Laser‐facilitated epicutaneous immunotherapy with hypoallergenic beta‐glucan neoglycoconjugates suppresses lung inflammation and avoids local side effects in a mouse model of allergic asthma

Abstract Background Allergen‐specific immunotherapy via the skin targets a tissue rich in antigen‐presenting cells, but can be associated with local and systemic side effects. Allergen‐polysaccharide neoglycogonjugates increase immunization efficacy by targeting and activating dendritic cells via C‐type lectin receptors and reduce side effects. Objective We investigated the immunogenicity, allergenicity, and therapeutic efficacy of laminarin‐ovalbumin neoglycoconjugates (LamOVA). Methods The biological activity of LamOVA was characterized in vitro using bone marrow‐derived dendritic cells. Immunogenicity and therapeutic efficacy were analyzed in BALB/c mice. Epicutaneous immunotherapy (EPIT) was performed using fractional infrared laser ablation to generate micropores in the skin, and the effects of LamOVA on blocking IgG, IgE, cellular composition of BAL, lung, and spleen, lung function, and T‐cell polarization were assessed. Results Conjugation of laminarin to ovalbumin reduced its IgE binding capacity fivefold and increased its immunogenicity threefold in terms of IgG generation. EPIT with LamOVA induced significantly higher IgG levels than OVA, matching the levels induced by s.c. injection of OVA/alum (SCIT). EPIT was equally effective as SCIT in terms of blocking IgG induction and suppression of lung inflammation and airway hyperresponsiveness, but SCIT was associated with higher levels of therapy‐induced IgE and TH2 cytokines. EPIT with LamOVA induced significantly lower local skin reactions during therapy compared to unconjugated OVA. Conclusion Conjugation of ovalbumin to laminarin increased its immunogenicity while at the same time reducing local side effects. LamOVA EPIT via laser‐generated micropores is safe and equally effective compared to SCIT with alum, without the need for adjuvant.

some proteins such as tropomyosin, 9 and upon ligand binding activates innate immune responses. Murine Dectin-1 is expressed on macrophages, neutrophils, and dermal DCs. 10 In humans, Dectin-1 is not restricted to cells of the myeloid lineage but can also be found on epithelial cells, 8,11,12 keratinocytes, 13 B cells, and subpopulations of T cells. 8 As Dectin-1 is expressed on dermal DCs, its use as a vaccination platform for targeted skin vaccination is highly promising.
Here, we used 4-5 kDa laminarin/ovalbumin (LamOVA) conjugates for allergen-specific immunotherapy (AIT). Laminarin has strong immunostimulatory properties and has been employed in tumor therapy for activation of the innate immune system. 14 Moreover, polysaccharides can mask epitopes on allergens preventing IgE cross-linking on mast cells and histamine release, decreasing the risk of side effects. 15 Skin has a protective function, and its outermost layer, the stratum corneum, represents a tight barrier that has to be overcome for antigen delivery. We use infrared laser to form micropores in the upper layers of the skin. 16 Besides facilitating vaccine penetration, the local tissue damage caused by laser treatment attracts large numbers of APCs and creates a pro-inflammatory milieu, thereby acting as a physical adjuvant for skin vaccinations. 17 In our current work, we evaluated the immunostimulatory capacity of LamOVA conjugates and their efficacy in a preventive and therapeutic BALB/c mouse model of allergic lung inflammation.
Laser-facilitated EPIT with LamOVA significantly reduced airway hyperresponsiveness and local side effects in vivo.

| MATERIAL S AND ME THODS
A more detailed version of materials and methods and a complete list of used antibodies and their dilutions (Table S1) can be found in the Supplementary Materials.

