Dissolving microneedle array patches containing mesoporous silica nanoparticles of different pore sizes as a tunable sustained release platform

Dissolving microneedle array patches (DMAPs) enable efficient and painless delivery of therapeutic molecules across the stratum corneum and into the upper layers of the skin. Furthermore, this delivery strategy can be combined with the sustained release of nanoparticles to enhance the therapeutic potential in a wide variety of pathological scenarios. Among the different types of nanoparticles that can be included in microneedle formulations, mesoporous silica nanoparticles (MSNs) of tuneable pore sizes constitute a promising tool as drug delivery systems for cargos of a wide range of molecular weights. However, the development of efficient methods to produce DMAP containing large amounts of MSNs of different pore sizes has not been reported. In this work, DMAP containing MSNs with varying pore sizes was prepared and characterized. After synthesizing and characterizing MSNs, the pore size of the nanoparticles (in the range of 3 to 13 nm for S-MSN and XL-MSN, respectively) was observed to influence the loading and release of both small and large molecules, using fluorescein and ovalbumin (OVA) as model cargos. Moreover, a new preparation method was developed to produce DMAP containing large amounts of these MSNs located mainly in the microneedle tips. The successful insertion of these DMAPs was confirmed in vitro (using Parafilm), ex vivo (using excised neonatal porcine skin) and in vivo (in the back of mice) models. The dissolution of the microneedles and deposition of the nanoparticles inside the skin were also confirmed both ex vivo and in vivo using fluorescent nanoparticles, with complete microneedle dissolution after 2 h of insertion in vivo. Through histological studies, the microneedle-delivered MSNs were found to end up inside antigen presenting cells in the skin tissue (either F4/80+ macrophages or CD11c+ dendritic cells). For this reason, the uptake and biological effect of the MSNs was evaluated in vitro in dendritic cells, showing that while smaller pore MSNs were taken up by cells more efficiently (with over 80 % of S-MSN uptake compared to ca. 55 % for XL-MSNs), the dendritic cells treated with OVA- loaded XL-MSNs underwent the largest degree of activation (inducing over 25 % of CD40 expression compared to less than 2 % for OVA-loaded S- MSNs). Finally, the immune response to OVA-loaded XL-MSNs in mice was evaluated after repeated administration either subcutaneously or through DMAP. The results of this experiment showed comparable levels of anti-ovalbumin immunoglobulin generation through both routes of administration (with significant production of OVA-specific IgG1 and IgG2b antibodies), highlighting the good potential of this delivery platform for vaccination or immunotherapy applications.


Introduc on
Dissolving microneedle array patches (DMAPs) are arrays of needle-like structures with microscale diameters and lengths up to 1 mm mainly made using polymers that dissolve with the inters al fluid a er their inser on into the skin [1].Given their small size, when they are inserted in the skin, while they enable the deposi on of drugs inside the upper layers of the skin (and across the external barrier of the stratum corneum), they do not reach blood vessels or pain receptors, not producing any bleeding or pain.For this reason, they are o en proposed as an alterna ve drug administra on op on without the need for conven onal injec ons [2][3][4].DMAPs have been proposed for a wide variety of therapeu c applica ons, such as vaccines [5][6][7], cancer treatment [8,9], migraines [10], fungal infec ons [11], psoriasis [12] or malaria [13], among many others.Within the polymer matrix that makes up the microneedles, different micro-or nano-par cle based formula ons can be included to improve the therapeu c performance of the formula on [14].For example, if nanopar cles with sustained drug release are administered through DMAP, their deposi on inside the skin creates a reservoir of the drug at the site of administra on, which is slowly released from the nanopar cles, reducing the need for con nuous re-administra on of the treatment.Based on this concept, DMAP has been prepared containing many types of nanopar cles, such as liposomes [15][16][17], cubosomes [18,19], polymeric nanopar cles [20][21][22], metallic nanopar cles [23] and mesoporous silica nanopar cles (MSNs) [24].Among these different types of nanopar cles, MSNs with tunable pore sizes (generally in the range of 2-20 nm in diameter [25]) cons tute a par cularly promising tool as drug delivery systems [25][26][27], as they present a large loading capacity for cargos of a wide range of molecular weights [28][29][30].MSNs have been proven to be safe and to undergo dissolu on in physiological environments, giving rise to nontoxic degrada on products such as silicic acid, which can be safely excreted in urine [31,32].Several microneedle arrays (either dissolving or nondissolving) containing MSNs have been previously reported [24,[33][34][35].However, the amount of MSNs contained in these previously reported formula ons was rela vely low and would likely not be enough to deliver a therapeu cally relevant dose of drug for most poten al applica ons.Furthermore, the possibility of preparing DMAP with MSNs of different pore sizes has not previously been reported to the best of our knowledge.The development of an efficient method to prepare DMAP containing large quan es of MSNs of different pore sizes would provide a tunable pla orm that could be adapted for many different therapeu c applica ons, as the pore size of the nanopar cles could be tailored for the desired drug [27], and might even allow for combina on therapies in which a mixture of different MSN formula ons (each op mized for a different drug) could be codelivered through a single DMAP.In this work we report for the first me a simple method to obtain DMAP containing large amounts of MSNs of tuneable mesopore sizes.Different nanopar cle-containing DMAP formula ons were prepared and characterized, and their therapeu c poten al was assessed through a variety of in vitro, ex vivo and in vivo methods.The modular pla orm presented here could be adapted to deliver sustained release formula ons of therapeu c molecules over a wide range of molecular weights, either as monotherapies or as combina on therapies with mul ple drugs.

