EHEC O157:H7 Sakai, Citrobacter. rodentium DBS100, Enterococcus. hirae ATCC9790RF, Lactobacillus. acidophilus ATCC4356, and Lactobacillus. casei ATCC393 strains were used. EHEC and C. rodentium were cultured in Luria-Bertani (LB) medium unless otherwise indicated. E. hirae was cultured in Brain Heart Infusion (BHI) medium. L. acidophilus and L. casei were cultured in De Man, Rogosa, and Sharpe (MRS) medium. Bacteria were cultured in glass tubes at 37°C. Caco-2 (ATCC HTB-37) cells and Vero cells, and human kidney cells (HTB-44) were cultured in Dulbecco’s modified Eagle medium (DMEM) and Eagle’s minimal essential medium (EMEM), respectively containing 10% HyClone FetalClone III serum (HyClone Laboratories, Inc., Logan, UT, United States) at 37°C and in an atmosphere of 5% CO₂. The MgOC material with a pore size of 150 nm, named MgOC₁₅₀, was obtained from Toyo Tanso (Osaka, Japan). The activated charcoal material (made from coconut shell) was obtained from UES Co., Ltd. (Wakayama, Japan). The purified Shiga toxin protein and lysozyme were obtained from Nacalai Tesque (Kyoto, Japan).

To estimate the amounts of Shiga toxins (Stx1 and Stx2), latex agglutination reagents (Denka Seiken Co. Ltd., Tokyo, Japan) were used. EHEC strains were cultured with shaking to the early stationary phase in Mueller–Hinton medium, and their cells and culture supernatants were separated by centrifugation at 13,000 g. Cell pellets were resuspended in EzBactYeastCrusher containing 60 mg/l lysozyme (ATTO, Tokyo, Japan) to extract intracellular proteins, and diluted into phosphate-buffered saline (PBS). These cell lysates (0.63 ng) and culture supernatants were serially diluted in 96-well round bottomed plates containing PBS, and an equal volume of latex suspension sensitized with the Stx1 or Stx2 antibody was then added. After incubation for 14 h at 4°C, titers were determined. The titers are presented as the reciprocal of the dilution of the last well before agglutinations were observed. To test the adsorption of Stx1 and Stx2 to MgOC₁₅₀ and activated charcoal, the cell-free culture supernatant from EHEC was incubated with and without these materials for 2 h at 4°C. The cell-free culture supernatant was prepared by centrifugation at 13,000 g and passing through a membrane filter (pore size: 0.22 μm). After removal of porous materials by centrifugation at 15,000 g, Stx1 and Stx2 in the supernatant were assayed using latex agglutination reagents. To test the adsorption of purified Shiga toxin to MgOC₁₅₀, the toxin (0.4 μg) was incubated with and without MgOC₁₅₀ for 2 h at 4°C. Non-adsorbed toxin in the supernatant was quantified in a Bio-Rad protein assay according to the Bradford method (Bio-Rad Laboratories, Hercules, CA, United States). The cytotoxicity of Shiga toxins in Vero cells and HTB-44 was measured. Bacteria-free culture supernatants were 10-fold diluted into DMEM and EMEM, respectively containing, 10% HyClone FetalClone III serum, and added to cultured Vero and HTB-44 cells in 96-well plates. As a control, a 10-fold diluted Mueller–Hinton medium in DMEM and EMEM containing 10% HyClone FetalClone III serum was added to host cells. After incubation for 48 h, the cell viabilities were determined as previously described with CellTiter-Glo Luminescent Cell Viability Assay (Promega Corp., Madison, WI, United States; Hirakawa et al., 2021). Cell viabilities were represented as relative light units (RLUs) by their ratios (%) to the RLU of the control sample.

