Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/yexcr

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ] 1

0014-4827/$ - see frohttp://dx.doi.org/10.1

nCorresponding autE-mail address: z

Please cite this artthe acute kidney

Research Article

Effect of erythropoietin on the migration of bonemarrow-derived mesenchymal stem cells to the acutekidney injury microenvironment

Nanmei Liua,b, Jun Tianb, Jin Chengb, Jinyuan Zhangb,n

aDepartment of Nephrology, Changzheng Hospital affiliated to the Second Military Medical University, Shanghai 200003 ChinabDepartment of Nephrology, the 455th hospital of PLA, Shanghai 200052 China

a r t i c l e i n f o r m a t i o n

Article Chronology:

Received 4 February 2013Received in revised form10 April 2013Accepted 11 April 2013

Keywords:

ErythropoietinBone marrow-derived mesenchymalstem cellsMigrationAcute kidney injuryHypoxia/re-oxygenation

nt matter & 2013 Elsevier016/j.yexcr.2013.04.008

hor. Fax: +86 21 32260890hangjinyuan455@gmail.co

icle as: N. Liu, et al., Effeinjury microenvironment

a b s t r a c t

Bone marrow-derived mesenchymal stem cells (BMSCs) preferentially migrate to the injured

tissue but with limited efficiency. Here we investigated the effect of erythropoietin (EPO)treatment on the BMSC migration to the acute kidney injury (AKI) microenvironment. Thepossible mechanisms were also discussed. A hypoxia/re-oxygenation (HR) model of renal tubularepithelial cells (RTECs) was established to generate AKI in vitro, and a chemotaxis experimentwas conducted using the transwell chamber. EPO treatment enhanced the BMSC migration to theHR-RTEC culturing chamber in a SDF-1 level-dependent manner, which was fully inhibited by thetreatment of anti-SDF-1 antibody. The BMSC migration could also be partly blocked by LY294002(phosphoinositide 3-kinase (PI3K) inhibitor) and PD98059 (MAPK inhibitor). Western blotanalysis showed that phosphorylated Akt and phosphorylated MAPK in BMSCs were enhancedby EPO treatment. In the in vivo experiment, BMSCs were transplanted into the AKI mice and EPOwas subcutaneously injected. The results showed that EPO injection increased the SDF-1 protein

expression and BMSC accumulation in the renal tissue, which was consistent with a decentimprovement of renal function. In addition, the BMSC accumulation in the renal tissue wasblocked by anti-SDF-1 antibody, LY294002 or PD98059. Our data suggest that AKI microenviron-ment had a directional chemotactic effect on BMSCs, which could be further enhanced by the EPOtreatment. The increased SDF-1 level in the AKI microenvironment and the activations of PI3K/AKT and MAPK in BMSCs were the possible mechanisms for the effect of EPO. Therefore, BMSCtransplantation combined with EPO injection can be a novel and effective approach for AKI repair.

& 2013 Elsevier Inc. All rights reserved.

Introduction

Acute kidney injury (AKI) is a common clinical disease and theprognosis and mortality are still significant issues. Previousstudies [1–6] have suggested that transplantation of bone

Inc. All rights reserved.

.m (J. Zhang).

ct of erythropoietin on th, Exp Cell Res (2013), http

marrow-derived mesenchymal stem cells (BMSCs) is helpful forthe repair of AKI because of their preferential migration to theinjured kidney and paracrine/autocrine mechanisms or differen-tiation potential. However, many transplanted BMSCs can still stayin the blood-rich organs. Therefore, to maintain the repair effect

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ]2

when a limited number of BMSCs is transplanted, it is vital toincrease the efficiency of the BMSC migration to the injuredkidney. Erythropoietin (EPO) is a glycoprotein that has beensuggested to be effective in treating AKI [7–9]. Although it hasbeen suggested that EPO has positive effects on the BMSCmigration [10,11], its effect on the kidney-directional migrationof BMSCs remains unclear.In this study, a hypoxia/re-oxygenation (HR) model of renal

tubular epithelial cells (RTECs) was established to generate AKIin vitro, and the effect of EPO treatment on the BMSC migration tothe AKI microenvironment was investigated. Furthermore, AKImouse models were constructed and the effect of EPO was furtherinvestigated through the in vivo experiment. The possiblemechanisms were also explored.

Materials and methods

Construction of the HR model of RTECs to generate AKIin vitro

Mouse RTECs, which were purchased from American TissueCulture Collection (ATCC) (Manassas, VA, USA), were divided into5 groups: the Control Group, the H6/R12 Group, the H12/R12 Group,the H16/R12 Group and the H24/R12 Group. For the Control Group,4�105/well RTECs were cultured in the 6-well plates under thenormoxia condition (95% air and 5% CO2) with low-glucoseDMEM (Invitrogen, San Diego, CA, USA). For the other 4 groups,4�105/well RTECs were cultured in the 6-well plates first underthe hypoxia condition (3% O2, 5% CO2 and 92% N2) using theAnaeroPack™ System (Mitsubishi Gas Chemical Co. Inc., Tokyo,Japan) with glucose-free DMEM (Invitrogen) for 6 h, 12 h, 16 hand 24 h, respectively. Then, they were all cultured under there-oxygenation condition (95% air and 5% CO2) with low-glucoseDMEM for 12 h.The general changes of cell morphology in the 5 groups were

monitored. An Annexin V-FITC/PI Apoptosis Detection Kit (Beyo-time Biotechnology Co., Ltd., Nanjing, China) was used to evaluatethe cell apoptosis. Briefly, after being rinsed with ice-cold PBS, thecells were resuspended in 200 μl binding buffer. After 5 μlAnnexin V-FITC was added, the cells were incubated at roomtemperature for 10 min. The cells were then further incubatedwith 10 μl propidium iodide, followed by an instant analysis usingFACScan.According to the changes of cell morphology and the apoptosis

rate of RTECs, the group that can maximally generate AKI in vitrowas determined, based on which HR-RTECs were prepared forfurther use. RTECs (4�105/well) cultured totally under thenormoxia condition, whose culturing time was consistent withthe hypoxia/re-oxygenation time used for HR-RTECs, wereselected as the control.

