Research ArticleStem Cells

Kynurenine signaling through the aryl hydrocarbon receptor maintains the undifferentiated state of human embryonic stem cells

See allHide authors and affiliations

Science Signaling  25 Jun 2019:
Vol. 12, Issue 587, eaaw3306
DOI: 10.1126/scisignal.aaw3306

Kynurenine metabolism in embryonic stem cells

Kynurenine, a tryptophan metabolite generated by the enzyme IDO1, is a ligand for the aryl hydrocarbon receptor (AhR). Yamamoto et al. found that IDO1, kynurenine, and the AhR were required for self-renewal and for AhR-mediated expression of self-renewal genes in undifferentiated human embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs). Kynurenine synthesis was required for maintenance of the undifferentiated state in ESCs, and the degradation of kynurenine was required for ESCs to undergo ectodermal differentiation. In addition to identifying kynurenine metabolism as an important factor in ESC maintenance and ectodermal differentiation, these findings show that the secretion of kynurenine and the kynurenine catabolite 2-AAA are biomarkers for undifferentiated stem cells and nascent ectoderm, respectively.

Abstract

Kynurenine, which is generated from tryptophan by indoleamine 2,3-dioxygenase 1 (IDO1), binds to the aryl hydrocarbon receptor (AhR). Here, we report that kynurenine was produced by undifferentiated human embryonic stem cells (hESCs) and by induced pluripotent stem cells (iPSCs). In undifferentiated hESCs, kynurenine stimulated the AhR to promote the expression of self-renewal genes. The kynurenine-AhR complex also stimulated the expression of IDO1 and AHR, activating a positive feedback loop. Inhibition of IDO1 activity reduced the proliferation of undifferentiated ESCs but did not stimulate their differentiation. Substantial amounts of free kynurenine were present in the culture medium, providing a paracrine signal for maintenance of the undifferentiated state. Kynurenine was not present in the medium of differentiated ESCs or iPSCs. When ESCs were induced to undergo ectodermal differentiation, the abundance of kynurenine in the medium was reduced through activation of the main kynurenine catabolic pathway mediated by kynurenine aminotransferase 2 (KAT2, also known as AADAT), resulting in the secretion of 2-aminoadipic acid (2-AAA) into the culture medium. Inhibition of KAT2 activity blocked ectodermal differentiation. Thus, kynurenine metabolism plays an important role in the maintenance of the undifferentiated state and in ectodermal differentiation. Furthermore, kynurenine in the culture medium is a biomarker for the undifferentiated state, whereas the presence of 2-AAA in the culture medium is a biomarker of ESCs and iPSCs that have committed to differentiate along the ectoderm lineage.

INTRODUCTION

Human embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) share two distinct features: the potential for self-renewal and the capacity to differentiate into cells of all three germ layers. Understanding the mechanisms by which ESCs and iPSCs maintain their undifferentiated state and how specific stimuli initiate cellular differentiation is important for manipulating stem cells in vitro and for developing potential stem cell–based therapies. Several groups have reported that cytokines such as transforming growth factor–β1 (TGF-β1) and fibroblast growth factor 2 (FGF2) are essential for delivering signals that support the expression of self-renewal factors (1, 2). Shiraki et al. (3) reported that methionine in the culture medium is also essential to maintain the undifferentiated state of ESCs and iPSCs. Moussaieff et al. (4) identified a link between cellular metabolism and epigenetic changes associated with differentiation by showing that glycolysis inhibition leads to histone deacetylation and differentiation of human and mouse ESCs. These studies are insightful, but, aside from whether genes encoding self-renewal factors are expressed or repressed, there is little understanding of how the self-renewing state is maintained and how cells depart from self-renewal and initiate differentiation in the presence of differentiation stimuli. The issue is particularly important in the development of cell therapies using ESC- or iPSC-derived cell products because the inclusion of differentiation-resistant cells in the final product could permit tumor formation.

Cell culture can be used as a selection system. We have shown that a population of cells having differentiation potential can be fractionated by selecting or fine-tuning culture conditions (5). Culture parameters such as the method of cell seeding, the type of culture medium, and the substrate determine the critical attributes of cells in culture. In this study, we explored metabolic pathways and related signals used by ESCs and iPSCs in the maintenance or loss of self-renewal. By analyzing metabolites that secreted into the culture medium and corresponding gene expression profiles, we identified kynurenine as a key metabolite to maintain self-renewal and found that the reduction of kynurenine was a key event for leaving the undifferentiated state in ESCs and iPSCs.

RESULTS

Secreted metabolites are markers for undifferentiated and early-differentiating ESCs and iPSCs

To permit investigation of the differentiation process of well-characterized undifferentiated cells, rather than cells of heterogenous differentiation status, as observed in feeder cell clump passaging (5), we seeded single cells of the human ESC and iPSC lines H9 and PFX#9, respectively, and cultured them in Essential 8 (Es8) medium on vitronectin N-terminal domain (VTN-N)–coated dishes. We set up a cytokine-induced lineage-specific differentiation system for ESCs and iPSCs in which ectodermal differentiation was induced with the TGF-β1 signaling inhibitors Noggin and the activin receptor–like kinase inhibitor SB431542, mesodermal differentiation was induced with bone morphogenetic protein 4 (BMP4), and endodermal differentiation was induced with activin-A, BMP4, and WnT3A (Wnt family members 3A) (Fig. 1A) to identify characteristic metabolites specific for the undifferentiated state and metabolites indicative of cellular differentiation to ectodermal, mesodermal, and endodermal lineages (68).

Fig. 1 Identification of secreted metabolites that are markers for undifferentiated or early differentiating ESCs and iPSCs.

(A) The protocol for culture, induction of differentiation, and supernatant collection for the analysis of metabolites in human H9 ESCs and PFX#9 iPSCs. Cells were seeded in chemically defined medium (Es8 for H9 cells and TeSR-Es8 for PFX#9 cells) and a ROCK inhibitor on dishes coated with recombinant human (rh) VNT-N. The cells were untreated or treated with the indicated cocktails of differentiation factors specific for each germ layer. The medium was removed for analysis each day and replaced with fresh medium for 6 days. The cell numbers increased exponentially during the experiment because the cultures were not split during medium replacement. The harvested medium was subjected to LC-MS/MS analysis for 95 metabolites (table S1). (B) Time-course profiles of the amounts of tryptophan, kynurenine, lysine, 2-AAA, lactic acid, and methionine relative to an internal control in supernatants from H9 ESCs (H9/Es8) and PFX#9 iPSCs (PFX#9/TeSR-E8) in the absence of differentiation factors (Undiff) and in the presence of ectodermal (Ecto), mesodermal (Meso), or endodermal (Endo) differentiation cocktails. Blank, medium only. n = 3 independent experiments. (C) Comparison of two different quantification methods for four metabolites in undifferentiated or differentiated cultures of H9 ESCs in Es8 medium. Metabolites were measured relative to an internal standard (the ratio of the area under the peak for each metabolite to the area under the peak for 2-isopropyl malic acid) and by an external calibration method using a standard calibration curve of known concentrations of each metabolite. n = 3 independent experiments. (D) The average amounts of kynurenine and 2-AAA secreted from a single H9 cell per hour in Es8 medium were estimated by the external calibration method for days 1 to 3 and days 4 to 6 [in μM per cell per hour = nmol (in 2 ml) per cell per hour: normalization]. n = 3 independent experiments.

