cr
ip t
265
12 Page 12 of 36
gives the worst predictions, and also, the values obtained for W10 as the analyte are mostly
295
poorer than the values calculated for other compounds. It seems to be logical that this
296
situation results from the particularly high pKa1 value of W10. Nevertheless, the calculations
297
performed for μeff obtained at pH 5.25 are relatively consistent with CE. It prompts us that a
298
choice of pH from the range between pKa of internal standard and analyte, is actually, the
299
most profitable. It can be also noted that the results obtained for WAR at pH around 6.0 are
300
worse than for hydroxywarfarins. In this case, putatively, possession of the second hydroxyl
301
group might be the main factor affecting accuracy of predictions.
302
In the case of pKa2 the discrepancy between the CE and IS-CE methods is significantly
303
higher, mostly over 0.2 pH units. As it can be easily concluded, the satisfactory results are
304
noted only if pKa2 of analyte is close to pKa2 of internal standard, like for W3 and W4, or W7
305
and W8. Nevertheless, the right qualitative trend between particular compounds is
306
indicated in each column (see Table 3). Owing to this qualitative accordance one can
307
speculate that only two analytical runs and the obtained very approximate pKa2 values
308
should suffice for proper identification of metabolites based on their migration order in CE
309
(compare with Section 3.6).
310
An overall evaluation of the IS-CE method is not straightforward. On the one hand, an
311
accurate pKa value is rather difficult to be obtained when properties of an internal standard
312
and an analyte differ from each other. On the other hand, for compounds of more similar
313
nature the outcomes are appreciably amended. Secondly, a rough pKa value or a direction of
314
its shift with regard to a reference molecule or state, can be obtained very fast. Even if IS-CE
315
is not a perfect tool, its potential of time saving merits attention during development of new
316
high throughput methods for pKa determination.
317
3.4. Thermodynamic pKa
318
Determination of thermodynamic pKa, contrary to apparent values discussed above, enables
319
an easy comparison between different experimental conditions. pKaT refers to the zero ionic
320
strength, and thus, it is defined by the activities of respective species, in this state equal to
321
their concentrations.
Ac ce p
te
d
M
an
us
cr
ip t
294
13 Page 13 of 36
In order to investigate the usefulness of the extended Debye-Hückel model for prediction of
323
pKaT values, the variation of apparent pKa1 with changing ionic strength has been studied
324
experimentally. For that purpose, the CE procedure has been repeated using diluted BGEs,
325
to 25 and 10 mM ionic strength, respectively. It allowed us to compare possible differences
326
between particular ionic strengths obtained experimentally with the values calculated from
327
Debye-Hückel model (see Section 2.3.4). The results have been collected in Table 4.
328
Table 4
329
They suggest that the current version of Debye-Hückel model suitably predicts the variation
330
of pKa1 with respect to changes in ionic strength. The differences between the theoretical
331
model and the values verified experimentally are in the order of only 0.01 pH units.
332
Therefore, the Debye-Hückel correction has been applied for calculation of pKa1T, and also
333
pKa2T. It is worth mentioning that the scale of this correction is greater for pKa2T than for
334
pKa1T, what can be derived from Eq.9 and Eq.10, due to the fact that second dissociation
335
occurs for the singly ionized species.
336
3.5. Theoretical and pharmacological considerations
337
We have further performed preliminary DFT/TZP/BLYP-D3 study using ADF program [29]
338
in order to characterize the structures of all compounds (WAR, W3, W4, W6, W7, W8 and
339
WAR-10) in gas phase. They are presented in Fig.2. It is clearly visible that all compounds
340
contain intramolecular hydrogen bonding of the type O…HO. More importantly, it has been
341
found that the bond lengths within the O…HO bridge are very similar (~1.0 Å for O–H and
342
~1.78 Å for H…O) in all compounds except W10, for which the shortest H…O distance (1.61
343
Å) has been found, see Fig.2. It shows that the O–H proton of W10 is engaged in the
344
strongest intramolecular interaction. It correlates with the trend in the pKa1 values, i.e.
345
markedly larger value of pKa1 is noted for W10 with respect to the remaining regio-isomers.
346
Accordingly, one could suggest that the characteristics of O…HO bonding could directly
347
determine the trend in pKa1 values.
