Both Hypoxia-Inducible Factor 1 and MAPK Signaling Pathway Attenuate PI3K/AKT via Suppression of Reactive Oxygen Species in Human Pluripotent Stem Cells

Mild hypoxia (5% O2) as well as FGFR1-induced activation of phosphatidylinositol-4,5-bisphosphate 3-kinase/protein kinase B (PI3K/AKT) and MAPK signaling pathways markedly support pluripotency in human pluripotent stem cells (hPSCs). This study demonstrates that the pluripotency-promoting PI3K/AKT signaling pathway is surprisingly attenuated in mild hypoxia compared to the 21% O2 environment. Hypoxia is known to be associated with lower levels of reactive oxygen species (ROS), which are recognized as intracellular second messengers capable of upregulating the PI3K/AKT signaling pathway. Our data denote that ROS downregulation results in pluripotency upregulation and PI3K/AKT attenuation in a hypoxia-inducible factor 1 (HIF-1)-dependent manner in hPSCs. Using specific MAPK inhibitors, we show that the MAPK pathway also downregulates ROS and therefore attenuates the PI3K/AKT signaling—this represents a novel interaction between these signaling pathways. This inhibition of ROS initiated by MEK1/2–ERK1/2 may serve as a negative feedback loop from the MAPK pathway toward FGFR1 and PI3K/AKT activation. We further describe the molecular mechanism resulting in PI3K/AKT upregulation in hPSCs—ROS inhibit the PI3K's primary antagonist PTEN and upregulate FGFR1 phosphorylation. These novel regulatory circuits utilizing ROS as second messengers may contribute to the development of enhanced cultivation and differentiation protocols for hPSCs. Since the PI3K/AKT pathway often undergoes an oncogenic transformation, our data could also provide new insights into the regulation of cancer stem cell signaling.


Mild
hypoxia (5% O 2 ) as well as FGFR1-induced activation of phosphatidylinositol-4,5-bisphosphate 3-kinase/protein kinase B (PI3K/AKT) and MAPK signaling pathways markedly support pluripotency in human pluripotent stem cells (hPSCs). This study demonstrates that the pluripotency-promoting PI3K/AKT signaling pathway is surprisingly attenuated in mild hypoxia compared to the 21% O 2 environment. Hypoxia is known to be associated with lower levels of reactive oxygen species (ROS), which are recognized as intracellular second messengers capable of upregulating the PI3K/AKT signaling pathway. Our data denote that ROS downregulation results in pluripotency upregulation and PI3K/AKT attenuation in a hypoxia-inducible factor 1 (HIF-1)-dependent manner in hPSCs. Using specific MAPK inhibitors, we show that the MAPK pathway also downregulates ROS and therefore attenuates the PI3K/AKT signaling-this represents a novel interaction between these signaling pathways. This inhibition of ROS initiated by MEK1/2-ERK1/2 may serve as a negative feedback loop from the MAPK pathway toward FGFR1 and PI3K/AKT activation. We further describe the molecular mechanism resulting in PI3K/AKT upregulation in hPSCs-ROS inhibit the PI3K's primary antagonist PTEN and upregulate FGFR1 phosphorylation. These novel regulatory circuits utilizing ROS as second messengers may contribute to the development of enhanced cultivation and differentiation protocols for hPSCs. Since the PI3K/AKT pathway often undergoes an oncogenic transformation, our data could also provide new insights into the regulation of cancer stem cell signaling.

INTRODUCTION
Human pluripotent stem cells (hPSCs) hold great promise for disease modeling, development of cell replacement therapies, and human embryology research due to their unique properties such as self-renewal and pluripotency. Traditionally, pluripotent stem cells can be divided into two distinct categories with very similar gene expression profiles: embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs). While ESCs can be derived directly from preimplantation blastocysts, iPSCs are created by reprogramming somatic cells (Takahashi et al., 2007). Preimplantation blastocysts used for the preparation of ESCs naturally reside in a low oxygen environment (Fischer and Bavister, 1993;Okazaki and Maltepe, 2006). It has been shown that low oxygen conditions prevent differentiation and support pluripotency in various stem cell populations (Ezashi et al., 2005;Forristal et al., 2010) and enhance the generation of human iPSCs from fibroblasts (Yoshida et al., 2009). Mathieu et al. (2013) have reported that the transition of committed cells to hypoxic conditions can reverse the differentiation commitment back to full pluripotency. Despite these findings, hPSCs are commonly maintained in a 21% O 2 (atmospheric oxygen concentration), although hypoxic conditions would seem more physiological (for the purposes of this publication, we will further refer to 5% O 2 as mild hypoxia). In order to maintain hPSCs self-renewal and pluripotency in vitro, fibroblast growth factor 2 (FGF2) is commonly used (Xu et al., 2005b;Levenstein et al., 2006;Eiselleova et al., 2009). Upon binding to fibroblast growth factor receptor (FGFR), FGF2 activates MAPK, phosphatidylinositol-4,5-bisphosphate 3-kinase/protein kinase B (PI3K/AKT), phospholipase Cγ as well as the Janus kinase/signal transducers and activators of transcription pathways.
