Graphical abstract

   graphic file with name fx1.jpg
   [41]Open in a new tab

Highlights

     * •
       Medium acidosis plays an active role in lineage-specific
       differentiation of hPSCs
     * •
       Acidic pH alone is sufficient to induce cardiac differentiation
     * •
       Prevention of acidification overrides Wnt inhibition to block
       cardiac differentiation
     * •
       Glycolysis inhibition rescues cardiac induction under alkaline pH
     __________________________________________________________________

   In this study, Chen and colleagues show that medium acidosis plays a
   key role in hPSC cell fate determination. Acidic pH alone can induce
   cardiac cell fate, while alkaline pH blocks cardiac differentiation
   even in the presence of Wnt inhibitor. Glycolysis inhibition rescues
   cardiac induction under alkaline pH. Through metabolic regulation,
   cardiac differentiation efficiency and consistency can be greatly
   enhanced.

Introduction

   Human pluripotent stem cells (hPSCs), including human embryonic stem
   cells (hESCs) and induced pluripotent stem cells (iPSCs), are important
   research models and starting materials for cell production in
   regenerative medicine. In order to achieve optimal maintenance or
   differentiation, proper signaling and nutritional inputs are provided
   to the hPSCs in cell culture. Activators and inhibitors of specific
   signaling pathways are commonly used to drive stem cells toward desired
   cell fates. On the other hand, accumulating evidence shows that
   metabolism of various nutrients is also closely linked to pluripotency,
   reprogramming, and differentiation, which may have an impact on cell
   fate transitions ([42]Dahan et al., 2019; [43]Folmes et al., 2011;
   [44]Moussaieff et al., 2015; [45]Teslaa and Teitell, 2015).

   Besides the exogenous factors present in cell culture ([46]Liu and
   Chen, 2021; [47]Liu et al., 2019), hPSCs themselves are also dynamic
   contributors to their local environment. For example, hPSCs produce
   autocrine mitogenic factors, such as insulin growth factors (IGFs) that
   affect cell fate specification in mesoderm lineages ([48]Yang et al.,
   2019). On the nutritional side, hPSCs not only consume large amount of
   nutrients such as glucose and glutamine but also remodel their
   surroundings through the autocrine secretion of metabolites ([49]Kumar
   et al., 2017; [50]Tatapudy et al., 2017). Numerous endogenous metabolic
   factors are continuously produced by hPSCs. However, it is unclear
   whether and how these metabolites play active roles in cell fate
   determination.

   We are interested in the differentiation of hPSCs toward mesendodermal
   cell fates because many important cell types such as cardiomyocytes and
   hepatocytes are generated through this lineage. For example,
   cardiomyocytes from hPSCs through targeted differentiation provide
   important tools for drug screening, toxicology studies, and cell
   therapy. Cardiomyocyte induction can be effectively achieved through
   modulation of WNT (Wingless/Integrated) signaling using growth factors
   or small chemicals. WNT activation drives hPSCs toward mesendodermal
   progenitors, and subsequent WNT inhibition leads to cardiomyocyte cell
   fate ([51]Burridge et al., 2014; [52]Lian et al., 2012). In comparison,
   if mesendodermal progenitors were allowed to spontaneously
   differentiate without WNT inhibitor, a variety of cell types will
   emerge, driven by endogenous signals from the cells. Suppression of
   endogenous IGF or transforming growth factor β (TGF-β) pathways also
   leads to cardiomyocyte differentiation ([53]Kattman et al., 2011;
   [54]Yang et al., 2019). These findings demonstrate the key roles of
   signal transduction pathways in mesendodermal cell fate specification.
   However, the roles of metabolic processes or metabolites in this
   process are less clear and need further investigation.

   Glycolysis and oxidative phosphorylation are essential metabolic
   processes to generate energy in hPSCs. In the pluripotent state,
   glycolysis is the main energetic metabolism process in hPSCs, and
   lactic acid is generated as a metabolic byproduct, acidifying the
   culture media. The balance of energetic metabolism shifts from
   glycolysis toward oxidative phosphorylation when hPSCs differentiate to
   specific cell types such as cardiomyocytes ([55]Cho et al., 2006;
   [56]Chung et al., 2007; [57]Cliff et al., 2017; [58]Shyh-Chang et al.,
   2013; [59]Tsogtbaatar et al., 2020). As a result, acid secretion is
   high in the early stages of differentiation, covering the critical
   period of cell fate determination. We reported previously that acidic
   environment is inhibitory to mesoderm induction by BMP4, while pH
   neutralization by NaHCO[3] enhances mesoderm induction ([60]Liu et al.,
   2018). In this study, we investigate the functional role of acids in
   cell fate determination from mesendoderm progenitors and how
   environmental pH controls lineage-specific differentiation. Cellular
   secretion of acids acidifies the culture medium and leads to pH
   fluctuations between daily medium changes. When the secreted acid is
   neutralized by excess buffering agents during differentiation, culture
   medium is no longer acidified, and cardiomyocyte generation is
   inhibited. This inhibition is strong enough to override the impact of
   WNT inhibitor IWP2, which is commonly used to induce cardiomyocytes. On
   the other hand, the addition of exogenous acids drives mesendoderm
   progenitors toward cardiomyocytes without the need for signaling
   modulators. Upon further investigation, we report that significant
   changes in signal transduction and metabolic processes occur as a
   result of pH modulation and contribute to the cell fate changes.

Results

Medium pH significantly affects cell fate determination in spontaneous
differentiation of mesendoderm progenitors

   Mesendoderm progenitor cells can differentiate to distinct cell types
   without exogenous stimuli ([61]Yang et al., 2019), and we used this
   platform to study the impact of environmental pH on cell fate
   determination. GSK3 inhibitor CHIR-99021, a WNT activator, was applied
   on H1 hESCs to induce mesendoderm progenitors. Cells were then allowed
   to spontaneously differentiate in the absence of exogenous growth
   factors or chemical inducers ([62]Figure S1A). Culture medium was
   changed every day. To characterize the impact of lactic acid secretion
   on the microenvironment, we monitored medium pH during differentiation
   and observed that medium pH decreased significantly within 24 h,
   fluctuating between 7.3 and 6.6 ([63]Figures 1A and [64]S1B). The
   amplitude of pH change was most prominent between day 2 and day 4,
   which coincides with the critical period of mesendodermal cell fate
   specification. We also observed that medium pH was acidified within 4 h
   of medium change and reached 6.8 after just 8 h ([65]Figure S1B). This
   suggested that hPSCs continuously modified medium pH through lactic
   acid release, and cells were in acidic condition for over 16 h on a
   24-h feeding schedule. We then showed that the pH fluctuation could be
   modulated by exogenous reagents, HCl and NaHCO[3] ([66]Figures S1A and
   S1C). Exogenous HCl kept medium pH at acidic levels ([67]Figure 1B),
   while NaHCO[3] prevented acidification ([68]Figure 1C).