348
Figure 2
Ac ce p
te
d
M
an
us
cr
ip t
322
14 Page 14 of 36
As far as the second pKa2 values are concerned, one shall note that changes in stability when
350
going from monoanion to dianion are of vital importance. We see that W3 and W4 exhibit
351
larger pKa2 values as compared to W6, W7 and W8. It is noticeable that in the former case
352
the hydroxyl is introduced at the single phenyl ring, whereas in the latter one various O–H
353
positions at coumarin moiety are considered. It could be the case that delocalization of
354
negative charge, leading to resonance and induction effects, is stronger for the two aromatic
355
coumarin rings than for the single phenyl ring. Coumarin moiety, in addition, contains the
356
electron withdrawing units, carbonyl and heteroatom. An important role of carbonyl unit in
357
enhancement of anion stability is known, it is reflected e.g. by dissociation of carboxyl acids.
358
Apart from these speculations, the increased electrostatic attraction of the proton caused by
359
the negative charge localized on the oxygen of dissociated OH group should be considered
360
for each hydroxywarfarin. Such effect can account for the higher pKa2 noted for W6, than for
361
W7 and W8.
362
We plan in the future to calculate free energy changes for all compounds, as well as the
363
related theoretical pKa values. It will be performed both in the gas phase and in the water.
364
Furthermore, the electronic structures that could help us to understand factors that
365
determine pKa values on a microscopic level will be also calculated.
366
Another important issue is an equilibrium between distinct tautomers of WAR and its
367
derivatives which can coexists in solution: the open form and the two diastereomeric
368
hemiketals, depicted in Fig.3. It is remarkable that different compounds may exhibit
369
different partition between particular tautomers, depending also on the solvent
370
composition [30]. Furthermore, amount and spatial distribution of the potential hydrogen
371
donors and acceptors varies between the open and the hemiketal forms. Therefore, acid-
372
base equilibrium defined by proximity of given molecule regions can be associated with
373
equilibrium between the tautomers. This aspect cannot be omitted during elucidation of
374
structure-related properties of WAR and its metabolites.
375
Figure 3
376
A pharmacological relevance of our results is in fact noticeable. The values of pKa1T and
377
pKa2T should be discussed in two different areas of interest. The first one is elucidation of
Ac ce p
te
d
M
an
us
cr
ip t
349
15 Page 15 of 36
WAR pharmacokinetics. It is commonly known that a role of the phase I oxidative
379
metabolism of drugs mediated by CYP enzymes is to increase a drug hydrophilicity and
380
solubility. This increase may stem from the two reasons: appearance of novel ionizable
381
groups yielding charge upon dissociation, or formation of groups which do not dissociate
382
but change an overall character of given molecule sites. One of the decisive factors for the
383
prevalence of one effect over other one is the relation between pKa of this group and pH of
384
environment. Formation of the second OH group during phase I metabolism of WAR, as can
385
be concluded from pKa2T values, does not trigger an increase in overall charge of molecule at
386
physiological pH. Yet more interesting conclusions can be drawn when referring to the
387
pharmacological relevance of pKa1T. Regarding the range of pH of natural intracellular
388
environment or body fluids like blood, urine or bile – generally from 5.5 to 8.0, one can
389
conclude that both parent compound and every hydroxylated metabolite occur rather in the
390
ionized form. The intriguing exception can be however W10 with pKa1≈ 6. In this case the
391
pKa1value may suffice for existence of significant fraction of the non-ionized form, far more
392
hydrophobic, especially in slightly acidic matrices like e.g. urine (pH ≈ 5.5). The conclusion
393
could be that phase I metabolism leading to W10 may cause in fact the adverse effects on
394
solubility in some matrices. This is of special relevance because (R)-W10 is one of the main
395
metabolites found in biological material collected from patients.
396
The second pharmacologically-relevant area of interest is, still enigmatic, the phase II
397
conjugative metabolism of WAR. It is quite likely that the reactions of glucuronidation
398
catalyzed by UDP-glucuronosyltransferases depend on hydroxylation site, thus, exert high
399
regioselectivity [31-33]. This may be in consequence one of the reasons why a diverse
400
profile of WAR phase II metabolites found in vivo is correlated with the expression of given
401
CYP enzymes. The last issue is a hypothetic modulation of WAR metabolism directly by
402
hydroxywarfarins. The two different phenomena should be considered, either a
403
transcriptional modulation of genes expression or a direct inhibition of CYP enzymes e.g. by
404
blocking of the active sites of enzymes. One may predict that these phenomena are
405
somehow related to structure and conformation of the particular metabolites and
406
respective stabilities of O–H bonds. In consequence, their pKa values would be of greater
407
biochemical relevance.
