Single-cell dynamics of pannexin-1-facilitated programmed ATP loss during apoptosis

  1. Hiromi Imamura  Is a corresponding author
  2. Shuichiro Sakamoto
  3. Tomoki Yoshida
  4. Yusuke Matsui
  5. Silvia Penuela
  6. Dale W Laird
  7. Shin Mizukami
  8. Kazuya Kikuchi
  9. Akira Kakizuka
  1. Graduate School of Biostudies, Kyoto University, Japan
  2. Graduate School of Engineering, Osaka University, Japan
  3. Department of Anatomy and Cell Biology, Schulich School of Medicine and Dentistry, University of Western Ontario, Canada
  4. Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Japan

Abstract

ATP is essential for all living cells. However, how dead cells lose ATP has not been well investigated. In this study, we developed new FRET biosensors for dual imaging of intracellular ATP level and caspase-3 activity in single apoptotic cultured human cells. We show that the cytosolic ATP level starts to decrease immediately after the activation of caspase-3, and this process is completed typically within 2 hr. The ATP decrease was facilitated by caspase-dependent cleavage of the plasma membrane channel pannexin-1, indicating that the intracellular decrease of the apoptotic cell is a ‘programmed’ process. Apoptotic cells deficient of pannexin-1 sustained the ability to produce ATP through glycolysis and to consume ATP, and did not stop wasting glucose much longer period than normal apoptotic cells. Thus, the pannexin-1 plays a role in arresting the metabolic activity of dead apoptotic cells, most likely through facilitating the loss of intracellular ATP.

Introduction

Living cells require energy that is provided by the principal intracellular energy carrier, adenosine-triphosphate (ATP). Free energy from ATP that is released upon hydrolysis is utilized in various vital processes including the generation and maintenance of the plasma membrane potential (Kaplan, 2002), remodeling of chromatin (Vignali et al., 2000), locomotion of molecular motors (Vale, 2003), protein degradation (Ciechanover, 1994), and metabolic reactions (Voet and Voet, 2010). Therefore, living cells must maintain high concentrations of intracellular ATP by continuously regenerating ATP from its hydrolysis products, adenosine-diphosphate (ADP) and phosphate ion, via energy metabolism that uses chemical energy stored in cellular nutrients, such as glucose. On the other hand, dead cells contain very little, or even no ATP.

Apoptosis is a form of programmed cell death, with important roles in development, tissue homeostasis, and immunity. It is characterized by distinctive morphological changes, such as membrane blebbing, nuclear condensation, and externalization of phosphatidylserine (Elmore, 2007). Although there are a variety of stimuli that can provoke apoptosis, these stimuli converge into the activation of a proteolytic cascade of cysteine proteases, called caspases. Because cleavage of specific target proteins by activated effector caspases triggers apoptotic events, including the characteristic morphological changes, apoptotic cell death is a systematically and genetically determined, or ‘programmed’, process. In contrast, necrosis is considered to be the ‘unprogrammed’, cataclysmic demise of the cell.

It has been reported that cells die from necrosis rather than apoptosis when intracellular ATP is depleted prior to otherwise apoptotic stimuli (Eguchi et al., 1997; Leist et al., 1997). It has been also reported that dATP/ATP is required for the formation of cytochrome c/Apaf-1/procaspase-9 complexes (Hu et al., 1999; Li et al., 1997). Moreover, it is suggested that ATP is required for chromatin condensation of apoptotic cells (Kass et al., 1996). Apoptosis is, thus, considered to be an energy-demanding process, requiring intracellular ATP for the execution of the cell death program. In spite of, or perhaps in part due to the requirement of ATP for apoptosis, the intracellular ATP in apoptotic cells is ultimately depleted. Therefore, it seems likely that both maintenance and reduction of the intracellular ATP level are systematically regulated during the progression of apoptosis. Intracellular ATP levels have been conventionally analyzed with the firefly luciferin-luciferase system (Lundin and Thore, 1975), liquid chromatography (Sellevold et al., 1986), or related methods, which use lysates of a large number of cells. Because apoptosis progresses differently between cells (for example see Goldstein et al., 2000), even those cultured and stimulated in identical conditions, it has been quite difficult to precisely understand the dynamics of the intracellular ATP level in each dying apoptotic cell, and difficult to tell whether the ATP decrease accompanies specific apoptotic events. Furthermore, the molecular mechanism of how intracellular ATP decreases in apoptotic cells also remains to be elucidated. In addition, it is totally unclear whether the depleted intracellular ATP in an apoptotic cell benefits the dying cell itself or the surrounding, healthy cells.

In this work, we established a method for imaging both ATP concentration and caspase-3 activity in a single apoptotic cell with newly developed genetically encoded Förster resonance energy transfer (FRET)-based biosensors for ATP and caspase-3 activity. We found that the intracellular ATP level starts to decrease following the activation of caspase-3 and that the caspase-triggered opening of the plasma membrane channel pannexin-1 (PANX1) is the major cause of the decrease in intracellular ATP.

Results

Development of FRET biosensors for dual imaging of ATP and caspase-3 activity of apoptotic cells

In general, the progression of apoptosis varies between individual cells, even in the same cell type. It is, thus, essentially difficult to understand how the intracellular ATP level in a particular cell changes during apoptosis using conventional biochemical analyses of pooled cells, such as firefly luciferase assays. To reveal the dynamics of intracellular ATP levels during apoptosis at the single-cell level, we first used a genetically encoded FRET-based ATP biosensor, ATeam (Imamura et al., 2009), which is comprised of a cyan fluorescent protein (CFP; mseCFP), an FoF1-ATP synthase ε subunit and yellow fluorescent protein (YFP; cp173-mVenus). Unfortunately, we found that the original ATeam (AT1.03) was cleaved into its constituent pair of separate fluorescent proteins in apoptotic cells, most probably by activated caspases (Figure 1—figure supplement 1). Thus, the FRET signals of the original biosensor were reduced in apoptotic cells irrespective of the ATP concentration. We replaced Asp-242 and Asp-339 of AT1.03, which we predicted were within the target sequences of the caspases, with Asn and Gly, respectively, and found that the altered ATeam was not cleaved inside apoptotic cells (Figure 1—figure supplement 1). We subsequently used this caspase-resistant ATeam (AT1.03CR) to study the ATP dynamics in apoptotic cells.

The dynamics of cytosolic ATP levels throughout the apoptotic process were investigated by imaging single human cervical adenocarcinoma (HeLa) cells expressing AT1.03CR. Overall, the intracellular ATP levels remained almost constant for several hours after stimulation. The cytosolic ATP levels in these cells started to decrease after a variable time interval (typically from 3 to 8 hr after apoptotic stimulation, see Figure 1—figure supplement 2). Once the intracellular ATP levels started to decrease, they were depleted within typically 0.5–2 hr.

The presence of a pan-caspase inhibitor zVAD-fmk almost completely blocked the cytosolic ATP decrease of anti-FAS-induced apoptotic cells (Figure 1A). Thus, the ATP decrease induced by apoptotic stimuli is most likely a caspase-dependent process. Next, we developed a FRET-based caspase-3 biosensor O-DEVD-FR by connecting an orange fluorescent protein mKOκ (Tsutsui et al., 2008) and a far-red fluorescent protein mKate2 (Shcherbo et al., 2009) by a Gly-Gly-Asp-Glu-Val-Asp-Gly-Thr linker containing a bona fide caspase-3 recognition sequence (Figure 1—figure supplement 3). Once caspase-3 is activated, it cleaves the linker in O-DEVD-FR, resulting in the separation of mKOκ and mKate2, and the consequent reduction in FRET signal in apoptotic cells (Figure 1B). It was recently demonstrated that mKOκ-mKate2 FRET pair is compatible with CFP-YFP FRET pair because they use different spectral windows (Watabe et al., 2020). Thus, it is possible to use both biosensors to fluorescently image ATP level and caspase-3 activity in the same apoptotic cell (Figure 1C,D). Activation of caspase-3 was clearly observed as a decrease in FRET signal (an increase in mKO/mKate ratio). We defined onset of caspase-3 activation as the frame immediately preceding the first frame in which the increase in mKO/mKate ratio was first observed (see an arrow in the inset of Figure 1D). It was observed that intracellular ATP started to decrease after the onset of caspase-3 activation, also supporting that the ATP decrease of the apoptotic cell is a caspase-dependent process. It should be noted that any increase or decrease in fluorescence intensity due to cell morphological change was offset because we monitored the ratios of fluorescence intensities of an acceptor and a donor of the FRET biosensors.

Figure 1 with 3 supplements see all
The initiation of cytosolic ATP decrease occurs almost simultaneously with that of caspase-3 activation.

(A) Single-cell ATP dynamics after anti-FAS treatment in the absence and the presence of zVAD-fmk. Each line represents the time course of the YFP/CFP ratio of AT1.03CR from a single cell (70 [DMSO] and 49 [zVAD] cells from three biological replicates). Cells were treated with anti-FAS and cycloheximide at time = 1.0 hr. DMSO (0.1%) or pan-caspase inhibitor zVAD-fmk (20 µM) was added just before the start of imaging experiment. (B) Schematic drawings of AT1.03CR and O-DEVD-FR. (C) Time-lapse images of the ATP level and caspase-3 activity of a single apoptotic HeLa cell expressing AT1.03CR and O-DEVD-FR. Pseudocolored ratio images of AT1.03CR are shown in the upper panel, and those of O-DEVD-FR are shown in the lower panel. Bar, 10 µm. (D) Time courses of ATP level and caspase-3 activity of a single apoptotic HeLa cell. YFP/CFP ratio of AT1.03CR and mKO/mKate ratio of O-DEVD-FR were shown in the upper and the lower panels, respectively. Arrow in the inset indicates the onset of caspase-3 activation. Fluorescence images were captured every 3 min.

Single-cell dynamics of cytosolic ATP of apoptotic PANX1-knockout cells

PANX1 belongs to the innexin/pannexin superfamily and forms a heptameric pore in the plasma membrane (Deng et al., 2020; Michalski et al., 2020; Qu et al., 2020), functioning as a large-pore channel capable of passing small molecules (Bao et al., 2004; Dahl and Muller, 2014; Panchin et al., 2000; Penuela et al., 2013). It has been reported that apoptotic cells release ATP, AMP, and also UTP through the PANX1 channel as ‘find-me’ signals to attract macrophages and that the PANX1 channel is opened by caspase-3/7, which cleaves the C-terminal region of the channel (Chekeni et al., 2010; Elliott et al., 2009; Yamaguchi et al., 2014). The previous cell population-based study has reported that accumulation of extracellular adenine nucleotides correlates with decreases in intracellular ATP during the apoptotic progression of Jurkat cells (Boyd-Tressler et al., 2014). In order to investigate the impact of PANX1 on intracellular ATP dynamics during apoptosis at a single-cell level with high temporal resolution, we utilized the dual imaging setup for ATP and caspase-3 activity to PANX1-knock out (KO) HeLa cell lines (PANX1-KO1 and PANX1-KO2), which were generated using a CRISPR-Cas9 system (Figure 2A). Strikingly, decreases in the intracellular ATP levels of PANX1-KO cells after caspase-3 activation were significantly slower than those of wild-type HeLa cells when apoptosis was induced by anti-FAS antibody (Figure 2B–D). Knockout of PANX1 apparently has no effect on the ability of the cells to undergo cell death. A marked suppression of ATP decrease by knockout of PANX1 was also observed when apoptosis was induced by staurosporine (Figure 2E,F). Moreover, knockout of PANX1 also suppressed the decrease in ATP during TRAIL-induced apoptosis of SW480 human colorectal adenocarcinoma cells (Figure 2—figure supplement 1). Thus, PANX1 is involved in the facilitation of intracellular ATP decreases during apoptosis in multiple cell types and on various apoptotic stimuli. Notably, the intracellular ATP levels of PANX1-KO cells were almost unchanged in the first 30–60 min after the activation of caspase-3, followed by a gradual ATP decline (Figure 2C–F, Figure 2—figure supplement 1). The lag in the cytosolic ATP decrease observed for PANX1-KO cells might be partially relevant to the previous observation by Zamaraeva (Zamaraeva et al., 2005), which suggested the enhancement of cytosolic ATP level after apoptotic stimulation. Single-cell imaging also provided unexpected observations that intracellular ATP concentration transiently and repeatedly re-elevated on the course of the gradual ATP decrease in some populations of the PANX1-KO cells (Figure 3). Although the mechanism for these fluctuations in the intracellular ATP concentrations is unknown at present, the fluctuations must reflect either fluctuation in the rate of regeneration of ATP from ADP, that of adenosine nucleotide synthesis through de novo/salvage pathways, or that of degradation/release of ATP, or a combination thereof.

