A novel versatile flow-donor chamber as biorelevant ex-vivo test assessing oral mucoadhesive formulations

Oral transmucosal drug delivery is a non-invasive administration route for rapid therapeutic onset and greater bioavailability avoiding the first-pass metabolism. Mucoadhesive formulations are advantageous as they may retain the drug at the administration site. Proper equipment to assess mucoadhesive properties and corresponding drug absorption is fundamental for the development of novel drug delivery systems. Here we developed a new flow-through donor chamber for well-established diffusion cells, and we tested the effects on drug and formulation retention in situ of adding mucoadhesive polymers or mesoporous silica particles to a reference formulation. Mesoporous silica particles are of particular interest as they may be used to encapsulate and retain drug molecules. Compared to other ex-vivo methods described in literature for assessing mucoadhesive performance and transmucosal drug delivery, this new donor chamber provides several advantages: i) it reflects physiological conditions better as a realistic saliva flow can be provided over the administration site, ii) it is versatile since it can be mounted on any kind of vertical diffusion cell allowing simultaneous detection of drug retention at the administration site and drug permeation through the tissue, and iii) it enables optical quantification of formulations residence time aided by image processing. This new flow-through donor diffusion cell set-up proved sensitive to differentiate a reference formulation from one where 20 %(w/w) Carbomer was added (to further improve the mucoadhesive properties), with respect to both drug and formulation residence times. We also found that mesoporous silica particles, investigated as particles only and mixed together with the reference formulation, gave very similar drug and formulation retention to what we observed with the mucoadhesive Carbomer. However, after some time ( > 30 min) it became obvious that the tablet excipients in the reference formulation promote particle retention on the mucosa. This work provides a new simple and versatile biorelevant test for the evaluation of oral mucoadhesive formulations and paves the way for further studies on mesoporous silica particles as valuable excipients for enhancing oral mucoadhesion.