We evaluated the morphology of H9 cells cultured in Es8 medium in the presence of specific cytokine combinations (fig. S1A) and the gene expression profiles of these cells during lineage induction at days 3 and 6 (fig. S1B). Spinous cells were observed in the rim of the cell colonies differentiated into ectoderm, whereas cobblestone-like cells were observed in the colonies differentiated into mesoderm or endoderm. Secreted metabolites in culture medium were measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS) on days 1 to 6 (fig. S1C). Heat maps created by hierarchical clustering analysis showed several differentiation lineage-specific metabolites. We applied principal components analyses (PCAs) and volcano plot analyses to compare metabolites in the culture medium of undifferentiated H9 cells versus H9 cells induced to differentiate into ectodermal, mesodermal, or endodermal lineages (fig. S1, D and E). The same experiments were conducted with PFX#9 cells cultured in TeSR-E8 medium (fig. S2, A to E). PCA showed that specific secreted metabolites of undifferentiated or lineage-differentiated cells could be identified by days 3 and 4. Volcano plot analysis of culture medium from undifferentiated versus ectoderm-differentiated, mesoderm-differentiated, or endoderm-differentiated cells showed that the presence of kynurenine in the culture medium was a marker for ESCs and iPSCs in an undifferentiated state and that 2-aminoadipic acid (2-AAA) in cell culture supernatants was a marker for cells committed to ectoderm differentiation. Among the metabolites measured, we did not detect a specific marker for mesodermal or endodermal differentiation of ESCs and iPSCs.

Kynurenine is a metabolite in the tryptophan catabolic pathway, whereas 2-AAA is generated as an intermediate metabolite in the lysine catabolic pathway or alternatively from α-ketoadipic acid (AKA) in the kynurenine catabolic pathway by kynurenine aminotransferase 2 [KAT2; also known as kynurenine or α-aminoadipate aminotransferase] (9). Time-course profiles of tryptophan, kynurenine, 2-AAA, and lysine secretion from H9 cells in Es8 medium and PFX#9 cells in TeSR-E8 medium (Fig. 1B) showed that ESCs and iPSCs took up tryptophan and secreted kynurenine when the cells were maintained in the undifferentiated state. The cells continued to take up tryptophan to a lesser extent and secreted 2-AAA when they committed to ectodermal differentiation, although the consumption of lysine was not specific to ectodermal differentiation. Undifferentiated ESCs and iPSCs activate the glycolytic pathway and secrete lactic acid into the culture medium (10). Secretion of lactic acid reduced as the cells differentiated (Fig. 1B), which agrees with a previous report (11). Although the consumption of methionine was previously reported to indicate the activity of a metabolic pathway specific to ESCs and iPSCs in the undifferentiated state (3), we observed comparable methionine consumption across the undifferentiated and the lineage-committed cells (Fig. 1B), suggesting that the consumption of methionine is unlikely to be specific to ESCs and iPSCs in the undifferentiated state during a short monitoring period.

In our LC-MS/MS analytic system, the amount of each targeted metabolite was determined using a dimensionless quantity called the area ratio obtained from the integrated peak areas of targeted compounds divided by that of the internal standard (Fig. 1C, internal standard). The area of the detected peaks varied depending on the components of the culture medium, and thus, area ratios varied with the culture system. The amount of each metabolite was also quantified using a calibration curve made with known concentrations of the metabolite of interest (Fig. 1C, external calibration). We found a good correlation between a curve made by an internal standard and that by an external calibration. A time-course profile of each metabolite (Fig. 1B) was generated with a culture supernatant from whole cells in culture. To estimate the amount of a metabolite secreted by an individual cell, we divided the calculated average area ratio with the calculated average cell number in the culture. The marked secretion of kynurenine by a single H9 cell in 1 hour was observed on both days 1 to 3 and days 4 to 6 when cells were maintained in the undifferentiated state, and this was attenuated because the cells commenced differentiation (Fig. 1D). In addition, the marked secretion of 2-AAA from a single cell in 1 hour was observed when cells were committed to ectoderm lineage differentiation (Fig. 1D).

In subsequent metabolic studies, we conducted LC-MS/MS analyses using only an internal standard for convenience. We preferred this approach because it eliminated the need to prepare an individual calibration curve specific to the metabolite of interest each time we measured it by LC-MS/MS. We generally used a chemically defined medium lacking bovine serum albumin (BSA) or human serum albumin (HSA) for the extraction of candidate metabolites. This method eliminated the high analytical background associated with high amounts of proteins in the medium. To control for the absence of BSA, we determined whether kynurenine and 2-AAA were accurate indicators of ESC and iPSC differentiation states when the cells were cultured in BSA-containing medium. Associations of kynurenine with the undifferentiated state and 2-AAA with ectodermal differentiation were also observed for PFX#9 iPSCs cultured in the BSA-containing medium mTeSR1 (fig. S3, A to D). For subsequent analytical studies, we used ESCs and a chemically defined medium to minimize analytical background.

Kynurenine generated by IDO1 mediates signaling that maintains the undifferentiated state

The uptake of tryptophan and the secretion of kynurenine are specific metabolic events observed during the culture of undifferentiated ESCs and iPSCs. In ESCs and iPSCs, there are three oxidoreductases that generate N-formyl kynurenine, a precursor of kynurenine, from tryptophan: indoleamine 2,3-dioxygenase 1 (IDO1), IDO2, and tryptophan 2,3-dioxygenase (TDO2). Among these enzymes, the major oxidoreductase in ESC H9 was IDO1, as determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR; Fig. 2A). To determine whether the tryptophan catabolic pathway leading to the secretion of kynurenine was critical for the proliferation and maintenance of ESCs in the undifferentiated state, we added a specific inhibitor of IDO1, INCB024360, which is also known as epacadostat (12), to H9 ESCs maintained on VNT-N–coated dishes in Es8 medium with no differentiation-inducing cytokines or inhibitors (Fig. 2B). Cells under these culture conditions were undifferentiated, as indicated by gene expression profiles at day 3, as determined by a qRT-PCR scorecard panel (Fig. 2C). However, the addition of the IDO1 inhibitor greatly suppressed the growth of H9 cells (Fig. 2D) and reduced both the secretion of kynurenine over 3 days of culture (Fig. 2E) and the secretion of kynurenine by a single cell over a period of 1 hour (Fig. 2F). These results suggest that the activity of the tryptophan-kynurenine catabolic pathway, which is mediated by IDO1, is important for the proliferation and the maintenance of the undifferentiated state of ESCs. In support of this, the addition of kynurenine 1 day after IDO1 inhibitor treatment (fig. S4A) resulted in recovery of the number of undifferentiated H9 cells (fig. S4, B and C).

