Thrombospondin-1 Activation of Signal-Regulatory Protein-A Stimulates Reactive Oxygen Species Production and Promotes Renal Ischemia Reperfusion Injury

BASIC RESEARCH www.jasn.org

Thrombospondin-1 Activation of Signal-Regulatory Protein-a Stimulates Reactive Oxygen Species Production and Promotes Renal Ischemia Reperfusion Injury

† ‡ ‡ Mingyi Yao,* Natasha M. Rogers,* Gábor Csányi,* Andres I. Rodriguez,* Mark A. Ross,§ † Claudette St. Croix,§ Heather Knupp,* Enrico M. Novelli,* Angus W. Thomson, ‡ †‡| Patrick J. Pagano,* and Jeffrey S. Isenberg*

*Vascular Medicine Institute, †Starzl Transplantation Institute, ‡Department of Pharmacology and Chemical Biology, §Center for Biologic Imaging, and |Division of Pulmonary, Allergy and Critical Care Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

ABSTRACT Ischemia reperfusion injury (IRI) causes tissue and organ injury, in part, through alterations in tissue blood flow a a and the production of reactive oxygen species. The cell surface receptor signal-regulatory protein- (SIRP- )is BASIC RESEARCH expressed on inflammatory cells and suppresses phagocytosis, but the function of SIRP-a in IRI has not been determined. We reported previously that the matricellular protein thrombospondin-1 is upregulated in IRI. Here, we report a novel interaction between thrombospondin-1 and SIRP-a on nonphagocytic cells. In cell-free experiments, thrombospondin-1 bound SIRP-a. In vascular smooth muscle cells and renal tubular epithelial cells, treatment with thrombospondin-1 led to phosphorylation of SIRP-a and downstream activation of Src homology phox domain 2–containing phosphatase-1. Thrombospondin-1 also stimulated phosphorylation of p47 (an organizer subunit for nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 1/2) and increased production of superoxide, both of which were abrogated by knockdown or antibody blockade of SIRP-a. In rodent aortic rings, treatment with thrombospondin-1 increased the production of superoxide and inhibited nitric oxide– mediated vasodilation in a SIRP-a–dependent manner. Renal IRI upregulated the thrombospondin-1–SIRP-a signaling axis and was associated with increased superoxide production and cell death. A SIRP-a antibody that blocks thrombospondin-1 activation of SIRP-a mitigated the effects of renal IRI, increasing blood flow, suppress- ing production of reactive oxygen species, and preserving cellular architecture. A role for CD47 in SIRP-a activation in these pathways is also described. Overall, these results suggest that thrombospondin-1 binding to SIRP-a on nonphagocytic cells activates NADPH oxidase, limits vasodilation, and promotes renal IRI.

J Am Soc Nephrol 25: 1171–1186, 2014. doi: 10.1681/ASN.2013040433

Thrombospondin-1 (TSP1) is a secreted matricel- Signal regulatory protein-a (SIRP-a)isacell lular protein produced by platelets, endothelial and surface receptor expressed on phagocytic and neu- vascular smooth muscle cells (VSMCs), and non- ronal cells and activated through interactions with vascular cells.1 TSP1 transduces signals from the extracellular to cellular components of tissues Received April 29, 2013. Accepted October 29, 2013. through binding to cell surface receptors, including M.Y. and N.M.R. contributed equally to this work. 2 the integrins, CD36 and CD47. We and others have Published online ahead of print. Publication date available at shown that TSP1 levels are increased in plasma and www.jasn.org. ﬂ in conditions associated with decreased blood ow, Correspondence: Dr. Jeffrey S. Isenberg, University of Pittsburgh such as ischemia reperfusion injury (IRI),3 athero- School of Medicine, Room E1258, 200 Lothrop Street, Pittsburgh, PA sclerosis,4 pulmonary hypertension,5,6 and sickle 15261. Email: [email protected] cell anemia.7 Copyright © 2014 by the American Society of Nephrology

J Am Soc Nephrol 25: 1171–1186, 2014 ISSN : 1046-6673/2506-1171 1171 BASIC RESEARCH www.jasn.org the cell surface protein CD47, by growth factors or integrin alter total SIRP-a protein levels (Figure 1D, densitometry pre- signaling.8,9 SIRP-a controls cell responses through the re- sented). cruitment and phosphorylation of Src homology domain 2– containing phosphatase-1 (SHP1) and -2 (SHP2).10 SIRP-a is TSP1 Activates a Downstream Target of SIRP-a classified as an inhibitory cell receptor, and SIRP-a–mediated Upon phosphorylation, SIRP-a activates the Src homology-2 signaling suppresses macrophage phagocytosis.11 However, (SH2) domain containing protein phosphatases SHP1 and/or little is known about the role of SIRP-a in vascular cells and SHP2.18 We tested whether TSP1 activates these downstream IRI. signal transducers in smooth muscle cells. Arterial VSMCs pre- Loss of nitric oxide (NO) signaling, including decreased NO incubated under growth factor-free and serum-free conditions bioavailability, is a major contributor to cardiovascular dis- (for 24 hours) and treated with TSP1 (2.2 nmol/L) displayed ease.12 NO reacts rapidly with the reactive oxygen species phosphorylation of SHP1 in a temporal fashion similar to that z2 (ROS) superoxide anion (O2 ) which dramatically limits its of SIRP-a (Figure 1F). Treatment of VSMCs with TSP1 did not biologic effect.13 This interaction becomes important after result in SHP2 phosphorylation (Figure 1G) and did not alter ischemia reperfusion, where pathologic ROS production, in- total SHP1 or SHP2 protein levels within the time course of the z2 cluding O2 , is increased. We have shown that TSP1 inhibits experiment (Figure 1, F and G, densitometry presented). NO signaling5 and limits blood flow,14–16 but the exact mechanisms are still unclear. Treatment of VSMCs with the NO Donor Sodium Our data demonstrate that TSP1 stimulates phosphoryla- Nitroprusside Does Not Stimulate Phosphorylation of tion of nonphagocytic SIRP-a and stimulates NADPH oxidase SIRP-a or SHP1 z2 (Nox)–mediated O2 production and that SIRP-a phosphor- The NO donor S-nitroso-N-acetylpenicillamine was reported ylation is absent upon CD47 deletion. In arteries, TSP1 inhib- to activate SHP1,19 and our laboratory reported that TSP1 its NO-mediated vasodilation through SIRP-a–dependent regulates NO signaling in vascular cells.20–22 We were inter- stimulation of ROS. IRI upregulates renal TSP1–SIRP-a sig- ested in exploring what role exogenous NO could play in our naling, increases pathologic ROS production, and promotes studies. We tested this in VSMCs by determining SIRP-a z2 cell death. Disruption of TSP1–SIRP-a signaling inhibits O2 phosphorylation in the presence of sodium nitroprusside production, promotes vasodilation, improves blood flow, and (SNP), a prodrug that is metabolized by VSMCs to NO.23,24 limits IRI. Once again, TSP1 phosphorylated SIRP-a and SHP1, whereas treatment with SNP had no effect on SIRP-a or SHP1 phosphorylation (Supplemental Figure 1, A and B). RESULTS TSP1-Stimulated Phosphorylation of VSMC SIRP-a May TSP1 Engages and Phosphorylates Nonphagocytic Not Completely Depend on CD47 Activation SIRP-a CD47 and SIRP-a are expressed on vascular cells,25 and SIRP- TSP1 can interact with several cell surface receptors,1 including a activation requires CD47 in phagocytic cells.25,26 This con- CD47,aswehavereported.17 However, it is not known clusion is based, in part, on studies using a TSP1-derived pep- whether TSP1 can interact with or signal through SIRP-a. tide (peptide 4N1K, KRFYVVMWKK) that was reported to In arterial VSMC lysates, SIRP-a was coprecipitated by a interact specifically with CD47.27 However, the validity of this TSP1 monoclonal antibody and, conversely, TSP1 was copre- contention is not clear because peptide 4N1K has effects on cipitated by a SIRP-a monoclonal antibody (Figure 1A). An CD47 null cells.28 Nevertheless, to test whether CD47 is re- isotype-matched control IgG antibody did not coprecipitate quired for TSP1 activation of SIRP-a,wefirst treated cells SIRP-a (Figure 1A) or TSP1. In cell-free preparations, low with a CD47-targeting morpholino oligonucleotide to block concentrations of immobilized TSP1 bound soluble human mRNA translation.29 Western blot confirmed suppression of SIRP-a (Figure 1B). In contrast, the signature domain of TSP1 CD47 protein (by 37%) (Figure 2A). A control nonspecific (E123CaG1), which contains the C-terminal of TSP1 and morpholino (scrambled) did not alter total CD47 expression binds CD47,17 did not bind to SIRP-a (Figure 1C). Extending in treated cells (Figure 2A), and neither the CD47 targeting these observations to cell culture systems, where endogenous nor the control morpholinos significantly changed total SIRP- TSP1 production was minimized by restricting serum and a expression (Figure 2B, densitometry presented). TSP1 in growth factors, we treated arterial VSMCs with exogenous CD47 gene-suppressed cells still phosphorylated SIRP-a (Fig- TSP1 (2.2 nmol/L) and assessed SIRP-a phosphorylation. ure 2C). In further experiments, we used a CD47 monoclonal TSP1 phosphorylated SIRP-a within 10 minutes, and this antibody (Ab) that blocks TSP1 binding to CD47,15–17, 30 as we persisted for at least 60 minutes (Figure 1D). Because these reported, and also blocks CD47 interaction with SIRP-a.31 experiments used a general phospho-tyrosine antibody, we Treating VSMCs with CD47-blocking Ab had no effect on confirmed our results by immunoprecipitating for SIRP-a basal or TSP1-mediated SIRP-a phosphorylation (Figure and then probing for changes in tyrosine phosphorylation (Fig- 2D) and did not alter total SIRP-a protein levels (Figure 2E, ure 1E). Finally, TSP1 treatment under these conditions did not densitometry presented). These data, although interesting,

