Role of Toll-Like Receptor 4 in Endotoxin-Induced AcuteRenal Failure1
Patrick N. Cunningham,2 Ying Wang, Rongqing Guo, Gang He, and Richard J. Quigg
Toll-like receptor 4 (TLR4) is present on monocytes and other cell types, and mediates inflammatory events such as the releaseof TNF after exposure to LPS. C3H/HeJ mice are resistant to LPS-induced mortality, due to a naturally occurring mutation inTLR4. We therefore hypothesized that LPS-induced acute renal failure (ARF) requires systemic TNF release triggered by LPSacting on extrarenal TLR4. We injected C3H/HeJ mice and C3H/HeOuJ controls with 0.25 mg of LPS, and sacrificed them 6 hlater for analysis of blood urea nitrogen (BUN) and kidney tissue (n � 8 per group). In contrast to C3H/HeOuJ controls, C3H/HeJmice were completely resistant to LPS-induced ARF (6-h BUN of 32.3 � 1.1 vs 61.7 � 5.6 mg/dl). C3H/HeJ mice released no TNFinto the circulation at 2 h (0.00 vs 1.24 � 0.16 ng/ml), had less renal neutrophil infiltration (6.4 � 1.0 vs 11.4 � 1.3 neutrophilsper high power field), and less renal apoptosis, as assessed by DNA laddering. Transplant studies showed that C3H/HeJ recipientsof wild-type kidneys (n � 9) were protected from LPS-induced ARF, while wild-type recipients of C3H/HeJ kidneys (n � 11)developed severe LPS-induced ARF (24-h BUN 44.0 � 4.1 vs 112.1 � 20.0 mg/dl). These experiments support our hypothesis thatLPS acts on extrarenal TLR4, thereby leading to systemic TNF release and subsequent ARF. Renal neutrophil infiltration andrenal cell apoptosis are potential mechanisms by which endotoxemia leads to functional ARF. The Journal of Immunology, 2004,172: 2629–2635.
A cute renal failure (ARF)3 occurs in up to 5% of hospitaladmissions, and is a leading cause of morbidity and mor-tality (1). A common cause of ARF is sepsis, which
results from overwhelming infection (2, 3). Although a variety ofbacterial products may cause the diffuse inflammatory responseseen in sepsis, one of the most important is endotoxin (LPS), acomponent of the cell wall of Gram-negative bacteria. Injection ofLPS into animals reproduces many of the manifestations of sepsis,including ARF. Cytokines such as TNF are thought to be key earlymediators of this syndrome. In previous work, we have shown thatmice deficient in TNFR1 are resistant to LPS-induced ARF, andthat TNFR1 mediates LPS-induced ARF within the kidney (4).
In recent years, much has been learned about the immediateevents following LPS exposure. Various bacterial products, in-cluding LPS, signal through a family of transmembrane proteinsknown as the Toll-like receptors (TLRs) (5). The importance of theTLRs in facilitating innate immunity is underscored by their highlyconserved presence in organisms ranging from Drosophila to hu-mans. TLR4 has been found to be the primary molecule throughwhich LPS activates cells, leading to the rapid release of cytokinessuch as TNF and IL-1. In addition, TLR4 binds an endogenousligand, heat shock protein 60, and this interaction may mediateinflammation seen in response to cellular damage (6). TLR4 is
most highly expressed in leukocytes, but is also present in a varietyof organs, including the kidney (7, 8). The key role for TLR4 invivo was illustrated when the C3H/HeJ strain of mice, long knownto be resistant to LPS-induced mortality, was found to be homozy-gous for a naturally occurring single base pair mutation in theTLR4 gene, leading to a complete absence of functional protein(9). When bone marrow-derived cells from these mice were trans-ferred to irradiated wild-type control mice, relative protection againstLPS-induced mortality resulted, implying that leukocyte TLR4 sig-naling is necessary to cause the systemic effects of LPS (10).
Nevertheless, TLR4-independent pathways of LPS action havealso been described (11, 12), including possible signaling throughTLR2 (13, 14). Additionally, in vitro work has documented effectsof LPS on renal tubular and mesangial cells (15–18). We thereforeundertook the following study to examine the role and site of ac-tion of TLR4 in LPS-induced ARF.
Materials and MethodsAnimal studies
C3H/HeJ and C3H/HeOuJ (wild-type control) mice were obtained fromThe Jackson Laboratory (Bar Harbor, ME). In all experiments, 8- to 10-wk-old male mice were injected i.p. with 0.25 mg of Escherichia coli LPS(Sigma-Aldrich, St. Louis, MO). This dose of LPS is higher than the 0.15mg dose used in previous studies, as preliminary work showed that theC3H/HeOuJ strain is somewhat less sensitive to LPS than the C57BL/6strain. Blood was obtained via retroorbital bleeding at the time of injectionand at various later time points. Blood urea nitrogen (BUN) concentrationswere used to determine renal function and were measured with a BeckmanCX5CE autoanalyzer (Beckman Coulter, Fullerton, CA). Mice were sac-rificed at 6, 24, or 48 h, with collection of blood as above and harvest ofkidney tissue. The above procedures were done under anesthesia using acontinuous inhalational isoflurane/oxygen mixture. The animals weremaintained and experiments were performed in accordance with the guide-lines set by the University of Chicago Institutional Animal Care and UseCommittee.
