Amniotic Fluid Derived Stem Cells with a Renal Progenitor Phenotype Inhibit Interstitial Fibrosis in Renal Ischemia and Reperfusion Injury in Rats

Objectives Mesenchymal stem cells derived from human amniotic fluid (hAFSCs) are a promising source for cellular therapy, especially for renal disorders, as a subpopulation is derived from the fetal urinary tract. The purpose of this study was to evaluate if hAFSCs with a renal progenitor phenotype demonstrate a nephroprotective effect in acute ischemia reperfusion (I/R) model and prevent late stage fibrosis. Methods A total of 45 male 12-wk-old Wistar rats were divided into three equal groups;: rats subjected to I/R injury and treated with Chang Medium, rats subjected to I/R injury and treated with hAFSCs and sham-operated animals. In the first part of this study, hAFSCs that highly expressed CD24, CD117, SIX2 and PAX2 were isolated and characterized. In the second part, renal I/R injury was induced in male rats and cellular treatment was performed 6 hours later via arterial injection. Functional and histological analyses were performed 24 hours, 48 hours and 2 months after treatment using serum creatinine, urine protein to creatinine ratio, inflammatory and regeneration markers and histomorphometric analysis of the kidney. Statistical analysis was performed by analysis of variance followed by the Tukey’s test for multiple comparisons or by nonparametric Kruskal-Wallis followed by Dunn. Statistical significance level was defined as p <0.05. Results hAFSCs treatment resulted in significantly reduced serum creatinine level at 24 hours, less tubular necrosis, less hyaline cast formation, higher proliferation index, less inflammatory cell infiltration and less myofibroblasts at 48h. The treated group had less fibrosis and proteinuria at 2 months after injury. Conclusion hAFSCs contain a renal progenitor cell subpopulation that has a nephroprotective effect when delivered intra-arterially in rats with renal I/R injury, and reduces interstitial fibrosis on long term follow-up.


Introduction
Acute kidney injury (AKI) is defined as a rapid and progressive decrease in glomerular filtration rate following an insult, which leads to a loss of renal function [1,2]. It is a pathology that affects neonatal to geriatric patients, present in 2% of all hospitalized patients [3] and up to two thirds of patients in an ICU setting [4]. Ischemia-reperfusion (I/R) injury remains one of the leading causes of AKI [5,6]. Despite current clinical treatment for AKI, some patients remain dialysis-dependent or develop progressive chronic kidney disease (CKD) necessitating transplantation [7,8]. Moreover, I/R injury is an inevitable consequence of kidney transplantation and can induce delayed graft function [9].
The pathophysiology of I/R injury is complex and mainly characterized by proximal tubular cell damage, loss and inflammation, leading to tissue injury and dysfunction. Prolonged ischemia is one of the major contributors to the development of primary kidney graft failure, delayed graft function and enhanced immunogenicity in kidney transplantation, and has a major impact on long-term outcome [10][11][12]. I/R may enhance the expression of MHC molecules in renal epithelium and could thereby trigger allogeneic acute/chronic rejection [13]. Given the shortage of donor organs and usage of marginal donor kidneys for transplantation, novel treatment options to minimize renal I/R injury are needed [14].
Stem cell-based therapy is a promising approach for kidney regeneration. Mesenchymal stem cells (MSCs) possess the ability to home to areas of tissue injury due to the local production of growth factors and cytokines. An extensive review of the working mechanisms of stem cells lies beyond the scope of this introduction. Briefly, these cells have been shown to influence angiogenesis [15], to have immunomodulatory properties [16], to activate the endogenous progenitors and to promote cellular reprogramming [17]. Additionally, stem cells have also antifibrotic, anti-oxidant, anti-inflammatory [18,19] and anti-apoptotic properties [20,21].
The aim of this study was to evaluate whether amniotic fluid derived stem cells with a renal progenitor phenotype have a nephroprotective effect in I/R injury and can prevent renal fibrosis as a late consequence.

