Mucoadhesive chitosan-coated nanostructured lipid carriers for oral delivery of amphotericin B 12

24 This study describes the properties of an amphotericin B-containing mucoadhesive 25 nanostructured lipid carrier (NLC), with the intent to maximize uptake within the gastrointestinal 26 tract. We have reported previously that lipid nanoparticles can significantly improve the oral 27 bioavailability of amphotericin B (AmpB). On the other hand, the aggregation state of AmpB within 28 the NLC has been ascribed to some of the side effects resulting from IV administration. In the 29 undissolved state, AmpB (UAmpB) exhibited the safer monomeric conformation in contrast to AmpB 30 in the dissolved state (DAmpB), which was aggregated. Chitosan-coated NLC (ChiAmpB NLC) presented a slightly slower AmpB release profile as compared to the uncoated formulation, 32 achieving 26.1 % release in 5 hours. Furthermore, the ChiAmpB NLC formulation appeared to 33 prevent the expulsion of AmpB upon exposure to simulated gastrointestinal pH media, whereby up 34 to 63.9 % of AmpB was retained in the NLC compared to 56.1 % in the uncoated formulation. The 35 ChiAmpB NLC demonstrated mucoadhesive properties in pH 5.8 and 6.8. Thus, the ChiAmpB NLC 36 formulation is well-primed for pharmacokinetic studies to investigate whether delayed 37 gastrointestinal transit may be exploited to improve the systemic bioavailability of AmpB, whilst 38 simultaneously addressing the side-effect concerns of AmpB.

tract. We have reported previously that lipid nanoparticles can significantly improve the oral 27 bioavailability of amphotericin B (AmpB). On the other hand, the aggregation state of AmpB within 28 the NLC has been ascribed to some of the side effects resulting from IV administration. In the 29 undissolved state, AmpB (UAmpB) exhibited the safer monomeric conformation in contrast to AmpB 30 in the dissolved state (DAmpB), which was aggregated. Chitosan-coated NLC (ChiAmpB NLC) 31 presented a slightly slower AmpB release profile as compared to the uncoated formulation, 32 achieving 26.1 % release in 5 hours. Furthermore, the ChiAmpB NLC formulation appeared to 33 prevent the expulsion of AmpB upon exposure to simulated gastrointestinal pH media, whereby up 34 to 63.9 % of AmpB was retained in the NLC compared to 56.1 % in the uncoated formulation. The 35

Introduction 42
Amphotericin B (AmpB) is a broad spectrum antifungal agent commonly used to treat invasive 43 systemic fungal infections and visceral leishmaniasis (Legrand et al. 1992). It has a large glycosylated 44 lactone ring, coupled with asymmetrical distribution of hydrophobic polyene chromophore and 45 hydrophilic polyhydroxyl groups (Jung et al. 2009; Silva et al. 2013). Due to its amphipathic nature, 46 AmpB tends to self-aggregate in aqueous solutions, forming, dimers or polyaggregates. The dimers 47 are usually associated with the most toxic properties of AmpB and unfortunately this is the 48 predominant state in the reconstituted marketed intravenous (IV) formulation, Fungizone ® (Barwicz et 49 al. 1992; Raquel Espada et al. 2008). AmpB exerts its antifungal properties by binding to ergosterol 50 within fungal membranes, forming transmembrane pores that allow depletion of intracellular ions, 51 which eventually lead to the cell death. In a similar fashion, AmpB also binds to mammalian 52 cholesterol within the plasma membrane, which lead to severe side effects, notably nephrotoxicity 53 (Butani et al. 2016). This is the hallmark of the toxic effects of the dimers mentioned above (Radwan 54 et al. 2017). There is evidence that when delivered orally, these side-effects are minimized primarily In terms of novelty on the current work and in advancement of the oral AmpB SLN formulation 73 developed in our labs (Amekyeh et al. 2015), we have sought to exploit a sluggish transit of the NLC 74 within the gastrointestinal tract in order to maximize uptake via the lymphatic pathway by way of 75 coating the NLC with chitosan. Chitosan, a natural, non-toxic, biocompatible polycationic 76 polysaccharide, derived from partial deacetylation of chitin was employed as the mucoadhesive 77 polymer coating (Sandri et al. 2017). Therefore, this piece of work reports on the formulation and 78 characterisation of uncoated and chitosan-coated AmpB-loaded NLC. We hypothesise that through 79 mucoadhesion of the NLC we will confer a prolonged gastrointestinal transit in the small intestine 80 which would result in improved uptake via lymph and hence improve bioavailability of AmpB. 81 Furthermore, we are cognizant of the correlation between the aggregated state of AmpB and its 82 toxicity. We exploited the pH-solubility and stability profiles of AmpB under alkaline conditions 83 during the formulation of the NLC, which assumes the monomeric configuration. This configuration 84 is known to manifest fewer side effects (Lance et al. 1995

