Renal Tubular Cell-Derived Extracellular Vesicles Accelerate the Recovery of Established Renal Ischemia Reperfusion Injury

BASIC RESEARCH www.jasn.org

† ‡ ‡ Jesus H. Dominguez,* Yunlong Liu, Hongyu Gao, James M. Dominguez II,* Danhui Xie,* and K. J. Kelly*

*Nephrology Division, Department of Medicine, and ‡Department of Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, Indiana; and †Roudebush Veterans Administration Hospital, Indianapolis, Indiana

ABSTRACT Ischemic renal injury is a complex syndrome; multiple cellular abnormalities cause accelerating cycles of inflammation, cellular damage, and sustained local ischemia. There is no single therapy that effectively resolves the renal damage after ischemia. However, infusions of normal adult rat renal cells have been a successful therapy in several rat renal failure models. The sustained broad renal benefit achieved by relatively few donor cells led to the hypothesis that extracellular vesicles (EV, largely exosomes) derived from these cells are the therapeutic effector in situ. We now show that EV from adult rat renal tubular cells significantly improved renal function when administered intravenously 24 and 48 hours after renal ischemia in rats. Additionally, EV treatment significantly improved renal tubular damage, 4-hydroxynanoneal adduct formation, neutrophil infiltration, fibrosis, and microvascular pruning. EV therapy also markedly reduced the large renal transcriptome drift observed after ischemia. These data show the potential utility of EV to limit severe renal ischemic injury after the occurrence.

J Am Soc Nephrol 28: ccc–ccc, 2017. doi: https://doi.org/10.1681/ASN.2016121278

Renal ischemia causes and frequently worsens exist- Ischemic preconditioning (IPC) is a highly con- ing AKI1 and CKD.2 AKI is morbid, potentially served adaptation in which transient, sublethal lethal,3–9 and an emergent medical problem.10 Un- ischemia protects from subsequent ischemic in- fortunately, there is no effective therapy for AKI.11 jury.21–25 However, translating the powerful effects The medical failure to affect ischemic AKI is not sur- of IPC to treat unpredictable renal ischemia is not prising given its unpredictability and inherent com- feasible.26 It is not known how IPC confers protec- plexity.12,13 Therefore, it is unlikely that this broad tion, although an important clue is that IPC can be organ malfunction can be corrected by targeting sin- transferred by the blood of IPC donors,27,28 i.e., gle dysfunctional processes.13 The realization that remote IPC (RIPC). One possible mechanism of renal ischemia also aggravates progressive CKD, re- RIPC transfer is by EV, released membrane-bound gardless of its initial cause, is an additional impetus in the search for solutions.2,14,15 In this report, we de- scribe the use of renal cell extracellular vesicles (EV) Received December 2, 2016. Accepted June 12, 2017. to treat ischemic renal injury in rats. We derived this Published online ahead of print. Publication date available at work from prior experimentation that showed im- www.jasn.org. proved renal function and structure by relatively Correspondence: Dr. K.J. Kelly, Department of Medicine (Ne- 16–20 small numbers of infused renal cells. We suggest phrology), Indiana University School of Medicine, 950 West that the beneﬁcial effect of those donor cells was due Walnut Street, RII 202, Indianapolis, IN 46202. Email: kajkelly@iu. to their EV, and tested the hypothesis that renal EV edu can protect ischemic kidneys. Copyright © 2017 by the American Society of Nephrology

J Am Soc Nephrol 28: ccc–ccc, 2017 ISSN : 1046-6673/2812-ccc 1 BASIC RESEARCH www.jasn.org nanovesicles that mediate cell-to-cell communication and convey the distinctive state of originating cells.29–31 In this work, we show that EV from adult rat renal tubular cells can be delivered by peripheral vein to successfully treat rats with established severe ischemia/reperfusion AKI. Furthermore, EV from hypoxia preconditioned (IPC) cells are more effective, providing greater improvements in renal phenotype and genotype than EV from normoxic cells.

RESULTS

EV We employed two sets of EV (exosomes) derived from the same primary rat renal tubular cells: one set maintained in normoxic conditions (EXO) and the other exposed to 1% O2 hypoxia for 4 hours before collection of EV (IPC EXO). Cell number and EV protein content, size, and vesicle number were not different between EXO and IPC EXO (Table 1). Nano tracker analysis (Supplemental Figure 1) and electron microscopy showed similar size distribution between EV groups. The EV contained mRNAs encoding the male determinant SRY gene as well as the transfected serum amyloidA1 Figure 1. Donor EV in recipient kidneys. (A and B) ICAM1 and transcript. These were used to track EV in recipient kidneys. In CD63 protein levels in normoxic EV (EXO) and IPC EV (IPC EXO), addition, EV tracking was achieved by labeling a 10% fraction measured on western blot, n=5. ICAM1 was higher (P=0.001) and of EV with red Exo-Glow. This maneuver showed the exosomes CD63 lower (P,0.01) in IPC EXO. (C) Electron microscopy broadly distributed within the kidney (Figure 1). Cell culture showed IPC EXO and EXO (not shown) averaged ,200 nM di- studies demonstrated that EV were rapidly taken up by proximal ameter. (D) RT-PCR was used to identify the SRY male de- tubular epithelia; IPC EXO more than EXO (Supplemental termining gene in EV (EXO) from male donor cells (positive Figure 2). This result is consistent with higher levels of adherent control), and also in kidneys of female EV recipients (IPC EXO intercellular cell adhesion molecule1 (ICAM1),32 which report- and EXO), but not in female kidneys of rats injected with saline edly promotes EV cell fusion33 (Figure 1). (saline). Additional SRY-positive controls were amplified from normal male rat kidneys. (E) Representative image of three rat Renal Function kidneys used to localize IPC EV labeled with red EXO-Glow. A 10% fraction of IPC EV was labeled, remixed with the unlabeled Four groups of five Sprague Dawley (SD) rats were studied: 90%, injected, and visualized by fluorescent microscopy in 50-mM sham surgery (controls), bilateral renal ischemia injected with vibratome sections. (F) RT-PCR was used to identify the murine saline vehicle (untreated ischemia), or with EV from renal cells SAA1 gene amplified in donor EV derived from cells transfected cultured under normoxic conditions (EXO), or with EV from with SAA1 (EXO), and also in recipient kidneys of rats injected posthypoxic kidney cells (IPC EXO). The sample size was de- with EXO and IPC EXO (EXO and IPC EXO), but not in kidneys of termined by power analysis.34 The goal was to treat after the rats injected with saline (saline). The negative and positive con- injury and EV or saline vehicle were given intravenously 24 trols were amplifications of pCDNA1 and SAA1-pCDNA1 plas- and 48 hours postischemia. There was a rapid and similar in- mids used for transfection. Ctrl, control; MW, molecular weight; crease in serum creatinine in all ischemic groups after 24 pos, positive. hours, and it decreased slowly in the ischemic rats given saline (Figure 2). In contrast, creatinine improved rapidly in ischemic rats injected with EXO, and the recovery was even faster in rats treated with IPC EXO (P,0.02 versus EXO). Table 1. Exosome parameters Creatinine levels in sham-operated rats remained within nor- Parameter EXO IPC EXO P Value mal limits (Figure 2). Similarly, serum urea nitrogen demonstrated accelerated recovery in the postischemic rats treated Cells/flask 12.93106673105 13.43106653105 NS with EXO or IPC EXO (not shown). Untreated ischemic rats Protein/flask, mg706791629 NS Vesicle number/flask 1.260.13107 1.160.073107 NS (saline) also had abnormally large kidneys, and EV, particu- , fi Mean EV size, nm 10063.94 10663.78 NS larly IPC EXO (P 0.01), signi cantly limited organomegaly, fl fl fi Data are mean6SEM. EV derived from normoxic (EXO) and IPC renal cells likely a re ection of improved in ammation and brosis. No were comparable in size and number. mortality was observed.

2 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 www.jasn.org BASIC RESEARCH

(HNE).37 Whereas renal HNE formation was nearly undetectable in sham rats, strong HNE formation existed in untreated ischemic rats (Figure 3B); EXO and IPC EXO significantly limited the HNE labeling. Analysis of specific redox transcripts indicates that ischemia alters redox balance as compared with either sham or IPC EXO groups. Thus, treatment with EV from preconditioned cells (IPC EXO) restores renal genetic profiles to near that of nonischemic (sham) animals. Renal inflammation complicates renal ischemia38 and, among the aggregates of renal inflammatory cells, short-lived neutrophils32 point to ongoing inflammation.38 Neutrophils in sham kidneys were rare, whereas kidneys from untreated ischemic rats contained large numbers of neutrophils (Figure 3C). In contrast, EXO and IPC EXO curtailed renal neutrophil Figure 2. Improved renal function and kidney weight after accumulation postischemia. Another sensitive indicator of in- treatment with EV. Left: Serum creatinine levels in the four groups flammation postischemia is upregulation of renal comple- of rats, n=5: Sham (open squares), untreated ischemic (filled tri- ment components.39 In Figure 3D, we show that glomerular angles, saline), ischemic treated with normoxic EV (filled circles, C3 immuno-reactivity, near background in sham operated EXO), and ischemic treated with IPC EV (filled diamonds, IPC rats, was markedly increased in untreated ischemia. In con- EXO). Time zero designates day of ischemia/reperfusion; EV or trast, EXO, and particularly IPC EXO, attenuated C3 protein saline were injected 1 and 2 days postischemia. (*) Next to sham expression. Moreover, the renal transcriptome in ischemia fi , symbols indicate signi cantly different than all others (P 0.05). (*) revealed broader activation, including C2, C3, and C4, with fi Next to IPC EXO symbols indicate signi cant difference from attenuation of the membrane attack complex inhibitor CD55. saline (P,0.05), and (§) indicates significantly different from EXO We also found activation of mRNAs encoding proinflamma- (P,0.05). (*) Next to EXO symbols designate significant difference from saline (P,0.05). Right: Kidney weights corrected for tory chemokines, cytokines, and adhesion molecules. These body weight. (*) On saline (untreated ischemia) bar indicates responses were suppressed by EXO and IPC EXO treatment, significantly different than all others. (*) On EXO and IPC EXO with significantly lower levels of proinflammatory transcripts indicate significantly different than sham (P,0.05). in the IPC EXO group. Substantial peritubular fibrosis followed renal ischemia within 6 days (Figure 3E). Treatment with EXO significantly Renal Phenotype and Genotype limited the broad activation of profibrotic genes and decreased Untreated renal ischemia caused severe disruption of renal renal fibrosis. Infusions of IPC EXO were even more effective architecture, including tubular congestion and simplification, at limiting fibrosis in postischemic kidneys. It is noteworthy intratubular obstruction with casts, and inflammation (as that ischemia upregulated genes that break down extracellular compared with sham-operated controls, Figure 3A). In con- matrix, including MMP2, 7, 11, and 14, but also the MMP trast, injection of EXO, and more so IPC EXO, limited each of inhibitor TIMP1. IPC EXO limited the amplitude of these these histologic abnormalities (Supplemental Figure 3). Heat- ischemic responses. inactivated EV were ineffective in limiting postischemic injury Renal ischemia severely compromised the renal microvas- (Supplemental Table 1), consistent with a role of EV RNA or culature, consistent with prior observations.40 Renal sections, specific proteins in the protection observed. Thus, compre- labeled with Lycopersicon Esculentum lectin (LEL), showed hensive genetic profiling was performed and revealed signifi- severe loss of capillaries, or of their lectin epitopes, in kidneys cant activation of both pro- and antiapoptotic transcripts in from untreated postischemic rats (Figure 3F). This vascular untreated postischemic kidneys, when compared with sham rarefication was prevented, in part, by EXO, and, for the most and IPC EXO groups (Figure 3). Renal injury in untreated part, by IPC EXO. The renal genotype revealed that ischemia ischemia also included enhanced cellular proliferation, as in- suppressed angiogenesis, in that angiostatic genes were acti- dicated by higher protein levels of proliferative cell nuclear vated whereas proangiogenic genes were inhibited. These an- antigen (PCNA, a cell proliferation marker), as reported.35 tiangiogenic effects were antagonized by IPC EXO. Although Treatment with EXO or IPC EXO limited the proliferative renal ischemia severely decreased glomerular number (2.496 postischemic response (Supplemental Figure 4). Although 0.2/hpf versus 5.5860.2 in sham), glomeruli were relatively tubular cell proliferation contributes to recovery, proliferation preserved after EV treatment (3.4660.2 in EXO; 4.1260.2 in of leukocytes and fibroblasts is detrimental,36 thus lower pro- IPC EXO; P,0.001 versus untreated ischemia). liferation in the treated groups is likely a manifestation of decreased tubular injury. Western Blots The role of EV on oxidant stress induced by ischemia was Three groups of proteins were measured on immunoblots; examined in sections stained for 4-hydroxynonenal adducts catalase (CAT) and SOD1 are constituents of the antioxidant

