Ellagitannins from Castanea sativa Mill. Leaf Extracts Impair H. pylori Viability and Infection-Induced Inflammation in Human Gastric Epithelial Cells

Helicobacter pylori (H. pylori) is an etiologic factor of peptic ulcer disease and gastric cancer. Virulent strains of H. pylori are correlated with the severity of gastritis, due to NF-κB activation and IL-8 expression at the epithelial level. Ellagitannins have been documented for antibacterial and anti-inflammatory activities, thus suggesting their potential use in gastritis. Recently, several authors, including our group, demonstrated that tannin-rich extracts from chestnut byproducts, at present considered agricultural waste, display promising biological activities. In this work, we detected high levels of polyphenols in hydroalcoholic extracts from chestnut leaves (Castanea sativa L.). Among polyphenols, the ellagitannin isomers castalagin and vescalagin (about 1% w/w of dry extract) were identified as potential bioactive compounds. In GES-1 cells infected by H. pylori, leaf extract and pure ellagitannins inhibited IL-8 release (IC50 ≈ 28 µg/mL and 11 µM, respectively). Mechanistically, the anti-inflammatory activity was partly due to attenuation of NF-κB signaling. Moreover, the extract and pure ellagitannins reduced bacterial growth and cell adhesion. A simulation of the gastric digestion suggested that the bioactivity might be maintained after oral administration. At the transcriptional level, castalagin downregulated genes involved in inflammatory pathways (NF-κB and AP-1) and cell migration (Rho GTPase). To the best of our knowledge, this is the first investigation in which ellagitannins from plant extracts have demonstrated a potential role in the interaction among H. pylori and human gastric epithelium.


Introduction
The discovery of Helicobacter pylori (H. pylori) infection by Warren and Marshall in 1983 is relatively recent, especially considering the subsequent efforts to convince the scientific community regarding its causal association with peptic ulcers, which led to the Nobel prize awarded in 2005 [1,2]. Gastric inflammation and oxidative stress are the clinical traits of acute and chronic gastritis, in which neutrophils, macrophages, and lymphocytes infiltrate the mucosa following chemokine attraction (for a comprehensive review, see Naito et al. [3]). The major inflammatory signal is the release of IL-8, whose levels correlate with H. pylori infection and gastric disease severity [4][5][6][7], but other mediators such as

Total Phenol Content Assay
Total polyphenol content was determined according to Folin-Ciocâlteu's method, as reported by Singleton and Rossi [29]. Freeze-dried leaf extracts (1 mg) were solubilized in 1 mL of water. Aliquots of 300 µL from different samples were mixed in test tubes with 1.5 mL of Folin-Ciocâlteu's reagent diluted 10 times and 1.2 mL of 7.5% (w/v) sodium carbonate. The absorbance was measured at 765 nm using a UV-Vis spectrophotometer (Victor TM X3, Perkin Elmer, Walthman, MA, USA). Gallic acid was used as the reference standard for the calibration curve (0-30 µg/mL). Results were expressed as the weight of gallic acid equivalents (GA eq.) per weight of dry extract (% w/w).
2.6. Bacterial Culture H. pylori cag+ strain 26695 was cultured in Petri dishes with Mueller-Hinton Broth medium, 5% agar, and 25% blood, for 72 h, under a microaerophilic atmosphere (5% O 2 , 10% CO 2 , and 85% N 2 ) at 37 • C, 100% humidity. For the biological experiments, the bacterium was collected from the Petri dishes and counted according to its optical density at 600 nm (an OD value of 5 corresponds to 2 × 10 8 bacteria).

Cell Treatment
GES-1 cells were seeded in a 24-well plate (3 × 10 4 cells/well) or 12-well plate (6 × 10 4 cells/well), according to the specific investigation. Cells were treated with the proinflammatory stimulus TNFα (10 ng/mL) or H. pylori (bacterium-to-cell ratio of 50:1) for 1 h or 6 h, along with the leaf extracts (Castanea sativa Mill. var. verdesa and var. venegon) at different concentrations. The day before the bacterial infection, serum starvation was performed using 0.5% serum medium, supplemented with 1% L-glutamine and 1% penicillin/streptomycin. All the infection treatments were conducted with serumand antibiotic-free medium, while TNFα treatments were conducted only with serumfree medium. During the treatment, cells were maintained in an incubator at 37 • C and 5% CO 2 . EGCG (50 µM) and procyanidin A2 (500 µM) were used as reference inhibitors of inflammatory markers and bacterial adhesion, respectively [11,[30][31][32].

