Materials

Gelatin B (average Mw 20 to 25 kDa. Bloom strength 75, Lot No.: MKBV7621V), Poloxamer 188 (Kolophor® P 188), 25% glutaraldehyde solution, branched polyethylene imine (25 kDa, PEI), sodium metabisulfite, trinitrobenzenesulfonic acid 5% in methanol, sodium hydrogen carbonate, lysozyme from chicken egg white, and solvents (HPLC grade) have been obtained from Sigma-Aldrich (Steinheim, Germany). Fluorescein isothiocyanate-dextran 70 kDa (FITC-dextran) was purchased from TdB (Upsala Sweden). Pall Minimate™ tangential flow filtration capsules 300 kDa were obtained from VWR International GmbH (Darmstadt, Germany). Silicon wafers used as substrates in AFM experiments were derived from Plano GmbH (Wetzlar, Germany). MLCT Cantilevers were purchased from Bruker France Nano Surfaces (Wissembourg, France).

The human cell line A549 ACC107 [20] (DSMZ, Braunschweig, Germany) was cultivated in RPMI 1640 (Roswell Park Memorial Institute 1640, gibco™, Fisher Scientific, USA) supplemented with 10% fetal bovine serum (South American origin, PAN-Biotech, Aidenbach, Germany) and 1% v/v antibiotics (penicillin (10.000 U/mL)/streptomycin (10.000 µg/mL), Gibco™, Fisher Scientific, Waltham, MA, USA).

The human alveolar epithelial lentivirus immobilized cell line hAELVi [21], was obtained from InSCREENeX GmbH (Braunschweig, Germany) and cultured in SAGM™ (Small Airway Epithelial Cell Growth Medium BulletKit™, Lonza, Basel, Switzerland) supplemented with 1% v/v fetal calf serum (South American origin, Superior; Biochrom, Berlin, Germany) and 1% v/v antibiotics (penicillin (10.000 U/mL)/streptomycin (10.000 µg/mL), Gibco™, Fisher Scientific, Waltham, MA, USA). All plastic devices used for hAELVi culture were precoated with a 1% v/v fibronectin (Corning, NY, USA)–collagen (Sigma-Aldrich, Darmstadt, Germany) solution.

Methods Preparation of gelatin nanoparticles

The preparation was based on the nanoprecipitation method described in [17]. In brief, 20 mg gelatin was dissolved in 1 mL of deionized water at 50 °C and added to the antisolvent phase, consisting of 2.8% poloxamer 188 dissolved in a mixture of acetone and deionized water in a ratio of 15:1, with an injection rate of 0.25 mL/min using a syringe pump (Legato 200, KD Scientific Inc., Holliston, MA, USA). Immediately after completion of GNP formation, the freshly prepared particles were stabilized by adding 537 µL of 1.74% glutaraldehyde in acetone. After incubation times of 0.5, 1.0, 2.0, and 3.0 h the crosslinking reaction was stopped [22] with sodium metabisulfite, which forms a complex with glutaraldehyde, thus, terminating the crosslinking process [23]. The complex, unreacted material, and excess poloxamer 188 were removed by purification via tangential flow filtration (Minimate TFF capsule, 300 kDa, Pall Corporation, Port Washington, New York, USA). The remaining poloxamer amount was determined with a colorimetric method [24]. After six purification cycles, all free poloxamer had been removed. For the preparation of FITC-dextran-loaded GNPs, respective amounts of a 5 mg/mL FITC-dextran 70 kDa, 5 mg/mL FITC-dextran 150 kDa, or lysozyme stock solution was added to 20 mg gelatin dissolved in Milli-Q® water giving a final concentration of 20 mg gelatin per milliliter containing 1, 2, or 3 mg drug per milliliter. All following preparation, purification, and characterization steps were performed as described for blank gelatin nanoparticles.