EHEC strains were cultured with shaking to an early stationary phase in Dulbecco’s modified Eagle medium (DMEM). Secreted proteins were precipitated from the supernatants using 10% trichloroacetic acid (TCA) and were dissolved in a Laemmli sample buffer (Bio-Rad Laboratories, Hercules, CA, United States). Bovine serum albumin (BSA) was used as a loading control and was added to the secreted protein samples prior to the precipitation with TCA. Intracellular proteins were resuspended in 50 mM phosphate buffer containing 8 M urea and then lysed by sonication. EspA and EspB were detected with their antisera, as previously described (Hirakawa et al., 2020a,b). To test the adsorption of EspA and EspB to MgOC₁₅₀, cell-free culture supernatant prepared by centrifugation of EHEC culture at 13,000 g and passing through a membrane filter was incubated with and without MgOC₁₅₀ for 2 h at 4°C. After removal of the MgOC₁₅₀ material by centrifugation at 15,000 g, EspA and EspB in the supernatant was measured by western blotting.

Three-week-old female C3H/HeJ mice were obtained from CLEA Japan (Tokyo, Japan).

MgOC₁₅₀ was orally administered with a feed during the experiment as performed in previous studies used AST-120, an approved porous carbon medicine (Yang et al., 2017; Nakada et al., 2019). In those studies, 5 to 8% (w/w) of AST-120 was contained in a feed. For this reason, we supplied 7.5% (w/w) of MgOC₁₅₀. The mice were housed for 7 days before infection (N = 5 control mice for non-infection and mice bled without MgOC₁₅₀ for infection, N = 6 mice bled with MgOC₁₅₀ for infection). C. rodentium DBS100 was cultured overnight in LB medium. Bacterial cells were resuspended in fresh LB medium at a concentration of 1 × 10⁹ CFU/ml, and 200 μl bacterial suspension (2 × 10⁸ CFU) was orally administered. As a control group, 200 μl bacteria-free LB medium was inoculated into mice. To measure the survival rates and body weight, mice were monitored daily for 21 days.

To test the toxicity of MgOC₁₅₀ in host cells, we used Caco-2 cells. MgOC₁₅₀ was added to cultured Caco-2 cells. After incubation for 24 h, the cell viabilities were determined with CellTiter-Glo Luminescent Cell Viability Assay. The cell viabilities were represented as relative light units (RLUs) by their ratios (%) to RLU of the sample incubated without MgOC₁₅₀.

To estimate the capability of MgOC₁₅₀ to adsorb antimicrobial agents, 1.25 mg aztreonam, ciprofloxacin, minocycline, trimethoprim, and sulfamethoxazole, 0.68 mg rifampicin, or 5 mg fosfomycin and amikacin in 5 ml aqueous solution were incubated with and without 30 mg MgOC₁₅₀ for 2 h. Drug amounts were calculated as described previously (Hirakawa et al., 2020a).

We used the Gehan–Breslow–Wilcoxon tests for mouse survival experiments and the unpaired t-tests for mouse body weight measurement, cytotoxicity assays, and adsorption experiments, and then determined p-values using GraphPad Prism version 6.00.

