EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx 1
Contents lists available at ScienceDirect
European Journal of Pharmaceutics and Biopharmaceutics journal homepage: www.elsevier.com/locate/ejpb
2
Research paper
7 4 8
6
Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy
9
M. Vogtherr, A. Marx, A.-C. Mieden, C. Saal ⇑
5
10 12 11 13 1 3 5 0 16 17 18 19 20 21 22 23 24 25 26 27 28 29
Merck KGaA, Darmstadt, Germany
a r t i c l e
i n f o
Article history: Received 4 November 2014 Revised 27 January 2015 Accepted in revised form 14 February 2015 Available online xxxx Keywords: Biophysical models Diffusion Dissolution Gastrointestinal Micelles NMR Taurocholate–lecithin molecular species
a b s t r a c t The interaction between an ampholytic and amphiphilic Active Pharmaceutical Ingredient (API) showing unusual pH dependent solubility and Fasted State Simulated Intestinal Fluid (FaSSIF) was studied by NMR spectroscopy. Solubility in FaSSIF was drastically increased, about 30 fold, compared to simulated gastrointestinal fluid without bile salts. Our studies aimed at understanding the mechanisms that lead to this drastic enhancement. All species present in solution at various concentrations of API were characterised by Diffusion Ordered Spectroscopy (DOSY) NMR measurements. These indicated the presence of mixed taurocholate–lecithin and pure taurocholate micelles in pure FaSSIF, and formation of mixed taurocholate–API micelles after addition of API. The formation of taurocholate–API micelles was also supported by Nuclear Overhauser Enhancement (NOE) contacts between taurocholate and the API. Formation of mixed taurocholate–API micelles took place at the expense of pure taurocholate micelles, whereas mixed taurocholate–lecithin micelles remained uninfluenced by the presence of API. Our results showed that the increase in solubility was due to similar amphiphilic properties of the API and taurocholate which enabled formation of mixed taurocholate–API micelles. From results of determination of solubility as well as NMR experiments a phase diagram comprising several micellar species was derived. Ó 2015 Elsevier B.V. All rights reserved.
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
49 50
1. Introduction
51
The pharmacokinetic behaviour of an API is governed by the complex interplay of absorption, distribution, metabolism and excretion (ADME). For many oral drugs, bioavailability is mainly governed by absorption, which in turn depends upon solubility, dissolution and permeability. Based on these properties the biopharmaceutical classification scheme (BCS) was introduced [1] and refined with respect to solubility and dissolution rate [2]. Further approaches to assess drug absorption at an early stage of pharmaceutical research and development comprise pharmacokinetic modelling [3–6]. Regardless of the method used, knowledge of solubility and dissolution rate is crucial to get a first understanding of drug absorption. This is particularly important as pharmaceutical industry faced a pronounced shift in research compounds towards poorly soluble molecules [7,8].
52 53 54 55 56 57 58 59 60 61 62 63 64
Abbreviations: DOSY, Diffusion Ordered Spectroscopy; TC, taurocholate; L, lecithin; NOE, Nuclear Overhauser Effect; FaSSIF, Fasted State Simulated Intestinal Fluid. ⇑ Corresponding author. Merck KGaA, Frankfurter Strasse 250, 64293 Darmstadt, Germany. Tel.: +49 6151 727634; fax: +49 6151 723073. E-mail address: [email protected] (C. Saal).
Conditions to assess solubility and dissolution rate need to mimic physiological conditions as closely as possible. The best approach, using ex vivo dissolution media [9], is only feasible in a very limited number of special cases. Therefore synthetic biorelevant dissolution media are available which simulate conditions in stomach (Fasted State Simulated Gastric Fluid – FaSSGF, Fed State Simulated Gastric Fluid – FeSSGF) and intestine (Fasted State Intestinal Fluid – FaSSIF, Fed State Intestinal Fluid – FeSSIF) under fasted and fed state conditions [10]. Beyond simulating pH and ionic strength under fasted and fed state conditions, intestinal media contain lecithin and taurocholate as bile salts. These exhibit surfactant properties such as formation of micelles that can interact with API molecules to aid solubilisation. Although biorelevant media are well established in pharmaceutical research to investigate solubility and dissolution rate of drugs and to represent a starting point to predict or compare human drug absorption, details of the interaction between an API and bile salts on a molecular level remain largely unknown. This holds especially true for the molecular interactions which are responsible for formation of micelles. On the one hand, large datasets are available comparing solubility in buffer systems and biorelevant media from which the authors also deduced
http://dx.doi.org/10.1016/j.ejpb.2015.02.016 0939-6411/Ó 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 2 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx
structure–property relationships [11]. On the other hand micelles in biorelevant media have been characterised with regard to their geometry and dynamic behaviour [12]. To our knowledge there is no in-depth characterisation of micelle formation in biorelevant media on a molecular basis so far. In general spectroscopic methods represent valuable tools to get such an understanding. In this contribution we show that NMR methods can be used to get insight into intermolecular bile-salt–API interactions which govern solubility of a certain APIs in biorelevant media. For this purpose one poorly soluble API was chosen. The structure of the API – (methyl-{3-[5-(20 -methyl-2-trifluoromethyl-biphenyl-4-yl)-[1,2,4]oxadiazol-3-yl]-benzyl}-amino)-acetic acid – is shown in the left part of Scheme 1. As the molecule represents an S1P1 agonist its structure mimics sphingosine – as depicted in the right part of Scheme 1 – and consists of a large lipophilic part and a smaller zwitterionic amino-acid part. Both parts together make the molecule amphiphilic and accordingly amendable for solubility enhancement by bile salts. The tendency of the API for micelle formation is also recognised in the unusual pH dependent solubility profile of the API which is further affected by variable concentrations of chloride ions. Diffusion-ordered spectroscopy is an NMR method for measurement of self-diffusion constants [13]. The latter differ characteristically between isolated molecules and larger aggregates. In addition to self-diffusion constants, DOSY can give information regarding the composition of a supramolecular species from its NMR spectra, in particular if the constituting molecules are known. Other NMR experiments, in particular Nuclear Overhauser spectroscopy (NOESY), are standard methods to characterise intermolecular interactions. The combination of DOSY and NOESY was widely used to characterise the solubilisation of APIs by surfactants [14,15] including taurocholate as a typical component of bile liquid [16]. Nevertheless so far no investigations for biorelevant dissolution media such as FaSSIF and FeSSIF mimicking in vivo-behaviour of drugs are available in the literature. In this study we use DOSY and NOESY experiments to characterise the interaction between the API 1 and bile salts at molecular level. FaSSIF was chosen as medium as it contains a sufficient amount of bile salts to exhibit considerable solubility enhancement. At the same time the concentration of bile salts is lower compared to FeSSIF which makes the NMR experiments more sensitive to detect interactions between API and bile salts.
2. Materials and methods
129
We used compound 1 as model compound to investigate interaction between an API and bile salts. Compound 1 was discovered in a program for S1P1 agonists. Therefore it contains a polar head group represented by an amino acid and a lipophilic tail consisting of three substituted phenyl rings and an oxadiazol. Due to its amino acid functionality 1 has protolytic activity with two pKa values, 4.2 and 7.9. At neutral pH compound 1 is present as a zwitterion. Log D at pH 7.4 is 3.8, as determined by the shake-flask method. Compound 1 was used for the investigations described in this work as a crystalline hydrochloride salt.
130
2.1. Preparation of aqueous buffer systems
140
To obtain pH dependent solubility, aqueous buffer systems were prepared covering the whole pH range between pH 1 and pH 9. Buffer systems were prepared as described by USP [17]. In some cases additional buffers were used which were obtained by addition of chloride to USP buffers as described below. In addition to the USP and modified USP buffers, sulphuric acid was used as a system representing a chloride free buffer with low pH. Acidic conditions were simulated using 0.1 M HCl (Merck KGaA, Art. No. 109060) with a pH of 1. For comparison of this buffer with a high chloride concentration, a chloride free medium of a similar pH consisting of 0.1 M H2SO4 (20 ml 0.5 M H2SO4 (Merck KGaA, Art. No. 109072) diluted to 100 ml with demineralised water) with a final pH of 0.7 was used. The intermediate pH range was simulated using sodium acetate buffer (0.598 g NaCH3COO ⁄ 3H2O (Merck KGaA, Art. No. 106267) and 7.8 ml 1 N acetic acid (Merck KGaA, Art. No. 137035) in 200 ml demineralised water) with a pH of 4.5, and sodium phosphate buffer (0.689 g NaH2PO4 ⁄ H2O (Merck KGaA, Art. No. 106346) and 3.75 ml 1 N NaOH (Merck KGaA, Art. No. 137031) in 100 ml demineralised water) with a final pH of 7.4. The upper pH range was simulated using sodium borate buffer (0.6185 g H3BO3 (Merck KGaA, Art. No. 100165) and 18.8 ml 0.2 N NaOH (Merck KGaA, Art. No. 109140) in 200 ml demineralised water) with a final pH of 9.0. 400 mM NaCl (Merck KGaA, Art. No. 106404) was added to the acetate, phosphate and borate buffer system to investigate the effect of variable chloride concentrations on the solubility of 1 at
141
OH HO
NH2
OH
N O N
F F
N O
F
Scheme 1. Left hand side: Compound 1. Right hand side: Sphingosine.
