The Fusion Protein rFlaA:Betv1 Modulates DC Responses by a p38-MAPK and COX2-Dependent Secretion of PGE2 from Epithelial Cells

Developing new adjuvants/vaccines and better understanding their mode-of-action is an important task. To specifically improve birch pollen allergy treatment, we designed a fusion protein consisting of major birch pollen allergen Betv1 conjugated to the TLR5-ligand flagellin (rFlaA:Betv1). This study investigates the immune-modulatory effects of rFlaA:Betv1 on airway epithelial cells. LA-4 mouse lung epithelial cells were stimulated with rFlaA:Betv1 in the presence/absence of various inhibitors with cytokine- and chemokine secretion quantified by ELISA and activation of intracellular signaling cascades demonstrated by Western blot (WB). Either LA-4 cells or LA-4-derived supernatants were co-cultured with BALB/c bone marrow-derived myeloid dendritic cells (mDCs). Compared to equimolar amounts of flagellin and Betv1 provided as a mixture, rFlaA:Betv1 induced higher secretion of IL-6 and the chemokines CCL2 and CCL20 from LA-4 cells and a pronounced MAPK- and NFκB-activation. Mechanistically, rFlaA:Betv1 was taken up more strongly and the induced cytokine production was inhibited by NFκB-inhibitors, while ERK- and p38-MAPK-inhibitors only suppressed IL-6 and CCL2 secretion. In co-cultures of LA-4 cells with mDCs, rFlaA:Betv1-stimulated LA-4 cells p38-MAPK- and COX2-dependently secreted PGE2, which modulated DC responses by suppressing pro-inflammatory IL-12 and TNF-α secretion. Taken together, these results contribute to our understanding of the mechanisms underlying the strong immune-modulatory effects of flagellin-containing fusion proteins.


Introduction
The current SARS-CoV-2 pandemic has clearly demonstrated the need to develop both safe and efficacious vaccines. Protein-antigen-based vaccines that typically have a low immunogenicity often need to be adjuvanted in order to induce robust immune responses.
In this context, some "Toll"-like receptor (TLR) ligands with intrinsic immune-activating properties are attractive adjuvant candidates. They act as pathogen-associated molecular patterns (PAMPs) which bind conserved pattern recognition receptors (PRRs) on immune cells to induce strong immune responses.
Currently, monophosphporyl lipid A (MPLA), a modified TLR4 ligand derived from the lipopolysaccharide of Salmonella minnesota R595 is the only TLR ligand that is already used as an adjuvant in licensed vaccines. Here, several vaccines containing MPLA as a component of more complex adjuvant systems have been licensed or are undergoing phase III clinical trials. These include Fendrix ® (for the prevention of hepatitis B), Cervarix ® (human papillomavirus-16 and papillomavirus-18), RTS,S(R) (malaria), and the allergen product Pollinex ® Quattro (pollen allergies) [1][2][3].
The success of MPLA as a vaccine adjuvant clearly demonstrates the immune-modulating potential of TLR-ligands. However, other TLR-ligands were less successful in pre-clinical trials. For example, nucleic acid-based TLR-ligands, such as CpG (activating TLR9), R848 (TLR7/8), or Poly I:C (TLR3), are potent immune activators, but are limited in their clinical efficacy due to problems with both toxicity and stability in vivo [4].
So far, the TLR5 ligand flagellin, a bacterial motility protein that forms the main body of the bacterial flagellum [5], is the only purely protein-based TLR ligand. In addition to TLR5, flagellin activates intracellular NOD-like receptor 4 (NLRC4) resulting in inflammation and IL-1β production [6]. Consequently, flagellin has repeatedly demonstrated mucosal adjuvant properties resulting in the induction of protective immune responses [7][8][9].
Mechanistically, TLR ligand:antigen fusion proteins effectively target the fused antigen to immune cells in vivo that express their respective TLR, resulting in more effective processing and simultaneous presentation of the enclosed antigen in the context of the TLR ligand-mediated immune cell activation [9,26].
We recently generated a fusion protein consisting of flagellin A from Listeria monocytogenes and the major birch pollen allergen Betv1 (rFlaA:Betv1) that was able to suppress allergic sensitization in vivo [22]. In vitro analyses in myeloid dendritic cells (mDCs) showed rFlaA:Betv1 to induce a pronounced mDC activation characterized by the secretion of both pro-(IL-6, IL-12, and TNF-α) and anti-inflammatory (IL-10) cytokines as well as pronounced surface expression of co-stimulatory molecules [22]. While the effects of flagellin antigen fusion proteins on DCs are well investigated, their effects on other cell types require further investigation.
Epithelial cells, which similarly express high levels of TLR5 [27] are the first line of defense against invading pathogens [28]. In addition to their function as part of the mechanical barrier, epithelial cells engage in the initiation and maintenance of immune responses [29,30]. Just as other immune cells, epithelial cells express pattern recognition receptors, become activated, and secrete both cytokines and chemokines to alert other immune cells to invading pathogens [28]. Therefore, epithelial cells play an integral part in the establishment of immune responses.
To our knowledge, the effects of fusion proteins on epithelial cells have been poorly studied. Maffía et al. showed that fusion of Cementoin to the secretory leukocyte proteinase inhibitor (SLPI) increased both SPLI's preoteolytic stability and surface binding to TNF-α-or LPS-pretreated A549 alveolar epithelial cells [31]. Moreover, Savar et al. in 2014 generated a FimH/tFliC fusion protein consisting of a truncated form of flagellin C (FliC) from enteroaggregative E. coli fused to FimH, the tip adhesion on type 1 fimbria, from uropathogenic E. coli to create a vaccine candidate against urinary tract infections [32]. Despite showing that the truncated form of FliC alone could induce IL-8 in the HT-29 epithelial cell line [32], no experiments using this fusion protein on epithelial cells were performed [32].
Therefore, as far as we know, the effects of flagellin:antigen fusion proteins on epithelial cells and intracellular signal transduction are currently unknown.
In the present study, we analyzed the effects of rFlaA:Betv1 on the mouse lung epithelial cell line LA-4. Here, we show that rFlaA:Betv1 is taken up more readily than the mixture of both proteins and induces a MAPK-and NFκB-dependent secretion of both cytokines and chemokines. Moreover, rFlaA:Betv1-stimulated LA-4 cells modulated cytokine secretion from BALB/c mDCs stimulated with rFlaA:Betv1. Supernatants derived from rFlaA:Betv1-stimulated LA-4 cells selectively suppressed rFlaA:Betv1-induced, mDCderived IL-12 and TNF-α secretion. These immune-modulatory effects of LA-4 cells were shown to depend on the p38-MAPK-and cyclooxygenase-2 (COX2)-dependent production of prostaglandin E2 (PGE 2 ).