To fully utilize the hPSCs potential of self-renewal and differentiation into the desired cell types, it is essential to understand the signaling pathways driving cell fate decision. Since hypoxic conditions support pluripotency maintenance (Mathieu et al., 2013), we investigated whether mild hypoxia modulates FGF2 signaling pathways and their crosstalk in hPSCs. Our data suggest that ROS serve as second messengers activating PI3K/AKT. To the best of our knowledge, we are the first to report that ROS downregulation caused by mild hypoxia and MAPK activation attenuates the pluripotency-maintaining PI3K/AKT signaling in hPSCs.
The membranes were blocked in 5% dried milk or bovine serum albumin in Tris-buffered saline containing 0.1% Tween-20 (TBS-T) for 1 h and incubated overnight with antibodies diluted in the respective blocking solutions at 4 • C. The following day, membranes were washed 3 × 15 min with TBS-T and incubated with secondary antibodies diluted in respective blocking solution for 1 h at room temperature followed by 5 × 10 min washes with TBS-T. Immobilon Western Chemiluminescent HRP Substrate (Merck Millipore, P90720) was used as a substrate for the luminescence reaction. Developing was done in G:Box Chemi (SYNGENE, Bangalore, India). Obtained images were adjusted using the GIMP2 software and analyzed in ImageJ. Uncropped blots with molecular markers are shown in Raw Images 1.
For the detection of PTEN, redox state cells were harvested in native lysis buffer [100 mM Tris pH 7, 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 0.1% Triton X-100] supplemented with the cOmplete Mini Protease Inhibitor Cocktail (Roche, Basel, Switzerland; 11836153001) and 50 mM N-ethylmaleimide (Sigma-Aldrich; 04259). Samples were sonicated, mixed with a non-reducing loading buffer, and resolved using 8% SDS-PAGE at 4 • C. Protein transfer and detection were performed as described above.
WB quantification analysis was performed using the ImageJ software. Mean gray values for individual proteins were measured using rectangle selection with a diameter fitted tightly to the largest band of that specific protein in a given WB. The same-sized rectangle selection was used for the measurement of all specific protein bands in the lane. Eventual background was subtracted using the same selection rectangle next to the specific protein band. Proteins of interest were then normalized to a loading control. Values chosen for relativization are indicated in the individual figure legends.

Analysis of Reactive Oxygen Species Generation
Cells grown on Matrigel-coated coverslips were treated with the above-described substances. Fifty minutes before the end of the treatment, CellROX Green (5 mM; Thermo Fisher Scientific, C10444) was added. Afterward, the cells were washed three times in PBS on ice and fixed with 4% paraformaldehyde (Sigma-Aldrich, 158127) for 30 min. Snapshots of the approximately same-sized cell clusters were taken with constant exposure time using a fluorescent LSM700 microscope (×40 1.3 oil differential interference contrast objective; Carl Zeiss, Oberkochen, Germany) within 6 h of fixation. ROS levels were determined as fluorescence divided by the fluorescence area in raw images using the ImageJ software.
of 1:200 overnight at 4 • C. Donkey antirabbit Alexa 594 (Thermo Fisher Scientific, A21207) and donkey antimouse Alexa 488 (Thermo Fisher Scientific, A21202) were used as secondary antibodies at a dilution of 1:500 at room temperature for 1 h. Nuclei were counterstained using 4 ′ ,6-diamidino-2-phenylindole (DAPI). Snapshots were taken with constant exposure time for individual channels using a fluorescent LSM700 microscope (×63 oil immersion objective; Carl Zeiss). Signal density per area was calculated for individual nuclei from raw images using ImageJ, and mitotic, overlapping, or otherwise irregular nuclei were left out of the analysis.

