Sphingosine kinase 1 protects against renal ischemia reperfusion injury in mice by sphingosine-1-phosphate 1 receptor activation

Transcription

1 & International Society of Nephrology Sphingosine kinase protects against renal ischemia reperfusion injury in mice by sphingosine--phosphate receptor activation Sang Won Park,, Minjae Kim,, Mihwa Kim, Vivette D. D Agati and H. Thomas Lee Department of Anesthesiology, College of Physicians and Surgeons of Columbia University, New York, New York, USA and Department of Pathology, College of Physicians and Surgeons of Columbia University, New York, New York, USA The roles of sphingosine kinases SK and SK in ischemia reperfusion injury have not been fully elucidated since studies have found beneficial effects of SK while others showed no role in this injury. To help resolve this, we used SK or SK knockout mice and confirmed that renal ischemia reperfusion injury induced SK, but not SK, in the kidneys. Furthermore, knockout or pharmacological inhibition of SK increased injury after renal ischemia reperfusion injury. In contrast, lack of SK conferred renal protection following injury. In addition, we used lentiviral gene delivery to selectively express enhanced green fluorescent protein (EGFP) or human SK coexpressed with EGFP (EGFP-huSK) in the kidney. Mice with kidney-specific overexpression of EGFP-huSK had significantly improved renal function with lower plasma creatinine, renal necrosis, apoptosis, and inflammation. Moreover, EGFP-huSK overexpression in cultured human proximal tubule (HK-) cells protected against peroxide-induced necrosis. Selective overexpression of EGFP-huSK led to increased HSP7 mrna and protein expression in vivo and in vitro. Functional protection as well as induction of HSP7 with EGFP-huSK overexpression in vivo was blocked with sphingosine-- phosphate- receptor (SP ) antagonism. Thus, our findings suggest that SK is renoprotective by SP activation and perhaps HSP7 induction. Kidney-specific expression of SK through lentiviral delivery may be a viable therapeutic option to attenuate renal ischemia reperfusion injury. Kidney International () 8, 5 7; doi:.8/ki..8; published online 7 August KEYWORDS: acute kidney injury; apoptosis; heat shock protein 7; inflammation; necrosis; sphingosine--phosphate Correspondence: H. Thomas Lee, Department of Anesthesiology, Anesthesiology Research Laboratories, Columbia University, P&S Box 6 (PH-5), 6 West 68th Street, New York, New York -78, USA. tl8@columbia.edu These authors contributed equally to this work. Received January ; revised June ; accepted 5 June ; published online 7 August Renal ischemia reperfusion injury (IRI) is a significant cause of acute kidney injury (AKI), a major clinical problem often leading to multiorgan dysfunction and systemic inflammation. The development of AKI is associated with increased morbidity and mortality among patients in the intensive care setting, and AKI requiring renal replacement therapy is an independent predictor of mortality after cardiac surgery. Despite continued research on renal protective agents, there are no proven therapies to reduce AKI in the perioperative setting. Sphingosine kinase (SK) is an important regulator of cellular signaling pathways and exists in two isoforms. 5,6 SK leads to the formation of sphingosine--phosphate (SP), a lysophospholipid functioning as both an extracellular ligand for specific G- protein-coupled receptors and an intracellular second messenger in promoting cell growth and survival and the inhibition of apoptosis. 5,7 Activation of the SP receptor (SP )hasbeen showntobeprotectiveincardiac 8 and renal IRI, 9 but the roles of SK and SK in IRI are controversial and have yet to be fully elucidated. A recent study suggests that constitutive expression of SK may contribute to reduced injury with no roleforskafterrenaliri. However, in acute lung injury, SK was protective and SK led to increased injury. Furthermore, in cardiac IRI, SK activation has a potent cytoprotective role. In this study, we determined the specific role of SK enzyme subtypes in renal IRI. We tested the hypothesis that SK activation is renoprotective and that selective renal overexpression of SK would improve renal function and reduce necrosis, inflammation, and apoptosis after IRI in mice. We used genetically modified strains of mice lacking SK (SK / ) and SK (SK / ) enzyme to study the effects of SK subtypes in renal IRI. In addition, we used intrarenal gene delivery in mice with a lentiviral vector that expresses either enhanced green fluorescent protein (EGFP) or human SK coexpressing EGFP (EGFP-huSK), as well as cultured human proximal tubule (HK-) cells infected with EGFP or EGFP-huSK lentivirus. RESULTS Differential induction of SK activity and mrna regulation after renal IRI We initially measured baseline renal SK activity in wild-type (WT), SK /, and SK / sham-operated mice (Figure a). Kidney International () 8, 5 7 5

2 SW Park et al.: Sphingosine kinase protects in renal ischemia a Sphingosine kinase activity (fold over WT sham) b 6 5 WT SK / SK / c Plasma Cr (mg/dl) WT SK / SK / min IRI min IRI GAPDH d Wild type SK / min IRI msk msk mrna/gapdh over WT sham WT sham SK / sham SK / sham 5 WT renal IRI SK / renal IRI msk mrna/gapdh over WT sham 5 SK / renal IRI WT SK / SK / msk e SK / Plasma Cr (mg/dl) Vehicle SKI-II CAY6 min IRI min IRI Figure Renal ischemia reperfusion injury (IRI) upregulates sphingosine kinase (SK)-, but not SK, in the kidney. (a) Relative SK activity (fold over wild-type (WT) sham) from kidneys of WT, SK /, and SK / mice subjected to sham operation or min of left renal ischemia, followed by h of reperfusion (renal IRI). N ¼ 5 6 per group. (b) Representative gel images and densitometric quantification of relative band intensities normalized to glyceraldehyde -phosphate dehydrogenase (GAPDH) of semiquantitative reverse transcriptase-pcr (RT-PCR; of four experiments) of murine (m)-sk, msk, and GAPDH from the kidneys of WT, SK /, and SK / mice after sham operation or renal IRI. (c) Plasma creatinine (mg/dl) was measured in WT, SK /, and SK / mice after sham operation, min of renal ischemia ( min IRI), or min of renal ischemia ( min IRI), followed by h of reperfusion (N ¼ 6 8 per