Synthesis of MSNs
MSNs of different pore sizes were prepared by a previously described biphasic method based on the condensa on of TEOS in a biphasic water/cyclohexane system, using triethanolamine as the base and CTAC as the structure-direc ng agent surfactant [25,26].The aqueous phase was composed of a mixture of 24 mL of a commercial aqueous solu on of CTAC (25% w/v)), 0.18 g of triethanolamine and 36 mL of deionized water.The organic phase consisted of 20 mL of a mixture of cyclohexane with TEOS.The concentra on of TEOS depended on the material to be prepared: 40% for S-MSNs, 20% for M-MSNs, 10% for L-MSNs and 5% for XL-MSNs.The synthesis reac on was carried out at 50°C for 24 h.Then, the surfactant was extracted by ion exchange with an ethanolic solu on of ammonium nitrate (10 mg/mL) at reflux for 1 h, followed by a second reflux for 2 h in an ethanolic solu on of 12 mM HCl.Finally, the material was washed with ethanol 3 mes to obtain the desired materials, which were dried and stored at room temperature un l further use.Fluorescent MSNs were also obtained by adding a mixture of 1.5 mg of fluorescein FITC or RITC and 15 µL of APTES in 1 mL of ethanol in the aqueous phase during MSN synthesis.

Cargo loading and release from MSNs
Fluorescein sodium salt (as a model for small molecule drugs) or OVA (as a model for therapeu c proteins) was loaded in MSNs by dispersing 10 mg of MSNs in a 10 mg/mL solu on of the cargo in PBS (10 mM, pH=7.4) and s rring overnight.Then, the loaded par cles were collected by centrifuga on, and the nonloaded cargo was quan fied from the supernatant by UV-Vis spectrophotometry.For release experiments, loaded par cles were suspended in PBS and s rred at 37°C.At different me points, the par cles were centrifuged, released cargo was quan fied by fluorimetry (fluorescein sodium salt, λEX=580 nm; λEM=520 nm) or UV-Vis spectrophotometry (OVA, λABS=280 nm), and the par cles were resuspended in fresh PBS to con nue s rring at 37°C.

Characteriza on techniques
Dynamic light sca ering (DLS) and Z-poten al measurements were performed with a Malvern Zetasizer Nano ZS90 instrument, checking both par cle size and surface charge.The instrument used was equipped with a "red laser" (ʎ = 300 nm), and DLS measurements were performed with a detec on angle of 90°, while the Smoluchowski approxima on was used for Z-poten al measurements.To check the morphology and the different pore sizes of the nanopar cles, the characteriza on of the nanopar cles was performed by transmission electron microscopy (TEM) on a Thermo Fisher Scien fic Tecnai G2 20 Twin using copper grids of mesh size 200 coated with a Formvar-Carbon film.Scanning electron microscopy (SEM) was carried out on an FEI Quanta-250 microscope (Thermo Fisher Scien fic, USA) a er coa ng the samples with a thin layer of gold under vacuum.Nitrogen adsorp on (in a Micromeri cs ASAP 2020 unit) measurements were carried out at the Central Research Support Services (SCAI) of the University of Malaga (UMA).Fluorimetry and UV-Vis spectrophotometry were carried out using a plate reader (FLUOstar Omega Microplate Reader, BMG LABTECH, Germany).DMAPs were visualized and imaged using a stereomicroscope (Leica EZ4 D, Leica Microsystems, Milton Keynes, UK).Constant compressive force was applied through a TA-TX2 Texture Analyser (Stable Microsystems, UK).Op cal coherence tomography (OCT) was carried out in an EX-101 device (Michelson Diagnos cs Ltd., Kent, UK).Fluorescence microscopy was carried out on an EVOS FL microscope (Thermo Fisher Scien fic, USA).Two-photon fluorescence microscopy was carried out in a Leica TCS SP8-MP mul photon excited fluorescence upright microscope (Leica Microsystems, UK).In vivo fluorescence was evaluated with In-Vivo Xtreme equipment (Bruker, Germany).Flow cytometry was carried out in a CytoFLEX cytometer (Beckman Coulter, USA).Confocal microscopy was performed using a Leica SP5 HyD Confocal Microscope (Leica, Germany).

