The stem bark of C. brasiliense Cambess. was collected in June 2010 by LMSL (authorization number 22698-1 Ministério do Meio Ambiente, Brazil), at the source of the Coxipó River (S15°38′40.8′′, W056°03′05.6′′), Cuiabá, MT, Brazil. A voucher specimen (# 37993) was deposited at the Herbarium of Federal University of Mato Grosso (UFMT), Brazil, and was identified by Harri Lorenzi M.Sc., Instituto Plantarum de Estudos da Flora, Nova Odessa, SP, Brazil. The preparation of HECb, as well as brasiliensic (Bras. acid) and isobrasiliensic acid (Isobras. acid) isolation process, were as previous described (Lemos et al., 2016).

All animal procedures were performed at the University of Liverpool with UK Home Office approval. In vivo experiments were performed in male INS-Gas mice on the FVB/N (Wang et al., 1993) background bred and maintained at the University of Liverpool Biomedical Services Unit. Primary gastric gland cultures were generated from male C57BL/6 mice acquired from Charles River, Margate, UK.

Helicobacter felis (ATCC 49179) was cultured for 72–96 h at 37°C on Columbia chocolate agar plates in a microaerophilic environment generated by Campygen atmosphere generating packs in an anaerobic jar (all Oxoid, Basingstoke, UK). For colonization of mice, the organism was harvested into tryptone soy broth and bacterial density was estimated by optical density at 600nm. An estimated bacterial density in excess of >10⁸ CFU/mL was required to progress to gavage.

Groups of at least five male INS-Gas aged 6 weeks were administered 0.5 ml H. felis suspension by oro-gastric gavage on three occasions over one week. Successful H. felis colonization was confirmed 2 weeks after the final gavage procedure by quantitative PCR for FlaA in fecal DNA (Duckworth et al., 2015b). At this time, drinking water was supplemented with 2% Tween 20, HECb 33 mg/L (lHECb, approximately 10 mg/kg/day) or HECb 133 mg/L (hHECb, approximately 40 mg/kg/day) and made available ad-libitum. Equivalent uninfected control groups were also studied. Animals were culled by cervical dislocation 6 weeks after H. felis colonization. Corpus and antrum mucosal samples were taken for histopathology and immunohistochemistry studies. The remainder of gastric tissue was homogenized in PBS with protease inhibitor for quantification of cytokines by electrochemiluminescent assay.

Gastric tissues were fixed in 4% formalin in PBS for a minimum of 12 h, processed into paraffin wax embedded blocks by standard methods and sectioned at 4 μm thickness for all staining techniques. Corpus and antrum were stained with hematoxylin and eosin (HE) for histopathological evaluations. Immunohistochemical analysis of corpus was also performed.

Gastric corpus mucosa was labeled by immunohistochemistry for proliferation (Ki67 primary antibody, AbCam, Cambridge, UK), apoptosis (cleaved caspase 3, AF835, R and D Systems, Minneapolis, MN, USA), and tyrosine 204 phosphorylation state specific ERK (sc-7383, SantaCruz Biotechnology, Dallas, TX, USA) by immunohistochemistry. All primary antibodies were raised in rabbit and were visualized using the Impress HRP system (Vector laboratories, Peterborough, UK).

Histological tissue sections were scored by an investigator blinded to sample identity, using a modified visual analog scale (Rogers, 2012). To quantify cell numbers in the gastric corpus mucosa, 10 areas per mouse, with well oriented gastric glands, forming a well visualized epithelial monolayer were chosen. Ki67 scoring was performed using a 10 mm × 10 mm eyepiece graticule divided into 1 mm squares which was overlapped along the chosen area using a ×40 objective. Number of positive cells per square were recorded as previously described (Burkitt et al., 2013). Apoptotic and ERK phosphorylation events were scored by examining the number of positively stained cells in 10 high powered fields per section, using a ×63 objective. All results were expressed as mean ± SEM.