Figure 2 with 3 supplements see all
Knockout of PANX1 suppresses the cytosolic ATP decrease during apoptotic progression.

(A) Western blot analysis of PANX1 expression of wild-type and PANX1-KO HeLa cells. (B) Time-lapse images of the ATP level of a wild-type cell and a PANX1-KO cell. Pseudocolored ratio images of AT1.03CR are shown. Apoptosis was induced by anti-FAS and cycloheximide. The onset of caspase-3 activation was set as time = 0. Bar, 10 µm. (C, E) Single cell ATP dynamics of wild-type and PANX1-KO cells during apoptosis. Apoptosis was induced by anti-FAS and cycloheximide (C; 39 [WT], 18 [PANX1-KO1] and 22 [PANX1-KO2] cells from three biological replicates), or staurosporine (E; 15 [WT], 27 [PANX1-KO1] and 19 [PANX1-KO2] cells from three biological replicates). Each line represents the time course of the YFP/CFP ratio of AT1.03CR from a single cell, and was adjusted by setting the onset of caspase-3 activation as time = 0. Traces from different replicates were labeled with bars of different shades. (D, F) Effect of PANX1 knockout on the decrease in cytosolic ATP levels. Changes in YFP/CFP ratios at indicated time after the onset of caspase-3 activation were calculated for each apoptotic cell. Apoptosis was induced by anti-FAS and cycloheximide (D), or staurosporine (F). Analysis of variance (ANOVA) followed by post-hoc Dunnett’s test (versus wild-type).

PANX1 knockout cells show fluctuations of cytosolic ATP levels after the activation of caspase-3.

(A) Fluctuation of cytosolic ATP level of a PANX1-KO2 cell. Pseudocolored YFP/CFP ratio images of AT1.03CR are shown. Apoptosis was induced by anti-FAS and cycloheximide. Bar, 10 µm. (B, C) Representative traces of YFP/CFP ratios of AT1.03CR from single PANX1-KO cells from three biological replicates are shown. Apoptosis was induced by anti-FAS and cycloheximide (B), or staurosporine (C). Each trace was adjusted by setting the onset of caspase-3 activation as time = 0.

Effect of exogenous expression of PANX1 on single-cell dynamics of cytosolic ATP during apoptosis

Next, we exogenously expressed wild-type PANX1 in PANX1-KO HeLa cells. Cells overexpressing wild-type PANX1 precipitously lost their intracellular ATP, concomitant with the onset of caspase-3 activation, for both anti-FAS- and staurosporine-induced apoptosis (Figure 4A–E), further confirming that the PANX1 channel plays a major role in facilitating intracellular ATP decrease of apoptotic cells.

Exogenous expression of PANX1 alters cytosolic ATP dynamics of apoptotic cells.

(A) Time-lapse images of the ATP level of PANX1-KO2 cells. Cells were transfected with either an empty vector (upper) or a vector expressing wild-type PANX1 (PANX1WT) (lower). Pseudocolored ratio images of AT1.03CR are shown. (B, D) Single-cell ATP dynamics of PANX1-KO cells expressing exogenous PANX1WT. Cells were transfected with either an empty vector or a vector expressing PANX1WT. Apoptosis was induced by either anti-FAS and cycloheximide (B; 20 [PANX1-KO1 + vector], 40 [PANX1-KO1 + PANX1WT], 38 [PANX1-KO2 + vector] and 50 [PANX1-KO2 + PANX1WT] cells from three biological replicates), or staurosporine (D; 27 [PANX1-KO1 + vector], 36 [PANX1-KO1 + PANX1WT], 20 [PANX1-KO2 + vector] and 26 [PANX1-KO2 + PANX1WT] cells from three biological replicates). Each line represents the time course of the YFP/CFP ratio of AT1.03CR from a single cell, and was adjusted by setting the onset of caspase-3 activation as time = 0. (C, E) Effect of exogenous expression of PANX1WT on the decrease in cytosolic ATP levels. Changes in YFP/CFP ratios for 15 min after the onset of caspase-3 activation were calculated for each apoptotic cell. Apoptosis was induced by anti-FAS and cycloheximide (C), or staurosporine (E). Student’s t-test. (F) Single-cell ATP dynamics of wild-type cells exogenously expressing a caspase-resistant mutant of PANX1. Wild-type HeLa cells were transfected with either an empty vector or a vector expressing a caspase-resistant mutant of PANX1 (PANX1CR). Apoptosis was induced by anti-FAS and cycloheximide. Each line represents the time course of the YFP/CFP ratio of AT1.03CR from a single cell, and was adjusted by setting the onset of caspase-3 activation as time = 0 (35 [vector] and 41 [PANX1CR] cells from three biological replicates). (G) Effect of exogenous expression of PANX1CR on the decrease in cytosolic ATP levels. Changes in YFP/CFP ratios at indicated time after the onset of caspase-3 activation were calculated for each apoptotic cell. Apoptosis was induced by anti-FAS and cycloheximide. P-values of Student’s t-test are shown.

To examine that the intracellular ATP decrease in apoptotic cells is dependent on caspase-3 activity, we investigated the intracellular ATP dynamics of single apoptotic wild-type HeLa cells overexpressing a D376A/D379A mutant of PANX1 (PANX1-CR), in which the caspase recognition sequence close to the C-terminus of PANX1 is mutated (Chekeni et al., 2010). When PANX1-CR is overexpressed, most of the intrinsic wild-type PANX1 molecules are predicted to form hetero-heptamer with overexpressed PANX1-CR. The PANX1 hetero-heptamer will be expected to have much less channel activity than the wild-type PANX1 homo-heptamer as caspase-3 cannot separate the inhibitory C-terminal region from PANX1-CR inside the hetero-heptamers. In fact, it has been reported that the overexpression of PANX1-CR significantly suppresses the release of ATP from apoptotic cells (Chekeni et al., 2010). Accordingly, the intracellular ATP depletion was significantly protracted by the overexpression of PANX1-CR (Figure 4F; Figure 4G), clearly indicating that the cleavage of the C-terminal region of PANX1 by caspases is required for the PANX1-dependent intracellular ATP decrease of apoptotic cells.

Single-cell dynamics of cytosolic ATP during apoptosis under an OXPHOS-dominant culture condition

It is known that cells in normal adult tissues preferentially use oxidative phosphorylation (OXPHOS) in mitochondria for the regeneration of ATP, while cells in embryonic tissues and tumors use glycolysis (Vander Heiden et al., 2009). In the experiments described thus far we used HeLa cells cultured in glucose-containing medium. The cells preferentially regenerate ATP by glycolysis rather than OXPHOS under these conditions. To examine the role of PANX1 in apoptosis of OXPHOS-dominant cells, we compared the dynamics of intracellular ATP in PANX1-KO HeLa cells with those in wild-type cells during apoptotic progression under an OXPHOS-dominant culture condition. The cytosolic ATP level after the onset of caspase-3 activation dropped more quickly in the OXPHOS-dominant condition than in the glycolysis-dependent condition (Figure 2—figure supplement 2), probably because mitochondrial membrane potential ∆Ψm, which is required for mitochondrial ATP regeneration, is almost lost upon activation of caspase-3 (Figure 2—figure supplement 3). Even in this condition, knockout of PANX1 also suppressed the decrease in cytosolic ATP levels of the cells (Figure 2—figure supplement 2). This result indicates that PANX1 also promotes the intracellular ATP reduction of OXPHOS-dominant cells.

Extracellular AMP suppresses the decrease in the cytosolic ATP

If efflux of adenine nucleotides through PANX1 channel causes the decrease in the cytosolic ATP concentrations of apoptotic cells, extracellular adenine nucleotides would suppress the decrease by counteracting the efflux of its cytosolic counterpart. We monitored cytosolic ATP dynamics in apoptotic cells in the presence of AMP, ADP, or ATP in the culture medium, and found that extracellular AMP suppressed the decrease in intracellular ATP levels of apoptotic cells, while extracellular ADP and ATP exhibited no or negligible effects (Figure 5). Thus, it is most likely that the intracellular ATP decrease of apoptotic cells is a result of a reduction in adenosine nucleotide pools inside apoptotic cells, which is caused, at least in part, by the release of AMP from the cells. This result is consistent with the previous reports that AMP constitutes a large part of adenine nucleotides released from apoptotic cells (Yamaguchi et al., 2014; Boyd-Tressler et al., 2014).

Extracellular AMP counteracts the decrease in the cytosolic ATP level of apoptotic cells.

Wild-type HeLa cells expressing AT1.03CR and O-DEVD-FR were imaged in the presence of an adenine nucleotide (1 mM) in the culture medium. Apoptosis was induced by anti-FAS and cycloheximide. (A) Each line represents the YFP/CFP ratio of AT1.03CR from a single apoptotic cell, and was adjusted by setting the onset of caspase-3 activation as time = 0 (40 [control], 50 [AMP], 39 [ADP] and 39 [ATP] cells from three biological replicates). (B) Effect of the addition of extracellular nucleotide on the decrease in cytosolic ATP levels. Changes in FRET/CFP ratios at 0.5, 1.0, and 1.5 hr after the caspase-3 activation were calculated for each apoptotic cell. ANOVA followed by post-hoc Dunnett’s test (versus control).

PANX1 activity regulates free Mg2+ dynamics, but not phosphatidylserine externalization, in apoptotic cells

Next, we investigated whether the PANX1 channel is involved in apoptotic events other than intracellular ATP reduction. First, we examined the Mg2+ dynamics of apoptotic cells. Mg2+ is an essential divalent cation in cells, required for various cellular processes, including the activity of endonucleases and the compaction of chromosomes during cell division (Hartwig, 2001; Maeshima et al., 2018). It is known that most of the intracellular ATP form a complex with Mg2+, due to the high affinity of ATP for Mg2+ (Gupta and Moore, 1980; Grubbs, 2002). Thus, ATP acts as a major intracellular chelator for Mg2+. We hypothesized that the PANX1-dependent cytosolic ATP decrease might affect free Mg2+ in apoptotic cells, and investigated the dynamics of free Mg2+ in single apoptotic HeLa cells using a Mg2+-sensing fluorescent probe MGH (Matsui et al., 2017). In wild-type cells, free Mg2+ was transiently decreased after shrinkage of the cells. Subsequently, free Mg2+ began to increase and often reached higher than the basal level. In contrast, the fluctuations in free Mg2+ were significantly suppressed in apoptotic PANX1-KO cells (Figure 6A). Although the cause of the Mg2+ decrease observed immediately after cell shrinkage is unclear, it might be possible that PANX1 transiently releases Mg2+ from cytosol to extracellular space. The Mg2+ increase in the second phase might be coupled with the decrease in ATP, an intracellular Mg2+ chelator. Taken together, PANX1 regulates the dynamics of free Mg2+ in apoptotic cells, likely in part by decreasing ATP concentrations inside cells. Second, we examined the role of PANX1 on the externalization of phosphatidylserine (PS) on plasma membrane, one of the hallmarks of apoptosis (Elmore, 2007; Nagata, 2018). Externalized PS functions as an ‘eat-me’ signal for phagocytosis of apoptotic cells by macrophages. We examined whether PANX1 channel affects the externalization of PS in the plasma membrane by quantifying the amount of externalized PS using fluorescently-labeled annexin-V. As a result, no significant difference in the externalized PS was observed between wild-type and PANX1-KO HeLa cells (Figure 6B), suggesting that neither the PANX1-dependent intracellular ATP reduction or PANX1 itself does not contribute to the externalization of PS during apoptotic progression.

