Distinct Renal Pathology and a Chemotactic Phenotype after Enterohemorrhagic Escherichia coli Shiga Toxins in Non-Human Primate Models of Hemolytic Uremic Syndrome

Abstract

Enterohemorrhagic Escherichia coli cause approximately 1.5 million infections globally with 176,000 cases occurring in the United States annually from ingesting contaminated food, most frequently E. coli O157:H7 in ground beef or fresh produce. In severe cases, the painful prodromal hemorrhagic colitis is complicated by potentially lethal hemolytic uremic syndrome (HUS), particularly in children. Bacterial Shiga-like toxins (Stx1, Stx2) are primarily responsible for HUS and the kidney and neurologic damage that ensue. Small animal models are hampered by the inability to reproduce HUS with thrombotic microangiopathy, hemolytic anemia, and acute kidney injury. Earlier, we showed that nonhuman primates (Papio) recapitulated clinical HUS after Stx challenge and that novel therapeutic intervention rescued the animals. Here, we present detailed light and electron microscopic pathology examination of the kidneys from these Stx studies. Stx1 challenge resulted in more severe glomerular endothelial injury, whereas the glomerular injury after Stx2 also included prominent mesangiolysis and an eosinophilic inflammatory infiltration. Both toxins induced glomerular platelet-rich thrombi, interstitial hemorrhage, and tubular injury. Analysis of kidney and other organs for inflammation biomarkers showed a striking chemotactic profile, with extremely high mRNA levels for IL-8, monocyte chemoattractant protein 1, and macrophage inflammatory protein 1α and elevated urine chemokines at 48 hours after challenge. These observations give unique insight into the pathologic consequences of each toxin in a near human setting and present potential pathways for therapeutic intervention.

Contamination of food and water sources with Shiga toxin-producing enterohemorrhagic Escherichia coli (EHEC) is a global cause of sporadic outbreaks of painful diarrhea and hemorrhagic colitis^1–3 with an estimated 176,000 cases in the United States annually and approximately one death for every 1000 infections.^4,5 Symptoms arise within 3 to 4 days after infection and most resolve, but 5% to approximately 10% of patients progress to develop hemolytic uremic syndrome (HUS).⁶ Postdiarrheal HUS is characterized by thrombocytopenia, nonimmune hemolytic anemia, and thrombotic microangiopathy, often progressing to acute renal injury with severe cases requiring renal dialysis.⁷ The most vulnerable to infection are the young and elderly,⁸ and EHEC infections are a leading cause of acute renal failure in otherwise healthy children in the United States.

EHEC bacteria attach to the intestinal epithelium with characteristic attaching and effacing lesions, which allows type III secretion of bacterial effector proteins and the Shiga toxin type-1 and type-2 toxins (Stx1, Stx2) and several variants into the host.⁹ Bacteremia is rare, and these toxins are primary contributors to the development of HUS and organ damage.¹⁰ The strain often associated with greatest severity is the O157:H7 serotype,¹¹ although there are dozens of pathogenic strains. New strains are emerging with greater virulence as experienced in Germany during summer 2011 when a rare enteroaggregative E. coli O104:H4 strain that causes otherwise self-limiting diarrhea acquired both a stx2 gene and aggressive virulence.^12–14 This is a matter of considerable concern because antibiotics increase HUS risk,¹⁵ and no toxin-specific therapies are available.

The relative contribution of the two toxins to organ injury is difficult to distinguish in patients because EHEC strains can secrete one or both toxins in differing ratios, and the EHEC strain may not be identified or reported. Organ injury is assumed to be roughly equivalent between the toxins, although postdiarrheal renal injury is more commonly associated with EHEC strains that secrete Stx2.¹⁰ There is suggestion that inhibition of only Stx2 is necessary for therapeutic relief,¹⁶ but no data are available that directly compares the toxins in an animal model that presents with full-spectrum HUS.

In ongoing studies to develop clinically relevant EHEC and HUS animal models, we are characterizing the pathophysiology elicited by Stx1 or Stx2 in juvenile baboons (Papio). We previously showed that, when the toxins are administered intravenously, they elicit thrombocytopenia, hemolytic anemia, thrombotic microangiopathy, and acute kidney injury, consistent with HUS.¹⁷ Using this model, we demonstrated rescue of the animals from an otherwise lethal Stx2 challenge and preservation of kidney function, with a custom-designed anti-Stx2 synthetic peptide.¹⁸ When comparing effects of the two toxins, we observed substantial distinguishing features, including different proinflammatory responses and different timing with delayed organ injury after Stx2 challenge. We present here detailed pathology examinations of kidney tissue from the animals challenged with Stx1 or Stx2 and cytokine analyses that extend our prior characterizations of kidney injury. In the baboon model, as in humans, the glomeruli are a particular target of the toxins, but injury is not exclusive to that structure. Stx1 and Stx2 had distinct effects on the glomeruli, with endothelial injury predominating with Stx1 and mesangial injury a predominant feature with Stx2. Both toxins elicited a marked chemotactic environment in the kidneys and other organs that may contribute to the pathophysiology.

Materials and Methods

Reagents

Purified recombinant Stx2 was obtained from BEI Resources (Manassas, VA) or was purchased from Phoenix Laboratory, Tufts-New England Medical Center Microbial Products and Services Facility (Tufts University School of Medicine). Stx1 was prepared from cell lysates obtained from E. coli DH5-harboring plasmid pCKS112, which contains the stx1 operon under the control of a thermoinducible promoter. Stx1 was purified from cell lysates by sequential ion-exchange and chromatofocusing chromatography. The purity of toxins was assessed by SDS-PAGE with silver staining and by Western blot analysis. Prepared toxins contained <0.1 ng endotoxin/mL, as determined by Limulus amoebocyte lysate assay (Associates of Cape Cod, Inc., East Falmouth, MA).