Figure 1.

   [69]Figure 1
   [70]Open in a new tab

   Medium pH significantly affects cell fate determination in spontaneous
   differentiation of mesendoderm progenitors

   (A–C) pH values of cell culture medium before and after daily medium
   change during the induction and spontaneous differentiation of
   mesendoderm progenitors. See [71]Figure S1A for procedure details. (A)
   Control condition, (B) HCl (8 mM) applied daily on day 2–7, (C)
   NaHCO[3](20 mM) applied daily on day 2–7. “DF” refers to
   differentiation medium (DF medium) (DMEM/F12, vitamin C, selenium,
   transferrin, lipid).

   (D) Heatmap of gene expression levels on day 10 of differentiation,
   measured by qPCR (n = 3 differentiations). Expression levels were
   calculated from 2ˆ(-ΔCt), normalized to GAPDH and scaled by column. See
   [72]Figure 3D for comparison.

   (E) Percentage of cells expressing TNNT2, AFP, and CD44, measured by
   flow cytometry analysis on day 10 of differentiation (n = 3
   differentiations). ^∗∗∗, p < 0.001 compared to control (one-way ANOVA
   with Dunnett’s multiple comparison test). NS, not significant.

   (F) Immunostaining showing expression of TNNT2, AFP, and CD44 on day 10
   of differentiation under control (Ctrl), HCl, and NaHCO[3] treatment.
   Scale bar, 100 μm.

   See also [73]Figure S1.

   We investigated how pH fluctuation could affect the emergence of
   various cell types in 10 days of differentiation, when pH was
   transiently modulated between day 2 and day 7 ([74]Figures 1A–1C).
   Under spontaneous differentiation condition, medium pH fluctuated daily
   ([75]Figure 1A), and various cell type-specific genes were detected
   ([76]Figure 1D), including markers for cardiomyocytes (ISL1, NKX2-5,
   TNNT2, MYL7, MYL2, MYH7), epicardium (WT1), mesenchymal cells (CD44),
   endothelium (PECAM, CDH5), hepatocytes (AFP), and intestinal cells
   (CDX2). The addition of HCl maintained a constantly acidic pH
   ([77]Figure 1B) and enhanced cardiac differentiation while suppressing
   other cell fates ([78]Figure 1D). Meanwhile, NaHCO[3] neutralized
   cellular acid secretion to prevent medium acidification
   ([79]Figure 1C), which promoted mesenchymal, endothelial, and
   endodermal gene expression but suppressed cardiac differentiation
   ([80]Figure 1D). Flow cytometry analysis and immunostaining also
   demonstrated that acidic condition enhanced cardiac differentiation,
   while alkaline pH suppressed cardiac cell fate and promoted other cell
   types ([81]Figures 1E and 1F). At the concentration of HCl (8 mM) and
   NaHCO[3] (20 mM) we were using, the percentage of dead cells was
   slightly higher under HCl treatment, but the majority (∼80%) of cells
   survived ([82]Figure S1D). Intracellular pH is lower under HCl
   treatment, as shown by staining with intracellular pH indicator
   ([83]Figure S1E). These data suggested that environmental pH plays a
   critical role in the generation of heterogeneous cell fates from
   mesendoderm progenitors. Exposure to low pH is necessary for the
   spontaneous generation of cardiomyocytes, and suppression of acidosis
   prevented cardiac cell fate. In contrast, a constantly acidic pH led to
   loss of diversity and promoted differentiation toward cardiomyocytes.

Acidic pH alone is sufficient to drive cardiac differentiation

   We then further investigated the aforementioned observation that
   application of acid drives cardiac differentiation. Considering that
   the medium acidification is caused by lactic acid from glycolysis, we
   compared HCl and lactic acid in their ability to modulate cell fate. We
   showed that lactic acid and HCl decreased medium pH in a dose-dependent
   manner with similar patterns ([84]Figure S2A). With decreased pH in
   spontaneous differentiation condition, cardiac gene expression was
   elevated by both lactic acid and HCl, while epicardial markers WT1 and
   TBX18 were suppressed ([85]Figures 2A and 2B). The cardiac induction by
   both acids was confirmed by flow cytometry ([86]Figure S2B). These
   results indicated that decreased pH was a functional factor modulating
   cell differentiation. To simplify the study, we used HCl as the main
   tool to study proton function in this paper. 8 mM HCl was used unless
   otherwise specified. We also showed that acidic medium induced
   cardiomyocytes from multiple hPSC lines, including H1 and H9 hESCs as
   well as NL-1 and NL-4 iPSCs ([87]Figures S2C–S2F).

Figure 2.

   [88]Figure 2
   [89]Open in a new tab

   Acidic pH drives cardiac differentiation

   (A) Heatmap of cardiac and epicardial gene expression levels on day 11
   of mesendoderm spontaneous differentiation showing the
   concentration-dependent effect of HCl treatment (day 2–7).

   (B) Heatmap of cardiac and epicardial gene expression levels on day 11
   of differentiation showing the concentration-dependent effect of lactic
   acid treatment (day 2–7). Gene expression was measured by qPCR (n = 3
   differentiations. Expression levels were calculated from 2ˆ(-ΔCt),
   normalized to GAPDH and scaled by row).

   (C) FACS analysis of TNNT2-positive cells on day 12 of cardiac
   differentiation, showing the impact of HCl treatment at different
   stages of the differentiation.

   (D) Expression of TNNT2 and NKX2-5 following HCl treatment at different
   stages of the differentiation, measured by qPCR (n = 4 technical
   replicates and are representative of 3 independent experiments).

   (E) Western blot showing the levels of non-phosphorylated (active)
   β-catenin and total β-catenin under NaHCO[3], control, or HCl
   treatment. Mesendoderm progenitors on day 2 of differentiation were
   subjected to treatment for 24 h and collected on day 3. β-actin was
   used as a loading control.

   (F) Immunostaining of NKX2-5 and TNNT2 in HCl-induced cardiomyocytes.
   Cells were passaged into confocal dishes on day 12 of differentiation
   and fixed for immunostaining the next day. Scale bar, 50 μm.

   (G) Heatmap showing row-scaled TPM (transcripts per kilobase million)
   values of differentially expressed genes (DEGs) under control, HCl, and
   NaHCO[3] treatment on day 0, 3, 5, 7, and 11 of spontaneous
   differentiation, compared to IWP2-induced cardiac differentiation.

   (H) Expression profile of representative genes from each cluster in the
   heatmap in (G).

