Research ArticleKidney Disease

Human pluripotent stem cell–derived erythropoietin-producing cells ameliorate renal anemia in mice

See allHide authors and affiliations

Science Translational Medicine  27 Sep 2017:
Vol. 9, Issue 409, eaaj2300
DOI: 10.1126/scitranslmed.aaj2300

Cell therapy for renal anemia

Erythropoietin dysregulation is a hallmark of renal anemia. Although recombinant erythropoietin treatment is beneficial and safe, more physiological therapies are required. Hitomi et al. have developed a differentiation protocol for erythropoietin-producing cells from human and mouse iPSCs and ESCs. These cells produced and secreted functional erythropoietin protein in a hypoxia-dependent manner. Transplantation of these cells into a mouse model ameliorated renal anemia. From the perspective of basic research, erythropoietin-producing cells may be a useful tool for investigating the molecular mechanisms of erythropoietin production and secretion. From the perspective of clinical research, these results may provide a physiological therapeutic agent for treating renal anemia.

Abstract

The production of erythropoietin (EPO) by the kidneys, a principal hormone for the hematopoietic system, is reduced in patients with chronic kidney disease (CKD), eventually resulting in severe anemia. Although recombinant human EPO treatment improves anemia in patients with CKD, returning to full red blood cell production without fluctuations does not always occur. We established a method to generate EPO-producing cells from human induced pluripotent stem cells (hiPSCs) by modifying previously reported hepatic differentiation protocols. These cells showed increased EPO expression and secretion in response to low oxygen conditions, prolyl hydroxylase domain–containing enzyme inhibitors, and insulin-like growth factor 1. The EPO protein secreted from hiPSC-derived EPO-producing (hiPSC-EPO) cells induced the erythropoietic differentiation of human umbilical cord blood progenitor cells in vitro. Furthermore, transplantation of hiPSC-EPO cells into mice with CKD induced by adenine treatment improved renal anemia. Thus, hiPSC-EPO cells may be a useful tool for clarifying the mechanisms of EPO production and may be useful as a therapeutic strategy for treating renal anemia.

INTRODUCTION

Erythropoietin (EPO), a potent regulator of erythropoiesis, is secreted by the kidney and liver. The kidney is the major site of EPO production in adults, and the liver is the primary organ of EPO production in the fetal and early neonatal periods, but EPO is produced by the adult liver in states of severe anemia (1, 2). EPO production is reduced in patients with chronic kidney disease (CKD), eventually resulting in renal anemia. Treatment with recombinant human EPO (rhEPO) improves the control of renal anemia compared to conventional therapies, such as blood transfusions, which have been associated with viral infections and iatrogenic hemosiderosis. However, although rhEPO is relatively safe, the requirement for infusions one to three times a week makes it difficult to return to physiological control of red blood cell production (3). In addition, the total cost of rhEPO is increasing due to the increasing number of CKD patients, and patients with anemia secondary to chronic diseases may not respond well to rhEPO (4). Finally, very rare cases of severe anemia caused by germline mutations in EPO (5) or by anti-rhEPO autoantibodies after the administration of rhEPO have also been reported (6, 7). Therefore, new physiological therapies for renal anemia are required.

The differentiation of induced pluripotent stem cells (iPSCs) (8, 9) and embryonic stem cells (ESCs) (10), which have unlimited self-renewal capability and the potential to differentiate into any cell type in the body, provides promising cell sources for regenerative medicine. The somatic cell types differentiated from these stem cells have the potential for clinical applications and practical use, including cell therapy, drug screening, toxicology, and disease modeling. Although vigorous efforts have been made to generate multiple somatic cell types from these stem cells, the directed differentiation of EPO-producing cells (EPO cells) from iPSCs or ESCs has not yet been achieved. EPO cells were originally identified in the mouse kidney (11), but the development of culture methods for these cells has been proven difficult. For this reason, we considered cultured EPO cells derived from iPSCs or ESCs for both basic and clinical applications. One basic research application of iPSC/ESC-derived EPO cells is to clarify the mechanisms of EPO production and secretion, given that EPO cells isolated from the kidney or liver are not readily available. Clinical applications of iPSC- or ESC-derived EPO cells could include transplanting these cells as a cell therapy for treating renal anemia.

Here, we aimed to establish a differentiation method for producing EPO cells from human iPSCs (hiPSCs) and human ESCs (hESCs). We modified previously reported differentiation protocols for hepatic lineages and generated cells expressing EPO from mouse iPSCs and ESCs (miPSCs/ESCs) and hiPSCs/ESCs. These cells could physiologically regulate EPO production, as evidenced by increased EPO expression in response to low oxygen conditions. EPO protein secreted by hiPSC-derived EPO (hiPSC-EPO) cells induced the differentiation of human umbilical cord blood progenitor cells into erythropoietic cells in vitro. Furthermore, we confirmed that the transplantation of hiPSC-EPO cells into a mouse model of CKD improved renal anemia.

RESULTS

EPO-producing cells can be generated from hiPSCs/ESCs

Given that the liver is the primary organ of EPO production in the fetal stage, we examined EPO mRNA expression in the human fetal liver. In particular, we measured EPO mRNA expression in various human tissues including human fetal liver and adult kidney (Fig. 1A). To confirm EPO expression during normal development, we examined EPO expression in the embryonic day 12.5 (E12.5) mouse fetal liver by immunostaining (Fig. 1B). The results showed that almost all hepatoblasts were positively stained for both α-fetoprotein (AFP) and EPO protein, suggesting that most or all hepatocytes were derived from EPO-producing hepatoblasts.

Fig. 1. Differentiation of hiPSCs/ESCs into EPO-producing cells.

(A) The expression of EPO mRNA was examined in human fetal and adult liver tissue, human fetal and adult kidney, and human adult brain tissue. Bacterial artificial chromosome (BAC) containing EPO complementary DNA (cDNA) was used as a positive control. (B) EPO protein expression was evaluated using immunocytochemistry. Fetal mouse liver (E12.5) was positively stained with anti-EPO (green) and anti-AFP (red) antibodies. (C) Expression of marker genes for endoderm and liver lineages and EPO in cultures of differentiated hiPSCs. Human liver specimens were used as a positive control. hiPSCs were differentiated into the hepatic lineage (stages 1 to 3), as shown in fig. S1. (D and E) Time-course analyses of the expression of EPO mRNA (D) and protein (E). The expression of EPO mRNA was measured using qRT-PCR in (D), and Western blotting was performed with anti-EPO antibodies in (E). Each value was normalized to that of the sample on day 0. (F) EPO protein expression was evaluated using immunocytochemistry. hiPSC-EPO cells were positively stained with anti-EPO (green) and anti-AFP (red) antibodies. (G) Secretory vesicles (black arrows) were observed in the cytoplasm of hiPSC-EPO cells by transmission electron microscopy. The right lower panel is an image of E11.5 fetal mouse liver for comparison. (H) A temporal analysis of the EPO protein concentrations in the culture medium of hiPSC-EPO cells measured using ELISA. (I) EPO mRNA expression was observed in the differentiation cultures of multiple hiPSC/ESC lines. Each value was normalized to that of the hiPSC 253G4 cell line on day 0. The data were obtained from four independent experiments and are means ± SEM in (D), (E), (H), and (I). *P < 0.05 versus day 0 in (D), (E), and (H); analysis of variance (ANOVA) with Bonferroni’s test. Scale bars, 40 μm (B and F) and 2 μm (G).

