FOXO nuclear shuttling dynamics are stimulus-dependent and correspond with cell fate

INTRODUCTION

The forkhead box O (FOXO) proteins are a family of transcription factors (TFs) that regulate homeostasis at both the cellular and tissue levels and have an evolutionarily conserved role in longevity (Kenyon et al., 1993; Timmers et al., 2019). There are four different FOXO genes in humans: FOXO1, FOXO3, FOXO4, and FOXO6. FOXO proteins are activated in response to a number of cellular stresses, including growth factor deprivation, metabolic stress, and oxidative stress. Upon activation, FOXO TFs up-regulate the transcription of hundreds of genes with diverse biological functions, including cell-cycle arrest (p21, p27), ROS scavenging (SOD2, CAT), autophagy (Bnip3, ATF4), inhibition of FOXO activity (PI3K, EGFR, RICTOR), and apoptosis (Bim, PUMA; Webb et al., 2016). The cellular consequences of FOXO activation are both stress- and cell type–specific. For example, during quiescence, FOXO proteins play a protective role, up-regulating ROS scavengers and autophagy genes that are required for the long-term viability of stem cells (Kops et al., 2002). In contrast, FOXO activity following H₂O₂ treatment or oncogene inhibition can lead to cell death through the up-regulation of apoptotic genes (Lehtinen et al., 2006; Bean et al., 2013).

The conflicting outcomes of FOXO activation suggest a need for precise control of FOXO activity to ensure that the response is commensurate with the stress. An attractive model for how input/output specificity is achieved by FOXO TFs is the idea of a “FOXO code” (Calnan and Brunet, 2008). In this model, posttranslational modifications (PTMs) alter the binding of FOXOs to different cofactors that control the level of FOXO TFs or their binding to the promoters of specific target genes. There is a lot of evidence to support this model. FOXO proteins can be modified by dozens of PTMs, and a subset of these PTMs have been shown to alter FOXO target gene expression (Wang et al., 2016). For example, SIRT1 can deacetylate FOXO proteins, and inhibition of SIRT1 increases the ability of ectopically expressed FOXO3 to induce cell-cycle arrest while inhibiting apoptosis (Brunet et al., 2004). Moreover, mutations in the FOXO1 gene that prevent acetylation lead to the uncoupling of weight gain and insulin resistance in mice (Banks et al., 2011).

In addition to altering target gene specificity, FOXO PTMs likely provide an additional layer of control by altering the dynamics of FOXO TF shuttling in and out of the nucleus. FOXO TF activity is primarily regulated by nuclear/cytoplasmic shuttling by the regulation of AKT kinase activity (Tzivion et al., 2011). Under progrowth conditions, AKT phosphorylates FOXO proteins at three serine/threonine residues (Brunet et al., 1999). Phosphorylation at these sites sequesters FOXOs in the cytoplasm by inhibiting the FOXO nuclear localization signal, activating the nuclear export signal, and promoting the binding of FOXO to 14-3-3 proteins (Brunet et al., 2002). Stresses activate FOXO by inhibiting AKT activity or through additional PTMs that prevent 14-3-3 binding and drive FOXO to the nucleus. Recent work has revealed that the dynamics of FOXO nuclear/cytoplasmic shuttling exhibits unique patterns that depend on treatment conditions. For example, when cells are serum-starved and then restimulated with different growth factors, FOXO1/FOXO3 shuttling to the cytoplasm shows growth factor–dependent dynamics. IGF1 induces sustained cytoplasmic FOXO, while other growth factors, including epiregulin and betacellulin, cause recurrent pulses of FOXO from the cytoplasm to the nucleus (Gross and Rotwein, 2015; Sampattavanich et al., 2018). Pulses of FOXO nuclear/cytoplasmic shuttling have also been observed during energy stress induced by iodoacetate, an alkylating agent that inhibits GAPDH (Hung et al., 2017). Similar patterns of stress-specific dynamics have been observed in other TFs, including p53, NF-KB, and MSN2, and have been linked to input/output specificity (Hoffmann et al., 2002; Batchelor et al., 2011; Hao and O’Shea, 2012; Purvis et al., 2012; Paek et al., 2016; Cheng et al., 2021).