RNA Isolation and Quantitative Real-Time PCR
Total RNA was isolated using the RNA Blue reagent (Top-Bio, Czech Republic) according to the manufacturer's protocol. Messenger RNA (mRNA) concentration and purity were determined using NanoDrop (NanoDrop Technologies, Wilmington, Germany). Two micrograms of total RNA were transcribed into complementary DNA (cDNA) using the Moloney Mouse Leukemia Virus (M-MLV) reverse transcriptase (Invitrogen, Carlsbad, CA, USA) and Oligo(dT) primers (Thermo Fisher Scientific Inc., USA) at 37 • C for 1 h followed by 5 min at 85 • C. Quantitative real-time PCR (qRT-PCR) was performed using the LightCycler R 480 DNA SYBR Green I Master (Roche) in a Light Cycler 480 instrument. The obtained data were normalized to glyceraldehyde 3phosphate dehydrogenase (GAPDH) mRNA expression and are presented as 2-cq. Sequences of the primers used were as follows: GAPDH (forward: AGCCACATCGCTCAGACACC; reverse: GTACTCAGCGCCAGCATCG), POU5F1 (forward: GCAAAGCAGAAACCCTCGT; reverse: ACACTCGGACC ACATCCTTC), Sox2 (forward: ATGCACCGCTACGACGTGA; reverse: CTTTTGCACCCCTCCCATTT), and Nanog (forward: CCTATGCCTGTGATTTGTGG, reverse: CTGGGACCTTGT CTTCCTTT).

Statistical Analysis
The number of independent experiments is indicated in the figure legends. Arithmetical means and SEM/SD were calculated using the GraphPad Prism 8 software (GraphPad Software, La Jolla, CA, USA). Statistical significance was determined using a two-tailed Mann-Whitney test, one-sample Student's ttest, or a paired two-tailed Student's t-test. Asterisks denote a significant difference as follows: * p < 0.05; * * p < 0.01; * * * p < 0.0001. Source data used to generate the graphs are shown in Supplementary Data 1.
imaging. Compared to 21% O 2 , the levels of ROS in CCTL14 cells cultivated in mild hypoxia were significantly decreased ( Figure 1C).
It has been previously described that ROS are able to activate receptor tyrosine kinases (RTKs), either directly causing their autophosphorylation (Chiarugi and Buricchi, 2007) or by inhibition of protein tyrosine phosphatases (Chiarugi and Cirri, 2003;Östman et al., 2011). Thus, to pinpoint the particular mechanism by which ROS modulate PI3K/AKT signaling in hPSCs, we looked upstream of AKT at FGF receptor 1 (FGFR1), the most abundantly expressed FGF receptor in hESCs (Dvorak et al., 2005). Using Western blot, we observed a significantly lower level of FGFR1 Tyr653/654 activating phosphorylation in mild hypoxia compared to 21% O 2 . The treatment with H 2 O 2 (0.5 mM/1 h) led to upregulation of FGFR1 phosphorylation in both 21 and 5% O 2 , while the GSH treatment (10 mM/1 h) led to a significant downregulation of FGFR1 phosphorylation only in 21% O 2 , in a fashion similar to AKT phosphorylation ( Figure 3A). This was also observed in the CCTL12 hESC (Supplementary Figure 3A) and AM13 iPSCs (Supplementary Figure 3B). To elucidate the role of ROS-induced FGFR1 dimerization and autophosphorylation under different oxygen concentrations, we compared FGFR1 phosphorylation with and without exogenous stimulation by FGF2 in 21% O 2 , 5% O 2 , and in H 2 O 2 (1 mM/1 h)-treated cells. Our results show that H 2 O 2 -mediated ROS induction is sufficient to induce FGFR1 phosphorylation regardless of the exogenous stimulation by FGF2. ROS, therefore, seem to induce dimerization and autophosphorylation of FGFR1 in hPSCs ( Figure 3B). Moreover, a stronger FGFR1 phosphorylation was detected in cells in 21% O 2 compared to 5% O 2 without FGF2 stimulation (Figure 3B), indicating that the difference in ROS concentration between 21% O 2 and 5% O 2 environment ( Figure 1C) is sufficient to promote FGFR autophosphorylation independent of exogenous stimuli.