1172 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

Figure 1. TSP1 binds to and activates nonphagocytic SIRP-a and its downstream signal transducer SHP1. (A) Coimmunoprecipitation in arterial VSMCs of TSP1 and SIRP-a. Immunoprecipitation was with monoclonal Ab to TSP1, SIRP-a, or an isotype-matched control IgG antibody. Results presented are representative of six separate experiments. Plastic wells coated with TSP1 (B) or the recombinant domain of TSP1 containing the C-terminal (E123CaG1, C) at the indicated concentrations were incubated with 125I-SIRP-a at room temperature. Bound radioactivity was quantiﬁed and data are presented as the mean6SEM. of three separate experiments. VSMCs were incubated in basal medium with TSP1 (2.2 nmol/L) for the indicated time points, cell lysate prepared, protein separated by SDS-PAGE electrophoresis, membranes probed with a phospho-tyrosine antibody (D) or lysates immunoprecipitated with a SIRP-a Ab, protein separated via SDS-PAGE electrophoresis, and nitrocellulose membranes probed with Ab to total SIRP-a and to phosphorylated tyrosine residues (E), as well as phosphorylated SHP1 (Y536) (F) and p-SHP2 (G). Densitometry is presented as mean ratios of p-tyr-SIRP-a to total SIRP-a and total SIRP-a to b actin (6SEM) (D), p-SHP1 to total SHP1, p-SHP1 to b actin, and total SHP1 to b actin (6SEM) (F), and p-SHP2 to total SHP2 and total SHP2 to b actin (6SEM) (G). Representative data from four independent experiments are presented. *Statistically signiﬁcant difference (P,0.05 compared with untreated).

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1173 BASIC RESEARCH www.jasn.org

cannot exclude a possible role for CD47 in TSP1-mediated phosphorylation of SIRP-a in VSMCs. That is, there is the possibility of cross-talk, especially in light of CD47 null data in rTEC (vide infra).

TSP1 Activation of SIRP-a Does Not Require b Integrins TSP1 has several functional domains, and we reported that TSP1, via its C-terminal domain, binds CD47.17 To test whether b integrin could be mediating SIRP-a phosphorylation, we used the TSP1-based peptide 753 (M.W. 685, sequence-acAELDVP) that is derived from the N-terminal domain of the protein and is known to activate integrins.32 Treating VSMCs with peptide 753 (10 mmol/L, 10 minutes) phosphorylated SIRP-a (Supplemental Figure 2A) without changing total SIRP-a protein levels (Supplemental Figure 2C, densitometry presented). The N-terminal domain of TSP1 is known to engage b integrins.32 To explore the role of b integrins in this process, we treated VSMCs with a b integrin– blocking Ab (clone Ha2/5, BD Biosciences) followed by TSP1 (2.2 nmol/L), and determined SIRP-a phosphorylation. Interest- ingly, TSP1-stimulated phosphorylation of SIRP-a was not altered by b integrin Ab blockade (Supplemental Figure 2B). Under basal conditions, treatment of VSMCs with the b integrin Ab alone did not increase SIRP-a phosphorylation and had no effect on total SIRP-a protein levels (Supplemen- tal Figure 2B, densitometry presented). Peptide 753–mediated activation of SIRP-a was also not inhibited with the b integrin– blocking Ab (Supplemental Figure 2C). In a Figure 2. TSP1-stimulated phosphorylation of VSMC SIRP-a may not depend com- contrast, a monoclonal SIRP- Ab blocked pletely on CD47 activation. (A) VSMCs were treated with a CD47-specific(10mM) or TSP1-mediated phosphorylation of both scrambled oligonucleotide morpholino as per the manufacturer instructions. Western SIRP-a (data not shown) and the down- blot analysis of CD47 was performed. Densitometry is presented as the mean ratio of stream signal transducer SHP-1 (Supple- CD47 to b actin (6SEM). (B) Densitometry of Western blots from untreated (control), mental Figure 2D). scrambled, and CD47 morpholino treated VSMC is presented as the mean ratio of total SIRP-a to b actin (6SEM). (C) VSMCs were pretreated with a CD47-specificorscrambled ·2 TSP1-Stimulated O2 Production in morpholino, then incubated in basal medium with TSP1 (2.2 nmol/L) for 10 minutes, VSMCs Requires SIRP-a lysate prepared and Western blots performed. Densitometry is presented as the mean We have reported concurrent upregulation ratio of p-SIRP-a to SIRP-a (6SEM). (D) VSMCs were incubated in basal medium, treated of TSP1 and ROS in ischemia33 and IRI.3,15 with a CD47-blocking Ab (1 mg/ml) for 20 minutes followed by TSP1 (2.2 nmol/L) for 10 To determine whether TSP1 is capable of minutes and cell lysate prepared. Western blot analysis of p-SIRP-a was performed. z2 Densitometry is presented as the mean ratio of p-SIRP-a to b actin (6SEM). (E) Densi- stimulating O2 production in VSMC via a fi tometry of Western blots from CD47 Ab and TSP1-treated VSMCs is presented as the activation of SIRP- , we used a speci c mean ratio of total SIRP-a to b actin (6SEM). For all measurements: *statistically signif- small interfering RNA (siRNA) (oligo 3) icant difference (P,0.05) compared with untreated and control morpholino-treated, and to suppress SIRP-a,thentreatedcells representative examples of four independent experiments are presented. with TSP1 (2.2 nmol/L for 60 minutes)

