Inactivated rotavirus vaccine by parenteral administration induces mucosal immunity in mice

To improve the safety and efficacy of oral rotavirus vaccines, we developed an inactivated rotavirus vaccine (IRV) for parenteral administration. Since it remains unknown whether parenteral vaccination can induce mucosal immunity, we performed a comprehensive assessment of immune responses to IRV in mice with an adjuvant-free dissolving polymer MN patch or by alum-adjuvanted IM injection. We demonstrated that IRV induced the expression of the gut homing receptor LPAM-1 on T and B cells in spleen and mLN of vaccinated mice. MN patch IRV vaccination induced a slight Th1 phenotype while IM vaccination induced a balanced Th1/Th2 phenotype. In addition, a dose-sparing effect was seen for rotavirus-specific serum IgG and neutralizing activity for both vaccination routes. Our study is the first to show that parenterally administered IRV can induce mucosal immunity in the gut, in addition to strong serum antibody response, and is a promising candidate vaccine in achieving global immunization against rotavirus.

infection, a very small number of circulating T and B cells are activated to express LPAM-1 and the LPAM-1 + B cells were shown to secrete RV-specific antibody [21][22][23] . However, there are no defined correlates of protection for LPAM-1 expression after natural RV infection or oral RV vaccination. No studies have investigated the role of B and T cells expressing LPAM-1 after parenteral RV vaccination.
In this study, we examined systemic and mucosal immune responses to IRV vaccination in mice using an adjuvant-free, dissolving MN patch or alum-adjuvanted IM injection. We showed that IM and MN patch vaccination triggered the activation of memory B and effector memory T helper cells expressing LPAM-1 primarily in spleen and mLN. In addition, RV-specific IgG antibodies and neutralizing activities in serum were expressed in a dose-dependent manner. The results of our study provide additional evidence for furthering development of CDC-9 IRV for parenteral vaccination against RV among children throughout the world.

Results
Determination of antigen dose administered by microneedle patches. MN patches used for RV vaccination were fabricated as 100 MN arrays, containing RV antigen in the MN tips. Imaging analysis showed sharply tapered MN tips on widely tapered base structures before use (Fig. 1a, left). Based on prior experience, approximately 50% of the antigen loaded in the patches is contained in the MN tip portion that penetrates the skin. Therefore, aimed antigen content in each patch was 2 µg (to deliver 1 µg antigen) and 10 µg (to deliver 5 µg antigen). Measurement of RV antigen showed that on average, patches contained 2.4 µg and 9.2 µg RV antigen (Fig. 1a, right). After insertion into the skin of mice, the inserted tips of MN patch dissolved away, and the wide base structures that were not inserted into the skin remained (Fig. 1b, left). Measurement of residual RV antigen in the used patches showed that the average dose delivered to the skin was 1.4 µg and 4.5 µg, aiming for 1 µg and 5 µg doses, respectively (Fig. 1b, right).

IRV vaccination leads to a shift of Th and Tc cells in peripheral blood.
To determine whether IRV vaccination led to any changes in the immune cell distribution, mice were inoculated with IRV either by IM injection or by MN patches on day 0, 21, and 42 and organs before vaccination (baseline), two days after each vaccination (day 2, 23, and 44), and at study end (day 63) were harvested and examined to assess systemic (blood and spleen) and intestinal (mLN) immune cell distribution by flow cytometry ( Supplementary  Figs 2 and 3). Approximately 40% of total cells in blood, spleen, and mLN of unvaccinated animals (baseline) are CD3-expressing T cells (38% in blood, 37% in spleen, and 42% in mLN). A more detailed analysis of T cells showed 59-63% T helper (Th) cells co-expressing CD3 and CD4, and 37-40% cytotoxic T (Tc) cells co-expressing CD3 and CD8 in blood, spleen and mLN, respectively. B cells expressing B220 were present in blood (33%), spleen (39%) and mLN (24%). We also analyzed organs for the presence of granulocytes (CD11b + GR-1 + ), NK cells (NK1.1 + ), dendritic cells (CD11b + CD11c + ), and myeloid cells (CD11b + F4/80 + ) and detected no elevations. We observed no differences in the distribution of T cells, B cells, dendritic cells, myeloid cells, and NK cells in blood, spleen and mLN of animals that received different IRV doses (1 µg, 5 µg) and placebo via IM or MN patch routes when compared to baseline values ( Supplementary Fig. 3). However, we observed that both vaccination routes independent of the vaccination dose induced a comparable shift in Th/Tc cell ratio towards a Th cell phenotype in blood on days 23, 44 and 63 post vaccination. No shift in Th/Tc ratio was seen in spleen and mLN of vaccinated mice.