Cytotoxicity Assay
The integrity of the cell morphology before and after treatment was assessed by light microscope inspection. Cell viability was measured, after 6 h treatment (24-well plate, 3 × 10 4 cells/well), using the 3,4,5-dimethylthiazol-2-yl-2-5-diphenyltetrazolium bromide (MTT) method [33]. This method evaluates the activity of a mitochondrial enzyme, which is an index of cell viability. Briefly, the medium was discarded; then, 200 µL of MTT solution (0.1 mg/mL, phosphate-buffered saline (PBS) 1×) was added to each well (45 min, 37 • C) and kept in darkness. Then, MTT solution was discarded, the remaining purple salt was dissolved in isopropanol/dimethyl sulfoxide (DMSO) (90:10 v/v), and the absorbance was read at 595 nm (Victor TM X3, Perkin Elmer, Walthman, MA, USA). The absorbance of the solvent (blank) was subtracted from each value.
Ellagitannins and leaf extracts had no significant impact on cell viability up to the concentrations of 50 µM and 100 µg/mL, respectively ( Figure S1). The impact on cell viability up to the concentrations used for 1 h treatments, i.e., 200 µM and 250 µg/mL, respectively, was also excluded.

Measurement of IL-8 Release
Cells were seeded at the density of 3 × 10 4 cells/well (24-well plate) for 48 h. The chemokine IL-8 was quantified in cell medium in at least three independent experiments, after 6 h treatments with TNFα or H. pylori and the extracts, using a sandwich enzyme-linked immunosorbent assay (human IL-8 ABTS ELISA Development Kit, Peprotech). The assay was conducted according to manufacturer's instructions, as previously reported [34]. The absorbance of the sample at 405 nm was compared to the absorbance of human recombinant standard IL-8 (0-1000 pg/mL). Data (mean ± SEM of at least three experiments) were expressed as the percentage relative to stimulated control, to which the value of 100% was arbitrarily assigned.

NF-κB Activation
The NF-κB pathway was evaluated by immunofluorescence and plasmid transfection, to measure the translocation of p65 subunit into cell nuclei or κB element-driven transcription, respectively, in at least three independent experiments.

Immunofluorescence
The immunofluorescence technique was applied for the evaluation of NF-κB (p65 subunit) translocation into GES-1 nuclei; cells were challenged with TNFα or H. pylori and treated with the extracts, as previously described [34]. Briefly, cells were cultivated (3 × 10 4 /well) on coverslips placed in 24-well plates for 24 h. H. pylori was stained with CFSE 5 mM (2 µL of CFSE: 5 × 10 8 bacteria) and incubated for 20 min at 37 • C; then, the bacterial suspension was supplemented with FBS, washed three times (PBS 1×), and centrifuged at 3150× g for 5 min to remove the excess of CFSE not bound to the bacterium. After 1 h treatment, the coverslips were washed (PBS 1×) and fixed with 4% formaldehyde solution for 15 min at r.t. A blocking solution (5% BSA) was added to the well and incubated at room temperature for 1 h. Cells were incubated with the primary antibody (NF-κB p65 (D14E12) XP ® Rabbit mAb #8242, Cell Signaling Technology, Danvers, MA, USA) diluted 1:400 v/v overnight at 4 • C, washed three times (PBS 1×), and then incubated with the secondary antibody (Alexa Fluor 647 conjugated with anti-rabbit immunoglobulin G (IgG) (heavy + light (H + L)), F(ab')2 Fragment #4414, Cell Signaling Technology, Danvers, MA, USA) diluted 1:1000 v/v. A drop of ActinRed reagent, previously diluted 1:5 in PBS 1×, was added 30 min before the end of incubation. After 2 h, coverslips were washed with PBS and mounted on slides with a drop of ProLong Gold Antifade Reagent with 4 ,6-diamidino-2-phenylindole (DAPI) (#8961, Cell Signaling Technology, Danvers, MA, USA), and then imaged using a confocal laser scanning microscope (LSM 900, Zeiss, Oberkochen, Germany).

Measurement of the NF-κB-Driven Transcription
GES-1 cells were seeded in 24-well plates for 48 h (3 × 10 4 cells), and then transiently transfected with a reporter plasmid, in which the luciferase gene was under the control of the E-selectin promoter containing three κB elements responsive to NF-κB (50 ng per well). Lipofectamine ® 3000 reagent was used for the transfection assays, according to the manufacturer's instructions. The plasmid was a gift from Dr. N. Marx (Department of Internal Medicine-Cardiology, University of Ulm; Ulm, Germany). The day after, cells were treated with TNFα in addition to the extracts for 6 h. At the end of the treatment, the luciferase activity in cells was measured using Britelite TM Plus reagent, according to the manufacturer's instructions, as previously described [25]. Results (mean ± SEM of at least three experiments) were expressed as the percentage relative to the stimulated control, to which a value of 100% was arbitrarily assigned.

