Purified recombinant Stx1 (cat. # NR-857) and Stx2a (cat. # NR-4478) were obtained from BEI resources. Stx was diluted in tissue culture grade PBS (pH 7.4) for all inoculations. Lipopolysaccharide (LPS) content was determined by the limulus amoebocyte lysate (LAL) assay (Lonza).

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Cincinnati, and conducted in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals. Outbred male CD-1 mice, 13–15 g obtained from Charles River Laboratories (Wilmington, MA) were housed in filter-top cages with access to food and water ad libitum. Mice were challenged by intraperitoneal (IP) injection. Mice were observed twice daily for signs of illness, and weighed once a day.

Mice were given Stx2a at 7 ng/ml and Stx1 at 1500 ng/ml. PBS alone served as the negative control. At the indicated times post-challenge, animals were anesthetize by IP injection of 100 μl, 2.5% avertin (2,2,2-Tribromoethanol) and perfused intracardially with 100 ml sterile PBS containing 20 units/ml heparin sodium salt to prevent clotting. Organs were fixed by perfusion with 200 ml of 4% paraformaldehyde prepared by dissolving 20 g of paraformaldehyde powder in 500 ml of heated PBS solution, cooled, adjusted to pH 6.9 and filter sterilized. After perfusion, organs were removed and stored in 4% paraformaldehyde.

For brain histology studies, the tissues were washed in PBS with 3 changes over a period of 15 min and embedded in 5.0% low gelling temperature agarose (Sigma Type XI, A-3038) in 20 mm peel-a-way disposable plastic tissue embedding molds (Thermo Scientific, cat no: 2219). Forty micron thick brain samples were sectioned on the vibratome and stored in 4% paraformaldehyde until stained. For kidney histology, thin sections of 5 micron thickness were cut using Leica CM 1900 cryostat and mounted on slide until stained. For hematoxylin and eosin staining, brain and kidneys were processed by fixing in 4% paraformaldehyde solution, and then dehydrated in graded alcohol, cleared by xylol and embedded in paraffin.

Rabbit polyclonal antibody against Iba1 (ionized calcium binding adaptor molecule-1) was used as the marker for microglial detection. The protocol was followed as laid down in the Vectastain Elite ABC system kit from Vector Laboratories. Briefly, brain sections were incubated with blocking buffer (1.5% normal goat serum (Vector Laboratories, cat no: S-1000) and 1% Bovine Serum Albumin in 1X PBS) for 2 h at room temperature (RT) followed by incubation with anti-Iba1 antibody, 0.5 μg/ml (Wako Catalog No. 019-19741) in blocking buffer (1.5% normal goat serum (Vector Laboratories, cat no: S-1000) and 1% Bovine Serum Albumin in 1X PBS) overnight at 4°C. The sections were washed with 1X PBS, 3 times and incubated with Biotinylated anti-Rabbit IgG Antibody (1:200) in blocking buffer for 1 h at RT and further washed with 1X PBS, 3 times prior to incubation with Vectastain Elite ABC (Reagent A (1:50) and Reagent B (1:50) in 0.01M PBS) for 1 h at RT. The sections were washed three times and incubated in peroxidase solution (0.01% hydrogen peroxide and 0.05% DAB (3, 3′-diaminobenzidine) HRP substrate in 0.05 M Tris buffer) for color development.

Forty eight to seventy two hours after IP injection with Stx2a (3 ng) or PBS animals were transferred to Cincinnati Children's Hospital and Medical Center, Department of Radiology, Imaging Research Center. All of the Stx2a-treated mice were symptomatic at the time of imaging, as evidenced by lack of weight gain, weight loss and/or lethargy, and the mice that were most severely affected were imaged first. For the toxin-treated animals, two were imaged on day 2, and two were imaged on day 3. For the control animals, two were imaged on day 2 and one was imaged on day 3. Animals received standard MRI brain scans and MRS data was collected from the thalamus and cortex. Unprocessed MRS signals were analyzed using the LCModel software package (version 6.2-0), data where the % standard deviation exceeded 15% were rejected.