Prepara on of MSN-loaded DMAP
MSN-loaded DMAP was prepared using a nega ve silicone mold with a design containing 600 pyramidal microneedles (750 µm in length) through the following procedure: i) the microneedle p region was filled with MSNs in powder form using a spatula; ii) a 20% (w/w) polymer solu on (PVA and PVP at a 1:1 weight ra o) in deionized water was added to each mold, followed by centrifuga on and removal of excess polymer solu on; iii) 800 µL of a 40% (w/w) polymer solu on (PVA and PVP at a 1:1 weight ra o) in deionized water was added to each mold, followed by centrifuga on; iv) the samples were le to dry for 24 h at room temperature and for 24 addi onal hours at 37°C.Then, the DMAP was removed from the mold and stored un l further use.Control DMAPs without MSNs were prepared by a similar procedure, skipping steps i) and ii).

2.6
In vitro evalua on of DMAP inser on and dissolu on First, DMAP inser on was evaluated in vitro using a Parafilm M® inser on model [36] by applying a compressive force of the DMAP against 8 layers of Parafilm M® for 30 seconds, either 32 N using a Texture Analyser or manually using thumb pressure (32 N was previously selected as a comparable force to the force a human produces when applying thumb pressure on microneedles [36,37]).The depth of inser on was then evaluated by examining the different Parafilm M® layers under a stereomicroscope.DMAP dissolu on was then evaluated by introducing DMAP in PBS and imaging the microneedles at different me points using a stereomicroscope.Finally, DMAP inser on and dissolu on were also evaluated in a 3% agarose gel.Five minutes a er inser on in the agarose gel, the baseplate of the DMAP was removed, and the fate of FITC-labelled or fluorescein sodium-loaded MSNs was evaluated 1 h later (a er incuba on at 37°C) using fluorescence microscopy.

Ex vivo experiments using neonatal porcine skin
Full thickness neonatal porcine skin was used as a skin model, with samples obtained from s llborn piglets and immediately (<24 h a er birth) excised.Skin samples were shaved and stored in sealed Petri dishes at −20°C un l use.Prior to use, skin samples were equilibrated in PBS.Inser on of DMAP into neonatal porcine skin was carried out as described for the Parafilm M® in vitro model.Inser on was evaluated by OCT, and dissolu on was evaluated a er different me points at 37°C.During DMAP in situ dissolu on, samples were kept in a sealed container where PBS-we ed paper was placed below the skin samples to prevent them from drying.Quan fica on of deposited fluorescent MSNs was carried out by fluorimetry following the extrac on of the MSNs from excised skin into PBS by thorough sonica on in an ultrasound bath.The diffusion of FITC-labelled MSNs or fluorescein sodium (from loaded MSNs) from DMAP across neonatal porcine skin was evaluated using Franz diffusion cells (Crown Glass Co. Inc., Sommerville, USA).Receptor compartments were filled with PBS, and the temperature was controlled during the experiment at 37°C.Skin samples were secured to the donor compartment of the diffusion cell using cyanoacrylate glue with the stratum corneum side facing the donor compartment.DMAPs were inserted as previously described into the center of the skin sample.DMAPs were kept in place during the experiment by a cylindrical metal weight (diameter 11 mm, mass 11.5 g) on their upper surface.With DMAP in place, donor compartments were mounted onto the receptor compartments of the Franz cells.Using a long needle, 0.2 mL of sample was removed from the receptor compartment at defined me intervals and replaced with an equal volume of PBS.Sink condi ons were maintained throughout the experiment.The concentra ons of FITC-labelled MSNs or fluorescein sodium in the receiver medium were determined by fluorimetry.

In vitro evalua on of MSNs in a model of dendri c cells
A mouse dendri c cell line (DC 2.4) was used to evaluate the immunological effect of MSNs [27,38].The day prior to the experiment, 50,000 DC2.4 cells were seeded in each well, using a 96 well plate.For cellular uptake experiments, DC 2.4 cells were incubated with RITC-labelled MSNs for 2 h at a concentra on of 10 µg/mL (in complete culture medium, RPMI-1640 supplemented with 10% fetal bovine serum, nonessen al amino acids, L-glutamine and βmercaptoethanol, as recommended by the distributor (Sigma-Aldrich)) at 37°C and 5% CO2.Twenty-four hours later, nanopar cle uptake was evaluated by flow cytometry and fluorescence microscopy.To evaluate the biological effect, changes in the expression of CD40 (a marker of dendri c cell ac va on) were assessed by flow cytometry a er incuba on with empty and OVAloaded nanopar cles (nonlabelled).