Western Blot
For Western blot experiments, LA-4 cells were seeded overnight at 1.66 × 10 5 cells/ 2 mL in 6-well plates (Thermo Scientific) with culture medium. On the next morning, cells were cultured for 3 h at 37 • C, 5% CO 2 in DMEM supplemented with 2% FCS (Sigma-Aldrich) for starvation, and subsequently, stimulated with the indicated proteins in DMEM for 30 min, washed with ice-cold PBS, and lysed with 200 µL lysis buffer (62.5 mM Tris-HCl (pH 6.8), 2% w/v SDS, 10% glycerol, 50 mM DTT, 0.01% w/v bromophenol blue) for 10 min on ice. Target proteins in lysates were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. After blocking with 5% non-fat milk, the Cells 2021, 10, 3415 5 of 25 membranes were incubated with the following primary antibodies from Cell Signaling Technologies overnight at 4 • C: phospho-MAPK family antibody sampler kit (#9910), NF-κB pathway sampler kit (#9936), mTOR Substrates Antibody Sampler Kit (#9862) and loading control anti-H3 antibody (#12648, HRP Conjugate). Detection was performed with the provided secondary antibodies using immobilon crescendo Western HRP substrate (#WBLUR0500, Merck, Darmstadt, Germany), and images were captured with either a Fusion-Fx7 spectra reader (Vilber Lourmat, Eberhardzell, Germany) or iBright™ CL1500 system (Thermo Fischer Scientific). Band intensities in Western blots were quantified with ImageJ software (imagej.nih.gov, version: 1.52a, accessed on 1 December 2021) as relative light unit (RLU) normalized to histone H3 loading control. Protein concentrations after Alexa Flour 488 labeling was confirmed by BCA (micro BCA protein assay kit; Pierce, Rockford, IL, USA). Subsequently, 6.25 × 10 4 LA-4 cells/mL were stimulated with equimolar amounts of the labeled proteins (rBetv1, rFlaA, rFlaA plus rBetv1, or rFlaA:Betv1) for 15 min at 37 • C either with or without 90 min of prestimulation with chloroquine or cytochalasin A and extensively washed with FACS buffer (PBS, 1% BSA, 0.3% sodium azide, and 24 mmol/L EDTA, pH 8.0). Unspecific binding was blocked using incubation of the cells with Fc-Block (eBioscience) for 30 min and then analyzed for protein uptake in LA-4 cells by means of flow cytometry.
For microscopy analysis, LA-4 cells were seeded overnight at 6.25 × 10 4 cells/mL in 24-well plates. The next day cells were pre-incubated with either cytochalasin A (1 µg/mL) or chloroquine (10 µM) for 90 min and subsequently stimulated with Alexa Fluor 488 labeled proteins for 15 min in DMEM supplemented with 2% FCS. Cells were then washed three times with PBS and fixed for 10 min at room temperature with 4% (w/v) paraformaldehyde solution (Thermo Scientific, Dreieich, Germany). After fixation, cells were washed three times with PBS, and stained with 2 µg/mL of the nuclear marker 4 ,6-diamidino-2phenylindole, dilactate (DAPI) (Thermo Scientific) for 5 min and subsequently washed three times with PBS. The uptake of the Alexa Fluor 488 labeled proteins was visualized under a BZ-X800 fluorescence microscope (Keyence, Neu-Isenburg, Germany) at 26× magnification, the addition of the scale bars and merging of figures were achieved using a BZ-X800 analyzer (Keyence, Neu-Isenburg, Germany).

Mice
BALB/c mice (Jackson Laboratories, Bar Harbor, ME, USA) were bred at the animal facility of the Paul-Ehrlich-Institut under specific pathogen-free conditions.

