PerspectiveCell Biology

Apoptosis--the Calcium Connection

See allHide authors and affiliations

Science  04 Apr 2003:
Vol. 300, Issue 5616, pp. 65-67
DOI: 10.1126/science.1083628

The cells of our body are able to quickly commit suicide in response to genetic or environmental cues, a process termed apoptosis. This process is essential for development, tissue homeostasis, and defense against pathogens. Organized life requires cell death, and execution of cell death relies on the very machinery of life. Mitochondria, the organelles that produce energy through cellular respiration, integrate death signals mediated by proteins belonging to the Bcl-2/Bax family, and kill cells by releasing critical factors such as cytochrome c that activate executioner caspase proteases (1, 2). Calcium ions (Ca2+), the cellular messengers that control every aspect of cell and tissue physiology, can be turned into death signals when delivered at the wrong time and place (3, 4). Mitochondria eventually decide whether Ca2+ signals are decoded as life or death signals (5), but it is not clear whether Ca2+ is an additional stress factor that “tips the balance” or is an obligatory signal for death. On page 135 of this issue, Scorrano et al. (6) demonstrate that the transfer of Ca2+ from the endoplasmic reticulum (ER) to the mitochondria is required for initiation of programmed cell death by some, but not all, apoptotic signals (see the figure). Their elegant approach of genetically inactivating crucial proteins and reconstituting them in specific organelles reveals that the Ca2+ content of the ER determines the cell's ability to commit suicide, defining the ER as a new gateway to apoptosis.

The ER Ca2+ apoptotic gateway.

(A) Under normal conditions, Ca2+ continuously cycles between the ER and mitochondria. Ca2+ is pumped into the ER by Ca2+ ATPases (SERCA), and released by IP3-gated channels (IP3R). Ca2+ enters mitochondria by a Ca2+ uniporter (mCU) and is released by a Na+/Ca2+ exchanger (mNCE). The ER Ca2+ load reflects the balance between Bcl-2 and Bax/Bak proteins. (B) Ablation of Bax/Bak decreases the ER Ca2+ load and protects cells from apoptosis (6). (C) Correction of the ER Ca2+ load by overexpressing SERCA proteins or selective expression of Bax in mitochondria (D) defines three classes of apoptotic stimuli: (i) stimuli that engage the ER Ca2+ gateway and do not require Bax/Bak in mitochondria, such as arachidonic acid, ceramide, and oxidative stress; (ii) stimuli that do not engage the ER Ca2+ gateway but require mitochondrial Bax/Bak, such as the “BH3-only” protein tBid; and (iii) stimuli that engage both pathways, such as T cell receptor activation, staurosporine, etoposide, and brefeldin A.


By using Ca2+ as an intracellular messenger, cells walk a tightrope between life and death. Because of the toxicity of Ca2+ ions, a low Ca2+ concentration must be maintained in the cytoplasm, and most of the cellular Ca2+ is stored in the ER. Ca2+ is pumped into the ER by SERCA ATPases (adenosine triphosphatases) and is released only transiently during bouts of signaling by the opening of inositol 1,4,5-trisphosphate (IP3) or ryanodine receptor (RyR) Ca2+-release channels (3). A significant fraction of the released Ca2+ is captured by mitochondria, which are strategically located near Ca2+-release channels (7). This close connection allows mitochondria to modulate, propagate, and synchronize Ca2+ signals (8) and to prevent ER depletion by recycling Ca2+ to the ER (9) (see the figure). This ER-mitochondria connection enables Ca2+ signals not only to fine-tune cellular metabolism (10) but also to modulate the ability of mitochondria to undergo apoptosis (5, 11). The switch from a life to a death signal involves the coincidental detection of Ca2+ and proapoptotic stimuli, and depends on the amplitude of the mitochondrial Ca2+ signal (12). The magnitude of mitochondrial Ca2+ signals, in turn, depends largely on the Ca2+ content of the ER, which is maintained by the balance between active Ca2+ pumping by SERCA and passive leakage through Ca2+-release channels (see the figure).

Several studies indicate that the Ca2+ content of the ER determines the cell's sensitivity to apoptotic stress. Procedures that decrease the Ca2+ concentration in the ER, such as genetic ablation of the ER Ca2+-buffering protein calreticulin or overexpression of plasma membrane Ca2+ ATPases, protect cells from apoptosis (13, 14). Conversely, procedures that increase the ER Ca2+ load, such as overexpression of SERCA or calreticulin, sensitize cells to apoptotic stress (14, 15). Sensitivity to apoptosis correlates with the total ER Ca2+ load, rather than with the free ER Ca2+ concentration, and depends on the ability of cells to transfer Ca2+ from the ER to the mitochondria. Accordingly, procedures that enhance the transfer of Ca2+ from the ER to mitochondria augment ceramide-induced cell death (16).