In vivo experiments in mice
Mouse studies were carried out following Spanish na onal and European regula ons (RD1201/2005, 32/2007, 2010/63/UE and RD53/2013).The mice were hosted at IBIMA-Plataforma BIONAND (Registra on No. ES 290670001687).All procedures followed the 3R principles and received appropriate regulatory approval before star ng (protocol 18/11/2021/180 approved by both the Ins tu onal Ethics Commi ee and by Consejería de Agricultura, Ganadería, Pesca y Desarrollo Sostenible, Junta de Andalucía).The mice were anaesthe zed during the different procedures and finally sacrificed by cervical disloca on.A er obtaining the corresponding samples, the mice were stored and incinerated according to ins tu onal guidelines.
To evaluate DMAP inser on and MSN deposi on in mice, DMAP loaded with FITC-labelled M-MSNs was used.Five-to six-week-old BALB/c mice (both male and female, n=3 mice per group) from Janvier Lab (Saint-Berthevin Cedex, France) were used.For DMAP administra on, back hair was first removed from the mice by using a hair-removal cream under intraperitoneal anaesthesia (xylazine + ketamine mixture).A er washing the skin with saline solu on to remove the cream and gently drying it with paper, the DMAP was inserted on the back skin by applying appropriate pressure for 30 seconds with the mice s ll under anaesthesia, and then adhesive tape was used to fix the DMAP in the same posi on.A er 2 h, the baseplates were removed, and excess polymer on the skin surface (not inserted) was removed with PBS.A er different me points, nanopar cle fluorescence in the mice was examined using an in vivo imaging system.At the endpoint (3 days a er DMAP administra on), the mice were euthanized by cervical disloca on under anaesthesia, and the skin was observed by fluorescence microscopy.The deposited MSN amount was quan fied as described for the ex vivo experiments.Finally, ssue sec ons from skin were obtained and evaluated under fluorescence confocal microscopy a er immunofluorescence staining for immune cells (using primary an bodies against F4/80 for murine macrophages and CD11c for mouse dendri c cells: rabbit an -mouse CD11c primary an body and goat an -mouse F4/80 primary an body; and fluorescent secondary an bodies for visualiza on: chicken an -rat IgG (H+L) cross-adsorbed secondary an body-Alexa Fluor™ 647; goat an -rabbit IgG (H+L) cross-adsorbed secondary an body-Alexa Fluor™ 555).
To evaluate the immunological effect of an gen-loaded MSNs administered through DMAP, OVAloaded XL-MSNs were administered either subcutaneously or in DMAP once a week for 3 weeks.Five-to six-week-old BALB/c mice (male, n=3 mice per group) from Janvier Lab (Saint-Berthevin Cedex, France) were used.One week a er the last administra on, the mice were anaesthe zed intraperitoneally, and blood was obtained from the retroorbital plexus before euthanizing the animals.Different specific an -OVA an bodies (IgG1, IgG2a, IgG2b and IgE) were determined from the mouse sera by ELISA using bio nylated rat an -mouse an bodies.For the ELISAs, high binding ELISA 96-well plates were coated with OVA.A er blocking the plate with a caseincontaining buffer solu on, the mouse sera were added (1:8 dilu on for IgE detec on, 1:50 dilu on for IgG detec on) and incubated overnight at 4°C.Then, bio nylated secondary an bodies were added, followed by the addi on of avidin-horseradish peroxidase (HRP).Finally, the results were obtained by measuring the colorimetric conversion of an HRP substrate (TMB) a er stopping the reac on with H2SO4 using a plate reader (λABS=450 nm).Thorough washing of the plate with PBS containing 0.05% Tween 20 was carried out between the different steps in the protocol.

