Materials and methods 1. Materials

TA was purchased from Hänseler AG (Herisau, Switzerland). Resomer^® RG 503H (RG 503H; 50/50 poly(D, L-lactide-coglycolide), MW 24-38 kDa, intrinsic viscosity 0.32-0.44 dL/g), Resomer^® R207 (R207; poly(D, L-lactide), MW 199.8 kDa, intrinsic viscosity 1.3-1.7 dL/g), fluorescein, and dichloromethane were purchased from Sigma-Aldrich Chemie (Buchs, Switzerland). Nile Red was purchased from TCI Europe N.V. (Zwijndrecht, Belgium).

Polyvinyl alcohol (PVA; MW 72kDa) was purchased from Axon lab AG (Baden, Switzerland). All other chemicals and solvents were of analytical grade.

2.2. Preparation of TA loaded MP

These were prepared by either the conventional oil in water (o/w) emulsion technique (OW1-TA10 (with RG 503H) and OW2-TA10 (with R207)) or the newly developed FFT (FFT1-TA10

and FFT2-TA20). The composition of each MP batch and the preparation method are given in Table 1.

135 subsequent slow addition to 25 mL of 1% (w/v) aqueous PVA solution under homogenization at 7,500 rpm for 1 min using a Polytron^® PT 2500 E Homogenizer (Kinematica AG; Luzern, Switzerland). The o/w emulsion was then transferred to 100 mL of 1% (w/v) aqueous PVA solution under homogenization at 5,000 rpm for 2 min. The final emulsion was left overnight at room temperature under continuous stirring to evaporate the organic solvent and to harden the MP. The resulting MP were centrifuged at 1620 × g for 5 min and the pelleted MP were washed twice with Milli-Q^® water and ethanol/water (25/75) mixture to remove unloaded TA.

The final microparticle pellet was redispersed in 1.0 mL Milli-Q^® water and dried at room

136 (450 mg of RG 503H for FFT1-TA10 and 300 mg of each polymer for FFT2-TA20) and TA (50 mg and 150 mg for FFT1-TA10 and FFT2-TA20, respectively) in dichloromethane (17 mL) at 40°C and then removing the organic solvent using a Rotavapor^® R-124 (Buchi Labortechnik AG; Switzerland) to obtain a solid solution of TA and polymer. The TA-polymer solid solution was freeze-fractured using a cryogenic mill (6770 Freezer/Mill^®, SPEX SamplePrep; Stanmore, UK). The mill was set for 10 cycles operating at an impaction rate of 15 cps with a run time of 1.5 min and a cool time of 2 min between each cycle. Liquid nitrogen, a cryogen, was added and grinding initiated. The fractured particles were then passed through a 90 µm sieve (#170 U.S. mesh size). Particles larger than 90 µm were processed again for 10 cycles. Particles were collected and stored for one night under vacuum at room temperature to remove any residual moisture.

2.3. Preparation of fluorescein/Nile Red loaded MP

Fluorescent dye loaded MP (FFT3-FL/NR) containing fluorescein (2 % w/w) and Nile Red (0.2 % w/w) were prepared using the same method in order to visualize the deposition and localization of MP in microporated skin and the release of their “cargo” – fluorescein – into the surrounding tissue by CLSM.

2.4. Analytical methods

A previously reported reversed-phase HPLC method was adapted and validated for quantification of TA in the in vitro release study [29]. The HPLC apparatus comprised a P680A LPG-4 pump in line with an ASI-100 autosampler, a TCC-100 thermostatted column compartment, and a UV170D detector (Dionex; Voisins LeBretonneux, France). Briefly, the samples were analyzed using a LiChrospher^® 100, RP-18 (4 x 250 mm and 5 µm particle size) column (BGB Analytik AG; Boeckten, Switzerland) with acetonitrile:water (60:40, v/v) as mobile phase (flow rate: 0.6 mL/min). TA was quantified using a detection wavelength of 266 nm; the retention time was 6.35 ± 0.1 min and the assay was linear (r² = 0.999) in the concentration range of 0.5 – 50 µg/mL. The specificity of the analytical method was tested in the presence of skin matrix compounds. An isocratic UHPLC-MS/MS method was developed and validated to determine the biodistribution of TA, that is, the amounts of TA deposited as a function of position in the different skin layers during skin transport studies. A Waters Acquity^® UHPLC^® core system (Baden-Dättwil, Switzerland) comprising a binary solvent pump and sample manager with a Waters XEVO^® TQ-MS tandem quadrupole detector