Statistical Analysis
Statistical analysis was performed with GraphPad Prism v6 to v8 for Mac or Windows using 2-way ANOVA tests with confidence intervals adjusted for multiple comparisons according to either Bonferroni or Turkey. For statistically significant results the following convention was used: *-p-value < 0.05, **-p-value < 0.01, ***-p-value < 0.001.

rFlaA:Betv1 Induces Chemokine and IL-6 Secretion from LA-4 Epithelial Cells
Initially, LA-4 cells were characterized for their expression of typical epithelial cell surface markers as well as for their expression levels of TLR5 and MHC I (Supplementary Figure S1). LA-4 cells expressed the type II alveolar epithelial markers EpCAM and CEACAM 1, as well as MHC I and TLR5, while only expressing low levels of both mucin 1 and pancytokeratin (Supplementary Figure S1).
In the first set of experiments, LA-4 cells were stimulated with equimolar amounts of either the mixture of rFlaA + rBetv1 or the fusion protein rFlaA:Betv1 for 2, 4, or 24 h, and checked for cytokine and chemokine secretion ( Figure 1A).

Statistical Analysis
Statistical analysis was performed with GraphPad Prism v6 to v8 for Mac or Windows using 2-way ANOVA tests with confidence intervals adjusted for multiple comparisons according to either Bonferroni or Turkey. For statistically significant results the following convention was used: *-p-value < 0.05, **-p-value < 0.01, ***-p-value < 0.001.

rFlaA:Betv1 Induces Chemokine and IL-6 Secretion from LA-4 Epithelial Cells
Initially, LA-4 cells were characterized for their expression of typical epithelial cell surface markers as well as for their expression levels of TLR5 and MHC I (Supplementary Figure S1). LA-4 cells expressed the type II alveolar epithelial markers EpCAM and CEA-CAM 1, as well as MHC I and TLR5, while only expressing low levels of both mucin 1 and pancytokeratin (Supplementary Figure S1).
In the first set of experiments, LA-4 cells were stimulated with equimolar amounts of either the mixture of rFlaA + rBet v 1 or the fusion protein rFlaA:Betv1 for 2, 4, or 24 h, and checked for cytokine and chemokine secretion ( Figure 1A). Supernatants were collected after the indicated time points and checked for the secretion of IL-6, CCL2, and CCL20 by ELISA (B). Data are mean results ± SD from three independent experiments with two technical replicates per experiment. Statistical significance indicated as: n.s.: p-value > 0.05, * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001.
Compared to the mixture of both proteins, rFlaA:Betv1 induced a significantly increased secretion of the pro-inflammatory cytokine IL-6 as well as the chemokines CCL2 and CCL20 (for the stimulation concentration of 27.4 µg/mL rFlaA:Betv1 induced a 562-fold higher secretion of IL-6, a 37-fold higher secretion of CCL2, and a 30-fold higher secretion of CCL2 24 h post-stimulation) ( Figure 1B). IL-6 secretion was detected as early as 4 h post-stimulation, while chemokine secretion was already detected 2 h post-stimulation ( Figure 1B). In this experimental setting no secretion of either GM-CSF, G-CSF, IL-33, TSLP, IL-10, TNF-α, IL-12, or IL-1β was detected from stimulated LA-4 cells (data not shown).
To exclude that the minute amounts of residual LPS contained within the used protein preparations may be responsible for the observed LA-4 activation, both LPS titration curves and stimulations of LA-4 cells were performed applying the amounts of endotoxin contained within the used stimulation concentrations (Supplementary Figure S2). Here, initial LPS-mediated LA-4 cell activation was observed with LPS concentrations around 10 ng/mL (Supplementary Figure S2) which is far above the low pg amounts contained within our protein preparations. Moreover, stimulation of LA-4 cells with the endotoxin amounts contained within the used rFlaA:Betv1 concentrations did not result in secretion of any of the investigated cytokines (Supplementary Figure S2).

rFlaA:Betv1-Mediated LA-4 Cell Activation Is TLR5-Independent
To address whether the observed LA-4 activation by the fusion protein is TLR5dependent, we generated a mutant fusion protein that harbors a modified sequence stretch of 8 amino acids (QRMRQLAV on position 87 to 94 replaced with DTVKVKAT) in the flagellin D1 domain (rFlaA *D1 :Betv1). This sequence alteration was recently described to abolish TLR5 activation [34]. Indeed, both rFlaA *D1 and rFlaA *D1 :Betv1 were unable to activate TLR5-expressing HEK293 reporter cells ( Figure 2A).  When LA-4 cells were stimulated with either the wild-type or mutant proteins, we observed no difference in CCL2, CCL20, or IL-6 secretion ( Figure 2B), indicating the observed LA-4 activation by rFlaA:Betv1 to be TLR5-independent.  To investigate the TLR5-dependency of rFlaA:Betv1mediated LA-4 activation, a mutant fusion protein lacking a described TLR5-activation motif was generated (rFlaA *D1 :Betv1). Wildtype and mutant rFlaA and rFlaA:Betv1 were tested for their capacity to activate TLR5-expressing HEK293 reporter cells (A). LA-4 cells were stimulated with protein amounts of both wildtype and mutant rFlaA + rBetv1 and rFlaA:Betv1 equimolar to 10 µg/mL rBetv1 and checked for chemokine and cytokine secretion after the indicated time points poststimulation (B). Data are mean results of three independent experiments ± SD with two technical replicates per experiment. Statistical significance indicated as: n.s. p-value > 0.05, *** p-value < 0.001.
When LA-4 cells were stimulated with either the wild-type or mutant proteins, we observed no difference in CCL2, CCL20, or IL-6 secretion ( Figure 2B), indicating the observed LA-4 activation by rFlaA:Betv1 to be TLR5-independent.