Bacterial Adhesion to Cells
The bacterial adhesion was measured using a cytofluorimetric method adapted from Messing and colleagues [30]. GES-1 cells were seeded in 12-well plates for 48 h (6 × 10 4 cells) before the infection with FITC-labeled H. pylori. FITC (2 µL) solution (1% in DMSO) was added to 10 8 bacteria suspended in PBS 1× and incubated for 45 min at 37 • C; then, the bacterial suspension was centrifuged (3150× g, 5 min) and washed twice (PBS 1×) to remove the excess probe, before resuspending in PBS 1× for cell infection. Cell treatment (at least three independent experiments) was carried out for 1 h with extracts and a reference inhibitor (procyanidin A2, 500 µM), before (pretreatment) or after (cotreatment) infection. After 1 h (37 • C), cells were washed twice with PBS 1×, collected through a scraper in PBS/EDTA 2 mM, and centrifuged (3150× g for 5 min); then, they were fixed by formaldehyde (4% in PBS) and incubated in an ice bath for 10 min. Finally, cells were centrifuged, washed, and resuspended in 0.5% BSA (PBS/EDTA 2 mM) for analysis using

Minimum Inhibitory Concentration (MIC)
The microbroth dilution method was performed according to the recommendations of the Clinical and Laboratory Standards Institute (CLSI) [35] and was used to evaluate the MIC value, in at least three independent experiments. Extracts at different concentrations and the positive control (tetracycline 0.125 µg/mL) were prepared in 5% FBS Brucella broth; next, 100 µL of each sample were placed in a 96-well U-bottom plate. Then, 100 µL of H. pylori suspension (OD = 0.1) prepared in the same medium was added to each well. After mixing well, the 96-well plate was incubated at 37 • C in a 5% CO 2 incubator under microaerophilic conditions. After 72 h, the rate of bacterial growth was measured using a microplate reader at 600 nm (Victor TM X3, PerkinElmer, Waltham, MA, USA).

In Vitro Simulated Gastric Digestion
The gastric digestion was mimicked by an in vitro simulation, as previously described [34]. In brief, the extracts (100 mg) were incubated for 5 min at 37 • C with reconstituted saliva (6 mL); then, gastric juice (12 mL) was added, and the sample was incubated for 2 h at 37 • C under shaking. Finally, the suspension was centrifuged (5 min, 3000× g), and the supernatant was freeze-dried.

RNA Sequencing
GES-1 cells were seeded in a 12-well plate at a density of 6 × 10 4 cells/well for 48 h and mRNA isolated using miRNeasy Mini Kit (Qiagen, Hilden, Germany). Cells were treated for 6 h with H. pylori, with or without castalagin (10 µM). Three replicates were used for each condition (ctrl, H. pylori treatment, H. pylori, and castalagin treatment).
The quality and the concentration of the mRNA were assessed with RNA Screen Tape (Agilent Technologies, Santa Clara, CA, USA). The mRNA (500 ng) was fragmented and converted into complementary DNA (cDNA) through Illumina ® Stranded mRNA Prep (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. The quality and concentration of the final dual-indexed libraries were checked with D1000 Screen Tape (Agilent Technologies, Santa Clara, CA, USA). The NextSeq 500/550 High-Output Kit v2.5 was used to perform the sequencing through NextSeq 550 instrument (Illumina, San Diego, CA, USA).

RNA Sequencing Data Analysis
For RNA sequencing data analysis, Illumina blc2fastq software was used to generate the Fastq files. The quality of each Fastq was inspected individually using the FASTQC tool vs. 0.11.9 [36], and Multiqc vs. 1.10.1 was then used to assess the overall good sequencing quality [37]. After the quality check, reads were aligned to the human reference genome GRCh38.p13 using STAR 2.7.9a [38], while Feature Counts 2.0.1 was used to obtain the counts for each sample [39] (see Table S1). The rlog function was applied to transform the count data, reducing differences among samples, and normalizing for the library size. Transformed data were used to perform principal component analysis (PCA) to evaluate the sample distribution and their clustering within groups representing the same condition. Normalization and differential gene expression analyses were carried out using the Bioconductor package DeSeq2 [40]. Following normalization, differentially expressed genes (DEGs) were detected. The Wald test in DESeq2 is the default test used for hypothesis testing. Genes were considered differentially expressed when they had AdjPval < 0.05 and log 2 FoldChange ≥ 0.58, indicating a fold change > 1.5 in either direction. The DEGs were then used to perform the enrichment analysis with the R package "EnrichR" [41]. The enriched pathways were determined using the GO_Biological_Process_2021 database.