group). (d) Representative photomicrographs of four experiments (hematoxylin and eosin staining, original magnifications and ) with WT (top panels), SK / (middle panels), and SK / (bottom panels) mice subjected to min of renal IRI. (e) Plasma creatinine in mice pretreated with vehicle or the SK inhibitors SKI-II (5 mg/kg subcutaneously (s.c.)) or CAY6 ( mg/kg intravenously (i.v.)) and subjected to sham operation, min of IRI, or min of IRI, followed by h of reperfusion (N ¼ 6 per group). For all graphs: Po. vs. WT sham group; Po. vs. respective sham groups; Po. vs. respective WT or vehicle IRI group; w Po.5 vs. respective WT or vehicle IRI group; z Po. vs. respective SK / IRI group. Data are presented as mean±s.e.m. SK / and SK / mice are congenically derived on a C57BL/6 background. Compared with WT mice, SK / mice had similar renal SK activity at baseline, whereas SK / mice had a large decrease (approximately threefold) in baseline renal SK activity. With min of renal ischemia and h of reperfusion (renal IRI), renal SK activity markedly increased (B 5-fold) in WT mice and SK / mice compared with sham-operated mice. In contrast, SK / mice subjected to renal IRI had no increase in renal SK activity compared with SK / sham-operated mice. These results demonstrate that there was a significant and selective upregulation of SK and not SK activity in mice in response to renal IRI. Similarly, there were increased SK mrna levels in WT and SK / mice after renal IRI compared with shamoperated mice (Figure b, representative of four experiments). However, SK mrna levels in WT and SK / mice were unchanged after renal IRI compared with shamoperated mice, further confirming selective renal SK upregulation in response to renal IRI. Differential roles of SK and SK in renal injury after IRI Baseline plasma creatinine (Cr) values (mice not subjected to anesthesia and surgery) were similar between WT (Cr ¼.±. mg/dl, N ¼ ), SK / (Cr ¼.±. mg/dl, N ¼ ), and SK / (Cr ¼.±. mg/dl, 6 Kidney International () 8, 5 7

3 SW Park et al.: Sphingosine kinase protects in renal ischemia N ¼ ) mice. Plasma creatinine levels were also similar (although slightly elevated compared with mice not subjected to surgery) in sham-operated (anesthesia, laparotomy, right nephrectomy, and recovery) WT, SK /, and SK / mice (Figure c). Plasma creatinine increased in WT, SK /, and SK / mice subjected to moderate ( min) or severe ( min) renal IRI compared with sham-operated mice. After both moderate and severe renal IRI, SK / mice had significantly increased renal injury compared with WT or SK / mice with higher plasma creatinine levels. In contrast, SK deficiency provided renal protection as SK / mice had significantly lower plasma creatinine levels compared with WT and SK / mice after moderate or severe renal IRI. Figure d demonstrates histological injury in mice after min of renal IRI. Compared with shamoperated mice (data not shown), the kidneys of WT mice developed mild injury with some tubular vacuolization and proteinaceous casts (Figure d, top panels). SK / mice developed more severe injury with increased tubular dilatation and congestion, vacuolization, and simplification (Figure d, middle panels). SK / mice had markedly reduced injury compared with SK / mice (Figure d, bottom panels). Because of the inherent concerns regarding the use of genetic knockout mice (for example, alterations in the expression of unrelated proteins), we treated some WT mice with a selective but nonspecific SK inhibitor (SKI-II) or an SK-specific inhibitor (CAY6) before induction of renal IRI. We determined that SKI-II inhibited mouse kidney SK activity in a dose-dependent manner (inhibition of 5±% and 85±% at 5 and 5 mg/ml, respectively, N ¼ ), and preliminary experiments (N ¼ ) showed that the dose of SKI-II used in our study reduced plasma SP levels in mice by 7%. In addition, vehicle, SKI-II, or CAY6 administration had no detrimental effects on renal function in shamoperated mice (Figure e). Mice subjected to moderate ( min) or severe ( min) renal IRI with SKI-II or CAY6 treatment had increased plasma creatinine levels compared with vehicle-treated mice (Figure e), providing further evidence that blocking the induction of SK activity worsened renal function after IRI. We also demonstrate that administration of a selective SP agonist, SEW87 ( mg/kg p.o. h before ischemia and 9 h after reperfusion), protected mice against renal IRI. The vehicle used to administer SEW87 (% dimethylsulfoxide, 5% Tween-) had no effects on renal function in sham-operated mice (Cr ¼.5±. mg/dl, N ¼ ), but caused an increase in plasma creatinine after renal IRI compared with mice not treated with vehicle (Cr ¼.8±.7 mg/dl, N ¼ 7 vs..5±. mg/dl, N ¼ 8, Po.). However, when mice were treated with SEW87, there was a significant decline in plasma creatinine (Cr ¼.±. mg/dl, N ¼ 8, Po. vs. vehicle IRI). Taken together, our results indicate that SK has protective effects (most likely via SP activation), whereas SK contributes to worsening renal injury after renal IRI. In vivo renal expression of EGFP or EGFP-huSK after intrarenal lentiviral gene delivery in WT mice We used lentiviral constructs encoding EGFP or EGFP-huSK to selectively express these gene products in the kidneys of WT mice. Intrarenal lentivirus injection had no adverse effects in mice. As previously demonstrated, 6 intrarenal delivery of lentivirus encoding EGFP or EGFP-huSK resulted in robust expression of EGFP in the corticomedullary junction (primarily S segments of proximal tubules as the major tubular segment represented) at 8 h after injection (Figure a, left panels, representative of four experiments) with minimal EGFP expression (similar to baseline green autofluorescence) in the contralateral (noninjected) kidney (data not shown). To confirm that the cells expressing EGFP-SK protein are proximal tubules, we performed Phaseolus vulgaris leucoagglutinin lectin staining in kidney sections. The brush border of proximal tubules stained heavily with P. vulgaris leucoagglutinin lectin (red, Figure a, middle panels), and the merged images demonstrate that nearly all of the EGFP-expressing cells are proximal tubules (Figure a, right panels). Brush-border stains were lush, consistent with S segment morphology. We confirmed the expression of EGFP and husk mrna in mice renally injected with EGFP and EGFP-huSK lentivirus using reverse transcriptase-pcr. Figure b (representative of four experiments) shows that the kidneys of mice injected with a lentiviral construct all expressed EGFP mrna. In contrast, only EGFP-huSK-injected mice expressed husk mrna. We failed to detect EGFP or husk mrna in the contralateral (noninjected) kidneys of lentivirus-injected mice (Figure b) as demonstrated previously. 