To determine the role of FOXO shuttling dynamics in the response to cellular stress, we used a combination of immunofluorescence and live-cell reporters to measure the temporal dynamics of FOXO1 and FOXO3 in response to H₂O₂-induced oxidative stress and serum starvation. These stresses were chosen due to the different FOXO–dependent cell outcomes they induce; H₂O₂ leads to FOXO–dependent cell death, while serum starvation largely results in FOXO–dependent arrest and protects cells from cellular senescence (Lehtinen et al., 2006; Imai et al., 2014). We focused on FOXO1 and FOXO3 due to the availability of good antibodies for staining and measuring their nuclear levels. We found that the dynamics of FOXO1/FOXO3 shuttling and cellular outcomes was stress-specific. In response to H₂O₂, FOXO enters the nucleus in a bimodal all-or-none manner: in some cells nearly all FOXO protein is in the nucleus, while other cells exhibit no accumulation of nuclear FOXO. For cells that show high levels of FOXO in the nucleus, it remains nuclear for either a short period of 4–5 h or an extended period of up to 48 h (though often shorter due to cell death). The fraction of cells with nuclear FOXO and the time it remained in the nucleus increased with the dose of H₂O₂ and was associated with more cell death. FOXO nuclear accumulation in response to H₂O₂ could be reversed by pretreating with the antioxidant N-acetylcysteine (NAC). However, adding NAC 2 h after H₂O₂ treatment had no effect on the FOXO response. Serum starvation, on the other hand, led to pulses of nuclear FOXO, with nuclear levels much lower than those measured in response to oxidative stress and fewer cell death events. Pulses of nuclear FOXO ceased rapidly after restimulation with serum. Differences in FOXO nuclear shuttling dynamics are reflected in the strength of AKT inhibition. H₂O₂ treatment results in a bimodal population of cells: one population with increased pAKT and low nuclear FOXO, and a separate population with low pAKT and high nuclear FOXO. In contrast, serum starvation causes graded levels of pAKT that are inversely correlated with nuclear FOXO levels.

RESULTS

The distribution of nuclear FOXO1 and FOXO3 is stress–specific

To determine if different stresses elicit different temporal dynamics of FOXO1 nuclear/cytoplasmic shuttling, we first performed an immunofluorescence time course of MCF7 cells exposed to increasing doses of H₂O₂ (Figure 1, A and B) or serum starvation (Figure 1, D and E) and measured the fraction of FOXO1 in the nucleus. The formula for calculating the nuclear fraction can be found in Supplemental Figure S1A. We observed a distinct pattern for each stress. H₂O₂ treatment elicited a bimodal staining pattern, in which some cells had mostly cytoplasmic FOXO1, similarly to untreated controls, while others had mostly nuclear FOXO1, similarly to cells treated with an AKT inhibitor (Figure 1B, bottom row). The proportion of cells with predominantly nuclear FOXO1 and their persistence over time increased with the H₂O₂ dose (Figure 1B). In contrast, serum withdrawal led to cells with a mix of nuclear and cytoplasmic FOXO1 with a wide unimodal distribution. The amount of nuclear FOXO1 increased slowly over time and peaked 24 h after serum withdrawal (Figure 1, D and E). To determine if these patterns are specific to FOXO1 or are shared by other isoforms, we costained cells with FOXO1 and FOXO3 antibodies. This revealed that the fractions of nuclear FOXO1 and FOXO3 were similar and correlated in individual cells (Figure 1, C and F; Supplemental Figure S1, C and F). We observed bimodal activation of FOXO1 in response to oxidative stress in the primary hTERT-immortalized human mammary epithelial cell line HME1 and in a two-dimensional human gut colonoid culture described recently, suggesting that this pattern of activation is not specific to MCF7 breast cancer cells (Supplemental Figure S1, B and E; Pond et al., 2022). In addition, serum withdrawal led to a unimodal accumulation of nuclear FOXO1 in HME1 cells, similarly to the MCF7 data (Supplemental Figure S1D).