   See also [90]Figure S2, [91]Videos S1 and [92]S2.
   Video S1. Beating cardiomyocytes induced by HCl, related to Figure 2

   Mesendoderm progenitors were generated from H1 cells using CHIR-99021
   (day 0) and then allowed to spontaneously differentiate under 8 mM HCl
   (applied day 2–7) in DF medium. Video taken on day 11.
   [93]Download video file^ (898.8KB, mp4)
   Video S2. Beating cardiomyocytes induced by lactic acid, related to
   Figure 2

   Mesendoderm progenitors were generated from H1 cells using CHIR-99021
   (day 0) and then allowed to spontaneously differentiate under 8 mM
   lactic acid (applied day 2–7) in DF medium. Video taken on day 11.
   [94]Download video file^ (1.7MB, mp4)

   We further analyzed cell differentiation under low pH. We showed that
   the timing of acid treatment was critical for the emergence of
   cardiomyocytes ([95]Figures 2C and 2D). When acid was applied before
   day 2, most cells died by day 3. If acid was applied after day 5, there
   was no cardiac induction. Acid treatment between day 2 and day 7
   yielded the best cardiac differentiation, so we used this schedule to
   induce cardiac differentiation in the rest of this study. We showed
   that acidic medium decreased the level of active (non-phosphorylated)
   β-catenin and total β-catenin, suggesting WNT pathway was inhibited
   under acid treatment ([96]Figure 2E). We then showed that WNT activator
   CHIR-99021 partially suppressed cardiac induction by HCl
   ([97]Figure S2G). These data suggested that acidic pH induced cardiac
   differentiation at least partially through WNT inhibition.

   Under acid treatment, most cells were positive in NKX2-5 and TNNT2
   expression by immunostaining ([98]Figure 2F), which was consistent with
   flow cytometry analysis ([99]Figure S2C). Expression of epicardium cell
   marker WT1 was suppressed by acid treatment ([100]Figures 2A and
   [101]S2H). Upon extended culture, well-organized myofibril structures
   formed in acid-induced cardiomyocytes as shown by α-actinin
   immunostaining, similar to the cells induced by WNT inhibitor IWP2
   ([102]Figure S2I). Patch clamp analysis and microelectrode array
   analysis demonstrated typical cardiomyocyte electrophysiological
   profiles similar to the cells induced by IWP2 ([103]Figures S2J and
   S2K). These data demonstrated that low pH was an inducer for cardiac
   cell fate in the absence of other exogenous signaling modulators.

Global gene expression is dynamically regulated by medium pH

   In order to understand the impact of medium pH on the differentiation
   process, a time course experiment was conducted to study global gene
   expression in spontaneous differentiation when different pH was applied
   to mesendoderm progenitors from day 2 to day 7. Principal-component
   analysis (PCA) showed that pH modulation led to significant
   transcriptome changes after just one day of treatment (day 3), and the
   difference became more prominent in the following days
   ([104]Figure S2L). At the end of the time course (day 11), gene
   expression was enriched in liver, kidney, and heart cell types under
   spontaneous condition without additional pH modulation
   ([105]Figure S2M, cluster 2 and 4). Acidic pH promoted global gene
   expression enriched in cardiac cell fate similar to Wnt inhibitor IWP2
   ([106]Figure S2M, cluster 3 and 5), while suppressing gene expression
   in the liver, smooth muscle, adipocytes, and lung ([107]Figure S2M,
   cluster 1, 2, and 4). Meanwhile, alkaline pH induced a diversity of
   mesodermal/endodermal cell types ([108]Figure S2M, cluster 1 and 2) but
   reduced cardiac gene expression ([109]Figure S2M, cluster 3 and 5).
   Further examination of cardiomyocyte markers confirmed the
   cardiac-promoting effect of HCl and the inhibitory effect of NaHCO[3]
   ([110]Figure S2N). These data supported the findings that medium pH
   played critical roles in cell fate determination in the absence of
   growth factors.

   An overview of the gene expression time course revealed that, during
   the differentiation process, following the loss of pluripotency in
   response to CHIR-99021 treatment ([111]Figures 2G and 2H, cluster 1),
   expression of primitive streak and progenitor cell markers was elevated
   on day 3–5 ([112]Figures 2G and 2H, cluster 2 and 3), and cell
   type-specific gene expression elevated between day 5–10
   ([113]Figures 2G and 2H, cluster 4, 5, and 6). HCl promoted the
   expression of cardiomyocyte markers (cluster 5), similar to IWP2. We
   then further investigated the gene expression on day 3, which was the
   critical period for cell fate specification ([114]Figure S2O). Kyoto
   Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showed
   that HCl treatment significantly downregulated gene expression in DNA
   replication, cell cycle, Wnt signaling, TGF-β signaling, and various
   metabolic pathways ([115]Figure S2O, cluster 1). HCl also upregulated
   gene expression in multiple signaling pathways, including calcium
   signaling pathway and mitogen-activated protein kinase (MAPK) and
   AMP-activated protein kinase (AMPK) signaling pathways, as well as
   cellular processes such as focal adhesion ([116]Figure S2O, cluster 3
   and 5). In comparison, NaHCO[3] treatment significantly suppressed gene
   expression in cyclic AMP and calcium signaling pathways
   ([117]Figure S2O, cluster 4 and 5). NaHCO[3] treatment significantly
   elevated gene expression in glycolysis, Wnt and TGF-β signaling, and
   other metabolic pathways ([118]Figure S2O, cluster 1). Further analysis
   of Wnt pathway components on day 3 showed that NaHCO[3] promoted
   expression of canonical Wnt pathway genes, and HCl elevated several Wnt
   target genes and negative regulators of Wnt pathway such as NKD2 and
   DDIT3 ([119]Figure S2P). Taken together, these data suggest that proton
   levels in the microenvironment affected important cellular processes
   and signaling pathways involved in cell fate determination, and acidic
   pH is capable of driving cell fate toward cardiomyocytes without other
   chemical inducers.

Alkaline pH interferes with signal transduction in cardiac differentiation

   Because acidic and alkaline conditions could modulate mesoderm
   differentiation, we investigated how they interacted with WNT pathway
   inhibition, which is widely used for inducing cardiomyocytes
   ([120]Figure S3A). In the presence of WNT inhibitor IWP2 (applied from
   day 2 to day 4), the medium was also acidified daily by hPSCs within a
   few hours, and medium pH decreased to 6.8 in around 8 h
   ([121]Figure S3B). Exogenous HCl kept the medium at acidic pH, and
   NaHCO[3] elevated medium pH ([122]Figures 3A–3C). The overall trend of
   daily medium acidosis was similar to the one under spontaneous
   condition ([123]Figures 1 and [124]S1). These data suggested that WNT
   inhibition did not affect proton generation from glycolysis in
   mesendoderm progenitors.

Figure 3.

   [125]Figure 3
   [126]Open in a new tab

   Alkaline pH interferes with signal transduction in cardiac
   differentiation

   (A–C) pH values of cell culture medium before and after daily medium
   change during IWP2-induced cardiac differentiation as shown in
   [127]Figure S3A. (A) Control condition, (B) HCl (8 mM) applied daily on
   day 2–7, (C) NaHCO[3] (20 mM) applied daily on day 2–7. “DF” refers to
   DF medium (DMEM/F12, vitamin C, selenium, transferrin, lipid).