6 There were similar SK activity levels in the kidneys of mice not subjected to lentiviral injection (data not shown), the contralateral (noninjected) kidneys (Figure c), or the kidneys renally injected with EGFP lentivirus (Figure c). However, when mice were renally injected with EGFP-huSK lentivirus, there was a significant increase in renal SK activity, confirming that the lentiviral construct encoded functional husk enzyme (Figure c). We measured SP levels in the injected kidneys (cortex and corticomedullary junction tissue) and plasma of mice after lentiviral injection using high-performance liquid chromatography. Compared with kidneys injected with EGFP lentivirus, there was a significant increase in renal SP levels after renal injection with EGFP-huSK lentivirus (SP level ¼.7±.-fold compared with EGFP lentivirus-injected samples, N ¼, Po.5). However, plasma levels of SP were not different in mice injected with EGFP lentivirus compared with mice injected with EGFP-huSK lentivirus (SP level ¼.95±.8-fold compared with EGFP lentivirus-injected samples, N ¼ ). Renal protection with EGFP-huSK expression in mice Mice renally injected with EGFP or EGFP-huSK lentivirus and subjected to sham operation had normal plasma creatinine similar to mice without lentiviral injection Kidney International () 8, 5 7 7

4 SW Park et al.: Sphingosine kinase protects in renal ischemia a EGFP-huSK LV-injected kidney EGFP Proximal tubule Merge C C C CMJ CMJ CMJ M M M Corticomedullary junction ( 6) b EGFP LV EGFPhuSK LV EGFPhuSK LV Noninjected kidney EGFP LV Injected kidney EGFP husk GAPDH Sphingosine kinase activity (fold over noninjected) EGFP-huSK LV Noninjected Injected Figure Kidney-specific reconstitution of lentivirus (LV) encoding enhanced green fluorescent protein (EGFP) human sphingosine kinase (husk) increases renal SK activity. (a) Representative kidney fluorescent photomicrographs of four experiments from mice 8 h after LV injection (frozen section, original magnification of (top panels) and 6 (bottom panels)) demonstrating predominant localization of EGFP expression (left panels) in the cortex (C) and corticomedullary junction (CMJ) with relative sparing of the medullary areas (M). The brush border of proximal tubules was stained using Phaseolus vulgaris leucoagglutinin lectin conjugated to a red fluorescent dye, demonstrating primarily S segments in this region (middle panels), and the merged images demonstrate that the majority of EGFP-expressing cells are proximal tubules (right panels). (b) Representative gel images of reverse transcriptase-pcr (RT-PCR; of four experiments) of EGFP, husk, and glyceraldehyde -phosphate dehydrogenase (GAPDH) from the injected and noninjected kidneys of sham mice renally injected with EGFP or EGFP-huSK LV 8 h before sample collection. (c) Relative SK activity (fold over noninjected kidney) from the injected and noninjected kidneys of mice renally injected with LV encoding EGFP or EGFP-huSK (N ¼ 6 to 7 per group). Po. vs. -injected group. Data are presented as mean±s.e.m. c (Figure ). Mice without lentiviral injection or mice renally injected with EGFP lentivirus 8 h prior had similar and significant increases in plasma creatinine after min of renal ischemia and h of reperfusion compared with shamoperated mice. In contrast, mice renally injected with lentivirus encoding EGFP-huSK showed significant renal protection after renal IRI. The protective effects of EGFPhuSK lentivirus were mediated via SP as treatment with an SP -specific antagonist (W6) given before renal IRI reversed the protection seen in mice renally injected with EGFP-huSK lentivirus (Figure ). Mice renally expressing EGFP-huSK show less renal tubular necrosis after renal IRI Compared with sham-operated mice (Figure a, representative of six experiments), there was significant renal injury in mice subjected to renal IRI (Figure b) and in mice renally injected with EGFP lentivirus 8 h before renal IRI (Figure c), as evidenced by severe tubular necrosis, corticomedullary congestion and hemorrhage, and development of proteinaceous casts. However, mice renally injected with lentivirus encoding EGFP-huSK had reduced renal tubular damage and necrosis after renal IRI (Figure d) 8 Kidney International () 8, 5 7

5 SW Park et al.: Sphingosine kinase protects in renal ischemia Plasma Cr (mg/dl) No virus injection EGFP-huSK LV W6 Figure Kidney-specific reconstitution of lentivirus (LV) encoding enhanced green fluorescent protein (EGFP) human sphingosine kinase (husk) protects mice against renal ischemia reperfusion injury (IRI) via activation of the sphingosine--phosphate--receptor (SP ). Plasma creatinine (Cr; mg/dl) was measured in mice without LV injection or in mice renally injected with LV encoding EGFP or EGFP-huSK after sham operation or min of left renal ischemia, followed by h of reperfusion (renal IRI). Some mice were treated with the selective SP antagonist W6 (. mg/kg) 5 min before renal IRI. N ¼ 6 8 per group. Po. vs. respective sham groups. Po. vs. renal IRI group. w Po. vs. EGFP-huSK LV renal IRI group. Data are presented as mean±s.e.m. compared with mice injected with EGFP lentivirus and mice not subjected to lentiviral injection. The Jablonski scale renal injury score histology grading 7 was used to grade renal tubular necrosis after renal IRI (Figure e). Mice injected with EGFP or EGFP-huSK encoding lentivirus and subjected to sham operation developed very mild tubular necrosis compared with shamoperated mice not subjected to lentiviral injection. Renal ischemia for min and reperfusion for h resulted in severe acute tubular necrosis in mice subjected to renal IRI and in mice injected with lentivirus encoding EGFP 8 h before renal IRI. In contrast, mice renally injected with lentivirus