The duration of nuclear FOXO1 after H ₂O ₂ treatment is dose-dependent and associated with cell death

The distinct nuclear accumulation pattern for each stress suggests that dynamics might play a role in dictating different FOXO-induced cell fates. To measure FOXO1 shuttling dynamics and cell-cycle arrest/death in single cells, we used a previously developed MCF7 cell line where CRISPR was used to add the Venus fluorescent protein to the C-terminus of FOXO1 and added an H2B-CFP tag for tracking nuclei (Stewart-Ornstein and Lahav, 2016). We verified that the distribution of nuclear accumulation of FOXO1 in response to H₂O₂, serum starvation, and AKT inhibition broadly matched the distributions observed by immunofluorescence of untagged FOXO1 (Figure 1, B and E; Supplemental Figure S2, A and B). However, the amplitude of the response was not as strong, possibly reflecting the decreased sensitivity of live-cell imaging as compared with immunofluorescence. In addition, there was an increase in the number of FOXO1 positive cells at 50 µM H₂O₂ (Figure 1, B and E; Supplemental Figure S2, A and B). This is likely due to the known sensitivity of the H₂O₂ response to cell number (Gülden et al., 2010; Lim et al., 2014).

We first looked at the dynamics of FOXO1 shuttling in response to H₂O₂. Cells were treated with four different doses of H₂O₂ and imaged every 20 min for 48 h (Supplemental Movie 1). Approximately 300 single cells were tracked for each dose, and we extracted the nuclear fraction of FOXO1-mVenus signal for each cell using previously developed software (Reyes et al., 2018). Cell death events were detected by morphology. The average FOXO1–mVenus nuclear fraction increases with H₂O₂ dose, consistent with our immunofluorescence data (Supplemental Figure S2C).

We observed three broad patterns of nuclear FOXO1 accumulation in single cells. In the first pattern, which we call non responders, cells show no or only a brief pulse of nuclear FOXO1 (<1 h, Figure 2, A and D, Left panel), with the majority of these cells surviving H₂O₂ treatment. The second pattern, which we call short-responders, exhibits a 1–5-h pulse of high nuclear FOXO1 levels followed by a rapid and sustained switch to the cytoplasm; ∼68% of these cells died (Figure 2, A and D, Middle). The final pattern, which we call long-responders, showed a sustained nuclear FOXO1 signal for long periods of time (>5 h, mean ∼18 h) that ended in cell death in 92% of cells (Figure 2, A and D, Right panel; see Supplemental Figure S2E for correlation of FOXO1 duration and cell death). It is important to note that cells were sorted into each category based on the length of the initial FOXO1 pulse, and some cells that died reactivated FOXO1 at later time points, perhaps through H₂O₂-induced stress that takes longer to accumulate or H₂O₂-independent stresses. The fraction of cells showing a long nuclear FOXO1 response increased with dose (Figure 2B). Additionally, the median trace of cells that survive H₂O₂ treatment exhibits both a lower amplitude and a shorter duration of nuclear FOXO1, and this pattern was consistent across H₂O₂ doses (Figure 2C).

The number of divisions a cell experienced after H₂O₂ treatment decreased with increasing dose, and even at lower doses, cells in the long responder category were less likely to divide again (Supplemental Figure S2D). Together, these data suggest that nuclear FOXO1 accumulation in response to H₂O₂ is all-or-none in single cells, and the time it remains in the nucleus increases with H₂O₂ dose. In addition, cells exhibit substantial heterogeneity in the time that FOXO1 remains in the nucleus, and cell death is correlated with the length of time that FOXO1 remains in the nucleus.

Accumulation of nuclear FOXO1 shows time-dependent irreversibility

The observation that H₂O₂ triggers an “all-or-none” nuclear accumulation of FOXO1 in single cells suggests that FOXO1 is controlled by a bistable system. One feature of bistable systems is that they exhibit hysteresis, meaning that the response of cells to a stimulus depends on whether the cell is in the on or off state. Cells in the off state require relatively high levels of stimulus to switch to the on state. In contrast, cells in the on state require relatively low levels of stimulus to maintain activation. At the extreme, bistable systems can show irreversibility, where the complete removal of the stimulus is insufficient for turning the system off (Ferrell, 2002). We hypothesized that nuclear accumulation of FOXO1 was irreversible, as previous measurements of H₂O₂ levels with single-cell sensors showed rapid clearance of H₂O₂ in less than 1 h, whereas nuclear accumulation of FOXO1 lasts several hours and in some cases days (Figure 2B; Ermakova et al., 2014). Indeed, using the HyPer-3 H₂O₂ reporter, we found that the reporter returned to baseline levels in less than 1 h for all doses tested (Figure 3A; Bilan et al., 2013).