J Am Soc Nephrol 28: ccc–ccc, 2017 Exosomes Treat Renal Ischemia 3 BASIC RESEARCH www.jasn.org

Figure 3. Improved renal structure after treatment with EV. (A) Representative images of renal sections stained with PAS. Left to right: sham rats, ischemic untreated rats (saline), ischemic rats treated with normoxic EV (EXO), and ischemic rats treated with IPC EV (IPC EXO). Tubular damage was widespread in untreated ischemia, including cystic-like tubular dilation. These changes were partly corrected by EXO, and even more by IPC EXO. In the bar diagram is shown the mean fraction of damaged tubules per field in each group: sham (n=20) 2 2 and saline groups (n=20) were significantly different (P=2.03310 16); sham and EXO (n=20; P=1.27 12) and sham and IPC EXO (n=20; 2 2 8.92 10) were also significantly different; EXO was significantly higher than IPC EXO (P=1.37 08). (*) Represents significantly different

4 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 www.jasn.org BASIC RESEARCH system. Components of the hypoxia inducible factor (HIF) Hsp27 upregulation (Supplemental Figure 5). Consistent with pathway, vascular endothelial growth factor (VEGF), glucose histologic and RNAseq data, immunoblots also showed in- transporter1 (GLUT1 or Slc2a1), and erythropoietin (EPO), creases in TGFb1, fas, and BIRC3 postischemia; these were were included. Heat shock protein 27 (Hsp27) was included all prevented with EV treatment (Supplemental Figure 6). as a presumptive reno-protective protein. At 6 days postischemia, levels of CAT, SOD1, VEGF, EPO, and GLUT1 were re- mRNA Pathways duced in untreated ischemic rats. However, treatment with Sequencing by RNAseq identiﬁed 12,159 renal transcripts. IPC EXO largely prevented loss of these proteins. On the other There were 3141 renal genes signiﬁcantly altered in untreated hand, protein levels of renal HIF were not different 6 days ischemia as compared with the normal (sham) group. The postischemia (not shown). In contrast, Hsp27 protein levels number of altered genes was reduced to 1599 in the EXO- were elevated postischemia, and IPC EXO treatment prevented treated postischemic group, and IPC EXO further reduced

from sham, and (§) indicates significant difference between IPC EXO and EXO, P,0.05 for both. Specific renal proapoptotic transcripts were significantly activated by untreated ischemia (NS) when compared with sham or IPC EV–treated ischemic rats (IPC) (P,0.05). (B) Representative images of renal HNE. Heavy formation of HNE adducts was seen in untreated ischemic (saline) rats, whereas it was nearly undetectable in sham rats. Treatment with normoxic EV (exosome) or IPC EV (IPC EXO) limited the formation of HNE adducts. The bar 2 diagram shows that HNE formation in the saline group (n=13) was significantly higher than in sham (n=5; P=2.74310 6); sham and EXO (n=5) groups were also significantly different (P=0.02), but sham and IPC EXO (n=6) groups were not (P=0.05). EXO (P,0.01) and EXO IPC groups (P,0.001) were lower than the untreated ischemic group, whereas the IPC EXO group was lower than the EXO group (P,0.05). (*) Significantly different than sham group, P,0.05. Specific redox-related gene transcripts were reduced in the untreated ischemia group with respect to sham and also IPC EXO groups, indicating partial restoration of redox gene expression by IPC EXO treatment. In contrast, the pro-oxidant gene Olr (or Lox1) was activated in ischemia with respect to sham, and it was reduced by IPC EXO treatment. (C) Rep- resentative images of renal sections labeled with anti-neutrophil antibody. Large congregations of neutrophils were detected in the interstitial and intraluminal spaces of kidneys from untreated ischemic rats (saline), whereas were barely seen in kidneys of sham rats. In contrast, ischemic rats treated with EXO or IPC EXO had much lower numbers of neutrophils. The bar diagram shows mean neutrophil number per 4003 field (PMN/hpf) in sham rats (sham, n=27 fields), untreated ischemic (saline, n=47), ischemic rats treated with normoxic 2 EV (EXO, n=31), and ischemic rats treated with IPC EV (IPC EXO, n=32). Sham and saline groups were significantly different (P=1.94310 5), 2 sham and EXO groups were also significantly different (P=1.05310 5), but sham and IPC EXO groups were not (P=0.65). In addition, 2 2 EXO (P=8.74310 5)andIPCEXO(P=2.72310 5) were significantly lower than saline, and IPC EXO was significantly lower than EXO (P,0.001). Proinflammatory renal transcripts, activated in untreated ischemia and compared with sham rats or IPC-treated ischemic rats by fold change, include cytokines, chemokines, complement components, and adhesion molecules. In most cases, IPC EXO partly corrected the activation induced by ischemia (NS). (D) Glomerular complement C3 expression in sham rats was nearly undetectable, but it was heavily expressed in untreated ischemic rat kidneys. Treatment with EXO and IPC EXO limited C3 expression in ischemic rats. The bar diagram shows average pixel density of C3 immunofluorescence per 4003 power field in sham rats (sham, n=24 fields), untreated ischemic (saline, n=16), ischemic rats treated with normoxic EV (EXO, n=8), and ischemic rats treated with IPC EV (IPC EXO, n=24). Saline was higher 2 2 than sham group (P=5.97310 26); sham and EXO were also significantly different (P=1.13310 14), as well as sham and IPC EXO 2 2 (P=1.78310 8). In addition, EXO (P,0.001) and IPC EXO (9.43310 10) were significantly lower than saline, and IPC EXO was significantly lower than EXO (P,0.01). (*) Significantly different than sham group, P,0.05; (§) indicates significant difference between IPC EXO and EXO, P,0.05. (E) Representative images of renal sections stained with Masson’s trichrome. In untreated ischemia (saline), interstitial renal fibrosis (blue) was extensive and replaced normal epithelia (red) when compared with controls (sham). Treatment of ischemia with normoxic EV (exosome) or IPC EV (IPC EXO) limited fibrosis. The bar diagram shows average fractional fibrosis in control (sham, n=16), untreated ischemia (saline, n=20), ischemia treated with normoxic EV (EXO, n=20), and ischemia treated with IPC EV (IPC EXO, n=20). Sham and 2 saline were significantly different (P=2.94310 6), sham and EXO were also significantly different (P,0.01), but sham and IPC EXO were 2 not (P=0.08). In addition, EXO (P,0.001) and IPC EXO (P=3.31310 6) were significantly lower than saline, and IPC EXO was significantly lower than EXO (P,0.02). (*) Significantly different than sham, P,0.05; (§) indicates significant difference between IPC EXO and EXO, P,0.05. Untreated ischemia (NS) activated a suite of profibrotic transcripts and the fold change among NS and sham and IPC EV (IPC EXO) is shown. These activated transcripts encoded extracellular matrix proteins, three members of the TGFb family, metalloproteinases, inhibitors of metalloproteinases, transcription factors, and other genes linked to renal fibrosis. (F) Representative images of renal peritubular microvasculature labeled with DyLight 594 conjugated Lycopersicon Esculentum (tomato lectin, LEL, red). The microvasculature was far less dense in untreated ischemia (saline, n=92) than in sham (n=62). However, treatment with normoxic EV (EXO, n=53) or IPC EV (IPC EXO, n=81) preserved microvascular networks in ischemic kidneys. The bar diagram shows the average fractional (fluo- 2 2 2 rescence) positivity; sham and saline were significantly different (P=2.03 16), EXO (P=1.27 12) and IPC EXO (P=8.92 10) were also sig- 2 2 nificantly lower than sham. However, EXO (P=8.81 11) and IPC EXO (P=7.80 17) were also significantly higher than saline, whereas IPC EXO 2 was significantly higher than EXO (P=1.37 08). (*) Significantly different than sham, P,0.05; (§) indicates significant difference between IPC EXO and EXO, P,0.05. Untreated ischemia (NS) inhibited key proangiogenic transcripts, but also activated others—Ang, Adra, Cyr61, as well as antiangiogenic Thbs1. The significant (P,0.05) fold change among NS and sham and IPC EV (IPC EXO) is shown. Hpf, high power field; PAS, periodic acid Schiff; PMN, polymorphonuclear; V, versus.

