Most vaccines including influenza are administered in liquid form using a hypodermic needle. In contrast, microneedle-based vaccination involves coating vaccine onto microneedles using a drying process, and thus represents the delivery of vaccine in a solid state (Fig. 1A). This drying process could damage antigen stability. Using formalin-inactivated whole influenza virus as a model antigen, we determined HA activity as an indicator of antigen structural integrity and vaccine stability, after coating influenza vaccines onto microneedles. As shown in Fig. 1B, vaccine coated onto microneedles was re-dissolved into PBS solution by soaking the coated needles in PBS. HA activity of the reconstituted influenza vaccine was then determined. After the drying process during coating, the influenza vaccine was found to lose most of its HA activity when reconstituted in PBS (Fig. 1B). To address this, we noted in the literature that trehalose, a disaccharide, was reported to stabilize an influenza subunit HA vaccine during freeze drying [26]. Guided by this, we found that the addition of 15% trehalose to the coating solution significantly improved the retention of HA activity of microneedle-coated vaccines up to 64% compared to an untreated vaccine (Fig. 1C). These results suggest that drying of the inactivated influenza virus has a significant damaging effect on vaccine integrity and that the addition of trehalose can largely stabilize the antigen HA activity during coating onto microneedles.

We next investigated the relationship between retention of HA activity of influenza vaccine and its immunogenicity after microneedle vaccination using both solid state formulations administered using MN or liquid state formulations intramuscularly. Groups of mice (n = 12) were immunized with a single dose of vaccine in the skin after coating the vaccine in the absence (MN) or presence of trehalose (MN+T) in the coating formulation (0.4 µg of viral protein). Also, to compare with the conventional intramuscular route of immunization (IM), two more groups (n = 12) were immunized with inactivated viral vaccine that was coated onto microneedles in the absence (IM) or presence of trehalose (IM+T), dissolved off the microneedles and then injected intramuscularly. A mock immunized control group with microneedle vaccination of coating buffer only in the skin was also included (n = 12).

We determined vaccine (A/PR8 virus) specific total IgG and isotypes IgG1, IgG2a and IgG2b antibodies in sera at weeks 1, 2 and 4 after a single IM or microneedle vaccination in the skin. Trehalose-stabilized microneedle vaccination induced the highest levels of antibodies among the groups (Fig. 2). Trehalose stabilized vaccines (MN+T, IM+T) induced significantly higher levels of virus-specific antibodies than those in the corresponding groups without trehalose (MN, IM) at weeks 1, 2, and 4 after a single immunization (p<0.005). Among the trehalose-stabilized groups, total IgG antibody responses were higher in the trehalose microneedle group (MN+T) than those in the trehalose-stabilized IM group (IM+T) at week 4 post immunization (p<0.05). For comparison, the levels of antibody responses after intramuscular immunization with unprocessed influenza vaccine were similar regardless of the addition of trehalose to the vaccine (Table 1), which suggests that trehalose does not behave as an adjuvant, but rather acts to protect the antigen from damage during microneedle coating.

The pattern of antibody isotypes indicates either T helper type 1 (Th1) or type 2 (Th2) like immune response corresponding to dominance of the IgG2a or IgG1 isotype antibody, respectively [27]. When isotype-switched antibodies were examined, we found a very striking pattern among the groups studied (Fig. 3 and Table 2). The ratios of IgG2a to IgG1 (IgG2a/IgG1) were analyzed based on isotype antibody levels determined by optical densities (Fig. 3E) and end-point dilution titers (Table 2). Both analyses showed similar results. The trehalose-stabilized vaccines (MN+T, IM+T) induced IgG2a as the dominant isotype (Fig. 3B, 3D, 3E, and Table 2). A similar pattern of IgG2a dominant immune responses was previously observed in mice intramuscularly immunized with intact influenza vaccine [28], [29]. In contrast, trehalose-negative vaccine groups (MN, IM) showed IgG1 as the dominant isotype antibody (Figs. 3A, 3C), which resulted in a switch in IgG1/IgG2a ratio of >1 for trehalose-containing groups (IM+T, MN+T) to <1 for groups without trehalose (IM, MN). Therefore, these results suggest that the retention of HA activity by trehalose in the influenza vaccine significantly affected the magnitude of humoral immune responses as well as the pattern of isotype switching.