Images of immunohistochemically developed slides were captured using an Aperio ScanScope slide scanner (Aperio, Vista, CA) to create whole-slide digital images. Images were subsequently analyzed using ImageScope Positive-Pixel Count Algorithm (Aperio) for systematic identification and quantitation of deposited DAB generating a pseudo-colored markup image representing staining intensities. Staining intensities were assigned using as parameters the following selections for input upper limits: weak positive (70, yellow on markup image), medium positive (100, orange) and strong positive (180, brown). This set of algorithm settings was used on all positive and negative control tissues with manual comparisons of multiple fields to ensure validity of the selected settings. Analysis of variance was used to determine the significant difference in the survival curve among intoxication groups as well as for the brain microglial study followed by Bonferroni post-tests (GraphPad Prism 5, GraphPad Software, Inc.). P-values < 0.05 were considered significant.

Human cerebral cortex microvascular endothelial cells (HBMEC, ACBRI 376, Lot 376.04.0H.0U.0Y) were obtained from Cell Systems Corporation (Kirkland, WA) and were propagated in Complete Classic Medium with serum, CultureBoost and Bac-Off® antibiotic (Cell Systems, Kirkland, WA). Cells were harvested with the Passage Reagent System (Cell Systems) according to manufactures protocol. The BALB/c bEnd.3 cell line was a gift from Dr. Jerry Lingrel and was maintained in DMEM with 10% FBS and Pen/Strep. Primary neonatal dermal microvascular endothelial cells (dHMEC, CC-2516, Lot 0000317328) were obtained from Clonetics (Lonza, Walkersville, MD) and propagated in Endothelial Cell Medium without phenol red (ScienCell, Carlsbad, CA). Cells were harvested with TrypKit (Lifeline Cell Technologies, Walkersville, MD) according to manufacturer's protocol. The CDC.HMEC-1 line was maintained in MCDB 151 medium supplemented with 10 mM L-glutamine, 10 ng/ml epidermal growth factor, 1 μg/ml hydrocortisone, 10% fetal bovine serum (FBS), 50 μg/ml gentamycin sulfate and 100 μg/ml kanamycin sulfate.

Stx was serially diluted in 20 μl of medium in sterile, black, clear, flat-bottom 96-well plates (Corning, Tewksbury, MA). Cells (in 60 μl of medium) were added to toxin-containing wells at subconfluent or confluent numbers, as follows: BMEC (subconfluent 1.5 × 10³ cells per well, confluent 8 × 10³ cells per well), bEnd.3 and dHMEC (subconfluent 6 × 10³ cells per well, confluent 12 × 10³ cells per well). Toxin-treated cells were incubated at 37°C in 5% CO₂. After 42 h the medium was removed and 50 μl of fresh medium containing 10% (vol/vol) alamarBlue (AbD Serotec, Raleigh, NC) was added. Cells were incubated in the presence of alamarBlue for a total of 3 h. Fluorescence was read at 590 nm on a FLX-800 fluorimeter (BioTek, Winooski, VT) every 30 min and values within the linear range, before depletion of substrate, are reported. In assays using TNF-α, cells were plated at the subconfluent or confluent densities in 60 μl of medium and incubated overnight. TNF-α (10 ng/ml) was added the next day and 24 h later toxin was added. Cells were incubated for an additional 42 h and analyzed as described above. Dose-response curves were plotted as the percent relative fluorescence units (RFU) of untreated cells verses toxin concentration.