1174 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

z2 and determined O2 levels via a cytochrome c reduction assay. whether the SIRP-a–SHP1 signaling cascade is present in ro- z2 TSP1 significantly stimulated VSMC O2 production (Figure dent arteries. Real-time PCR confirmed the presence of SIRP- 3A). Gene suppression of SIRP-a using siRNA abrogated a, SHP1, and SHP2 in arterial rings (Figure 5D). To investigate z2 TSP1-stimulated O2 generation (Figure 3B). Conversely, a the role SIRP-a plays in controlling vascular tone, we treated control siRNA (scrmb siRNA) had no effect on TSP1-stimu- endothelial-free arteries with the SIRP-a–blocking Ab. Impor- z2 lated O2 production (Figure 3B). Western blot analysis tantly, TSP1-mediated inhibition of NO-stimulated vasodila- confirmed a reduction in SIRP-a expression by the targeting tion was attenuated in vessels treated with the SIRP-a–blocking siRNA (40%) (Figure 3C). Treatment of VSMCs with the Ab (Figure 5E). SIRP-a siRNA (oligo 3) did not change CD47 protein expres- ·2 sion (Figure 3D). Corroborating these data, TSP1-stimulated TSP1 Stimulates Phosphorylation of SIRP-a and O2 z2 O2 production was inhibited by more than half in cells treated Production in Human Renal Tubular Endothelial Cells with a SIRP-a Ab (Figure 3E). Renal tubular epithelial cells (rTECs) express Noxs,40,41 including Nox1 and 2,42,43 and are a target of cell injury driven z2 3 TSP1, via SIRP-a, Phosphorylates the Nox1/2 by O2 . To determine whether TSP1–SIRP-a–mediated phox z2 Organizer Subunit p47 stimulation of O2 is a general finding and extends to other A major NADPH oxidase family member found in rodent large parenchymal cells, we treated human rTECs with TSP1. Hu- vessel arterial VSMCs is Nox1.34 Activation of Nox1 (and man rTECs expressed SIRP-a and SHP1 (Figure 6, A and B), z2 Nox2) and subsequent enzymatic O2 formation occur on and TSP1 (2.2 nmol/L) treatment promoted rapid phosphor- assembly of several subunit proteins, including p47phox (re- ylation of SIRP-a and SHP1 (Figure 6C) concurrent with 35 z2 viewed by Lassègue and Griendling ). This latter process re- stimulating increased O2 production (Figure 6D). On the quires serine phosphorylation of p47phox.36 TSP1 treatment other hand, TSP1-stimulated phosphorylation of SIRP-a and (2.2 nmol/L) increased p47phox phosphorylation in VSMCs SHP1 was blocked in rTECs treated with a SIRP-a Ab (Figure (Figure 4A). Treatment with the SIRP-a–blocking Ab reduced 6C). In these experiments a general phospho-tyrosine anti- TSP1-mediated phosphorylation of p47phox (Figure 4B), sug- body was again employed. Therefore we confirmed these re- gesting that SIRP-a is upstream of TSP1-mediated phosphor- sults by immunoprecipitating for SIRP-a and probing for ylation of the Nox subunit. changes in tyrosine phosphorylation. Here, too, TSP1 treatment increased p-SIRP-a expression in human rTECs (Figure ·2 3 TSP1–SIRP-a Signaling Stimulates VSMC O2 6E). Having previously reported rTEC express CD47, we Production and Inhibits Vasodilation tested whether CD47 is required for TSP1-mediated SIRP-a TSP1 is a large, 480-kDa protein found in the blood of healthy phosphorylation in rTECs. Wild-type (CD47+/+) and null 2 2 persons at 100-picomolar concentrations21 and at low nano- (CD47 / ) rTECs obtained from kidneys of the respective molar concentrations in disease.7 It is unknown whether TSP1 strains of mice were pre-treated with a SIRP-a antagonist can cross the subendothelial barrier to gain access to the arte- Ab (1 mg/ml) for 15 minutes followed by exogenous TSP1 rial VSMC compartment. Totest this, aortic endothelium-intact (2.2 nmol/L) for 60 minutes. TSP1 treatment stimulated arterial rings were incubated with TSP1 (2.2 nmol/L) for 60 SIRP-a phosphorylation in wild-type rTECs (Figure 6F), minutes and tissue sections cut from the middle of the vessels which was inhibited by SIRP-a Ab treatment. However, 2.2 prepared. TSP1-treated arterial rings showed increased TSP1 nmol/L TSP1 treatment did not increase p-SIRP-a levels in on both luminal endothelial cells and throughout the suben- CD47 null rTEC (Figure 6F). Total SIRP-a and SHP1 protein dothelial VSMC compartment (Figure 5A; arrows indicate ves- levels did not vary between wild-type and CD47 null rTEC or sel lumen). treatment groups (data not shown). rTEC undergo apoptotic NO stimulates VSMC relaxation and arterial vasodilation.20 cell death post IRI44 and a TSP1-derived peptide is known to z2 45 NO also rapidly interacts with O2 to generate peroxyni- promote endothelial cell death. We tested whether TSP1, via trite,37 limiting the bioavailability of NO. On the basis of SIRP-a, promoted rTEC apoptosis. Cells were treated with z2 our finding that TSP1 can activate SIRP-a to increase O2 TSP1 (2.2 nmol/L) for 24 hours, and apoptosis and cell via- production in arterial VSMCs, we hypothesized that this bility were assessed. Under these conditions TSP1 did not might, in part, account for our previous observation that stimulate apoptosis or decrease cell viability (Supplemental TSP1 inhibits NO-mediated arterial vasodilation.5 We tested Figure 3, A and B). whether TSP1 inhibited arterial vasodilation via stimulation z2 of VSMC O2 production in endothelial-free arterial rings, TSP1–SIRP-a Signaling Is Upregulated in Renal IRI usingSNP,which,onconversiontoNO,stimulatesvasodilation.38 Concurrent with Increased ROS Production, Decreased In both murine (Figure 5B) and rat (Figure 5C) endothelial-free Blood Flow, and Increased Tissue Injury arterial rings, TSP1 (2.2 nmol/L) inhibited SNP-mediated IRI is a major cause of cellular and organ dysfunction and, z2 39 46,47 vasodilation whereas the O2 scavenger Tempol significantly in transplantation, of graft failure. Pathologic ROS con- attenuated this effect. Interestingly, Tempol alone also decreased tribute significantly to this process,48,49 and ROS mitigation SNP-mediated vasodilation (data not shown). It is unknown is a therapeutic goal for limiting tissue injury in several

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1175 BASIC RESEARCH www.jasn.org

conditions, including myocardial infarction, stroke, and organ transplantation.50 To test whether TSP1–SIRP-a signaling is upregulated in IRI, we challenged mice with 20 minutes of renal ischemia followed by 24 hours of reperfusion. Our data demonstrate concordant upregulation of renal TSP1, p-SIRP-a, and p-SHP1 protein (Figure 7A) and SIRP-a and SHP1 mRNA (Figure 7B) after IRI. Immunofluorescent staining of renal corticomedullary sections after IRI demonstrated increased TSP1 expression in the renal tubules compared with sham control (Figure 7C). SIRP-a was expressed in both proximal and distal tubules regardless of injury state. Analysis of merged images suggests colocalization of TSP1 and SIRP-a in renal tubules after IRI. Histology confirmed significant renal tubular injury and decreased renal function (as assessed by el- evated serum urea and creatinine) in wild- type kidneys and mice after IRI (Figure 7D). As we reported previously,3 real-time laser Doppler analysis demonstrated marked de- terioration of renal blood flow in post-IRI animals compared with sham-operated mice (data not shown) and greater cell death on TUNEL staining (Figure 7F). As- sessment of ROS using the cell-permeable agent dihydroethidium (DHE) demonstrated increased fluorescence in renal tissue after IRI (Figure 7E). Because DHE can

z2 be oxidized by cells to several fluorescent Figure 3. TSP1-stimulated O production in VSMCs requires SIRP-a.(A)VSMCs 2 products, only one of which is specific for were incubated with vehicle or TSP1 (2.2 nmol/L) for 60 minutes. Superoxide z2 51 production was measured in the 28,000g membrane fraction and calculated from ethidium-O2 interaction, we also evalu- the initial linear rate of superoxide dismutase–inhibitable cytochrome c reduction. ated tissue oxidative stress using 4-hydroxy- Data represent the rate of superoxide production (6SEM). *Statistically significant nonenal (4-HNE), a stable aldehyde difference (P,0.05) compared with untreated. (B) VSMCs were transfected with formed by the degradation of polyunsatu- SIRP-a siRNA (oligo 3) or a control scrambled (scrmb) siRNA and incubated with rated fatty acids during lipid peroxida- TSP1 (2.2 nmol/L) for 60 minutes. Superoxide production was measured as de- tion.52 4-HNE immunostaining was greater scribed above. Data represent therateofsuperoxideproduction(6SEM). *Statis- after IRI and was minimally detected in re- # tically significant difference (P,0.05) compared with scrmb; statistically significant nal tissue sections from sham-operated , z2 difference (P 0.05) compared with scrmb siRNA+TSP1. (C) VSMCs were trans- mice (Figure 7E). A byproduct of O re- a 2 fected with several SIRP- or scrambled (scrmb) siRNA oligos. Western blot analysis acting with NO is the reactive nitrogen was performed for SIRP-a. A representative blot from three separate experiments is 2 species peroxynitrite (ONOO ), which presented. Densitometry is presented as the mean ratio of total SIRP-a to b actin (6SEM). *Statistically significant difference (P,0.05) compared with scrambled can adversely alter protein function a through nitration.53 Consistent with in- control. (D) VSMCs were transfected with SIRP- (oligo 3) or control scrambled z2 siRNA. Western blot analysis was performed for CD47. Densitometry is presented creased in vivo O2 and reactive nitrogen asthemeanratiooftotalCD47tob actin (6SEM) from three experiments. (E) species, renal tissue levels of 3-nitrotyrosine VSMCs were treated with a SIRP-a–blocking Ab (1 mg/ml) for 10 minutes, followed were increased after IRI (Figure 7A). Inter- by TSP1 (2.2 nmol/L) for 60 minutes. Superoxide production was measured. Data estingly, Nox1 protein expression was in- represent the rate of superoxide production (6SEM). *Statistically significant dif- creased (Figure 7A, blot and densitometry) # ference (P,0.05) compared with untreated control; statistically significant differ- and Nox2 remained unchanged (Figure 7A, , ence (P 0.05) compared with TSP1 treated. densitometry alone), while Nox1 mRNA levels were unchanged whereas Nox2

