Phospholipid formulation of curcumin (which is called “Meriva”) was purchased from sigma Aldrich; pure curcumin was synthesized as described in (Pabon, 1964; Gordon et al., 2013) and was used as a control. Curcumin and its analogues were synthesized following a modified procedure originally developed by Pabon (Pabon, 1964; Gordon et al., 2013). Curcumin has been prepared from vanillin and acetylacetone in the presence of tri-sec. butyl borate and butylamine. The reaction was carried out at room temperature in ethyl acetate. Other compounds were prepared using suitable aldehyde (0.2 mole), tributyl borate (0.4 mole) and the reaction product of acetylacetone (0.1 mole) and boric anhydride (0.07 mole) were dissolved in 100 ml of dry ethyl acetate. Butylamine was added (0.5 ml every 10 min; total 2 ml) while stirring (Pabon, 1964; Gordon et al., 2013).

H. pylori strains 26695, (Chaturvedi et al., 2015; Jang et al., 2017) PMSS1, (Chaturvedi et al., 2015; Jang et al., 2017) PZ5056G, (Chaturvedi et al., 2015; Jang et al., 2017) and B128 7.13 (Chaturvedi et al., 2015; Jang et al., 2017) were maintained on plates of tryptic soy agar containing 5% sheep blood. Before experiments, bacteria were grown overnight in Brucella broth supplemented with 10% FBS at 37°C (Johnson-Henry et al., 2004). Escherichia coli strain DH 5-α and Citrobacter rodentium strain ICC168 (Petty et al., 2010) were maintained on LB plates, and starter cultures were grown overnight in LB broth at 37°C with or without agitation, respectively (all the chemicals procured from sigma St. Louis).

Bacterial growth curves were determined. As 0.1 optical density at 560 nm (OD560) H. pylori strains were used to inoculate new Brucella broth, and bacterial growth was monitored by optical density for 24 h. Some cultures were supplemented with synthetic curcumin, (Pabon, 1964) or Meriva, and differences in growth were measured. The effect of Meriva on Citrobacter rodentium and Escherichia coli was determined in LB broth cultures begun at 0.05 OD560, and bacterial growth was monitored by optical density.

Samples were extracted using Waters HLB cartridges and dissolved in 50 μl of acetonitrile/water (1:1) for LC-MS analysis. Samples were electro sprayed into a Thermo Vantage triple quadruple mass spectrometer operated in the negative ion mode and coupled to a Waters Acquity UPLC system with a Waters Symmetry Shield C18 3.5 μm column (2.1 x 100 mm) using a gradient of 5% to 95% acetonitrile in water 0.1% formic acid within 3 min and a flow rate of 0.4 μl/min (Luis et al., 2018) (all the chemicals procured from sigma St. Louis).

AGS cells were grown in (DMEM) Advanced Dulbecco’s Modified Eagle’s Medium/F12 medium (sigma) supplemented with 10% FBS and 2 mM glutamine and antibiotics. H. pylori were grown for 6 h with or without exposure to curcumin or Meriva, at which time the bacteria were collected and washed with PBS. AGS cells were infected with treated H. pylori at a multiplicity of infection (MOI) which was determined of 10. MOI was measured as the number of pathogen that are added per cell during infection. If one million pathogens are added to one million cells, the MOI is one (Bussiere et al., 2006).

mRNA Expression of CXCL8 the gene encoding for IL-8, was measured in AGS epithelial cells exposed to H. pylori strains. Total RNA was isolated 6 h after exposure, using TriZOL (Invitrogen), and cDNA was synthesized from 1 µg RNA, as described in kits protocol (Invitrogen). Real-time PCR was performed, and relative gene expression was calculated from second derivative maximum values using β-actin as a reference gene. CXCL8 primers were 5′-TAGCAAAATTGAGGCCAAGG-3′ and 5′-AAACCAAGGCACAGTGGAAC-3′.

