Cell shrinkage is an incipient hallmark of apoptosis and is accompanied by potassium release that decreases the concentration of intracellular potassium and regulates apoptotic progression. The plasma membrane K+channel recruited during apoptosis has not been characterized despite its importance as a potential therapeutic target. Here we provide evidence that two-pore domain K+ (K2P) channels underlie K+ efflux during apoptotic volume decreases (AVD) in mouse embryos. These K2P channels are inhibited by quinine but are not blocked by an array of pharmacological agents that antagonize other K+ channels. The K2P channels are uniquely suited to participate in the early phases of apoptosis because they are not modulated by common intracellular messengers such as calcium, ATP, and arachidonic acid, transmembrane voltage, or the cytoskeleton. A K+channel with similar biophysical properties coordinates regulatory volume decreases (RVD) triggered by changing osmotic conditions. We propose that K2P channels are the pathway by which K+ effluxes during AVD and RVD and that apoptosis co-opts mechanisms more routinely employed for homeostatic cell volume regulation.
- self-referencing electrode
- cell volume
- cell shrinkage
apoptosis occurs in most cell types, including preimplantation mammalian embryos, and can be triggered by a variety of physiological and pathological stimuli. Independent of the cell type or inducer, this type of orchestrated cell death is nearly always accompanied by cell shrinkage (see Fig. 1 A; Refs. 15,35). Apoptotic volume decreases (AVD) precede other better understood processes of apoptosis such as cytochromec release, mitochondria membrane potential dissipation, caspase activation, and DNA fragmentation (15, 27, 35,41). The changes in intracellular ion concentrations accompanying AVD modulate caspase activity and thereby regulate the progression of apoptosis (4, 7, 16, 18). Consequently, the molecules underlying AVD might serve as suitable therapeutic targets for inducing apoptosis in cancer cells or preventing apoptosis in degenerative diseases. Despite its significance, the mechanisms underlying AVD are poorly understood (6, 15, 35, 44). Emerging evidence suggests that potassium release from cells participates in AVD (6, 7, 15, 27, 41,44); however, the responsible plasma membrane K+transporter or channel has not been characterized (4, 15,41). Here we report electrophysiological and pharmacological evidence demonstrating that apoptosis recruits two-pore domain K+ (K2P) channels.
Similarities exist between AVD and regulatory volume decreases (RVD) in response to hypoosmotic shock, predicting that AVD and RVD employ similar mechanisms (15, 35). Hypoosmotic conditions cause cells to swell initially, but in a compensatory action, cells regain their original volume by effluxing K+, Cl−, and organic osmolytes, which results in “osmotically obliged” water loss and RVD (15, 34, 35, 37). K+ can efflux from swollen cells through any of three mechanisms: electroneutral K+-Cl− symporters, an K+/H+ and Cl−/HCO antiporter system, or independent K+ channels (35, 44). In particular, calcium-independent, quinine-inhibitable K+channels participate in K+ efflux during RVD (11, 34,37). These K+ channels are voltage insensitive and resistant to several K+ channel antagonists such as charybdotoxin (11, 34, 35, 37). Volume decreases accompanying apoptosis might co-opt mechanisms more routinely employed during homeostatic cell volume regulation (as suggested in Refs. 15 and 21).
Interestingly, the emerging family of K2P channels possesses biophysical properties similar to the K+ channels participating in RVD (10, 23, 36). The K2Pchannels are calcium- and voltage insensitive and are not inhibited by traditional K+ channel antagonists (e.g., charybdotoxin). In addition to being modulated by phospholipids, some K2Pchannels are sensitive to environmental stresses such as heat and plasma membrane deformations (29, 30, 36). These K2P channels integrate stimuli, adjust membrane conductances accordingly, and therefore are uniquely poised to participate in AVD.