Isolation, expansion and characterization of stem cells from amniotic fluid
The Ethics Committee of the UZ Leuven approved the research program on the regenerative potential of amniotic fluid derived cells isolated from discarded amniotic fluid obtained after clinically indicated amniocentesis procedures. Consenting women in this study were between 15-22 weeks of pregnancy. Cell lines were excluded retrospectively when infection or genetic abnormalities were detected at a later stage. Briefly, amniotic fluid was filtered with a 40 μm strainer (BD bioscience, Erembodegem, Belgium) and centrifuged. The cell pellet was re-suspended in expansion medium consisting of α-MEM (Invitrogen, Ghent, Belgium), 15% fetal bovine serum (Invitrogen, Ghent, Belgium), 1% L-glutamine (Invitrogen, Ghent, Belgium), 1% penicillin/streptomycin (Invitrogen, Ghent, Belgium) and 18% Chang B and 2% Chang C (Irvine Scientific, Brussels, Belgium), plated in a petri dish and incubated at 37°C in 5% CO 2 . As soon as single cells attached to the dish, medium was replaced to remove debris and unwanted epithelial cells. All individual cells were monitored daily using a light microscope. After a few days, single colonies were mechanically transferred into a 96 well plate and expanded as monoclonal populations [44].
For flow cytometry, 100,000 hAFSCs at passage 4 were stained for mesenchymal markers CD117, CD44, CD73, HLA-ABC, CD24 (BD bioscience, Erembodegem, Belgium), CD29 (Acris, Herford, Germany), and the hematopoietic markers CD34 and CD45 (BD bioscience, Erembodegem, Belgium) (see Table 1). Cells were washed three times with PBS and 1% BSA AlexaFluor 594 goat anti-rabbit and then incubated with primary antibodies for 20 minutes at 4°C in the dark. Cells were then washed 3 times with PBS and then re-suspended in 200 μl of PBS. Cells were incubated with respective isotype controls and unstained cells were used to define reading settings. Analysis was performed with the FACS Canto (BD bioscience, Erembodegem, Belgium). Differentiation protocols (were performed as previously described [44]): Osteogenic differentiation. Osteogenic differentiation was induced by culturing hAFSCs at 70% confluence for 4 weeks in 'osteogenic differentiation medium' (Invitrogen). Differentiation was assessed by Alizarin staining (Sigma Aldrich, Diegem, Belgium) of the calcified extracellular matrix deposition.
Adipogenic differentiation. To induce adipogenic differentiation, cells were cultured at 100% confluence and subsequently differentiated with adipogenic differentiation medium for 14 days. This medium was composed of 10% FBS, 10 6 M dexamethasone, 0.5 M 3-isobutylmethylxanthine, 10 mg/mL insulin, and 200 mM indomethacin in Dulbecco's Modified Eagle Medium high glucose (Invitrogen). Differentiation into the adipogenic lineage was determined by Oil Red O staining (Sigma-Aldrich, MO, USA).
Chondrogenic differentiation. To induce chondrogenic differentiation, cells were cultured in high-density pellet mass cultures for 14 days. 20 mL droplets of cells suspension (400,000 cells resuspended in PBS) were seeded into individual wells of a 24-well plate. Cells were allowed to attach without medium for 3 h at 37°C in a 5% CO2 incubator and then cultured for 24h in growth medium. 24h later, the medium was replaced with chondrogenic differentiation medium (Invitrogen). Chondrogenic differentiation was determined by Alcian Blue staining (Sigma-Aldrich).
qPCR was performed for detection of differentiation markers on cells before and after differentiation. Briefly, RNA was extracted using TRIPURE (Roche, Vilvoorde, Belgium) according to the manufacturer's instructions and cDNA was synthesized from 1μg total RNA using TaqMan Reverse Transcriptase kit (Applied Biosystem, Gent, Belgium). qPCR assay consisted of initial UDG incubation and denaturation (50°C for 2min, 95°C for 2 min) followed by 40 cycles of a 2-step PCR (95°C for 15 sec, 60°C for 45 sec) using primers for the gene of interest, Platinum SyBR Green qPCR supermix-UDG (Invitrogen, Gent, Belgium) and 2μl cDNA. Primers used for amplification are provided in Table 2. GAPDH was used as a housekeeping gene to normalize mRNA levels and results were computed using the Delta Ct (Ct gene of interest -Ct GAPDH ) formula and subsequently normalized to the level before differentiation.

