www.kidney-international.org basic research
Generation of renal Epo-producing cell lines by conditional gene tagging reveals rapid HIF-2 driven Epo kinetics, cell autonomous feedback regulation, and a telocyte phenotype Faik Imeri1,2,5, Karen A. Nolan1,2,5, Andreas M. Bapst1,2, Sara Santambrogio1,2, Irene Abreu-Rodrı´guez1,2, Patrick Spielmann1,2, Svende Pfundstein1,2, Silvana Libertini1,2, Lisa Crowther1,2, Ilaria M.C. Orlando1,2, Sophie L. Dahl1,2, Anna Keodara1,2, Willy Kuo1,2, Vartan Kurtcuoglu1,2, Carsten C. Scholz1,2, Weihong Qi3, Edith Hummler2,4, David Hoogewijs1,2,6 and Roland H. Wenger1,2
1Institute of Physiology, University of Zurich, Zurich, Switzerland; 2National Center of Competence in Research “Kidney.CH”, Zurich, Switzerland; 3Functional Genomics Center Zurich, University of Zurich, Zurich, Switzerland; and 4Department of Pharmacology and Toxicology, University of Lausanne, Lausanne, Switzerland
Erythropoietin (Epo) is essential for erythropoiesis and is and to elucidate the molecular mechanisms of HIF mainly produced by the fetal liver and the adult kidney stabilizing drugs currently in phase III studies to treat following hypoxic stimulation. Epo regulation is anemia in end-stage kidney disease. commonly studied in hepatoma cell lines, but differences Kidney International (2019) 95, 375–387; https://doi.org/10.1016/ in Epo regulation between kidney and liver limit the j.kint.2018.08.043 understanding of Epo dysregulation in polycythaemia KEYWORDS: anemia; cell signaling; erythropoietin; hypoxia and anaemia. To overcome this limitation, we have Copyright ª 2018, International Society of Nephrology. Published by generated a novel transgenic mouse model expressing Elsevier Inc. All rights reserved. Cre recombinase specifically in the active fraction of renal Epo-producing (REP) cells. Crossing with reporter fi fi mice con rmed the inducible and highly speci c tagging rythropoietin (Epo)-regulated red blood cell homeo- of REP cells, located in the corticomedullary border stasis is crucial for oxygen delivery in vertebrates, and region where there is a steep drop in oxygen E 1,2 Epo dysregulation causes anemia or polycythemia. bioavailability. A novel method was developed to While the liver is the main site of Epo production in the selectively grow primary REP cells in culture and to embryo, the kidney produces about 90% of circulating Epo generate immortalized clonal cell lines, called in the adult, and only the remaining 10% is of hepatic fibroblastoid atypical interstitial kidney (FAIK) cells. FAIK 3,4 a origin. Local tissue oxygenation appears to be the major cells show very early hypoxia-inducible factor (HIF)-2 factor triggering Epo production. Intriguingly, Epo is induction, which precedes Epo transcription. Epo almost exclusively regulated on the mRNA level by induction in FAIK cells reverses rapidly despite ongoing hypoxia-inducible factor (HIF)-dependent transcriptional hypoxia, suggesting a cell autonomous feedback activation.5 Whereas hepatic and neuronal cell lines are mechanism. In contrast, HIF stabilizing drugs resulted in used to study Epo gene regulation,6,7 a reliable cell culture chronic Epo induction in FAIK cells. RNA sequencing of model derived from renal Epo-producing (REP) cells is three FAIK cell lines derived from independent kidneys currently not available. revealed a high degree of overlap and suggests that REP Under normal conditions, REP cells are located in the cells represent a unique cell type with properties of peritubular interstitial space of the inner cortex.8 Anumber fi pericytes, broblasts, and neurons, known as telocytes. of attempts to generate REP cell models have been reported. These novel cell lines may be helpful to investigate Cell lines derived from renal cell carcinoma occasionally fi myo broblast differentiation in chronic kidney disease produce Epo, but oxygen-regulated Epo expression has not been found in these cancer cells.9 Transgenic mice bearing a Epo Correspondence: Roland H. Wenger, Institute of Physiology, University of SV40largeTantigeninthe locus principally allowed for Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland. E-mail: the isolation and immortalization of REP cells, but cultured [email protected] cells showed no inducible Epo or large T expression.10 5These authors share first authorship. Kidney-derived mesenchymal progenitor cells were differ- in vitro 6Current address: Department of Medicine/Physiology, University of Fri- entiated to produce cell lines capable of oxygen- 11 bourg, 1700 Fribourg, Switzerland. regulated Epo expression. Transgenic mouse lines fl Received 15 January 2018; revised 18 August 2018; accepted 23 August expressing uorescent proteins under the control of regu- 2018; published online 27 November 2018 latory elements derived from the Epo locus (knock-in allele
Kidney International (2019) 95, 375–387 375 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
in a background of severe neonatal anemia) or from the RESULTS Col1a1 locus were used to isolate primary REP cells by flow Specific conditional targeting of active REP cells in vivo þ þ cytometry.12,13 Human mesenchymal-like CD133 /CD73 ACreERT2 expression vector was constructed using 220 kb of the progenitor cells isolated from the inner medulla showed mouse Epo locus as shown in Supplementary Figure S1.This increased Epo production following hypoxic stimulation.14 construct was first tested in vitro to confirm hypoxia-inducible Finally, several conditional knockout mouse models with Cre expression and tamoxifen-dependent Cre activation constitutive HIF (over-)expression resulted in ectopic (Figure 1a). Several transgenic founder lines were then gener- renal Epo synthesis, including models that involve the ated and crossed with reporter mice, which allowed for the promoters derived from the genes encoding for CD68, permanent tagging of REP cells by red fluorescent tdTomato renin, connexin 40, PDGFRb, and FOXD1 to drive Cre expression. While under normoxic conditions no or only very – expression.15 21 few fluorescent cells could be observed in the absence or In conclusion, none of these attempts resulted in a reliable presence, respectively, of tamoxifen (data not shown), this renal cell culture with permanent capability of regulated Epo number increased following exposure to hypoxic conditions expression, maybe due to the transient nature of Epo locus (Figure 1b). Pixel area quantification demonstrated that hypoxia activation by differentiation and oxygen-dependent signals. clearly activated more REP cells in founder lines 1, 240, and 241 ERT2 We hence reasoned that isolation of freshly isolated cells than in 244 (Figure 1c). Epo-Cre 1 and 241 were chosen for acutely tagged for an active Epo locus may enhance the chance subsequent analyses. As depicted in Figure 1d, tdTomato of obtaining REP cells. Therefore, we generated novel trans- expression overlaps with immunoreactivity of CD73, an genic mouse lines to conditionally tag REP cells and analyzed established marker of REP cells.22 Furthermore, the CD73 specific targeting of “on” REP cells by spatial comparison with surface marker revealed the long processes typical for REP oxygen bioavailability and hypoxic Epo mRNA expression. cells.8 High-resolution microscopy showed in more detail the Using this mouse model, primary REP cultures as well as irregular shape and the long processes of REP cells, extending clonal REP cell lines can be generated, recapitulating hall- between multiple tubular cells (Figure 1e), which further con- marks of Epo regulation. firms the reported phenotype.8
Figure 1 | Conditional targeting of renal Epo-producing (REP) cells in vivo. (a) In vitro analysis of the Epo-CreERT2 bacterial artificial chromosome transgenic vector in Hep3B cells by cotransfection of the bacterial artificial chromosome construct together with Cre-inducible and constitutive reporter genes. Dual luciferase reporter gene assays were performed following exposure to tamoxifen and/or hypoxia (0.2% O2 for 20 hours). (b) Red fluorescent tdTomato expression in REP cells of Epo-CreERT2#241xtdTomato mice following exposure to tamoxifen and hypoxia (0.1% CO for 4 hours). Nuclei were stained with 40,6-diamidino-2-phenylindole (DAPI; blue), and tubuli were visualized by their autofluorescence (green). (c) REP cell labeling of the initial Epo-CreERT2 founder lines following crossing with tdTomato reporter mice and exposure to tamoxifen and hypoxia. tdTomato was detected using an anti-red fluorescent protein antibody, and 7 to 11 sections of 2 female mice (3–5 months old) for each founder line were scanned on a slide scanner. The pixel areas of tdTomato and DAPI fluorescence over a manually defined threshold were quantified, and the tdTomato area is shown as percentage of the DAPI area (mean þ SEM). (d) REP cell co- staining (yellow) by anti-CD73/ecto-5’-nucleotidase antibodies (green) together with tdTomato fluorescence (red) following exposure to ERT2 tamoxifen and hypoxia (8% O2 for 16 hours) of Epo-Cre #1xtdTomato mice. Triple staining together with the DAPI nuclear stain (blue) results in white signals. (e) Deconvolution of laser scanning fluorescence microscopy of Epo-CreERT2#241xtdTomato mice following exposure to tamoxifen and hypoxia (0.1% CO for 4 hours). To optimize viewing of this image, please see the online version of this article at www.kidney- international.org.
376 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
Whole-genome DNA sequencing revealed that both REP cells localize to areas with abrupt decline in O2 strains show single integration sites, but whereas the vector bioavailability integrated in strain 1 into an intergenic region of chro- Renal Epo expression is exceptionally hypoxia-sensitive even mosome 14 with the loss of only 330 bp, in strain 241 it though the HIF system and its hundreds of target genes are 5 integrated into chromosome 9 with the loss of 52.2 kb, ubiquitously expressed. While the underlying molecular including exons 4 to 13 of the Pou2f3 gene. The remaining mechanisms remain to be elucidated, the unique oxygen Epo flanking regions in the integrated bacterial artificial distribution in the adult kidney is likely involved in REP cell chromosome vectors were 45.9 kb and 98.2 kb in 1, and activation. Indeed, mice ubiquitously expressing a fusion 75.2 kb and 52 kb in 241, upstream and downstream, protein between the HIF-1a oxygen-dependent degradation respectively, of the Epo-coding region (Supplementary (ODD) domain and the luciferase (Luc) reporter gene showed Figure S2A). Both, the 5’ and 3’ boundaries of the inte- a strikingly kidney-specific bioluminescence, as reported 24 grated vectors were confirmed by polymerase chain reac- previously. Only distal extremities and the normally hypoxic tion (examples shown in Supplementary Figure S2B). testis but no other inner organs showed a comparable Although the gross phenotypes of both founder strains hypoxia-induced bioluminescence (Figure 2a). Because were indistinguishable from that of wild-type mice, and whole-body imaging cannot reveal the intra-organ ODD-Luc despite identical REP cell tagging by tdTomato expression, distribution, we analyzed fresh organ slices of the kidneys strain 1 was chosen for further analyses due to the gene- shown in Figure 2a. Despite the now uniform tissue thickness, disrupting nature of the vector integration into strain bioluminescence intensity in the medulla was still much 241, which potentially could have deleterious effects.23 stronger than in the cortex. Both medulla and cortex
Figure 2 | Hypoxic activation and localization of renal Epo-producing (REP) cells. (a) Bioluminescence in vivo imaging following subcutaneous injection of luciferin into a male mouse ubiquitously expressing the oxygen-labile ODD-Luc fusion protein. (b) Bioluminescence imaging of a transversal thick slice (500 mm) of the kidney shown in a. Bright field microscopy (left), in vivo imaging system (IVIS) camera in bright light (middle), and IVIS bioluminescence (right) images are shown, demonstrating that the highest bioluminescence (red) is located in the medulla, medium bioluminescence (green) in the cortex, and lowest bioluminescence (blue) in the large vessels and outer cortical rim. (c) Immunofluorescence microscopy to localize tdTomato-positive REP cells detected using an anti-red fluorescent protein antibody (white) in a kidney derived from a Epo-CreERT2#241xtdTomato mouse after tamoxifen treatment (daily oral gavage of 5 mg tamoxifen for 5 consecutive days) followed by a repeated exposure to 0.1% CO for 4 hours with 1 week recovery in between. Kidney Epo mRNA in situ hybridization of Epo- ERT2 Cre #241 mice exposed to (d) CO as described above or (e) inspiratory hypoxia (8% O2) for 3 hours. (f) Relative numbers of tdTomato-positive REP cells in non-treated (n.t.) Epo-CreERT2#1xtdTomato mice, or after vehicle (DMSO; Veh) or tamoxifen (Tmx) pre-treatment (daily oral gavage of 2 mg tamoxifen for 4 consecutive days) followed by exposure to inspiratory hypoxia as indicated. Each bar represents average values þSEM of 6 to 20 slices derived from 1 mouse as indicated. No red fluorescent cells were detected in wild-type (wt) control mice. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.