encoding EGFP-huSK 8 h before renal IRI had significantly lower renal injury scores compared with mice injected with lentivirus encoding EGFP and mice not subjected to lentiviral injection (Figure e). Mice renally expressing EGFP-huSK show reduced proinflammatory gene expression in the kidney after renal IRI There were no differences in the mrna expressions of tumor necrosis factor-a, intercellular adhesion molecule-, monocyte chemoattractive protein-, macrophage inflammatory protein-, and interleukin-6 between sham-operated mice renally injected with EGFP lentivirus or EGFP-huSK lentivirus (Figure 5). We found significantly increased mrna expression of these proinflammatory genes in mice subjected to renal IRI after renal injection with EGFP lentivirus compared with sham-operated mice. However, mice renally injected with EGFP-huSK lentivirus before renal IRI had significantly reduced expression of these proinflammatory renal IRI Renal injury score No virus injection EGFP-huSK LV EGFP-huSK LV renal IRI Figure Reduced kidney injury with lentiviral gene delivery of human sphingosine kinase (husk)- after renal ischemia reperfusion injury (IRI). Representative photomicrographs of six experiments (hematoxylin and eosin staining, original magnification ) with mice subjected to (a) sham operation or (b) renal IRI, or mice renally injected with (c) enhanced green fluorescent protein (EGFP) or (d) EGFP-huSK lentivirus (LV) 8 h before renal IRI. Pictures of the outer medulla of the kidneys are shown. (e) Jablonski scale renal injury score in mice without LV injection or mice renally injected with lentivirus encoding EGFP or EGFP-huSK after sham operation or renal IRI. Note that the no-virus-injection sham group renal injury score ¼ ±. Po. vs. respective sham group. Po. vs. no-virus renal IRI and renal IRI group. Data are presented as mean±s.e.m. mrnas compared with EGFP lentivirus-injected mice undergoing renal IRI. Mice renally expressing EGFP-huSK show reduced renal neutrophil infiltration after renal IRI Compared with sham-operated mice (Figure 6a, representative of four experiments), there was an increase in neutrophil infiltration in mice subjected to renal IRI (Figure 6b) and in mice renally injected with EGFP lentivirus 8 h before renal IRI (Figure 6c). In contrast, mice renally injected with lentivirus encoding EGFP-huSK had reduced neutrophil infiltration after renal IRI (Figure 6d and e). Kidney International () 8, 5 7 9

6 SW Park et al.: Sphingosine kinase protects in renal ischemia TNF-α mrna/gapdh over sham EGFP-huSK LV MCP- mrna/gapdh over sham IL-6 mrna/gapdh over sham ICAM- mrna/gapdh over renal IRI MIP- mrna/gapdh over sham Mice renally expressing EGFP-huSK show reduced renal apoptosis after renal IRI We failed to detect significant terminal deoxynucleotidyl transferase biotin-dutp nick end labeling (TUNEL)-positive cells in kidney sections from sham-operated mice (Figure 7a, representative of four experiments) or in mice subjected to sham operation after renal injection with EGFP or EGFPhuSK lentivirus (data not shown). Mice subjected to renal IRI (Figure 7b) or mice renally injected with EGFP lentivirus before renal IRI (Figure 7c) showed many TUNEL-positive cells in the corticomedullary junction. However, mice renally injected with lentivirus encoding EGFP-huSK had reduced TUNEL-positive renal tubule cells (Figure 7d and e). Selective renal expression of husk upregulates HSP7 in the mouse kidney Renal expression of EGFP-huSK increased renal heat shock protein 7 (HSP7) mrna (Figure 8a, representative of four experiments) and protein (Figure 8b, representative of four ND Figure 5 Lentivirus-mediated expression of human sphingosine kinase (husk)- reduces proinflammatory gene expression after renal ischemia reperfusion injury (IRI). Densitometric quantification of relative band intensities normalized to glyceraldehyde -phosphate dehydrogenase (GAPDH; fold over enhanced green fluorescent protein (EGFP) lentivirus (LV) sham group) from semiquantitative reverse transcriptase-pcr (RT-PCR) reactions for proinflammatory mrnas from C57BL/6 mice renally injected with EGFP or EGFP-huSK LV before sham operation or renal IRI (N ¼ 5 per group). Intercellular adhesion molecule- (ICAM-) mrna expression was not detected (ND) in sham mice, and hence values are expressed as fold over the renal IRI group. Abbreviations: IL-6, interleukin-6; MCP-, monocyte chemoattractant protein ; MIP-, macrophage inflammatory protein ; TNF-a, tumor necrosis factor-a. Po. vs. respective sham group. Po.5 vs. respective sham group. Po. vs. renal IRI group. w Po. vs. renal IRI group. z Po.5 vs. renal IRI group. Data are presented as mean±s.e.m. Neutrophils/hpf renal IRI EGFP-huSK LV renal IRI No virus injected EGFP-huSK LV Figure 6 Reduced neutrophil infiltration with lentiviral (LV) gene delivery of human sphingosine kinase (husk)- after renal ischemia reperfusion injury (IRI). Representative photomicrographs of four experiments ( ) of immunohistochemistry for neutrophil infiltration in the outer medulla of the kidneys of mice subjected to (a) sham operation or (b) renal IRI, or mice renally injected with (c) enhanced green fluorescent protein (EGFP) or (d) EGFP-huSK LV 8 h before renal IRI. Secondary antibody conjugated to horseradish peroxidase was developed with diaminobenzidine to stain neutrophils dark brown. (e) Quantification of neutrophils per high-powered field (hpf) from the kidneys of mice without LV injection after renal IRI or mice renally injected with EGFP or EGFP-huSK LV before renal IRI. Po. vs. no-virus-injected renal IRI group. Po. vs. renal IRI group. Data are presented as mean±s.e.m. experiments) expressions in mice. As expected, renal IRI led to an increase in HSP7 expression as EGFP lentivirusinjected mice subjected to renal IRI had higher HSP7 mrna (B-fold) and protein (B-fold) expression than EGFP lentivirus-injected mice subjected to sham operation. Surprisingly, in sham-operated mice, EGFP-huSK lentiviral injection alone led to a sharp increase in HSP7 mrna (B6-fold) and protein (B-fold) compared with mice injected with EGFP lentivirus. Furthermore, induction of HSP7 mrna (B.-fold) and protein (B.-fold) were enhanced even further in EGFP-huSK lentivirus mice subjected to renal IRI compared with mice renally injected with EGFP lentivirus and subjected to renal IRI. Finally, induction of HSP7 with EGFP-huSK lentiviral injection Kidney International () 8, 5 7