PANX1 regulates the cytosolic free Mg2+ level, but not the externalization of phosphatidylserine, in apoptotic cells.

(A) HeLa cells expressing Halo-tag were loaded with Halo-TMR and a magnesium indicator MGH(AM), followed by induction of apoptosis. Each trace represents a time course of normalized MGH/TMR fluorescence ratios of individual cell from three biological replicates. Timing of the initiation of cell shrinkage was defined as time = 0. Reddish and bluish traces represent wild-type cells (n = 22) and PANX1-KO2 cells (n = 20), respectively, from three biological replicates. Traces from different biological replicates were shown in different color codes. (B) Apoptosis of HeLa cells was induced by anti-FAS and cycloheximide in the presence of Alexa647-annexinV and propidium iodide. Fluorescence intensity of Alexa647-annexinV of each cell that does not show propidium iodide fluorescence was quantified using a fluorescent microscopy at 6 hr after induction of apoptosis (359 [WT] and 322 [PANX1-KO2] cells from three biological replicates).

PANX1 activation prevents glucose expenditure by apoptotic cells

We showed above that dying apoptotic cells retained intracellular ATP levels for longer periods when the PANX1 channel is lost or suppressed. These observations suggest that an intracellular system for regenerating ATP from ADP and phosphate may still be active in dying apoptotic cells. In living cells, glycolysis and OXPHOS play prominent roles in the regeneration of ATP. To examine whether these ATP regenerating pathways are active, we treated PANX1-KO cells with either an inhibitor for glycolysis or OXPHOS after activation of caspase-3 while monitoring the dynamics of cytosolic ATP levels (Figure 7A–D). We used 2-deoxyglucose (2DG) or sodium oxamate, which are an inhibitor for hexokinase and lactate dehydrogenase, respectively, to inhibit glycolysis, whereas used oligomycin A, which is an inhibitor for FoF1-ATP synthase, to inhibit OXPHOS. Treatment with either 2DG or sodium oxamate induced a rapid decrease in intracellular ATP concentration (Figure 7B,C). In contrast, treatment with oligomycin A, an inhibitor of OXPHOS, seemed to have only a small effect on intracellular ATP dynamics under this condition (Figure 7D). These observations indicate that apoptotic processes do not disrupt the glycolytic system of the cells and that even dying apoptotic cells retain the ability to regenerate ATP by glycolysis. The rapid decrease in cytosolic ATP concentration of apoptotic PANX1-KO cells by inhibition of glycolysis also implies that at least some of the intracellular ATP-utilizing systems are active during apoptosis if sufficient intracellular ATP is present. It has been previously reported that apoptotic cells treated with a PANX1 inhibitor showed continuous and extensive blebbing (Poon et al., 2014), which is dependent on myosin ATPase (Coleman et al., 2001). Consistently, we also observed that PANX1-KO cells showed more extensive blebbing than wild-type cells during apoptosis (Videos 1 and 2). Moreover, forced ATP depletion of apoptotic PANX1-KO cells by 2DG leaded to the reduction of the blebbing of the cells (Videos 3 and 4). It is also likely that the reduction of ATP during apoptosis leads to the decrease in the activities of other ATPases because ATPase activity depends on the concentration of ATP. We expected that if the intracellular ATP concentration of apoptotic cells is not depleted, the cycle of ATP consumption and regeneration will continue, resulting in a continuous glucose consumption by the cells. To examine glycolytic activity in apoptotic cells, we quantified the consumption of glucose and the release of lactate by wild-type and PANX1-KO HeLa cells after induction of apoptosis. Both cells exhibited similar glucose consumption and lactate production rates when apoptosis was not induced (Figure 7E). Glucose consumption and lactate production by wild-type cells had almost ceased by 16 hr after induction of apoptosis, while those by PANX1-KO cells continued for at least 32 hr when apoptosis was induced by anti-FAS (Figure 7F). Trends of glucose consumption and lactate production by those cells were quite similar when apoptosis was induced by ultraviolet (Figure 7G). Thus, apoptotic cells with deficient PANX1 activity have a prolonged glycolytic activity compared to normal apoptotic cells. Taken together, activation of PANX1 channels thwarts glucose expenditure of apoptotic cells, most likely by rapidly depleting intracellular ATP reserves (Figure 8).

PANX1 is required for suppressing consumption of glucose by apoptotic cells.

(A–D) Effect of metabolic inhibitors on cytosolic ATP of apoptotic cells. Apoptosis of PANX1-KO2 cells expressing AT1.03CR and O-DEVD-FR were induced by anti-FAS and cycloheximide. After activation of caspase-3, the cells were untreated (A) or treated either with 10 mM 2-deoxyglucose (B), 50 mM sodium oxamate (C), or 1 µg/ml oligomycin A. Each trace represents the YFP/CFP ratio of AT1.03CR from a single apoptotic cell, and was adjusted by setting the onset of caspase-3 activation as time = 0. A period of inhibitor treatment was indicated as a bar on each plot. Representative traces from three biological replicates are shown. (E) Glycolytic activity of living wild-type and PANX1-KO cells. Consumption of glucose (left) and release of lactate (right) in 8 hr is shown. (F–G) Glycolytic activity of apoptotic cells. Apoptosis was induced either by anti-FAS/cycloheximide (left) or ultraviolet (right) at time = 0. Consumed glucose (F) and released lactate (G) by wild-type Hela cells (black, dashed line) and PANX1-KO2 cells (red, solid line) are shown. Means ± s.d. (three biological replicates) and p-values of paired t-test were shown (E–G).

Proposed model for PANX1-dependent control of the cytosolic ATP level and glucose consumption of apoptotic cells.
Video 1
Time-lapse fluorescent movie of an apoptotic wild-type HeLa cell.

YFP (FRET) fluorescence from ATeam is shown. Apoptosis was induced by anti-FAS and cycloheximide. Time = 0 indicates the onset of caspase-3 activation.

Video 2
Time-lapse fluorescent movie of an apoptotic PANX1-KO HeLa cell.

The same experimental conditions as Figure 1—figure supplement 1.

Video 3
Time-lapse florescent movie of an untreated apoptotic PANX1-KO HeLa cell.

Apoptosis was induced by anti-FAS and cycloheximide. Time = 0 indicates the onset of caspase-3 activation.

Video 4
Time-lapse fluorescent movie of a 2DG-treated apoptotic PANX1-KO HeLa cell.

Apoptosis was induced by anti-FAS and cycloheximide. Time = 0 indicates the onset of caspase-3 activation. 2DG was added between 60 and 63 min.

Discussion

In this study, we developed two genetically encoded FRET-based biosensors AT1.03CR and O-DEVD-FR, which enabled dual imaging of both ATP levels and caspase-3 activities during the apoptotic process at the single-cell level. This method allowed us to analyze the single-cell dynamics of cytosolic ATP level after caspase-3 activation, which occurs at different times in different cells. It was clearly shown that the cytosolic ATP level remained almost constant until caspase-3 was activated (Figure 1C,D). We did not observe any profound immediate ATP changes upon induction of apoptosis (Figure 1—figure supplement 2), in contrast to the previous intracellular ATP analysis from populations of cells using firefly luciferase that has suggested acute ATP production when apoptosis is induced (Zamaraeva et al., 2005). This discrepancy is not clear at present. Once caspase-3 was activated, cytosolic ATP level of the cells started to drop and was typically depleted within 2 hr (Figure 1C,D). Previous studies have shown that sufficient levels of intracellular ATP are required for progression of apoptosis (Eguchi et al., 1997; Hu et al., 1999; Leist et al., 1997; Li et al., 1997; Zamaraeva et al., 2005) and that pre-reduction of intracellular ATP inhibits the activation of caspase-3 (Zamaraeva et al., 2005). Moreover, apoptosome formation has been shown to require dATP/ATP in vitro (Hu et al., 1999; Li et al., 1997). Taken together, it is likely that maintenance of high intracellular ATP is critical to activate caspases. Further analysis on cytosolic ATP dynamics demonstrated that PANX1 channels play a major role in intracellular ATP depletion in apoptotic cells (Figures 2 and 4, S4 and S5). It has been reported that PANX1 releases adenine and uridine nucleotides upon activation by cleavage of the C-terminal cytosolic region of the protein by effector caspases (caspase-3 and 7) and that the released nucleotides act as ‘find-me’ signals for attracting macrophages, which engulf apoptotic cells (Chekeni et al., 2010; Yamaguchi et al., 2014). In this study, we show that the caspase-dependent cleavage of PANX1 is also crucial for the intracellular ATP depletion, meaning that intracellular ATP depletion in apoptotic cells is a ‘programmed’ process rather than a passive phenomenon. The depletion of intracellular ATP is most likely the result of the decrease in intracellular adenine nucleotide pool, caused by the release of AMP from the cytosol to the extracellular space (Figure 8). It should be noted that the intracellular ATP decrease in apoptotic cells could not be completely stopped by knockout of PANX1 (Figure 2). Therefore, there must be an additional mechanism of decreasing the intracellular ATP level in apoptotic cells, which is also likely to contribute to the kinetic variations in cytosolic ATP reduction among individual apoptotic cells (Figure 2C–F, Figure 2—figure supplement 1).

Metabolism is one of the major biological activity of living systems. The metabolic activity of cells is an indicator of ‘the state of being alive’. In other words, in order for a cell to ‘die a complete death’, metabolism must stop. However, it has not been well understood how metabolism of dead cells ceases. Our results showed that PANX1 plays a major role in stopping glycolysis, a central part of metabolism, of apoptotic cells, most likely through facilitating ATP loss. The suppression of glucose expenditure by apoptotic cells may benefit surrounding live cells because the resource of glucose for surrounding live cells would be limited if dying cells continued to consume glucose. Besides, macrophages that eat apoptotic cells may encounter a risk of energy deprivation if the eaten cells actively consume glucose inside the macrophages.

Because the removal of apoptotic cells by macrophages is critical to suppress inflammation and autoimmune diseases (Nagata, 2018), stimulating local macrophages by ATP and related nucleotides released through PANX1 channels should be a biologically important process in apoptosis (Chekeni et al., 2010). However, it is not clear why apoptotic cells use ATP and related nucleotides as major chemoattractants for macrophages over others, such as lysophosphatidylcholine (Ousman and David, 2000). Although it is difficult to say whether the primary function of PANX1 opening in apoptosis is to deplete intracellular ATP reserves or to release chemoattractant, or both, one possible scenario may be that cells first developed the function of PANX1 for depleting intracellular ATP, followed by effective utilization of the released adenine nucleotides as chemoattractants.