Statistical Analysis
All biological results were expressed as the mean ± SEM of at least three independent experiments; the confidence interval related to the half-maximal inhibitory concentrations (IC 50 ) calculation is reported in Section 3. Data were elaborated through an unpaired ANOVA test and Bonferroni post-hoc analysis. Statistical assessment and IC 50 calculation were conducted using GraphPad Prism 8.0 software (GraphPad Software Inc., San Diego, CA, USA). Values of p < 0.05 were considered statistically significant.

Phytochemical Characterization
Castanea sativa Mill. leaf is a well-known source of polyphenols, which include tannins and flavonoids [23,24]. According to our previous work on chestnut-derived extracts [25], we selected leaves from two varieties (Castanea sativa Mill. var. venegon and Castanea sativa Mill. var. verdesa) to perform a polar extraction with hydroalcoholic solvent (50:50 water/ethanol). The extraction yields were 16.24% and 21.88% (wt. of dry extract/wt. of dry plant material), respectively.
According to the literature, tannins represent 6-8% w/w of polar extracts from leaves [23]. The chemical nature of tannins from leaves has been poorly investigated in comparison to the bark, which should contain high levels of hydrolyzable tannins (up to 10%), including ellagitannins, such as the isomers castalagin and vescalagin, and their hydrolysis products, castalin and vescalin [20]. For this reason, we firstly characterized the extracts from leaves using LC-MS analysis, thus confirming the presence of castalagin and vescalagin.
The analytes, castalagin and vescalagin, were identified by comparing the retention time and the m/z with the respective standard compounds, using the following mass transitions: 933/631 (castalagin) and 933/613 (vescalagin) ( Figure S2). The quantitative analysis was validated in terms of LOD and LOQ; thus, the ellagitannins were measured using the linear regression curve of the respective analytical standards (Table 1). In parallel, we measured the total phenolic content in the extracts from venegon and verdesa varieties, which was 26.59% and 28.97% of weight GA eq./weight of dried extract, respectively ( Table 2). The analysis was repeated after an in vitro simulated gastric digestion (see Section 2), and the process slightly diminished the total phenol level in both varieties, with statistically significant difference just for verdesa variety. Quantitative analysis using LC-MS/MS of castalagin and vescalagin showed that the two isomers overall represented 1.07% (10.69 µg/mg) and 1.24% (12.36 µg/mg) of the extracts from verdesa and venegon varieties, respectively. To gain more insight into the gastric stability of the two ellagitannins, their amount was also measured following the simulated gastric digestion as well. Reduction was statistically significant for both verdesa and venegon varieties: a fraction of castalagin (around 40% and 70%, respectively) was preserved, while vescalagin was not detectable after gastric digestion. The results are summarized in Table 3.