7 SW Park et al.: Sphingosine kinase protects in renal ischemia Apoptotic cells/hpf renal IRI EGFP-huSK LV renal IRI No virus injected EGFP-huSK LV Figure 7 Reduced apoptosis with lentiviral (LV) gene delivery of human sphingosine kinase (husk)- after renal ischemia reperfusion injury (IRI). Representative fluorescence photomicrographs (of four experiments) of kidney sections illustrating apoptotic nuclei (terminal deoxynucleotidyl transferase biotin-dutp nick end labeling (TUNEL) fluorescence staining, ) in the outer medulla of the kidneys of mice subjected to (a) sham operation or (b) renal IRI, or mice renally injected with (c) enhanced green fluorescent protein (EGFP) or (d) EGFP-huSK LV 8 h before renal IRI. (e) Quantification of TUNELpositive cells per high-powered field (hpf) from the kidneys of mice without LV injection after renal IRI or mice renally injected with EGFP or EGFP-huSK LV before renal IRI. Po. vs. novirus-injected renal IRI group. Po. vs. renal IRI group. Data are presented as mean±s.e.m. (Figure 8c, mrna and protein) was lost when mice were treated with the SP antagonist W6 before and after EGFP-huSK lentivirus injection. Overexpression of husk protects HK- cells against H O -induced necrosis and upregulates HSP7 Stable overexpression of EGFP-huSK in HK- cells was protective against H O -induced necrosis compared with HK- cells overexpressing EGFP, confirming our previous results. 8 EGFP HK- cells had higher lactate dehydrogenase (LDH) release (% LDH released ¼ 8.±.7, N ¼ ) after h of mmol/l H O treatment compared with EGFP-huSK HK- cells (% LDH released ¼.9±.8, N ¼, Po.). We also found that EGFP-huSK HK- cells had significantly increased HSP7 and SK mrna expression (Figure 9a, representative of four experiments) and protein expression (Figure 9b, representative of four experiments) compared with EGFP HK- cells. DISCUSSION The major findings of this study are as follows: () renal IRI led to significant upregulation of SK mrna expression and activity in the kidney, whereas SK was not affected; () this induction of SK was important in mediating protective effects after renal IRI, as both SK / mice and mice treated with the SK inhibitor SKI-II or the selective SK inhibitor CAY6 had increased renal injury, whereas SK / mice had a reduction in injury; () selective renal overexpression of husk protected mice against renal IRI in vivo via activation of SP ; and () SK overexpression resulted in induction of HSP7 mrna and protein expression in vivo and in vitro. AKI remains a significant clinical problem in the perioperative period, and the development of postoperative AKI requiring renal replacement therapy carries a 6% mortality rate. 9 AKI is an inflammatory process involving multiple cellular and systemic responses, including activation of proinflammatory cytokines and chemokines, and infiltration by leukocytes such as neutrophils, macrophages, and T cells. The inflammatory response is not limited to the kidney, but affects other organ systems, including the liver, intestine, lung, and heart. Multiorgan dysfunction associated with AKI leads to greatly increased morbidity and mortality. 5 Our studies demonstrate a renal protective role for SK, and intrarenal lentiviral delivery of SK may be a therapeutic option to protect against renal IRI. SK, the enzyme catalyzing the conversion of sphingosine to SP, is an important regulator of the putative sphingolipid rheostat that balances the cytoprotective effects of SP against the proapoptotic effects of ceramide and sphingosine. 7,6 SP binds to specific G-protein-coupled receptors, of which five are known (SP 5 ). 7 The exact roles for the two subtypes of SK in tissue injury and survival remain controversial. Of the two known isoforms, SK is primarily known to be cytoprotective, promoting cell survival via the generation of SP that signals in an autocrine/paracrine manner to mediate SP-receptor-activated signaling. 7 The role of SK is not as clear as some studies demonstrate a proapoptotic and antiproliferative role for SK, 8 which may function independently of SP receptors, whereas other studies show an antiapoptotic role for SK, as inhibiting SK or suppressing its expression enhanced apoptosis in various cancer cells. Complicating matters further, some experimental models demonstrate a proinflammatory role of SK, as SK inhibition or mice lacking Kidney International () 8, 5 7

8 SW Park et al.: Sphingosine kinase protects in renal ischemia GAPDH HSP7 EGFP LV sham EGFPhuSK LV sham renai IRI EGFP-huSK LV renai IRI GAPDH mhsp7 mrna/gapdh over sham 6 EGFP-huSK LV veh sham veh sham EGFP-huSK LV veh sham EGFP-huSK LV veh sham W6 LV sham W6 LV sham EGFP-huSK LV W6 sham EGFP-huSK LV W6 sham HSP7 β-actin HSP7 sham HSP7 protein/β-actin over sham EGFP-huSK LV sham renal IRI EGFP-huSK LV EGFP-huSK LV renal IRI β-actin HSP7 mhsp7 mrna/gapdh over vehicle Vehicle W6 HSP7 protein/β-actin over vehicle Vehicle EGFP-huSK LV Figure 8 Selective renal expression of human sphingosine kinase (husk)- upregulates heat-shock protein (HSP)-7 expression. (a) Representative gel images (of four experiments) and densitometric quantification of relative band intensities normalized to glyceraldehyde -phosphate dehydrogenase (GAPDH) of semiquantitative reverse transcriptase-pcr (RT-PCR) of mouse HSP7 and GAPDH from the renal cortices of C57BL/6 mice renally injected with enhanced green fluorescent protein (EGFP) or EGFP-huSK lentivirus (LV) and subjected to sham operation or renal IRI. (b) Representative immunoblot images (of four experiments) and densitometric quantification of relative band intensities normalized to b-actin from immunoblot images of mouse HSP7 and b-actin from the renal cortices of C57BL/6 mice renally injected with EGFP or EGFP-huSK LV and subjected to sham operation or renal IRI. (c) Representative gel (RT-PCR) and immunoblot images (of four experiments) of mouse HSP7 and GAPDH (RT-PCR) or HSP7 and b-actin (immunoblot) from the renal cortices of C57BL/6 mice renally injected with EGFP or EGFP-huSK LV and treated with vehicle (veh) or W6 (. mg/kg) 5 min before and h after lentivirus injection. Po. vs. sham or vehicle sham groups. z Po.5 vs. sham group. Po. vs. respective sham group. w Po.5 vs. respective sham group. Po. vs. renal IRI group. Data are presented as mean±s.e.m. W6 SK enzyme had reduced inflammation and colon damage in a model of Crohn s disease, and blockade of SK reduced the inflammatory response in endotoxin-mediated sepsis. 