Chemotaxis assay

The chemotaxis experiment was conducted using the transwellchamber with an 8-μm pore size polycarbonate filter (CoringIncorporated, NY, USA). Upon completion of the preparations ofHR-RTECs and RTECs, the 6-well plates were removed from theincubator and the transwell chambers were inserted into theplates. Mouse BMSCs (ATCC) resuspended in the medium were

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

plated in the upper chambers and the plating density was1.8�105/chamber. The chemotaxis assay was used for the follow-ing groups: the Control group (BMSCs/RTECs), the HR group(BMSCs/HR-RTECs) and the EPO groups (different concentrations(1 IU/ml, 5 IU/ml, 10 IU/ml, 50 IU/ml) of recombinant human EPO(rhEPO, Roche, China) were added into the HR-RTEC culturemedium, then, the transwell chambers were inserted into theplates and BMSCs were added into the upper chambers).

Stromal cell-derived factor 1 (SDF-1) inhibition experiment wasalso performed. 10 IU/ml rhEPO and 5 μg/ml anti-mouse SDF-1monoclonal antibody (R&D systems, Wiesbaden, Germany) wereboth added into the HR-RTEC culture medium. For the other twoinhibition experiments, the BMSCs in the EPO (10 IU/ml) groupwere first preincubated with 20 μM LY294002 (phosphoinositide3-kinase (PI3K) inhibitor) or 5 μM PD98059 (MAPK inhibitor)(both were purchased from Sigma-Aldrich, St Louis, MO, USA) for30 min and then added into the upper chambers. All the groupswere then incubated at 37 1C for 6 h in a humidified atmospherewith 5% CO2. After completion of the co-culturing, the nonmi-grating BMSCs from the upper surface of the polycarbonatemembrane were wiped out with a cotton bud. The BMSCs underthe membrane were stabilized with 4% paraformaldehyde for20 min and stained with 1% crystal violet for 30 min. In the finalstep, PBS cleaning was carried out till the violet color became dim.Photographs were taken for each membrane under the micro-scope, and the numbers of cells in 5 non-overlapped visual fieldswere counted and the average number of migrating cells wascalculated.

SDF-1 levels in RTECs or HR-RTECs

Protein expression: After 6 h co-culturing, the transwell cham-bers were removed. Cell lysate was added into the 6-well platesfor the full lysis of RTECs (or HR-RTECs). The supernatant wastaken after centrifugation and the protein concentration wasmeasured. After sodium dodecyl sulfate polyacrylamide gel elec-trophoresis (SDS-PAGE), the protein was transferred to the PVDFmembrane. After being sealed, the PVDF membrane was incu-bated with rabbit anti-mouse SDF-1 monoclonal antibody at 4 1Covernight. HRP labeled goat anti-rabbit IgG (Santa Cruz Biotech-nology Inc., CA, USA) was added and the membrane was incu-bated for 1 h at room temperature. The mixture of the reacted ECLwas added to the PVDF membrane for 1–2 min, and then thePVDF membrane was placed into the Fluorchem HD2 gel imageanalysis system for analysis. The intensity of the band wasmeasured. Tubulin was used as the internal reference. The proteinexpression was the ratio of the two band gray values.

Supernatant concentration: We removed the transwell cham-bers, and the culture supernatants of RTECs and HR-RTECs in the6-well plates were collected to measure the SDF-1 concentrationusing the SDF-1 Enzyme-linked Immunosorbent Assay (ELISA)Detection Kit (Round Record Biotech Co., Ltd, Shanghai, China)according to the manufacturer's instructions.

Phosphorylations of AKT and MAPK in BMSCs

After completion of the co-culturing, the transwell chamberswere transferred to the new blank 6-well plates, and cell lysatewas added for the full lysis of BMSCs. Western blot was employedas described above with rabbit anti-mouse phospho-Akt (pAKT)

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

Fig. 1 – h-RTEC construction. Cell morphology was observed under the inverted microscope (A) and cell apoptosis was evaluated byflow cytometry analysis (B, C). (B) LL: viable cells (Annexin V-FITC-/PI−); LR: early apoptotic cells (Annexin V-FITC +/PI−); UR: lateapoptotic or necrotic cells (Annexin V-FITC +/PI+). (C) apoptotic proportion of each group (n¼6/group). The results are presentedas means7standard deviation. nPo0.05, compared to the Control Group.

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ] 3

and phospho-MAPK (pMAPK) monoclonal antibodies (both werepurchased from CST Inc., MA, USA) as the primary antibodies.GAPDH was used as the internal reference and the proteinexpression was the ratio of the two band gray values.