When inhibited, the PI3K's regulatory subunit p55/85 is bound to the p110 catalytic subunit. Upon phosphorylation, p55/85 releases p110, which then catalyzes the conversion of phosphatidylinositol 4,5-diphosphate to phosphatidylinositol 3,4,5-triphosphate (PIP 3 ). Figure 3C shows that PI3K p85 phosphorylation was upregulated in 21% O 2 when compared to mild hypoxia, and H 2 O 2 led to a massive upregulation of PI3K p85 phosphorylation under both conditions. Treatment with GSH resulted in a significant downregulation of PI3K p85 phosphorylation ( Figure 3C) corresponding with the GSH-induced changes in FGFR1 ( Figure 3A and Supplementary Figures 3A,B) and AKT phosphorylation ( Figure 2C and Supplementary Figures 2A,B).
These observations were also made in CCTL12 hESCs (Supplementary Figure 3C) and AM13 iPSCs (Supplementary Figure 3D). Taken together, these findings suggest that ROS upregulate FGFR1 phosphorylation either directly or via inhibition of phosphatases, which subsequently leads to PI3K and AKT phosphorylation.

ROS Are Attenuated in HIF-1-Dependent Manner
To understand what contributes to the ROS downregulation in mild hypoxia, we looked at the effect of HIFs, the main facilitators of cellular adaptation to hypoxic conditions (Semenza, 2001;Keith et al., 2012). First, we employed CoCl 2 (50 mM), a hypoxia mimetic capable of stabilizing the alpha subunits of HIFs in 21% O 2 (Figure 4A). In our cells, CoCl 2 treatment led to a decrease in AKT phosphorylation, comparable to the decrease observed in mild hypoxia (Figures 1A,B and Supplementary Figures 1A,B). It has been previously established that CoCl 2 stabilizes alpha subunits of all HIFs, but only HIF-1α is transcriptionally active under CoCl 2 treatment (Befani et al., 2013). Next, we silenced HIF-1α expression using endoribonuclease-prepared shortinterfering RNA (esiRNA). This caused a decrease in ERK1/2 phosphorylation and an increase in AKT phosphorylation independent of FGF2 presence (Figure 4B), which was also observed in AM13 iPSCs (Supplementary Figure 4A). To determine if HIF-1α-mediated phospho-AKT downregulation is also associated with the changing levels of ROS, we compared ROS levels in cells maintained in 5% O 2 , in hESCs treated and untreated with esiRNA targeting HIF-1α, and in cells grown in 21% O 2 . The silencing of HIF-1α expression induced a significant increase in ROS in 5% O 2 , an amount comparable to that observed in 21% O 2 ( Figure 4C). These results suggest that the upregulation of AKT phosphorylation observed in hPSCs following HIF-1α silencing is associated with ROS upregulation.

MAPK Downregulate PI3K/AKT via ROS Attenuation
Since we observed a simultaneous increase in AKT phosphorylation and a decrease in ERK1/2 phosphorylation following the HIF-1α knockdown by esiRNA, we wondered whether MAPK could play a role in PI3K/AKT regulation in hPSCs. Crosstalk between the MAPK and PI3K/AKT pathway was previously described (Aksamitiene et al., 2012). In order to analyze the effect of MAPK on PI3K/AKT, we used two small molecule inhibitors of mitogen-activated protein kinase kinase 1 and 2 (MEK1/2), PD184161 (PD18), and PD0325901 (PD03) to treat cells cultivated either in 21% O 2 or 5% O 2 . Comparing the results, it can be seen that the downregulation of ERK1/2 phosphorylation leads to the upregulation of AKT phosphorylation in 21% O 2 in the presence of FGF2 in CCTL14 and CCTL12 hESCs and AM13 iPSCs (Figure 5A and Supplementary Figures 5A,B). These results suggest a negative impact of FGF2-induced MAPK signaling on PI3K/AKT pathway activation in hPSCs.
To assess whether MAPK signaling modulates ROS, we analyzed ROS levels in hESC line CCTL14 treated with a MEK1/2 inhibitor PD03 under both 21 and 5% O 2 . MEK1/2 inhibition leads to a statistically insignificant (p = 0.145) upregulation of ROS in 21% O 2 and a significant upregulation of ROS in 5% O 2 ( Figure 5B) compared to untreated cells maintained in corresponding O 2 concentrations. Considering that both the maintenance of cells in 21% O 2 and MEK1/2 inhibition in mild hypoxia were sufficient to raise ROS and elevated ROS are associated with increased AKT phosphorylation, it is possible that MAPK suppresses PI3K/AKT signaling via modulation of ROS levels as second messengers.