5 Despite these differences, there is redundancy between SK and SK as both SK / and SK / mice have no gross phenotypic abnormalities, whereas mice lacking both SK and SK are not viable. 6,7 The regulation of SK varies depending on the organ. Cardiac IRI downregulated SK activity but did not affect SK in the hearts of mice. In the lung, both SK and SK were upregulated after lipopolysaccharide-induced acute lung injury. 8 The role that each SK subtype has in renal IRI is not fully elucidated. We have previously demonstrated that the volatile anesthetic isoflurane increased plasma membrane translocation of SK to mediate protection against renal IRI in mice. 9 In contrast, a recent study by Jo et al. suggests that although SK was upregulated after renal IRI, mice lacking SK did not have increased renal dysfunction after renal IRI. On the contrary, a lack of SK resulted in increased renal injury and neutrophil infiltration after renal IRI. We confirm in this study that only SK is upregulated in the kidney after renal IRI. However, we demonstrate a protective role for endogenous SK and a detrimental role for endogenous SK in both models of moderate ( min) and severe renal ischemia ( min). We used a model of contralateral nephrectomy with unilateral ischemia, whereas Jo et al. used a model of bilateral ischemia. The model of AKI used may result in different outcomes, as it was previously demonstrated that unilateral or bilateral ischemia resulted in different effects on subsequent resistance of proximal tubules to hypoxia. Differences in anesthetic agents used (pentobarbital in our study vs. ketamine, xylazine, and acepromazine in study by Jo et al. ) may impact the outcome after renal IRI as several anesthetics (for example, volatile anesthetics) are known to modulate sphingolipid metabolism. 9, It is possible that operative conditions Kidney International () 8, 5 7

9 SW Park et al.: Sphingosine kinase protects in renal ischemia HSP7 mrna/gapdh over EGFP HSP7 protein/β-actin over EGFP 75 kda 5 kda 5 kda EGFP HK- HK- EGFP HK- HK- EGFP-huSK HK- GAPDH HSP7 SK EGFP SK mrna/gapdh over EGFP and/or immunomodulatory effects of anesthetics may affect the outcome of lentiviral overexpression as a therapeutic modality. In addition, the SK / mice that we used and the SK tr/tr mice from the Jo et al. study were obtained from different sources 7, and there may be subtle differences in local laboratory conditions that may affect outcome. It will be necessary to reconcile the differences between our models in future studies. Our findings are consistent with other models of inflammation as mice lacking SK had increased injury with ischemic preconditioning in cardiac IRI and mice EGFP-huSK HK- SK protein/β-actin over EGFP EGFP-huSK HK- β-actin HSP7 SK-b SK-a EGFP EGFP-huSK HK- Figure 9 Overexpression of human sphingosine kinase (husk)- in cultured human proximal tubule (HK-) upregulates heat-shock protein (HSP)-7 expression. (a) Representative gel images (of four experiments) and densitometric quantification of relative band intensities normalized to glyceraldehyde -phosphate dehydrogenase (GAPDH) of semiquantitative reverse transcriptase-pcr (RT-PCR) of HSP7, husk, and GAPDH from HK- cells infected with lentivirus encoding enhanced green fluorescent protein (EGFP HK-) or EGFP-huSK (EGFP-huSK HK-). (b) Representative immunoblot images (of four experiments) and densitometric quantification of relative band intensities normalized to b-actin from immunoblot images of HSP7, SK, and b-actin from EGFP HK- or EGFP-huSK HK- cells. Po.5 vs. EGFP HK- group. Po. vs. EGFP HK- group. Data are presented as mean±s.e.m. lacking SK enzyme had poor recovery from anaphylaxis with delayed histamine clearance, whereas mice lacking SK enzyme had rapid recovery from anaphylaxis. Overexpression of SK via adenoviral vectors has been used to demonstrate the protective effects of SK in cardiac IRI and acute lung injury. 8 We used a lentiviral vector to selectively and stably express the desired gene products (EGFP and EGFP-huSK) in the renal cortex and corticomedullary junction of kidneys, 6 as well as in HK- cells. 8 There is selective protein expression in injected kidneys when compared with noninjected kidneys, and mice injected with a lentiviral construct show robust expression of the targeted gene products. 6 The main site of injury in our model of warm renal IRI is the corticomedullary junction, composed mainly of S segments of the proximal tubule 5,6 and HK- cells correspond to the cell type in the kidney that is most susceptible to necrosis after renal IRI. 7,8 Using P. vulgaris leucoagglutinin lectin staining to identify proximal tubule brush border, we demonstrated that the majority of cells expressing EGFP were S segments of the proximal tubule, the cells most damaged after ischemic injury. Therefore, using our method of lentiviral gene delivery, we were able to directly test the hypothesis that selective expression of husk in the kidney and in HK- cells protects against renal IRI in vivo and H O -induced necrosis in vitro. Selective expression of EGFP-huSK protected against renal IRI as evidenced by a marked reduction in plasma creatinine levels. The beneficial effects of renal EGFP-huSK expression were dependent on SP activation, as mice treated with an SP antagonist were not protected against renal IRI after EGFP-huSK lentivirus injection. Indeed, our current and previous studies have implicated the renal protective role of the SP, as SP or SP agonists such as FTY7 or SEW87 were protective in in vivo models of renal IRI via peripheral lymphopenia 9 and a reduction in early renal T-cell infiltration. 9 However, a recent study showed that the protective effects of SP agonism occurred independently of lymphocytes through direct effects on renal proximal tubule epithelial cells via activation of the cytoprotective Akt/ERK pathways. 