In Vivo Toxin Challenge Procedures

The animal studies were performed at the University of Oklahoma Health Sciences Center animal annex with the use of previously published procedures.^17,19,20 Six- to 8-kg juvenile baboons (Papio c. cynocephalus or Papio c. anubis) were housed and used in accordance with approved protocols from the Institutional Animal Care and Use Committees and the Institutional Biosafety Committees of Boston University School of Medicine and the University of Oklahoma Health Sciences Center. Death is not an endpoint, and baboons were euthanized according to established criteria if deemed necessary, before the end of the 7-day experimental period. Animals (n = 6 per group) were injected i.v. with Stx1 at 10, 50, or 100 ng/kg or Stx2 at 10, 50 ng/kg on day 0. The highest toxin dose is lethal; toxin dose-response profiles showing development of HUS are published for these models.^17,18 All animals received 10 mg/kg Baytril (enrofloxacin antibiotic) i.m. on day 0 before toxin challenge. Baboons then received either 3.5 mg/kg prophylactic Levaquin (levofloxacin antibiotic) i.v. bolus or 10 mg/kg Baytril i.m. each day over the experimental period. Toxin-induced hypovolemia was controlled on the basis of criteria developed and used in previous studies.^17,18,20

Histopathology

Organs were harvested at necropsy or at 7 days after challenge, and tissues were fixed in 10% phosphate-buffered formalin, embedded in paraffin, and sectioned. Sections of kidney tissue were stained with H&E or PAS.

Electron Microscopy

Tissue for electron microscopy was fixed in half-strength Karnovsky solution (1% paraformaldehyde and 1.25% glutaraldehyde in 0.1 mol/L sodium cacodylate buffer, pH 7.0). Tissue was postfixed in 1% osmium tetroxide for 2 hours, dehydrated in graded ethanols, and embedded in epoxy resin. Thin sections were stained with uranyl acetate and lead citrate and examined with a JEOL 1010 transmission electron microscope (Peabody, MA). Kidney tissue examination included at least two and usually three or more glomeruli per animal, as well as a survey of cortical and medullary interstitium, tubules, and blood vessels.

qPCR on Baboon Tissues

Primers for cloning baboon tumor necrosis factor-α (TNFα), IL-6, IL-8 (CCL2), IL-12p35, macrophage inflammatory protein 1α (MIP-1α CCL3), monocyte chemotactic protein 1 (MCP-1, CCL2), and vascular endothelial growth factor (VEGF) were designed with the use of Clustal W alignments of human, chimpanzee, and rhesus nucleotide sequences of these cytokines. Primers were chosen on the basis of these sequence alignments to have a Tm of approximately 55°C, 12 to 18 nucleotides, and 20% to 80% guanidine and cytosine content. Amplicons were generated from baboon cDNA (Expand High Fidelity PCR Kit; Applied Biosystems, Foster City, CA) and cloned into pCR4 vector with the use of the ZeroBlunt TOPO cloning kit for sequencing (Invitrogen, Carlsbad, CA). Sequence quality was determined from the trace files with the use of 4 Peaks sequence analysis software (Groothuis AGand Mekentosj B.V.; 4 Peaks, The Netherlands). Contigs were assembled with Geneious bioinformatics software version 4.6 (Biomatters Ltd., Auckland, New Zealand). Baboon TNFα and IL-12p35 sequences available in the GenBank database were used (NM_001112648 and AY234219.1, respectively). Obtained sequences were used to design baboon-specific quantitative real-time PCR (qPCR) primers.

Primer sets for qPCR assays were designed (Table 1) with the use of standard parameters with the Primer3 plug-in of Geneious bioinformatics software. Validation of assay efficiency and specificity for each primer set was determined by linear regression and melting curve analysis.

Table 1.

Primers for Cytokine/Chemokine qPCR

Species	Target	Forward	Reverse
Baboon	TNFα	5′-TTCAGCTGGAGAAGGGTGAT-3′	5′-CCAAAGTAGACCTGCCCAGA-3′
	IL-6	5′-ATGCAATAACCACCCCTGAA-3′	5′-CTGCAGCCACTGGTTCTGT-3′
	IL-8	5′-CCTTTCCACCCCAAATTTATC-3′	5′-TTCTGTATTGACGCAGTGTGG-3′
	IL-12p35	5′-CCACAAAAATCCTCCCTTGA-3′	5′-CCGAATTCTGAAAGCATGAAG-3′
	MIP-1α	5′-CCCGGTGTCATCTTCCTAAC-3′	5′-CCACTCCATACTGGGGTCAG-3′
	MCP-1	5′-AAGCAGAAGTGGGTTCAGGA-3′	5′-AGGCTTCGGAGTTTGGATTT-3′
	VEGF	5′-AGGCCAGCACATAGGAGAGA-3′	5′-CCTCGGCTTGTCACATTTTT-3′
Human	IL-8	5′-GGTATCCAAGAATCAGTGAAGA-3′	5′-CTACAACAGACCCACACAATA-3′
	MCP-1	5′-GTCTCTGCCGCCCTTCTGTG-3′	5′-AGGTGACTGGGGCATTGATTG-3′
	VEGF	5′-CGAGGGCCTGGAGTGTGT-3′	5′-CCGCATAATCTGCATGGTGAT-3′
	IL-6	5′-TCAGATTGTTGTTGTTAATGG-3′	5′-TAGTGTCCTAACGCTCAT-3′
	β-actin	5′-CCTGGCACCCAGCACAAT-3′	5′-GCCGATCCACACGGAGTACT-3′