To test whether the FOXO1 response exhibited hysteresis in response to H₂O₂, we used N-acetylcysteine (NAC), an antioxidant that rapidly reduces H₂O₂ and other reactive oxygen species (Pedre et al., 2021). We first verified that pretreating cells with NAC prevented nuclear FOXO1 accumulation in response to H₂O₂ (Supplemental Figure S3A). We then performed a time-lapse experiment where cells were first treated with H₂O₂ and with NAC at different time points. While treatment with NAC 1 h after H₂O₂ causes FOXO1 to exit the nucleus and accumulate in the cytoplasm, adding NAC 2 h after treatment is no longer capable of reversing the FOXO1 response (Figure 3B). This indicates that the nuclear accumulation of FOXO1 shows time-dependent irreversibility and that H₂O₂ is no longer needed to maintain nuclear FOXO1 2 h after treatment.

FOXO1 shows a reversible pulsatile activation pattern in response to serum starvation

We next measured the dynamics of FOXO1–mVenus shuttling in response to serum starvation. MCF7 FOXO1–mVenus H2B–ECFP cells were imaged every 20 min for 125 h. For the first 5 h, cells were grown in 5% serum to get a baseline, after which the level of serum was dropped from 5% to 0.02%, and cells were imaged for 72 h and then restimulated with 2% serum to determine if the FOXO1 response was reversible (Supplemental Movie 2).

Example single-cell traces can be seen in Figure 4A. In response to serum withdrawal, cells exhibit a synchronized short spike in nuclear FOXO1 that lasts ∼20–30 min, after which FOXO1 returns to the cytoplasm. FOXO1 then reenters the nucleus after a delay that varies from cell to cell (14.9 ± 13.7 h); in some cells FOXO1 returns to the nucleus within a few hours, while other cells require more than 24 h for FOXO1 to return to the nucleus (Figure 4B). The median trace of all cells predominantly reflects the fraction of cells in which FOXO1 has returned to the nucleus (Figure 4C). Once FOXO1 returns to the nucleus, it begins a series of pulses in and out of the nucleus for the remainder of the time that cells are under serum withdrawal conditions (Figure 4, A and B). In response to serum withdrawal, cells slow cell divisions and ∼32% of cells die. The concentration of FOXO1 in the nucleus was higher in apoptotic cells than in surviving cells, yet the fraction of total FOXO1 levels in the nucleus was similar between the two populations (Supplemental Figure S4, A and B). Interestingly, FOXO1 exits the nucleus rapidly upon restimulation with 2% FBS, which is associated with a rapid decline in cell death and a slower increase in mitotic events (Figure 4, C–E). Together, these data show that the FOXO1 response to serum withdrawal is different from that to H₂O₂ treatment. While H₂O₂ leads to an all-or-none accumulation of nuclear Foxo1 with time- dependent irreversibility, serum withdrawal causes stochastic pulses of nuclear FOXO1 that are rapidly reversible.

We next tested how H₂O₂ and serum starvation impact gene expression. We measured expression of two FOXO target genes involved in cell-cycle arrest (p21, p27) and two apoptotic target genes (BIM and PUMA). While expression of all four genes increased with both increasing duration of serum starvation (Figure 4F) and increasing concentration of H₂O₂ (Figure 4G), both PUMA and p21 levels were substantially higher in H₂O₂-treated cells than in serum-starved cells. These data allow a direct comparison of transcriptional activation of the FOXO targets after 5 h of treatment for both serum starvation and H₂O₂. For example, after 5 h of H₂O₂ treatment, PUMA levels increased 8–32 fold (50–100 µM), whereas 5 h of serum starvation led to a fourfold increase in PUMA levels. This is in line with the increased level and speed of cell death occurring after treatment with 50–100 μM H₂O₂ compared with after serum starvation.

