Response: Granzyme A: cell death–inducing protease, proinflammatory agent, or both?

We recently showed that granzyme A (GzmA) cleaves poly(adenosine 5′-diphosphate-ribose) polymerase-1 (PARP-1) to promote cell death and inhibit DNA repair.¹  PARP-1 cleavage is required for apoptotic cell death, which reduces inflammation, because scavenger macrophages phagocytose apoptotic cells.

Our paper discussed a recent study from Froelich and colleagues²  that asserts that native GzmA from human killer cells is not cytotoxic at nanomolar concentrations, contradicting our results using recombinant GzmA (rGzmA). There is considerable evidence that native GzmA and GzmB both independently activate cell death.³  Mice genetically deficient in either GzmA or GzmB are essentially immunocompetent, whereas mice lacking both Gzms are profoundly immunodeficient. Moreover, as shown in Figure 7 of our paper,¹  cytotoxic T lymphocytes (CTLs) lacking GzmA or GzmB induce comparable cytotoxicity.

The Froelich study also could not verify previous work that native GzmA cleaves and activates pro–IL-1β.⁴  However, they showed that native human GzmA was proinflammatory, because it induced monocytes to secrete IL-1β. Because the GzmA concentration that activates cytokine secretion was approximately 2 to 3 logs less than the cytotoxic concentration, they speculated that GzmA's primary function is proinflammatory, rather than cytotoxic. However, because Gzms are secreted into a small enclosed space (the immunologic synapse) and only small amounts leak out into extracellular fluids, the local synapse concentration is relatively much higher. During chronic inflammation (rheumatoid arthritis joints, AIDS patient blood), GzmA can reach low-nanomolar extracellular levels (vs < 1pM in normal blood).⁵  Based on the GzmA yield from killer cells (∼ 20 μg/10⁹ cells)⁶  and conservative estimates that approximately one-tenth of CTL granule contents are released into a single synapse with a volume of less than 5 μm³, we estimate that GzmA synapse concentrations are vastly greater (∼ 2 × 10⁻⁹ μg/5 μm³ or ∼ 8μM). Therefore, the local synapse concentration is more than adequate for inducing cytotoxicity, activated at 250nM concentrations (Figure 1).^1,7-9 

In the absence of a defined mechanism, we are uncertain whether GzmA has additional proinflammatory effects, besides activating pro–IL-1β⁴  (which we recently confirmed using rGzmA). Monocytes/macrophages are exquisitely endotoxin-sensitive. It is impossible to verify endotoxin elimination from GzmA preparations. This is an issue for both recombinant and purified enzymes, considering that buffers, water, and glassware can be contaminated. Commercial kits do not completely remove endotoxin. Moreover, the Limulus clot assay for endotoxin, which measures tryptase activity, cannot be performed on GzmA preparations. Therefore, because there may be residual endotoxin in all GzmA preparations, it will be important to define the molecular basis for additional proinflammatory mechanisms of GzmA.

The Froelich group recently kindly provided us with the native natural killer (NK) cell GzmA that they used. We compared the cytotoxicity of human GzmA from 4 sources⁶ : rGzmA,¹  native NK92 cell GzmA purified by us and Froelich,²  and native lymphokine-activated killer cell GzmA purified by us. rGzmA induced significant cytotoxicity at approximately 250nM concentrations (consistent with our previous publications^1,7-9 ), whereas native GzmA had limited cytotoxic activity (consistent with Froelich's published study² ) (Figure 1A). rGzmA was also substantially more active at cleaving SET, the inhibitor of GzmA-activated DNA damage⁷  (Figure 1B). Moreover, rGzmA and native GzmB from YT-Indy NK cells⁶  were active at similar concentrations (∼ 250nM; Figure 1C). Therefore native GzmA from human killer cell granules is less cytolytic than rGzmA.

Further work is needed to understand why native GzmA has reduced cytotoxicity. GzmA is synthesized as an inactive pro-enzyme. One potential explanation could be that GzmA might be stored mostly as the pro-enzyme, which is activated only during CTL degranulation.