We aimed to find materials that adsorb proteins responsible for EHEC pathogenicity and attenuate the virulence of this bacterium. For this purpose, we used one MgOC material (MgOC₁₅₀) with an average pore size of 150 nm because this size is predicted to highly adsorb protein molecules that are more than 50,000 Da (Funabashi et al., 2017). First, we tested the capability of MgOC₁₅₀ to adsorb the Shiga toxins produced by EHEC. The bacteria-free supernatant from an EHEC culture was incubated with MgOC₁₅₀, and Stx1 and Stx2 titers after MgOC₁₅₀ removal were measured in latex agglutination assays. The agglutination titers of Stx1 and Stx2 from the EHEC supernatant were 64 and 128, respectively. However, no agglutination of Stx1 and Stx2 in the supernatant when incubated with 30 mg MgOC₁₅₀ was observed (Table 1). To compare the capability of MgOC₁₅₀ with ordinal activated charcoal, the bacteria-free supernatant was incubated with 30 mg activated charcoal made from coconut shells. The agglutination titers of Stx1 and Stx2 were then measured. These titers were the same as those in supernatants incubated without the activated charcoal (Table 1). These observations indicated that MgOC₁₅₀, not activated charcoal, highly adsorbs Shiga toxins. To estimate the adsorption affinity of MgOC₁₅₀ to Shiga toxins, a purified Shiga toxin standard was incubated with different amounts of MgOC₁₅₀. More than 60% of the toxin (0.4 μg) in a solution was adsorbed when incubated with at least 0.04 mg of MgOC₁₅₀, and more than 95% of the toxin was adsorbed when incubated with 0.2 mg of MgOC₁₅₀ (Figure 1A). We also tested the ability of MgOC₁₅₀ to adsorb lysozyme, a small-sized protein. Its predicted molecular size is approximately 14,300 Da thus, the ability of MgOC₁₅₀ to adsorb this protein molecule may be low. As predicted, more than 90% of the lysozyme protein (50 μg) remained in a solution even after being incubated with 25 mg of MgOC₁₅₀ (Figure 1B). We next measured the levels of Shiga toxins in bacterial cultures. EHEC was cultured with MgOC₁₅₀, and agglutination titers were then determined. No agglutinations of Stx1 and Stx2 in the EHEC supernatant cultured with 30 mg MgOC₁₅₀ were observed, whereas there was no apparent difference in these titers between the cell lysates from strains cultured with and without MgOC₁₅₀ (Table 2). Bacterial colony-forming units (CFUs) were essentially the same when cultured with and without MgOC₁₅₀ (data not shown). Therefore, MgOC₁₅₀ could reduce extracellular Stx1 and Stx2 levels in EHEC without suppressing bacterial growth. These results indicated that MgOC₁₅₀ adsorbs the Stx1 and Stx2 secreted by EHEC and decreases the extracellular Stx1 and Stx2 levels but does not impair Shiga toxin production and EHEC growth.

Shiga toxins cause damage to human renal cells (Kiyokawa et al., 1998). We tested whether the removal of Shiga toxins by MgOC₁₅₀ treatment reduces toxicity to human HTB-44 renal cells. The addition of a bacteria-free supernatant from an EHEC culture killed approximately 80% of HTB-44 cells, whereas 70% of the host cells survived when incubated with EHEC supernatants cultured with 30 mg MgOC₁₅₀ (Figure 2). The cytotoxicity of Shiga toxins is also commonly evaluated in Vero cells as alternative cells (Konowalchuk et al., 1977). The toxicity of EHEC supernatants cultured with and without MgOC₁₅₀ to Vero cells was also examined. The addition of the supernatant killed approximately 85% of Vero cells. Similar to HTB-44 cells, the cytotoxicity of the supernatant cultured with 30 mg MgOC₁₅₀ was modest (58% of the cells survived; Figure 2). These results suggested that MgOC₁₅₀ decreases Shiga toxin-associated cytotoxicity.

Type III secretory proteins are the other subset of proteins required for EHEC virulence. EspB is a member of these proteins, and it is required to translocate other effector proteins into host epithelial cells. This protein also has an effector activity (Taylor et al., 1999; Kodama et al., 2002). The EspB level in EHEC supernatants and cell extracts cultured with and without MgOC₁₅₀ was measured. The addition of at least 2 mg MgOC₁₅₀ reduced the EspB level in the supernatant (Figure 3A). In contrast, the EspB level in the cell extract was not reduced even when 30 mg MgOC₁₅₀ was present (Figure 3B). The capability of MgOC₁₅₀ to adsorb the EspB protein was also assessed. The EHEC supernatant was incubated with MgOC₁₅₀, and the EspB level in the supernatant was measured after MgOC₁₅₀ removal. No residual EspB protein was observed in the supernatant after incubation with at least 2 mg MgOC₁₅₀ (Figure 3C). In addition to EspB, the adsorption and extracellular accumulation of EspA another protein secreted via the Type III secretion system were examined. MgOC₁₅₀ also adsorbed the EspA protein and reduced its extracellular level (Figures 4A,B).