Identification of stem cells with a renal phenotype
To show the nephrogenic potential of a subtype of hAFSCs, multiple monoclonal populationsisolated from different gestational ages (from 15 until 22 and the 26th)-were analyzed by flow  [42] and by qPCR for the renal marker Kidney Specific Protein (KSP) [45][46][47], as well as for SIX2 and PAX2 (Table 2). cDNA from human embryonic kidney (hEK) from the 19 th gestational week was used as a control (hEK RNA was donated by Prof. Paul Winyard, University College London, UK). Among the 120 hAFSCs populations analyzed, one population was chosen for further experiment in vitro and in vivo. This population of hAFSC was negative for the hematopoietic marker CD34, had a high proliferative capacity (> 30 passages in culture) and was positive (>50%) for the markers CD24, KSP, SIX2 and PAX2.
For flow cytometry analysis 100,000 cells were used for each condition. Cells were fixed in 2% PFA for 15 minutes and then permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, Diegem, Belgium) for 15 minutes. Cells were blocked in buffer containing 1% BSA in PBS for 15 minutes on ice and then incubated with primary antibodies for SIX2 and PAX2 (1:100) in permeabilizing solution for 30 minutes on ice. Specific isotypes were used and are listed in Table 1. Secondary Ab were added at a concentration of 1:50 in 0.5% Tween 20 in PBS for 30 minutes. Cells were then washed twice with 1% BSA in PBS and resuspended in 200 μl of PBS and read with the FACS Canto.

Labelling and transplantation of isolated hAFSCs
To track the cells in vivo, a lentiviral vector platform was used to introduce the reporter gene β-Galactosidase (LacZ) into the selected monoclonal population of hAFSCs, as described previously [44]. Briefly, lentiviral vector particles were added to the cell culture with a multiplicity of infection of 10. X-gal staining was performed to investigate the transduction efficiency of lentiviral vector in hAFSCs. Cells were fixed in 25% glutaraldehyde and 37% of methanol for 2 minutes and incubated with X-gal solution (Sigma-Aldrich, Diegem, Belgium) overnight at 37°C. Stained cells were counted using light microscopy and expressed as a 'positive percentage' in relation to the total cell population (% positive cells). To study the localization of injected cells in vivo, 1x10 6 LacZ transduced hAFSCs were injected into abdominal aorta 6 hours after ischemia and reperfusion (I/R) injury [48]. Rats were harvested 6 and 24 hours after injection. Kidneys, lungs, heart, spleen and liver were frozen in OCT medium. Cryosections of 5 μm thickness were incubated overnight at 37°C with X-gal solution pH 8.5 (vWR International, Heverlee, Belgium), counterstained with paracarmine and mounted in Mowiol.

Genetic stability testing of hAFSCs
Chromosomal microarray analysis (CMA) was performed on the cell population selected for the in vivo experiment to exclude unbalanced chromosomal aberrations. Cells were trypsinized, centrifuged and cell pellets washed twice in PBS. Genomic DNA was extracted using the DNA mini kit (QIAGEN, Venlo, Netherlands) following the manufacturer's recommendations, with a final elution volume of 60 μl. 500 ng sample of DNA and sex-mismatched reference DNA were labelled and co-hybridized to an 8x60K chromosomal microarray (Oxford Gene Technology, OGT, Oxford, UK) as previously described [49]. Genomic DNA was labeled in Cy3 or Cy5 for 4 hours using the CytoSure Labelling Kit (Oxford Gene Technology), with no enzyme digestion. Hybridization was performed for 40 hours in a rotator oven (SciGene, CA, USA) at 65°C. Washing of arrays was performed using the Little Dipper (SciGene) with Agilent wash buffer solutions and subsequently dried using acetonitrile. Arrays were scanned using an Agilent microarray scanner at 2-μm resolution, followed by calculation of signal intensities using Feature Extraction software (Agilent Technologies, Dieghem, Belgium). Visualization of results and data analysis were performed using the CytoSure Interpret Software (Oxford Gene Technology), which uses the circular binary segmentation (CBS) algorithm. Genomic coordinates were based on build hg19. Quality control metrics were also monitored with the CytoSure Interpret software (Oxford Gene Technology).
Selected hAFSCs cell lines were analysed at an early passage in culture (passage 14) and at a late one (passage 33) to exclude acquired genetic aberrations. To exclude lentiviral induced genotoxicity the labeled LacZ-hAFSC cell line was also tested.