J Am Soc Nephrol 28: ccc–ccc, 2017 Exosomes Treat Renal Ischemia 5 BASIC RESEARCH www.jasn.org the number to 71 alteredtranscripts, oronly 2% of those altered KEGG pathways (http://www.genome.jp/kegg/pathway.html; in the untreated ischemic group. Hence, renal ischemia mark- Kyoto University, Japan). The pathways altered postischemia edly altered the normal renal transcriptome. However, EXO included, in descending order of statistical significance, met- treatment partly corrected and IPC EXO nearly fully corrected abolic, cell proliferation, focal adhesion, solute absorption, the large drift caused by ischemia. These effects can be visual- cell cycle, extracellular matrix receptor interaction, oxidative ized on a multidimensional scaling plot which reports the phosphorylation, glutathione metabolism, phagosome, distances, or the inverse of similarities, among the groups apoptosis, cytokine-cytokine receptor interaction, and DNA (Figure 4). Distances on the multidimensional scaling plot replication (Supplemental Table 2). Compared with sham, correspond to leading log–fold changes between each pair of renal ischemia caused the largest changes in metabolic path- samples. The leading log–fold change is the average (root ways. Specifically, the largest number of altered transcripts mean square) of the largest absolute log–fold changes between were involved in metabolism and achieved the highest statis- each pair of samples. The dimensions on the plot are leading tical significance. Treatment with IPC EXO effectively correc- log–fold changes, and demonstrate the genetic drift postische- ted the perturbations in metabolic gene expression such that mia and its almost complete correction by IPC EXO. It is also expression was similar to sham (P.0.05). Postischemia, renal noted that significant differences in gene expression among transcripts were partly normalized by IPC EXO treatment; the four groups were determined by the false discovery rate 1526 were significantly higher and 1593 lower than in un- (FDR,0.05), and only those genes that met this cutoff were treated rats. A subset of these transcripts was also corrected included. by EXO treatment. In analyzing the transcripts protected by All transcriptomes were then used for pathway analysis with both IPC EXO and EXO treatment we found 39 antiapoptotic WebGestalt (Vanderbilt University, Nashville, TN), and the and 29 anti-inflammatory transcripts. The most significant generated data sets were incorporated into well defined KEGG pathway was metabolic and by GO analysis it was transporter activity (Supplemental Figure 7). RNAseq was used to measure 312 EV transcripts in RNA extracted from EXO and IPC EXO. There were 24 differentially expressed transcripts between EXO and IPC EXO (Supplemen- tal Table 3). Ribosomal RNAs were abundant in renal EV, as shown by others.41 mRNAs encoding cytoplasmic ribosomal proteins Rps6 (2.3-fold) and Rps13 (1.3-fold) were also upregulated in IPC EXO. In contrast, mRNAs for structural Rn28S (0.28) and Rn45S (0.30) appeared suppressed (Supplemental Table 3). IPC EXO also carried higher mRNA levels for Drd4 (78-fold increased), a gene that protects from ischemic injury,42 as well as Lrrc56 (16-fold increased), and Ptdss2 (eight-fold increased). Interestingly, Ptdss enhances recovery from ischemic injury.43 IPC EXO also contained higher levels of Eps8l2 (4.5-fold increased), a transcript involved in actin remodeling.44 Phrf1, a transcript that participates in RNA processing,45 was increased 3.9-fold. Ugt2b1 and Ugt2b10, genes that encode detoxification enzymes reportedly suppressed by hypoxia,46 were each increased 3.2-fold. Rassf7, a transcript 47 Figure 4. Correction of transcriptome alterations after treatment activated by hypoxia that regulates microtubule dynamics, with EV. Top, significantly altered genes with respect to control was increased 2.9-fold. Although our EV mRNA compendium (sham) in untreated ischemia (saline = 3141 altered genes), is not complete and its contents likely reflect only a portion of treatment with normoxic EV (EXO = 1599 altered genes), and the entire renal EV library,41 we detected significant differences treatment with IPC EV (IPC EXO = 71 altered genes). A total of between IPC EXO and EXO among the 312 EV transcripts mea- 12,159 transcripts were sequenced and identified per group. sured. The most significantly altered transcripts detected by Bottom, multidimensional scaling plot, where distances corre- Kegg pathway analysis were associated with the ribosome – spond to leading log fold changes between each pair of sam- whereas GO analysis detected alterations in RNA processing – ples. The leading log fold change is the average (root mean (Supplemental Figure 8). square) of the largest absolute log–fold changes between each pair of samples. The larger shapes represent the mean for the transcriptomes in each group and the smaller shapes represent DISCUSSION individual rat transcriptomes. Filled diamonds = sham; gray tri- angles = untreated ischemic; striped circles = ischemic treated with normoxic, or neutral, EV; and speckled squares = treated Ischemic injury to the kidney and other organs can be deadly. with IPC EV. FC, fold change; Vs, versus. Even with subsequent recovery, AKI is still linked to short- and

6 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 www.jasn.org BASIC RESEARCH long-term mortality.7,48 Chronic/recurrent renal ischemia also The renal transcriptome analysis revealed a major drift in promotes the progression of CKD to ESRD.2,14,15 Moreover, transcriptomes postischemia away from normal (sham), ischemic AKI is common in transplanted kidneys, and is the whereas IPC EV largely contained the drift. Gene pathway main cause of delayed graft function.49,50 analysis revealed drastic modifications by ischemia whereas Unfortunately, there is no specific therapy for ischemic AKI, IPC EXO prevented most alterations. These findings are con- and supportive care has not improved outcomes.51 Thus, new sistent with a state of mixed renal dysfunction and dormancy/ approaches to AKI are needed. Animal and human data are “stunning” for up to 24 hours after severe ischemia, which is consistent, showing that no single pathogenic event can be partly resolved by IPC EXO. We presume that structural and targeted to completely halt ischemic renal injury. Instead, cargo differences between EXO and IPC EXO might account data suggest that multiple abnormalities are intertwined in for what may be differences in activity, i.e., due to greater an expanding cycle of ischemia and inflammation.52 membrane fusion potential, via ICAM1, or to transcripts To provide a multifaceted therapy, we have transplanted that stabilized dormant ischemic cells in the IPC EXO group. adult renal tubular cells in experimental AKI and We hypothesize that abundant ribosomal transcripts in IPC CKD.16,17,19,20 The advantages of organ-specific cells for repair EXO and EXO might facilitate the transition from dormancy have been documented.53–55 These studies led to the hypothesis to restoration of function by reversing hypoxia-dependent in- that the relatively small number of donor cells infused17,20 acted hibition of protein synthesis.67 In summary, treatment with at a distance on endogenous cells. Accordingly, we described the EV from adult renal cells applied well after renal ischemia presence of growth factors and SAA1 protein in conditioned improved multiple parameters of structure, function, and medium of donor cells.20 This report is consistent with data of transcriptome profile. others using conditioned media.56–65 We further hypothesized that exosomes or EV released from transplanted cells, and in conditioned medium, serve as a therapeutic agent that improves CONCISE METHODS kidney function. Furthermore, we postulated that EV commu- nicate an “IPC state.” Unfortunately, translating the powerful Primary Renal Tubular Cells and EV effect of IPC to the clinical arena is not feasible, because most Primary renal tubular cells were obtained from male SD rats (Charles ischemia is unpredictable, and inducing ischemia in a vital organ River, Wilmington, MA). All experiments were conducted in confor- to prevent possible future injury is unethical.66 There are dozens mity with the “National Institutes of Health guide for the care and use of conflicting RIPC human trials, using diverse strategies (Clin- of laboratory animals.” The rats were euthanized by removing both icalTrials.gov). However, not knowing the mechanism of RIPC kidneys under general anesthesia. The kidney cortices were minced in hinders design and implementation.24 Here, we tested the hy- S1 medium (below), and digested with type 4 collagenase (Worthing- pothesis that EV transfer is one mechanism of RIPC. ton, Lakewood, NJ), 6 mg/dl, at 37°C in 38% O2 and 5% CO2 for 50 There are several points to be emphasized: Ischemic damage minutes. The renal tubules were then separated by percoll gradient,17 was marked by oxidative stress footprints as well as renal neu- cultured for 3 days, and the tubular cells, now in monolayers, were trophil infiltration and complement activation, likely in situ. transfected with pcDNA3.1-SAA1 (2 mg/ml) plasmid. pcDNA3.1- Within 6 days, extensive fibrosis and loss of peritubular capil- SAA1 was manufactured and sequenced in our laboratory, and its laries40 and glomeruli were seen. These postischemic changes construction was previously reported.16,17 This plasmid is used be- were largely abrogated by EXO treatment, and greater im- cause tubulogenic SAA1 can be used to track transfer to host kidney provements were seen with IPC EXO in virtually all parame- cells19 from donor EV (Figure 1). In addition, male determinant gene ters evaluated. Additionally, when the two groups, EXO and SRY was used to track donor male EV in recipient female kidneys. IPC EXO, are taken together (ten rats), the comparison Transfection was performed with lipofectamine 2000 (Life Technol- strongly indicates that exosomes are superior to either vehicle ogies, Grand Island, NY); efficiencies were .70%.19 The transfected (saline) or heat-inactivated exosomes postischemia. We ac- cells were cultured in S1 medium: Each 2 L contains F-12 HAM, 10.7 g; knowledge similarities in EXO and IPC EXO EV protection, DMEM, 8.32 g; L-glutamine, 0.29 g; HEPES, 4.78 g; sodium selenite, 1.7 suggesting that EXO EV were derived from cells that might mg; sodium pyruvate, 0.110 g; 3.2 ml phenol red and pH was adjusted to have had some degree of hypoxic stress in normal culture. 7.4 with sodium bicarbonate (Sigma). S1 medium was supplemented Multilayered ischemic damage was also manifested by sup- with hepatocyte growth factor, 200 ng/ml, and EGF, 400 ng/ml (R&D pressed antioxidant and HIF pathway proteins which were Systems, Minneapolis, MN). The medium also contained hydrocorti- improved with IPC EXO treatment. It is also noteworthy sone, 100 mg/ml; insulin, 35 mg/ml; transferrin, 32 mg/ml; sodium sel- that EV limited ischemia-induced activation of renal HSP27, enite, 42 ng/ml (Sigma, St. Louis, MO), with 10% FCS. The following suggesting that some stress responses were not required after day, renal tubular cells were divided into two groups: one group was EV treatment. It is remarkable that EV administration was placed in a hypoxic chamber (SCI-tive Dual; Baker Ruskinn, Sanford successful well after renal ischemia, indicating that EV infu- MA), 1% O2/5% CO2, and the other continued in 38% O2/5% CO2 for sions are potentially therapeutic. This latent period also sug- 4 more hours. The culture medium of both groups was then changed to gests that endogenous renal mechanisms allow survival for a oxygenated S1 medium with 10% FCS previously depleted of EV by time before treatment. 100,000 3 g centrifugation twice for 90 minutes. The cells were cultured