1176 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

24 hours (Figure 8A), concurrent with decreased p-SIRP-a, p-SHP1, total SIRP-a, Nox1 and 3-nitrotyrosine protein expression (protein and/or densitometry shown) and decreased TSP1, SIRP-a and Nox1 mRNA (Figure 8B). Nox2 protein levels were unchanged in both treatment groups. In IRI-challenged animals, treatment with a SIRP-a Ab decreased oxidative stress as quantified by DHE and 4-HNE immunofluorescence (Figure 8C), concordant with less renal tubular injury and neutrophil in- filtration (Figure 8D), decreased proinflam- matory cytokine and chemokine transcript expression (CCL2, CXCL2, IL1-b, TNF-a) (Figure 8E), and less cell death (as detected by TUNEL staining with results in kidneys from SIRP-a Ab–treated animals [Figure 8F] approaching levels found in sham-operated kidneys [Figure 7E]). From a functional perspective, serum urea and creatinine levels were lower after IRI in animals treated with the SIRP-a Ab compared with control IgG Ab–treated animals (Figure 8G). Previously we reported that CD47 null mice were protected from renal IRI.3 Therefore, we treated CD47 null mice with the SIRP-a antagonist Ab and challenged them with renal IRI. Interestingly, we found no further improvement in serum urea and creati- Figure 4. TSP1, via SIRP-a, phosphorylates the Nox1/2 organizer subunit p47phox.(A) nine levels in SIRP-a Ab–treated CD47 VSMCs were treated in minimal medium with TSP1 (2.2 nmol/L) for 60 minutes or (B) null mice after IRI compared with un- aSIRP-a–blocking Ab (1 mg/ml) for 10 minutes with or without TSP1 (2.2 nmol/L) for 60 phox treated CD47 null animals (Supplemen- minutes, cell lysates prepared, immunoprecipitated with a p47 Ab, resolved via tal Figure 4). gel electrophoresis, and blots probed with a phospho-serine Ab. A representative blot from four separate experiments is presented. Densitometry is presented as the mean ratio of total p-serine to p47phox (6SEM). *Statistically significant difference (P,0.05) compared with untreated control; #statistically significant difference (P,0.05) com- DISCUSSION pared with TSP1 treated. Densitometry of Western blot of total p47phox expression from indicated treated cell groups not significantly different from control. In a myeloid leukemia cell line, SIRP-a inhibits expression of the Nox2 subunit and is associated with decreased hydrogen per- mRNA levels were decreased post IRI in wild-type kidneys oxide production.54 However, nonphagocytic SIRP-a has not compared with sham controls (Figure 7E). previously been linked directly to pathologic ROS production. Our present results implicate a role for SIRP-a in the stimu- z2 Disrupting TSP1–SIRP-a Signaling Suppresses Nox1 lation of Nox-mediated O2 production in multiple nonpha- Transcription and Oxidative Stress, Increases gocytic cell types. The data indicate that cells treated with a Reperfusion Blood Flow, and Limits Renal IRI Injury physiologically relevant concentration of TSP1 demonstrate a z2 Tofurther assess the role of the TSP1–SIRP-a signaling cascade SIRP-a–dependent increase in O2 production. Either siRNA in promoting renal IRI, we treated mice with the SIRP-a–blocking knockdown or antibody blockade of SIRP-a inhibited both Ab that we have shown herein prevents TSP1-stimulated TSP1-mediated phosphorylation of the Nox organizer subunit z2 phox z2 phosphorylation of SIRP-a and inhibits subsequent O2 pro- p47 and O2 production without changing basal ROS duction. Following IRI, SIRP-a Ab treated mice demonstrated production, consistent with SIRP-a playing an active role z2 restoration of kidney blood flow to near preischemic levels after in stimulating O2 production.

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1177 BASIC RESEARCH www.jasn.org

cells, TSP1 stimulates the phosphorylation of SIRP-a and its downstream effector SHP1 through a process that may depend on interaction with CD47. Gene suppression of CD47 by 37% and CD47 monoclonal Ab blockade did not prevent TSP1-mediated activation of SIRP-a in VSMCs. It is still possible that a residual amount of CD47 in gene-silenced cells is still sufficient to mediate TSP1-induced SIRP-a phosphorylation. Moreover, the pharmacologic prop- erties of the employed CD47 Ab has to be more fully characterized using protein binding experiments to determine whether the Ab inhibits TSP1–SIRP-a binding or the Ab modifies CD47–SIRP-a interaction. Interestingly, treatment of murine CD47 null rTEC with 2.2 nmol/LTSP1 for 60 minutes did not stimulate SIRP-a phosphorylation. Also, CD47 null mice treated with a SIRP-a antibody did not exhibit further improvement in serum creatinine after renal IRI compared with isotype control-treated animals. Together, these data suggest an interaction between TSP1 and SIRP-a, and they also suggest an important role of CD47. Indeed, our previous studies showed that TSP1 activates Nox1-dependent ROS production in a CD47-dependent manner.58 In conjunction with the current data, it a z2 Figure 5. TSP1, via SIRP- , stimulates VSMC O2 production and inhibits vasodila- would appear that TSP1 could mediate its tion. (A) Murine aortic segments were freshly harvested from wild-type male C57BL/6 ROS-inducing effect via several mecha- mice, incubated in standard myograph buffer with TSP1 (2.2 nmol/L, 60 minutes), and nisms, including (1) parallel activation of fl tissue sections prepared for immuno uorescence. Representative images of three CD47 and SIRP-a with these two pathways independent experiments are presented. TSP1 appears red. The red arrows highlight exhibiting cross-talk or (2) activation of the vessel lumen. Murine (B) and rat (C) aortic rings were prepared from freshly har- CD47 followed by downstream activation vested thoracic aortas. The endothelium was removed and the resulting arterial rings a were incubated with vehicle, TSP1 (2.2 nmol/L, 60 minutes), or TSP1+Tempol (30 mM) of SIRP- . Future studies are required to 2 and constricted with phenylephrine (3310 7 M). Endothelium-independent vasodi- further delineate the precise molecular 2 2 lation was stimulated by SNP (10 9 to 10 5 M). Data represent the means6SEM of mechanisms by which TSP1 stimulates four experiments. *Statistically significant difference (P,0.05) in relaxation between ROS generation. TSP1 versus TSP1 + Tempol. (D) RT-PCR assessment of SIRP-a, SHP1, and SHP2 In human umbilical vein endothelial mRNA of murine thoracic aorta. Data were normalized to the housekeeping gene cells, angiotensin II–stimulated ROS pro- HPRT1 and represent the means6SEM of five separate samples. (E) Murine thoracic duction was associated with increased aortas were harvested and endothelial-free arterial rings were incubated with vehicle, SHP1 activity.59 We now find in two dis- TSP1 (2.2 nmol/L, 60 minutes), or TSP1+a SIRP-a–blocking Ab (1 mg/ml) and con- 2 tinct nonphagocytic cell types, VSMCs and stricted with phenylephrine (3310 7 M). Endothelium-independent vasodilation was 2 2 rTECs, that TSP1 treatment stimulates stimulated by SNP (10 9 to 10 5 M). Data represent the means6SEM of four ex- SIRP-a and downstream SHP1 and in- periments. *Statistically significant difference (P,0.05) in relaxation between TSP1 z2 versus TSP1+SIRP-a Ab. creased O2 production. VSMCs and rTECs, depending on the vascular bed, express both Nox1 and Nox2, and the assembly of both of these enzyme complexes is Signaling through SIRP-a is reported to occur via interac- promoted by the serine phosphorylation of the p47phox sub- tions with integrins,55 growth factors and CD47.56,57 In mye- unit.60,61 TSP1 phosphorylated p47phox, whereas a SIRP-a Ab loid cells, SIRP-a phosphorylation has been demonstrated in blocked TSP1-mediated phosphorylation of this essential Nox the absence of CD47.55 Our data indicate that in nonmyeloid organizing subunit, and both the SIRP-a–blocking Ab and

1178 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

z2 Figure 6. TSP1 stimulates SIRP-a phosphorylation and O2 production in human rTECs. (A) Human rTECs were stained for SIRP-a (red) and nuclear protein (blue). Representative images are presented at a magnification of 320. (B) Human rTECs were cultured in basal medium, lysate prepared and analyzed for total SIRP-a and SHP1 protein expression. A representative blot from three separate experiments is presented. (C) Human rTECs were incubated with vehicle or TSP1 (2.2 nmol/L) with or without a SIRP-a Ab (1 mg/ml) for 60 minutes, and protein expression of p-SIRP-a and p-SHP1 was determined. Data represent the mean ratios of target protein to b actin (6SEM). *Statistically significant difference (P,0.05) compared with untreated. **Statistically significant difference (P,0.05) compared z2 with TSP1 treated. (D) Human rTECs in minimal medium were treated with TSP1 (2.2 nmol/L) for 45 minutes and O2 production measured as described previously. Data represent the rate of superoxide production (6SEM). *Statistically significant difference (P,0.05) compared with untreated. (E) Human rTECs were incubated in basal medium with TSP1 (2.2 nmol/L), cell lysate prepared and immunoprecipitated with a SIRP-a Ab, protein separated via SDS-PAGE electrophoresis, and nitrocellulose membranes probed with Ab to total SIRP-a and to phosphorylated tyrosine residues. A representative blot from three separate experiments is presented. Data represent the mean ratios of target protein to SIRP-a (6 SEM). *Statistically significant difference (P,0.05) compared with untreated (control). (F) Wild-type and CD47 null murine rTECs were treated with TSP1 (2.2 nmol/L) with or without a SIRP-a Ab (1 mg/ml) for 60 minutes and protein expression of p-SIRP-a determined. A representative blot from three separate experiments is shown. Data are presented as the mean ratio of target protein to b actin (6SEM). *Statistically significant difference (P,0.05) compared with untreated (control). **Statistically significant difference (P,0.05) compared with TSP1 treated.

z2 SIRP-a siRNA abrogated TSP1-stimulated O2 production. epithelial cells. Further work will deﬁne the dominant Nox iso- These data identify SIRP-a as a new, physiologically important form responsible for TSP1–SIRP-a–induced ROS following re- z2 activator of Nox-derived O2 production in vascular and nal IRI.