Pathology
For routine histologic analysis, kidneys were sectioned coronally, fixed inmethyl Carnoy’s solution at 4°C for 48 h, embedded in paraffin, and stained
Section of Nephrology, Department of Medicine, University of Chicago, Chicago, IL60637
Received for publication January 3, 2003. Accepted for publication December12, 2003.
The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by National Institutes of Health Grants R01DK41873,R01DK55357, and K08DK61375.2 Address correspondence and reprint requests to Dr. Patrick N. Cunningham, Sectionof Nephrology MC 5100, Room S511, University of Chicago, 5841 South MarylandAvenue, Chicago, IL 60637. E-mail address: [email protected] Abbreviations used in this paper: ARF, acute renal failure; BUN, blood urea nitro-gen; LM, ligase mediated; TLR, Toll-like receptor.
with periodic acid-Schiff base. Sections were scored for tubular injury in ablinded fashion, as previously described (4).
For immunohistochemistry, 4-�m kidney cryostat sections were fixedwith ether/ethanol, incubated with 0.3% H2O2 for 30 min, and blocked withdilute horse serum. Sections were stained for neutrophils by sequentialincubation with rat anti-mouse neutrophil (anti-Gr-1; BD PharMingen, SanDiego, CA) at 1/60 for 30 min, followed by HRP-conjugated rabbit anti-ratIgG (Sigma-Aldrich) at 1/60 for 30 min, and diaminobenzidine reagent(Vector Laboratories, Burlingame, CA) for 10 min. A blinded observercounted the number of neutrophils per high power field and recorded theaverage of 10 fields for each sample.
TUNEL staining
Tissue sections were stained for apoptotic nuclei via the TUNEL technique,using a commercially available kit (Trevigen, Gaithersburg, MD), accord-ing to manufacturer instructions. In brief, kidney cryosections were cut into10-�M sections and mounted on charged slides. Slides were dried over-night, rehydrated through graded alcohols to PBS, and fixed in 10% For-malin for 10 min at room temperature. This was followed by proteinase Ktreatment (20 �g/ml) for 15 min at 37°C and incubation in 3% H2O2/methanol for 5 min. Specimens were incubated with TdT/Br-dNTP mixtureat 37°C for 60 min, followed by anti-5-bromo-2�-deoxyuridine at 1/150 for60 min at 37°C. This was followed by incubation in streptavidin-HRP at1/500 for 30 min, followed by detection with TACS blue label for 5 min.Sections were counterstained with nuclear fast red for 10 min, washed, andmounted under coverslips.
Ligase-mediated (LM) PCR
At time of sacrifice, kidneys were cut into thirds and snap frozen in liquidnitrogen. From this, genomic DNA was purified by means of the DNeasyDNA purification system (Qiagen, Valencia, CA), according to the man-ufacturer’s instructions, and was quantitated spectophotometrically. Theextent of DNA laddering was amplified and detected via LM-PCR using acommercially available kit (Clontech Laboratories, Palo Alto, CA), accord-ing to the manufacturer’s instructions, as follows. DNA isolated from eachanimal was incubated with supplied primer targets and T4 DNA ligase for18 h at 16°C. A total of 20 �g of this ligated DNA was then used as thetemplate for PCR, using supplied primers and Advantage DNA polymerase(Clontech Laboratories) for 21 cycles at 94°C (1 min)/72°C (3 min). Thereaction product for each animal was electrophoresed through a 1.3% aga-rose gel, and ethidium bromide-stained bands were detected with UV lightillumination. To ensure an equivalent amount of genomic DNA templatewas used for each animal, standard PCR for the gene En-2 was performed,using primers contained in the above kit.
TNF ELISA
TNF levels were determined from sera obtained at baseline and 2 and 6 hafter LPS administration using a commercially available ELISA kit formouse TNF (R&D Systems, Minneapolis, MN), according to the manu-facturer’s instructions. The limit of sensitivity of this assay was 10 pg/ml.
Real-time PCR
A portion of whole kidney obtained at sacrifice from each mouse wasfrozen in liquid nitrogen. This was later placed in TRIzol reagent (LifeTechnologies, Grand Island, NY), from which total RNA was purified ac-cording to the manufacturer’s instructions. To remove all traces of genomicDNA, samples were then treated with RNase-free RQ1 DNase (Promega,Madison, WI; 1 U/4 �g RNA) in 10 �l of reaction buffer (final concen-tration 40 mM of Tris-HCl, 10 mM of MgSO4, 1 mM of CaCl2, pH 8.0),at 37°C for 30 min. This was followed by addition of 1 �l of 20 mMEGTA, pH 8.0, to stop the reaction, and incubation at 65°C for 10 min toinactivate the DNase. cDNA was generated from RNA using random hex-amers as primers with the SuperScript first-strand synthesis kit (Life Tech-nologies), according to the manufacturer’s instructions, and diluted 5-foldbefore analysis.