Castanea sativa Mill. Extracts Impair Inflammatory Markers Typically Increased during Gastritis
H. pylori infection and antibacterial effectors, such as TNFα, induce the release of IL-8 from gastric epithelial cells through the modulation of pathways converging in the NF-κB activation [6,42,43]. Thus, we carried out experiments to investigate the bioactivity of leaf extracts in both TNFα-and H. pylori-induced GES-1 cells ( Figure 1). Firstly, we evaluated the effect of the leaf extracts on IL-8 release in cells challenged with TNFα (10 ng/mL): both the extracts inhibited IL-8 release in a concentration-dependent fashion ( Figure 1A), with IC 50 lower than 10 µg/mL. Then, the effect was also demonstrated on H. pylori-infected cells at slightly higher concentrations (IC 50 < 30 µg/mL) ( Figure 1B).   The in vitro simulated gastric digestion showed a moderate impact on the inhibitory activity ( Figure 1C,D), as evident by the comparison of the IC 50 values reported in Table 4. Since castalagin and vescalagin were found in leaf extracts, and castalagin was partially stable after the simulated digestion, we addressed their potential effect on IL-8 secretion triggered by inflammatory stimuli or H. pylori infection. Ellagitannins strongly inhibited TNFα-induced IL-8 release with comparable IC 50 (0.27 and 0.22 µM, respectively) ( Figure 2A). Notably, they also inhibited H. pylori-induced IL-8 release, albeit at higher concentrations ( Figure 2B). The IC 50 values of castalagin and vescalagin with these parameters are summarized in Table 5. Results are expressed as the mean ± SEM (n = 3) of the relative percentage in comparison to stimulus (black bar), to which the value of 100% was arbitrarily assigned. ** p < 0.01, and *** p < 0.001 vs. stimulus. Our group previously reported that ellagitannins can interfere with the NF-κB pathway in gastric epithelial models [25,44]. Consequently, we wondered whether the impairment of NF-κB pathway contributes to the IL-8 inhibition observed in the presence of the leaf's extracts and pure ellagitannins. Firstly, we tested the underlying mechanism using the plasmid transfection assay, observing that extracts (25-100 µg/mL) and pure ellagitannins (0.5-10 µM) counteracted the NF-κB-driven transcription induced by TNFα ( Figure S3). The bioactivity of the extracts was maintained after the simulated digestion, albeit at higher IC50 (>100 µg/mL), thus remarking that gastric environment may cause a partial reduction in chemical stability, in line with data regarding IL-8 release (Table 5). However, castalagin and vescalagin exhibited IC50 lower than 1 µM toward NF-κB-driven Results are expressed as the mean ± SEM (n = 3) of the relative percentage in comparison to stimulus (black bar), to which the value of 100% was arbitrarily assigned. ** p < 0.01, and *** p < 0.001 vs. stimulus. Our group previously reported that ellagitannins can interfere with the NF-κB pathway in gastric epithelial models [25,44]. Consequently, we wondered whether the impairment of NF-κB pathway contributes to the IL-8 inhibition observed in the presence of the leaf's extracts and pure ellagitannins. Firstly, we tested the underlying mechanism using the plasmid transfection assay, observing that extracts (25-100 µg/mL) and pure ellagitannins (0.5-10 µM) counteracted the NF-κB-driven transcription induced by TNFα ( Figure S3).
The bioactivity of the extracts was maintained after the simulated digestion, albeit at higher IC 50 (>100 µg/mL), thus remarking that gastric environment may cause a partial reduction in chemical stability, in line with data regarding IL-8 release (Table 5). However, castalagin and vescalagin exhibited IC 50 lower than 1 µM toward NF-κB-driven transcription, thus reflecting the inhibitory effect observed for leaf extracts (IC 50 values are summarized in Table S2).
The impact on the NF-κB pathway was further evaluated by confocal microscopy experiments. At the selected concentration (200 µg/mL), the extracts interfered with NF-κB (subunit p65) translocation into the nucleus, during both TNFα (Figure 3) and H. pylori challenges (Figure 4). At the same concentration, the bioactivity of the extracts subjected to simulated gastric digestion was moderately affected, as previously observed.  In this case, we also confirmed the parallelism between the biological effects of the extracts and pure ellagitannins during H. pylori infection. As expected, castalagin and vescalagin impaired the nuclear translocation of p65 at the highest concentration responsible for the total inhibition of IL-8 release (50 µM) ( Figure 5). The overall data sustain that the extracts may counteract IL-8 release by interfering with the NF-κB pathway, and that pure ellagitannins present in leaves may be responsible for this inhibitory mechanism.
For this reason, we proceeded by comparing leaf extracts and ellagitannins for potential antibacterial effect against H. pylori, such as bacterial proliferation and adhesion to gastric cells. Figure 3. Effect of Castanea sativa Mill. leaf extracts on NF-κB (p65) nuclear translocation induced by TNFα. GES-1 cells were treated for 1 h with TNFα (10 ng/mL) in addition to leaf extracts (200 µg/mL) before and after in vitro simulated digestion. NF-κB (red) nuclear translocation from the cell cytoplasm into the nuclei (DAPI, blue) was detected by confocal immunofluorescence microscopy (63× objective). Merged images include β-actin staining (green). VR, Castanea sativa Mill. var. verdesa leaf extract; VN, Castanea sativa Mill. var. venegon leaf extract; dig., in vitro simulated digestion. In this case, we also confirmed the parallelism between the biological effects of the extracts and pure ellagitannins during H. pylori infection. As expected, castalagin and vescalagin impaired the nuclear translocation of p65 at the highest concentration responsible for the total inhibition of IL-8 release (50 µM) ( Figure 5). The overall data sustain that the extracts may counteract IL-8 release by interfering with the NF-κB pathway, and that pure ellagitannins present in leaves may be responsible for this inhibitory mechanism. For this reason, we proceeded by comparing leaf extracts and ellagitannins for potential antibacterial effect against H. pylori, such as bacterial proliferation and adhesion to gastric cells.