Kidney International (2019) 95, 375–387 377 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
Figure 3 | Generation and characterization of fibroblastoid atypical interstitial kidney (FAIK) cell lines derived from primary renal Epo- producing (REP) cells. (a) Bright field (upper panel) and tdTomato fluorescence (lower panel) microscopy of the same primary REP cell preparation 1.5, 7, and 10.5 days after isolation following 3 days of selection with diphtheria toxin. (b) Hypoxic Epo mRNA induction in primary REP cells 6 days after cell isolation by exposure to 0.2% O2 for 24 hours. Ratios between Epo (measured by exon 4–5 (continued)
378 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
displayed a quite homogenous bioluminescence distribution, recombination, followed by treatment with diphtheria toxin. with a steep O2 bioavailability gradient mainly located in the While only few red cells were visible after 1.5 days of isolation, corticomedullary border region (Figure 2b). the abundance of tdTomato-positive REP cells increased after 7 We next analyzed the localization of REP cells in Epo- days of normoxic cultivation (Figure 3a). Most but not all of the ERT2 Cre #1xtdTomato mice. In order to compare the delayed viable cells were tdTomato-positive after 10.5 days, suggesting reporter protein accumulation with the short-lived Epo that some cells may not have recombined the tdTomato allele. mRNA levels, we used 2 relatively strong hypoxic stimuli Epo mRNA was induced by hypoxia in primary REP cells after 6 (each 4 hours of 0.1% CO inspiration) with 1 week recovery days of isolation (Figure 3b), demonstrating that these prolif- in between. While the first stimulus was used to perma- erating primary REP cell cultures maintained the capacity to nently tag REP cells, the subsequent second stimulus was express and regulate the Epo gene. used to analyze acutely induced Epo mRNA. Both tdTomato (Figure 2c) as well as endogenous Epo mRNA (Figure 2d) Oxygen, DNA methylation, and HIF-2 mediated Epo localized to peritubular interstitial cells of the cortico- regulation in REP cell-derived fibroblastoid atypical medullary border region. Strikingly, a much higher number interstitial kidney (FAIK) cell lines of REP cells was observed by histochemical Epo mRNA in Because primary REP cells ceased to proliferate and lost situ hybridization than tdTomato immunofluorescence, inducible Epo expression after approximately 12 to 14 days most likely due to the single-molecule sensitivity of the of culture (data not shown), we generated permanent REP former method, which detects Epo mRNA levels far below cell lines. Therefore, independent primary REP cell prep- the corresponding Cre expression threshold required for arations were immortalized with SV40 large T antigen 3 or recombination of the tdTomato reporter locus. Therefore, we 6 days after isolation. Limited dilution cloning resulted in analyzed Epo mRNA by the same technique following a FAIK cell lines. Three independent clones were chosen for more physiological hypoxic stimulus (3 hours of 8% O2 further analysis. FAIK cells maintained hypoxia-inducible inspiration). As shown in Figure 2e, this treatment results in Epo protein for at least 30 passages (Figure 3cand an REP cell pattern similar to the tdTomato-tagged REP cells Supplementary Figure S3). shown in Figure 2c. While no red cells were detected in wild- Because the Epo signal intensity was rather weak, we type mice, only sporadic cells were seen in transgenic mice reasoned that during propagation some of the cells may following tamoxifen treatment. This number did not further have lost Epo expression due to epigenetic silencing by Epo increase after 4 hours of 8% O2 inspiration (Figure 2f), DNA methylation of Epo-regulatory loci or the locus suggestingthatthisstimulusissufficient for Epo mRNA itself. DNA methyltransferase inhibition may hence restore induction but borderline for functional Cre expression. Epo expression in these cells as recently shown in fibrotic However, there was a clear increase in REP cell numbers kidneys in vivo.13 Indeed, 5-azacytidine could enhance both following 16 hours of 8% O2 inspiration, demonstrating that basal and hypoxia-inducible Epo protein levels in FAIK3-5 this mouse model faithfully recapitulates the spatio- cells (Figure 3c). These results were confirmed by enzyme- temporal conditional regulation of the Epo gene. linked immunosorbent assay using supernatants as well as cell lysates derived from the same cells (Figure 3d). A Isolation of primary REP cells by negative selection typical well-known feature of Epo gene expression is the Because the results shown above revealed that only a very small early decrease after an initial strong induction, despite fraction of all cells of the kidney are active “on” REP cells, we ongoing hypoxia and long before any change in hematocrit reasoned that selection against Epo-negative cells may be a more occurs.26,27 Interestingly, this feature could also be seen in suitable strategy to isolate primary REP cells than selection for FAIK cell lines as exemplified by Epo immunoblot detec- ERT2 Epo-positive cells. Therefore, Epo-Cre #1xtdTomato reporter tion in FAIK3-5 cells independent of whether 1 or 5 pas- mice were crossed with the Terminator strain, which allowed for sages went by since the 5-azacytidine treatment (Figure 3e). the killing of non–Cre-expressing cells using diphtheria toxin25 Bisulfite sequencing of a short Epo promoter region (Supplementary Figure S1). To select the acutely Epo-expressing revealed mostly methylation-free CpG dinucleotides prox- cells, freshly isolated renal cell suspensions were immediately imal to the transcription start site, whereas the more 5’ exposed to hypoxia and tamoxifen to induce Cre-mediated CpGs remained mostly methylated, independent of hypoxia = Figure 3 | (continued) reverse-transcription quantitative polymerase chain reaction) and ribosomal protein L28 mRNA levels were normalized to the normoxic control of each individual experiment. The mean induction factor þ SEM is shown (n ¼ 6). (c) Immunoblot detection of Epo protein in FAIK cell lines. FAIK3-5 cells at passage 9 or 10 were either left untreated or were pre-incubated with 1 mM 5-azacytidine (5-Aza) for 6 days, passaged once more, and then exposed to hypoxia (0.2% O2) for 24 hours. (d) Enzyme-linked immunosorbent assay measurements of mouse Epo (mEpo) protein in the supernatant (upper panel) and cell lysates (lower panel) of the same cells as shown in c (nd, not detectable). (e) Time course of hypoxic Epo protein induction in FAIK3-5 cells after pre-incubation with 5-Aza as in c, followed by exposure to hypoxia for the indicated time periods 1 or 5 passages later. (f) CpG methylation of the Epo promoter region in FAIK1-10 and FAIK3-5 cells determined by bisulfite sequencing. An example polymerase chain reaction product sequencing track is shown in the bottom panel (red arrows, CpG sites). (g) Knock-down of HIF-2a or Epo by short hairpin RNA expression (shCtrl, unrelated control short hairpin RNA). b-Actin served as loading and blotting control in all im- munoblots. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.
Kidney International (2019) 95, 375–387 379 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
or 5-azacytidine treatments (Figure 3f), suggesting that induction of HIF-2a protein often observed in cancer cell increased Epo expression following 5-azacytidine treatment lines.28 However, such a delayed HIF-2a induction would is not due to direct action on the Epo promoter in FAIK not be consistent with the rapid induction of its target cells. gene Epo in REP cells in vivo and FAIK cell lines in vitro. The mRNA levels in FAIK3-5 cells could efficiently be Intriguingly, while HIF-1a showed maximal protein levels blocked using lentiviral transduction with short hairpin RNA after 12 hours, HIF-2a induction was even faster in REP constructs (Figure 3g and Supplementary Methods and cor- cells, reaching maximal levels after 3 to 6 hours of hypoxia responding Supplementary Figures). shHIF-2a but not shCtrl (Figure 4a). Because such rapid kinetics is dominated by or shHIF-1a blunted Epo expression, confirming HIF-2– theslowoxygendiffusionandexchangerates,29 these dependent Epo regulation. experiments were repeated using cell culture media that had been pre-equilibrated with 0.2% O2.WhileHIF-2a Faithful recapitulation of hypoxic Epo expression kinetics in was already induced at the earliest time point (1.5 hours), FAIK cell lines the phosphorylated high molecular weight form of HIF- A well-established difference between HIF-1a and HIF-2a 1a maximally accumulated after 12 hours (Figure 4b, is the delayed (after 2 to 3 days) and prolonged hypoxic upper panel). Under these conditions, Epo protein
Figure 4 | Hypoxia-inducible factor (HIF) and Epo expression kinetics. Time course of hypoxic HIF-1a,HIF-2a, and Epo levels in fibroblastoid atypical interstitial kidney (FAIK)3-5 cells (a) following exposure to 0.2% O2 in a conventional incubator or (b) after medium replacement by pre- equilibrated medium in a hypoxic workstation. (c) Enzyme-linked immunosorbent assay determination of Epo concentrations (left panel) and total amounts (right panel) in lysates and supernatants (n above background is indicated) prepared from cells treated as in b. Shown are mean values þ SEM of n ¼ 4 independent experiments. Hypoxic increase versus normoxic control was evaluated using unpaired 2-tailed t-tests (*P < 0.05). (d)Ratios between Epo (measured by exon 3/4-4 reverse-transcription quantitative polymerase chain reaction) and ribosomal protein S12 mRNA of cells treated as in b (shown are mean hypoxic induction factors þ SEM; n ¼ 3). (e)TimecourseofhypoxicHIF-1a,HIF-2a, and Epo protein induction in FAIK3-5 cells following exposure to FG-4592 (roxadustat) or 0.2% O2.(f) Parallel Epo mRNA (upper panel) and protein (lower panel) detection in FAIK3-5 cells treated with FG-4592 for up to 72 hours. Shown are triplicate mRNA measurements þ SEM relative to the vehicle (DMSO) treated control. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.
