A Murine Model of HUS: Shiga Toxin with Lipopolysaccharide Mimics the Renal Damage and Physiologic Response of Human Disease
Tiffany R. Keepers,
Mitchell A. Psotka,
Lisa K. Gross and
Tom G. Obrig
Department of Internal Medicine, Division of Nephrology, University of Virginia, Charlottesville, Virginia
Address correspondence to: Dr. Tom Obrig, University of Virginia, Department of Internal Medicine, Division of Nephrology, Box 800133, Charlottesville, VA 22908. Phone: 434-982-1063; Fax: 434-924-5848; E-mail: to3e{at}virginia.edu
Received for publication May 1, 2006.
Accepted for publication September 26, 2006.
Hemolytic uremic syndrome (HUS), which is caused by Shiga toxin–producingEscherichia coli infection, is the leading cause of acute renalfailure in children. At present, there is no complete smallanimal model of this disease. This study investigated a mousemodel using intraperitoneal co-injection of purified Shiga toxin2 (Stx2) plus LPS. Through microarray, biochemical, and histologicanalysis, it was found to be a valid model of the human disease.Biochemical and microarray analysis of mouse kidneys revealedthe Stx2 plus LPS challenge to be distinct from the effectsof either agent alone. Microarrays identified differentiallyexpressed genes that were demonstrated previously to play arole in this disease. Blood and serum analysis of these miceshowed neutrophilia, thrombocytopenia, red cell hemolysis, andincreased serum creatinine and blood urea nitrogen. In addition,histologic analysis and electron microscopy of mouse kidneysdemonstrated glomerular fibrin deposition, red cell congestion,microthrombi formation, and glomerular ultrastructural changes.It was established that this C57BL/6 mouse is a complete modelof HUS that includes the thrombocytopenia, hemolytic anemia,and renal failure that define the human disease. In addition,a time course of HUS disease progression that will be usefulfor identification of therapeutic targets and development ofnew treatments for HUS is described.
Shiga toxin–producing Escherichia coli (STEC), also knownas enterohemorrhagic E. coli, is the primary cause of diarrhea-associatedhemolytic uremic syndrome (D+HUS) (1–3). This pathogenalso is the major cause of acute renal failure in young children(4). It generally is accepted that systemic toxemia and subsequentrenal disease is due to a combined action of Shiga toxins (Stx1,Stx2) that are the primary virulence factors of STEC and bacterialLPS (1–3,5,6). The series of events that precede acuterenal failure in D+HUS and the respective roles of Stx and LPSin the disease remain to be elucidated. A complete understandingof these events is essential for identification of new drugtargets and development of therapeutics for D+HUS, because thelatter do not currently exist. It is hopeful that such therapeuticsmay be used in the clinical setting, where a 3- to 5-day "windowof opportunity" generally exists before acute renal failure.Toward this goal, in this study we report on the temporal seriesof host response events, including renal gene activation, ina small animal model that exhibits the hallmarks of HUS, thrombocytopenia,microangiopathic hemolytic anemia, and acute renal failure.We also demonstrate that both virulence factors of STEC—Stxand LPS—are required to elicit the triad of HUS symptoms.
Shiga Toxin Purification
Stx2 was purified by immunoaffinity chromatography from celllysates (provided by Alison O’Brien) of E. coli DH5 thatcontained the Stx2-producing pJES120 plasmid (7). Briefly, Stx2was purified using 11E10 antibody (8) that was immobilized usingan AminoLink Plus Kit (Pierce Biotechnology, Rockford, IL) accordingto the manufacturer’s instructions. Endotoxin was removedusing De-toxi-Gel (Pierce Biotechnology) as per the manufacturer’sinstructions. Stx2 purity was assessed by SDS-PAGE and determinedto be endotoxin-free, and activity was measured in a Vero cellcytotoxicity assay. Stx2 was chosen because it is far more frequentlyassociated with HUS clinical isolates than Stx1 (9,10).
Animal Studies
C57BL/6 male mice that weighed 22 to 24 g were purchased fromCharles River Laboratories (Wilmington, MA). Mice received anintraperitoneal injection of a low sublethal dose of LPS at300 µg/kg (O55:B5; Sigma-Aldrich, St. Louis, MO), 225ng/kg Stx2 (two times the LD50), or both. Saline injection wasused for control mice. At 0, 2, 4, 6, 8, 12, 24, 48, and 72h after injection, two mice per time point were killed and kidneyswere processed as described next. These experiments were repeatedthree times. In separate experiments, blood samples were obtainedfrom mice at 0, 4, 8, 12, 24, 36, 48, 60, and 72 h after injection.Mice were weighed every 12 h for 3 d to determine the percentageof weight loss. All animal procedures were done in accordancewith University of Virginia Animal Care and Use Committee policies.
Blood Analysis
Blood was collected from mice with EDTA-treated or with non–EDTA-treatedcapillary tubes via retroorbital bleed. Blood that was collectedwithout anticoagulant was either smeared on microscope slidesor allowed to clot for 30 min at room temperature. Dried bloodsmears were flooded with Wright-Giemsa stain (Sigma-Aldrich)for 1 min and rinsed with distilled water for 2 min. After clotting,blood was centrifuged at 2000 x g for 15 min at 4°C. Theserum layer was removed and stored at –80°C untilanalysis. Creatinine was determined using Cayman Chemical CreatinineAssay Kit (Ann Arbor, MI) as per the manufacturer’s instructions.Blood urea nitrogen (BUN) was determined spectrophotometricallywith VetScan (Idexx Corp., Westbrook, ME). For reticulocytecounts, three drops of EDTA-treated blood was mixed with twodrops of Reticulocyte Stain (Sigma-Aldrich) for 10 min at roomtemperature. Mixtures then were smeared on a microscope slide,dried, and coverslipped. Percentage of reticulocytes was determinedby counting the number of reticulocytes per 1000 red blood cells(RBC). A complete blood count (CBC) was performed on 20 µlof the EDTA-treated blood using MASCOT HEMAVET 850 (CDC Technologies,Oxford, CT) according to the manufacturer’s instructions.
