Nanoaggregation of inclusion complexes of

1

1

Nanoaggregation of inclusion complexes of glibenclamide with cyclodextrins

2 3

David Lucioa, Juan Manuel Iracheb, María Fontc, María Cristina Martínez-Ohárriza

4 5

a

6

31080, Navarra. Spain.

7

b

8

Navarra. Irunlarrea s/n. Pamplona 31080, Navarra. Spain.

9

c

10

Department of Chemistry, Faculty of Sciences. University of Navarra. Irunlarrea s/n. Pamplona

Department of Pharmacy and Pharmaceutical Technology, Faculty of Pharmacy. University of

Department of Organic and Pharmaceutical Chemistry. University of Navarra. Irunlarrea s/n.

Pamplona 31080, Navarra. Spain.

11 12

Tel.: +34 948425600 (6378); fax: +34 948425649; e-mail address:

13

[email protected]; [email protected]; [email protected]; [email protected]

1

Abbreviations: Glibenclamide (GB), α-cyclodextrin (αCD), β-cyclodextrin (βCD), 2-hydroxypropyl-β-

cyclodextrin (HP-βCD), randomly methyl-β-cyclodextrin (RM-βCD), γ-cyclodextrin (γCD), dynamic light scattering (DLS), transmission electron microscopy (TEM), complexation efficiency (CE).

2 14

Abstract

15

Glibenclamide is a sulfonylurea used for the oral treatment of type II diabetes mellitus. This drug

16

shows low bioavailability as consequence of its low solubility. In order to solve this problem, the

17

interaction with cyclodextrin has been proposed. This study tries to provide an explanation about

18

the processes involved in the formation of GB-βCDs complexes, which have been interpreted in

19

different ways by several authors. Among native cyclodextrins, βCD presents the most

20

appropriate cavity to host glibenclamide molecules showing AL solubility diagrams (K1:1 ≈ 1700

21

M-1).

22

However, AL solubility profiles were found for βCD derivatives, highlighting the coexistence of

23

several phenomena involved in the drug solubility enhancement. At low CD concentration, the

24

formation of inclusion complexes can be studied and the stability constants can be calculated

25

(K1:1 ≈ 1400 M-1). Whereas at high CD concentration, the enhancement of GB solubility would

26

be mainly attributed to the formation of nanoaggregates of CD and GB-CD complexes (sizes

27

between 100 and 300 nm). The inclusion mode into βCD occurs through the cyclohexyl ring of

28

GB, adopting a semi-folded conformation which maximizes the hydrogen bond network.

29

As consequence of all these phenomena, a 150-fold enhancement of drug solubility has been

30

achieved using β-cyclodextrin derivatives. Thus, its use has proven to be an interesting tool to

31

improve the oral administration of glibenclamide in accordance with dosage bulk and

32

dose/solubility ratio requirements.

33

Keywords: Glibenclamide, cyclodextrins, inclusion complexes, non-inclusion phenomena, self-

34

association, nanoaggregates.

35

3 36

1. Introduction

37

Glibenclamide (GB) or glyburide is an oral hypoglycemic agent from the second generation of

38

sulfonylureas. This drug is widely used for the treatment of type II diabetes mellitus (non-insulin

39

dependent). In addition, recent studies have demonstrated its ability to prevent cerebral ischemia

40

and hemorrhagic stroke1, 2. In 2015, glibenclamide was included in the World Health

41

Organization model List of Essential Medicines3. From a physico-chemical point of view,

42

glibenclamide shows a lipophilic character and it is ascribed to the group II of the

43

Biopharmaceutical Classification System (low solubility, high permeability). The main issue of

44

its oral administration is its low bioavailability, consequence of its low solubility in physiological

45

media. Therefore, the dissolution of glibenclamide is considered to be the rate limiting step, as its

46

absorption after oral administration reaches 45% of the initial amount of drug4.

47

Different methods and strategies have been proposed to increase the solubility of this type of

48

lipophilic drugs, including the particle size reduction5, incorporation of hydrophilic polymers

49

(i.e. PEG, PVP, HPMC, alginate, starch derivatives, etc.)6, formation of inclusion complexes7 or

50

incorporation in solid dispersions8, nanoemulsions9 or nanoparticles10. In this context, the

51

inclusion complex method has been proven to be one of the most effective when trying to

52

increase aqueous solubility and release rate in a wide variety of drugs11. Furthermore, it allows

53

the removal of unpleasant odors and flavors while increasing chemical and physical stability of

54

certain drugs12, 13.

