A. kunkeei PF12, PL13, and PF15 and H. uvarum AN8Y27B belonging to the Culture Collection of the Department of Soil, Plant and Food Science, University of Bari Aldo Moro (Bari, Italy), were used as mixed starters for BCP fermentation. Their aptitude to drive standardized fermentation of BCP at laboratory, pilot-plant, and full-scale levels was preliminarily verified (Di Cagno et al., 2019a; Giuliani et al., 2020). Cultures were maintained as stocks in 15% (v v⁻¹) glycerol at −80°C and routinely propagated. Fructophilic lactic acid bacteria were cultured at 30°C for 24 h in fructose-yeast extract-polypeptone (FYP) broth (10 g D-fructose, 10 g yeast extract, 5 g polypeptone, 2 g sodium acetate, 0.5 g Tween 80, 0.2 g MgSO₄·7H₂O, 0.01 g MnSO₄·4H₂O, 0.01 g FeSO₄·7H₂O, and 0.01 g NaCl per liter of distilled water). H. uvarum AN8Y27B was cultivated at 30°C for 36 h in yeast extract-peptone-dextrose (YPD) broth (10 g yeast extract, 20 g bacteriological peptone, and 20 g dextrose per liter of distilled water).

Bee-collected ivy pollen (BCP) was collected during September–October 2018 from organic fields of the Apulia region (Italy). BCP was fermented under a standardized protocol previously described by Di Cagno et al. (2019a), which included a mixed inoculum of A. kunkeei PF12, PL13, and PF15 strains, and H. uvarum AN8Y27B. Briefly, microorganisms were cultivated until the late exponential growth phase was reached, washed twice in 50 mM phosphate buffer (pH 7.0), and used to inoculate the BCP at the final density of ca. 8 Log CFU g⁻¹. BCP was added with sterile water to reach the final water content of 40% (w w⁻¹), inoculated with the mixed starter, placed into sealed tubes, and incubated at 30°C for 216 h. BCP fermented by the selected mixed starter (Started-BCP) was characterized as described below. BCP treated under the same conditions except for the use of microbial starters (Unstarted-BCP), and fresh BCP without any treatment [raw bee-collected pollen (Raw-BCP)] were used as controls. Unstarted-BCP underwent spontaneous fermentation to partially mimic the spontaneous fermentation of beebread within hive cells. Mesophilic lactic acid bacteria and yeast cell densities were monitored through plate counts on FYP agar containing 0.1% of cycloheximide (Sigma Aldrich, Saint Louis, MO, United States) incubated at 30°C for 48 h, and on YPD agar added of 0.1% of chloramphenicol incubated at 25°C for 72 h, respectively. Freeze-dried BCP samples were used for the subsequent assays.

To evaluate the nutrients and bioactive compounds bioaccessibility in Raw-, Unstarted-, and Started-BCP, we chose phenolics as target compounds. Phenolics bioaccessibility in BCP samples was investigated through an in vitro gastrointestinal batch digestion process according to Eid et al. (2014) and Celep et al. (2015), with few modifications. BCP (10 g) was homogenized in distilled water (50 ml) for 2 min using a stomacher. Then, 20 mg of α-amylase were dissolved in 6.25 ml of CaCl₂ (1 mM) solution and added to the mixture. The mixture was incubated at 37°C for 30 min under stirring condition (100 rpm) to mimic the oral digestion phase. To simulate gastric digestion, pepsin (2.7 g) was dissolved in 25 ml of 0.1 M HCl and added to mixture. Then, the pH value was adjusted to 2.0 using 6 M HCl and the mixture was incubated at 37°C for 3 h under stirring condition. To simulate small intestine conditions, pancreatin (560 mg) and bile (3.5 g) were dissolved in 125 ml of 0.1 M NaHCO₃ and added to the mixture. Value of pH was slowly adjusted to 7.0 by using 6 M NaOH and a segment of cellulose dialysis tubing (molecular weight cut off 12 kDa) was placed inside the beaker. The mixture was incubated at 37°C for 3 h under stirring condition (100 rpm). After the incubation, the solution that diffused into the dialysis tubing was taken as the serum-available fraction. The latter was centrifuged at 16,000 rpm, filtered through a nylon syringe filter with a pore size of 0.45 μm, and total phenolic compounds were assayed according to Folin-Ciocalteu method (Singleton and Rossi, 1965; Singleton et al., 1999). Data were expressed as gallic acid g equivalents per liter of digested BCP. In vitro digested BCP samples were freeze-dried and used for the subsequent assays.