Formulation of AmpB-loaded NLC (AmpB NLC) 96
AmpB was incorporated during the formulation of the NLC either in a dissolved state 97 (DAmpB) or undissolved state (UAmpB), at the initial step of preparation (method 1) or at the final 98 step in the NLC formulation (method 2). DAmpB comprised of 10 mg/mL of AmpB in 0.1 M NaOH. 99 Method 1: Briefly, the oily phase (OP), comprised of 290 mg of beeswax and 10 mg of coconut 100 oil was melted at 70°C. AmpB either in undissolved or dissolved state was added to the melted lipids 101 and mixed. The aqueous phase (AP), consisted of 50 mg of lecithin, 50 mg of Tween -80 and 10 mL 102 of deionised water was stirred at 500 rpm using a magnetic stirrer for 45 minutes at 70°C and then 103 added to the melted OP above. The resulting mixture was homogenised at 12,400 rpm for 8 minutes 104 using a high speed homogeniser (Ultra-Turrax T25, Germany). The coarse emulsion was further 105 subjected to probe ultrasonication (Q500 QSonica, Newtown, CT, USA) at 20 % amplitude for 8 106 minutes. Finally, the emulsion was added into sufficient deionised water (4°C) under stirring to a 107 total of 100 mL. 108 Method 2: The preparation of AmpB NLC formulation was similar to method 1 except that the 109 incorporation of AmpB occurred at the final step of the formulation, in which the AmpB was added to only DAmpB was utilised. 113

Formulation of chitosan-coated AmpB-loaded NLC (ChiAmpB NLC) 114
The physical adsorption of chitosan on the formulated NLC was done by addition of 0.2 %w/v 115 chitosan solution in 1 % acetic acid dropwise into the NLC formulations at a ratio of 1:40 v/v under 116 mechanical stirring at 250 rpm for 15 minutes at room temperature. 117 118 119

2.4.
Physical properties of the formulations 120

Particle size, polydispersity index (PDI) and zeta potential 121
The particle size, polydispersity index (PDI) and zeta potential (ζ) of the AmpB NLC and 122 ChiAmpB NLC formulations were measured using Zetasizer Nano ZS (Malvern, UK). Prior analysis, 123 the samples were diluted appropriately using deionised water to avoid multiple scattering. All 124 measurements were carried out in triplicate at 25°C and results were expressed as mean ± standard 125 deviation. 126

2.4.2.Aggregation states of AmpB NLC and ChiAmpB NLC formulations 127
The predominant aggregation state of AmpB in the different formulations was characterised using 128 a UV-visible spectrometer (Epoch Microplate Spectrophotometer, Bio Tek Instruments, USA). 129 Absorption spectra from the formulations were recorded from 300-450 nm with a resolution of 1 nm 130 at room temperature. All formulations were diluted with deionised water (1:10 v/v) so that results 131 were within the linear sensitivity of the instrument. The absorbance from the blank formulations was 132 also measured in order to eliminate effects due to artifacts. The predominant aggregation state of 133 AmpB within each spectrum was determined by calculating the ratio of absorbance at 332 nm (peak 134 of the dimer) to that at 407 nm (peak of the monomer). 135

2.4.3.Morphology and topography 136
The morphology and topography of the formulations were examined using a scanning 137 transmission electron microscopy (STEM) system, Quanta 400F (FEI Company, USA). Prior to 138 analysis, the undiluted samples were applied on a formvar-coated copper grids without fixation and 139 air-dried. Samples were observed under scanning transmission of 20 kV in high vacuum. 140