Antibacterial Effect
A previous paper demonstrated the direct antibacterial effect of common monomeric ellagitannins on H. pylori, with MIC values below 25 µM [18]. However, to the best of our knowledge, castalagin and vescalagin have never been investigated for their antibacterial properties against H. pylori. Moreover, the antibacterial effect in coculture models of H. pylori infection has never been previously reported. For this reason, we decided to

Antibacterial Effect
A previous paper demonstrated the direct antibacterial effect of common monomeric ellagitannins on H. pylori, with MIC values below 25 µM [18]. However, to the best of our knowledge, castalagin and vescalagin have never been investigated for their antibacterial properties against H. pylori. Moreover, the antibacterial effect in coculture models of H. pylori infection has never been previously reported. For this reason, we decided to evaluate the impact of leaf extracts and their ellagitannins on bacterial growth and adhesion to gastric epithelial cells.
Both the extracts impaired bacterial growth with MICs of 100 µg/mL, although the simulated digestive process caused a significant decrease in the bioactivity, increasing the MICs to 200 µg/mL ( Figure 6A,B). Of note, castalagin and vescalagin showed MIC values of 25 µM (Figure 6C), similarly to other monomeric ellagitannins [18]. H. pylori adhesion to gastric epithelial cells is a crucial event for the pathogenesis of gastritis. Accordingly, we wondered whether leaf extracts and pure ellagitannins may counteract the bacterial adhesion through a direct effect on H. pylori or by acting on the adhesion machinery of gastric epithelial cells. Thus, we treated GES-1 cells for 1 h after (cotreatment) or before (pretreatment) H. pylori infection; the first experimental setting involved the interaction of the natural compounds with either the bacterium or human cell, whereas the second setting excluded the direct interaction with the bacteria, thus testing the involvement of variables related to the host cells.
Following the first experimental setting, the treatment with leaf extracts from verdesa and venegon varieties impaired the bacterial adhesion in a concentration-dependent fashion; the highest inhibitory effect was observed at 200 µg/mL (−42.5% and −30.1%, respectively), with comparable values after the simulation of gastric digestion (−44.5% and 52.8%, respectively) ( Figure 7A,B). Castalagin and vescalagin showed an inhibitory activity within the concentration range of 50-200 µM (−46.5% and −41.4%, respectively, at the concentration of 200 µM) ( Figure 7C). Of note, the inhibitory activity was observed at concentrations comparable to the MIC values, thus suggesting that antiadhesive properties may be due, at least in part, to the antibacterial activity.  H. pylori adhesion to gastric epithelial cells is a crucial event for the pathogenesis of gastritis. Accordingly, we wondered whether leaf extracts and pure ellagitannins may counteract the bacterial adhesion through a direct effect on H. pylori or by acting on the adhesion machinery of gastric epithelial cells. Thus, we treated GES-1 cells for 1 h after (cotreatment) or before (pretreatment) H. pylori infection; the first experimental setting involved the interaction of the natural compounds with either the bacterium or human cell, whereas the second setting excluded the direct interaction with the bacteria, thus testing the involvement of variables related to the host cells.
Following the first experimental setting, the treatment with leaf extracts from verdesa and venegon varieties impaired the bacterial adhesion in a concentration-dependent fashion; the highest inhibitory effect was observed at 200 µg/mL (−42.5% and −30.1%, respectively), with comparable values after the simulation of gastric digestion (−44.5% and 52.8%, respectively) ( Figure 7A,B). Castalagin and vescalagin showed an inhibitory activity within the concentration range of 50-200 µM (−46.5% and −41.4%, respectively, at the concentration of 200 µM) ( Figure 7C). Of note, the inhibitory activity was observed at concentrations comparable to the MIC values, thus suggesting that antiadhesive properties may be due, at least in part, to the antibacterial activity.
When experiments were carried out following the pretreatment setting, the extracts showed significant antiadhesive properties at 200 µg/mL (−53.8% and −27.5%, respectively) ( Figure 8A), thus suggesting not only a direct effect on H. pylori during infection but also a potential impact on host cell variables. However, the simulated digestion abolished this effect ( Figure 8B). Accordingly, castalagin and vescalagin were ineffective until the highest concentration tested (200 µM) ( Figure 8C). This observation led us to hypothesize that direct antibacterial activity might represent the main explanation for the antiadhesive properties. However, it is not possible to exclude potential interference with the plethora of virulence factors involved during bacterial adhesion to host cells, as further mechanisms. This intriguing aspect is still poorly investigated referring to ellagitannins and related plant sources, thus demanding further host-pathogen interaction studies.
cell, whereas the second setting excluded the direct interaction with the bacteria, thus testing the involvement of variables related to the host cells.
Following the first experimental setting, the treatment with leaf extracts from verdesa and venegon varieties impaired the bacterial adhesion in a concentration-dependent fashion; the highest inhibitory effect was observed at 200 µg/mL (−42.5% and −30.1%, respectively), with comparable values after the simulation of gastric digestion (−44.5% and 52.8%, respectively) ( Figure 7A,B). Castalagin and vescalagin showed an inhibitory activity within the concentration range of 50-200 µM (−46.5% and −41.4%, respectively, at the concentration of 200 µM) ( Figure 7C). Of note, the inhibitory activity was observed at concentrations comparable to the MIC values, thus suggesting that antiadhesive properties may be due, at least in part, to the antibacterial activity.  When experiments were carried out following the pretreatment setting, the extracts showed significant antiadhesive properties at 200 µg/mL (−53.8% and −27.5%, respectively) ( Figure 8A), thus suggesting not only a direct effect on H. pylori during infection but also a potential impact on host cell variables. However, the simulated digestion abolished this effect ( Figure 8B). Accordingly, castalagin and vescalagin were ineffective until the highest concentration tested (200 µM) ( Figure 8C). This observation led us to hypothesize that direct antibacterial activity might represent the main explanation for the antiadhesive properties. However, it is not possible to exclude potential interference with the plethora of virulence factors involved during bacterial adhesion to host cells, as further mechanisms. This intriguing aspect is still poorly investigated referring to ellagitannins and related plant sources, thus demanding further host-pathogen interaction studies.