Immunohistochemistry
One half of a mouse kidney was fixed in 4% paraformaldehydefor 24 h, processed, and embedded in paraffin. Three-micron-thicksections were cut and placed onto charged slides. Martius Yellow-BrilliantCrystal Scarlet-Aniline Blue (MSB) differential staining procedurewas performed as described previously (11). Martius yellow andphosphotungstic acid in alcoholic solution stain red cells,brilliant crystal scarlet stains muscle and mature fibrin, andaniline blue stains collagen. Glomeruli that were positive forfibrin staining were quantified by counting three sets of 20glomeruli per slide and averaging the percentage positive forfibrin at each time point. Immunohistochemistry for plateletswas performed using polyclonal goat anti-human integrin-3 antibodythat cross-reacts with the mouse protein (Santa Cruz Biotechnology,Santa Cruz, CA) (12). Primary antibody was detected using anavidin/biotin horseradish peroxidase system (Vector Laboratories,Burlingame, CA) and diaminobenzidine. Sections then were counterstainedin hematoxylin, dehydrated, and mounted.
Electron Microscopy
Kidneys were harvested from killed mice 0, 24, 48, and 72 hafter Stx2 plus LPS challenge, cut into blocks approximately1 to 2 mm3, and fixed overnight at 4°C in 4% paraformaldehydeand 2.5% glutaraldehyde in 1x PBS. The fixed tissue subsequentlywas processed by the University of Virginia Advanced MicroscopyFacility. The tissue blocks were washed with PBS at 24°Cand postfixed for 1 h in 1.0% osmium tetroxide. The blocks thenwere washed in distilled water, dehydrated using a graded acetoneseries, embedded in epoxy resin (EMBED 812; Electron MicroscopySciences, Hatfield, PA), and polymerized for 2 d at 60°C.Ultrathin sections approximately 70 nm in thickness, obtainedwith a diamond knife (Diatome, Hatfield, PA) on a Leica UltracutUCT ultramicrotome (Leica, Bannockburn, IL), were collectedon 200-mesh copper grids, contrast-stained with uranyl acetateand lead citrate according to routine procedures, and examinedin a JEOL 1230 transmission electron microscope (JEOL, Peabody,MA). Digital images were acquired using an SIA L3-C digitalcamera (Scientific Instruments and Applications, Duluth, GA).At least four glomeruli from each of three mice were examinedper time point.
cRNA Synthesis and Microarray Hybridization
One half of a mouse kidney was stored in 2 ml of RNALater (Ambion,Austin, TX) at 4°C until RNA extraction. Total RNA was isolatedusing the RNeasy Midi Kit (Qiagen, Santa Clarita, CA) accordingto the manufacturer’s instructions. Total kidney RNA fromcontrol and treated mice was compared with GeneChip ExpressionAnalysis using Mouse Genome 430A 2.0 Arrays (Affymetrix, SantaClara, CA) by the Biomolecular Research Core Facility (Universityof Virginia, http://www.healthsystem.virginia.edu/internet/biomolec/).
Microarray Analysis and Quantitative Reverse Transcriptase–PCR
Microarray probe–level determinations were made by theProbe Logarithmic Intensity Error estimation method, part ofthe ArrayAssist Lite software package (Stratagene, La Jolla,CA). DNA-Chip Analyzer (dChip) (13) model-based estimation wasused to normalize the signal levels to the median arrays foreach series of experiments. Multiclass significance analysisacross the time course was performed with the Significance Analysisof Microarrays software (14). Genes with q values (false discoveryrate) of <5% were determined to be altered significantlyby challenge. From this set, only genes that were altered 2.0-foldor greater at any time point compared with controls were usedfor further analysis. Cluster analysis was performed on thesedata set with GeneCluster 2.0 software (15), and dChip was usedfor gene ontology analysis of the clusters (13). Expressionpatterns of select genes from each cluster were verified byquantitative real-time PCR using the iScript cDNA SynthesisKit, iQ SYBR Green Supermix and iCycler Thermal Cycler (BioRad,Hercules, CA).
Statistical Analyses
All statistics (excluding those that were used for microarrayanalysis) were performed using single-factor ANOVA followedby two-sample t test, and P < 0.05 was considered significant.
Stx2 Plus LPS Causes Diminished Renal Function
Mice an intraperitoneal injection of a low sublethal dosageof LPS (300 µg/kg), 225 ng/kg Stx2, or both agents. Thedosage of Stx2 chosen is the minimum 100% lethal dosage (twotimes the LD50), resulting in lethality within 4.5 d (Figure 1A).When this dosage of Stx2 was combined with the sublethal dosageof LPS, the time to death was decreased by 1 d. Mice were weighedevery 12 h to determine percentage of weight loss caused bythe toxins (Figure 1B). LPS induced weight loss early in thetime course, whereas Stx2 induced weight loss late in the timecourse. Stx2 plus LPS caused weight loss both early and latein the time course. Mice were evaluated for kidney function,peripheral cell count abnormalities, and structural kidney changes.These results are summarized in Table 1. We determined thatonly mice that received Stx2 plus LPS exhibited all of the signsof clinical HUS. Therefore, we chose to present the data thatare relevant to the complete Stx2 plus LPS mouse model of HUS.Stx2 plus LPS co-administration intraperitoneally at these dosageswas used for all subsequent experiments except where noted.
Figure 1. Survival and weight loss of mice that were challenged with Shiga toxin 2 (Stx2), LPS, and Stx2 plus LPS. Mice received an injection of 225 ng/kg Stx2, 300 µg/kg LPS, or both. (A) Survival curve is representative of three experiments. Data contain six mice per group. (B) Mice were weighed every 12 h after injection for 72 h. Data contain 10 to 12 mice per group. *P < 0.002 significantly decreased from weight at time 0 h; #P < 0.002 significantly increased from weight at time 0 h evaluated by ANOVA and t test.