2.4.4.Encapsulation efficiency (% EE) and drug loading (% DL) 141
A direct method was used to determine the encapsulation efficiency (% EE) and drug loading 142 (% DL) in AmpB NLC and ChiAmpB NLC formulations. Briefly, 1 mL of the formulation was 143 precipitated by addition of 10 mL of acetonitrile. The resulting mixture was centrifuged at 20 000 rpm 144 for 10 minutes at 4°C. The supernatant was decanted and DMSO: MeOH (1:1) was added to the 145 pellet containing the encapsulated drug which was heated at 70°C. The amount of AmpB was 146 measured using high performance liquid chromatography (HPLC) system (1260 Series, from Agilent 147 technologies, Waldbronn, Germany, equipped with a 15 cm x 4.6 mm reversed-phase C-18 column, 148 Hypersil Gold, ThermoFisher Scientific, Waltham, United States, 5 µm particle size stationary phase). where, WT is the amount of AmpB in the system, WS is the amount of AmpB detected in the sediment 157 and WN weight of nanoparticles obtained from freeze-dried sediments. 158

In vitro release studies 159
A 50 µl aliquot of AmpB NLC or ChiAmpB NLC formulations was mixed with 950 µl 160 phosphate buffered saline (pH 7.4) containing 1 % Tween -80 into six seeded tubes and rotated at 120 161 rpm in a rotary shaker (WiseCube ® , Witeg Inc., Germany) maintained at 37 o C. Sink conditions for 162 AmpB were maintained in each tube. At predetermined time intervals (15 min, 1, 2, 3, 4 and 5 hour), 163 one tube was removed and the nanoparticles were precipitated using 1000 µl acetonitrile, followed by 164 centrifugation at 20 000 rpm for 10 minutes at 4°C to pellet the particles. The amount of AmpB 165 released was determined by analyzing the supernatant using the HPLC system described above after 166 correction for free AmpB. Three independent runs were conducted and the results are expressed as 167 mean ± standard deviation. 168 Since the formulations were designed for absorption from the upper gastrointestinal tract, the 169 effect of variable pH on the retention of AmpB within the AmpB NLC and ChiAmpB NLC 170 formulations were investigated. A 1:20 v/v dilution of the formulations in phosphate buffer pH 5.8 171 (British Pharmacopeia) representing the proximal small intestine were firstly incubated for 2 hours, 172 followed by adjustment of the pH to 6.8 (distal small intestine) with 6 µl of 3M NaOH and further 173 incubated for 4 hours (Ovesen et al. 1986;Evans et al. 1988). The percentage of AmpB retained in the 174 formulations were determined similarly as described above. 175

Mucoadhesion studies 176
The mucoadhesive properties of the AmpB NLC and ChiAmpB NLC formulations were 177 determined turbidimetrically after dispersing the formulations in type III porcine gastric mucin 178