J Am Soc Nephrol 28: ccc–ccc, 2017 Exosomes Treat Renal Ischemia 7 BASIC RESEARCH www.jasn.org for 3 more days, and cell medium collected for EV isolation. The time counted in blinded PAS-stained sections. Renal sections were periods of hypoxia and exosome collection were chosen on the basis of immunostained for HNE to estimate tissue peroxidation. The pri- pilot studies in which survival of hypoxic cultured cells and exosome mary HNE rabbit antibody was applied at 1:300 dilution number were maximized. The medium was subjected to sequential (ABIN873270; antibodies-online.com). Horse anti-rabbit antibody centrifugation, 300 3 g for 10 minutes, recovering supernatant; 2000 was used to develop: ImmPRESS HRP Anti-Rabbit IgG (Peroxidase) 3 g for 10 minutes, recovering supernatant; 10,000 3 g for 30 minutes, Polymer Detection Kit (MP-7401; Vector). Additional kidney sec- recovering supernatant; and then 100,000 3 g for 70 minutes twice, tions made to visualize the microvasculature by immunofluores- recovering the pellet. The EV pellet was suspended in 500 ml of PBS, and cence: kidneys were immersion-fixed in 3.8% paraformaldehyde, filtered through a 1.0 mM filter and then a 0.2 mM filter. The EV protein followed by cryoprotection in 20% sucrose/PBS, and mounting in was measured with the Pierce Coomassie Plus (Bradford) Assay Kit optimal cutting temperature compound. Serial cross-sections (10 (23236; Thermo Fisher Scientific). The EV were characterized by elec- mm thick) were cut on a cryostat and mounted, and fixed to slides tron microscopy and by detecting EV CD63 on immunoblots (Figure in 100% methanol at 220°C for 10 minutes and PBS-washed twice, 1).30 Nanotrack analysis was performed using ZetaView (Particle Met- followed by treatment with 0.3% Triton X-100 for 5 minutes and then rix, Mebane, NC) according to the supplier’s instructions. A third group PBS-washed twice. Sections were incubated with Lycopersicon Escu- of EV consisted of IPC EV heated at 60°C for 45 minutes to inactivate lentum (Tomato) Lectin, conjugated with Alexa Fluor 594 (DL-1177; mRNA. Vector) at 1:300, overnight at 4°C. DAPI (Molecular Probes, Eugene, OR) was applied at 300 nM for 5 minutes at room temperature. Animal Protocols Images were taken through an inverted microscope (Zeiss Axio Im- At 8 weeks of age, female SD rats were anesthetized with inhaled ager.D2, Oberkochen, Germany). Positive fluorescence was quanti- isoflurane (0.5%–1%) and placed on a homeothermic table to fied from .10 sections per animal by Aperio’s Positive Pixel Count maintain a body core temperature at approximately 37°C. In three algorithm (Aperio Technologies, Inc., Vista, CA). Negative controls groups of rats (below) both renal pedicles were occluded for 50 min- for all targets were analyzed to ensure that the algorithm detected utes with microaneurysm clamps, as described.16–18 In a fourth rat minimal false-positives and autofluorescence. Microvascular density group, sham rats had an identical surgical procedure except that the was expressed as fractional positivity (positive pixels/total pixels on renal pedicles were not clamped. The rats received EV 24 and 48 each of 4003 fields per rat kidney). hours (equivalent quantities [100 mg] protein in 0.5 ml of sterile physiologic saline) after surgery via the tail vein. Blood samples Immuno (Western) Blotting were obtained daily from the saphenous or tail veins. Two doses Kidney cortices were homogenized in 25 mM Tris, pH 7.6, 150 mM were given to increase efficacy because we hypothesize that exosome NaCl, 1% deoxycholate, 1% P-40, 0.1% SDS, and 23 Halt Protease release from engrafted cell transplants occurred over time on the basis Inhibitor Cocktail (Thermo Scientific, Rockford, IL) and adjusted of prior studies.16–20 Euthanasia occurred 6 days after ischemia: after to a protein concentration of 2 mg/ml. The homogenates (20 mg) ensuring adequate anesthesia with isoflurane, organs were removed were fractionated by electrophoresis through 16.5% polyacrylamide and fixed in 3.8% paraformaldehyde. A fifth group was subjected to Tris-Tricine gels. After electrophoresis, proteins were transferred to a ischemia (as above) and received heat-inactivated EV at 24 and 48 nitrocellulose membrane. Blocking was carried out in 1% casein, 13 hours after surgery. PBS for 1 hour. Incubation with primary antibody diluted in 13 PBS was for 1 hour. The following rabbit primary antibodies were from Histology and Immunohistochemistry Abcam, Cambridge, MA: anti-SOD (1:1000, ab52950); anti-catalase Kidneys were fixed in 3.8% paraformaldehyde, paraffin embedded, (1:2000, ab16731); anti-erythropoietin (1:500, ab135390); anti- and 5-mM sections obtained for Masson’s trichrome to stain connec- GLUT1 (1:250, ab652); anti-HSP27 (1:1000, ab5579); anti-actin tive tissue, and periodic acid Schiff (PAS) to image cellular morphol- (1:1000, ab8227), and mouse anti-actin (1:50,000, Ab6276). Anti- ogy. The percentage of tubules in the outer medulla that showed VEGF (1:500, sc507) and anti-CD63 (1:250, sc15363) were from epithelial necrosis, luminal debris, and/or tubular dilation was esti- Santa Cruz Biotechnology, Santa Cruz, CA. Goat anti-ICAM1 mated on coded sections without the knowledge of the experimental (1:200, af583) was from R&D Systems. Mouse anti-PCNA (1:3000, group to which the animals belonged.68 Further quantification of p8825) was from Sigma. Secondary antibodies were from Li-COR congestion, intratubular obstruction (casts), inflammation, and tu- Biosciences, Lincoln, NE: IRDYE800CW donkey anti-rabbit (green, bule simplification was graded in cortex and medulla on the following 1:15,000, 926–32213), IRDYE 680RD donkey anti-rabbit (red, scale: 0 = none, 1+ = ,10%, 2+ = 10%–25%, 3+ = 26%–75%, 4+ = 1:20,000, 926–68073), IRDYE800CW donkey anti-mouse (green, .75%.68 Renal neutrophils were visualized with primary anti-rat 1:15,000, 926–32212), IRDYE 680RD donkey anti-mouse (red, neutrophil rabbit antibody (ABIN2586050; antibodies-online.com) 1:20,000, 926–68072), and IRDYE 680RD donkey anti-goat (red, and secondary antibody with HRP tag (SK-4805; Vector, Burlingame, 1:20,000, 926–68074). The membrane was then washed in 13 PBS CA). The neutrophils were counted in blinded renal sections. The and incubated with secondary antibodies diluted in 13 PBS for 1 areas of glomerular and peritubular fibrosis were quantified using hour. Secondary antibodies were IRDye 680RD goat anti-rabbit IgG Metamorph imaging image processing software (Sunnyvale, CA) (1:15,000) (Li-COR Biosciences) and IRDye 800 CW goat anti- and expressed as fractional areas per 2003 microscope field, covering mouse IgG (1:20,000) (Li-COR Biosciences). After washing in 13 all available surfaces in all coded kidneys. Atrophic tubules were PBS, the filter was scanned using an Odyssey Infrared Imaging System

8 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 www.jasn.org BASIC RESEARCH