Crosslinking degree

Glutaraldehyde is a frequently used crosslinker in the preparation of GNPs. Each molecule of glutaraldehyde reacts with two free amine groups [25]. Translating this to gelatin, it randomly reacts with the free aliphatic amine of lysine side chains. Therefore, it is appropriate to quantify the degree of crosslinking by determining free lysine in the particles. Colorimetric determination with trinitrobenzenesulfonic acid (TNBS) is a common method [26]. Under alkaline conditions, the primary amines and TNBS react to trinitrophenyl derivatives and sulfite. The formed compound can be quantitatively measured by UV–vis spectrometry at a wavelength of λ = 349 nm related to the number of primary amines present in the equal amount of uncrosslinked gelatin. In brief, the method was as follows: 10 to 12 mg of freeze-dried particles was resuspended in 1 mL of a 4% NaHCO3 solution. Subsequently, an equal volume of 0.5 M TNBS in NaHCO3 was added. The mixture was stirred at 40 °C for 4 h. After incubation, full hydrolysis was achieved by adding 3 mL of a 6 M hydrochloride solution and autoclaving at 121 °C for 1 h. Unreacted TNBS was removed by extraction with ethyl acetate until the organic phase remained uncolored. Samples were then diluted tenfold, and the absorbance was measured at λ = 349 nm in an Infinite M200 plate reader (Tecan group, Männerdorf, Switzerland). The same procedure was followed with the same amount of pure uncrosslinked gelatin, which served as a reference.

Measurement of size and zeta-potential

Particle size and size distribution were measured based on dynamic light scattering using a ZetaSizer® Ultra (Malvern Panalytics, Malvern, United Kingdom). 50 µL of GNP dispersion was diluted 20-fold, the pH was adjusted to 7.5, and the samples were measured in the backscatter mode. The size was evaluated as z-average, and the size distribution is displayed as polydispersity index (PdI). The zeta potential was measured by mixed measurement mode phase analysis light scattering (M3-PALS). All measurements were performed in capillary cells with a technical and an experimental triplicate.

Atomic force microscopy

For AFM measurements, GNPs were electrostatically fixed on positively coated silica specimens. Samples for AFM measurements were prepared according to the following protocol: Silica wafers were cleaned in an ultrasonic bath (Elmasonic B, Elma Schmidbauer GmbH, Singen, Germany) at a frequency of 37 kHz in pure ethanol for 5 min. Subsequently, wafers were dried under a flow of nitrogen and covered with an 1% aqueous solution of branched polyethyleneimine (PEI), 25 kDa, for 15 min. To remove unbound PEI, the wafers were rinsed with deionized water. Coated wafers were kept in deionized water and used on the same day. GNPs were incubated for 1 min to allow for a sufficient nanoparticle deposition without overloading the substrate surface. The supernatant was washed away with deionized water, and the samples were subsequently kept in liquid and measured on the same day.

AFM measurements were carried out with a JPK NanoWizard® 3 AFM (JPK Instruments, Berlin, Germany) using the MLCT cantilever tip D (Bruker France Nano Surfaces, Wissembourg, France) with a nominal resonance frequency of 15 kHz and a spring constant of 0.03 N/m. Before each measurement, the actual sensitivity and the spring constant of the used cantilever were calibrated on a cleaned silica wafer by the thermal noise method by Hutter et al. [27] using a correction factor of 0.251. The data was acquired using the quantitative imaging mode (QI™) with image sizes of 5 × 5 µm and a resolution of 128 × 128 pixels. A loading force of 1 nN was applied and the pixel time was set to 50 ms per pixel. The setup allowed for imaging and force measurements in parallel. Thus, particles were not exposed to any forces before measuring the elastic values, reducing any measuring artefacts due to possible time-dependent elastic behavior.

Data processing of AFM

AFM data were processed using the JPK SPM Data Processing program (DP) version 6.1.111. Force curves were extracted from the generated files, and four force–distance curves per particle were selected from pixels representing the middle of single nanoparticles. For the values of Young’s moduli, the respective curves were treated as follows: The determined spring constant and sensitivity must be applied to calibrate the cantilever deflection. To correct the vertical offset, a baseline subtraction is applied, and the contact point is determined. The curve obtained during the AFM experiment is based on the piezo movement. The piezo movement is larger than the indentation into the nanoparticle as the cantilever bends to the opposite direction. This effect is corrected by the so-called tip–sample separation, leading to the actual indentation curve. Subsequently, the force–distance curves are fitted by the Hertz equation modified to use square-pyramidal probes according to Bilodeau [28]. Thus, Young’s moduli are obtained. Subsequently, averages of the selected curves were calculated per particle. Therefore, per batch at least 30 particles where selected. Each particle represents the mean value of four pixels located in the middle of the particle resulting in at least 120 force–distance-curves per batch. Finally, the mean value of each formulation was determined using OriginPro software.