Fig. 2 Kynurenine generated by IDO1 mediates a signal that maintains the undifferentiated state.

(A) Expression of IDO1, IDO2, and TDO2 in undifferentiated H9 cells cultured on VNT-N in Es8 medium as determined by qRT-PCR and shown as the fold increase compared with that of TDO2. n = 3 independent experiments. (B) Protocol for the addition of the IDO1 inhibitor INCB024360 analog (IDO1i) or vehicle [dimethyl sulfoxide (DMSO)] to the H9 culture system. Media were changed every day. (C) Gene expression profiles of H9 cells treated with IDO1i or DMSO as determined by qRT-PCR scorecard panel. n = 3 independent experiments. (D) Morphology of H9 cells that were untreated or treated with vehicle (DMSO) or IDO1i at day 3. Cell counts in one well of a six-well plate are noted in each image. n = 3 independent experiments. Scale bar, 200 μm. (E) Time-dependent changes in the concentration of kynurenine, 2-AAA, and tryptophan in culture medium of H9 cells treated with vehicle or IDO1i, as measured by LC-MS/MS. Blank, medium only. n = 3 independent experiments. (F) The amount of kynurenine secreted from a single H9 cell in a 1-hour period under the indicated conditions was estimated from the average of the normalized values and cell numbers from day 1 to day 3 (D1 to D3). A representative result is shown. n = 3 independent experiments.

The tryptophan-IDO1-kynurenine-AhR cascade maintains the undifferentiated state

Kynurenine reportedly exerts its biological effect through its binding to the aryl hydrocarbon receptor (AhR) (13), which is present in the cytoplasm and translocates into the nucleus, where it controls the expression of target genes, upon ligand binding (14). Mouse Ahr and human AHR are both expressed in embryos and ESCs, but their gene expression profiles at developmental stages are noticeably different (fig. S5). To examine the biological response mediated by the kynurenine-AhR complex in ESCs, we added to cell cultures an AhR inhibitor (CH223191) that is known to block kynurenine-AhR binding (Fig. 3A) (15). The cells were cultured in Es8 medium on VTN-N, conditions that allow only undifferentiated cells to grow as evidenced by gene expression profiles (Fig. 3B). The addition of CH223191 reduced cell growth (Fig. 3C), as well as tryptophan uptake and kynurenine secretion (Fig. 3D).

Fig. 3 The IDO1-kynurenine-AhR loop sustains the undifferentiated state of ESCs.

(A) Protocol for the addition of the AhR inhibitor CH223191 to the H9 culture system. DMSO, vehicle control. Medium was changed daily. (B) Gene expression profiles of H9 cells treated with CH223191 or DMSO at day 3. n = 3 independent experiments. (C) Morphology of H9 cells on day 3. The number in each image represents the cell count in one well of six-well plate. Scale bar, 200 μm. Data are representative of n = 3 independent experiments. (D) Changes in kynurenine and tryptophan concentrations over time in culture medium of H9 cells with or without CH223191. Blank, medium only. n = 3 independent experiments. (E) Expression of IDO1 and AHR in H9 cells 3 days after adding CH223191 and 6 days after adding the IDO1 inhibitor to the culture. Transcript abundances were determined by qRT-PCR and are shown as the fold change of expression in untreated cells. n = 3 independent experiments. (F) Expression of IDO1, IDO2, TDO2, and AHR, as determined by qRT-PCR, in undifferentiated H9 cells and at days 3 and 6 after induction of ectodermal differentiation. Data are shown as the fold change in IDO1 expression in H9 cells during ectoderm induction day 3 or AHR in untreated, undifferentiated H9 cells. n = 3 independent experiments. (G) Representative Western blot for IDO1, IDO2, and TDO2 in undifferentiated H9 cells. n = 3 independent experiments.

These data suggested that the maintenance of undifferentiated ESCs was mediated by signals generated by the tryptophan-IDO1-kynurenine-AhR pathway. To test this hypothesis, we measured the expression of AHR and IDO1 in H9 cells treated with either the IDO1 inhibitor INCB024360 or the AhR inhibitor CH223191 (Fig. 3E). The observed reduction in AhR expression by the addition of INCB024360 and the reduction in IDO1 expression by the addition of CH223191 (Fig. 3E) support the idea that the tryptophan-IDO1-kynurenine-AhR cascade delivers a sustainable signal that maintains the undifferentiated state of ESCs and that tryptophan is acquired from the culture medium. It took 3 to 6 days before the presence of the IDO1 inhibitor elicited a suppressive effect on AHR expression. The amount of kynurenine present in the cytoplasm and culture medium before the addition of the IDO1 inhibitor was likely high enough (Fig. 2F) to support kynurenine-AhR signaling for a substantial period until the abundance of kynurenine fell below a threshold. When H9 cells committed to ectodermal differentiation after the addition of Noggin and an inhibitor of TGF-β1 signaling (Fig. 1A and fig. S1B), a strong reduction in the expression of both IDO1 and AHR was evident 3 days after addition of differentiation cocktail (Fig. 3F). Western blotting indicated that IDO1 was present in undifferentiated H9 cells, but IDO2 and TDO2 were not (Fig. 3G).

The kynurenine-AhR complex mediates a signal that supports self-renewal

Kynurenine was detected by immunohistochemistry in both the cytoplasm and the nucleus when H9 cells were in the undifferentiated state, and its abundance reduced in cells treated with the AhR inhibitor CH223191, or when cells committed to ectodermal differentiation (Fig. 4A). In the same context, AhR in the cytoplasm and the nucleus was reduced by the addition of the IDO1 inhibitor (Fig. 4B). These observations support the qRT-PCR results (Fig. 3E) and suggest the presence of an IDO1-kynurenine-AhR-IDO1 loop. Our qRT-PCR experiments showed that expression of AHR in H9 cells was suppressed when cells committed to ectodermal differentiation (Fig. 3F). In agreement, flow cytometric analysis showed that the abundance of AhR reduced when the cells differentiated into ectoderm (Fig. 4C). Immunoprecipitation followed by Western blotting indicated an association between kynurenine (25 kDa) (16) and AhR in undifferentiated cells (Fig. 4D). Overexpressed AhR fused to green fluorescent protein (AhR-GFP) preferentially accumulated in the nucleus in undifferentiated H9 cells but not when the cells were treated with the IDO1 inhibitor (Fig. 4E). These results suggested that formation of the kynurenine-AhR complex is essential for the delivery of self-renewal signaling. We next performed chromatin immunoprecipitation (ChIP)–qPCR on AhR and control [immunoglobulin G (IgG)] immunoprecipitates from H9 cells that were untreated or treated with the IDO1 inhibitor using primers directed against promoter and enhancer regions of POU5F1 (also called OCT4), NANOG, CYP1A1, EP300, and IDO1. The AhR complex preferentially bound to putative promoter and enhancer regions of POU5F1, NANOG, and IDO1 genes and to the promoter regions of AHR, CYP1A1, and EP300 in both IDO1 inhibitor–treated and vehicle-treated cells (Fig. 4F). Although treatment with the IDO1 inhibitor reduced the number of undifferentiated cells in culture, the profile of AhR binding sites in the culture was essentially the same. This supports the idea that the self-renewal state in ESCs is maintained by self-sustaining IDO1-kynurenine-AhR-IDO1 and IDO1-kynurenine-AhR-AHR loops that are distinct from the signal transduction pathway mediated by TGF-β1 and FGF2 (1, 2). TGF-β1 and FGF2 required continuous or periodic application to sustain cell growth (fig. S9A) and maintain the expression of self-renewal genes in H9 cells (fig. S9B).