   (D) Heatmap of gene expression levels on day 10 of differentiation,
   measured by qPCR (n = 3 differentiations). Expression levels were
   calculated from 2ˆ(-ΔCt), normalized to GAPDH and scaled by column. See
   [128]Figure 1D for comparison.

   (E) Percentage of cells expressing TNNT2, measured by flow cytometry
   analysis on day 10 of differentiation (n = 3 differentiations). ^∗∗,
   p < 0.01; ^∗∗∗, p < 0.001 compared to IWP2 condition (one-way ANOVA
   with Dunnett’s multiple comparison test).

   (F) Flow cytometry analysis of TNNT2-positive cells on day 10 of
   differentiation (n = 3 differentiations). Cells were treated with or
   without IWP2 (day 2–4) and NaHCO[3] (applied during different time
   periods).

   (G) Western blot showing the levels of non-phosphorylated (active)
   β-catenin and total β-catenin under IWP2, IWP2/NaHCO[3], or IWP2/HCl
   treatment. Mesendoderm progenitors on day 2 of differentiation were
   subjected to treatment for 24 h and collected on day 3. β-actin was
   used as loading control.

   (H) Flow cytometry analysis of TNNT2 expression on day 10, showing the
   inhibitory effect of CHIR-99021 (applied day 2–4) on IWP2-induced
   cardiac differentiation and its rescue by HCl (applied day 2–7).

   See also [129]Figure S3.

   We investigated how pH modulation could affect the emergence of
   cardiomyocytes under WNT inhibition. WNT inhibitor and acid
   synergistically induced cardiomyocytes according to real-time PCR
   ([130]Figure 3D) and flow cytometry analysis ([131]Figure 3E). In
   contrast, NaHCO[3] suppressed cardiac differentiation even in the
   presence of WNT inhibitor IWP2, and it promoted differentiation to
   other mesoderm and endoderm cell fates ([132]Figures 3D and 3E).
   Similar impacts of acidic and basic pH on WNT inhibitor-induced cardiac
   differentiation were observed with H9 hESCs and NL-1 and NL-4 iPSCs
   ([133]Figures S4C–S4E). To further evaluate the impact of
   alkalinization, we applied NaHCO[3] for different lengths of time
   during IWP2-induced cardiac differentiation and found that NaHCO[3]
   treatment during day 2–4 is sufficient to suppress cardiomyocyte
   generation by IWP2, and the effect is similar to day 2–7 treatment.
   NaHCO[3] treatment during day 5–7 did not have much impact
   ([134]Figure 3F). These results suggest that medium acidification
   between day 2 to day 4 is a necessary factor for cardiomyocyte cell
   fate specification. To better understand the mechanism of the pH
   impact, we examined WNT signaling 24 h after IWP2 and pH treatment. We
   showed that HCl further suppressed active β-catenin in the presence of
   IWP2, while the level of active β-catenin was maintained in the
   presence of NaHCO[3] ([135]Figure 3G). WNT pathway activator CHIR-99021
   fully suppressed IWP2-induced cardiac differentiation, but HCl was able
   to partially rescue cardiac differentiation under CHIR-99021 treatment
   ([136]Figure 3H). It implied that acidic treatment could induce cardiac
   fate synergistically with WNT inhibition through another pathway.

   In addition to differentiation in DMEM/F12-based media, we also carried
   out spontaneous and IWP2-induced differentiation in a DF medium
   prepared using RPMI1640 medium (RPMI1640, vitamin C, selenium,
   transferrin, lipid) and observed similar impacts of acidic and basic pH
   on cardiac differentiation ([137]Figure S3F). Similar phenotypes were
   also reproduced when mesendoderm progenitors were induced with a
   different protocol using BMP4/Activin A ([138]Figure S3G).

   PCA showed that HCl led to a transcriptome trajectory similar to IWP2,
   but NaHCO[3] led cells to another path even in the presence of IWP2
   ([139]Figure S3H). Day 11 RNA sequencing (RNA-seq) data demonstrated
   that IWP2 induced gene expression enriched in cardiac fate
   ([140]Figure S3I, cluster 2), but NaHCO[3] suppressed cardiac gene
   expression and promoted gene expression enriched in the liver, lung,
   small intestine, and kidney both in the presence and in the absence of
   IWP2 ([141]Figure S3I, cluster 1 and 3). These data suggested that
   elevated pH was able to override signaling modulator IWP2 and suppress
   cardiac fate induced by WNT inhibition.

pH status influences metabolic changes during differentiation

   To understand how alkaline pH overrides WNT inhibition to control cell
   fate, we then investigated the gene expression profile on day 3 of
   differentiation under IWP2+NaHCO[3] treatment ([142]Figure S3J).
   Compared to the IWP2 group, addition of NaHCO[3] promoted gene
   expression in the Ras signaling pathway, regulation of actin
   cytoskeleton, DNA replication, cell cycle, and Wnt signaling pathway
   ([143]Figure S3J, cluster 2 and 3). In addition to the impact on
   signal transduction and cellular processes, KEGG pathway enrichment
   analysis also showed involvement of pH-induced differentially expressed
   genes (DEGs) in multiple metabolic pathways, including purine and
   pyrimidine metabolism, carbon metabolism, and amino acid metabolism
   ([144]Figures S4A and S4B, see also [145]Figures S2O and [146]S3J).
   Based on these findings, we explored whether environmental pH may
   control cell fate partially through metabolic changes. We examined the
   expression levels of glycolysis and tricarboxylic acid (TCA) pathway
   genes through the differentiation process. On day 3 of differentiation,
   NaHCO[3] treatment significantly elevated the expression of genes in
   the glycolysis pathway, coinciding with early mesendoderm cell fate
   commitment ([147]Figure 4A). In contrast, differences in TCA cycle
   genes appeared later in the differentiation process around day 7–11,
   and levels are elevated in HCl- and IWP2-treated groups, probably as a
   result of cardiac cell generation ([148]Figure 4B).

Figure 4.

   [149]Figure 4
   [150]Open in a new tab

   pH-induced metabolic changes during differentiation

   (A) Heatmap showing the time course of gene expression in the
   glycolysis pathway based on RNA-seq data (TPM values scaled by row).

   (B) Heatmap showing the time course of gene expression in the TCA
   pathway based on RNA-seq data (TPM values scaled by row).

   (C) Impact of pH on metabolite levels, measured by LC-MS. Mesendoderm
   progenitors were subjected to control (Ctrl), HCl, or NaHCO[3]
   treatment on day 2 of differentiation. Spent medium was collected for
   LC-MS analysis on day 3 and compared with fresh DF medium to calculate
   the level changes. Data presented as mean ± SD and are normalized to
   control, n = 3 differentiations. ^∗∗, p < 0.01; ^∗∗∗, p < 0.001
   compared to control well (one-way ANOVA with Dunnett’s multiple
   comparison test).

   (D) Glycolysis Stress Test showing the impact of pH modulation on
   glycolysis.

