The Effect of a Lipid Surface Coating on the Permeation of Upconverting Nanoparticles through a 3D Human Lung Epithelial Model

Preclinical studies of nanoparticles for pulmonary therapeutics are often performed on 2D cell cultures or in vitro models that do not include a mucus barrier. However, the mucus layer lining the lungs is an essential barrier for drugs to permeate in order to exert a therapeutic effect. Herein, the role of surface coating of lanthanide‐doped upconverting nanoparticles (UCNPs) and their interaction with the mucus barrier are explored using a patient‐derived 3D cell culture model. The upconverted emissions from the UCNPs are used to track them throughout the 3D model and study their localization as a function of administration time and mucus thickness. Positively charged, ligand‐free, and negatively charged, supported lipid bilayer‐coated UCNPs are evaluated. A substantial difference in the residence time in mucus and mucociliary clearance of each type of UCNP is observed in a realistic and relevant model. As such, these results underscore the need for preclinical investigations in tissue models, especially with respect to the surface properties of the nanoparticles under study.


Introduction
The widespread paradigm for studying nanoparticle-cell interactions based on two-dimensional cell cultures is presently beset by challenges with regard to clinical translation.Increasingly, researchers advocate for the use of refined models to bridge the in vitro -in vivo gap, [1] that can reproduce physiological conditions and generate reliable and reproducible long-term measurements under standard laboratory conditions. [2]OI: 10.1002/adfm.4][5][6] To understand their specific roles and design successful nanoparticle delivery strategies, it is of primary importance to be able to isolate each parameter.Lanthanide-doped nanoparticles are attractive candidates for studying the interaction of nanoparticles in a complex microenvironment, as the lanthanides do not photobleach, and exhibit narrow emission bands that are easily isolated for imaging, and can undergo photon upconversion.[ 7,8] LiYF 4 :Yb 3+ ,Tm 3+ nanoparticles are particularly interesting because of the wide range of upconverted emissions obtained from the Tm 3+ /Yb 3+ pairing, including ultraviolet, blue, red, and nearinfrared (NIR) emissions upon excitation at 980 nm. [9]The use of 980 nm excitation minimizes deleterious autofluorescence from tissues, and the large anti-Stokes shift of the upconverted blue emissions from Tm 3+ do not overlap with common fluorophores such as green fluorescent protein (GFP). [10,11]14] Herein, we focus on the role of the nanoparticle surface coating with respect to their interaction with a physiological barrier of  3+ ,Tm 3+ UCNPs and d) corresponding upconversion emission spectra.The orange trace is collected from a colloidal dispersion of UCNPs (1 mg/mL) upon continuous wave (CW) 980 nm excitation and acquired by a high-resolution spectrometer.The blue trace is the average signal extracted from a laser-scanning microscopy image of dried LiYF 4 :Yb 3+ ,Tm 3+ deposited on a microscopy substrate upon femtosecond pulse excitation at 980 nm.The upconverted emission is epi-collected and directed into the microscope hyperspectral detection unit, with spectral resolution set at 2.5 nm.e) Schematic of the configuration used for the mucus-only measurements (experiments 1 and 2).Hyperspectral images at 2 h after the aerosol deposition of f) SLB-and g) LF-UCNPs on the top of the mucus layer.Images were acquired within a 400-590 nm spectral range with spectral resolution of 6.0 nm upon 980 nm excitation.Signals at 430-485 nm correspond to UCNPs; 520-590 nm correspond to two-photon excited fluorescence (TPEF) from the porous membrane upon 800 nm excitation.The dark region between the UCNPs and the membrane defines the volume occupied by the mucus.h) Time-resolved trace of the cumulative signal of UCNPs collected in phosphate buffered saline (PBS) below the porous membrane as a function of time after exposure to SLB-UCNPs (red trace) and LF-UCNPs (yellow trace).Data shown in the traces is normalized to the maximum cumulative emission intensity (12 h, SLB-UCNPs).