Mouse one-cell embryos (zygotes) exhibit robust RVD (19,37) and AVD (see Fig. 1; Refs. 25, 26,41) and provide an excellent system for analyzing K+ channels underlying volume changes, because the volume of these large (80-μM diameter) cells can be reliably monitored optically and zygotes are amenable to electrophysiological techniques (19, 31, 41). Oxidative stress (200 μM H2O2) evokes rapid AVD and K+efflux from zygotes (41). The K+ efflux associated with AVD can be noninvasively quantified by self-referencing a potassium-selective electrode within the diffusive boundary layer and thereby comparing the concentration of K+([K+]) in medium nearby embryos to that remote from embryos (for a description of the self-referencing electrode technique see Refs. 40 and 41). Using self-referencing electrode technology we demonstrate here that the K+ channels responsible for K+ efflux during AVD have pharmacological properties similar to those known to participate in RVD and consistent with K2P channels, suggesting that apoptosis co-opts existing mechanisms for homeostatic cell volume control.
Animals, embryo collection and culture.
Female B6C3F1 mice (6 wk old) were purchased from Charles River Laboratory (Boston, MA) and subjected to a 14:10-h light-dark cycle for at least 1 wk before use. Animals were cared for according to procedures approved by the Marine Biological Laboratory and Women and Infants Hospital Animal Care Committees. Zygotes were collected 21–22 h after injection of human chorionic gonadotropin from pregnant mare serum gonadotropin-primed female mice mated with males. After cumulus removal, zygotes were cultured at 37°C in humidified air (7% CO2) in modified potassium simplex optimized medium (KSOM) supplemented with nonessential amino acids and 2.5 mM HEPES. The zona pellucida (ZP) was removed mechanically after mild treatment with pronase. No differences in physiological signals were observed between ZP-intact and ZP-free zygotes.
Self-referencing potassium electrode technique and image acquisition.
The self-referencing system used to monitor [K+] around embryos and oocytes was identical to that previously described (40, 41). Physiological measurements were conducted at 37°C in covered glass-bottom petri dishes coated with poly-l-lysine (MatTek, Ashland, MA) with 4 ml of HEPES-buffered KSOM (HKSOM) containing reduced NaHCO3(4 mM) and elevated HEPES (14 mM). Potassium-selective electrodes (tip diameter of 3–5 μm) were fabricated using K+ ionophore I-cocktail B (Fluka, Milwaukee, WI) and backfilled with 100 mM KCl. All electrodes were calibrated and confirmed to be Nerstian before use (38, 39, 41). A silver/silver chloride reference electrode completed the circuit in solution by way of a 3 M NaCl-3% agar bridge. During recording, the potassium-selective electrode was oscillated in a dampened square wave parallel to the electrode axis over a distance of 10 μm with a frequency of 0.3 Hz. The near position of this oscillation was 5 μm from the ZP, or the plasma membrane in cases where the ZP was removed. Data acquisition and manipulation were performed as described previously (40, 41). The hardware and software controlling electrode movements, signal amplification, and data acquisition were designed and constructed by the BioCurrents Research Center at the Marine Biological Laboratory (Woods Hole, MA; www.mbl.edu/BioCurrents).
The morphometric features of zygotes were analyzed from digital images captured with a microscope-mounted Cohu analog video camera (Cambridge Research Instruments, Cambridge, MA) and a personal computer running Metamorph Software (Universal Imaging, West Chester, PA). Morphometric measurements were obtained from images with Metamorph. All physiology and morphometric data were processed in Excel (Microsoft, Seattle, WA) and Sigma Plot (SPSS Science, Chicago, IL). Data from individual embryos were interpolated and averaged using a program employing the interpolation function of MatLab software (MatLab, Cambridge, MA). Interpolation between data points gathered from any individual embryos allowed calculation of the average and standard deviation from a group of zygotes from which data were gathered at different times relative to the pharmacological treatment. All data are presented as means ± SD, and statistical comparisons (ANOVA) were done in Excel with statistical significance defined asP < 0.05.