How does the intracellular ATP concentration change in other types of cell death, such as necrosis and pyroptosis? Because a forced decrease of intracellular ATP levels reportedly switches the cell death fate from apoptosis to necrosis, it is believed that the intracellular ATP level will decrease in the early stage of necrosis (Eguchi et al., 1997; Leist et al., 1997; Tsujimoto, 1997). However, a precise investigation of when and how the intracellular ATP level changes in the necrotic process is still lacking, especially at single-cell resolution. In pyroptosis, gasdermin-D has been reported to form pores as large as 10 nm in diameter in the plasma membrane and to release adenine nucleotides (Liu et al., 2016; Russo et al., 2016). However, it is not clear whether a drop in the intracellular ATP level coincides with the gasdamin-D pore formation. The techniques used in this study might be useful for uncovering the mechanism of how ATP decreases in other types of cell death.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Cell line (H.sapiens)HeLaShin Yonehara LabAuthenticated by STR profiling; mycoplasma tested
Cell line (H.sapiens)HeLa-PANX1-KO1This studyPANX1-knockout line
Cell line (H.sapiens)HeLa-PANX1-KO2This studyPANX1-knockout line
Cell line (H.sapiens)SW480ATCCATCC: CCL-228; RRID:CVCL_0546
Cell line (H.sapiens)SW480-PANX1-KOThis studyPANX1-knockout line
Antibodyanti-FAS (Mouse monoclonal)Molecular biology laboratoryCat. #:SY-001, RRID:AB_591016
Antibodyanti-His tag (Mouse monoclonal)R and D SystemsCat. #: MAB050-100
Antibodyanti-PANX1 (Rabbit monoclonal)Cell SignallingCat# 91137, RRID:AB_2800167WB (1:1000)
AntibodyHRP-conjugated anti-β-actin (Mouse monoclonal)Santa Cruz BiotechnologyCat# sc-47778 HRP, RRID:AB_2714189WB (1:1000)
Recombinant DNA reagentpcDNA3.1-AT1.03 (plasmid)Imamura et al., 2009Original ATP biosensor
Recombinant DNA reagentpcDNA3.1-AT1.03CR (plasmid)This paperCaspase-resistant ATP biosensor
Recombinant DNA reagentpcDNA3.1-O-DEVD-FR (plasmid)This paperCaspase-3 biosensor
Recombinant DNA reagentIRAK027A14 (plasmid)Riken Bioresource CenterCat. #: HGX010814Human pannexin-1 cDNA clone
Recombinant DNA reagentpIRES2-Sirius (plasmid)This paper
Recombinant DNA reagentpIRES2-Sirius-PANX1 (plasmid)This paperWild-type human PANX1
Recombinant DNA reagentpIRES2-Sirius-PANX1CR (plasmid)This paperCaspase-resistant human PANX1 mutant
Recombinant DNA reagentpcDNA4-SCAT3.1 (plasmid)Nagai and Miyawaki, 2004Caspase-3 biosensor
Recombinant DNA reagentPannexin-1 CRISPR/Cas9 KO Plasmid (h)Santa Cruz BiotechnologyCat. #: sc-401635
Peptide, recombinant proteinTRAIL, humanR and D SystemsCat. #:
375-TL-010
Commercial assay or kitGlucose CII-test WakoWako Pure ChemicalsCat. #:
300167002
Commercial assay or kitLactate Assay Kit-WSTDojindoCat. #:
L256
Chemical compound, drugMGHMatsui et al., 2017
Chemical compound, drugHalo-TMRPromegaCat. #:
G8251
Chemical compound, drugAnnexin-V, Alexa647 conjugateMolecular ProbesCat. #:
A23204
Chemical compound, drugTetramethylrhodamine ethyl esterMolecular ProbesCat. #:
T669
Software, algorithmMetamorphMolecular DevicesRRID:SCR_002368

Materials

Oligomycin A, ATP, ADP, and AMP were obtained from Sigma-Aldrich. Solutions of ATP, ADP, and AMP were neutralized with sodium hydroxide before use. Anti-FAS antibody was from Molecular Biology Laboratory (Nagoya, Japan). Cycloheximide (CHX) was from Calboichem. Tetramethylrhodamine ethyl ester (TMRE) and alexa647-annexin-V were purchased from Molecular Probes. Recombinant human Trail and anti-His tag antibody were purchased from R and D. Trail was dissolved in DPBS(-) supplemented with 0.1% BSA at a concentration of 20 µg/mL, and then stored at −80 °C. Anti-His antibody was dissolved in DPBS(-) at a concentration of 1 mg/mL, and then stored at −80 °C. 2-Deoxyglucose were from Wako Pure Chemicals (Osaka, Japan). Other chemicals were purchased from Nacalai Tesque (Kyoto, Japan) unless otherwise noted.

Mammalian cell culture and gene knock out

Request a detailed protocol

The HeLa cell line was a kind gift from Prof. Shin Yonehara. The cell line was authenticated by short-tandem repeat profiling and was checked for mycoplasma infection. The cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, 1 g/L glucose; Nacalai Tesque) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich). The SW480 cell line was obtained from American Type Culture Collection. The cells were cultured in Leibovitz’s L-15 medium (Wako Pure Chemicals, Japan) supplemented with 10% FBS. Apoptosis of HeLa cells was initiated either by adding anti-FAS antibody (125 ng/mL) and CHX (10 µM) or by adding staurosporine (1 µM). Apoptosis of SW480 cells was initiated by adding Trail (50 ng/mL) and anti-His tag antibody (1 µg/mL). Knockout of PANX1 gene was carried out using pre-designed PANX1-KO CRISPR-Cas9 plasmids (Santa Cruz Biotechnology). Briefly, cells were transfected with CRISPR-Cas9 plasmids using PEI-Max (Polysciences Inc) as described previously (Morciano et al., 2020). After 2 days, each individual cell with strong GFP fluorescence was sorted into a well in 96-well plates using a cell sorter (SH800S, Sony), and then was cultured. Knockout of PANX1 gene in each cultured line was verified by western blotting and by sequencing of the targeted region of the genomic DNA.

Plasmids

The expression vector for the caspase-resistant AT1.03 (pcDNA-AT1.03CR) was constructed by introducing caspase-resistant mutations (D242N/D339G) into pcDNA-AT1.03 (Imamura et al., 2009) using PCR-based mutagenesis. O-DEVD-FR cDNA was constructed by fusing mKOκ and mKate2(V94S) through a Gly-Gly-Asp-Glu-Val-Asp-Gly-Thr linker using PCR. The amplified cDNA was cloned between XhoI and HindIII sites of pcDNA3.1(-) (Thermo Scientific) to obtain a mammalian expression vector pcDNA-O-DEVD-FR. Human PANX1 cDNA (Riken Bioresource Center) was amplified by PCR, and was cloned between XhoI and EcoRI sites of pIRES2-Sirius, a custom made vector, in which EGFP cDNA of pIRES2-EGFP (Clontech) was replaced by Sirius fluorescent protein cDNA (Tomosugi et al., 2009), to obtain a pIRES2-Sirius-hPANX1 plasmid. Caspase-resistant mutations (D376A/D379A) in PANX1 were introduced by PCR-based mutagenesis.

Fluorescence imaging of ATP levels and caspase-3 activities

Request a detailed protocol

Cells were transfected with the AT1.03CR plasmid and the O-DEVD-FR plasmid using PEI-Max as described previously (Morciano et al., 2020). For exogenous expression of PANX1, the plasmid encoding PANX1 cDNA was co-transfected with the AT1.03CR and the O-DEVD-FR plasmids. One day after transfection, cells were trypsinized and plated on a collagen-coated glass-bottom 4-compartment dish (0.16–0.19 mm thick; Greiner). Two days after transfection, cells cultured in phenol red-free DMEM supplemented with 10% FBS (HeLa) or phenol red-free Leibovitz’s L-15 medium supplemented with 10% FBS (SW480) were subjected to imaging. For OXPHOS-dependent cell culture, phenol red- and glucose-free DMEM (Gibco) supplemented with 10% FBS, 10 mM sodium lactate (Sigma-Aldrich), and 10 mM sodium dichloroacetate (an inhibitor of pyruvate dehydrogenase kinase, Sigma-Aldrich) was used. Cells were visualized with a Ti-E inverted microscope (Nikon, Tokyo, Japan) using a Plan Apo 40×, 0.95 numerical aperture, dry objective lens (Nikon). Cells were maintained on a microscope at 37°C with a continuous supply of a 95% air and 5% carbon dioxide mixture by using a stage-top incubator (Tokai Hit). All filters used for fluorescence imaging were purchased from Semrock (Rochester, NY): for dual-emission ratio imaging of AT1.03CR, an FF01-438/24 excitation filter, an FF458-Di02 dichroic mirror, and two emission filters (an FF02-483/32 for CFP and an FF01-542/27 for YFP); dual-emission ratio imaging of O-DEVD-FR, an FF01-543/22 excitation filter, an FF562-Di02 dichroic mirror, and two emission filters (an FF01-585/22 for mKOκ and an FF01-660/52 for mKate2). Cells were illuminated using a 75 W xenon lamp through 25% and 12.5% neutral density filters. Fluorescence emissions from cells were imaged using a Zyla4.2 scientific CMOS camera (Andor Technologies). The microscope system was controlled by NIS-Elements software (Nikon). Image analysis was performed using MetaMorph software (Molecular Devices). First, a background fluorescence intensity, which was measured from a region within image where no cell exist, was subtracted from an entire image. Next, the intensity of a donor fluorophore (YFP or mKate2) of a cell was divided by the intensity of an acceptor fluorophore (CFP or mKOκ) to obtain the emission ratio. The detailed method for the image analysis has been described previously (Morciano et al., 2020). Because fluorescent biosensors were abruptly dissipated from the cell when the plasma membrane of the apoptotic cell was collapsed, data after fluorescence values from cells were less than half of the maximum were excluded from the analysis. Single-cell ATP traces were visualized using PlotTwist (Goedhart, 2020). Plots of single-cell ATP changes were generated using PlotsOfData (Postma and Goedhart, 2019).

Fluorescence imaging of caspase-3 activity and ∆Ψm

Request a detailed protocol

HeLa cells were transfected with the pcDNA-SCAT3.1 plasmid (Nagai and Miyawaki, 2004) using Lipofectamine 2000 (Thermo Scientific). One day after transfection, cells were trypsinized and plated on a collagen-coated glass-bottom dish (0.16–0.19 mm thick; MatTek). Two days after transfection, the medium was replaced by phenol red-free DMEM containing 10% FBS and 50 nM TMRE. Then, the cells were visualized with a Ti-E inverted microscope (Nikon, Tokyo, Japan) using a Plan Apo 40×, 0.95 numerical aperture, dry objective lens (Nikon). Cells were maintained on a microscope at 37°C with a continuous supply of a 95% air and 5% carbon dioxide mixture by using a stage-top incubator (Tokai Hit). All filters used for fluorescence imaging were purchased from Semrock (Rochester, NY): for dual-emission ratio imaging of SCAT3.1 biosensors, an FF01-438/24 excitation filter, an FF458-Di02 dichroic mirror, and two emission filters (an FF02-483/32 for CFP and an FF01-542/27 for YFP); for imaging of TMRE, an FF01-562/40 excitation filter, an FF593-Di02 dichroic mirror, and an FF01-641/75 emission filter. Cells were illuminated using a 75 W xenon lamp through 25 and 12.5% neutral density filters. Fluorescence emissions from cells were imaged using a Zyla4.2 scientific CMOS camera (Andor Technologies). The microscope system was controlled by NIS-Elements software (Nikon). Image analysis was performed using MetaMorph software (Molecular Devices).