380 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
Kidney International (2019) 95, 375–387 381 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
followed the HIF-2a kinetics with a maximal level already Established HIF target genes,33 including VEGFa, Glut1, after 1.5 hours followed by a gradual decrease, as shown DEC2, and NDRG1 showed a very robust hypoxia-inducible by immunoblotting (Figure 4b, lower panel), enzyme- expression, while housekeeping ribosomal protein rpL28 linked immunosorbent assay (Figure 4c), and mRNA was constitutively expressed (Figure 5a). However, CAIX, quantification (Figure 4d). Of note, this rapid reversal of which is commonly induced in virtually all cancer cell lines, is hypoxic Epo induction could not be observed using the almost completely absent even in hypoxic FAIK cells (data not prolyl-4-hydroxylase domain (PHD) inhibitor FG-4592 shown), underlining the non-transformed nature of these cell (roxadustat), which led to a permanent Epo expression lines. Of the mRNAs involved in the oxygen signaling cascade, until the end of the observation period after 48 hours, PHD2 was most prominently expressed, in line with its pre- while the hypoxic levels were decreased after 24 hours dominant function in renal Epo regulation.34 PHD2 and (Figure 4e). As shown in Figure 4f, a single bolus of FG- PHD3, but not PHD1, were significantly induced by hypoxia, 4592 maintained Epo mRNA and protein levels for up to consistent with a role in an intrinsic negative feedback loop.30 72 hours, suggesting that chemical PHD inhibition blocks Intriguingly in this context, also VHL (but not FIH) was a potential negative feedback loop seen in prolonged induced by hypoxia, at least in the FAIK cell lines analyzed hypoxia.30 (Figure 5b). Epo mRNA did not reach the threshold of 10 normalized sequence reads. Considering (i) the basal levels Transcriptomics of FAIK cell lines reveals metabolic close to zero, (ii) the nonoptimal time point for Epo mRNA adaptation to hypoxia induction, (iii) the small size of the Epo mRNA, and (iv) the The 3 independent FAIK cell lines were further phenotyped low Epo versus housekeeping control ratio usually observed by RNA sequencing following exposure to hypoxia for 24 by reverse-transcription quantitative polymerase chain reac- hours. Statistical evaluation confirmed the reproducibility tion, RNA sequencing was not sensitive enough for Epo of the REP cell isolation method (Supplementary detection under these experimental conditions. However, Figure S4A). Nevertheless, unsupervised cluster analysis manual inspection revealed Epo mRNA sequencing reads showed that FAIK2-5 were more distant from FAIK1-10 below the threshold in all 3 FAIK cell lines under hypoxic andFAIK3-5,withnormoxicandhypoxicFAIK2-5being conditions only (data not shown). more closely related than the distinct normoxic and hyp- oxic clusters formed by the 2 other cell lines FAIK cells resemble the telocyte subtype of pericytes and (Supplementary Figure S4B). However, with a total of fibroblasts 16,357 expressed mRNA and lncRNA species, the majority To characterize the developmental origin of FAIK cell lines, a (84.8%) was commonly expressed in all 3 cell lines. There panel of cell-type–specific markers was analyzed. All FAIK cell waslessoverlapinthesequencereadnumbersofthe lines were positive for the fibroblast markers PDGFRb,19 hypoxically regulated (i.e., at least a 2-fold change) RNA CD73,22 and Col1A113 (Figure 5c). Also FSP135 and the species between the 3 cell lines, with 14.7% regulated, myofibroblast marker aSMA36 were expressed, indicating that 21.1% induced, and 8.8% repressed RNA species, respec- at least a subpopulation of these cells may have differentiated tively (Supplementary Figure S4C). A rank order list of the during cell culture. 100 most strongly induced genes is shown in REP cells have recently been shown to express the neuronal Supplementary Figure S5. In addition to the approximately markers MAP2 and NFL.8 While we could confirm the 700 to 950 mRNA species also 120 to 200 lncRNA species expression of MAP2, NFL was not found in FAIK cell lines, in accumulated under hypoxic conditions in the 3 FAIK cell line with a previous report.19 However, they were clearly posi- lines (Supplementary Figure S4D), in line with our previ- tive for the neuronal markers NGF, BDNF, and nestin, as well as ous reports on hypoxic inducibility of lncRNAs.31,32 Gene for the pericyte marker NG2/CSPG4 (Figure 5d) that has ontology analysis revealed metabolic reprogramming to- recently been used to trace brain pericytes as a major source of ward HIF-1–dependent glycolysis under hypoxic condi- Epo in the mouse brain.37 The neural crest markers P0 and tions (Supplementary Table S1). Pax3, which have been used for the lineage tracing of REP
= Figure 5 | Oxygen-regulated and cell type–specific marker gene expression in fibroblastoid atypical interstitial kidney (FAIK) cell lines. Normalized RNA sequence read numbers in FAIK1-10, FAIK2-5 and FAIK3-5 cell lines determined by RNA sequencing as shown in Supplementary Figures S4 and S5.(a) Transcript levels of established hypoxia-inducible and control genes. (b) Transcript levels of genes involved in the oxygen signaling cascade. (c) Transcript levels of marker genes of the fibroblast lineage. (d) Transcript levels of marker genes of the neuronal lineage. (e) Transcript levels of hematopoietic marker genes. Mean values þ SEM of 3 independent experiments are shown. Hypoxic regulation was evaluated using paired 2-tailed t-tests (ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001). aSMA, a smooth muscle actin; BDNF, brain-derived neurotrophic factor; CD, cluster of differentiation; Col1A1, collagen 1A1; FIH, factor inhibiting hypoxia-inducible factor; FoxD1, forkhead box D1; FSP, fibroblast-specific protein; Glut1, glucose transporter 1; HSA, mouse heat-stable antigen; MAP2; microtubule-associated protein 2; NDRG1, N-myc downstream regulated 1; NG2, neuron-glial antigen 2; NGF, nerve growth factor; PDGFRb, platelet-derived growth factor receptor b; PHD, prolyl-4-hydroxylase domain; rpL28, ribosomal protein L28; VEGFa, vascular endothelial growth factor A; VHL, von Hippel-Lindau; WT1, Wilms tumor protein 1.
382 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
Kidney International (2019) 95, 375–387 383 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
cells,38,39 could either not be detected in FAIK cells or only at “telopodes.”43 Of note, while REP cells in vivo contain far less low levels (data not shown). mitochondria than the neighboring tubular epithelial cells, The transcription factors FoxD1 and WT1 are markers of mitochondria can also be detected in the membrane pro- the mesenchymal stroma precursor cells of the developing trusions of REP cells (Figure 6e). Finally, all 3 FAIK cells lines kidney that give rise to renal interstitial cells, including fibro- are positive for CD73 as well as for CD34 and PDGFRb blasts and pericytes. FoxD1 has previously been used to trace (Figure 6f), consistent with marker expression in telocytes.45,47 REP cells.20 Both markers are expressed in all FAIK cell lines, and WT1 was strongly induced by hypoxia. Because WT1 is DISCUSSION known to regulate the Epo promoter,40 hypoxia-inducible WT1 The renal Epo source has been discussed in the literature may potentiate Epo induction in REP cells. The macrophage quite controversially over the past decades. Not only peri- – marker CD68 is weakly expressed in a hypoxia-inducible tubular interstitial cells,22,50 52 but also proximal and distal manner in all FAIK cell lines. Notably, CD68 promoter- convoluted tubules, intercalated cells of the collecting ducts, – driven Cre transgenic mice have previously been used to and glomerular cells53 56 have been reported to contain Epo trace REP cells.15 Furthermore, FAIK cells were also positive for mRNA. These findings were apparently consistent with the stemness marker CD24, also known as mouse heat-stable transgenic mouse models expressing reporter genes under the antigen,41 and the hematopoietic progenitor marker CD34,42 control of DNA fragments derived from the Epo locus.57,58 although the latter could not be detected in FAIK2-5 cells However, these data were all obtained by “classical” mRNA (Figure 5e). However, endothelial cell marker CD31 as well as in situ hybridization or by using genomic Epo DNA fragments markers of leukocytes (CD45), dendritic cells (CD8a, CD11b, of limited length. With the recent refinement of the in situ and CD103), and macrophages (Mac1, F4/80) could not be hybridization technique, exclusively peritubular interstitial detected, clearly demonstrating that FAIK cell lines are not of cells at the corticomedullary border were found to be Epo myeloid origin. mRNA positive.59 These findings have been confirmed by The morphology of primary REP cells and FAIK cell lines transgenic mouse models expressing reporter genes under the showed some characteristic features (e.g., few long and thin control of longer regions of the Epo locus,10,60 with membrane protrusions partially forming cell-cell connec- approximately 200-kb long bacterial artificial chromosomes tions, large nuclei, and an apparent lack of nucleoli in most faithfully recapitulating the endogenous Epo expression 40,6-diamidino-2-phenylindole [DAPI] stainings) that are pattern.8 The results presented in our study confirm corti- reminiscent of a subpopulation of pericytes and fibroblasts comedullary peritubular interstitial cells as sole REP cell type. “ ”43 also referred to as telocytes. Telocytes can be found in The corticomedullary region showed the deepest drop in O2 most organs, including in the kidney cortex (but not me- bioavailability, which may expand into the cortex upon – dulla).44 46 Telocytes display a number of phenotypical and decreased tissue oxygenation, a prerequisite for the recruit- transcriptomic properties that distinguish them from “clas- ment of additional REP cells.61 sical” fibroblasts.43,47,48 As shown in Table S2, there is an Our animal model allowed for the conditional tagging of intriguing overlap between markers expressed in telocytes and acutely recruited REP cells, whereas in previously established in FAIK cell lines. These markers were compiled from reports constitutively Cre-driven REP models it remained unknown on telocytes derived from various organs, and some of the whether labeled REP cells were still active.8,12,39,62 Whole- ERT2 mismatches may hence be explained by the fact that only little genome sequencing of 2 Epo-Cre founder strains information about telocytes isolated from the kidney is revealed that 45.9 kb 5’ and 52 kb 3’ flanking regions are available. sufficient to confer accurate tissue-specific and oxygen- FAIK cell lines kept in low-density culture commonly form regulated Epo gene expression in the kidney. While Epo is long and thin membrane protrusions (nanotubes), some of known to be expressed in several other organs,2 we only oc- which apparently establish cell-cell connections (Figure 6a). casionally detected tagged cells in liver and brain (data not These nanotubes can reach lengths of more than 100 mmand shown). This may be explained by comparably low Epo levels are positive for TOM20 (mitochondria), b-actin (microfila- per cell, but quite high overall Epo expression per organ, ments), and a-tubulin (microtubuli) (Figure 6b–d), typical for because a far higher percentage of liver (hepatic and stellate) tunneling nanotubes.49 Such mitochondria-containing nano- and brain (neuronal, glial, and pericytic) cells are (mildly) tubes represent another hallmark of telocytes referred to as Epo-positive compared with the proportion of (strongly)
= Figure 6 | Telocyte phenotype of fibroblastoid atypical interstitial kidney (FAIK) cell lines. (a) Live microscopy showing thin membrane protrusions (nanotubes) formed in low-density FAIK cell culture. (b–d) Localization of mitochondria in FAIK cell lines following immunode- tection by polyclonal (pAb) or monoclonal (mAb) anti-TOM20 antibodies using epifluorescence (b,c) or confocal (d) microscopy in combination with anti-b-actin (b,d) or anti-a-tubulin (c) immunodetection. (a–d) Arrows indicate nanotubes containing mitochondria, arrowheads indicate nanotubes forming cell-cell contacts; tube lengths are indicated in mm. (e) TOM20 immunofluorescence of Epo-CreERT2#1xtdTomato mouse kidney. Arrows indicate mitochondria in renal Epo-producing cells. (f) Immunofluorescence detection of the telocyte markers CD34 and PDGFRb. F-Actin was stained with phalloidin. To optimize viewing of this image, please see the online version of this article at www.kidney- international.org.