Stability studies 192
The AmpB NLC and ChiAmpB NLC formulations were stored at 4°C and protected from light. 193 Aliquots were withdrawn at appropriate time intervals and the particle size, PDI, ζand aggregation 194 state were evaluated. 195 The effect of variation in pH on the physical properties of AmpB NLC and ChiAmpB NLC 196 formulations was also studied by means of changes in the particle size and ζ. 50 µl of the formulations 197 were mixed in 950 µl of phosphate buffer, pH 5.8 and 6.8 (British Pharmacopeia). The samples were 198 incubated at 37°C using a rotary shaker operated at 120 rpm for 2 hours. The changes in the particle 199 size andζwere evaluated using Zetasizer Nano ZS (Malvern, UK). ChiAmpB NLC formulated with DAmpB by almost five-fold (1141 ± 28 nm) using method 1 and an 235 increase of 35 nm using method 2. Further characterisation on the DAmpB NLC using method 1 was 236 not carried out because of the significant increase in size of the NLC after the coating which is not 237 have to be contended with. An increase in AmpB cargo within the NLC was carried out on UAmpB 239 (method 1) and DAmpB (method 2) at a 50 mg AmpB threshold and a stability study was carried out 240 on the resulting formulations. The results on particle size, PDI andζof the AmpB NLC and the 241 chitosan-coated counterpart (ChiAmpB NLC) from both methods are presented in Table 2  There was a slight decrease in particle sizes of the uncoated NLC series (AmpB NLC) formulated 245 using method 1 with 50 mg UAmpB compared to the previous uncoated series formulated with 10 mg 246 of UAmpB (Table 1). However as in the initial series (Table 1), there was a decrease in size of the 247 NLC formulated using method 2 (DAmpB) compared to the method 1 (UAmpB). Interestingly, the 248 size of the DAmpB NLC remained essentially unchanged after we increased the AmpB load from 10 249 to 50 mg. Furthermore, the particle size of the DAmpB NLC at 50mg AmpB load also remained 250 essentially unchanged during storage from both methods over the 120-day study period: method 1 (p 251 = 0.102) and method 2 (p = 0.428). On the other hand, after coating with chitosan there was an 252 increase in the sizes of the ChiAmpB NLC at 50 mg AmpB load in both methods of NLC formulation. 253 Furthermore, there was a significant increase in size of the ChiAmpB NLC formulated via method 2 254 as a function of storage time (p = 0.039), indicating that the formulations from this method may be 255 unstable upon long-term storage (Tan & Billa 2014). It is worth noting that there was a progressive 256 reduction in the particle size andζof the ChiAmpB NLC formulated via method 1, from day 1 to 120. 257 This can be explained by the fact that in order to lower the surface free energy of the system, the 258 particles tend to minimize their surface area to volume ratio and thus, form larger particles at the 259 expense of smaller ones. However, we inferred that in the case of the ChiAmpB NLC system, the absorption spectra for ChiAmpB NLC prepared via method 1 presented a similar pattern as the 285 uncoated NLC formulation but with lower absorbance intensities. Likewise, the absorption spectra 286 from ChiAmpB NLC prepared by method 2 mirrors the spectra of the uncoated NLC which is 287 attributable to the dimer conformation. 288 The aggregation ratio within each formulation was obtained as the ratio of the absorbance at 289 332 nm (peak of the dimer) to that at 407 nm (peak of the monomer). An aggregation ratio > 1 290 indicates that more than 50 % of the AmpB is in the aggregated state while ratio < 0.2 reflects nearly 291 100 % monomer form. The aggregation ratios of AmpB in uncoated and chitosan coated NLC 292 prepared by method 1 were all below 1 over a 120-day period, indicating a predominantly monomeric 293 conformation ( Figure 2). 2008). On the other hand, the extent of AmpB aggregation in the NLC formulation prepared by 302 method 2 decreased over time from a highly aggregated state (3.9) that fell to 0.2 on day 35, which is 303 indicative of growth of AmpB to the monomeric state during storage. There was a slight increase in 304 the aggregation ratio on day 120 however this was still below the threshold 1. However, after coating 305 with chitosan, AmpB (method 2) remained in the aggregated state throughout the 120-day study 306 period. In fact, it appears that the aggregated state grew over time. Clearly, the mode of incorporation 307 of AmpB is crucial in determining the ultimate conformation. Due to the dimer conformation of 308 AmpB and the instability of the formulations from method 2, further studies were discontinued using 309 this method and focus is now on AmpB NLC and ChiAmpB NLC formulations using method 1. 