Ac ce p
te
d
M
an
us
cr
ip t
378
16 Page 16 of 36
3.6. pKa-related migration profile
409
Apart from the theoretical structure-activity relationships and the pharmacological
410
importance of the data obtained in this work, another interesting issue is a separation of all
411
compounds from each other mixed in a single sample. To this end, the CE method based on
412
differences in already known pKa values seems to be the simplest opportunity. To attempt
413
this approach, CE-based separation has been conducted at pH 10.4. This pH has been
414
selected as optimal for obtaining different ionization levels of the particular compounds,
415
thus also high selectivity. In Fig.4 the migration profile has been depicted.
416
Figure 4
417
It has occurred that the individual migration times properly reflect the order of pKa2 values.
418
In particular, almost total but still slightly different ionization can be predicted from the
419
position of W8 and W7 peaks, the ionization level below 50% from W6 peak, while yet
420
lower but still visible ionization from W3 and W4 peaks. The lack of separation between W3
421
and W4 proves that their pKa2 is very similar. Interestingly, W10 has been separated from
422
WAR, and this is probably caused not by different ionization, but rather by a different
423
hydrodynamic resistance during electromigration. Hypothetically, W10 as capable of
424
forming stronger intramolecular hydrogen bonds may migrate in a more occlusive way, and
425
thus, actually a more mobile form than WAR. This hypothesis is supported by different
426
shape of the W10 molecule presented in Fig.2. Further investigation of these analytically
427
important conundrums will yet be the aim of our future work.
428
4. Conclusions
429
The classical CE method has allowed us to determine thermodynamic pKa values with the
430
aid of extended Debye-Hückel model, upon its experimental verification. The CE-DAD
431
method has occurred to be very useful for determination of pKa1, and provides totally
432
different physicochemical background than CE. Its additional advantage is that information
433
about concentration of analyte and absorptivity coefficient values is not needed. Both
434
methods are complementary and can be applied interchangeably. Accuracy of the IS-CE is
435
strictly dependent on degree of similarity between an internal standard and an analyte, and
436
this method appears to be especially useful for qualitative predictions. Due to vast
Ac ce p
te
d
M
an
us
cr
ip t
408
17 Page 17 of 36
economization of time and materials, however, its potential is still worth to be appreciated
438
and further tested.
439
One may conclude that the values of pKa1T and pKa2T obtained in the current work are
440
important from the analytical, pharmacological and theoretical points of view. The acid-
441
base properties depend on hydroxylation site. In particular, relatively large variations are
442
observed between pKa2T values. W10 exhibits specific properties, namely it undergoes only
443
one dissociation contrary to other hydroxywarfarins, and its pKa1T surpasses the values
444
noted for any other compounds. These features are related to the specific molecular
445
structure and alcoholic character of the second OH group. It turns out that phase I
446
metabolism leads to creation of more hydrophilic molecules, however still only singly
447
ionized at physiological pH, similarly as the parent drug. The specific conformations and
448
hydrogen binding energies, as well as the actual proximity of the given hydrogen donors
449
and acceptors, may be very important factors for explanation of regioselective phase II
450
metabolism and modulation of CYP-mediated metabolism, not studied in depth so far.
451
Selection of proper pH enables the efficient separation by CE of almost all tested
452
compounds mixed together, based on their pKa2 values.
453
The authors declare no competing financial interest.
454
Acknowledgments
455
Author Paweł Nowak has received the financial support from Krakowskie Konsorcjum
456
“Materia-Energia-Przyszłość” within the subsidy KNOW.
457
The research was carried out with equipment purchased with financial support from the
458
European Regional Development Fund within the framework of the Polish Innovation
459
Economy Operational Programme (contract no. POIG.0 2.01.00-12-0 23/08).
Ac ce p
te
d
M
an
us
cr
ip t
437
460
18 Page 18 of 36
References
461
[1] L.S. Kaminsky, Z-Y. Zhang, Human P450 Metabolism of Warfarin, Pharmacol. Ther. 73
462
(1997) 67–74
463
[2] M. Gebauer, Synthesis and structure-activity relationships of novel warfarin derivatives,
464
Bioorg. Med. Chem. 15 (2007) 2414–2420.