384 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
Epo-positive cells in the kidney. Overall, these findings sug- Cell culture gest that our novel conditional mouse model selectively labels Primary REP cells and clonal cell lines were generated and charac- only those cells with strong acute Epo transcription rates, terized as outlined in Supplementary Methods. ideally suited for the generation of primary active “on” REP Protein and transcript analyses cells. Mouse Epo protein was detected by enzyme-linked immunosorbent While we did not obtain any detectable Epo expression in 71,72 11,63 assay and immunoblotting as described previously. mRNA levels previously published REP cell lines (data not shown), the were quantified by reverse-transcription quantitative polymerase strong transcriptomic correlation of 3 independent REP- chain reaction as described previously28 using the polymerase chain derived FAIK cell lines demonstrated the reproducibililty of reaction primers provided in Supplementary Table S3. Poly-A the described procedure. More than 50% of the isolated FAIK selected RNA was sequenced as detailed in Supplementary clones still showed hypoxia-inducible Epo expression, but Methods, and the RNA sequencing data were deposited in the Eu- even these clonal cell lines apparently kept the typical REP cell ropean Nucleotide Archive (www.ebi.ac.uk/ena; study accession property of the “on-off” behavior of Epo gene expression.64,65 number PRJEB19328). This REP cell feature is poorly understood and may involve DISCLOSURE intrinsic transcriptional negative feedback loops as well All the authors declared no competing interests. as epigenetic mechanisms.13,30 Indeed, treatment with 5-azacytidine increased Epo levels, suggesting an at least ACKNOWLEDGMENTS partial epigenetic silencing in “off” REP cells by DNA The authors wish to thank T. Buch and E. Campeau for the gift of methylation. Intriguingly, FG-4592/roxadustat, the most plasmids; L.G. Cantley for the Terminator mouse strain; and the advanced PHD inhibitory drug currently in clinical phase 3 Zurich Integrative Rodent Physiology facility, the Transgenic and Reproductive Techniques group of the University of Lausanne, the studies for the treatment of renal anemia,66 led to a consti- Functional Genomics Center Zurich, and the Center for tutive Epo induction in FAIK cells, whereas hypoxia only Microscopy and Image Analysis for expert contributions to this transiently induced Epo, suggesting differences in the negative work. This study was supported by the NCCR “Kidney.CH,” by the feedback regulation and/or interference with myofibroblast Swiss National Science Foundation (grant nos. 165679 to RHW and differentiation.67 However, our findings on HIF-2a/Epo ki- 153523 to VK), by the Hartmann Müller-Stiftung (grants to SL and Epo KAN), and by the EU’s seventh FP for research, technological netics clearly demonstrate that the transient expression in development, and demonstration (grant agreement no. 608847 to chronic hypoxia is a cell autonomous effect and does not IAR and KAN). require, for example, altered blood composition or micro- environmental changes. SUPPLEMENTARY MATERIAL RNA sequencing of FAIK cells allowed for a thorough Supplementary Methods. characterization of the cellular identity of REP cells. Based on Figure S1. Generation of a mouse line to conditionally target renal these data, we suggest that REP cells represent a unique subset Epo-producing (REP) cells and to select against non-REP cells. of interstitial pericytes and fibroblasts that are also known as Figure S2. Genotyping of Epo-CreERT2 transgenic mouse lines. telocytes and that were found to express a number of Figure S3. Persistent hypoxia-inducible Epo expression in fibro- blastoid atypical interstitial kidney (FAIK)3-5 cells. neuronal markers. While further analyses will be required to fi fi Figure S4. RNA sequencing of broblastoid atypical interstitial kidney con rm this suggestion, it is an intriguing idea that also (FAIK) cell lines. telocytes residing in other organs may represent major Figure S5. Heat map of hypoxia-inducible gene expression in fibro- sources of nonrenal Epo. Of note, while both neurons and blastoid atypical interstitial kidney (FAIK) cell lines. astrocytes have been well-established as Epo-producing cells Figure S6. Testing of antibodies to detect mouse Epo and HIFa. of the brain,68,69 a recent report demonstrated that brain Table S1. List of significantly (P < 0.05) enriched Kyoto Encyclopedia pericytes actually represent a major source of cerebral Epo of Genes and Genomes (KEGG) pathways in the 3 fibroblastoid upon hypoxic stimulation.37 It will be of interest to analyze atypical interstitial kidney (FAIK) cell lines. The input list contained all genes that were at least 2-fold hypoxically induced (P # 0.001) as whether Epo-producing pericytes of the brain also share determined by RNA sequencing. features of telocytes. Table S2. Comparison of marker gene expression in telocytes and fibroblastoid atypical interstitial kidney (FAIK) cell lines. MATERIALS AND METHODS Table S3. Primers used for polymerase chain reaction. Animals Supplementary References (References for the Supplementary ERT2 Epo-Cre transgenic mice were generated according to previously Methods and Supplementary Tables S2 and S3). published methods,70 using a newly constructed expression vector as Supplementary material is linked to the online version of the paper at detailed in Supplementary Methods. All animal experiments were www.kidney-international.org. approved by the veterinary office of the canton Zurich (license numbers ZH126/13 and ZH233/15). REFERENCES 1. Semenza GL. Involvement of oxygen-sensing pathways in physiologic Tissue analyses and pathologic erythropoiesis. Blood. 2009;114:2015–2019. fl in 2. Wenger RH, Kurtz A. Erythropoietin. Compr Physiol. 2011;1:1759–1794. Mouse kidney sections were analyzed by immuno uorescence or 3. Schooley JC, Mahlmann LJ. Erythropoietin production in the anephric rat. situ hybridization using RNAscope technology (see Supplementary I. Relationship between nephrectomy, time of hypoxic exposure, and Methods for antibody and imaging details). erythropoietin production. Blood. 1972;39:31–38.
Kidney International (2019) 95, 375–387 385 basic research F Imeri et al.: Epo kinetics in renal telocyte-like cell lines
4. Fried W. The liver as a source of extrarenal erythropoietin production. 29. Wenger RH, Kurtcuoglu V, Scholz CC, et al. Frequently asked questions in Blood. 1972;40:671–677. hypoxia research. Hypoxia. 2015;3:35–43. 5. Wenger RH, Hoogewijs D. Regulated oxygen sensing by protein 30. Stiehl DP, Wirthner R, Köditz J, et al. Increased prolyl 4-hydroxylase hydroxylation in renal erythropoietin-producing cells. Am J Physiol Renal domain proteins compensate for decreased oxygen levels. Evidence for Physiol. 2010;298:F1287–F1296. an autoregulatory oxygen-sensing system. J Biol Chem. 2006;281:23482– 6. Goldberg MA, Glass GA, Cunningham JM, et al. The regulated expression 23491. of erythropoietin by two human hepatoma cell lines. Proc Natl Acad Sci 31. Wollenick K, Hu J, Kristiansen G, et al. Synthetic transactivation screening USA. 1987;84:7972–7976. reveals ETV4 as broad coactivator of hypoxia-inducible factor signaling. 7. Stolze I, Berchner-Pfannschmidt U, Freitag P, et al. Hypoxia-inducible Nucl Acids Res. 2012;40:1928–1943. erythropoietin gene expression in human neuroblastoma cells. Blood. 32. Lelli A, Nolan KA, Santambrogio S, et al. Induction of long noncoding 2002;100:2623–2628. RNA MALAT1 in hypoxic mice. Hypoxia. 2015;3:45–52. 8. Obara N, Suzuki N, Kim K, et al. Repression via the GATA box is essential 33. Wenger RH, Stiehl DP, Camenisch G. Integration of oxygen signaling at for tissue-specific erythropoietin gene expression. Blood. 2008;111:5223– the consensus HRE. Sci STKE. 2005;2005:re12. 5232. 34. Takeda K, Aguila HL, Parikh NS, et al. Regulation of adult erythropoiesis 9. Sherwood JB, Shouval D. Continuous production of erythropoietin by an by prolyl hydroxylase domain proteins. Blood. 2008;111:3229–3235. established human renal carcinoma cell line: development of the cell 35. Higgins DF, Kimura K, Bernhardt WM, et al. Hypoxia promotes line. Proc Natl Acad Sci USA. 1986;83:165–169. fibrogenesis in vivo via HIF-1 stimulation of epithelial-to-mesenchymal 10. Maxwell PH, Osmond MK, Pugh CW, et al. Identification of the renal transition. J Clin Invest. 2007;117:3810–3820. erythropoietin-producing cells using transgenic mice. Kidney Int. 1993;44: 36. Souma T, Nezu M, Nakano D, et al. Erythropoietin synthesis in renal 1149–1162. myofibroblasts is restored by activation of hypoxia signaling. J Am Soc 11. Plotkin MD, Goligorsky MS. Mesenchymal cells from adult kidney support Nephrol. 2016;27:428–438. angiogenesis and differentiate into multiple interstitial cell types 37. Urrutia AA, Afzal A, Nelson J, et al. Prolyl-4-hydroxylase 2 and 3 including erythropoietin-producing fibroblasts. Am J Physiol Renal coregulate murine erythropoietin in brain pericytes. Blood. 2016;128: Physiol. 2006;291:F902–F912. 2550–2560. 12. Pan X, Suzuki N, Hirano I, et al. Isolation and characterization of renal 38. Asada N, Takase M, Nakamura J, et al. Dysfunction of fibroblasts of erythropoietin-producing cells from genetically produced anemia mice. extrarenal origin underlies renal fibrosis and renal anemia in mice. J Clin PLoS One. 2011;6:e25839. Invest. 2011;121:3981–3990. 13. Chang YT, Yang CC, Pan SY, et al. DNA methyltransferase inhibition 39. Suzuki N, Hirano I, Pan X, et al. Erythropoietin production in restores erythropoietin production in fibrotic murine kidneys. J Clin neuroepithelial and neural crest cells during primitive erythropoiesis. Nat Invest. 2016;126:721–731. Commun. 2013;4:2902. 14. Bussolati B, Lauritano C, Moggio A, et al. Renal CD133þ/CD73þ 40. Dame C, Kirschner KM, Bartz KV, et al. Wilms tumor suppressor, Wt1, is a progenitors produce erythropoietin under hypoxia and prolyl transcriptional activator of the erythropoietin gene. Blood. 2006;107: hydroxylase inhibition. J Am Soc Nephrol. 2013;24:1234–1241. 4282–4290. 15. Franke K, Kalucka J, Mamlouk S, et al. HIF-1a is a protective factor in 41. Hahne M, Wenger RH, Vestweber D, et al. The heat-stable antigen can conditional PHD2-deficient mice suffering from severe HIF-2a-induced alter very late antigen 4-mediated adhesion. J Exp Med. 1994;179:1391– excessive erythropoiesis. Blood. 2013;121:1436–1445. 1395. 16. Kurt B, Paliege A, Willam C, et al. Deletion of von Hippel-Lindau protein 42. Furness SGB, McNagny K. Beyond mere markers. Immunol Res. 2006;34: converts renin-producing cells into erythropoietin-producing cells. JAm 13–32. Soc Nephrol. 2013;24:433–444. 