Introduction
Oral transmucosal administration such as sublingual and buccal (through cheek) offers a non-invasive, cost effective and patient-friendly way to administer active pharmaceutical ingredients (API) into the body.The API is directly absorbed through the non-keratinized mucosa, particularly under the tongue which is vascularized with the superior vena cava.Indeed, the key advantages of transmucosal administration in pharmaceutical treatment include: i) rapid API absorption where the fast onset of the drug action is critical; ii) avoiding the acidic and highly enzymatically active gastrointestinal environment that can degrade the API; and iii) bypassing the first-pass effect that can metabolize the API, creating toxic intermediates and inducing drug-drug interactions (Elsner et al., 2011;Lyseng-Williamson, 2013;Yang and Deeks, 2012).Besides these benefits, several challenges need to be considered.The surface area of the non-keratinized epithelia in the mouth (i.e.sublingual and cheeks) is small (100 cm 2 ) and the dissolution volume is low (0.8 mL, the residual volume of the saliva) (Patel et al., 2011).The constant swallowing of the saliva also results in a limited residence time of the drug at the absorption site (Perioli and Pagano, 2013).Therefore, the API should be rapidly released and dissolved in the mouth, so that absorption takes place before inevitable swallowing.To overcome this limitation, mucoadhesive components have been developed to increase drug retention time at the mucosal surface.Mucoadhesion is a complex phenomenon that results from a combination of several mechanisms.Firstly, wetting takes place, swelling the dosage form, after which noncovalent bonds are created with the mucus (electrostatic and adsorption processes) and finally the ingredients interpenetrate in the mucus gel (diffusion process) resulting in formation of an interpenetration layer (Khutoryanskiy, 2011).Mucoadhesive agents often form viscous gels that increase retention at the oral surface (Han et al., 1999;Khutoryanskiy, 2011;Lee et al., 2000;Perioli et al., 2004;Sudhakar et al., 2006).In a recent report from our group we show that water sorption by the formulation should also be considered a factor that influences buccal adhesion (Ali et al., 2018).Numerous alternative polymers have been used to improve mucoadhesion, and for a more thorough description of the use of such polymers in transmucosal delivery the reader is referred to e.g., the extensive reviews by Montenegro-Nicolini and Morales, Alaei and Omidian (Alaei and Omidian, 2021;Montenegro-Nicolini and Morales, 2017).Among the most widely studied mucoadhesive excipients are the high molecular weight cross-linked poly(acrylic acid) polymers called carbomers, which form viscous hydrogels at pH above 4-6.(Mastropietro et al., 2017).Additionally, nano-and micro-particles have been shown to increase the retention time and absorption of drugs across gastrointestinal mucosa (Hua, 2015;Hua et al., 2015).In this context, mesoporous silica particles (MSP) have emerged as a promising material with unique specific properties that may be utilized in pharmaceutical products as a functional excipient to enhance drug bioavailability (Li et al., 2019).MSP are biocompatible particles composed of amorphous silicon dioxide (SiO 2 ) that contain thousands of pores within the "meso" range (i.e., 2 -50 nm) offering vast surface areas (up to 1000 m 2 /g) and large pore volumes (up to 1.5 cm 3 /g) and allowing high loading (up to 50% w/w of drug content).Moreover, MSP consist of amorphous silicon dioxide which has GRAS status (Generally Regarded As Safe) from the FDA as an excipient for oral pharmaceutical dosages.Previous studies from our group have demonstrated that encapsulation and confinement of poorly soluble drugs inside MSP can enhance the dissolution kinetics and subsequent bioavailability after oral administration (Angiolini et al., 2018;Stjern et al., 2017;Valetti et al., 2017c).Indeed, the enhanced drug dissolution kinetics offer the potential to increase the portion of dissolved drug in the oral cavity prior to saliva swallowing.Some studies have demonstrated improved mucoadhesive properties of silica nanoparticles through chemical functionalisation of their surfaces (M Ways et al., 2020).In particular, mucus-penetrating particles were developed with PEGylation and similar polymers (Desai et al., 2016;Mansfield et al., 2015).It has furthermore been shown that the penetration of silica microparticles across a mucus layer is reduced increasing the particle size (Sotres et al., 2017).Alongside the development of mucoadhesive drug carriers, proper equipment to assess mucoadhesive properties is fundamental for the development of novel drug delivery systems.Numerous tests (in-vivo, ex-vivo, and in-vitro) are described in literature for assessing the mucoadhesive performance of drug delivery systems (Alaei and Omidian, 2021).In-vivo tests are expensive and therefore not suitable for early discovery phase.Ex-vivo and in-vitro methods include tensile and rheological studies, visual detachment, dissolution rate measurements and flow retention techniques (Hoffmann and Daniels, 2018;Khutoryanskiy, 2011).In a typical experiment, the test material is applied on a mucosa anchored to a surface and subjected to simulated saliva flow.Despite a variety of systems being described in literature, it is difficult to find examples where temperature and simulated biological fluid (content, volume and flow rate of saliva) mimic the physiological conditions.Recently, Hoffmann et al. reported a novel flow through method considering biorelevant conditions such as saliva flux and composition (Hoffmann and Daniels, 2018).However, this method does not consider or allow for quantification of the permeated drug.Diffusion cells, such as Franz or flow-through cells, are recognized from international guidelines (OECD, 2011;WHO, 2006) as a standardized method to evaluate permeation of drugs across biological tissue.To the best of our knowledge, there is no report showing how to improve these standard ex-vivo release/permeation methods for studying the effect of formulation retention and mucoadhesion in buccal or sublingual drug delivery.
In this context, we aimed to improve the established test methods by developing a new donor chamber that can be mounted on any diffusion cell, enabling us to mimic physiological conditions and to quantify residence time of both formulation and API, including drug mass balance thereof.The suitability of this setup was assessed using a set of test formulations based on a conventional mixture of excipients used for sublingual tablets of (Park and Munday, 2002;Zhou, 2019).A Carbomer (i.e., Carbopol 980) was added to further improve the mucoadhesive properties in one formulation, and providing a positive control for evaluating the effect of MSP on formulations residence time on the mucosa.Eletriptan hydrobromide is used for treatment of migraine and it was chosen as a relevant API.In fact, oral transmucosal administration, and particularly the sublingual route, is of high relevance for migraine management as it has high potential to decrease onset time and improve drug bioavailability.