H. pylori 26695 was grown for 6 h with or without Meriva. Total RNA was extracted from 10⁸ bacteria using TriZOL (Invitrogen), and cDNA was synthesized from 1 µg RNA (Rokbi et al., 2001). Real-time PCR was performed, and relative gene expression was calculated using 16S rRNA as a reference gene (Rokbi et al., 2001). cagA E, M and 16S primers (Barry et al., 2011) were cagA 5′-ACCAACAAGGTAACAATGTGGC-3′ and 5′-TCGTTGTGAGCCTGTGAGTTGGT-3′. cagE 5′-CAATGGGTGGGGAGTATGTC-3′ and 5′-TGCTCCATTGTTGCATTTGT-3′. cagM 5′-GGTTGCGTTTGGAGTTTTGTCGGC-3′ and 5′-AGCGTCTTCTTTTGCGGCCACT-3′. 16S 5′-CAGCTCGTGTCGTGAGATGT-3′ and 5′-CGTAAGGGCCATGATGACTT-3′.

The cells were removed by trypsinization and lysates were prepared in buffer composed of 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% (v/v) NP-40, and 0.1% (w/v) sodium dodecyl sulfate. 20 µg of protein per lane was resolved by electrophoresis on 10% Tris-HCl polyacrylamide gels (Bio-Rad) and transferred overnight onto PVDF and blocked in BSA. Translocation of CagA in gastric epithelial cells was detected by level of phosphorylated (phospho)-CagA. Upon injection, CagA undergoes tyrosine phosphorylation by host kinases, and therefore, in our experiments the translocated CagA was detected by anti-phospho-CagA. Phosphorylated (phospho)-CagA, and β-actin levels as loading controls were detected with 1∶300 diluted anti-phosphotyrosine antibodies (Santa Cruz Biotechnology,PY20), 1∶2000 polyclonal anti-H. pylori CagA antibody (Austral Biologicals HPA-5000-4), and 1∶20000 anti-β-actin antibody (Sigma A2228), respectively. In some experiments, H. pylori co-cultured AGS cell lysates were also probed for levels of phosphorylated-cSrc and cSrc proteins with mouse anti-phospho-cSrc (Y418, dilution 1:300; cell signaling), and mouse anti-cSrc (dilution 1:500; cell signaling), respectively.

Mouse stomach glands were isolated as described (Wroblewski et al., 2014) by ligating at the esophago-gastric and gastro-duodenal junctions. Released glands were plated in Matrigel (BD Biosciences) containing N-acetylcysteine (1 µM), gastrin (10 nM), epidermal growth factor 10 (50 ng/mL), R-spondin 1 (500 ng/mL), Noggin (100 ng/mL), fibroblast growth factor 10 (100 ng/mL), rWnt3A (100 ng/mL), and Y27632 (10 µM) as described (Shibata et al., 2017) (DMEM)/F12 medium supplemented with B27, N2, penicillin/streptomycin, N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid (10 mM) and Glutamax (2 mM) was overlaid on the Matrigel.

The H. pylori were grown in Brucella broth with 10% fetal bovine serum for 6 h with and without Meriva, harvested and microinjected into the lumen of gastroids at a multiplicity of infection of 100:1 (Bussiere et al., 2006). Infected gastroids were cultured for 4 hr in Advanced DMEM/F12 medium without penicillin/streptomycin. Some of the wells were fixed in 4% formaldehyde and embedded in paraffin blocks. Five μM thin sections were cut, and gastroids were stained with H&E and also for PAS-AB as described(Shibata et al., 2017) Some of the gastroids were also processed for immunofluorescence.

Gastroids were cultured on MatTek dishes (MATTEK Corporation) and stained for phospho-cSrc. In brief, gastroids were washed twice in 1X PBS, and formalin-fixed. Cells were permeabilized using 1X PBS containing 0.1% Triton X-100 (30 minutes, room temperature), and washed three times in 1X PBS. Gastroids were stained with mouse anti-phospho-cSrc (Y 418: Cell Signaling) followed by Alexa Fluor 488-conjugated anti-mouse IgG antibody (1:200, Molecular Probes). Nuclei were stained with DAPI (4 μg/mL). Images were captured using an Olympus FV-1000 Inverted Confocal Microscope. Image acquisition was performed using Fluoview FV10-ASW 1.7 software.