5 × 10⁴ cells were cooled on ice for 30 min and all subsequent incubations were carried out on ice. Cells were incubated for 1 h with antibody to Gb3 (clone 38-13, Accurate Chemical and Scientific Corporation, Westbury, NY) or PBS for controls. Cells were washed, incubated for 1 h with fluorescein isothiocyanate (FITC) conjugated α-rat IgM secondary antibody diluted 1:500 (AbD Serotec, product number MCA189F, Bio-Rad, Raleigh, NC) and analyzed by flow cytometry (BD FACSCalibur; Becton Dickinson, Franklin Lakes, NJ). Due to the limited number of cells, analysis was performed by collecting 1 × 10⁴ events or all events collected for 5 min. Unstained cells (for labeled primary antibody) or cells incubated with secondary antibody (for unlabeled primary antibody) were used as negative controls. Cells were gated for the live cell population based on the forward and side scatter. Statistical analysis for flow cytometry was performed by paired Student's t-test using Prism5 (GraphPad Software, La Jolla, CA).

To assess the role of TNF-α on Gb3 expression, cells were plated in 12-well plates and allowed to adhere overnight. TNF-α (10 ng/ml) was added and 24 h later cells were harvested and treated as described above. As a positive control for TNF-α activity, stimulated HBMECs were stained for induction of ICAM-1 expression and analyzed by flow cytometry. Allophycocyanin (APC) labeled ICAM-1 (CD54) antibody was diluted 1:200 (clone HA58, product number 353111, BioLegend, San Diego, CA).

Stx toxicity was validated using standard assays. Groups of mice were injected IP with PBS, purified Stx1 (1500 ng) or purified Stx2a (7 ng), and weighed every 24 h post-injection (Figure 1). Mice injected with PBS gained weight over the 72 h period (Figure 1, closed circles). As observed previously, injection of 7 ng of Stx2a resulted in greater than 50% mortality (Fuller et al., 2011) and statistically significant weight loss was observed (Figure 1, closed triangles). In this study injection of 1500 ng of Stx1 did not result in statically significant weight loss (Figure 1, closed squares). These results demonstrate that Stx1 is not as toxic as Stx2a. Stx2a, the more important toxin in human disease, was the primary focus for subsequent studies.

Kidney damage was assessed at 72-h post-injection (Figure 2). Red blood cell congestion was observed in the Stx1 and Stx2a-treated mice (Figure 2, yellow arrows), but not the control mice. The failure of extensive perfusion in the presence of the anticlotting agent, heparin, to eliminate the red blood cells from the tissues suggests the presence of preexisting clots. Increased spacing in the Bowman's capsule was observed in the glomeruli of the Stx2a-treated mice (Figure 2C, blue arrows). In addition, the Stx2a-treated mice displayed diffuse tubular dilation in the renal cortexes, and minimal-to-moderate acute tubular necrosis of distal tubules, characterized by tubules lined with degenerating, necrotic, or sloughed epithelial cells (Figure 2F, green arrows). Kidney lesions were present in the Stx1-treated mice (Figures 2B,E, yellow and green arrows), but were less severe than those observed in the Stx2a treated animals. Thus, as observed in previous studies, Stx2a is more potent than Stx1, with the smaller dose of Stx2a (7 ng) causing greater weight loss and kidney pathology than the much larger dose of Stx1 (1500 ng).

Coronal sections of mouse brain tissue were stained with H&E (Figure 3), and few changes were observed. However, as observed in the kidney, red blood cells were present in the brains of the Stx2a-treated mice. They first appeared at 24 h post-injection (data not shown), and at 48 and 72 h post-injection, large areas of red blood cell congestion and clumping were evident throughout the cortex (Figures 3C,F, green arrows). RBC congestion was not observed in the brains of the Stx1-treated or control mice.

Microglial cell responses are early indicators of neural injury. These responses include up-regulation of surface-marker expression and stereotypical morphological changes, which include conversion from the resting, ramified form (characterized by numerous, long branching processes and a small cellular body), to the macrophage phagocytic form (characterized by a large, ameboid shape, with few processes). We used Iba1, the most commonly used marker to stain for microglial activation (Hoogland et al., 2015). Compared to PBS controls (Figure 4), mice injected with Stx2a showed increased Iba1 staining, and mild morphological changes associated with microglial activation (thickening of processes), which was more pronounced at 72 h compared to 48 h.