Fluorescence imaging of free Mg2+

Request a detailed protocol

Mg2+ imaging was performed using a synthetic Mg2+ indicator MGH (Matsui et al., 2017). HeLa cells maintained in 10% FBS in DMEM (Invitrogen) at 37°C under 5% CO2 were transfected with the plasmid pcDNA-3.1-(+)-Halo-Tag using PEI-Max (Polysciences). After 48 hr, the cells were washed twice with HBSS and incubated with 5 µM MGH(AM) for 30 min, then 50 nM Halo-TMR for 100 min. After washing the cells twice with HBSS, the medium was replaced by phenol red-free DMEM containing 10% FBS. After 4 hr, the cells were washed with HBSS. Imaging of the cells was started just after replacing the medium by phenol red-free DMEM containing 10% FBS, 50 ng/mL anti-FAS and 10 µM cycloheximide. Cells were visualized with a Ti-E inverted microscope using a Plan Apo 40×, 0.95 numerical aperture, dry objective lens (Nikon). Cells were maintained on a microscope at 37°C with a continuous supply of a 95% air and 5% carbon dioxide mixture by using a stage-top incubator (Tokai Hit). For imaging of MGH, an FF01-497/16 excitation filter, an FF516-Di01 dichroic mirror and an FF01-535/22 emission filter were used. For imaging of TMR, an FF01-562/40 excitation filter, an FF593-Di02 dichroic mirror, and an FF01-641/75 emission filter were used. All filters were purchased from Semrock.

Fluorescence imaging of phosphatidylserine

Request a detailed protocol

HeLa cells were cultured on a 35 mm glass-bottom culture dish. At 6 hr before imaging, the medium was replaced by 2 mL of phenol red-free DMEM supplemented with FBS (10%), CaCl2 (1 mM), propidium iodide (1 µg/mL), Annexin-V-Alexa647 (10 µL), anti-FAS (250 ng/mL) and CHX (10 µM). Cells were visualized with a Ti-E inverted microscope (Nikon, Tokyo, Japan) using a Plan Apo 20×, 0.75 numerical aperture, dry objective lens (Nikon). All filters used for fluorescence imaging were purchased from Semrock (Rochester, NY): for imaging of propidium iodide, an FF01-504/12 excitation filter, an FF593-Di02 dichroic mirror, and an FF01-562/40 emission filter; for imaging of Annexin- V-Alexa647, an FF02-628/40 excitation filter, an FF660-Di02 dichroic mirror, and an FF01-692/40 emission filter. Fluorescence emissions from cells were imaged using a Zyla4.2 scientific CMOS camera (Andor Technologies). Image analysis was performed using MetaMorph software (Molecular Devices). Fluorescence intensity of Alexa647 within a whole-cell area of each cell that was shrunk but did not exhibit propidium iodide fluorescence was quantified.

Glucose and lactate assay

Request a detailed protocol

HeLa cells (1.5 × 106) were cultured in 60 mm dish in DMEM (1 g/L glucose) supplemented with 10% FBS. After 24 hr, the medium was replaced by HBSS. Subsequently, apoptosis was induced either by replacing the medium with 4 mL of phenol red-free DMEM (1 g/L glucose) supplemented with 10% FBS, 250 ng/mL anti-FAS antibody and 10 µM cycloheximide, or by irradiating the cells with 20 mJ UV-C, followed by replacing the medium with 4 mL of phenol red-free DMEM (1 g/L glucose) supplemented with 10% FBS. Small aliquots of culture medium from the cell cultures were sampled at defined intervals. After centrifugation at 3000 x g for 3 min at 4 °C, the supernatant from each sample was stored at −30°C until the glucose and lactate quantification assay. Glucose and lactate concentrations of the aliquots were determined using Glucose CII-test Wako (Wako Pure Chemicals, Osaka, Japan) and Lactate Assay Kit-WST (Dojindo, Kumamoto, Japan), respectively.

Western blotting

Request a detailed protocol

Cleavage of ATeam in apoptotic cells was examined by western blotting. Briefly, apoptosis was induced in HeLa cells, which were previously transfected with pcDNA-AT1.03 or pcDNA-AT1.03CR, by adding anti-FAS antibody and CHX. The cells were harvested 12 hr after anti-FAS stimulation and lysed. After separation of the cell lysate by SDS-PAGE, ATeam and actin proteins were detected by western blotting using a polyclonal anti-GFP antibody (Invitrogen), which cross-reacts with both CFP and YFP, and a monoclonal anti-actin antibody (Millipore), respectively. Protein expression levels of PANX1 were examined by western blotting using a rabbit monoclonal antibody against human PANX1 (D9M1C, Cell Signaling). Actin was also detected as a control by western blotting using a mouse monoclonal anti-β-actin antibody (Santa Cruz). Horseradish peroxidase (HRP)-labeled anti-mouse or anti-rabbit IgG antibody (GE healthcare) was used as a secondary antibody. Chemi-Lumi One L reagent (Nacalai tesque) was used as a HRP substrate. A LAS4000 imager (GE healthcare) was used to detect the luminescence.

Data availability

All data analyzed during this study are included in the manuscript. Source data files have been provided for Figure 6 and 7. DNA and amino acid sequences for a developed caspase-3 biosensor is presented in Figure 1-figure supplement 3.

References

    1. Eguchi Y
    2. Shimizu S
    3. Tsujimoto Y
    (1997)
    Intracellular ATP levels determine cell death fate by apoptosis or necrosis
    Cancer Research 57:1835–1840.
    1. Gupta RK
    2. Moore RD
    (1980)
    31p NMR studies of intracellular free Mg2+ in intact frog skeletal muscle
    The Journal of Biological Chemistry 255:3987–3993.
    1. Hartwig A
    (2001) Role of magnesium in genomic stability
    Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 475:113–121.
    https://doi.org/10.1016/S0027-5107(01)00074-4
  1. Book
    1. Voet D
    2. Voet JG
    (2010)
    Biochemistry (Fourth Edition)
    Wiley Global Education.

Decision letter

  1. David Wallach
    Reviewing Editor; The Weizmann Institute of Science, Israel
  2. Jonathan A Cooper
    Senior Editor; Fred Hutchinson Cancer Research Center, United States
  3. George Dubyak
    Reviewer; Case Western Reserve University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

By designing an assay that allows continuous real-time dual imaging of intracellular ATP level and caspase-3 activity in single apoptotic cultured human cells, the authors have gained detailed knowledge of the kinetics of the decline in cellular ATP relative to caspase activation. This allowed the authors to nail down that adenine nucleotide via caspase activation of pannexin-1 channels early in apoptosis not only recruits phagocytes to the apoptotic cells, but also makes a major contribution to the eventual arrest of energy-dependent functions in the dying cells.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Pannexin-1 mediates programmed intracellular ATP decrease during apoptosis" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: George Dubyak (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Your manuscript "Pannexin-1 mediates programmed intracellular ATP decrease during apoptosis" has been examined by four expert reviewers whose detailed comments are enclosed. As you can see, the reviewers were very appreciative of the means that you have established for monitoring the dynamics of changes in the cytosolic ATP concentration on a single-cell level during apoptosis. It is certainly a valuable addition to the analytical toolbox applied in the study of cell death. The reviewers were reserved, however, about the fact that too much of the paper (including its title) concerns the finding that Pannexin-1 mediates the decrease of ATP in apoptosis – a finding that has been already documented before. They also raised many questions and requests about the evidence for the new biological insights that the paper aimed to present.

Since the breath of the requests made by the reviewers is too wide to allow addressing them within the two month period allowed by eLife, I am unfortunately unable to recommend further handling of current submission. However, we do hope that you will submit the paper anew to eLife after addressing the comments of the reviewers.

A simpler and quicker alternative would be to resubmit the paper, not as Research Article, but in the Tools and Resources category (after adding more detailed description of the ATP monitoring technique). This category is appropriate for a new technique that will be generally useful to a field, and although new biological insight is not required in this category you should be able also to include your major observations in the "Tools and Resources" format.

Reviewer #1:

Overall this is an interesting study that adds incremental insight to the original story from Chekeni et al., 2010, follow up study by Sandilos et al., 2012, and other more recent work. The use of imaging to assay ATP levels creates a potential for revealing spatial information but this potentially novel aspect is unfortunately not fully realized. Most importantly, clear interpretations of some of the key pieces of imaging data are hampered by the lack of controls, including controls demonstrating that the loss of signal is not a loss of cell area/cell collapse. Further, rather than relying primarily on the use of the blocker probenecid for a large portion of the study, the systematic use of a more complete genetic deletion model like Crispr Cas9, which should be readily feasible in the Hela model, would also make the interpretation and model more convincing. There are also some kinks in the logic/model that could be smoothened out (clarified) by more detailed explanations. In summary, there is potential here, but the manuscript seems premature.

Detailed comments:

Abstract: I'm not sure this sentence in the Abstract makes sense (nor is it substantiated by the data) – "Thus, the pannexin-1-mediated "programmed intracellular ATP decrease" of the apoptotic cell might benefit surrounding healthy cells by preserving environmental glucose."

Results first paragraph – Why is cell-specific information important? This isn't very well explained.

The first sentence should specify whether "between cells" means between cell types or between cells of a given cell type to emphasize the point being made.

Variable time interval range should be described in the text.

“It should be noted that the timing and the speed of the decreases in ATP did vary between cells, emphasizing the advantages of single-cell ATP analysis of apoptotic cells over the conventional cell population-based methods.” Great, but why is that precise information important?

Figure 1—figure supplement 2 should include a negative control consisting of a cohort of cells treated with a control antibody to control for cell death due to imaging conditions (note that during ATP/apoptosis imaging it's not mentioned in the Materials and methods whether the cells are being maintained at 37°C and in 5% CO2, however this is noted in the Mg2+ level measurements section). Some minimal details of the imaging/acquisition should be provided in the main body of the text.

Figure 1: The data shown in B for a single cell could also be aggregated/averaged and shown for a larger cohort of cells by adjusting the data as in part C. Part C should be added to Figure 1—figure supplement 2 as is the same data presented with an adjusted time scale, not clear how this really adds anything to the story. The images in Part A are much too small to appreciate and spatial information.

Perhaps a more important point is that the images provided in Figure 1A suggest that cell has collapsed between minute 297 and 300, rather than lost ATP, as it appears the FRET signal and TMRE signal are both now (at minute 300) quite spatially restricted but it is not clear that the space where the signal is absent is still "cell". Without a cell membrane marker like WGA (which can be used in live cells), or a DIC image (or similar) it is impossible to ascertain whether or not this is the case.

It seems that Figures 1 and 2 could be collapsed into one figure.

It seems that Figures 3 and 4 should be combined. To more convincingly show that PANX1 is the mediator of the caspase mediated ATP release, the authors should completely delete it using Crispr-Cas9 and then re-express the caspase resistant mutant (and compare with re-expressed WT). There should be at least 3 biological replicates and the traces from the individual biological replicates should be distinguishable from one another.

Figure S6: how many biological replicates here? Reporting on number of cells and number of biological replicates should be consistent throughout the figures.

Figure 5: The logic behind the interpretation of the results of Figure 5 isn't clear, please elaborate on "Strikingly, extracellular AMP and ATP, but not ADP, suppressed the decrease in intracellular ATP levels of apoptotic cells (Figure 5). Thus, it is most likely that the intracellular ATP decrease of apoptotic cells is a result of a reduction in adenosine nucleotide pools inside apoptotic cells, which is caused, at least in part, by release of ATP and AMP from the cells."

Figures 6 and 7: Cytosolic free Mg2+ and glucose consumption this also be assayed with Crispr-Cas9 deletion of Panx1, and the compared with wildtype and mutant caspase-defective PANX1 expression.