rFlaA:Betv1 Is Taken up More Strongly by LA-4 Cells
Since the activation of LA-4 cells by rFlaA:Betv1 was shown to be TLR5 independent (Figure 2), we checked if an increased uptake of the fusion protein compared to the equimolar mixture of both single proteins might account for the observed stronger cell activation ( Figure 3). For this, we labeled the fusion protein and the respective controls with Alexa Fluor 488 (achieving comparable degrees of labeling) and checked for the uptake of the different proteins into LA-4 cells by either flow cytometry or microscopy ( Figure 3A). Here, flow cytometry showed the fusion protein to be taken up with a higher frequency than the equimolar mixture of both proteins ( Figure 3B). These results were confirmed by microscopy analysis where we observed extensive uptake of the fusion protein but not the respective controls into LA-4 cells ( Figure 3C).   Figure  S3A,B) of the inhibitor of actin polymerization cytochalasin A and the inhibitor of endosomal acidification chloroquine strongly inhibited rFlaA:Betv1 uptake ( Figure 3C) and dose-dependently (albeit not significantly for chloroquine) reduced rFlaA:Betv1-induced IL-6 secretion ( Figure 3D). In this experimental setting, neither cytochalasin A nor chloroquine affected CCL2 or CCL20 secretion (Supplementary Figure S3C). Controls for the microscopy analysis are shown in Supplementary Figure S4   Both LPS and rFlaA:Betv1 induced a pronounced phosphorylation of all investigated MAP kinases (p38 MAPK, p42/44 MAPK (also called ERK1/2), and SAP/JNK MAPK) ( Figure 4B,C). Moreover, the NF B subunit p65 was phosphorylated while I Bα levels were strongly reduced ( Figure 4B,C), concordant with an activation of NF B signaling. In contrast, at equimolar concentrations, the controls rFlaA, rBet v 1, and rFlaA + rBet v 1 showed no changes in the levels of investigated proteins ( Figure 4B,C).
Our previous results from myeloid dendritic cells (mDCs) showed rFlaA:Betv1 to induce both mammalian target of rapamycin (mTOR)-dependent metabolic changes and IL-10 secretion [22]. Therefore, we also checked if mTOR-signaling is activated by rFlaA:Betv1 in epithelial cells. However, there was no pronounced difference in the phosphorylation levels of the mTOR target protein p70 S6 kinase in the different stimulation groups compared to unstimulated controls (Supplementary Figure S5). Moreover, the Both LPS and rFlaA:Betv1 induced a pronounced phosphorylation of all investigated MAP kinases (p38 MAPK, p42/44 MAPK (also called ERK1/2), and SAP/JNK MAPK) ( Figure 4B,C). Moreover, the NFκB subunit p65 was phosphorylated while IκBα levels were strongly reduced ( Figure 4B,C), concordant with an activation of NFκB signaling. In contrast, at equimolar concentrations, the controls rFlaA, rBetv1, and rFlaA + rBetv1 showed no changes in the levels of investigated proteins ( Figure 4B,C).
Our previous results from myeloid dendritic cells (mDCs) showed rFlaA:Betv1 to induce both mammalian target of rapamycin (mTOR)-dependent metabolic changes and IL-10 secretion [22]. Therefore, we also checked if mTOR-signaling is activated by rFlaA:Betv1 in epithelial cells. However, there was no pronounced difference in the phosphorylation levels of the mTOR target protein p70 S6 kinase in the different stimulation groups compared to unstimulated controls (Supplementary Figure S5). Moreover, the mTOR inhibitor rapamycin had no significant effect on rFlaA:Betv1-induced cytokine and chemokine secretion (Supplementary Figure S6).  Figure S7).
3.6. NFκB-Signaling Contributes to Both rFlaA:Betv1-Induced Pro-Inflammatory IL-6 and CCL2/CCL20 Chemokine Secretion in LA-4 Cells The contribution of NFκB-signaling to rFlaA:Betv1-mediated cytokine and chemokine secretion was analyzed by pre-treatment of LA-4 cells with either the NFκB-inhibitor dexamethasone or the IκBα-inhibitor TPCA-1 ( Figure 6A). Toxic effects of the applied inhibitor concentrations on the LA-4 cells were excluded by live/dead staining ( Figure S8).