Determination of FITC-dextran loading

The loading of FITC-dextran was determined by dissolving the gelatin network using enzymatic degradation. The fluorescence intensity of the resulting solution was measured [18]. Therefore, 5 mg of freeze-dried GNPs were dispersed in 5 mL PBS containing 2.5 mg trypsin at room temperature. After 6 h of incubation, samples were centrifuged. The fluorescence intensity of supernatants was measured with the spectrofluorometer Duetta™ (Horiba Europe GmbH, Oberursel, Germany) using an excitation wavelength of 493 nm and an emission wavelength of 519 nm. Concentrations were calculated using a calibration row made by different FITC-dextran concentrations in PBS.

Determination of lysozyme loading

The loaded amount of lysozyme was investigated using the indirect method by reverse-phase high-pressure liquid chromatography (HPLC) with a method previously reported and validated by our group [12] using an Ultimate 3000 HPLC System (Thermo Fisher Scientific, Waltham, MA, USA) with a quaternary pump, a column oven, a LichroSphere® 100 RP 18 column (Merck KGaA, Darmstadt, Germany), and a UV–vis detector. The mobile phase consisted of solvent A: 10% acetonitrile, 90% MiIli-Q® water and 0.1% trifluoro acetic acid and B: 90% acetonitrile, 10% Milli-Q® water and 0.1% trifluoro acetic acid. The flow rate was 0.8 mL/min, and a gradient was set from 100% A to 100% B in 15 min. The concentration was determined using a calibration row by different lysozyme concentrations and the resulting AUC/min extracted from the chromatogram covering the relevant concentrations.

Cell viability

For determining the cell viability after particle exposure, MTT assays were performed. Therefore, 2 × 105 cells/mL of A549 or hAELVi were seeded in a 96-well plate (Greiner Bio-one, Frickenhausen, Germany) in a medium volume of 200 µL. After 24 h, the cells were visualized by light microscopy (PrimoVert, Zeiss, Oberkochen, Germany) to ensure that nearly 100% confluence of the epithelial cells in the 96-well plate was reached. Before starting particle exposure, the cells were washed twice with Hanks Balanced Salt Solution (HBSS). Afterwards, the appropriate particle concentration in 200 µL HBSS was applied. After an incubation time of 4 or 24 h on a shaker with 35 rpm at 37 °C, the particle dispersion was removed, and the cells were washed once with HBSS. The MTT reagent (methylthiazolyldiphenyl tetrazolium bromide, Acros organics, USA) was added in a concentration of 0.5 mg/mL and the cells were incubated for 4 h on a shaker with 35 rpm at 37 °C protected from light. Subsequently, the absorbance was measured at 550 nm with a plate reader (Synergy 2, BioTek Instruments GmbH, Bad Friedrichshall, Germany). The resulting cell viability after substance exposure was calculated based on the absorbance measurements obtained from MTT assay by a percentual ratio to the positive control of 1% Triton X-100 (PanReac AppliChem ITW Reagents, Darmstadt, Germany) and the negative control treated with HBSS according to Equation 1.

(1) Statistical analysis

If not stated differently, results are displayed as mean values (n) ± standard deviation (SD). A minimum of three independent experiments was performed for each nanoparticle formulation. Statistical significance was tested by an analysis of variance (ANOVA). It was performed as one-way ANOVA followed by Bonferroni multiple comparisons test to assess the differences between the different examined formulations. Statistical difference was assumed when p < 0.05. Statistical evaluations were made by OriginPro Version 2019b (OriginLab Corporation, Northampton, MA, USA).