Preparation of the AKI mouse models and experimentalsettlement

Mouse BMSCs were labeled with bromodeoxyuridine (BrdU, Sigma-Aldrich) by culturing themwith low-glucose DMEM containing 10 μMBrdU for 72 h, and the labeling efficiency was determined by theproportion of BrdU-labeled BMSCs (BrdU-BMSCs) using the immu-nohistochemistry (IHC) method. C57BL/6 mice (Experimental AnimalCenter of the Second Military Medical University, 2072 g) were usedto develop the AKI models by clamping the bilateral renal pedicles for30min and reopening for 30min. The AKI mice were divided into 4groups: the AKI model group (n¼5), the BMSCs group (n¼6, 0.2 mlPBS containing 2�106 BrdU-BMSCs was injected into the tail vein),the EPO group (n¼6, 3000 IU/kg rhEPO was subcutaneously injectedfor 3 consecutive days) and the EPO–BMSCs group (n¼6, in additionto the injection of 2�106 BrdU-BMSCs, 3000 IU/kg rhEPO was alsosubcutaneously injected for 3 consecutive days). For the inhibitionexperiments, 3 other groups were included: the EPO–BMSCs+SDF-1antibody group (n¼6, 1 mg/kg anti-mouse SDF-1 monoclonal

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

antibody was intraperitoneally injected into the AKI mice for 3consecutive days and the other protocols were identical to those inthe EPO–BMSCs group), the EPO–BMSCs+LY294002 group and theEPO–BMSCs+PD98059 group (n¼5 in both groups, BrdU-BMSCswere first suspended in the medium containing 20 μM LY294002 or5 μM PD98059 at 37 1C for 30min and then injected into the tail vein,and 3000 IU/kg rhEPO was subcutaneously injected for 3 consecutivedays). All mice were housed at favorable temperatures and humiditywith an unlimited supply of water and food, and sacrificed at 7 d afterthe injection of BrdU-BMSCs. Blood samples were collected and theserumwas isolated and stored at −20 1C. After perfusion with PBS viathe heart, the bilateral kidneys were separated and further perfusedwith PBS. The corticomedullary junction of the kidneys was cut andfixed with 10% formalin for pathological and IHC analysis. Allprocedures were performed in accordance with the principles ofthe Guidelines for Animal Experimentation of The Second MilitaryMedical University (Shanghai, China).

Blood biochemical indicators

Blood urea nitrogen (BUN) and serum creatinine (Scr) concentra-tions were measured using the Beckman Automatic BiochemistryAnalyzer (Beckman Coulter Inc.).

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

Fig. 2 – BMSC migration in vitro. HR: the HR group. BMSCs and HR-RTECs were co-cultured together with the transwell chamber,and BMSCs were plated in the upper chamber with HR-RTECs being in the lower chamber. The EPO groups: differentconcentrations of rhEPO (1 IU/ml, 5 IU/ml, 10 IU/ml, 50 IU/ml) were added into the HR-RTECs culture medium and the otherprotocols were identical to the HR group. (A) Images of the migrating BMSCs under the microscope (�200). (B) HR-RTECs hadchemotaxis on BMSCs and EPO treatment further increased the BMSC migration in a concentration-dependent manner (up to10 IU/ml) (n¼6/group). (C) BMSC migration in the EPO (10 IU/ml) group was totally blocked by the anti-SDF-1 antibody (n¼6/group). (D) LY294002, a PI3K inhibitor, partly blocked the BMSC migration (n¼6/group). (E) PD98059, an MAPK inhibitor, partlyblocked the BMSC migration (n¼6/group). The results are presented as means7standard deviation. aPo0.05, compared to theControl group; bPo0.05, compared to the HR group; cPo0.05, compared to the EPO (1 IU/ml) group; dPo0.05, compared to the EPO(10 IU/ml) group.

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ]4

Fig. 3 – SDF-1 levels in RTECs or HR-RTECs of each group. (A)

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ] 5

Pathological changes

The fixed samples were used to make paraffin sections. After HEstaining, the degree of tubular necrosis was scored using theblinding method. Briefly, for each section, 15 non-overlapped viewfields were chosen under the magnification of 200� for scoring:0¼normal, 1¼minor injury (injured tubular o5%), 2¼mild injury(injured tubular 5–25%), 3¼medium injury (injured tubular 25–75%), 4¼severe injury (injured tubular 475%). Semi-quantitativeanalysis was performed and the mean score value was used toindicate the degree of acute tubular necrosis (ATN scoring).

SDF-1 level in the renal tissue

The samples were treated with 1 ml lysis buffer and centrifugedat 20,000 rpm/min (4 1C) for 10 min. The supernatant was cen-trifuged for another 10 min at 12,000 rpm/min (4 1C). The proteinconcentration was measured. Western blot was employed asdescribed above using the anti-mouse SDF-1 monoclonal anti-body as the primary antibody. β-actin was used as the internalreference and the protein expression was the ratio of the twoband gray values.

Assessment of BMSC migration in vivo

The IHC method was employed. The paraffin sections were de-waxed with dimethylbenzene and dehydrated with graded etha-nol. After the endogenous peroxidase was blocked with 0.3%H2O2, the sections were placed in a container filled with citratebuffer. After adding the antigen restored liquid, the container washeated to 98 1C for 3 min to expose the antigen. The sections wereincubated with BrdU monoclonal antibody (Sigma-Aldrich) at 4 1Cfor 16 h. After washed with PBS, the sections were incubated withHRP labeled anti-mouse lgG (Santa Cruz Biotechnology Inc.) at37 1C for 1 h. Then the sections were incubated in DAB chromo-genic substrate liquid for 15 min, and dyed with haematoxylin.After sealing with resin, the sections were stored at roomtemperature for use. 15 non-overlapped view fields (�400) werechosen for each section to determine the proportion of BrdU-BMSCs in the corticomedullary junction of the kidneys. The meanvalue of the proportion of BrdU-BMSCs was used for statisticalanalysis.