The conclusions drawn from Figure 7 don't make intuitive sense to me. While the cells aren't consuming the ATP they are still in effect losing ATP by releasing it, so less ATP is available for cellular processes, so they should still "want to consume" glucose.

Finally, given the nature of their studies the authors should take the following papers into consideration and add genetic models to strengthen their investigation of these aspects:

1) Probenecid interferes with renal oxidative metabolism: A potential pitfall in its use as an inhibitor of drug transport https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1572299/

2) Probenecid potentiates MPTP/MPP+ toxicity by interference with cellular energy metabolism. https://www.ncbi.nlm.nih.gov/pubmed/23802648

Reviewer #2:

The major novelty of this paper is the generation of new probes to study intracellular ATP, and using this probe the authors address the involvement of Pannexin-1 in decreasing intracellular ATP (ATPi) during apoptotic cell death via FRET-based single cell imaging. Additionally, albeit in less detail, the authors begin to show that apoptotic cells can use glycolysis during the cell death process but the loss of intracellular ATP impedes this.

Overall usage of the probe to track single cells and their intracellular ATP and some of the controls are well done and the findings are backed up with experimental evidence. However, as the authors acknowledge and based this work on, it is known that ATP is released via pannexin channels during apoptosis and this occurs via caspase-mediated cleavage of the channel. What is potentially novel is the possible differential involvement of glycolysis and OXPHOS during apoptosis when Panx1 function is blocked – however, the data supporting this claim is weak and will require additional experiments to fully understand the contributions of these metabolic processes. Additionally, the authors do not begin to explore what impact this process has on the neighboring tissue, and what the larger importance of it could be. If the authors can successfully address the major concerns, especially major concern 1, 2, and 3, then this work would potentially be a strong candidate for publication in eLife as it adds another level to the understanding of cell death process.

1) Although the authors show intracellular ATP decreases during apoptosis, what is the importance of this process? They speculate that there will be more glucose for surrounding cells (which makes them healthier?) because the apoptotic cells cannot use it as they will have lost their ATPi. However, they never actually provide evidence for this and by mentioning it in the Abstract it is misleading. This needs to be addressed much more in detail.

2) This paper primarily addresses one form of death – anti-Fas + methotrexate treatment of HeLa cells. The authors make much larger claims on what this would mean for the larger process of apoptosis and the glycolysis vs. OXPHOS may be impactful for the tissue, etc. But these are not justified. At the minimum, the authors should use one or two other cell lines and another form of cell death.

3) Lastly, the authors provide some indirect evidence for the function of glycolysis during apoptosis. By inhibiting Panx1 channels, thereby increasing ATPi , the authors show an increase in glucose uptake. However, this does not necessarily indicate increased glycolysis. Can the authors measure lactate release to more directly show that there is increased glycolysis during Panx1 inhibition (or even better via Seahorse Assay)? Interestingly, this may indicate that glycolysis is still occurring during apoptosis even if Panx1 is active although it may be with lower levels of ATPi and therefore to a less extent. This is an important point and needs to be proven.

4) The authors attempt to rule-out the involvement of mitochondrial depolarization as a mechanism for decreased ATPi by treatment of CCCP on live cells and showing that ATPi does not decrease. However, this is not a fair comparison because this does not indicate that it cannot be occurring during apoptosis (or in dying cells). Can the authors more clearly rule out this issue? For example, can the authors induce cell death, but inhibit mitochondrial depolarization, and show that ATPi is still decreasing regardless of mitochondria membrane potential? Does this affect the dynamics of ATPi decrease?

5) The authors state that "the ATP decrease in apoptotic cells is a caspase-3-dependent process", but they reach this conclusion based on timing events (correlational). Can the authors inhibit caspases with zVAD/QVD and show that ATPi levels do not decrease after anti-Fas treatment to directly prove this statement?

6) Can the authors show that a cellular ATPi levels still drop in a cell that endogenously/naturally uses OXPHOS and that this is still dependent on Panx1 activation? In this paper they use a system where glucose has been depleted and this can impact results. Also, does Panx1 inhibition still increase glycolysis in these cells? It would be interesting to know whether these cells attempt to switch their metabolism during cell death.

Reviewer #3:

This study describes a novel on-line optical probe to track the kinetics of intracellular ATP concentration in single HeLa cells during the induction/progression of apoptosis by Fas activation. HeLa are "type II" cells wherein extrinsic apoptotic (i.e., receptor-initiated) activation of the executioner caspase-3 is indirectly mediated by a caspase-8> Bid > mitochondrial permeabilization> caspase-9 cascade. This contrasts with type I cells wherein extrinsic apoptosis involves a direct caspase-8 > caspase-3 cascade. The major findings are that the decrease in intracellular ATP: 1) occurs shortly after collapse of mitochondrial membrane potential and accumulation of active caspase-3; and 2) is mediated by caspase-3 mediated proteolytic activation of plasma membrane pannexin-1 channels as conduits for efflux of ATP and other adenine nucleotides. Importantly, genetic or pharmacological suppression of pannexin-1 function markedly delays the decrease in intracellular ATP even when mitochondrial function or ATP synthesis is inhibited. That proteolytically-gated pannexin-1 channels can be major determinants of intracellular and extracellular adenine nucleotide metabolism/ distribution basically confirms several previous studies with populations of apoptotic cells. However, extending such findings to the single cell level is a significant contribution. One major issue requires attention.

Major concern: In the Figure 7 experiments, the authors use 2-deoxyglucose (2DG) as a "glycolysis inhibitor" to conclude that ongoing ATP synthesis via glycolysis is sufficient to sustain the intracellular ATP levels of apoptotic HeLa cell for several hours when pannexin-1-mediated ATP efflux is blocked by probenecid. There's major problem with this interpretation. 2DG acts an intracellular ATP "sink". Hexokinase readily phosphorylates 2DG, but 2DG-phosphate can't be further converted to fructose-6P. Thus, it's difficult to deconvolute the ATP-consuming action of 2DG from subsequent suppression of glycolytic ATP production. This needs to be addressed and alternative experimental support for the stated conclusion is required. Inhibition of glucose uptake by cytochalasins or other GLUT blockers might be an approach.

Reviewer #4:

Too much of this paper is dedicated to demonstration of the role of PANX1 cleavage in ATP release, and in the resulting depletion of the cellular ATP pool, which are, both, not new. (As the authors write "…it has (already) been reported that over-expression of PNAX1-CR (a dominant negative mutant) significantly suppresses the release of ATP from apoptotic cells").

What the paper should have focused at, are the following two things that are new and potentially important:

a) The more accurate information of the kinetics of ATP-loss during apoptosis, and its relation to the kinetics of other changes, which one can gain by applying the ATP

sensor that the authors devised to monitor the changes on a single-cell level.

b) Novel knowledge of the functional significance of the changes in the cellular ATP level in the course of apoptosis.

Both subjects are addressed, but in a way that is not thorough enough to yield solid conclusions.

With regard to the documentation of the kinetics of ATP loss, the paper claims to present findings that challenge a prior claim that the occurrence of apoptosis requests maintenance, or in fact – elevation, of the ATP cellular levels till a rather late stage in the process (Zamaraeva et al., 2005). However, the quality of the evidence that the current paper presents in support of its claims is harmed by the following shortcomings:

a) The means applied for arrest of PANX1 function have limitations. Knock down is incomplete. Probenecid has unrelated effects. The authors should have knocked out

the PANK1 gene by Crispr/Cas9 instead.

b) The cause for the marked intercellular variations in the effect of PANK1 inhibition on the kinetics of ATP decrease is not clear. The authors did not exclude contribution of variation in the extent of arrest of PANK1 function/expression, which makes it particularly important to use cells in which PANK1 was knocked out.

c) The apparent lag in the decrease of ATP pursuant to caspase activation/mitochondrial potential loss (Figure 1B), which might actually be related to the observations of Zamaraeva et al., and the intercellular variation in the length of this lag, requests further analysis and documentation. Does the length of the lag depend on the mode of cell death induction, on the type of cell, on growth condition? The possible mechanisms for this lag should be discussed.

Particularly in view of the dissimilarity to the study of Zamaraeva et al., where various kinds of cells and various means of death induction have been examined, there is need also to apply the tests of the current study in various cells and with various means of death induction. It is of particular importance to repeat the tests in cells induced to die through the type 1 extrinsic apoptotic pathway – a direct Fas (or TNF) >casp8>casp3>Panx1 cascade independent of perturbed mitochondrial function.

d) The evidence presented for some of the major conclusions in this paper (e.g. Figure 1B) is much too limited, constituting just of data acquired with a single cell.

e) It is not clear to which extent does the apoptotic volume-decrease, such as that seen in Figure 1A, affect the interpretation of the data acquired with the ATP sensor. To

which extent does the decrease of the signal indeed reflect decrease of ATP and not decrease in cell area? A clearer more complete description of the methodology e.g. more details of how the background subtractions were made, should be presented

With regard to the functional significance of ATP loss in apoptosis, the paper should have more thoroughly addressed the following three issues:

a) Apoptosis is largely believed to depend on maintenance of high ATP levels in the cells. At which exact stage of the apoptotic process does the maintenance of high ATP in the cells turn, from being requested for the apoptotic process, to become inhibitory to some of its manifestations?

b) Does the observed increase in glucose consumption indeed reflect enhanced glycolysis (just monitoring glucose levels does not suffice to tell that), and is this increase also occurring in other cell types, including ones that normally resort to oxidative phosphorylation, and in cells dying in response to other triggers?

c) Which of the various manifestations of apoptosis depend on the decrease of ATP. The authors suggest that, besides its contribution to arrest of glucose consumption, this decrease accounts for the role of PANX1 in restricting blebbing (Yoon et al., Nature 507: 329) and that it also contributes to DNA fragmentation/condensation through

increase of magnesium. However, they provide no evidence for that. One would expect comprehensive assessment of the impact of arrest of ATP decrease (e.g. by incubating the dying cells in growth medium containing ATP or AMP) on a wide range of functional and mechanistic events that are known to be associated with apoptosis.

https://doi.org/10.7554/eLife.61960.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

Detailed comments:

Abstract: I'm not sure this sentence in the Abstract makes sense (nor is it substantiated by the data) – "Thus, the pannexin-1-mediated "programmed intracellular ATP decrease" of the apoptotic cell might benefit surrounding healthy cells by preserving environmental glucose."

We have removed this discussion from the Abstract.

Results first paragraph – Why is cell-specific information important? This isn't very well explained.

The first sentence should specify whether "between cells" means between cell types or between cells of a given cell type to emphasize the point being made.

We have changed the sentence as follows: “In general, the progression of apoptosis varies between individual cells, even in the same cell type.”

Variable time interval range should be described in the text.

We have added the time information in the text as follows: “The cytosolic ATP levels in these cells started to decrease after a variable time interval (typically from 3 to 8 hours after apoptotic stimulation, see Figure 1—figure supplement 2).”

“It should be noted that the timing and the speed of the decreases in ATP did vary between cells, emphasizing the advantages of single-cell ATP analysis of apoptotic cells over the conventional cell population-based methods.” Great, but why is that precise information important?

This sentence has been removed because it was considered disruptive to the flow of the story.

Figure 1—figure supplement 2 should include a negative control consisting of a cohort of cells treated with a control antibody to control for cell death due to imaging conditions (note that during ATP/apoptosis imaging it's not mentioned in the Materials and methods whether the cells are being maintained at 37°C and in 5% CO2, however this is noted in the Mg2+ level measurements section). Some minimal details of the imaging/acquisition should be provided in the main body of the text.

Thank you for the suggestions. We added a negative control experiment (Figure 1—figure supplement 2). The cells did not die without the apoptotic stimulus. During imaging, cells were maintained at 37˚C and in 5% CO2. We described the conditions in Materials and methods section.