55

For this purpose, inclusion complex based on the use of cyclodextrins are largely preferred.

56

Cyclodextrins are cyclic oligosaccharides with a hydrophilic outer surface which are able to host

57

non-polar groups of hydrophobic molecules in their internal cavity. Additionally, its oral

58

administration has shown to be practically non-toxic14, 15, making them particularly interesting

59

for its use in food and pharmaceutical applications. Among the different classes of cyclodextrins,

60

oligosaccharides based on β-cyclodextrin (βCD) have shown to be the most appropriate to host

4 61

molecules with aromatic rings. However, the use of βCD in solid oral dosage forms is

62

conditioned by its low aqueous solubility. For this reason, soluble derivatives of βCD have been

63

developed over the past few years. Some worth-mentioning examples are 2-hydroxypropyl--

64

cyclodextrin (HP-βCD)16 and random methylated-β-cyclodextrin (RM-βCD)17.

65

The interaction of GB with different cyclodextrins has been previously reported and the use of

66

βCD has shown to increase in vivo bioavailability of GB18. However, discrepancies arise when

67

interpreting solubility diagrams and regarding the mechanism of drug inclusion into the

68

cyclodextrin cavity for GB:βCD and GB:HP-βCD systems. AL type19, 20, 21, 22, 23 and B type 24

69

solubility diagrams were reported for GB:βCD systems, whereas AL type4, 21, 23, 25 and AP type19

70

were reported for GB:HP-βCD systems. As the type of solubility diagram indicates the possible

71

stoichiometry of the inclusion complexes26, different interpretations arise about the mechanism

72

of complexation between cyclodextrins and glibenclamide. Moreover, the aforementioned

73

studies reported inconsistencies in stability constants, which is common when assessing drugs

74

with very low aqueous solubility.

75

The aim of this investigation was to study in depth the complexation process between

76

glibenclamide and different cyclodextrins (native and derivatives) and to establish the different

77

processes involved in the enhancement of glibenclamide solubility, that could be explained by

78

inclusion and non-inclusion processes simultaneously.

79

5

80

2. Materials and methods

81

2.1. Materials

82

Glibenclamide was supplied by Sigma-Aldrich (Spain). Native cyclodextrins (α, β and γCD),

83

random methylated-β-cyclodextrin (RM-βCD) and 2-hydroxypropyl--cyclodextrin (HP-βCD),

84

with an approximate substitution degree of 12 and 4.5 respectively, were supplied by Cyclolab

85

(Hungary). All aqueous solutions were prepared with deionized water obtained from a

86

commercial Millipore Elix 3 system. (0.1 mS/cm conductivity).

87

2.2. Phase solubility studies

88

The effects of cyclodextrins on the solubility of glibenclamide were studied in phosphate buffer

89

solutions (pH 7.4). An excess of GB was added to different solutions containing increasing

90

amounts of αCD, βCD, γCD (from 1 to 12 mmol/L for native CDs), HP-βCD and RM-βCD

91

(from 1 to 80 mmol/L for βCD derivatives). Sealed glass containers were magnetically stirred at

92

constant temperature (37°C) until equilibrium was reached (4 days). After equilibrium, an aliquot

93

of solution (3 mL) was withdrawn with a syringe filter (pore size 0.45 μm) and GB concentration

94

was determined at 300 nm by UV-visible molecular absorption spectrophotometry (Hewlet

95

Packard 8452A diode-array spectrophotometer). Each experiment was performed in triplicate

96

(coefficient of variation CV0.996). Nevertheless, the affinity of

157

βCD for GB was significantly higher than that of αCD or γCD. This fact may be related with the

158

dimensions of the internal cavity of the cyclodextrins. Thus, the diameter of this cavity in βCD

159

(6-6.5 Å) would be the most appropriate to host the non-polar groups of glibenclamide. αCD and

160

γCD show cavity diameters of 4.7-5.3 Å and 7.7-8.3 Å, respectively; these cavities were not

161

adequate to host GB moieties, resulting in weak drug-cyclodextrin interactions. For this reason,

9 162

native βCD and βCD derivatives have been chosen for a complete study about the complexation

163

with GB.