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1179 BASIC RESEARCH www.jasn.org

We have reported that TSP1 limits arterial vasodilation via Santa Cruz Biotechnology, and the b integrin–blocking Ab (clone inhibition of NO signaling.14,62 Results from our present study Ha2/5) was purchased from BD Biosciences. suggestthatTSP1inhibitsarterial vasodilation, in part by z2 z2 stimulating pathologic O production. The O scavenger 2 2 Animals Tempol decreased TSP1-mediated inhibition of NO-driven Male C57BL/6 wild-type mice and Sprague–Dawley rats were vasodilation in murine and rat arteries, supporting a role for obtained from The Jackson Laboratory (Bar Harbor, ME) and z2 a TSP1 in promoting arterial O2 production. A SIRP- Ab Taconic (Hudson, NY), respectively. All studies were performed using reversed TSP1-mediated inhibition of vasodilation, placing protocols approved by the Institutional Animal Care and Use a SIRP- upstream in this process. These data obtained using Committee at the University of Pittsburgh and in accordance whole arteries are unlikely to be merely the result of TSP1- with National Institutes of Health guidelines. mediated inhibition of NO bioavailability because before analysis, vessels were rendered endothelial free and thus lacked an endogenous source of NO. Importantly, these data are, to our Murine rTEC Cultures knowledge, the first to characterize SIRP-a as an acute regu- Kidneys from age-matched male wild-type and CD47 null mice were lator of arterial vasodilation. excised and digested, and primary rTEC were harvested and cultured 66 fi IRI is characterized by defects in tissue blood flow, as previously described with several minor modi cations. increased thrombosis, pathologic ROS production,63 and cell death.64 We herein identify SIRP-a as a new receptor Protein-Binding Assays for the matricellular protein TSP1. These results are also Recombinant SIRP-a was labeled using NaI125 by the iodogen the first to identify a soluble ligand of SIRP-a.TSP1,onen- method as we previously published.17 Immulon 2HB Removawells gaging SIRP-a in nonphagocytic cells, stimulates NADPH (Thermo, Franklin, MA) were coated with 50 ml of TSP1 at several z2 – fi oxidase–derived O2 production. Contrary to previous concentrations for 16 20 hours at 4°C. Nonspeci cbindingwas work (reviewed by Barclay et al.2 and Matozaki et al.65), we blocked by incubating the wells with DPBS with Ca2+ and Mg2+ identify SIRP-a as an activating, rather than inhibitory, re- and 1% BSA. Radiolabeled SIRP-a was then added and incubation ceptor in two nonphagocytic cell types, specifically VSMCs conducted for 2 hours at room temperature. Following extensive and rTECs. In vivo, TSP1, SIRP-a, and SHP1 were induced by washing, bound radioactivity was quantified. and promoted IRI. Consistent with this notion, blocking TSP1–SIRP-a signaling with a SIRP-a monoclonal Ab inhibited z2 Coimmunoprecipitation IRI-mediated expression of these proteins, lowered O2 pro- Immunoprecipitation was performed as we have described, with duction in cells and tissues, restored NO-mediated vasodila- minor modification.67 tion (in ex vivo vessel preparations), augmented reperfusion blood flow, preserved end-organ integrity, and improved renal function after IRI. Consistent with our previous findings, Western protein analysis CD47 is likely to play a parallel and supportive role in this Tissue or cells were homogenized in ice-cold lysis buffer. Super- fi process. Which of the two pathways predominates is still open natants were collected and lysates quanti ed using a Bradford assay to question. (Bio-Rad, Hercules, CA). Thirty micrograms of total protein were resolved by SDS-PAGE and transferred onto nitrocellulose membrane (Bio-Rad). In blots for CD47, nonreducing Laemmli buffer was CONCISE METHODS used with 8% SDS-PAGE. Blots were probed with primary antibody to the respective proteins and visualized on an Odyssey Imaging System(Licor,Lincoln,NE).Theintensityofthebandswasquantified Detailed methods are available are available in the Supplemental Ma- using ImageJ. terial.

Reagents and Cells RNA Extraction and Quantification by Real-Time PCR Human rTECs and rat aortic VSMCs were purchased from Lonza Total RNA was extracted using Qiagen RNeasy Mini Kits (Qiagen, (Switzerland) and maintained in recommended medium. Cells were Hilden, Germany) as per the manufacturer’s instructions. RNA was used between passages 3 and 7. TSP1 was purchased from Athens quantified using the Take3 Gen5 spectrophotometer (BioTek, Winooski, Research & Technology (Athens, GA). CD47 morpholino oligonu- VT). One microgram of RNA was treated with DNase I (amplification cleotides and corresponding mismatched control morpholino were grade; Invitrogen, Carlsbad, CA) and then reverse-transcribed us- purchased from GeneTools, Inc. (Philmonth, Oregon). The TSP1- ing the Superscript III First Strand Synthesis Supermix (Invitrogen). derived peptide 753 was synthesized by Dr. Henry C. Krutzsch and cDNA was amplified using Platinum Quantitative PCR SuperMix- kindly provided by Dr. David D. Roberts (National Cancer Institute, UDG(Invitrogen)in10-ml volumes in triplicate with gene-specific National Institutes of Health, Bethesda, MD). The SIRP-a monoclo- primers and probed on the ABI Prism 7900HT Sequence Detection nal Ab (clones C20 for treatment applications and A1 for Western System (Applied Biosystems, Foster City, CA), according to the blot) and monoclonal CD47 Ab (clone OX101) were purchased from manufacturer’s instructions.

1180 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

Figure 7. TSP1-SIRP-a signaling is upregulated in renal IRI concurrent with increased ROS production, decreased blood flow, and increased tissue injury. Male C57BL/6 mice were challenged with 20 minutes of unilateral renal ischemia and 24 hours of reperfusion (n=8) or sham surgery (n=6) and Western blot for (A) TSP1, SIRP-a,p-SIRP-a, p-SHP1, Nox1, Nox2, 3-nitrotyrosine, and b actin was performed. Data shown are mean6SEM. *Statistically significant (P,0.001) IRI compared with sham-operated kidneys. (B) RT-PCR analysis of renal mRNA expression of the indicated genes in IRI and sham-operated kidneys. Data shown are mean6SEM. *Statis- tically significant (P,0.05) IRI compared with sham. (C) Representative kidney tissue sections from IRI and sham-operated mice stained for TSP1 and SIRP-a. TSP1 colored green; SIRP-a colored red; nuclei colored blue. Scale bar, 50 mm(magnification 320). (D) Quantitative analysis and photomicrographs of tubular damage and neutrophil invasion in juxtamedullalary sections from IRI and sham-operated mice are shown (magnification, 3200; arrow highlights tubular injury); serum urea and creatinine from the same. Data shown are means 6 SEM. *Statistically significant (P,0.05) IRI compared with sham. (E) Representative tissue sections and quantitative analysis from IRI and sham-operated mice stained with DHE or 4-HNE and visualized by microscopy (magnification, 3200). RT- PCR analysis of Nox1 and Nox2 mRNA from sham and post-IRI kidneys. Data shown are means6SEM. *Statistically significant (P,0.05) IRI compared with sham. (F) Representative tissue sections and quantitative analysis from IRI and sham-operated mice stained by TUNEL assay and visualized by microscopy (magnification, 3200). *Statistically significant (P,0.05) IRI compared with sham.

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1181 BASIC RESEARCH www.jasn.org