Total RNA was isolated from baboon tissues stored in RNA later. Tissues (<25 mg) were ground with a mortar and pestle in a small amount of liquid nitrogen. Proteinase K (20 μg/mL) was added, and samples were incubated for 1 hour at 37°C. RNA was extracted with TRIzol and isolated (Pure link-RNA mini kit; Ambion, Carlsbad, CA). Samples were treated with DNase, and total RNA was quantified with nanophotometer (Implen Inc., Westlake Village, CA). cDNA was reverse transcribed from RNA (High-Capacity cDNA reverse transcription kit; Applied Biosystems). qPCR of experimental samples (tissues from 100 ng/kg Stx1, n = 3; or 50 ng/kg Stx2, n = 4), healthy samples (tissues from a healthy adult baboon), and negative controls was performed with 250 nmol/L baboon-specific and cytokine-specific primers, 2× SYBR Green Fast PCR Master Mix (Applied Biosystems) with the use of a StepOne Plus system (Applied Biosystems). Reactions were cycled as follows: 95°C for 5 minutes, then 40 cycles of 95°C for 1 second followed by 60°C for 3 seconds. To ensure specificity, a melting curve analysis was performed with each run. Quantitation was based on standards from dilutions of human DNA along with human glyceraldehyde-3-phosphate dehydrogenase primers in each test run. Ct values were obtained with StepOne software, and a standard curve was plotted with Ct versus DNA concentration. The Ct values of test samples were extrapolated, and the amount of RNA was expressed as ng/μg total RNA. Fold change in the expression of each cytokine was calculated relative to untreated (healthy) sample values.

Cell Culture

Primary human renal glomerular endothelial cells (Cascade Biologics, Portland, OR) were grown in EC medium with endothelial cell growth supplement (ScienCell, Carlsbad, CA) on tissue culture plates coated with 0.1% gelatin. Cells (passages 2 to 5) were incubated with 400 ng/mL Stx1 or Stx2 for 24 hours (>90% viable). Total RNA was isolated with RNAqueous Kit (Applied Biosystems). RNA was reverse-transcribed into cDNA with the use of QuantiTect Reverse Transcription Kit (Qiagen, Valencia, CA). qPCR reactions were run with the fast SYBR Green master mix (StepOne Plus; Applied Biosystems) according to the manufacturer’s protocol. Data analyses were performed with StepOne Software 2.0 (Applied Biosystems). The ratio of mRNA was calculated by the comparative threshold cycle method with β-actin (ACTB) as control. Primers for each human cytokine gene were designed with Primer Express 3.0 (Applied Biosystems); primer sequences can be found in Table 1.

Cytokine Biomarker Analyses

Cytokine protein levels were quantified by xMAP multiplex fluorescent bead-based assays with the use of a Luminex 200 IS system (Millipore, Billerica, MA), Luminex xPONENT software version 3.1 (Luminex, Austin, TX), and nonhuman primate cytokine panel kits (Millipore) as described previously.¹⁷ For each sample, median fluorescent intensity was analyzed with a weighted five-parameter logistic (Milliplex Analyst; Millipore) and quantified relative to the standard curve for that cytokine. Student’s t-test was used to determine statistical differences between toxins and from time zero (T0) values.

Results

Previously, we showed that challenge of baboons with Stx1 or Stx2 elicited shared and distinct effects in a dose-dependent manner.¹⁷ Both toxins elicited the spectrum of clinical manifestations, characteristic of HUS, including thrombocytopenia, thrombotic microangiopathy, and acute renal injury with progression to renal failure. Detailed pathologic examinations are presented here, with a focus on the kidneys as a primary target organ of the toxins.

Stx1 Induces Severe Glomerular Endothelial Injury

Kidneys from 13 Stx1-challenged animals (three at 10 ng/kg; five at 50 ng/kg; and five at 100 ng/kg) were examined by light microscopy. Light microscopic changes in the kidneys from Stx1-challenged animals were characteristic of severe injury of the glomerular filtration barrier with dose-dependent severity. In the low-dose (10 ng/kg) Stx1 animals, only mild/minor changes were observed in the glomeruli, including mild glomerular hypertrophy (three of three animals), mild mesangial expansion (two animals), or mild glomerular hypoperfusion (one animal). No significant pathologic changes were observed in the tubulointerstitium or vasculature (Figure 1, A–C; Table 2). More substantial pathologic changes were observed in animals exposed to higher doses of Stx1 (50 ng/kg, Figure 1, D–F; 100 ng/kg, Figure 1, G–I). The changes were similar in nature and severity for both doses.

Figure 1.

Kidneys from Stx1-challenged animals show dose-dependent injury of the glomerular filtration barrier and tubules. Light microscopic images shown represent the most severe changes seen at each dose. Stx1 dose indicated at left for each row. Scale bars (for each column): 50 μm (G); 100 μm (H); 0.5 mm (I). A, D, and G: PAS-stained kidney cortex. Glomeruli show a membranoproliferative pattern of injury at higher doses of Stx1, with increased lobularity, capillary wall thickening and double contouring, loss of capillary lumens, mild mesangial expansion, and focal hypoperfusion. Proteinaceous debris is seen in Bowman’s space at higher doses. B, E, and H: H&E-stained kidney cortex shows features of acute tubular injury at higher doses of Stx1, including tubular luminal distension and epithelial flattening. Focal protein reabsorption is also present in tubular epithelia. C, F, and I: H&E-stained kidney tissue. Medullary columns show increasing proportion of distal tubules occluded with proteinaceous casts, with increasing Stx1 dose. Vasculature (arterial and venous) showed no significant pathologic changes.

Table 2.