Synthesis and characteriza on of MSNs
MSNs with different pore sizes (from smaller to larger: S-MSN, M-MSN, L-MSN and XL-MSN) were prepared and characterized.The size histograms obtained by DLS show peak par cle diameters of approximately 100 nm (Figure 1A-D), with Z average values in the range of 110-160 nm and narrow size distribu ons (polydispersity index, PDI, below 0.2) for all the obtained par cles (Table 1).The nanopar cle surface charge was nega ve for all the prepared MSNs, as would be expected by the presence of silanol groups on the external surface of silica nanopar cles (Table 1).The round morphology and porosity of the MSNs could also be observed in the TEM micrographs of the prepared nanopar cles (Figure 1E-H).The textural proper es of the prepared MSNs were further confirmed by N2 adsorp on (Table 1).All the materials presented large surface areas (in the range of 300-700 m 2 /g), typical of mesoporous silica materials.The pore diameters obtained by N2 adsorp on confirm the successful prepara on of MSNs with 4 different pore sizes, in the range of 3 to 12 nm (S-MSN<M-MSN<L-MSN<XL-MSN).These results were in good agreement with the characteris cs of MSNs prepared by other authors using the same method [26].The effect of pore size on cargo loading and release was evaluated using fluorescein sodium salt as a model for small molecule drugs and OVA as a model for therapeu c proteins and macromolecules (Figure 2).The loading of OVA was maximum for extralarge pore par cles (XL-MSN) and decreased as pore size was reduced.This result was in good agreement with previous reports that have also shown that MSNs with extralarge pore sizes presented increased OVA loading compared to par cles with smaller pores [27].On the other hand, fluorescein sodium salt loading was maximum for par cles with small or medium pores (S-MSN and M-MSN) and dras cally decreased for par cles with larger pores.These data indicate that to obtain op mal cargo loading in MSNs, the pores should be large enough to accommodate the cargo, but if the pore size is too large, the loading efficiency decreases.Thus, the pore size should be tuned depending on the molecular weight of the cargo, requiring smaller pores for small molecules and larger pores for macromolecules such as proteins.Despite this, when cargo release experiments were carried out, larger pore par cles presented faster release kine cs for both model cargos, as the larger pores provide easier accessibility to the solvent, which drives release.This result is also in good agreement with previous reports [27].Furthermore, taking these results into account, a combina on therapy scheme could be envisioned where drugs of different molecular weights can be coadministered in a cocktail of MSNs of varying pore sizes, each one of which is op mized for one of the drugs in the combina on.

Prepara on and characteriza on of MSN-loaded DMAP
Then, DMAP-containing MSNs of different pore sizes were prepared using PVP and PVA (2 watersoluble polymers).When a emp ng to disperse MSNs in a solu on of PVP and PVA to prepare DMAP, we found that the maximum weight % of MSNs that could be added in the mixture that s ll enabled filling up the molds used to prepare DMAP was 30%.When larger percentages of MSNs were included in the polymer mixture, its viscosity was too high to properly fill the mold used to obtain DMAP.As our aim was to prepare DMAP with a large amount of MSNs located in the microneedle ps, we developed an alterna ve method by first filling the mold with MSNs in powder form and later adding the polymers in solu on.A er following this process, the stereomicroscopy images of the obtained DMAP (Figure 3) confirm the successful prepara on of the desired arrays presen ng well-defined microneedles of the expected length (ca.750 µm) and with intact ps.No morphological differences were found between the blank DMAP (without nanopar cles) and those containing the different types of MSNs.Furthermore, the SEM micrographs (Figure 3P-Y) confirm the presence of MSNs, which make up most of the microneedle ps, as seen in the higher magnifica on micrographs (Figure 3U-Y).These results confirm the prepara on of DMAP in which the majority of the microneedle ps are composed of MSNs, in contrast with previous reports, where the amount of MSNs included in DMAP formula ons was rela vely low [24].Furthermore, as only the microneedle ps will be inserted inside the skin upon DMAP applica on, the MSNs should be selec vely located in the microneedle ps to avoid unnecessary wastage of drug-loaded par cles in future therapeu c applica ons of these formula ons.To confirm the loca on of the MSNs within the patches, FITClabelled M-MSNs were prepared and used to obtain DMAP.The stereomicroscopy and twophoton fluorescence microscopy images obtained from these formula ons (Figure S1) confirm the presence of the MSNs only in the microneedle ps.