We, therefore, combined the ROS scavenger GSH (10 mM) and MEK1/2 inhibitor PD03 in 21% O 2 . This led to a rescue of the PD03-induced AKT phosphorylation (Figure 5C), implicating that MAPK downregulates AKT phosphorylation via modulation of ROS. Same results were acquired in CCTL12 hESCs (Supplementary Figure 5C) and AM13 iPSCs (Supplementary Figure 5D). To further validate this hypothesis, we also looked at FGFR1 Tyr653/654-activating phosphorylation, which we previously detected to be upregulated by ROS (Figures 3A,B and Supplementary Figures 3C,D). Our results show that FGFR1 phosphorylation is also upregulated by MEK1/2 inhibition, and the addition of GSH rescues this effect in 21% O 2 in all three hPSCs lines used in this study (Figure 5D and Supplementary Figures 5C,E). In summary, our results suggest that MAPK can downregulate PI3K/AKT phosphorylation via the downregulation of second messengers-intracellular ROS.

SHIP2 and PP2A Are Not Involved in the Immediate Mild Hypoxia-Induced Downregulation of AKT Phosphorylation
SHIP2 is an antagonist of PI3K functionally similar to PTEN that is described to be regulated by ROS .
To analyze whether SHIP2 facilitates the downregulation of AKT phosphorylation in mild hypoxia, we treated hPSCs in 5% O 2 with a specific SHIP2 inhibitor AS1938909 (AS19, 10 µM/2 h). hPSCs treated with AS19 did not display a significantly elevated AKT phosphorylation than untreated control ( Figure 6F and Supplementary Figures 7A,B), suggesting that SHIP2 is not involved in the observed downregulation of PI3K/AKT in mild hypoxia. PP2A directly regulates AKT and ERK1/2 phosphorylation and is also described to be regulated by ROS (Rao and Clayton, 2002;Raman and Pervaiz, 2019). To analyze its role in the regulation of AKT in hPSCs, we used a 2-h okadaic acid (OKA) treatment in the 0.5 nM concentration specific for PP2A and in the 10 nM concentration specific for protein phosphatase 1 (PP1). Treatment with OKA in both concentrations did not significantly enhance AKT or ERK1/2 phosphorylation in hPSCs regardless of O 2 concentration ( Figure 6G and Supplementary Figures 7C,D). This, together with the fact that PP2A targets both AKT and ERK1/2 and that we did not see ERK1/2 downregulation in mild hypoxia, suggests that it has no role in the observed hypoxia-mediated downregulation of AKT phosphorylation in hPSCs.
requires longer incubation times (Forristal et al., 2010;Mathieu et al., 2014). We, therefore, analyzed the Oct-3/4 and Nanog levels after 24 and 48 h in the 5% O 2 environment. Upon nuclear signal quantification, we observed a significant increase in the nuclear Oct-3/4 and Nanog signal in 5% O 2 after 24 and 48 h as well as in the GSH-treated cells compared to control in 21% O 2 ( Figure 7A). To distinguish between the amount of protein and changes in the pluripotency markers' gene expression, we also performed quantitative real-time PCR (qRT-PCR) of NANOG, POU5F1, and SOX2 comparing cells cultivated in 21 and 5% O 2 and cells with ROS scavenged by GSH (10 mM) in 21% O 2 ( Figure 7B). We observed an increase in NANOG, POU5F1, and SOX2 expression following the GSH treatment comparable to levels of expression seen in cells grown in 5% O 2 . qRT-PCR results correlate with the observed significant changes in protein amount, emphasizing the importance of ROS signaling in pluripotency maintenance.

ROS Upregulate AKT Phosphorylation in hPSCs and Are Downregulated in Mild Hypoxia
We show that the 5% O 2 environment (referred to as mild hypoxia for the purpose of this study) stabilizes HIF-1α and downregulates AKT phosphorylation in hPSCs, while it does not have a significant effect on MEK1/2-ERK1/2 pathway (Figures 1A,B and Supplementary Figures 1A,B). Hypoxia was previously associated with decreased ROS levels (Kučera et al., 2017), and ROS were shown to regulate various cellular signaling pathways (Rhee, 2006;Genestra, 2007;Zhang et al., 2016). ROS were even found to be upregulating the MAPK and PI3K/AKT signaling in mESCs where hypoxia was associated with downregulation of these pathways (Kučera et al., 2017).