Induction of ischemia reperfusion injury and stem cell injections in the rat model
All procedures involving animals were approved by the Local Ethics Committee for animal experimentation of the Catholic University of Leuven (KU Leuven) (project number: 088/ 2013). Animals were housed at constant temperature and humidity, with 12:12-h light-dark cycles, and unrestricted access to standard diet and water. All experiments were performed in cohorts of adult male Wistar rats, 3 months old, weighing 200-300g. Rats (n = 10/group) were randomly assigned as follows: rats subjected to I/R injury and treated with fresh expansion media (= group I/R+ vehicle); rats subjected to I/R injury and treated with hAFSCs (= group I/ R + hAFSC) and sham-operated animals (= group no injury). The experimental time line is shown in S1 Fig.
I/R was induced in isoflurane-anesthetized rats as described previously [48]. After a midline laparotomy, the renal pedicles were located and isolated by blunt dissection. Renal pedicles were clamped with atraumatic vascular clamps for 50 min. Kidney reperfusion was confirmed visually after the clamps were removed. The timing of ischemia was optimized to obtain a reversible model of ischemic AKI with a minimum of vascular thrombosis without animal mortality. At six hours after I/R injury, 800 μl of cell suspension (1x10 6 cells) in fresh expansion media was administered intra-arterially in two steps. Briefly, the abdominal aorta was caudally dissected around the level of the renal arteries. In order to direct the flow of the injected cells towards the kidney, the aorta was occluded with a vascular clamp 1 cm caudal of the renal arteries and veins. Subsequently, the right vascular pedicle was isolated and both the right renal artery and vein were temporarily clamped for 10-20 seconds before injection of half the volume of cells. After injection of the initial 400 μL, the clamp was released and placed on the left renal pedicle, followed by injection of the remaining 400 μL. Fresh expansion media was injected in vehicle group rats using the same method. The operators were blinded to the actual administered products. A sham or 'no injury group' was operated in identical fashion, except that the renal pedicles were not clamped and no treatment was injected as previously described [50][51][52][53]. Peri-operative pain-relief of 0.05 mg/kg IP q12h of buprenorphine was administered for 2 days. The rats received 5 ml saline after the surgery and kept on a heating pad at 37°C until recovery of locomotion.

Assessment of renal function after I/R injury
Serum creatinine was measured by colorimetric assay (modified Jaffé technique) at 3 different time points: 24 hours, 48 hours and 2 months after I/R injury. Urine samples were collected at 24 hours (n = 5/group), 48 hours (n = 5/group) and 2 months (n = 5/group) and analyzed for creatinine concentration and urine protein-to-creatinine ratio determination. Total protein concentration was determined by silver staining according to the modified method of Heukeshoven and Dernick [54]. Each urine sample was diluted in water to normalize creatinine levels and treated with loading buffer containing 2% β-mercaptoethanol and Laemmli (Sigma-Aldrich, Diegem, Belgium) and boiled for 5 minutes at 100°C. Protein fractions were separated in a 10% SDS polyacrylamide gel by electrophoresis followed by silver staining. Gels were soaked in 50% methanol, 12% acetic acid, 0.1% formaldehyde and 0.1% glutaraldehyde solution for 45 minutes and then washed 3 times in 50% of methanol for 5, 10 and 15 minutes. Subsequently they were soaked for 45 seconds in 0.2g/L Na 2 S 2 O 3 and washed 3 times in distilled water. Finally, gels were stained in 2g/L silver nitrate for 20 minutes and washed in distilled water. Developing solution was composed of 60g/L Na 2 CO 3 and 0.1% formaldehyde. All products were obtained from Sigma-Aldrich, Belgium.