Table 1. Summary of pathologic conditions that were induced by Stx2, LPS, or Stx2 plus LPS injectiona
Increases in serum levels of creatinine and BUN suggest decreasedglomerular filtration and were used as indicators of abnormalrenal function (Figure 2). After administration of Stx2 plusLPS, creatinine levels were increased significantly comparedwith saline at 12 h and continued to rise with a maximal averageconcentration of 0.92 mg/dl at 72 h. Similarly, BUN levels wereincreased in mice that received Stx2 plus LPS in a time-dependentmanner. BUN was elevated significantly at 8 h after injectionand continued to increase until death, with a maximal averageBUN concentration of 114.33 mg/dl by 72 h.
Figure 2. Serum creatinine and blood urea nitrogen (BUN) analysis. Serum from mice that received Stx2 plus LPS was analyzed for creatinine (A) and BUN (B) concentration. Each circle represents one mouse, and data represent three separate experiments. Black bars are the average of all mouse samples. *P < 0.05 significantly increased over saline by ANOVA and t test.
Stx2 Plus LPS Alters White Blood Cells
A CBC was performed on mice that received Stx2 plus LPS throughoutthe 72-h time course (Figure 3, A through D). Mice exhibitedneutrophilia, lymphocytopenia, and mild monocytosis. Rise inperipheral neutrophil levels was significant at 4 h and maximalat 8 h after Stx2 plus LPS injection, after which levels graduallyreturned to normal at 72 h (Figure 3A). Neutrophils expandedfrom 0.67 ± 0.17 to 2.09 ± 0.48 K/µl andincreased from 12.4 ± 2.1% of the total white blood cell(WBC) population to 56.8 ± 8.9%. Consequently, the lymphocytepopulation decreased from 82.9 ± 4.7% of total WBC withsaline injection to 37.7 ± 5.8% at 8 h after Stx2 plusLPS injection (Figure 3B). Similarly, total lymphocyte levelswere minimal at 4 to 8 h after injection with a significantdrop from 4.69 ± 0.92 to 1.30 ± 0.28 K/µl.Lymphocyte levels then gradually increased throughout the timecourse with a recovery at 72 h to approximately half the normallevel. The monocyte cell population expanded with respect tothe other WBC populations as reflected by the increase in thepercentage of monocytes of total WBC throughout the time course.Whereas the total number of monocytes was not altered significantlyafter Stx2 plus LPS injection, the percentage of monocytes increasedfrom 3.17 ± 1.0% with saline injection to 5.92 ±2.8% 8 h after Stx2 plus LPS injection (Figure 3C).
Figure 3. Peripheral blood cells in mice that received Stx2 plus LPS. (A) Total neutrophil counts (•) and percentage of neutrophils of total white blood cells (WBC; ). *P < 0.01 significantly increased compared with saline. (B) Total lymphocyte counts (•) and percentage of lymphocytes of total WBC (). *P < 0.0001, #P < 0.05 significantly decreased versus saline. (C) Total monocyte counts (•) and percentage of monocytes of total WBC (). *P < 0.05 significantly increased versus saline. (D) Percentage of peripheral reticulocytes. *P < 0.05 significantly increased versus saline. Data are the average of seven to 18 mice and three separate experiments. All data are the average ± SD, and all statistical analysis was performed using ANOVA and t test. (E through H) Blood smears of mice that were administered Stx2 plus LPS at different times after injection: 0 h, control shows no irregularities (E); 24 h after injection, increased neutrophils (arrowhead) and monocytes (arrow) (F); 8 h after injection, appearance of reticulocytes as indicated by arrows (G); 12 h after injection, appearance of red blood cells (RBC) with Howell-Jolly bodies as indicated by arrows (H). All pictures were taken under oil immersion, and (H) was enlarged two-fold. Magnification, x1000.
Blood smears from mice at various time points after Stx2 plusLPS administration demonstrated changes in WBC morphology. Figure 3Fis a blood smear that was taken 24 h after Stx2 plus LPS injectionand depicts a segmented neutrophil as well as an activated monocyte.Monocytes and neutrophils increased in smears beginning at 12h after injection and thereafter were present throughout thetime course. Monocytes displayed evidence of activation, including,increased cell size, linearized chromatin, granules, and pseudopodformation. Neutrophil activation was indicated by increasedcell size, granules, and segmentation as well as hypersegmentation.Plasmacytoid lymphocytes also were present in blood smears.
Stx2 Plus LPS Causes Signs of Hemolytic Anemia
Although a CBC of Stx2 plus LPS mice during the 72-h time courseshowed moderately increased hematocrit and hemoglobin levels(data not shown), blood smears demonstrated signs of anemiaand hemolysis. Hematocrit reached a plateau at 36 h after injectionof 52.0 ± 2.6% compared with the control of 44.3 ±4.3%. However, manual peripheral reticulocyte counts were increasedat 12 h after Stx2 plus LPS injection and remained elevatedthroughout the time course (Figure 3, D and G). Reticulocytesare newly developed immature RBC that appear blue/purple andare a sign of anemia. It also was noted that serum that wascollected from mice that received Stx2 plus LPS had a red discoloration,presumably from hemoglobin as a result of RBC hemolysis.
In addition, blood smears revealed several RBC morphologic abnormalities(Figure 3, G and H) compared with normal (Figure 3E). Howell-Jollybodies are nuclear fragmentations of DNA that appear as small,round, blue structures in erythrocytes in anemic states. Thesewere found in smears beginning at 8 h (Figure 3G, arrowhead),were more numerous at 12 h (Figure 3H, arrows), and continuedto increase through 72 h after Stx2 plus LPS injection. Echinocytesare a morphologic change in which the RBC has uniform spikesor burrs on its surface, indicating uremia. At 24 h after Stx2plus LPS injection, echinocytes were evident in blood smears.
