点击显示 收起
【摘要】 Acute renal failure (ARF) markedly sensitizes mice to endotoxin (LPS), as evidenced by exaggerated renal cytokine/chemokine production. This study sought to further characterize this state by testing the following: 1 ) does anti-inflammatory heme oxygenase-1 (HO-1) upregulation in selected ARF models prevent this response? 2 ) Is the ARF hyperresponsive state specifically triggered by LPS? 3 ) Does excess iNOS activity/protein nitrosylation participate in this phenomenon? and 4 ) are upregulated Toll receptors involved? Mice with either 1 ) rhabdomyolysis-induced ARF (massive HO-1 overexpression), 2 ) cisplatin nephrotoxicity, 3 ) or HO-1 inhibition (Sn protoporphyrin) were challenged with either LPS (a TLR4 ligand), lipoteichoic acid (LTA; a TLR2 ligand), or vehicle. Two hours later, renal and plasma TNF- /mRNA, MCP-1/mRNA, renal nitrotyrosine/iNOS mRNA, and plasma cytokines were assessed. Renal TLR4 was gauged by mRNA and Western blot analysis. Both ARF models markedly hyperresponded to both LPS and LTA, culminating in exaggerated TNF-, MCP-1, and iNOS/nitrotryosine increments. This was despite the fact that HO-1 exerted anti-inflammatory effects. TLR4 levels were either normal (cisplatin), or markedly depressed ( 50%; rhabdomyolysis) in the ARF kidneys, despite the LPS hyperresponsive state. 1 ) The ARF kidney can hyperrespond to chemically dissimilar Toll ligands; 2 ) HO-1 does not prevent this response; 3 ) excess NO/protein nitrosylation can result; and 4 ) this hyperresponsiveness can be expressed with either normal or reduced renal TLR4 expression. This suggests that diverse signaling pathways may be involved.
【关键词】 endotoxin lipoteichoic acid iNOS tumor necrosis factor monocyte chemoattractant protein
IT HAS BECOME INCREASINGLY recognized that acute renal failure (ARF) is an independent risk factor for mortality in critically ill patients ( 4, 5, 12 ). This is despite the use of aggressive renal replacement therapy to prevent the adverse consequences of uremia. Indeed, it is now apparent that even mild, transient, acute renal insufficiency, e.g., following radiocontrast administration or cardiac surgery, can negatively impact long-term patient survival ( 14 ). The explanation for these remarkable clinical observations remains unknown. However, the data suggest that ARF, in some way, induces extra-renal tissue damage, culminating in poor patient outcomes.
In an effort to gain initial mechanistic insights into these clinical observations, this laboratory recently tested the following hypothesis: that the acutely damaged ARF kidney has greatly heightened sensitivity to systemic stressors (e.g., endotoxin), culminating in exaggerated renal cytokine and chemokine production ( 36, 37 ). With renal cytokine/chemokine efflux into the systemic circulation, extra-renal tissue damage might result. To initially test this hypothesis, mice with postischemic ARF were subjected to Escherichia coli endotoxin (LPS) injection, followed by assessment of plasma and renal cortical TNF- levels and renal cortical TNF- mRNA ( 36 ). Those studies demonstrated that the presence of preexistent ischemic renal damage did indeed heighten renal sensitivity to LPS, culminating in about two- to threefold greater increases in renal and plasma TNF- levels, compared with sham-treated, LPS-injected controls. That renal cortical TNF- mRNA levels paralleled the exaggerated plasma/renal TNF- cytokine increases suggested that the kidney either caused, or contributed to, the heightened TNF- production state ( 36 ). In a second study ( 37 ), we demonstrated that LPS hyperresponsiveness is not specific for the postischemic kidney, based on observations that both nephrotoxic ARF (cisplatin) and unilateral ureteral obstruction each evoked this same exaggerated renal LPS response. Furthermore, it was demonstrated that this hyperreactive state was not restricted to TNF-, given that both IL-10 and monocyte chemoattractant protein (MCP-1) paralleled the elevated TNF- levels. When cultured human proximal tubular (HK-2) cells were sublethally damaged with either antimycin A ( 36 ) or cisplatin ( 37 ), hyperresponsiveness to LPS was, once again, observed. Thus these in vitro findings provided direct support for the concept that damaged proximal tubules are the site of, or contribute to, the enhanced renal cytokine/chemokine production state.
The present study was undertaken to further explore the above phenomenon. Specifically, the following questions have been addressed: 1 ) Is the ARF-hyperresponsive state specifically triggered by endotoxemia or might other pathophysiological "stressors" evoke the same result? (If so, this could greatly expand the potential clinical relevance of this phenomenon, i.e., beyond the realm endotoxemia/Gram-negative sepsis). 2 ) Heme oxygenase-1 (HO-1) is induced in several forms of acute renal injury ( 15, 17, 18, 23 ), and it has been reported to confer striking protection against LPS ( 3, 7, 9, 10, 19, 21, 22, 24, 25, 30 - 32 ). Thus might HO-1 induction in selected ARF models counteract the renal injury-induced cytokine-hyperresponsive state? 3 ) Given that LPS signals through the Toll-like receptor 4 (TLR4) pathway, might renal injury induce TLR4 overexpression, thereby providing a seemingly straightforward explanation for renal injury-induced hypersensitivity to LPS? Noteworthy in this regard are recent reports of increased TLR4 expression in the delayed aftermath of both ischemic- and cyclosporine-induced ARF ( 8, 11, 33 ); and 4 ) Because LPS-induced TLR4 signaling can evoke nitric oxide (NO) production, is increased NO expression an additional provocateur in the ARF-induced hyperinflammatory state? Studies designed to answer each of these questions form the basis of this report.
METHODS
Animal Utilization
All animal protocols employed male CD-1 mice (25-35 g) that were obtained from Charles River Laboratory (Wilmington, MA). They were maintained under routine vivarium conditions with free food and water access. Experimental protocols were approved by/in conjunction with the Institutional Animal Care and Use Committee at the authors? institution.
LPS-Mediated TNF- /MCP-1 Responses in the Glycerol Model of ARF
The first goal of these studies was to test whether the glycerol ARF model, which induces profound renal HO-1 overexpression ( 15, 17, 18, 23 ), is protected against renal injury-induced hypersensitivity to LPS. Toward this end, 32 mice were lightly anesthetized with isoflurane and injected with 50% hypertonic glycerol (10 ml/kg im in equally divided doses into the hind limbs). Sixteen additional mice, briefly anesthetized with isoflurane, served as controls. Approximately 18 h later, they were placed into cylindrical restrainers. Half received a tail vein injection of E. coli LPS (2 mg/kg; 0111:B4; L -2630; Sigma, St. Louis, MO; in 80 µl of saline). The remaining mice received a sham LPS injection ( 80 µl of saline). This created the following experimental groups ( 18 h between 1st and 2nd interventions): 1 ) no glycerol tail vein saline injection ( n = 8); 2 ) glycerol treatment tail vein saline injection ( n = 8); 3 ) no glycerol LPS injection ( n = 8); and 4 ) glycerol injection LPS injection ( n = 8).