IRV vaccination induces B and T cells expressing the gut-homing receptor LPAM-1.
Immune cells expressing the gut-homing receptor LPAM-1 have an important role in clearing infections in the intestine and intestine-associated mucosa 24,25 . To analyze if IRV vaccination led to the expression of LPAM-1 on immune cells, we analyzed B cells, Th cells, and Tc cells two days after each vaccination in blood, spleen and mLN of mice (Figs 2 and 3; Supplementary Fig. 4). Since LPAM-1 positive cells are only present in a very low percentage in naïve animals, gating strategy is shown from a representative animal receiving 5 µg IRV delivered by MN patch (Figs 2a and 3a).
We found that LPAM-1 expression on B cells of vaccinated mice depended, to a certain degree, on dosing and vaccination route (Fig. 2). LPAM-1 expression on B cells in peripheral blood was significantly elevated (p = 0.0286 for 1 µg and 5 µg IRV) compared with MN placebo controls as early as two days after the first MN patch vaccination and then declined to baseline levels (Fig. 2b). The second and third vaccinations did not boost the expression of LPAM-1 on B cells in the peripheral blood. A significant increase in the expression of LPAM-1 on B cells appeared at day 2, post dose 1, in spleen (p = 0.0175 for 5 µg) and in mLN (p = 0.0159 for 1 µg and 5 µg) after MN patch vaccination, this expression further increased post dose 2 in spleen (p = 0.0313 for 1 µg and 5 µg) and in mLN (p = 0.0161 for 1 µg and p = 0.0038 for 5 µg) and post dose 3 in spleen (p = 0.0079 for 1 µg and 5 µg) and in mLN (p = 0.0119 for 1 µg and p = 0.0159 for 5 µg). On day 63, the proportions of LPAM-1 positive B cells dropped to 11% in mLN of mice that received 1 µg and 5 µg doses by MN patch, and 7%-8% in spleen of 1 µg and 5 µg IM vaccinated animals, respectively. Similar patterns were observed in IM vaccinated animals (Fig. 2c). However, we only saw a significant increase in the proportion of LPAM-1 positive B cells in mLN from post dose 2 (day 23) to study end (day 63), mostly in mice receiving 5 µg IRV (p = 0.0159, p = 0.0161, and p = 0.0119 for day 23, day 44, and day 63 respectively). In spleen, we only observed at study end (day 63) a significant elevation of LPAM-1 positive B cells in mice receiving high (p = 0.0119) and low (p = 0.0313) dose IRV via the IM route. Administration of 1 µg or 5 µg IRV by both vaccination routes induced a comparable increase in the proportions of activated, LPAM-1-expressing B cells in blood and immune organs of mice, resulting in significantly elevated LPAM-1-expressing B cell levels in spleen (p = 0.0159 for IRV MN; p = 0.0286 for IRV IM) and mLN (p = 0.0079 for IRV MN; p = 0.0159 for IRV IM) at study end (day 63). Thus, we demonstrated a potential dose sparing effect for parenterally administered IRV.
We observed similar patterns for effector memory Th cells expressing LPAM-1 (Fig. 3). Both vaccination routes at low and high doses induced similar increase of those cells in blood on day 2 (post dose 1), which declined to baseline levels during study period. In spleen and mLN of MN patch-vaccinated mice, we observed a significant increase in LPAM-1-expressing effector memory Th cells from day 23 (post dose 2; p = 0.0119 for 1 µg and 5 µg in spleen; p = 0.0286 for 1 µg and p = 0.0159 for 5 µg in mLN) through day 44 (post dose 3; p = 0.0119 for 1 µg and p = 0.0079 for 5 µg in spleen; p = 0.0159 for 1 µg and p = 0.0079 for 5 µg in mLN) (Fig. 3b). Comparable percentages of positive cells were detected in spleen of IM vaccinated mice, whereas substantial smaller percentage of positive cells were detected in mLN of IM vaccinated mice ( Fig. 3c). At study endpoint (day 63), both vaccination routes led to significantly increased LPAM-1-expressing effector memory Th cells in spleen (p = 0.0159 for 1 µg and 5 µg IM; p = 0.0119 for 1 µg and 5 µg MN) and mLN (p = 0.0117 for 1 µg and p = 0.0114 for 5 µg IM; p = 0.0119 for 1 µg and 5 µg MN) compared to placebo controls. No differences in LPAM-1 expressing effector memory Th cells were observed in spleen and mLN between IM and MN patch vaccinations.
In addition, both MN patch and IM vaccinations induced elevated expression of LPAM-1 positive effector memory Tc cells in blood on day 2, which then declined during the study period ( Supplementary Fig. 4). LPAM-1 expression on effector memory Tc cells showed only slight increase in spleen and mLN after MN patch or IM vaccination over the time course of this study. At study end (day 63), only high and low dose MN patch vaccinated mice showed a significant increase in LPAM-1-expressing effector memory Tc cells in spleen (p = 0.0119 for 1 µg and 5 µg) and mLN (p = 0.0498 for 1 µg and p = 0.0159 for 5 µg) compared to placebo controls. No significant increase could be shown for IM vaccination routes in blood, spleen, and mLN as well as blood from MN patch vaccinated mice.