164 165

Figure 1. Solubility diagrams of GB:αCD (►), GB:βCD (■) and GB:γCD (□).

166 167

The solubility diagrams for βCD soluble derivatives (HP-βCD and RM-βCD) are shown in

168

Figure 2. In both cases, a very similar behaviour was observed; although, a slightly higher drug

169

solubility enhancement was achieved in the presence of RM-βCD than with HP-βCD.

170 171 172

Figure 2. Solubility diagrams of GB:HP-βCD (○) and GB:RM-βCD (●).

10 173

The experimental data for GB:HP-βCD and GB:RM-βCD systems were fitted to AL type

174

solubility diagrams. However, deviations from the linear behaviour took place when the

175

concentration of both CDs was approaching to zero. AL4, 21, 23, 25 and AP19 type behaviors have

176

been previously described for GB:HP-βCD systems. At this point it is noticeable that the type of

177

solubility diagrams obtained by different authors could depend on the cyclodextrin concentration

178

range studied. On the one hand, AL diagrams were usually obtained when an extensive range of

179

CD concentrations was assessed (typically from 0 to 80 mmol/L) but the experimental data at

180

low CD concentrations were scarce. On the other hand, AP solubility diagrams were obtained if

181

the concentration range evaluated was under 20 mmol/L. To the best of our knowledge, GB:RM-

182

βCD system has not been previously studied.

183

Data fitting to the equations that describe AL (Eq. 2) and AP type solubility profiles (Eq. 3) were

184

carried out for GB:HP-βCD and GB:RM-βCD systems32.

185

GB + CD

186

 K1:1 ·[GB]0  ·[CD]TOT [GB]TOT  [GB]0    1  K1:1 ·[GB]0 

187

GB-CD + CD

GB-CD

GB-CD2

K1:1 

[GB  CD] [GB]·[CD]

Eq. 2

K1:2 

[GB  CD2 ] [GB  CD]·[CD]

188

 [GB]TOT  [GB]0     K1:1 ·[GB]0  K1:1 ·K1:2 ·[GB]0 ·[CD]TOT [CD]TOT  

189

Initially, it was intended to fit all experimental data to equations 12 and 23 (AL and AP type

190

respectively). Nevertheless, poor determination coefficients (R20.996), negative stability constants were obtained based on slope and intercept

193

from linear fitting (Eq.4) as consequence of negative intercept values33.

Eq. 3

11

slope intercept·(1  slope)

194

K1:1 

195

Obviously, negative stability constants indicate that the intercept value cannot be used to

196

calculate K1:1 in these systems. Thus, and in order to elucidate this inconsistency, first approach

197

was to substitute the intercept value by the experimental aqueous solubility of GB in absence of

198

cyclodextrins ([GB]0 = S0,GB = 2  10-2 mmol/L) (Eq. 5)33.

199

K1:1 

200

It is not clear why the intercept of the phase solubility diagram is below S0,GB. Some authors

201

have reported similar behaviours for water-insoluble drugs, which have been described in

202

literature as AL solubility profiles34, typically for drugs presenting low intrinsic solubility below

203

to 0.1 mmol/L (0.02 mmol/L in case of glibenclamide). This behaviour has been justified as a

204

consequence of different phenomena which could modify the host-guest interaction and drug

205

solubility behaviour35. Among others, these phenomena include the non-ideality of water as

206

solvent due to its highly ordered structure36, self-association of drug molecules to form dimers37,

207

non-inclusion complex formation38, 39 and auto-aggregation of cyclodextrins40. For all these

208

reasons, the stability constant calculated includes not only inclusion complex formation but also

209

the associated non-inclusion phenomena.

210

The values of stability constants for all GB:CD complexes obtained from the solubility diagrams,

211

using either the intercept (Sint) value or S0,GB, according to eq. 4 and 5 respectively, are shown in

212

Table 1.

213 214

slope S 0,GB  (1  slope)

Eq. 4

Eq. 5

12

215

Table 1. Apparent stability constants calculated using either Sint or S0,GB (37°C; pH 7.4).

216

*

Negative values for stability constant have no physical sense.