biological and clinical relevance.Specifically, we consider aerosolbased nanoparticle deposition at the apical side of lung epithelial tissue coated with mucus, an adhesive secretion that lines the surface of epithelial tissue and protects them from exogenous particulates.While assuming the role of an essential protective mechanism for the body, mucus also represents a major obstacle in delivering cargo to internal organs. [15]We compared the interactions of ligand-free (LF) and supported lipid bilayer (SLB)-coated lanthanide-doped upconverting nanoparticles (UC-NPs).[18] Moreover, the use of a SLB coating facilitates biocompatibility, owing to their similar properties to cell membranes. [19]Of preclinical relevance, we use MucilAir, a standardized 3D in vitro model of lung epithelial tissue derived from human-origin primary cells to accurately mimic the morphology and function of the native tissue.MucilAir and similar tran-swell culture systems have been available on the market for many years, and are recognized as reliable toxicological screening platforms, as exemplified by their recent endorsement by the Organisation for Economic Co-operation and Development (OECD) to replace 90 day rat inhalation testing . [20][23] Despite their prevalence in toxicology, to date, they are very sparsely adopted by the bio-nano technology community to evaluate the tissue interactions of nanoparticles designed for biomedical applications.Thus, we intend to bridge this gap by taking advantage of this important in vitro technology.As reported in Figure 1a,b, the differentiated epithelium of MucilAir consists of basal, ciliated, and goblet cells cocultured with human primary fibroblasts.Importantly, the mucus barrier used in this model is obtained from healthy donors, which provides a reliable representation of the complex lung environment.

Results and Discussion
LiYF 4 :Yb 3+ ,Tm 3+ UCNPs were used owing to their uniform size (91.0 ± 4.5 nm), morphology, and luminescence properties (Figure S1, Supporting Information).The lowest-energy crystal plane of the tetragonal LiYF 4 host is the 011 plane, which results in the square bipyramidal morphology observed (Figure 1c). [24]his enables the formation of the SLB on the UCNP surface because the self-assembly of the continuous coverage generated by the phospholipid leaflets is preferred to occur on materials that exhibit the same surface energy on all faces of the nanoparticle. [25,26]onveniently, the two-photon excited fluorescence (TPEF) of the epithelium is observed in the green spectral region under 800 nm excitation, whereas the Tm 3+ emissions are observed in the blue region upon excitation at 980 nm corresponding to the Tm 3+ : 1 D 2 → 3 F 4 and 1 G 4 → 3 H 6 transitions (Figure 1d; Figures S2 and S3, Supporting Information).Thus, the location of UCNPs can be spectrally correlated with the structure and constituents of the three-dimensional model without the need for additional cell staining.
We first evaluated the permeation of UCNPs exclusively through a mucus layer, the first barrier encountered upon apical delivery (Experiments 1 and 2 in Table 1).For these measurements, mucus collected from MucilAir was deposited on the membrane of the transwell insert (Figure 1e).UCNPs were nebulized and deposited on top of the mucus by aerosol delivery to ensure uniform coverage of the sample (Figure S4, Supporting Information).Mucus is largely comprised of negatively-charged mucin glycoproteins which form a mesh-like structure with pores up to hundreds of nanometers in diameter [27][28][29] and are therefore comparable to or larger than the UCNPs used herein.Also, since 95% of mucus is composed of water, [30] a hydrophilic UCNP surface can facilitate mucus penetration.Figure 1f,g displays the spatial distribution of SLB-and LF-UCNPs 2 h after aerosol deposition.The dark region between the signal from the UCNPs and the membrane defines the volume occupied by the mucus, which is not luminescent.2 h post-deposition, a fraction of SLB-UCNPs has successfully penetrated the mucus layer and reached the membrane at the bottom, whereas the LF-UCNPs remain above the mucus and exhibit minimal penetration.
To evaluate the kinetics of UCNP mucus permeation, we quantified the upconverting luminescence signal in phosphate buffered saline (PBS) solution below the porous transwell membrane (2×10 6 pores/cm 2 ) over the course of 24 h (Figure 1h).For SLB-UCNPs, the cumulative signal increased rapidly during the first 12 h and plateaued thereafter.In contrast, LF-UCNPs showed minimal signal detection throughout, indicating a weak presence and limited penetration through the mucus layer.As the size of SLB and LF-UCNPs are identical, but smaller than the upper limit reported for particle diffusion in mucus (≈200 nm), [31] we ascribe the differences observed to the surface coatings.SLB-UCNPs are hydrophilic and display a negative surface charge (−30 mV) owing to the phosphate headgroups of the outer phospholipid leaflet, while LF-UCNPs are hydrophilic and positively charged (+34 mV).An electrostatic interaction between the LF-UCNPs and the negatively charged mucin proteins likely obstructs their movement within the layer and extends the time of particle retention within mucus.As such, a high positive charge may preclude the permeation of nanoparticles through a mucus layer.
Next, we investigated the permeation of SLB-and LF-UCNPs in the full 3D model, where mucus hindrance is combined with ciliary clearance and the differentiated epithelial tissue is present (Experiment 3, Table 1).The model is grown on top of the porous transwell membrane, facilitating exchange with the cell medium below.When imaged using TPEF, the tissue appears as a homogeneous layer with a typical thickness of 30-60 μm (Figure S2, Supporting Information), and ciliary beating is observed by brightfield microscopy (Movie S1, Supporting Information).The bronchial samples used for this experiment are characterized by a mucus layer of ≈80 μm thick.