Materials
d-Mannitol (Mannite), microcrystalline cellulose (MCC, Avicel PH-113 ~ Ø 50 µm) and magnesium stearate (Mg stearate) were purchased from Sigma Aldrich AB (Stockholm, Sweden).Carbomer (CAR, Carbopol 980 NF), hydroxypropyl cellulose (HPC, Klucel HF Pharm) and phosphate buffered saline (PBS, 137 mM NaCl, 2.7 KCl and 10 mM phosphate buffer pH 7.4, E404-100TABS) were sourced from VWR chemicals (Spånga, Sweden).Methanol, acetonitrile and formic acid (>99%) was purchased from VWR chemicals (Spånga, Sweden).Eletriptan hydrobromide was kindly provided by Orexo AB (Uppsala, Sweden).Mesoporous silica particles (MSP) and mesoporous silica particles loaded with eletriptan hydrobromide were provided by Nanologica AB (Södertälje, Sweden).Details about the MSP properties, synthesis and drug loading procedure have been published previously (Valetti et al., 2021;).Briefly, MSP were synthesized in an emulsion based, sol-gel process using a silica precursor to produce spherical particles in the 5-micron size range.The porosity in these particles occurs during the condensation and drying of the silica gel which yields a broader pore size compared to the templated synthesis route with an average pore size of 100 Å.After calcination step that removed organic residues (600 C for 6 h) we obtained particles with a mean diameter of 7 µm, and monodisperse size distribution D90/D10=2.3(based on Elzone analysis).

Collection of saliva from human volunteers
Saliva was collected and kindly provided by CTC Clinical Trial Consultants AB (Uppsala, Sweden).The donated saliva was unstimulated saliva collected from de-identified volunteers.The volunteers rinsed the mouth with water and waited for 10 minutes before starting to collect the saliva.They were instructed to let the saliva flow naturally to the front of the mouth and drip into a plastic tube for a period of approximately 10 minutes.The saliva was kept on ice, and centrifugated at 4 • C at 2000 g (2500 g if too viscous) for 15 minutes before it was placed in a freezer at -70 • C. To homogenize the saliva lot, at least 3 tubes of saliva collected from different volunteers were pulled together.

Preparation of porcine mucosa
Porcine esophagus, considered as an appropriate and more accessible replacement for buccal tissue (Diaz del Consuelo et al., 2005;Padula et al., 2013), was obtained from a local slaughterhouse (Strömbecks Gårdsslakt & Chark, Sweden) as byproduct in food processing, and kept at 4 • C during transport to the laboratory.The esophagus was separated from trachea and dissected longitudinally using a pair of surgical scissors.Next, the epithelium was separated from underlying tissue using a scalpel.It was gently washed with PBS to remove food debris but leaving the mucus layer on the surface.The tissue was then placed on a filter paper soaked in PBS, wrapped in aluminum foil and stored at -80 • C until use (not more than one month).Each piece was marked with animal and tissue part origin (i.e., close to the mouth, middle, close to the stomach).Prior to starting an experiment, the tissue was thawed and hydrated on PBS for 30 minutes.