Cyclodextrin

Slope

R2

K1:1 (L/mol) using Sint

α-CD

0.0057

0.996

266

275

β-CD

0.0388

0.997

2103

1945

HP-β-CD

217

using S0,GB

0.0391

0.999

-863

*

1960

*

2159

RM-β-CD

0.0418

0.998

-732

γ-CD

0.0049

0.999

250

235

S0,GB = 10.25 mg/L.

218 219

It is noteworthy that both K1:1 values obtained for complexes with native CDs (,  and CD) are

220

similar. This fact did not take place in case of complexation with CD derivatives, where K1:1

221

values were strongly dependent on the method used to calculate the apparent stability constants

222

(eq. 4 or eq. 5).

223

In this sense and in order to obtain a better estimation of stability constants for GB complexation

224

with βCD derivatives, this value should be calculated using low CD concentrations close to ideal

225

infinite dilution, where only host-guest interactions happen and non-inclusion phenomena are

226

avoided. Stability constant values obtained for the complexes with βCD, HP-βCD and RM-βCD

227

at low CD concentrations (0-5 mmol/L) were 1743, 1385 and 1432 L/mol (R20.998)

228

respectively using equation 4. These values are on good agreement if comparing the two

229

different ways to calculate the stability constant (using Sint or S0,GB).

230

However, the amount of cyclodextrin needed in pharmaceutical formulations do not usually

231

approach ideal infinite-dilution, on the contrary, high CDs concentrations are needed to achieve

232

the intended solubilization of the drug. In this sense, the solubilizing effect is an important aspect

233

to be considered and, thus, the concept of complexation efficiency (CE)32 gains special

13 234

importance in the pharmaceutical field. CE is calculated from the slope of solubility diagram and

235

its value does not depend on either S0,GB or Sint (Eq. 6).

236

CE 

237

Based on CE determinations, it is possible to estimate the molar ratio (Eq. 7) and the dosage bulk

238

for these complexes33. Dosage bulk parameter should be considered for the use of drug-

239

cyclodextrin complexes as solid dosage forms.

240

Molar Ratio  GB : CD  1 : (1 

241

The stability constant, calculated close to ideal infinite dilution conditions, corresponds to the

242

contribution of inclusion complexation; whereas complexation efficiency, molar ratio and dosage

243

bulk values include all the processes involved on glibenclamide solubilization (Table 2).

[GB  CDn ] slope  S 0,GB ·K1:1  [CD] 1  slope

Eq. 6

1 ) CE

Eq. 7

244 245

Table 2. Stability constants values for GB:βCDs inclusion complexes and parameters involving

246

inclusion and non-inclusion phenomena (complexation efficiency, molar ratio and dosage bulk). Cyclodextrin

247

Stability Constant*

CE

Molar Ratio Dosage bulk (mg)

K1:1 (L/mol)

R2

β-CD

1743

0.9995

0.040

1:26

300

HP-β-CD

1385

0.9999

0.041

1:26

370

RM-β-CD

1432

0.9988

0.044

1:23

300

*

Using 0-5 mmol/L CD concentration range, where only inclusion phenomena take place.

248 249

For GB:CD and GB:HP-βCD systems, the calculated CE values were 0.040 and 0.041,

250

respectively. That mean that 26 molecules of either βCD or HP-βCD were involved in the

251

solubilization process of one GB molecule, whereas 23 RM-βCD molecules (CE=0.044) are

252

needed for the same process.

14 253

In this regard, it is well known that two requisites should be considered for the use of drug-CD

254

systems in solid oral dosage forms41: an upper limit in dosage bulk of about 800 mg and a

255

dose/solubility ratio ≤ 250 mL. As it can be observed in Table 2, the dosage bulk needed was

256

clearly under 800 mg. The initial (2.5 mg) or the usual dose of GB (5 mg) led to a dose:solubility

257

ratio about 250 mL and 500 mL in absence of CD respectively, which was 2-fold higher than the

258

limit usually established. However, considering the ability of βCDs to increase GB solubility,

259

only 1 mmol/L of any βCDs would be enough to meet the requirement for the higher dose of

260

glibenclamide. In summary, all GB:βCD inclusion complexes tested in this study fulfill the two

261

requirements aforementioned in relation to solid oral dosage forms. But it is worthy to note that

262

the solubility of the CD employed must be also considered and, in this sense, cyclodextrin

263

derivatives provide an advantage over the use of the native cyclodextrin.