137 UHPLC-MS/MS analysis together with a Waters XBridge™ BEH C18 column (2.1 x 50 mm and 2.5 µm particle size). The mobile phase comprised Milli-Q^® water (0.1% formic acid):acetonitrile (55:45, v/v) maintained at 30 °C with a flow rate of 0.3 mL/min. The injection volume was 5 µL. The Waters XEVO^® TQ-MS detector was operated in positive ion mode using multiple reaction monitoring. The mass transition ion-pair was m/z 435.277

→415.154 (removal of hydrogen fluoride). MS source parameters were as follows: ion spray capillary voltage, 1.8 kV; cone voltage, 19 V; desolvation gas temperature, 500 °C; cone gas flow rate, 1000 L/h; and collision energy, 2 V. Data acquisition was carried out using MassLynx V4.1 software.

2.5. Characterization of MP

TA content in the MP was determined by dissolving approximately 5 mg of MP in 1 mL of dichloromethane and diluting 25 times with methanol before HPLC analysis. The drug loading and encapsulation efficiencies of MP were calculated using eqs 1 and 2, respectively.

Drug loading (%) =mass of drug in microparticles

mass of microparticles x 100 (1)

Encapsulation efficiency (%) =experimental drug loading

theoretical drug loading x 100 (2)

To measure MP size and size distribution, approximately 15 mg of MP were suspended in 7 mL of 1% aqueous PVA solution and analyzed using a Mastersizer S (Malvern Instruments Ltd; Malvern, UK). Surface morphology of the MP prepared was studied using scanning electron microscopy (SEM) (JEOL JSM-7001FA, JEOL USA, Inc.; CA, U.S.A.) after mounting the MP on a conductive carbon surface and coating with a thin layer of gold (~21.1 nm) using a high vacuum sputter coater (Leica EM SCD500, Anatech USA; CA, U.S.A.) prior to microscopy. Powder X-ray diffraction (PXRD) patterns were measured using an Agilent Supernova Diffractometer (CA) and Cu radiation by loading the samples in a 100 µm MiTeGen cryoloop using fomblin oil. A rotation of 180° was made during 6 min. The pattern was then integrated using the CrysAlisPro Agilent software. MP phase transitions were analyzed by differential scanning calorimetry (DSC) (SSC/5200, Seiko Instruments; Cheshire, UK). Approximately 5 mg of each MP was weighed in an aluminum pan and the thermograph was recorded over a heating range of 30°C to 320°C at a heating rate of 10 °C/min.

138 Phosphate buffered saline (PBS; pH 7.4) containing 1% Tween 80 as solubility enhancer was selected as the dissolution medium for the in vitro release study. A TA suspension, with a composition similar to the Kenalog^® -40 injection (Bristol-Myers Squibb), was prepared as a control formulation. FFT1-TA10 and FFT2-TA20 (5 mg; – containing 0.5 and 1 mg TA, respectively) and 12.5 µL of TA suspension (TA, 0.5 mg) were dispersed in 50 mL of dissolution medium. All samples were kept in a shaker bath maintained at 34.5 ± 1 °C and 300 rpm. Aliquots (1 mL) were withdrawn at predetermined time intervals and centrifuged (Eppendorf Centrifuge 5804; Germany) at 10621 x g for 5 min. An aliquot (0.5 mL) of supernatant was taken and analyzed for TA content using HPLC while the remaining 0.5 mL was diluted with a fresh 0.5 mL of dissolution media and vortexed to resuspend the MP and transferred back to the dissolution container. All samples were analyzed in triplicate. The changes in the physical properties of the MP after the in vitro release study were also monitored. For this, the MP incubated in release medium for one week were collected by centrifugation (1620 x g, 5 min) and dried under vacuum overnight. These MP were then analyzed using PXRD, DSC and SEM as described above.

2.7. Preparation of skin and P.L.E.A.S.E.^® microporation

Porcine ears were obtained shortly after sacrifice from a local abattoir (CARRE; Rolle, Switzerland). After cleaning with running cold water, the skin from the outer region of ears was carefully excised from the underlying cartilage using a scalpel. The excised full thickness skin samples (2 mm) were then punched into 32 mm circular discs and then stored at -20 °C for no more than 1 month before use. Frozen skin samples were thawed and equilibrated in 0.9 % NaCl solution for 30 min before microporation using an Er:YAG laser (P.L.E.A.S.E.^® , Pantec Biosolutions AG; Ruggell, Liechtenstein). Skin surface moisture was removed and then the samples were mounted on a custom designed assembly. Microporation parameters were set to provide 300 pores/cm² (15% pore density) at a fluence of 90 J/cm.