465
[3] S. Babić, A.J.M. Horvat, D. Mutavdžić-Pavlović, M. Kaštelan-Macan, Determination of pKa
466
values of active pharmaceutical ingredients, Trac-Trend. Anal. Chem. 26, (2007) 1043–
467
1061.
468
[4] I. Ghosh, W.M. Nau, The strategic use of supramolecular pKa shifts to enhance the
469
bioavailability of drugs, Adv. Drug Deliver. Rev. 64 (2012) 764–783.
470
[5] H. Wan, A. Holmen, M. Någård, W. Lindberg, Rapid screening of pKa values of
471
pharmaceuticals by pressure-aassisted capillary electrophoresis combined with short-end
472
injection, J. Chromatogr. A 979 (2002) 369–377.
473
[6] J.M. Cabot, E. Fuguet, C. Ràfols, M. Rosés, Fast high-throughput method for the
474
determination of acidity constants by capillary electrophoresis. II. Acidic internal standards,
475
J. Chromatogr. A 1217 (2010) 8340–8345.
476
[7] M. Shalaeva, J. Kenseth, F. Lombardo, A. Bastin, Measurement of Dissociation Constants
477
(pKa Values) of Organic Compounds by Multiplexed Capillary Electrophoresis Using
478
Aqueous and Cosolvent Buffers, J. Pharm. Sci. 97 (2008) 2581–2606.
479
[8] Y. Ishihama, M. Nakamura, T. Miwa, T. Kajima, N. Asakawa, A Rapid Method for pKa
480
Determination of Drugs Using Pressure-Assisted Capillary Electrophoresis with Photodiode
481
Array Detection in Drug Discovery, J. Pharm. Sci. 91 (2002) 933–942.
482
[9] P. Nowak, M. Garnysz, M.P. Mitoraj, F. Sagan, M. Woźniakiewicz, P. Kościelniak,
483
Analytical aspects of achiral and cyclodextrin-mediated capillary electrophoresis of
484
warfarin and its two main derivatives assisted by theoretical modelling, J. Chromatogr. A
485
1377 (2015) 106–113.
Ac ce p
te
d
M
an
us
cr
ip t
460
19 Page 19 of 36
[10] J. Reijenga, H. van Arno, L. van Antonie, B. Teunissen, Development of Methods for the
487
Determination of pKa Values. Anal. Chem. Insights 8 (2013) 53−71.
488
[11] P. Nowak, M. Woźniakiewicz, P. Kościelniak, Application of capillary electrophoresis in
489
determination of acid dissociation constant (pKa) values, J. Chromatogr. A 1377 (2015) 1–
490
12.
491
[12] S.K. Poole, S. Patel, K. Dehring, H. Workman, C.F. Poole, Determination of acid
492
dissociation constants by capillary electrophoresis, J. Chromatogr. A 1037 (2004) 445–454.
493
[13] M. Riesová, J. Svobodová, K. Ušelová, Z. Tošner, I. Zusková, B. Gaš, Determination of
494
thermodynamic values of acidic dissociation constants and complexation constants of
495
profens and their utilization for optimization of separation conditions by Simul 5 Complex,
496
J. Chromatogr. A 1364 (2014) 276–288.
497
[14] E.C. Demiralay, Z. Ustun, Y.D. Daldal, Estimation of thermodynamic acidity constants of
498
some penicillinase-resistant penicillins, J. Pharmaceut. Biomed. 91 (2014) 7–11.
499
[15] J.L. Wang, X.J. Xu, D.Y. Chen, Determination of pK(a) values of a triptolide derivative
500
and its impurities by pressure-assisted capillary electrophoresis, J. Pharmaceut. Biomed. 88
501
(2014) 22–26.
502
[16] V. Solínová, V. Kašička, Determination of acidity constants and ionic mobilities of
503
polyprotic peptide hormones by CE, Electrophoresis 34 (2013) 2655–2665.
504
[17] H.X. Ren, L.C. Wang, X.S. Wang, X. Liu, S.X. Jiang, Measurement of acid dissociation
505
constants and ionic mobilities of 3-nitro-tyrosine and 3-chloro-tyrosine by capillary zone
506
electrophoresis, J. Pharmaceut. Biomed. 77 (2013) 83–87.