Two hours post-LPS or -saline injection, the mice were deeply anesthetized with pentobarbital sodium (3-4 mg ip). The abdominal cavity of each mouse was opened, and the right kidney was rapidly excised. A heparinized blood sample was collected immediately at the site of the severed right renal pedicle. The left kidney was then excised. Both kidneys were immediately iced, and the cortices were dissected. Right renal cortical samples were used for total RNA extraction ( 35 ); the left renal cortices were extracted for protein in the presence of protease inhibitors ( 35 ). [Note: the purpose of resecting the right kidney and then collecting plasma, rather than first phebotomizing the mice and then removing the kidney, was to avoid a period of hemorrhagic hypotension before right renal resection (which could theoretically alter mRNA levels).] Plasma and left renal protein extracts were assayed for TNF- and MCP-1 using commercially available ELISA kits (TNF-, R&D Systems, Minneapolis, MN; MCP-1, BD Biosciences, San Diego, CA). Renal RNA samples were used to assess TNF-, MCP-1, and GAPDH mRNAs by RT-PCR ( 36, 37 ). The TNF- /MCP-1 results were expressed as ratios to the GAPDH product. The severity of ARF was assessed by blood urea nitrogen (BUN) concentrations. [Note: the employed glycerol ARF protocol has previously been confirmed in this laboratory ( 35 ) to induce massive HO-1 protein overexpression, as noted by others.]
Sn Protoporphyrin-Induced HO-1 Inhibition: Renal Responses to LPS
The following experiment was conducted to directly test whether HO-1 impacts renal responses to LPS. Eight mice were lightly anesthetized with isoflurane and injected intraperitoneally with the HO-1 inhibitor Sn protoporphyrin (SnPP; 50 µmol/kg; stock solution, 5 µmol per ml of 0.01 N NaOH/saline). This regimen completely inhibits mouse kidney HO-1 (personal communication, Nath KA; Mayo Clinic, Rochester, MN). An additional eight mice served as controls (injected intraperitoneally with vehicle). Approximately 18 h later, each mouse was injected with LPS, as noted above. Two hours later, renal tissues were extracted and assayed for TNF- and TNF- mRNA. A blood sample was simultaneously obtained for plasma TNF- assessments. To determine SnPP effects independent of LPS injection, mice were injected with SnPP or its vehicle ( n = 4 each); 18 h later, plasma and renal tissue samples were obtained for the above TNF- assessments. (SnPP effects on renal cortical TLR4 abundance/TLR4 mRNA were also determined.)
Glycerol and Cisplatin-Induced ARF: Responses to Lipoteichoic Acid
Glycerol model of ARF. Lipoteichoic acid (LTA) is a cytotoxic/proinflammatory constituent of gram-positive bacteria that is released from cell walls (i.e., a gram-positive bacterial "equivalent" to gram-negative bacterial LPS). However, unlike LPS, LTA is thought to signal via the TLR2, rather than the TLR4, pathway (e.g., Refs. 20 and 26 ). In light of these considerations, the following experiment was conducted to ascertain: 1 ) whether LTA, like LPS, induces renal TNF- /MCP-1 generation and 2 ) whether the ARF kidney hyperresponds to LTA in a fashion analogous to LPS. To these ends, 10 mice were injected with glycerol, and 10 mice served as controls. Approximately 18 h later, half of the control mice and half of the postglycerol-treated mice were subjected to tail vein injections of LTA (from S. aureus; 10 mg/kg; L2515; Sigma) or its vehicle ( 80 µl of saline). Two hours posttail vein injection, the mice were deeply anesthetized, followed by blood and kidney tissue extractions, as above. TNF-, MCP-1, and their respective renal cortical mRNA levels were assessed.
Cisplatin model of ARF. It was previously established that by either 18 or 72 h after mice have received cisplatin (CP; 30 mg/kg), their kidneys hyperrespond to LPS (as gauged by TNF-, MCP-1, IL-10, and each of their respective mRNAs) ( 37 ). The following experiment was undertaken to ascertain whether a similar hyperresponsive state exists when the challenge is LTA. Sixteen mice were briefly anesthetized with isoflurane to permit intraperitoneal CP injection (30 mg/kg; Sigma; stock solution, 1 mg/ml saline). Sixteen additional mice served as controls, receiving an equal volume intraperitoneal injection of saline. Approximately 18 h later, half the CP-injected mice and half the control mice received either tail vein LTA or LTA vehicle (saline) injection. Two hours post-LTA/vehicle injections, the mice were anesthetized with pentobarbital sodium, followed by renal extraction and blood collection, as previously detailed. Plasma and renal protein/RNA samples were assayed for TNF-, MCP-1, and their respective renal cortical mRNAs ( 36, 37 ).
Renal iNOS Responses to LPS Injection: Impact of Prior Renal Injury
The following experiment was undertaken to ascertain whether heightened NO expression is another component of the ARF-induced renal cytokine/chemokine hyperresponsive state. To this end, the following experimental groups ( n = 6-8 mice per group) were established: groups 1 and 2 ): control mice ± LPS challenge; groups 3 and 4 ): 18 h postglycerol mice ± LPS challenge; groups 5 and 6 ): 18 h post-CP mice ± LPS challenge; and groups 7 and 8 ): 72 h post-CP mice ± LPS challenge. Two hours post-LPS/vehicle injections, the mice were anesthetized, and the kidneys were extracted for both protein and RNA. The following assessments were made: iNOS mRNA assessments: Right renal cortical RNA from each mouse was assayed for iNOS and GAPDH mRNAs by RT-PCR, using general methods as previously described ( 36, 37 ), and employing the conditions/primers presented in Table 1. Results were expressed as ratios to the GAPDH product.