phox Assessment of p47 Phosphorylation with ethanol. Slides were blocked with 5% goat serum and then phox Wedeterminedthephosphorylationofp47 (criticalcytosolicsubunit incubated overnight at 4°C with SIRP-a Ab. Nuclei were stained with phox of Nox) in p47 immunoprecipitates using an antiphosphoserine Ab Hoechst dye. Fluorescent images were captured with an Olympus as described.68 Briefly, cells were lysed and protein from the cytosolic Fluoview 1000 confocal microscope (software version 1.7a). Cryostat phox fraction incubated with anti-p47 Ab. Immune complexes were re- sections (5 mm) of mouse kidneys were fixed for 20 minutes with 2% covered with Protein G Plus-Agarose. Immunoprecipitates were then paraformaldehyde. They were then washed, blocked with 2% donkey subjected to Western blotting with a phospho-serine Ab. serum in BSA, incubated with combined primary antibodies for TSP1 and SIRP-a, washed and incubated with secondary antibody. Nuclei ·2 Measurement of O2 via Cytochrome c Reduction were stained with Hoescht dye. After rinses with PBS, sections were Cells were suspended in 200 ml of ice-cold disruption buffer. The coverslipped with Gelvatol mounting media. Fluorescent images suspension was lysed by five freeze/thaw cycles and passed were captured with an Olympus Fluoview 1000 confocal microscope through a 30-gauge needle five times.58 The cell lysate was centrifuged (software version 2.01). at 1000g for 10 minutes at 4°C to remove intact cells, nuclei, and debris. The supernatant was transferred to another Eppendorf tube Cell Apoptosis and Viability Assays and was centrifuged at 28,000g for 15 minutes at 4°C to collect the Human rTECs were grown to 80% confluence, treated with TSP1 membrane fraction pellet, which was resuspended in 50 ml disruption (2.2 nmol/L) with or without SIRP-a antibody (1 mg/ml) for 24 hours, buffer. Membrane fractions (2 mg/well) were added to cytochrome lysates prepared, proteins then separated by SDS-PAGE, transferred c–containing oxidase assay buffer. After a 5-minute baseline measure- to nitrocellulose membrane and probed for caspase-3. Cell viability m z– ment, NADPH (180 M) was added to initiate the reaction. O2 was was assessed in a 96-well plate using the LIVE/DEAD Viability/ measured as the initial linear rate of superoxide dismutase–inhibitable Cytotoxicity kit (Molecular Probes). Calcein (0.5 mM) and ethidium cytochrome c reduction.69 homodimer-1 (2 mM) were added to the cells for 30 minutes and fluorescence read using a microplate reader as per manufacturer siRNA Transfection instructions. Cells were plated the day before transfection to achieve 30%–50% fl con uence in DMEM medium containing 10% fetal calf serum with- Arterial Myography out antibiotics. The siRNA transfections were performed by using Myography of arterial segments was performed as previously pub- Lipofectamine 2000 (Invitrogen) in Opti-MEM according to the lished5,70 with minor modifications. Animals were anesthetized with manufacturer’s instructions. pentobarbital (50 mg/kg intraperitoneally). Thoracic aortas of mice and rats were cleared of adherent adipose tissue and excised. The Morpholino Oligonucleotide Protein Suppression of endothelial layer was removed by gently rubbing the luminal side of CD47 the vessel along the rough surface of a blunt needle. Arterial Cells were seeded onto 60-mm2 plates in DMEM medium containing segments were mounted on myograph pins (Danish Myo Technol- 10% fetal calf serum. CD47 morpholino was transfected with a final ogy, Atlanta, GA) in 5-ml incubation buffer maintained at 37°C, concentration of 10 mM using the delivery agent Endoporter (6 ml/ml; pH 7.4, gassed with 95% O2 and 5% CO2,andbroughttoanoptimal Genetools, Philomath, OR) according to protocol instructions. resting tension. Viability of the vessels was ascertained by a contrac- tile response to potassium chloride. Phenylephrine (Sigma-Aldrich) 2 2 concentration-response curves (10 9 to 10 5 M) were generated by Tissue Histology and Microscopy measuring contraction plateaus at each concentration. In precontracted Kidneys embedded in paraffin were sectioned at 3 mm and stained vessels, endothelium-independent vasodilation to SNP (Sigma-Aldrich; with hematoxylin and eosin by standard methods. Markers of tubular 2 2 10 10 to 10 5 M) with or without the indicated treatments was then damage (tubular dilatation, cell necrosis, infarction, and cast forma- tested. tion) were scored in “blinded” fashion on randomly selected corticomedullary fields (magnification, 3200). Light microscopy images Detection of ROS in Renal Tissue Sections were acquired under identical settings using a Zeiss Axiovert 40CFL ROS in renal tissue sections was assessed using several techniques. The microscope and Axiovision software, version 4.8 (Carl Zeiss, Oberkochen, cell-permeable agent dihydroethidium was applied to unfixed frozen Germany). sections that were incubated in a light-protected humidified chamber at 37°C for 30 minutes, washed with PBS, and mounted with fluo- Immunofluorescence Staining of Arterial Rings, Human rescent mounting medium. Cryosections of kidneys were washed rTECs, and Murine Kidneys with PBS, followed by 0.5% BSA in PBS. Sections were blocked Cryostat sections of vessels were washed with PBS, followed by 0.5% with 2% BSA solution. They were incubated at room temperature BSA in PBS. Sections were blocked with 2% BSA solution. The slides with primary antibody for 4-HNE. Slides were incubated with CY3 were incubated at room temperature with primary anti-TSP1 Ab. goat anti-rabbit secondary antibody (Jackson ImmunoResearch) in Slides were washed with BSA solution and incubated with a CY3 goat combination with F-actin dye rhodamine phalloidin. Nuclei were anti-rabbit secondary Ab in combination with the F-actin dye stained with Hoescht dye. Fluorescent images were captured with an rhodamine phalloidin. For rTEC, cytospins were generated and fixed Olympus Fluoview 1000 confocal microscope (software version 1.7a).

1182 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

Figure 8. Disrupting TSP1–SIRP-a signaling suppresses Nox1 transcription and oxidative stress, increases reperfusion blood flow, and limits renal IRI injury. Wild-type mice were treated with a SIRP-a–blocking Ab or isotype control (CTRL) Ab and challenged with renal ischemia and 24-hour reperfusion (n=8 per group). (A) Laser Doppler images of renal blood flow. Red indicates blood flow. Results represent the means6SEM of four measurements from five mice in each group. *Statistically significant difference (P,0.001) for SIRP-a compared with isotype control at 30 minutes; **statistically significant difference (P,0.001) for SIRP-a compared with isotype control at 24 hours of reperfusion. (B) Western and RT-PCR analysis was performed for the indicated proteins and genes. Data shown are means6SEM, with representative Western blots.*Statistically significant difference (P,0.001) for SIRP-a compared with isotype control Ab. (C) Quantification and representative tissue sections from SIRP-a or control Ab treated mice after renal IRI stained with DHE or 4- HNE and visualized by microscopy (magnification, 3200) (n=5 per group). (D) Representative tissue sections and analysis of renal tubular damage and neutrophilic invasion from SIRP-a and isotype control Ab–treated mice (magnification, 3200; inset 3400). Data shown are means6SEM, *statistically significant difference (P,0.001) for SIRP-a compared with control Ab. Arrows highlight rTEC death and cast formation. (E) RT-PCR analysis of inflammatory cytokines and chemokines. Data shown are means6SEM. *Statistically significant difference (P,0.001) for SIRP-a compared with control Ab. (F) Quantification and tissue sections from control or SIRP-a Ab– treated mice after renal IRI mice stained by TUNEL assay (magni fication, 3200). Data shown are means6SEM (n=6 per group). *Statistically significant difference (P,0.001) for SIRP-a compared with control Ab. (G) Serum urea and creatinine from mice. Data shown are means6SEM (n=6 per group). *Statistically significant difference (P,0.05) for SIRP-a compared with control Ab.

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1183 BASIC RESEARCH www.jasn.org