Human Caco-2 cells were obtained from ATCC (ATCC® HTB-37™) and were used between passages 15 and 35 for all experiments. Cells were cultured in high glucose DMEM supplemented with 10% (v v⁻¹) Fetal Bovine Serum (FBS), 1% (v v⁻¹) HEPES, NEAA, and 1% (v v⁻¹) penicillin/streptomycin, maintained at 37°C in a 5% CO₂ incubator, and were sub-cultured at 80–90% confluence every 3–4 days.

Colon adenocarcinoma cell line-2 cells metabolic activity was assessed through the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Cells were seeded at a concentration of 2 × 10⁴ in a 96-well plate. After confluence was reached, cells were treated with Raw-, Unstarted-, and Started-BCP at different concentrations (10–500 μg ml⁻¹) for 24 h. At the end of the treatment exposure, medium was discarded and replaced by 100 μl of 0.5 mg ml⁻¹ MTT solution. After 2 h of incubation at 37°C, the MTT solution was discarded, and formazan crystals were dissolved in 100 μl dimethyl sulfoxide, and absorbance was read in a multiwall plate reader (Bio-Rad, Hercules, CA, United States) at 570 nm. Percent viability was calculated relative to untreated Caco-2 cells.

Cytotoxicity of Raw-, Unstarted-, and Started-BCP was determined by measuring the release of lactate dehydrogenase (LDH) into the Caco-2 culture medium. Caco-2 cells were seeded in a 96-well plate at a density of 1.5 × 10⁴ cells/well in 200 μl medium. On the fifth day of culture, the cells were treated with various concentrations of RW pollen or fermented pollen. After 24 h incubation, supernatants were collected and centrifuged at 200 × g for 5 min at room temperature and assayed using Pierce LDH cytotoxicity assay kit (Thermo Fisher Scientific, Waltham, MA, United States) according to the manufacturer’s instructions to evaluate LDH concentration in the culture medium. Percent cytotoxicity was calculated as follows:

%Cytotoxicity=BCP-treatedLDHactivity−SpontaneousLDHactivity/MaximumLDHactivity−SpontaneousLDHactivity×100

where untreated cells are referred to as Spontaneous LDH Activity Control and 5% triton x-100 treated cells are referred to as Maximum LDH Activity Control.

Interleukin-8, IL-6, MCP-1, TNF-α, and PGE-2 release by Caco-2 cells after 24 h of treatment with Raw-, Unstarted-, and Started-BCP (100 μg ml⁻¹) was quantified using commercial ELISA kits. Caco-2 cells pre-treated with BCP for 6 h were further stimulated with an inflammatory stimulus and then incubated for others 18 h. The inflammatory stimulus for IL-8, IL-6, MCP-1, and TNF-α was the interleukin-1β (IL-1β) at the concentration of 25 ng ml⁻¹. A cytokines mix (LPS, 10 ng ml⁻¹; TNF-α, 50 ng ml⁻¹; and IL-1β, 25 ng ml⁻¹) was the inflammatory stimulus for PGE-2. ELISA kit (R&D Systems) were used according to manufacturer’s instructions. Optical density was read with a microplate reader (Bio-Rad) at 450 nm.

Intracellular ROS content was assessed using 2',7'–dichlorodihydrofluorescein diacetate acetyl ester (DCFH-DA; Invitrogen). Cells were incubated with 100 μg ml⁻¹ of Raw-, Unstarted-, and Started-BCP for 24 h, and a set of samples was exposed to 50 μM H₂O₂ in the last 6 h to induce ROS production. At the end of the treatment, cells were rinsed twice with PBS and incubated with 80 μM DCFH-DA (Life Technologies), prepared in complete cell culture medium, for 30min at 37°C. They were harvested by scraping, and DCFH-DA fluorescence intensity (FI) was measured at λ_exc/λ_em 480/570 nm (Agilent Technologies). After background removal (λ_exc/λ_em 480/650 nm), DCF fluorescence was normalized to protein concentration.