Figure 1: The data shown in B for a single cell could also be aggregated/averaged and shown for a larger cohort of cells by adjusting the data as in part C. Part C should be added to Figure 1—figure supplement 2 as is the same data presented with an adjusted time scale, not clear how this really adds anything to the story. The images in Part A are much too small to appreciate and spatial information.

The old Figure 1 has been removed because we no longer used TMRE as an indicator of caspase-3 activation in the revised experiments.

Perhaps a more important point is that the images provided in Figure 1A suggest that cell has collapsed between minute 297 and 300, rather than lost ATP, as it appears the FRET signal and TMRE signal are both now (at minute 300) quite spatially restricted but it is not clear that the space where the signal is absent is still "cell". Without a cell membrane marker like WGA (which can be used in live cells), or a DIC image (or similar) it is impossible to ascertain whether or not this is the case.

The fluorescence might have appeared to be localized to a part of the cell because the apoptotic cells shrank. There is no spatial restriction of the fluorescent signals. We used larger fluorescent images of the cells in the revised manuscript to clearly show cell shrinkage.

It seems that Figures 1 and 2 could be collapsed into one figure.

These figures have been removed in the revised manuscript.

It seems that Figures 3 and 4 should be combined. To more convincingly show that PANX1 is the mediator of the caspase mediated ATP release, the authors should completely delete it using Crispr-Cas9 and then re-express the caspase resistant mutant (and compare with re-expressed WT). There should be at least 3 biological replicates and the traces from the individual biological replicates should be distinguishable from one another.

We used PANX1-knockout cell lines, instead of using siRNA and inhibitors, in the revised manuscript. Because there are many panels, we decided not to combine figures. Traces from individual replicates were labeled with bars of different shades.

Figure S6: how many biological replicates here? Reporting on number of cells and number of biological replicates should be consistent throughout the figures.

Number of cells and number of biological replicates were presented throughout the revised manuscript.

Figure 5: The logic behind the interpretation of the results of Figure 5 isn't clear, please elaborate on "Strikingly, extracellular AMP and ATP, but not ADP, suppressed the decrease in intracellular ATP levels of apoptotic cells (Figure 5). Thus, it is most likely that the intracellular ATP decrease of apoptotic cells is a result of a reduction in adenosine nucleotide pools inside apoptotic cells, which is caused, at least in part, by release of ATP and AMP from the cells."

I agree that the explanation in the previous manuscript was ambiguous. We have added the following sentences in the revised manuscript: “If efflux of adenine nucleotides through PANX1 channel causes the decrease in the cytosolic ATP concentrations of apoptotic cells, extracellular adenine nucleotides would suppress the decrease by counteracting the efflux of its cytosolic counterpart.”

Figures 6 and 7: Cytosolic free Mg2+ and glucose consumption this also be assayed with Crispr-Cas9 deletion of Panx1, and the compared with wildtype and mutant caspase-defective PANX1 expression.

We re-measured glucose consumption and free Mg2+ dynamics using PANX1-knockout cells.

The conclusions drawn from Figure 7 don't make intuitive sense to me. While the cells aren't consuming the ATP they are still in effect losing ATP by releasing it, so less ATP is available for cellular processes, so they should still "want to consume" glucose.

In general, activity of enzymes, including ATPases, highly depends on substrate concentration. Although I agree that some ATPases with very low Km will retain their activity even at low ATP levels, the decreased intracellular ATP levels will reduce the net ATPase activity of the cell. We added the following sentence in the revised manuscript: “It is also likely that the reduction of ATP during apoptosis leads to the decrease in the activities of other ATPases because ATPase activity depends on the concentration of ATP.”

Finally, given the nature of their studies the authors should take the following papers into consideration and add genetic models to strengthen their investigation of these aspects:

1) Probenecid interferes with renal oxidative metabolism: A potential pitfall in its use as an inhibitor of drug transport https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1572299/

2) Probenecid potentiates MPTP/MPP+ toxicity by interference with cellular energy metabolism. https://www.ncbi.nlm.nih.gov/pubmed/23802648

Thank you for raising the concern. We decided to remove the data that used probenecid.

Reviewer #2:

[…]

1) Although the authors show intracellular ATP decreases during apoptosis, what is the importance of this process? They speculate that there will be more glucose for surrounding cells (which makes them healthier?) because the apoptotic cells cannot use it as they will have lost their ATPi. However, they never actually provide evidence for this and by mentioning it in the Abstract it is misleading. This needs to be addressed much more in detail.

Metabolism is one of the major biological activity of living systems. The metabolic activity of cells is an indicator of "the state of being alive". In other words, in order for a cell to "die a complete death", metabolism must stop. However, it has not been well understood how metabolism of dead cells ceases. Our results showed that PANX1 plays a major role in stopping glycolysis, a central part of metabolism, of apoptotic cells, through facilitating ATP loss. This is the biological importance of ATP depletion in apoptosis that we want to emphasize in the revised work. We discuss this statement in the revised manuscript. We agree that the model that PANX1 stops the waste of glucose by apoptotic cells to benefit surrounding cells is a discussion/speculation. We have deleted this point from Abstract.

2) This paper primarily addresses one form of death – anti-Fas + methotrexate treatment of HeLa cells. The authors make much larger claims on what this would mean for the larger process of apoptosis and the glycolysis vs. OXPHOS may be impactful for the tissue, etc. But these are not justified. At the minimum, the authors should use one or two other cell lines and another form of cell death.

In the revised manuscript, we analyzed staurospoine-induced apoptosis (Figure 2E-F, 4DE), as well as anti-FAS + cycloheximide-induced apoptosis for HeLa cells. In addition,

TRAIL-induced apoptosis was analyzed for SW480 cells (Figure 2—figure supplement

1).

3) Lastly, the authors provide some indirect evidence for the function of glycolysis during apoptosis. By inhibiting Panx1 channels, thereby increasing ATPi , the authors show an increase in glucose uptake. However, this does not necessarily indicate increased glycolysis. Can the authors measure lactate release to more directly show that there is increased glycolysis during Panx1 inhibition (or even better via Seahorse Assay)? Interestingly, this may indicate that glycolysis is still occurring during apoptosis even if Panx1 is active although it may be with lower levels of ATPi and therefore to a less extent. This is an important point and needs to be proven.

Thank you for the suggestion. Lactate release was also measured in the revised manuscript. Consistent with glucose consumption data, lactate release did not cease in PANX1-KO cells. To further support that existence of glycolytic activity, we showed that lactate dehydrogenase inhibitor sodium oxamate also reduces intracellular ATP, like hexokinase inhibitor 2-deoxyglucose does. Please note that we did not show that glycolysis is increased, just showed that glycolysis is retained at substantial levels in apoptotic cells.

4) The authors attempt to rule-out the involvement of mitochondrial depolarization as a mechanism for decreased ATPi by treatment of CCCP on live cells and showing that ATPi does not decrease. However, this is not a fair comparison because this does not indicate that it cannot be occurring during apoptosis (or in dying cells). Can the authors more clearly rule out this issue? For example, can the authors induce cell death, but inhibit mitochondrial depolarization, and show that ATPi is still decreasing regardless of mitochondria membrane potential? Does this affect the dynamics of ATPi decrease?

We have no idea to perform the suggested experiment. The relevant text and data have been removed in the revised version for clarity. We do not think that the removal of this result affects the major conclusion of this study.

5) The authors state that "the ATP decrease in apoptotic cells is a caspase-3-dependent process", but they reach this conclusion based on timing events (correlational). Can the authors inhibit caspases with zVAD/QVD and show that ATPi levels do not decrease after anti-Fas treatment to directly prove this statement?

We have confirmed that treatment with zVAD inhibit ATPi decrease after anti-FAS stimulation (Figure 1A).

6) Can the authors show that a cellular ATPi levels still drop in a cell that endogenously/naturally uses OXPHOS and that this is still dependent on Panx1 activation? In this paper they use a system where glucose has been depleted and this can impact results. Also, does Panx1 inhibition still increase glycolysis in these cells? It would be interesting to know whether these cells attempt to switch their metabolism during cell death.

As far as we know, cultured mammalian cells that are generally used for apoptosis research predominantly use glycolysis for ATP synthesis rather than OXPHOS. Therefore, it is difficult to conduct experiments to solve this question at present.

Reviewer #3:

[…]

Major concern: In the Figure 7 experiments, the authors use 2-deoxyglucose (2DG) as a "glycolysis inhibitor" to conclude that ongoing ATP synthesis via glycolysis is sufficient to sustain the intracellular ATP levels of apoptotic HeLa cell for several hours when pannexin-1-mediated ATP efflux is blocked by probenecid. There's major problem with this interpretation. 2DG acts an intracellular ATP "sink". Hexokinase readily phosphorylates 2DG, but 2DG-phosphate can't be further converted to fructose-6P. Thus, it's difficult to deconvolute the ATP-consuming action of 2DG from subsequent suppression of glycolytic ATP production. This needs to be addressed and alternative experimental support for the stated conclusion is required. Inhibition of glucose uptake by cytochalasins or other GLUT blockers might be an approach.

We agree that 2DG can potentially act as an ATP sink. In the revised manuscript, we also used sodium oxamate, an inhibitor for lactate dehydrogenase. Supply of NAD+, which is required for the reaction of GAPDH, is inhibited by this chemical. Consistent with our model, the addition of sodium oxamate to the apoptotic PANX1-KO cells rapidly reduced the intracellular ATP levels (Figure 7C).

Reviewer #4:

Too much of this paper is dedicated to demonstration of the role of PANX1 cleavage in ATP release, and in the resulting depletion of the cellular ATP pool, which are, both, not new. (As the authors write "…it has (already) been reported that over-expression of PNAX1-CR (a dominant negative mutant) significantly suppresses the release of ATP from apoptotic cells").

The reviewer might think that nucleotide release is synonymous with depletion of intracellular ATP concentration, but we do not think that the idea is correct. For example, astrocytes and vascular endothelial cells are known to release ATP frequently, but the ATP release is unlikely to deplete ATP in these cells. Most of previous PANX1-apoptosis works have focused on nucleotide release. Therefore, we disagree with the reviewer's view that there is no scientific novelty in examining in detail whether the mechanism of intracellular ATP depletion in apoptosis is related to the mechanism of nucleotide release.

What the paper should have focused at, are the following two things that are new and potentially important:

a) The more accurate information of the kinetics of ATP-loss during apoptosis, and its relation to the kinetics of other changes, which one can gain by applying the ATP

sensor that the authors devised to monitor the changes on a single-cell level.

b) Novel knowledge of the functional significance of the changes in the cellular ATP level in the course of apoptosis.

Both subjects are addressed, but in a way that is not thorough enough to yield solid conclusions.

Thank you for the positive comments on our work. We believe that the revised work has strengthen these conclusions.

With regard to the documentation of the kinetics of ATP loss, the paper claims to present findings that challenge a prior claim that the occurrence of apoptosis requests maintenance, or in fact – elevation, of the ATP cellular levels till a rather late stage in the process (Zamaraeva et al., 2005). However, the quality of the evidence that the current paper presents in support of its claims is harmed by the following shortcomings:

a) The means applied for arrest of PANX1 function have limitations. Knock down is incomplete. Probenecid has unrelated effects. The authors should have knocked out

the PANK1 gene by Crispr/Cas9 instead.

b) The cause for the marked intercellular variations in the effect of PANK1 inhibition on the kinetics of ATP decrease is not clear. The authors did not exclude contribution of variation in the extent of arrest of PANK1 function/expression, which makes it particularly important to use cells in which PANK1 was knocked out.

c) The apparent lag in the decrease of ATP pursuant to caspase activation/mitochondrial potential loss (Figure 1B), which might actually be related to the observations of Zamaraeva et al., and the intercellular variation in the length of this lag, requests further analysis and documentation. Does the length of the lag depend on the mode of cell death induction, on the type of cell, on growth condition? The possible mechanisms for this lag should be discussed.