264

3.2. Inclusion phenomena

265

As has been mentioned, significant differences between the stability constant values for

266

cyclodextrin derivatives were observed when non-appropriate concentration range was studied.

267

Due to this reason, the mechanisms which govern the solubility enhancement of GB have to be

268

explained as a combination between the inclusion complex formation and other phenomena, such

269

as self-aggregation of drug-cyclodextrin complexes, so different techniques were employed to

270

study each of these cases. The formation of inclusion complexes at low CD concentrations has

271

been tested by molecular docking and NMR studies. Non-inclusion or aggregation processes

272

taking place at high CD concentrations were studied by transmission electron microscopy (TEM)

273

and dynamic light scattering (DLS) techniques. It is worthy to note that at high cyclodextrin

274

concentration both, inclusion and non-inclusion phenomena take place.

275

3.2.1. Molecular docking

276

In order to verify the experimental results obtained in solubility assays and to fix the

277

complexation mechanism between GB and CDs, six initial configurations were constructed (Fig.

15 278

3). All of them in accordance with a design in which the glibenclamide rings were included in

279

the cyclodextrin cavity through the major or minor faces of βCDs.

280 281

Figure 3. Initial structural conformations of GB:βCDs complexes with the cyclohexyl C ring, the

282

central aromatic B ring or the external aromatic A ring inside the cyclodextrin cavity.

283 284

After the optimization, glibenclamide adopts a semi-folded conformation when it is incorporated

285

inside the cavity of cyclodextrins (Figure 4). With respect to the data obtained for native β-

286

cyclodextrin, two preferred configurations with a better pose scoring value were obtained. Both

287

complexes show the C ring (cyclohexyl) included in the βCD inner space whereas the rest of the

288

molecule crosses the major face (54% of the poses obtained, Figure 4A) or the minor face of the

289

host (40% of the poses, Figure 4B). The complexes appear stabilized by a network of hydrogen

290

bonds established between the sulfonylurea, amide and methoxy groups of GB and the hydroxyl

291

moieties placed on the outer faces of βCD.

16

292 293

Figure 4. Models of preferred configurations obtained from docking calculations of GB:βCD (A

294

and B), GB:RM-βCD (C), and GB:HP-βCD (D).

295 296

In the case of RM-βCD, the most stable complexes also showed the cyclohexyl C ring included

297

in RM-βCD (65% of the obtained poses), with a preferred orientation of the rest of drug

298

molecule directed towards the minor face of the host (Figure 4C). The network of hydrogen

299

bonds appeared to be lower than in the case of GB:βCD, but the hydrophobic interactions

300

established with the host methoxy groups also contributed to stabilize the complex.

301

With respect to the data obtained for GB:HP-βCD, and similarly to the GB:βCD complexes, two

302

preferred configurations with a better pose scoring value were obtained. Both complexes showed

303

the cyclohexyl C ring included in the HP-βCD cavity and the rest of the molecule crossing the

304

major face (62% of the poses obtained, Figure 4D) or the minor face of the host (30% of the

305

poses). The hydrogen bond network established between GB and HP-βCD is significantly

306

stronger than the established with RM-βCD but weaker than the established with native βCD.

307

17 308

3.2.2. NMR analysis

309

1

310

investigated in order to explore the inclusion mode of complex formation (Fig. 5).

311

For glibenclamide and due to the presence of two amide groups, 4 different tautomeric forms that

312

could modify the interaction with cyclodextrins can be presented42. If the tautomeric equilibrium

313

takes place in the amide group near the external aromatic ring, this configuration change could

314

difficult the formation of inclusion complexes between CD and the external aromatic A ring of

315

GB. In fact, this would be due to a higher steric hindrance if hydrogen bonds between enol and

316

methoxy groups were established. Additionally, electrostatic and volume constraints would

317

preclude the inclusion of the chlorine atom, located at the opposite side of the aromatic A ring,

318

into the cyclodextrin cavity. For these reasons, to facilitate interpretation of the results provided

319

by NMR spectra, the analysis was based on the most likely structures proposed by molecular

320

modelling.