Statistical analysis

The results are expressed as means7standard deviation. Student'st-test was performed to analyze the differences between the twogroups. Multiple-group comparison was performed using one-way analysis of variance (ANOVA) followed by a Student–New-man–Keuls test. SPSS17.0 statistical software was used for theanalysis. Values of Po0.05 were considered statisticallysignificant.

The SDF-1 protein expression in the cells. (B) SDF-1concentration in the culture supernatant. EPO treatmentsignificantly increased the SDF-1 levels inside the cells and inthe culture supernatants compared with the HR group (n¼6/group). The results are presented as means7standarddeviation. aPo0.05, compared to the Control group; bPo0.05,compared to the HR group; cPo0.05, compared to the EPO(1 IU/ml) group.

Results

HR-RTEC construction

The morphology of RTECs in the H6/R12 Group and the H12/R12

Group were ideal under the inverted microscope. However, the

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

RTECs in the H16/R12 Group showed obvious shrinkage, andnecrotic cells appeared and floated. In the H24/R12 Group, theRTECs completely floated (Fig. 1A).Analysis of Annexin V-FITC/PI staining revealed that

3.2571.01% of the RTECs appeared apoptotic in the H6/R12 Group,which showed no difference from the Control Group. The propor-tion of apoptotic cells significantly increased to 32.8271.16% inthe H12/R12 Group, which was consistent with the apoptoticproportion of renal tubules in the AKI mice [12]. Longer hypoxiaculturing in the H16/R12 Group and the H24/R12 Group resulted ingreater proportions of apoptotic cells, especially late apoptotic ornecrotic cells (Fig. 1B and C).Thus, it can be seen that the RTECs in the H12/R12 Group are

optimal to generate AKI in vitro, and subsequently the HR-RTECswere prepared based on the H12/R12 Group for further use.The RTECs cultured under the normoxia condition for 24 h wereused as the normal control.

Effect of EPO on the BMSC migration in vitro

A transwell chamber-based migration assay was established toquantitatively evaluate the BMSC migration in vitro. The lower

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ]6

chamber was loaded with RTECs (cultured under the normoxiacondition for 24 h) or HR-RTECs (cultured under the hypoxiacondition and the re-oxygenation condition, respectively, for12 h), and different concentrations of rhEPO were added intothe HR-RTEC culture medium. The upper chamber was loadedwith BMSCs. The HR model of RTECs showed an obvious chemo-tactic effect on BMSCs, as indicated by the increased number ofmigrating cells (the HR group vs. the Control group: 19.0072.37vs. 10.8372.42) (Fig. 2B). The EPO treatment further increased thenumber of migrating BMSCs in a concentration-dependent man-ner up to 10 IU/ml (46.6777.37) (Fig. 2B). Further increase of theEPO concentration (e.g., 50 IU/ml) did not show a greater effect onthe BMSC migration (Fig. 2B). However, the BMSC migration wasblocked by the anti-SDF-1 monoclonal antibody and the migrat-ing number (9.5072.43) was even close to that of the Controlgroup (10.8372.42) (Fig. 2C).

SDF-1 levels

Analysis of Western blot and ELISA detection showed that the HRmodel significantly enhanced the SDF-1 protein levels inside theRTECs (ratio of gray values between the HR group and the Controlgroup: 0.3870.01 vs. 0.2570.01) and in the culture supernatant(61.6474.88 pg/ml vs. 35.2678.78 pg/ml). EPO treatment furtherincreased the SDF-1 protein levels both in the cells and in theculture supernatant, and the changes showed a concentration-dependent manner, with the highest SDF-1 levels appearing inthe EPO (10 IU/ml) group (ratio of gray values inside the cells:0.7170.02; the concentration in the culture superna-tant:173.53714.66 pg/ml, respectively). The EPO (50 IU/ml) groupand EPO (10 IU/ml) group showed no significant differences in theSDF-1 levels (Fig. 3A and B).

PI3K/Akt and MAPK are involved in the EPO-induced BMSCmigration in vitro

To explore whether EPO can enhance the BMSC migrationthrough the PI3K/Akt-dependent or MAPK-dependent pathways

Fig. 4 – Phosphorylations of AKT and MAPK in BMSCs. (A) Phosphincreased the protein expressions of phospho-Akt and phospho-MAresults are presented as means7standard deviation. aPo0.05, compcPo0.05, compared to the EPO (1 IU/ml) group.

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

in BMSCs, we tested the effects of LY294002 (PI3K/Akt inhibitor)and PD98059 (MAPK inhibitor) on the EPO-induced BMSC migra-tion. As shown in Fig. 2D and E, application of the inhibitorssignificantly attenuated the effect of EPO on the BMSC migration.However, the numbers of the migrating cells in the presences ofthe two inhibitors (EPO (10 IU/ml)+LY294002 group: 31.5574.81,EPO (10 IU/ml)+PD98059 group: 27.4273.57, respectively) werestill significantly higher than that of the Control group(10.8372.42), indicating that each of the two signal pathwaysonly played a partial role in the EPO-induced BMSC migration.Western blot analysis (Fig. 4A and B) showed that the levels ofphospho-Akt and phospho-MAPK were increased in the BMSCsco-cultured with HR-RTECs (the HR group vs. the Control group:ratio of gray values in pAkt: 0.3870.02 vs. 0.2970.01, ratio of grayvalues in pMAPK: 0.2970.02 vs. 0.1970.02, respectively). Theapplication of EPO at low concentration (e.g., 1 IU/ml) furtherenhanced the phosphorylation of MAPK (ratio of gray values:0.4870.02) but not Akt. As the EPO concentration increased (e.g.,5 IU/ml or 10 IU/ml), the expression levels of phospho-Akt andphospho-MAPK in the BMSCs both increased in a concentration-dependent manner until the EPO reached the concentration of50 IU/ml.