In Co-Cultures with BALB/c mDCs LA-4 Cells Modulate mDC-Derived, rFlaA:Betv1-Induced Cytokine Secretion
To investigate the functional consequences of rFlaA:Betv1-mediated epithelial cell activation, we checked for the effects of rFlaA:Betv1-stimulated LA-4 cells on mDCs. For this, BALB/c mDCs or LA-4 cells were either cultured alone or as a co-culture, stimulated with rFlaA:Betv1 or their respective controls, and checked for their respective cytokineand chemokine-secretion ( Figure 7A).
Compared to LA-4 cells, mDCs produced significantly higher levels of IL-6 upon stimulation with either rFlaA + rBetv1 or rFlaA:Betv1 in all tested dose ranges ( Figure 7B). Here, IL-6 secretion was not significantly different between mDCs cultured alone or together with LA-4 cells ( Figure 7B). Moreover, levels of either IL-1β-and IL-10-secretion induced by rFlaA:Betv1 did not differ between mDCs cultured alone or in the presence of LA-4 cells, while LA-4 cells did not produce these cytokines upon stimulation with the tested proteins ( Figure 7B). . Supernatants were collected and checked for the secretion of IL-6, CCL2, and CCL20 by ELISA (B). Data are mean results ± SD from three independent experiments with two technical replicates per experiment. Statistical significance indicated as: n.s.: no statistically significant difference, p-value > 0.05, * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001

Induced Cytokine Secretion
To investigate the functional consequences of rFlaA:Betv1-mediated epithelial cell activation, we checked for the effects of rFlaA:Betv1-stimulated LA-4 cells on mDCs. For this, BALB/c mDCs or LA-4 cells were either cultured alone or as a co-culture, stimulated with rFlaA:Betv1 or their respective controls, and checked for their respective cytokineand chemokine-secretion ( Figure 7A). Compared to LA-4 cells, mDCs produced significantly higher levels of IL-6 upon stimulation with either rFlaA + rBet v 1 or rFlaA:Betv1 in all tested dose ranges ( Figure  7B). Here, IL-6 secretion was not significantly different between mDCs cultured alone or together with LA-4 cells ( Figure 7B). Moreover, levels of either IL-1β-and IL-10-secretion Secretion of the cytokines IL-12 and TNF-α was exclusively observed from mDCs stimulated with rFlaA:Betv1 ( Figure 7B). Interestingly, co-culture of mDCs with LA-4 cells resulted in highly significant reductions in the secretion of both cytokines compared to mDCs stimulated alone (IL-12 reduction by 70% to 92%, TNF-α reduction by 62% to 67%, Figure 7B).
In contrast, the chemokines CCL2 and CCL20 were produced in higher quantities by LA-4 cells ( Figure 7B), while mDCs produced a maximum of 2 ng of CCL2 after stimulation with rFlaA:Betv1 and no CCL20 secretion was detected ( Figure 7B). While CCL20 levels were consistently higher in co-cultures of mDCs and LA-4 cells compared to LA-4 cells alone, CCL2 levels were slightly decreased upon co-culture of both cell types ( Figure 7B).

Epithelial Cell-Derived Soluble Factors Modulate mDC-Responses to rFlaA:Betv1
We were interested, if the immune-modulating properties of rFlaA:Betv1-stimulated LA-4 cells on mDCs (reduction in IL-12 and TNF-α secretion, see Figure 7B) depend on secreted factors rather than cell-cell contact-dependent mechanisms. Therefore, we incubated mDCs with either 1, 10, or 100 µL of supernatant derived from LA-4 cells that were either unstimulated or stimulated with rFlaA:Betv1. Afterwards, we re-stimulated the mDCs with rFlaA:Betv1 and checked for the effects of LA-4 supernatants on mDC-derived cytokine secretion ( Figure 8A). 7B).

Epithelial Cell-Derived Soluble Factors Modulate mDC-Responses to rFlaA:Betv1
We were interested, if the immune-modulating properties of rFlaA:Betv1-stimulated LA-4 cells on mDCs (reduction in IL-12 and TNF-α secretion, see Figure 7B) depend on secreted factors rather than cell-cell contact-dependent mechanisms. Therefore, we incubated mDCs with either 1, 10, or 100 µL of supernatant derived from LA-4 cells that were either unstimulated or stimulated with rFlaA:Betv1. Afterwards, we re-stimulated the mDCs with rFlaA:Betv1 and checked for the effects of LA-4 supernatants on mDC-derived cytokine secretion ( Figure 8A).  . Subsequently, supernatants were collected and checked for the secretion of TNF-α, IL-12, IL-6, IL-1β, and IL-10 by ELISA (B). Data are mean results ± SD from three independent experiments with two technical replicates per experiment. Statistical significance indicated as: n.s.: no statistically significant difference, p-value > 0.05, ** p-value < 0.01, *** p-value < 0.001. Supernatant derived from unstimulated LA-4 cells slightly, but not statistically significantly, suppressed rFlaA:Betv1-induced secretion of both IL-12 and IL-6 from mDC cultures, while having no effect on rFlaA:Betv1-induced IL-1β secretion ( Figure 8B). In contrast, supernatants of unstimulated LA-4 cells dose-dependently and significantly suppressed rFlaA:Betv1-induced TNF-α secretion (62% reduction for the addition of 100 µL supernatant compared to mDC cultures stimulated with rFlaA:Betv1 in the absence of supernatant) ( Figure 8B).