The new flow-through donor chamber diffusion set-up
A new flow-through donor chamber for diffusion cells (e.g., Franz cells) was designed in house and custom made by Prototypverkstaden AB (Lund, Sweden).The donor chamber has an inlet and an outlet, making it possible to flush liquid through the chamber and mimic the saliva flow in the mouth (Fig. 1).
In this study we used standard 6 mL mantled Franz cells with two sampling ports, a 9 mm diameter permeation orifice, and magnetic stirring in the receptor chamber (Permegear Inc, USA).PBS pH 7.4 was used as receptor media and the temperature was controlled to 37±0.5 • C to mimic physiological conditions.The new flow-through donor chamber was clamped to the Franz cell receptor chamber with a mucosal membrane in between (23 mm in diameter), dorsal side facing upwards.Saliva or PBS was flushed through the donor chamber connected by Teflon tubes to a peristatic pump (Biolab Minipuls 3, Gilson Inc, USA).The set-up allows us to monitor the complete drug distribution (i.e., the amount of drug flushed away, retained at the mucosal surface, absorbed in the mucosa and permeated across the tissue) and conduct a full drug mass balance calculation.

Ex-vivo release, residence time and mucoadhesion
Each Franz cell was loaded with 10 mg of a test formulation (see section 2.7) corresponding to around 16 mg/cm 2 .The setup was closed by attaching the donor chamber on top of the receptor chamber with a clamp.The donor chamber was then filled with saliva (pH 6.9 ± 1) and the formulation was left to equilibrate for 3 min.Thereafter, the flow rate was set to 0.3 mL/min to simulate a normal physiologically relevant saliva flow (Humphrey and Williamson, 2001).The variations in flow rate together with length and inner diameter of the tubing were evaluated and reported in supplementary data (Table S1).After 15 minutes the saliva was exchanged for PBS (pH 7.4).We believe that the change of solution would not affect the saliva pellicle formed at the mucosa layer since previous reports have shown that saliva pellicles are resistant to washing (Cárdenas et al., 2007;Cárdenas et al., 2008).After 45 min, the flow rate was set to 1.5 mL/min to simulate drinking or food intake.

Optical evaluation of residence time on the mucosa
After each time point, the tissue was removed and the formulation residence time on the mucosa was evaluated by taking images with a microscope digital camera (MU500 USB2.0 DC 5V 250 mA, Amscope, China) during the dissolution experiment, and then analyzing the images with ImageJ (version 2.0.0-rc-69/1.52p2018).The area of interest was completely covered by formulation at start, and the area still covered with formulation after a certain time was calculated based on white resolution.The total % coverage was calculated from covered area vs. area of interest.

Test formulations
The reference placebo test formulation (REF, Table 1) consisted of a conventional mixture of excipients used for sublingual tablets, including microcrystalline cellulose, mannitol, hydroxypropyl methyl cellulose and magnesium stearate (Zhou, 2019).Carbomer (i.e., Carbopol 980) was added to further improve the mucoadhesive properties (Park and Munday, 2002) and provides a positive control for evaluating the effect of mesoporous silica particles (MSP) on residence time on mucosa.These two formulations are referred to as REF+CAR and REF+MSP, respectively.The model drug, eletriptan hydrobromide (EB), was added as free drug or encapsulated in MSP's using a solvent impregnation method previously employed with other drug molecules, such as metronidazole (Stjern et al., 2017).By encapsulation in MSP, EB was stabilized in its amorphous state.Test formulations with EB were prepared with and without carbomer (REF and REF+CAR, respectively) as well as with MSP, where the drug was either encapsulated in the MSP's (REF+MSP (E)) or just added together with the MSP as a physical mixture (REF+MSP(M)) (Table 1).We have previously shown (Valetti et al., 2017c) that the drug does not become encapsulated in the silica particles mixing only.All compositions were blended until homogenous mixtures.

Drug distribution and mass balance
The current set-up allows us to monitor the complete drug distribution (i.e., the amount of drug flushed away with the donor flow, retained at the at the surface, absorbed in the mucosa and permeated across the tissue, respectively) and conduct a full drug mass balance calculation.The drug that was flushed away with the donor flow was collected continuously in 10 minutes intervals during the experiment, and the drug that permeated across the mucosa was monitored by collecting 0.5 mL aliquots with a syringe from the receptor compartment at the same time points.At the end of the experiment the drug remaining on the mucosa was collected by washing the surface with 4 mL PBS.In particular, the drug absorbed in the tissue was extracted by immersing the mucosa in methanol and sonicate for 10 hours.All collected fractions (aliquots from donor and receptor, washing solutions and tissue extract) were vortexed for 3 minutes to dissolve any drug precipitated or encapsulated in the particles, followed by centrifugation (20 minutes at 1500g) to remove debris from the supernatant.The supernatant was then filtered using a 0.2 μm PVDF filter (VWR chemicals, Spånga, Sweden) before HPLC analysis.