SDF-1 protein expression in the renal tissue

Quantitative analysis of SDF-1 protein expression relative to β-actinin the renal tissue is shown in Fig. 5A. No difference was foundbetween the BMSCs group and the AKI model group. However,exogenous EPO significantly enhanced the protein expression ofSDF-1. The EPO group and the EPO–BMSCs group both showedsimilar increases (ratio of gray values: 0.7470.01 in the EPO group,0.7970.05 in the EPO–BMSCs group, respectively) [13].

BrdU labeling of BMSCs

BMSCs were labeled with BrdU. This method is not only simpleand convenient, but also has a high labeling efficiency. The result

orylation of Akt. (B) Phosphorylation of MAPK. EPO treatmentPK in the BMSCs co-cultured with HR-RTECs (n¼6/group). Theared to the Control group; bPo0.05, compared to the HR group;

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ] 7

of IHC indicated that BrdU was brown and surrounding thenucleus of BMSCs (Fig. 5B). The proportion of BrdU-BMSCs was98.7170.32%.

BMSC migration to the renal tissue after AKI

Some cells in the renal tissue showed brown nuclei in the BMSCsgroup, indicating that BrdU-BMSCs migrated to the renal tissue.EPO injection increased the proportion of BrdU-BMSCs in therenal tissue (14.2871.63% vs. 8.5971.12%). Administrations ofanti-SDF-1 antibody, LY294002 and PD98059 all resulted insignificant reductions of the BrdU-BMSC accumulation, whichwere even lower than that of the BMSCs group (Fig. 5C and D).

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

Blood biochemical indicators and ATN scoring

BMSC transplantation was helpful for the AKI repair with asignificant decrease in the BUN level, while there were nosignificant changes in the Scr level and the ATN scoring. EPOinjection alone showed no significant effect on these indicators. IfBMSC transplantation was combined with EPO injection, the BUNand Scr levels and the ATN scoring all showed sharp decreases(BUN: 7.7371.30 mmol/l, Scr: 28.4372.78 μmol/l; ATN scoring:1.6770.24) (Fig. 5E) [13].

Discussion

As mentioned before, BMSC transplantation is helpful for AKItreatment, but the progress is slow due to the low repairefficiency. In our animal experiment, also, the Scr level and theATN scoring showed no significant differences between theBMSCs group and the AKI model group. The limited kidney-directional migration of the transplanted BMSCs should be themain reason.When injury occurs, exogenously transplanted BMSCs will

preferentially migrate to the injured tissue, especially to theischemia tissue [14,15]. The kidney is an ischemia-sensitive organand AKI will occur after ischemia. Apoptosis of RTECs is the mainmanifestation of AKI, and hypoxia/re-oxygenation (HR) is themain reason resulting in the cell apoptosis. In this experiment, weestablished a HR model of RTECs to generate AKI in vitro. Achemotaxis assay was performed using the transwell chamberand the migration of BMSCs to the HR-RTEC culturing chamberwas enhanced, which also confirmed that BMSCs preferentiallymigrate to the injured tissue. However, a large amount of thetransplanted BMSCs still reside in the blood-rich organs, e.g., lung,liver and spleen [16], and thus the AKI repair efficiency of BMSCtransplantation will obviously be mitigated. Therefore, to improvethe AKI repair efficiency, it is critical to increase the kidney-directional migration of BMSCs.BMSC migration may involve various chemokines, cyokines, and

integrins. Among the chemokines, SDF-1 and its cellular receptor,

Fig. 5 – Results of the in vivo experiment. (A) The SDF-1 proteinexpression in the renal tissue was determined by Western blot. (B)BrdU-labeled BMSCs (BrdU-BMSCs) (�200). BrdU showed brownaround the nucleus of positive cells. (C) Accumulation of BrdU-BMSCs in the corticomedullary junction of the kidney (�400).The brown nucleus in the cells indicated BrdU-BMSCs (arrowsindicated). (D) The image analysis showed that the proportion ofBrdU-BMSCs in the EPO–BMSCs group was higher and decreasedsignificantly after the administration of anti-SDF-1 antibody,LY294002 (PI3K inhibitor) and PD98059 (MAPK inhibitor). (E) Theblood biochemical indicators and the ATN scoring of each group.The lower BUN and Scr levels and the lower ATN scoring wereshown in the EPO–BMSCs group. N¼5 in the AKI model group,n¼6 in the BMSCs group, n¼6 in the EPO group, n¼6 in the EPO–BMSCs group, n¼6 in the EPO–BMSCs+SDF-1 antibody group,n¼5 in the EPO–BMSCs+LY294002 group and n¼5 in the EPO–BMSCs+PD98059 group. The results are presented asmeans7standard deviation. aPo0.05, compared to the AKI modelgroup; bPo0.05, compared to the BMSCs group; cPo0.05,compared to the EPO group.