We thus aimed to generate EPO cells from the hiPSC 253G4 cell line (12) by modifying previously reported protocols for hepatic lineage differentiation (fig. S1) (13, 14). As a result, hepatic lineage cells derived from hiPSCs expressed EPO mRNA at stage 2, and the expression disappeared at stage 3 (Fig. 1C). The expression of both SOX17 and AFP was also detected in stage 2 cells, confirming that the differentiated cells obtained in the hepatic differentiation step (stage 2) but not in the maturation step (stage 3) mainly produced EPO. We next examined the time course of EPO mRNA expression in stage 2 cells and found that the expression gradually increased with statistical significance on culture day 8 (55.49 ± 6.22–fold increase on day 8 versus day 0; P < 0.05; Fig. 1D). Consistent with the results of the quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis, the increase in EPO protein expression became statistically significant on day 12 (3.42 ± 0.53–fold increase versus day 0; P < 0.05) according to Western blot analysis (Fig. 1E). The EPO protein expression was also confirmed by immunostaining, which showed signals in the cytoplasm of the induced cells at stage 2 and confirmed that the EPO cells were also positive for the hepatic lineage marker AFP (Fig. 1F). AFP and another hepatic lineage marker, albumin, along with the endoderm markers hepatocyte nuclear factor-1β (HNF-1β), HNF-4, Sal-like protein 4 (SALL4), GATA4, and cytokeratin 19 (CK19), were expressed in almost all hiPSC-EPO cells (fig. S2). In addition, EPO expression, as analyzed by immunostaining, was similar across the hiPSC-EPO cell population (fig. S2). Ultrastructural observations using transmission electron microscopy revealed that EPO cells contained numerous vesicles in the cytoplasm, but these were less dense than in the mouse fetal liver (Fig. 1G). Next, we examined EPO protein secretion by hiPSC-EPO cells into the culture medium. The concentration of EPO protein measured using the enzyme-linked immunosorbent assay (ELISA) showed a statistically significant increase in the culture medium of hiPSC-EPO cells at stage 2 after 8 days of culture (0.20 ± 0.05 ng/ml on day 8 versus day 0; P < 0.05; Fig. 1H).

Different hiPSC/ESC lines vary in their differentiation potential (14, 15). We thus confirmed the generation of EPO cells from multiple hiPSC/ESC lines using our differentiation protocol, including eight hiPSC lines (201B6, 201B7, 253G1, 253G4, 585A1, 604A3, 606A1, and 606B1) (8, 12, 14) and two hESC lines (KhES3 and H9) (10, 16). The results suggested that EPO cells can be differentiated from multiple hiPSC/ESC lines using our modified hepatic differentiation protocol, although the EPO mRNA expression was variable among the cell lines (Fig. 1I). We then investigated the correlation of EPO mRNA expression and protein secretion and mRNA expression of the hepatoblast markers AFP and DLK1 (delta-like noncanonical Notch ligand 1) in the 253G4, 585A1, and KhES3 cell lines. EPO mRNA expression and protein secretion correlated with the expression of hepatoblast markers in the 253G4 and 585A1 cell lines but not when the 585A1 and KhES3 cell lines were compared, suggesting that both the differentiation potential of hiPSC/ESC-EPO cells and EPO production and secretion capacity were variable among the cell lines (fig. S3).

hiPSC-EPO cells show long-term proliferative and EPO-producing capacity in vitro

We evaluated the long-term proliferation and EPO-producing capacity of hiPSC-EPO cells in vitro. The number of hiPSC-EPO cells increased in a time-dependent manner (Fig. 2A), and some hiPSC-EPO cells at stage 2 on day 12 of culture stained positively for the cell proliferation marker Ki67 (Fig. 2B). By developing an in vitro maintenance culture method for hiPSC-EPO cells using a stage 2 culture medium and gentle cell passaging with Accutase treatment, we could increase the number of cells about 1000 times after 10 cell passages in 70 days (Fig. 2C), during which time both EPO mRNA expression and protein secretion were maintained (Fig. 2, D and E). These results suggest that hiPSC-EPO cells have a long-term proliferative and EPO-producing capacity in vitro.

Fig. 2. Cell proliferation and EPO-producing capacity of hiPSC-EPO cells.

(A) Number of stage 2 hiPSC-EPO cells during culture. hiPSCs were differentiated into EPO-producing cells (stage 2), as shown in fig. S1. (B) The expression of a cell proliferation marker, Ki67, was measured by immunocytochemistry. (C) Number of hiPSC-EPO cells during maintenance culture. (D and E) Time-course analyses of EPO mRNA expression (D) and EPO protein secretion (E) using RT-PCR and EPO ELISA, respectively. Each value was normalized to that of the stage 2 sample on day 0 (D). The data from four independent experiments are means ± SEM in (A), (C), (D), and (E). *P < 0.05 versus day 0 in (A), (D), and (E); ANOVA with Bonferroni’s test. Scale bars, 40 μm (B).

EPO-producing cells can be generated from miPSCs/ESCs

We next examined whether miPSCs/ESCs could also be induced to differentiate into EPO cells. By modifying our differentiation method for driving hiPSCs to become EPO cells, we developed a differentiation protocol for obtaining EPO cells from miPSCs/ESCs (fig. S4). qRT-PCR analyses showed that the miPSC cell line 492B-4 treated with our differentiation protocol expressed Epo mRNA during stage 2 of development and that the expression became statistically significant on culture day 10 (5.88 ± 1.02–fold increase versus day 0; P < 0.05; Fig. 3A). We subsequently confirmed that our differentiation protocol for inducing EPO cells also worked with the mESC D3 cell line (Fig. 3B). EPO protein secretion into the culture medium was also confirmed by ELISA, which showed that the protein concentration increased in a time-dependent manner (Fig. 3C). These results indicate that our differentiation protocol could be used to generate EPO cells from both miPSCs/ESCs and hiPSCs/ESCs.

Fig. 3. Differentiation of EPO-producing cells from miPSCs/ESCs.

(A) Temporal pattern of mouse Epo mRNA expression in the differentiation culture of an miPSC line, 494B-4, measured using qRT-PCR. Each value was normalized to that of the sample on day 0. (B) Epo mRNA expression observed in the differentiation culture of an mESC line, D3. Each value was normalized to that of the sample on day 0. (C) Mouse EPO protein concentrations in the cell culture medium were measured using ELISA. The data were obtained from four independent experiments and are means ± SEM. *P < 0.05 versus day 0; Student’s t test (B) and ANOVA with Bonferroni’s test (C).

IGF-1 treatment augments EPO production and secretion in hiPSC-EPO cells

To increase EPO production and improve the differentiation efficiency of EPO cells from hiPSCs/ESCs, we evaluated the effects of various growth factors and chemicals on EPO expression during stage 2 culture. The screening results for 46 different factors showed that the addition of human insulin-like growth factor 1 (IGF-1) most potently augmented EPO mRNA expression (fig. S5). We next compared the effects of IGF-1 on EPO expression with that induced by related factors, such as IGF-2 and insulin (Fig. 4A). IGF-1 treatment at 100 ng/ml increased EPO gene expression about sixfold compared to that observed in controls treated with no additional factors, whereas the same concentration of IGF-2 or insulin had little or no effect. Western blot analyses confirmed that stage 2 hiPSC-EPO cells expressed the IGF-1 receptor (Fig. 4B). We then examined the dose response and time course of EPO gene expression and protein secretion in the culture medium after IGF-1 treatment. Both EPO gene expression and protein secretion were augmented by IGF-1 treatment in a dose-dependent and time-dependent manner (Fig. 4, C to F). IGF-1 treatment at 100 ng/ml resulted in a significant increase in EPO mRNA expression after 4 days (82.14 ± 10.86–fold increase on day 4 versus day 0; P < 0.05) and EPO protein secretion after 8 days (0.19 ± 0.02 ng/ml on day 8; P < 0.05), which continued to increase up to day 12. Neither AFP nor ALBUMIN mRNA expression was significantly changed by IGF-1 treatment, excluding the possibility that IGF-1 altered the differentiation state of the hiPSC-EPO cells (Fig. 4, G and H). These results indicate that IGF-1 treatment promoted EPO production by hiPSC-EPO cells.