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
131 132 133 134 135 136 137 138 139
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx 167 168
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
constant pH. An effect of chloride on solubility of 1 might be expected as consequence of a common-ion effect. 2.2. Preparation of biorelevant media The impact of micelles consisting of bile salts and lecithin on the solubility of 1 was analysed using FaSSIF. To investigate the effect of the bile salts and lecithin, solubility was also assessed using the same medium – a phosphate buffer containing sodium chloride – without these components. A phosphate buffer containing sodium chloride was used to prepare FaSSIF: 3.438 g NaH2PO4 (Merck KGaA, Art. No. 106370) and 6.186 g of NaCl (Merck KGaA, Art. No. 106404) were dissolved in 900 ml demineralised water. pH was adjusted to 6.5 by addition of 0.1 N NaOH solution (Merck KGaA, Art. No. 109141) and finally demineralised water was added to a volume to 1000 ml. FaSSIF containing 3 mM sodium taurocholate and 0.75 mM lecithin was prepared by dissolving SIF powder (Phares AG, Muttenz, Switzerland, Art. No. PHA S 1108) in the phosphatesodium chloride buffer as described above. 2.24 g of SIF powder was dissolved in 500 ml of the phosphate-sodium chloride buffer. After dispersion by stirring, further phosphate-sodium chloride buffer was added to yield 1 l total volume. The solution was left to equilibrate for 2 h. The solution had a slightly opalescent appearance.
3
2.5. NMR spectroscopy
225
NMR spectra were recorded on a Bruker Avance III 700 spectrometer equipped with a cryoprobe at 700 MHz proton resonance frequency. All experiments were performed at 300 K. Samples were equilibrated for 30 min prior to acquisition of DOSY spectra. Water suppression in 1D and NOESY spectra was accomplished by excitation sculpting [18]. NOESY spectra were recorded with 80 ms mixing time and excitation sculpting water suppression. Diffusion constants were obtained from a stimulated echo pulse sequence with bipolar gradient pulses and longitudinal echo delay [19], using diffusion time D = 100 ms and gradient length d = 4 ms. Water signal was suppressed by selective presaturation. Gradients were generated with a maximum gradient strength of 5.35 G/mm. Typically series of 16 spectra with gradient strength variation from 5% to 95% (0.26–5.0825 G/mm) were recorded. Spectra were Fourier transformed using topspin 2.1 (Bruker Biospin). Peak intensities were fitted using the program Dynamics Center 2.0.4 (Bruker Biospin) as a function of gradient strength to obtain diffusion constants. Eq. (1) was used for fitting.
226
2
2 2
IðGÞ ¼ I0 expðc G d ðD d=3Þ
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243
244
ð1Þ
246
where I(G) is the signal intensity at gradient strength G, I0 the signal intensity without gradient, c the gyromagnetic ratio of protons, G the gradient strength, D the diffusion time (100 ms), and d is the gradient length (4 ms).
247
2.6. Powder X-ray diffraction
251
Samples were prepared between amorphous films. Measurements were performed in transmission geometry with Cu Ka1 radiation on a Stoe StadiP 611 KL diffractometer. Intensities have been recorded with a PSD. An angular range of 1–65° 2H with a step width of 0.05° 2H was used.
252
2.7. Determination of pKa values
257
pKa values were determined by potentiometric titration on a GLpKa instrument (Sirius) using 0.15 M KCl solution. Titration was carried out using 0.5 M HCl and 0.5 M KOH in the pH range 1.8–11.0. Due to the low solubility of 1, methanol was used as a cosolvent at three concentration levels and extrapolation to pure aqueous conditions was carried out using Yesuda-Shedlovski approach. Potentiometric titration at each methanol concentration level was carried out in triplicate, and results were averaged.
258
3. Results and discussion
266
248 249 250
190
2.3. Measurement of solubility
191
209
Solubility of 1 in USP buffers with and without addition of NaCl, FaSSIF and the corresponding phosphate-sodium chloride buffer was measured by the shake flask method and HPLC with UV detection. The range of buffers used was designed to obtain the pH-dependent solubility profile in the presence and absence of 400 mM chloride. Accordingly, with this choice of buffers the effect of pH and chloride concentration could be assessed independently. To establish equilibrium conditions, ca. 4 mg 1 (introduced as hydrochloride salt) was incubated with the respective solvent and shaken with 450 rpm at 37 °C for 24 h in WhatmanÒ UniPrep Syringeless Filters (PTFE, 0.2 lm, Whatman, Art. No. UN513EORG). After shaking, the suspension was filtered using the same filters and diluted 1:1 with methanol avoiding precipitation of 1. The concentration of 1 was measured by HPLC (Agilent 1100 HPLC with ChromolithÒ performance RP-18e 100-3 mm column; detection wavelength 228 nm; 37 °C). A reference solution of 1 in acetonitrile (Merck KGaA, Art. No. 100016) was used for quantification of 1. Solubility measurements were carried out as triplicate measurements.