Figure 2 shows 3D images acquired at the center of the Mu-cilAir inserts at 2 and 24 h after aerosol deposition of the UCNPs.This timeframe is compatible with the continuous secretion of mucus and clearance by cilial beating, which in vivo leads to a complete turnover of the airway mucus within several hours. [32]t 2 h, both SLB-and LF-UCNPs are mostly localized on the top of the mucus layer, displaying a homogenous and dense coverage, as clearly observed in the vertical transverse maximum intensity projections (Figure 2a,b).Compared to the penetration observed at 2 h for the mucus only experiment, there is no clear indication of SLB-UCNPs penetrating through the mucus when cilial beating is present.However, at 24 h, the SLB-UCNPs have penetrated the mucus and form dense regions in direct contact with the epithelial cells.In contrast, at this time point, LF-UCNPs are not detected on the mucus layer, with only minimal detection at or within the epithelium.
To investigate the scenario of progressive clearing by the cilia, we imaged the edges of the MucilAir inserts at 24 h (right image panels in Figure 2, corresponding to region B in Figure 1a).LF-UCNPs appear as dense aggregates.Note that the horizontal striped pattern is associated with the interplay between scanning speed and the luminescence lifetime of the UCNPs (Figure S3, Supporting Information).In contrast, the spatial pattern of the SLB-UCNPs signal is similar at the edges and center of the model, suggesting that the penetration of the SLB-UCNPs through the mucus to the epithelium occurs on a timescale faster than the complete removal by the cilia.The overall particle diffusion in the central region and the edge of the insert is summarized by plotting the spatially integrated UCNP signal and tissue autofluorescence as a function of depth for the two observation times (Figure 2c,d).For the SLB-UCNPs, at 24 h, we observe a partial overlap between the nanoparticle and tissue signals both at the center and the edge of the insert, confirming SLB-UCNPs have penetrated the epithelial layer.The intensity of the UCNP signal is significantly reduced and localized at the top of the epithelium at 24 h, indicating a fraction of the nanoparticles either have penetrated through the 3D model, or have undergone mucociliary clearance.In contrast, there is only a minimal overlap between the LF-UCNPs and tissue autofluorescence at any depth in the central region of the insert, with a prominent signal around the epithelium at the edge of the insert.
Since the efficiency of mucociliary clearance (acting parallel to the epithelium surface) is in competition with the (vertical) speed of mucus penetration, it is expected that the mucus layer thickness assumes a major role in the permeation.To investigate this, the same protocol used for Figure 2 was applied to samples from a different donor characterized by a thinner mucus layer (Experiment 4, Table 1).As reported in Figure S5 (Supporting Information), in this case, at 2 h post-treatment, we do not observe a homogeneous layer of particles on top of the mucus.Rather, a similar fraction of both LF-and SLB-UCNPs penetrated the mucus and were able to reach the epithelium surface, despite having a higher ciliary beating frequency than in the sample with thicker mucus (9.5 ± 0.6 Hz vs 8.0 ± 0.3 Hz).However, after 24 h, the situation is drastically different, as LF-UCNPs were cleared from the center of the insert, while the SLB-UCNPs remained in contact with the epithelium, in agreement with the results obtained with thicker mucus.Studies performed on a sample with an intermediate mucus thickness of 60 μm (Experiment 5, Figure S6, Supporting Information) corroborate these results.Thus, we can conclude that the thickness of the mucus affects the rate of mucus penetration, but ultimately does not appear to alter the fate of nanoparticles in terms of mucociliary clearance.It is important to recall that the mucus layer thickness is strongly dependent on a multiplicity of factors [33] (epithelium location (nasal, bronchial), donor, health state) suggesting that preclinical studies may only provide general trends for the permeation process.
We then investigated the interaction of UCNPs with the epithelial cells.To ensure unambiguous identification of the tissue constituents (mucin fibers, goblet cells, basal cells, fibroblasts, etc.), histological sectioning of tissues 24 h post-treatment was carried out (Experiments 6 and 7 in Table 1, Figures 3a;  Figures S7-S10, Supporting Information).We restricted the analysis to the center of the insert, which has been most relevant for observing the mucociliary clearance effect thus far.We visually investigated a total of 200 histological slides (100 SLB-UCNPs, 100 LF-UCNPs) using upconversion microscopy, and can confidently report a higher presence of SLB-UCNPs in cilia.This is likely due to the LF-UCNPs being trapped in the mucus layer, which was removed prior to the histological sectioning.As discussed previously, one would expect that electrostatic interactions would favor positively charged nanoparticles (LF-UCNPs) binding to cells, [34,35] however here the major obstacle is represented by the mucus and the electrostatic interaction with the negatively-charged mucins, which will inevitably occur prior to interaction with a cell membrane upon apical administration.