43. Cretoiu SM, Popescu LM. Telocytes revisited. Biomol Concepts. 2014;5: 17. Kurt B, Gerl K, Karger C, et al. Chronic hypoxia-inducible transcription 353–369. factor-2 activation stably transforms juxtaglomerular renin cells into 44. Qi G, Lin M, Xu M, et al. Telocytes in the human kidney cortex. J Cell Mol fibroblast-like cells in vivo. J Am Soc Nephrol. 2015;26:587–596. Med. 2012;16:3116–3122. 18. Gerl K, Miquerol L, Todorov VT, et al. Inducible glomerular erythropoietin 45. Li L, Lin M, Li L, et al. Renal telocytes contribute to the repair of production in the adult kidney. Kidney Int. 2015;88:1345–1355. ischemically injured renal tubules. J Cell Mol Med. 2014;18:1144–1156. þ 19. Gerl K, Nolan KA, Karger C, et al. Erythropoietin production by PDGFR-b 46. Rusu MC, Mogoanta L, Pop F, et al. Molecular phenotypes of the human cells. Eur J Physiol. 2016;468:1479–1487. kidney: Myoid stromal cells/telocytes and myoepithelial cells. Ann Anat. 20. Kobayashi H, Liu Q, Binns TC, et al. Distinct subpopulations of FOXD1 2018;218:95–104. stroma-derived cells regulate renal erythropoietin. J Clin Invest. 2016;126: 47. Bei Y, Zhou Q, Fu S, et al. Cardiac telocytes and fibroblasts in primary 1926–1938. culture: different morphologies and immunophenotypes. PLoS One. 21. Gerl K, Steppan D, Fuchs M, et al. Activation of Hypoxia Signaling in 2015;10:e0115991. Stromal Progenitors Impairs Kidney Development. Am J Pathol. 2017;187: 48. Zheng Y, Zhang M, Qian M, et al. Genetic comparison of mouse lung 1496–1511. telocytes with mesenchymal stem cells and fibroblasts. J Cell Mol Med. 22. Bachmann S, Le Hir M, Eckardt KU. Co-localization of erythropoietin 2013;17:567–577. mRNA and ecto-5’-nucleotidase immunoreactivity in peritubular cells of 49. Vignais ML, Caicedo A, Brondello JM, et al. Cell connections by tunneling rat renal cortex indicates that fibroblasts produce erythropoietin. nanotubes: effects of mitochondrial trafficking on target cell metabolism, J Histochem Cytochem. 1993;41:335–341. homeostasis, and response to therapy. Stem Cells Int. 2017;2017:6917941. 23. Matsumoto I, Ohmoto M, Narukawa M, et al. Skn-1a (Pou2f3) specifies 50. Koury ST, Koury MJ, Bondurant MC, et al. Quantitation of erythropoietin- taste receptor cell lineage. Nat Neurosci. 2011;14:685–687. producing cells in kidneys of mice by in situ hybridization: correlation 24. Safran M, Kim WY, O’Connell F, et al. Mouse model for noninvasive with hematocrit, renal erythropoietin mRNA, and serum erythropoietin imaging of HIF prolyl hydroxylase activity: assessment of an oral agent concentration. Blood. 1989;74:645–651. that stimulates erythropoietin production. Proc Natl Acad Sci USA. 51. Eckardt KU, Koury ST, Tan CC, et al. Distribution of erythropoietin 2006;103:105–110. producing cells in rat kidneys during hypoxic hypoxia. Kidney Int. 25. Guo JK, Shi H, Koraishy F, et al. The Terminator mouse is a diphtheria 1993;43:815–823. toxin-receptor knock-in mouse strain for rapid and efficient enrichment 52. Maxwell PH, Ferguson DJ, Nicholls LG, et al. Sites of erythropoietin of desired cell lineages. Kidney Int. 2013;84:1041–1046. production. Kidney Int. 1997;51:393–401. 26. Eckardt KU, Dittmer J, Neumann R, et al. Decline of erythropoietin 53. Shanks JH, Hill CM, Lappin TR, et al. Localization of erythropoietin gene formation at continuous hypoxia is not due to feedback inhibition. Am J expression in proximal renal tubular cells detected by digoxigenin- Physiol. 1990;258:F1432–F1437. labelled oligonucleotide probes. J Pathol. 1996;179:283–287. 27. Tan CC, Eckardt KU, Firth JD, et al. Feedback modulation of renal and 54. Suliman HB, Ali M, Piantadosi CA. Superoxide dismutase-3 promotes hepatic erythropoietin mRNA in response to graded anemia and full expression of the EPO response to hypoxia. Blood. 2004;104:43– hypoxia. Am J Physiol. 1992;263:F474–F481. 50. 28. Stiehl DP, Bordoli MR, Abreu-Rodríguez I, et al. Non-canonical HIF-2a 55. Beirao I, Moreira L, Barandela T, et al. Erythropoietin production by distal function drives autonomous breast cancer cell growth via an AREG- nephron in normal and familial amyloidotic adult human kidneys. Clin EGFR/ErbB4 autocrine loop. Oncogene. 2012;31:2283–2297. Nephrol. 2010;74:327–335.
386 Kidney International (2019) 95, 375–387 F Imeri et al.: Epo kinetics in renal telocyte-like cell lines basic research
56. Nagai T, Yasuoka Y, Izumi Y, et al. Reevaluation of erythropoietin 64. Souma T, Suzuki N, Yamamoto M. Renal erythropoietin-producing cells in production by the nephron. Biochem Biophys Res Commun. 2014;449: health and disease. Front Physiol. 2015;6:167. 222–228. 65. Suzuki N, Yamamoto M. Roles of renal erythropoietin-producing (REP) 57. Loya F, Yang Y, Lin H, et al. Transgenic mice carrying the cells in the maintenance of systemic oxygen homeostasis. Eur J Physiol. erythropoietin gene promoter linked to lacZ express the reporter in 2016;468:3–12. proximal convoluted tubule cells after hypoxia. Blood. 1994;84:1831– 66. Haase VH. Therapeutic targeting of the HIF oxygen-sensing pathway: 1836. lessons learned from clinical studies. Exp Cell Res. 2017;356:160–165. 58. Haidar MA, Loya F, Yang Y, et al. Electron microscopic localization of lacZ 67. Humphreys BD, Lin SL, Kobayashi A, et al. Fate tracing reveals the expression in the proximal convoluted tubular cells of the kidney in pericyte and not epithelial origin of myofibroblasts in kidney fibrosis. Am transgenic mice carrying chimeric erythropoietin/lacZ gene constructs. J Pathol. 2010;176:85–97. J Struct Biol. 1997;118:220–225. 68. Marti HH, Wenger RH, Rivas LA, et al. Erythropoietin gene expression in 59. Farsijani NM, Liu Q, Kobayashi H, et al. Renal epithelium regulates human, monkey and murine brain. Eur J Neurosci. 1996;8:666–676. erythropoiesis via HIF-dependent suppression of erythropoietin. J Clin 69. Bernaudin M, Bellail A, Marti HH, et al. Neurons and astrocytes express Invest. 2016;126:1425–1437. EPO mRNA: oxygen-sensing mechanisms that involve the redox-state of 60. Semenza GL, Koury ST, Nejfelt MK, et al. Cell-type-specific and hypoxia- the brain. Glia. 2000;30:271–278. inducible expression of the human erythropoietin gene in transgenic 70. Johansson T, Broll I, Frenz T, et al. Building a zoo of mice for genetic mice. Proc Natl Acad Sci USA. 1991;88:8725–8729. analyses: a comprehensive protocol for the rapid generation of BAC 61. Nolan KA, Wenger RH. Source and microenvironmental regulation of transgenic mice. Genesis. 2010;48:264–280. erythropoietin in the kidney. Curr Opin Nephrol Hypertens. 2018;27:277–282. 71. Nytko KJ, Maeda N, Schläfli P, et al. Vitamin C is dispensable for oxygen 62. Yamazaki S, Souma T, Hirano I, et al. A mouse model of adult-onset sensing in vivo. Blood. 2011;117:5485–5493. anaemia due to erythropoietin deficiency. Nat Commun. 2013;4:1950. 72. Schörg A, Santambrogio S, Platt JL, et al. Destruction of a distal hypoxia 63. Olmos G, Muñoz-Félix JM, Mora I, et al. Impaired erythropoietin synthesis response element abolishes trans-activation of the PAG1 gene mediated in chronic kidney disease is caused by alterations in extracellular matrix by HIF-independent chromatin looping. Nucl Acids Res. 2015;43:5810– composition. J Cell Mol Med. 2018;22:302–314. 5823.
Kidney International (2019) 95, 375–387 387 SUPPLEMENTARY METHODS
Reagents Tamoxifen (T5648), 4-hydroxytamoxifen (H6278), diphtheria toxin, puromycin and 5- azacytidine were purchased from Sigma-Aldrich (St Louis, MO, USA). For in vivo applications, tamoxifen at 50 mg/ml in 10% ethanol and 90% sunflower oil was ultrasonicated (Bioruptor Plus with built-in cooling system; Diagenode, Seraing, Belgium) using 5 to 10 cycles (30 seconds on/off) until in solution. If not noted otherwise, high-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS, 50 IU/ml penicillin and 100 µg/ml streptomycin (Sigma-Aldrich) was used for cell culture.
Construction of the Epo-CreERT2 BAC vector The bacterial artificial chromosome (BAC) clone RP24-338L14 was obtained from the BACPAC Resources Center (Oakland, CA, USA). The CreERT2 cDNA sequence, combined with a neomycin/kanamycin resistance cassette flanked by FRT sites, was amplified by PCR using the BBV14_pNNCre19-E vector as template. The PCR primers (Table S3) contained specific 50 nt homology arms targeting the 2nd exon and 4th intron of the mouse Epo gene, respectively, like previously reported for another BAC clone.s1 Following transformation of E.coli containing the L-arabinose- inducible recombinase-carrying pSC101-BAD-gbaA-tetra plasmid, the CreERT2 cDNA was inserted into the Epo BAC clone RP24-338L14 by homologous recombination. The Neo cassette in the targeted BAC was excised using the FLP-FRT system by transforming electrocompetent bacteria with the FLP-expressing plasmid 706-FLP (all plasmids and bacteria were kindly provided by T. Buch, Zurich, Switzerland).
In vitro evaluation of the Epo-CreERT2 BAC vector The functionality of the transgenic vector was evaluated by co-transfection of 1.5 µg Epo-CreERT2 BAC together with 0.5 µg p133_pSV-STOP-luc and 10 ng pRLSV40- Renilla luciferase control plasmid (Promega, Madison, WI, USA) into Hep3B cells using Targefect-BAC (Targeting Systems, El Cajon, CA, USA). 1.5 hours after transfection, 450 nM 4-hydroxytamoxifen was added and the cells were exposed to normoxic or hypoxic (0.2% O2) conditions for 20 hours in a hypoxia workstation (Invivo2 400, Baker Ruskinn, Bridgend, UK). Dual luciferase reporter gene assays were performed in triplicates as described previously.s2
Genotyping and mouse whole genome sequencing In total, eight independent founder lines of Epo-CreERT2 BAC transgenic mice (B6D2;C57BL6N-Tg (EPO::Cre)nRhw) were generated and two of them were further analysed by whole genome sequencing. Mouse genomic DNA was isolated from ear punches and amplified by PCR using an Express Extraction and HotStart PCR kit (Kapa Biosystems, London, UK) and the primers provided in Table S3. The integration site of the Epo-CreERT2 BAC vector was determined by whole genome DNAseq of kidney DNA derived from Epo-CreERT2 strains #1 and #241. Sequence reads were aligned to the reference genome (mouse reference genome GRCm38) using bwa (version 0.7.12-r1039).s3 Alignment results were processed using Delly (version 0.7.6) and LUMPY (version 0.2.13) to identify structural variations.s4,s5 Translocation events that involve the chr5 region corresponding to the BAC were manually inspected using IGV to identify the integration sites.s6,s7 Genomic coordinates of breakpoints within the BAC were back-translated using chr5 coordinates based on BLAST alignments. Correctness of the back translation was 1 confirmed by aligning the fusion consensus sequences around the breakpoints to the BAC sequences.