2.8. MP deposition in P.L.E.A.S.E.^® porated skin and CLSM

The P.L.E.A.S.E.^® porated skin samples were mounted in Franz diffusion cells (area 2.0 ± 0.1 cm²) and silicone grease was applied at the edges to ensure a water tight seal. The receiver compartments were filled with release medium (10 mL PBS at pH 7.4 with 1% Tween 80) and skin was equilibrated for 30 min. For the confocal microscopy experiments, 2.5 mg of FFT3-FL/NR was suspended in 200 µL of Milli-Q^® water and applied to the external surface

139 experiments to study MP deposition and release of fluorescein in the skin were performed for two formulation application times of 30 min and 48 h. After completion of the experiments, the diffusion cells were disassembled and the skin samples were gently dried with a paper towel. The microporated area of the skin was dissected and snap-frozen in 2-methylbutane cooled with liquid nitrogen at -160 °C. Then the skin was sliced into 40 µm thick sections using a cryotome (Microm HM 560 Cryostat; Walldorf, Germany). The sections were then fixed in 4% paraformaldehyde and counterstained with Hoechst blue 33258 to visualize nuclei. Finally, the stained tissue sections were visualized under a confocal laser scanning microscope (LSM 700, Zeiss; Germany); Nile Red enabled localization of the MP and the release and diffusion of fluorescein from the MP was monitored using its characteristic green fluorescence. The confocal images were analyzed using Zen software (Carl Zeiss, Germany) and processed using Image J 1.45s software. The microporated area was also visualized under an optical microscope to examine MP deposition.

2.9. Biodistribution of TA in the P.L.E.A.S.E.^® porated skin

The same experimental set-up was also used to study the biodistribution of TA. A solution containing 2.5 mg of FFT1-TA10 (corresponding to 0.25 mg of TA) suspended in 200 µL of Milli-Q^® water was applied to the microporated skin for 48 h. The receiver phase was maintained at 34.5 ± 1 °C. Cutaneous delivery was compared with that from a TA suspension (200 µL, 0.25 mg of TA). After completion of the study, skin biodistribution of TA released from MP was investigated as a function of depth by quantifying the amount of TA present in five lamellae each with a thickness of 100 µm going from the skin surface to a nominal depth of 500 µm. TA was also extracted from the remaining dermis. TA from each cryotomed skin lamella was extracted using 4 mL methanol:water (1:1) mixture. The amount of TA diffused from the skin into the receiver compartment after 48 h was also determined. TA biodistribution study samples were analyzed by the UHPLC-MS/MS method.

2.10. Statistical analysis.

Data were expressed as mean ± SD. Outliers determined using the Grubbs test were discarded. Results were evaluated statistically using analysis of variance (ANOVA followed by Student–Newman–Keuls test) or Student t-test. The level of significance was fixed at α=0.05.

140 3.1. Characterization of MP

The size distribution and drug encapsulation efficiencies of MP are shown in Table 1. The polymers were selected on the basis of their degradation behavior and their ability to control drug release. PLGA polymers with a carboxylic acid end group, e.g. RG 503H, generally provide faster release in comparison to PLA polymers, e.g. R207. This can be further prolonged when the end group is capped with an ester function as in R207 which can result in extended release for periods of up to several months. Therefore, it was decided to also investigate a mixture of R207 and RG 503H (1:1) in order to achieve an intermediate release profile.

It is known that the size of MP prepared by the o/w emulsion technique is a function of several variables including polymer concentration [30], solvent, drug content, stabilizer molecular weight [31], stabilizer concentration, and solvent evaporation rate [29]. In contrast, controlling the size of MP prepared by the FFT is more straightforward. The drug-polymer matrix is processed in a cryomill until the desired MP size is achieved and only the number of milling cycles needs to be optimized at a given set of operating parameters such as frequency and impact duration along with the cool time to harden the polymer-drug mixture between each cycle. Furthermore, TA-MP prepared by the FFT, FFT-TA10 and FFT-TA20, had encapsulation efficiencies of effectively 100 % (99.9 ± 1.7 % and 101.6 ± 2.1 %, respectively). In contrast, TA-MP prepared by the o/w emulsion technique, OW1-TA10 (with RG 503H) and OW2-TA10 (with R 207), showed much lower encapsulation efficiencies of 5.4

± 0.3 % and 6.8 ± 0.2 %, respectively.