Twelve cytokines were measured in the same samples of gastric homogenate of mice infected or not with H. felis by multiplexed electrochemiluminescence cytokine immunoassays (Meso Scale Discovery, Gaithersburg, MD, USA). Specifically, these were a Th1/Th2 standard 10-plex panel consisting of IFN-γ, IL-1β, IL-2, IL-4, IL-5, KC-GRO, IL-10, IL-12 p70, IL-13, and TNF. In addition, simplex IL-17 and IL-23 assays were performed in parallel, as Th17 responses are strongly associated with Helicobacter infections. The part of the stomach between corpus and antrum of H felis infected and uninfected INS-Gas mice was homogenized twice in PBS with protease inhibitors (SigmaFast, Sigma–Aldrich, UK), using the TissueLyserII (QIAGEN, Tokyo, Japan) at 25 Hz for 3 min. After centrifugation (4°C, 12,000 rpm, 10 min), supernatants were transferred to a clean tube, and stored at −80°C until use. Immediately before analysis, samples were clarified by further centrifugation (4°C, 12,000 rpm, 10 min). Electrochemiluminescence analysis was performed according to the manufacturer’s instructions. A standard curve for each analyte was curve-fitted and allowed determination of the concentration in pg cytokine/mL.

Gastric epithelial cultures were generated as previously described (Duckworth et al., 2015a) and were maintained in 12-well tissue culture plates on glass cover slips (Appleton Woods, Selly Oak, UK) that contained 1.0 mL/well DMEM-Ham’s F-12 mix (Sigma–Aldrich), 10% fetal calf serum (Invitrogen, Paisley, UK), 1.25% L-glutamine (Sigma–Aldrich), and 1% antibiotic/antimycotic mixture (Sigma–Aldrich). Following digestion and plating, glands were maintained at 37°C in a humidified environment containing 5% CO₂ for 24 h. Media was changed to fresh complete media after a further 24 h. Forty-eight hours after initial plating, cells were treated with HECb, brasiliensic and isobrasiliensic acids (12.5 to 100 μg/mL) for 24 h, 2 h before fixation EdU was added to the culture media. Cells were fixed in 2% formaldehyde for 30 min followed by three washes in PBS. Treatments were repeated a minimum of four times using glands extracted from a different mouse on each occasion.

Primary glands were immunolabeled for cleaved caspase 3 and EdU. Two hours before the end of treatment, 200 μL of treatment media was replaced by 200 μL of 10 μM EdU and incubated to complete the treatment. Cultures were washed and fixed with 2% paraformadehyde in PBS for 30 min. Following fixation, EdU intercalation was labeled using the Click-iT EdU Alexa Fluor 594 Imaging Kit (Invitrogen, Paisley, UK) as per protocol. Subsequently non-specific protein binding was blocked with 10% goat serum and apoptotic cells were labeled with a rabbit anti-cleaved caspase 3 antibody (AF835, Cell Signalling, Beverly, MA, USA) and visualized with Alexa fluor 488 conjugated donkey anti-rabbit immunoglobulins (Invitrogen). Coverslips were mounted with Vectashield with DAPI (Vector Labs). Slides were observed using a standard Nikon fluorescent microscope, and proliferative and apoptotic events were quantified as previously described (Duckworth et al., 2015a).

The human gastric adenocarcinoma cell line AGS (ATCC CRL 1739) were grown in complete medium, consisting of Dulbecco’s Modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum, 1% L-glutamine and 1% penicillin/streptomycin at 37°C in 5% CO₂ atmosphere with humidity.

AGS cells (10⁶ cells/well) were seeded in 12-well plates, then treated or not with HECb, brasiliensic, and isobrasiliensic acids (12.5, 25, and 50 μg/mL) for 24 h. Cells were harvested, washed with phosphate-buffered saline (PBS), fixed with cold 70% ethanol and kept at –20°C until use. Cells were washed three times with PBS and stained with a solution of ribonuclease A (R4875, Sigma–Aldrich, Sao Paulo, BR, USA) at 50 μg/mL and propidium iodide (P4170, Sigma–Aldrich, Sao Paulo, BR, USA) at 20 μg/mL in PBS for 90 min, cell cycle distribution was determined by flow cytometric analysis using a BD Accuri™ C6 (BD, Piscataway, NJ, USA).