Histology and immunohistochemistry of renal tissue
Forty-eight hours and 2 months after I/R injury, rats were euthanized and the kidneys were perfused with saline to remove blood from the vascular beds. The kidney specimens were fixed in 4% neutral formalin and embedded into paraffin and then sectioned to 4 μm slides and processed for periodic acid-Schiff (PAS), Masson's Trichrome, Picrosirius red and immunostaining. All slides were evaluated by experienced renal pathologists (D.L.G) who were blinded to the sample identification. Images were analyzed qualitatively (scores) by independent observers and quantitatively using the ImageJ software. The tubular necrosis, cast and fibrosis scoring system ranged from 0 to 5+ (0, no changes; 1+, very occasional tubular profiles (usually <3/ section and <2/HPF) affected by lesion; 2+, more evident lesions affecting 25% of the HPF; 3+, lesions affecting between 25% and 50% of the HPF; 4+, lesions affecting between 50% and 75% of the HPF and 5+, lesion in more than 75% of the HPF).
All morphometric analyses were performed using ImageJ software. Twenty non-overlapping fields on the coronal sections from both kidneys were selected for quantification of acute kidney injury (AKI) at 24 hours, 48 hours and fibrosis at 2 months I/R. PAS stained slides were used for AKI quantification. The number of luminal hyaline casts and cell loss (denudation of the tubular basement membrane) were quantified per field using a 40X objective (high-power field [HPF]). Fibrotic surface area was quantified in the Masson's trichrome stained slides at 2 months after I/R injury. The digital color images were segmented (color deconvolution plugin) and further binarized in order to measure the percentage of the area stained in blue.
Immunostaining was performed on paraffin sections. After deparaffinization, endogenous peroxidase activity was blocked with 0.5% H 2 O 2 in PBS for 20 min at room temperature. Sections were then heated at 98°C for 1 hour in citrate buffer (10mmol/L, pH 6) to enhance antigen retrieval. Non-specific binding was minimized by incubating sections in 1% BSA and 2% milk in PBS-0.1% tween 80 for 30 min. Sections were then incubated overnight at 4°C with the primary polyclonal antibodies against α-smooth muscle actin at 1:100 dilution (α-SMA-clone 1A4, DAKO, Heverlee, Belgium), Ki67 at 1:50 dilution (ki-67 clone MIB-5, DAKO, Heverlee, Belgium) and CD68 at 1:200 (ED-1, ab31630, Abcam, Cambridge, UK). Negative controls included buffer alone. Specific labeling was detected with EnVision/HRP Detection Kit (DAKO, Heverlee, Belgium). The color reaction was developed with 3,3'-diaminobenzidine (Sigma-Aldrich, Diegem, Belgium) and sections were counterstained with Mayer hematoxylin. Sections were then dehydrated through graded ethanol, cleared in xylene, and mounted in dePex (BDH, vWR international, Belgium). Immunostained slides were quantified using Ima-geJ software. Twenty non-overlapping fields on the coronal section from both kidneys were selected for quantification. For CD68 and α-SMA images were assessed at 200X magnification. While for Ki67 expression, positive tubular cells were quantitatively assessed at 400X magnification. The digital color images were processed and analyzed as described above.

Molecular analysis of inflammation and injury/toxicity
Kidney samples harvested at 48 hours were snap frozen in liquid nitrogen and total RNA and cDNA synthesis were performed as described above. The list of primers used to detect mRNA expression are listed in Table 3. Three different reference genes were tested to find the most stable between biological groups (gapdh, pdhb and sdha). Gen Norm software was used to ascertain the most stable gene between groups. The geometric mean of pdhb and sdha were used to normalize mRNA level. Tgf-β1,Tnf-α, CD204 and CD206 were investigated to assess inflammation, eNOS and iNOS to assess oxidative stress, caspase-3 to quantify apoptosis and Kim1 to quantify injury and toxicity level in the samples. Results were computed using the Delta Ct (Ct gene of interest -Ct GAPDH ) formula and subsequently normalized to the level of the "No injury" group.

Statistics
Results were expressed as mean ± SE. Analysis of variance followed by Tukey's test for multiple comparisons was used to analyze renal function, proliferation, and morphometric parameters. Renal histology was analyzed by nonparametric Kruskal-Wallis followed by Dunn testing. qPCR data were analyzed by nonparametric t-test. Statistical significance level was defined as p <0.05.