Stx2 Plus LPS Causes Thrombocytopenia in Mice
Platelet levels significantly decreased in mice that were administeredStx2 plus LPS beginning at 4 h after injection and continuedto decline through the time course (Figure 4A). The minimalplatelet level was 392.7 ± 104.9 K/µl at 36 h afterinjection compared with 820.8 ± 134.6 K/µl withsaline injection. In addition, platelet clumping in blood smearswas most apparent at 72 h after Stx2 plus LPS injection. Inthe kidney, glomerular platelet aggregation was increased at2 h after Stx2 plus LPS injection (Figure 4B). After a slightdecline at 12 h, platelet clumping began to increase again untilthe end of the time course (Figure 4, B through D).
Figure 4. Peripheral and kidney platelets in mice that received Stx2 plus LPS. (A) Peripheral platelet levels indicate thrombocytopenia. Data are the average of nine to 14 mice and three separate experiments. *P < 0.005 significantly decreased versus saline. (B) Glomeruli that were positive for platelet clumping in kidney. *P < 0.05 significantly decreased versus control. Three sets of 20 glomeruli were counted for platelet clumps per time point. Immunohistochemistry demonstrating platelet clumping in the kidney at 0 h (C) and 2 h (D) after injection. Platelets are stained brown, and arrows indicate platelet clumping. Data are the average ± SD, and statistics were performed using ANOVA and t test. Magnification, x400.
Stx2 Plus LPS Causes Changes in the Mouse Kidney
The MSB differential stain for RBC, fibrin, and collagen wasperformed on fixed mouse kidney tissue during the time course(Figure 5). In normal kidney tissue (Figure 5A), collagen wasdistributed throughout the inner tubules of the cortex, medulla,and papilla as indicated by light blue staining. There was nosignificant change in collagen levels after Stx2 plus LPS injection.
Figure 5. Fibrin deposition and RBC congestion accumulates over time in mice that were administered Stx2 plus LPS. Martius Yellow-Brilliant Crystal Scarlet-Aniline Blue staining of mouse kidneys at 0 h (A), 4 h (B), 48 h (C), and 72 h (D) after injection (arrow indicates thrombi in arteriole). Bright red staining is fibrin deposition, RBC are stained yellow, nuclei are stained dark blue/purple, and collagen is stained blue. Magnifications: x200 in A through C; x400 in D.
RBC, stained yellow, were widely dispersed throughout the normalkidney in small numbers. There was a general increase in RBCin the entire kidney and RBC clustering in the cortex beginningat 4 h after Stx2 plus LPS injection (Figure 5B). RBC congestionin glomeruli, capillaries, and intertubular spaces continuedto increase during the time course. Specifically, RBC beganto accumulate in the medulla at 6 h after injection, glomerularRBC congestion was evident at 8 h after injection, and largeareas of RBC congestion and clumping were evident throughoutthe cortex at 24, 48, and 72 h after injection.
Significant changes in fibrin staining were evident at 8 h afterStx2 plus LPS injection. Fibrin was increased in the cortexof the kidney in the intertubular spaces, glomeruli, and capillaries.It was dispersed throughout the papilla and medulla. Glomerulithat were positive for fibrin also increased at this time pointfrom 23.3% positive at 0 h to 75.0% positive for fibrin at 8h after injection. Fibrin continued to accumulate with maximalstaining at 48 h after Stx2 plus LPS challenge (Figure 5C).At this time, the medulla and cortex showed a rise in the numberand in the size of concentrated areas of fibrin deposition,and 93.3% of glomeruli were positive for fibrin. In addition,thrombi of RBC and fibrin were seen in the glomerular arterioles(Figure 5D).
Transmission electron microscopy of glomeruli from mice thatreceived Stx2 plus LPS demonstrated significant ultrastructuralchange to the capillary loops, endothelial cells, and podocytescompared with controls (Figure 6). RBC congestion and electron-denseflocculent material, which likely is a combination of collectedserum proteins and fibrin, was present from 24 to 72 h (Figure 6,B and C). Increasingly large extranuclear endothelial inclusionswere observed beginning at 24 h and extending through 72 h afterchallenge (Figure 6B). These inclusions are bounded by two membranesand are anuclear, which suggests that they are remnants of endocytosedRBC, platelets, or other cellular debris. Furthermore, endothelialcell fenestrations were irregular and occasionally partiallydetached from the glomerular basement membrane (Figure 6C).Podocytes appeared swollen, diminishing the urinary space, andsome contained extensive vacuoles of unknown significance (Figure 6D).None of the pathologic findings was observed in control mousekidneys (Figure 6A).
Figure 6. Stx2 plus LPS administration causes glomerular ultrastructural changes in mice. Representative transmission electron micrographs of mouse glomeruli from control (A), and 72 h after challenge (B through D). (B) Endothelial cell with arrows indicating extranuclear inclusions. (C) Glomerular capillary loop with endothelial detachment (arrow). (D) Abnormal podocyte with cytoplasmic vacuolation (arrows). Bars = 2 µm. E, endothelial cell; P, podocyte; R, RBC. Magnifications: x5000 in A; x10,000 in B through D.
Stx2, LPS, and Stx2 Plus LPS Challenges Alter Renal Gene Expression
Microarrays were used to analyze gene expression in the wholemouse kidney in response to Stx2, LPS, or Stx2 plus LPS challenges.Analysis of the data revealed 136, 737, and 722 significantdifferentially expressed genes for the Stx2, LPS, and Stx2 plusLPS challenges, respectively (Figure 7). These genes representedapproximately 1% (Stx2) and 5% (LPS and Stx2 plus LPS) of the14,000 genes that were included on the array. The Venn diagramin Figure 7 demonstrates that, although many of the genes whoseexpression was altered by each challenge were affected by theother challenges, there also was a distinct set of genes thatwere uniquely significant for each challenge. For example, ofthe 136 genes that were differentially expressed in responseto Stx2 alone, 74 also were differentially expressed in responseto LPS alone, 75 also in response to Stx2 plus LPS, and 57 alsowere differentially regulated by both LPS and Stx2 plus LPSwhereas 44 were unique to Stx2 alone.