Table 1. Primers/conditions used for RT-PCR analyses of iNOS and TLR4 mRNA
Nitrotyrosine assessments. Renal cortical protein nitrosylation was assessed by Western blotting ( 34 ). Protein samples (75 µg/lane) were electrophoresed into 4-12% precast gels (from Bio-Rad, Hercules, CA; catalog no. 345-0124). Nitrotyrosine was probed with a rabbit polyclonal antinitrotyrosine antibody (Abcam; cat. no. 23704 Cambridge, MA; 1:500 dilution). Secondary detection was performed with a horseradish peroxidase (HRP)-labeled antirabbit IgG (from donkey; Amersham Biosciences; 1:12,000) and enhanced chemiluminescence (ECL). Two approaches were used: 1 ) proteins were electrophoresed into the gels for just 15 min to allow protein entry, but not sufficiently long to permit wide separation of the protein bands (which would hinder overall quantification of degrees of nitrosylation of the entire protein extract by densitometric analysis); and 2 ) full electrophoresis ( x 60 min) to assess which protein band(s) were predominately affected by the nitrosylation process. (Note: a 70-kDa band was the dominant nitrotyrosylated product. Thus it, as well as total lane densities, was assessed.) Equal protein transfer was confirmed by subsequent India ink staining.
TLR4 Expression in Response to Glycerol- or CP-Induced Renal Injury
The following experiments were undertaken to determine whether alterations in renal TLR4 expression correlate with the ARF-induced LPS hyperresponsive state. Renal tissues from 18-h glycerol mice, 18-h CP mice, and their controls ( n = 6 each) were subjected to the following assessments.
TLR4 mRNA expression. TLR4 mRNA was determined in the above kidneys by RT-PCR ( 36, 37 ) employing primers and conditions presented in Table 1. Results were expressed as ratios to simultaneously determined GAPDH product.
Western blotting for renal cortical TLR4 expression. Renal cortical protein extracts (50 µg) were electrophoresed into 4-12% precast gels (60-min electrophoresis) and then probed for TLR4 with a rabbit antimouse/antihuman TLR4 antibody (Imgenex, San Diego, CA; Cat. no. 578A; dilution, 1:200). Anti-TLR4 detection was performed with HRP-labeled donkey antirabbit IgG (1:12,000 dilution) followed by ECL reaction and quantitation by densitometry. To confirm equal protein transfer, two approaches were taken: first, protein staining with India ink was staining; and second, in representative runs, a second antibody (rabbit anti-beta tubulin; Abcam; 1:3,000 dilution) was incubated along with the anti-TLR4 antibody (beta tubulin serving as a housekeeping protein, serving as an internal control). TLR4 and beta tubulin were detected at 95 and 55 kDa, respectively.
TLR4 Expression in Injured Cultured Proximal Tubular (HK-2) Cells
Because renal cortical results (as obtained from the above experiments) need not reflect proximal tubular cell events, the following experiment was conducted to more directly assess the impact of cell injury on proximal tubular cell TLR4 expression. In brief, human derived proximal tubular (HK-2) cells were grown in 27 T75 Costar flasks with keratinocyte serum-free medium (K-SFM), as previously described ( 36 ). At near confluence, the 27 flasks were divided into three experimental groups ( n = 9 each): group 1 ) control incubation; group 2 ) incubation with 7.5 µM antimycin A (previously been shown to sensitize HK-2 cells to LPS-mediated cytokine generation) ( 36 ); or Fe gluconate complex (Ferrlecit; Watson Laboratories; 0.5 mg/ml Fe content) to simulate glycerol-mediated/Fe-driven oxidative stress ( 35 ). After 18 h under these conditions, the media were removed, the cells were recovered by scraping with a rubber "policeman," and following washing and cell pelleting, protein extracts were prepared in the presence of protease inhibitors. The protein extracts were analyzed for TLR4 and beta tubulin using the conditions and reagents noted above for mouse TLR4 detection.
Calculations and Statistics
All values are given as means ± SE. Statistical comparisons were performed by unpaired Student's t -test, unless stated otherwise. If multiple statistical comparisons were made between groups, the Bonferroni correction was applied. Significance was judged by a P value of <0.05.
RESULTS
Glycerol ARF: Responses to LPS
Renal cortical TNF- protein/mRNA assessments. Glycerol injection induced severe ARF, as assessed by BUN concentrations (129 ± 11 vs. controls, 26 ± 1 mg/dl; P < 0.001). Despite this injury, no increase in renal cortical TNF- ( Fig. 1, left ) or TNF- mRNA resulted ( right ) compared with control mouse kidney values.
Fig. 1. TNF- and TNF- mRNA levels in normal and postglycerol acute renal failure (ARF) renal cortex under basal conditions and following LPS injection. By 18 h post-glycerol (Gly) injection, no increase in either renal cortical TNF- or TNF- mRNA was apparent, compared with values in control (C) mouse renal cortex. LPS injection into control mice significantly increased both TNF- and its mRNA (* P < 0.01 vs. control values). When LPS was injected into glycerol-induced ARF mice, a 2- to 3-fold greater increase in both TNF- and TNF- mRNA developed vs. LPS-injected controls. NS, not significant.
LPS injection into control mice significantly increased renal cortical TNF- and TNF- mRNA (* P < 0.01 vs. controls). When injected into glycerol ARF mice, two- to threefold greater TNF- and TNF- mRNA increases were observed vs. those seen in LPS-injected controls. Thus, although glycerol injection did not independently alter TNF- /TNF- mRNA, it sensitized the mice to LPS-induced TNF- cytokine/mRNA increases.
Renal cortical MCP-1 protein/mRNA assessments. The glycerol model slightly increased renal cortical MCP-1 ( Fig. 2, left ). A corresponding increase in MCP-1 mRNA was also observed ( Fig. 2, right ).
Fig. 2. Monocyte chemoattractant protein-1 (MCP-1) and MCP-1 mRNA levels in normal and postglycerol-induced ARF renal cortex under basal conditions and following LPS injection. By 18 h post-Gly injection, modest increases in renal cortical MCP-1 and MCP-1 mRNA levels were apparent compared with C values. When LPS was injected into the glycerol ARF mice, significantly greater increases in renal cortical MCP-1 and MCP-1 mRNA levels developed, compared with LPS-injected controls.
LPS injection raised both MCP-1 and its mRNA in all mice. However, the increases were 2 x as great in the glycerol ARF mice vs. their LPS-injected controls. Thus the MCP-1 results mirrored the above noted TNF- results, i.e., glycerol-induced injury hypersensitized the kidney to LPS-mediated cytokine/chemokine increases.
Plasma cytokine levels in control and glycerol ARF mice. The glycerol ARF model caused no discernible increase in TNF- levels (<5 pg/ml for glycerol and control mice; the lower limit of assay detection; Fig. 3 ). Plasma MCP-1 levels were slightly higher in glycerol ARF mice vs. their controls (0.2 ± 0.02 vs. 0.1 vs. 0.05 ng/ml, respectively; P < 0.05). However, these absolute values were quantitatively trivial and thus appear as "zero" in Fig. 3.