Fig. 4 The kynurenine-AhR complex mediates a signal that supports self-renewal.

(A) Immunohistochemical detection of the intracellular distribution of kynurenine (red) in undifferentiated (Undiff.) H9 cells treated with vehicle only (DMSO) or with the AhR inhibitor CH223191 for 24 hours and at day 6 after ectoderm induction. Nuclei are stained with DAPI (4′,6-diamidino-2-phenylindole) (blue). Scale bars, 5.0 and 2.5 μm (zoom). (B) Immunohistochemical detection of the intracellular distribution of AhR (red) in undifferentiated H9 cells treated with DMSO alone or with the IDO1 inhibitor INCB024360 analog (IDO1i) for 24 hours. Nuclei are stained with DAPI (blue). n = 3 independent experiments. Scale bar, 5.0 μm. (C) Abundance of AhR in undifferentiated H9 cells or in H9 cells 6 days after ectoderm induction as determined by flow cytometry. Undifferentiated H9 cells stained with an isotype control antibody were used as control. n = 3 independent experiments. (D) Western blot for kynurenine in lysates of undifferentiated H9 cells before and after immunoprecipitation (IP) with an antibody against AhR or a control IgG antibody. n = 3 independent experiments. (E) GFP fluorescence in H9 cells expressing AhR-GFP or GFP alone (control) and treated with DMSO or the IDO1 inhibitor INCB024360 analog (IDO1i). n = 3 independent experiments. Scale bar, 10 μm. (F) ChIP-qPCR assay for POU5F1, NANOG, IDO1, AHR, CYP1A1, and EP300 in DNA immunoprecipitated from control and IDO1i-treated H9 cells using an antibody recognizing AhR or a control antibody and primers recognizing the promoters or downstream regions of each gene. qPCR values using primers for the promoter regions in the AhR immunoprecipitate were compared to those in the control immunoprecipitant and shown as fold enrichment (orange bars). qRCR values using primers for the downstream regions in the AhR immunoprecipitant were compared to those of the control immunoprecipitant and shown as fold enrichment (black bars). n = 3 independent experiments with three technical replicates per experiment. Gene expression profiling of the cells used in this experiment is also shown.

The addition of kynurenine 1 day after application of the IDO1 inhibitor (fig. S4A) rescued the reduction in undifferentiated cells by the IDO1 inhibitor (fig. S4, B and C), supporting the notion that this autonomous loop is essential to support the undifferentiated state in ESCs. Excess kynurenine not bound to AhR is likely secreted into the culture medium, where it acts as a long-lived self-renewal signal. Therefore, kynurenine is a biomarker for the presence of undifferentiated ESCs and iPSCs. This model also implies that ESCs and iPSCs must reduce the amount of kynurenine in the cytoplasm to leave the undifferentiated state. Accordingly, we observed the secretion of 2-AAA, a product of kynurenine catabolism, within 2 days of stimulating ectodermal differentiation (Fig. 1, B and C). However, kynurenine abundance did not drop to background levels until 4 days after stimulating ectodermal differentiation (Fig. 1, C and D). Thus, kynurenine-AhR continues signaling after kynurenine has started to degrade and does not stop until kynurenine abundance drops to the background.

Secretion of 2-AAA is a benchmark for the initiation of ectodermal differentiation

2-AAA is generated either by the aldehyde dehydrogenase 7 family member A1 (ALDH7A1) in the lysine catabolic pathway or from AKA in the kynurenine catabolic pathway by KAT2 (17). Lysine consumption rates were independent of cell differentiation status and the culture system (Fig. 1B), which suggested that the lysine catabolic pathway might not be responsible for the secretion of 2-AAA when cells committed to ectodermal differentiation. Therefore, we examined the role of KAT2 in ectodermal differentiation by using a KAT2 inhibitor (KI:PF-04859989, referred to hereafter as KI) (9). KI was added to H9 cell culture medium together with the ectoderm-inducing factors Noggin and SB431542 (Fig. 5A). KI moderately inhibited ectodermal differentiation of H9 cells (Fig. 5B and fig. S6). KI also perturbed ectoderm differentiation even in cells already committed to differentiation (fig. S7, A to D). The addition of KI reduced the secretion of 2-AAA by single H9 cells (Fig. 5C) and whole cultures (Fig. 5D) treated with Noggin and SB431542. Experimental results showed that ectoderm differentiation, as characterized by gene expression profiles, was associated with a reduction in the abundance of kynurenine in the culture medium (Figs. 1, B and D, and 4A) and with the KAT2-dependent secretion of 2-AAA into the culture medium (Fig. 5D). Together, it would be reasonable to propose that KAT2-mediated kynurenine catabolism was responsible for the reduction in cytoplasmic kynurenine; thus, KAT2 supports the initiation of ectoderm differentiation by attenuating the signal mediated by the kynurenine-AhR complex. However, the addition of KI (PF-04859989) could not completely block ectodermal differentiation process at gene expression level (Fig. 5D and figs. S6 and S7, C and D). Therefore, cells cultured with ectoderm-inducing factors and KI at days 2 to 4 were ectoderm lineage–committed cells and ceased to secrete kynurenine into culture medium. Because culture medium was changed every day with fresh medium containing ectoderm-induction molecules, we could not expect to detect kynurenine even by the addition of KI in the culture (Fig. 5D).

Fig. 5 Secretion of 2-AAA into the culture medium is a biomarker for ectoderm differentiation.

(A) Protocol for testing the effect of the KAT2 inhibitor PF-04859989 (KI) on ectodermal differentiation of H9 cells. (B) Gene expression profiles of H9 cells treated with ectoderm stimuli in the presence or absence of KI were determined by qRT-PCR scorecard panel. n = 3 independent experiments. (C) The amount of 2-AAA secreted from a single H9 cell in 1 hour (in area ratio per cell per hour) under the indicated conditions was determined from the normalized values from day 1 to day 4 (D1 to D4) of culture. n = 3 independent experiments. (D) Quantification of the abundances of 2-AAA, kynurenine, tryptophan, lactate, and lysine in culture medium from day 1 to day 4 of ectodermal differentiation was determined by LC-MS/MS. Culture medium was replaced with fresh medium every day. n = 3 independent experiments.