RNA-Seq Analysis
We then wondered whether the gastroprotective properties of ellagitannins may be partly due to or be mediated by changes in the epithelial cell transcriptome. Therefore, we carried out RNA-seq analysis of GES-1 cells infected with H. pylori without or upon treatment with castalagin (10 µM).
The principal component analysis ( Figure S4) showed that replicates clustered very well according to their condition (H. pylori vs. castalagin); in fact, the analysis showed a clear separation between replicates of infected cells and replicates of infected cells subjected to castalagin treatment. The normalized counts obtained through DESeq2 analysis are reported in the Table S1.
A total of 36 differentially expressed genes were identified with three upregulated and 33 downregulated genes in RNA samples from H. pylori-infected cells treated with castalagin compared to RNA samples from infected cells with no other treatment ( Figure  9A,B).

RNA-Seq Analysis
We then wondered whether the gastroprotective properties of ellagitannins may be partly due to or be mediated by changes in the epithelial cell transcriptome. Therefore, we carried out RNA-seq analysis of GES-1 cells infected with H. pylori without or upon treatment with castalagin (10 µM).
The principal component analysis ( Figure S4) showed that replicates clustered very well according to their condition (H. pylori vs. castalagin); in fact, the analysis showed a clear separation between replicates of infected cells and replicates of infected cells subjected to castalagin treatment. The normalized counts obtained through DESeq2 analysis are reported in the Table S1.
A total of 36 differentially expressed genes were identified with three upregulated and 33 downregulated genes in RNA samples from H. pylori-infected cells treated with castalagin compared to RNA samples from infected cells with no other treatment ( Figure 9A,B). Of note, considering the same subgroup of genes, most of them (29 out of 36) were regulated by H. pylori versus control uninfected cells, thus indicating that castalagin was  Lastly, the 33 downregulated genes were used to perform an enrichment analysis to evaluate which pathways were over-represented; the identified significant pathways with the corresponding number of genes are reported in Table 7. The enriched pathway analysis suggested that the main transcriptional targets of castalagin are as follows: signals involved in response to cytokines, including transcription factors (AP-1 and NF-κB) and MAPKs (ERK1/2), transport and signals related to divalent ions, and DNA-regulatory proteins, such as targets of the polymerase II ( Figure S5).