Figure 7. Venn diagram of genes that were altered 2.0-fold or greater by each challenge with q < 5% in multiclass ANOVA analysis. The total genes used from each challenge are outside the circles. Each challenge alters both an overlapping set of genes and a distinct gene set, compared with the other challenges.
Stx2, LPS, and Combination Challenges Provoke Distinct Renal Expression Patterns
The Stx2, LPS, and Stx2 plus LPS challenges exhibited distincttemporal alterations in gene expression over the time course.Figure 8A is a plot of the number of genes whose transcriptionwas altered, either increased or decreased, at each time pointby each challenge. LPS affected a rapid, early change in geneexpression, whereas Stx2 caused a later, gradual change. Thedifferentially expressed genes from Stx2 plus LPS challengecreated a pattern that shared the respective temporally earlyand late changes in gene expression from the individual agents.To discover classes of genes with similar expression patterns,we performed self-organizing map clustering, an unsupervisedlearning method, with the GeneCluster program. Within each challenge,genes were grouped together on the basis of the similarity oftheir expression patterns within the 72-h time course (Figure 8B).These groups were labeled on the basis of the time point ofmaximum expression change over the course of the pattern. Sixmajor gene expression patterns were observed for LPS challenge.Genes whose maximum differential expression occurred at 2 to4 h after injection were termed "immediate," those at 6 to 8h were termed "early," and those at 12 to 24 h were termed "late."In contrast, Stx2 challenge created only an upregulated anddownregulated pattern, each reaching maximum differential expressionat 48 to 72 h. These clusters therefore were termed "very late"to contextualize them in light of the LPS expression patterns.The Stx2 plus LPS challenge creates an arrangement of gene expressionpatterns that combines each of the Stx2 plus LPS patterns together(data not shown). Supplemental Table 1 contains a list of themost differentially expressed genes in each cluster from Figure 8B.Many of these gene products are upregulated in patients withHUS, most notably IL-6 and monocyte chemoattractant protein-1,thereby adding validity to the HUS mouse model (16,17).
Figure 8. (A) The total number of genes that were altered at each time point by each challenge. Temporal changes in gene expression as a result of LPS and Stx2 are distinct, and the Stx2 plus LPS time course incorporates both of these alterations (time course to scale). (B) Temporal clusters that were created by GeneCluster for LPS and Stx2 challenges. The average expression pattern for each cluster across the 72-h time course is shown (time course not to scale). Inset is the number of significant 2.0-fold changed genes in each cluster. LPS and Stx2 cause distinct alterations in patterns of gene expression.
Stx2 and LPS Alter Overlapping and Unique Groups of Genes
To discern the functions of the genes that were differentiallyexpressed in response to challenge, we subjected each clusterthat was described in Figure 8B to gene ontology analysis (Table 2).dChip categorizes genes on the basis of molecular function,biologic process, and cellular component using GeneOntology(http://www.geneontology.org/) terms and information from theNational Center for Biotechnology Information LocusLink databaseand calculates a P value for overrepresentation of that clusterin the gene set. The LPS-upregulated gene clusters containedsignificantly large alterations in genes that are involved inthe following biologic processes: Inflammatory response, immuneresponse, and regulation of transcription, with more minor changesof genes involved in apoptosis, cell differentiation, proliferation,and regulation of cell cycle. Molecular function analysis ofthe LPS-upregulated genes identified numerous cytokines, transcriptionfactors, cell surface ligands, and complement components. Incontrast, the gene biologic processes that are upregulated byStx2 did not include the inflammatory and immune responses,although Stx2 did cause an upregulation of genes that are involvedin cell proliferation, differentiation, and regulation of cellcycle and transcription. Molecular function analysis of theStx2-upregulated genes included some cytokines and transcriptionfactors but not any complement components. Downregulated byboth challenges were solute and macromolecule transporters thatare necessary for normal renal function. The Stx2 plus LPS challengealtered the combined functional groups from each of the individualchallenges (data not shown).
This Stx2 plus LPS mouse model of HUS recapitulates the humandisease in both its signs and symptoms, including the clinicaldiagnostic triad of renal failure, thrombocytopenia, and hemolyticanemia (1–3). Increased serum creatinine and BUN in thesemice demonstrate kidney dysfunction. Low platelet counts thatwere found in our model are the definition of thrombocytopenia.Reticulocytes and Howell-Jolly bodies that were found in theblood smears, an increase in the percentage of reticulocytes,and red discoloration of the serum that was caused by free hemoglobinare evidence of hemolytic anemia. Although no schistocytes wereseen in the blood smears, we found that, for unknown reasons,in published mouse models of hemolytic anemia, schistocytesare an uncommon finding (18–20). The mice also exhibitedthe neutrophilia and monocytosis that is found clinically (21),and fibrin and red cell staining of kidneys show thrombus formationin the microvasculature, as seen in patients with HUS (22).Furthermore, electron microscopy showed ultrastructural changesin the glomeruli that are consistent with HUS. As indicatedin Table 1, neither Stx2 nor LPS alone was able to mimic HUSin the mouse; however, injection of both agents together elicitedthe diagnostic triad as well as the other associated clinicalsigns. This reinforces evidence from human patients that bothStx and LPS are involved in the development of HUS (6,23).