Fig. 3. Plasma TNF- and MCP-1 concentrations in glycerol-induced ARF and control mice ± LPS injection. Left : TNF- was barely detectable (<5 pg/ml) in plasma harvested 18 h postglycerol injection or in normal mice (appearing as zero values in figure). LPS induced marked TNF- increases in both C and Gly mice, but the values were 300% higher in the glycerol ARF group. Right : MCP-1 levels were significantly elevated in the non-LPS-challenged glycerol ARF mice vs. controls (0.2 ± 0.02 vs. 0.1 vs. 0.05 ng/ml; P < 0.05), but given the low values, they appear as "zero." LPS markedly increased plasma MCP-1 levels but to comparable degrees in the 2 groups.
LPS induced dramatic increases in TNF- and MCP-1 plasma levels. The degree of the TNF- increase was 3 x greater in glycerol ARF mice vs. LPS-injected controls (again consistent with increased sensitivity to LPS). Conversely, the LPS-induced plasma MCP-1 increases were comparable for the two groups (consistent with prior observations that renal cortical MCP-1 and MCP-1 mRNA increases in response to LPS do not correspond well to changes in plasma MCP-1 levels) ( 37 ).
SnPP-Induced HO-1 Inhibition: Renal Responses to LPS
As shown in Table 2, in the absence of LPS, SnPP caused an approximate 50% increase in renal cortical TNF- without altering TNF- mRNA. This occurred without any change in plasma TNF- levels (undetectable; <5 pg/ml).
Table 2. SnPP effects on TNF-, MCP-1, and TLR4 ± LPS
SnPP also sensitized to LPS, significantly increasing LPS-mediated TNF- /TNF- mRNA levels. An increase in plasma TNF- levels also occurred (LPS ± SnPP 2,517 ± 45 vs. LPS alone, 1,930 ± 29 pg/ml, P < 0.01). These changes were not associated with any change in TLR4 mRNA levels ( Table 2 ) or TLR4 protein density, as assessed by Western blotting (controls, 3,345 ± 150 densitometry units; SnPP, 3,154 ± 158; NS).
Glycerol ARF: Responses to LTA
TNF- protein: renal cortical and plasma assessments. Control values for renal cortical TNF- /TNF- mRNA (data from above experiments) are depicted in Fig. 4 by the horizontal dotted lines (upper limit of 95% confidence band). LTA injection significantly increased renal cortical ( Fig. 4, left ) and plasma TNF- concentrations ( Fig. 4, right ) in all mice (* P < 0.01 vs. control values). However, glycerol mice clearly hyperresponded, as evidenced by 50% greater increases in renal cortical TNF-, and 20 x greater plasma TNF- increases, compared with LTA-injected controls. Conversely, LTA induced comparable TNF- mRNA responses in both groups of animals ( Fig. 4, middle ).
Fig. 4. Lipoteichoic (LTA) injections: effects on TNF- expression in normal mice and glycerol ARF mice. Left : LTA injection increased renal cortical TNF- levels in both normal and postglycerol ARF mice (* P < 0.01 vs. normal concentrations). The increase was statistically greater ( P < 0.01) in the glycerol ARF group (dotted line represents upper limit of normal values). Right : LTA injection had virtually no effect on plasma TNF- levels in control mice. Conversely, the glycerol ARF mice manifested a profound TNF- increase in response to LTA injection. Middle : above noted preferential increases in renal cortical and plasma TNF- levels in the glycerol ARF mice were expressed in the absence of a difference in renal TNF- mRNA expression. LTA significantly increased TNF- mRNA in both groups (* P < 0.01) compared with normal mean values.
MCP-1 protein: renal cortical and plasma assessments. Control renal cortical MCP-1/MCP-1 mRNA values (data from above LPS experiments) are depicted in Fig. 5 by the horizontal dotted lines (upper limit of 95% confidence band). LTA injection into normal mice induced slight increases in both renal cortical MCP-1 and MCP-1 mRNA (* P < 0.05 vs. controls). The glycerol mice hyperresponded to LTA, approximately doubling renal cortical MCP-1 protein/mRNA values vs. those seen in the LTA-injected controls. Although LTA increased plasma MCP-1 levels in both normal and glycerol ARF mice (* P < 0.05 vs. control values), comparable increases were observed in both sets of animals (again consistent with above noted observations: that plasma MCP-1 levels do not correlate well with renal cortical MCP-1/ MCP-1 mRNA changes).
Fig. 5. LTA injections: effects on MCP-1 expression in normal mice and mice with glycerol-induced ARF. Left : LTA increased renal cortical MCP-1 levels, with the degree of increase being 2 x greater in the glycerol ARF vs. the control group ( P < 0.0025; normal values depicted by dotted line). Middle 2 x greater in the glycerol ARF group (thereby mirroring the MCP-1 protein responses). Right : despite the preferential MCP-1/mRNA increases in the glycerol vs. control group, plasma MCP-1 concentrations were equally raised by LTA injection. *Statistically greater than normal values ( P < 0.05).
Cisplatin ARF: Responses to LTA
Renal cortical TNF- responses. By 18 h post-CP treatment, modest renal insufficiency had developed (BUNs, 42 ± 5 vs. 26 ± 2 mg/dl for controls; P < 0.01). This injury was associated with a significant increase in TNF- mRNA ( Fig. 6, right ), but not in renal cortical TNF- protein levels ( Fig. 6, left ).
Fig. 6. Impact of cisplatin (CP) nephrotoxicity on TNF- responses to LTA. Left : CP did not significantly increase renal cortical TNF- levels in C mice. However, it dramatically sensitized to LTA-induced renal cortical TNF- induction. Right : CP also sensitized the kidney to LTA-initiated TNF- mRNA increases.
LTA injection raised both TNF- and TNF- mRNA. The CP-treated mice hyperresponded to LTA injection, as denoted by approximately twofold greater increases in both TNF- cytokine and its mRNA, vs. values seen in LTA-injected controls.
Renal cortical MCP-1 responses. CP treatment alone induced quantitatively trivial, but statistically significant, increases in renal cortical MCP-1 ( Fig. 7, left ) and MCP-1 mRNA ( Fig. 7, right ).
Fig. 7. Impact of CP nephrotoxicity on MCP-1 responses to LTA. CP alone caused only small increases in MCP-1 protein and mRNA expression. However, CP strongly sensitized to LTA-driven MCP-1 protein/mRNA increases.
All mice responded to LTA injection with significant increases in both MCP-1 protein and MCP-1 mRNA (<0.01 vs. control values). Once again, the CP-treated mice hyperresponded to LTA, with approximately two- to threefold greater increases in both MCP-1 and MCP-1 mRNA levels being observed, vs. LTA-injected controls. Thus the results depicted in Figs. 6 and 7 indicate that by 18 h post-CP treatment, mice hyperrespond to LTA, as denoted by exaggerated increases in renal cortical TNF- /mRNA and MCP-1/mRNA vs. their LTA-injected controls.