Fig. 4. Effects of IGF-1 treatment on EPO expression and secretion in hiPSC-EPO cells.

(A) Effects of IGF-1, IGF-2, and insulin treatment on the EPO mRNA expression for stage 2 hiPSC-EPO cells on culture day 8 using qRT-PCR analysis. Each value was normalized to that of control samples (no growth factor). A sample of human fetal liver was used as a positive control. (B) IGF-1 receptor expression was measured using Western blot analysis. Representative data are shown for four independent experiments. (C and D) Concentration-dependent effects of IGF-1 treatment on EPO mRNA expression (C) and protein secretion (D) for stage 2 hiPSC-EPO cells on culture day 8. Each value was normalized to that of control hiPSCs treated with no IGF-1. (E and F) Time-course analyses of the IGF-1–induced EPO mRNA expression (E) and protein secretion (F) for hiPSC-EPO cells. Each value was normalized to that of samples on day 0 in (E). (G and H) The mRNA expression of AFP (G) and ALBUMIN (H) was analyzed using qRT-PCR. Each value was normalized to control stage 2 hiPSC-EPO cells on culture day 8 not treated with IGF-1. The data were obtained from four independent experiments and are means ± SEM in (A) and (C) to (H). *P < 0.05 versus control samples without IGF-1 treatment in (A) and (C) to (F); ANOVA with Bonferroni’s test.

Hypoxia augments EPO production by hiPSC-EPO cells

It has been reported that EPO production is regulated by oxygen concentrations through hypoxia-inducible factors (HIFs) and their regulators, prolyl hydroxylase domain–containing enzymes (PHDs) (17). We examined whether this was also the case with hiPSC-EPO cells and evaluated the effects of hypoxia on EPO gene expression and protein secretion in hiPSC-EPO cells (Fig. 5, A and B). The results showed that both EPO mRNA expression and protein secretion gradually increased even under normoxic condition (21% oxygen) for the first 8 days of stage 2 development in parallel with the AFP and ALBUMIN mRNA expression (Fig. 5, C and D), indicating that the differentiation or maturation of hiPSC-EPO cells proceeded to some extent during these 8 days. Nevertheless, hypoxia (1% oxygen) increased EPO mRNA expression and EPO protein concentration in the culture medium more than under normoxic conditions, with statistically significant differences in EPO mRNA expression on day 6 and in EPO protein secretion on day 10 (P < 0.05; Fig. 5, A and B). In addition, the sensitivity of hiPSC-EPO cells to the hypoxic conditions was evaluated. Compared to the normoxic condition (21% oxygen), two hypoxic conditions (1 and 5% oxygen) augmented EPO mRNA expression and protein secretion dose-dependently (fig. S6, A and B). In contrast, hypoxia did not significantly alter AFP or ALBUMIN mRNA expression, excluding the possibility that hypoxia affected the differentiation state of the hiPSC-EPO cells (Fig. 5, C and D). These results suggest that hiPSC-EPO cells are functionally similar to EPO cells in vivo in that they produce and secrete EPO in response to the oxygen concentration.

Fig. 5. Stimulation of EPO expression and secretion under hypoxic culture conditions.

(A and B) Time-course analyses of EPO mRNA expression (A) and protein secretion (B) in hiPSC-EPO cells cultured under low oxygen (1%) conditions compared to control hiPSC-EPO cells cultured under normal oxygen (21%) conditions. EPO mRNA expression was measured using qRT-PCR, whereas the EPO protein concentrations in the culture medium were analyzed using ELISA. Each value was normalized to that of samples on day 0 in (A). (C and D) AFP (C) and ALBUMIN (D) mRNA expression was analyzed using qRT-PCR. Each value was normalized to control stage 2 hiPSC-EPO cells on culture day 8 cultured under normoxic conditions. (E and F) The effects of PHD inhibitors on EPO mRNA expression (E) and protein secretion (F) in HepG2 cells and in stage 2 hiPSC-EPO cells on culture day 8 were analyzed using qRT-PCR and ELISA, respectively (100 μM DFO, 50 μM FG4592, and 1 mM DMOG). Each value was normalized to control samples (no PHD inhibitor) in (E). (G) Effects of an HIF-1 dimerization inhibitor, Acriflavine (10 μM), on EPO protein secretion by hiPSC-EPO cells. (H) The nuclear translocation of HIF-1α and HIF-2α in hiPSC-EPO and HepG2 cells under hypoxic (1%) conditions was evaluated by immunocytochemistry. The data from four independent experiments are means ± SEM. *P < 0.05 versus control samples under normoxic conditions in (A) and (B) or without PHD inhibitors in (E) to (G); **P < 0.05 versus control samples (HepG2 cells) in (E) and (F); ANOVA with Bonferroni’s test (A, B, E, and F) and Student’s t test (G). Scale bars, 20 μm (H).

We then evaluated the effects of three PHD inhibitors: dimethyloxalylglycine (DMOG), deferoxamine (DFO), and FG4592 (18). First, we confirmed the expression of HIF subunits. HIF-1α, HIF-2α, HIF-3α, HIF-1β, and HIF-2β mRNA expression was detected in both hiPSC-EPO and HepG2 cells, an immortalized human cell line that secretes EPO protein in an oxygen-dependent manner (fig. S6C). We also confirmed the protein expression of HIF-1α and HIF-2α by immunostaining in hiPSC-EPO cells (fig. S6D). In addition, both hiPSC-EPO and HepG2 cells expressed PHD1, PHD2, and PHD3 mRNAs, sharing the expression of the same HIF and PHD genes with the human fetal liver (fig. S6C). In hiPSC-EPO cells, DFO and FG4592 significantly augmented EPO production in a dose-dependent manner, whereas DMOG did not (Fig. 5, E and F, and fig. S6, E and F). Furthermore, an inhibitor of HIF-1 dimerization, Acriflavine, attenuated EPO production in hiPSC-EPO cells (Fig. 5G). In contrast, in HepG2 cells, DFO and DMOG augmented EPO production, but FG4592 did not (Fig. 5, E and F). We also confirmed that hypoxic culture conditions stabilized HIF-1α and HIF-2α, as evidenced by the augmentation of nuclear signals revealed using anti–HIF-1α and anti–HIF-2α immunostaining of hiPSC-EPO and HepG2 cells (Fig. 5H). IGF-1 treatment also stabilized HIF-1α and HIF-2α and increased EPO expression (Fig. 4D and fig. S6, G and H). These data suggest that hiPSC-EPO cells produced EPO via the HIF-PHD pathway and that hiPSC-EPO and HepG2 cells may use different regulatory machineries to express EPO.