AGS cells (2 × 10⁶ cells/well) were seeded in 6-well plates, pretreated with HECb (12.5, 25, and 50 μg/mL) and a highly selective MEK1 inhibitor (PD98059) at 10 mM for 24 h. Cultures were subsequently infected with H. pylori at a MOI of 300:1 for 1 h. After incubation, cells were lyzed in ice-cold RIPA buffer supplemented with protease cocktail and phosphatase inhibitors (Sigma Fast, 10 mM sodium orthovanadate, 10 mM sodium pyrophosphate, and sodium fluoride 100 mM).

Protein lysates were subjected to SDS-PAGE before being immobilized onto nitrocellulose membranes (Biorad, USA). After transfer, membranes were blocked (20 mM Tris-HCl, pH 7.4, 125 mM NaCl, 0.2% Tween 20, 1% bovine serum albumin, 3% non-fat milk) for 1 h at room temperature and incubated for 4 h at 4°C with specific primary antibodies: p-ERK1/2 and β-actin (as above, Santa Cruz Biotechnology, Dallas, TX, USA). Blots were incubated with secondary antibody rabbit anti-mouse IGG-HRP (sc-358914, Santa Cruz Biotechnology, Dallas, TX, USA) and immunoreactive bands were visualized by chemiluminescence (ECL Amersham, USA) and detected with ChemiDoc XRS system™ software and subsequently analyzed with Image Lab™ (Biorad, Irvine, CA, USA).

Statistical analysis was performed using GraphPad Prism 6 software. Data represent mean ± SEM. Comparisons were made using 1-way or 2-way ANOVAs and Tukey’s or Sidak’s post hoc analysis as appropriate. p < 0.05 was considered significant.

To determine whether the administration of HECb affected the outcome of H. felis induced gastric pre-neoplasia, groups of five 6-week old male INS-Gas mice were infected with H. felis by oro-gastric gavage, two weeks later drinking water was supplemented with hHECb, lHECb, or vehicle (2% Tween 20). This treatment was maintained until the end of the procedure 6 weeks after the final dose of H. felis. At the end of the procedure, animals were culled and gastric epithelial tissues prepared for quantitative histopathology and electrochemiluminescent cytokine analysis. Control groups that underwent changes to their water supply, but did not receive H. felis were also maintained.

Pathological gastric lesions were quantified using an established visual analog scoring tool. Uninfected animals had low combined pathology scores (mean score 2.2 ± 0.58), with no significant differences observed in mice exposed to HECb compared to those receiving vehicle. The administration of H. felis led to the development of marked gastric corpus pathology in the vehicle group (5.6 ± 0.87). lHECb appeared to have no impact on development of gastric corpus neoplasia at this timepoint, with similar composite pathology scores (7.2 ± 0.37) compared to vehicle treated, H. felis infected mice. In contrast, composite pathology scores were partially attenuated in animals co-administered hHECb (4.0 ± 0.45) (Figures 1A,B). At this time-point, treatment with HECb had no discernable effect on inflammatory cell infiltration or parietal cell loss (Figures 1C,D), however, the gastric mucosa was 1.9 times thicker in vehicle treated H. felis infected mice compared to vehicle treated uninfected mice (p < 0.05). In HECb treated mice H. felis infection did not significantly alter mucosal thickness (Figure 1E), and mucous metaplasia was decreased on morphological scoring criteria in H. felis infected mice receiving hHECB (Figure 1F).

To further characterize the gastric epithelial immune response to H. felis infection in this model mucosal cytokine abundance was determined by electrochemiluminescent assay. Colonization with H. felis induced a Th17 polarized immune response, as previously demonstrated in this and other mouse models of Helicobacter induced gastric pre-neoplasia (Figures 2A,B).