Transepithelial electrical resistance (TEER) was used as a measure of cell monolayer integrity and was assessed before and after all treatments. Caco-2 cells were seeded at a density of 2 × 10⁵ cells per ml onto polycarbonate membrane Transwell inserts with 0.4 μm pore size (Corning, Inc.; Lowell, MA). Cells were cultured for 21 days to reach differentiation, and growth media were refreshed every 2–3 days. Fully differentiated monolayers showed TEER values of 500–700 Ohms. Differentiated Caco-2 monolayers were treated with Raw-, Unstarted-, and Started-BCP at a concentration of 100 μg ml⁻¹ for 24 h, and a set of samples was exposed to a mix of inflammatory cytokines in the last 18 h to mimic chronic intestinal barrier dysfunction associated with inflammation. Cytomix consisted in IL-1β (25 ng ml⁻¹), TNF-α (50 ng ml⁻¹), and IFN-γ (50 ng ml⁻¹). At the end of the experiment, plates were then transferred to a Thermoplate set at 37°C and TEER was measured using an epithelial volt-ohm meter with a chopstick electrode (Millicell ERS-2, EMD Millipore, Billerica, MA). The electrode was immersed at a 90° angle with one tip in the basolateral chamber and the other in the apical chamber. Care was taken to avoid electrode contact with the monolayer. An insert without cells was used as a blank and its mean resistance was subtracted from all samples. TEER was expressed as Ohms × cm².

Fluorescein isothiocyanate-dextran (FITC-dextran; MW 4 kD; Sigma Aldrich) was used as a paracellular marker for Caco-2 cell monolayers. At the end of 24 h of TEER measurement, cells were washed with PBS and 1 mg ml⁻¹ of FITC-dextran in PBS was added to the apical side of the cell monolayer, and in the basolateral compartment only PBS. Two hundred microliters of samples were collected from the basolateral compartment 2 h later and transferred into 96-well plates, and the diffused fluorescent tracer was measured by fluorometry (λ_exc/λ_em 485/528 nm). The intensity of FITC-dextran fluorescence was measured by a fluorescence spectrophotometer (Agilent Technologies, Santa Clara, CA, United States) at excitation/emission of 495/525 nm.

Normal human keratinocyte NCTC 2544 (National Institute on Cancer Research, Italy) were cultured at 5% CO₂, 37°C on RPMI medium containing 2 mM l-glutamine, 1% of penicillin (100 U ml⁻¹), and streptomycin (100 U ml⁻¹), supplemented with 10% FBS (basal medium). Cells were incubated in 25 cm² surface culture flasks at 37°C with 5% CO₂ until ca. About 80% of confluence was reached. Cells were then harvested with trypsin/EDTA and seeded at a density of 5 × 10⁴ cells per well into 96-well plates for MTT assay and 1 × 10⁶ cells per well into 12-well plates for qRT-PCR, respectively.

Twenty-four hours after seeding on 12-well plates, NCTC2544 80% confluent cells were exposed to Raw-, Unstarted-, and Started-BCP (100 μg ml⁻¹) for 16–24 h. A set of samples was simultaneously exposed to LPS 10 μg ml⁻¹. RPMI medium with 2.5% FBS, 2 mM l-glutamine, and 1% of penicillin (100 U ml⁻¹) and streptomycin (100 U ml⁻¹) was used as basal medium. Cells in basal medium were used as negative control and cells incubated only with LPS were used as the positive control. After the medium treatment, RNA for qRT-PCR analysis was extracted. Tri Reagent (Sigma Aldrich) method as described by Chomczynski and Mackey (1995) was used. cDNA was then synthesized from 2 μg RNA template in a 20 μl reaction volume, using the PrimeScript RT-PCR Kit (Takara, Japan). cDNA was amplified and detected by the Stratagene Mx3000P Real-Time PCR System (Agilent Technologies). The amplification of cDNA from NCTC2544 cells was conducted using the following Taqman gene expression assays: Hs00174128_m1 (TNF-α) as target gene and Hs999999 m1 [human glyceraldehyde-3-phosphate dehydrogenase (GAPDH)] as housekeeping genes, respectively. PCR amplifications were carried out in a 20 μl of total volume. The mixture of reaction contained 10 μl of 2X Premix Ex Taq (Takara), 1 μl of 20× TaqMan gene expression assay, 0.4 μl of RoX Reference Dye II (Takara), 4.6 μl of water, and 4 μl of DNA. PCR conditions were the following: 95°C for 30 s followed by 40 cycles of 95°C for 5 s, 60°C for 20 s. PCR reactions were performed using a MX3000p PCR machine (Stratagene, La Jolla, CA). Δ cycle threshold (Vigetti et al., 2008) was used for the calculation of the relative abundance in the expression of each gene.