Real-time PCR was performed using the Prism 7700 reactor and theSybrGreen intercalating dye method with HotStar DNA polymerase (Ap-plied Biosystems, Foster City, CA). Each reaction was conducted in a totalvolume of 50 �l with primers at 200 nM, 1 mM of dNTPs, 3 mM of MgCl2,and 10 �l of sample or standard cDNA. PCR was conducted with a hot startat 95°C (5 min), followed by 45 cycles at 95°C (15 s)/60°C (30 s). For eachsample, the number of cycles required to generate a given threshold signal(Ct) was recorded. Using a standard curve generated from serial dilutionsof kidney cDNA, the ratio of ICAM-1 expression relative to GAPDH ex-pression was calculated for each experimental animal, and normalized rel-ative to an average of ratios from the control group. Measurements of TNF
mRNA expression were performed in an analogous fashion, except that theannealing temperature was 58°C. Products of each reaction yielded a singleband when run on agarose gel, confirming specific amplification. Primerswere synthesized by Integrated DNA Technologies (Coralville, IA), withsequences as follows: GAPDH forward primer, 5�-GGC AAA TTC AACGGC ACA GT-3�; GAPDH reverse primer, 5�-AGA TGG TGA TGG GCTTCC C-3�; TNF forward primer, 5�-CCG ATG GGT TGT ACC TTG TC-3�; TNF reverse primer, 5�-GTG GGT GAG GAG CAC GTA GT-3�;ICAM-1 forward primer, 5�-CGC AAG TCC AAT TCA CAC TGA-3�;ICAM-1 reverse primer, 5�-CAG AGC GGC AGA GCA AAA G-3�.
Western blotting
Selected organs were isolated from each animal at sacrifice and snap frozenin liquid nitrogen. These were later homogenized in a buffer consisting of200 mM of NaCl, 10 mM of Tris-HCl (pH 7.0), 5 mM of EDTA, 10%glycerol, 1 mM of PMSF, 20 �M of pepstatin, 20 �M of leupeptin, and 0.1�M of aprotinin. The total protein concentration in each homogenate wasdetermined by means of the bicinchoninic acid method (Pierce, Dallas,TX). Samples (20 �g/lane) were electrophoresed through a 10% SDS-PAGE gel under nonreducing conditions. Proteins were transferred to anImmobilon-P nitrocellulose membrane (Millipore, Bedford, MA), andblocked in 5% milk/TBST overnight. This membrane was then incubatedwith goat anti-mouse TNF (R&D Systems) at 1/1000, followed by HRPrabbit anti-goat IgG (Sigma-Aldrich) at 1/2000, and activity was detectedusing the Supersignal West Pico chemiluminescent kit (Pierce). Three an-imals were randomly selected from each group for analysis.
Kidney transplantation
To define the level of action of the TLR4, single kidney transplants wereperformed between C3H/HeJ and wild-type mice, and their response toLPS injection was studied. To provide adequate controls, four groups ofmice were studied: C3H/HeJ to wild type, wild type to C3H/HeJ, C3H/HeJto C3H/HeJ, and wild type to wild type. Each individual mouse was usedonly as a recipient or a donor, at an age of 10 wk. Mice were anesthetizedwith 65 mg/kg i.p. pentobarbital. The left kidney of donor mice was per-fused via the renal artery with 0.3 ml of cold saline, and resected withartery, vein, and ureter attached. The kidney was stored at 4°C until timeof anastomosis. Next, the recipient underwent midline abdominal incision,followed by suprarenal clamping of aorta and inferior vena cava. The donorkidney was placed in the right flank, and its artery and vein were attachedwith 10-0 nylon suture via side-to-end anastomoses with the recipient aortaand inferior vena cava. The bladder was punctured with a 21-gauge needle,and the ureter was sewn in place. Both native kidneys were then resected.Animals were allowed to recover 10 days after surgery before LPS injec-tion. Preliminary dosing experiments showed that mouse kidney transplantrecipients have a stable, mild elevation of BUN and creatinine, but aresomewhat more sensitive to LPS; thus, a dose of 0.125 mg was used inthese studies.
Statistics
Data were analyzed with Minitab software (State College, PA). Unlessnoted otherwise, data are given as mean � SEM. Groups were comparedby two-tailed t test, or ANOVA using the Dunn-Sidak correction for mul-tiple comparisons in cases in which more than two groups were compared.When BUN levels pre- and post-LPS were compared within a given groupof mice, a paired t test was used. When comparing the severity of ARFbetween two groups, the slope of BUN vs time was determined for eachindividual animal by least squares regression, and the individual slopes inthe two groups were compared with the Mann-Whitney rank sum test. A pvalue �0.05 was considered significant.
ResultsC3H/HeJ mice are resistant to LPS-induced ARF
To determine whether the protection against LPS-induced mortal-ity previously observed in C3H/HeJ mice extends to LPS-inducedARF, C3H/HeJ mice and wild-type controls were injected with asublethal dose of LPS, and BUN was measured at various timepoints up through 48 h. Mice of the C3H/HeOuJ strain, which arehomozygous for a fully functional TLR4, were used as the wild-type control group. Although wild-type controls developed severeARF manifested by an abrupt rise in BUN, C3H/HeJ mice showedabsolutely no evidence of ARF (24-h BUN of 131.6 � 24.8 mg/dlvs 28.8 � 1.8 mg/dl, p � 0.001, n � 4 per group; Fig. 1A). Inaddition, renal histology of wild-type controls at time of sacrifice
showed a moderate degree of tubular injury (Fig. 1D), while renalhistology in C3H/HeJ mice was completely normal (Fig. 1E).