   (E) Mito Stress Test showing the impact of pH modulation on
   mitochondrial respiration. Mesendoderm progenitors were treated with
   HCl or NaHCO[3] on day 2 of differentiation, and the assays were
   carried out on day 3. Data are normalized to total protein (by Bradford
   assay) and presented as mean ± SD, n = 4 differentiations. ^∗,
   p < 0.05; ^∗∗, p < 0.01; ^∗∗∗, p < 0.001 compared to control (one-way
   ANOVA with Dunnett’s multiple comparison test). NS, not significant.

   See also [151]Figure S4.

   We then examined the impact of pH modulation on hPSC metabolism by
   liquid chromatography-mass spectrometry (LC-MS) and demonstrated that
   metabolite levels were affected by pH on day 3 of differentiation
   ([152]Figure 4C). Acidic environment suppressed glucose consumption,
   glutamine consumption, and lactic acid release, while NaHCO[3] enhanced
   these processes. Pyruvate consumption was not significantly changed. In
   comparison, IWP2 treatment did not significantly change day 3
   metabolite levels ([153]Figure S4C). Seahorse extracellular flux
   analysis showed that acidic environment significantly suppressed
   glycolysis and glycolytic capacity in differentiating cells
   ([154]Figure 4D) but did not affect mitochondrial respiration
   ([155]Figure 4E). On the other hand, alkaline environment led to
   elevated glycolysis and glycolytic capacity ([156]Figure 4D), as well
   as elevated basal and spare respiratory capacity and ATP production
   ([157]Figure 4E). Similar phenotypes were observed in the presence of
   WNT inhibitor IWP2 ([158]Figures S4D and S4E). These data suggested
   that glycolysis and oxidative phosphorylation were affected by proton
   levels in the stem cell microenvironment.

Glycolysis inhibition promotes cardiac cell fate

   Considering that the level of glycolysis decreases during stem cell
   differentiation ([159]Shyh-Chang et al., 2013), we hypothesized that
   NaHCO[3] may interfere with cardiac differentiation by maintaining
   glycolysis. We examined whether glycolysis inhibition could rescue
   cardiac differentiation under NaHCO[3]. Glycolysis inhibitor
   2-deoxy-D-glucose (2-DG) was applied on mesendoderm progenitors in the
   presence of IWP2 and NaHCO[3]. Glycolysis Stress Test showed that 2-DG
   suppressed glycolysis and glycolytic capacity that were elevated by
   NaHCO[3] ([160]Figures 5A and [161]S5A). In Mito Stress Test, 2-DG
   elevated spare respiratory capacity but not basal mitochondrial
   respiration, proton leak, or ATP production ([162]Figures 5B and
   [163]S5B). When applied during IWP2-induced cardiac differentiation,
   2-DG rescued expression of cardiac markers TNNT2 and NKX2.5 under
   NaHCO[3] ([164]Figure 5C). Flow cytometry confirmed that 2-DG increased
   TNNT2-positive cells in the presence of NaHCO[3] ([165]Figure 5D).
   Transcriptome analysis showed that 2-DG shifted the differentiation
   trajectory that was affected by NaHCO[3] ([166]Figure S5C). Gene
   Ontology term enrichment analysis on day 3 samples showed that 2-DG
   treatment elevated gene expression associated with cardiac muscle
   tissue development and suppressed glycolytic process which was elevated
   by NaHCO[3] ([167]Figure S5D). Cell type analysis on day 11 showed that
   2-DG rescued cardiac differentiation in the presence of IWP2/NaHCO[3]
   and promoted development of cardiac muscle fiber, ventricle, and heart
   bulk tissue ([168]Figure S5E). These data suggested that high pH
   suppressed cardiac differentiation by maintaining glycolysis, so the
   inhibition of glycolysis could rescue cardiac differentiation at high
   medium pH, confirming that metabolic control is part of the mechanism
   through which NaHCO[3] overrides WNT inhibition in cardiac
   differentiation.

Figure 5.

   [169]Figure 5
   [170]Open in a new tab

   Glycolysis inhibition promotes cardiac cell fate through AMPK signaling

   (A) Glycolysis Stress Test showing the impact of 2-DG (4 mM) on
   glycolysis during IWP2-induced cardiac differentiation under alkaline
   conditions.

   (B) Mito Stress Test showing the impact of 2-DG (4 mM) on mitochondrial
   respiration during IWP2-induced cardiac differentiation under alkaline
   conditions. Mesendoderm progenitors were treated with IWP2 with or
   without NaHCO[3] or NaHCO[3]/2-DG on day 2 of differentiation, and the
   assays were carried out on day 3. Data are normalized to total protein
   (by Bradford assay) and presented as mean ± SD, n = 4 differentiations.
   See [171]Figures S5A and S5B for data analysis results.

   (C) qPCR analysis of NKX2-5 and TNNT2 expression on day 10 of cardiac
   differentiation, normalized to GAPDH. IWP2, 3 μM (day 2–4); NaHCO[3],
   20 mM (day 2–7); 2-DG, 4 mM (day 2–4). Data presented as mean ± SD of
   four technical replicates and are representative of three independent
   experiments. ^∗∗∗, p < 0.001 (one-way ANOVA with Dunnett’s multiple
   comparison test).

   (D) Flow cytometry analysis of the percentage of TNNT2^+ cells on day
   10 of IWP2-induced cardiac differentiation. IWP2, 3 μM; NaHCO[3],
   20 mM; 2-DG, 4 mM. Data presented as mean ± SEM of four independent
   experiments. ^∗∗, p < 0.01 (one-way ANOVA with Dunnett’s multiple
   comparison test).

   (E) Flow cytometry analysis of TNNT2^+ cells on day 10 of mesendoderm
   spontaneous differentiation, showing the impact of 2-DG treatment
   without IWP2. Data presented as mean ± SD, n = 3 differentiations.
   ^∗∗∗, p < 0.001 (one-way ANOVA with Dunnett’s multiple comparison
   test).

   (F) Western blot showing the levels of phosphorylated AMPK and total
   AMPK under control (Ctrl), HCl, or NaHCO[3] treatment. Mesendoderm
   progenitors on day 2 of differentiation were subjected to treatment for
   24 h and collected on day 3. β-actin was used as loading control.

   (G) Western blot showing the effect of 2-DG treatment on AMPK
   phosphorylation in cells differentiating under control (Ctrl), 2-DG,
   IWP2, IWP2/NaHCO[3], or IWP2/NaHCO[3]/2-DG treatment. Mesendoderm
   progenitors on day 2 of differentiation were subjected to treatment for
   24 h and collected on day 3. β-actin was used as loading control.

   (H) Flow cytometry analysis of the percentage of TNNT2^+ cells on day
   10 of IWP2-induced cardiac differentiation in the presence or absence
   of NaHCO[3] and AMPK activator A769662. Data presented as mean ± SEM of
   three independent experiments. ^∗∗, p < 0.01 (one-way ANOVA with
   Dunnett’s multiple comparison test).