Plasma cytokine responses. Both the 18-h CP-pretreated mice and the LTA-injected control mice had virtually no detectable TNF- in plasma (<5 pg/ml; appearing as zero/near zero in the figure). However, when injected into CP-pretreated mice, LTA strikingly increased plasma TNF- levels, rising to 800 pg/ml ( Fig. 8, left ).
Fig. 8. CP toxicity: impact on plasma TNF- /MCP-1 responses to LTA. Left : TNF- levels were at, or below, the level of detection in the C and CP mice (appearing as 0 in the figure). Control mice failed to mount a TNF- response to LTA. However, when injected into CP-pretreated mice, LTA induced dramatic TNF- increases. Right : CP-treated mice manifested a quantitatively trivial, but statistically significant, increase in plasma MCP-1 concentrations. However, CP treatment sensitized the mice to LTA-induced plasma MCP-1 increases, doubling the values seen in control mice injected with LTA.
LTA injection into control mice raised plasma MCP-1 levels ( P < 0.01 vs. normal values). The CP-pretreated mice hyperresponded to LTA injection (2 x increase over values observed in LTA-injected controls; Fig. 8, right ).
iNOS Expression in Control and ARF Kidneys
Assessments of iNOS mRNA. Despite substantial renal injury, neither glycerol alone nor CP treatment alone increased renal cortical iNOS mRNA levels (controls, 0.5 ± 0.005; 18 h postglycerol, 0.5 ± 0.1; 18 h CP, 0.2 ± 0.02; 72 h CP, 0.5 ± 0.2). BUNs were controls 26 ± 1 mg/dl, 18 h glycerol mice 128 ± 12, 18 h CP mice 42 ± 4, and 72 h CP mice 134 ± 34 mg/dl.
The control mice responded to LPS injections with 3 x increases in renal cortical iNOS mRNA ( Fig. 9, left ). The 18-h CP mice also manifested a 3 x increase in iNOS mRNA with the LPS challenge, mirroring values in the controls. However, both the 18-h glycerol mice and the 72-h CP mice massively overresponded to LPS, with iNOS mRNAs reaching values that were 5 x (glycerol) and 10 x (CP) higher than those seen in the LPS-injected controls.
Fig. 9. LPS effects on renal cortical iNOS mRNA and nitrotyrosine expression. Left : renal cortical iNOS mRNA levels in postglycerol ARF and 72 h post-CP-induced ARF did not differ from control values. LPS injection caused a significant increase in iNOS message in all mice. However, the relative LPS-induced increases were 5- to 10-fold greater in the injured kidneys compared with the values observed in LPS-injected controls. Right : total baseline renal cortical nitrotyrosine levels were 50% lower in 18 h postglycerol and 72 h post-CP-treated mice compared with controls. The control mice did not develop a detectable nitrotyrosine increase in response to LPS. However, when administered to the glycerol and 72 h CP ARF mice, a doubling of total nitrotyrosine expression resulted.
Renal nitrotyrosine analyses. Nitrotyrosine was observed in all renal cortical samples. By 2 h post-LPS injection, no increase in nitrotyrosine levels was seen in control kidneys ( Fig. 9, right ).
Total baseline nitrotyrosine expression was significantly depressed (by 50%) in both the 18-h glycerol and in the 72-h CP kidney extracts (vs. normal kidney values). However, unlike the control kidneys, both the glycerol and 72-h CP kidneys manifested dramatic responses to LPS injection, doubling their total nitrotyrosine levels ( Fig. 9, right ). The bulk of the nitrotyrosine increase appeared as an approximate 70-kDa band ( Fig. 10; rows 4 and 5; *). When the Western blots were run with a longer separation time to further isolate this 70-kDa band (depicted in line 6 of Fig. 10 ), statistically greater levels were observed in the LPS ± glycerol and LPS ± CP kidneys vs. any other groups [controls, 202 ± 82 vs. LPS, 253 ± 63 (NS); glycerol, 292 ± 57 vs. glycerol ± LPS 482 ± 45; P < 0.01; CP 161 ± 40 vs. CP ± LPS 443 ± 104; P < 0.01]. Thus a correlate of the preferential increase in iNOS mRNA in the LPS-challenged injured kidneys was an approximate doubling of nitrotyrosine levels (as assessed by either total values or 70-kDa values). This was compared with no changes for LPS-injected controls.
Fig. 10. Representative Western blots of TLR4 and nitrotyrosine expression. Row 1 : TLR4 in glycerol (G) and control (C) kidneys. TLR4 appeared as a doublet at 95 kDa (reflecting different degrees of glycosylation; Ref. 6 ). There was an approximate 50% reduction in its expression in the postglycerol kidney samples (see Fig. 11 ). This affected both of the protein bands. Row 2 : TLR4 in 18 h CP and control kidneys. No significant differences were observed between the groups. Row 3 : TLR4 expression in culture proximal tubular (HK-2) cells following 18-h treatments with either Fe gluconate treatment ("Fe"; used to simulate rhabdomyolysis-induced, Fe-mediated oxidant stress) or antimycin A (AA). As with the in vivo blots of TLR4, a doublet appearance was observed. AA had no effect on either TLR4 band. Conversely, Fe treatment caused a statistically significant TLR4 reduction, principally involving the lower molecular wt. band. Rows 4 and 5 : total nitrotyrosine (NT) expression in control kidneys (C), 72 h post-CP kidneys (CP), and postglycerol (G) kidneys. LPS failed to raise NT levels in control mice. NT levels were suppressed in both CP and glycerol mice in the absence of LPS (compared with controls). However, with LPS, both CP and glycerol mouse mounted a NT response that was observed at 70 kDa (denoted by *). Row 6 : separate analyses of this 70-kDa band demonstrated its existence almost exclusively in the CP ± LPS and the glycerol ± LPS groups (see RESULTS for values).
TLR4 Analyses
Renal cortical Western blot analyses. Representative blots are presented in Fig. 10 ( rows 1 and 2 ). TLR4 appeared as a doublet (reflecting differing degrees of glycosylation) ( 6 ) at the relevant molecular size of 95 kDa. By 18 h postglycerol injection, an approximate 50% reduction in TLR4 expression was apparent, reflecting decreases in both bands ( row 1 ). CP toxicity, assessed at 18 h postinjection, did not significantly alter TLR4 expression ( Fig. 10, row 2 ). Statistical comparisons of the Western blot TLR4 densities are provided in Fig. 11, left. Glycerol induced a statistically significant ( 50%) TLR4 reduction. CP treatment had no effect.