To assess the in vivo effectiveness of MgOC₁₅₀ that attenuates bacterial virulence, a murine intestinal infection model with C. rodentium was used. C. rodentium is a natural pathogen of mice, and it causes typical diarrheal symptoms (Mundy et al., 2005). This pathogen produces one subset of orthologous Type III secretory proteins including EspA and EspB, but does not have genes that encode Shiga toxins. For this reason, C. rodentium is used to evaluate Type III secretion system-associated virulence in mice. To treat mice with MgOC_150, it was administered with their feed. The mice did not adequately consume the MgOC₁₅₀-containing diet on the first few days, but they consumed the diet normally after that. Thus, the mice were pretreated with MgOC₁₅₀ for 7 days before infection and the treatment was maintained until the end of experiments. The C. rodentium DBS100 strain that is highly virulent in C3H/HeJ mice was used. A decrease in body weight was observed after 4 days of infection in non-treated mice, and all mice died within 9 days post-infection (Figures 5A,B). When MgOC₁₅₀ was administered with the feed, the mice infected with C. rodentium exhibited no body weight decrease until 11 days post-infection, and mice survived significantly longer (Figures 5A,B). The median duration of survival could be prolonged up to 12.5 days post-infection. These results suggested that MgOC₁₅₀ administration reduces the virulence of C. rodentium in mice. No significant difference in weight gain between non-infected control mice fed with a regular diet and a MgOC₁₅₀-containing diet was observed (Figure 5C).

We used Caco-2 cells to test the cytotoxicity of MgOC₁₅₀ in human intestinal epithelial cells. The host cells were incubated with MgOC₁₅₀ for 24 h. MgOC₁₅₀ did not exhibit cytotoxicity because no significant reduction of viable cell numbers was observed after incubation with MgOC₁₅₀ (Figure 6A). When MgOC₁₅₀ is orally administered, it may disturb normal bacterial flora in the intestinal tract by adsorbing some beneficial compounds and/or bacteria. To test this hypothesis, the growth of Enterococcus hirae, Lactobacillus acidophilus, and Lactobacillus casei, which are commonly isolated from the normal intestine, was examined after being cultured with and without MgOC₁₅₀ for 8 and 24 h_. When the bacterial strains were cultured without MgOC₁₅₀, the CFUs of E. hirae, L. acidophilus, and L. casei reached 8.2 × 10⁸, 2.4 × 10^8, and 4.6 × 10⁸ after 8 h, and 1.0 × 10⁹, 1.8 × 10⁸ and 9.6 × 10⁸ after 24 h, respectively (Figures 6B–D). The CFUs of these bacteria reached similar values even when cultured with 30 mg MgOC₁₅₀ (8.0 × 10⁸, 1.5 × 10⁸, and 3.5 × 10⁸ after 8 h, and 1.0 × 10⁹, 1.8 × 10⁸, and 9.0 × 10⁸ after 24 h, respectively). These results indicated that MgOC₁₅₀ does not disturb E. hirae, L. acidophilus, and L. casei.

MgOC₁₅₀ may adsorb some antimicrobial agents. This property may impair a combination therapy that uses MgOC₁₅₀ and an antimicrobial agent. We tested whether MgOC₁₅₀ adsorbs antimicrobial agents that are generally active in EHEC. Each antimicrobial agent was incubated with and without 30 mg MgOC₁₅₀ for 2 h and the amount of each agent in the supernatant was measured after MgOC₁₅₀ removal. Undesirably, aztreonam, ciprofloxacin, rifampicin, minocycline, trimethoprim, and sulfamethoxazole were highly adsorbed by MgOC₁₅₀, as these amounts after incubation with MgOC₁₅₀ were lower than 20% compared to the MgOC₁₅₀-free control (Figure 7A). In contrast, fosfomycin and amikacin retained more than 80% even after incubation with MgOC₁₅₀ (Figure 7A). The in vitro activities of fosfomycin and amikacin in EHEC were examined when cultured with MgOC₁₅₀. Bacterial cell deaths were observed as a significant reduction in bacterial CFUs was found when cultured with 32 μg/ml fosfomycin and 64 μg/ml amikacin. Similar reductions in CFUs were also observed even when MgOC₁₅₀ was present (Figure 7B). Therefore, MgOC150 does not impair the antimicrobial activities of fosfomycin and aminoglycosides, such as amikacin.