HPLC analysis
Eletriptan hydrobromide was analyzed by HPLC/UV-vis using an Agilent 1100 spectrometer (Agilent Technologies, Australia) with a SVEA HPLC column (4.6*100 mm, 3.5 μm; Nanologica AB) and a Phenomenex Security Guard precolumn (Cartridges C18 4 * 3.0 mm ID; AJO-4287 Phenomenex, Sweden).The flow rate of the mobile phases was 0.8 mL/min and the injection volume was 100 μL.Mobile phase A comprised 0.1% formic acid in water, while mobile phase B comprised 0.1% formic acid in water: acetonitrile (5:95 v/v).Mobile phases were degassed before use.The gradient steps consisted of 0-0 min 15% of B, 0-6 min 95% of B and 7-9 min 15% of B. The detection wavelength was λ=274 nm, and retention time for the drug was 4.5 minutes under these conditions.A calibration curve was performed by preparing standard solutions in the range from 3.7-500 μg/mL in the same medium that was used during the experiments.LOD and LOQ were 0.872 and 2.643 μg/ mL calculated with the standard deviation of the response and the slope according to the International Conference of Harmonization (ICH) (ICH, 2005).

Statistical analysis
The data were analyzed using Excel (Microsoft Office 2018) and DataGraph version 4.5.1.For all pair-wise comparison of means, Student's independent t-test was performed with significance levels of p<0.05 or p<0.01.

Formulations residence time on the mucosa
Formulations residence time on mucosa was evaluated in the new flow-through donor diffusion cell set-up using the test formulations listed in Table 1.The reference placebo test formulation (REF), reflecting a conventional recipe (Zhou, 2019), was compared to one where 20% (w/w) Carbomer (CAR) was added to further improve the mucoadhesive properties (REF+CAR) (Park and Munday, 2002) and serve as positive control for evaluating the effect on residence time.Fig. 2A shows that 68% of the REF+CAR formulation was retained at the administration site (i.e., on the mucosa surface) after 5 minutes with a saliva flow of 0.3 mL/min, which was significantly higher than the 38% obtained with the REF formulation (p<0.01).The amount of REF+CAR formulation residing at the mucosa was still significantly higher after 30 minutes at the same flow conditions (p<0.01).After 45 minutes the donor flow rate was increased to 1.5 mL/min to simulate e.g. the effect of drinking, which resulted in increased wash-off of both formulations from the mucosa.
The mucoadhesive properties of MSP were investigated with particles only (MSP 100%) and with MSP formulated in REF, i.e., as REF+MSP (Table 1).The results obtained with both MSP 100% and REF+MSP (Fig. 2B) are very similar to what we observed with REF+CAR up to 30 minutes (Fig. 2A) and statistically different from REF on the

Table 1
Compositions of test formulations (expressed as mass and in % w/w).The reference placebo test formulation (REF), which consisted of a conventional mixture of excipients used for sublingual tablets (Zhou, 2019), was further modified by adding either Carbomer, to improve mucoadhesive properties, or MSP.Eletriptan hydrobromide (EB) was blended with the excipients, except where the drug was first encapsulated in amorphous state in the MSP (i.e., REF+MSP(E)).p<0.05 level.However, after 30 minutes it becomes obvious that tablet excipients promote retention of MSP on the mucosa since REF+MSP show a higher retention time than MSP 100% (58% vs 37%).