Light Microscopic Findings after Challenge with Stx1 or Stx2

Glomeruli						Tubulointerstitium
Toxin	Dose (ng/kg)	Survival (h)	Endocap infil	Mesangial apoptosis	Mesangial mitosis	Distension	Epithelial flattening	Tubular pyelo. (poly)	Interstit hemor	Tubular epithelial prolif./mitoses
Stx1	10	168∗	0	0	0	0	0	0	0	0
	10	168∗	0	0	0	0	0	0	0	0
	10	168∗	0	0	0	0	0	0	0	0
	50	58	0	0	0	1	1	0	0	0
	50	72	0	0	0	1	1	0	0	0
	50	72	0	1	0	1	1	0	0	0
	50	74	0	0	0	1	1	2F	0	0
	50	168∗	0	0	0	0	0	0	0	0
	100	48	0	0	0	1	1	0	0	0
	100	49.5	0	0	0	1	1	0	0	0
	100	57.5	0	0	0	2	2	0	0	0
	100	60	0	0	0	1	1	0	0	0
	100	72	0	0	0	2	2	0	0	0
Stx2	10	111.2	0	3	0	1	1	2F	3	0
	10	288∗	0	0	0	0	0	0	0	0
	10	672∗	0	0	0	0	0	0	0	0
	10	720∗	1F(1)	0	0	0	0	0	0	0
	50	84	0	1FS	0	1	1	0	2†	+/−
	50	96	0	2	1	1	0	0	2†	0
	50	96	0	0	0	0	0	1F	0	0
	50	112	0	4	0	2	2	0	4+	3
	50	121.6	0	4	0	2	2	1F	4+	2
	50	121.3	0	4	0	2	3	0	4+	4+
	50	128	0	3	0	2	2	1F	1F	2

Categories scored: +/− (occasional), 0 to 4+ where 0 = not observed and 4+ = most severe.

F, focal (otherwise diffuse); S, segmental (otherwise global).

^∗Time at necropsy of survivor (survivor is ≥168 hours).

^†Medulla only.

The glomeruli in most of the high-dose Stx1 animals showed changes characteristic of a membranoproliferative pattern of injury, including mild mesangial expansion and increased lobularity of the tuft (seven animals), glomerular basement membrane (GBM) thickening (six animals), and mild-to-moderate double contouring (six animals) (Figure 1, D and G). Focal glomerular hypoperfusion was noted in six animals. All animals in the high-dose groups showed at least one of the above described glomerular pathologic changes at the light microscopic level. Proteinaceous or granular debris was often seen in Bowman’s space of glomeruli in all high-dose animals, regardless of whether the involved glomeruli showed other pathology.

Changes in the tubules were more consistent among the high-dose animals and included mild tubular injury (nine animals), proteinaceous cast formation in distal tubules (nine animals), most notable in the medullary columns, and focally prominent tubular epithelial protein reabsorption (nine animals) (Figure 1, E and H). Focal mononuclear and/or polymorphonuclear interstitial inflammation was observed in two high-dose Stx1 animals. Aside from mild arterial and arteriolar sclerosis seen in two high-dose Stx1 animals, the vasculature (arterial and venous) in Stx1-treated animals showed no significant pathologic changes (Figure 1, F and I).

Stx2 Induces Glomerular Injury and Mesangiolysis

Kidneys from 11 Stx2-challenged animals (four at 10 ng/kg; seven at 50 ng/kg) were examined by light microscopy. Compared with Stx1-challenged animals, Stx2-challenged animals exhibited pathologic changes in the kidney, indicative of severe acute tubulointerstitial injury, including hemorrhage, with less obvious glomerular changes at the light microscopic level (Figure 2 and Table 2). One important exception was that kidneys of animals challenged with Stx2, including one of four low-dose and all high-dose Stx2-challenged animals, exhibited extensive, usually diffuse mesangial cell pyknosis (Figure 2, A and D). This finding was not seen in Stx1-challenged animals. Otherwise, kidneys from three of four low-dose Stx2 animals showed mild glomerular pathology, including mild mesangial expansion (three of four), segmental double contour formation (two of four), and GBM thickening (one of four).

Figure 2.

Kidneys from Stx2-challenged animals show interstitial hemorrhage, inflammation, and mesangiolysis. Light microscopic images shown represent the most severe changes seen at each dose; features described were seen in a subset of low-dose specimens and all high-dose specimens. Stx2 dose indicated at left for each row. Scale bars (for each column): 50 μm (D); 100 μm (E); 0.5 mm (F). A and D: PAS-stained kidney cortex. Glomeruli show mild mesangial expansion and widespread mesangial cell pyknosis and karyorrhexis (arrows). There is also segmental thickening of glomerular capillary walls, and focal inflammatory cell margination. B and E: H&E-stained kidney cortex show features of focal acute tubular injury, focal interstitial hemorrhage, and focal interstitial inflammation with many eosinophils, more prominent at higher doses of Stx2. Focal protein reabsorption is also present in tubular epithelia. C and F: H&E-stained kidney tissue. Renal medullary columns show extensive interstitial hemorrhage. Vasculature showed no significant pathologic changes.

Similar changes were seen in glomeruli of high-dose Stx2 animals. Mild mesangial expansion was observed in all six high-dose Stx2 animals, whereas three animals exhibited focal glomerular hypertrophy, two showed segmental double contour formation, and two exhibited GBM thickening. Two high-dose Stx2 animals also showed focal glomerular hypoperfusion. Isolated mitotic figures were observed in the mesangium in one high-dose Stx2 animal.