Administration of either vehicle or HECb did not alter this overall response, however, there were subtle changes in the abundance of individual cytokines. Amongst H. felis infected mice IFN-γ, TNF, and IL-6 were all less abundant (3.3-fold, 3.1-fold, and 1.7-fold, respectively) in animals administered hHECb compared to those treated with vehicle. Treatment with lHECb also induced an apparent, though statistically not-significant, reduction in IFN-γ and IL-6 abundance, supportive of a dose response effect for HECb on production of these cytokines. lHECb had minimal impact on TNF abundance compared to vehicle (Figures 2C–E). The abundance of KC-GRO, a mouse homolog for IL-8, in vehicle treated mice was 10.9-fold more abundant in mice colonized with H. felis, compared to the uninfected group, this cytokine was unaffected by the administration of HECb (Figure 2F). These observations suggest that treatment with HECb minimally attenuates the inflammatory response induced by H. felis.

The observation that H. felis induced metaplasia was less abundant in mice treated with hHECb led us to hypothesize that HECb treatment might influence epithelial remodeling, either impacting de-differentiation of mature cell lineages, or influencing epithelial cell turnover. To characterize this, quantitative histology was used to determine the number of Ki67 positive proliferating cells (Figure 3A), and cleaved caspase 3 positive (Figure 3C) apoptotic cells in the gastric corpus mucosa of mice.

A mean apoptotic index of 3.8 (±0.53) cells per high powered field (hpf) was demonstrated in vehicle treated, uninfected mice. This did not change significantly following H. felis infection, or administration of HECb (Figure 3D). In contrast H. felis infection had a profound impact on abundance of Ki67 immunopositive cells. Uninfected mice treated with Tween20 had a proliferation index of 20.5 (±3.2), which was similar to the proliferation index of uninfected mice treated with HECb. Following H. felis infection vehicle treated mice exhibited a 2.8-fold increase in proliferative index (p < 0.01, Figure 3B). When HECb was administered at either dose Ki67 abundance was significantly lower in infected mice compared to vehicle treated controls (2.3-fold and 2.2-fold reductions for hHECb and lHECb, respectively, p < 0.01), leading to proliferation indices similar to those seen in uninfected mice.

To determine whether the anti-proliferative effect of HECb observed in H. felis infected INS-Gas mice was driven purely by its apparently modest influence on inflammation, or through an immune cell independent mechanism, primary cultures of murine gastric glands were generated. In our hands these cultures can be maintained for in excess of 5 days, and have previously been shown to contain cells of each of the major gastric epithelial lineages (Duckworth et al., 2015a).

Cultures were generated from male C57BL/6 mice. On the third day of culture, gastric glands were treated with rising concentrations of HECb or its constituents, brasiliensic acid, and isobrasiliensic acid. Cells were fixed at 24 h. Epithelial cell proliferation was assayed by quantifying the percentage of cells that had intercalated EdU, apoptosis was quantified by immunofluorescence for cleaved caspase-3. Each treatment was performed on cultures derived from four individual mice.

In untreated cultures, 8.2% (±0.58) of cells intercalated EdU into their DNA. At different doses of treatment with HECb, brasiliensic acid, and isobrasiliensic acid all suppressed proliferation (Figures 4A–C). Significant suppression of proliferation was observed following treatment with HECb at doses in excess of 25 μg/mL. Brasiliensic acid partially suppressed proliferation at 12.5 μg/mL and had more pronounced effects at doses in excess of 25 μg/mL. Isobrasiliensic acid suppressed proliferation at doses of 50 μg/mL and 100 μg/mL.

All three compounds also induced apoptotic responses. Cytotoxicity in this model occured following treatment with 100 μg/mL of HECb or Brasiliensic acid. 100 μg/mL HECb induced a 13.3-fold increase in apoptosis compared to untreated glands (42% apoptotic cells, p < 0.001), whilst 100 μg/mL Brasiliensic acid triggered 55% (p < 0.0001) of cells to become apoptotic. Isobarasiliensic acid treatment induced apoptosis at both 50 μg/mL and 100 μg/mL with, respectively, 31 and 52% of cells shown to be apoptotic (Figures 4D–F). These observations demonstrate that HECb and its constituents induce cell cycle arrest and apoptosis in untransformed epithelial cell culture, suggesting that there is a direct epithelial effect of these compounds.