Selection of hAFSCs with a renal phenotype
Stemness characteristics of hAFSCs were proven by differentiation into adipogenic, osteogenic and chondrogenic lineages as depicted in Fig 1E. Differentiation was quantified by dye extraction ( Fig 1D, right panel) and by qPCR analysis. Upon chondrogenic differentiation, there was an induction of Sox9 (3-fold), an important gene in the early phase of chondrogenesis and of  Collagen II (9-fold), a major protein component of cartilage [55]. Upon osteogenic differentiation there was minor increase in Runx2 and Osteocalcin while ALP expression was significantly increased (17-fold), which suggests that hAFSCs started the deposition of matrix proteins [56] (Fig 1C, middle panel). 61% of the isolated hAFSCs populations expressed the markers KSP, SIX2 and PAX2 (S2 Fig). hAFSCs were morphologically uniform and showed a characteristic trumpet shape ( Fig  1A, insert). The selected population expressed MSCs markers and was negative for the hematopoietic ones (Fig 1B). The monoclonal hAFSCs population used for the in vitro and in vivo experiments, expressed SIX2 and PAX2 by immunofluorescence staining (Fig 1C, left panel) and by qPCR (Fig 1D). Labelling of hAFSCs for in vivo tracking hAFSCs were efficiently transduced by the lentiviral vector platform (Fig 2A) as shown by the X-gal staining, which detected the nuclear activity of β-galactosidase in >95% of the cells  ( Fig 2B). These labeled hAFSCs maintained their mesenchymal profile when analyzed by flow cytometry (data not shown) and were still positive for nuclear X-gal staining after 20 passages in culture. Lac-Z labeled hAFSCs were tracked in both kidneys and lungs at 6 hours after injection ( Fig 2C) and only in the lungs and spleen at 24 hours after injection (data not shown). Tracking of cells at later time points (48h and 2m) did not reveal the presence of injected hAFSCs in the renal tissue (data not shown).
The selected cell population was analyzed for the occurrence of spontaneous chromosomal abnormalities or lentiviral mediated genotoxicity (Fig 2D). The imbalances observed for the X and Y chromosomes are due to sex-mismatch of the reference DNA sample which acts as an internal experimental control. No abnormal molecular karyotypes were detected in early passage in culture (passage 14) nor after expansion (passage 33). Lentiviral manipulation (Fig 2D) did not affect karyotyping stability.

Protective effect of stem cells in experimental renal I/R injury
To evaluate the effect of hAFSCs on AKI, we used a surgical model to induce ischemia/reperfusion damage in male Wistar rats [48]. Histological analysis of the kidneys of animals at 48h after injury showed proximal tubular injury characterized by tubular necrosis, denudation of the basal membranes and vacuolization of tubular epithelial cells (Fig 3A). In addition, proximal tubular cells showed a loss of brush borders and the presence of intratubular hyaline casts (Fig 3). The morphometric evaluation of AKI in the cohort treated with hAFSCs revealed a significantly lower number of hyaline cast-containing tubules (5.99 vs 11.87 cast/HPF, p<0.0001) The renal function, as measured by serum creatinine level, was impaired in the short term after I/R. The serum creatinine level in rats treated with hAFSC were significantly reduced at 24h, compared to animals treated with vehicle alone (1.7 versus 2.8 mg/dl, p = 0.0002). In addition, treatment with stem cells resulted in the normalization of the creatinine levels at 48h, whereas controls still had an increased creatinine level compared to uninjured animals (1.0 versus 1.2 mg/dl, p = 0.0412) (Fig 3F and 3G).

Effect of hAFSCs on tubular proliferation, inflammation and myofibroblast formation
To assess the regenerative response of kidney tissue in stem cells injected animals, we quantified Ki-67, CD68 and α-SMA expression at 48 hours after injury (Fig 4). hAFSCs enhanced the proliferation rate of tubular cells after I/R injury as detected by Ki67 expression compared to vehicle treated controls (14.2 vs 10.8, not significant) (Fig 4A-4C). Animals with no injury had negligible Ki67 expression in renal tissue. Moreover, in comparison with non-injured animals, I/R injury induced a massive infiltration of macrophages and an increased presence of myofibroblasts in the kidneys (Fig 4D-4I). A significant decrease of macrophage infiltration and myofibroblast formation was observed in rats treated with stem cells (0.76 vs 0.27, p< 0.0001) and (1771.63 vs 105.548, p<0.0001) respectively. The injury level quantified by qPCR showed an increased expression of Kim1 and Caspase-3 compared to the no-injury group. No statistical difference between treated groups was observed (Kim1: p = 0.9307; Caspase 3: p = 0.6623). To quantify the M1-type macrophage infiltration, iNOS and Tnf-α were assessed, whereas CD 206 and 204 were used to measure the presence of M2-type macrophages. Animals treated with hAFSC showed a significant down-regulation of iNOS, but not Tnf-α expression (iNOS: p = 0.0286; Tnf-α: p = 0.8413), nor did we observe any difference in the expression of mannose receptor (CD206) and scavenger receptor (CD204) (CD206: p = 0.9048; CD204: p = 0.1905). There was no difference in expression of Tgf-β1 (Tgf-β1: p = 0.1255) or eNOS (p = 0.1905) (S5 Fig).