Subsequent experiments focused on events occurring in the first6 h after LPS injection, during which ARF becomes established. Inboth wild-type and C3H/HeJ mice, serum levels of TNF were un-detectable at baseline. Two hours after LPS injection, there was aprofound increase in serum TNF levels in wild-type mice, whichwas completely absent in C3H/HeJ mice (1.24 � 0.16 ng/ml vs0.00 ng/ml, p � 0.001; Fig. 2). By 6 h, these levels had returnedto near baseline in wild-type mice, and remained undetectable in
C3H/HeJ mice (data not shown). In parallel with this extreme dif-ference in TNF release, all wild-type mice had developed ARF at6 h after LPS injection, while C3H/HeJ mice had no evidence ofARF (6-h BUN of 61.7 � 5.6 mg/dl in wild-type mice vs 32.3 �1.1 mg/dl in C3H/HeJ mice, n � 8 per group, p � 0.001). Thissupports our hypothesis that TLR4-mediated TNF release has aprimary pathogenic role in LPS-induced ARF.
Renal neutrophil infiltration
Our previous work has shown an influx of neutrophils into thekidney 48 h after LPS administration. Therefore, we hypothesizedthat C3H/HeJ mice would be resistant to LPS-induced neutrophilinfiltration. Six hours after LPS administration, a robust neutro-philic infiltration into renal parenchyma was observed in wild-typemice, illustrating that the neutrophilic infiltration occurred earlyand coincident with development of renal functional impairment.This infiltration was present diffusely throughout the renal cortex,especially prominent in venules, but also observed in peritubularareas and occasionally in glomeruli (Fig. 3, A–C). The extent ofrenal neutrophil infiltration in C3H/HeJ mice was significantly lessthan in wild-type controls (6.4 � 1.0 vs 11.4 � 1.3 neutrophils perhigh power field, p � 0.05), and the magnitude of the change frombaseline was also less (Fig. 3D). Interestingly, C3H/HeJ mice didexhibit an increase from 3.5 � 0.7 neutrophils per high power fieldat baseline to 6.4 � 1.0 neutrophils per high power field after LPS,although this did not reach statistical significance ( p � 0.23).
FIGURE 1. C3H/HeJ mice are resistant to LPS-induced ARF. A, In contrast to the acute rise in BUN seen in wild-type mice, BUN levels werecompletely unchanged in C3H/HeJ mice after LPS injection. n � 4 per group. Normal kidney histology is shown for wild-type and C3H/HeJ mice atbaseline in B and C, respectively. D, By 48 h after LPS injection, pathology in wild-type mice showed modest tubular injury, consisting of tubular dilationand flattening (asterisk) and degenerated tubules with small vacuoles (arrows). E, In C3H/HeJ mice, renal pathology after LPS injection was indistin-guishable from baseline. Magnification �200, periodic acid-Schiff stain.
FIGURE 2. Serum TNF release after LPS administration. Serum TNFlevels as determined by ELISA are shown. At baseline, the amount of TNFin both C3H/HeJ and wild-type mice was undetectable. Two hours afterLPS injection, there was a profound increase in serum TNF in wild-typemice, but not in C3H/HeJ mice. n � 8 per group.
The influx of inflammatory leukocytes into a variety of tissuesafter endotoxin administration has been attributed to LPS-inducedup-regulation of various adhesion molecules and chemotactic che-mokines (19, 20). One of the most important molecules in thisadhesion process is ICAM-1 (21). To investigate the role of TLR4in ICAM-1-mediated neutrophil infiltration, whole kidney taken atsacrifice 6 h after LPS administration was analyzed for ICAM-1mRNA by means of real-time PCR. This showed a profound in-crease in renal ICAM-1 transcription in wild-type mice after LPSadministration (20.5 � 2.9-fold increase, p � 0.001; Fig. 4). Thelevel of ICAM-1 mRNA in C3HeJ mice after LPS was much less,
although was definitely elevated compared with baseline (2.9 �0.4-fold increase, p � 0.05).
Renal apoptosis
In previous studies, we have also shown that LPS administrationcauses renal cell apoptosis in vivo, and that the extent of renalapoptosis strongly correlates with the degree of renal failure (4). Inthis study, we again observed minimal apoptosis at baseline onTUNEL-stained kidney cryosections, in contrast to a clear increasein apoptotic bodies seen in wild-type mice 6 h after LPS, early inthe course of LPS-induced ARF (Fig. 5, A and B). Apoptotic nucleiwere primarily seen in both tubular and peritubular locations; thelatter may represent apoptosis of peritubular capillary cells, con-sistent with what has been observed in other vascular beds afterLPS administration (22). We further quantitated renal apoptosis inC3H/HeJ and wild-type mice by means of LM-PCR, which am-plifies the characteristic DNA laddering seen in apoptotic cells. Incontrast to the scant amount of apoptosis seen at baseline in bothC3H/HeJ and wild-type mice (Fig. 5C), which probably reflectsnormal cell turnover, apoptosis was markedly increased in wild-type mice after LPS (Fig. 5D). This stands in contrast to the sig-nificantly lesser amount of apoptosis seen in C3H/HeJ kidney fol-lowing LPS (Fig. 5D).