507
[18] K. Vceláková, I. Zusková, E. Kenndler, B. Gaš. Determination of cationic mobilities and
508
pK(a) values of 22 amino acids by capillary zone electrophoresis, Electrophoresis 25 (2004)
509
309–317.
510
[19] V. Solínová, V. Kašička, D. Koval, M. Cesnek, A. Holý. Determination of acid-base
511
dissociation constants of amino- and guanidinopurine nucleotide analogs and related
512
compounds by capillary zone electrophoresis, Electrophoresis 27 (2006) 1006–1019.
Ac ce p
te
d
M
an
us
cr
ip t
486
20 Page 20 of 36
[20] Z. Glatz, Application of short-end injection procedure in CE, Electrophoresis 34 (2013)
514
631–642.
515
[21] X. Wu, S. Gong, T. Bo, Y. Liao, H. Liu, Determination of dissociation constants of
516
pharmacologically active xanthones by capillary zone electrophoresis with diode array
517
detection, J. Chromatogr. A 1061 (2004) 217–223.
518
[22] S.P. Ozkorucuklu, J.L. Beltrán, G. Fonrodona, D. Barrón, G. Alsancak, J. Barbosa,
519
Determination of Dissociation Constants of Some Hydroxylated Benzoic and Cinnamic Acids
520
in Water from Mobility and Spectroscopic Data Obtained by CE-DAD, J. Chem. Eng. Data 54
521
(2009) 807–811.
522
[23] E. Fuguet, C. Ràfols, M. Rosés, A fast high throughput method for the determination of
523
acidity constants by capillary electrophoresis. 3. Basic internal standards, J.
524
Chromatography A 1218 (2011) 3928–3934.
525
[24] J.M. Cabot, E. Fuguet, C. Ràfols, M. Rosés, Determination of acidity constants by the
526
capillary electrophoresis internal standard method. IV. Polyprotic compounds, J.
527
Chromatogr. A 1279 (2013) 108–116.
528
[25] R.S. Gravador, E. Fuguet, C. Ràfols, M. Rosés, Temperature variation effects on the
529
determination of acidity constants through the internal standard–capillary electrophoresis
530
method, Electrophoresis 34 (2013) 1203–1211.
531
[26] J.M. Cabot, E. Fuguet, M. Rosés, Internal Standard Capillary Electrophoresis as a High
532
Throughput Method for pKa Determination in Drug Discovery and Development, ACS Comb.
533
16 (2014) 518–525.
534
[27] J.M. Cabot, E. Fuguet, M. Rosés, Determination of acidity constants of sparingly soluble
535
drugs in aqueous solution by the IS-CE method, Electrophoresis 35 (2014) 3564–3569.
536
[28] P. Debye, P. Huckel, On the theory of electrolytes, Phyz. Z. 24 (1923) 305–325.
537
[29] G. teVelde, F.M. Bickelhaupt, E.J. Baerends, C. Fonseca Guerra, S.J.A. van Gisbergen, J.G.
538
Snijders, T. Ziegler, Chemistry with ADF, J. Comput. Chem. 22 (2001) 931–967.
Ac ce p
te
d
M
an
us
cr
ip t
513
21 Page 21 of 36
[30] B. Castleberry, E.J. Valente, D.S. Eggleston, Open-cyclic warfarin isomerism: 5-
540
hydroxywarfarin, J. Cryst. Spectrosc. 20 (1990) 583–593.
541
[31] C.P. Pugh, D.L. Pouncey, J.H. Hartman, R. Nshimiyimana, L.P. Desrochers, T.E. Goodwin,
542
G. Boysen, G.P. Miller, Multiple UDP-glucuronosyltransferases in human liver microsomes
543
glucuronidate both R- and S-7-hydroxywarfarin into two metabolites, Arch. Biochem.
544
Biophys. 564 (2014) 244–253.
545
[32] D.R. Jones, G.P. Miller, Assays and applications in warfarin metabolism: What we know,
546
how we know it and what we need to know, Expert Opin. Drug Met. 7 (2011) 857–874.
547
[33] D.R. Jones, J.H. Moran, G.P. Miller, Warfarin and UDP-glucuronosyltransferases: Writing
548
a new chapter of metabolism, Drug Met. Rev. 42 (2010) 53–59.
an
us
cr
ip t
539
549
Ac ce p
te
d
M
550
22 Page 22 of 36
550 Fig.1. Absorbance spectra obtained during CE-DAD separation of WAR at pH 2.5 and 8.0,
552
respectively. Under these conditions only one form: the acidic (protonated) or the basic
553
(deprotonated) exists in solution. Vertical dashed lines indicate the wavelengths (280 and
554
308 nm) for which the absorbance values were recorded and further processed.