The Modulation of mDC Responses by rFlaA:Betv1-Stimulated LA-4 Cells Is Dependent on the p38-MAPK-Signaling Pathway
To further investigate the intracellular signaling pathways involved in the production of DC-modulating factors by rFlaA:Betv1-stimulated LA-4 cells, mDCs were treated with supernatants from rFlaA:Betv1-stimulated LA-4 cells that had also been pre-treated with the MAPK-or NFκB-inhibitors previously investigated in Figures 5 and 6 (Figure 9A). Prostaglandin E2 (PGE2) is a naturally occurring lipid mediator generated from arachidonic acid. This reaction is catalyzed by either the cyclooxygenase-1 or -2 (COX1/2) enzymes [36]. As an inflammatory mediator, PGE2 is also known to influence DC activa- Among the tested supernatants, LA-4 cell-derived supernatants pre-treated with the p38 MAPK-inhibitor SB202190 lost their capacity to suppress mDC-derived cytokine secretion in terms of TNF-α, IL-6, and IL-12 (58.8-fold higher IL-12, 2.4-fold higher TNF-α, and 1.8-fold higher IL-6 levels in mDC + SB202190 sup. vs. rFlaA:Betv1 sup.) ( Figure 9B).

Modulation of mDC Responses by rFlaA:Betv1-Stimulated LA-4 Cells Is Dependent on the COX2/PGE 2 Pathway
Prostaglandin E2 (PGE 2 ) is a naturally occurring lipid mediator generated from arachidonic acid. This reaction is catalyzed by either the cyclooxygenase-1 or -2 (COX1/2) enzymes [36]. As an inflammatory mediator, PGE 2 is also known to influence DC activation, migration, and stimulatory capacity [37].
Finally, we incubated mDCs with 100 µL of supernatant derived from LA-4 cells which were pre-treated with increasing concentrations of NS-398, and then checked for mDC-derived cytokine secretion upon re-stimulation with rFlaA:Betv1 ( Figure 10D).