The tubulointerstial changes in Stx2-challenged animals appeared to exhibit a greater degree of dose dependence than did the glomerular changes. Only one of the low-dose Stx2 animals showed significant tubulointerstitial pathology, whereas all of the high-dose Stx2 animals showed significant changes. The involved kidneys showed features of tubular injury, including mild-to-moderate tubular distension and epithelial flattening; focal tubular epithelial protein reabsorption was also seen in the same specimens (Figure 2, B and E). Four of the six high-dose Stx2 animals also showed tubular regenerative changes, including numerous epithelial cell mitotic figures. Scattered proteinaceous casts were observed in one low-dose and four high-dose Stx2 specimens.

One low-dose and five high-dose Stx2 animals showed extensive interstitial hemorrhage that extended into the medulla in some cases (Figure 2, B, C, E, and F). Focal tubulointerstitial inflammation, often with a prominent eosinophilic infiltrate, was noted in three low-dose and four high-dose Stx2 specimens. Mild arterial and arteriolar scleroses were seen in the kidneys of one low-dose and four high-dose Stx2 animals.

Electron Microscopic Findings after Stx1 or Stx2

Kidney tissue from one high-dose (50 ng/kg) Stx1-challenged animal was available for examination by electron microscopy. The glomeruli in this animal showed features of severe endothelial injury at the ultrastructural level, including diffuse endothelial cell swelling and loss of fenestrations, extensive detachment of the endothelial cells from their basement membranes, focal endothelial denudation, and platelet thrombus formation and fibrin activation in numerous capillary loops (Figure 3, A–C). Other structures in the glomeruli, including the podocytes, GBM, and mesangium, did not show significant pathologic changes. The peritubular capillaries also exhibited signs of severe endothelial injury, including diffuse swelling of endothelial cells, focal endothelial denudation, focal fibrin activation, and occasional marginating inflammatory cells (Figure 3D). Mild features of endothelial injury were also noted in arterioles, including mild endothelial swelling and focal inflammatory cell margination. The tubules exhibited features of focal injury, including vacuolization, loss of brush border, and cell debris in some tubular lumens. Focal tubular epithelial protein reabsorption was also seen, as represented by the presence of electron dense lysosomes in tubular epithelial cells.

Figure 3.

Electron microscopy of kidney tissue from a Stx1-challenged animal shows acute endothelial injury and platelet thrombus formation. Tissue from a Stx1-challenged animal was available for examination by electron microscopy. A: Typical glomerular segment shows thrombosis and occlusion of capillary loops and accumulation of proteinaceous fluid in Bowman’s space. B: Capillary loop segment with endothelial loss and fibrin accumulation (arrow). Podocyte foot processes are intact. C: Capillary loop occluded by swollen cells, platelets, and fibrin. D: Peritubular capillary exhibiting endothelial cell swelling, focal endothelial denudation (arrowheads), focal fibrin activation (arrow), leukocyte margination, and blood flow stasis. Scale bars: 4 μm (A and D); 1 μm (B); 2 μm (C).

Kidney tissue from five animals challenged with Stx2, including one animal challenged with 10 ng/kg and four animals challenged with 50 ng/kg Stx2, were examined by electron microscopy. The animal challenged with 10 ng/kg Stx2 showed no significant pathologic changes in the kidney at the ultrastructural level. However, kidneys from all of the high-dose Stx2-challenged animals (Figure 4, A–D) showed features of moderate-to-severe glomerular endothelial injury, including cell swelling and loss of fenestrations, focal and segmental cell detachment, widening of the subendothelial space, and subendothelial accumulation of loose granular material. In one instance, a mitotic figure was observed in a glomerular endothelial cell. The endothelial changes were accompanied by mild and focal GBM irregularities, including isolated segmental basement membrane redundancy (double contours) in three of four animals. Focal platelet microthrombus formation was noted in glomerular capillary loops of two high-dose Stx2-challenged animals.

Figure 4.

Electron microscopy of kidneys from Stx2-challenged animals show acute endothelial injury and mesangial apoptosis. Kidney tissues from four animals challenged with 50 ng/kg Stx2 were examined by electron microscopy. A: Glomerular segment, typical of all specimens examined, shows features of endothelial injury, including cell swelling and loss of fenestrations. Podocyte foot processes show mild morphologic irregularity but do not exhibit outright effacement. A pyknotic mesangial cell is noted at lower right. B: Mesangial area shows dissolution of the mesangial matrix (mesangiolysis) and mesangial cell pyknosis and karyorrhexis (arrows). C: Capillary loop at left shows endothelial cell swelling, detachment, and segmental denudation. A platelet thrombus is attached to the endothelial surface (arrow). Podocytes show mild foot process irregularities, microvillous degeneration, and electron dense lysosomes (protein reabsorption granules), indicating protein loss through the glomerular barrier. D: Peritubular capillary at top of image shows endothelial cell swelling and mitosis (arrow). Endothelial vacuolization and leukocyte margination is also evident. Red blood cells are scattered throughout the loose, edematous connective tissue of the interstitium, indicative of interstitial hemorrhage. Scale bars: 4 μm (A–D).

In each of the five high-dose Stx2 animals, the mesangial areas showed various amounts of mesangiolysis, characterized by replacement of mesangial matrix with loose granular material. Three high-dose animals showed various degrees, ranging from focal to diffuse, of mesangial cell pyknosis and karyorrhexis, characteristic of apoptosis.

Although podocytes in low-dose Stx2 animals showed no pathologic changes, in all high-dose animals, the podocytes showed features of mild injury, including scattered foot processes with irregular appearing cross sections (with little or no effacement per se) in four animals, mild microvillous degeneration in three animals, and focal protein reabsorption in one animal; the latter indicative of protein loss through the glomerular barrier.

