Diphtheria toxoid and TT (for in vitro assays) were obtained from Staten Serum Institute (Copenhagen, Denmark) and imiquimod (IMQ) Vaccigrade was obtained from Invivogen. Trifluoroacetic acid (TFA), 3,3′,5,5′-tetramethylbenzidine (TMB) and bovine serum albumin (BSA), and 0.4% (w/v) were obtained from Sigma Aldrich. High-performance liquid chromatography (HPLC)-R grade acetonitrile (ACN) was from Biosolve and phosphate-buffered saline (PBS; pH 7.2, 1.5 mM KH₂PO₄, 2.7 mM Na₂HPO₄–7H₂O, and 155 mM NaCl) was from Gibco (ThermoFisher Scientific). IRdye800cw carboxylic acid N-hydroxy succinimide ester (IRdye800cw-NHS) was purchased from Li-cor (Lincoln, NE, USA). Dexdomitor was purchased from Orion Corporation, Narketan ketamine from Vétoquinol, and Atipam was purchased from Dechra. Goat anti-mouse immune globulin G (IgG) total horseradish peroxidase (HRP) (GaM-IgG total HRP), goat anti-mouse IgG1 HRP (GaM-IgG1 HRP), and goat anti-mouse IgG2a HRP (GaM-IgG2a HRP) were obtained from Southern Biotech and microtiter plates 9610 used for ELISA were from Corning Costar.

Nanoporous microneedle arrays were produced by using a double replication technology as previously described (12). In brief, from an inverse silicon master a first positive PDMS mold was created, from which a second inverse PDMS mold was produced. Alumina npMNAs were fabricated at LouwersHanique B.V. from the second PDMS mold according to MyLife Technologies’ proprietary manufacturing procedure (18), using a slurry that contains alumina nanoparticles with an average size of 300 nm and a plasticizer. After controlled drying, the resulting MNAs were removed from the PDMS mold and were sintered at 1,450°C. This results in removal of the plasticizer and the formation of nanoporous alumina material with an average pore size of 80 nm and a porosity of approximately 40% (10, 12). Microneedles used in this study had a length of 475 µm and a density of 150 microneedles/cm² on a back plate of 0.7 cm² (105 microneedles/array; Figures 1A,B). The total volume in the nanopores of only the tips of the microneedles of a single MNA was calculated to be 0.25 µL. Bruker Nano Surface analysis was performed to characterize the geometry of the npMNAs.

To only load the tips of the microneedles, microneedles were pierced through a foil (Parafilm^®) by using a UAFM-V1 electrical applicator (uPRAX Microsolutions) at a velocity of 65 cm/s. Next, a drop of 5 µL drug formulation was applied onto the foil-pierced npMNAs to absorb a drug/vaccine formulation into the microneedle tips. After 5 s, the surplus drop of drug formulation and the foil were sequentially removed from the npMNAs. To confirm that only the microneedle tips can be loaded with a drug formulation, the tips of npMNAs were loaded with a 0.4% trypan blue solution as described earlier.

To test the piercing ability, a npMNA was applied trice on the dorsal side of ex vivo murine ears (Balb/C), which were collected from surplus mice, by using 3D-printed uPRAX impact applicators (Figure 1C) having an average holding force of 4.08 ± 0.75 N (mean ± SD, n = 19). Subsequently, the three pierced ex vivo mouse ears were incubated with 50 µL of a 0.4% trypan blue solution at room temperature. After 30 min, the trypan blue was removed, and the ears were washed in 10 mL PBS. Finally, the blue dots (piercings) were counted, from which the penetration efficiency was calculated.

To determine whether the npMNAs are suitable devices to load and release subunit antigens DT and TT, the hydrodynamic diameter and size distribution of DT and TT were determined by using dynamic light scattering on a Zetasizer Nano (Malvern Instruments). For these measurements, DT and TT were at a concentration of 0.8 and 0.4 mg/mL, respectively.

Imiquimod has the potential to enhance the immunogenicity of antigens in the skin (19). To evaluate how IMQ is released from IMQ-loaded nanoporous material in the presence of DT and TT, npMNAs were loaded by applying a drop (on the microneedle side of a npMNA that were not pierced through a foil) of 5 µL PBS containing only 2.5 µg IMQ, or 2.5 µg IMQ and 2.5 Lf DT or TT. Such a drop is absorbed into an npMNA within seconds because of capillary forces. Next, the IMQ-loaded npMNAs were incubated in 2.5 mL release buffer (PBS) under shaking at 300 rpm, and samples of 75 µL were taken in duplicate at different time points (1, 5, 10, 30, 60, 120, and 240 min). The IMQ concentration in the release buffer was determined by using HPLC analysis on an Agilent 1100 series HPLC equipped with a UV detector using a Phenomenex Kinetex 150 mm × 4.6 mm 2.6 μm EVO C18 column. A linear gradient of 5% solvent A (ACN with 0.1% TFA) to 68% solvent B (milliQ with 0.1% TFA) from 0 to 12 min was detected at a wavelength of 242 nm at a retention time of 9.8 min.