Inhibition of the progression to chronic kidney disease
To assess if the nephroprotective effect observed in the early phase would also result in a reduced fibrogenesis in the long term, interstitial fibrosis was evaluated by Masson's Trichrome and Picrosirius red staining at 2 months after injury (Fig 5). Our analysis demonstrated significantly less interstitial fibrosis in the hAFSC-treated group compared to I/R+vehicle group (17.49 vs 13.24, p<0.05) (Fig 5A and 5B). These results were in line with fibrosis scores given by the pathologist (S4B Fig). Although there was no significant difference between the creatinine levels of the groups at 2 months (0.736 vs 0.755 mg/dl, p = 0,0616), we could observe an increase in the microproteinuria levels in the group IR+vehicle, while hAFSC-treated rats had urine protein levels comparable to the no injury group as shown by silver staining (Fig 5, bottom right panel). Hence, we observed an inhibitory effect of the stem cells on the progression of CKD after renal I/R injury.

Discussion
hAFSC represent a more heterogeneous population of stem cells when compared to many adult cell types such as bone marrow or adipose derived stem cells. The fetus sheds cells from the renal, gastrointestinal, dermal, respiratory system and even from the amniotic membranes [57] at different rates and ratios depending on the specific stages of organ development. Isolating stem cells of renal origin will thus depend on the structural and functional development of the renal tract. During nephrogenesis, after the initial contact of the ureteric bud with the metanephric mesenchyme, the condensed mesenchyme generates a population of stem and progenitor cells, which are found at the periphery of the nephrogenic zone in the developing kidney [58]. Such stem-like cells have been demonstrated to be present in the amniotic fluid and they were shown to be able to differentiate in vitro into podocytes [43]. Our experiments corroborate these findings. We have identified cells that behave as mesenchymal stem cells based on their expression markers and proliferation and differentiation capacity, but that are also positive for markers that are found in the fetal kidney.
In this study we used cells obtained from standard clinical amniocentesis procedures between 15-22 weeks of gestation. At this stage of the pregnancy a large part of the amniotic fluid consists of fetal urine [59,60]. In our samples most of the isolated colonies of stem cells at this stage expressed KSP and the renal progenitor markers SIX2 and PAX2. From these specific lineages we selected the clone with the highest proliferative capacity (passage number >30) and the highest expression of CD24, KSP, SIX2 and PAX2 (but not late renal differentiation markers data not shown). It has been shown that AFSCs express mesenchymal markers and differentiate in the mesenchymal lineages (osteogenic, chondrogenic and adipogenic), while at the same time they are able to differentiate into other tissue types (such as neurons,. . . [33]). In the previous studies, a stem-like cell population derived from human amniotic fluid and possessing characteristics of podocyte precursors were selected based on the expression of CD24 and OB-cadherin [42,43].
In the second part of our experiments, we investigated the therapeutic potential of these cells in an animal model of AKI. As we intended to locate the injected cells in our animal model, we introduced a reporter gene into our cell line. We opted for a stable genetic introduction as some cell populations are known to release live-cell-dyes [61,62] and that the transient overexpression of reporter gene decreases after every cell division, which leads to loss of signal. We ensured that the CD marker expression, the differentiation capacity and the karyotypic stability of the cell line remained unchanged after this genetic manipulation. We detected the injected cells in the kidney at 6h after injection but no longer at 24 or 48h. Initial research of MSC based therapy demonstrated their usage as safe with little to no host response [63]. However, the full range of MSC-mediated immune-modulation remains incompletely understood, as emerging reports also reveal that MSCs can adopt an immunogenic phenotype [64]. In our experiments we did not study the possible immune responses of the host towards the injected cells. We did not assume that they would be present for longer periods in a xenograft context with an immunocompetent host. We have chosen an arterial delivery route, as the IA administration of MSCs has been shown to result in a more significant nephroprotection [65][66][67].