EPO protein produced by hiPSC-EPO cells shows erythropoietic activity in vitro

EPO is essential for erythropoiesis and is the primary regulator of this process. To evaluate the erythropoietic potential of human EPO protein produced by hiPSC-EPO cells (hiPSC-EPO protein), we performed clonogenic hematopoietic progenitor cell assays in methylcellulose with purified and concentrated hiPSC-EPO protein (Fig. 6A). Similar to observations for rhEPO, treatment with hiPSC-EPO protein significantly increased the number of erythroid burst-forming unit colonies [BFU-E; 7.33 ± 1.20 colonies with hiPSC-EPO protein (2.8 ng/ml) (n = 3) versus 2.00 ± 0.00 colonies without EPO treatment (n = 3); P < 0.05; Fig. 6, B and C] but not the other types of colonies, such as granulocyte-macrophage colony-forming units [CFU-GM; 9.33 ± 1.45 colonies with hiPSC-EPO protein (2.8 ng/ml) (n = 3) versus 6.00 ± 1.16 colonies without EPO treatment (n = 3); P > 0.05], differentiated from cord blood CD34+ hematopoietic progenitor cells (Fig. 6C and table S1). The efficiency of BFU-E formation induced by hiPSC-EPO protein was similar to that achieved with rhEPO protein [6.67 ± 0.33 colonies with rhEPO protein (2.8 ng/ml); n = 3; P > 0.05]. However, the erythropoietic effects of the hiPSC-EPO protein (2.8 ng/ml) were markedly suppressed by adding human EPO–specific antibodies (Fig. 6D and table S1). These results suggest that hiPSC-EPO cells secrete functional EPO protein that can stimulate erythropoiesis to a similar extent as conventional rhEPO protein.

Fig. 6. Effects of hiPSC-EPO protein on erythropoiesis in vitro.

(A) Schematic of clonogenic hematopoietic progenitor assays using methylcellulose-based semisolid medium. (B) Representative images of BFU-E induced by rhEPO (left) and hiPSC-EPO protein (right). (C) Number of clonal colonies on semisolid medium containing ST3 supplemented with rhEPO (0.28 and 2.8 ng/ml) or hiPSC-EPO protein (0.28 and 2.8 ng/ml) (n = 3). (D) Number of clonal colonies on semisolid medium containing different concentrations of neutralizing antibodies against human EPO in addition to ST3 and hiPSC-EPO protein (2.8 ng/ml) (n = 3). The data were obtained from three independent experiments and are means ± SEM in (C) and (D). Scale bars, 200 μm (B).

EPO protein produced by hiPSC-EPO cells shows erythropoietic activity in vivo

A renal anemia mouse model was developed using adenine treatment of C57BL/6 mice (19, 20). An oral gavage of adenine (50 mg/kg body weight daily for 4 weeks) decreased hematocrit and hemoglobin concentration. The mice were then further treated after dividing them into the following five groups: group 1, no adenine treatment (control; n = 6); group 2, rhEPO (28 ng), subcutaneous injection (n = 10); group 3, rhEPO (0.56 ng), subcutaneous injection (n = 10); group 4, hiPSC-EPO protein (0.56 ng), subcutaneous injection (n = 10); and group 5, saline, subcutaneous injection (n = 10). Notably, the subcutaneous injection of rhEPO and hiPSC-EPO protein into the mice for 4 weeks resulted in recovery of both the hematocrit [normal range in C57BL/6 mice, 36.8 to 52.7%; 39.78 ± 0.51% with hiPSC-EPO protein (n = 10) versus 33.85 ± 0.52% with saline (n = 10); P < 0.05; Fig. 7, A and B] and hemoglobin concentration (normal range in C57BL/6 mice, 11.0 to 15.2 g/dl; 11.40 ± 0.16 g/dl with hiPSC-EPO protein versus 9.52 ± 0.27 g/dl with saline; P < 0.05; Fig. 7C) (21) to within the normal range. However, this treatment did not increase the number of white blood cells or platelets (Fig. 7D), suggesting that hiPSC-EPO protein induces the differentiation of only erythrocyte progenitor cells. The adenine-treated mice did not show any significant changes in mean corpuscular volume or mean corpuscular hemoglobin concentration (Fig. 7E) or body weight (fig. S7A) after 4 weeks of treatment with rhEPO or hiPSC-EPO protein. Whereas the mouse EPO concentrations measured using ELISA were not affected by human EPO treatment in the adenine-treated mice (Fig. 7F), the human EPO concentrations increased after treatment with rhEPO (28 ng) or hiPSC-EPO protein (0.56 ng), suggesting that the mouse renal anemia was improved by treatment with human EPO protein (Fig. 7G). We also confirmed that both hiPSC-EPO protein and rhEPO improved renal anemia in a dose-dependent manner (fig. S7B). The efficiency of renal anemia recovery was calculated on the basis of the increase in hematocrit per week. Treatment with hiPSC-EPO protein more efficiently improved the anemic state than did rhEPO treatment [3.46 ± 0.47%/week with hiPSC-EPO protein (0.56 ng) versus 1.00 ± 0.52%/week with rhEPO (0.56 ng); Fig. 7H]. To elucidate the molecular mechanisms responsible, we measured the glycosylation pattern of hiPSC-EPO protein by lectin microarray analysis (Fig. 7I). Although the glycosylation pattern resembled that of two clinically available rhEPOs (epoetin-α and epoetin-β), the glycosylation of hiPSC-EPO protein was more complicated, as evidenced by its reactivity with numerous lectins. Because glycosylation modulates the efficacy and degradation of EPO (22), we also compared half-lives in vivo and found that hiPSC-EPO had a slightly longer half-life than did epoetin-β in mice, although the difference may not be clinically significant (Fig. 7J). These data suggest that EPO protein produced by hiPSC-EPO cells was capable of inducing erythropoiesis in vivo and in vitro and that different glycosylation patterns may help explain why hiPSC-EPO protein induced a slightly greater hematocrit in mice with renal anemia than did rhEPO. However, we should note that there may be no biological difference between the two EPO proteins because increased doses of rhEPO resulted in bioequivalence with hiPSC-EPO protein (fig. S7B).

Fig. 7. Therapeutic effects of hiPSC-EPO protein on renal anemia in adenine-treated mice.

(A to H) Renal anemia was induced by adenine treatment (50 mg/kg body weight daily for 4 weeks) in male C57BL/6 mice. The mice were then treated with rhEPO or hiPSC-EPO protein. (A) Time-course analyses of the hematocrit (Hct) measured using glass capillary tubes. (B) Hematocrit after 4 weeks of treatment with rhEPO or hiPSC-EPO protein. (C) Hemoglobin concentrations after 4 weeks of treatment with rhEPO or hiPSC-EPO protein analyzed by ELISA. The gray shaded areas indicate the normal hematocrit in C57BL/6 mice in (A) and (B) and the normal hemoglobin concentration in C57BL/6 mice in (C). (D) Number of red blood cells (RBC), white blood cells (WBC), and platelets after 4 weeks of treatment with rhEPO or hiPSC-EPO protein. (E) Mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), and mean corpuscular hemoglobin concentration (MCHC) after 4 weeks of treatment with rhEPO or hiPSC-EPO protein. (F and G) The concentrations of mouse (F) and human (G) EPO protein in mouse serum were measured after 4 weeks of treatment with rhEPO or hiPSC-EPO protein using ELISA. (H) The efficiency of renal anemia recovery after treatment with 0.56 ng of rhEPO or hiPSC-EPO protein was calculated on the basis of the increase in hematocrit per week (ΔHct/week). (I) Glycosylation patterns of hiPSC-EPO and rhEPO protein were measured using lectin microarray assays. The data are means ± SEM (n = 4). Abbreviations are defined in table S3. (J) Half-lives of the hiPSC-EPO protein and rhEPO in vivo were evaluated by measuring EPO protein concentrations in mouse serum after subcutaneous injection of hiPSC-EPO or rhEPO using ELISA. The data from three independent experiments are means ± SEM in (A) to (H) and (J). n = 6 for control and n = 10 for rhEPO (28 ng), rhEPO (0.56 ng), hiPSC-EPO protein (0.56 ng), and saline. *P < 0.05 versus saline in (A) to (D), control in (G), and rhEPO in (H); ANOVA with Bonferroni’s test (A to G) and Student’s t test (H).