The kynurenine catabolic pathway is involved in the initiation of ectodermal differentiation

Given that AKA is an intermediate metabolite in the main kynurenine catabolic pathway and is converted to 2-AAA by KAT2, we asked whether ectodermal differentiation could sequentially trigger the activation of the main kynurenine catabolic pathway from kynurenine to AKA and the production of 2-AAA through the enzymatic activities of kynurenine 3-monooxidase (KMO), kynureninase (KYNU), 3-hydroxyanthranilic acid 3,4-dioxygenase (HAAO), α-amino-β-carboxymuconate-semialdehyde decarboxylase (ACMSD), and oxoglutarate dehydrogenase (OGDH) (17). Because there are no reliable assay systems to measure all of these enzymatic activities or technology to measure the accumulation of intermediate metabolites in the cytoplasm, we measured expression of the genes encoding these enzymes by qRT-PCR. Expression of each gene increased over time as cells committed to ectoderm differentiation (Fig. 6A). Expression of ALDH7A1, which mediates the generation of 2-AAA in the lysine catabolic pathway, was not changed by ectoderm differentiation stimuli (Fig. 6A). On the basis of these results and the observation that lysine consumption did not increase during differentiation (Fig. 1B), we postulate that the generation of 2-AAA during ectodermal differentiation is due to the activation of the main kynurenine catabolic pathway mediated by KAT2. KAT2 is also known to mediate the conversion of kynurenine into kynurenate and 3-hydroxy kynurenate into xanthurenic acid, besides conversion of AKA into 2-AAA. The concentration of kynurenate, 3-hydroxy kynurenate, or xanthurenic acid in culture medium before or after ectoderm induction was determined by LC-MS/MS. 3-Hydroxy kynurenate or xanthurenic acid was not detected before or after ectoderm induction only a small amount of kynurenate before ectoderm differentiation was detected, but it ceased to be detected after induction (fig. S10). Supported by this experiment, we presumed that KAT2 activates the main kynurenine catabolic pathway that uses kynurenine as a starting substrate and produces 2-AAA as a final product.

Fig. 6 The kynurenine catabolic pathway is involved in the initiation of ectodermal differentiation.

(A) Gene expression profiles (qRT-PCR scorecard panels) and expression of genes in the kynurenine catabolic pathway (HAAO, KMO, KYNU, ACMSD, OGDH, and KAT2) and the lysine catabolic pathway (ALDH7A1) in undifferentiated H9 cells and at days 3 and 6 of ectodermal differentiation. Gene expression was measured by qRT-PCR and shown as the fold change compared to the undifferentiated state. n = 3 independent experiments. (B) Photos and gene expression profiles of H9 cells at day 6 of ectodermal differentiation in the absence (Ecto diff.) or presence of 100 μM 2-AAA (Ecto diff. + 2-AAA). Undifferentiated H9 cells with no treatment were used as a control. The number of cells in one well of a six-well plate for each condition is noted in each image. n = 3 independent experiments. Scale bar, 200 μm. (C) Putative schema for metabolic pathways in undifferentiated ESCs and iPSCs and in ESCs and iPSCs committed to ectodermal differentiation. Kynurenine produced from tryptophan by IDO1 binds to AhR in the cytoplasm, and the kynurenine-AhR complex translocates to the nucleus, where it promotes the transcription of self-renewal factors, IDO1, and AHR to maintain self-renewal and the IDO1-kynurenine-AhR loop. Excess kynurenine not used for signaling is secreted into the culture medium, where it acts as a sustainable self-renewal signal. When cells commit to ectodermal differentiation, the kynurenine pool decreases through the activation of the kynurenine catabolic pathway. KAT2 converts AKA to 2-AAA, which is secreted instead of being used for additional pathways. Gene expression profiles and relevant time-lapse metabolic analyses of kynurenine and 2-AAA with LC-MS/MS are included in the schema.

Next, we addressed the biological role of 2-AAA in ectoderm differentiation. The concentration of 2-AAA in the culture medium of H9 cells at day 3 after the induction of ectoderm differentiation was 6 to 8 μM (Fig. 1C). We examined the role of 2-AAA by adding 10, 50, or 100 μM 2-AAA to H9 cells in Es8 medium and evaluating gene expression profiles by qRT-PCR scorecard panel. Even 100 μM 2-AAA failed to induce ectodermal differentiation of unstimulated H9 cells or to enhance ectodermal differentiation of cells stimulated with Noggin and SB431542, although it moderately reduced the cell number (Fig. 6B). This experiment provided supportive evidence that 2-AAA could be a final product that is not further metabolized and that overdosing of 2-AAA attenuated the main kynurenine catabolic pathway, leading to mild suppression of steady cell growth in ectoderm differentiation. 2-AAA is known as an intermediate metabolite in the lysine catabolic pathway, leading to the generation of acetyl-CoA (coenzyme A). 2-AAA is generated from L-2-aminoadipate δ-semialdehyde by ALDH7A1 and catabolized to AKA (17). Because ALDH7A1 was not induced (Fig. 6A) and the lysine catabolic pathway did not seem to be activated, we believe that 2-AAA was produced only or mainly by KAT2 from AKA and accumulated without being further metabolized. Being a final product makes 2-AAA a good biomarker for cells that initiate ectodermal differentiation. The background noise (detection limit) for 2-AAA with LC-MS/MS is 0.02 by area ratio, which corresponds to 0.7 μM determined with an external standard calibration method. Therefore, using LC-MS/MS, we observed that 2-AAA concentrations exceeding 0.7 μM induced ectodermal differentiation mediated by KAT2 activation. Furthermore, ectodermal differentiation was fully activated when most of the cells leave the undifferentiated state, which occurs 3 days after the addition of ectoderm-inducing stimuli. At that point, kynurenine was barely detected in the culture medium of differentiated cells (Fig. 1C). If the kynurenine level fell below 0.2 by area ratio, then most of the cultured cells were differentiated (Fig. 1B). Together, our data suggested that the kynurenine metabolic pathway is likely to be involved in the maintenance of the undifferentiated state, whereas the kynurenine catabolic pathway is involved in the initiation of ectoderm differentiation in ESCs (Fig. 6C).