In vitro and ex vivo evalua on of MSN-loaded DMAP
The mechanical proper es of the prepared DMAP and whether they allow for their inser on in skin were first evaluated in vitro using a previously reported Parafilm M® model [36].DMAP inser on was evaluated either using a Texture Analyser applying 32 N for 30 seconds or by manually applying thumb pressure for 30 seconds.The results (Figure 4) show that the inser on of all the prepared DMAP was similar when inserted in the same way, regardless of whether the DMAP contained any kind of MSN.Furthermore, the DMAPs were inserted more efficiently by manual applica on (successfully piercing through the 3 rd Parafilm M® layer, with a depth of 378 µm) compared to the ones inserted using the Texture Analyser equipment (which only reached the 2 nd layer, at a depth of 252 µm).These results are in good agreement with previous reports of microneedle patches that show similar inser on in the Parafilm M® model, as well as deeper inser on upon manual force [39].Finally, the mechanical proper es of the prepared DMAP were also proven to be adequate for further evalua on, as there was no significant change (p>0.54) in microneedle length a er inser on in the Parafilm M® model, neither by manual nor Texture Analyser applica on (Figure 4B).The in vitro dissolu on of the different DMAP formula ons in aqueous media was also confirmed by immersing them in PBS.Three minutes a er immersion, the microneedles of all DMAP were fully dissolved, as observed by stereomicroscopy (Figure S2).
By dissolving in PBS DMAP containing FITC-labelled M-MSNs and then measuring the fluorescence of the resul ng suspensions, the amount of nanopar cles loaded in each DMAP could be es mated to be 2.33 ± 0.04 mg M-MSNs/DMAP.The next step was to evaluate the inser on and dissolu on of DMAP in neonatal porcine skin, which will provide more relevant informa on towards the poten al use of these formula ons in humans.A er manual ex vivo applica on in porcine skin, the OCT images confirm the successful inser of the microneedles inside the porcine skin, with an average inser on length of 472 ± 26.7 µm (62.9 ± 3.5% of the total microneedle length) (Figure 5).Furthermore, the successful dissolu on of the microneedles inserted in the skin was also observed, with par al dissolu on taking place in DMAP inserted for 30 min at 37°C and reaching almost complete dissolu on at 60 min (Figure 5B-C).At this me (60 min), the amount of FITC-labelled M-MSNs that had been deposited in the skin was 0.75 ± 0.02 mg, which was 20.9 ± 7.26% of the total amount present in the DMAP (Figure 5D-F).As seen in the results shown in Figure 2, cargo loading and release in MSNs is strongly affected by the interac on between cargo molecular weight and the pore size of the nanopar cles.For this reason, we postulate the poten al use of combina ons of MSNs of various pore sizes for combined drug codelivery, where each type of MSN is op mized for one drug in the mix.With this goal in mind, the next step was to evaluate whether the method developed to prepare DMAP would enable patches with combina ons of different MSNs to be obtained.To evaluate this, we selected MSNs that presented op mal loading for each of the model molecules previously used: M-MSNs (which presented the largest fluorescein sodium salt loading capacity) and XL-MSNs (which presented the largest OVA loading capacity).To visualize each type of par cle independently, we prepared and used differently labelled MSNs: FITC-labelled M-MSNs and RITC-labelled A er mixing (in suspension) both types of MSNs in a 1:1 weight ra o and drying the combina on, the MSN mix was used to prepare DMAP following the same method as previously described.The characteriza on of these new DMAPs confirms the successful prepara on of DMAPs containing a combina on of MSNs of different pore sizes (Figure 6, Figure S3, Table S1).Furthermore, DMAP containing a combina on of FITC-labelled M-MSNs and RITClabelled XL-MSNs presented analogous inser on and nanopar cle deposi on in neonatal porcine skin as those previously evaluated containing only one MSN type (Figure S4, Table S2).In a poten al future applica on of these DMAPs, a er inser on and microneedle dissolu on, drug-loaded MSNs will be deposited inside the skin.The fate of both MSNs and their cargo a er deposi on in the skin should therefore also be evaluated to determine the poten al these formula ons.To evaluate the diffusion of both the nanopar cles and their cargo, DMAP with 2 different nanopar cle systems was prepared: FITC-labelled M-MSNs (to analyse the fate of the nanopar cles) and fluorescein sodium salt-loaded M-MSNs (to analyse the fate of the fluorescent cargo as a model for a drug being released from the nanopar cles).In the first experiment, an agarose gel was used as a model for skin ssue.A er inser on of DMAP, the microneedles dissolve and leave the MSNs inside the agarose gel.One hour later, sec ons of the gel were cut and evaluated by fluorescence microscopy.The results (Figure S5) show that while the MSNs remain in the site of deposi on, the cargo (fluorescein sodium salt) diffuses out of the nanopar cles and into the gel.This would indicate that the nanopar cles would remain in the site of inser on, ac ng as a depot system and providing a sustained release of the drug to the surrounding skin ssue.To confirm these results, a second experiment was carried out in a Franz cell setup.The same DMAP as those from the previous experiment (plus an addi onal control DMAP with nonencapsulated fluorescein sodium salt) were inserted in neonatal porcine skin that acted as the membrane separa ng the donor and receptor