Levels of functional antibodies measured as hemagglutination inhibition (HAI) and/or virus neutralizing activities in immune sera are known to be better immune correlates for protection [1], [30]. As shown in the Fig. 4, highest titers of HAI and neutralization activity were induced by microneedle vaccination in the skin using stabilized vaccine (MN+T). As shown in Fig. 4A, mice immunized with the trehalose-stabilized vaccines showed considerably higher neutralizing activities, showing a 6-fold comparing MN and MN+T groups (270 vs 1620 titers of 50% plaque reduction, p<0.005) and 2-fold between IM and IM+T groups (270 vs 540 titers of 50% plaque reduction, p<0.05). Similarly higher HAI titers were observed in the trehalose-stabilized vaccine compared to corresponding groups without trehalose in the formulation (Fig. 4B, p<0.001 between MN and MN+T; p<0.05 between IM and IM+T). These results indicate that microneedle delivery of stabilized vaccines can be superior to IM immunization in inducing neutralizing and hemagglutination inhibiting antibodies, and that maintaining HA functional activity in the influenza vaccine is critical to inducing high levels of functional antibodies.

To determine whether vaccine stability measured by HA activity or the routes of immunization (i.e., microneedle versus intramuscular) affects protective immunity, vaccinated mice including a mock control were challenged with a lethal dose of A/PR8 virus (20× LD₅₀) at 5 weeks after a single microneedle vaccination in the skin or intramuscular immunization (Fig. 5). All mock-immunized control mice rapidly lost body weight and died by day 6 post lethal-challenge. Groups of mice immunized with either inramuscular or microneedle vaccination without trehalose formulation (MN and IM) showed significant illness as determined by loss in body weight up to 15 to 17% whereas the intramuscular group with trehalose (IM+T) showed a only transient body weight loss of 5%. Importantly, the trehalose-microneedle group (MN+T) did not show any loss in body weight. As expected from body weight changes, survival rates of groups without trehalose were lower (70% for MN and 80% for IM) than those groups with trehalose, which showed 100% survival (MN+T, IM+T). Therefore, these results support the conclusion that stabilization of influenza vaccine by trehalose during microneedle coating is critically important for inducing protective immunity.

To better appreciate the efficacy of protection, we determined the viral titers and inflammatory cytokine levels in lungs at day 4 post challenge infection. The groups of mice with microneedle or intramuscular vaccination (MN, IM) had lower lung viral titers by 20- and 75-fold compared to those in the mock-immunized control (p<0.01), but still showed lung viral titers in a range of from 2.9×10⁵ to 1×10⁶ pfu/g lung tissue (Fig. 6A). The IM immunized group with trehalose (IM+T) showed lung viral titers of 2.4×10⁵ pfu/g lung tissue, 100-fold lower compared to the mock control (Fig. 6A, p<0.005). Levels of lung viral titers in the group of microneedle delivery with trehalose (MN+T) were lowest and were under the detection limit (<270 pfu/g lung tissue).

Acute infection by pathogenic influenza virus causes severe lung inflammation [31], [32]. Thus, levels of inflammatory cytokines can be used as an additional parameter in evaluating protective efficacy. The control mice showed the highest levels of interferon-γ (IFN-γ) and interleukin-6 (IL-6) pro-inflammatory cytokine levels in lungs at day 4 post challenge. Microneedle delivery of trehalose stabilized vaccine (MN+T) was found to be most effective in lowering inflammatory cytokines whereas high levels of inflammatory cytokines were detected in the trehalose-negative microneedle group (MN), showing a correlation between lung viral titers and inflammatory cytokines (Fig. 6B and 6C). Also, groups of mice immunized intramuscularly with trehalose-stabilized vaccine (IM+T) showed moderate or low levels of cytokines compared to the trehalose-negative groups (p<0.05) (Fig. 6B and 6C). In summary, these results show that microneedle delivery of stabilized influenza vaccine induced superior protective immunity in controlling lung viral replication of challenge virus compared to microneedle delivery of unstabilized vaccine and to intramuscular immunization using stabilized or unstabilized vaccine.

To better understand the significantly improved viral clearance in the hemagglutinin-stabilized microneedle group (MN+T), we determined virus-specific antibody levels in lungs and sera at day 4 post challenge infection (Fig. 7). Interestingly, the group of mice that received microneedle vaccine formulated with trehalose (MN+T) showed highest levels of both IgG1 and IgG2a antibodies specific to virus in lungs at day 4 post challenge infection (p<0.001) (Fig. 7A). Microneedle vaccination without trehalose (MN) induced moderately high levels of IgG1 antibody as a dominant isotype in this group. In contrast, inramuscular immunized groups (IM, IM+T) did not induce significant levels of recall antibody responses specific to virus in lungs.