| Generation and analysis of allergencarbohydrate neoglycoconjugates
Laminarin from laminaria digitata (Sigma, L9634, batch SLBP4829V) was dialyzed with a cut-off of 2 kDa to remove low molecular weight impurities and partially oxidized with sodium (meta)periodate (Sigma). 15 Briefly, 400 mg of laminarin were reacted with 95.2 mg sodium (meta)periodate for 1 hour at RT followed by dialysis against 50 mM phosphate buffer, pH 7. Laminarin and its oxidized variant were analyzed by two-dimensional NMR spectroscopy. Oxidized laminarin was coupled to endotoxin-free ovalbumin (OVA, EndoFit™ Ovalbumin, InvivoGen) using 2-step reductive amination with 50 mM sodium cyanoborohydride (NaCNBH 3 ) and 10 mM sodium borohydride (NaBH 4 ) as previously described. 15  Carbohydrate concentration in conjugates was estimated using anthrone method 18 and OVA concentration was determined by amino acid analysis. 19 The hydrodynamic radius of conjugates was analyzed by dynamic light scattering using a Zetasizer Nano ZS with a DTS1070 capillary cell (Malvern Instruments).
Biological activity of laminarin in LamOVA conjugates was analyzed by ELISA using a soluble murine Fc-Dectin-1a receptor (InvivoGen) and by Microscale Thermophoresis (MST) using a Monolith NT.115 RED instrument (Nanotemper, Munich, Germany). Hypoallergenicity was analyzed in vitro as a measure of rat basophil leukemia cell (RBL-2H3) activation by β-hexosaminidase release assay. 15

| Animal experiments
Female BALB/c mice aged 6-8 weeks were obtained from Janvier To study immunogenicity of the conjugates, mice (n = 6) were immunized three times with 20 µg of OVA, LamOVA conjugates (20 µg OVA coupled to 49 µg laminarin), a mixture of equivalent amounts of OVA + laminarin, or laminarin and PBS (sham controls).
The antigens were applied to laser-microporated skin on days 1, 15, and 29. Microporation was performed using a PLEASE® professional infrared laser device (Pantec Biosolutions, Ruggell, Liechtenstein) with a total fluence of 8.3 J/cm 2 as previously described. 15 After the third immunization, mice were sensitized two times in a ten-day interval (days 36 and 46) by i.p. injections of 1 µg EndoFit™ OVA adsorbed to 50% (v/v) alum (Alu-Gel-S, Serva) in endotoxin-free PBS. To induce lung inflammation, animals were challenged by exposure to aerosolized grade V OVA from Sigma (A5503) that has been purified from endotoxin using Triton-X114 method 22 (10 mg/mL in 0.9% NaCl, LPS <0.39 pg per 1 µg OVA), for 30 minutes on days 53-55 using a Pariboy SX jet nebulizer with a Pari LL nebulizer head (Pari GmbH).
To study therapeutic efficacy, animals (n = 12) were sensitized by two i.p. injections with 10 µg of EndoFit™ OVA/alum on days

| Analysis of local side effects
For analysis of erythema severity, mice were photographed on days 47 or 54 with a SONY ILCE-7M2 camera, 28-70 mm zoom lens with a constant source of light using.JPG format. MATLAB R2019a was used for analysis. Briefly, the area of laser microporation (region of interest, ROI) was defined by an investigator blinded to the treatment. Based on the gray value of the selected ROI, a threshold was manually selected, to define the erythema that was darker than healthy skin. Erythema size and mean blob size were calculated from the selected ROI. Erythema size is calculated as the ratio of pixels above the chosen threshold to the number of pixels inside the ROI. A blob is defined as the sum of directly connected pixels of the segmented erythema. The mean value of the five biggest blobs was evaluated.