5 Selective expression of EGFP-huSK in the kidneys of mice led to an increase in SK activity in the kidney, which resulted in a localized increase in kidney SP levels, but not in plasma SP levels. Increased renal SP after SK lentiviral transduction could have decreased leukocyte-mediated renal injury in addition to providing the direct cytoprotective effects on kidney proximal tubule cells via SP activation after IR. However, it is possible that, as our targeted gene product was of human origin, cellular regulation of husk may be different from regulation of endogenous murine SK. The renal protection occurring with EGFP-huSK lentiviral injection was associated with a decrease in the renal expression of proinflammatory cytokines, including tumor necrosis factor-a, intercellular adhesion molecule-, monocyte chemoattractive protein-, macrophage inflammatory protein-, and interleukin-6. In addition, there was reduced Kidney International () 8, 5 7

10 SW Park et al.: Sphingosine kinase protects in renal ischemia neutrophil infiltration, as well as significantly less renal tubular necrosis of the outer stripe of the outer medulla, and reduced renal tubular apoptosis. Consistent with the finding of increased plasma creatinine in SK / mice, there were significant increases in the kidney expression of some proinflammatory mrnas in SK / mice after renal IRI compared with WT mice (data not shown). We also confirmed our previous finding that HK- cells overexpressing husk were protected against H O -induced necrosis. 8 In our study, selective overexpression of husk strongly induced HSP7 in the kidney and in HK- cells, which was reversed with SP blockade, suggesting an additional potential mechanism of cytoprotection with SK overexpression. As we used an overexpression model of husk, we must be cautious in implicating this pathway as the only protective mechanism in the physiology of the normal mouse. It is possible that HSP7 may be one of several protective factors that the SK-SP pathway regulates to induce renal protection after IRI. HSPs are a group of molecular chaperones that are upregulated after various cellular stresses, including heat shock, hypoxia, ischemia, and toxin exposures. 5,5 Increased HSP7 expression protects a cell from injury or death by functioning as a chaperone to facilitate proper polypeptide folding, assembly of oligomers, transport, and conformational changes. We recently demonstrated that selective renal overexpression of HSP7 in the kidney protected mice against renal IRI by increasing F-actin cytoskeletal preservation. 5 In addition, translocation of SK to the plasma membrane has been shown to be coordinated with the actin cytoskeleton in macrophages. 5 The induction of HSP7 by SK could enhance cytoskeletal integrity allowing translocation and activation of SK. SP has been shown to induce HSP7 via the p8 mitogenactivated protein kinase 5 and the phosphatidylinositol -kinase/akt 55,56 pathways in osteoblasts. Further study will be necessary to determine the precise relationship between SK and HSP7 in mediating renal protection after IRI. In conclusion, we show in this study that endogenous SK had protective effects, whereas endogenous SK had deleterious effects in a model of murine renal IRI. In addition, we show that selective renal expression of the EGFP-huSK transgene via lentiviral vector is possible in the mouse kidney, and that this selective renal overexpression protects mice against renal IRI via activation of SP and perhaps via induction of HSP7. These findings may have important clinical implications, as they imply that kidney-specific expression of SK through lentiviral delivery is a viable therapeutic option in attenuating the effects of renal IRI. MATERIALS AND METHODS Materials SKI-II [-[[-(-Chlorophenyl)--thiazolyl]amino]phenol] and SEW87 (5-[-Phenyl-5-(trifluoromethyl)--thienyl]--[-(trifluoromethyl)phenyl]-,,-oxadiazole) were purchased from Tocris Bioscience (Ellisville, MO). CAY6 [,-dimethyl-s- (-oxo--hexadecyn--yl)-,-dimethylethyl ester--oxazolidinecarboxylic acid] was purchased from Cayman Chemical (Ann Arbor, MI). W6 [(R)--Amino-(-hexylphenylamino)--oxobutylphosphonic acid] was purchased from Avanti Polar Lipids (Alabaster, AL). Unless otherwise specified, all other reagents were purchased from Sigma (St Louis, MO). Intrarenal lentivirus delivery in mice Lentivirus encoding EGFP-huSK was generated by subcloning husk cdna into a modified shuttle vector CMV-pLL.7, in which the insert expression is driven by a CMV promoter, followed by an IRES EGFP sequence for simultaneous coexpression of the EGFP reporter gene and hemagglutinin (HA) tag. When expressed, this vector yields two independent proteins, EGFP and the husk protein fused with the HA tag. Lentivirus was produced by a triple transfection of CMV-huSK-pLL.7 or CMV-pLL.7 ( mg), pvsvg (Vesicular stomatitis virus G; Invitrogen, Carlsbad, CA; 7 mg), and pd8.9 (from Dr Van Parjs, MIT, Cambridge, MA; 5 mg) as previously described. 57,58 In vivo virus transduction in mice was performed as described. 6 Determination of transgene expression with intrarenal lentivirus injection We detected the expression of EGFP and EGFP-huSK 8 h after intrarenal injection of lentivirus in mice as described previously. 6 We also identified proximal tubules by labeling kidney cryosections using P. vulgaris leucoagglutinin (PHA-L) lectin as described previously. 59 PHA-L lectin binds avidly to the brush border of proximal tubules. 6 Briefly, fixed cryosections were treated with.% H O in methanol for min, % bovine serum albumin for min, and then PHA-L conjugated to a fluorescent dye (Alexa Fluor 59, Invitrogen) for h at room temperature. Overlap of proximal tubules (red) with EGFP expression (green) was examined with an Olympus IX8 (Center Valley, PA) epifluorescence microscope and analyzed with Slidebook software (Intelligent Imaging Innovations, Denver, CO). We also extracted total RNA from renal cortices and performed reverse transcriptase-pcr assays for EGFP, GAPDH (glyceraldehyde -phosphate dehydrogenase), and husk as described previously 58,6 (Table ). In addition, we confirmed the functional nature of the transgene product by measuring total SK activity as described previously. 