Fig. 11. Renal TLR4 quantitation, as assessed by Western blotting and mRNA analyses. Left : by 18 h postglycerol injection, renal cortical TLR4 levels were reduced by 50%, as gauged by Western blotting. Conversely, no significant change in TLR4 was observed in 18 h post-CP-treated kidneys. Right : both the glycerol and CP-treated mice manifested an approximate doubling of renal cortical TLR4 mRNA compared with normal mice.
Renal cortical TLR4 mRNA expression. By 18 h postglycerol or CP injection, 2 x increases in TLR4 mRNA were observed vs. time-matched control values ( Fig. 11, right ).
HK-2 cell TLR4 Western blot assessments. As in renal cortex, HK-2 cell TLR4 was again detected as a doublet at 95 kDa (see line 3, Fig. 10 ). Fe-mediated oxidative stress caused an approximate 40% reduction in TLR4 (density units: controls, 701 ± 52; Fe, 414 ± 51; P < 0.005). Thus this was highly analogous to the 50% glycerol ARF-mediated TLR4 reductions, noted above. Conversely, antimycin A (AA), known to predispose to LPS-mediated inflammation ( 37 ), did not significantly alter HK-2 cell TLR4 levels (721 ± 53; NS vs. the above controls).
DISCUSSION
TLRs are plasma membrane glycoproteins that recognize and bind pathogen-associated molecules. Via recruitment of two proximate "downstream" adaptor proteins, MyD88 and TRIF, they signal the production of inflammatory molecules that are part of the innate immune response ( 1, 2, 28, 29 ). Although these pathways may confer survival advantages, when uncontrolled, an exaggerated inflammatory state may result, potentially culminating in multiorgan dysfunction and cell death. It is through the TLR4 pathway that LPS is thought to exert its toxic effects. Factors that protect against LPS-mediated TLR4 signaling remain poorly defined. However, recent observations suggest that HO-1, a potent cytoprotectant with antioxidant, vasodilator, and anti-inflammatory activities (reviewed in Ref. 16 ) can curb LPS/TLR-initiated toxicity. This conclusion is based on observations that HO-1 "knockout" mice have increased vulnerability to LPS ( 32 ) and that HO-1 overexpression, e.g., as induced via heme protein injection ( 21 ) or genetic manipulation ( 38 ), mitigates LPS-triggered TNF- production and lethality.
In our previous studies ( 36, 37 ), we demonstrated that the ARF kidney hyperresponds to LPS, with exaggerated cytokine and chemokine production being the result. However, because HO-1 confers protection against LPS-initiated TLR signaling (e.g., Ref. 30 - 32 ), we hypothesized that an ARF model with robust HO-1 overexpression should be resistant to an LPS-initiated hyperinflammatory response. To test this hypothesis, we selected the glycerol ARF model because it evokes the most profound renal HO-1 increases documented to date (e.g., Refs. 15, 17, 23, 35 ). Thus HO-1 upregulation might block LPS hyperresponsiveness in this particular ARF model. Despite this theoretical consideration, we observed that LPS induced approximately two- to threefold greater cytokine/chemokine increases in glycerol ARF mice compared with their LPS-treated controls. Indeed, the levels of these increases are highly comparable to those previously noted in our prior CP and postischemic ARF studies ( 36, 37 ). To prove that HO-1 can, indeed, protect the kidney against LPS-induced cytokine generation, mice were treated with SnPP to inhibit HO-1, and then they were challenged with LPS. SnPP significantly increased renal cortical TNF- under basal conditions, and it enhanced renal TNF- /and renal TNF- mRNA responses to LPS. It also exaggerated LPS-induced plasma TNF- increases. Thus, despite HO-1's protective influence, confirmed herein, it is clearly insufficient to override a glycerol-induced, ARF-initiated, proinflammatory state.
In our prior studies, ARF-induced cytokine hyperresponsiveness was defined solely by responses to LPS. Therefore, the second goal of the present study was to ascertain whether acute renal injury might also sensitize the kidney to a non-LPS stressor that signals via an alternative, non-TLR4, pathway. Toward this end, mice with either glycerol-induced ARF or CP nephrotoxicity were challenged with S. aureus LTA, a TLR2 ligand ( 20, 26 ). As with LPS, both the glycerol- and CP-ARF kidneys dramatically hyperresponded to LTA injection. This is graphically illustrated by plasma TNF- assessments. Whereas neither LTA alone, nor ARF alone, raised plasma TNF- 2-5 pg/ml), when LTA was injected into CP- or glycerol-ARF mice, massive plasma TNF- increases resulted ( 800 pg/ml). Supporting the notion that the kidney induces/participates in this response is that LTA-injected ARF mice manifested about twofold greater increases in renal cortical TNF- /mRNA and MCP-1/mRNA than did LTA-injected controls. Indeed, these LTA results could have substantial clinical relevance: they imply that more than one stressor (i.e., beyond LPS) can initiate excessive cytokine/chemokine production in the ARF kidney. Of note, multiple bacterial products, degraded tissue constituents (e.g., RNA, DNA), and stress proteins (e.g., heat shock proteins) can act as TLR ligands ( 2, 16, 28, 29 ). This raises the possibility that patients who sustain extrarenal tissue injury might release such ligands into the systemic circulation, and with renal access, an exaggerated TLR-based inflammatory response could result. Indeed, such factors might also induce extrarenal TLR4 signaling, thereby contributing to increases in circulating as well as intrarenal cytokine levels. To the degree that cytokines are excreted in the urine (e.g., Ref. 36 ), ARF per se might further contribute to elevated plasma cytokine levels.
To date, only TNF-, MCP-1, and IL-10 have been documented to participate in the ARF-initiated renal "hyperresponsive" state ( 36, 37 ). Each is thought to be a downstream product of the MyD88 adaptor arm of the TLR signaling cascade. Conversely, LPS-mediated NO generation is thought to arise via the TLR-TRIF adaptor pathway ( 27, 39 ). Therefore, to more fully define the biological scope, and signaling pathways, of the ARF-initiated proinflammatory state, NO expression by the ARF kidney and its responsiveness to LPS were assessed. Three notable results arose from these investigations. First, despite the induction of substantial, if not massive, renal injury, neither rhabdomyolysis nor CP toxicity increased baseline renal iNOS mRNA. Second, despite essentially identical baseline iNOS mRNAs for the control, postglycerol, and post-CP kidneys, the latter two massively overresponded to LPS, with their iNOS mRNA levels rising 5 to 10 times more than LPS-challenged controls. Third, a correlate of this LPS-initiated iNOS mRNA hyperresponsiveness was an increase in renal nitrotyrosine formation. Whereas LPS failed to increase nitrotyrosine levels in control kidneys, it doubled total, as well as 70-kDa, renal nitrotyrosine in the setting of glycerol- or CP-induced ARF. This was despite the fact that baseline total nitrotyrosine levels in the glycerol and CP kidneys were actually suppressed vs. controls. Clearly then, LPS-mediated NO hyperproduction, with increased protein nitrosylation, appears to be concomitants of the ARF-initiated hyperresponsive state. These data also imply that the TRIF pathway likely participates in this phenomenon.