compartments.At different me points, samples were taken from the receptor compartment, and the amount of FITC-labelled M-MSNs or free fluorescein sodium salt was evaluated by fluorimetry.The results (Figure 7) confirm those from the experiment, showing that MSNs are not capable of diffusing through the skin ssue, as no nanopar cle fluorescence was detected in the receptor compartment.On the other hand, fluorescein sodium salt crosses through the skin ssue a er being deposited with DMAP.Importantly, encapsula on of the dye inside M-MSNs provided sustained release compared to the use of DMAP with a nonencapsulated fluorophore.Taken together, these results indicate that the MSNs deposited in the skin through DMAP remain at the site of inser on and act as a drug reservoir.Then, the sustained release of the drug would provide a con nuous flow of the drug towards both the surrounding ssue or even systemic circula on (depending on the release kine cs and dose of the drug or nanopar cles included in the patches).Using the DMAP-inserted skin samples from the previous experiment, the fate of the deposited nanopar cles was also evaluated by confocal microscopy using immunofluorescent labelling.In previous in vitro and ex vivo experiments, we observed that deposited MSNs remain in the site of inser on without significant diffusion into the surrounding skin ssue.However, these were performed in the absence of living cells in the skin ssue that might engulf the nanopar cles and determine their final fate and biological performance.Two different types of fluorescent labels were used: F4/80 staining (a tradi onal marker for macrophages [40,41]) and CD11c staining (a tradi onal marker for dendri c cells) 40,42]).In agreement with previous reports, we found (Figure 9) that the majority of the dermal cells in mouse skin express F4/80, while a much smaller percentage express CD11c [40].In any case, most of the dermal cells are an gen-presen ng cells, either F4/80 + or CD11c + .Green fluorescence from the nanopar cles was found in all ssue sec ons around cell nuclei (stained with DAPI), indica ng that the nanopar cles are taken up by dermal cells a er deposi on in the skin from DMAP.Nanopar cle fluorescence was observed inside an gen-presen ng cells (which were also posi ve for either F4/80 or CD11c, Figure 9E, F, K, L).As the MSNs deposited in the skin through DMAP were then taken up by an gen presen ng cells, a poten al applica on of these systems would be in vaccina on or immunotherapy as carriers of different an gens to the target immune cells in the skin.To evaluate the biological effect of an gen-loaded MSNs in an gen-presen ng cells, an in vitro experiment was first carried out using a mouse cell line commonly employed as a model for dendri c cells (DC2.4 [27]).The results (Figure S6) show that DC2.4 cells efficiently uptake MSNs, with a clear pore size effect, as XL-MSNs presented a lower uptake % compared to all other types of MSNs.When the ac va on of DC2.4 cells was evaluated (by quan fying the % of cells expressing CD40 by flow cytometry), the results showed that OVA-loaded XL-MSNs induced the largest degree of dendri c cell ac va decreasing the % of CD40 + cells as the nanopar cle pore size decreased.Furthermore, treatment with free OVA in the absence of a nanocarrier, even at high concentra ons (up to 4 µg/mL), did not induce significant cell ac va on.These results are in good agreement with previous reports that showed that MSN pore size regulates their an gen delivery efficiency [27], with extralarge pore MSNs loaded with OVA (similar to the ones prepared in this work) being the op mal an gen delivery system.In previous reports, the good poten al of XL-MSNs as an gen delivery systems [27,43] was evaluated by subcutaneous injec on in mice.While subcutaneous injec on is known to produce a strong immune response, the need for trained personnel and the pain associated with the injec on could be improved with a needle-free administra on system such as the one proposed here.In some applica ons, there are also safety concerns associated with subcutaneous administra on, such as in allergy immunotherapy, in which an undesired allergic reac on to the treatment can be triggered in subcutaneous immunotherapy (with some reports indica ng an improved safety profile of microneedle-assisted allergy immunotherapy [44]).To evaluate the prospects of MSN-loaded DMAP for immuniza on, OVA-loaded XL-MSNs were administered 3 mes (weekly) in BALB/c mice either subcutaneously or in DMAP (equivalent nanopar cle dose in both types of administra on).OVA-loaded XL-MSNs were used since despite presen ng a lower uptake %, their capacity to ac vate DC2.4 cells was higher.One week a er the last administra on, blood was obtained, and OVA-specific an bodies were evaluated in sera by ELISA.The results (Figure 10) show comparable humoral responses between subcutaneous and DMAP-mediated administra on of OVA-loaded XL-MSNs, with significant produc on of an -OVA IgG1 and IgG2b compared to nontreated control mice.On the other hand, no significant differences were found between control and nanopar cle-treated mice in the produc on of specific an -OVA IgG2a or IgE.These results highlight the promising nature of the developed pla orm for vaccina on or immunotherapy through needle-free administra on.Future work will explore the combina on of MSNs carrying an immunomodulatory adjuvant with an gencarrying MSNs for different therapeu c applica ons, as this work has demonstrated the possibility of preparing DMAP with combina ons of different types of MSNs tailored for various therapeu c molecules of a wide range of molecular weights.