Renal transplant studies
When put in the context of our previous work, the above datasupport the hypothesis that LPS acts through TLR4 to trigger sys-temic release of TNF, which reaches the kidney through the cir-culation and acts through renal TNFR1 to cause ARF (4). How-ever, because TLR4 is expressed in the kidney, it is conceivablethat LPS could be working through renal TLR4 as well. It has beendemonstrated that TNF is synthesized in the kidney following LPS
FIGURE 3. Renal neutrophil infiltration followingLPS administration. A, At baseline, few neutrophilswere present in renal tissue. After LPS administrationto wild-type mice, abundant neutrophils were founddiffusely throughout the renal cortex, especially in re-nal venules (B) and peritubular capillaries (C, arrows).A similar distribution of neutrophils was observed inC3H/HeJ mice, although this was quantitatively less.D, Quantitation of neutrophil infiltration in the fourgroups is shown. �, p � 0.05 vs wild-type baseline,p � 0.23 vs C3H/HeJ baseline; ��, p � 0.05 vs allother groups. Magnification �200 in A and B, and�400 (high power field) in C.
FIGURE 4. Renal ICAM-1 expression after LPS administration. Real-time PCR was used to determine ICAM-1 mRNA levels in whole kidney6 h after LPS administration. Each data point represents an individualanimal, and the dotted line represents the average baseline expression forwild-type mice. Although ICAM-1 expression was significantly increasedin both wild-type and C3H/HeJ mice after LPS injection compared withbaseline, this up-regulation was markedly greater in wild-type mice. �, p �0.01 vs baseline; ��, p � 0.01 vs all other groups.
administration (23). In support of this, real-time PCR showed aprofound increase in renal TNF mRNA 6 h after LPS administra-tion in wild-type mice (19.5 � 2.3-fold increase, p � 0.01), whichwas significantly less in C3H/HeJ mice (Fig. 6A). Western blottingsubstantiated this at the protein level, showing that while the pri-mary sites of TNF protein production 6 h after LPS administrationare the liver and small intestine, there is also a significant amountthat is synthesized in the kidney (Fig. 6B).
To determine the site at which TLR4 mediates LPS-inducedARF, single kidney transplants were performed between C3H/HeJand wild-type mice, and their response to LPS injection was stud-ied. C3H/HeJ kidneys in wild-type recipients still developed se-vere ARF after LPS injection (n � 11), while wild-type kidneys inC3H/HeJ recipients were largely resistant to LPS-induced ARF( p � 0.01, n � 9) (Fig. 7). Additionally, tubular injury scoresshowed significantly greater pathologic injury in C3H/HeJ kid-neys in wild-type recipients vs wild-type kidneys in C3H/HeJrecipients (tubular injury scores 3.5 � 0.4 vs 1.7 � 0.2, p �0.01). This proves that the TLR4 that mediates LPS-inducedARF is located in a primarily extrarenal location. However, it isnotable that even in C3H/HeJ recipients, wild-type kidneys hada significant increase in BUN after LPS administration (41.4 �
2.5 mg/dl at baseline, vs 61.0 � 8.9 mg/dl 12 h after LPS, p �0.05 by paired t test) before declining to baseline levels, unlikethe C3H/HeJ3C3H/HeJ control group, which had no increasein BUN above baseline at any time point.
FIGURE 7. TLR4 mediates LPS-induced ARF primarily from outsidethe kidney. Although C3H/HeJ kidneys implanted into wild-type recipients(C3H/HeJ3WT, n � 11) sustained severe ARF after LPS injection, wild-type kidneys implanted into C3H/HeJ recipients (WT3C3H/HeJ, n � 9)were protected. �, p � 0.001 for wild-type recipients vs C3H/HeJ recipi-ents. Nevertheless, even wild-type kidneys in C3H/HeJ recipients had atransient rise in BUN at 12 h after LPS in comparison with baseline (p �0.05). Syngeneic transplants of WT3WT (n � 7) and C3H/HeJ3C3H/HeJ (n � 6) mice are depicted as controls.
FIGURE 5. Renal apoptosis after LPS administration. As depicted in A,TUNEL staining of wild-type kidney cryosections revealed only rare ap-optotic nuclei at baseline. In contrast, LPS injection induced scattered ap-optosis (arrows), as depicted in B. Renal apoptosis was further quantitatedby LM-PCR, shown in C and D. Each lane represents whole kidney takenfrom an individual animal 6 h after LPS injection. C, There is little apo-ptosis at baseline in either wild-type or C3H/HeJ kidney, in contrast towild-type kidney following LPS injection (far right lane). As shown in D,LPS-induced apoptosis was greater in wild-type mice as compared withC3H/HeJ mice.