Decreased ROS levels are associated with mild hypoxia in hESCs as well ( Figure 1C). Their selective downregulation by GSH (Figure 2A) led to PI3K/AKT downregulation independent of oxygen status (Figure 2C and Supplementary Figures 2A,B). Vice versa, selective upregulation of ROS by H 2 O 2 ( Figure 2B) independent of oxygen concentration led to PI3K/AKT and MAPK upregulation (Figure 2C and Supplementary Figures 2A,B). Considering that ROS levels directly respond to oxygen concentration (Figure 1B), ROS seem to act as second messengers responsible for PI3K/AKT upregulation in response to oxygen level. H 2 O 2 also upregulated ERK1/2 phosphorylation in hPSCs, but ERK1/2 phosphorylation changed neither upon O 2 concentration changes ( Figure 1A and Supplementary Figures 1A,B) nor in the presence of GSH in hPSCs ( Figure 2C and Supplementary Figures 2A,B) contrary to mESCs (Kučera et al., 2017). A possible explanation is that MAPK serves different roles in mESCs and hESCs. Upregulation of MAPK in mESCs leads to lineage commitment (Kunath et al., 2007), but self-renewal of hPSCs relies on highly active MAPK (Dvorak et al., 2005;Eiselleova et al., 2009). Thus, high MAPK activity upon FGF2 induction possibly makes hPSCs less sensitive to subtle changes in ROS induced by O 2 concentration changes. Above that, our data also indicate that the MEK1/2-ERK1/2 pathway attenuates ROS ( Figure 5B); the existence of a positive feedback loop from ROS toward mitogenic pathways is, therefore, possible, perhaps on the level of the FGF2 receptor.
O 2 Activates FGFR1 and PI3K via ROS in hPSCs ROS have been described to activate receptor tyrosine kinases (RTKs) through dimerization induced by the oxidationmediated formation of disulfide bonds between the cysteines of neighboring monomers (Chiarugi and Buricchi, 2007) or by inhibition of protein tyrosine phosphatases via cysteine oxidation (Chiarugi and Cirri, 2003;Chiarugi and Buricchi, 2007). Indeed, phosphorylation of FGFR1, the most abundantly expressed FGFR in hESCs (Dvorak et al., 2005), was upregulated by both ROS and O 2 in hPSCs (Figures 3A,B and Supplementary Figures 3A,B), suggesting this may contribute to the ROS-mediated upregulation of AKT phosphorylation. The upregulation of FGFR1 phosphorylation by H 2 O 2 and O 2 was observed without the exogenous FGF2 stimulation as well (Figure 3B), implying that oxidation mediates the FGFR1 phosphorylation either via promoting its dimerization and autophosphorylation or via inhibition of FGFR1-associated phosphatases.
We also observed ROS-induced phosphorylation of the PI3K regulatory subunit p85, which is necessary for the release and activation of the catalytic p110 subunit. Possibly, this is a result of upstream ROS-mediated FGFR1 phosphorylation. Nevertheless, the lack of O 2 -induced FGFR1-MEK1/2-ERK1/2 response denotes that ROS could directly modulate PI3K p85 in hPSCs (Figure 3C and Supplementary Figures 3C,D) as previously described in human mammary epithelial cells (Okoh et al., 2013). ROS-induced p85 phosphorylation would explain the ROS-induced increase in AKT phosphorylation ( Figure 2C). Alternatively, ROS could inhibit phosphatases involved in PI3K/AKT regulation, for example PTEN, PP2A, or SHIP2. Such mechanism has been described in mouse skeletal muscle cells, embryonic fibroblasts, or HeLa cells (Lee et al., 2002;Zhang et al., 2007;Kim et al., 2018;Raman and Pervaiz, 2019).