Discussion
Castanea sativa Mill. is a well-known source of polyphenols, including gallotannins and ellagitannins, with potential gastroprotective properties. It is remarkable that byproducts from chestnut harvesting, such as chestnut peel and leaves, represent a source of bioactive compounds with potential nutraceutical use. We previously reported that the outer tissues from chestnuts, belonging to var. venegon and var. verdesa (Castanea sativa Mill.), are rich in proanthocyanidins, showing anti-inflammatory effects on gastric epithelial cells [25]; however, chestnuts are devoid of ellagitannins, contrarily to what reported for the bark [19,20].
Ellagitannins have been reported to possess anti-ulcer and antibacterial properties [12][13][14]18], but their potential role in H. pylori-related gastritis is still poorly investigated. In this work, we demonstrated the presence of ellagitannin isomers, castalagin and vescalagin, in hydroalcoholic extracts from leaves belonging to both varieties from Castanea sativa Mill. We validated their pharmacological activity in a model of the nontumoral gastric epithelium (GES-1), challenged with TNFα or H. pylori infection.
Extracts from leaf and pure ellagitannins that are typically found in Castanea sativa Mill. inhibited the release of IL-8 from GES-1 cells, with more pronounced activity against TNFα challenge (Tables 5 and 6). The mechanism of action may be ascribed, at least in part, to the impairment of the NF-κB pathway (Figures 3-5 and S3), a known crossroad of TNFα-and H. pylori-induced molecular signaling [43]. However, it is plausible that other inflammatory targets were involved since, in our experiments, IL-8 release was always inhibited at lower concentrations than those impairing the NF-κB cascade.
The same parallelism among leaf extracts and pure ellagitannins was observed in respect to the direct antibacterial and the antiadhesive properties against H. pylori. Our results are in line with those obtained by other authors, reporting that casuarinin, structurally similar to castalagin and vescalagin, could inhibit the growth of H. pylori at 13.35-26.70 µM [18]. Of note, the same authors suggested a higher specificity for H. pylori than E. coli (MIC > 100 µg/mL). Further investigations should clarify the antibacterial mechanism of action and the potential selectivity.
Regarding the possibility to attribute the bioactivity of leaf extracts to castalagin and vescalagin as main compounds, it is worth mentioning that the anti-inflammatory properties of the extracts showed a clear parallelism with the activities exhibited by ellagitannin isomers. The latter were selected as available compounds, representative for the class of ellagitannins, widely reported in Castanea sativa Mill. Indeed, especially during H. pylori challenge, pure ellagitannins exhibited their inhibitory effects at concentrations higher than those present in the extracts, thus suggesting the contribution of other hydrolyzable tannins, including ellagitannins, to the biological effect of the extract.
To obtain a more comprehensive profile of the biological activities of ellegitannins, castalagin (10 µM) was selected as a candidate compound for RNA-seq experiments. This high-throughput approach allows capturing transcriptional changes and identifying target genes and pathways. A small set of genes (n = 36) was reverted by castalagin treatment with respect to H. pylori infection; most of them were downregulated (33), thus suggesting the prevalence of an inhibitory effect on transcription. Of note, a subset of downregulated genes (tfrc, spry4, gbp1, gbp3, prag1, amigo2, and ptpre) were related to the Rho GTPase pathway; among them, amigo2 has been suggested as a biomarker of poor prognosis in gastric adenocarcinoma [45][46][47]. Despite the restricted number of modulated genes, PCA showed that the group of data related to castalagin treatment clustered from H. pylori infection, thus reflecting a significant difference among groups.
The impact of ellagitannins on the Rho GTPase pathway is still poorly described, although other sources of ellagitannins have been investigated for their anti-inflammatory and antitumoral effect at the gut level [48]. Regarding inflammation, we could not exclude, on the basis of the literature concerning ellagitannins [16], that post-transcriptional and enzymatic mechanisms may have contributed to IL-8 inhibition in our setting.
Altogether, our data suggest that the use of the whole extract, accurately standardized for bioactive compounds such as castalagin and vescalagin, may exert more beneficial activity with respect to single compounds in gastritis. Other polyphenols, whose presence has been documented in Castanea sativa Mill. leaves, such as gallotannins or flavonolderivatives [23,24], have previously been reported for anti-inflammatory and antibacterial properties at the gastric level [14,49]. In summary, our work attributed a relevant role to ellagitannins in the impairment of H. pylori viability and infection-induced inflammation in human gastric epithelial cells.

Conclusions
Our work sustains, for the first time, the characterization of ellagitannins as bioactive principles in Castanea sativa Mill. leaf extracts. We suggest that leaves should be taken into consideration as suitable plant material to produce sustainable and bioactive extracts. Moreover, to the best of our knowledge, this is the first study in which the bioactivity of ellagitannins and related plant extracts has been validated in a model mimicking the gastric epithelium-H. pylori interaction.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/nu15061504/s1: Figure S1. Effect of Castanea sativa Mill. leaf extracts and ellagitannins on GES-1 viability (MTT test); Table S1. Total number of raw and normalized counts of the RNA-seq analysis for each sample; Figure S2. LC-MS chromatograms identifying the presence of castalagin and vescalagin in leaf extracts; Figure S3. Effect of Castanea sativa Mill. leaf extracts and ellagitannins on NF-κB driven transcription; Table S2. IC 50 (µg/mL) of NF-κB driven transcription: extracts and ellagitannins from Castanea sativa Mill. leaves; Figure S4