555
Fig.2. DFT/TZP/BLYP-D3 optimized structures of WAR and hydroxywarfarins together with
556
the selected bond distances (in Å).
557
Fig.3. Tautomeric forms of WAR, potential hydrogen donors and acceptors have been
558
pointed by red and blue colors, respectively.
559
Fig.4. Electropherogram presenting the separation of all analytes mixed together,
560
conducted at pH 10.4 in 25 mM ionic strength (in the
an
us
cr
ip t
551
Ac ce p
te
d
M
561
23 Page 23 of 36
561
Table 1. Composition and predicted pH of all buffering solutions prepared for experiments,
562
calculated for 50 mL total volume and 100 mM ionic strength.
564 565
cr
ip t
NaH2PO4 (100 mM) 4.80 CH3COONa (500 mM) 10.00 10.00 Na2HPO4 (100 mM) 4.70 13.29 16.25 NaOH (1 M) 1.02 2.13 2.50 KCl (1 M) 4.36 0.45
an
M
d
Ac ce p
563
H3PO4 (100 mM) 18.52 CH3COOH (500 mM) 14.16 2.52 NaH2PO4 (100 mM) 3.59 1.01 0.12 Na2B4O7·10H2O (50 mM) 39.77 28.67 25.05 NaOH (1 M) 0.64 4.55
te
Phosphate buffer I 2.50 Acetic buffer 4.50 5.25 Phosphate buffer II 6.00 7.00 8.00 Borate buffer 9.40 10.00 11.00 NaOH / KCl 12.00 13.00
Buffer composition [mL]
us
pH
24 Page 24 of 36
565
Table 2. Comparison of CE and CE-DAD methods used for pKa1 determination
CE-DAD
CE-DAD
CE-DAD
4.87 ± 0.02
4.71 ± 0.03
4.86 ± 0.03
4.86 ± 0.03
W3
4.87 ± 0.04
4.77 ± 0.02
4.89 ± 0.02
4.88 ± 0.02
W4
4.93 ± 0.04
4.77 ± 0.03
4.91 ± 0.03
4.91 ± 0.03
W6
4.92 ± 0.07
4.62 ± 0.02
4.92 ± 0.02
4.92 ± 0.02
W7
5.09 ± 0.07
5.08 ± 0.04
5.10 ± 0.03
5.10 ± 0.04
W8
4.88 ± 0.06
4.64 ± 0.03
4.84 ± 0.03
4.84 ± 0.03
W10
5.80 ± 0.03
5.68 ± 0.03
5.79 ± 0.02
5.79 ± 0.02
an
566
us
WAR
cr
CE (ref)
ip t
pKa1
M
567
Ac ce p
te
d
568
25 Page 25 of 36
Table 3. Apparent pKa values (in 100 mM ionic strength) obtained by the IS-CE method for
569
all possible combinations between potential internal standards and analytes, based on the
570
selected data obtained by the CE method. Internal standard pH of partial
WAR
W3
W4
W6
W7
4.88
4.91
4.90
4.94
4.85
4.84
4.85
4.71
4.71
4.67
ionization 4.87
W6
W7
W8
W10
4.92
4.67
4.66
4.90
4.89
4.93
4.88
5.11
4.87
4.90
4.88
4.95
4.83
4.82
5.06
4.93
4.97
4.92
5.14
4.94
4.94
5.01
4.89
4.89
4.88
5.12
4.96
4.91
5.14
4.94
4.92
4.99
4.93
4.92
5.15
5.04
5.27
5.07
5.14
5.08
5.32
4.86
pH ≈ 5.25
4.89
pH ≈ 6.00
5.03
4.83
pH ≈ 4.50
4.89
4.91
pH ≈ 5.25
4.96
4.93
4.93
4.96
pH ≈ 6.00
5.09
4.93
pH ≈ 4.50
4.89
4.90
4.92
pH ≈ 5.25
4.94
4.92
4.91
pH ≈ 6.00
5.12
4.96
4.96
pH ≈ 4.50
5.02
5.03
5.05
5.05
pH ≈ 5.25
5.09
5.06
5.06
5.07
pH ≈ 6.00
5.29
5.13
5.13
5.08
pH ≈ 4.50
4.86
4.87
4.89
4.89
4.93
pH ≈ 5.25
4.90
4.87
4.87
4.88
4.90
pH ≈ 6.00
5.09
4.93
4.93
4.88
4.89
pH ≈ 4.50
5.55
5.56
5.59
5.58
5.62
5.57
pH ≈ 5.25
5.75
5.72
5.72
5.73
5.75
5.73