Finally, to affirm the role of the mucus, we evaluated UCNP permeation through the epithelium in the presence and absence of such a layer (Experiments 8 and 9, Figure 3b).Mucus-covered and mucus-free MucilAir were incubated for 24 h with SLB-and LF-UCNPs, and the presence of UCNPs in the medium beneath the model was assessed by averaging the cumulative upconversion signals obtained by imaging the underside of well plates over large fields of view (Figure 3b).While the permeation of SLB-UCNPs remains similar with and without mucus present, LF-UCNPs exhibit a over a ten-fold increase in penetration when the mucus is absent, highlighting the importance of both the mucus layer and the nanoparticle coating in nano-bio interactions.In the absence of mucus, positively-charged nanoparticles (such as the LF-UCNPs) are known to more effectively interact with cells [34][35][36] relative to their negatively-charged counterparts.However, in the presence of mucus, similar degrees of cell interaction are observed for the LF-UCNPs and SLB-UCNPs (Figure S9, Supporting Information), despite their different surface charges.As such, the role of the mucus layer in trapping the LF-UCNPs suggests the usefulness of positively-charged nanoparticles may be hindered in situations where mucus permeation is required.Such differences cannot be discerned in models without a mucus layer.Although our study does not provide a quantitative comparison of the fraction of UCNPs crossing the barrier and of the transport mechanisms involved (e.g., transcellular vs paracellular), our observations altogether underscore the inherent limitations associated with studying nanoparticle interactions using 2D cell cultures, as they fail to consider the intricate structural and chemical complexity of the surrounding microenvironment, especially with respect to the role of mucociliary clearance.

Conclusion
The significance of this work is multifaceted: i) UCNPs are a popular platform studied for biomedical applications; ii) lipid-based nanoparticles/coatings are a highly effective approach to permeate the mucus layer, and iii) aerosol delivery is an important administration route for pulmonary therapies such as treatment of chronic obstructive pulmonary disease, asthma and respiratory infections.Therefore, employing aerosol delivery in this study provides a realistic and clinically relevant approach, closely mimicking the actual procedure used in pulmonary therapies. [37]Our results shed light on a significant disparity in the behavior of SLBand LF-UCNPs in a complex microenvironment, despite their identical core size and morphology.The key distinctions are primarily attributed to the particles' ability to overcome the barrier formed by the mucus layer which consequently influences the effectiveness of ciliary clearance.This knowledge was obtained using time-dependent, non-invasive, non-destructive upconversion microscopy imaging in three dimensions, enabling clear visualization of these behaviors.These results align with the existing body of literature on the interaction between lipid-based nanoparticles and epithelial tissues, [38][39][40] with regards to the positive influence of a hydrophilic and negatively-charged surface for permeation of body epithelia lined with mucus.We therefore strongly advocate for the use of 3D in vitro models that generate realistic microenvironments as a means to facilitate preclinical translation of nanomedicines.
It is worth emphasizing that while the overall outcomes remain consistent, the specifics of the nano-bio interactions can differ between donor samples.To address this variability, we ensured that for each human donor sample (i.e., Experiments 1-9 in Table 1) that experiments were performed using both SLB-and LF-UCNPs, enabling direct comparisons between the UCNPs for each patient donor sample.This approach minimizes the influence of donor variability while focusing on the specific effect of the UCNP coating on permeation.Moreover, our results are consistent across models derived from six different donors, suggesting that the trends observed are of widespread importance.This work highlights the potential importance of personalized medical solutions and tailoring delivery approaches to the characteristics of the patient.Finally, since mucus secretion occurs in various mammalian epithelia, including the respiratory, urogenital, and gastrointestinal systems, the results presented here hold broader significance for treating other tissues.This not only enhances the translational value of the study but also enables researchers to better understand and evaluate the efficacy of therapeutic interventions in a manner that closely aligns with real-world clinical scenarios.