(Li-COR Biosciences) for visualization of immunoreactive proteins. preparation and amplification using Ion OneTouch 2 (Life Technolo- All steps were carried out at room temperature. gies), followed by ISP loading onto a PI chip and sequencing on an Ion Proton semiconductor (Life Technologies). Each PI chip allows loading Electron Microscopy of about 140 million ISP templates, generating approximately 80–100 EV were fixed in 2% paraformaldehyde/2% glutaraldehyde/0.1 M million usable reads, up to 10–15 Gb. The reads were then mapped to the phosphate and then adsorbed to a 200–400 mesh carbon/formvar rat genome rn5 using TopHat2.70 Gene-based expression levels were coated grid and the negative stain (NANOVAN; Nanoprobes, Ya- quantified with NGSUtils.71 Read counts were normalized to the total phank, NY) added. The EV were visualized by electron micros- number of sequencing reads falling into annotated gene regions in each copy18 (Tecnai G2 12 Bio TWIN Microscope [FEI, Hillsboro, OR] sample, and further scaled on the basis of a trimmed mean of log- equipped with an AMT CCD camera [Advanced Microscopy Tech- transformed counts per million (CPM) to correct for the variability niques, Danvers, MA]). of RNA composition in each sample.72 The genes with less than one read per million mappable reads in no less than half of samples in at Renal and EV SAA1 and SYR mRNA least one condition were removed. The scaled CPM was used as gene- RT-PCR was also used to amplify SAA1 mRNA in recipient kidneys level quantification in each sample. The package “edgeR” was uti- and EV.RNA extracted from homogenized renal cortices in lysis buffer lized to identify the genes that were differentially expressed between fi was isolated with a puri cation kit as recommended by the vendor different biologic conditions.73 Benjamini and Hochberg’salgo- – (CAT # 12183 555; Invitrogen), and cleaned with an RNeasy Mini kit rithm was used to control the FDR. We determined FDR#0.05 as as recommended by the vendor (CAT # 74104; Qiagen, Valencia, CA). the statistical cutoff for differential expression among different m For RT-PCR, 2 g of total RNAwere used to synthesize cDNAwith an conditions. WEB-based GEne SeT AnaLysis Toolkit WEBGestalt fi Af nityScript QPCR cDNA Synthesis Kit as recommended by the (http://bioinfo.vanderbilt.edu/webgestalt/)74 was used to integrate vendor (CAT # 600559; Agilent Technologies, Santa Clara, CA). the differentially expressed transcripts to the centrally curated da- fi 19 The murine SAA1 mRNA was ampli ed as previously described tabase, KEGG (Kyoto Encyclopedia of Genes and Genomes), and to using the following primers: create functional categories with those differentially expressed transcripts. We performed several comparisons of renal transcripts: 9 9 F: 5 -GA GTC TGG GCT GAG AA-3 sham (control) and untreated ischemic; sham and normoxic EV (EXO); or sham and IPC EV. The IPC EV transcripts were separately R: 59-TG TCT GTT GGC TTC CTG GT-39 compared with the normoxic EV (EXO). The PCR products were separated in a 2.5% agarose gel. The specificSRYDNAwasamplified from extracted kidney and EV Statistical Analyses fi using the following primers19: Sample size of ve animals in each group gave 90% power to detect differences of 0.3 mg/dl in serum creatinine (with a set at 0.05).75 Data 6 F:59-AAGCGCCCCATGAATGC-39 are expressed as mean 1 SEM. ANOVA was used to determine if differences among mean values reached statistical significance. Paired R: 59-AGCCAACTTGCGCCTCTCT-39 t test (two tailed, two sample, unequal variance) was used for comparisons between groups (GraphPad Prism, LaJolla, CA). Tukey’s test The PCR products were separated in a 2.5% agarose gel. was used to correct for multiple comparisons. The null hypothesis was rejected at P,0.05. RNAseq and Differential Gene Expression Whole renal and EV transcriptomes by RNAseq analysis39,69 were obtained by the Center for Medical Genomics, Indiana University School of Medicine sequencing core facility. Kidney total RNA and ACKNOWLEDGMENTS EV RNAwere first evaluated for quantity and quality, using an Agilent Bioanalyzer. For kidney RNA quality, an RNA integrity number of We thank Caroline Miller of the Indiana University (IU) Electron seven or higher was required. The starting amount of RNA for library Microscopy Center and Jennifer Nelson and Dr. Sarah Tersey of the preparation was 500 ng for EV RNA and 1000 ng for kidney total Islet and Physiology Core of the Center for Diabetes and Metabolic RNA. PolyA mRNA capture of total RNA was performed using a Diseases at IU for nanotrack analysis. Dynabeads mRNA DIRECT Micro Kit (61021; Ambion) as the first This work was supported by a National Institutes of Health grant step. cDNA library preparation followed, inclusive of enzymatic R01 DK 082739, a research grant from Dialysis Clinics Inc. Paul fragmentation, hybridization and ligation of adaptors, reverse tran- Teschan Research and Development Fund, and a biomedical research scription, size selection, and amplification with barcode primers, fol- grant from the IU School of Medicine to K.J.K. and Veterans Affairs lowing the Ion Total RNA-Seq Kit v2 User Guide, Pub. No. 4476286 Merit Review funds to J.H.D. Rev. E (Life Technologies). Each resulting barcoded library was quantified and its quality accessed by an Agilent Bioanalyzer, and multiple libraries pooled in equal molarity. Eight microliters of 100 pM pooled DISCLOSURES libraries were then applied to the Ion Sphere Particles (ISP) template None.

J Am Soc Nephrol 28: ccc–ccc, 2017 Exosomes Treat Renal Ischemia 9 BASIC RESEARCH www.jasn.org

REFERENCES 20. Kelly KJ, Zhang J, Wang M, Zhang S, Dominguez JH: Intravenous renal cell transplantation for rats with acute and chronic renal failure. Am J Physiol Renal Physiol 303: F357–F365, 2012 1. Bonventre JV, Yang L: Cellular pathophysiology of ischemic acute 21. Bonventre JV: Kidney ischemic preconditioning. Curr Opin Nephrol kidney injury. JClinInvest121: 4210–4221, 2011 Hypertens 11: 43–48, 2002 2. Heyman SN, Khamaisi M, Rosen S, Rosenberger C: Renal parenchymal 22. Islam CF, Mathie RT, Dinneen MD, Kiely EA, Peters AM, Grace PA: Is- hypoxia, hypoxia response and the progression of chronic kidney dis- chaemia-reperfusion injury in the rat kidney: The effect of preconditioning. ease. Am J Nephrol 28: 998–1006, 2008 Br J Urol 79: 842–847, 1997 3. Brivet FG, Kleinknecht DJ, Loirat P, Landais PJ; French Study Group on 23. Jonker SJ, Menting TP, Warlé MC, Ritskes-Hoitinga M, Wever KE: Acute Renal Failure: Acute renal failure in intensive care units–causes, Preclinical evidence for the efficacy of ischemic postconditioning outcome, and prognostic factors of hospital mortality; a prospective, against renal ischemia-reperfusion injury, a systematic review and multicenter study. Crit Care Med 24: 192–198, 1996 meta-analysis. PLoS One 11: e0150863, 2016 4. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW: Acute 24. Liu Z, Gong R: Remote ischemic preconditioning for kidney protection: kidney injury, mortality, length of stay, and costs in hospitalized pa- GSK3b-centric insights into the mechanism of action. AmJKidneyDis tients. J Am Soc Nephrol 16: 3365–3370, 2005 66: 846–856, 2015 5. Coca SG, Yusuf B, Shlipak MG, Garg AX, Parikh CR: Long-term risk of 25. Wever KE, Menting TP, Rovers M, van der Vliet JA, Rongen GA, mortality and other adverse outcomes after acute kidney injury: A sys- Masereeuw R, Ritskes-Hoitinga M, Hooijmans CR, Warlé M: Ischemic tematic review and meta-analysis. AmJKidneyDis53: 961–973, 2009 preconditioning in the animal kidney, a systematic review and meta- 6. Horkan CM, Purtle SW, Mendu ML, Moromizato T, Gibbons FK, analysis. PLoS One 7: e32296, 2012 Christopher KB: The association of acute kidney injury in the critically ill 26. Kierulf-Lassen C, Nieuwenhuijs-Moeke GJ, Krogstrup NV, Oltean M, and postdischarge outcomes: A cohort study*. Crit Care Med 43: 354– Jespersen B, Dor FJ: Molecular mechanisms of renal ischemic condi- 364, 2015 tioning strategies. Eur Surg Res 55: 151–183, 2015 7. Hoste EA, Bagshaw SM, Bellomo R, Cely CM, Colman R, Cruz DN, 27. Dickson EW, Lorbar M, Porcaro WA, Fenton RA, Reinhardt CP, Edipidis K, Forni LG, Gomersall CD, Govil D, Honoré PM, Joannes- Gysembergh A, Przyklenk K: Rabbit heart can be “preconditioned” via Boyau O, Joannidis M, Korhonen AM, Lavrentieva A, Mehta RL, transfer of coronary effluent. Am J Physiol 277: H2451–H2457, 1999 Palevsky P, Roessler E, Ronco C, Uchino S, Vazquez JA, Vidal Andrade 28. Dickson EW, Reinhardt CP, Renzi FP, Becker RC, Porcaro WA, Heard E, Webb S, Kellum JA: Epidemiology of acute kidney injury in critically SO: Ischemic preconditioning may be transferable via whole blood ill patients: The multinational AKI-EPI study. Intensive Care Med 41: transfusion: Preliminary evidence. J Thromb Thrombolysis 8: 123–129, 1411–1423, 2015 1999 8. Lassnigg A, Schmidlin D, Mouhieddine M, Bachmann LM, Druml W, 29. Burger D, Viñas JL, Akbari S, Dehak H, Knoll W, Gutsol A, Carter A, Bauer P, Hiesmayr M: Minimal changes of serum creatinine predict Touyz RM, Allan DS, Burns KD: Human endothelial colony-forming cells prognosis in patients after cardiothoracic surgery: A prospective cohort protect against acute kidney injury: Role of exosomes. Am J Pathol 185: study. JAmSocNephrol15: 1597–1605, 2004 2309–2323, 2015 9. Liaño F, Junco E, Pascual J, Madero R, Verde E: The Madrid Acute Renal 30. Kourembanas S: Exosomes: Vehicles of intercellular signaling, Failure Study Group: The spectrum of acute renal failure in the intensive biomarkers, and vectors of cell therapy. Annu Rev Physiol 77: 13– care unit compared with that seen in other settings. Kidney Int Suppl 66: 27, 2015 – S16 S24, 1998 31. Ratajczak J, Miekus K, Kucia M, Zhang J, Reca R, Dvorak P, Ratajczak 10. Siew ED, Davenport A: The growth of acute kidney injury: A rising tide MZ: Embryonic stem cell-derived microvesicles reprogram hemato- – or just closer attention to detail? Kidney Int 87: 46 61, 2015 poietic progenitors: Evidence for horizontal transfer of mRNA and 11. Faubel S, Chawla LS, Chertow GM, Goldstein SL, Jaber BL, Liu KD: Acute protein delivery. Leukemia 20: 847–856, 2006 Kidney Injury Advisory Group of the American Society of Nephrology: 32. Tak T, Tesselaar K, Pillay J, Borghans JA, Koenderman L: What’syour – Ongoing clinical trials in AKI. Clin J Am Soc Nephrol 7: 861 873, 2012 age again? Determination of human neutrophil half-lives revisited. J 12. Agarwal A, Dong Z, Harris R, Murray P, Parikh SM, Rosner MH, Kellum Leukoc Biol 94: 595–601, 2013 JA, Ronco C; Acute Dialysis Quality Initiative XIII Working Group: 33. Lee HM, Choi EJ, Kim JH, Kim TD, Kim YK, Kang C, Gho YS: A mem- Cellular and molecular mechanisms of AKI. JAmSocNephrol27: branous form of ICAM-1 on exosomes efficiently blocks leukocyte ad- – 1288 1299, 2016 hesion to activated endothelial cells. Biochem Biophys Res Commun – 13. Eltzschig HK, Eckle T: Ischemia and reperfusion from mechanism to 397: 251–256, 2010 – translation. Nat Med 17: 1391 1401, 2011 34. Landis SC, Amara SG, Asadullah K, Austin CP, Blumenstein R, 14. Fine LG, Orphanides C, Norman JT: Progressive renal disease: The Bradley EW, Crystal RG, Darnell RB, Ferrante RJ, Fillit H, Finkelstein – chronic hypoxia hypothesis. Kidney Int Suppl 65: S74 S78, 1998 R, Fisher M, Gendelman HE, Golub RM, Goudreau JL, Gross RA, 15. Tanaka S, Tanaka T, Nangaku M: Hypoxia as a key player in the AKI-to- Gubitz AK, Hesterlee SE, Howells DW, Huguenard J, Kelner K, CKD transition. Am J Physiol Renal Physiol 307: F1187– F1195, 2014 Koroshetz W, Krainc D, Lazic SE, Levine MS, Macleod MR, McCall 16. Kelly KJ, Kluve-Beckerman B, Dominguez JH: Acute-phase response JM,MoxleyRT3rd,NarasimhanK,NobleLJ,PerrinS,PorterJD, protein serum amyloid A stimulates renal tubule formation: Studies in Steward O, Unger E, Utz U, Silberberg SD: A call for transparent vitro and in vivo. Am J Physiol Renal Physiol 296: F1355–F1363, 2009 reporting to optimize the predictive value of preclinical research. 17. Kelly KJ, Kluve-Beckerman B, Zhang J, Dominguez JH: Intravenous cell Nature 490: 187–191, 2012 therapy for acute renal failure with serum amyloid A protein-reprogrammed 35. Witzgall R, Brown D, Schwarz C, Bonventre JV: Localization of pro- cells. Am J Physiol Renal Physiol 299: F453–F464, 2010 liferating cell nuclear antigen, vimentin, c-Fos, and clusterin in the 18. Kelly KJ, Zhang J, Han L, Kamocka M, Miller C, Gattone VH 2nd, postischemic kidney. Evidence for a heterogenous genetic response Dominguez JH: Improved structure and function in autosomal re- among nephron segments, and a large pool of mitotically active and cessive polycystic rat kidneys with renal tubular cell therapy. PLoS One dedifferentiated cells. J Clin Invest 93: 2175–2188, 1994 10: e0131677, 2015 36. Leung KC, Tonelli M, James MT: Chronic kidney disease following 19. Kelly KJ, Zhang J, Han L, Wang M, Zhang S, Dominguez JH: Intravenous acute kidney injury-risk and outcomes. Nat Rev Nephrol 9: 77–85, 2013 renal cell transplantation with SAA1-positive cells prevents the progres- 37. Gao G, Wang W, Tadagavadi RK, Briley NE, Love MI, Miller BA, Reeves sion of chronic renal failure in rats with ischemic-diabetic nephropathy. Am WB: TRPM2 mediates ischemic kidney injury and oxidant stress through J Physiol Renal Physiol 305: F1804–F1812, 2013 RAC1. J Clin Invest 124: 4989–5001, 2014