Conclusions
In this work, dissolving microneedle array patches containing large amounts of mesoporous silica nanopar cles of different pore sizes were prepared and characterized.The developed method also enabled the prepara on of microneedle array patches containing a combina on of different mesoporous silica nanopar cles, which could be useful for the development of combina on therapies codelivering different therapeu c cargos.The successful inser on, dissolu on and nanopar cle deposi on from the microneedles was confirmed through a series of in vitro, ex vivo and in vivo experiments.As the microneedle-delivered mesoporous silica nanopar cles were found to end up inside an gen presen ng cells in the skin ssue, a par cularly promising applica on of these systems would be in vaccina on or immunotherapy applica ons.To confirm this poten al applica on, the immune response to ovalbumin-loaded mesoporous silica nanopar cles in mice was evaluated, showing comparable levels of specific an body genera on a er subcutaneous or microneedle-mediated delivery.Based on the promising results presented here, future work will further evaluate the therapeu c poten al of this pla orm for immunotherapy in different disease scenarios, by developing microneedle codelivery systems containing an genic and adjuvant molecules encapsulated in op mized mesoporous silica nanopar cles.

Figure 2 .
Figure 2. Cargo loading and release experiments in/from MSNs of different pore sizes.Loading of fluorescein sodium salt (A) and OVA (B); Release of fluorescein sodium salt (C) and OVA (D).Data are means ±SDs, n=3.

Figure 4 .
Figure 4. DMAP inser on in the Parafilm M® model either using a Texture Analyser (TAI) (con nuous lines) or through manual inser on (MI) (do ed lines) (A).Blank (not containing MSNs) or MSN-containing DMAP were evaluated.Microneedle lengths before and a er inser on in the Parafilm M® model (B).Data are means ±SDs, n=3.

Figure 5 .
Figure 5. DMAP inser on and dissolu on ex vivo using neonatal porcine skin.OCT image of M-MSN-containing DMAP inserted in neonatal porcine skin (A).Photographs of FITC-labelled M-MSN-containing DMAP a er inser on in skin for 30 (B) or 60 (C) min.Photographs of neonatal porcine skin a er removal of FITC-labelled M-MSN-containing DMAP that had been inserted for 30 (D) or 60 (E) min.Quan fica on of FITC-labelled M-MSN deposited inside neonatal porcine skin at different inser on mes (F).Data are means ±SDs, n=3.

Figure 7 . 3 . 4
Figure 7. Franz cell experiment with neonatal porcine skin.Fluorescence microscopy images of the different DMAPs used in the experiment: FITC-labelled M-MSNs (A), fluorescein sodium saltloaded M-MSNs (B), and nonencapsulated fluorescein sodium salt (C).Schema c representa on of the Franz cell setup (D).Fluorescence intensity in the receptor compartment (indica ng transdermal delivery) at different me points a er DMAP inser on in neonatal porcine skin (E).Amount (in µg) of total fluorescein sodium salt delivered transdermally at different me points (F).Percentage of total fluorescein sodium salt delivered transdermally at different me points (G).Data are means ±SDs, n=5.3.4In vivo evalua on of MSN-loaded DMAPOnce the characteris and performance of MSN-loaded DMAP had been evaluated through the different in vitro and ex vivo experiments described above, a series of in vivo tests were carried out using a mouse model to examine the therapeu c poten al of the developed pla orm.First, the in vivo inser on, microneedle dissolu on and nanopar cle deposi on were

Figure 8 .
Figure 8. DMAP inser on, dissolu on and MSN deposi on in vivo.In vivo fluorescence imaging of both controls and FITC-labelled M-MSN-loaded DMAP-inserted male and female mice taken 3 days a er DMAP administra on (top).Ex vivo fluorescence stereomicroscopy of skin excised from the same mice showing fluorescence of the deposited FITC-labelled M-MSNs (bo om).

Figure 9 .
Figure 9. Confocal microscopy images of immunofluorescently labelled mouse skin sec ons 3 days a er FITC-labelled M-MSN-loaded DMAP administra on.Female (A,B,G,H) and male (C,D,I,J) mice were used, either as controls (A,C,G,I) or with DMAP applica on (B,D,H,J).Inserts at higher magnifica on showing double-posi ve cells: MSN + F4/80 + (E, F) or MSN + CD11c + (K, L).Blue signal represents DAPI nuclei staining, green signal represents MSN fluorescence, red signal represents F4/80 staining (A-F) and yellow signal (highlighted with white arrows) represents CD11c staining (G-L).

Table 1 .
Characteriza on of the prepared MSNs by DLS, Z poten al and N2 adsorp on.