321

The first evidence of inclusion complexation between glibenclamide and cyclodextrins is the

322

displacement in chemical shift of the signals corresponding to the internal protons of the

323

oligosaccharide. When the chemical shifts of pure GB were compared with the corresponding

324

GB:CD complex signals, two different phenomena were observed. On the one hand, 1H-aromatic

325

signals did not experience any appreciable shift. The mixture of signals at 7.5 ppm corresponding

326

to glibenclamide aromatic protons (H3’, H5’ and H6) did not change when GB:CD ratio was

327

modified. This fact evidences that the external aromatic A ring was not include into the CD

328

cavity. On the other hand, a displacement in the signals corresponding to H9' (duplet, 6.30 ppm)

329

and a modification of the cyclohexyl C ring signals H2''-H6'' (multiplet, 1.62 ppm and multiplet,

330

1.19 ppm) were observed. These observations would be clear evidences of the inclusion of the

331

cyclohexyl C ring into the cyclodextrin cavity.

332

H and 2D NMR spectra of glibenclamide, cyclodextrins and GB:CDs complexes were

18

333 334

Figure 5. Structure of GB (A) and HP-βCD (B) molecules for the identification of 1H-NMR

335

signals (GB in horizontal axis and HP-βCD in vertical axis) and partial NOESY spectrum of

336

GB:HP-βCD (C).

337 338

Bidimensional analysis was carried out in order to confirm unambiguously the host-guest

339

interactions suggested by 1H-NMR analysis. Both intramolecular and intermolecular interactions

340

can be observed in NOESY analysis (Fig. 5). The strongest intramolecular couplings can be

341

observed for the signals corresponding to GB ring protons among themselves. In this sense, the

19 342

interaction between methoxy and amide groups (H8 and H3) can also be detected in

343

bidimensional analysis.

344

Intermolecular couplings allowed us to confirm the mechanism of complex formation.

345

Interactions between both, H9’ (-CONH-) and cyclohexyl C ring protons of GB (H2''-H6''), with

346

internal protons of CD cavity and CH3 protons of 2-hydroxypropyl groups has been detected,

347

proving the inclusion of the cyclohexyl group of GB into CD cavity. The methoxy group also

348

showed interaction with -CH3 of 2-hydroxypropyl groups, which could be due to the folded

349

conformation that GB molecule adopts when an inclusion complex is formed, contributing to the

350

stabilization of the complex. These interactions of H9’ and methoxy group of GB with -CH3 of

351

CD also proved the orientation of cyclodextrin molecule suggested by molecular docking, with

352

the major face of CD near to the rest of glibenclamide molecule.

353 354

3.3. Self-aggregation

355

Inclusion complexes can coexist with non-inclusion phenomena, such as the self-aggregation of

356

cyclodextrins and their complexes. The aggregates formed can be able to promote the additional

357

solubilization of guest molecules through non-inclusion complexation and/or micelle-like

358

structures38, 43.

359

As calculated from the complexation efficiency data, more than 20 cyclodextrins were involved

360

in the solubilization of each individual glibenclamide molecule. This fact evidences an important

361

change in the chemical environment of the drug molecules (at high cyclodextrin concentrations),

362

which would be related with a self-aggregation phenomena. In order to confirm this idea, DLS

363

and TEM were employed.

364

20 365

3.3.1. Dynamic Light Scattering (DLS)

366

DLS measurements were carried out using the same GB-CD ratios employed in the solubility

367

diagrams. The most representative changes in size distribution with increasing CD concentration

368

are shown in Figure 6.

369 370

Figure 6. Size distribution for saturated solutions of GB in presence of HP-βCD (■) and RM-

371

βCD (□) at CD concentrations of 1, 5 and 30 mmol/L.

372 373

From the analysis of size distributions, a unimodal population corresponding to inclusion

374

complex with monomeric CDs and free CD molecules can be observed44. It presents a mean

375

hydrodynamic radius below 1 nm for CD concentration lower than 5 mmol/L (mean RH=0.74 nm

376

for GB:RM-βCD and 0.86 nm for GB:HP-βCD at 1 mmol/L).

21 377

At higher CD concentrations (5-30 mmol/L) a new size distribution appears. This population

378

could be attributed to self-aggregation of cyclodextrins and drug-cyclodextrin complexes. The

379

mean radius of the second distribution appeared around 50 nm and this signal increased with CD

380

concentration, whereas the intensity of the first population (RH