To characterize whether the impact of HECb and its constituents on gastric epithelial cell proliferation was an isolated phenomenon in the ex-vivo culture setting, or whether the same effects are identifiable in transformed cell lines, the cell cycle dynamics of AGS cells treated with HECb and its constituent chromanones was characterized by propidium iodide FACS analysis. Experiments were repeated a total of 4 times for each treatment.

In untreated AGS cultures, 0.11% (±0.02 %) of cells were identified in the pre-G1 apoptotic phase, 37.7% (±1.14%) were in the G1 phase, 21.8 % (±2.52 %) of cells were in S phase and 35.7% (±1.7 %) were in G2M phase (Figure 5). No increase in the proportion of cells in pre-G1 was observed when AGS cells were treated with HECb or its constituents at 25 or 50 μg/mL. This is in keeping with our findings in primary cell culture where apoptosis was not induced when these concentrations were tested.

Hexane extract of Calophyllum brasiliense stem bark (HECb) significantly decreased the proportion of cells in S-phase at both 25 and 50 μg/mL (9.9 and 8.3%, of cells in S-phase, respectively, Figure 5A), with a reciprocal increase in proportion of cells in G1 (45.5 and 54.9%, p < 0.05 and p < 0.0001 respectively for 25 and 50 μg/mL HECb). Brasiliensic acid induced a similar reduction in percentage of cells in S-phase at a dose of 50 μg/mL (13.0% of cells in S-phase, p < 0.01 Figure 5B), with an increase in proportion of cells in G1 observed (52.3%, p < 0.01). Both 25 and 50 μg/mL isobrasiliensic acid also reduced the proportion of cells in S-phase compared to control (11.5 and 9.4% of cells in S-phase, respectively, both p < 0.01 Figure 5C). Intriguingly, however, lower dose isobrasiliensic acid induced an increase in proportion of cells in G1 (52.9 %, p < 0.001), similar to that observed in cells treated with either HECb or brasiliensic acid, whilst higher dose isobrasiliensic acid appeared to induce G2M arrest with an increase in the number of cells in this phase (45.4%, p < 0.05).

These data demonstrate that HECb and its constituents are capable of inducing cell cycle arrest in transformed cell lines. The evidence that isobrasiliensic acid induced G1 arrest at low dose and G2M arrest at higher doses attests to these compounds potentially acting through more than one mechanism, dependent upon the drug dosing regime.

To further characterize how HECb affects Helicobacter induced proliferation, we pre-treated AGS cells with either HECb or the MEK 1 inhibitor PD98025 for 24 h. Subsequently cells were co-cultured with H. pylori at a multiplicity of infection of 300:1 for 1 h. A well characterized strain of H. pylori was used for these asssays, rather than H. felis as the effects of this organism on human cell culture are better characterized than those of H. felis, and H. pylori infection is more relevant to human disease.

The abundance of p-ERK in whole cell lysates was quantified by Western blotting. All experiments were repeated three times. Blot densitometry was performed to quantify relative expression of p-ERK compared to pan-actin abundance.

Exposure of AGS cells to H. pylori for 1h induced phosphorylation of ERK 1 and 2. When cells were pre-treated with HECb we observed significantly less phosphorylation of ERK at all doses that were administered (Figures 6A,B).

To determine whether ERK phosphorylation was also involved in the reduction of gastric epithelial cell proliferation in response to H. felis infection, in vivo gastric corpus tissue samples from mice infected with H. felis or not, and treated with hHECb, lHECb or vehicle were immunostained for p-ERK. The number of cells expressing p-ERK was determined by quantitative immunohistochemistry.

The gastric corpus of vehicle treated, uninfected mice exhibited 1.3 (±0.2) p-ERK positive cells per high power field. Administration of HECb did not significantly influence this in uninfected mice, however, administration of H. felis induced a 6.4-fold (p < 0.001) increase in p-ERK positive cells in mice treated with vehicle. H. felis induced phosphorylation of ERK was almost entirely suppressed by treatment with either of the tested doses of HECb (Figures 6C,D).

This suggests that regulation of a classical MAPK pathway may be targeted directly or indirectly by HECb administration both in vitro and in vivo.