The fact that the cells cannot be detected during later time points in the kidney suggests that hAFSCs are less likely to have an effect on the pathophysiological processes by direct integration into the kidney, but may play a paracrine role during the early phases of the ischemic events.
When we appraised the effect of the cells in the early phase of the ischemic event, we observed nephroprotective effects at 24h as shown by a significantly lower level of serum creatinine in the hAFSCs group compared to the controls. In this sublethal ischemia model the most pronounced increase in creatinine is noted at 24h, with normalization between 3 to 5 days [68], which is in line with the creatinine value at 48h in our study. We next determined the degree of tubular necrosis and tubular cast formation and detected a marked difference between the cohorts treated with hAFSC and the controls both at 24 and 48h. We also investigated the possible pathways by which the cells could exert their protective effects. Previous studies have focused on the proliferation of epithelial cells, inflammation and myofibroblast differentiation in the affected kidneys [69]. We noted a marked increase in the presence of proliferating cells in the renal tissue 48h after the ischemic episode. The levels of Ki67 positive cells were higher in both injured groups, but in the hAFSC cohort this was even higher. This suggests that the hAFSCs exert a part of their effect through stimulating proliferation and modulation of the regenerative processes in the damaged tissue. The presence of macrophages were also evaluated as a marker for the inflammatory response. The injection of stem cells led to a decreased infiltration of M1 macrophages into the ischemic organs as seen by lower expression of CD68 and iNOS. Mesenchymal stem cells are known to induce the switch from M1 to M2 macrophages [70] but no upregulation was seen in our hAFSCs treated cohort. Other research has shown that M2 macrophages are present in the late phases of the repair [71,72]. Finally we assessed the myofibroblast differentiation in the injured cohorts and noticed that the stem cells exert a negative effect on this process. The lower expression of α-SMA is indicative of inhibition of fibrosis, as myofibroblasts increase the deposition of extracellular matrix (especially type I collagen) leading to glomerular and tubular atrophy [73].
In the final part of this study we investigated the long-term effect of the hAFSCs injection on post-ischemic injury. We quantified the presence of subcapsular and interstitial fibrosis by morphometry. Based on our observations we concluded that hAFSCs significantly decrease fibrotic process in the renal I/R injury model in rats. These data are in line with research by Baulier E. et al., who show an inhibition of fibrosis in a porcine model of kidney transplantation treated with AFSCs at 6 days after transplantation [40]. At the functional level we noticed, in the untreated animals, a higher level of proteinuria, which was not present in the hAFSCs treated cohort. This suggests an inhibition of the progression to CKD after renal I/R injury since microproteinuria can indicate early glomerular injury [74].
The renoprotective mechanism of stem cell therapy has not been fully elucidated. Some studies suggest that MSCs are capable to fuse with resident tubular cells [75,76], others suggest that stem cells are able to differentiate into resident kidney cells [43,77]. Today the most prominent mechanism of action seems to constitute a paracrine effect [78][79][80]. Kidney constitutively express eNOS that plays an important role in the vascular homeostasis by promoting vasodilatation [81]. After renal I/R a downregulation of eNOS is observed which can exacerbate the injury by vasoconstriction and endothelial dysfunction. The increase of eNOS expression in the stem cell treated group seems to be one of the possible pathways involved in the nephroprotective effect of hAFSC in the I/R model. Further research will need to elucidate the exact pathways that are responsible for this effect and investigate if AFSC are indeed a promising experimental therapy for the treatment of AKI in men.

Conclusion
hAFSCs contain a subpopulation of clonal lineage with a renal progenitor phenotype that have a nephroprotective effect when delivered intra-arterially in rats with renal I/R injury, leading to an inhibition of interstitial fibrosis on long term follow up.

Acknowledgments
Our thanks go to Prof. Paul Winyard for kindly donating human embryonic kidney RNA sample used as control for qPCR analysis. We acknowledge Gabriela Bomfim Ferreira, Dries Braeken, Mary Patrice Eastwood and Katrien Luyten for their technical support.