Male C57BL/6 mice aged 6–8 weeks were infected as described (Arnold et al., 2011) H. pylori was grown for 2 days on serum plates, harvested, and suspended in brucella broth, and the final concentration was adjusted to 10⁷bacteria/200 μl. Mice were inoculated orogastrically with 0.2 ml of bacterial suspension (10⁷bacteria) (Arnold et al., 2011). Briefly, H. pylori PMSS1, a clinical isolate, was grown for 6 h with or without Meriva, and 2×10⁷ bacteria were delivered intra-gastrically on days 0, 2 and 4. After one month, mice were sacrificed (Conlee et al., 2005), and colonization was determined by serial dilution and plating of gastric homogenates. After one week, colonies were counted to determine bacterial load, and single colonies were stored as output strains (D’costa et al., 2018). AGS cells were also co-cultured with output strains from the mouse infection experiment, and cell lysates were probed for levels of phospho-cSrc and cSrc. This study was carried out following recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The Institutional Animal Care and Use Committee of Vanderbilt University and the Institutional Animal Care and Use Committee of Jawaharlal Nehru University New Delhi approved the protocol (protocol No M/12/046).

Data are expressed as means ± standard error. Student’s t test was used for pairwise comparisons. Data from more than two groups were analyzed by ANOVA, followed by the Newman-Keuls post hoc multiple comparisons tests.

Curcumin, a non-antibiotic compound is effective against H. pylori without any toxicity and resistance (Sarkar et al., 2016). Since the bioavailability of pure curcumin is low its ability to inhibit growth of H. pylori was compared to a phospholipid complexed formulation (Meriva) with improved bioavailability and pharmacokinetics (Cuomo et al., 2011; Franceschi et al., 2016). To compare the effects on H. pylori growth four prototypic strains were selected, 26695, PZ5056G, PMSS1, and 7.13. All H. pylori strains were partially inhibited at 40 μM curcumin resulting in a modest increase in generation time, whereas 20 μM did not show an effect ( Figures 1A, C ). Meriva caused a significant decrease of the growth of all four H. pylori strains at 20 μM ( Figure 1B ). This concentration of Meriva also increased the generation time of two representative strains of H. pylori at 17.4 h for PMSS1 and at 6.5 h for 7.13 ( Figure 1C and Supplementary Figure 1 ). We observed Meriva inhibit the growth of H. pylori at low concentration, which indicates it is more toxic than curcumin for H. pylori ( Kidd, 2009 ). Quantification of curcumin inside the bacteria revealed that the levels were significantly higher in bacteria treated with Meriva (18.6 ± 2.4 ng/10⁸ bacteria) compared to control curcumin (0.5 ± 0.1 ng/10⁸ bacteria) when a concentration equivalent to 20 μM curcumin was applied ( Supplementary Figure 2 ) (Blanchard and Nedrud, 2012). This suggested an increase in bioavailability of curcumin upon encapsulation with phospholipids, and this could explain the more growth inhibitory effect compared to pure curcumin.

Metabolism affects the biological activity of curcumin with reduction of the double bonds as well as conjugation decreasing activity (Anand et al., 2007). Reduction of curcumin in E. coli, is catalyzed by the enzyme NADPH-dependent curcumin reductase (CurA) yielding dihydro-curcumin and tetrahydrocurcumin (Hassaninasab et al., 2011). Phylogenetic analysis showed the widespread presence of CurA in enteric bacteria ( Supplementary Figure 3 ). To understand the effect of reductive metabolism we studied the effect of Meriva on the growth of H. pylori, E. coli, and the enteropathogenic rodent equivalent, Citrobacter rodentium. The growth of E. coli and Citrobacter rodentium was not inhibited compared to H. pylori at 20 μM Meriva ( Figures 2A–C ). We detected high levels of intracellular curcumin and the reduced metabolites, dihydro-curcumin and tetrahydro-curcumin, in E. coli and Citrobacter rodentium ( Figures 2D, E ), while only curcumin was detected in H. pylori ( Figure 2F ). This indicated that H. pylori was not able to metabolize curcumin by reduction, and suggested that, conversely, reductive metabolism of curcumin enabled E. coli and Citrobacter rodentium to evade its antibacterial effect.