9 Briefly, kidney tissues from mice were collected and homogenized in cell lysate buffer ( mmol/l sucrose, mmol/l EGTA, mmol/l MOPS, ph 7., 5% Percoll,.% digitonin, protease (Calbiochem, La Jolla, CA), and phosphatase inhibitors) on ice. After a g spin for 5 min to pellet cellular debris, protein concentrations were determined and SK activity was measured as described by Vessey et al. 6 Total SK activity was also measured in the kidneys of WT, SK /, and SK / mice after sham operation or renal IRI. To confirm our results, we supplemented the assay buffer with.5% Triton X- or mol/l KCl to create SK- and SK-specific assay buffer conditions, respectively, 6 and we found that the rise in SK activity following renal IRI was entirely due to an increase in SK activity and not SK activity (data not shown). Our results show that mice with only SK enzyme (SK / mice) did not have increased SK activity after renal IRI. In contrast, mice with only SK enzyme (SK / mice) had significantly increased SK activity after renal IRI (5.9±.7- fold, N ¼, Po.5 vs. sham). We measured SP levels in the kidneys and plasma of mice after lentiviral injection using highperformance liquid chromatography as described previously. 6 Kidney International () 8, 5 7

11 SW Park et al.: Sphingosine kinase protects in renal ischemia Table Primers used to amplify mrnas encoding SK, SK, GAPDH, and proinflammatory cytokines based on published GenBank sequences for mice Primer Accession number Sequence (sense/antisense) Product size (bp) Cycle number Annealing temp (C) Human SK NM_97 5 -ATCTCCTTCACGCTGATGC GTGCAGAGACAGCAGGTTCA- Human HSP7 NM_5 5 -CACGAGGAGCGGCAGGACGAG AGTGGCGGCAGCAGGGGTGG- Mouse SK NM_5 (variant ) 5 -GATGCATGAGGTGGTGAATG- 7 6 NM_567 (variant ) 5 -GCCCACTGTGAAACGAATC- Mouse SK NM_8 (variant ) 5 -ACTGCTCGCTTCTTCTCTGC NM_ (variant ) 5 -ACCATTGAGGGACAGGTCAG- Mouse MIP- X CCAAGGGTTGACTTCAAGAAC AGCGAGGCACATCAGGTACG- Mouse ICAM- X56 5 -TGTTTCCTGCCTCTGAAGC CTTCGTTTGTGATCCTCCG- Mouse TNF-a X6 5 -TACTGAACTTCGGGGTGATTGGTCC CAGCCTTGTCCCTTGAAGAGAACC- Mouse MCP- NM_ 5 -ACCTGCTGCTACTCATTCAC TTGAGGTGGTTGTGGAAAAG- Mouse IL-6 NM_68 5 -CCGGAGAGGAGACTTCACAG GGAAATTGGGGTAGGAAGGA- Mouse HSP7 NM_ 5 -CCTAAGGTCTGGCATGGTA AGGAAGCTCGTTGTTGAAGC- GAPDH M ACCACAGTCCATGCCATCAC CACCACCCTGTTGCTGTAGCC- Abbreviations: bp, base pairs; GAPDH, glyceraldehyde -phosphate dehydrogenase; HSP7, heat shock protein 7; ICAM-, intercellular adhesion molecule-; IL-6, interleukin-6; MCP-, monocyte chemoattractant protein ; MIP-, macrophage inflammatory protein ; SK, sphingosine kinase; TNF-a, tumor necrosis factor-a. Respective anticipated RT-PCR product size, PCR cycle number for linear amplification, and annealing temperatures used for each primer are also provided. Murine model of renal IRI All animal protocols were approved by the institutional animal care and use committee of Columbia University (New York, NY). We used male C57BL/6 (Harlan, Indianapolis, IN) or SK / and SK / mice (kindly provided by RL Proia, NIH, Bethesda, MD). The generation and initial characterization of SK / 6 and SK / 7 mice have been described previously. These mice are congenically derived on a C57BL/6 background. In adult ( 5 g) male C57BL/6, SK /, or SK / mice, - or thirty-min left renal IRI was performed under pentobarbital anesthesia as described. 65 For lentiviral experiments, renal IRI was performed 8 h after intrarenal injection of EGFP or EGFP-huSK lentivirus in C57BL/6 mice as described.,5 All mice underwent right nephrectomy at the time of sham operation or renal IRI. To test the effects of SK inhibition, the SK inhibitors SKI-II or CAY6 were administered to mice subjected to renal IRI. SKI-II (5 mg/kg) was administered subcutaneously 5 min before ischemia and h after reperfusion. SKI-II is an SK-selective inhibitor with minimal effects on other kinases, 66 and this dose was shown to have effective inhibition of activity without significant toxicity. 67 CAY6, an SK-specific inhibitor with no inhibitory effect on SK, 68 was administered intravenously ( mg/kg) min before ischemia. To test the effects of SP agonism, a selective SP agonist, SEW87 ( mg/kg), was administered to mice h before ischemia and 9 h after reperfusion. SEW87 was administered via oral gavage in % dimethylsulfoxide/5% Tween- as it has low solubility in aqueous solution. To test whether SP antagonism blocked the renal protective effects of husk overexpression, the SP -specific inhibitor W6 (see ref ) (. mg/kg) was administered intraperitoneally 5 min before sham operation or renal ischemia. To test the effects of SP antagonism on HSP7 mrna and protein induction after husk overexpression, W6 was administered intraperitoneally 5 min before and h after lentivirus injection. SKI-II, CAY6, and W6 were dissolved in % dimethylsulfoxide in water (vehicle). Plasma creatinine level determination Plasma creatinine levels were measured with an enzymatic creatinine reagent kit according to the manufacturer s instructions (Thermo Fisher Scientific, Waltham, MA). Histological detection of necrosis Morphological assessment was performed by an experienced renal pathologist (VDD) who was unaware of the treatment that each animal had received. A grading scale of, as outlined by Jablonski et al., 7 was used to assess the degree of renal tubular necrosis in the outer medullary area after renal IRI as described previously. 7,7 Detection of renal tubular apoptosis after renal IRI We detected apoptosis after renal IRI with TUNEL staining. In situ labeling of fragmented DNA was performed with TUNEL stain (green fluorescence) using a commercially available in situ cell-death detection kit (Roche, Indianapolis, IN) according to the instructions provided by the manufacturer. TUNEL-positive cells per highpowered field were counted to quantify apoptosis. Assessment of renal inflammation Renal inflammation after IRI was determined () with detection of neutrophil infiltration with immunohistochemistry h after renal IRI and () by measuring mrna-encoding markers of inflammation, including interleukin-6, intercellular adhesion molecule-, monocyte chemoattractive protein-, macrophage inflammatory protein-, and tumor necrosis factor-a. Immunohistochemistry for Kidney International () 8, 5 7 5