To assess the release of antigen from antigen-loaded nanoporous material, npMNAs (that were not pierced through a foil) were loaded with 15 µL PBS that contained 5 Lf DT or TT as described earlier. Next, antigen-loaded npMNAs were incubated in 4 mL release buffer while shaking at 300 rpm. At different time points (1, 5, 10, 30, 60, 120, and 240 min), samples of 500 µL were taken in which the released amount of antigen was quantified by measuring the intrinsic fluorescence (emission wavelength of 348 nm) at an excitation wavelength of 280 nm and using standard concentrations of DT and TT ranging from 0.01 to 50 ng/mL on a Tecan Infinite M1000 plate reader. The release of antigen in the presence of IMQ was not investigated, because IMQ is fluorescent at similar wavelengths (excitation at 260 nm and emission at 340 nm) that are used to measure the intrinsic fluorescence of the antigens (20).

To quantify the amount of DT and TT that is delivered from DT- and TT-loaded npMNAs into skin, DT and TT were labeled with a near-infrared fluorescent dye (IRdye800cw-NHS). To this end, 1 mg/mL solutions of DT and TT in a 100 mM carbonate buffer pH 8.5 were prepared. Subsequently, 1 mL of each of these solutions was added to 500 µg of IRdye800cw-NHS. After 1 h shaking (300 rpm) at 37°C, the free dye was removed and the carbonate buffer was exchanged by PBS using a Zeba™ spin desalting column with a molecular weight cutoff (MWCO) of 7 kDa (Thermo Fisher Scientific). Next, IRdye800cw-labeled antigens were concentrated approximately 50 times by using 0.5 mL Amicon (Millipore) 10 kDa MWCO filters. Finally, the concentration of IRdye800cw-labeled DT and TT was determined by using a calibration curve of non-labeled antigens and measuring the intrinsic fluorescence (as described earlier).

To quantify the delivered amount of DT and TT into murine skin, npMNAs of which only the tips were loaded (using foil piercing) with fluorescently labeled antigens were prepared by using 5 µL of 12 Lf/μL (DT) and 6 Lf/μL (TT), as described earlier. The microneedles were applied by impact application and retained onto the skin by using a uPRAX 3D-printed applicator. After 30 min at room temperature, the antigen-loaded npMNAs were removed from the ears, and their fluorescence was compared with the fluorescence of standard solutions having known amounts of fluorescently labeled DT and TT, by using a IVIS^® lumina II equipped with an ICG filter set. The intradermally delivered amounts of DT and TT were quantified by using Living Image^® software (version 4.3.1).

Subunit vaccine formulations of DT and TT for loading the microneedle tips were prepared from antigen stock solutions (2.0 and 0.7 Lf/μL, respectively). Antigen stock solutions were concentrated (6–30×) by using 0.5 mL Amicon (Millipore) 10 kDa MWCO filters. Next, the concentration of the concentrates was determined by measuring the intrinsic fluorescence as described earlier. Finally, the antigen concentration was adjusted by diluting the concentrates in PBS to a concentration of 12 and 6 Lf/μL for DT and TT, respectively.

Seven-week-old Balb/C female mice (10 mice per group) obtained from Charles River (France) were immunized with 1.2 Lf (~0.50 μg) DT and 1.5 Lf (~0.77 μg) TT, or with 0.6 Lf DT and 0.75 Lf TT adjuvanted with 0.5 µg IMQ, on days 0, 21, and 42. The vaccine was administered via a SC injection of 100 µL in the neck using traditional hypodermic needles, or by dermal administration in the ear pinnae by using microneedles of which only the tips were loaded with vaccine formulation. Before each microneedle-based immunization, mice were anesthetized with 30 mg/kg ketamine and 0.1 mg/kg Dexdomitor by intraperitoneal injection. After the microneedles were removed, the anesthetic was antagonized with 0.4 mg/kg Atipam. On each ear, a DT- and TT-loaded npMNA was applied for 30 min by using a uPRAX applicator (Figure 1C). As a negative control mice were mock immunized via a PBS-loaded npMNA. Blood samples were collected from the tail vein 1 day before each immunization and serum was obtained by centrifugation; spleens were collected at day 47 (Figure 1D).