210
2.4. NMR sample preparation
3.1. API solubility in aqueous buffers
267
211
NMR samples were prepared either in FaSSIF medium or in the corresponding phosphate-sodium chloride buffer as described above. For NMR purposes each of these was lyophilised and re-dissolved in D2O to minimise the residual water signal. 0.1 mM trimethylsilyl propionic acid (Merck KGaA, Art. No. 108652) was added as NMR frequency standard for frequency calibration of selected spectra. Compound 1 as hydrochloride salt was weighed directly into a 5 mm NMR tube to yield final concentrations of 1.0, 2.0, 3.0, 4.0, 5.0 or 6.0 mM and dissolved in 500 ll of FaSSIF or the corresponding phosphate-sodium chloride buffer. To prepare saturated solutions of 1 in FaSSIF or the corresponding phosphate-sodium chloride buffer, excess amounts of 1 were suspended and the undissolved residue removed by centrifugation.
Compound 1 is characterised by two proteolytically active sites with two pKa values, pKA1 = 4.2 and pKA2 = 7.9 which were determined by potentiometric titration. Therefore its solubility is expected to be strongly dependent on pH, with an expected solubility profile for an ampholyte as depicted in the left side of Fig. 1. pH-dependent solubility of ampholytes is generally discussed in [20]. From theory, solubility of 1 is high below pKA1, where the protonated, cationic form of 1 is present in solution, and above pKA2, where the deprotonated, anionic form is present in solution. Poor solubility in the pH range between pKA1 and pKA2 where the zwitterionic form is prevalent in solution is expected. For measurement of pH-dependent solubility profiles aqueous buffers were chosen to obtain a pH profile between pH 1 and 9. Additionally solubility in the presence of chloride was measured
268
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
212 213 214 215 216 217 218 219 220 221 222 223 224
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
253 254 255 256
259 260 261 262 263 264 265
269 270 271 272 273 274 275 276 277 278 279 280 281
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 4
M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx
Fig. 1. Left: pH-dependent solubility profile expected for ampholytic behaviour of 1 as described in [20]. Right: experimentally observed pH dependent solubility of 1 (grey: chloride-free buffers, black: same buffers as grey curve with addition of 0.4 M NaCl).
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
to investigate the effect of chloride on solubility, as e.g. expected by a common ion effect for the hydrochloride salt. Experimental solubility of 1 is summarised in Table 1 and plotted in Fig. 1 at the right side as a function of pH. The highest solubility of 1 is found in the intermediate pH range, especially for buffers containing 0.4 mM NaCl. This pH-dependence of solubility is in strong contrast to the expectations for an ampholytic compound, for which solubility is lowest in the intermediate pH range where the uncharged, zwitterionic species is dominating in solution. By contrast, 1 has a solubility maximum in the intermediate pH range. This is particularly true for buffers containing 400 mM NaCl. Very low solubility independent of chloride concentration was observed under strongly acidic conditions (pH < pKA1 = 4.2), i.e. in 0.1 M HCl and 0.1 M H2SO4. Solubility in hydrochloric acid and sulphuric acid was similar. Conversion of the poorly soluble chloride salt of 1 into the equally insoluble sulphate salt occurred. Presence of a different solid state form was confirmed by powder X-ray diffraction of the precipitate. Fig. 2 shows a comparison of PXRD patterns as obtained from the parent form of 1, the sodium-salt of 1 as well as the chloride and sulphate salt of 1. In the intermediate pH range (pKA1 = 4.2 < pH < pKA2 = 7.9) a more than 1000 fold increased solubility was observed. This increase in solubility occurred in acetate buffer (pH 4.5) only in the presence of 400 mM NaCl. Irrespective of chloride concentration, this gain in solubility was observed in phosphate buffer at pH 7.4. Solubility of 1 in phosphate buffer with and without 400 mM NaCl was similar high. No effect of NaCl on solubility at this pH was observed. Finally at high pH (pH > pKA2 = 7.9), a decrease in solubility compared to the intermediate pH range was seen. This showed that the original hydrochloride of 1 was converted into a less soluble sodium salt as was also shown by powder X-ray diffraction (data not shown). Increased solubility in the presence of NaCl as measured by HPLC was always accompanied with opalescence and foam formation. This was indicative of micelle formation under these conditions. Apparently micelle formation was favoured by chloride ions. Micelle formation of the zwitterionic form of 1 explains the deviation between experimental pH dependence of solubility and the expectations.
Fig. 2. PXRD patterns obtained from precipitates obtained from solubility determinations. Top to bottom: Sulphate salt of 1, chloride salt of 1, sodium salt of 1, parent form of 1.
To get a closer understanding of the underlying mechanisms of the abnormal pH dependent solubility of 1, solutions were characterised by NMR spectroscopy.