Kynurenine and 2-AAA in culture medium are robust markers for the assessment of the differentiation state of cultured PSCs

The data above showed that measuring the concentration of kynurenine in culture medium is useful for monitoring the undifferentiated state of ESCs and iPSCs, that the amount of lactic acid in culture medium provides an index for the growth of undifferentiated cells that predominantly use the glycolytic pathway, and that the presence of 2-AAA in culture medium is a marker for the commitment to differentiate into ectoderm. We explored the use of these markers during the differentiation of H9 ESCs in embryoid body (EB) formation (Fig. 7A), monitoring both the morphology (Fig. 7B and fig. S8) and gene expression profiles (Fig. 7C) at days 3 and 6. In EB formation cultures, kynurenine and lactic acid secretion decreased over the 6 days, whereas 2-AAA secretion increased over the same time period (Fig. 7D). These data indicated that H9 cells in culture had exited self-renewal (decreased kynurenine and lactic acid) and initiated ectodermal differentiation (increased 2-AAA) during EB formation. The changes were apparent within the first 3 days of EB formation, indicating that these biomarkers can be used to predict the onset of differentiation of ESCs in EB formation cultures at a very early stage. The ratios of kynurenine to 2-AAA in the EB assay culture medium decayed exponentially (Fig. 7D), supporting our finding that the reduction in kynurenine during ectoderm differentiation is predominately mediated by the main kynurenine catabolic pathway, starting with the consumption of reserved kynurenine and ending with the production and secretion of 2-AAA when de novo production of kynurenine is blocked by the attenuation of IDO1 activity.

Fig. 7 Kynurenine and 2-AAA in the culture medium are robust markers for the assessment of the differentiation state of cultured stem cells.

(A) Protocol for EB formation. H9 cells were transferred to low-attachment plates in Essential 6 (Es6) medium at day 0. The medium was removed daily for LC-MS/MS analysis and replaced with fresh medium. (B) Phase contrast imaging of EBs at days 3 and 6. n = 3 independent experiments. Scale bar, 200 μm. (C) Gene expression profiles of EBs at days 3 and 6 as determined by a qRT-PCR scorecard panel. n = 3 independent experiments. (D) Quantification of kynurenine, 2-AAA, and lactic acid and the ratio of kynurenine to 2-AAA secreted into the culture medium by EBs over days 1 to 6 by LC-MS/MS. n = 3 independent experiments with three technical replicates per condition for each time point.

DISCUSSION

We showed the involvement of the IDO1-kynurenine-AhR-AHR and IDO1-kynurenine-AhR-IDO1 autonomous loops in the maintenance of self-renewal in ESCs and iPSCs. Signal transmission by the AhR complex is a unique signaling system that directly links a metabolite in the cytoplasm to the transcriptional machinery. The AhR is an evolutionarily conserved basic-helix-loop-helix Per-Arnt-Sim transcription factor (14). Various internal and external ligands bind to several binding pockets in the AhR, generating various types of ligand-AhR complexes that translocate to the nucleus and bind to the promoters or enhancers of specific target genes together with other transcriptional regulatory molecules. The transcriptional output varies depending on the ligand, transcriptional cofactors, and cell type. The tryptophan-derived metabolite ITE [2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid methyl ester] can bind to the AhR and enhances AhR binding to the promoter of POU5F1, thus suppressing the expression of this target gene transcription in cancer cells (18). Kynurenine is also a tryptophan derivative, and the kynurenine-AhR complex exerts distinct biological effects. This characteristic is a core feature of the AhR as a hub molecule that can deliver a range of signals by binding to and shuttling various molecules to different DNA binding sites.

Another issue that merits discussion is the fact that AhR complex signals are species specific. Ahr−/− mice are born with a normal appearance despite the fact that several organs are dysfunctional (19). There was a marked difference in the gene expression profiles of mouse Ahr in embryos, ESCs, and EBs and human AHR in embryos, ESCs, iPSCs, and EBs (fig. S5). Expression of Ahr in mouse peaks at the one-cell stage, is then suppressed throughout early embryonic stages up to the blastocyst stage and then increases again after implantation. Ahr expression is repressed in mouse ESCs. In contrast, human AHR is maintained throughout early embryonic stages and in ESCs, iPSCs, and EBs. Further, mouse ESCs are naïve, equivalent to preimplantation inner cell mass cells, whereas human ESCs are similar to mouse stem cells derived from the epiblast only after implantation. Ko et al. (20) argue that repression of Ahr expression is required to prevent loss of ESC pluripotency. However, the expression of Ahr in mouse blastocysts and ESCs is naturally very low. The biological role of kynurenine-Ahr signaling in normal mouse ESCs might not be important, and this could explain why some Ahr−/− mice are viable.

It has been reported that some cancer cells exclusively use the tryptophan-kynurenine-AhR pathway to promote proliferation by inhibiting the activation of tumor-suppressive genes and prodifferentiation genes (13, 21). There is a notable similarity between cancer cells and undifferentiated ESCs and iPSCs, especially in their ability to proliferate without differentiating. However, the kynurenine-AhR complex has different transcriptional targets in cancer cells and in ESCs and iPSCs. Moreover, the loci related to the restriction of proliferation and promotion of differentiation are hypermethylated in cancer cells but bivalently methylated in ESCs and iPSCs (22). This could explain the unique characteristic of ESCs and iPSCs simultaneously having the capacities for both self-renewal and differentiation. The kynurenine secreted from cancer cells is reported to modulate the immunosurveillance system of the host, promoting cancer cell survival and tumor progression (13). A similar mode of immunosurveillance modulation is observed in mammalian conceptuses, which are genetically distinct from the mother (allogenic). They are protected from T cell–mediated rejection by tryptophan catabolites that are present in the trophoblast (23). Thus, both cancer cells and embryos establish a microenvironment conducive to survival and proliferation. Similarly, the secretion of kynurenine by ESCs and iPSCs in culture generates an environment that preserves self-renewal without being affected by incidental differentiation stimuli such as a temporary nutrient deprivation or contact inhibition resulting from overgrowth. It is important to understand that cell culture is a cell selection system: Only the cells that are fit for a given culture can survive and proliferate. Thus, critical qualities of the cells are mostly determined by the specifics of a given cell culture system and culture conditions.

It has been reported that TGF-β1 and FGF2 signalings synergize to inhibit the BMP signaling that triggers differentiation and to sustain the expression of self-renewal circuitry genes such as NANOG, POU5F1, and SOX2 (1, 2). There are several reports stating that there are interactions between TGF-β1– or FGF2-mediated and AhR-mediated signals (2427). However, these reports found the signal interaction in Sertoli cells, fibroblasts, myoblasts, and hepatocytes but not in ESCs or iPSCs. We believe that kynurenine cooperates with TGF-β1 and FGF2 to maintain ESCs and iPSCs in the undifferentiated state. Our experiments suggested that ESCs and iPSCs cannot be maintained in the undifferentiated state without FGF2 or TGF-β1 (fig. S9) or tryptophan (3). How these signals work cooperatively is an open question. The cytokine-mediated signals are rather transient. Consequently, periodic (repeated) stimulation is needed to maintain the signaling, whereas the kynurenine-AhR complex delivers a sustainable signal to maintain the undifferentiated state through the IDO1-kynurenine-AhR-AHR and IDO1-kynurenine-AhR-IDO1 loops. We believe that these signals can work synergistically in a complementary fashion because of differences in their mode of actions.