| Lung function analysis
The day after the aerosol challenge, lung function was analyzed by whole-body plethysmography (WBP) using a Buxco 6-chamber unrestrained WBP system (Data Sciences International). The next day, lung resistance/compliance measurement was performed using a FinePointe Series Resistance and Compliance (RC) site (Data Sciences International). Bronchoalveolar lavage fluid (BALF) was collected, and its cellular composition was analyzed by flow cytometry staining for Siglec F, CD45, CD4, CD8, CD19, and Ly6G/Ly6C (Gr-1).
F I G U R E 1 (A) Structure of laminarin as determined by NMR. Glucose residues that are distinguishable by NMR spectroscopy are labeled with small letters a to h and the mannitol is abbreviated with M. β1-3 and β1-6 glycosidic bonds are indicated as β3 and β6, respectively. (B) Schematic model depicting OVA (PDB: 1OVA) with 6 covalently linked chains of laminarin. The six lysine residues that have a fractional accessible surface area >0.5 at least in one of the four structures in the asymmetric unit of the crystal structure are shown in red, the other lysines in blue. A potential cross-linking of two OVA molecules is shown 2.7 | Analysis of OVA-specific serum IgG/IgE and cell-bound IgE OVA-specific IgG isotypes in sera were analyzed by direct ELISA.
OVA-specific serum IgE was measured using a rat basophil leukemia cell (RBL-2H3) assay. Briefly, cells were incubated for 2 hours with mouse sera diluted 1:150 in medium, followed by three washing steps and 1 hour stimulation with 0.3 µg/mL EndoFit™ OVA.
β-hexosaminidase levels in supernatants were measured and degranulation was calculated as percentage of total lysis from control wells lysed with Triton X-100.
Cell-bound IgE was analyzed ex vivo by basophil activation test from whole blood drawn one day after the second aerosol challenge.
Cells were stimulated in the presence or absence of autologous serum with 2 ng/mL OVA for 2 hours at 37°C, 5% CO 2 , 95% humidity. After the incubation, cells were stained for IgE, CD4, CD19, and CD200R and analyzed by flow cytometry on a FACS Canto II flow cytometer (BD Biosciences).

| Statistical analysis
Data were statistically analyzed using GraphPad Prism 7 using oneway ANOVA with Tukey's post hoc test unless otherwise specified.

| Generation of laminarin-OVA conjugates
NMR analysis of the used laminarin batch (SLBP4829V) indicated approx. 70% M-series (terminated with 1-linked d-mannitol) and 30% G-series (containing only glucose) in the fresh and also in the dialyzed laminarin sample ( Figure 1A and Figures S1 and S2). Approx.
1.5 branch sites per molecule were observed. These findings are in agreement with previously published data (Table S2). 23

| Laminarin is biologically active after conjugation
To ensure that the structure of laminarin was not destroyed during mild periodate oxidation and that the polysaccharide remained biologically active after coupling, binding to Dectin-1 was assessed by ELISA and MST. As shown in Figure 2A, plate-bound LamOVA was recognized by Dectin-1, whereas no binding to unconjugated OVA was observed. To verify that the observed differences were not due to relative coating efficiencies of OVA and LamOVA or unspecific binding of LamOVA due to denaturation and aggregation, we confirmed these results in solution using MST ( Figure 2B) and included an unspecific soluble receptor (TLR5) as a control ( Figure S4).  Figure S4), this was still sixfold lower compared with its affinity for Dectin-1, confirming the specificity. The partial binding of LamOVA to TLR-5 was probably due to the OVA moiety of the conjugates, as unconjugated OVA showed the highest affinity for TLR-5 (0.8 ± 0.3 µM).

| LamOVA conjugates activate RBL cells to a fivefold lesser extent than unmodified OVA
Hypoallergenicity is a prerequisite for AIT via the skin, as administration of native allergen to barrier disrupted skin has been shown to induce local or even systemic side effects. 2 We hypothesized that laminarin polysaccharide chains (6 kDa) would mask IgE epitopes of OVA and prevent mast cell and basophil degranulation in vivo. To prove that conjugates were less allergenic than unconjugated OVA, rat basophil leukemia (RBL) cells were incubated with sera of highly sensitized mice and stimulated with LamOVA or OVA. Beta-hexosaminidase secretion after antigen stimulation is a measure of IgE cross-linking. As shown in Figure 2C, a fivefold higher dose of LamOVA was required to activate IgE-loaded RBL cells compared with OVA thus confirming hypoallergenicity of LamOVA. In contrast, neither EndoFit™ OVA nor laminarin induced significant cytokine or chemokine secretion in either BMDC type.