Analysis of genetically modified mice Animals were housed by the laboratory animal service center (LASC) and animal experiments were conducted at the Zurich integrative rodent physiology (ZIRP) facility. To functionally analyse Epo-CreERT2 founder lines, they were crossed with Ai14 (B6.Cg-Gt(ROSA)26Sor
In vivo imaging of O2 bioavailability ODD-Luc mices9 were obtained from Jackson Laboratory (FVB.129S6- Gt(ROSA)26Sor
Tissue processing and fluorescence microscopy Mice were killed by cervical dislocation, the kidneys excised, cut in half and fixed with 4% paraformaldehyde. Cryosections (12-20 µm) were prepared and blocked with 5% goat serum/0.3% Triton X-100/PBS solution. Primary antibodies were applied in 1% BSA/0.3% Triton X-100/PBS at 4°C overnight. Following three washes in PBS, secondary antibodies were applied for 1 hour at room temperature. Sections were counterstained with DAPI (0.5 µg/ml) and mounted in Mowiol/DABCO (Millipore, Darmstadt, Germany, and Sigma-Aldrich, respectively) medium. Cultured cells were analyzed by immunofluorescence after fixation with 4% PFA. Fluorescent signals were recorded using an Axio Scan.Z1 microscope (Zeiss, Feldbach, Switzerland) or a SP8 upright confocal laser scanning microscope (Leica, Heerbrugg, Switzerland) or a slide scanner (Axio Scan.Z1; Zeiss). The following antibodies were used: rabbit anti- RFP polyclonal antibody (600-401-379; Rockland, Limerick, PA, USA), rat monoclonal anti-CD73 (550738; BD Biosciences, San Jose, CA, USA), mouse monoclonal anti-TOM20 (sc-17764; Santa Cruz Biotechnology, Dallas, TX, USA), rabbit polyclonal anti-TOM20 (sc-11415; Santa Cruz Biotechnology), rabbit polyclonal anti-α-tubulin (2144; Cell Signalling Technology, Leiden, The Netherlands), mouse monoclonal anti-β-actin (A5441; Sigma-Aldrich), rat monoclonal anti-CD34 (ab8158; Abcam) and rabbit monoclonal anti-PDGFRβ (ab32570; Abcam, Cambridge, UK). Secondary goat antibodies were coupled to Alexa488, Alexa568, Alexa635 or
2
Alexa647 (Thermo Fisher Scientific). F-actin was stained using Alexa488-coupled phalloidin (A12379; Thermo Fisher Scientific).
In situ hybridization Epo mRNA was detected by in situ hybridization according to the manufacturer’s instructions (RNAscope; Advanced Cell Diagnostics, Hayward, CA, USA) as reported previously.s10 Briefly, the Epo target probe (Cat No. 315501) consists of 12 double Z probe pairs and targets the region between 39 and 685 of mouse Epo mRNA. Mouse Ppib (Cat No. 31391) and DapB (Cat No. 310043) were used as positive and negative control probes, respectively. Tissues were fixed and prepared according to the manufacturer’s recommendations, and embedded in OCT or paraffin. Cryosections (12 µm) or paraffin sections (5 µm) were pre-treated according to the manufacturer’s recommendations. The probes were hybridized for 2 hours at 40°C in a HybEZ oven (Advanced Cell Diagnostics), followed by a series of signal amplification and washing steps. For horseradish peroxidase labelled hybridization probes, signals were detected with 3,3'-diaminobenzidine using the 2.5 HD reagent kit BROWN (Advanced Cell Diagnostics). Tissue sections were counterstained with 50% Gill’s hematoxylin, dehydrated by baking at 60°C for 30 minutes, and mounted with the xylene based mounting medium Eukitt (Sigma-Aldrich).
Generation and manipulation of REP cell-derived cell lines Epo-CreERT2#1 x tdTomato reporter mice were crossed with B6.129- Gt(ROSA)26Sor
Bisulfite sequencing Bisulfite conversion of genomic DNA was performed using the EpiTect bisulfite kit (Qiagen, Hilden, Germany). The Epo promoter region was amplified by bisulfite specific PCR. Bisulfite compatible primers (Table S3) were designed using Methyl Primer Express software (Thermo Fisher Scientific). PCR products were purified (NucleoSpin PCR clean-up kit; Macherey-Nagel, Düren, Germany) and sequenced (Microsynth, Balgach, Switzerland). For the quantification of CpG methylation, the
3 signal peak height of cytosine was divided by the sum of signal peak heights of cytosine and tyrosine as described previously.s11
RNA sequencing and bioinformatical analysis The quantity and quality of the isolated RNA was determined with a Qubit fluorometer (Life Technologies, Carlsbad, CA, USA) and a Bioanalyser 2100 (Agilent), respectively. The TruSeq Stranded mRNA Sample Prep Kit (Illumina, San Diego, CA, USA) was used in the subsequent steps. Briefly, total RNA samples (100 ng) were poly-A selected and reverse-transcribed into double-stranded cDNA with actinomycin added during first-strand synthesis. The cDNA samples were fragmented, end- repaired and adenylated before ligation of TruSeq adapters. The adapters contain the index for multiplexing. Fragments with TruSeq adapters on both ends were selectively enriched with PCR. The quantity and quality of the enriched libraries was validated as above (average fragment size of approx. 360 bp). The libraries were normalized to 10 nM in 10 mM Tris-HCl, pH 8.5, containing 0.1% Tween 20. The TruSeq SR Cluster Kit v4-cBot-HS (Illumina) was used for cluster generation using 8 pM of pooled normalized libraries on the cBOT. Sequencing was performed on the HiSeq 4000 single end 125 bp using the TruSeq SBS Kit v4-HS (Illumina). Bioinformatic data analysis was performed using SUSHI.s12 In detail, the raw reads were quality checked using Fastqc and FastQ Screen (www.bioinformatics.babraham.ac.uk). Quality controlled reads (adaptor trimmed, first 5 and last 6 bases hard trimmed, minimum average quality Q10, minimum tail quality Q10, minimum read length 20 nt) were aligned to the reference genome (Ensembl GRCm38, not patched) using STAR aligner,s13 with at least 30 bp matching and max. 10% mismatches. Alignments to more than 50 different genomic loci were ignored. Transcript reads were normalized using the trimmed mean of M-values (TMM) method.s14 Expression counts and differential expression was computed using featureCounts in the Bioconductor package Subread and the DESeq2 package, respectively.s15,s16 Gene ontology pathway enrichment analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) based on the DAVID 6.8 bioinformatics database.s17,s18 P-values were corrected using the Benjamini- Hochberg correction for multiple testing.
Immunoblotting The following antibodies were used for immunoblotting: rabbit polyclonal anti-Epo (ab65394; Abcam; discontinued. Figures 3C, 3E, 4A, 4B and 4E); mouse monoclonal anti-Epo (7D10, sc-80995; Santa Cruz Biotechnology. Figures 3G and S3A); mouse monoclonal anti-Epo (B4, sc-5290; Santa Cruz Biotechnology. Figure S6A); rabbit polyclonal anti-Epo (STJ27630; St John's Laboratory, London, UK. Figures 4F, S3B, S6A-C); rabbit recombinant monoclonal anti-HIF-2α (BL-95-1A2; Bethyl Laboratories, Montgomery, TX, USA); mouse monoclonal anti-HIF-1α (clone 54; BD Transduction Laboratories, San Jose, CA, USA); mouse monoclonal anti-β-actin (A5441; Sigma- Aldrich). Secondary antibodies were HRP-conjugated goat anti-rabbit or anti-mouse IgG (#31460 and #31430, respectively; Thermo Fisher Scientific). A pre-stained 10 - 250 kDa marker was purchased from Thermo Fisher Scientific. Testing of antibodies against Epo and HIF-2α using FAIK3-5 knock-down cells is shown in Figures 3G and S6
4
mouse Epo gene BAC clone 101 kb 119 kb (RP24-338L14)
PCR product ERT2 R (BBV14_pNNCre19-E) Cre Neo homologous recombination
recombination ERT2 R intermediary product Cre Neo Flp recombination ERT2 Epo-Cre ERT2 transgenic vector Cre hypoxia tamoxifen conditional gene activation CreERT2
tdTomato reporter CAGpr (Ai14 mouse strain) Stop tdTomato WPRE Rosa26 Cre recombination locus permanent tdTomato expression CAGpr tdTomato WPRE
DTR selection/GFP reporter Rosa26 SA DTR pA PGK-neo pA IRES-GFP pA (Terminator mouse strain ) Cre recombination loss of DTR expression permanent GFP expression Rosa26 SA IRES-GFP pA
Figure S1. Generation of a mouse line to conditionally target renal Epo-producing (REP) cells and to select against non-REP cells. The Epo-CreERT2 BAC transgenic vector was generated by replacing mouse Epo exons 2 to 4 with a CreERT2 cassette as indicated. Conditional Cre expression by hypoxia and Cre activation by tamoxifen was analysed in vivo following crossing of Epo-CreERT2 transgenic mice into the tdTomato reporter mouse strain. The crossing of Epo-CreERT2 transgenic mice into the Terminator mouse strain allowed for the permanent activation of the tdTomato and GFP reporter alleles and the inactivation of the diphtheria toxin receptor (DTR) allele following conditional induction of the Cre recombinase by simultaneous hypoxia and tamoxifen treatment. CAGpr, tripartite CAG promoter; WPRE, Woodchuck hepatitis virus posttranscriptional regulatory element; SA, splice acceptor; PGK, phosphoglycerate kinase 1; pA, polyadenylation signal; IRES, internal ribosome entry site. A Founder strain Epo-CreERT2#1
chr14:56,337,390 EpoCreERT2 BAC vector (chr5) chr14:56,337,060
ERT2 non-coding Gnb2 Gigyf1 Pop7 EpoCre Zan PCR (519 bp) non-coding
ATG TGA
45.9 kb 98.2 kb
Founder strain Epo-CreERT2#241