Lanthanide-Doped Upconverting Nanoparticles: LiYF 4 :Yb 3+ ,Tm 3+ upconverting nanoparticles were prepared according to a previously established one-pot thermal decomposition protocol. [41]Briefly, in a 250 mL 3-neck round bottom flask, 1.25 mmol of lanthanide oxides (0.9375 mmol Y 2 O 3 , 0.3125 mmol Yb 2 O 3 and 0.0025 mmol Tm 2 O 3 ) were mixed with 10 mL of 1:1 v/v H 2 O:trifluoroacetic acid and stirred at 80 °C under reflux, until the solution became transparent and colorless.The resulting solution was dried to yield lanthanide trifluoroacetate precursors as a white powder.20 mL of oleic acid and 20 mL of 1-octadecene were added to the reaction flask containing the lanthanide trifluoroacetate precursors, along with 2.5 mmol of lithium trifluoroacetate.The resulting solution was stirred at 350 rpm and heated to 120 °C in vacuo at 10 mbar for 30 min to remove residual water and gases.The solution was then brought to ambient pressure and an Argon flow was introduced.The solution was then heated to 315 °C at a rate of 10 °C min −1 and held at 315 °C for 1 h before cooling back to room temperature.The resulting light yellow solution was centrifuged at 4000 rpm for 15 min to yield a white pellet.The supernatant was discarded and the pellet was redispersed in 10 mL hexanes and precipitated through the addition of 45 mL 99% ethanol.The mixture was centrifuged again at 4000 rpm for 15 min to purify the nanoparticles.This process was repeated three times to yield oleate-capped LiYF 4 :Yb 3+ ,Tm 3+ upconverting nanoparticles.
Transmission electron microscopy (TEM) images were acquired using a ThermoScientific Talos L120C electron microscope operating at an accelerating voltage of 100 kV.TEM samples were prepared by depositing 10 μL of a 1 mg/mL dispersion of oleate-capped upconverting nanoparticles in hexanes onto a 300-mesh formvar-coated copper grid.
Preparation of Ligand-Free Nanoparticles: The oleate capping ligand was removed from the surface of the LiYF 4 :Yb 3+ ,Tm 3+ upconverting nanoparticles via simple acid-base chemistry. [42]50 mg (wet mass) of oleate-capped nanoparticles were dispersed in 5 mL of hexanes and sonicated for 15 min.The dispersion was then stirred at 900 rpm and 10 mL of pH 2 (adjusted with HCl) water was added to the dispersion.After 4 h of stirring, all nanoparticles were transferred to the aqueous layer as evidenced by the presence of upconverting luminescence upon 976 nm excitation in the aqueous layer, and absence of it in the organic layer.The aqueous layer was then collected and centrifuged at 13 300 rpm to yield a white pellet.The supernatant was discarded and the pellet was redispersed in 1 mL deionized water, followed by the addition of 1 mL of ethanol and centrifuged again to collect the oleate-free nanoparticles.Zeta potential was confirmed to be +33.0mV, indicating the positive charge of the ligand-free nanoparticles.Ligand-free nanoparticles were stored as a stock solution in sterile deionized water at 4 °C for use in the biological experiments.
Zeta potential was measured at 25 °C using a 0.5 mg mL −1 dispersion of SLB-or LF-UCNPs in HEPES buffer (pH 7.4) on a Malvern Zetasizer Nano ZSM using a DTS1070 disposable folded capillary cell.Zeta potential measurements were performed under identical conditions for SLB and LF-UCNP samples.
Preparation of the Supported Lipid-Bilayer Coated-UCNPs: The supported lipid bilayer is added to the surface of the LiYF 4 :Yb 3+ ,Tm 3+ nanoparticles with the oleate capping ligand still present, as per the previously established protocol. [26]2 mg of oleate-capped nanoparticles were dispersed in 2 mL chloroform and sonicated for 10 min.700 μL of the lipid mixture (29 mol.% cholesterol, 64 mol.%DOPA, 7 mol.%DOPC in chloroform) was added to the nanoparticle dispersion and sonicated for an additional 10 min.The solvent was then evaporated under reduced pressure to form a dry lipid film, and then rehydrated with 1 mL chloroform and re-evaporated under reduced pressure.This process was repeated a total of three times.After the third evaporation cycle, the dry lipid film was left under vacuum for an additional 45 min to ensure all traces of solvent were removed.The film was then hydrated in HEPES buffer (5 mm HEPES buffer, 5 mm sodium acetate, 30 mm NaCl, adjusted to pH 7.4 with NaOH) for 16 h.The resulting dispersion was then vortexed for 10 min followed by sonication for 10 min.The solution was then extruded 41 times though a 0.2 micron polycarbonate membrane filter.The lipid-coated nanoparticles were then collected by centrifugation at 12 000 rpm for 30 min followed by 20 min at 5000 rpm to yield a transparent pellet.The pellet was then redispersed in 1 mL HEPES buffer and used for in vitro experiments.The dispersion was stored at 4 °C and used within one month of preparation.