| LamOVA activated BMDCs are potent inducers of T-cell responses
Laminarin conjugation to OVA significantly increased uptake and induced activation of BMDCs and secretion of pro-inflammatory cytokines. These properties of LamOVA conjugate also resulted in enhanced stimulation of OVA-specific naïve T cells co-cultured with BMDCs. Both, FL-and GM-BMDCs incubated with LamOVA significantly activated T-cell proliferation and cytokine secretion ( Figure 3B and C). GM-BMDCs loaded with conjugates were more potent in inducing DO11.10 T-cell proliferation than FL-BMDCs, which may be due to their higher basal activation status ( Figure S5). This is also supported by the fact that only GM-BMDCs could induce activation of naïve T cells ( Figure 3B) and significant secretion of cytokines (IFN-γ and IL-9, Figure 3C) in the presence of Endofit™ OVA. In both cell culture models, conjugates significantly induced IL-2, IL-6, and IL-22 production (P < .0001, Figure 3C). Interestingly, LamOVA loaded FLand GM-BMDCS displayed a different TH-polarizing potential as FL-BMDCs mainly generated TH1 cells (IFN-γ and TNF-α, P < .0001 and P < .001), whereas GM-BMDCs favored upregulation of IL-2, which correlated with the higher T-cell proliferation rate.

| Epicutaneous immunization with LamOVA induces a strong antibody response
We have previously shown that epicutaneous immunization with mannan-conjugated allergen via laser-generated micropores (EPI) is considerably more immunogenic compared to immunization with F I G U R E 5 Therapeutic epicutaneous immunization via laser-generated micropores (EPIT  unconjugated allergen. 15 In a prophylactic immunization experiment ( Figure 4A), we could confirm that EPI with LamOVA induces ~threefold higher IgG1 antibody titers compared with OVA after two immunizations ( Figure 4B). This immune potentiating effect was dependent on the covalent linkage between OVA and laminarin since a mix of uncoupled OVA with laminarin did not enhance antibody responses. After a second booster immunization, all groups displayed similar antibody titers ( Figure 4C). IgG2a levels were close to the detection limit (data not shown).

| Preventive LamOVA immunizations reduce allergic lung inflammation and airway hyperresponsiveness
To investigate whether LamOVA-EPI would protect from allergic sensitization and lung inflammation, vaccinated mice were sensitized by two i.p. injections of Endofit™ OVA/alum followed by three aerosol challenges. In summary, our data show that despite their hypoallergenicity, conjugates are more potent in the production of OVA-specific IgG antibodies and can prevent allergic sensitization and airway hyperresponsiveness. In a next step, we therefore evaluated safety and potency of LamOVA conjugates in a therapeutic mouse model of allergic asthma.