Discussion
In the present study, we analyzed the immune modulating properties of a flagellin:antigen fusion protein on epithelial cells. While the effects of flagellin on epithelial cells are well-known [43][44][45][46][47][48][49][50][51] and the interaction of flagellin with epithelial cells has often been described as a key factor for its adjuvant activity [52], our study is the first to characterize the response of epithelial cells to a flagellin:antigen fusion protein. We could show that, upon stimulation with a flagellin:antigen fusion protein, epithelial cells can modulate pro-inflammatory mDC responses by a both p38 MAPK-and COX2-dependent production of PGE 2 .
In our study, we observed that compared to the mixture of both proteins, the flagellin:allergen fusion protein rFlaA:Betv1 induced a significantly increased secretion of the myeloid chemo-attractants CCL2 and CCL20 as well as the cytokine IL-6 as early as two hours post-stimulation. This strongly enhanced epithelial cell activation is in accordance with our previous results showing rFlaA:Betv1 to also activate mDCs [22,23] and macrophages [35] more strongly than equimolar amounts of both single proteins. Of note, we observed no secretion of either GM-CSF, G-CSF, IL-33, TSLP, IL-10, TNF-α, IL-12, or IL-1β from LA-4 cells after rFlaA:Betv1-stimulation (data not shown).
Mechanistically, rFlaA:Betv1 was taken up into LA-4 cells more strongly than the mixture of both proteins and rFlaA:Betv1-mediated epithelial cell activation was shown to result from activation of both intracellular MAPK-(p38-, p42/44-, and SAP/JNK-) and NFκB-signaling as inhibition of both pathways prevented cytokine and chemokine secretion. Here, inhibition of NFκB, IκBα, and all types of MAPK dose-dependently inhibited rFlaA:Betv1-induced IL-6 secretion. Interestingly, rFlaA:Betv1-induced CCL2 and CCL20 secretion were shown to be suppressed by either NFκB or IκBα inhibition. In contrast, MAPK-signaling did not contribute to the secretion of the investigated chemokines (the suggested molecular mechanism of rFlaA:Betv1-mediated epithelial cell activation is shown in Figure 11). Interestingly, treatment with both proteins alone or as a mixture did not result in comparable MAPK-or NFκB-activation, cytokine-, or chemokine secretion.
Van Maele et al. showed that non-hematopoietic cells stimulated with flagellin to play a key role in TLR5-dependent, flagellin-mediated adjuvant activity by triggering CCL20 secretion from epithelial cells [55]. In line with this, titration rFlaA on LA-4 cells also resulted in the dose-dependent secretion of both CCL2 and CCL20 when applying higher amounts of FlaA (>17.4 µg/mL, data not shown). These results show, that in higher stimulation concentrations, rFlaA can also activate the investigated epithelial cells. However, it needs to be considered that both studies were performed with different subtypes of flagellin. secretion were shown to be suppressed by either NF B or I Bα inhibition. In contrast, MAPK-signaling did not contribute to the secretion of the investigated chemokines (the suggested molecular mechanism of rFlaA:Betv1-mediated epithelial cell activation is shown in Figure 11). Interestingly, treatment with both proteins alone or as a mixture did not result in comparable MAPK-or NF B-activation, cytokine-, or chemokine secretion. Van Maele et al. showed that non-hematopoietic cells stimulated with flagellin to play a key role in TLR5-dependent, flagellin-mediated adjuvant activity by triggering CCL20 secretion from epithelial cells [55]. In line with this, titration rFlaA on LA-4 cells also resulted in the dose-dependent secretion of both CCL2 and CCL20 when applying higher amounts of FlaA (>17.4 µg/mL, data not shown). These results show, that in higher Using a fusion protein mutant unable to activate TLR5 we showed the rFlaA:Betv1mediated LA-4 cell activation to be TLR5 independent. Although flagellin is a TLR5 ligand, flagellin may achieve its effects TLR5 independently. In previous studies, TLR5 activation by flagellin was found to be MyD88-, MAPK-, and NFκB-dependent [44,45,50,51,53]. Nevertheless, while some TLR5-expressing epithelial cell lines showed a strong activation after flagellin-stimulation (HT29 and A549), others showed only a weak (HeLa, 293T) or no response (T98G) to flagellin [53]. In line with these results, Tallant et al. suggested that the activation of other factors besides TLR5 could be necessary for a complete response to flagellin [53]. Recent studies indicate, for example, that TLR11, a TLR that is highly expressed in various epithelial cells [56,57], can recognize flagellin TLR5 independently [58]. Interestingly, Hatai and colleagues found that TLR11 only acted as a binding receptor for FliC under acidic conditions (pH 6.0), whereas TLR5 interacted with FliC at pH 6.0 and pH 7.0, suggesting that TLR11 recognizes FliC endolysosomatically [59]. As intracellular processing of the fusion protein was not the focus of this study, further research will be necessary to determine more precisely how flagellin:allergen fusion proteins like rFlaA:Betv1 MAPK-, NFκB-, and COX2-dependently activate epithelial cells after their uptake.
According to previous reports, the stronger LA-4 cell activation by the fusion protein may result from the observed high-molecular aggregation of rFlaA:Betv1 [22], potentially resulting in higher densities of the fusion protein on the cell surface, facilitating protein uptake, and thus stronger activating signals being transmitted to the respective cell's nucleus. In line with this, inhibition of rFlaA:Betv1 uptake into LA-4 cells reduced fusion protein-induced IL-6 secretion but had no effect on chemokine production. Therefore, we speculate that other, yet unknown mechanisms are likely also engaged in rFlaA:Betv1mediated epithelial cell activation.
In our previous studies, stimulation of mDCs and macrophages with rFlaA:Betv1 also resulted in an mTOR-dependent metabolic shift towards Warburg metabolism (also called aerobic glycolysis) [22,23,35]. However, in rFlaA:Betv1-stimulated LA-4 cells, we did not observe activation of mTOR signaling. Moreover, the mTOR inhibitor rapamycin had no significant effect on rFlaA:Betv1-induced cytokine and chemokine secretion, suggesting only a minor contribution of increased glucose metabolism to rFlaA:Betv1-mediated epithelial cell activation. These results might be explained by our observation, that LA-4 cells predominantly rely on oxidative phosphorylation for energy generation, while the overall metabolism of both mDCs and macrophages is rather glycolytic (data not shown).
Both CCL2 and CCL20 are myeloid chemo-attractants that alert other immune cells to the site of infection [60,61]. The pronounced activation of and chemokine secretion from rFlaA:Betv1-stimulated LA-4 cells prompted us to investigate, if rFlaA:Betv1-activated LA-4 cells could modulate subsequent immune responses induced by the fusion protein.