Peritubular capillaries in all high-dose Stx2 animals showed a variable extent of endothelial injury, with focal mild-to-diffuse moderate endothelial swelling and vacuolization, associated with focal blood flow stasis. Focal thrombosis and inflammatory cell margination were also noted within peritubular capillaries in three of four high-dose Stx2 animals. In one high-dose animal, a mitotic figure was noted in a peritubular capillary endothelial cell (Figure 4D). High dose animals also showed generally mild interstitial inflammation and hemorrhage. Minimal endothelial injury was observed in arterioles. Focal tubular injury was observed in all high-dose animals, with epithelial flattening, vacuolization, and/or loss of brush border noted in scattered tubules.

Both Toxins Increase Chemokine mRNA and Urinary Chemokine Levels

Animals challenged with either Stx1 or Stx2 showed considerable inflammatory cell margination of mononuclear and polymorphonuclear inflammatory cells (Table 3), some of which were eosinophilic (Figure 5). This suggests the activation of leukocyte recruitment mechanisms.

Table 3.

Renal Interstitial Inflammatory Cell Infiltrates

Toxin	Dose ng/kg (No.)	Survival, average hours (range)	Mono (Range, 0–4)	Poly (Range, 0–4)
Stx1	10 (3)	168∗	0 (0)	0 (0)
	50 (5)	69 (58–168)	0.20 (0–1F)	0.6 (0–2FE)
	100 (5)	57.4 (48–72)	0 (0)	0 (0)
Stx2	10 (4)	153.8 (111–168)	1.5 (0–2F)	1.0 (0–2FE)
	50 (7)	108.4 (84–128)	0.42 (0–2F)	1.3 (0–2FE)

E, infiltrate includes eosinophils; F, focal (otherwise diffuse); Mono, mononuclear; Poly, polymorphonuclear.

^∗Animal survived to day 7 (necropsy at approximately 168 hours).

Figure 5.

Eosinophilic inflammatory infiltrates after Stx1 challenge. Kidney tissue from an animal challenged with 50 ng/kg Stx1 (euthanized at 168 hours after challenge) was examined by light microscopy after H&E staining. Black boxes: perivascular infiltrates with a predominance of eosinophils. White box: PMN infiltrates.

To evaluate this further, mRNA levels for several common cytokines and chemokines were determined with tissues obtained at necropsy from experimental animals (100 ng/kg Stx1, n = 3; 50 ng/kg Stx2, n = 4) and baboon-specific primers. Data were normalized for values obtained with tissues from a healthy baboon. In the kidneys, high levels of mRNA were found for chemokines IL-8 (CXCL8), MCP-1 (CCL2), and (MIP-1α (CCL3) from animals challenged with high-dose Stx1 or Stx2 (Figure 6). Stx2 (mean, 861.1; range, 148.7 to 1562.0) increased VEGF mRNA levels significantly higher than Stx1 (mean, 5.7; range, 0.07 to 16.2; P < 0.05). In comparison, changes in mRNA for the inflammatory cytokines TNFα and IL-12p35 were modest, and no differences were observed between Stx1 or Stx2 challenge.

Figure 6.

Chemokine mRNA is highly up-regulated in kidneys after Stx1 or Stx2 challenge. Baboon kidney tissues obtained at necropsy after challenge with 100 ng/kg Stx1 (black) or 50 ng/kg Stx2 (striped) were processed to obtain total RNA, then cDNA was generated, and qPCR was performed for each mediator shown as described in Materials and Methods. Data were normalized to values obtained from normal healthy baboon kidney tissue to obtain fold increase. Data shown are means ± SD obtained from multiple animals (Stx1, n = 3; Stx2, n = 4).

Stored tissues suitable for quantification of cytokine/chemokine protein were not available to confirm protein production from the increased mRNA levels, but timed urine samples were part of the original experimental design to characterize pathophysiology of Stx challenge. Urine samples from the same animals as those presented here for pathologic examination were tested for cytokine/chemokine protein levels. Timed samples showed increases in urine chemokines within 24 to 48 hours after toxin challenge (Figure 7A). Compared with urine collected at T0 just before toxin injection (Figure 7B), most biomarker levels shown were significantly elevated at 48 hours after challenge with either toxin (IL-6, P < 0.01; IL-8, P < 0.05; MCP-1, P < 0.00001; VEGF, P < 0.001). Urinary MIP-1α levels did not change, and TNFα was below detection limits. When comparing differences between the toxins, urinary VEGF (P < 0.01) and proinflammatory IL-6 levels (P < 0.05) were higher after Stx1 at 48 hours after challenge. Stx1 induces a more rapid inflammatory response and time to euthanasia compared with Stx2 challenge in the baboon models.¹⁷

Figure 7.

Urine chemokine levels increase after Stx challenge. A: Urine obtained at the indicated time points after high-dose Stx challenge (100 ng/kg Stx1 or 50 ng/kg Stx2) was analyzed for biomarkers with the use of multiplex immunoassays as described in Materials and Methods. Mean values from five animals are shown. B: Urine biomarkers at T0 just before toxin (X) or at 48 hours after toxin from individual animals. Student’s t-test, difference between Stx1 and Stx2. ∗P < 0.05, ∗∗P < 0.01. IL-8, squares; MCP-1, upward pointing triangles; MIP-1α, downward pointing triangles; VEGF, diamonds; IL-6, circles.

To confirm Stx ability to increase chemokine mRNA expression, human renal microvascular endothelial cells were incubated in vitro with either toxin for 24 hours (>95% viability). Stx1 or Stx2 increased IL-8 mRNA expression levels 50-fold to ∼70-fold, but there was little effect of either toxin on MCP-1 or VEGF (Figure 8).

Figure 8.

Chemokine mRNA expression in cultured renal endothelial cells. Human microvascular renal endothelial cells were incubated for 24 hours with 400 ng/mL Stx1 (black) or Stx2 (striped). cDNA was reverse transcribed from total RNA preparations and qPCR was performed with biomarker-specific primer pairs. The mRNA ratio was calculated by the comparative threshold cycle method with β-actin (ACTB) as control. Values are means ± SD of triplicate wells.