Ectodermal differentiation is the first observed differentiation event in EBs (Fig. 7, C and D) and in embryogenesis. In this report, we provide several lines of evidence that ectodermal differentiation of ESCs is coupled to the activation of the main kynurenine catabolic pathway leading to the production of 2-AAA. There are several reports that the kynurenine pathway of tryptophan catabolism produces the potent neurotoxic metabolites quinolinate (28) and kynurenate (29, 30). A strong suppression of kynurenate production is coupled with the increased production of 2-AAA in neonate rat brain tissue, which supports neurite outgrowth after birth (29, 31). A plausible explanation for the activation of the main kynurenine catabolic pathway during normal differentiation of ESCs would be that ESCs must shut off the production of neurotoxic metabolites such as kynurenate or quinolinate generated in the branch pathway by activating the main KAT2-dependent kynurenine catabolic pathway for normal neural development, which requires ectoderm differentiation. Kynurenate, which has neurotoxic properties, was detected in the culture medium of undifferentiated cells, but it was no longer detectable when the cells differentiated (fig. S10). Given that KAT2 has versatile functions depending on the developmental stage and neural cell type (32), further study will be required to elucidate the mechanism by which KAT2 activity is suppressed or fine-tuned in the undifferentiated state and then activated upon ectodermal differentiation. Those findings will provide further insights into sequential developmental events occurring in embryos after implantation.

MATERIALS AND METHODS

Cell culture

Two pluripotent stem cell lines were used: the ESC line H9 (WiCell) (33) and the iPSC line PFX#9 [Foundation for Biomedical Research and Innovation (FBRI)] (34). Single-cell suspensions of H9 and PFX#9 cells were seeded at 1 × 105 per well in a six-well plate (Corning) coated with recombinant human VTN-N (Thermo Fisher Scientific) and cultured with TeSR-E8 (4 ml per well) or Es8 medium. The culture medium was changed every day, and the cells were cultured in an incubator (Model 3110, Thermo Fisher Scentific) at 37°C in a 5% CO2 atmosphere. Three days after seeding, cells were harvested using TrypLE Select (Thermo Fisher Scientific) for single-cell passaging.

Morphologies were observed by phase contrast microscopy (IX81, Olympus). EB formation was conducted by culturing iPSCs or ESCs with Es6 medium (Thermo Fisher Science) that lacked FGF2 and TGF-β1 on an ultralow-attachment surface dish (Corning) for 14 days. Reagents such as 20 μM IDO1 inhibitor (INCB024360, Cosmo Bio), 5 μM AhR inhibitor (CH223191, Sigma-Aldrich), and 11.5 μM KAT2 inhibitor (PF-04859989, Sigma-Aldrich) were added to Es8 culture medium of iPSCs and ESCs for inhibitor supplementation experiments.

Quantitative reverse transcription polymerase chain reaction

Total RNA was extracted using the RNeasy Micro Kit (QIAGEN) according to the manufacturer’s instructions. qRT-PCR analysis was performed using the TaqMan hPSC Scorecard Panel (Thermo Fisher Scientific) containing 96 genes according to the manufacturer’s instructions. Interpretation of self-renewal and ectodermal, mesodermal, and endodermal lineage differentiation was based on a previous study (35). For qRT-PCR, 500 ng of deoxyribonuclease-treated RNA was reverse-transcribed into complementary DNA using the QuantiTect Reverse Transcription Kit (QIAGEN). qRT-PCR was performed in triplicate using the SYBR Select Master Mix and the TaqMan Fast Advanced Master Mix on the StepOnePlus Real-Time PCR System (all from Thermo Fisher Scientific). The specific primers used are listed in table S1. Relative quantification was calculated using the 2−ΔΔCt method after normalization to glyceraldehyde-3-phosphate dehydrogenase expression.

Chromatin immunoprecipitation quantitative polymerase chain reaction

ChIP-qPCR was performed in accordance with the manual for the SimpleChIP Plus Enzymatic Chromatin IP Kit (Cell Signaling Technology). Briefly, cell lysates from 8 × 106 H9 cells cultured with Es8 medium on VTN-N–coated dish were sonicated with Qsonica (WakenBtech) after a protein DNA cross-linking step. Samples were then subjected to immunoprecipitation either with normal rabbit IgG antibody or with an antibody to AhR (both from Cell Signaling Technology). DNA from the precipitates was purified and used as templates for qPCR with primers designed to cover promoter regions located 150 base pairs (bp) to 2.0 kilo–base pairs (kbp) upstream of the transcription starting sites of EP300, AHR, CYP1A1, NANOG, POU5F1, and IDO1, and 10 kbp downstream of the coding region for these genes. Primers used for this assay are listed in table S1. RT2 SYBR Green qPCR Master Mix (QIAGEN), EpiTect ChIP qPCR Primer (QIAGEN), and StepOnePlus Real-Time PCR System (Thermo Fisher Scientific) were used for each PCR step.

Immunoprecipitation and Western blotting

Cell lysates for immunoprecipitation were prepared from 1 × 106 H9 cells with 10× Cell Lysis Buffer (Cell Signaling Technology). Lysate (200 μl) was incubated with AhR antibody (Cell Signaling Technology) or mouse control IgG1 antibody (Abcam) for 24 hours, followed by incubation with 20 μl of protein A (Cell Signaling Technology) for 3 hours. The lysate was centrifuged at 14,000g for 30 s at room temperature. The supernatant (20 μl) was subjected to Western blotting with mouse antibody specific for kynurenine antibody (ImmuSmol) and visualized with anti-mouse IgG [Heavy and Light Chains (H + L)] secondary antibody conjugated to horseradish peroxidase (Abcam) and with Chemi-Lumi One Super reagents (Nacalai Tesque).

Cell lysate (40 μg) from undifferentiated H9 cells was used for Western blotting of IDO1, IDO2, and TDO2 using the IDO1 antibody ab55305 (Abcam), the IDO2 antibody 25053-1-AP (Proteintech), and the TDO2 antibody SAB2102400 (Sigma-Aldrich), respectively, and visualized with anti-mouse IgG (H + L) secondary antibody conjugated to horseradish peroxidase (Abcam) and Chemi-Lumi One Super reagents (Nacalai Tesque).

Immunohistochemistry

Ectoderm-differentiated H9 cells at day 6 were fixed with 4% paraformaldehyde (PAF), permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (PBS), and stained with a mouse antibody to L-kynurenine (ImmuSmol). Stained cells were visualized with Alexa Fluor 594 and DAPI (both from Thermo Fisher Scientific) and observed with confocal microscopy (FV1200, Olympus) using a ×60 objective lens.