The dynamics of FOXO1 shuttling in response to H ₂O ₂ and serum starvation are reflected by the strength of AKT inhibition

What accounts for the different dynamic patterns in cells treated with H₂O₂ and in serum-starved cells? Since AKT phosphorylation of FOXO is known to keep FOXOs in the cytoplasm, we sought to determine whether a decrease in AKT activity was responsible for the bimodal activation of FOXO1 in response to H₂O₂. We first measured phosphorylation of AKT at serine 473 (pAKT-S473), a marker for AKT activity, by Western blot. H₂O₂ treatment led to a slight increase at lower concentrations but little change over untreated cells at higher concentrations (Figure 5A). The increase in pAKT-S473 is likely due to the activation of receptor tyrosine kinases by H₂O₂ (Sundaresan et al., 1995; Heppner and van der Vliet, 2016; Dustin et al., 2020)_.

The Western blot data suggested that AKT remained active in cells treated with H₂O₂. Yet given the cell-to-cell heterogeneity in FOXO1 accumulation, coupled with the increase in pAKT-S473 at low levels of H₂O₂, it is possible that the Western blot data are misleading due to averaging cell populations. Indeed previous studies have shown that high levels of H₂O₂ can inhibit AKT (Martín et al., 2002; Murata et al., 2003). To test if AKT was active in cells with nuclear FOXO1, we measured both pAKT-S473 and FOXO1 nuclear accumulation by immunofluorescence in single cells following H₂O₂ treatment. Strikingly, these data revealed that cells with high nuclear FOXO1 levels showed low levels of pAKT-S473, while cells with cytoplasmic FOXO1 showed pAKT-S473 levels higher than those in untreated controls (Figure 5, B and C). Averaging the pAKT-S473 levels of the immunofluorescence data revealed a modest decrease at high levels of H₂O₂ due to the existence of the two different cell populations: cells with low pAKT-S473 and nuclear FOXO1 and cells with increased pAKT473 and cytoplasmic FOXO1 (Figure 5C). Note that some cells had high levels of pAKT-S473 and nuclear FOXO1 (Figure 5C, 75 µM). These cells are likely in a transition state, where AKT is being reactivated and FOXO1 is in the process of movement to the cytoplasm. In line with this, there was a downward slope in this cell population, consistent with these cells being in the process of transitioning from low AKT activity and nuclear FOXO1 to high AKT activity and cytoplasmic FOXO1. Moreover, the fraction of these cells decreased with H₂O₂ dose when fewer cells are expected to be making the transition between states. The density plots for FOXO1 and pAKT-473 can be seen individually in Supplemental Figure S5A.

We also measured pAKT-473 and FOXO1 nuclear accumulation in response to serum starvation. In contrast to H₂O₂ treatment, serum starvation caused a graded decrease in pAKT-S473, with nuclear FOXO1 levels showing an inverse correlation with pAKT-S473 levels (Figure 5D). Consistent with previous data, nuclear FOXO1 increased, while pAKT-S473 decreased over time in response to serum starvation (Figure 5D; see Supplemental Figure S5B for individual distributions).

To further test whether AKT inhibition was required for H₂O_2-induced nuclear accumulation of FOXO1, we introduced a myristoylated AKT construct (Myr-AKT), which drives AKT to the membrane, leading to hyperactivation of AKT (Martz et al., 2014). Western blots confirmed hyperactivation of AKT as measured by pAKT-S473 (Figure 5A). High levels of H₂O₂ slightly decreased pAKT-S473, yet the levels were still substantially higher than in control cells lacking the Myr-AKT construct (Figure 5A). Immunofluorescence experiments revealed that H₂O₂ was still capable of inducing nuclear FOXO1 accumulation, yet cells with nuclear accumulation of FOXO1 had reduced pAKT-S473 despite the presence of the Myr-AKT construct (Figure 5E; see Supplemental Figure S5C for individual distributions). This suggests that H₂O₂ is capable of inhibiting both wild-type AKT and the Myr-AKT construct. In contrast, serum starvation did not alter nuclear accumulation of FOXO1 or pAKT-S473 levels in cells harboring the Myr-AKT (Figure 5F; see Supplemental Figure S5D for individual distributions). Together these data suggest that the mechanisms of AKT inactivation are different under H₂O₂ and serum starvation conditions. H₂O₂ causes a bimodal response, with high pAKT-S473 and low nuclear FOXO1 levels in one population, while other cells have a robust decrease in pAKT-S473 and high nuclear FOXO1 levels. Interestingly, H₂O₂ can still inactivate Myr-AKT. In contrast, serum starvation causes graded inactivation of pAKT-S473 and nuclear FOXO1 accumulation that can be reversed completely by Myr-AKT.