ERT2 chr9:43,159,262 EpoCre BAC vector (chr5) chr9:43,211,463 PCR (174 bp) Actl6b Pop7 EpoCreERT2 Zan Pou2f3 Gnb2 Gigyf1 non-coding
intron 3 ATG TGA
75.2 kb 52.0 kb
B Epo-CreERT2#1 Epo-CreERT2#241 wt/wt tg/tg wt/tg wt/tg wt/wt tg/tg tg -519 bp tg -174 bp wt -255 bp wt - 92 bp
Figure S2. Genotyping of Epo-CreERT2 transgenic mouse lines. (A) Scheme depicting the genomic integration sites of the Epo-CreERT2 BAC vector (blue) derived from chromosome (chr) 5 into chromosomes 14 and 9 (red) of founder strains #1 and #241, respectively, obtained by whole genome DNAseq. The remaining 5' and 3' flanking regions upstream of the ATG start codon and downstream of the TGA stop codon, respectively, of the Epo gene are indicated in green. Nucleotide numbers are given according to the mouse reference genome GRCm38. (B) Agarose gel electrophoresis of PCR products to confirm the sequence boundaries of the integrated vectors predicted by whole genome DNAseq. The corresponding PCR primers are indicated in (A). A FAIK3-5 (0.2% O2 non-pre-equilibrated) passage 10 passage 15 passage 21 hours 0 12 24 0 12 24 0 12 24 HIF-2a
Epo 7D10 sc-80995 b-actin
B FAIK3-5 (0.2% O2 non-pre-equilibrated) passage 19 passage 30 hours 0 3 6 12 24 0 3 6 12 24
HIF-1a
HIF-2a
Epo STJ27630 b-actin
C FAIK3-5 (0.2% O2 non-pre-equilibrated) passage 30 hours 0 1.5 3 6 12 24 48 HIF-2a
b-actin
Epo 7D10 sc-80995 b-actin
Figure S3. Persistent hypoxia-inducible Epo expression in long-term cultured FAIK3- 5 cells. Examples of immunoblot detection of HIFa and Epo protein following exposure to hypoxia (0.2% O2) for the indicated time periods using non-pre- equilibrated media. FAIK3-5 cells were passaged twice weekly and used after passage 10, 15 and 21 (A) or 19 and 30 (B, C). Two different anti-Epo antibodies were used as indicated. b-Actin served as loading and blotting control. A 16413 of 43068 mRNAs and lncRNAs B N 1.000 N AIK3-5
N F hypoxia N
AIK3-5 0.975
F N AIK1-10 F
normoxia N H 0.950 H hypoxia AIK1-10
H F AIK2-5 F 0.925 H
normoxia H AIK3-5
H F 0.900 H hypoxia H AIK2-5 AIK1-10 F F H 0.875 normoxia N normoxia hypoxia normoxia hypoxia normoxia hypoxia N
FAIK1-10 FAIK2-5 FAIK3-5 AIK2-5 N F
0.05 0.04 0.03 0.02 0.01 0.00 height
C expressed regulated induced repressed FAIK1-10 FAIK2-5 FAIK1-10 FAIK2-5 FAIK1-10 FAIK2-5 FAIK1-10 FAIK2-5
465 285 393 268 242 720 146 172 487 116 66 208
13876 559 368 154 433 255 278 223 98 78 131 105
650 1518 395 964 mRNAs and lncRNAs FAIK3-5 FAIK3-5 FAIK3-5 FAIK3-5
D FAIK1-10 FAIK2-5 FAIK3-5 FAIK1-10 FAIK2-5 FAIK3-5 1500
400 increased increased 1000 200 500 0
0 decreased decreased 200 500
1000 400 number of regulated mRNAs 1500 number of regulated lncRNAs 600
Figure S4. RNAseq of FAIK cell lines. (A) RNA isolated from three different FAIK cell lines following exposure to hypoxia (0.2% O2) for 24 hours was subjected to RNAseq and read numbers were determined for each RNA species (43,068 RNAs annotated; 16,413 mRNAs/lncRNAs expressed; threshold: 10 reads). Shown are Spearman correlation coefficients of mRNA and lncRNA read numbers between the three FAIK cell lines. (B) Dendrogram of an unsupervised cluster analysis of the mRNAs and lncRNAs expressed in the three FAIK cell lines. (C) Venn diagrams showing the overlap of all mRNAs and lncRNAs expressed (threshold: 10 reads); regulated, induced or repressed (threshold: log2 ratio >1, p<0.001) in the three FAIK cell lines indicated (n=3). (D) Number of mRNAs (upper panel) and lncRNAs (lower panel) hypoxically regulated in the three FAIK cell lines indicated by a log2 ratio of >1. FAIK1-10 FAIK2-5 FAIK3-5 Induction Gene Normoxia Hypoxia Induction Gene Normoxia Hypoxia Induction Gene Normoxia Hypoxia 84.39 Gipr 110.51 Gpr35 235.73 Aire 83.17 2610528A11Rik 105.42 Gipr 113.69 2610528A11Rik 47.21 Cdh17 104.26 2610528A11Rik 51.91 Evpl 36.18 Fibcd1 92.6 Aire 47.34 Fam159b 35.36 Aire 73.21 Frem2 44.29 Gipr 31.96 Sprr1a 50.04 Kbtbd11 40.53 4930538K18Rik 29.2 Trpm2 47.05 Stc1 40.06 Ankk1 26.89 Ndrg1 42.49 Fgf11 36.89 Selenbp1 26.08 Gpr35 41.79 Fibcd1 35.83 Gpr35 23.7 Acap1 39.1 Rnf152 26.19 Ccno 22.83 Dlx3 35.09 Muc2 23.9 Acer2 22.55 Tph1 34.9 AU021092 23.88 Tnfrsf18 21.24 Frem2 34.58 Vwa1 23.52 Acap1 21.21 4930538K18Rik 34.15 Tph1 23.04 Exoc3l2 19.24 Vwa1 33.78 4930538K18Rik 22.09 Dlx3 18.23 Bend5 31.02 Ndrg1 20.97 Ndrg1 18.11 Sh3gl3 29.77 Atp7b 20.27 Lama3 17.86 1600014C23Rik 29.1 Lppr3 19.75 Pvrl4 17.38 Stc1 28.23 Fam19a2 19.13 Prss27 16.9 Lppr3 27.47 Selenbp1 18.34 Prr15l 16.53 Phospho1 26.98 Trpm2 18.28 Frem2 16.22 Hhatl 26.37 Slc38a3 18.11 Tmem74b 15.35 Smtnl2 25.51 Ppp1r3c 17.35 Snai3 14.69 Tenm4 24.49 Prtn3 16.92 Casq2 14.46 Aox4 24 Cbln3 16.78 Cnga3 14.05 Mgarp 21.86 Mgarp 16.74 Atf3 13.81 Kbtbd11 21.5 Atg9b 15.65 Lman1l 13.31 Ppp1r3c 20.82 Lama3 15.62 Apln 13.14 Gm853 19.84 Epm2a 15.4 Muc2 12.68 Vldlr 19.31 Kcnb1 15.07 Fgf18 12.54 Ciart 19.28 Il10 14.82 Pga5 12.35 Ero1l 18.86 Ush2a 14.24 Cbln3 12.27 Wt1 17.51 Vldlr 13.51 Ankrd65 11.62 Tmem74b 17.5 Ero1l 13.48 Egln3 11.35 Prtn3 17.5 Celsr3 13.33 Tbx2 10.88 Rnf152 17.18 Kctd16 13.18 Fgf11 10.7 Rapsn 16.76 Nrxn3 12.94 Stc1 10.69 Apln 15.92 Smtnl2 12.9 Ciart 10.52 Muc2 15.37 Reep1 12.85 Acox2 10.18 Cbln3 15.3 Dlx3 12.7 Zfp750 9.43 Egln3 15.1 Cdh17 12.56 Ero1l 8.97 Pgbd1 14.84 Gm853 12.41 Sytl1 8.94 Ankrd37 14.82 Rgcc 12.19 Kctd11 8.75 Ccdc13 14.6 Myo18b 12.16 Slc29a4 8.49 Pfkp 14.47 Acer2 12.15 Olfr456 8.34 Galr2 13.83 Ndufa4l2 12.12 Fosb 8.14 Epm2a 13.63 Ankrd37 11.72 Sprr2h 7.99 Celsr3 13.24 Wt1 10.99 Cdh17 7.97 Kctd11 12.92 Ascl2 10.87 Ankrd37 7.88 Ppp1r3b 12.79 Egln3 10.7 Vwa1 7.71 H2 Q7 12.53 Gls2 10.57 Atg9b 7.65 Fam65b 12.47 Aif1 10.48 Adrb2 7.59 Proser2 12.18 Phospho1 10.2 Pinlyp 7.56 Prss27 12.13 Ppp1r3b 10.04 Cdcp1 7.56 Sprr2h 12.05 Cacna1h 10.04 Smtnl2 7.32 Aldoc 12.02 Car12 9.88 1600014C23Rik 7.23 Mt1 11.94 Tmem74b 9.75 Hhatl 7.22 Adm2 11.8 Apln 9.74 Il17f 7.12 Cdhr1 11.75 Dpf1 9.53 Phospho1 7.09 Gls2 11.66 Galr2 9.23 Rab44 7.01 Atg9b 11.54 Rasgrf1 9.15 Jag2 6.88 Gck 10.94 4732471J01Rik 9.14 Sdr42e1 6.85 Trib3 10.86 Tchh 9.1 Wnt9a 6.71 Higd1a 10.85 Dgkg 9.04 Otog 6.68 Amz1 10.76 Ciart 8.99 Egln1 6.63 Egln1 10.66 Gbe1 8.98 Rapsn 6.59 Tmem45a 10.43 Higd1a 8.92 1700007K13Rik 6.53 Bsg 10.13 Fam65b 8.74 Gapdhs 6.52 Car12 10.11 Aox4 8.71 Vmn1r44 6.46 Ache 10.05 Ermap 8.65 Ppp1r3b 6.36 Gbe1 10.04 Hilpda 8.65 B3gnt8 6.24 Pgm2 9.89 Cr2 8.59 Mgarp 6.23 Fam19a2 9.8 Rgs11 8.57 Klf2 6.22 Reep1 9.74 Hs3st6 8.52 Dusp2 6.21 Il10 9.71 Megf6 8.5 Aldh1a3 6.19 Il17f 9.61 Ache 8.44 Sbp 6.18 Fkbp1b 9.42 Acap1 8.39 Nppb 6.18 Isg20 9.41 Pfkp 8.33 Ccdc153 6.16 Pgf 9.18 Gper1 8.28 Soat2 6.16 Olfr456 9.01 Tenm4 8.23 Fam222a 6.11 Shc4 8.93 Arrb1 7.97 Lppr3 6.1 Trappc6a 8.92 Egln1 7.97 Fibcd1 6.09 Slc2a1 8.76 Lelp1 7.9 Pmaip1 6.02 Fgf11 8.44 Mpp2 7.85 Itga2 6 Syt17 8.33 Fam184b 7.84 Aknad1 5.91 P4ha2 8.29 Mfsd9 7.78 Ak4 5.86 Hpse 8.26 Lrp2bp 7.73 Slc2a1 5.86 Unc13a 8.04 Kank3 7.72 Trpm2 5.8 Trim72 7.96 Tmem191c 7.72 Rasd1 5.79 Ackr3 7.84 P4ha2 7.56 Selenbp2 5.79 Aldh1l2 7.75 Cdhr1 7.55 Slc38a3 5.79 4732471J01Rik 7.7 Sprr1a 7.43 Cmah 5.64 Crhbp 7.63 Adrbk2 7.4 Grhl3 5.57 Atf3 7.58 Aldh1l2 7.4 Npas1 5.55 Ascl2 7.55 Fkbp1b 7.39 Fank1 5.54 Hk1 7.42 Lingo1 7.33 Vwa7 5.48 Tmem71 7.38 Syt17 7.32 Krt16 5.47 Mpp2 7.34 Bend5 7.2 Nanos1 5.46 Matk 7.32 Kctd11 7.06 Gna15 5.45 Ush2a 7.26 Tet1 7.04 Apol8
normalized reads [log2] 0 5 10 15 Figure S5. Heat map of hypoxia-inducible gene expression in FAIK cell lines. FAIK1- 10, FAIK2-5 and FAIK3-5 cells were exposed to normoxic or hypoxic (0.2% O2) conditions for 24 hours and transcript levels determined by RNAseq (n=3). Shown are normalized RNA sequence read numbers in log2 scale, ranked according to hypoxic induction factors.
Figure S5. Heat map of hypoxia-inducible gene expression in FAIK cell lines. FAIK1- 10, FAIK2-5 and FAIK3-5 cells were exposed to normoxic or hypoxic (0.2% O2) conditions for 24 hours and transcript levels determined by RNAseq (n=3). Shown are normalized RNA sequence read numbers in log2 scale, ranked according to hypoxic induction factors. A FAIK3-5 shCtrl shHIF-2a#5
0.2% O2 0 3 6 12 24 0 3 6 12 24 hours HIF-2a
Epo STJ27630 b-actin
Epo B-4 sc-5290
b-actin
B FAIK3-5 shCtrl shHIF-2a#4
0.2% O2 0 3 6 12 24 0 3 6 12 24 hours HIF-2a
Epo STJ27630 b-actin
C FAIK3-5 shCtrl shHIF-1a#1
0.2% O2 0 3 6 12 24 0 3 6 12 24 hours HIF-1a
b-actin
Epo STJ27630 b-actin
Figure S6. Testing of antibodies to detect mouse Epo and HIFa. Lentiviral shRNA expression was used to knock down HIF-2a (A, B; two different shRNA constructs) and HIF-1a (C). Epo was detected by different antibodies as indicated. Note that multiple isoforms of glycosylated Epo with different molecular weights can be detected, but shown is always the heaviest isoform which consistently runs at an apparent molecular weight of approx. 50 kDa relative to the pre-stained molecular weight marker. b-Actin served as loading and blotting control. Table S1. List of significantly (p < 0.05) enriched KEGG pathways in the three FAIK cell lines. The input list contained all genes that were at least two-fold hypoxically induced (p ≤ 0.001) as determined by RNAseq.