322
3.2. API solubilisation by FaSSIF
325
The abnormal pH-dependence of solubility prompted the question how bile salts influence the solubility of 1. Solubility of 1 increased from
No. of Pages 10, Model 5G
26 February 2015 European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx 1
Contents lists available at ScienceDirect
European Journal of Pharmaceutics and Biopharmaceutics journal homepage: www.elsevier.com/locate/ejpb
2
Research paper
7 4 8
6
Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy
9
M. Vogtherr, A. Marx, A.-C. Mieden, C. Saal ⇑
5
10 12 11 13 1 3 5 0 16 17 18 19 20 21 22 23 24 25 26 27 28 29
Merck KGaA, Darmstadt, Germany
a r t i c l e
i n f o
Article history: Received 4 November 2014 Revised 27 January 2015 Accepted in revised form 14 February 2015 Available online xxxx Keywords: Biophysical models Diffusion Dissolution Gastrointestinal Micelles NMR Taurocholate–lecithin molecular species
a b s t r a c t The interaction between an ampholytic and amphiphilic Active Pharmaceutical Ingredient (API) showing unusual pH dependent solubility and Fasted State Simulated Intestinal Fluid (FaSSIF) was studied by NMR spectroscopy. Solubility in FaSSIF was drastically increased, about 30 fold, compared to simulated gastrointestinal fluid without bile salts. Our studies aimed at understanding the mechanisms that lead to this drastic enhancement. All species present in solution at various concentrations of API were characterised by Diffusion Ordered Spectroscopy (DOSY) NMR measurements. These indicated the presence of mixed taurocholate–lecithin and pure taurocholate micelles in pure FaSSIF, and formation of mixed taurocholate–API micelles after addition of API. The formation of taurocholate–API micelles was also supported by Nuclear Overhauser Enhancement (NOE) contacts between taurocholate and the API. Formation of mixed taurocholate–API micelles took place at the expense of pure taurocholate micelles, whereas mixed taurocholate–lecithin micelles remained uninfluenced by the presence of API. Our results showed that the increase in solubility was due to similar amphiphilic properties of the API and taurocholate which enabled formation of mixed taurocholate–API micelles. From results of determination of solubility as well as NMR experiments a phase diagram comprising several micellar species was derived. Ó 2015 Elsevier B.V. All rights reserved.
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
49 50
1. Introduction
51
The pharmacokinetic behaviour of an API is governed by the complex interplay of absorption, distribution, metabolism and excretion (ADME). For many oral drugs, bioavailability is mainly governed by absorption, which in turn depends upon solubility, dissolution and permeability. Based on these properties the biopharmaceutical classification scheme (BCS) was introduced [1] and refined with respect to solubility and dissolution rate [2]. Further approaches to assess drug absorption at an early stage of pharmaceutical research and development comprise pharmacokinetic modelling [3–6]. Regardless of the method used, knowledge of solubility and dissolution rate is crucial to get a first understanding of drug absorption. This is particularly important as pharmaceutical industry faced a pronounced shift in research compounds towards poorly soluble molecules [7,8].
52 53 54 55 56 57 58 59 60 61 62 63 64
Abbreviations: DOSY, Diffusion Ordered Spectroscopy; TC, taurocholate; L, lecithin; NOE, Nuclear Overhauser Effect; FaSSIF, Fasted State Simulated Intestinal Fluid. ⇑ Corresponding author. Merck KGaA, Frankfurter Strasse 250, 64293 Darmstadt, Germany. Tel.: +49 6151 727634; fax: +49 6151 723073. E-mail address: [email protected] (C. Saal).
Conditions to assess solubility and dissolution rate need to mimic physiological conditions as closely as possible. The best approach, using ex vivo dissolution media [9], is only feasible in a very limited number of special cases. Therefore synthetic biorelevant dissolution media are available which simulate conditions in stomach (Fasted State Simulated Gastric Fluid – FaSSGF, Fed State Simulated Gastric Fluid – FeSSGF) and intestine (Fasted State Intestinal Fluid – FaSSIF, Fed State Intestinal Fluid – FeSSIF) under fasted and fed state conditions [10]. Beyond simulating pH and ionic strength under fasted and fed state conditions, intestinal media contain lecithin and taurocholate as bile salts. These exhibit surfactant properties such as formation of micelles that can interact with API molecules to aid solubilisation. Although biorelevant media are well established in pharmaceutical research to investigate solubility and dissolution rate of drugs and to represent a starting point to predict or compare human drug absorption, details of the interaction between an API and bile salts on a molecular level remain largely unknown. This holds especially true for the molecular interactions which are responsible for formation of micelles. On the one hand, large datasets are available comparing solubility in buffer systems and biorelevant media from which the authors also deduced