Table 1 lists the pharmacological agents and concentrations used to determine the K+ channel subtype recruited during apoptosis. These concentrations used are consistent with those that effectively modulate K+ channels in other cell types. Stock solutions of agents were prepared in solvents (water, DMSO, or EtOH) and diluted in media to the working concentration on the day of use. The concentration of solvents did not exceed 0.1%, and control solvent applications were conducted on the same batches of embryos. Self-referencing technology was used to monitor [K+] near embryos before and during a 40-min exposure to each pharmacological agent. To test whether agents modulated H2O2-induced K+ efflux, zygotes were pretreated with each K+ channel inhibitor for 10–15 min followed by addition of 200 μM H2O2. For caged calcium experiments, zygotes were preloaded with 1-(4,5-dimethoxy-2-nitrophenyl)-1,2-diaminoethane-N,N,N′,N′-tetraacetic acid, tetra(acetoxymethyl ester) (DMNP-EDTA-AM; Molecular Probes, Eugene, OR) with standard procedures and calcium was uncaged by photolysis with 368-nm light (14, 32). Various patterns of photolysis were tested including continual light exposure for 10 s and pulsed exposure (5–10 pulses of 200- to 500-ms duration at 1 Hz) (32). Data were analyzed by ANOVA (Excel), and statistical significance was defined as P < 0.05.
The ratiometric fluorescent dye fura 2-AM was used to measure intracellular Ca2+ concentration ([Ca2+]). Zygotes were incubated with 2.5 μM fura 2-AM (Molecular Probes) in KSOM for 30 min at 37°C and then washed, and calcium imaging was conducted at excitation wavelengths of 334/380 nm and measured at an emission wavelength of 520 nm with the Attofluor imaging system. Free calcium measurements were taken every 15 s for 10 min before application of H2O2 and for 40 min subsequently. No transient rise in calcium was observed.
Treatment of one-cell mouse zygotes with 200 μM H2O2 evoked rapid AVD and K+ efflux (Fig. 1). During the first 25–35 min of H2O2 exposure, dying zygotes shrunk to ∼70% of their original cross-sectional area, which equated to a >40% decline in cell volume (Fig. 1). Concurrently, the [K+] near embryos gradually increased to a peak [K+] of 1.6 ± 0.5 μM above that of the bulk media (Fig. 1 B). By using the Fick equation (40), this change in [K+] can be equated to an efflux of 0.4 pmol of K+ · embryo−1 · min−1. K+ efflux ceased with extended H2O2exposure, and [K+] near embryos returned to homeostatic levels by 40–60 min (Fig. 1 B).
Previously, the broad-spectrum K+ channel blocker tetraethylammonium (TEA) was shown to inhibit H2O2-induced K+ efflux from zygotes, suggesting that the K+ effluxed through K+ channels in the plasma membrane (41). To delineate the specific class of K+ channel participating in AVD we used a series of pharmacological agents that target K+ channel subtypes (Table 1). Agonists and antagonists to ATP-sensitive, voltage-sensitive, and calcium-activated K+channels failed to modulate AVD associated K+ efflux (Table1), suggesting that apoptosis was not recruiting these channel types. The constancy of [Ca2+] during the first 40 min of H2O2 exposure when K+ efflux occurred (75 ± 14 nM 10 min before H2O2, 94 ± 14 nM after H2O2; n = 9) and the failure of uncaged calcium, strontium, and thapsigargin to modulate AVD (Table 1) indicated that the K+ channels underlying AVD were calcium independent. These pharmacological properties are consistent with those of K2P channels (10, 23,36).
K2P channels typically set membrane leak conductances as they allow K+ to move down its electrochemical gradient (23). To maintain high intracellular [K+], these leak conductances are counterbalanced by ATPase activity (e.g., Na+-K+-ATPase), such that at homeostasis there is no net K+ efflux or K+ influx. Block of the Na+-K+-ATPase with a cocktail of 50 μM strophanthidin and 1 mM ouabain (2, 3, 43) uncovered the resting leak potassium conductance indicative of the presence of K2P channels. On strophanthidin-ouabain exposure, a slight but statistically significant K+ efflux increased [K+] near embryos (0.13 ± 0.06 μM [K+] above background; n = 9,P < 0.003). This cocktail, however, did not inhibit H2O2-induced K+ efflux (1.4 ± 0.3 μM [K+] above background).