Higher levels of both recall serum IgG1 and IgG2a antibodies were also observed in the group immunized by microneedle delivery with trehalose-formulated vaccine (MN+T) than those in the corresponding inramuscular immunization groups (p<0.05 between MN+T and IM+T for both IgG1 and IgG2a) (Fig. 7B and Table 2). Increases in serum IgG1 antibodies were also prominent in the group immunized by microneedle delivery after challenge infection (p<0.005 between MN and IM or IM+T for IgG1, Fig. 7B).

To determine T cell responses, spleen cells were harvested from immunized mice at day 4 post challenge infection. After in vitro stimulation of splenocytes with A/PR/8/34 hemagglutinin-specific MHC I or MHC II peptides, cytokine secreting cells were quantified (Fig. 7C). The microneedle vaccine formulated with trehalose (MN+T) showed the highest levels of MHC II peptide-specific IFN-γ secreting splenocytes (p<0.05 among the immunized groups). In the inramuscular groups (IM, IM+T), there were no significant differences in levels of IFN-γ secreting splenocytes with or without trehalose formulation.

The lung is the primary site for replication of influenza virus, and neutralizing antibodies in lungs are expected to play a significant role in clearing the viruses. Thus, neutralizing titers were determined in lung samples collected at day 4 post challenge (Fig. 7D). The highest neutralizing titers of 45 with 80% plaque reduction were detected in lung samples after microneedle vaccination with trehalose formulation (MN+T), which was more than 3-fold higher compared to those in the corresponding unstabilized or other groups. Without trehalose stabilization, both inramuscular and microneedle groups (IM, MN) showed lower levels of neutralizing titers in lungs. These lung neutralizing titers showed correlations with lung viral titers post challenge as shown in Fig. 6A. These results suggest that microneedle delivery of vaccine stabilized by trehalose is superior in inducing rapid recall mucosal and systemic immune responses upon challenge infection compared to microneedle delivery of unstabilized vaccine and to intramuscular immunization using stabilized or unstabilized vaccine.

Influenza virus, A/PR/8/1934 (H1N1, abbreviated as A/PR8), was grown in 10-day old embryonated hen's eggs and purified from allantoic fluid by using a discontinuous sucrose gradient (15%, 30% and 60%) layers. The purified virus was inactivated by mixing the virus with formalin at a final concentration of 1∶4000 (v/v) as previously described [51], [52]. Inactivation of the virus was confirmed by plaque assay on confluent monolayer Madin-Darby canine kidney (MDCK) cells and inoculation of the virus into 10-day old embryonated hen's eggs. Inactivated whole virus vaccine (A/PR8) was used to coat solid metal microneedles for vaccination in the skin. For use in challenge experiments, mouse adapted A/PR8 virus was prepared as lung homogenates from mice infected 4 days earlier [53], [54].

Stainless steel microneedles of approximately 700 µm in length (Fig. 1A) were fabricated by laser cutting and electro-polishing technology as described previously [23]. A manual dip-coating device was used to coat a vaccine onto microneedles by dipping and air-drying six times at 25°C [23], [55]. The vaccine coating solution was composed of 1% (w/v) carboxymethylcellulose sodium salt (Carbo-Mer, San Diego, CA), 0.5% (w/v), Lutrol F-68 NF (BASF, Mt. Olive, NJ), and inactivated virus (A/PR8, 1 mg/ml based on total protein contents) in phosphate-buffered saline (PBS) with or without 15% (w/v) trehalose (Sigma Aldrich, St. Louis, MO). Half of the microneedles coated with inactivated virus vaccine were de-coated to reconstitute vaccines in PBS for intramuscular immunization controls and the other half were used for microneedle immunization (Fig. 1B).

To determine the amount of inactivated virus vaccine coated on microneedles, vaccine-coated microneedles were incubated in PBS for 12 h at 4°C and the total protein content of the reconstituted vaccine from an array of five microneedles was measured by a DC protein assay kit (Bio-Rad, Irvine, CA). The retained HA activities were also determined in the reconstituted vaccines. Dissolved vaccines in PBS (50 µl) was serially diluted in 50 µl of PBS mixed with an equal volume of a fresh 0.5% suspension of chicken red blood cells (Lampire Biological Laboratories, Pipersville, PA) and incubated for 1 h at 25°C. The titers were determined as the endpoint dilutions inhibiting the precipitation of red blood cells.