Transplantation of hiPSC-EPO cells improves renal anemia in mice

Immunodeficient (NOD.CB17-Prkdcscid/J) mice were treated with adenine by oral gavage (50 mg/kg body weight daily for 5 weeks) to induce renal anemia. To examine the feasibility of hiPSC-based cell therapy for treating renal anemia, we transplanted 20 aggregates of hiPSC-EPO cells (5.0 × 105 cells per aggregate) under the kidney subcapsules of the anemic mice. Notably, transplantation of hiPSC-EPO cells significantly improved renal anemia after 4 weeks compared with saline treatment [hematocrit, normal range in NOD.CB17-Prkdcscid/J mice, 42.3 to 48.1%; 42.56 ± 1.74% with hiPSC-EPO cells (n = 6) versus 34.60 ± 5.23% with saline (n = 6); P < 0.05; Fig. 8A] (21). The human EPO concentration in the host mouse serum at 4 weeks after transplantation was detectable using ELISA and was significantly increased in the mice that underwent transplantation with hiPSC-EPO cells [1.42 ± 0.23 ng/ml with hiPSC-EPO cells (n = 6) versus 0.07 ± 0.02 ng/ml with saline (n = 6); P < 0.05; Fig. 8B]. Mice with renal anemia transplanted with hiPSC-EPO cells did not develop polycythemia during the observation period (for up to 28 weeks after transplantation), and their hematocrit remained within the normal range. In contrast, renal anemia persisted in adenine-treated mice that did not undergo transplantation (46.40 ± 3.49% with hiPSC-EPO cells (n = 4) versus 37.23 ± 2.86% with saline (n = 4) at 28 weeks after transplantation; P < 0.05; Fig. 8C). In addition, human EPO protein in mouse serum was also detectable for up to 28 weeks after transplantation (Fig. 8D). We performed histological analyses of the hiPSC-EPO cell aggregates injected under the renal subcapsule of immunodeficient mice. Hematoxylin and eosin (H&E) staining revealed structures consisting of homogeneous cells in the grafts (Fig. 8E). Immunohistochemical analysis revealed that the transplanted hiPSC-EPO cell aggregates maintained EPO expression in the subcapsule of host mouse kidneys for at least 28 weeks after transplantation (Fig. 8E). Although AFP expression was decreased 8 weeks after transplantation, albumin expression was detectable for 28 weeks after transplantation (Fig. 8E). Blood vessel–like tubular structures were found inside the grafts and contained red blood cells (Fig. 8F). The new vasculature derived from host mice was confirmed by the expression of mouse CD31/platelet endothelial cell adhesion molecule-1 (PECAM-1), indicating that mouse host blood vessels may have supplied nutrition and oxygen to the grafts. EPO secretion from transplanted hiPSC-EPO cells after phlebotomy was evaluated (Fig. 8G). We found that moderate phlebotomy (with the hematocrit decreased from 44% to 34%) in the group transplanted with hiPSC-EPO cell aggregates resulted in an increase in circulating human EPO, but this did not occur in control untransplanted mice (Fig. 8G). These data indicate that hiPSC-EPO cells can survive in host mice with renal anemia, can secrete functional human EPO protein, and can improve renal anemia in these animals.

Fig. 8. Therapeutic effects of the transplantation of hiPSC-EPO–producing cells on renal anemia in adenine-treated mice.

Renal anemia was induced using adenine treatment (50 mg/kg body weight daily for 5 weeks) in immunodeficient mice (NOD.CB17-Prkdcscid/J mice). Twenty aggregates of hiPSC-EPO cells (5.0 × 105 cells per aggregate) were transplanted into the kidney subcapsules of mice with renal anemia. (A) Hematocrit was examined during the first 4 weeks after transplantation using glass capillary tubes. (B) Human EPO concentrations in mouse serum at 4 weeks after transplantation were measured using ELISA. (C) Hematocrit was examined for up to 28 weeks after transplantation. The gray shaded areas in (A) and (C) indicate the normal hematocrit range in NOD.CB17-Prkdcscid/J mice. (D) Human EPO concentrations in mouse serum after transplantation were measured using ELISA. (E) The hiPSC-EPO–producing cell grafts were evaluated using immunohistochemistry for EPO (green), AFP (red) and ALBUMIN (red) and using H&E staining. (F) The new vasculature in the grafts derived from host mice was examined by anti-mouse CD31/PECAM-1 immunostaining and H&E staining. (G) The human EPO concentrations in host mouse serum after phlebotomy were measured using ELISA. The data from three independent experiments are means ± SEM; n = 6 for hiPSC-EPO–producing cells and saline in (A) and (B). The data from two independent experiments are means ± SEM; n = 4 for hiPSC-EPO cells and saline in (C), (D), and (G). *P < 0.05 versus control; ANOVA with Bonferroni’s test (A, C, D, and G) and Student’s t test (B). Scale bars, 40 μm (E) and 20 μm (F).

DISCUSSION

Although rhEPO treatment is beneficial for CKD patients with renal anemia, several problems remain to be addressed. First, the increasing number of CKD patients is expanding the demand for rhEPO treatment, which, in turn, increases the total cost of this therapy. Second, it is difficult to physiologically control renal anemia using rhEPO treatment. The intermittent administration of rhEPO causes fluctuations in hemoglobin concentrations (3), which is associated with an increased incidence of cardiovascular events (23). In addition, the target hemoglobin concentration for rhEPO treatment remains controversial, and hemoglobin concentrations in most patients are lower than those observed in healthy subjects (24). The physiological control of renal anemia based on a stable, normal range of hemoglobin concentrations may help in the treatment of CKD patients.

hiPSCs/ESCs are potential cell sources for regenerative medicine. Here, we generated EPO-producing cells from hiPSCs/ESCs and miPSCs/ESCs. These cells expressed EPO mRNA and protein and increased the production and secretion of EPO protein in response to low oxygen conditions. The secreted EPO protein induced the erythropoietic differentiation of human umbilical cord blood CD34+ hematopoietic cells in vitro and improved renal anemia in adenine-treated mice in vivo. Furthermore, the transplantation of hiPSC-EPO cells reversed renal anemia, with a return to the normal range for the hematocrit in immunodeficient mice. Therefore, EPO cells generated from hiPSCs/ESCs may be useful as a cell therapy for treating renal anemia.

rhEPO treatment is associated with a very rare side effect of severe anemia due to the generation of anti-rhEPO autoantibodies (6). Although further studies are needed to clarify the immunological effects of transplanted hiPSC-EPO cells, the cell types generated from a patient’s own iPSCs may provide a safer cell therapy in terms of the immune response. Moreover, we confirmed that hiPSC-EPO cells secreted functional EPO protein for at least 70 days using regular in vitro culture conditions. In addition, hiPSC-EPO cells have the potential to survive in a long term (28 weeks) after transplantation in the body.

In addition to their application for clinical use, hiPSC/ESC-derived EPO-producing cells will be valuable for basic research. Although much is known about the possible mechanisms of EPO production and secretion, such as the involvement of HIF-2α (17), the detailed molecular mechanisms have not yet been fully elucidated because of the limited access to isolated and cultured EPO-producing cells, especially human cells. EPO cells have been identified in the peritubular interstitial space in mouse kidneys (11). Moreover, EPO cells were recently isolated from primary cultures of human (25) and mouse kidneys (26). These cells exhibited gene expression and protein secretion of EPO in an oxygen-dependent manner, suggesting that EPO cells removed from the kidneys can be used to investigate the molecular mechanisms of human EPO production. However, it is difficult to obtain an adequate amount of EPO cells because access to human kidney specimens is limited. In addition, primary cultured cells isolated from normal human tissues are often difficult to expand in vitro in a long term. HepG2 cells are widely used to investigate EPO production (27, 28); however, because they are derived from hepatoma tissues, EPO production and secretion may differ from that observed in healthy, nontumor cells. Here, hypoxia activated EPO secretion from the hiPSC-EPO cells, and the secreted EPO protein induced erythropoietic differentiation both in vitro and in vivo, suggesting that hiPSC-EPO cells may be functionally similar to their in vivo counterparts. Given that these EPO cells can be readily derived from hiPSCs/ESCs and miPSCs/ESCs and display robust in vitro expansion potential, they could be reliable cellular models for investigating the intracellular signaling pathways involved in EPO production and secretion.