Some K+ channels underlying RVD are quinine sensitive (12, 37), and quinine (1 mM) reduced both AVD and K+ efflux from zygotes exposed to H2O2 (Fig. 2; Table 1). After pretreatment with quinine, H2O2evoked only modest shrinkage of zygotes to 93% of their original cross-sectional area (Fig. 2, A and B). [K+] of the medium near quinine/H2O2-treated embryos increased to only 0.1 μM above background compared with 1.6 μM in nearby control H2O2-treated zygotes (Fig.2 C; Table 1), indicating an attenuated H2O2-induced K+ efflux. Likewise, the Cl− channel antagonists 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) and DIDS, known to inhibit RVD, also reduced H2O2-evoked K+ efflux (Table 1), presumably by feedback between these interrelated ion systems. Likewise, clofilium tosylate and NPPB strongly inhibited AVD (NPPB = 99.7 ± 5.1% and clofilium = 95.0 ± 3.4% change in cross-sectional area, compared with 70% in control zygotes).
The cytoskeleton, particularly actin filaments, is disrupted by H2O2 (8, 42) and cleaved by apoptotic enzymes (caspases) (9, 13, 33) and modulates the activity of some K+ channels (28). However, K2P channels are not modulated by the cytoskeleton (23, 30, 36), and inhibition of caspases does not block AVD (5, 7, 16, 27). Direct disassembly of microfilaments in zygotes with cytochalasin D (2 μg/ml) did not evoke K+efflux (0.07 ± 0.08 μM [K+] above background), attenuate H2O2-induced K+ efflux (1.4 ± 0.2 μM [K+] above background), or prevent AVD (73% original cross-sectional area) (Fig.3 C; Table1). Apoptosis-associated membrane blebbing was blocked by cytochalasin D (Fig. 3 A), as reported in other cells (17). These results demonstrate that the K+channel underlying AVD is not modulated by the actin cytoskeleton, compatible with the properties of K2P channels.
Although ATP and mercury modulate cell volume changes, these agents did not inhibit K+ efflux associated with apoptosis. Application of 5 mM ATP effectively reduced RVD in zygotes (19,37). Similarly, ATP attenuated AVD but uncoupled K+efflux from AVD (Fig. 3, A and B). Zygotes preincubated with 5 mM ATP followed by H2O2shrunk to only 87% their original cross-sectional area compared with ∼70% in control zygotes (Fig. 3 B). Nevertheless, K+ efflux (1.5 ± 0.2 μM [K+] above background) and membrane blebbing occurred (Fig. 3 B). Similarly, the mercurial compound (30 μM HgCl2) known to block aquaporins (20) also failed to inhibit H2O2-induced K+ efflux (1.8 ± 0.3 μM [K+] above background), but remarkably, this K+ efflux was associated with cell swelling rather than AVD (Fig. 4). The time course of HgCl2/H2O2-induced cell swelling was similar to that of H2O2-induced AVD (Fig. 4 B), suggesting that independent channels move water and ions across the plasma membrane during apoptosis. Separate water and ion mechanisms also are likely to be employed during RVD, because RVD is similarly inhibited by ATP (37). The uncoupling of ion fluxes from volume changes challenges the direct role of ion fluxes in eliciting osmotically obliged water loss during RVD and AVD. Regardless, these results showed that the K+ channels underlying AVD are not modulated by ATP or mercury, in agreement with their identity as K2Pchannels.
Few pharmacological tools are available to directly probe K2P channels, but some of the six K2P channel family members are activated by arachidonic acid (AA) (10, 23,30, 36). AA (10 μM) failed to evoke K+ efflux, potentiate H2O2-induced K+ efflux, or attenuate AVD in mouse zygotes (Table 1). Several members of the K2P channel family are inhibited by protein kinase C (PKC) (23, 30, 45), and PKC activators suppress apoptosis and reduce RVD (16). Activating PKC with the phorbol ester phorbol 12-myristate 13-acetate (PMA) effectively inhibited both AVD and K+ efflux from zygotes induced by oxidative stress (Fig. 3, A and D). In the presence of PMA, H2O2 exposure evoked less shrinkage to only 88% of the zygote's original volume (Fig. 3,A and D) and significantly reduced K+efflux such that [K+] near embryos rose to only 0.3 μM above background (Fig. 3 D, Table 1). These observations suggest that a specific subtype of K2P channel, modulated by PKC but not AA, underlies AVD.