Undifferentiated H9 cells (with or without 20 μM IDO1 inhibitor CH223191) were fixed with 4% PAF, treated with 0.2% Triton X-100 in PBS, and stained with mouse antibody recognizing AhR (Cell Signaling Technology). Stained cells were visualized with Alexa Fluor 594 and DAPI (both from Thermo Fisher Scientific).

Flow cytometry

Undifferentiated H9 cells or ectoderm-differentiated H9 cells at day 6 were permeabilized with a BD Cytofix fixation buffer (BD Biosciences), and then the cells were stained with either isotype control IgG1 antibody (BD Biosciences) or anti-AhR antibody AF6185 (R&D Systems) and a secondary antibody conjugated to Alexa Fluor 488 (Thermo Fisher Scientific), followed by analysis with the FACS ARIA II (BD Biosciences).

Lipofection of plasmids

H9 cells (200,000) were treated with either 20 μM IDO1 inhibitor (INCB024360-analog) or DMSO. Eight hours after the treatment, 2 μg of pAHR-GFP or pMaxGFP was introduced to 2 × 105 H9 cells with Lipofectamine 2000 (both from Thermo Fisher Scientific) in accordance with the manufacturer’s protocol. Twenty-four hours after lipofection, the cells were fixed with 4% PFA, stained with DAPI (Dojindo Molecular Technologies), and observed with a fluorescence microscope (BX-51, Olympus).

Collection of culture supernatant

All the culture media were collected for metabolite analysis and replaced with fresh media every day without splitting the cultured cells (days 1 to 6). Culture supernatants were transferred to a polypropylene tube every 24 hours. Centrifugal separation (800g, 1 min, at room temperature) was performed to remove dead cells and debris, and the supernatant was transferred to a new tube for storage at −80°C until used.

Pretreatment of culture supernatant samples for LC-MS/MS analysis

We conducted this experiment using chemically defined culture medium (Es8 or TeSR-E8) to eliminate high analytical background noise generated by medium components such as BSA or HSA. A pretreatment procedure was required to remove proteins from culture supernatants. Twenty microliters of an internal standard solution (0.5 mM 2-isopropylmalic acid; 333115-100MG, Sigma-Aldrich) dissolved in ultrapure water (39253-1L-R, Sigma-Aldrich) and 200 μl of LC-MS grade acetonitrile (34967-2.5L, Sigma-Aldrich) were added to 100 μl of centrifuged supernatant. The solution was thoroughly agitated and centrifuged (20400g, 15 min, at room temperature). The supernatant (100 μl) was diluted with 900 μl of ultrapure water and thoroughly agitated for LC-MS/MS analysis.

LC-MS/MS analysis

Pretreated samples were analyzed in multiple reaction monitoring mode using triple quadrupole LC-MS/MS (LCMS-8050) and the Cell Culture Profiling Method Package (Shimadzu Corp). This combination enabled us to simultaneously analyze 95 compounds, including basal medium components and secreted metabolites in 17 min. The 95 metabolites and compounds analyzed are listed in table S2. For normalization of the measured area ratio, culture medium was collected every 24 hours, and the cells were refed with fresh medium. The exponential cell growth curve was generated by plotting cell number counted at every medium collection. The amount of secreted metabolite from a single cell in 1 hour was calculated by dividing the area ratio or micromolar measured by LC-MS/MS with the area defined by the cell number curve of the past 24 hours [in area ratio per cell per hour or in nmol (in 2 ml of culture volume) per cell per hour]. LC-MS/MS data for each experiment are summarized in data file S1. The raw LC-MS/MS data are available in data file S2.

Multivariate analysis

Hierarchical clustering analysis, PCA, and statistical analyses such as volcano plots of LC-MS/MS data were performed using R version 3.1.2. P values were calculated by Student’s t test, and false discovery rates (FDRs) were controlled by the Benjamini-Hochberg method in volcano plots, where substantially changed metabolites were defined as metabolites with P of <0.05, FDR of <0.2, and fold change of 1.2 in an area ratio of LC-MS/MS data.

Morphological analysis of EBs with vision tool

An image of the total area of a 30-mm dish used for EB formation was scanned with Vision Tool (Olympus). The number, area, and roundness of individual EBs were analyzed from the captured images and displayed in a two-dimensional plot as area (x axis) and roundness (y axis) with HCS Studio and CellInsight CX5 (Life Technologies).

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/587/eaaw3306/DC1

Fig. S1. Biomarkers for the differentiation status of H9 cells cultured in Es8 medium.

Fig. S2. Biomarkers for the differentiation status of PFX#9 cells cultured in TeSR-E8.

Fig. S3. Biomarkers for the differentiation status of PFX#9 cells cultured in mTeSR1.

Fig. S4. Exogenous kynurenine partially rescues IDO1 inhibitor–induced suppression of cell proliferation.

Fig. S5. Gene expression profiles of mouse Ahr and human AHR.

Fig. S6. qRT-PCR scorecard panel for H9 cells under ectoderm induction.

Fig. S7. KAT2 inhibition blocks ectodermal differentiation of H9 cells.

Fig. S8. Characteristics of EBs.

Fig. S9. TGF-β1 and FGF2 positively regulate the transcription of NANOG.

Fig. S10. Kynurenic acid (kynurenate) in culture medium.

Table S1. Primers used in this study.

Table S2. Target metabolites and compounds analyzed by LC-MS/MS.

Data file S1. Summary of LC-MS/MS data for each figure.

Data file S2. Raw LC-MS/MS data.

REFERENCES AND NOTES

Acknowledgments: We thank T. Hine, E. Murakami, N. Yajima, K. Tsubasa, and K. Yamamoto of FBRI and S. Gomi of Tokyo Electron Ltd. for qRT-PCR work and technical support and H. Hashimoto of FBRI (Kobe, Japan) and F. McKenzie of Sakarta Co. (Edinburgh, UK) for critical reading of the manuscript and valuable comment. Funding: This work was partially supported by the Japanese research grant AMED “Points to consider for the safety test of the Pluripotent Stem Cell-derived cell product” (grant no. JP18bk0104059), Regulatory Science Regenerative Medicine 2016–2018. Ethics statement: All the experiments conducted in this study used human ESCs and human iPSCs. The work was reviewed and approved by the ethical committee of the FBRI. This study does not include human participants or the establishment of new iPSC cell lines from human tissues. Author contributions: T.Y. conducted all the experiments except LC-MS/MS analysis, which was done by K. Hatabayashi, K. Hara, M.A., and T.S. N.Y. conducted the plasmid transfection. K. Hatabayashi designed Fig. 1. C.T. conducted experiments shown in Fig. 7 (A to C). T.S. developed the LC-MS/MS methods. M.T. summarized the data from T.S. and reported to K.K. K.K. summarized the data from K. Hatabayashi and reported to S.K. Y.O. summarized the data from K. Hara and reported to S.K. S.K. directed and reviewed all the experiments and prepared the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The microarray data were deposited into the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/gds/) with the identifier GSE129507. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article