Biomarker mRNA levels were similarly quantified by qPCR in tissue samples from other organs taken at necropsy from these animals for comparison (Figure 9). Overall, the changes were similar to that observed in the kidneys, albeit to a lesser degree. Up-regulation of mRNA levels for IL-8, MCP-1, and more variably for MIP-1α, was observed in small intestine, colon, lung, and spleen. Modest or minimal changes were observed for IL-12p35 and TNFα in these organs with no consistent differences elicited by Stx1 or Stx2.

Figure 9.

Chemokine mRNA increases in multiple baboon organs. Small intestine, colon, lung, and spleen tissue samples obtained at necropsy after challenge with 100 ng/kg Stx1 (black) or 50 ng/kg Stx2 (striped) were processed to obtain total RNA for generation of cDNA and qPCR for each mediator shown as described in Materials and Methods. Data were normalized to values obtained from normal healthy baboon kidney tissue to obtain fold increase. Data shown are means ± SD obtained from multiple animals (Stx1, n = 3; Stx2, n = 4).

Discussion

The development of HUS and acute kidney injury after infection with Shiga toxin-producing E. coli is a clinically important complication associated with increased risk of patient morbidity and mortality. The present study presents a detailed pathologic examination of the changes induced in kidneys as a consequence of Stx1 or Stx2 in an animal model that replicates the HUS manifestations observed in patients during EHEC infection. Our primary observations are that the pattern of glomerular and tubulointerstitial injury differed between the two toxins in our baboon models and that challenge with the toxins elicits leukocyte chemotactic responses in kidney and other tissues that may contribute to the pathophysiology.

Glomerular injury after Stx1 challenge featured prominent HUS-like capillary wall changes, including thickening of the GBM and double contouring with prominent thrombus formation, and severe endothelial injury with diffuse cell swelling and focal endothelial denudation. Podocytes did not appear to be injured after Stx1, but after Stx2 podocytes showed features of mild injury at the ultrastructure level. In patients, the renal pathology as a consequence of EHEC infection includes cortical necrosis, glomerular thromboses, and congestion with widened subendothelial space, endothelial cell swelling, neutrophilia, and occasional mesangiolysis.^21,22 We also observed these consequences with the toxins, although mesangiolysis and interstitial hemorrhage were more predominant after Stx2.

Stx1 induced only mild mesangial swelling, in contrast to the prominent loss of mesangial cells after Stx2. Mesangial cells provide structural support for glomerular capillary loops, but also have metabolic functions by generating a unique mesangial matrix, contributing to regulation of glomerular capillary flow and ultrafiltration, and providing growth factors. The loss of mesangial cells can be preceded by damage to glomerular endothelial cells and loss of platelet-derived growth factor B, which is an essential mesangial survival factor. Glomerular endothelial dysfunction in other diseases, such as diabetic nephropathy²³ or Plasmodium vivax malaria,²⁴ often correlates with thrombotic microangiopathy, glomerular capillary microaneurysms, and mesangiolysis. The severe glomerular endothelial injury in Stx1 animals, but selective mesangiolysis in Stx2 animals, suggests that other mechanisms may contribute; but there is sufficient overall cellular damage to impair glomerular function during development of HUS.

Renal inflammatory cell infiltrates observed more frequently after Stx2 challenge included a surprising number of eosinophils. Clinical eosinophilia (>6% of total white blood cell count) is reportedly a common finding in renal diseases with the highest correlation in patients with vascular disease (diabetes, hypertension, cholesterol emboli) and no association with allergies or fungal or parasitic infections.²⁵ Eosinophilia is not observed in baboon Stx models,¹⁷ and IL-5, which induces eosinophil maturation from CD34⁺ precursor cells, is not elevated in the studied models (data not shown). However, eosinophilia was observed in Dutch belted rabbits challenged with EHEC O157:H7,²⁶ and eosinophilic infiltration of the lamina propria accompanied severe intestinal damage of piglets after gastric challenge with a Stx2e-producing EHEC strain.²⁷ Kidney eosinophil infiltration is also an independent predictor of severe renal dysfunction in patients with autoimmune antineutrophil cytoplasmic antibody-associated vasculitis,²⁸ but there is little pathologic data available to describe a presence of eosinophils in kidneys of patients with EHEC infections. We cannot rule out the presence of a coinfection that contributed to the predominance of eosinophils, but the baboons used in the present study are screened routinely for viral and parasitic infections and were clinically healthy.²⁹ This was not observed after Stx1 challenge or low-dose Stx2, and we have not observed eosinophilic infiltration in baboon kidneys after other infectious challenges otherwise.^19,20,30 It is also possible that eosinophilic infiltration after Stx2 is a feature of the baboon toxin model and not of human EHEC infection. Eosinophils may contribute to the thrombotic microangiopathy by promoting a procoagulant environment. Although controversial, one study showed that eosinophils have high concentrations of active tissue factor, the in vivo initiator of coagulation.³¹ It is possible that this tissue factor is from monocyte contamination or acquired passively from other cells,³² but other eosinophil products such as eosinophil peroxidase could contribute to cellular oxidation and inflammation, both of which are procoagulant.³³