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ]8

CXCR4, are extensively studied and have been demonstrated todirect the migration of stem cells associated with injury repair inmany species and tissue types [17,18]. SDF-1 is a CXC sub-familymember that is expressed in many organs [19], and the ischemiamicroenvironment in the injured tissue including AKI will up-regulate its expression [20–22]. The increased SDF-1 levels werealso observed in our in vitro and in vivo experiments. CXCR4, thereceptor of SDF-1, is localized both on the surface of BMSCs andinside the cells [23]. However, it usually exists inside the cells [24]and is translocated to the cell surface by stimulation of specificcytokines. SDF-1 can act as this mobilization cytokine [25] andthen bind with the surface CXCR4 to direct the BMSC migration.Previous studies have shown that this directional migrationis related with the SDF-1 level in the injured tissue [24,26].Therefore, increasing the SDF-1 level in the injured kidney willhave a promising effect on the kidney-directional migrationof BMSCs.Previous studies on the acute myocardial infarction rat model

[27–29] have confirmed that EPO can enhance the SDF-1 expres-sion in the ischemia cardiomyocytes via the EPO receptor (EPOR),which is also expressed on RTECs [30]. In our experiment, theSDF-1 levels in the HR-RTECs and in the culture supernatant wereboth increased by EPO treatment in a concentration-dependentmanner (not greater than 10 IU/ml). A similar result was alsoobserved in our animal experiment. We consider that EPO tookthis effect via the EPOR of RTECs. We note that EPO at theconcentration of 50 IU/ml had no greater effect on the SDF-1 level,which might be due to the saturation of the connection of EPOand its receptor. Based on these results, EPO might be effective ininducing the BMSC migration to the AKI microenvironment,which is well consistent with our experimental results. Forinstance, the numbers of migrating BMSCs in our in vitro andin vivo experiments were both significantly increased after EPOtreatment. Additionally, this was matched with a greaterimprovement in renal function, including lower levels of BUN,Scr and lower ATN scoring. Comparisons among the AKI modelgroup, the EPO group and the EPO–BMSCs group showed that thisgreater improvement of renal function in the EPO–BMSCs groupwas not caused by the direct effect of EPO on treating AKI, and itcould be due to the increased kidney-migration of BMSCs. We alsoshowed that the EPO-induced BMSC migration was SDF-1 level-dependent, which was confirmed by the correlated changingpatterns of the migrating number and the SDF-1 level in vitro,as well as the full inhibition of the migration by anti-SDF-1antibody.Both the PI3K/Akt and the MAPK signal pathways have been

shown to mediate the cell migration induced by chemokines orcytokines in different cell types [31–33]. Previous research on theBMSC migration also confirmed that activations of AKT and MAPKcan increase the secretion of actin and myosin involved in theBMSC movement. They can also contribute to the secretion ofmatrix metalloproteinase (MMP), which can degrade the extra-cellular matrix and basilar membrane, and thus is beneficial forthe BMSC migration [34]. In addition, the phosphorylations ofAKT and MAPK can stimulate the secretion of fibrinolytic enzymeto degrade clots, which is beneficial for BMSCs to enter theinjured tissue [35]. In our experiment, the Western blot analysisshowed that the levels of phospho-Akt and phospho-MAPK wereenhanced with varying degrees after EPO treatment. The migra-tion assays with PI3K-specific inhibitors (LY294002) and MAPK

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

inhibitor (PD98059) also showed that applications of theseinhibitors significantly attenuated the EPO-induced BMSC migra-tion both in vitro and in vivo. However, the numbers of themigrating cells in vitro were still higher than that of the Controlgroup. Therefore, it seems that the PI3K/Akt and MAPK pathwaysare both likely involved in the enhanced BMSC migration inducedby EPO treatment.

Conclusions

In conclusion, our experiment suggested that EPO treatmentenhanced the BMSC migration to the AKI microenvironment witha greater improvement in renal function. The enhanced SDF-1level in the AKI microenvironment, as well as the activations ofPI3K/Akt and MAPK pathways in the BMSCs, are the possiblemechanisms. In addition, the effect and security of EPO in treatingAKI have been widely accepted. As a result, a preliminaryconclusion can be reached that the combined therapy of BMSCtransplantation and EPO injection can be a novel and effectivetherapy for AKI repair.

Acknowledgments

We gratefully acknowledge the financial supports from ShanghaiRising-star Program (12QA1405000), Shanghai Municipal NaturalScience Foundation (11ZR1449600) and Shanghai Key Projects ofBasic Research (12DJ1400203).

r e f e r e n c e s

[1] B.A. Yeagy, S. Cherqui, et al., Kidney repair and stem cells: acomplex and controversial process, Pediatr. Nephrol. 26 (2011)1427–1434.

[2] P. Chhabra, K.L. Brayman, et al., The use of stem cells in kidneydisease, Curr. Opinion Organ Transplant. 14 (2009) 72–78.

[3] A.L. Ponte, E. Marais, N. Gallay, A. Langonné, B. Delorme, O.Hérault, P. Charbord, J. Domenech, et al., The in vitro migrationcapacity of human bone marrow mesenchymal stem cells:comparison of chemokine and growth factor chemotactic activ-ities, Stem Cell 25 (2007) 1737–1745.

[4] F. Tögel, J. Isaac, Z. Hu, K. Weiss, C. Westenfelder, et al., RenalSDF-1 signals mobilization and homing of CXCR4-positive cells tothe kidney after ischemic injury, Kidney Int. 67 (2005) 1772–1784.

[5] M.B. Herrera, B. Bussolati, S. Bruno, V. Fonsato, G.M. Romanazzi,G. Camussi, et al., Mesenchymal stem cells contribute to the renalrepair of acute tubular epithelial injury, Int. J. Mol. Med. 14(2004) 1035–1041.