Female inbred BALB/c mice (Charles River, Wilmington, MA) aged 6 to 8 weeks were used. Groups of mice (12 mice per group) were immunized with coated microneedle vaccine (0.4 µg total protein) for delivery to the skin or reconstituted vaccines (0.4 µg protein/100 µl) for intramuscular immunization in the upper quadriceps muscles of mice (both legs each with 50 µl). The four groups of immunized mice were designated as trehalose-negative formulated microneedle vaccine (MN), trehalose-positive formulated microneedle vaccine (MN+T), trehalose-negative reconstituted vaccine (IM) and trehalose-positive reconstituted vaccine (IM+T) for intramuscular injection. For microneedle delivery, mice were anesthetized with ketamine (110 mg/kg, Abbott Laboratories, Chicago, IL) mixed with xylaxine (11 mg/kg, Phoenix Scientific, St. Joseph, MO). Hair on the dorsal surface of mice was removed by a hair-removing cream (Nair, Church and Dwight Company, Princeton, NJ) with a moisturized cotton stick. After cleaning with a soaked cotton ball (70% ethanol) and drying with a hair dryer, an array of vaccine-coated microneedles was inserted into the skin and left in place for 10 min for release of the vaccine antigen from the microneedle.

For challenge infections, mice lightly anesthetized with isoflurane were intranasally infected with a lethal dose of A/PR8 virus (20× LD50) in 50 µl of PBS at 5 weeks after a single dose immunization. Mice were observed daily to monitor changes in body weight and to record mortality. Mice with >25% loss of body weight were sacrificed to prevent undue suffering. These studies were approved by Emory University IACUC.

Influenza virus-specific antibodies of different isotypes (IgG, IgG1, IgG2a and IgG2b) were determined by enzyme-linked immunosorbent assay (ELISA) plates coated with A/PR8 viral antigen and by using anti-mouse IgG isotype specific secondary antibodies as described previously [30], [54]. Antibody concentrations (ng per g lung tissues or ng per ml sera) were determined using standard curves for mouse IgG1 and IgG2a antibodies. For determination of hemagglutination-inhibition (HAI) titers, serum samples were first treated with receptor destroying enzyme (Denka Seiken, Kayabacho,Chuo-ku,Tokyo) by incubation overnight at 37°C, and then incubated 30 min at 56°C. Sera were serially diluted, mixed with 4 HA units (HAU) of influenza A/PR8 virus, and incubated for 30 min at room temperature prior to adding 0.5% chicken red blood cells. The highest serum dilution preventing hemagglutination was scored as the HAI titer as described [30].

Virus neutralization assay was performed using MDCK cells (American Type Culture Collection, VA, USA) following a previously described procedure [30], [54]. The neutralization activity was expressed as the percentage of plaque reduction. Lung viral titers at day 4 post challenge were determined by counting plaques formed on the MDCK cells as previously described [30], [54]. Inflammatory cytokines (IL-6, IFN-γ) in lungs collected at day 4 post challenge were analyzed by Ready-Set-Go cytokine kits (eBioscience, San Diego, CA) following the manufacturer's procedure as previously described [54].

Recall immune responses were determined from serum, lung, and spleen at day 4 post-challenge. To determine T cell responses, spleen cells were harvested from immunized and mock control mice at day 4 post-challenge and were used to determine cytokine producing T cell responses by ELISPOT as previously described [54]. Briefly, spleen cells (10⁶ cells per well of 96 well plates) were stimulated in vitro with A/PR8 hemagglutinin specific peptides [56], [57]; a mixture of two major histocompatibility complex class I (MHC-I) peptides (IYSTVASSL and LYEKVKSQL) or a pool of five MHC-II peptides (SFERFEIFPKE, HNTNGVTAACSH, CPKYVRSAKLRM, KLKNSYVNKKGK, and NAYVSVVTSNYNRRF). Regents for cytokine ELISPOT were purchased from BD/PharMingen (San Diego, CA).

All parameters were recorded for individuals within all groups. A two-tailed Student's t-test was performed when comparing two different conditions. Statistical comparisons of three or more conditions were carried out using the correlation and regression test of the PC-SAS system (SAS Institute, Cary NC). A P value less than 0.05 was considered to be significant.