The presence of urine chemokines and increased chemokine mRNA in kidney tissue after Stx challenge is consistent with the leukocyte infiltration observed by pathologic examination of the baboon kidneys. Challenge with a high dose of either toxin resulted in increased chemokine mRNA levels in the kidneys and urinary protein levels, particularly IL-8 (CXCL8), MCP-1 (CCL2), and MIP-1α (CCL3). The cell source of the kidney and urine chemokines is difficult to identify in these animals, but our in vitro data suggest that at least renal glomerular endothelial cells are not a likely source of MCP-1 after Stx challenge. This chemotactic environment would favor influx of activated leukocytes that may carry toxin and impose collateral leukocyte-mediated cell injury. MCP-1 attraction of activated monocyte-expressing procoagulant tissue factor could contribute to development of thrombi. EHEC patient neutrophils stain for Stx antigen,³⁴ and kidney IL-8 may recruit activated neutrophils that carry toxin from the intestinal epithelia for deposition at the glomerular endothelium.^35,36 Other studies have shown that the chemokine receptors CXCR4 and CXCR7 with lymphotactic ligand CXCL12 (stromal cell-derived factor-1) contribute to Stx injury of microvascular endothelial cells and are pathogenic in murine Stx2 models.³⁷ Cultured microvascular endothelial cells treated with Stx2 released CXCL12 ligands, and Stx2 increased expression of platelet P-selectin³⁸ which could contribute to thrombotic microangiopathy. Collectively, the prediction is that targeted inhibition of IL-8 or other chemokines may prevent EHEC-HUS or may reduce disease severity, and these studies are in progress.

The outcome after challenge with either toxin is acute kidney injury, but we consistently observe differences in host responses elicited between the two toxins, either tissue pathology or inflammation markers. Mechanisms that could account for these differences may lie in structural differences between the toxins, receptor heterogeneity, and/or receptor microenvironment in a particular tissue. The prototype Stx1 and Stx2 molecules are only 56% identical at the amino acid level, and each has multiple variants that are associated with differing degrees of clinical disease.³⁹ Targeted loss of the Gb3 glycosphingolipid receptor (CD77) results in resistance to Stx intoxication in mice,⁴⁰ but in vivo susceptibility during infection is likely governed by more subtle mechanisms. Both toxins bind to Gb3 with similar specificity, but the Stx1 interaction has a 10-fold higher affinity and the Stx2 interaction has a slower off-rate.⁴¹ A related glycolipid, Gb4, has high affinity for the Stx2e variant to cause edema in a piglet model but little affinity for other Stx2 molecules.⁴² There is also heterogeneity in the fatty acyl chains of Gb3 with respect to chain length and degree of unsaturation that contributes to differential toxin binding and intracellular effects.^43,44 Variability in receptor expression and density on endothelial cells is well known, either in vitro with degree of cell confluence, or in vivo with respect to vascular bed⁴⁵ and presence of inflammatory signals such as TNFα.⁴⁶ In an earlier model of baboon Stx1 toxemia, preconditioning the animal with lipopolysaccharide up-regulated toxin receptor expression in kidneys and increased Stx1-mediated disease severity.⁴⁷ Finally, these glycolipid toxin receptors are enriched in lipid rafts of caveolae, which are plasma membrane microenvironments that support receptor avidity, signaling, and intracellular toxin trafficking.^48,49 Additional studies with isolated baboon glomerular endothelial cells and tubular epithelial cultures are necessary to identify contributions of receptor heterogeneity and toxin binding kinetics to differential toxin responses.

Priming of the Stx system with lipopolysaccharide (endotoxin), TNFα, or ADP potentiates toxin chemotactic and inflammatory responses in vitro and in vivo,^50–52 but the relevance to disease pathogenesis in patients is not clear. Addition of lipopolysaccharide priming to Stx2 challenge in mice will induce thrombocytopenia,⁵¹ but this cause is disseminated intravascular coagulation^53,54 rather than HUS, and they have different coagulation profiles. To investigate pathophysiology induced by only the toxins, we made significant effort to reduce endotoxin contamination. Like humans, baboons respond to endotoxin with early increases in TNFα and other inflammatory cytokines.⁵⁵ Lack of detectable TNFα protein in blood^17,56 or urine of our Stx-challenged baboons is consistent with our effort to monitor only responses to Stx. Under these conditions, we consistently observe a modest proinflammatory response to the toxins¹⁷ and development of a robust chemotactic environment.

Limitations of this study include that these are nonhuman primates receiving toxins, not enteric bacterial infection, so correlates with natural EHEC infection are limited, and the toxin doses are likely higher than that experienced by patients. We also observed discrepancy between tissue mRNA and urine protein levels of some biomarkers. For example, VEGF mRNA is highly up-regulated in kidneys after Stx2, but urine VEGF levels are higher after Stx1. It is not possible to know how much of the elevated baboon kidney mRNA levels are translated to protein, and, if so, how much is released into the circulation or urine. The mechanism for this discrepancy is not known, but a contributor may be differential mRNA instability because of the ribosome-inactivating toxins. It is also relevant to note that overall time to euthanasia was earlier for high-dose Stx1-challenged animals at 2 to 3 days than at 4 to 5 days for animals that received high-dose Stx2.¹⁷ This time difference may contribute to some differences observed in tissue mRNA levels, urine biomarkers, and development of tissue pathophysiology. In addition, the development of HUS in the baboons is more rapid than that in patients, which is closer to 7 to 12 days after onset of diarrhea; an enteric EHEC infection model in the baboons is needed and is under development.

In summary, the Stx toxins from EHEC have distinct effects on renal pathology but sufficient shared effect on renal function to result in loss of kidney function during development of HUS. Therapeutic approaches to target only Stx2 may overlook the significant renal injury that Stx1 can elicit, and, although disease progression may not be as severe, long-term kidney function may still be impaired. This is an important consideration because 25% to 36% of surviving patients have long-term kidney sequelae.⁵⁷ Both toxins elicit a strong chemotactic response in tissues, particularly the kidneys, and this environment is likely a primary contributor to toxin distribution and local inflammatory responses, possibly enhancing Stx effects and increasing disease severity.