DISCUSSION

TFs are often activated by multiple signals, and upon activation can up-regulate a number of downstream programs with different phenotypic outcomes. Which programs are activated typically depends on both the type and magnitude of the signal. This multiple-input, multiple-output structure is particularly common amongst TFs that respond to cellular stress, including p53, NF-κB, MSN2, and FOXO TFs.

One way in which these TFs achieve input/output specificity is through linking inputs to specific dynamic patterns of TF activation (Purvis et al., 2013). For example, activation of p53 by DNA double-strand breaks leads to regular pulses of p53 accumulation with relatively stable amplitude and frequency, which typically leads to transient cell-cycle arrest or senescence (Purvis et al., 2012). If p53 levels are sustained by pharmacological inhibition of Mdm2, this forces cells into senescence. In contrast, other DNA-damaging agents such as cisplatin cause a single pulse of p53 accumulation, and the rate of p53 accumulation is linked to cell death (Paek et al., 2016). NF-κB and Msn2 also exhibit pulsatile or sustained activation that is stimulus-dependent and linked to the expression of downstream target genes (Hao and O’Shea, 2012; Hansen and O’Shea, 2016; Mitchell et al., 2016; Cheng et al., 2021).

By measuring the shuttling of FOXO1 and FOXO3 in single cells, we found that nuclear accumulation of these two TFs is also stimulus-dependent. H₂O₂ leads to an all-or-none accumulation of FOXO1/3 in single cells, while serum starvation results in weaker nuclear FOXO1/3 that pulses into and out of the nucleus. The low-amplitude pulses of nuclear FOXO induced by serum starvation predominantly resulted in cell-cycle arrest, while the high levels of nuclear FOXO after H₂O₂ were associated with cell-cycle arrest and high levels of cell death.

There are several mechanisms that could contribute to the difference in cell fates associated with H₂O₂ and serum starvation. The low levels of nuclear FOXO during serum starvation could be insufficient to induce cell death through preferential binding to other FOXO targets, or alternatively, FOXO could bind to apoptotic promoters and enhancers with similar strength, but the low levels of FOXO lead to insufficient up-regulation of apoptotic genes. In addition, up-regulation of apoptotic genes could require high nuclear levels of FOXO for long durations to disrupt nucleosomal histone–DNA interactions in the promoters/enhancers of apoptotic genes, as shown recently for Msn2 and NF-κB (Hao and O’Shea, 2012; Cheng et al., 2021). This is consistent with the observation that the amount of cell death in response to H₂O₂ increased with the duration of nuclear FOXO1. In addition, FOXO1 has been shown to perturb nucleosome binding to a FOXO1 response element in vitro, suggesting that it can act as a pioneer factor (Hatta and Cirillo, 2007). However, genome-wide studies in human cells found that FOXO3 mostly increases the transcription of target genes already in an open chromatin state, arguing against FOXO TFs acting as pioneer factors (Eijkelenboom et al., 2013a,b). It is also likely that FOXO could be modified in a stress-specific manner, either by posttranslational modifications or by cofactor binding to selectively up-regulate specific target genes. In addition, some of the cell death events caused by H₂O₂ are likely independent of FOXO, as we observed cell death events that occurred well after FOXO1 was switched off at lower concentrations of H₂O₂. Although previous studies found a strong reduction of H₂O₂-nduced cell death events in FOXO knockouts, some death events remained (Lehtinen et al., 2006; Yamagata et al., 2008). p53 is also activated in response to H₂O₂, and induces apoptotic genes (Shi et al., 2021; Hanson and Batchelor, 2022). However, p53 knockdown experiments in these studies showed mixed results in the effect of H₂O_2-induced cell death.