KEGG Term FAIK1-10 FAIK2-5 FAIK3-5 HIF-1 signaling pathway 4.78E-07 3.69E-05 7.37E-06 Glycolysis / Gluconeogenesis 2.94E-10 2.91E-07 2.23E-08 Pentose phosphate pathway 2.98E-02 n.s. n.s. Fructose and mannose metabolism 1.40E-02 4.24E-02 n.s. Galactose metabolism 1.09E-02 3.29E-02 n.s. Starch and sucrose metabolism 1.09E-02 2.09E-03 1.64E-02 Amino sugar and nucleotide sugar metabolism 1.59E-02 2.15E-02 n.s. Glycosaminoglycan degradation n.s. n.s. 4.66E-02 Metabolic pathways 3.85E-04 7.46E-04 1.80E-03 Biosynthesis of antibiotics 1.86E-05 3.55E-03 3.24E-03 Carbon metabolism 1.91E-06 5.70E-04 5.87E-04 Biosynthesis of amino acids 1.18E-03 2.82E-02 9.32E-03 MAPK signaling pathway 3.66E-02 n.s. n.s. p53 signaling pathway n.s. n.s. 2.22E-02 Phagosome 2.03E-02 n.s. n.s. PI3K-Akt signaling pathway n.s. 2.21E-02 n.s. Focal adhesion n.s. 1.31E-02 n.s. ECM-receptor interaction n.s. 2.41E-02 4.72E-02 Cell adhesion molecules (CAMs) 2.71E-02 n.s. n.s. Antigen processing and presentation 8.05E-03 n.s. n.s. Circadian rhythm n.s. 7.95E-03 n.s. Insulin resistance n.s. 9.08E-02 n.s. Type I diabetes mellitus 1.50E-02 n.s. n.s. HTLV-I infection n.s. n.s. 1.54E-02 Pathways in cancer 4.78E-02 2.69E-02 1.52E-02 Small cell lung cancer n.s. 4.05E-02 n.s. Central carbon metabolism in cancer 1.27E-03 2.80E-03 1.68E-02 Autoimmune thyroid disease 2.58E-02 n.s. n.s. Allograft rejection 9.92E-03 n.s. n.s. Graft-versus-host disease 1.98E-02 n.s. n.s. Viral myocarditis 1.53E-02 n.s. n.s.
5
Table S2. Comparison of marker gene expression in telocytes and FAIK cell lines.
Marker TCs FAIK cell lines Tissue 1-10 2-5 3-5 CD34 kidney,s19,s20 heart,s21,s22 guts23 CD117/c-Kit/SCFR kidneys19,s20 CD31/PECAM-1 kidneys19,s20 D2-40/podoplanin kidneys19 vimentin kidney,s19 livers22 αSMA/Acta2 kidneys20 SMM/Myh11 kidney,s20 placentas24 endoglin/Cd105 kidney,s20 skeletal muscles25 desmin kidney,s20 placentas24 tryptase/Tpsab1 kidneys19 collagen type IV kidney,s26 lungs27 nestin kidney,s20 placentas24 MMP3 kidneys26 MMP10 kidneys26 TIMP1 kidneys26 TIMP3 lungs27 PDGFRβ livers22 CD29/ITGB1 hearts28 Sca1 spleens29 PDGFRα heart,s30 gut,s23,s31 livers22 CD45 placentas24 calveolin-1 placentas24 CD44 placentas24 CD90 placentas24 cytokeratin 7 placentas24 NOS2 placentas24 NSE/Eno2 placentas24 S-100 placentas24 VEGF placentas24 Ki67/MKI67 skeletal muscles25 Oct4 skeletal muscles25 MME/Cd10 mammary glands32 Ctgf lungs27 tagln lungs27 Nid1 lungs27 Myh14 lungs27 calreticulin guts31 KCNN3 guts31 FOXL1 guts23
TCs, telocytes; red, high (frequency of reads > 100); pink, low; grey, below threshold (frequency of reads < 10). 6
Table S3. Primers used for PCR.
BAC vector construction PCR (underlined: Cre vector) Epo, 2nd exon 5'-cgtcccaccctgctgcttttactctccttgctactgattcctctgggcctcttaattaaccaccatgggc-3' Epo, 4th intron 5'-ctgtcaagagagaataggatgtcactggagtgtggagggctacagataagcgcgccgcgacattttgaag-3'
Genotyping PCR CreERT2 fwd 5'-cgtactgacggtgggagaat-3' (909 bp) rev 5'-gatctccaccatgccctcta-3' CreERT2 fwd 5'-gcggtctggcagtaaaaactatc-3' (102 bp) rev 5'-gtgaaacagcattgctgtcactt-3' Rosa26 (wt) fwd 5'-aagggagctgcagtggagta-3' (297 bp) rev 5'-ccgaaaatctgtgggaagtc-3' Ai14 (tdTomato) fwd 5'-ctgttcctgtacggcatgg-3' (196 bp) rev 5'-ggcattaaagcagcgtatcc-3' Terminator (DTR) fwd 5'-accatgaagctgctgccgtc-3' (630 bp) rev 5'-tcagtgggaattagtcatgc-3' EpoCreERT2#1 (519 bp) fwd 5'-aggattgggggaaagttgtgt-3' Chr14 (255 bp) fwd 5'-aggagttgacagatggtgctg-3' Chr14 rev 5'-atgaggtatggcgttcacgg-3' EpoCreERT2#241 (174 bp) rev 5'-tgtatccagccatccattca-3' Chr9 (92 bp) rev 5'-agccctaacccttgagggaa-3' Chr9 fwd 5'-ccacacccaacaagataccca-3' RT-qPCR Epo exon 4-5 fwd 5'-ggccatagaagtttggcaag-3' (302 bp) rev 5'-cctctcccgtgtacagcttc-3' Epo exon 3/4-4s10 fwd 5'-aatggaggtggaagaacagg-3' (174 bp) rev 5'-acccgaagcagtgaagtga-3' mS12 exon 3-5s33 fwd 5'-gaagctgccaaagccttaga-3' (215 bp) rev 5'-aactgcaaccaaccaccttc-3' mL28 exon 3-5 fwd 5'-gcaaaggggtcgtggtagtt-3' (196 bp) rev 5'-ttctggcttcgaaggatggc-3' Bisulfite sequencing Epo promoter fwd 5'-tgtgtgtttggaatagtttgtt-3' (226 bp) rev 5'-aacacaataaaacacaaccacc-3' Product lengths are given in parentheses (fwd, forward; rev, reverse).
7
SUPPLEMENTARY REFERENCES s1. Obara N, Suzuki N, Kim K, et al. Repression via the GATA box is essential for tissue-specific erythropoietin gene expression. Blood. 2008;111:5223-5232. s2. Schörg A, Santambrogio S, Platt JL, et al. Destruction of a distal hypoxia response element abolishes trans-activation of the PAG1 gene mediated by HIF-independent chromatin looping. Nucl Acids Res. 2015;43:5810-5823. s3. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26:589-595. s4. Rausch T, Zichner T, Schlattl A, et al. DELLY: structural variant discovery by integrated paired-end and split-read analysis. Bioinformatics. 2012;28:i333-i339. s5. Layer RM, Chiang C, Quinlan AR, et al. LUMPY: a probabilistic framework for structural variant discovery. Genome Biol. 2014;15:R84. s6. Robinson JT, Thorvaldsdottir H, Winckler W, et al. Integrative genomics viewer. Nat Biotechnol. 2011;29:24-26. s7. Thorvaldsdottir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Brief Bioinform. 2013;14:178-192. s8. Marti HH, Wenger RH, Rivas LA, et al. Erythropoietin gene expression in human, monkey and murine brain. Eur J Neurosci. 1996;8:666-676. s9. Safran M, Kim WY, O'Connell F, et al. Mouse model for noninvasive imaging of HIF prolyl hydroxylase activity: assessment of an oral agent that stimulates erythropoietin production. Proc Natl Acad Sci USA. 2006;103:105-110. s10. Gerl K, Nolan KA, Karger C, et al. Erythropoietin production by PDGFR-β+ cells. Eur J Physiol. 2016;468:1479-1487. s11. Jiang M, Zhang Y, Fei J, et al. Rapid quantification of DNA methylation by measuring relative peak heights in direct bisulfite-PCR sequencing traces. Lab Invest. 2010;90:282-290. s12. Hatakeyama M, Opitz L, Russo G, et al. SUSHI: an exquisite recipe for fully documented, reproducible and reusable NGS data analysis. BMC bioinformatics. 2016;17:228. s13. Dobin A, Davis CA, Schlesinger F, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15-21. s14. Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol. 2010;11:R25. s15. Liao Y, Smyth GK, Shi W. The Subread aligner: fast, accurate and scalable read mapping by seed-and-vote. Nucl Acids Res. 2013;41:e108. s16. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. s17. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature protocols. 2009;4:44-57. s18. Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucl Acids Res. 2009;37:1-13. s19. Qi G, Lin M, Xu M, et al. Telocytes in the human kidney cortex. J Cell Mol Med. 2012;16:3116-3122. s20. Rusu MC, Mogoanta L, Pop F, et al. Molecular phenotypes of the human kidney: Myoid stromal cells/telocytes and myoepithelial cells. Ann Anat. 2018;218:95-104.
8 s21. Rusu MC, Hostiuc S, Vrapciu AD, et al. Subsets of telocytes: Myocardial telocytes. Ann Anat. 2017;209:37-44. s22. Wang F, Song Y, Bei Y, et al. Telocytes in liver regeneration: possible roles. J Cell Mol Med. 2014;18:1720-1726. s23. Shoshkes-Carmel M, Wang YJ, Wangensteen KJ, et al. Subepithelial telocytes are an important source of Wnts that supports intestinal crypts. Nature. 2018;557:242-246. s24. Suciu L, Popescu LM, Gherghiceanu M, et al. Telocytes in human term placenta: morphology and phenotype. Cells Ttissues Organs. 2010;192:325- 339. s25. Bojin FM, Gavriliuc OI, Cristea MI, et al. Telocytes within human skeletal muscle stem cell niche. J Cell Mol Med. 2011;15:2269-2272. s26. Li L, Lin M, Li L, et al. Renal telocytes contribute to the repair of ischemically injured renal tubules. J Cell Mol Med. 2014;18:1144-1156. s27. Zheng Y, Zhang M, Qian M, et al. Genetic comparison of mouse lung telocytes with mesenchymal stem cells and fibroblasts. J Cell Mol Med. 2013;17:567-577. s28. Bei Y, Zhou Q, Fu S, et al. Cardiac telocytes and fibroblasts in primary culture: different morphologies and immunophenotypes. PLoS One. 2015;10:e0115991. s29. Chang Y, Li C, Gan L, et al. Telocytes in the Spleen. PLoS One. 2015;10:e0138851. s30. Zhou Q, Wei L, Zhong C, et al. Cardiac telocytes are double positive for CD34/PDGFR-α. J Cell Mol Med. 2015;19:2036-2042. s31. Vannucchi MG, Traini C, Manetti M, et al. Telocytes express PDGFRα in the human gastrointestinal tract. J Cell Mol Med. 2013;17:1099-1108. s32. Petre N, Rusu MC, Pop F, et al. Telocytes of the mammary gland stroma. Folia Morphol. 2016;75:224-231. s33. Müller-Edenborn K, Léger K, Glaus Garzon JF, et al. Hypoxia attenuates the proinflammatory response in colon cancer cells by regulating IκB. Oncotarget. 2015;6:20288-20301.
9