Drug distribution and mass balance
Drug distribution and mass balance analysis were made comprising the total amount of drug found in four different compartments: i) flushed away (i.e., swallowed) with the saliva flow (flow-through donor fraction), ii) retained on the on top of the mucosa (surface fraction); iii) absorbed into the tissue (extracted from the membrane); iv) permeated through the mucosa (receptor fraction).The donor chamber has to be a closed compartment to allow saliva to flow through, and in our first design we had a fixed cover glass to seal the top.The loading of the drug/formulation then evidently had to be performed before attaching the donor chamber to the rest of the cell.On testing the donor chamber with fixed cover glass we obtained a total mass recovery of about 51% (Fig. 3A), which is not satisfactory with respect to the OECD recommendations of 100 ± 20% (OECD, 2011).This could e.g., be due difficulties to load the formulation on the permeation area only, and/or unintentional sliding of the mucosa during assembly of the chambers.
Drug mass balance chart from an experiment with formulation REF+MSP(M) after 30 minutes using the flow-through donor chamber with A) fixed cover glass or B) removable cover glass.Drug permeation was negligible in this experiment.Data are presented as mean ± SD (n=4).
Therefore, we modified the design and introduced a cover glass that could be removed.This version allows loading the sample after the donor and receptor compartments are clamped together, reducing the risk that the formulation is being placed outside the area of application.
Here we obtained a total drug recovery of 86%, which is in agreement with the OECD guidelines (Fig. 3B).
After optimizing the flow-through donor chamber, we tested if alternative formulations (i.e., REF, REF+CAR, REF+MSP(M) and REF+MSP(E), Table 1) had different drug residence time when exposed to saliva flow.In the REF+MSP(E) formulation, the encapsulated eletriptan hydrobromide is rapidly released within the first 5 minutes (supplementary data -Fig.S2).This immediate drug release is in line with previous reports where small and more hydrophilic drug, metronidazole, was rapidly released from the MSPs (Valetti et al., 2021).Fig. 4 shows the result obtained after 10 minutes.Evidently addition of Carbopol to the reference formulation (REF) had a statistically significant effect (p<0.05) with 92% drug retained for REF+CAR, compared to 61% for REF.Both MSP formulations (i.e., drug encapsulated in MSP (REF+MSP(E)) or mixed with MSP (REF+MSP(M), Table 1) also showed an increased drug residence time compared to REF (70% and 84%, respectively), with a statistical significant higher retention in case of REF+MSP(M).To avoid data misinterpretation due to poor drug recovery, the threshold for the data used in Fig. 4 was set to 95% recovery.Nevertheless, when including all samples that complied with OECD recommendations of a total mass recovery of 100 ± 20%, the same trends were found (see supplementary data -Fig.S3).Here REF+MSP (E) also provided a statistically significant higher drug retention compared to REF (p<0.05).As process control, we found that having saliva or just PBS as flowing medium does not influence the drug distribution obtained with a formulation comprising eletriptan encapsulated in MSP (supplementary data -Fig.S4).The set-up allowed also the simultaneous detection of the drug absorbed and permeated across the tissue although differences between the formulations were not observed in the current study (Fig. 4B).This is probably due to the slow permeation of eletriptan hydrobromide, which is charged at physiological pH and the short time point.