IRI REFERENCES Mice were anesthetized using isoflurane and oxygen titrated to effect, and body temperature was maintained at 36°C with the aid of a rectal 1. Lawler J: Thrombospondin-1 as an endogenous inhibitor of angio- temperature probe, warming pad, and warming lamp. A microaneu- genesis and tumor growth. JCellMolMed6: 1–12, 2002 rysm clip was placed to occlude the left, or both, renal pedicles for 2. Chen H, Herndon ME, Lawler J: The cell biology of thrombospondin-1. – 20 minutes, after which the clip was removed. The abdomen was Matrix Biol 19: 597 614, 2000 fi 3. Rogers NM, Thomson AW, Isenberg JS: Activation of parenchymal closed with 5/0 mono lament suture. Mice were used for laser Dopp- CD47 promotes renal ischemia-reperfusion injury. J Am Soc Nephrol ler experiments 30 minutes or 24 hours after reperfusion, and then 23: 1538–1550, 2012 euthanized. Blood was collected and kidney tissue snap frozen, placed 4. Riessen R, Kearney M, Lawler J, Isner JM: Immunolocalization of in RNAlater, embedded in optimal cutting temperature compound, thrombospondin-1 in human atherosclerotic and restenotic arteries. – or fixed in 10% neutral buffered formalin. Am Heart J 135: 357 364, 1998 5. Bauer EM, Qin Y, Miller TW, Bandle RW, Csanyi G, Pagano PJ, Bauer PM, Schnermann J, Roberts DD, Isenberg JS: Thrombospondin-1 Laser Doppler Blood Flow Analysis supports blood pressure by limiting eNOS activation and endothelial- Real-time kidney perfusion and blood flow was measured using laser dependent vasorelaxation. Cardiovasc Res 88: 471–481, 2010 Doppler imaging (MoorLDI-2l; Moor Instruments, Devon, UK). 6. Moura R, Tjwa M, Vandervoort P, Cludts K, Hoylaerts MF: Thrombo- Briefly, animals were anesthetized and core temperature maintained spondin-1 activates medial smooth muscle cells and triggers neointima formation upon mouse carotid artery ligation. Arterioscler Thromb at 36°C. Organ blood flow was assessed at baseline, in response to Vasc Biol 27: 2163–2169, 2007 ischemia and reperfusion at 30 minutes and 24 hours. Results are 7. Novelli EM, Kato GJ, Ragni MV, Zhang Y, Hildesheim ME, Nouraie M, expressed as the percentage change from baseline control of the re- Barge S, Meyer MP, Hassett AC, Gordeuk VR, Gladwin MT, Isenberg JS: gion of interest. Plasma thrombospondin-1 is increased during acute sickle cell vaso- occlusive events and associated with acute chest syndrome, hydroxy- urea therapy, and lower hemolytic rates. Am J Hematol 87: 326–330, In Vivo SIRP-a Ab Treatment 2012 Age-matched male wild-type C57BL/6 or CD47 null mice were 8. Kharitonenkov A, Chen Z, Sures I, Wang H, Schilling J, Ullrich A: A family randomly assigned to receive a SIRP-a monoclonal Ab (clone C20; of proteins that inhibit signalling through tyrosine kinase receptors. Nature 386: 181–186, 1997 Santa Cruz Biotechnology; 0.4 mg/g body weight intraperitoneally in m 9. Stofega MR, Argetsinger LS, Wang H, Ullrich A, Carter-Su C: Negative 100 l sterile PBS) or an IgG isotype-matched control Ab (Santa Cruz regulation of growth hormone receptor/JAK2 signaling by signal reg- Biotechnology) 90 minutes before surgery. ulatory protein alpha. JBiolChem275: 28222–28229, 2000 10. Gresham HD, Dale BM, Potter JW, Chang PW, Vines CM, Lowell CA, Lagenaur CF, Willman CL: Negative regulation of phagocytosis in Statistical Analyses murine macrophages by the Src kinase family member, Fgr. JExpMed Statisticalanalyseswere performedusing GraphPadPrismsoftware. 191: 515–528, 2000 Data were analyzed by one-way ANOVA followed by the Tukey test 11. Yamao T, Noguchi T, Takeuchi O, Nishiyama U, Morita H, Hagiwara T, for multiple comparisons. For grouped analysis, data were analyzed Akahori H, Kato T, Inagaki K, Okazawa H, Hayashi Y, Matozaki T, Takeda K, Akira S, Kasuga M: Negative regulation of platelet clearance and of by two-way ANOVA followed by Bonferroni post hoc test. A P value , fi the macrophage phagocytic response by the transmembrane glyco- of 0.05 was assumed to indicate a statistically signi cant differ- protein SHPS-1. J Biol Chem 277: 39833–39839, 2002 ence. 12. Chirkov YY, Horowitz JD: Impaired tissue responsiveness to organic nitrates and nitric oxide: A new therapeutic frontier? Pharmacol Ther 116: 287–305, 2007 13. Wink DA, Cook JA, Pacelli R, Liebmann J, Krishna MC, Mitchell JB: Nitric oxide (NO) protects against cellular damage by reactive oxygen ACKNOWLEDGMENTS species. Toxicol Lett 82-83: 221–226, 1995 14. Isenberg JS, Hyodo F, Matsumoto K, Romeo MJ, Abu-Asab M, Tsokos M, This work was supported by National Institutes of Health (NIH) Kuppusamy P, Wink DA, Krishna MC, Roberts DD: Thrombospondin-1 grant R01-HL108954, American Heart Association (AHA) grant limits ischemic tissue survival by inhibiting nitric oxide-mediated vascular smooth muscle relaxation. Blood 109: 1945–1952, 2007 11BGIA7210001 (J.S.I.), and NIH grant R01-HL079207 (P.J.P.); AHA 15. Isenberg JS, Maxhimer JB, Powers P, Tsokos M, Frazier WA, Roberts award 10POST3030009 and K99-HL114648 (G.C.S.); and a C.J. DD: Treatment of liver ischemia-reperfusion injury by limiting throm- Martin Award and AHA award 13POST14520003 (N.M.R.). This bospondin-1/CD47 signaling. Surgery 144: 752–761, 2008 work was also supported by the Institute for Transfusion Medicine, 16. Maxhimer JB, Shih HB, Isenberg JS, Miller TW, Roberts DD: Throm- the Hemophilia Center of Western Pennsylvania, and the Vascular bospondin-1/CD47 blockade following ischemia-reperfusion injury is – Medicine Institute (J.S.I., P.J.P.). tissue protective. Plast Reconstr Surg 124: 1880 1889, 2009 17. Isenberg JS, Annis DS, Pendrak ML, Ptaszynska M, Frazier WA, Mosher DF, Roberts DD: Differential interactions of thrombospondin-1, -2, and -4 with CD47 and effects on cGMP signaling and ischemic injury re- – DISCLOSURES sponses. JBiolChem284: 1116 1125, 2009 18. Barclay AN: Signal regulatory protein alpha (SIRPalpha)/CD47 in- J.S.I. is chair of the Scientific Advisory Boards of Vasculox, Inc. (St. Louis, teraction and function. Curr Opin Immunol 21: 47–52, 2009 MO), and Radiation Control Technologies, Inc. (Rockville, MD), and holds 19. Palen DI, Belmadani S, Lucchesi PA, Matrougui K: Role of SHP-1, equity interests in the same. Kv.1.2, and cGMP in nitric oxide-induced ERK1/2 MAP kinase

1184 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 www.jasn.org BASIC RESEARCH