Particularly in view of the dissimilarity to the study of Zamaraeva et al., where various kinds of cells and various means of death induction have been examined, there is need also to apply the tests of the current study in various cells and with various means of death induction. It is of particular importance to repeat the tests in cells induced to die through the type 1 extrinsic apoptotic pathway – a direct Fas (or TNF) >casp8>casp3>Panx1 cascade independent of perturbed mitochondrial function.

Because our results indicate that cytosolic ATP is maintained until caspase-3 is activated (Figure 1C-D), we think that our results do not fully challenge the previous claim that the occurrence of apoptosis requests maintenance of ATP. However, we did not observe an apparent increase in cytosolic ATP just after induction of apoptosis, unlike the observation by Zamaraeva and colleagues. The cause of this discrepancy is unknown at present. According to the reviewer’s suggestion, we have performed the same experiments again using PANX1 knockout cells and using different types of apoptotic stimulations. Again, we observed the short lag in ATP reduction and the kinetic variations in cytosolic ATP reduction after caspase-3 activation. The possible causes for the lag and the variations were discussed in the revised manuscript as described below.

“Notably, the intracellular ATP levels of PANX1-KO cells were almost unchanged in the first 30-60 min after the activation of caspase-3, followed by a gradual ATP decline (Figure 2C-F, Figure 2—figure supplement 1). The lag in the cytosolic ATP decrease observed for PANX1-KO cells might be partially relevant to the previous observation by Zamaraeva (Zamaraeva et al., 2005), which suggested the enhancement of cytosolic ATP level after apoptotic stimulation.”

In addition, we performed the experiments using SW480 cells (Figure 2—figure supplement 1), which is known to die through type 1 cell death pathway when stimulated with TRAIL (Özören and El-Deiry 2002, Neoplasia).

d) The evidence presented for some of the major conclusions in this paper (e.g. Figure 1B) is much too limited, constituting just of data acquired with a single cell.

We disagree with this assertion. We think that the power of single-cell analysis has led to these results. In addition, the single-cell ATP imaging technology, which is the basis of this research, has been proven to be practical by number of researchers over the past decade.

e) It is not clear to which extent does the apoptotic volume-decrease, such as that seen in Figure 1A, affect the interpretation of the data acquired with the ATP sensor. To

which extent does the decrease of the signal indeed reflect decrease of ATP and not decrease in cell area? A clearer more complete description of the methodology e.g. more details of how the background subtractions were made, should be presented

Since we monitor the ratio of YFP fluorescence and CFP fluorescence as a signal, any increase or decrease in fluorescence intensity due to cell shrinkage is offset. That is, this method is much more robust against morphological changes in cells than methods such as obtaining a single fluorescence wavelength. We added the following sentence. “It should be noted that any increase or decrease in fluorescence intensity due to cell morphological change was offset because we monitored the ratios of fluorescence intensities of an acceptor and a donor of the FRET biosensors.”

Details of background subtractions and ratio calculation were described in Materials and methods.

With regard to the functional significance of ATP loss in apoptosis, the paper should have more thoroughly addressed the following three issues:

a) Apoptosis is largely believed to depend on maintenance of high ATP levels in the cells. At which exact stage of the apoptotic process does the maintenance of high ATP in the cells turn, from being requested for the apoptotic process, to become inhibitory to some of its manifestations?

Previous studies have shown that the pre-reduction of ATP inhibits the activation of caspase-3 (Zamaraeva et al., 2005) and switches cell death form from apoptosis to necrosis (Eguchi, Shimizu and Tsujimoto, 1997; Leist et al., 1997). It is also known that dATP/ATP is also important for the formation of apoptosomes (Li et al., 1997; Hu et al., 1997). Our results demonstrated that cytosolic ATP is retained at high level until caspase-3 is activated. Thus, it is very likely that maintenance of high ATP is required until the activation of caspases. We added the above points into the revised manuscript (Discussion paragraph one). We found that PANX1-KO did not affect phosphatidylserine exposure, indicating that ATP loss is not critical to this process (Figure 6B). Apoptotic PANX1-KO cells did not immediately die a complete death because they did not completely stop glycolytic metabolism (Figure 7B). Also, apoptotic PANX1-KO cells continued to waste environmental glucose for more than 32 hours, which potentially causes starvation of the surrounding cells (Figure 7F-G and Figure 8).

b) Does the observed increase in glucose consumption indeed reflect enhanced glycolysis (just monitoring glucose levels does not suffice to tell that), and is this increase also occurring in other cell types, including ones that normally resort to oxidative phosphorylation, and in cells dying in response to other triggers?

Glycolytic activity of apoptotic cell is not enhanced, as evident from Figure 6E-G: Although apoptotic PANX1-KO cells retain glycolytic activity, the glucose consumption rate and the lactate release rate are slower than the wild-type cells.

c) Which of the various manifestations of apoptosis depend on the decrease of ATP. The authors suggest that, besides its contribution to arrest of glucose consumption, this decrease accounts for the role of PANX1 in restricting blebbing (Yoon et al., Nature 507: 329) and that it also contributes to DNA fragmentation/condensation through

increase of magnesium. However, they provide no evidence for that. One would expect comprehensive assessment of the impact of arrest of ATP decrease (e.g. by incubating the dying cells in growth medium containing ATP or AMP) on a wide range of functional and mechanistic events that are known to be associated with apoptosis.

We added new videos as supplementary materials, which clearly shows that an apoptotic PANX1-KO cell immediately stop blebbing after the forced decrease in intracellular ATP concentration by 2DG treatment. We think that this video strongly supports the idea that PANX1-enhanced decrease in intracellular ATP restricts excessive blebbing of apoptotic cells.

https://doi.org/10.7554/eLife.61960.sa2

Article and author information

Author details

  1. Hiromi Imamura

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Writing - original draft, Project administration
    Contributed equally with
    Shuichiro Sakamoto
    For correspondence
    imamura.hiromi.7a@kyoto-u.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1896-0443
  2. Shuichiro Sakamoto

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Writing - review and editing
    Contributed equally with
    Hiromi Imamura
    Competing interests
    No competing interests declared
  3. Tomoki Yoshida

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Data curation, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Yusuke Matsui

    Graduate School of Engineering, Osaka University, Suita, Japan
    Contribution
    Data curation, Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Silvia Penuela

    Department of Anatomy and Cell Biology, Schulich School of Medicine and Dentistry, University of Western Ontario, London, Canada
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4829-5517
  6. Dale W Laird

    Department of Anatomy and Cell Biology, Schulich School of Medicine and Dentistry, University of Western Ontario, London, Canada
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Shin Mizukami

    Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai, Japan
    Contribution
    Resources, Data curation, Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
  8. Kazuya Kikuchi

    Graduate School of Engineering, Osaka University, Suita, Japan
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
  9. Akira Kakizuka

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Conceptualization, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3513-0334

Funding

Japan Society for the Promotion of Science (22687011)

  • Hiromi Imamura

Japan Society for the Promotion of Science (24657101)

  • Hiromi Imamura

Japan Society for the Promotion of Science (16K14709)

  • Hiromi Imamura

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Shin Yonehara for technical advice and for providing HeLa cells, Hiroyuki Noji, Takeharu Nagai and Kenta Saito for technical advice, and James Alan Hejna and Kunio Takeyasu for critical assessment of this manuscript. We are grateful to Takeharu Nagai and Atsushi Miyawaki for sharing the SCAT3.1 plasmid. This work was supported by JSPS KAKENHI Grant Number 22687011, 24657101 and 16K14709 (to HI), and by the Platform Project for Supporting Drug Discovery and Life Science Research (Platform for Dynamic Approaches to Living Systems) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Japan Agency for Medical Research and Development (AMED).

Senior Editor

  1. Jonathan A Cooper, Fred Hutchinson Cancer Research Center, United States

Reviewing Editor

  1. David Wallach, The Weizmann Institute of Science, Israel

Reviewer

  1. George Dubyak, Case Western Reserve University, United States

Publication history

  1. Received: August 10, 2020
  2. Accepted: September 25, 2020
  3. Version of Record published: October 14, 2020 (version 1)
  4. Version of Record updated: October 19, 2020 (version 2)

Copyright

© 2020, Imamura et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,561
    Page views
  • 351
    Downloads
  • 20
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Hiromi Imamura
  2. Shuichiro Sakamoto
  3. Tomoki Yoshida
  4. Yusuke Matsui
  5. Silvia Penuela
  6. Dale W Laird
  7. Shin Mizukami
  8. Kazuya Kikuchi
  9. Akira Kakizuka
(2020)
Single-cell dynamics of pannexin-1-facilitated programmed ATP loss during apoptosis
eLife 9:e61960.
https://doi.org/10.7554/eLife.61960

Further reading

    1. Cell Biology
    2. Stem Cells and Regenerative Medicine
    Terren K Niethamer, Lillian I Levin ... Edward E Morrisey
    Research Article

    Following acute injury, the capillary vascular bed in the lung must be repaired to reestablish gas exchange with the external environment. Little is known about the transcriptional and signaling factors that drive pulmonary endothelial cell (EC) proliferation and subsequent regeneration of pulmonary capillaries, as well as their response to stress. Here, we show that the transcription factor Atf3 is essential for the regenerative response of the mouse pulmonary endothelium after influenza infection. Atf3 expression defines a subpopulation of capillary ECs enriched in genes involved in endothelial development, differentiation, and migration. During lung alveolar regeneration, this EC population expands and increases the expression of genes involved in angiogenesis, blood vessel development, and cellular response to stress. Importantly, endothelial cell-specific loss of Atf3 results in defective alveolar regeneration, in part through increased apoptosis and decreased proliferation in the endothelium. This leads to the general loss of alveolar endothelium and persistent morphological changes to the alveolar niche, including an emphysema-like phenotype with enlarged alveolar airspaces lined with regions that lack vascular investment. Taken together, these data implicate Atf3 as an essential component of the vascular response to acute lung injury that is required for successful lung alveolar regeneration.

    1. Cell Biology
    2. Microbiology and Infectious Disease
    Emma M Briggs, Catarina A Marques ... Keith R Matthews
    Research Article Updated

    African trypanosomes proliferate as bloodstream forms (BSFs) and procyclic forms in the mammal and tsetse fly midgut, respectively. This allows them to colonise the host environment upon infection and ensure life cycle progression. Yet, understanding of the mechanisms that regulate and drive the cell replication cycle of these forms is limited. Using single-cell transcriptomics on unsynchronised cell populations, we have obtained high resolution cell cycle regulated (CCR) transcriptomes of both procyclic and slender BSF Trypanosoma brucei without prior cell sorting or synchronisation. Additionally, we describe an efficient freeze–thawing protocol that allows single-cell transcriptomic analysis of cryopreserved T. brucei. Computational reconstruction of the cell cycle using periodic pseudotime inference allowed the dynamic expression patterns of cycling genes to be profiled for both life cycle forms. Comparative analyses identify a core cycling transcriptome highly conserved between forms, as well as several genes where transcript levels dynamics are form specific. Comparing transcript expression patterns with protein abundance revealed that the majority of genes with periodic cycling transcript and protein levels exhibit a relative delay between peak transcript and protein expression. This work reveals novel detail of the CCR transcriptomes of both forms, which are available for further interrogation via an interactive webtool.