Differences in Th1 and Th2 cytokine induction after MN patch and IM vaccination.
To determine if the induction of effector memory Th cells by IRV vaccination was associated with the presence of pro-and anti-inflammatory cytokines, peripheral blood of vaccinated animals was examined for 7 cytokines (Table 1) Table 1). These data indicate that MN patch vaccination induces a slight Th1 cytokine profile, whereas IM vaccination induces both Th1 and Th2 cytokine profiles in mice.

IRV by MN patch and IM vaccination induces comparable levels of IgG and neutralizing antibodies.
We analyzed the levels of RV-specific IgG antibodies in sera of IM and MN patch-vaccinated mice collected on days 0, 21, 42, and 63 (Fig. 4a). RV-specific IgG was detected after one dose vaccination, and its levels increased after two vaccinations and remained elevated up to 63 days. Placebo vaccinated mice had no elevated IgG levels throughout the study. Of note, low dose and high dose IRV induced comparable IgG levels on days 42, and 63 for IM vaccination. MN 1 µg IRV induced slightly lower levels of total IgG antibodies compared to MN 5 µg IRV over the time course of this study, but those differences were not statistically significant. In addition, no significant differences were seen between IM and MN vaccination for both doses at study end (p = 0.0503 for 1 µg; p = 0.0527 for 5 µg). These data indicate a dose-sparing effect independent on the vaccination route.
We further analyzed sera for IgG subclasses, IgG 1 and IgG 2a (Fig. 4b). Both vaccination routes induced IgG 1 and IgG 2a in a dose-dependent manner. Overall, IM vaccination induced higher levels of IgG 1 and IgG 2a antibodies than MN patch vaccination. However, vaccination with 1 µg IRV by MN patch and IM injection led to comparable IgG 2a titers at study end. MN patch vaccination at both doses resulted in a higher IgG 2a :IgG 1 ratio than IM vaccination (Fig. 4c). Overall, these data show that MN patch vaccination with IRV induced a slightly more dominant Th1 response, whereas IM vaccination led to a balanced Th1/Th2 response. Of note, no elevated IgA levels compared to baseline control were detected in serum after MN patch and IM vaccination (data not shown). Last, we assessed the neutralizing activities in sera of IRV-vaccinated mice (Fig. 5). Mice that received two doses of 1 µg and 5 µg IRV by MN patch or IM injection had comparable levels of neutralizing activities against the homotypic RV strain Wa. For vaccination with the MN patch delivering 1 µg, no boosting effect was observed after two additional vaccinations (days 42 and 63). Of note, on day 63 neutralizing activity titer for groups receiving 1 µg dose by both MN patch and IM vaccination declined, but remained significantly above the levels of the appropriate placebo controls. Only in IM 5 µg vaccinated animals, Wa-neutralizing antibody titers remained significantly elevated (p = 0.0277).