dephosphorylation in rat vascular smooth muscle cells. Cardiovasc Res 40. Djamali A, Vidyasagar A, Yagci G, Huang LJ, Reese S: Mycophenolic 68: 268–277, 2005 acid may delay allograft fibrosis by inhibiting transforming growth 20. Isenberg JS, Frazier WA, Roberts DD: Thrombospondin-1: A physiological factor-beta1-induced activation of Nox-2 through the nuclear factor- regulator of nitric oxide signaling. Cell Mol Life Sci 65: 728–742, 2008 kappaB pathway. Transplantation 90: 387–393, 2010 21. Isenberg JS, Martin-Manso G, Maxhimer JB, Roberts DD: Regulation of 41. Xu Y, Yang S, Huang J, Ruan S, Zheng Z, Lin J: Tgf-b1 induces au- nitric oxide signalling by thrombospondin 1: Implications for anti- tophagy and promotes apoptosis in renal tubular epithelial cells. Int J angiogenic therapies. Nat Rev Cancer 9: 182–194, 2009 Mol Med 29: 781–790, 2012 22. Isenberg JS, Ridnour LA, Perruccio EM, Espey MG, Wink DA, Roberts 42. Khanna AK, Xu J, Baquet C, Mehra MR: Adverse effects of nicotine and DD: Thrombospondin-1 inhibits endothelial cell responses to nitric immunosuppression on proximal tubular epithelial cell viability, tissue oxide in a cGMP-dependent manner. Proc Natl Acad Sci U S A 102: repair and oxidative stress gene expression. J Heart Lung Transplant 13141–13146, 2005 28: 612–620, 2009 23. Palmer RF, Lasseter KC: Drug therapy. Sodium nitroprusside. NEnglJ 43. Simão S, Fraga S, Jose PA, Soares-da-Silva P: Oxidative stress plays a Med 292: 294–297, 1975 permissive role in alpha2-adrenoceptor-mediated events in immortal- 24. Reves JG, Sheppard LC, Wallach R, Lell WA: Therapeutic uses of so- ized SHR proximal tubular epithelial cells. Mol Cell Biochem 315: 31– dium nitroprusside and an automated method of administration. Int 39, 2008 Anesthesiol Clin 16: 51–88, 1978 44. Du C, Wang S, Diao H, Guan Q, Zhong R, Jevnikar AM: Increasing re- 25. Maile LA, Badley-Clarke J, Clemmons DR: The association between sistance of tubular epithelial cells to apoptosis by shRNA therapy integrin-associated protein and SHPS-1 regulates insulin-like growth ameliorates renal ischemia-reperfusion injury. Am J Transplant 6: factor-I receptor signaling in vascular smooth muscle cells. Mol Biol Cell 2256–2267, 2006 14: 3519–3528, 2003 45. Guo N, Krutzsch HC, Inman JK, Roberts DD: Thrombospondin 1 and 26. Oldenborg PA, Zheleznyak A, Fang YF, Lagenaur CF, Gresham HD, type I repeat peptides of thrombospondin 1 specifically induce apo- Lindberg FP: Role of CD47 as a marker of self on red blood cells. Sci- ptosis of endothelial cells. Cancer Res 57: 1735–1742, 1997 ence 288: 2051–2054, 2000 46. Halloran PF, Homik J, Goes N, Lui SL, Urmson J, Ramassar V, Cockfield 27. Maile LA, Clemmons DR: Integrin-associated protein binding domain SM: The “injury response”: a concept linking nonspecific injury, acute of thrombospondin-1 enhances insulin-like growth factor-I receptor rejection, and long-term transplant outcomes. Transplant Proc 29: 79– signaling in vascular smooth muscle cells. Circ Res 93: 925–931, 2003 81, 1997 28. Tulasne D, Judd BA, Johansen M, Asazuma N, Best D, Brown EJ, Kahn 47. Gobé G, Willgoss D, Hogg N, Schoch E, Endre Z: Cell survival or death M, Koretzky GA, Watson SP: C-terminal peptide of thrombospondin- in renal tubular epithelium after ischemia-reperfusion injury. Kidney Int 1 induces platelet aggregation through the Fc receptor gamma- 56: 1299–1304, 1999 chain-associated signaling pathway and by agglutination. Blood 98: 48. Ambrosio G, Zweier JL, Duilio C, Kuppusamy P, Santoro G, Elia PP, – 3346 3352, 2001 Tritto I, Cirillo P, Condorelli M, Chiariello M, et al: Evidence that mito- 29. Heasman J: Morpholino oligos: Making sense of antisense? Dev Biol chondrial respiration is a source of potentially toxic oxygen free radicals – 243: 209 214, 2002 in intact rabbit hearts subjected to ischemia and reflow. JBiolChem 30. Isenberg JS, Romeo MJ, Abu-Asab M, Tsokos M, Oldenborg A, Pappan 268: 18532–18541, 1993 L, Wink DA, Frazier WA, Roberts DD: Increasing survival of ischemic 49. Li C, Jackson RM: Reactive species mechanisms of cellular hypoxia- – tissue by targeting CD47. Circ Res 100: 712 720, 2007 reoxygenation injury. Am J Physiol Cell Physiol 282: C227–C241, 2002 31. Vernon-Wilson EF, Kee WJ, Willis AC, Barclay AN, Simmons DL, Brown 50. Chatterjee PK: Novel pharmacological approaches to the treatment of MH: CD47 is a ligand for rat macrophage membrane signal regulatory renal ischemia-reperfusion injury: A comprehensive review. Naunyn – protein SIRP (OX41) and human SIRPalpha 1. Eur J Immunol 30: 2130 Schmiedebergs Arch Pharmacol 376: 1–43, 2007 2137, 2000 51. Laurindo FR, Fernandes DC, Santos CX: Assessment of superoxide 32. Calzada MJ, Zhou L, Sipes JM, Zhang J, Krutzsch HC, Iruela-Arispe ML, production and NADPH oxidase activity by HPLC analysis of dihy- Annis DS, Mosher DF, Roberts DD: Alpha4beta1 integrin mediates droethidium oxidation products. Methods Enzymol 441: 237–260, selective endothelial cell responses to thrombospondins 1 and 2 in vitro 2008 and modulates angiogenesis in vivo. Circ Res 94: 462–470, 2004 52. Liu J, Ormsby A, Oja-Tebbe N, Pagano PJ: Gene transfer of NAD(P)H 33. Isenberg JS, Romeo MJ, Maxhimer JB, Smedley J, Frazier WA, Roberts oxidase inhibitor to the vascular adventitia attenuates medial smooth DD: Gene silencing of CD47 and antibody ligation of thrombospondin- muscle hypertrophy. Circ Res 95: 587–594, 2004 1 enhance ischemic tissue survival in a porcine model: Implications for 53. Nediani C, Raimondi L, Borchi E, Cerbai E: Nitric oxide/reactive oxygen human disease. Ann Surg 247: 860–868, 2008 species generation and nitroso/redox imbalance in heart failure: From 34. Hamilton CA, Brosnan MJ, Al-Benna S, Berg G, Dominiczak AF: NAD(P) molecular mechanisms to therapeutic implications. Antioxid Redox H oxidase inhibition improves endothelial function in rat and human Signal 14: 289–331, 2011 blood vessels. Hypertension 40: 755–762, 2002 54. van Beek EM, Zarate JA, van Bruggen R, Schornagel K, Tool AT, 35. Lassègue B, Griendling KK: NADPH oxidases: functions and pathologies Matozaki T, Kraal G, Roos D, van den Berg TK: SIRPa controls the ac- in the vasculature. Arterioscler Thromb Vasc Biol 30: 653–661, 2010 tivity of the phagocyte NADPH oxidase by restricting the expression of 36. Huang J, Kleinberg ME: Activation of the phagocyte NADPH oxidase gp91(phox). Cell Rep 2: 748–755, 2012 protein p47(phox). Phosphorylation controls SH3 domain-dependent 55. Johansen ML, Brown EJ: Dual regulation of SIRPalpha phosphorylation binding to p22(phox). JBiolChem274: 19731–19737, 1999 by integrins and CD47. JBiolChem282: 24219–24230, 2007 37. Squadrito GL, Pryor WA: The formation of peroxynitrite in vivo from 56. Brown EJ, Frazier WA: Integrin-associated protein (CD47) and its li- nitric oxide and superoxide. Chem Biol Interact 96: 203–206, 1995 gands. Trends Cell Biol 11: 130–135, 2001 38. Medina P, Segarra G, Martínez-León JB, Vila JM, Otero E, Lluch S: 57. Maile LA, Allen LB, Hanzaker CF, Gollahon KA, Dunbar P, Clemmons Relaxation and cGMP formation in response to sildenafil and sodium DR: Glucose regulation of thrombospondin and its role in the modu- nitroprusside in saphenous veins from normotensive and hypertensive lation of smooth muscle cell proliferation. Exp Diabetes Res 2010: patients. Am J Hypertens 15: 798–802, 2002 2010 39. Simonsen U, Christensen FH, Buus NH: The effect of tempol on en- 58. Csányi G, Yao M, Rodríguez AI, Al Ghouleh I, Sharifi-Sanjani M, dothelium-dependent vasodilatation and blood pressure. Pharmacol Frazziano G, Huang X, Kelley EE, Isenberg JS, Pagano PJ: Thrombo- Ther 122: 109–124, 2009 spondin-1 regulates blood flow via CD47 receptor-mediated activation

J Am Soc Nephrol 25: 1171–1186, 2014 TSP1–SIRP-a Signaling Regulates ROS and Blood Flow 1185 BASIC RESEARCH www.jasn.org

of NADPH oxidase 1. Arterioscler Thromb Vasc Biol 32: 2966–2973, 66. Wu H, Chen G, Wyburn KR, Yin J, Bertolino P, Eris JM, Alexander SI, 2012 Sharland AF, Chadban SJ: TLR4 activation mediates kidney ischemia/ 59. Sohn HY, Raff U, Hoffmann A, Gloe T, Heermeier K, Galle J, Pohl U: reperfusion injury. JClinInvest117: 2847–2859, 2007 Differential role of angiotensin II receptor subtypes on endothelial su- 67. Bauer PM, Bauer EM, Rogers NM, Yao M, Feijoo-Cuaresma M, Pilewski peroxide formation. Br J Pharmacol 131: 667–672, 2000 JM, Champion HC, Zuckerbraun BS, Calzada MJ, Isenberg JS: Acti- 60. Briones AM, Tabet F, Callera GE, Montezano AC, Yogi A, He Y, Quinn vated CD47 promotes pulmonary arterial hypertension through tar- MT, Salaices M, Touyz RM: Differential regulation of Nox1, Nox2 and geting caveolin-1. Cardiovasc Res 93: 682–693, 2012 Nox4 in vascular smooth muscle cells from WKY and SHR. JAmSoc 68. Touyz RM, Yao G, Schiffrin EL: c-Src induces phosphorylation and Hypertens 5: 137–153, 2011 translocation of p47phox: Role in superoxide generation by angio- 61. Kawahara T, Ritsick D, Cheng G, Lambeth JD: Point mutations in the tensin II in human vascular smooth muscle cells. Arterioscler Thromb proline-rich region of p22phox are dominant inhibitors of Nox1- and Vasc Biol 23: 981–987, 2003 Nox2-dependent reactive oxygen generation. J Biol Chem 280: 69. Csányi G, Cifuentes-Pagano E, Al Ghouleh I, Ranayhossaini DJ, Egaña 31859–31869, 2005 L, Lopes LR, Jackson HM, Kelley EE, Pagano PJ: Nox2 B-loop peptide, 62. Isenberg JS, Wink DA, Roberts DD: Thrombospondin-1 antagonizes Nox2ds, speciﬁcally inhibits the NADPH oxidase Nox2. Free Radic Biol nitric oxide-stimulated vascular smooth muscle cell responses. Car- Med 51: 1116–1125, 2011 diovasc Res 71: 785–793, 2006 70. Cascino T, Csanyi G, Al Ghouleh I, Montezano AC, Touyz RM, Haurani 63. Baud L, Ardaillou R: Involvement of reactive oxygen species in kidney MJ, Pagano PJ: Adventitia-derived hydrogen peroxide impairs re- damage. Br Med Bull 49: 621–629, 1993 laxation of the rat carotid artery via smooth muscle cell p38 mitogen- 64. Padanilam BJ: Cell death induced by acute renal injury: A perspective activated protein kinase. Antioxid Redox Signal 15: 1507–1515, 2011 on the contributions of apoptosis and necrosis. Am J Physiol Renal Physiol 284: F608–F627, 2003 65. Matozaki T, Murata Y, Okazawa H, Ohnishi H: Functions and molecular mechanisms of the CD47-SIRPalpha signalling pathway. Trends Cell This article contains supplemental material online at http://jasn.asnjournals. Biol 19: 72–80, 2009 org/lookup/suppl/doi:10.1681/ASN.2013040433/-/DCSupplemental.

1186 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 1171–1186, 2014 Erratum

CORRECTION Production and Promotes Renal Ischemia Reperfusion Injury. J Am Soc Nephrol 25: 1171–1186, 2014. Yao M, Rogers NM, Csányi G, Rodriguez AI, Ross MA, St. Please note the following should have been included in the Croix C, Knupp H, Novelli EM, Thomson AW, Pagano PJ, Acknowledgments section for the abovepublishedarticle in the Isenberg JS. Thrombospondin-1 Activation of Signal- June 2014 issue of JASN. “This work was also supported by Regulatory Protein-a Stimulates Reactive Oxygen Species P30-DK079307 (J.S.I.).”

1884 ISSN : 1046-6673/2508-1884 J Am Soc Nephrol 25: 1884, 2014