pH ≈ 6.00
5.77
5.61
5.61
5.57
5.57
5.56
4.87
4.87 4.87
te
W4
4.85
pH ≈ 4.50
Ac ce p
W3
4.89
4.87
M
pH ≈ 6.00
5.12
us
pH ≈ 5.25
d
WAR
W10
4.89
an
pH ≈ 4.50
W8
cr
Analyte
ip t
568
4.92
5.09
5.11 4.88
4.95 5.12 5.80
26 Page 26 of 36
pH ≈ 10.50
11.23
pH ≈ 11.00
W7
10.90 10.36 10.37
11.23
10.93
pH ≈ 9.90
11.22
10.82 10.23 10.22
pH ≈ 10.50
11.20 11.22 10.87 10.34 10.35
pH ≈ 11.00
11.22
pH ≈ 9.90
11.09
11.09
pH ≈ 10.50
11.02
11.04 10.69 10.15 10.16
pH ≈ 11.00
10.99
11.00
pH ≈ 9.90
10.38
10.38
pH ≈ 10.50
10.26
10.27
pH ≈ 9.90
10.49
10.50
10.10
9.51
pH ≈ 10.50
10.36
10.37
10.03
9.49
10.91
pH ≈ 11.00
9.98 9.93
9.39
9.38 9.40
9.50
d
W8
10.10 10.09
M
pH ≈ 11.00
cr
W6
11.25
us
W4
10.83 10.24 10.24
an
W3
11.23
ip t
pH ≈ 9.90
As pKa of the internal standard, the values calculated from the CE method (in 100 mM ionic
572
strength) have been used.
574
Ac ce p
573
te
571
27 Page 27 of 36
Table 4. Apparent (valid at given ionic strength) and thermodynamic (standardized to zero ionic strength) pKa
pKa2 100 mM
pKa2T
4.87 ± 0.02
4.89 ± 0.04
4.99 ± 0.03
4.99
4.87 ± 0.04
4.89 ± 0.04
4.93 ± 0.03
4.97
11.23 ± 0.06
11.55
W4
4.93 ± 0.04
4.98 ± 0.05
4.97 ± 0.04
5.03
11.22 ± 0.06
11.54
W6
4.92 ± 0.07
4.94 ± 0.06
4.97 ± 0.04
5.01
10.69 ± 0.08
11.01
W7
5.09 ± 0.07
5.09 ± 0.04
5.10 ± 0.04
5.16
W8
4.88 ± 0.06
4.88 ± 0.03
4.93 ± 0.04
4.97
W10
5.80 ± 0.03
5.93 ± 0.03
5.92 ± 0.05
W3
Δcalculated
0.03
Δpredicted
0.04 Δcalculated
581 582
9.71
9.50 ± 0.12
9.82
5.95
0.03
Δpredicted 0.02 has been calculated as an average from three values obtained for three distinct ionic strength by
Ac ce p
576 577 578 579 580
pKa1T
9.39 ± 0.16
d
WAR
cr
pKa1T
us
pKa1 10 mM
an
pKa1 25 mM
M
pKa1 100 mM
ip t
values.
te
574 575
applying respective Debye-Hückel corrections; Δcalculated denotes an average difference between apparent pKa for the same compound and obtained experimentally for two distinct ionic strengths; Δpredicted denotes a difference between apparent pKa for the same compound and predicted for two distinct ionic strengths from Debye-Hückel model.
long-end injection mode).
28 Page 28 of 36
Ac
ce
pt
ed
M
an
us
cr
i
*Graphical Abstract
Page 29 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1 bw print only
Page 30 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 1 color web only
Page 31 of 36
Ac ce p
te
d
M
an
us
cr
ip t
Figure 2 revised color- web only
Page 32 of 36
Ac ce p
te
d
M
an
us
cr
ip t
Figure 2 revised bw- print only
Page 33 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3 bw print only
Page 34 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3 color web only
Page 35 of 36
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4
Page 36 of 36