Analyses were carried out in triplicate on three biological replicates for each condition. Data were subjected to ANOVA test for multiple comparisons (one-way ANOVA followed by Tukey’s procedure at p < 0.05), using the statistical software, Statistica 7.0 (Statsoft).

Started-BCP was inoculated at the final density of ca. 8 Log CFU g⁻¹ with the selected mixed starter composed by A. kunkeei PF12, PL13, and PF15, and H. uvarum AN8Y27B. The aptitude of strains as well the process parameters were previously investigated by Di Cagno et al. (2019a). The initial cell densities of lactic acid bacteria and yeasts in the Unstarted-BCP were 5.41 ± 0.21 and 6.53 ± 0.27 Log CFU g^−1, respectively. During fermentation of Started-BCP, the cell density of the lactic bacteria in BCP reached ca. 9 Log CFU g⁻¹ after 96 h, remained almost stable until 144 h, then decreased (p < 0.05) to 7.15 ± 0.29 Log CFU g⁻¹ throughout the incubation time. On the other side, during the spontaneous fermentation of Unstarted-BCP, lactic bacteria reached a cell density of ca. 9 Log CFU g⁻¹ only after 120 h, and suddenly decreased (p < 0.05) to 4.0 ± 0.31 Log CFU g⁻¹. During the first 24 h of incubation, cell density of yeast slightly increased (p > 0.05) both in Started- and Unstarted-BCP, then progressively decreased throughout the incubation time until ca. 4 Log CFU g⁻¹.

Bioaccessibility of nutrients and functional compounds, intended as the fraction of their total amount that is available for human metabolism, represent a critical issue for BCP deserving investigation. This usually entails implementation of in vitro simulated gastrointestinal digestion. Following pancreatic digestion and dialysis, the total phenolics in the serum-available fraction obtained from Raw-BCP was 1.67 ± 0.07 g l⁻¹, whereas serum-available phenolics from Started-BCP were significantly (p < 0.05) higher of ca. 22%. Phenolics availability did not change (p > 0.05) in Unstarted-BCP.

Preliminarily, the cytotoxicity of BCP samples was investigated through the MTT and LDH assays (Figure 1). A slight dose-dependent decrease of cell viability was observed in Caco-2 cells due to the exposure to BCP (Figure 1A). BCP treatment up to 100 μg ml⁻¹ showed no cytotoxic effects. Higher dose treatment (500 μg ml⁻¹) with Raw-BCP significantly (p < 0.05) but slightly impaired Caco-2 cells proliferation, whereas cells viability still approached the control level following the treatment with 500 μg ml⁻¹ of Unstarted- and Started-BCP. Regarding the LDH release (Figure 1B), membrane stability strongly decreased following the treatment with 500 μg ml⁻¹ of BCP, whereas negligible decreases were detected up to 100 μg ml⁻¹. With both 10 and 500 μg ml⁻¹ treatments, Started-BCP led to lower LDH release compared to Raw- and Unstarted-BCP. To avoid cytotoxicity from BCP, a concentration of 100 μg ml⁻¹ was selected for the subsequent assays on Caco-2 cells as the highest dose without substantial cytotoxic effects.