Microarray expression analysis reinforces the mouse as a modelfor human disease and provides new insight into the detailsof HUS pathogenesis. Specific gene products that are known tobe increased in patients with HUS are upregulated in these mice,such as IL-6 and monocyte chemoattractant protein-1, and thesewere shown to be part of a general inflammatory response byevaluation of gene expression clusters. Furthermore, globalexpression analysis demonstrates that there are temporal wavesof transcriptional response for distinct types of genes. A closerlook at the specific differentially expressed genes allows formationof multiple testable hypotheses that should be useful to thescientific community, such as that the formation of the fibrin-richclots that are typical of HUS may be heavily affected by theupregulation of all three types of fibrinogen in the kidney(Supplemental Table 1). Although these gene clusters were formedin an unbiased manner by grouping similar expression patterns,it is noteworthy that the quoted classifications of these clustersdo not necessarily describe the complete functional potentialof these genes. Nevertheless, the classifications cited in Table 2do provide insight into what a transcript or group of transcriptsdoes. It also is important to note that the numbers of genesthat are altered by each challenge does not reflect the importanceof those genes or of the challenge. For instance, even thoughStx2 alters fewer transcripts than LPS in this model (Figure 7),much of the HUS pathology in Table 1 can be attributed to Stx2.In addition, this leads to the hypothesis that pharmacologicor other alteration of relatively few gene products may be ableto alter significantly the course of this disease. Overall,the gene expression analysis should serve as a starting pointfor numerous avenues of future research.
This model may be complicated by dehydration of the mice, aconsequence that does not occur in the human disease (2). Althoughpatients with HUS normally have decreased hematocrit, mice thatreceive Stx2 plus LPS display moderately increased hematocritand hemoglobin. In the mouse, this likely is hemoconcentrationas a result of dehydration, as suggested by the weight lossthat is sustained after injection (Figure 1). We contend thatall of our significant findings are evidence of HUS as causedby Stx2 plus LPS and are distinct from the vascular volume depletion.Specifically, even though the mice that are challenged withLPS alone lose as much initial weight as the Stx2 plus LPS mice,they exhibit no increased serum creatinine and no signs of hemolyticanemia (Table 1). Furthermore, the mice that are given Stx2alone develop increased serum creatinine and reticulocytosisat 4 to 12 h after injection but do not lose weight until 48h after injection. Because LPS induces early weight loss withouta rise in creatinine and Stx2 causes a rise in creatinine andreticulocytosis 36 h before weight loss, we conclude that thedehydration that results from combinatorial challenge with Stx2and LPS is distinct from the signs of HUS.
On the basis of our model, we propose a time course of diseaseprogression in the mouse. Two hours after Stx2 plus LPS injection,platelet levels rise in the kidney followed by peripheral thrombocytopeniathat persists throughout the time course. After the rise inrenal platelets, RBC infiltrate the kidney, form clumps, andcongest the glomeruli and vasculature of the kidney. Fibrindeposition follows RBC infiltration and leads to thrombus formationin the renal microvasculature. This thrombosis likely causesglomerular filtration failure, uremia, and eventual death ofthe mouse. During this time course, we also ascertained theprogression of gene expression in the affected kidney tissue.Inflammatory signaling that is induced primarily by LPS andcellular damage and repair pathways that are induced by bothStx2 and LPS are activated in these model HUS kidneys. Furthermore,the downregulation of genes that are necessary for normal renalfunction is evidence for activation of a de-differentiationand repair pathway that has been described in response to otherrenal insults (24). Modulation of upregulated genes that areinvolved in cell proliferation, apoptosis, the cell cycle, andrepair could be able to alter this disease course. This preciseestablishment of the physiologic and molecular progression ofHUS in the mouse model will allow identification of novel therapeuticstrategies to treat this disease.
Although other animal models have been reported, this mousemodel of HUS is the most economical, practical, and completemodel of HUS so far described. Although the baboon (25) andcanine (26) models of HUS mimic the human disease, these largeanimal models are expensive and impractical for the common researcher.Other rodent models of HUS have been described (27–32),but none has completely investigated the full pathophysiologyof the disease. Mouse models using oral bacterial inoculationrequire antibiotic pretreatment and do not result in a 100%rate of infection (33). Hence, these models can be inefficientand impractical. In addition, HUS is a toxemia and not a bacteremia;therefore, a model does not require live bacteria (2). In summary,we have shown that concurrent intraperitoneal injection of bothStx2 and LPS is sufficient to reproduce HUS in the C57BL/6 mouse.The data indicate this and offer a reproducible model with whichto study HUS and identify potential therapeutic targets andfor testing of new therapies.
Acknowledgments
This research was supported by funding from US Public HealthService grants AI24431, AI054782, and AI057168 (T.G.O.).
We thank Yongde Bao and the UVa Biomolecular Research Core Facilityfor help with microarray analysis, Regina Seaner for help withRNA isolation, and Kenneth Tung for interpretation of electronmicrographs.
Footnotes
Published online ahead of print. Publication date availableat www.jasn.org.
T.R.K. and M.A.P. contributed equally to this work.
Cleary TG: The role of Shiga-toxin-producing Escherichia coli in hemorrhagic colitis and hemolytic uremic syndrome.
Semin Pediatr Infect Dis 15
: 260
–265, 2004[CrossRef][Medline]
Banatvala N, Griffin PM, Greene KD, Barrett TJ, Bibb WF, Green JH, Wells JG: The United States National Prospective Hemolytic Uremic Syndrome Study: Microbiologic, serologic, clinical, and epidemiologic findings.
J Infect Dis 183
: 1063
–1070, 2001[CrossRef][Medline]
Melton-Celsa AR, O’Brien AD: Shiga toxins of Shigella dysenteriae and Escherichia coli. In:
Handbook of Experimental Pharmacology, edited by Aktories K, Just I, Freiburg, Springer, 2000
, pp 385
–406
Perera LP, Marques LR, O’Brien AD: Isolation and characterization of monoclonal antibodies to Shiga-like toxin II of enterohemorrhagic Escherichia coli and use of the monoclonal antibodies in a colony enzyme-linked immunosorbent assay.
J Clin Microbiol 26
: 2127
–2131, 1988[Abstract/Free Full Text]
O’Brien AD, Tesh VL, Donohue-Rolfe A, Jackson MP, Olsnes S, Sandvig K, Lindberg AA, Keusch GT: Shiga toxin: Biochemistry, genetics, mode of action, and role in pathogenesis.