FIGURE 6. Renal TNF synthesis. A, Real-time PCR was used to deter-mine TNF mRNA levels in whole kidney 6 h after LPS administration.Each data point represents an individual animal, and the dotted line rep-resents the average baseline expression of wild-type kidney. Although TNFexpression was significantly increased in both wild-type and C3H/HeJmice after LPS injection compared with baseline, this up-regulation wasmarkedly greater in wild-type mice. �, p � 0.01 vs baseline; ��, p � 0.01vs all other groups. B, Western blots of various organs 6 h after LPSinjection. After LPS administration, the majority of TNF was induced inliver and small intestine, but considerable renal synthesis was also seen inwild-type mice.
DiscussionThe above data build upon our previous work and strengthen ourunderstanding of how LPS causes ARF. In this paradigm, admin-istered LPS acts through extrarenal TLR4, leading to the rapidrelease of TNF into the circulation within the first several hours. Inturn, this circulating TNF acts through renal TNFR1 to cause ARF,through a variety of mechanisms that may involve renal neutrophilinfiltration and renal apoptosis (4). In addition, given that LPSinjection causes transient ARF in wild-type kidneys transplantedinto C3H/HeJ recipients, there appears to be a novel, albeit lesserrole for intrarenal TLR4 in this LPS-induced ARF. Thus, LPS alsoacts in part through renal TLR4, presumably leading to local TNFsynthesis, as shown above, followed by paracrine action on renalTNFR1 and subsequent ARF.
The above data are consistent with our previous study in sug-gesting an important role for the mechanisms of renal apoptosisand neutrophil infiltration. Previously, we focused on each of thesetwo processes 48 h after LPS exposure, while in this study weshow that both renal apoptosis and neutrophil infiltration are wellunderway 6 h after LPS injection, during the period when ARF isinitiated. The fact that these processes were found to correlate withLPS-induced ARF in both the C3H/HeJ and C57BL/6 mousestrains also strengthens their likely importance.
Apoptosis of kidney cells is a mechanism of cellular death thatis difficult to discern with routine histology, but that may contrib-ute importantly to loss of renal function. Because apoptotic cellsare rapidly and efficiently cleared by neighboring cells, a smallamount of apoptosis may add up over the course of time to asignificant amount of cellular loss. TNF administration to a varietyof cell types, including renal epithelial cells, has been shown toinitiate apoptosis (24, 25). TNF may trigger apoptosis in a recep-tor-mediated fashion, via TNFR1 activation of initiator caspase-8and caspase-10. Additionally, activation of TNFR1 may cause ap-optosis indirectly through various inflammatory events, such asNO release, local hypoxia, and reactive oxygen species released byinvading neutrophils, all of which may bring about apoptosisthrough the mitochondrial pathway (26–28).
Infiltration of various tissues such as liver, lung, and kidney byneutrophils and other inflammatory leukocytes has been well dem-onstrated to occur after LPS administration (21, 29, 30). LPS, aswell as cytokines such as TNF, are known to up-regulate chemo-tactic chemokines and adhesion molecules, such as ICAM-1,which promote leukocyte adhesion and migration (19–21). Oncein the tissue, these inflammatory cells may cause tissue injury viarelease of injurious proteases and reactive oxygen species (31).Additionally, there is evidence that neutrophils may mediatechanges in local vascular tone by signaling through ICAM-1 or byreleasing reactive oxygen species (32). In this study, we demon-strate invasion of the renal parenchyma by neutrophils soon afterLPS administration, in parallel with a profound increase inICAM-1 expression. Additionally, LPS and cytokines have beenshown to directly activate neutrophils, priming them for adhesive-ness and activity (33).
Interestingly, the LPS-resistant C3H/HeJ mice also had in-creased neutrophilic infiltration into the kidney after LPS. It waspreviously reported that a closely related strain of TLR4-deficientmice showed an exaggerated influx of neutrophils into the perito-neal cavity after LPS injection, demonstrating that LPS triggersneutrophil chemotaxis in part via a TLR4-independent pathway(14). The mechanism for this neutrophil chemotaxis in TLR4-de-ficient mice is not known. However, the fact that we observedrenal neutrophil infiltration without any elevation in BUN in C3H/HeJ mice shows that the mere presence of this number of neutro-
phils is not sufficient to cause LPS-induced ARF. It is likely thatadditional TLR4-dependent events, such as the stimulation of theneutrophil respiratory burst or the involvement of key adhesionmolecules or chemokines, are also necessary for neutrophils to causetissue damage. Alternatively, renal neutrophil infiltration may simplybe an associated finding without a direct pathogenic role.
Given the fact that mice lacking a functional TLR4 are fullyresistant to LPS-induced mortality as well as LPS-induced ARF,TLR4 clearly plays a primary role in the response to LPS in vivo.However, it is noteworthy that renal TNF and ICAM-1 expressionwere significantly increased after LPS administration in C3H/HeJmice, albeit much less so than in wild-type mice. This implies thatthere is some response to LPS through TLR4-independent path-ways, as has been described elsewhere (11, 12). Although earlystudies seemed to clearly show that LPS also interacts and signalsthrough TLR2 (13, 34), a subsequent study ascribes the effects ofTLR2 to contamination of commercial LPS preparations withother inflammatory bacterial products (35). Although we cannotrule out that a component of the increases in TNF and ICAM-1expression that we observed was due to LPS impurities, it seemsunlikely that all the changes we observed were entirely due to thisartifact.