Discussion
Diffusion cells are frequently used in both research and development applications to study drug uptake over time through biological tissue, such as mucosa or skin.Normally, the donor formulation is placed on top of the tissue and kept at static conditions during the whole experiment.However, in oral applications saliva flow may have a large impact on the duration of the formulation at the mucosa and thus the eventual drug bioavailability.To include this aspect ex vivo and better mimic the physiological situation we designed a new donor chamber with two ports which allowed for saliva or buffer to flow over a formulation applied on the mucosa (Fig. 1).The final design of this donor cell also has a transparent lid which can be opened separately.Thus, the test formulation can be applied after assembling the equipment, and the effects of saliva flow on formulation coverage can be quantified over time from the top with a camera and subsequent image analysis.The method is particularly relevant for estimating mucoadhesive properties of a test formulation designated for sites which are highly affected by liquid flow, that can wash off the formulation, such as in sublingual administration (Khutoryanskiy, 2011).Method suitability was evaluated by comparing i) the residence time on the mucosa of a reference test formulation (REF) with one that also contained 20% Carbomer (REF+CAR) known to facilitate mucoadhesion, and ii) the drug distribution in the system obtained with the alternative formulations (i.e., flushed away, retained at the surface, absorbed in the mucosa and permeated through the tissue).Being aware that for an immediate drug release often desired in sublingual drug delivery, dissolution and permeation will happen in the first 10 minutes, we investigated the residence on the mucosa over a rather longer period of time (45 minutes) to offer a model system that consider also sustained release approach (Alaei and Omidian, 2021).After 45 minutes the donor flow rate was Fig. 3. Optimization of the new donor chamber to fulfill mass balance according OECD specifications.Drug mass balance chart from an experiment with formulation REF+MSP(M) after 30 minutes using the flow-through donor chamber with A) fixed cover glass or B) removable cover glass.Drug permeation was negligible in this experiment.Data are presented as mean ± SD (n=4).Fig. 4. Effects of formulation excipients on drug distribution after 10 minutes with a saliva flow of 0.3 mL/min.Test formulations with eletriptan hydrobromide as model drug (Table 1), were prepared with and without carbomer (REF (white bar) and REF+CAR (black bar), respectively) as well as with MSP, where the drug was either encapsulated in the MSP (REF+MSP(E) (dark gray bar)) or just added together with the MSP as a physical mixture (REF+MSP(M) (light grey bar)).For all the samples minimum recovery (mass balance) was set to 95%.Data are presented as means ± SEMs, n=3-4.Statistically significant difference is marked by *; p< 0.05 (t-student).
increased to simulate e.g. the effect of drinking.We found a statistically significant increase (p<0.01) in formulation retention at the mucosa over 30 minutes with a flow of 0.3 mL/min from including Carbomer (Fig. 2A), an effect that may be attributed both to mucoadhesive properties of the Carbomer and the fact that it may form a hydrogel at saliva pH.Although this method does not discriminate between a monolayer surface coverage or multilayers of various thickness, the system allows for a reliable comparison when the amount of powder is kept constant.
Here we decided on a dose of 15.6 mg/cm 2 , based on a 1.56 g formulation applied over the buccal surface area of the 100 cm 2 or on a 400 mg tablet applied on the sublingual surface area of 25 cm 2 (Kraan et al., 2014).We also found a statistically significant increase (p<0.05) in the amount of drug, Eletriptan hydrobromide, retained at the mucosa after 10 minutes including Carbomer, while no difference could be detected in the amount absorbed in the tissue from the two formulations.Nonetheless, for a longer timepoint we do expect that a prolonged retention time at the mucosa surface would promote build-up of a depo in the tissue, which eventually can increase the permeated drug amount even after all the formulation has been cleared from the administration site.A smaller receiver chamber would allow for a better detection of low permeants such as eletriptan hydrobromide, but also potentially cause absence of sink conditions for hydrophobic and high permeant molecules.Of particular interest, we also observed a prolonged and similar retention time at the surface when Carbomer was replaced with MSP in the formulation (REF+MSP), statistically different from REF on the p<0.05 level over 30 minutes with a flow of 0.3 mL/min (Fig. 2B).It is well known that macromolecules, such as histatin and mucins comprised in saliva, are able to adsorb to silica surfaces forming a multilayer film (Hyltegren et al., 2016;Lindh et al., 2007;McColl et al., 2007;Svendsen et al., 2007) and adhesion forces between silica surfaces and mucus have previously been described (Sotres et al., 2017), indicating that silica nanoparticles can diffuse in and interact with mucus film lining the mucosa (Andreani et al., 2015).When repeating the same experiment with mesoporous silica particles only (MSP 100%), the results followed those of REF+MSP (Fig. 2B).However, beyond 30 minutes it becomes obvious that tablet excipients promote retention of MSP at the mucosa.This effect is most probably due to the presence of hydroxy propyl cellulose (HPC, Table 1).Although the HPC concentrations used correspond to the ones suggested for oral tablets as binder for granulation (Rowe et al., 2009) it may form a hydrogel on dissolution which may increase retention of the MSP.We also found that the two alternative MSP formulations, where Eletriptan hydrobromide was either encapsulated in MSP (REF+MSP(E)) or simply mixed with MSP (REF+MSP (M), Table 1), generated an increased drug residence time compared to REF (Fig. 4A).The increase was statistically significant for REF+MSP (M) where the ingredients were mixed without encapsulating the drug in the particles.A contributing factor may be that the drug can adsorb to the silica surface through electrostatic interactions with negatively charged silanol groups present at the particle surface at physiological pH.In fact, the MSPs´also have a large and negative electrostatic potential inside the particles, as we showed in a previous work combining a numerical model with titration techniques (Valetti et al., 2017a), that is affecting the drug release profile (Angiolini et al., 2018).In case of REF+MSP(E), particles could be washed-out prior drug release.Although eletriptan hydrobromide is rapidly released within the first 5 minutes in aqueous solution (supplementary data -Fig.S2) some MSP could be swallowed with the saliva prior drug release.Another possible explanation might be that although the drug retention is reported as a relative Fig., the REF+MSP(E) with encapsulated drug comprised less than half the amount of drug compared to what was included in REF+MSP(M).However, also for REF+MSP(E), increase in drug retention at the surface was statistically significant (p<0.05) when the including all samples that complied with OECD recommendations of a total mass recovery of 100 ± 20%, and not with the stricter requirement of 100 ± 5%.Further investigations on the mucoadhesive properties of MSP and differences in formulation strategies comprising MSP are currently in process in our group.