10 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 www.jasn.org BASIC RESEARCH

38. Kelly KJ, Burford JL, Dominguez JH: Postischemic inflammatory syn- 56. Cheng K, Rai P, Plagov A, Lan X, Kumar D, Salhan D, Rehman S, drome: A critical mechanism of progression in diabetic nephropathy. Malhotra A, Bhargava K, Palestro CJ, Gupta S, Singhal PC: Trans- Am J Physiol Renal Physiol 297: F923–F931, 2009 plantation of bone marrow-derived MSCs improves cisplatinum- 39. Kelly KJ, Liu Y, Zhang J, Dominguez JH: Renal C3 complement com- induced renal injury through paracrine mechanisms. Exp Mol Pathol 94: ponent: Feed forward to diabetic kidney disease. Am J Nephrol 41: 48– 466–473, 2013 56, 2015 57. Du T, Ju G, Wu S, Cheng Z, Cheng J, Zou X, Zhang G, Miao S, Liu G, Zhu 40. Basile DP, Friedrich JL, Spahic J, Knipe N, Mang H, Leonard EC, Y: Microvesicles derived from human Wharton’s jelly mesenchymal Changizi-Ashtiyani S, Bacallao RL, Molitoris BA, Sutton TA: Impaired stem cells promote human renal cancer cell growth and aggressiveness endothelial proliferation and mesenchymal transition contribute to through induction of hepatocyte growth factor. PLoS One 9: e96836, vascular rarefaction following acute kidney injury. Am J Physiol Renal 2014 Physiol 300: F721–F733, 2011 58. Li J, Ariunbold U, Suhaimi N, Sunn N, Guo J, McMahon JA, McMahon 41. Miranda KC, Bond DT, Levin JZ, Adiconis X, Sivachenko A, Russ C, AP, Little M: Collecting duct-derived cells display mesenchymal stem Brown D, Nusbaum C, Russo LM: Massively parallel sequencing of cell properties and retain selective in vitro and in vivo epithelial ca- human urinary exosome/microvesicle RNA reveals a predominance of pacity. J Am Soc Nephrol 26: 81–94, 2015 non-coding RNA. PLoS One 9: e96094, 2014 59. Machiguchi T, Nakamura T: Cellular interactions via conditioned media 42.ShimadaS,HirabayashiM,IshigeK,KosugeY,KiharaT,ItoY:Ac- induce in vivo nephron generation from tubular epithelial cells or tivation of dopamine D4 receptors is protective against hypoxia/ mesenchymal stem cells. Biochem Biophys Res Commun 435: 327– reoxygenation-induced cell death in HT22 cells. J Pharmacol Sci 333, 2013 114: 217–224, 2010 60. Tögel F, Zhang P, Hu Z, Westenfelder C: VEGF is a mediator of the 43. Zhang W, Liu J, Hu X, Li P, Leak RK, Gao Y, Chen J: n-3 polyunsaturated renoprotective effects of multipotent marrow stromal cells in acute fatty acids reduce neonatal hypoxic/ischemic brain injury by promoting kidney injury. JCellMolMed13[8B]: 2109–2114, 2009 phosphatidylserine formation and Akt signaling. Stroke 46: 2943–2950, 61. van Koppen A, Joles JA, van Balkom BW, Lim SK, de Kleijn D, Giles RH, 2015 Verhaar MC: Human embryonic mesenchymal stem cell-derived con- 44. Vaggi F, Disanza A, Milanesi F, Di Fiore PP, Menna E, Matteoli M, Gov ditioned medium rescues kidney function in rats with established NS, Scita G, Ciliberto A: The Eps8/IRSp53/VASP network differentially chronic kidney disease. PLoS One 7: e38746, 2012 controls actin capping and bundling in filopodia formation. PLOS 62. Wang DH, Li FR, Zhang Y, Wang YQ, Yuan FH: Conditioned medium Comput Biol 7: e1002088, 2011 from renal tubular epithelial cells stimulated by hypoxia influences rat 45. Yuryev A, Patturajan M, Litingtung Y, Joshi RV, Gentile C, Gebara M, bone marrow-derived endothelial progenitor cells. Ren Fail 32: 863– Corden JL: The C-terminal domain of the largest subunit of RNA po- 870, 2010 lymerase II interacts with a novel set of serine/arginine-rich proteins. 63. Yuen DA, Connelly KA, Advani A, Liao C, Kuliszewski MA, Trogadis Proc Natl Acad Sci U S A 93: 6975–6980, 1996 J, Thai K, Advani SL, Zhang Y, Kelly DJ, Leong-Poi H, Keating A, 46. Schults MA, Sanen K, Godschalk RW, Theys J, van Schooten FJ, Chiu Marsden PA, Stewart DJ, Gilbert RE: Culture-modified bone mar- RK: Hypoxia diminishes the detoxification of the environmental muta- row cells attenuate cardiac and renal injury in a chronic kidney gen benzo[a]pyrene. Mutagenesis 29: 481–487, 2014 disease rat model via a novel antifibrotic mechanism. PLoS One 5: 47. Recino A, Sherwood V, Flaxman A, Cooper WN, Latif F, Ward A, e9543, 2010 Chalmers AD: Human RASSF7 regulates the microtubule cytoskel- 64. Zarjou A, Kim J, Traylor AM, Sanders PW, Balla J, Agarwal A, Curtis LM: eton and is required for spindle formation, Aurora B activation and Paracrine effects of mesenchymal stem cells in cisplatin-induced renal chromosomal congression during mitosis. Biochem J 430: 207–213, injury require heme oxygenase-1. Am J Physiol Renal Physiol 300: 2010 F254–F262, 2011 48. U.S. Renal Data System: USRDS 2015 Annual Data Report: Atlas of 65. Zhou Y, Xiong M, Fang L, Jiang L, Wen P, Dai C, Zhang CY, Yang J: miR- End-Stage Renal Disease in the United States, Bethesda, Md, National 21-containing microvesicles from injured tubular epithelial cells pro- Institutes of Health, National Institute of Diabetes and Digestive and mote tubular phenotype transition by targeting PTEN protein. Am J Kidney Diseases, 2015 Pathol 183: 1183–1196, 2013 49. Ojo AO, Wolfe RA, Held PJ, Port FK, Schmouder RL: Delayed graft 66. Tapuria N, Kumar Y, Habib MM, Abu Amara M, Seifalian AM, Davidson function: Risk factors and implications for renal allograft survival. BR: Remote ischemic preconditioning: A novel protective method Transplantation 63: 968–974, 1997 from ischemia reperfusion injury–areview.JSurgRes150: 304–330, 50. Weber M, Dindo D, Demartines N, Ambühl PM, Clavien PA: Kidney 2008 transplantation from donors without a heartbeat. N Engl J Med 347: 67. Staudacher JJ, Naarmann-de Vries IS, Ujvari SJ, Klinger B, Kasim M, 248–255, 2002 Benko E, Ostareck-Lederer A, Ostareck DH, Bondke Persson A, 51. Brown JR, Rezaee ME, Marshall EJ, Matheny ME: Hospital mortality in Lorenzen S, Meier JC, Blüthgen N, Persson PB, Henrion-Caude A, the United States following acute kidney injury. BioMed Res Int 2016: Mrowka R, Fähling M: Hypoxia-induced gene expression results from 4278579, 2016 selective mRNA partitioning to the endoplasmic reticulum. Nucleic 52. Schnaper HW, Hubchak SC, Runyan CE, Browne JA, Finer G, Liu X, Acids Res 43: 3219–3236, 2015 Hayashida T: A conceptual framework for the molecular pathogenesis 68. Kelly KJ, Meehan SM, Colvin RB, Williams WW, Bonventre JV: Pro- of progressive kidney disease. Pediatr Nephrol 25: 2223–2230, 2010 tection from toxicant-mediated renal injury in the rat with anti-CD54 53. Huang GT, Gronthos S, Shi S: Mesenchymal stem cells derived from antibody. Kidney Int 56: 922–931, 1999 dental tissues vs. those from other sources: Their biology and role in 69. Kelly KJ, Liu Y, Zhang J, Goswami C, Lin H, Dominguez JH: Com- regenerative medicine. JDentRes88: 792–806, 2009 prehensive genomic profiling in diabetic nephropathy reveals the 54. Li TS, Cheng K, Malliaras K, Smith RR, Zhang Y, Sun B, Matsushita N, predominance of proinflammatory pathways. Physiol Genomics 45: Blusztajn A, Terrovitis J, Kusuoka H, Marbán L, Marbán E: Direct com- 710–719, 2013 parison of different stem cell types and subpopulations reveals superior 70.KimD,PerteaG,TrapnellC,PimentelH,KelleyR,SalzbergSL: paracrine potency and myocardial repair efficacy with cardiosphere- TopHat2: Accurate alignment of transcriptomes in the presence of derived cells. JAmCollCardiol59: 942–953, 2012 insertions, deletions and gene fusions. Genome Biol 14: R36, 2013 55. Tedesco FS, Dellavalle A, Diaz-Manera J, Messina G, Cossu G: Re- 71. Breese MR, Liu Y: NGSUtils: A software suite for analyzing and ma- pairing skeletal muscle: Regenerative potential of skeletal muscle stem nipulating next-generation sequencing datasets. Bioinformatics 29: cells. JClinInvest120: 11–19, 2010 494–496, 2013