310 Figure 3 shows the STEM images of the AmpB NLC and ChiAmpB NLC formulations 311 whereby the nanoparticles appeared to be spherical and discrete. The particle sizes from both AmpB 312 NLC and ChiAmpB NLC formulations are in agreement with the results obtained in Table 2. The 313 average encapsulation efficiency of AmpB NLC formulation was 83.4 ± 0.72 % with drug loading of 314 12.3 ± 0.11 % (Table 3). This high % EE and % DL can be attributed to the high lipophilicity of 315 12.3 ± 0.11 11.0 ± 0.04 Table 3. Encapsulation efficiency and drug loading of AmpB NLC and ChiAmpB NLC formulations. The release studies of AmpB from the NLC was carried out on AmpB NLC and ChiAmpB 331 NLC formulations prepared using method 1. The release medium was phosphate buffer pH 7.4 332 containing 1% Tween -80. Free AmpB (control) showed rapid release, 97.6 ± 0.1% of AmpB being 333 released within 15 minutes and this is in accordance with study by Jain et al. 2014. In contrast, both 334 formulations showed a biphasic release profile, with a burst release observed initially, followed by a 335 more sustained release, as presented in Figure 4a  The burst release during the first 15 minutes suggests the presence of some of the coconut oil 344 at the surface of the NLC due to a difference in the melting points of the beeswax and coconut oil so 345 that the former begins to crystallise before the latter during the cooling process of the NLC. The 346 crystallisation extrudes some of the coconut oil to the surface as it carries part of the dissolved AmpB 347 along. Therefore there is a regional accumulation of AmpB at the outer region of NLC which is 348 released as in a burst (Hu et al. 2005;Teeranachaideekul et al. 2007). Furthermore, the release profiles 349 of AmpB from AmpB NLC and ChiAmpB NLC formulation were superimposable, albeit a slightly 350 lower release from the ChiAmpB NLC which can be attributed to impedance in diffusion of AmpB by 351 the chitosan barrier coating. 352 The extent to which AmpB was retained within the formulations after dispersion in media at 353 pH 5.8 and 6.8 media is presented in Figure 4b. pH 5.8 is the typical pH of the proximal small 354 intestine whilst pH 6.8 represents the distal small intestine. Only about 20 % of AmpB was expelled 355 from both AmpB NLC and ChiAmpB NLC formulations during incubation in pH 5.8 in the first two 356 hours (p = 0.484). In order to mimic the gastrointestinal transit from duodenum, jejunum to ileum, the 357 pH of the medium was raised to 6.8, by the addition of 6 µl of 3M NaOH. After only 30 minutes of 358 incubation in media with pH 6.8, the percentage of AmpB retained in the AmpB NLC and ChiAmpB 359 NLC formulations were 56.1 ± 1.8 % and 63.9 ± 2.8 % respectively. Thus it apparent that the chitosan 360 coating shielded AmpB from expulsion from the NLC (Yang et al. 2012). 361 In a parallel study, the effect of variable pH on changes in the physical properties of the 362 formulations in terms of particle size andζwere conducted and presented in Figure 5. There was no significant change in the particle size of the AmpB NLC formulation in pH 5.8 369 medium (p = 0.332). Although there was a slight increase in particle size in pH 6.8 (p=0.003), it was 370 not aggregation-related since the final particle size remained at 208.1 ± 5.0 nm. Despite a decrease 371 inζvalues of AmpB NLC formulation of approximately 19 mV, to -16.9 ± 0.8 (pH 5.8) and -16.5 ± 0.3 372 mV (pH 6.8), the formulation remained essentially stable as reflected via the particle sizes, which 373 suggests that adequate electrostatic repulsion was maintained among the particles, that prevented 374 agglomeration (Amekyeh et al. 2017). In pH 5.8 and 6.8 media, the ChiAmpB NLC formulation 375 registered a two-fold increase in particle size with a marked drop in ζ, from +18.8 ± 0.3 to -8.1 ± 1.4 376 (pH 5.8) and -10.9 ± 1.1 mV (pH 6.8). This suggests the neutralisation of the positive charge density 377 on fresh ChiAmpB NLC by the anions present in the phosphate buffer. The reduction in the ζ favour 378 the van der Waals interactions which result in the increase in particle size of the nanoparticles 379 (Bhattacharjee 2016). 380 The mucoadhesive properties of two types of NLC formulations were evaluated based on 381 turbidimetric measurements which measures the increase in turbidity of the system as particles absorbance values compared to AmpB NLC formulation in both pH conditions. We may conclude 393 that mucoadhesion properties of the ChiAmpB NLC formulation is mostly driven by electrostatic 394 interactions between positively charged chitosan and negatively charged (COO -) mucin protein (He et 395 al. 1998;Rençber et al. 2016). 396

Conclusions 397
An AmpB-containing NLC was successfully formulated which demonstrated mucoadhesive 398 properties at pH values representing possible absorption regions in the small intestine. Based on a 399 previous study, we believe this formulation has the potential for improved uptake from the small 400 intestine due to mucoadhesion and hence improved bioavailability of AmpB. Furthermore, the 401 UAmpB-containing NLC formulations is more stable and presented the safer conformation of AmpB 402 compared to DAmpB-containing NLC formulation. Therefore, this formulation is primed for studies 403 to affirm the improved pharmacokinetics of AmpB whilst at the same time the toxicity concerns have 404 been addressed. 405 Declaration 406