Diphtheria toxoid- and TT-specific antibody titers were determined using ELISA. ELISA plates were coated with 0.2 µg DT or 0.2 µg TT for 30 min and then blocked with 1% BSA in PBS for 15 min. Thereafter, 50 µL of serum sample at dilutions ranging from 1:25 until 1:200 (day 0), or from 1:200 until 1:25,000 (day 20, 41, and 47) were added, for 30 min. After extensive washing with 0.01% Tween 20 in PBS (PBST), wells were incubated for 30 min with GaM-IgG total HRP (1:5,000), GaM-IgG1 HRP (1:3,000), or GaM-IgG2a HRP (1:5,000). After extensive washing with PBST, antibody titers were quantified by adding 50 µL of stock TMB. Reactions were stopped after 60 s, with 100 µL of 1 M H₂SO₄, and the absorption was measured at a wavelength of 450 nm, with a reference wavelength of 650 nm, on a Microplate reader 550 (Bio-Rad). Titers of all animals at all time points for each isotype were measured in one experiment per antigen.

From 4 optical density (OD) values (at a wavelength of 450 nm) of diluted serum samples, the EC₅₀ midpoint titers were determined using GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA, v6.05). Immunized mice showing OD values below the mean OD values + three times the SD measured for PBS treated mice were considered as non-responders and were given an arbitrary value of 0, equal to ¹⁰Log value of 1. Statistical differences between immunization groups were determined using a non-parametric one-way ANOVA Kruskal–Wallis test with a Dunn’s test for multiple comparisons, and statistical significance was presented as follows: *p < 0.05; **p < 0.01; ***p < 0.001, and ns, not significant. Ratio of IgG1:IgG2a was determined by dividing midpoint titers of individual isotypes, and if animals were considered as non-responders for one of the isotypes (see above), they were excluded from ratio analysis.

Nanoporous microneedle arrays fabricated from alumina nanoparticles as previously reported (10), were characterized for geometry and dimensions via surface Bruker analysis, which showed that the ceramic microneedles had an average length of 475 µm and a needle shaft diameter of 275 µm (Figure 2A). For economic reasons, loading of only the tip of the microneedle with vaccine is an advantage, as residual vaccine quantities in the npMNAs will be strongly reduced. To investigate the possibility to only load the microneedle tips, npMNAs were pierced through foil and tips were subsequently loaded with a trypan blue solution for visualization. Brightfield microscopy showed successful tip loading and no loading of the backplate of the npMNA (Figure 2B).

The ability of npMNAs to penetrate the skin is essential for ID antigen delivery. To determine whether the npMNAs are strong enough to penetrate the skin effectively and reproducibly, skin piercing was evaluated in ex vivo murine ear skin using a trypan blue assay (Figure 2C). Using the npMNAs resulted in an average piercing efficiency of 87 ± 17% (mean ± SD, n = 3). No visual breakage or reduction in microneedle strength or sharpness was observed. Together, these data show that the developed npMNA can be used to repeatedly penetrate the skin without breakage. To determine whether the npMNAs, having an average pore size of 80 nm, are suitable to be loaded with subunit vaccine antigens, the hydrodynamic diameter of DT and TT were determined by using dynamic light scatter. This revealed that DT and TT had a hydrodynamic diameter of 8.7 ± 2.8 nm (mean ± SD, n = 3) and 13.5 ± 5.6 nm (mean ± SD, n = 3), respectively, (Figures 3A,B). Therefore, the npMNAs should be suitable to be loaded with DT and TT into their nanopores.

After the npMNAs were loaded with either one of the antigens, the release of these antigens from the npMNAs in a release buffer was determined by measuring the intrinsic fluorescence of the antigens. After antigen-loaded npMNAs were incubated in release buffer for 30 min, 30% of both DT and TT were released from the npMNAs (Figure 3C). Besides, the release of IMQ from IMQ-loaded npMNAs was quantified after incubating them in a release buffer and using HPLC with UV detection. This revealed that approximately 50% of the npMNA-loaded IMQ was released after 30 min. Furthermore, it was observed that the co-delivery of IMQ and DT or TT did not result in a decreased release rate (Figure 3D). The release of IMQ reached a plateau at 60%, which indicates that IMQ partially adsorbs onto the npMNA. This was confirmed by incubating non-loaded npMNAs in an IMQ-containing buffer, having the same amount of IMQ as the IMQ-loaded npMNAs. The concentration of IMQ in the buffer decreased from 100 to 60% over time, showing that 0.2 µg IMQ was adsorbed onto the npMNA surface (data not shown). The effect of IMQ on the release of DT or TT could not be assessed due to interference with the fluorescence of IMQ (20). Together, these data showed that ceramic alumina npMNAs are suitable to be loaded with the subunit vaccine proteins DT and TT and the adjuvant IMQ, and that the antigens and adjuvant are released in vitro.