Curr Top Microbiol Immunol 180
: 65
–94, 1992[Medline]
Ostroff SM, Tarr PI, Neill MA, Lewis JH, Hargrett-Bean N, Kobayashi JM: Toxin genotypes and plasmid profiles as determinants of systemic sequelae in Escherichia coli O157:H7 infections.
J Infect Dis 160
: 994
–998, 1989[Medline]
Pusey W, Edwards CA: A modification of the martius yellow-scarlet red-aniline blue method for fibrin.
Stain Technol 53
: 58
–59, 1978[Medline]
Artoni A, Li J, Mitchell B, Ruan J, Takagi J, Springer TA, French DL, Coller BS: Integrin beta3 regions controlling binding of murine mAb 7E3: Implications for the mechanism of integrin alphaIIbbeta3 activation.
Proc Natl Acad Sci U S A 101
: 13114
–13120, 2004[Abstract/Free Full Text]
Li C, Wong WH: Model-based analysis of oligonucleotide arrays: Expression index computation and outlier detection.
Proc Natl Acad Sci U S A 98
: 31
–36, 2001[Abstract/Free Full Text]
Tusher VG, Tibshirani R, Chu G: Significance analysis of microarrays applied to the ionizing radiation response.
Proc Natl Acad Sci U S A 98
: 5116
–5121, 2001[Abstract/Free Full Text]
Reich M, Ohm K, Angelo M, Tamayo P, Mesirov JP: GeneCluster 2.0: An advanced toolset for bioarray analysis.
Bioinformatics 20
: 1797
–1798, 2004[Abstract/Free Full Text]
Proulx F, Toledano B, Phan V, Clermont MJ, Mariscalco MM, Seidman EG: Circulating granulocyte colony-stimulating factor, C-X-C, and C-C chemokines in children with Escherichia coli O157:H7 associated hemolytic uremic syndrome.
Pediatr Res 52
: 928
–934, 2002[CrossRef][Medline]
Korcheva V, Wong J, Corless C, Iordanov M, Magun B: Administration of ricin induces a severe inflammatory response via nonredundant stimulation of ERK, JNK, and P38 MAPK and provides a mouse model of hemolytic uremic syndrome.
Am J Pathol 166
: 323
–339, 2005[Abstract/Free Full Text]
Bishop DF, Johansson A, Phelps R, Shady AA, Ramirez MCM, Yasuda M, Caro A, Desnick RJ: Uroporphyrinogen III synthase knock-in mice have the human congenital erythropoietic porphyria phenotype, including the characteristic light-induced cutaneous lesions.
Am J Hum Genet 78
: 645
–658, 2006[CrossRef][Medline]
Peters LL, Lane PW, Andersen SG, Gwynn B, Barker JE, Beutler E: Downeast anemia (dea), a new mouse model of severe nonspherocytic hemolytic anemia caused by hexokinase (HKI) deficiency.
Blood Cells Mol Dis 27
: 850
–860, 2001[CrossRef][Medline]
Pavia AT, Nichols CR, Green DP, Tauxe RV, Mottice S, Greene KD, Wells JG, Siegler RL, Brewer ED, Hannon D, et al.: Hemolytic-uremic syndrome during an outbreak of Escherichia coli O157:H7 infections in institutions for mentally retarded persons: Clinical and epidemiologic observations.
J Pediatr 116
: 544
–551, 1990[CrossRef][Medline]
Hosler GA, Cusumano AM, Hutchins GM: Thrombotic thrombocytopenic purpura and hemolytic uremic syndrome are distinct pathologic entities. A review of 56 autopsy cases.
Arch Pathol Lab Med 127
: 834
–839, 2003[Medline]
Chaisri U, Nagata M, Kurazono H, Horie H, Tongtawe P, Hayashi H, Watanabe T, Tapchaisri P, Chongsa-nguan M, Chaicumpa W: Localization of Shiga toxins of enterohaemorrhagic Escherichia coli in kidneys of paediatric and geriatric patients with fatal haemolytic uraemic syndrome.
Microb Pathog 31
: 59
–67, 2001[CrossRef][Medline]
Bonventre JV: Dedifferentiation and proliferation of surviving epithelial cells in acute renal failure.
J Am Soc Nephrol 14[Suppl 1]
: S55
–S61, 2003
Taylor FB Jr, Tesh VL, DeBault L, Li A, Chang AC, Kosanke SD, Pysher TJ, Siegler RL: Characterization of the baboon responses to Shiga-like toxin: Descriptive study of a new primate model of toxic responses to Stx-1.
Am J Pathol 154
: 1285
–1299, 1999[Abstract/Free Full Text]
Raife T, Friedman KD, Fenwick B: Lepirudin prevents lethal effects of Shiga toxin in a canine model.
Thromb Haemost 92
: 387
–393, 2004[Medline]
Sugatani J, Igarashi T, Munakata M, Komiyama Y, Takahashi H, Komiyama N, Maeda T, Takeda T, Miwa M: Activation of coagulation in C57BL/6 mice given verotoxin 2 (VT2) and the effect of co-administration of LPS with VT2.
Thromb Res 100
: 61
–72, 2000[Medline]
Ikeda M, Ito S, Honda M: Hemolytic uremic syndrome induced by lipopolysaccharide and Shiga-like toxin.
Pediatr Nephrol 19
: 485
–489, 2004[CrossRef][Medline]
Karpman D, Connell H, Svensson M, Scheutz F, Alm P, Svanborg C: The role of lipopolysaccharide and Shiga-like toxin in a mouse model of Escherichia coli O157:H7 infection.
J Infect Dis 175
: 611
–620, 1997[Medline]
Taylor CM, Williams JM, Lote CJ, Howie AJ, Thewles A, Wood JA, Milford DV, Raafat F, Chant I, Rose PE: A laboratory model of toxin-induced hemolytic uremic syndrome.