http://dx.doi.org/10.1016/j.ejpb.2015.02.016 0939-6411/Ó 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 2 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx
structure–property relationships [11]. On the other hand micelles in biorelevant media have been characterised with regard to their geometry and dynamic behaviour [12]. To our knowledge there is no in-depth characterisation of micelle formation in biorelevant media on a molecular basis so far. In general spectroscopic methods represent valuable tools to get such an understanding. In this contribution we show that NMR methods can be used to get insight into intermolecular bile-salt–API interactions which govern solubility of a certain APIs in biorelevant media. For this purpose one poorly soluble API was chosen. The structure of the API – (methyl-{3-[5-(20 -methyl-2-trifluoromethyl-biphenyl-4-yl)-[1,2,4]oxadiazol-3-yl]-benzyl}-amino)-acetic acid – is shown in the left part of Scheme 1. As the molecule represents an S1P1 agonist its structure mimics sphingosine – as depicted in the right part of Scheme 1 – and consists of a large lipophilic part and a smaller zwitterionic amino-acid part. Both parts together make the molecule amphiphilic and accordingly amendable for solubility enhancement by bile salts. The tendency of the API for micelle formation is also recognised in the unusual pH dependent solubility profile of the API which is further affected by variable concentrations of chloride ions. Diffusion-ordered spectroscopy is an NMR method for measurement of self-diffusion constants [13]. The latter differ characteristically between isolated molecules and larger aggregates. In addition to self-diffusion constants, DOSY can give information regarding the composition of a supramolecular species from its NMR spectra, in particular if the constituting molecules are known. Other NMR experiments, in particular Nuclear Overhauser spectroscopy (NOESY), are standard methods to characterise intermolecular interactions. The combination of DOSY and NOESY was widely used to characterise the solubilisation of APIs by surfactants [14,15] including taurocholate as a typical component of bile liquid [16]. Nevertheless so far no investigations for biorelevant dissolution media such as FaSSIF and FeSSIF mimicking in vivo-behaviour of drugs are available in the literature. In this study we use DOSY and NOESY experiments to characterise the interaction between the API 1 and bile salts at molecular level. FaSSIF was chosen as medium as it contains a sufficient amount of bile salts to exhibit considerable solubility enhancement. At the same time the concentration of bile salts is lower compared to FeSSIF which makes the NMR experiments more sensitive to detect interactions between API and bile salts.
2. Materials and methods
129
We used compound 1 as model compound to investigate interaction between an API and bile salts. Compound 1 was discovered in a program for S1P1 agonists. Therefore it contains a polar head group represented by an amino acid and a lipophilic tail consisting of three substituted phenyl rings and an oxadiazol. Due to its amino acid functionality 1 has protolytic activity with two pKa values, 4.2 and 7.9. At neutral pH compound 1 is present as a zwitterion. Log D at pH 7.4 is 3.8, as determined by the shake-flask method. Compound 1 was used for the investigations described in this work as a crystalline hydrochloride salt.
130
2.1. Preparation of aqueous buffer systems
140
To obtain pH dependent solubility, aqueous buffer systems were prepared covering the whole pH range between pH 1 and pH 9. Buffer systems were prepared as described by USP [17]. In some cases additional buffers were used which were obtained by addition of chloride to USP buffers as described below. In addition to the USP and modified USP buffers, sulphuric acid was used as a system representing a chloride free buffer with low pH. Acidic conditions were simulated using 0.1 M HCl (Merck KGaA, Art. No. 109060) with a pH of 1. For comparison of this buffer with a high chloride concentration, a chloride free medium of a similar pH consisting of 0.1 M H2SO4 (20 ml 0.5 M H2SO4 (Merck KGaA, Art. No. 109072) diluted to 100 ml with demineralised water) with a final pH of 0.7 was used. The intermediate pH range was simulated using sodium acetate buffer (0.598 g NaCH3COO ⁄ 3H2O (Merck KGaA, Art. No. 106267) and 7.8 ml 1 N acetic acid (Merck KGaA, Art. No. 137035) in 200 ml demineralised water) with a pH of 4.5, and sodium phosphate buffer (0.689 g NaH2PO4 ⁄ H2O (Merck KGaA, Art. No. 106346) and 3.75 ml 1 N NaOH (Merck KGaA, Art. No. 137031) in 100 ml demineralised water) with a final pH of 7.4. The upper pH range was simulated using sodium borate buffer (0.6185 g H3BO3 (Merck KGaA, Art. No. 100165) and 18.8 ml 0.2 N NaOH (Merck KGaA, Art. No. 109140) in 200 ml demineralised water) with a final pH of 9.0. 400 mM NaCl (Merck KGaA, Art. No. 106404) was added to the acetate, phosphate and borate buffer system to investigate the effect of variable chloride concentrations on the solubility of 1 at
141
OH HO
NH2
OH
N O N
F F
N O
F
Scheme 1. Left hand side: Compound 1. Right hand side: Sphingosine.