The final goal of the present study was to ascertain whether increased TLR4 expression might be a simple explanation for why acute renal injury sensitizes to LPS. Noteworthy in this regard are prior studies that have reported increased TLR4 abundance in response to chronic cyclosporine administration and during the late recovery stage of postischemic ARF (assessed by immunohistochemistry or Western blots) ( 8, 11, 13, 33 ). Because LPS hyperresponsiveness is present at 18 h post-glycerol or CP injection, this time point was chosen for our TLR4 assessments. Surprisingly, glycerol (Fe mediated) ARF caused a 50% reduction in renal cortical TLR4 expression, as assessed by Western blotting. To confirm that these results reflected, at least in part, a proximal tubule event, cultured HK-2 cells were challenged with Fe, and a 40% TLR4 reduction resulted. Conversely, neither in vivo CP toxicity nor in vitro AA toxicity (which also increase LPS responsiveness) ( 36, 37 ) increased TLR4 expression. Clearly then, these results indicate that ARF-induced LPS hyperresponsiveness cannot simply be explained by TLR4 upregulation (at least during the early injury vs. the late injury phase) ( 8, 11, 13, 33 ). It is notable that TLR4 mRNA was increased in both glycerol- and CP-treated kidneys, despite reduced or normal TLR4 protein expression. This underscores that TLR4 mRNA levels cannot be used as a surrogate marker for TLR4 protein levels. Indeed, the TLR4 mRNA increases might simply represent a compensatory response to injury-associated TLR4 protein destruction.
In conclusion, the present results expand on our previous observations of ARF-induced cytokine hyperresponsiveness ( 36, 37 ) in several important ways. First, we provided the first demonstration that the glycerol ARF kidney markedly hyperresponds to LPS, despite the fact that robust HO-1 induction, a hallmark of this ARF model, has been proven to dampen LPS renal cytokine signaling. This indicates that acute renal injury is clearly sufficient to override HO-1's anti-inflammatory effects, culminating in an exaggerated cytokine/chemokine production state. Second, the present study indicates that ARF-induced cytokine hyperresponsiveness is not limited to LPS as the inciting agent. That S. aureus LTA, a TLR2 ligand, can recapitulate LPS actions raises the distinct possibility that multiple stressors/Toll ligands may be capable of triggering the ARF-associated hyperinflammatory response. Third, this study provides the first evidence that iNOS is an active participant in the "effector arm" of the ARF cytokine hyperresponsive state, presumably by acting via the TRIF (as opposed to the MyD88) adaptor protein pathway. The biological relevance of this finding is underscored by the results of the nitrotyrosine assays, given that both ARF models, but not controls, increased renal nitrotyrosine levels in response to LPS; and fourth, ARF-induced hyperresponsiveness is not determined by, or predicated on, increased TLR4 expression. Indeed, exaggerated LPS responsiveness can be expressed even in the setting of markedly reduced TLR4 levels (e.g., in glycerol ARF). However, these findings do not exclude increased TLR signaling, as it remains possible that the activity of residual TLR units in ARF kidneys could be enhanced. Alternatively, LPS and LTA could be mounting exaggerated cytokine/chemokine responses via non-TLR pathways. Exploration of this issue remains an important subject for future study, in part, because new therapeutic approaches for blocking these presumed maladaptive cytokine responses could result.
GRANTS
This work was supported by National Institutes of Health Research Grants R37-DK-38432-18 and R01-DK-68520-01.
ACKNOWLEDGMENTS
The authors thank A. Wolf for assistance with manuscript preparation.
【参考文献】
Akira S and Takeda K. Toll-like receptor signaling. Nat Rev Immunol 4: 499-511, 2004.
Beutler B. Innate immunity: an overview. Mol Immunol 40: 845-859, 2004.
Brouard S, Berberat PO, Tobiasch E, Seldon MP, Bach FH, and Soares MP. Heme oxygenase-1-derived carbon monoxide requires the activation of transcription factor NF- B to protect endothelial cells from tumor necrosis factor- -mediated apoptosis. J Biol Chem 277: 17950-17961, 2002.
Chertow GM, Burdick E, Honour M, Bonventre JV, and Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol 16: 3365-3370, 2005.
Chertow GM, Levy EM, Hammermeister KE, Grover F, and Daley J. Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 104: 343-348, 1998.
Correia JS and Ulevitch RJ. MD-2 and TLR4 N -linked.glycosylations are important for functional lipopolysaccharide receptor. J Biol Chem 277: 1845-1854, 2002.
Fujii H, Takahashi T, Nakahira K, Uehara K, Shimizu H, Matsumi M, Morita K, Hirakawa M, Akagi R, and Sassa S. Protective role of heme oxygenase-1 in the intestinal tissue injury in an experimental model of sepsis. Crit Care Med 31: 893-902, 2003.
Kim SB, Lim SW, Kim JS, Sun BY, Ahm KO, Han SW, Kim J, and Yang CW. Ischemia-reperfusion injury activates innate immunity in rat kidneys. Transplantation 79: 1370-1377, 2005.
Kushida T, Li Volti G, Goodman AI, and Abraham NG. TNF- -mediated cell death is attenuated by retrovirus delivery of human heme oxygenase-1 gene into human microvessel endothelial cells. Transplant Proc 34: 2973-2978, 2002.
Kushida T, Li Volti G, Quan S, Goodman A, and Abraham NG. Role of human heme oxygenase-1 in attenuating TNF- -mediated inflammation injury in endothelial cells. J Cell Biochem 87: 377-385, 2002.
Leemans JC, Stokman G, Claessen N, Rouschop KM, Teske GJ, Kirschning CJ, Akira S, van der Poll T, Weening JJ, and Florquin S. Renal-associated TLR2 mediates ischemia/reperfusion injury in the kidney. J Clin Invest 115: 2894-2903, 2005.
Levy EM, Viscoli CM, and Horwitz RI. The effect of acute renal failure on mortality. A cohort analysis. JAMA 275: 1489-1494, 1996.
Lim SW, Li C, Ahn KO, Kim J, Moon IS, Ahn C, Lee JR, and Yang CW. Cyclosporine-induced injury induces Toll-like receptor, and maturation of dendritic cells. Transplantation 80: 691-699, 2005.