Both clinical studies and in vivo research using experimental animals have reported that IGF-1 is involved in EPO production and secretion. One clinical study revealed that IGF-1 stimulates EPO secretion from the kidneys of patients with insulin resistance (29). On the other hand, the acute injection of IGF-1 negatively controls EPO secretion from the mouse kidney (30). In addition, IGF-1 suppresses EPO secretion from the kidney but increases its secretion from the liver (31). Here, we show that IGF-1 augmented EPO production in hiPSC-EPO cells. Combined, these data suggest that IGF-1 is an important regulatory factor for EPO production and secretion.

Several limitations of our study deserve mention. First, the EPO cells in our study were generated from hiPSCs/ESCs through hepatic lineage differentiation. The liver is a major organ of EPO production in the fetal and early neonatal periods. Although EPO production shifts to the kidney in late gestation and adulthood (32), previous studies have reported that EPO production by the adult liver can be induced under anemic or hypoxic conditions (1, 2, 33). Nevertheless, methods that induce hiPSCs/ESCs into becoming EPO-producing cells via differentiation through the kidney lineage would be helpful for elucidating the mechanisms of EPO production, especially considering that gene regulation is different between the mouse kidney and the liver (34). It has recently been reported that EPO-producing fibroblasts in the normal adult mouse kidney had originated from myelin protein zero–Cre lineage-labeled extrarenal cells, which entered the embryonic kidney on E13.5 (35). Because myelin protein zero is expressed in migrating neural crest cells in early embryonic stages, the origin of EPO cells in the kidney may be the neural crest. Therefore, differentiation methods that drive hiPSCs into the neural crest lineage and then into EPO cells could provide insights into the kidney-derived EPO production.

Second, the mechanisms underlying the variable responses seen with PHD inhibitors and the different responses between hiPSC-EPO and HepG2 cells are still unclear. We evaluated the effects of three different PHD inhibitors (DMOG, DFO, and FG4592) on EPO expression and secretion. Previous studies reported that these PHD inhibitors induced different responses in the expression or function of HIF-PHD pathway molecules and their biological effects (36, 37). Furthermore, PHD inhibitors have variable selectivity and inhibitory effects on different cell types, such as liver and retinal cells, and on PHD subtypes, such as PHD1, PHD2, and PHD3 (37, 38). Together, these differences among PHD inhibitors may at least, in part, explain the variable responses observed between hiPSC-EPO and HepG2 cells. Nevertheless, a recent report indicated that FG4592 has more efficient and broader inhibitory effects on various cell types compared to DMOG (37); several clinical trials are using FG4592 (39, 40). The present findings that FG4592 augmented EPO production only in hiPSC-EPO cells, but not in HepG2 cells, suggest that hiPSC-EPO cells may provide a good model for screening PHD inhibitors for their effects on renal anemia.

One final limitation of our study is that a large number of hiPSC-EPO cells need to be used. At the beginning of this study, we decided on the cell number for transplantation based on EPO secretion in vitro. The results of EPO secretion into the cell culture medium detected by ELISA indicated that 1 × 107 hiPSC-EPO cells secreted about 12 ng/day of EPO protein. The findings of in vivo dose-response experiments (fig. S7B) indicated that this amount of EPO protein was enough to reverse renal anemia. Therefore, we transplanted 1 × 107 hiPSC-EPO cells under the left renal subcapsule of each immunodeficient mouse. However, future studies should optimize EPO production capacity and the number of hiPSC-EPO cells for transplantation. In addition, our data suggest that there was variability in both the differentiation potential and EPO secretion capacity among the different hiPSC/ESC lines. Further optimization and improvement of the differentiation protocol for EPO cells will be required to achieve broad application.

In summary, we successfully generated EPO cells from hiPSCs/ESCs and miPSCs/ESCs. These cells produced and secreted functional EPO protein in a hypoxia-dependent manner, and IGF-1 treatment augmented EPO production. From the point of view of basic research, iPSC/ESC-derived EPO cells may be a powerful tool for investigating the molecular mechanisms of EPO production and secretion. hiPSC/ESC-derived EPO-producing cells may potentially provide a useful cell population for cell therapy for the treatment of renal anemia in CKD patients.

MATERIALS AND METHODS

Study design

The aim of this study was to establish a differentiation method that produces EPO cells from hiPSCs/ESCs and miPSCs/ESCs and to evaluate the transplantation of the resulting hiPSC-EPO cells in a CKD mouse model. EPO production and secretion were evaluated by RT-PCR, immunostaining, and ELISA. All mice analyzed in this study were handled and the experimental procedures were performed under the guidelines for the care and use of animals established by Kagawa University and Kyoto University. All animal experiments were approved by the Animal Care and Use Committee for Kagawa University and the Center for iPS Cell Research and Application Animal Experiment Committee. The sample size (n = 6 to 10) for the animal experiments was based on the results of preliminary experiments. Exact numbers for each experiment are described in the figure legends. The investigators were not blinded when evaluating the experiments. The mice were randomly assigned to the treatment and control groups.

Cell culture

hiPSCs (201B6, 201B7, 253G1, 253G4, 585A1, 604A3, 606A1, and 606B1) (8, 14) and hESCs (KhES-3 and H9) (10, 16) were grown on feeder layers of mitomycin C–treated SNL cells in Primate ES medium (ReproCELL) supplemented with penicillin/streptomycin (500 U/ml) (Thermo Fisher Scientific) and recombinant human basic fibroblast growth factor (4 ng/ml) (Wako), as previously described (41). The colonies of hiPSCs were dissociated using an enzymatic method with CTK solution consisting of 0.25% trypsin (Thermo Fisher Scientific), 0.1% collagenase IV (Thermo Fisher Scientific), 20% knockout serum replacement (KSR; Thermo Fisher Scientific), and 1 mM CaCl2 in phosphate-buffered saline (PBS).

miPSCs (492B-4) and mESCs (D3) were maintained on feeder layers of mitomycin C–treated SNL cells in Dulbecco’s modified Eagle’s medium (DMEM; Nacalai Tesque) supplemented with 15% fetal bovine serum (FBS; Hyclone Laboratory), penicillin/streptomycin (500 U/ml), 0.1 mM of nonessential amino acids (Thermo Fisher Scientific), 2 mM glutamine (Thermo Fisher Scientific), 0.55 mM 2-mercaptoethanol (Thermo Fisher Scientific), and leukemia inhibitory factor. miPSCs/ESCs were passaged using enzymatic dissociation with 0.25% trypsin/EDTA. HepG2 cells were cultured in DMEM with 10% FBS and penicillin/streptomycin (500 U/ml).

