The stealthy assassin creeps up to the victim and quickly injects a dose of deadly poison. No, it's not a scene from a spy novel, but one of the ways in which the immune system's natural assassins — cytotoxic T lymphocytes and natural killer cells — can kill their target cells. The deadly poison in this case is a serine protease called granzyme B, and the killing mechanism is discussed by Christopher Froelich and colleagues in The Journal of Cell Biology.

Much is known about granzyme B — there's evidence that it kills with or without caspase activation, and that it causes mitochondrial permeabilization with or without members of the BCL-2 family. So Froelich and co-workers set out to assess the relative contributions of these pathways, to come up with a probable scheme for how granzyme B normally acts. To do this, they measured several different hallmarks of apoptosis — loss of mitochondrial membrane potential (ψm), DNA fragmentation and activation of pro-caspase precursors.

Most previous studies have been done on cell lysates, after granzyme B has been delivered to its target. By contrast, Froelich and colleagues studied whole cells after delivery of the physiological form of granzyme B, a granzyme B–serglycin (SG) complex. Previous studies with lysates had indicated that granzyme B-treated cells rapidly acquire cleaved (and presumably active) pro-caspase-3. Mature caspase-3 and granzyme B then process caspase-7 to complete the initial phase of apoptosis. The authors therefore asked how the absence of pro-caspase-3 might alter the pattern of apoptosis in whole cells after delivery of granzyme B–SG.

The authors compared MCF-7 cells that lacked pro-caspase-3 (MCF-7vec) with cells that contained a stable transfectant that expressed this zymogen (MCF-7casp-3), after delivery of granzyme B–SG. They saw DNA fragmentation and a loss of ψm in the MCF-7casp-3 cells, but not in the MCF-7vec cells, indicating that caspase-3 is essential for starting apoptosis. The authors also showed that the requirement for caspase-3 was not due simply to the fact that caspase-7 could no longer be activated in the caspase-3-deficient cells.

If caspase-3 is the crucial initiator, what's the contribution of the BCL-2-family members? First, the pro-apoptotic members — such as BAX, BAK and the so-called 'BH3-only' protein BID — might be involved in mitochondrial permeabilization by forming large ion channels in the outer mitochondrial membrane. Froelich and colleagues looked into this by studying the sensitivity of Bax/Bak-deficient mouse embryo fibroblasts (MEFs) to granzyme B-mediated killing. The Bax/Bak-deficient MEFs showed lower levels of active caspase-3 intracellularly and lower rates of ψm loss than wild-type MEFs, which indicates that these proteins are required for the response to granzyme B. The authors also studied whether BID is cleaved in response to granzyme B. Unlike previous studies, however, they could not detect the truncated (active) form of BID when precautions were taken to minimize proteolysis during processing of cells for immunoblots.

A second role for certain members of the BCL-2 family is in preventing cell death. The eponymous BCL-2 has been shown to block granzyme B-mediated apoptosis, and the authors wondered whether it might do this by acting directly on caspase-3. They therefore looked at the effects of granzyme B–SG in a Jurkat cell line that overexpressed BCL-2. In contrast to wild-type cells, the JurkatBCL-2 cells showed no caspase-3 activity, loss of ψm or DNA fragmentation. BCL-2 did not inhibit isolated active caspase-3, and processed caspase-3 was observed by immunoblot, leading the authors to conclude that BCL-2 might suppress apoptosis by blocking the assembly of processed caspase-3.

On the basis of these and several other observations, Froelich and colleagues suggest that the predominant granzyme B death pathway starts with the activation of pro-caspase-3, and that the death signal is maximized by mitochondria through caspase-mediated engagement of the BH3-only/BAX/BAK pathway. The BH3-only protein that's cleaved by caspase-3 might be BID, but Froelich and colleagues' results indicate that there might be other participants too.