J Am Soc Nephrol 28: ccc–ccc, 2017 Exosomes Treat Renal Ischemia 11 BASIC RESEARCH www.jasn.org

72. Robinson MD, Oshlack A: A scaling normalization method for differ- 75. Kadam P, Bhalerao S: Sample size calculation. Int J Ayurveda Res 1: 55– ential expression analysis of RNA-seq data. Genome Biol 11: R25, 2010 57, 2010 73. Robinson MD, McCarthy DJ, Smyth GK: edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26: 139–140, 2010 74. Wang J, Duncan D, Shi Z, Zhang B: WEB-based GEne SeT AnaLysis Toolkit This article contains supplemental material online at http://jasn.asnjournals. (WebGestalt): Update 2013. Nucleic Acids Res 41: W77–W83, 2013 org/lookup/suppl/doi:10.1681/ASN.2016121278/-/DCSupplemental.

12 Journal of the American Society of Nephrology J Am Soc Nephrol 28: ccc–ccc,2017 Figure 1S. Representative Nanotrack Plots and image of EV showing size distribution of EV. Figure 2S. Representative images of EV uptake by cultured renal tubular cells. EV, labeled with green exoglow, were rapidly taken up by cultured tubular cells exposed to Hoescht to label nuclei blue. Uptake of IPC EXO was greater than than of EXO (p<0.05). Insets show areas at arrows. Figure 3S. Quantification of Histological Injury. Tubular congestion, intratubular obstruction (casts), inflammation and tubule simplification in blinded sections of cortex and medulla were graded on the follow scale: 0=none; 1+ = <10%, 2+ = 10-25%, 3+ = 26-75%, 4+ = >75%. Figure 4S. Representative western blot of proliferative cell nuclear antigen (PCNA) protein levels in the four groups of rats (green), n = 2 blots: sham control, untreated ischemic (saline), ischemic treated with normoxic exosomes (exo) and ischemic treated with IPC exosomes (IPC exo). The corresponding actin levels are also included (red). Figure 5S. Left, representative western blot of proteins suppressed by ischemia and partly supported by treatment with normoxic exosomes (exo) and IPC exosomes (IPC exo), n = 5: superoxide dismutase 1 (SOD, sham vs saline and IPC exo vs saline, p < 0.05); catalase (CAT, sham vs saline and IPC exo vs saline, p < 0.05); vascular endothelial growth factor (VEGF, IPC exo vs saline, p < 0.01); erythropoietin (EPO, sham vs saline and IPC exo vs saline, p < 0.05); glucose transporter 1 (GLUT1 or Slc2a1, sham vs saline and IPC exo vs saline, p < 0.05). In ischemia, heat shock protein 27 (hsp27) was increased (P<0.05), IPC exo limited its activation (p<0.05). Protein levels were all normalized for actin levels (red). Figure 6S. Representative western blots: A, transforming growth factor (TGF); B, pro-apoptotic fas; and C, anti-apoptotic birc3 protein levels in four groups of rat kidneys (green): sham control, untreated postischemic (saline), postischemic treated with normoxic EV (EXO) and postischemic treated with IPC EV (IPC EV). The graphs shows the fold change in mean TGF/actin (n=3), fas/actin (n=2) and birc3/actin (n=2) band density as compared to sham. Kegg pathway Number of transcripts Metabolicp=2.79e15 Anti-apoptotic 39 GO analysis Anti-inflammatory 29 Transporteractivity Pro-vasculature 8 p=1.02e10 Anti-fibrosis 12

Figure 7S. Many transcripts were altered in untreated renal ischemia: either suppressed (left) or activated (right). IPC EXO treatment significantly restored 1526 suppressed genes, and significantly restored 1593 activated genes. Whereas EXO restored 440 suppressed genes and restored 209 activated genes (p<0.05 for all treatment groups with respect to untreated ischemia). Most genes restored by EXO overlapped the genes restored by IPC. The most significantly altered Kegg pathway was metabolic. The table categorizes 89 restored transcripts in both IPC EXO and EXO in potentially beneficial systems. RNASeq in exosomes IPC EXO

Kegg pathway Ribosome p=0.0003 GO analysis RNA processing p=0.002

Figure 8S. RNA Seq in EV showed 24 differentially expressed transcripts between IPC and neutral EXO with the most altered pathway being the ribosome. heat inactivated P value v P value v EXO group saline EXO 24 h creatinine (mg/dl) 2.2±0.1 NS NS ccs HH 48 h creatinine (mg/dl) 1.62±0.3 NS 0.07 Weight (mg/300g body wt) 2.6±0.2 NS <0.05 fraction damaged tubules 0.72±0.04 <0.05 <0.05 PMN/hpf C, control RNA 5.1±0.9 <0.05 <0.05 S, size standard C3 pixel density 65±2.4 NS <0.05 Fibrosis (% area) 0.19±0.001 NS <0.05 H, heated IPC EXO RNA Microvascular density 0.86±0.05 NS NS

Table 1S. Heat-inactivated EV are not protective. The table shows multiple parameters in postischemic rats that had been treated with heated EV at 24 and 48 hours postischemia. Although protein levels were not significantly different between control and heated EV, RNA was not found in EV after heat inactivation (right). NS v EXO v IPC V IPC v NS SHAM SHAM SHAM Pathway p value p value p value p value Metabolic Pathways 9.47 e-92 2.75 e-35 --- 1.39 e-41 Proliferation* 1.16 e -29 4.66 e-18 --- 3.59 e-16 Focal adhesion 8.97 e-25 1.31 e-16 0.0029 3.51 e-09 Filtrate absorption# 5.83 e-21 8.32 e-15 0.0002 7.61 e-12 Cell cycle 3.88 e-20 1.20 e-16 --- 8.18 e-17 ECM-receptor 1.14 e-17 3.55 e-12 5.08 e-06 1.34 e-07 interaction Oxidative 1.08 e-16 ------0.0155 phosphorylation Glutathione 1.74 e-14 ------2.28 e-08 metabolism Phagosome 1.99 e-14 1.76 e-10 --- 2.42 e-11 Apoptosis 2.27 e-10 1.47 e-05 --- 2.42 e-06 Cytokine-cytokine R 5.74 e-14 8.15 e-11 --- 1.33 e - interaction 09 DNA replication 2.67 e-07 -- --- 1.48 e-08

In Kegg classification, pathway listed as cancer (*) protein digestion and absorption (#)