High resolution upconversion emission spectroscopy was performed on a 1 mg mL −1 dispersion of SLB-UCNPs in HEPES buffer (pH 7.4) in a 1 cm path length quartz cuvette (Thorlabs).A handheld 980 nm diode laser (SkyLaser, 1 W, 0.450 W cm −2 ) was used to excite the nanoparticles.Their emissions were collected perpendicular to the excitation source with a 600 μm optical fiber (OceanOptics Inc.) and filtered through a 330-807 nm band pass filter (Newport Inc. 10-CLVR-3) before being detected with a Princeton Instruments FERGIE BRX-VR UV-NIR spectrograph fitted with a 250 grooves/mm grating blazed at 550 nm with a 50 μm slit at the entrance of the spectrograph.
Cultures of Human 3D Lung Epithelial Models: MucilAir is a human airway epithelial tissue model developed by Epithelix Sàrl.Once differentiated, MucilAir is composed of basal cells, cilial cells, goblet cells.MucilAir-HF is a co-culture model with human fibroblasts on the baso-lateral side of the insert.
Cells used in this study were obtained from patients undergoing surgical polypectomy.All experimental procedures were explained in full, and all subjects provided informed consent.These studies were conducted according to the Declaration of Helsinki on biomedical research (Hong Kong amendment, 1989), and received approval from a Comission cantonale d'éthique de la recherche scientifique de Genève (15-062).All samples have been obtained with informed consent as part of studies or other processes that have had ethical review and approval.Collection protocols include i) obtaining donor medical history to be sure there are no issues that would put the donor or those that use the cells at risk, and ii) review of procedures by an Institutional Review Board.[45] More precisely, primary human bronchial/nasal epithelial cells were cultured on top of the microporous membrane of the transwell insert (Oxyphen, Wetzikon).After several days of culture, the insert was switched to air-liquid interface for at least 28 d to fully differentiate into ciliated epithelium.700 μL of basolateral culture medium (EP04MM, Epithelix Sàrl) was renewed twice a week.For the co-culture model, the lung fibroblasts were seeded onto a 6.5 mm transwell insert with differentiated epithelia, on the baso-lateral side and allowed to attach for 3 h.For the maintenance, the inserts were placed in an incubator (37 °C, air/CO 2 5% environment) and strictly treated under sterilized conditions.Pure mucus was collected by centrifugation of MucilAir placed in a 50 mL Falcon tube, at 2000 rpm for 5 min.
For this study, the primary cells of MucilAir originated from several different donors.Donor A was a 51-year-old male, non-smoker and had no pathologies reported.Information of donor B was not provided.Donor C was a 61-year-old female, African-American, non-smoker and had no pathologies reported.Donor D was a 42-year-old male, Hispanic, nonsmoker and had no pathologies reported.Donor E was a 39-year-old female, Caucasian, non-smoker and had no pathologies reported.Donor F was a 48-year-old male, non-smoker and had no pathologies reported.Their cilia beating frequency and tissue integrity (Table 1) were measured before the aerosol deposition.
Processing, Embedding, Sectioning, and Staining of MucilAir: Mucus layers were removed from the epithelial tissue before tissue fixation to avoid any permeation of UCNPs during the sectioning process (Experiments 6 and 7).Each insert of MucilAir was placed in a 50 mL Eppendorf tube and four inserts were placed in the centrifuge (Eppendorf Centrifuge 5702) for 2 min at 2000 rpm.Successively, each insert was placed in a clean 24 well plate and were fixed by complete immersion in formaldehyde 4% solution in PBS for 20 min at room temperature.A paraffin block was prepared after detaching the membrane from the plastic transwell chamber.The block was then cut using a microtome (Leica RM2135) into 5 μm thick slices that were mounted on glass microscope slides.As a result, 800 slides, each with four MucilAir inserts, were prepared.Successively, 50 slides were randomly selected from 200 consecutive slides of each Mu-cilAir insert for imaging.100 slides of SLB-UCNPs deposited MucilAir and 100 slides of LF-UCNPs deposited MucilAir were analyzed by multiphoton microscopy to detect the presence of upconversion signal.After the multiphoton imaging procedure, the sections were deparaffinized with Histo-SAV and stained with Periodic Acid and Schiff reagent.Bright field images of each section were acquired using a widefield microscope (Nikon Eclipse 80i).
Aerosol Exposure: Stock solutions of LF-UCNPs were sonicated for 10-15 min and diluted to the desired concentration of 2.0 mg mL −1 in deionized water.The dispersions were again sonicated for 10 min prior to aerosol deposition.The SLB-UCNP stock solution was vortexed for 10 s prior to the dilution to 2.0 mg mL −1 and immediately used for aerosol deposition.To ensure aerosolized deposition stability, 20 μL of sodium chloride 0.9% was added to each solution in accordance with manufacturer recommendations resulting in final concentrations for each nanoparticle suspension of 1.9 mg mL −1 .