In order to analyze the role of CurA in the reduction of curcumin and evasion of growth inhibition, we generated an isogenic deletion mutant of curA in E. coli strain DH5 -α (E. coli ΔcurA).The deletion of curA was done by PCR splicing method in which a pair of primers flanking the region where the deletion to be made, two complementary primers comprising a region of -15 bp to +15 bp related to the junction point were used and a high fidelity polymerase was used for PCR. The transformed E. coli were used for the assay. Curcumin inhibited the growth of the E. coli ΔcurA deletion mutant to a similar extent as observed for H. pylori ( Figure 3A ). The reduced metabolites dihydro- and tetrahydro-curcumin were absent in the curA deletion mutant despite high levels of curcumin in the cells ( Figure 3B ) while wild type E. coli mainly contained the reduced metabolites tetra- and hexahydro-curcumin and negligible amounts of curcumin ( Figure 3B ). PCR analysis (Hassaninasab et al., 2011) of genomic DNA confirmed the presence of the CurA reductase in E. coli DH5-α and showed absence of the gene in H. pylori ( Figure 3C ). The NCBI BLAST (Default parameter) showed no hits against H. pylori. We also checked the curA in 18 clinical isolates with E. coli as positive control ( Supplementary Figure 6 ). CurA gene was not detetected in any of the isolates. These data strongly suggested a role of CurA and reductive metabolism in evading growth inhibition by curcumin in enteric bacteria.

Reductive metabolism of curcumin has been shown to limit its biological effects (Ireson et al., 2002), whereas oxidative transformation appears to be required to affect protein targets with redox-active cysteine residues (Ketron and Osheroff, 2014; Schneider et al., 2015). Since reductive metabolism decreased growth inhibition, we next determined the role of oxidative metabolism of curcumin. To accomplish that, we compared curcumin to its derivatives, diacetalcurcumin, acetalcurcumin, and 4’,4”-dimethylcurcumin. The latter three do not or not readily undergo oxidative transformation (Edwards et al., 2017). Diacetalcurcumin and acetalcurcumin, similar to reduced curcumin, did not inhibit the growth of H. pylori at 40 μM concentration, whereas the same concentration of curcumin inhibited growth significantly ( Figure 4A ). Growth inhibition by curcumin was associated with the formation of bicyclopentadione (BCP), the most abundant stable end product of oxidative transformation, (Luis et al., 2017) that was present in about 10% abundance relative to curcumin inside the bacteria ( Figure 4B ). The curcumin/BCP ratio was significantly higher in E. coli and Citrobacter rodentium than in H. pylori ( Figure 4C ), indicating a greater contribution of the oxidative pathway to the metabolism in H. pylori. This was consistent with the hypothesis that growth inhibition is dependent on oxidative transformation of curcumin.

H. pylori translocates the bacterial protein CagA into infected gastric epithelial cells using a type IV secretion system, encoded by the cag pathogenicity island (Odenbreit et al., 2000). Translocation of CagA initiates numerous signaling events that contribute to H. pylori pathogenesis. CagA translocation and its subsequent phosphorylation occur in a time dependent manner (Backert et al., 2000). When H. pylori was grown in 20 μM Meriva for 6 h and then transferred to only broth, growth recovered to the same rate as for untreated bacteria within the next 24 h. When H. pylori were grown in the presence of Meriva for 12 h, bacteria did not recover when transferred to only broth ( Supplementary Figure 4 ), indicating that exposure of H. pylori to curcumin for 6 h resulted in reversible growth inhibition (Barry et al., 2011). We analyzed early events occurring during reversible growth inhibition. Treatment of H. pylori strains PZ5056G and PMSS1 with curcumin for 6 h resulted in reduced translocation of CagA into infected gastric epithelial cells and subsequent phosphorylation ( Figure 5A ).