   (I) Model diagram illustrating the mechanism of pH impact on cell fate.
   CHIR, CHIR-99021; 2-DG, 2-deoxy-D-glucose; CM, cardiomyocytes.

   See also [172]Figure S5.

   We further examined the impact of glycolysis inhibition on cell fate
   determination in spontaneous differentiation from mesendoderm
   progenitors (without IWP2). 2-DG alone was sufficient to promote
   cardiac gene expression, while suppressing gene expression in
   epicardial, mesenchymal, and endothelial lineages ([173]Figure S5F).
   The induction of cardiac differentiation by 2-DG was confirmed by flow
   cytometry analysis and immunostaining ([174]Figures 5E and [175]S5G).
   These data suggested that glycolysis inhibition promoted cardiac cell
   fate, and metabolic regulation played an active role in cardiac fate
   induction.

Glycolysis inhibition induces cardiomyocytes through AMPK activation

   In order to understand how metabolic modulation controls cell fate, we
   examined the impact of 2-DG on gene expression in key signaling
   pathways involved in cardiac differentiation. RNA-seq data showed that
   HCl treatment elevated the expression of PRKAA2 which encodes AMPK.
   NaHCO[3] suppressed its expression. Under IWP2 treatment, NaHCO[3]
   treatment suppressed PRKAA2 expression, which was rescued by 2-DG
   ([176]Figure S5H). Consistently, acidic pH upregulated gene expression
   in AMPK pathway in spontaneous differentiation ([177]Figure S2O). We
   hypothesized that pH-associated metabolic changes might alter cell fate
   through AMPK signaling, which is a master regulator of cellular
   metabolism and cardiac differentiation ([178]Sarikhani et al., 2020).
   We then examined the level of AMPK phosphorylation in cells
   differentiating under different pH. Acidic pH (HCl) promoted AMPK
   phosphorylation, while basic pH (NaHCO[3]) had an inhibitory effect
   ([179]Figure 5F). 2-DG promoted AMPK phosphorylation both by itself and
   under IWP2/NaHCO[3] treatment ([180]Figure 5G). These data suggested
   that glycolysis inhibition by acidic pH or 2-DG activated AMPK pathway.
   When AMPK activator A769662 was applied, it rescued cardiac
   differentiation in the presence of IWP2/NaHCO[3] as predicted
   ([181]Figure 5H). Taken together, these data suggest that pH-induced
   metabolic changes act through AMPK to control cell fate. A model
   diagram summarizing the mechanisms of pH impact on cell fate can be
   found in [182]Figure 5I.

   In summary, our results show that medium pH has a significant impact on
   mesendoderm differentiation through regulation of signaling pathways
   and metabolism. Modulation of environmental pH and cell metabolism
   provide promising tools for stem cell manipulation and cell fate
   determination.

Discussion

   The cellular microenvironment has a significant impact on stem cell
   function. In this report, we demonstrate that environmental pH plays an
   active role in cell fate determination. Medium acidosis and
   alkalinization significantly shift cell fate through changes in
   metabolism and signal transduction, and proper environmental pH is a
   prerequisite for cell type-specific differentiation driven by classic
   inducers.

   Energetic metabolism in hPSCs and mesendoderm progenitors are
   characterized by high levels of glycolysis, which produces large
   amounts of lactic acid as a byproduct in a matter of hours. In reality,
   cells could be in acidic medium most of the time during differentiation
   when cells are cultured at high density. Our results show that, during
   in vitro spontaneous differentiation, the daily pH fluctuation is
   essential for the emergence of diverse cell types. Disturbances of this
   fine pH balance can lead to significant changes in differentiation
   trajectories. Continuously acidic pH environment is sufficient to drive
   mesendoderm progenitors toward cardiomyocytes without WNT inhibition,
   while elevated pH promotes other cell types but inhibits cardiac fate.
   The elevated pH can override WNT inhibition in cardiac induction. These
   outcomes are realized partially through WNT pathway and glycolysis
   regulation. It is still to be explored how these processes interact
   with each other under pH regulation.

   Most stem cell manipulations focus on the modulation of signaling
   pathways, and it is quite surprising that elevated pH could override
   conventional signal modulation in cardiac induction. In order to
   examine the role of pH in other differentiation platforms, we
   characterized the role of medium pH during mesoderm induction
   ([183]Figure S5I). Under BMP4 treatment, expression of mesoderm genes
   such as TBXT and MIXL1 was elevated by alkaline pH and was inhibited by
   acidic pH. At the same time, acidic pH promoted the expression of
   extraembryonic markers TROP2 and CGB ([184]Figure S5J). Apparently,
   environmental pH is involved in cell fate determination not only in
   mesendoderm fates but also in the specification of mesoderm and
   extraembryonic lineages. These findings support the idea that metabolic
   feedback from medium acidity plays important roles in various cell fate
   determination processes.

   Fluctuations in pH have significant impacts on growth, reproduction,
   and development through key cellular processes, including cellular
   metabolism, protein synthesis, and cell cycle ([185]Ceccarini and
   Eagle, 1971; [186]Fukamachi et al., 2013; [187]Karagiannis and Young,
   2001; [188]Mackenzie et al., 1961; [189]Patel et al., 1973). pH has
   also been reported to control major signal transduction pathways, such
   as MAPK ([190]Stathopoulou et al., 2006), TGF-β ([191]Fang et al.,
   2017), and Wnt pathways ([192]Melnik et al., 2018; [193]Oginuma et al.,
   2017, [194]2020; [195]White et al., 2018). Elevated intracellular pH
   was shown to compromise β-catenin stability by promoting β-TrCP binding
   in Madin-Darby canine kidney (MDCK) cells ([196]White et al., 2018).
   Oginuma et al. reported that culturing chicken embryos in low-pH buffer
   reduced the expression of Wnt target AXIN2 and favored differentiation
   toward neural cell fate, while a basic culture environment promoted
   paraxial mesoderm differentiation ([197]Oginuma et al., 2020). In
   another report, acidification of intracellular pH was reported to
   induce the expression of DDIT3 and cause Wnt inhibition, leading to
   decreased stemness in cancer cells ([198]Melnik et al., 2018). In our
   study, we observed inhibition of Wnt signaling by HCl along with
   elevated levels of DDIT3. Elevation of pH by NaHCO[3], on the other
   hand, promoted canonical Wnt signaling and elevated the level of AXIN2
   ([199]Figure S2P). Our findings show that the acid-base balance is a
   key factor of the stem cell niche and determines cell fate through WNT
   and metabolic activities. It is likely that pH could be a pleiotropic
   modulator for stem cell manipulation.