McCullough PA, Wolyn R, Rocher LL, Levin RN, and O?Neill WW. Acute renal failure after coronary intervention: incidence, risk factors, and relationship to mortality. Am J Med 103: 368-375, 1994.
Nath KA. Heme oxygenase-1: a redoubtable response that limits reperfusion injury in the transplanted adipose liver. J Clin Invest 104: 1485-1486, 1999.
Nath KA. Heme oxygenase-1: a provenance for cytoprotective pathways in the kidney and other tissues. Kidney Int In press.
Nath KA, Balla G, Vercellotti GM, Balla J, Jacob HS, Levitt MD, and Rosenberg ME. Induction of heme oxygenase is a rapid, protective response in rhabdomyolysis in the rat. J Clin Invest 90: 267-270, 1992.
Nath KA, Haggard JJ, Croatt AJ, Grande JP, Poss KD, and Alam J. The indispensability of heme oxygenase-1 in protecting against acute heme protein-induced toxicity in vivo. Am J Pathol 156: 1527-1535, 2000.
Ohta K, Kikuchi T, Arai S, Yoshida N, Sato A, and Yoshimura N. Protective role of heme oxygenase-1 against endotoxin-induced uveitis in rats. Exp Eye Res 77: 665-673, 2003.
Opitz B, Schröder NWJ, Sprietzer I, Michelsen KS, Kirschning CJ, Hallatschek W, Zähringer U, Hartung T, Göbel UB, and Schumann RR. Toll-like receptor-2 mediates treponema glycolipid and lipoteichoic acid-induced NF- B translocation. J Biol Chem 276: 22041-22047, 2001.
Otterbein L, Chin BY, Otterbein SL, Lowe VC, Fessler HE, and Choi AM. Mechanism of hemoglobin-induced protection against endotoxemia in rats. A ferritin-independent pathway. Am J Physiol Lung Cell Mol Physiol 272: L268-L275, 1997.
Pileggi A, Molano RD, Berney T, Cattan P, Vizzardelli C, Oliver R, Fraker C, Ricordi C, Pastori RL, Bach FH, and Inverardi L. Heme oxygenase-1 induction in islet cells results in protection from apoptosis and improved in vivo function after transplantation. Diabetes 50: 1983-1991, 2001.
Platt JL and Nath KA. Heme oxygenase: protective gene or Trojan horse? Nat Med 4: 1364-1365, 1998.
Sarady JK, Zuckerbraun BS, Bilban M, Wagner O, Usheva A, Liu F, Ifedigbo E, Zamora R, Choi AM, and Otterbein LE. Carbon monoxide protection against endotoxic shock involves reciprocal effects on iNOS in the lung and liver. FASEB J 18: 854-856, 2004.
Sawle P, Foresti R, Mann BE, Johnson TR, Green CJ, and Motterlini R. Carbon monoxide-releasing molecules (CO-RMs) attenuate the inflammatory response elicited by lipopolysaccharide in RAW2647 murine macrophages. Br J Pharmacol 145: 800-810, 2005.
Schroder NW, Diterich I, Zinke A, Eckert J, Draing C, von Baehr V, Hassler D, Priem S, Hahn K, Michelsen KS, Hartung T, Burmester GR, Gobel UB, Hermann C, and Schumann RR. Heterozygous Arg753Gln polymorphism of human TLR-2 impairs immune activation by Borrelia burgdorferi and protects from late stage Lyme disease. J Immunol 175: 2534-2540, 2005.
Sugiyama T, Fujita M, Koide N, Mori I, Yoshida T, Mori H, and Yokochi T. 2-Aminopurine inhibits lipopolysaccharide-induced nitric oxide production by preventing IFN- production. Microbiol Immunol 48: 957-963, 2004.
Takeda K and Akira S. Toll receptors and pathogen resistance. Cell Microbiol 5: 143-153, 2003.
Takeda K, Kaisho T, and Akira S. Toll-like receptors. Annu Rev Immunol 21: 335-376, 2003.
Tobiasch E, Gunther L, and Bach FH. Heme oxygenase-1 protects pancreatic beta cells from apoptosis caused by various stimuli. J Investig Med 49: 566-571, 2001.
Vicente AM, Guillen MI, Habib A, and Alcaraz MJ. Beneficial effects of heme oxygenase-1 upregulation in the development of experimental inflammation induced by zymosan. J Pharmacol Exp Ther 307: 1030-1037, 2003.
Wiesel P, Patel AP, DiFonzo N, Marria PB, Sim CU, Pellacani A, Maemura K, LeBlanc BW, Marino K, Doerschuk CM, Yet SF, Lee ME, and Perrella MA. Endotoxin-induced mortality is related to increased oxidative stress, and end organ dysfunction, not refractory hypotension, in heme oxygenase-1 deficient mice. Circulation 102: 3015-3022, 2000.
Wolfs TH, Buurman WA, Van Schadewijk A, de Vries B, Daemen MA, Hiemstra PS, and Van?t Veer C. In vivo expression of Toll-like receptor 2 and 4 by renal epithelial cells: IFN- and TNF- mediated upregulation during inflammation. J Immunol 168: 1286-1293, 2002.
Zager RA, Johnson ACM, and Hanson SY. Sepsis syndrome stimulates proximal tubule cholesterol synthesis and suppresses the SR-B1 cholesterol transporter. Kidney Int 63: 123-133, 2003.
Zager RA, Johnson ACM, and Hanson SY. Parenteral iron therapy exacerbates experimental sepsis. Kidney Int 65: 2108-2112, 2004.
Zager RA, Johnson ACM, Hanson SY, and Lund S. Ischemic proximal tubular injury primes mice to endotoxin induced TNF- generation and release. Am J Physiol Renal Physiol 289: F289-F297, 2005.
Zager RA, Johnson AC, Hanson SY, and Lund S. Acute nephrotoxic and obstructive injury primes the kidney to endotoxin-driven cytokine/chemokine production. Kidney Int 69: 1181-1188, 2006.
Zampetaki A, Minamino T, Mitsialis SA, and Kourembanas S. Effect of heme oxygenase-1 overexpression in two models of lung inflammation. Exp Biol Med 228: 442-446, 2003.
Zughaier SM, Zimmer SM, Datta A, Carlson RW, and Stephens DS. Differential induction of the toll-like receptor 4-MyD88-dependent and -independent signaling pathways by endotoxins. Infect Immun 73: 2940-2950, 2005.
作者单位：1 Department of Medicine, University of Washington, and the 2 Fred Hutchinson Cancer Research Center, Seattle, Washington