Further studies suggest that the balance between pro- and antiapoptotic Bcl-2/Bax family members regulates the ER Ca2+ content. Overexpression of the antiapoptotic protein Bcl-2 decreases the ER Ca2+ load and protects cells from death (17, 18). Apoptosis is restored by correcting ER Ca2+ levels with SERCA and correlates with an increase in mitochondrial Ca2+ (14). Conversely, Bax/Bak overexpression favors the transfer of Ca2+ from ER to mitochondria and induces cell death (19, 20). How Bax, Bak, and Bcl-2 interfere with the ER Ca2+ load is uncertain, but clearly this interference depends on the ER-mitochondria connection. Bcl-2 protects from death only when targeted to the ER, but kills when directed at mitochondria (21). ER-targeted Bcl-2 protects from death induced by mitochondria-targeted Bax (22), suggesting that the ER exerts a dominant role in its coupling to mitochondria.

All of these gain-of-function studies point to the central part played by Bcl-2/Bax in regulating ER Ca2+ and suggest that the transfer of Ca2+ from the ER to mitochondria induces apoptosis. But do specific apoptotic signals really require this Ca2+ connection to kill? Scorrano et al. use a unique model to answer this question. They discovered that mice deficient in both Bax and Bak, so-called double knockouts, are resistant to a wide range of apoptotic stimuli (23) and show a decrease in ER Ca2+ load. Double knockout cells from these mice exhibit no intrinsic defects in the handling of mitochondrial Ca2+ but cannot deliver enough Ca2+ to the mitochondria because their ER Ca2+ content is too low. Genetic correction of the ER Ca2+ load by overexpression of SERCA restored the killing ability of Ca2+-mobilizing apoptotic agents. This demonstrates that Ca2+ acts on mitochondria independently of Bax or Bak. Conversely, expression of mitochondria-targeted Bax restored death selectively to the “BH3-only” protein tBid. This indicates that tBid requires mitochondrial Bax but not Ca2+ transfer from the ER to mitochondria to induce cell death. The ability to rescue apoptosis by expressing organelle-targeted proteins in cells deficient in both Bax and Bak neatly defines which apoptosis signals are controlled by ER Ca2+ and which by mitochondrial Bax. From a practical standpoint, this provides a useful classification of apoptotic stimuli into three categories (see the figure).

The first category comprises apoptotic signals—such as arachidonic acid, ceramide, and oxidative stress—that engage the ER Ca2+ gateway but do not require Bax/Bak action in mitochondria. These agents release Ca2+ themselves and kill more efficiently when Ca2+ is further increased by physiological or pathological stimuli, accounting for the “Ca2+-preconditioning” observed in previous studies (5). Killing absolutely requires an increase in mitochondrial Ca2+, and thus strictly depends on ER Ca2+ levels. In the second category are agents, such as tBid, that require the presence of Bax or Bak in the mitochondria but do not engage the ER Ca2+ gateway. These agents do not require mitochondrial Ca2+ and kill efficiently at all ER Ca2+ loads. The third category is constituted of agents—such as etoposide, staurosporine, brefeldin A, and T cell receptor activation—that engage both pathways. These agents require both Ca2+ and the presence of Bax or Bak in mitochondria, and both ER Ca2+ and Bax/Bak levels modulate their killing potency.

The Bax/Bak-deficient mouse cells of Scorrano et al. are the first loss-of-function model in which an alteration in Ca2+ handling is causally linked to cell killing, but the mechanism leading to decreased ER Ca2+ is not established. The presence of normal amounts of Ca2+ signaling proteins in Bax/Bak-deficient cells suggests that the defect is either directly caused by the Bax/Bak proteins themselves or is mediated by a change in activity, rather than content, of a Ca2+ handling protein. A possible candidate for such modulation is the IP3 receptor, the principal Ca2+-release channel of the ER, whose activity undergoes complex regulation by Ca2+, ATP, and phosphorylation. SERCA proteins are also subject to modulation, and it will be interesting to see whether Bax/Bak inactivation is associated with changes in activity of the IP3 channel or of SERCA. Another likely partner is the antiapoptotic protein Bcl-2. The effects of Bak/Bax inactivation mimic those of Bcl-2 overexpression, suggesting that the balance between Bax/Bak and Bcl-2, rather than the amounts of the individual proteins, determines ER Ca2+ load. Manipulation of Bcl-2 expression in Bax/Bak-ablated cells will allow researchers to test directly this “rheostat” model, and to confirm whether Bcl-2 and Bax/Bak indeed coregulate ER Ca2+.

The Scorrano et al. study defines a new role for the ER-mitochondria Ca2+ connection. The ER is now envisioned as a gun pointed at the mitochondria, which can be loaded and unloaded with Ca2+ by Bax and Bcl-2 proteins. Some, but not all, apoptotic signals are able to pull the ER Ca2+ trigger, and hence to kill cells in a strictly Ca2+-dependent manner. Future studies will determine whether this mechanism also occurs when Bcl-2 family members are expressed at physiological levels in vivo, and whether physiological death signals are able to pull the Ca2+ trigger.


View Abstract

Navigate This Article