The two different dynamic patterns of FOXO1/3 shuttling are likely due to their different modes of activation. Nuclear accumulation of FOXO1 in response to both H₂O₂ and serum starvation was associated with low AKT activity, although the mechanism of AKT inhibition was likely different. The pulsatile accumulation of FOXO1 in response to serum starvation is also consistent with the many known mechanisms of negative feedback in the PI3K/AKT pathway (Manning and Toker, 2017; Mukherjee et al., 2021). The all-or-none irreversible accumulation of FOXO1 and FOXO3 suggests positive feedback in the mode of activation, though it is not clear what the mechanism of positive feedback is. There is evidence that in the presence of high H₂O₂ levels, Cys-297 and Cys-311 of AKT form a disulfide bond that shuts off AKT kinase activity (Murata et al., 2003). This direct inactivation of AKT could explain why myristolated AKT, which is normally constitutively active, is inhibited by high concentrations of H₂O₂. There are also several separate mechanisms that have been linked to nuclear FOXO accumulation by ROS, including activation of the stress-activated MAP kinases p38 and JNK, the mammalian Ste20-like kinases MST1/2, activation of the arginine methyltransferae PRMT1, and formation of a disulfide bond with the nuclear import proteins Importin-7 and Importin-8 (Essers et al., 2004; Lehtinen et al., 2006; Yamagata et al., 2008; Putker et al., 2015; Hopkins et al., 2018).

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

Cell culture

MCF7 breast cancer cells and HME-1 hTert primary breast epithelial cells were grown in RPMI containing 10% fetal bovine serum (FBS), 100 units/ml penicillin, 100 μg/ml streptomycin, and 25 ng/ml amphotericin B. When needed for selection, puromycin, neomycin, and hygromycin were used at final concentrations of 500 ng/ml, 400 μg/ml, and 50 μg/ml, respectively.

Cell treatments

H₂O₂ experiments were performed by serially diluting stock H₂O₂ in PBS, until at 10× the concentration desired for the experiment. The 10× H₂O₂ solution was added directly to media for a final concentration of 1×. The stock H₂O₂ (Fisher Scientific AAL13235AP) was replaced monthly. Serum starvation experiments were performed by removing media, rinsing with PBS, and then adding 0.2% FBS media. MK-2206 (ApexBio A3010) was used from a stock concentration of 25 mM diluted in PBS before being added to media.

Cell line construction

myr-FLAG-AKT1-pcw107 was a gift from David Sabatini and Kris Wood (Addgene plasmid 64606; http://n2t.net/addgene:64606; RRID:Addgene_64606) and was used to make the MCF7 myr-AKT line. The H2B-ECFP lentiviral plasmid was described previously (Chakrabarti et al., 2018). Lentiviral production and infection was performed as described in a previous publication (Tiscornia et al., 2006). The pC1-HyPer-3 plasmid was a gift from Vsevolod Belousov (Addgene plasmid 42131; http://n2t.net/addgene:42131; RRID:Addgene_42131).

Immunofluorescence

Cells were plated in glass-bottomed multiwell plates (CellVis), treated, and then fixed with 2% paraformaldehyde, permeabilized with 0.1% Triton X-100 in PBS, blocked with 2% BSA in PBS, and incubated with primary antibodies overnight at 4°C in 2% BSA and 0.1% Tween 20 in PBS. Secondary antibodies are incubated for 1 h at room temperature in 2% BSA and 0.1% Tween 20 in PBS, stained with DAPI, and imaged in PBS. FOXO1 primary antibody: Cell Signaling 2880S; FOXO3 primary antibody: Cell Signaling 2497. To obtain single-cell correlation data between FOXO1 and FOXO3, after the FOXO3 primary and secondary incubations, samples were incubated with Rabbit IgG (Millipore-Sigma), and then a conjugated version of the FOXO1 primary antibody was used. Nuclei were segmented and mean nuclear FOXO1 and FOXO3 intensities were measured using CellProfiler (McQuin et al., 2018). To obtain cytoplasmic FOXO1/FOXO3, a ring three pixels wide was drawn around the nuclear mask and mean cytoplasmic FOXO1/FOXO3 was extracted using this mask. Plots were created using MATLAB.