AKT Phosphorylation Is Upregulated by ROS-Mediated Downregulation of PTEN Activity
To elucidate the mechanism behind the downregulation of AKT phosphorylation by ROS, we focused on PTEN-a well-established PI3K antagonist (Leslie and Downes, 2002). Indeed, PTEN is probably involved in the attenuation of AKT phosphorylation in hPSCs because the downregulation of AKT phosphorylation can be reverted by PTEN silencing (Figure 6C and Supplementary Figures 6E,G). According to the literature, ROS-mediated cysteine oxidation renders PTEN inactive (Lee et al., 2002;Covey et al., 2007). Furthermore, we found the effect of PTEN silencing on AKT phosphorylation to be more profound in mild hypoxia than in 21% O 2 ( Figure 6C and Supplementary Figures 6E,G) and confirmed that ROS are capable of oxidizing PTEN in hPSCs ( Figure 6E). ROS scavenging by GSH in 21% O 2 also led to lower AKT phosphorylation, which was rescued by PTEN silencing (Figure 6D and Supplementary Figures 6F,G). These findings further support the hypothesis that ROS mediate downregulation of PTEN activity in hPSCs. AKT itself was also shown to be reversibly oxidized by ROS, which strengthens its PIP 3 binding pocket, recruitment to the plasma membrane, and its activation (Su et al., 2019). Our results show that the possible AKT oxidation does not induce phosphorylation without the activity of PI3K ( Figure 2D). Seemingly, the ROS-mediated strengthening of the PIP 3 binding pocket on the AKT molecule (Su et al., 2019) cooperates with the reversible oxidation of PTEN to induce AKT phosphorylation.
PTEN was shown to be also regulated on the level of transcription, protein stability, and localization (Leslie and Downes, 2002). Since the total amount of PTEN remained stable when comparing 21 and 5% O 2 (Figure 6A and Supplementary Figures 6A,B) and cells treated with GSH and H 2 O 2 ( Figure 6B and Supplementary Figures 6C,D), downregulation of AKT phosphorylation in hPSCs does not appear to be driven by changes in PTEN total amount or stability. The extent of PTEN phosphorylation can impact its localization (Vazquez et al., 2000), but similarly to the total amount, we did not detect any changes in PTEN phosphorylation in different oxygen tensions (Figure 6A and Supplementary Figures 6A,B) or in the presence of a different amount of ROS ( Figure 6B and Supplementary Figures 6C,D). These data together suggest the involvement of only cysteine oxidation in the ROS-mediated regulation of PTEN activity in hPSCs.
We also analyzed the role of SH2-domain-containing inositol-5 ′ -phosphatase (SHIP2). This enzyme is known to antagonize PI3K in a manner similar to PTEN and is described to be regulated by ROS . As expected, 2-h inhibition of SHIP2 by AS19 had no effect on AKT phosphorylation in hPSCs ( Figure 6F and Supplementary Figures 7A,B) since the effect of SHIP2 on AKT phosphorylation starts to appear after 6 h (Fafilek et al., 2018). While SHIP2 probably does play a role in long-term hypoxia-induced regulation of AKT, our data suggest that it is not involved in the immediate dynamic response to O 2 -induced ROS regulation of PI3K/AKT pathway in hPSCs.
ERK1/2 and AKT dephosphorylation have also been previously attributed to protein phosphatase 2A (PP2A). PP2A has been described to be negatively regulated by ROS, which might contribute to its activation in mild hypoxia (Rao and Clayton, 2002;Raman and Pervaiz, 2019). However, PP2A inhibition had no significant effect on AKT phosphorylation in hPSCs cultivated in 5 or 21% O 2 ( Figure 6G and Supplementary Figures 7C,D). Similarly, no changes in ERK1/2 phosphorylation were detected in our cells ( Figure 1A and Supplementary Figures 1A,B). Therefore, PP2A does not seem to play a significant role in O 2 -induced ROS regulation of the PI3K/AKT pathway in hPSCs.
Even though our results stem from an artificial MEK1/2-ERK1/2 manipulation, MEK1/2-ERK1/2-induced inhibition of ROS suggest this mechanism may serve as a negative feedback loop from MAPK toward FGFR1. A similar negative feedback loop involving p38 has already been described in the literature (Zakrzewska et al., 2019). The detailed mechanism by which MEK1/2-ERK1/2 contributes to the ROS downregulation is so far unclear, but the MAPK pathway was shown to play an important role in metabolic reprogramming and upregulation of glycolysis (Papa et al., 2019). Glycolysis provides its intermediates to the pentose phosphate pathway (PPP), a significant NADPH source. NADPH is known to contribute to the ROS scavenging by providing its reductive potential to GSH and thioredoxins, consequently utilized to neutralize ROS (Hanschmann et al., 2013). It is, therefore, possible that MEK1/2-ERK1/2 may downregulate ROS in hPSCs via upregulation of NADPH production in PPP.