Kidney Int 55
: 1367
–1374, 1999[CrossRef][Medline]
Rutjes NW, Binnington BA, Smith CR, Maloney MD, Lingwood CA: Differential tissue targeting and pathogenesis of verotoxins 1 and 2 in the mouse animal model.
Kidney Int 62
: 832
–845, 2002[CrossRef][Medline]
Palermo MS, Alves Rosa MF, Van Rooijen N, Isturiz MA: Depletion of liver and splenic macrophages reduces the lethality of Shiga toxin-2 in a mouse model.
Clin Exp Immunol 116
: 462
–467, 1999[CrossRef][Medline]
Wadolkowski EA, Sung LM, Burris JA, Samuel JE, O’Brien AD: Acute renal tubular necrosis and death of mice orally infected with Escherichia coli strains that produce Shiga-like toxin type II.
Infect Immun 58
: 3959
–3965, 1990[Abstract/Free Full Text]
This article has been cited by other articles:
C. C. Toledo, T. J. Rogers, M. Svensson, R. Tati, H. Fischer, C. Svanborg, and D. Karpman Shiga Toxin-Mediated Disease in MyD88-Deficient Mice Infected with Escherichia coli O157:H7
Am. J. Pathol.,
November 1, 2008;
173(5):
1428 - 1439.
[Abstract][Full Text][PDF]
K. A. D. Sauter, A. R. Melton-Celsa, K. Larkin, M. L. Troxell, A. D. O'Brien, and B. E. Magun Mouse Model of Hemolytic-Uremic Syndrome Caused by Endotoxin-Free Shiga Toxin 2 (Stx2) and Protection from Lethal Outcome by Anti-Stx2 Antibody
Infect. Immun.,
October 1, 2008;
76(10):
4469 - 4478.
[Abstract][Full Text][PDF]
C. Zanchi, C. Zoja, M. Morigi, F. Valsecchi, X. Y. Liu, D. Rottoli, M. Locatelli, S. Buelli, A. Pezzotta, P. Mapelli, et al. Fractalkine and CX3CR1 Mediate Leukocyte Capture by Endothelium in Response to Shiga Toxin
J. Immunol.,
July 15, 2008;
181(2):
1460 - 1469.
[Abstract][Full Text][PDF]
K. A. Eaton, D. I. Friedman, G. J. Francis, J. S. Tyler, V. B. Young, J. Haeger, G. Abu-Ali, and T. S. Whittam Pathogenesis of Renal Disease Due to Enterohemorrhagic Escherichia coli in Germ-Free Mice
Infect. Immun.,
July 1, 2008;
76(7):
3054 - 3063.
[Abstract][Full Text][PDF]
T. Shea-Donohue, K. Thomas, M. J. Cody, Aiping Zhao, L. J. DeTolla, K. M. Kopydlowski, M. Fukata, S. A. Lira, and S. N. Vogel Mice deficient in the CXCR2 ligand, CXCL1 (KC/GRO-{alpha}), exhibit increased susceptibility to dextran sodium sulfate (DSS)-induced colitis
Innate Immunity,
April 1, 2008;
14(2):
117 - 124.
[Abstract][PDF]
M. K. Stone, G. L. Kolling, M. H. Lindner, and T. G. Obrig p38 Mitogen-Activated Protein Kinase Mediates Lipopolysaccharide and Tumor Necrosis Factor Alpha Induction of Shiga Toxin 2 Sensitivity in Human Umbilical Vein Endothelial Cells
Infect. Immun.,
March 1, 2008;
76(3):
1115 - 1121.
[Abstract][Full Text][PDF]
T. R. Keepers, L. K. Gross, and T. G. Obrig Monocyte Chemoattractant Protein 1, Macrophage Inflammatory Protein 1{alpha}, and RANTES Recruit Macrophages to the Kidney in a Mouse Model of Hemolytic-Uremic Syndrome
Infect. Immun.,
March 1, 2007;
75(3):
1229 - 1236.
[Abstract][Full Text][PDF]
L. Bao, Y. Wang, A. Chang, A. W. Minto, J. Zhou, H. Kang, M. Haas, and R. J. Quigg Unrestricted C3 Activation Occurs in Crry-Deficient Kidneys and Rapidly Leads to Chronic Renal Failure
J. Am. Soc. Nephrol.,
March 1, 2007;
18(3):
811 - 822.
[Abstract][Full Text][PDF]
J. K. Roche, T. R. Keepers, L. K. Gross, R. M. Seaner, and T. G. Obrig CXCL1/KC and CXCL2/MIP-2 Are Critical Effectors and Potential Targets for Therapy of Escherichia coli O157:H7-Associated Renal Inflammation
Am. J. Pathol.,
February 1, 2007;
170(2):
526 - 537.
[Abstract][Full Text][PDF]
Top
Abstract
Introduction
that contained the Stx2-producing pJES120 plasmid (7). Briefly, Stx2 was purified using 11E10 antibody (8) that was immobilized using an AminoLink Plus Kit (Pierce Biotechnology, Rockford, IL) according to the manufacturer’s instructions. Endotoxin was removed using De-toxi-Gel (Pierce Biotechnology) as per the manufacturer’s instructions. Stx2 purity was assessed by SDS-PAGE and determined to be endotoxin-free, and activity was measured in a Vero cell cytotoxicity assay. Stx2 was chosen because it is far more frequently associated with HUS clinical isolates than Stx1 (9,10). 
3 antibody that cross-reacts with the mouse protein (Santa Cruz Biotechnology, Santa Cruz, CA) (12). Primary antibody was detected using an avidin/biotin horseradish peroxidase system (Vector Laboratories, Burlingame, CA) and diaminobenzidine. Sections then were counterstained in hematoxylin, dehydrated, and mounted. 
). *P < 0.01 significantly increased compared with saline. (B) Total lymphocyte counts (•) and percentage of lymphocytes of total WBC (