| Epicutaneous immunotherapy with LamOVA induces blocking IgG and suppresses lung inflammation
In a therapeutic experiment, mice were sensitized by two i.p. injections of 10 µg Endofit™ OVA/alum followed by intranasal (i.n.) OVA challenge to induce lung inflammation. Based on WBP data, mice were stratified into five treatment groups with similar means and distribution of Penh. RBL assay confirmed that all animals had OVA-specific IgE antibodies, however, treatment arms-by chance-showed slightly higher IgE levels before therapy compared with the control groups ( Figure S7) LamOVA EPIT group and OVA s.c. group showed the highest levels of blocking IgG ( Figure S8B).
EPIT groups displayed the lowest number of eosinophils in BALF and significantly lower numbers in lung tissue compared with sham-treated mice ( Figure 5D). No significant differences in the BALF levels of IL-4, IL-5, IL-13, and eotaxin were found (data not shown).
OVA treatment significantly improved lung function measured by WBP and again the EPIT groups showed the lowest Penh values ( Figure 5E). Though not statistically significant, a trend for reduced resistance and increased compliance in the EPIT groups was confirmed by invasive R/C measurement ( Figure 5E).
Restimulation of splenocytes with OVA (10 µg/mL) induced increased expression of TH2 associated cytokines IL-4, IL-5, IL-10, and IL-13 but also IFN-γ, IL-2, IL-6, and IL-22 in the s.c. OVA group and surprisingly also in the group treated with laminarin alone. EPIT groups showed no such boost in cytokine responses and remained at similar levels compared with the PBS-treated group ( Figure S9).
This effect was also seen on the level of transcription factor expression Tbet, GATA3, and RORγT. No difference in the number of FoxP3 + CD25+ Tregs was found in the spleens ( Figure S10). In summary, s.c. injection with OVA/alum as well as treatment with laminarin boosted mainly TH2, but also TH1 and TH22 responses, whereas EPIT with OVA or LamOVA induced no increase in cytokine responses compared to the untreated group.

| Conjugation of polysaccharide to OVA reduces local side effects in vivo
In mice, local side effects are dependent on the used allergen.
While in previous studies with recombinant Phl p 5 28 or depigmented house dust mite extract, 29 no local side effects were observed, application of OVA to the skin of sensitized mice led to significant skin erythema and scab formation ( Figure 6A). We quantitated local skin reactions by calculating total erythema size and blob size ( Figure 6B and C) and found that LamOVA immunizations induced significantly lower skin reactions, confirming the hypoallergenicity observed in vitro. The use of laser with PBS and laminarin alone did not induce any side effects, confirming the antigen-specificity of the response. in allergic patients. 36 In our current study, we found 4-5 kD laminarin conjugates to be fivefold less allergenic in vitro ( Figure 2B) and to induce significantly less local side effects in sensitized mice during therapy ( Figure 6). Coupling of allergens to β-glucan molecules with a higher molecular weight could potentially further decrease allergenicity as we have previously seen a >1000-fold reduction of IgE cross-linking capacity of 30-40 kD mannan conjugates. 15 However, a careful balance of protein/carbohydrate ratio has to be maintained, as a higher polysaccharide content in conjugates potentially decreases IgG production (unpublished observation).