Table 2S. Differentially expressed Kegg pathways in untreated and EXO and EXO IPC treated renal ischemia. Data include the entire set of transcriptomes in the four groups of rats fold gene change P value Lrrc56 16.19684 3.26E-07 Drd4 78.49808 6.06E-06 Ptdss2 8.283269 2.55E-05 Phrf1 3.93936 0.0004027 Rn28s 0.270428 0.0005771 Rn45s 0.296246 0.0009219 Eps8l2 4.501065 0.0014974 Tmem80 10.53305 0.0022572 Ugt2b1 3.238392 0.002616 Ugt2b10 3.238402 0.002628 Mir3574 20.63152 0.0027636 Ddx39b 0.39594 0.0043904 Fau 0.689794 0.0055902 Rps6 2.338801 0.0061934 Rps13 1.380476 0.0085318 LOC102549726 0.307007 0.0151298 Rassf7 2.915573 0.0208718 Minos1 0.507992 0.0217366 Anxa1 1.300553 0.0255786 LOC100911483 0.411889 0.027246 Pfn1 0.68795 0.0275072 Mtap 0.233285 0.0393567 Clic1 0.629333 0.04683 Rnh1 0.512802 0.0486232 Tubb5 0.682995 0.0508805 Cox7a2l 1.670835 0.0540679 Eif3m 1.372874 0.0651085 Fam129b 0.774576 0.0735416 Ltb 0.29788 0.0866235 Atp6v1g2 0.29779 0.0877447 RT1-CE10 0.24278 0.0929345 Rps8 1.254032 0.0933281 Rps15a 0.754515 0.0955748 Eif3k 1.920594 0.1102808 Rpl17 1.272594 0.112006 Cd63 0.648305 0.112629 RT1-CE2 0.331922 0.1185701 Dynll1 0.650672 0.1206252 Ddt 0.503037 0.1212101 Cfdp1 1.415932 0.1224761 Atp5g3 0.693078 0.1231189 Rpl27 1.237502 0.1236026 Taldo1 1.604253 0.1238283 Rps3a 1.237234 0.1290179 Nrep 0.47773 0.1314657 Rbm3 1.341765 0.1332759 RT1-CE12 0.37312 0.1364004 Ube2s 0.610861 0.1428263 Pabpc1 1.257365 0.1488621 Rps24 0.821989 0.1493474 Cd24 1.250384 0.1529263 Psmb1 1.461866 0.153649 RT1-CE1 0.390259 0.1550898 Rpl30 1.303759 0.1646442 Ppp1ca 1.424261 0.1678137 Hnrnpa2b1 0.748745 0.175859 Hnrnpa1 1.245239 0.1778314 Skp1 1.304902 0.1832875 Pla2g16 1.719821 0.184658 Atp5b 0.8284 0.1946005 Eef2 0.822913 0.2015693 Rpsa 1.197804 0.2037134 Txn1 0.686598 0.2058584 Hsp90aa1 1.212961 0.2105127 Atp5g2 0.739137 0.2166733 Fkbp1a 0.806462 0.2166977 Atp6v1g1 0.743283 0.2295635 Fmr1 1.744195 0.2407201 Uqcr10 1.955778 0.242598 Akr1a1 0.846383 0.2449587 Cox7a2 0.756662 0.2480775 Rpl37 0.784878 0.2525435 Slc25a5 1.376195 0.2621689 S100a10 1.148694 0.2640821 Rps26 1.284187 0.2652901 Rps21 1.209257 0.2665313 Rpl24 1.153315 0.2680652 Eif3h 0.802052 0.2687002 Calm1 1.190372 0.2707449 Rps11 1.167072 0.2709647 Npm1 1.203589 0.2722887 Eif5a 0.856515 0.2767545 Eefsec 2.007712 0.2782998 Ccl2 0.703501 0.2837439 Rps20 0.834438 0.2862785 Gstp1 1.18054 0.295941 S100a11 0.856361 0.2967193 Anp32a 1.169359 0.2984425 Ncl 1.179824 0.3067983 Eif1 0.844571 0.3098891 Gabarap 1.229229 0.3104298 RT1-CE5 0.643873 0.3191237 Psma7 1.311407 0.3191301 Cox7c 1.314293 0.3209647 Tpm4 0.864358 0.3211267 Psmb4 0.838391 0.3238268 Hspa1b 0.764356 0.3257658 Gpx4 0.76898 0.3302108 Rps27a 1.16923 0.332307 Rplp1 1.157418 0.3365376 Oaz1 1.141339 0.3366418 Rpl31 1.16545 0.3380286 Sumo2 1.299053 0.3381638 Naca 0.885164 0.3398111 Irf7 0.413371 0.342284 Nap1l1 1.146845 0.3453079 Taf15 1.241188 0.3522772 Ptma 1.135475 0.3567916 Bex1 0.773795 0.3580473 Ap2m1 0.812035 0.3606806 Pgam1 0.818519 0.361135 Atp6v0c 0.769529 0.3666519 Tagln2 1.162377 0.3720678 Phf5a 0.684477 0.3747457 Btf3 1.144621 0.3779764 Rps23 1.117397 0.3789969 Cdc42se1 0.806276 0.3790978 Tceb2 0.781878 0.3807514 Rpl35a 0.825949 0.3985928 RT1-CE7 0.594312 0.4030576 Clta 0.845074 0.4047845 RT1-CE11 0.597632 0.4049716 LOC688869 0.787394 0.4211151 Rpl10a 0.893571 0.4272844 Ran 1.12744 0.427463 Cox5a 1.206173 0.4297347 Rpl29 1.12809 0.4352454 Myl12b 0.876199 0.4396517 Eif3g 0.825507 0.4412582 Ubxn1 1.143078 0.4422194 Pkm 0.837445 0.4492772 Rps25 1.131239 0.4495111 Ftl1 1.125947 0.4502224 Arpc3 1.231225 0.453335 Krt8 0.876578 0.4565726 Rpl5 1.124749 0.4582947 Cox4i1 0.84107 0.4584987 Eif2s1 0.817872 0.4587609 Rps18 1.102073 0.4589225 Tuba1c 0.885939 0.4617327 Mir1843b 0.513414 0.4633045 S100a6 1.239133 0.4651814 Tagln 0.845279 0.4759817 Mllt11 0.570966 0.480076 Gnb2l1 1.10397 0.4843559 Rps15 1.126447 0.4862987 Hint1 1.245795 0.4926176 Anxa5 0.872895 0.4939825 Slc25a4 1.164717 0.5028693 Rpl14 1.09488 0.5078502 RGD1560212 0.752566 0.5094949 Ywhae 0.844116 0.5124516 Hsp90ab1 1.09692 0.5128616 Rpl39 1.131763 0.5147275 Bc1 1.225677 0.5155934 Arpc2 0.881537 0.5170883 Nedd8 0.837991 0.5177774 Ost4 1.207691 0.5218714 Rps10 0.906872 0.5263327 Vamp8 0.668615 0.5275603 Sema6c 0.879467 0.5326785 Atp5h 0.800661 0.5533324 Net1 0.918908 0.556796 Setdb1 0.626909 0.5582528 Prr13 0.821144 0.5591506 Rpl22 1.130581 0.5647353 Ppia 1.087658 0.5656269 Actb 0.921363 0.5685723 Atp5g1 0.774983 0.5693493 Aldoa 0.926856 0.5742425 Gpx1 1.159759 0.5765184 Atpif1 1.197276 0.577025 Rpl26 1.079297 0.5790565 Dynlrb1 0.856332 0.5803291 Ubb 1.080049 0.5831251 Atp5i 0.823757 0.5832494 Rps7 1.074743 0.5854094 Rpl18a 0.918343 0.5869659 Rpl34 0.922835 0.587038 Mrpl17 1.19489 0.5889211 Hras 1.215426 0.5905863 Anxa2 0.927109 0.5943318 Gas5 0.761059 0.596497 Rpl19 0.921853 0.5974182 Ybx1 1.071817 0.598626 Ybx1-ps3 1.071805 0.5988658 Rpl36 0.918713 0.6059161 Mir186 1.490943 0.6061228 Rpl22l1 0.887405 0.607431 Gapdh 1.07394 0.6119164 Rpl18 0.936493 0.6121435 Mir7a-1 1.527482 0.6184054 Mgp 1.370084 0.6196138 Arpc1b 0.925357 0.6203871 Tpi1 1.094767 0.6272966 Fth1 1.087925 0.631166 Kif1c 0.937281 0.6403018 Rpl23 1.068558 0.6450173 Cdc37 0.907195 0.6451968 Cers2 0.903562 0.6490774 Prdx6 0.867922 0.6551477 Pebp1 0.903556 0.6553389 Atp5e 1.128412 0.6586591 Nsa2 1.124802 0.6593335 Deaf1 1.51687 0.6597719 Rps4x 0.936393 0.6661336 LOC689574 1.209821 0.6700299 Snrpd2 0.872997 0.6705857 Rpl10 0.942869 0.6768858 Atp5l 1.120157 0.6807349 Tuba4a 1.147543 0.6847105 Eef1a1 1.060202 0.6908095 Rpl7 1.06568 0.7003146 Apol9a 1.305528 0.7003316 Tnf 0.793562 0.7084428 Glrx3 1.079026 0.7093485 Rplp0 0.948217 0.7100793 Rpl35 1.08981 0.7117531 Spp1 1.087624 0.7177463 Rpl13 1.050367 0.7199709 Kctd10 1.049838 0.7221687 RT1-CE6 0.739749 0.7294701 Nme2 0.940854 0.7322649 Anp32b 1.06156 0.7323684 H2afz 0.922357 0.7332644 Atp5j 0.881965 0.7339682 Prdx2 1.052526 0.7448864 Ndufv3 1.168331 0.7453687 Actg1 0.952824 0.7459469 Rpl36al 0.912745 0.7519278 Pgk1 0.934738 0.7566777 Rpl37a-ps1 0.912298 0.7594105 Rpl21 0.950298 0.7597947 Rpl38 0.919155 0.7605254 Cox6c 0.921217 0.7606448 Lgals3 0.931137 0.7637967 Tpt1 1.046282 0.7694822 Csnk2b 1.089492 0.7708616 Calr 0.9617 0.7708902 Mir207 0.775739 0.7719874 Rps14 0.962285 0.7727725 Lta 0.835344 0.7737796 Rpl4 1.042309 0.7792766 Mir3064 1.240944 0.7817236 Rps12 1.035005 0.782746 Rpl32 1.036145 0.7861538 Mif 0.929525 0.7877495 Arpp19 0.849313 0.7889673 Cox7b 1.069265 0.8013971 Uba52 0.96798 0.8103173 Dstn 1.034023 0.8111642 Rpl41 0.964556 0.8118491 Gpx2 0.938774 0.8130104 Sat1 1.04026 0.8149929 Cfl1 0.974649 0.8201299 Myeov2 0.902782 0.8250237 Rps27 1.04381 0.8271308 Cdc42 1.037834 0.8358183 Tp53inp2 1.026886 0.8376532 Fxyd2 1.084099 0.8381056 Rpl6 0.973352 0.8382874 Coa3 1.07556 0.8384823 Chmp3 0.953033 0.8406023 Ccng1 0.968435 0.8428834 Atp5d 1.061233 0.8514214 Myl12a 1.023074 0.8532691 Atp5o 0.954268 0.8703125 Ndufa1 1.108824 0.8724803 Eno1 0.978673 0.8755616 Pmepa1 1.047201 0.876194 Tuba1a 0.974824 0.877689 Rps3 1.022198 0.8797707 RT1-CE3 1.122018 0.8829357 Rpl28 0.983611 0.8904055 Ndufb4 0.928273 0.8907963 Tomm20 1.036334 0.8942282 Tmsb4x 0.97827 0.8987523 Rpl3 1.01825 0.8990121 Ivns1abp 0.961778 0.9011038 Psmb7 1.029249 0.9038833 Tuba1b 0.98305 0.9061238 Mt2A 0.958631 0.9063594 Mir3583 1.096327 0.9178901 Rpl13a 0.984344 0.9241297 Snrpg 1.026437 0.9304076 Rpl9 0.98469 0.9334294 Prdx1 0.987126 0.9345581 Tmsb10 1.012851 0.9426112 Rpl23a 0.990667 0.9453443 Hspa8 0.991364 0.9453964 LOC252890 0.965323 0.9485038 Rps5 1.007585 0.9507977 Rps29 1.009315 0.9566838 Rps2 0.991934 0.9594823 Eef1g 0.993331 0.960309 Eif4a1 1.008239 0.9609216 Nfkbil1 0.974439 0.962564 Psma2 1.015447 0.9669022 Rps17 0.995146 0.9777611 Fam63a 1.005073 0.9827277 Tomm7 1.008497 0.983385 Prelid1 1.004061 0.9879881 Tpm1 0.997898 0.9890512 Tbca 0.99672 0.9916308 Cox17 1.004673 0.9931595 Rab13 1.001251 0.9948289 Ldha 0.999074 0.9953579 Rpl12 1.000357 0.997858 Uqcr11 0.99946 0.998225 Wbp5 0.999652 0.9988384 Psma6 0.999706 0.9988979 Rps28 0.999894 0.9995591

Table 3S. RNASeq in IPC and normoxic exosomes: fold change of IPC exosomes/ normoxic exosomes and p-values are presented.