We prepared gastroids from mouse stomach as described previously in order to test a physiologically more relevant model (Wroblewski et al., 2014; Pompaiah and Bartfeld, 2017). Gastroids were stained with H&E to confirm the morphology and PAS-AB to confirm mucin production as a marker of a functionally viable model ( Supplementary Figure 5 ). To analyze the effect of Meriva on the ability of H. pylori to activate biological signaling in gastroid epithelial cells, we microinjected H. pylori grown for 6 h in control medium or in medium supplemented with 20 μM Meriva. Microinjection of H. pylori grown in control broth caused phosphorylation of c-Src as detected by immunofluorescence ( Figure 5B ). Phosphorylation of c-Src was reduced in gastroid epithelial cells microinjected with H. pylori grown in the presence of Meriva. We confirmed the immunofluorescence data by Western blotting showing reduced phosphorylation of c-Src in gastroid lysates from H. pylori treated with Meriva ( Figure 5C ). H. pylori virulent strains encode the cag (cytotoxin-associated genes) pathogenicity island (cagPAI). The cagPAI expresses a type IV secretion system (T4SS). This T4SS injects virulence factors such as the CagA effector protein into host target cells. This is achieved by a number of T4SS proteins (Barrozo et al., 2013) and to assess the effect of curcumin on Cag PAI-associated genes (Ikenoue et al., 2001) we performed quantitative analysis for the levels of mRNA of cag A, F, T, Y, L, N, Z, M, D, and E. Curcumin dose-dependently decreased levels of cagF and cagE mRNA, which is a virulence factor for translocation of CagA ( Figure 5D ). It has been reported that both the genes are not co-transcribed (Ta et al., 2012). There might be possible both the genes are involved in virulence and upon treatment with curcumin both the genes get downregulated, the further investigation need to be done.

We compared the effects of curcumin, acetalcurcumin, and diacetalcurcumin on the ability of H. pylori to translocate CagA inside host epithelial cells. Translocation was determined via the levels of phosphorylated CagA inside infected gastric epithelial cells. Treatment of H. pylori with curcumin resulted in near complete inhibition of CagA phosphorylation. This effect was attenuated when H. pylori was grown in acetalcurcumin or diacetalcurcumin such that there was no difference in CagA phosphorylation compared to vehicle ( Figures 6A, B ). Treatment of H. pylori with curcumin also decreased the levels of phosphorylated c-Src in gastric epithelial cells, and this decrease was not observed by treatment with acetalcurcumin or diacetalcurcumin ( Figures 6A, C ). Levels of phosphorylated CagA have been shown to correlate with the levels of interleukin-8 (IL-8) since phosphorylation induces cellular responses, including IL-8 secretion. Consistent with the results on CagA and c-Src phosphorylation, H. pylori grown in the presence of curcumin reduced the levels of CXCL8 mRNA in gastric epithelial cells whereas the levels induced by H. pylori grown in the presence of acetal- or diacetalcurcumin were comparable to vehicle control ( Figure 6D ). These findings are consistent with a role of oxidative transformation of curcumin on the growth of H. pylori and its ability to regulate host cell responses.

We treated H. pylori-infected mice with chow containing 1% Meriva for 4 weeks. The output H. pylori strains isolated from Meriva-treated mice attenuated phosphorylation of c-Src ( Figures 7A, B ) and reduced levels of CXCL8 mRNA expression when co-cultured with AGS cells ( Figure 7C ). There was no difference in colonization or inflammation in the stomach of animals treated with vehicle or 1% Meriva ( Figure 7D ).