SLB-and LF-UCNPs were delivered to the MucilAir inserts under identical conditions using Vitrocell Cloud as nebulizer device.Prior to deposition, up to three MucilAir inserts (Experiments 3-9) and the transwell inserts filled with 50 μL of mucus (Experiments 1 and 2) were placed in the middle of the wells.520 μL of nanoparticle solution was injected into the nebulizer, allowing the production of an aerosol cloud in the chamber for 3-5 min.After this time, the deposition chamber was kept isolated for an additional 10 min to allow for gentle nanoparticle deposition.As a result, ≈1.0 μL of UCNPs solution were delivered to each insert.
Aerosol delivery of UCNPs onto MucilAir inserts resulted in homogeneous deposition across the entire sample surface, displaying a marked difference with the results of dropwise delivery, as reported in Figure S4 (Supporting Information).
For experiment 2, the medium (PBS) under the transwell insert was collected at 6, 12, and 24 h post-treatment with UCNPs.After collection, the medium was replaced with fresh PBS after each timepoint.To ensure even distribution and homogeneous sedimentation of the nanoparticles, the collected medium from each timepoint was then sonicated for 10 min before being deposited into a clean 24-well cell culture plate.The plate was left to stand for 48 h before imaging the underside of the plate.The same protocol was conducted in Experiments 8 and 9 at 24 h post-treatment with UCNPs.
Multiphoton Microscopy and Image Analysis: In Experiment 3-5, epithelial tissue images were acquired using a Leica SP8DIVE coupled with a Spectra Physics Insight X3+ultrafast tunable laser (120fs, 80 MHz, 680-1300 nm) equipped with a HC FLUOTAR L 25X/0.95W VISIR objective and a set of Hybrid PMT detectors.Upconversion and autofluorescence signals were epi-collected within the 430 -480 nm and 530 -600 nm wavelength ranges upon excitation at 980 and 800 nm, respectively.2D images were acquired using 1024 × 1024 pixels and a dwell time of 21.5 μs.The dwell time was established as a balance between sensitivity, scanningrelated elongation of long-lived UCNP emission and overall acquisition time (Figure S3, Supporting Information).Three-dimensional images were obtained by Z-stacking a 2D collection of consecutive scans with 3 μm step-size.In experiments 3, 4, and 5 (Table 1), images were acquired from different inserts to avoid any tissue damage due to a long image acquisition time.
To optimize image quality, upconversion signals in Figure 2, Figures S5 and S6 (Supporting Information) were acquired under different laser powers.Acquisition settings are provided in Table S1 (Supporting Information).
In Experiments 6 and 7, histology slides were imaged with Nikon A1R multiphoton microscope equipped with a Spectra Physics MaiTai DeepSee Ti:Sapphire laser oscillator (100 fs, 80 MHz, 690-1040 nm).Detection of upconversion signals were obtained by using a 480-30 band pass filter and autofluorescence signals were obtained by using a 520-40 band pass filter under 980 nm excitation.
In Figure 1h, hyperspectral images were acquired using a hyperspectral detection unit in a Nikon A1R multiphoton microscope from four different regions in PBS below two MucilAir inserts.Spectral intensity within the 430-485 nm range was then summed in each image after subtracting the background signal and their average value and standard error was calculated.The background signal was defined from the signal of the bottom of the well plate filled with PBS in the absence of nanoparticles.Background signal subtraction resulted in negative values for the spectral intensity of LF-UCNPs.Values at 12 h are the summation of signals acquired between 0-6 h and 6-12 h.Values at 24 h are the accumulation of the entire time span of the measurement.
In order to assess the upconversion and autofluorescence signals shown in Figure 2c,d, the average intensity of two channels were initially computed for each frame within the 3D image stacks.The background signals were determined as the lowest average intensity among the image stacks treated with SLB and LF-UCNPs at the corresponding time point and subtracted from the 3D image stacks.
For assessing UCNPs permeation through the sample, as reported in Figure 3b, for each experimental condition (SLB-and LF-UCNPs, mucus and no mucus) four 1.2 × 1.2 mm 2 hyperspectral images of the bottom surface of the medium below MucilAir (using two inserts per condition) were acquired.The bars in the plots on the right side of the panel showed the spectral intensity summed within the 430-485 nm interval and averaged among each set of four images.The error bars correspond to the standard error calculated in the same way.