[6] M. Morigi, B. Imberti, C. Zoja, D. Corna, S. Tomasoni, M. Abbate, D.Rottoli, S. Angioletti, A. Benigni, N. Perico, M. Alison, G. Remuzzi,et al., Mesenchymal stem cells are renotropic, helping to repairthe kidney and improve function in acute renal failure, J. Am. Soc.Nephrol. 15 (2004) 1794–1804.

[7] K. Rjiba-Touati, I. Ayed-Boussema, A. Belarbia, A. Achour, H.Bacha, et al., Protective effect of recombinant human erythro-poietin against cisplatin-induced oxidative stress and nephro-toxicity in rat kidney, Int. J. Toxicol. 30 (2011) 510–517.

[8] F.L. Yang, Y.M. Subeq, Y.H. Chiu, R.P. Lee, C.J. Lee, B.G. Hsu, et al.,Recombinant human erythropoietin reduces rhabdomyolysis-induced acute renal failure in rats, Injury 43 (2012) 367–373.

[9] Y.R. Song, T. Lee, S.J. You, H.J. Chin, D.W. Chae, C. Lim, K.H.Park, S. Han, J.H. Kim, K.Y. Na, et al., Prevention of acute

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008

E X P E R I M E N T A L C E L L R E S E A R C H ] ( ] ] ] ] ) ] ] ] – ] ] ] 9

kidney injury by erythropoietin in patients undergoing coronaryartery bypass grafting: a pilot study, Am. J. Nephrol. 30 (2009)253–260.

[10] K.J. Zwezdaryk, S.B. Coffelt, Y.G. Figueroa, J. Liu, D.G. Phinney, H.L.LaMarca, L. Florez, C.B. Morris, G.W. Hoyle, A.B. Scandurro, et al.,Erythropoietin, a hypoxia-regulated factor, elicits a pro-angiogenic program in human mesenchymal stem cells, Exp.Hematol. 35 (2007) 640–652.

[11] A. Daga, A. Muraglia, R. Quarto, R. Cancedda, G. Corte, et al.,Enhanced engraftment of EPO-transduced human bone marrowstromal cells transplanted in a 3D matrix in non-conditionedNOD/SCID mice, Gene Ther. 9 (2002) 915–921.

[12] N.M. Liu, J. Tian, W.W. Wang, J. Cheng, D.Y. Hu, J.Y. Zhang, et al.,Effects of mesenchymal stem cells on renal tubular epithelialcells apoptosis in mice under ischemia/reperfusion and itsmechanism, Chin. J. Integrated Traditional West. Nephrol. 11(2010) 581–585.

[13] N.M. Liu, G.F. Han, J. Cheng, J. Huang, J. Tian, Erythropoietinpromotes the repair effect of acute kidney injury by bone-marrowmesenchymal stem cells transplantation, Exp. Biol. Med. (in press).

[14] J. Chen, Y. Li, L. Wang, M. Lu, X. Zhang, M. Chopp, et al.,Therapeutic benefit of intracerebral transplantation of bonemarrow stromal cells after cerebral ischemia in rats, J. Neurol. Sci.189 (2001) 49–57.

[15] I.M. Barbash, P. Chouraqui, J. Baron, M.S. Feinberg, S. Etzion, A.Tessone, L. Miller, E. Guetta, D. Zipori, L.H. Kedes, R.A. Kloner, J.Leor, et al., Systemic delivery of bone marrow-derivedmesenchymal stem cells to the infarcted myocardium: feasibility,cell migration, and body distribution, Circulation 108 (2003)863–868.

[16] S.M. Devine, C. Cobbs, M. Jennings, A. Bartholomew, R. Hoffman,et al., Mesenchymal stem cells distribute to a wide range oftissues following systemic infusion into nonhuman primates,Blood 101 (2003) 2999–3001.

[17] J. Yamaguchi, K.F. Kusano, O. Masuo, A. Kawamoto, M. Silver, S.Murasawa, M. Bosch-Marce, H. Masuda, D.W. Losordo, J.M. Isner,T. Asahara, et al., Stromal cell-derived factor-1 effects on ex vivoexpanded endothelial progenitor cell recruitment for ischemicneovascularization, Circulation 107 (2003) 1322–1328.

[18] F. Floridi, F. Trettel, S. Di Bartolomeo, M.T. Ciotti, C. Limatola, et al.,Signalling pathways involved in the chemotactic activity ofCXCL12 in cultured rat cerebellar neurons and CHP100 neuroe-pithelioma cells, J. Neuroimmunol. 135 (2003) 38–46.

[19] R. Horuk, et al., Chemokines beyond inflammation, Nature 393(1998) 524–525.

[20] N. Li, X. Lu, X. Zhao, F.L. Xiang, A. Xenocostas, M. Karmazyn, Q.Feng, et al., Endothelial nitric oxide synthase promotes bonemarrow stromal cell migration to the ischemic myocardium viaupregulation of stromal cell-derived factor-1alpha, Stem Cells 27(2009) 961–970.

[21] P. Lai, T. Li, J. Yang, C. Xie, X. Zhu, H. Xie, X. Ding, S. Lin, S. Tang, etal., Upregulation of stromal cell-derived factor 1 (SDF-1)expression in microvasculature endothelial cells in retinalischemia–reperfusion injury, Graefes Arch. Clin. Exp. Ophthal-mol. 246 (2008) 1707–1713.