In conclusion, the above experiments show that extrarenal TLR4is crucial in mediating LPS-induced ARF, via systemic cytokinerelease and subsequent intrarenal events such as renal cell apopto-sis and renal neutrophil infiltration. The distal events occurring inthe kidney are likely to be complex and interrelated, and the extentto which apoptosis and neutrophil infiltration individually contrib-ute to LPS-induced ARF, as well as other possible effects of TNFon the kidney, is the subject of ongoing work. It is hoped that abetter understanding of this process will lead to therapies that caneffectively prevent or reverse the ARF associated with sepsis.
References1. Gesualdo, L., G. Grandaliano, L. Mascia, G. Pertosa, and F. P. Schena. 1999.
Acute renal failure in critically ill patients. Intensive Care Med. 25:1188.2. Groenveld, A. B., D. D. Tran, J. van der Meulen, J. J. Nauta, and L. G. Thijs.
1991. Acute renal failure in the intensive care unit: predisposing, complicatingfactors and outcome. Nephron 59:602.
3. Neveu, H., D. Kleinknecht, F. Brivet, P. Loirat, and P. Landais. 1996. Prognosticfactors in acute renal failure due to sepsis: results of a prospective multicentrestudy. Nephrol. Dial. Transplant. 11:293.
4. Cunningham, P. N., H. M. Dyanov, P. Park, J. Wang, K. A. Newell, andR. J. Quigg. 2002. Acute renal failure in endotoxemia is caused by TNF actingdirectly on TNF receptor-1 in kidney. J. Immunol. 168:5817.
5. Medzhitov, R. 2001. Toll-like receptors and innate immunity. Nat. Rev. Immunol.1:135.
6. Ohashi, K., V. Burkart, S. Flohe, and H. Kolb. 2000. Heat shock protein 60 is aputative endogenous ligand of the Toll-like receptor-4 complex. J. Immunol.164:558.
7. Medzhitov, R., P. Preston-Hurlburt, and C. A. Janeway, Jr. 1997. A human ho-mologue of the Drosophila Toll protein signals activation of adaptive immunity.Nature 388:394.
8. Wolfs, T. G., W. A. Buurman, A. van Schadewijk, B. de Vries, M. A. Daemen,P. S. Hiemstra, and C. van’t Veer. 2002. In vivo expression of Toll-like receptor2 and 4 by renal epithelial cells: IFN-� and TNF-� mediated up-regulation duringinflammation. J. Immunol. 168:1286.
9. Poltorak, A., X. He, I. Smirnova, M. Y. Liu, C. V. Huffel, X. Du, D. Birdwell,E. Alejos, M. Silva, C. Galanos, et al. 1998. Defective LPS signaling in C3H/HeJand C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282:2085.
10. Michalek, S. M., R. N. Moore, J. R. McGhee, D. L. Rosenstreich, andS. E. Mergenhagen. 1980. The primary role of lymphoreticular cells in the me-diation of host responses to bacterial endotoxin. J. Infect. Dis. 141:55.
11. Mokuno, Y., T. Matsuguchi, M. Takano, H. Nishimura, J. Washizu, T. Ogawa,O. Takeuchi, S. Akira, Y. Nimura, and Y. Yoshikai. 2000. Expression of Toll-likereceptor 2 on �� T cells bearing invariant V�6/V�1 induced by Escherichia coliinfection in mice. J. Immunol. 165:931.
12. Inohara, N., Y. Ogura, F. F. Chen, A. Muto, and G. Nunez. 2001. Human Nod1confers responsiveness to bacterial lipopolysaccharides. J. Biol. Chem. 276:2551.
13. Yang, R. B., M. R. Mark, A. Gray, A. Huang, M. H. Xie, M. Zhang, A. Goddard,W. I. Wood, A. L. Gurney, and P. J. Godowski. 1998. Toll-like receptor-2 me-diates lipopolysaccharide-induced cellular signalling. Nature 395:284.
14. Haziot, A., N. Hijiya, S. C. Gangloff, J. Silver, and S. M. Goyert. 2001. Inductionof a novel mechanism of accelerated bacterial clearance by lipopolysaccharide inCD14-deficient and Toll-like receptor 4-deficient mice. J. Immunol. 166:1075.
15. Murray, P. T., M. E. Wylam, and J. G. Umans. 1997. Endotoxin impairs agonist-induced calcium mobilization in rat mesangial cells. Am. J. Respir. Crit. CareMed. 156:1846.
16. Banu, N., and C. M. Meyers. 1999. IFN-� and LPS differentially modulate classII MHC and B7-1 expression on murine renal tubular epithelial cells. Kidney Int.55:2250.
17. Joly, V., Y. Bergeron, M. G. Bergeron, and C. Carbon. 1991. Endotoxin-tobra-mycin additive toxicity on renal proximal tubular cells in culture. Antimicrob.Agents Chemother. 35:351.