   During in vivo embryogenesis, the early development process from the
   zygote to the blastocyst occurs in the fallopian tube and the uterus.
   The pH of the follicular fluid and uterine fluid is alkaline (up to 7.7
   in human, 7.8 in monkeys, and 7.9 in rabbits and pigs) and rises higher
   during ovulation and early pregnancy ([200]Ng et al., 2018;
   [201]Vishwakarma, 1962), suggesting that embryogenesis takes place in
   an environment less acidic than in vitro cell culture conditions. Our
   results show that an alkaline environment permits the development of
   heterogeneous cell types; the in vivo environment for early
   embryogenesis, being alkaline in pH, therefore supports the development
   of a variety of cells. One interesting finding from this study is that
   cardiac development under alkaline conditions requires not only WNT
   pathway inhibition but also signals from other pathways, for example
   AMPK activation, and potentially other necessary signals. This
   indicates that knowledge on differentiation gained from cell culture
   studies may need to be re-examined in the context of cell
   microenvironment, which is under continuous modulations by cells
   themselves. As in vitro differentiation platforms often involve pH
   fluctuations and a mostly acidic environment, it is possible that some
   key factors important for differentiation may have been overlooked in
   the current research platforms. Further studies examining the
   interaction between medium pH and cell fate determination may reveal
   new mechanisms and provide novel tools for future applications.

   Our results show that changes in medium pH have a significant impact on
   cell fate. However, the impact of pH is often overlooked in common
   practice. pH control provides a new angle to improve stem cell
   differentiation. For example, the efficiency of cardiac induction with
   WNT inhibition often varies significantly from batch to batch, subject
   to plating density, growth rate, cell death, and many other processes
   in the differentiation period. We observed that, when the
   differentiation conditions are not optimal, acidic medium pH can often
   enhance the efficiency of cardiac induction. For example, when cells
   are plated at sub-optimal densities for cardiac differentiation,
   cardiomyocytes generated by WNT inhibition can be as low as <5%. Yet,
   with the modulation of pH by HCl, the efficiency of cardiac induction
   can be increased by 2- to 10-fold ([202]Figure S5K). Interestingly,
   current cardiac differentiation protocols often call for high plating
   density, which may reflect the empiric utilization of acidified medium
   at high cell densities. Our results potentially explain the mechanism
   underlying this common practice. Taken together, pH modulation can be a
   powerful tool to enhance the consistency of stem cell applications.

   In summary, our work reveals the significant impact of environmental pH
   in hESCs and demonstrates effective manipulation of cell fate through
   pH and metabolic modulations. This study provides a valuable tool to
   guide stem cell differentiation for research and clinical applications.

Experimental procedures

Resource availability

Lead contact

   Requests for resources, reagents, and further information should be
   directed to and will be fulfilled by the lead contact, Guokai Chen
   (GuokaiChen@um.edu.mo).

Materials availability

   All reagents generated in this study will be made available by the
   [203]lead contact upon request.

Data and code availability

     * •
       RNA-seq data have been deposited at GEO and are publicly available
       as of the date of publication. The accession number is GEO:
       [204]GSE211196
       .
     * •
       This paper does not report original code.

Experimental model and subject details

Cell lines

   hESC lines H1 and H9 (WiCell Research Institute) and human iPSC lines
   NL-1 and NL-4 (National Institutes of Health) were used in this study
   between passages 25–50. The cell lines were routinely tested for
   mycoplasma status, sterility, genomic integrity, and pluripotency. hESC
   and iPSC studies were approved by the Institutional Review Board at the
   University of Macau.

Method details

Maintenance of hPSCs

   hESCs and iPSCs were maintained in an E8 medium as previously described
   ([205]Beers et al., 2012). Briefly, cells were cultured on
   matrigel-coated 6-well plates in 37°C, 5% CO[2] incubators. The medium
   was changed daily until cells reach 60%–70% confluence. Cells were
   passaged every 3–4 days using the EDTA method in the presence of 5 μM
   Y-27632 (for 24 h after re-plating) at a split ratio of 1:6 to 1:12.

Differentiation of hPSCs

   For spontaneous differentiation of mesendoderm progenitors, hPSCs at
   60%–70% confluence were passaged 1:6 onto matrigel-coated 12-well
   plates in an E8 medium with 5 μM Y-27632 on day 2. The medium was
   changed on day 1 with fresh E8. Cells were switched to a DF medium
   (DMEM/F12, L-ascorbic acid [64 mg/L], sodium selenite [13.6 μg/L],
   transferrin [10μg/mL], chemically defined lipid concentrate (1×),
   penicillin/streptomycin) containing 5 μM CHIR-99021 on day 0 followed
   by a DF medium containing insulin (1 mg/L) on day 1. Starting day 2,
   the resulting mesendoderm progenitor cells were allowed to
   spontaneously differentiate in DF medium without chemical inducers.
   Specific treatments (pH modulation or 2-DG) were applied for designated
   periods of time starting from day 2. 8 mM HCl and 20 mM NaHCO[3] were
   used for pH modulation unless otherwise specified. The medium was
   changed daily. Cells were collected for analysis between day 10–12.

   For cardiac differentiation under WNT inhibition, the same procedures
   were used to induce mesendoderm progenitors with CHIR-99021 as
   described earlier. Starting at day 2, cells were cultured in a DF
   medium with IWP2 (3 μM, applied day 2, 3, and 4) to drive cardiac
   differentiation. Other treatments were applied as specified. The medium
   was changed daily. Cells were collected for analysis between day 10–12.

   For BMP4-induced mesoderm commitment, hESCs at 60%–70% confluence were
   passaged 1:18 onto matrigel-coated 6-well plates in an E8 medium with
   5 μM Y-27632 on day 1 and switched to an E8 medium with 20μg/mL BMP4 on
   day 0. The medium was changed daily. Cells were collected for RNA
   extraction on day 5.

qPCR

   Cells were collected in RNAiso Plus reagent (Takara), and total RNA was
   extracted following the manufacturer’s instructions. cDNA was generated
   by reverse transcription using high-capacity cDNA reverse transcription
   kit (Applied Biosystems) and used for qPCR with SYBR Premix Ex Taq kit
   (Takara). Gene expression was normalized to the level of GAPDH.
   Heatmaps were generated using the online software Heatmapper
   ([206]Babicki et al., 2016).

Western blotting

   Cells were collected in 2× Laemmli buffer, and total protein levels
   were measured using bicinchoninic acid (BCA) protein assay kit (Thermo
   Fisher). After addition of bromophenol blue and β-mercaptoethanol,
   protein lysate was boiled at 95°C for 2 min, and 15-20 μg total
   protein/sample was loaded for SDS-PAGE. After the gel run, proteins
   were transferred onto polyvinylidene fluoride membrane using Bio-Rad
   Mini Trans-Blot cell. The membrane was blocked for 1 h in 5% milk,
   1×TBST (Tris-Buffered Saline with Tween 20), and incubated with primary
   antibodies (diluted in 1% BSA, 1xTBST) at 4°C overnight, followed by
   horseradish peroxidase-conjugated secondary antibodies (diluted in 1%
   BSA, 1×TBST) for 1 h at room temperature. Chemiluminescent signals were
   generated using SuperSignal West Pico PLUS chemiluminescent substrate
   and detected using ChemiDoc MP system.