Discussion
The present study is the first to perform a comprehensive evaluation of systemic and mucosal immunity to an IRV by IM injection and dissolvable MN patch application. IRV vaccination using MN patch and IM injection induced LPAM-1-expressing Th and B cells in spleen and mLN of vaccinated mice. Expression of LPAM-1 especially in mLN is crucial to protect the vaccinated individual from infection with gastrointestinal viruses 26,27 . While MN patch vaccination induced a comparable level of LPAM-1-expressing B cells and Th cells in spleen and mLN, IM vaccination led to a higher percentage of LPAM-1 positive B cells in mLN and more Th cells in the spleen. This is in accordance with published data that generally showed the recruitment of LPAM-1 positive memory T cells to mucosa-associated lymphoid organs 20 . We analyzed LPAM-1 positive effector memory Th cells, which are known for their potential to migrate especially to peripheral tissues including intestinal mucosa, skin or lungs 28 and can secrete effector cytokines like IFN-γ and IL-4 29 . We observed LPAM-1 induction on B and Th cells in the blood as early as 2 days post vaccination but we did not observe a boosting effect for this cell population in the blood. Most leucocytes pass through the blood stream only once during their migration 30 , which may explain the phenomenon observed in this study. LPAM-1 expression on B and Th cell is induced potentially in draining LN after contact with IRV followed by a migration towards the intestinal mucosa and the spleen. Future studies would need to examine if LPAM-1-expressing B and Th cells are migrating through the blood to mLN and spleen at other time points after IRV vaccination.
MN patches facilitate vaccination via the skin, which has the advantage of directly targeting APC and triggering a strong immune response with a low antigen dose 14,31,32 . In the present study, MN patch-and IM-administered IRV at low and high dose induced comparable humoral and cellular immune responses as well as similar induction of LPAM-1-expressing B and effector memory Th cells. Our observations of dose-sparing potential hold promise for a low cost IRV for use in global immunization.
MN patch vaccination appeared to induce higher production of IFN-γ and RV-specific IgG 2a , thus a Th1 immune response, whereas IM vaccination induced comparable Th1 and Th2 cytokines, and similar levels of IgG 2a and IgG 1 antibodies, indicating a balanced Th1 and Th2 response. Our findings are in agreement with those in mice that received RV VLPs by IM injection or IRV using a coated stainless steel MN patch 14,33 . MN patch vaccination in mice also led to a strong induction of Th1 response to an H1N1 flu vaccine and production of fewer pro-inflammatory cytokines than IM vaccination 34,35 . Th cell responses differ according to the antigen uptake and the nature of stimulating agent. Thus, it is likely that IM and MN patch administrations used here contributed to different uptake by APC and could explain the differences in Th1 and Th2 response.
We observed some reduction in neutralizing activity in MN patch-vaccinated mice at study end despite comparable titers of total IgG in IM and MN patch vaccinated animals. The reasons for this drop in neutralizing activity are not known and may reflect a property of RV vaccination in skin. However, studies with other vaccines have shown long duration of immunity after MN patch vaccination compared to IM delivery [36][37][38] . In addition, MN patch vaccination in this study was performed without an adjuvant, whereas IM vaccination was adjuvanted with alum, which has been shown to enhance and shape innate and adaptive immunity 39 . Of note, our early studies showed that second and third dose of CDC-9 IRV when formulated with alum and administered IM further boosted neutralizing activity to homotypic and heterotypic human rotavirus strains in guinea pigs and gnotobiotic piglets 18,19,40 .
In conclusion, we demonstrated that CDC-9 IRV by MN patch vaccination without an adjuvant or IM administration with alum-adjuvantation induced RV-specific systemic and intestinal immune responses, including the induction of effector memory cells expressing the gut homing receptor LPAM-1 in mice. MN patch vaccination produced more Th1 phenotype, whereas IM vaccination induced a more balanced Th1/Th2-phenotype, as indicated by cytokine and IgG subclass induction. The findings of LPAM-1-expressing B and Th cells in blood, spleen and mLN of IRV-vaccinated mice could help us investigate and define correlates of protection from parenteral RV vaccination in animals and humans. IRV by IM administration or by MN patch is a promising alternative and improvement to oral RV vaccination among children throughout the world.