18. Jo, S. K., D. R. Cha, W. Y. Cho, H. K. Kim, H. K. Chang, S. Y. Yun, andN. H. Won. 2002. Inflammatory cytokines and lipopolysaccharide induce fas-mediated apoptosis in renal tubular cells. Nephron 91:406.
19. Koide, N., K. Abe, K. Narita, Y. Kato, T. Sugiyama, T. Yoshida, and T. Yokochi.1997. Expression of intercellular adhesion molecule-1 (ICAM-1) on vascularendothelial cells and renal tubular cells in the generalized Shwartzman reactionas an experimental disseminated intravascular coagulation model. FEMS Immu-nol. Med. Microbiol. 18:67.
20. Maier, S., K. Emmanuilidis, M. Entleutner, N. Zantl, M. Werner, K. Pfeffer, andC. D. Heidecke. 2000. Massive chemokine transcription in acute renal failure dueto polymicrobial sepsis. Shock 14:187.
21. Kamochi, M., F. Kamochi, Y. B. Kim, S. Sawh, J. M. Sanders, I. Sarembock,S. Green, J. S. Young, K. Ley, S. M. Fu, and C. E. Rose, Jr. 1999. P-selectin andICAM-1 mediate endotoxin-induced neutrophil recruitment and injury to the lungand liver. Am. J. Physiol. 277:L310.
22. Bannerman, D. D., and S. E. Goldblum. 2003. Mechanisms of bacterial lipopo-lysaccharide-induced endothelial apoptosis. Am. J. Physiol. 284:L899.
23. Donnahoo, K. K., X. Meng, L. Ao, A. Ayala, B. D. Shames, M. P. Cain,A. H. Harken, and D. R. Meldrum. 2001. Differential cellular immunolocalizationof renal tumor necrosis factor-� production during ischaemia versus endotoxae-mia. Immunology 102:53.
24. Glynne, P. A., and T. J. Evans. 1999. Inflammatory cytokines induce apoptoticand necrotic cell shedding from human proximal tubular epithelial cell mono-layers. Kidney Int. 55:2573.
25. Peralta-Solter, A., J. M. Mullin, K. A. Knudsen, and C. W. Marano. 1996. Tissueremodeling during tumor necrosis factor-induced apoptosis in LLC-PK1 renalepithelial cells. Am. J. Physiol. 270:F869.
26. Saikumar, P., Z. Dong, M. Mikhailov, M. Denton, J. M. Weinberg, andM. A. Venkatachalam. 1999. Apoptosis: definition, mechanisms, and relevance todisease. Am. J. Med. 107:489.
27. Bialik, S., D. L. Geenen, I. E. Sasson, R. Cheng, J. W. Horner, S. M. Evans,E. M. Lord, C. J. Koch, and R. N. Kitsis. 1997. Myocyte apoptosis during acutemyocardial infarction in the mouse localizes to hypoxic regions but occurs in-dependently of p53. J. Clin. Invest. 100:1363.
28. Lieberthal, W., S. A. Menza, and J. S. Levine. 1998. Graded ATP depletion cancause necrosis or apoptosis of cultured mouse proximal tubular cells.Am. J. Physiol. 274:F315.
29. Abraham, E., A. Carmody, R. Shenkar, and J. Arcaroli. 2000. Neutrophils asearly immunologic effectors in hemorrhage- or endotoxemia-induced acute lunginjury. Am. J. Physiol. 279:L1137.
30. Essani, N. A., M. A. Fisher, C. A. Simmons, J. L. Hoover, A. Farhood, andH. Jaeschke. 1998. Increased P-selectin gene expression in the liver vasculatureand its role in the pathophysiology of neutrophil-induced liver injury in murineendotoxin shock. J. Leukocyte Biol. 63:288.
31. Linas, S. L., D. Whittenburg, and J. E. Repine. 1991. Role of neutrophil derivedoxidants and elastase in lipopolysaccharide-mediated renal injury. Kidney Int.39:618.
32. Wang, Q., and C. M. Doerschuk. 2002. The signaling pathways induced by neu-trophil-endothelial cell adhesion. Antioxid. Redox Signal. 4:39.
33. Dusi, S., V. Della Bianca, M. Donini, K. A. Nadalini, and F. Rossi. 1996. Mech-anisms of stimulation of the respiratory burst by TNF in nonadherent neutrophils:its independence of lipidic transmembrane signaling and dependence on proteintyrosine phosphorylation and cytoskeleton. J. Immunol. 157:4615.
34. Heine, H., C. J. Kirschning, E. Lien, B. G. Monks, M. Rothe, andD. T. Golenbock. 1999. Cutting edge: cells that carry A null allele for Toll-likereceptor 2 are capable of responding to endotoxin. J. Immunol. 162:6971.
35. Hirschfeld, M., Y. Ma, J. H. Weis, S. N. Vogel, and J. J. Weis. 2000. Cuttingedge: repurification of lipopolysaccharide eliminates signaling through both hu-man and murine Toll-like receptor 2. J. Immunol. 165:618.