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
131 132 133 134 135 136 137 138 139
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx 167 168
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
constant pH. An effect of chloride on solubility of 1 might be expected as consequence of a common-ion effect. 2.2. Preparation of biorelevant media The impact of micelles consisting of bile salts and lecithin on the solubility of 1 was analysed using FaSSIF. To investigate the effect of the bile salts and lecithin, solubility was also assessed using the same medium – a phosphate buffer containing sodium chloride – without these components. A phosphate buffer containing sodium chloride was used to prepare FaSSIF: 3.438 g NaH2PO4 (Merck KGaA, Art. No. 106370) and 6.186 g of NaCl (Merck KGaA, Art. No. 106404) were dissolved in 900 ml demineralised water. pH was adjusted to 6.5 by addition of 0.1 N NaOH solution (Merck KGaA, Art. No. 109141) and finally demineralised water was added to a volume to 1000 ml. FaSSIF containing 3 mM sodium taurocholate and 0.75 mM lecithin was prepared by dissolving SIF powder (Phares AG, Muttenz, Switzerland, Art. No. PHA S 1108) in the phosphatesodium chloride buffer as described above. 2.24 g of SIF powder was dissolved in 500 ml of the phosphate-sodium chloride buffer. After dispersion by stirring, further phosphate-sodium chloride buffer was added to yield 1 l total volume. The solution was left to equilibrate for 2 h. The solution had a slightly opalescent appearance.
3
2.5. NMR spectroscopy
225
NMR spectra were recorded on a Bruker Avance III 700 spectrometer equipped with a cryoprobe at 700 MHz proton resonance frequency. All experiments were performed at 300 K. Samples were equilibrated for 30 min prior to acquisition of DOSY spectra. Water suppression in 1D and NOESY spectra was accomplished by excitation sculpting [18]. NOESY spectra were recorded with 80 ms mixing time and excitation sculpting water suppression. Diffusion constants were obtained from a stimulated echo pulse sequence with bipolar gradient pulses and longitudinal echo delay [19], using diffusion time D = 100 ms and gradient length d = 4 ms. Water signal was suppressed by selective presaturation. Gradients were generated with a maximum gradient strength of 5.35 G/mm. Typically series of 16 spectra with gradient strength variation from 5% to 95% (0.26–5.0825 G/mm) were recorded. Spectra were Fourier transformed using topspin 2.1 (Bruker Biospin). Peak intensities were fitted using the program Dynamics Center 2.0.4 (Bruker Biospin) as a function of gradient strength to obtain diffusion constants. Eq. (1) was used for fitting.
226
2
2 2
IðGÞ ¼ I0 expðc G d ðD d=3Þ
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243
244
ð1Þ
246
where I(G) is the signal intensity at gradient strength G, I0 the signal intensity without gradient, c the gyromagnetic ratio of protons, G the gradient strength, D the diffusion time (100 ms), and d is the gradient length (4 ms).
247
2.6. Powder X-ray diffraction
251
Samples were prepared between amorphous films. Measurements were performed in transmission geometry with Cu Ka1 radiation on a Stoe StadiP 611 KL diffractometer. Intensities have been recorded with a PSD. An angular range of 1–65° 2H with a step width of 0.05° 2H was used.
252
2.7. Determination of pKa values
257
pKa values were determined by potentiometric titration on a GLpKa instrument (Sirius) using 0.15 M KCl solution. Titration was carried out using 0.5 M HCl and 0.5 M KOH in the pH range 1.8–11.0. Due to the low solubility of 1, methanol was used as a cosolvent at three concentration levels and extrapolation to pure aqueous conditions was carried out using Yesuda-Shedlovski approach. Potentiometric titration at each methanol concentration level was carried out in triplicate, and results were averaged.
258
3. Results and discussion
266
248 249 250
190
2.3. Measurement of solubility
191
209
Solubility of 1 in USP buffers with and without addition of NaCl, FaSSIF and the corresponding phosphate-sodium chloride buffer was measured by the shake flask method and HPLC with UV detection. The range of buffers used was designed to obtain the pH-dependent solubility profile in the presence and absence of 400 mM chloride. Accordingly, with this choice of buffers the effect of pH and chloride concentration could be assessed independently. To establish equilibrium conditions, ca. 4 mg 1 (introduced as hydrochloride salt) was incubated with the respective solvent and shaken with 450 rpm at 37 °C for 24 h in WhatmanÒ UniPrep Syringeless Filters (PTFE, 0.2 lm, Whatman, Art. No. UN513EORG). After shaking, the suspension was filtered using the same filters and diluted 1:1 with methanol avoiding precipitation of 1. The concentration of 1 was measured by HPLC (Agilent 1100 HPLC with ChromolithÒ performance RP-18e 100-3 mm column; detection wavelength 228 nm; 37 °C). A reference solution of 1 in acetonitrile (Merck KGaA, Art. No. 100016) was used for quantification of 1. Solubility measurements were carried out as triplicate measurements.