To investigate the functional consequences of fusion protein-mediated epithelial cell activation, we co-cultured BALB/c mDCs with either LA-4 cells or supernatants derived from rFlaA:Betv1-stimulated LA-4 cell cultures.
Upon co-culture of the two cell types, both rFlaA:Betv1-induced IL-12 and TNF-α secretion from mDCs were reduced while the levels of other mDC-derived cytokines (IL-1β, IL-6, and IL-10) remained unchanged. Furthermore, levels of LA-4 cell-derived CCL2 were slightly decreased upon co-culture of both cell types, suggesting that mDCs might metabolize some of the CCL2 produced by the epithelial cells.
Of note, we also observed a suppression tendency of rFlaA:Betv1-induced TNF-α secretion from ex vivo-isolated lung DCs (Supplementary Figure S10), suggesting that our findings can at least in part be transferred to the in vivo situation. Here, lung DCs did not produce IL-12 upon stimulation with rFlaA:Betv1 (Supplementary Figure S10).
To further investigate if the immune modulation by LA-4 cells depends on either direct cell-cell contact or soluble factors, mDCs were incubated with cell-free supernatants of rFlaA:Betv1-stimulated LA-4 cells. Our results showed the suppression of IL-12 and TNF-α from BALB/c mDCs to be dependent on LA-4-derived soluble factors. Interestingly, the lower secretion of rFlaA:Betv1-induced cytokines IL-12 and TNF-α from BALB/c mDCs was paralleled by unchanged levels of IL-1β and IL-10. Taken together, these results suggest, that factors secreted from rFlaA:Betv1-stimulated LA-4 cells can specifically modulate mDC responses.
Since rFlaA:Betv1-induced CCL2, CCL20, and IL-6 secretion from LA-4 cells were not responsible for the modulation of mDC responses, other yet unidentified factors might be engaged. PGE 2 is a naturally occurring lipid mediator generated by COX2-mediated conversion of arachidonic acid [36]. PGE 2 is known as an inflammatory mediator that influences DC activation, migration, and stimulatory capacity [37]. In allergy, it was shown to suppress allergic airway responses by inhibiting group 2 innate lymphoid cell (ILC2) activation [62], and decreasing the production of the Th2 cytokine IL-13 [63]. PGE 2 was reported to be secreted by epithelial cells upon LPS stimulation in a COX2dependent manner [38,39]. Here, COX2 activation was shown to be located downstream of the p38-MAPK pathway [40,64]. In accordance with the results presented in our study, PGE 2 was shown to suppress TLR2-and TLR4-induced IL-12 and TNF-α secretion from both DCs and macrophages in vitro [39,41,42].
In our study, rFlaA:Betv1 induced a higher secretion of PGE 2 from LA-4 cells compared to either unstimulated controls or the mixture of both proteins. Finally, pre-treatment of LA-4 cells with either the COX2-inhibitor NS-398 or the p38 MAPK-inhibitor SB202190 dose-dependently suppressed rFlaA:Betv1-induced PGE 2 production and significantly reduced the inhibitory capacity of supernatants from FlaA:Betv1-treated LA-4 cells on mDCs. Of note, inhibition of either p42/44 MAPK-, SAP/JNK MAPK-, NFκB-, or IκBαactivation in LA-4 cells did not have comparable effects, suggesting that these pathways do not contribute to the production of immune-modulating, soluble factors from LA-4 cells.
It is tempting to speculate that COX2 in epithelial cells is either directly or indirectly engaged in the immune modulating capacity of rFlaA:Betv1. The importance of COX2 activation is highlighted by the current discussion if administration of COX2 inhibitors such as ibuprofen around the time of vaccination decreases long-term responses to the COVID-19 vaccines.
In this context, it would be possible that the pronounced anti-allergic effects of different flagellin:allergen fusion proteins observed before [17,18,[20][21][22] may at least in part result from a reduced production of pro-inflammatory cytokines (IL-12 and TNF-α) from DCs if these cells are confronted with epithelial cell-secreted factors in flagellin:allergen fusion protein-treated animals. This effect could be further promoted by the unchanged secretion of the anti-inflammatory cytokine IL-10 ( Figure 11).
One shortcoming of our study is the exclusive usage of a cell line which may or may not reflect the in vivo situation. Future studies will need to investigate the effects of rFlaA:Betv1 on primary epithelial cells. Moreover, investigating the modulation of rFlaA:Betv1-induced epithelial cell responses in an air-liquid interface setting would be of interest to better understand the effects of the fusion protein, but could not be achieved within the limits of this study.
In summary, we investigated the activation of the mouse lung epithelial cell line LA-4 by a vaccine candidate consisting of the TLR5-ligand flagellin and the major birch pollen allergen Betv1. We observed, that rFlaA:Betv1, but not the mixture of both single proteins, triggered an increased uptake and a MAPK-, NFκB-, and COX2-dependent activation of epithelial cells, characterized by a pronounced secretion of the cytokine IL-6, the myeloid chemo-attractants CCL2 and CCL20, as well as PGE 2 . Furthermore, rFlaA:Betv1-stimulated LA-4 cells modulated the activation of mDCs, resulting in lower secretion of the pro-inflammatory cytokines IL-12 and TNF-α, while the levels of the IL-1β and IL-10 remained unchanged. Our current mechanistic understanding of the events contributing to rFlaA:Betv1-mediated LA-4 activation and its potential consequences for subsequent immune responses induced by rFlaA:Betv1 is depicted in Figure 11.
Finally, the results further establish epithelial cells as important target cells for vaccination approaches and increase our understanding of the mechanisms underlying the strong immune-modulatory effects of flagellin-containing fusion proteins observed in vivo. Our work will likely contribute to a potentially safe and efficient application of such vaccines in the future.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cells10123415/s1, Figure S1: Flow cytometric characterization of the used LA-4 cell line, Figure S2: The amounts of LPS contained within the used rFlaA:Betv1 preparations do not induce LA-4 cells activation, Figure S3. Cytotoxicity and effect on chemokine secretion of the used uptakeinhibitors, Figure S4. rFlaA:Betv1 is taken up more strongly than the mixture of both single proteins, Figure S5: mTOR signaling is not activated by rFlaA:Betv1 in LA-4 cells, Figure S6: mTOR signaling does not contribute to rFlaA:Betv1-induced cytokine and chemokine production from LA-4 cells, Figure S7: Cytotoxicity of the used MAPK-inhibitors on LA-4 cells, Figure S8: Cytotoxicity of the used NFκB-inhibitors on LA-4 cells, Figure S9: Cytotoxicity and effect of the COX2 inhibitor NS-398 on chemokine and IL-6 secretion from LA-4 cells, Figure S10. LA-4 cells also suppress rFlaA:Betv1induced TNF-α secretion from lung dendritic cells.

Acknowledgments:
The authors would like to thank Nadine Duschek for expert technical assistance and Maren Krause for providing animals.

Conflicts of Interest:
The authors have no conflict of interest to declare.