Live-cell microscopy

Cells were plated in a glass-bottomed 24-well plate (CellVis) at ∼5000 cells per well, and grown in Roswell Park Memorial Institute, 10% FBS, 100 units/ml penicillin, 100 μg/ml streptomycin, and 25 ng/ml amphotericin B for 48 h and then rinsed with PBS and given DMEM Fluorobrite (ThermoFisher) medium with 5% FBS, 100 units/ml penicillin, 100 μg/ml streptomycin, 25 ng/ml amphotericin B, and 1× Glutamax (ThermoFisher). Cells were imaged every 20 min for 48–125 h by a Nikon Eclipse Ti microscope. The Nikon Perfect focus system was used to maintain focus. An OKO labs incubation system was used to maintain humidity, 37°C, and 5% CO₂. H2B-CFP was imaged using the C-FL AT ECFP/Cerulean Filter Set (Chroma) for 20–40 ms. FOXO1–mVenus was imaged using (Chroma) ET-EYFP Filter Set for 600 ms. HyPer3 probe was imaged using C-FL AT EGFP/FITC/CY2/Alexa Fluor 488 Filter Set (Chroma) for 300 ms. Fluorescence imaging used the Lumencor SOLA light engine. Images were taken using the Hamamatsu ORCA-Flash 4.0 camera with 16-bit depth and 2 × 2 binning. Movies were analyzed using single cell tracking software in MATLAB (Reyes et al., 2018).

Western blot

Cells were rinsed and covered with PBS, scraped off the plate, and spun down, the liquid was aspirated, and the cells were frozen in liquid nitrogen and stored at –80°C. Cells were then thawed on ice and lysed (25 mM Tris pH 7.6, 150 mM NaCl, 1% NP-40, 1% Na-deoxycholate, and 0/1% SDS in water + protease inhibitor cocktail [Sigma] + phosphatase inhibitor cocktail [Sigma-Aldrich] + okadaic acid + sodium fluoride). Sample protein concentrations were measured with a Bradford assay (Bio-Rad) and equal protein concentrations were run on NuPAGE 4–2% Bis–Tris gels (Invitrogen). Protein was transferred to a nitrocellulose membrane and incubated in blocking solution (PBS, 5% BSA, 0.1% Tween 20) for 30 min at room temperature. The membrane was incubated with primary antibodies in blocking solution at 4°C overnight, rinsed three times in PBS with 0.1% Tween 20, and then incubated in secondary antibody for 1 h at room temperature. Antibodies: total AKT (Cell Signaling), phosphorylated AKT at S473 (Cell Signaling), 680LT secondary (LICOR-IRDye), 800CW secondary (LICOR-IRDye), actin (Sigma). The blot was imaged on the LI-COR Odyssey.

Quantitative PCR

After treatment, cells were rinsed with PBS, trypsinized, spun down, and flash frozen in liquid nitrogen. RNA was extracted using the RNeasy mini-kit (Qiagen). RNA concentration was measured via absorbance at 260 nm. RNA was converted to a cDNA library using the High-Capacity cDNA reverse transcription kit (Applied Biosciences). Quantitative PCR was performed in triplicate on the qTOWER³ (analytikjena) with SYBR green master mix (Applied Biosciences), 16.8 ng cDNA, and 100 nM primers. The primer sequences used were p27 forward: TCTGAGGACACGCATTTGG, p27 reverse: TGTTCTGTTGGCTCTTTTGTTT, p21 forward: TGTCACTGTCTTGTACCCTTG, p21 reverse: GGCGTTTGGAGTGGTAGAA, BIM forward: TGGAGACGAGTTTAACGCTTAC, BIM reverse: CCGCAAAGAACCTGTCAATG, PUMA forward: CGACCTCAACGCACAGTACG, PUMA reverse: GGGTGCAGGCACCTAATTGG, actin forward: TGCAGAAAGAGATCACCGC, and actin reverse: CCGATCCACACCGAGTATTTG.

FOOTNOTES

ACKNOWLEDGMENTS

REFERENCES