210
2.4. NMR sample preparation
3.1. API solubility in aqueous buffers
267
211
NMR samples were prepared either in FaSSIF medium or in the corresponding phosphate-sodium chloride buffer as described above. For NMR purposes each of these was lyophilised and re-dissolved in D2O to minimise the residual water signal. 0.1 mM trimethylsilyl propionic acid (Merck KGaA, Art. No. 108652) was added as NMR frequency standard for frequency calibration of selected spectra. Compound 1 as hydrochloride salt was weighed directly into a 5 mm NMR tube to yield final concentrations of 1.0, 2.0, 3.0, 4.0, 5.0 or 6.0 mM and dissolved in 500 ll of FaSSIF or the corresponding phosphate-sodium chloride buffer. To prepare saturated solutions of 1 in FaSSIF or the corresponding phosphate-sodium chloride buffer, excess amounts of 1 were suspended and the undissolved residue removed by centrifugation.
Compound 1 is characterised by two proteolytically active sites with two pKa values, pKA1 = 4.2 and pKA2 = 7.9 which were determined by potentiometric titration. Therefore its solubility is expected to be strongly dependent on pH, with an expected solubility profile for an ampholyte as depicted in the left side of Fig. 1. pH-dependent solubility of ampholytes is generally discussed in [20]. From theory, solubility of 1 is high below pKA1, where the protonated, cationic form of 1 is present in solution, and above pKA2, where the deprotonated, anionic form is present in solution. Poor solubility in the pH range between pKA1 and pKA2 where the zwitterionic form is prevalent in solution is expected. For measurement of pH-dependent solubility profiles aqueous buffers were chosen to obtain a pH profile between pH 1 and 9. Additionally solubility in the presence of chloride was measured
268
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
212 213 214 215 216 217 218 219 220 221 222 223 224
Please cite this article in press as: M. Vogtherr et al., Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy, Eur. J. Pharm. Biopharm. (2015), http://dx.doi.org/10.1016/j.ejpb.2015.02.016
253 254 255 256
259 260 261 262 263 264 265
269 270 271 272 273 274 275 276 277 278 279 280 281
EJPB 11841
No. of Pages 10, Model 5G
26 February 2015 4
M. Vogtherr et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2015) xxx–xxx
Fig. 1. Left: pH-dependent solubility profile expected for ampholytic behaviour of 1 as described in [20]. Right: experimentally observed pH dependent solubility of 1 (grey: chloride-free buffers, black: same buffers as grey curve with addition of 0.4 M NaCl).
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
to investigate the effect of chloride on solubility, as e.g. expected by a common ion effect for the hydrochloride salt. Experimental solubility of 1 is summarised in Table 1 and plotted in Fig. 1 at the right side as a function of pH. The highest solubility of 1 is found in the intermediate pH range, especially for buffers containing 0.4 mM NaCl. This pH-dependence of solubility is in strong contrast to the expectations for an ampholytic compound, for which solubility is lowest in the intermediate pH range where the uncharged, zwitterionic species is dominating in solution. By contrast, 1 has a solubility maximum in the intermediate pH range. This is particularly true for buffers containing 400 mM NaCl. Very low solubility independent of chloride concentration was observed under strongly acidic conditions (pH < pKA1 = 4.2), i.e. in 0.1 M HCl and 0.1 M H2SO4. Solubility in hydrochloric acid and sulphuric acid was similar. Conversion of the poorly soluble chloride salt of 1 into the equally insoluble sulphate salt occurred. Presence of a different solid state form was confirmed by powder X-ray diffraction of the precipitate. Fig. 2 shows a comparison of PXRD patterns as obtained from the parent form of 1, the sodium-salt of 1 as well as the chloride and sulphate salt of 1. In the intermediate pH range (pKA1 = 4.2 < pH < pKA2 = 7.9) a more than 1000 fold increased solubility was observed. This increase in solubility occurred in acetate buffer (pH 4.5) only in the presence of 400 mM NaCl. Irrespective of chloride concentration, this gain in solubility was observed in phosphate buffer at pH 7.4. Solubility of 1 in phosphate buffer with and without 400 mM NaCl was similar high. No effect of NaCl on solubility at this pH was observed. Finally at high pH (pH > pKA2 = 7.9), a decrease in solubility compared to the intermediate pH range was seen. This showed that the original hydrochloride of 1 was converted into a less soluble sodium salt as was also shown by powder X-ray diffraction (data not shown). Increased solubility in the presence of NaCl as measured by HPLC was always accompanied with opalescence and foam formation. This was indicative of micelle formation under these conditions. Apparently micelle formation was favoured by chloride ions. Micelle formation of the zwitterionic form of 1 explains the deviation between experimental pH dependence of solubility and the expectations.
Fig. 2. PXRD patterns obtained from precipitates obtained from solubility determinations. Top to bottom: Sulphate salt of 1, chloride salt of 1, sodium salt of 1, parent form of 1.
To get a closer understanding of the underlying mechanisms of the abnormal pH dependent solubility of 1, solutions were characterised by NMR spectroscopy.
322
3.2. API solubilisation by FaSSIF
325
The abnormal pH-dependence of solubility prompted the question how bile salts influence the solubility of 1. Solubility of 1 increased from