SEM images showed that OW1-TA10 and OW2-TA10 were spherical, whereas FFT-TA10 and FFT-TA20 were irregularly shaped (Figure 1). In the case of the o/w emulsion technique, TA, which is practically insoluble in water, precipitated in the aqueous phase and crystals were observed together with the TA-MP in the SEM images (Figure 1A,b and 1B,b). This effect was probably due to transient diffusion of dichloromethane into the aqueous phase, which caused a temporary increase in TA solubility. As the dichloromethane slowly left the aqueous phase, TA precipitated out as its solubility decreased [32]. Therefore, the final two washings were done with ethanol/water mixture (25/75) to dissolve and so remove the precipitated TA.

141 Figure 1. SEM images of (A) PLGA (RG 503H) microparticles prepared by the oil in water (o/w) emulsion technique: (a) OW1-Placebo, (b) OW1-TA10, before washing with ethanol/water (25/75), (c) OW1-TA10, after washing with ethanol/water (25/75). (B) PLA (R 207) microparticles prepared by the o/w emulsion technique: (a) Placebo, (b) OW2-TA10, before washing with ethanol/water (25/75), and (c) OW2-TA10, after washing with ethanol/water (25/75). (C) PLGA microparticles prepared by the freeze fracture technique: (a) FFT1-TA10, (b) FFT1-TA10, individual microparticle before release study, (c) FFT1-TA10, individual microparticle after release study. (D) PLGA/PLA microparticles prepared by the freeze fracture technique: (a) FFT2-TA20, (b) FFT2-TA20, individual microparticle before release study, and (c) FFT2-TA20, individual microparticle after release study.

142 solid solution where drug and polymer(s) are co-dissolved in an organic solvent, which is subsequently removed to form a drug-polymer solid matrix. The FFT is based on the concept of ‘cryogenic hardening’, which involves cooling a substance to cryogenic temperatures (CT,

< −150 °C). Decreased molecular mobility of the polymer matrix under these conditions results in a decrease in the fracture resistance of the polymers [33]. Therefore, when impact forces are applied to the polymer it is less able to undergo plastic deformation and low intensity forces are sufficient to induce fracturing. The SPEX^® 6770 Freezer/Mill^® used in the present study for MP preparation is a small cryogenic mill that is specially designed for cryogenic grinding of tough and/or temperature sensitive materials.

The crystalline states of TA, RG 503H, R207/RG 503H mixture, TA-polymer physical mixtures, and FFT1-TA10 and FFT2-TA20 were assessed by PXRD (Figure 2A). Crystal diffraction peaks were clearly visible in the diffraction pattern of pure TA (Figure 2A,a) and they were also observed in the physical mixtures of TA-RG 503H and TA-R207/RG 503H (Figure 2A,d and 2A,e, respectively). However, they were absent in the FFT1-TA10 and FFT2-TA20 samples (Figure 2A,f and 2A,g, respectively) confirming that TA had been completely transformed into an amorphous form in these TA-MP and that no drug was present on the MP surface.

DSC thermograms were also recorded for TA, RG 503H, R207/RG 503H mixture, TA-polymer physical mixtures, and FFT1-TA10 and FFT2-TA20 in order to assess the physical state of TA in the TA-MP. As shown in Figure 2B, the endothermic peak of pure TA was at 271.5 °C which corresponds to its melting point. The glass transition points for RG 503H and the R207/RG 503H mixture were 53.2 °C and 60.2 °C, respectively. The DSC thermograms of the physical mixtures of TA with RG 503H and R207/RG 503H also exhibited the TA melting peak but instead of a sharp endotherm, a broad peak was observed. These broad peaks were found in the range from 260 °C to >300 °C due to solid state interactions between TA and the polymer upon heating [34]. The melting endotherm of TA was shifted to a lower temperature in FFT2-TA20 (241 °C) but no TA melting endotherm was seen in FFT1-TA10. The absence of a TA melting transition (Tm) in FFT1-TA10 was due to the complete dissolution of the 10% TA load inside the MP formulation. The TA content in FFT2-TA20

was twice as high (i.e. 20 %) and given that TA has a limited solubility in R207 and that the polymer also has a higher molecular weight and is more viscous (MW 199.8 kDa and intrinsic

143 underwent a melting transition.

Figure 2. Physical characterization of MP. (A) Powder X-ray diffraction patterns of (a) triamcinolone acetonide (TA), (b) RG 503H polymer, (c) R207/RG 503H polymer mixture (50:50), (d) TA-RG 503H physical mixture, (e) TA-R207/RG 503H physical mixture, (f) FFT1-TA10, individual microparticle before the release study, (g) FFT2-TA20, individual microparticle before the release study, (h) FFT1-TA10, individual microparticle after the release study, (i) FFT2-TA20, individual microparticle after the release study. (B) Differential scanning calorimetry thermograms of (a) triamcinolone acetonide (TA), RG 503H polymer, TA-RG 503H physical mixture, FFT1-TA10 before the release study, and FFT1-TA10 after the release study; (b) triamcinolone acetonide (TA), R207/RG 503H (50:50) polymer mixture, ,

144 after the release study.