Conclusion
In this study we have made an improvement to well-established diffusion cells by developing a new flow-through donor chamber for better mimicking the physiological situation.Compared to other ex-vivo methods described in literature for assessing mucoadhesive performance and transmucosal drug delivery, this new set-up provides several advantages: i) it reflects physiological conditions better as a realistic saliva flow can be provided over the administration site, ii) it is versatile as the new donor chamber can be mounted on any kind of vertical diffusion cell allowing simultaneous detection of drug retention at the administration site and drug permeation through the tissue, and iii) it enables optical quantification of formulations residence time aided by image processing.During the development we modified the donor chamber design and introduced a cover glass that could be removed.This version allows loading the sample after the donor and receptor compartments are clamped together and had a large impact on meeting the mass balance requirements (100±20%) stipulated by OECD and ICH guidelines.This new flow-through donor diffusion cell set-up further proved sensitive to differentiate a reference test formulation from one where 20 % (w/w) Carbomer was added (to further improve the mucoadhesive properties), with respect to both drug and formulation residence times.We also found that mesoporous silica particles, investigated as particles only or mixed with the reference test formulation, gave very similar results to what we observed with the mucoadhesive polymer Carbomer up to 30 minutes.However, after 30 minutes it became obvious that tablet excipients promote the retention of mesoporous silica particles at mucosa.Even simple mixing (without encapsulation) of a drug with silica particles in conventional formulation could be an interesting strategy for prolonged duration in situ of a readily water-soluble and positively charged drug, like eletriptan hydrobromide.

Fig. 1 .
Fig. 1.The new flow-through donor chamber for diffusion cells, like Franz cells.Numbers are expressed in mm.

Fig. 2 .
Fig. 2. Formulation residence time on administration site (mucosa surface) under flow.A) Reference placebo test formulation without (REF) or with Carbomer (REF+CAR) over time, n=3.B) Reference placebo test formulation without (REF) or with mesoporous silica particles (MSP) or mesoporous silica particles only (MSP 100%) over time, n=3-6.Values are presented as mean ± SD, statistical differences are marked by **; p<0.01, *; p<0.05 (t-student).C) Representative optical images of the formulation at the mucosa for: REF(left), REF+CAR (center) and REF+ MSP (right) after 15 minutes with a saliva flow of 0.3 mL/min.