Differentiation protocols for EPO cells

The differentiation of EPO cells from hiPSCs/ESCs (fig. S1) was performed by modifying previously reported protocols into hepatic lineages (13, 14, 41). Colonies of hiPSCs/ESCs grown on an SNL feeder layer were first dissociated, according to an enzymatic method, with CTK dissociation solution to remove SNL cells. Then, the cells were dissociated to single cells via gentle pipetting after treatment with Accutase (Innovative Cell Technologies) for 20 min and seeded on Matrigel-coated plates (BD Biosciences) at a density of 4.5 × 105 cells/cm2 with stage 1 medium containing RPMI 1640 (Nacalai Tesque) supplemented with penicillin/streptomycin (500 U/ml), B27 supplement (Thermo Fisher Scientific), recombinant human/mouse/rat activin A (100 ng/ml) (R&D Systems), and 1 μM CHIR99021 (Wako). Y-27632 (10 μM; Wako) was added to the stage 1 medium for the first 24 hours. After 24 hours, the stage 1 medium was supplemented with 0.5 mM sodium butyrate (NaB; Sigma-Aldrich) until culture day 3. On day 7, the medium was changed to stage 2 medium containing KnockOut DMEM (Thermo Fisher Scientific) supplemented with penicillin/streptomycin (500 U/ml), 20% KSR, 1% dimethyl sulfoxide, 1 mM l-glutamine, 1% nonessential amino acid, and 0.1 mM β-mercaptoethanol. For hepatocyte differentiation, the cells were shifted to the hepatocyte maturation step after 7 days of stage 2 treatment using incubation with stage 3 medium containing hepatocyte culture medium (Lonza) supplemented with hepatocyte growth factor (10 ng/ml) (R&D Systems) and oncostatin M (20 ng/ml) (R&D Systems). For cell passaging, stage 2 cells were dissociated to single cells via gentle pipetting after treatment with Accutase for 8 min and split on Matrigel-coated plates at a ratio of 1:4 with stage 2 medium containing Y-27632 (10 μM).

The differentiation of EPO cells from miPSCs/ESCs was performed using a similar protocol to that for hiPSCs/ESCs (fig. S4). Briefly, miPSCs/ESCs dissociated to single cells with Accutase treatment were cultured with stage 1 medium containing RPMI 1640, penicillin/streptomycin (500 U/ml), B27 supplement, recombinant activin A (100 ng/ml), and 1 μM CHIR99021. After 2 days, the medium was supplemented with 0.5 mM NaB until day 4. On day 5, the medium was changed to stage 2 medium.

RT-PCR and real-time qRT-PCR

Total RNA was isolated using the RNeasy kit (Qiagen) according to the manufacturer’s recommendations, followed by cDNA synthesis using standard protocols. Briefly, first-strand cDNA was synthesized from 1 μg of total RNA using ReverTra Ace (Toyobo). The cDNA samples were subjected to PCR amplification using a thermal cycler (Veriti 96-Well Thermal Cycler; Thermo Fisher Scientific), and PCR was performed using the Ex-Taq PCR kit (Takara Bio) according to the manufacturer’s protocol. The PCR cycles were as follows: For β-ACTIN, initial denaturation at 94°C for 2.5 min, followed by 25 cycles of 94°C for 30 s, 60°C for 1 min, 72°C for 30 s, and final extension at 72°C for 10 min. For the other genes, the cycles consisted of initial denaturation at 94°C for 2.5 min, followed by 30 to 40 cycles of 94°C for 30 s, 58° to 62°C for 30 s, 72°C for 30 s, and final extension at 72°C for 7 min. qRT-PCR was performed using the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific) and SYBR Green PCR Master Mix (Takara Bio). Denaturation was performed at 95°C for 30 s, followed by 45 cycles at 95°C for 5 s and at 60°C for 30 s. The threshold cycle method was used to analyze the data for the gene expression levels and calibrated to those of either housekeeping gene β-ACTIN or GAPDH. Primer sequences are described in table S2.

Immunostaining

Immunostaining of the cultured cells was carried out as previously described (41). Briefly, the cells were fixed with 4% paraformaldehyde/PBS for 20 min at 4°C. After washing with PBS, the cells were blocked with 1% normal donkey serum and 3% bovine serum albumin (Nacalai Tesque)/PBST (PBS/0.25% Triton X-100) for 30 min at room temperature. A fetal mouse liver (E12.5) and transplanted aggregates were also fixed with 4% paraformaldehyde/PBS for 20 min at 4°C, embedded in paraffin, and sectioned into 4-μm slices. The sections were stained with H&E. The samples were incubated overnight at 4°C with the following primary antibodies: anti-EPO (Santa Cruz Biotechnology), AFP (Sigma-Aldrich), albumin (Bethyl Laboratories), HNF-1β (Santa Cruz Biotechnology), HNF-4 (Santa Cruz Biotechnology), SALL4 (Abcam), GATA4 (Santa Cruz Biotechnology), CK19 (Dako), Ki67 (BD Biosciences), CD31 (Dianova), HIF-1α (Sigma-Aldrich), and HIF-2α (Novus Biologicals). Secondary antibodies (Alexa Fluor antibodies; Thermo Fisher Scientific) were incubated for 30 min at room temperature. The stained cells were evaluated using fluorescence microscopy (BZ-9000; Keyence).

Measurement of EPO protein expression

The concentrations of EPO protein in the culture medium were measured using ELISA according to the manufacturer’s protocol (ALPCO Diagnostics for human EPO and Abnova for mouse EPO). Briefly, the supernatant of the EPO cell cultures was added to a 96-well plate with enzyme-labeled anti-human or anti-mouse EPO antibodies. The plates were incubated for 2 hours at room temperature. Then, the substrate was added, and the reaction was stopped with 0.5 mM sulfuric acid. The plates were read at 450 nm with a microplate reader (2104 EnVision; PerkinElmer), and the concentration of EPO was calculated on the basis of the standard curve of lyophilized synthetic EPO protein.

For the Western blot analysis, the EPO protein expression in the cell lysate was measured as previously described (42). Briefly, the cells were lysed, and solubilized proteins were isolated via centrifugation and quantified using a Bradford assay. The proteins were separated using SDS–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (GE Healthcare). After blocking with Odyssey blocking buffer (LI-COR Biosciences), blots were incubated with anti-EPO (Santa Cruz Biotechnology) or anti–IGF-1 receptor (Cell Signaling Technology) antibodies overnight at 4°C. Blots with embedded infrared dye were visualized with the Odyssey Infrared Imaging System (LI-COR Biosciences). To confirm the loading of equal amounts of protein, each membrane was reprobed with anti–β-ACTIN antibodies (Sigma-Aldrich). The band intensity was quantified according to immunoblot densitometry using the ImageJ software (National Institutes of Health).

Electron microscopy

For observation with transmission electron microscopy, the cells were fixed with 4% formaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer at 4°C overnight and washed in 0.1 M phosphate buffer. Then, the cells were refixed with phosphate-buffered 1% OsO4, dehydrated in a graded series of ethanol, and embedded in LUVEAK-812 (Nacalai Tesque). Thin sections were cut with an EM UC6 ultramicrotome (Leica Microsystems), stained with uranyl acetate and lead citrate, and observed using a Hitachi H-7650 electron microscope (Hitachi).

Lectin microarray

The glycosylation patterns of hiPSC-EPO and rhEPO proteins were measured by lectin microarray assays. hiPSC-EPO proteins were concentrated using a centrifugal filter device (Amicon Ultra-15; Merck Millipore) and then purified and isolated with biotin-labeled anti-EPO antibody (Fitzgerald Industries) using Dynabeads MyOne Streptavidin T1 (Thermo Fisher Scientific). Lectin microarray assays were performed by GlycoTechnica, as previously reported (43).