3.2. In vitro drug release

Complete release of TA from the suspension formulation, which was similar in composition to Kenalog^® -40 injection, was achieved in 90 min (Figure 3). MP formulations prepared by FFT displayed sustained release profiles; for FFT1-TA10, 91.83 ± 1.72 % was released after 7 days, and in the case of FFT2-TA20, 50.33 ± 2.24 % was released after 14 days (Figure 3).

The differences between the two polymers were due to differences in the inherent viscosity, molecular weight and the nature of the end-groups. Polymer matrices with free carboxyl groups such as PLGA (RG 503H) undergo faster water absorption, hydrolysis and erosion than end-capped polymers with an ester termination (e.g. PLA (R207)) [35]. Therefore, for FFT1-TA10 a tri-phasic release profile was observed. First, hydrophobic TA diffused through the external surface of the polymeric MP immediately after coming in contact with release media; this gave an initial release of 8.24 ± 0.62 % within the first 2 h. During the second phase of the FFT1-TA10 tri-phasic release profile, constant release was observed between days 1 to 4. This second phase involved formation of small water-filled pores and hydrolytic degradation that led to the development of a porous connected network inside the MP matrix and ensured rapid TA release. The third phase corresponded to TA release due to surface erosion of the MP and diffusion of TA from the MP core. Since water absorption was slower in FFT2-TA20 because of its more hydrophobic polymer mixture, only 2.65 ± 0.14 % TA diffused through the exposed external region of MP in 2 h followed by constant, slow release up to the end of the study period of 14 days.

MP degradation was visualized by SEM analysis of MP recovered upon completion of the release study (Figure 1C,c and 1D,c). For FFT1-TA10, the integrity was markedly impaired as several “pores” were observed on the surface possibly contributing to the observed faster degradation of polymer and release of the drug (Figure 1C,c). In contrast, the integrity of FFT2-TA20 was intact and TA crystals were found in large numbers on their surface because of slow diffusion and the high TA content in the particles (Figure 1D,c). The slower degradation of the FFT2-TA20 formulation was correlated to the higher molecular weight and lower porosity of the polymer matrix due to R207 [36]. An inverse relation between drug release and polymer molecular weight or viscosity has also been reported [37]. Increasing

145 drug release process [30,38].

Figure 3. Release profiles of triamcinolone acetonide (TA) from the microparticle formulations, FFT1-TA10, FFT2-TA20 and a drug suspension during (A) 12 h and (B) 14 days.

Data represent mean values of three replicates ± standard deviation.

The in vitro drug release studies showed that the TA-MP provided sustained release under sink conditions but these obviously do not reflect the microenvironment of the skin.

According to some studies, vascular density in keloid scars is less than in hypertrophic scars and of course healthy skin [39,40]. It was reported that keloid scars lack microvascular connections and suffer from inadequate blood supply because of excessive collagen growth.

Deficient vascularization might prove to be an advantage for sustained drug release from MP since, in the absence of blood capillaries, drug levels can be maintained much longer resulting in a more prolonged local action.

0 20 40 60 80 100

0 2 4 6 8 10 12 14

TA Release (%)

Time (days)

TA Suspension FFT1-TA10 FFT2-TA20 0

20 40 60 80 100

0 2 4 6 8 10 12

TA release (%)

Time (h)

A

B

146 TA in both FFT formulations suggesting that TA released from the TA-MP during the release study crystallized on the surface of the particles (Figure 2A,h and 2A,i). This was consistent with the SEM images (Figure 1C,c and 1D,c). By the end of the release study, TA precipitated on the surface of both MP formulations and gave rise to small Tm peaks at

~231.0 °C together with the pure TA Tm at 271.5 °C (Figure 2B). During the release study TA was released into the dissolution media and the ratio of TA to polymer in the formulation was reduced. The decrease in Tm depends on the ratio of TA to the external phase present (i.e. the polymer matrix), thus the Tm is reduced to even lower temperature after the release study [41]. These observations were in agreement with a published report where the

~231.0 °C together with the pure TA Tm at 271.5 °C (Figure 2B). During the release study TA was released into the dissolution media and the ratio of TA to polymer in the formulation was reduced. The decrease in Tm depends on the ratio of TA to the external phase present (i.e. the polymer matrix), thus the Tm is reduced to even lower temperature after the release study [41]. These observations were in agreement with a published report where the