Methods
Vaccine preparation. CDC-9, a human G1P 8 , RV strain was cultivated in Vero cells 14 . Triple-layered particles (TLPs) were purified from cell culture supernatants by using isopycnic CsCl gradient centrifugation and inactivated at 60 °C for 4 hours 2 . TLPs in HBSS supplemented with 10% sorbitol were mixed with 2.96% w/w aluminum hydroxide (Alhydrogel, Accurate Chemical & Scientific Corp, Westbury NY) on a rocking platform overnight at 4 °C. The vaccine was stored at 4 °C before use for intramuscular (IM) injection.
Inactivated TLPs were used for the fabrication of dissolving polymer MN patches 41 . The fabrication was a two-step molding process on PDMS molds. Patches contained 100 conical MN in a 10 × 10 pattern (850 µm height and 500 µm diameter at base). HBSS was used to perform all necessary dilutions. IRV was formulated at a concentration of 0.4 or 2 mg/ml IRV with 1% (w/v) sodium carboxymethylcellulose, and 5% (w/v) sorbitol, and cast on PDMS molds. Vacuum was applied to ensure filling of the solution into the mold. Excess solution was removed after 40 minutes from the mold and the filled mold was left on vacuum for 20 additional minutes. The second filling solution, containing 18% PVA and 18% sucrose, was cast on the PDMS mold under vacuum for 3 hours and dried overnight at 35 °C. Placebo MN patches, containing no IRV, were prepared in the same manner. MN patches were produced two days before vaccination of animals and stored at room temperature. Delivery efficiency was expected to be ~50%, therefore patches were designed to contain 2 µg IRV and 10 µg IRV for the delivery of 1 µg and 5 µg, respectively. Four patches were analyzed for antigen concentration before application into mice with Premier Rotaclone EIA (Meridian Bioscience, Cincinnati, OH).

Vaccination of mice.
Balb/c mice (8-10 weeks old) were purchased from Charles River Laboratories (Wilmington, MA). Mice were anesthetized with intraperitoneal injection of 110 mg/kg ketamine (Vet One, Boise, ID) and 11 mg/kg xylazine (Akorn Animal Health, Lake Forest, IL). For MN patch application, the back of mice was shaved with an electric shear and remaining hair was removed using a depilatory cream (Nair, Church & Dwight, Princeton, NJ) two days before patch application. Pre-bleed was taken from the submandibular vein. Serum and organs were collected from 4 unvaccinated mice for baseline assessment. Mice in groups of 20 were anesthetized before each vaccination and vaccinated with 1 µg or 5 µg IRV formulated with alum or alum only (placebo control) by IM injection into the upper quadriceps muscle, or with MN patch containing 2 µg or 10 µg IRV MN patch or placebo patch, which was applied with pressure for 30 seconds on the back of the animals. After 15 minutes on the skin, MN patches were removed, analyzed microscopically for structural integrity, and then reconstituted in 0.5 ml HBSS to determine the residual antigen amount by Premier Rotaclone EIA (Meridian Bioscience, Cincinnati, OH). Measured antigen concentration of 4 patches and the determined residual antigen amount were used to calculate the delivered dose. Five mice from each group were sacrificed 2 days after each vaccination to examine changes in serum cytokine profiles, immune cell distribution, and gut-homing marker expression. All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the CDC and conducted in accordance with the ethical guidelines for animal experiments and safety guidelines.
Quantification of cellular immune responses by flow cytometry. Blood was collected at the indicated time points by submandibular vein puncture and sera were prepared. Murine cytokines in sera were measured by using the BD Cytometric Bead Array for Murine Th1/Th2/Th17 cytokines (BD Bioscience, San Jose, CA) according to manufacturer's instructions. Murine IFN-α in sera was detected by using VeriKine Mouse IFN Alpha ELISA Kit (PBL Assay Science, Piscataway, NJ) according to manufacturer's instructions.
Cell isolation. Following euthanasia of mice with CO 2 , spleen and mesenteric lymph nodes (mLN) were harvested. Splenocytes were isolated by dissecting the spleen capsule and pressing with a pipette. Cells were filtered through a 70 µm cell strainer and centrifuged at 1200 rpm for 5 minutes followed by red blood cell lysis (Red Blood Cell Lysing Buffer Hybri-Max, Sigma-Aldrich, St. Louis, MO). Single cell suspension from mLN was prepared by mashing mLN through a 70 µm cell strainer followed by centrifugation at 1200 rpm for 5 minutes.
Flow cytometric analysis. Single cell suspensions isolated from spleen and mLN, and whole blood containing peripheral blood mononuclear cells (PBMC) were incubated with cell surface marker-specific Figure 5. IRV vaccination induces elevated levels of rotavirus-specific neutralizing activity in mice. Peripheral blood was collected before the mice received the first (day 0), the second (day 21) and third (day 42) vaccination dose, and at the study endpoint (day 63). Levels of rotavirus-specific neutralizing activity in serum was determined by a microneutralization assay as described in the text. Data show mean and s.d. of n = 5. Mann-Whitney test was used for statistical analysis. MN: microneedle; IM: intramuscular; ns: not significant; **p ≤ 0.01.