Clonogenic hematopoietic progenitor assay

Human CD34+ cells were isolated from cord blood via immunomagnetic bead separation (Miltenyi Biotec) according to the manufacturer’s instructions. The purity of the isolated CD34+ cells was >95%, as analyzed using flow cytometry. A total of 1 × 105 CD34+ cells were incubated with rhEPO or hiPSC-EPO protein from hiPSC-EPO cells at 0.28 and 2.8 ng/ml for 2 hours. hiPSC-EPO protein was purified and concentrated using Amicon Ultra-30 (Merck Millipore). Hematopoietic cells were cultured at a low cell density, with a concentration of 1 × 104 cells/ml in 35-mm petri dishes (Becton Dickinson) using MethoCult GF+ semisolid medium (1 ml per dish) (STEMCELL Technologies) including stem cell factor, thrombopoietin, and interleukin-3 (ST3), as previously described (44). The number of colonies was counted after 14 to 21 days of culture, and the colony types [CFU-Mix (mixed colony-forming units), BFU-E, and CFU-GM] were determined with in situ observation using an inverted microscope, according to the criteria described previously (44).

Animal experiments

All experimental procedures were performed under the guidelines for the care and use of animals established by Kagawa University and Kyoto University. Six-week-old male C57BL/6 mice (CLEA Japan) were used for the adenine-induced renal anemia model. Renal injury was induced by oral gavage with adenine (50 mg/kg body weight in 0.5% methylcellulose) daily for 4 weeks. The presence of renal anemia was confirmed by measuring hematocrit and hemoglobin levels and red blood cell count.

One week after the end of adenine treatment, the following five mouse groups were prepared: group 1, no adenine treatment (control; n = 6); group 2, rhEPO (28 ng), subcutaneously injected after adenine treatment [rhEPO (28 ng); n = 10]; group 3, rhEPO (0.56 ng), subcutaneously injected after adenine treatment [rhEPO (0.56 ng); n = 10]; group 4, hiPSC-EPO protein (0.56 ng), subcutaneously injected after adenine treatment [hiPSC-EPO protein (0.56 ng); n = 10]; and group 5, saline injection after adenine treatment (saline; n = 10). The rhEPO and hiPSC-EPO protein treatments were performed via subcutaneous injection three times a week for 4 weeks. For EPO protein dose-response experiments in vivo, the following nine mouse groups were prepared: group 1, no adenine treatment (control; n = 6); group 2, hiPSC-EPO protein (0.56 ng), subcutaneously injected after adenine treatment (n = 6); group 3, hiPSC-EPO protein (5.6 ng), subcutaneously injected after adenine treatment (n = 6); group 4, hiPSC-EPO protein (14 ng), subcutaneously injected after adenine treatment (n = 6); group 5, hiPSC-EPO protein (28 ng), subcutaneously injected after adenine treatment (n = 6); group 6, rhEPO (0.56 ng), subcutaneously injected after adenine treatment (n = 6); group 7, rhEPO (5.6 ng), subcutaneously injected after adenine treatment (n = 6); group 8, rhEPO (14 ng), subcutaneously injected after adenine treatment (n = 6); and group 9, rhEPO (28 ng), subcutaneously injected after adenine treatment (n = 6). To measure the half-life of EPO protein, rhEPO and hiPSC-EPO proteins were subcutaneously injected (28 ng) into an adenine-induced renal anemia model. Serum EPO concentrations were measured using ELISA at 1, 6, 24, and 48 hours after injection. The half-life of EPO was evaluated by decreases in serum EPO protein concentration after injection. Blood samples were collected to measure the hematocrit and hemoglobin levels and EPO protein concentrations. The hematocrit levels were determined in blood samples withdrawn into glass capillary tubes. The hemoglobin levels were measured using ELISA (Abcam), and the numbers of red blood cells, white blood cells, and platelets were counted using a Coulter Counter Multisizer 3 (Beckman Coulter).

For the transplantation experiments, renal anemia was induced using adenine treatment (50 mg/kg body weight daily for 5 weeks) in immunodeficient mice (NOD.CB17-Prkdcscid/J). To prepare the EPO cell aggregates, EPO cells were dissociated to single cells via pipetting after incubation with Accutase for 10 min. Then, the EPO cells were split into spindle-bottom low-adhesion 96-well plates at a density of 5.0 × 105 cells per well and incubated in stage 2 medium supplemented with 10 μM Y-27632 for 48 hours. The formation of cellular aggregates was confirmed with microscopic observation. After washing the aggregates with saline, 20 EPO cell aggregates were transplanted into the renal subcapsules of immunodeficient mice with renal anemia. The hematocrit levels were measured for up to 28 weeks after transplantation. The human EPO concentrations were also measured using ELISA. For immunohistochemistry analysis, mice were sacrificed 2, 4, 8, 16, and 28 weeks after transplantation, and the serial sections of transplanted aggregates were examined by immunostaining. To elucidate the effects of phlebotomy, serum EPO concentrations of control (no transplantation) and hiPSC-EPO cell–transplanted mice were measured by a Goldenrod animal lancet (MEDIpoint).

Statistical analysis

Results were expressed as the mean ± SEM. Multiple group comparisons were made using one-way or two-way ANOVA, followed by Bonferroni’s test. Student’s t tests were performed to compare the mean values when the experimental design was composed of two individual groups. P < 0.05 was considered statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/409/eaaj2300/DC1

Fig. S1. Differentiation method for generating EPO-producing cells from hiPSCs/ESCs.

Fig. S2. Expression of hepatic lineage and endoderm markers in hiPSC-EPO cells.

Fig. S3. Variable expression and secretion of EPO and expression of hepatoblast markers among three hiPSC/ESC lines.

Fig. S4. Differentiation method for generating EPO-producing cells from miPSCs/ESCs.

Fig. S5. Effects of 46 different factors on EPO mRNA expression in the hiPSC-EPO cells.

Fig. S6. The HIF-PHD pathway regulates EPO production induced by hypoxia or IGF-1 treatment.

Fig. S7. Effects of hiPSC-EPO protein on body weight and renal anemia in adenine-treated mice.

Table S1. Effects of hiPSC-EPO protein on in vitro erythropoiesis (related to Fig. 6).

Table S2. The sequences of sense and antisense primers used for RT-PCR in this study.

Table S3. A list of lectins and their specificity for microarray analysis.

REFERENCES AND NOTES

  1. Acknowledgments: We thank T. Toyoda, A. Watanabe, and M. Nagao for the helpful suggestions; T. Sudo, H. Tanaka, Y. Fujita, and M. A. Sufiun for their excellent technical assistance; and P. Karagiannis for reading the manuscript. Pathological analysis was done by Center of Anatomical, Pathological and Forensic Medical Research, Graduate School of Medicine, Kyoto University. Funding: This work was supported by the Japan Agency for Medical Research and Development through its research grant “Core Center for iPS Cell Research, Research Center Network for Realization of Regenerative Medicine,” by a grant-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (#22790786, #22790792, #24591204, and #15K09266), by the iPS Cell Research Fund, by a Sanju Alumni Research Grant, by the Kanae Foundation for the Promotion of Medical Science, and by the Daiichi Sankyo Foundation of Life Science. Author contributions: H.H., A.N., and K.O. conceived the project and designed the experiments. H.H., T.K., N.K., A.H., S.M., M.K., T.T., A.R., D.N., A.N., M.K.S., T.N., A.N., and K.O. performed the experiments. H.H., S.M., A.N., M.K.S., T.N., A.N., and K.O. analyzed the data. H.H., A.N., M.K.S., T.N., A.N., and K.O. wrote the manuscript. Competing interests: H.H. and K.O. are co-inventors on patent numbers US61/621,256, PCT/JP2013/060878, JP2014-509231, US14/390853, US9334475, and EP13773017.2 “Method for inducing erythropoietin-producing cells.” All other authors declare that they have no competing interests.
View Abstract

Navigate This Article