Flow cytometry

   For flow cytometric analysis, cells in culture plates were washed with
   DPBS/EDTA (0.5 mM), treated with TrypLE select enzyme for 10 min at
   37°C, neutralized with DMEM/F12 + 10% BSA, and dissociated by repeated
   pipetting. Cells were washed with 1xPBS, fixed in 1%
   paraformaldehyde/1xPBS for 10 min at 37°C, and washed again with 1xPBS.
   Cells were then permeabilized in 90% methanol/10% 1xPBS on ice for
   30 min, washed in fluorescence-activated cell sorting (FACS) buffer (1%
   BSA, 0.05% Triton X-100, 1xPBS), and incubated with primary antibodies
   (diluted in FACS buffer) at 4°C overnight. After washing with FACS
   buffer, cells were incubated with secondary antibodies (diluted in FACS
   buffer) for 1 h at room temperature in the dark and then washed with
   1xPBS and analyzed on a CytoFLEX S flow cytometer.

Immunostaining

   Cells were either fixed in-well or passaged into matrigel-coated
   confocal dishes before fixation. Briefly, cells were rinsed with 1×PBS,
   fixed in 4% paraformaldehyde/1×PBS for 15 min at room temperature,
   washed with 1×PBS, and permeabilized in 0.5% Triton X-100/1×PBS for
   15 min at room temperature. After washing, cells were incubated with
   primary antibodies (diluted in 1% BSA/1xPBS) at 4°C overnight, washed,
   and then incubated with secondary antibodies at room temperature for
   1 h in the dark. Cell nuclei were stained with Hoechst for 10 min.
   Imaging was carried out using EVOS FL auto imaging system (Thermo
   Fisher Scientific) or Nikon A1R confocal microscope.

RNA-seq and bioinformatic analysis

   Cells were collected on day 0, 3, 5, 7, and 11 of differentiation, and
   total RNA was extracted using RNAiso Plus reagent (Takara).
   Next-generation sequencing library was prepared using VAHTS Universal
   V8 RNA-seq library prep kit for Illumina, and sequencing was carried
   out using a 2 × 150 bp paired-end configuration on an Illumina NovaSeq
   system. Clean data were aligned to reference genome using the software
   HISAT2 (v2.0.1).

   PCA and differential expression comparison were performed using R
   package PCAtools and DEseq2. For global expression filtration in PCA,
   genes with total expression in all groups greater than one were
   selected. Differential expression calculations were conducted across
   various groups in day 3, 5, and 11 samples. Genes with | log2 (fold
   change) | ≥1.5 and padj ≤0.05 were selected as significant genes. Z
   scores were used to plot heatmaps using the R package pheatmap.
   Function cutree was exerted to split genes into clusters. DEGs were
   used for cell type enrichment and KEGG pathway enrichment analyses by
   Enrichr ([207]Chen et al., 2013; [208]Kuleshov et al., 2016; [209]Xie
   et al., 2021) using ARCHS4_Tissues, Human Gene Atlas, Tabula Sapiens,
   and KEGG 2021 Human.

Accession number

   The accession number for RNA-seq data reported in this paper is GEO:
   [210]GSE211196
   .

LC-MS analysis

   Quantification of metabolites was carried out using LC-MS as previously
   described ([211]Xu et al., 2021). Briefly, H1 hESCs were induced toward
   mesoderm progenitors as described earlier and subjected to control,
   HCl, or NaHCO[3] treatment on day 2 of differentiation. Spent medium
   was collected after 24 h and centrifuged at 1,000 × g for 5 min. The
   supernatant was mixed with nine volumes of medium extraction solution,
   vortexed, and centrifuged at 18,000 × g at 4°C for 10 min. The
   supernatant was used for LC-MS analysis. Samples were run on Waters
   Xevo TQD coupled with Waters ACQUITY UPLC system, and data were
   analyzed using MassLynx with previously reported settings ([212]Xu
   et al., 2021). Changes in metabolite levels were calculated by
   subtracting the level of metabolite in fresh DF medium.

Seahorse extracellular flux analysis

   Extracellular flex assays were carried out using Seahorse XFe96
   analyzer (Agilent). H1 cells were seeded on matrigel-coated XF96
   microplate at 20,000 cells/well and induced toward mesendoderm
   progenitors as described earlier. Cells were subjected to control, HCl,
   NaHCO[3], 2-DG, and IWP2 treatment or their combinations on day 2 of
   differentiation. Metabolic flux assays were carried out on day 3 of
   differentiation.

   For Mito Stress Test, cells were switched to XF Cell Mito Stress Test
   assay medium (Seahorse XF base medium, 10 mM glucose, 1 mM pyruvate,
   2 mM glutamine) with the same treatments applied (the concentrations of
   HCl and NaHCO[3] were decreased to 1 mM and 2 mM, respectively, to
   maintain the pH of the assay medium at similar level as that of the
   HCl- or NaHCO[3]-supplemented DF medium) and were incubated in a
   non-CO[2] incubator at 37°C for 1 h before loading onto the analyzer
   for test. The chemicals used for the assay were oligomycin (2 mM),
   carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP, 1 mM), and
   rotenone/antimycin A (0.5 mM) from the Seahorse XF Cell Mito Stress
   Test kit. After the test, spent medium was removed, cells were lysed in
   lysis buffer (10 mM Tris pH 7.5, 0.1% Triton X-100) on ice, and
   Bradford reagent was added. Data were normalized to absorbance at
   595nm.

   For the Glycolysis Stress Test, cells were switched to XF Glycolysis
   Stress Test assay medium (Seahorse XF base medium, 2 mM glutamine) with
   the same treatments applied (the concentrations of HCl and NaHCO[3]
   were decreased to 1 mM and 2 mM, respectively, to maintain the pH of
   the assay medium at similar level as that of the HCl- or
   NaHCO[3]-supplemented DF medium) and were incubated in a non-CO[2]
   incubator at 37°C for 1 h before loading onto the analyzer for test.
   The chemicals used for the assay were glucose (10 mM), oligomycin
   (2 mM), and 2-DG (50 mM) from the Seahorse XF Glycolysis Stress Test
   kit. Cell lysis and data normalization were similar to the
   aforementioned Mito Stress Test.

Quantification and statistical analysis

   Flow cytometry data are presented as mean ± standard deviation (SD) of
   three independent experiments unless specified. qPCR data are presented
   as mean ± SD of three technical replicates and are representative of
   three or more independent experiments unless specified. Statistical
   significance was determined using unpaired two-tailed Student’s t test
   for experiments with two treatment groups, and one-way ANOVA with
   Dunnett’s multiple comparison test for experiments with multiple
   groups. ^∗, p < 0.05; ^∗∗, p < 0.01; ^∗∗∗, p < 0.001; NS, not
   statistically significant.

Acknowledgments