ROS Scavenging Upregulates Pluripotency Markers
Several studies have shown that a hypoxic environment improves hPSCs pluripotency (Ezashi et al., 2005;Mathieu et al., 2013), which was linked to HIFs-induced pluripotency gene expression (Forristal et al., 2010;Mathieu et al., 2014). Mild hypoxia-induced ROS downregulation might contribute to the pluripotency maintenance, at least according to our data showing that ROS scavenging with GSH in 21% O 2 upregulates pluripotency markers on levels similar to those observed in 5% O 2 (Figures 7A,B). This is in concert with a previous study showing that ROS are able to induce hPSCs differentiation (Ji et al., 2010). The role of PI3K/AKT attenuation due to ROS downregulation in pluripotency maintenance is unclear, but it could be an interesting topic of further studies, since PI3K/AKT signaling directs cell fate decision upon differentiation priming (Yu and Cui, 2016).
Another possible mechanism for ROS-mediated regulation of pluripotency is integrative nuclear FGFR1 signaling (INFS), which was shown to downregulate the expression of core pluripotency genes and to be instrumental in neural differentiation (Terranova et al., 2015). INFS-induced localization of FGFR1 in the nucleus is induced by activation of cell surface receptors (Stachowiak et al., 2007) and may include interaction with p85α (Dunham et al., 2004). Cell surface FGFR1 phosphorylation can be promoted by ROS as shown by us (Figures 3A,B) and others (Chiarugi and Cirri, 2003;Chiarugi and Buricchi, 2007). Similarly, the nuclear localization of FGFR2 negatively regulates HIFs in prostate cancer (Lee et al., 2019). We observed both the cytoplasmic and nuclear localization of FGFR1 in hPSCs using ICC (Supplementary Figure 8). We did not observe downregulation of FGFR1 nuclear localization after GSH or 5% O 2 treatment, but it is possible that this mechanism is employed in hPSCs but is under the ICC detection threshold. Such a mechanism would lead to FGFR1-activation-dependent downregulation of the pluripotency signaling network in response to ROS and consequent differentiation.
Taken together, our data show that the PI3K/AKT pathway in hPSCs is upregulated by ROS. The upregulation is secured by an increase in FGFR1-activating Tyr653/654 phosphorylation, PI3K p85 phosphorylation, and a decrease in PTEN activity. We further show that MAPK pathway and also mild hypoxia (in HIF-1-dependent manner) attenuate ROS and thus downregulate but do not completely shut off the PI3K/AKT signaling (Figure 8)a pathway essential for hPSCs pluripotency maintenance, selfrenewal, and cell fate decision. Such mechanism differs from observations made in mESCs (Kučera et al., 2017), possibly due to a different pluripotency status. Pluripotency maintenance and iPSCs reprogramming was shown to benefit from mild hypoxia, an environment native to the blastocyst; therefore, it is counterintuitive that it leads to PI3K/AKT downregulation. Higher PI3K/AKT activity in 21% O 2 could theoretically compensate for the missing hypoxia-induced pluripotency signaling. On the other hand, the PI3K/AKT pathway has also been implicated in the regulation of differentiation, as summarized by Yu and Cui (Yu and Cui, 2016). It seems that a precise balance in signaling pathways activity helps to maintain pluripotency, and swings or disbalances in their activity induced by signaling molecules and external factors can affect the fragile balance between pluripotency and differentiation. We hypothesize that a mildly hypoxic environment via HIF-1 and also the MEK1/2-ERK1/2 pathway might help to maintain such balance in PI3K/AKT (and possibly other ROS-sensitive pathways) activity in hPSCs by controlling levels of second messengers-ROS. We propose that introducing ROS control into the hPSCs maintenance and differentiation protocols might thus lead to their significant improvement (Figures 7A,B). Because of the similarity between hPSCs and cancer stem cells and because the PI3K/AKT pathway often undergoes an oncogenic transformation, our results might also help deepen the current understanding of cancer biology.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Materials, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
PF performed cell cultivation, WB with analysis, ROS analysis, and ICC analysis. DB performed qRT-PCR and analyzed the data. KH performed cell cultivation and Western blot analysis. MŠ performed WB and ICC. VR and PF wrote the manuscript and designed all experiments. All authors contributed to the article and approved the submitted version.