We demonstrate here that the channel responsible for K+ efflux during apoptosis has pharmacological and electrical characteristics similar to those known to participate in RVD and consistent with K2P channels, suggesting that apoptosis recruits existing mechanisms for homeostatic cell volume control. These K2P channels are not modulated by ATP, AA, calcium, the actin cytoskeleton, or transplasma membrane voltage, and therefore AVD can occur independent of other cellular signaling cascades. This independence could prevent apoptosis from being erroneously triggered in healthy cells or mistakenly attenuated in cells destined to die.
The K+ channels underlying AVD exhibit properties consistent with K2P channels. Both K2P channels and those participating in apoptosis are calcium- and voltage independent, resistant to several K+ channel antagonists such as charybdotoxin, dendrotoxin, and iberiotoxin, not modulated by the actin cytoskeleton, and inhibited by PKC (10, 23, 30, 36,45). Although specific pharmacological tools are not presently available for K2P channels, both K2P channels and the K+ channels underlying AVD are blocked by quinine derivatives (10, 23). One member of the K2Pchannel family, TRAAK, is activated by AA (10, 23, 30,36); however, AA failed to evoke K+ efflux or potentiate H2O2-induced K+ efflux or AVD (Table 1). The biophysical properties of the K2Pchannel underlying AVD most closely resemble those of the AA-insensitive K2P channel family member, TREK (23). Interestingly, TREK is nearly ubiquitously expressed and might provide a common mechanism by which apoptosis coordinates AVD in many cell types.
The K2P channel we describe here as participating in AVD exhibits biophysical properties remarkably similar to those of the K+ channel involved in RVD. Both AVD and RVD recruit a channel that is calcium- and voltage insensitive and inhibited by quinine derivatives. We propose that apoptosis is co-opting K+ channels more typically regulating homeostatic cell volume (15, 21). By employing these channels, a new cell volume set point is acquired by physiologically agreeable means. In support of the convergence of AVD and RVD mechanisms, osmotic changes can induce apoptosis (15, 21, 35).
In addition to participating in AVD and RVD, the K2Pchannel we characterize here might also participate in cell cycle regulation. Osmotic changes alter the cell cycle through yet unidentified mechanisms, and early stages of apoptosis are frequently accompanied by cell cycle arrest (1, 22, 24). When populations of somatic cells are treated with apoptotic agents, not all of the cells respond (7, 15, 18), which may reflect differences in cell cycle progression. In zygotes, Day et al. (9a, 9b) described a K+ channel that is tightly regulated by the cell cycle and is calcium- and charybdotoxin insensitive. Although we did not notice any relationship between the cell cycle and modulation of H2O2-induced AVD or K+ efflux, present studies are underway to investigate whether the K2Pchannel we characterize here links AVD, RVD, and cell cycle regulation progression.
None of the pharmacological agents that reduced K+ efflux evoked by H2O2 fully rescued embryo development (data not shown). H2O2 acts independently on several pathways (DNA damage, mitochondrial damage), only one of which is K2P channels and AVD. Because quinine only inhibits K+ efflux and AVD, it is not surprising that we were unable to completely rescue embryos from H2O2 insult with quinine alone. Nonetheless, the inhibition of K+efflux by quinine establishes a quinine-sensitive process as underlying AVD, one piece of evidence consistent with participation of K2P channels in apoptotic K+ efflux and AVD (23).
In summary, we demonstrate that the K+ channel responsible for K+ efflux during apoptosis in zygotes shares biophysical features with K2P channels. Furthermore, we suggest that apoptosis co-opts K2P channels usually responsible for homeostatic regulation of cell volume.
We thank Kasia Hammer, Richard Sanger, and Jane McLaughlin for assistance with the self-referencing technique and pharmacology and Gaudenz Danuser for interpolation support.
This work was supported by the Lalor Foundation (D. L. Keefe and J. R. Trimarchi) and National Center for Research Resources Grant P41-RR-01395 (P. J. S. Smith).
Address for reprint requests and other correspondence: D. L. Keefe, Laboratory for Reproductive Medicine, Lillie Bldg., Marine Biological Laboratory, 7 MBL St., Woods Hole, MA 02543 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2002 the American Physiological Society