[22] H. Ohnishi, S. Mizuno, T. Nakamura, et al., Inhibition of tubularcell proliferation by neutralizing endogenous HGF leads to renalhypoxia and bone marrow-derived cell engraftment in acuterenal failure, Am. J. Physiol. Renal Physiol. 294 (2008) F326–335.

[23] J.F. Ji, B.P. He, S.T. Dheen, S.S. Tay, et al., Interactions ofchemokines and chemokine receptors mediate the migration of

Please cite this article as: N. Liu, et al., Effect of erythropoietin on ththe acute kidney injury microenvironment, Exp Cell Res (2013), http

mesenchymal stem cells to the impaired site in the brain afterhypoglossal nerve injury, Stem Cells 22 (2004) 415–427.

[24] Y. Wang, Y. Deng, G.Q. Zhou, et al., SDF-1α/CXCR4-mediatedmigration of systemically transplanted bone marrow stromalcells towards ischemic brain lesion in a rat model, Brain Res. 1195(2008) 104–112.

[25] R.F. Wynn, C.A. Hart, C. Corradi-Perini, L. O’Neill, C.A. Evans, J.E.Wraith, L.J. Fairbairn, I. Bellantuono, et al., A small proportion ofmesenchymal stem cells strongly expresses functionally activeCXCR4 receptor capable of promoting migration to bone marrow,Blood 104 (2004) 2643–2645.

[26] M. Li, J. Yu, Y. Li, D. Li, D. Yan, Z. Qu, Q. Ruan, et al., CXCR4 positivebone mesenchymal stem cells migrate to human endothelial cellstimulated by ox-LDL via SDF-1 alpha/CXCR4 signaling axis, Exp.Mol. Pathol. 88 (2010) 250–255.

[27] J.S. Lin, Y.S. Chen, H.S. Chiang, M.C. Ma, et al., Hypoxic precon-ditioning protects rat hearts against ischaemia–reperfusioninjury: role of erythropoietin on progenitor cell mobilization,J. Physiol. 586 (2008) 5757–5769.

[28] C. Klopsch, D. Furlani, R. Gäbel, W. Li, E. Pittermann, M.Ugurlucan, G. Kundt, C. Zingler, U. Titze, W. Wang, L.L. Ong, K.Wagner, R.K. Li, N. Ma, G. Steinhoff, et al., Intracardiac injection oferythropoietin induces stem cell recruitment and improvescardiac functions in a rat myocardial infarction model, J. Cell.Mol. Med. 13 (2009) 664–679.

[29] S. Brunner, J. Winogradow, B.C. Huber, M.M. Zaruba, R. Fischer, R.David, G. Assmann, N. Herbach, R. Wanke, J. Mueller-Hoecker, W.M. Franz, et al., Erythropoietin administration after myocardialinfarction in mice attenuates ischemic cardiomyopathy associatedwith enhanced homing of bone marrow-derived progenitor cellsvia the CXCR-4/SDF-1 axis, FASEB J. 23 (2009) 351–361.

[30] E. Spandou, I. Tsouchnikas, G. Karkavelas, E. Dounousi, C.Simeonidou, O. Guiba-Tziampiri, D. Tsakiris, et al., Erythropoietinattenuates renal injury in experimental acute renal failureischaemic/reperfusion model, Nephrol. Dial. Transplant. 21(2006) 330–336.

[31] S. Barbero, R. Bonavia, A. Bajetto, C. Porcile, P. Pirani, J.L. Ravetti,G.L. Zona, R. Spaziante, T. Florio, G. Schettini, et al., Stromal cell-derived factor 1alpha stimulates human glioblastoma cell growththrough the activation of both extracellular signal-regulatedkinases 1/2 and Akt, Cancer Res. 63 (2003) 1969–1974.

[32] S. Bobis-Wozowicz, K. Miekus, E. Wybieralska, D. Jarocha, A.Zawisz, Z. Madeja, M. Majka, et al., Genetically modified adiposetissue-derived mesenchymal stem cells overexpressing CXCR4display increased motility, invasiveness, and homing to bonemarrow of NOD/SCID mice, Exp. Hematol. 39 (2011) 686–696.

[33] S. Dimmeler, A. Aicher, M. Vasa, C. Mildner-Rihm, K. Adler, M.Tiemann, H. Rütten, S. Fichtlscherer, H. Martin, A.M. Zeiher, et al.,HMG-CoA reductase inhibitors (statins) increase endothelialprogenitor cells via the PI 3-kinase/Akt pathway, J. Clin. Invest.108 (2001) 391–397.

[34] O. Kollet, S. Shivtiel, Y.Q. Chen, J. Suriawinata, S.N. Thung, M.D.Dabeva, J. Kahn, A. Spiegel, A. Dar, S. Samira, P. Goichberg, A.Kalinkovich, F. Arenzana-Seisdedos, A. Nagler, I. Hardan, M.Revel, D.A. Shafritz, T. Lapidot, et al., HGF, SDF-1, and MMP-9 areinvolved in stress-induced human CD34+ stem cell recruitmentto the liver, J. Clin. Invest. 112 (2003) 160–169.

[35] S. Neuss, R.K. Schneider, L. Tietze, R. Knüchel, W. Jahnen-Dechent, et al., Secretion of fibrinolytic enzymes facilitateshuman mesenchymal stem cell invasion into fibrin clots, CellsTissues Organs 191 (2010) 36–46.

e migration of bone marrow-derived mesenchymal stem cells to://dx.doi.org/10.1016/j.yexcr.2013.04.008