Next, the delivery of fluorescently labeled antigens from npMNAs into ex vivo murine skin was investigated. The antigen dose delivered into the skin was quantified after the application of the fluorescently labeled antigen-loaded npMNAs onto mouse ears. The delivery of the antigens into the ears was quantified by using infrared fluorescence imaging (Figures 4A,B) and was compared with a gradient of known amounts of fluorescently labeled antigens (Figures 4C–E). The delivery of DT was 0.61 ± 0.44 Lf (which corresponds with ~0.25 ± 0.18 μg) per MNA (Figure 4A), and the delivery of TT was 0.77 ± 0.23 Lf (~0.38 ± 0.11 μg) per MNA (Figure 4B).

Delivery efficiencies of DT and TT from the npMNAs were calculated by using the geometric values of the npMNAs. With an estimated total tip pore volume of 0.25 µL, the loaded amount of DT was calculated as 0.25 µL × 12 Lf/μL = 3 Lf/MNA. With a release of 0.61 Lf out of 3 Lf loaded, 20% delivery efficiency was achieved for DT, after 30 min of application onto the skin. For TT, an amount of 0.25 µL × 6 Lf/μL = 1.5 Lf/MNA was loaded and with a release of 0.77 Lf, the delivery efficiency was 0.77 Lf/1.5 Lf = 51%, after 30 min. For immunization studies, two arrays per mouse were used per antigen and this resulted in a delivery of 1.25 Lf (~0.5 μg) DT and a delivery of 1.53 Lf (~0.77 μg) TT. These values of delivered doses correspond with 26 and 31%, respectively, of the currently used human vaccination dose (5 Lf).

To determine whether npMNA-mediated ID delivery induces antigen-specific immunity, mice were immunized with both DT (1.2 Lf) and TT (1.5 Lf) using antigen-loaded npMNAs (ID administration) or using a needle and a syringe (SC injection). After each immunization, antibody titers in the serum were determined (Figure 1D). As expected, no DT- and TT-specific antibodies were detected 1 day before the first immunization (data not shown). At day 20 after immunization, approximately 50% of the immunized mice showed detectable IgG titers against DT, which were increased after the first boost measured at day 41. After the second boost (measured at day 47), all mice produced antibodies against DT (Figures 5A–C). IgG titers against TT were slightly higher compared with DT-specific titers (Figures 5D–F). When comparing ID administration with SC injection, no statistical differences were found for DT-specific titers. IgG titers against TT were slightly higher upon SC delivery as compared with microneedle-mediated delivery (significant after the boost and second boost) (Figures 5D–F).

For both immunization routes, also DT/TT combinations adjuvanted with toll-like receptor (TLR) 7 agonist IMQ were tested to determine whether adjuvants may modify the quality of vaccination-induced antibody responses. Antigen quantities loaded in combination with IMQ were half the antigen dose loaded into npMNAs in the absence of this TLR7 agonist (0.6 Lf DT, 0.75 Lf TT, and 0.5 µg IMQ). Despite the lower antigen dose, similar antibody titers were detected in mice immunized with adjuvanted compared with unadjuvanted vaccine, with the exception of DT-specific responses measured at day 41 where 3/10 in the adjuvanted group compared with 1/10 mice in the unadjuvanted group had failed to respond (Figure 5). Overall, observing TT-specific antibody titers, the subcutaneously injected mice showed slightly higher titers than the intradermally immunized mice.

To determine whether IMQ skews the induced DT- and TT-specific response, relative quantities of the IgG isotypes IgG2a and IgG1, which serve as markers for T helper 1 and T helper 2 type lymphocytes, respectively, were determined after boost immunization. Relative quantities could not be calculated after prime immunization because of non-responders. Overall, ratios between DT and TT-specific IgG1:IgG2a, after first and second boost immunization, indicated that IgG1, and thus Th2 cell responses, prevailed (Figure 6; Figures S1B,C and S2B,C in Supplementary Material). Remarkably, addition of IMQ enhanced vaccine-induced DT-specific IgG1 responses (day 41) in SC but not in ID immunized mice. After the second booster immunization (day 47) these differences disappeared, and similar ratios between DT-specific IgG1:IgG2a were observed in all four immunization groups (Figure 6; Figures S1B,C and S2B,C in Supplementary Material). For TT specific isotypes, the kinetics were different. While after the first boost immunization, in all mouse groups, ratios between IgG1 and IgG2a were similar, both ID immunized mice receiving the adjuvanted vaccine and SC immunized mice receiving the unadjuvanted vaccine showed increased IgG1:IgG2a ratios after boost (Figure 6; Figures S1E,F and S2E,F in Supplementary Material). Taken together, although no major differences in IgG isotype ratios between groups were observed, these data demonstrate that followed vaccination regimen induces a predominantly Th2-skewed lymphocyte response.