| D ISCUSS I ON
Due to the low immunogenicity of existing allergy vaccines, AIT requires repetitive administrations of high allergen doses.
More potent and controlled APC activation may be achieved via specific targeting of surface receptors, 6 for example, C-type lectin receptors. 37 Here, we used laminarin to target Dectin-1, which is expressed on various cell types including CD11b + dermal DCs and Langerhans cells in mice. 38  that oxidation of mannan impaired uptake of conjugates, 36 we have shown that mild oxidation of laminarin for allergen coupling does not influence its biological activity, as confirmed by its binding to Dectin-1 ( Figure 2A) and activation of BMDCs ( Figures S5 and S6).
We previously made similar observations for mannan 15 and other carbohydrates. 35 We employed two different in vitro models for studying antigen uptake, DC activation, and stimulation of naïve T cells, that is, Flt3L derived BMDCs and GM-CSF derived BMDCs.
While FL-BMDCs more closely represent steady state naïve DCs, GM-BMDCs rather resemble monocyte-derived inflammatory DCs Laminarin is a β-1,3-glucan and, depending on the mode of uptake,  Figure S5) and transport into the endosome ( Figure 3A). Yet, in contrast to data from others, 41 we clearly observed the induction of inflammatory pathways in LamOVA stimulated BMDCs. In line with the BMDC activation status, only GM-BMDCs could activate naïve OVA-specific T cells without additional stimuli ( Figure 3B and C). In contrast, BMDCs pulsed with LamOVA were potent activators of naïve T-cell proliferation and cytokine secretion. As components of fungal cell walls, β-1,3-glucans are known to trigger TH1/TH17 adaptive immunity. 38,42 This could be confirmed in T cells stimulated with LamOVA loaded FL-BMDCs, which secreted high levels of IFN-γ, TNF-α, and IL-22, and low levels of IL-17 ( Figure 3C). Interestingly, although GM-DCs induced stronger T-cell proliferation and IL-2 secretion, they were less potent in inducing TH1 cytokine secretion, which also correlates with their lower levels of IL-12 secretion upon LamOVA stimulation compared to FL-BMDCs ( Figure S6).
We have recently demonstrated that application of allergens to laser-microporated skin can induce humoral immune responses 16,29,43 and that conjugation to mannan potentiates antibody responses. 15 As shown previously for mannan, laminarin also boosted IgG1 responses threefold after EPI and covalent linkage of the carbohydrate to OVA was essential for this effect ( Figure 4B). This can be explained by the multimeric state of laminarin, that is more potent in stimulating Dectin-1 than soluble laminarin ( Figures S5 and S6), the enhanced uptake due to receptor-mediated endocytosis of the complexes ( Figure 3A), and a potentially slower draining to SDLNs as previously observed for Bet-mannan conjugates. 15 A similar observation was made by Xie et al after s.c. immunization; however, they also included poly I:C as an adjuvant. 39  We have previously shown in a therapeutic mouse model of allergic lung inflammation that laser-mediated EPIT with Phl p 5 or house dust mite extract can successfully reduce lung inflammation and improve lung function. Therapeutic efficacy was associated with induction of high levels of blocking IgG and a general downregulation of cytokine responses. 28,29 In the current study, we show for the first time the therapeutic efficacy of a hypoallergenic laminarin-allergen conjugate. While EPIT with OVA was similarly effective in suppressing allergic lung inflammation and airway hyperresponsiveness compared with LamOVA, it was associated with severe local side effects (Figure 6), replicating findings from clinical trials with grass pollen extract. 2 In contrast, LamOVA treated mice showed significantly ameliorated skin erythema. At the same time, conjugation to laminarin significantly boosted IgG1 (compared to OVA EPIT) to the same levels as achieved by s.c. injections together with alum ( Figure 5A), while inducing lower levels of therapy-associated IgE ( Figure 5C). Furthermore, both EPIT groups showed lower levels of lung inflammation compared to s.c.
treated mice ( Figure 5D and E). Although LamOVA-induced inflammation and TH1/TH17 polarization in vitro, EPIT with LamOVA did not result in enhanced inflammatory cytokine secretion in splenocytes from treated animals ( Figure S9). In contrast, SCIT with OVA boosted TH2 responses, an early therapy effect that is well known in the clinics. 44 Though we did not find increased numbers of FoxP3 + CD25+ regulatory T cells in the spleen after treatment, EPIT significantly reduced the number of activated T cells compared with SCIT ( Figure S10). Taken together, this confirms our previous findings that EPIT via laser-generated micropores can downregulate established allergic responses and mainly results in the induction of blocking IgG and a general suppression of cytokines 28,29 rather than immune deviation toward TH1 or TH17, even in the presence of an immunostimulatory Dectin-1 agonist. While the first clinical studies of SCIT with mannan allergoids are currently under way (NCT02654223, NCT02661854), we believe that combining a C-type lectin receptor-based DC targeting approach with a method to deliver a hypoallergenic vaccine to upper skin layers rich in APCs is the key to patient friendly, safe and effective immunotherapy. Our current proof of concept study demonstrates that laser-facilitated EPIT with β-glucan-allergen conjugates represents one such feasible approach.

ACK N OWLED G M ENT
This work was supported by the Austrian Science Fund (FWF; grant no. W 1213).

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflict of interest.