Digital imaging was used to quantify changes over larger and more diverse areas of the brain. The data was examined in two different ways. Positive pixel count assesses signal above background, corresponding to the area occupied by microglial cells without regard to intensity. Total pixel intensity indicates the amount of IbaI expression, and corresponds to the level of microglial activation. Stx2a-treatment resulted in statistically significant increased positive pixel numbers in the cortex and nucleus amygdala at both 48 and 72-h post-injection (Figure 5A). Significantly increased intensities were seen in all areas of the brain, except the nucleus at 72-h post-injection (Figure 5B). Stx1 only caused statistically significant increases in positive pixel numbers for the cortex (Figure 5A).

The brains of control mice and mice given a sublethal dose of Stx2a (3 ng) were examined by MRI. Brain chemistry changes upon death. Mice treated with Stx2a can deteriorate rapidly, and a sublethal dose of toxin was used to ensure viability of the animals; however all of the mice were symptomatic at the time of imaging, as evidenced by weight loss and/or lethargy. Brain scans from both the control mice and mice treated with Stx2a were unremarkable and the Stx2a treated animals did not exhibit hyperintensities in the thalamus or any other regions of the brain (data not shown), consistent with the lack of obvious pathologic changes in the H&E stained sections.

Brain metabolite concentrations in the thalamus and cortex regions were determined by MRS, and values within the limit of detection and with an acceptable range of standard deviation are plotted (Figure 6). In the thalamus, the Stx2a-treated mice had lower concentrations of all of the metabolites, but only phosphocreatine reached statistical significance (Figure 6A, PCr). No significant differences were seen in the metabolites of the cortex (Figure 6B).

We tested an immortalized line, bEnd.3, from BALB/c mice and primary human microvascular endothelial cells (HBMECs) isolated from the cerebral cortex for sensitivity to Stx1 and Stx2a. Cells were incubated with 10-fold dilutions of purified Stx1 or Stx2a, ranging in concentration from 0.5 to 5 × 10⁻⁶ μg/ml for bEnd.3 cells or 0.5 to 5 × 10⁻⁷ μg/ml for HBMECs. After 42 h, mitochondrial metabolic activity was assessed by the reduction of resazurin to the fluorescent resorufin. No loss of metabolic activity was seen in the toxin-treated cells for either the mouse bEnd.3 immortalized line (Figure 7A) or the primary human brain endothelial cells (Figure 7B). Previous reports suggested that sub-confluent cells are more sensitive to Stx than confluent cells (Obrig et al., 1988), however in our studies the level of cellular density did not alter susceptibility (Figures 7A,B). In addition, previous studies reported pre-treatment with TNF-α can sensitize cells to Stx2a (Ramegowda et al., 1999; Eisenhauer et al., 2001). Pre-incubation of human brain endothelial cells with TNF-α (10 ng/ml) did not sensitize the cells to the effects of Stx (Figure 7B). However, the endothelial cells were responding to TNF-α, as evidenced by up-regulation of cell-surface expression of ICAM-1, following treatment with TNF-α (Figure 7E). Both primary neonatal dermal microvascular endothelial cells (dHMEC; Figure 7C) and immortalized human dermal microvascular endothelial cells (CDC.HMEC-1; Figure 7D) showed a dose-dependent reduction in metabolic activity following treatment with Stx1 or Stx2a, confirming the activity of both toxins.

To determine the molecular basis for the lack of response to Stx, flow cytometry was used to determine if both the mouse and human cells were resistant to the effects of Stx due to a lack of Gb3 receptor expression. For the bEnd.3 cells, equivalent staining was seen in the presence or absence of the primary antibody to Gb3, suggesting the mouse cells lacked expression of Gb3 (Table 1). Slightly increased values of the geometric mean fluorescence were observed for the human BMECs both without and with TNF-α stimulation. Mean fluorescence for bEnd.3 cells or HBMECs stained for Gb3 was not significantly greater than unstained control cells (P > 0.05 paired t-test). CDC.HMEC-1 cells expressed Gb3 in agreement with the susceptibility of these cells to toxin (Table 1).

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