To evaluate potential interactions between BCP treatments and sinthesis of pro-inflammatory mediators by Caco-2 cells, their secretion in the culture medium was quantified by ELISA. Under regular physiological conditions, Raw- and Unstarted-BCP treatments slightly (p < 0.05) enhanced the release of IL-8 compared to the untreated Caco-2 cells (negative control; Figure 2). A slight but not significant (p > 0.05) increase was induced by the Started-BCP (Figure 2). Further treatment with the pro-inflammatory IL-1β stimulated the secretion of IL-8 compared to the negative control. The effect of the pro-inflammatory stimulus was significantly (p < 0.05) counteracted by the Raw- and Unstarted-BCP treatments, and was fully inhibited by the Started-BCP. A similar trend was observed for IL-6 (Figure 2). BCP-treatments lightly induced the IL-6 release under regular physiological conditions, but strongly counteracted the stimulus of IL-1β, with the highest inhibitory effect on IL-6 release observed with the Started-BCP treatment (Figure 2).

Bee-collected pollen treatments did not significantly (p < 0.05) modified the secretion of MCP-1, TNF-α, and PGE2 by Caco-2 cells under regular physiological conditions, but strongly counteracted the pro-inflammatory stimulus induced by the IL-1β or the cytokines mix (Figures 2, 3). Overall, Started-BCP exterted the highest (p < 0.05) inhibitory effect, followed by Raw- and Unstarted-BCP.

To gain insights into protective effects of BCP against oxidative stress, we investigated the accumulation of ROS within Caco-2 cells under the regular redox cell state as well under H₂O₂-induced oxidative stress. Intracellular ROS were measured by the probe 2',7'dichlorofluorescindiacetate (DCFH-DA) oxidation (Figure 4). BCP treatments did not induce any oxidative stress in Caco-2 cells. Intracellular ROS significantly (p < 0.05) increased when the cells were treated with H₂O₂ (164 ± 22 FI) compared to the negative control (23 ± 4 FI). Pretreatment with BCP significantly counteracted the increase of ROS, in spite of the oxidative stress induced by H₂O₂, with the lowest (p < 0.05) level of ROS (63 ± 16 FI) detectable with Started-BCP (Figure 4).

The epithelial barrier function was monitored through the measurement of inflammation-induced changes of TEER and permeability in Caco-2 cells monolayers. Without a pro-inflammatory stimulus, BCP did not affect significantly (p > 0.05) the TEER with respect to the negative control (untreated cells; Figure 5). The exposure of Caco-2 cells to the mix of inflammatory cytokines induced a significant (p < 0.05) decrease of TEER (ca. 68%; Figure 5). Pretreatment of cells with BCP exerted a protective effect. After incubation with the pro-inflammatory stimulus, the lowest (p < 0.05) decrease of TEER was found with the Started-BCP treatments (ca. 26%; Figure 5). The protective effect was less effective when Caco-2 cells were treated with Raw- and Unstarted-BCP, leading to a decrease of TEER of ca. 43 and 37%, respectively (Figure 5).

Accordingly to the changes of TEER, the exposure of Caco-2 cells to the pro-inflammatory cytomix significantly increased the permeability of cells monolayers (ca. 537%; Figure 6). BCP treatments significantly attenuated the inflammatory-induced permeability, with the lowest increase observed with the Started-BCP (ca. 113%), and followed by the Raw- and Unstarted-BCP treatments (ca. 334 and 265%, respectively; Figure 6).

Preliminarily, the cytotoxicity of BCP samples toward human keratinocytes was investigated through the MTT assay. BCP treatment up to 100 μg ml⁻¹ showed no cytotoxic effects, thus a concentration of 100 μg ml⁻¹ was selected for the subsequent assays on keratinocyte cells. The inflammatory response induced by LPS was evaluated by measuring the relative expression of TNF-α gene (Figure 7). Results showed that the levels of TNF-α mRNA were augmented by LPS-induced inflammation. Cells treatment with Started-BCP and, to a lesser extent, with Unstarted-BCP reduced (p < 0.05) the transcriptional levels of pro-inflammatory cytokine TNF-α (Figure 7).

DP was employed by the company Giuliani S.p.A (Milan, Italy).

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.