The average number of upconversion signal occurrences in a MucilAir slide is summarized graphically in Figure S9 (Supporting Information) and was determined as reported in Figure S10 (Supporting Information).
The thickness of the mucus layer was determined by measuring the vertical distance between the highest point of UCNP signal detection, assumed to represent the top layer of the mucus, and the highest point of autofluorescence detection, indicating the top of the epithelium.This measurement was conducted using images taken 2 h after the aerosol deposition (Experiments 3, 4, and 5, Table 1, main text).Histology staining was performed on the MucilAir insert of donor D without removing the mucus layer on top, followed by staining with Periodic Acid and Schiff's reagent.This staining process was employed to assess and determine the thickness of the mucus layer (Figure S11, Supporting Information).

Figure 1 .
Figure 1.a) Diagram of the MucilAir human-derived model in a transwell insert.The images displayed in Figure 2 (experiment 3) were acquired at the center (region A) and at the border (region B) of the insert as shown here.b) MucilAir cross sectional illustration.gob: goblet cells, ci: ciliated cells, bas: basal cells, PM: porous membrane, CM: culture medium c) TEM image of LiYF 4 :Yb 3+ ,Tm 3+ UCNPs and d) corresponding upconversion emission spectra.The orange trace is collected from a colloidal dispersion of UCNPs (1 mg/mL) upon continuous wave (CW) 980 nm excitation and acquired by a high-resolution spectrometer.The blue trace is the average signal extracted from a laser-scanning microscopy image of dried LiYF 4 :Yb 3+ ,Tm 3+deposited on a microscopy substrate upon femtosecond pulse excitation at 980 nm.The upconverted emission is epi-collected and directed into the microscope hyperspectral detection unit, with spectral resolution set at 2.5 nm.e) Schematic of the configuration used for the mucus-only measurements (experiments 1 and 2).Hyperspectral images at 2 h after the aerosol deposition of f) SLB-and g) LF-UCNPs on the top of the mucus layer.Images were acquired within a 400-590 nm spectral range with spectral resolution of 6.0 nm upon 980 nm excitation.Signals at 430-485 nm correspond to UCNPs; 520-590 nm correspond to two-photon excited fluorescence (TPEF) from the porous membrane upon 800 nm excitation.The dark region between the UCNPs and the membrane defines the volume occupied by the mucus.h) Time-resolved trace of the cumulative signal of UCNPs collected in phosphate buffered saline (PBS) below the porous membrane as a function of time after exposure to SLB-UCNPs (red trace) and LF-UCNPs (yellow trace).Data shown in the traces is normalized to the maximum cumulative emission intensity (12 h, SLB-UCNPs).

Figure 2 .
Figure 2. Images of aerosol-deposited a) SLB-UCNPs and b) LF-UCNPs acquired 2 and 24 h after delivery.The labels (SLBN, LFN (N = 1-3)) allow for tracking the specific sample used for each measurement.The upconversion signal (blue) is collected in the 430-480 nm region upon 980 nm excitation.The tissue autofluorescence (green) is collected in the 530-600 nm region upon two-photon excitation at 800 nm.On the right of each image panel, we report the transverse maximum intensity projection.The crosshatched areas were inspected before acquiring the volumetric image and no signals from the UCNPs were detected.Penetration kinetics of c) SLB-and d) LF-UCNPs in epithelial tissue at different intervals after delivery of UCNPs.Green filled areas: autofluorescence from epithelial tissue and porous membrane.Pink and yellow filled areas: UC signal of SLB-and LF-UCNPs, respectively.The zero of the depth axis is set at the maximum of the autofluorescence peak, corresponding to the position of the porous membrane.

Figure 3 .
Figure 3. a) Images of histological sections (bright field (left) and multiphoton (right)) of MucilAir inserts treated with SLB-UCNPs.The upconversion signal (blue) was obtained by using a 480-30 band pass filter and tissue autofluorescence (green) was obtained by using 520-40 band pass filter under 980 nm excitation.SLB-UCNPs were detected within the mucus layer (A), in the periciliary region (B) and within epithelial cells (C).b) Transepithelial transport of SLB-(pink) and LF-UCNPs (yellow) across the MucilAir insert with and without preliminary mucus removal.Large field of view (1.2 × 1.2 mm 2 ) images of the bottom of the culture medium acquired 24 h after deposition of the UCNPs.The bar plots display the average UC signal intensity for each experimental condition.Data are normalized to the value obtained for SLB-UCNPs in a sample with mucus.

Table 1 .
Summary of the parameters associated with each experiment completed in this work.The corresponding donor information can be found in the supporting information.