Reply to 'Small molecules not direct activators of caspases'

Paul J Hergenrother, Karson S Putt & Joseph S Sandhorst reply

In our paper¹ we presented several lines of evidence suggesting that PAC-1–induced cell death occurs through the direct activation of procaspase-3. For example, the order and timing of apoptotic events is dramatically different in PAC-1–treated cells as compared to cells treated with a known proapoptotic agent: Caspase-3- and caspase-7-like activity can be detected within 1 h of PAC-1 treatment of HL-60 cells, with a corresponding reduction of poly(ADP-ribose)polymerase (PARP) activity, whereas depolarization of the mitochondrial membrane does not appreciably occur until after 4 h; in contrast, the timing is reversed in etoposide-treated cells. The strong correlation of procaspase-3 levels in primary tumor samples with the toxicity of PAC-1 further suggested a direct activation of procaspase-3. Additionally, in new experiments, we have synthesized a large series of PAC-1 derivatives and evaluated them for their ability to activate procaspase-3 in vitro and to induce death in HL-60 cells. We observed a strong trend whereby those compounds that are able to activate procaspase-3 in vitro also induce death in the cancer cells in culture, and those compounds that do not activate procaspase-3 in vitro have no effect on the cancer cells (unpublished data).

In our experience, there is no question that PAC-1 activates procaspase-3 in vitro: data from recent, independent experiments show clear and unambiguous activation of procaspase-3 treated with PAC-1 as compared with vehicle-treated procaspase-3 (see Fig. 1a for representative curves). To further investigate the action of PAC-1, we have now extended our original study to investigate the parameters for in vitro activation of procaspase-3 by PAC-1. One of the major conclusions of this work is that the effect of PAC-1 on procaspase-3 activity in vitro is proportional to procaspase-3 concentration (unpublished data). With this in mind, we note that the experiments carried out by Denault et al. do not replicate our conditions: they used both different buffers and different procaspase-3 concentrations, which our new data indicate could significantly affect the activation (or lack thereof) that they observe. Similarly, it is difficult for us to comment on the data in Figure 1c provided by Denault et al., as we have not performed this experiment. However, given the significantly reduced concentration of procaspase-3 in the lysate as compared to the intact cell, these conditions could negatively affect the ability of PAC-1 to activate procaspase-3.

Figure 1: In vitro activation of procaspase-3 by PAC-1.

(a) Progress curves showing the effect of PAC-1 on procaspase-3. 45 μl of procaspase-3 (1 μM) in 50 mM Tris, pH 8.0, 300 mM NaCl, 500 mM imidazole was incubated with 100 μM PAC-1 or vehicle in a 384-well plate at 37 °C. After 2 h, 5 μl of Ac-DEVD-pNA in 50 mM HEPES pH 7.4, 100 mM NaCl, 0.10% CHAPS, 10 mM DTT, 0.1 mM EDTA, 10% glycerol was added to each well and the absorbance monitored at 405 nm every 30 s. The final concentration of Ac-DEVD-pNA is 200 μm. Data shown are raw data from an experiment performed in triplicate. (b) PAC-1–promoted autoprocessing of procaspase-3 as monitored by western blotting. A solution of procaspase-3 (30 μM, with a C-terminal His6 tag) in 50 mM Tris, pH 8.0, 300 mM NaCl, 500 mM imidazole was incubated in the presence of 50 μM PAC-1 or vehicle at 37 °C. After 12 h, 2× SDS sample buffer was added (50 mM Tris, pH 6.8, 100 mM DTT, 2% (w/v) SDS, 0.1% bromophenol blue, 10% (v/v) glycerol) and the samples heated to 95 °C. Bands were detected via western blotting for the His6 epitope.

Denault et al. assert that there is a discrepancy between the 3–4.5-fold activation of our published in vitro activity assays and the ∼50% processing observed by western blotting, but unfortunately, they seem to be misinterpreting our data. It is well documented that procaspase-3 is cleaved in vitro at three different sites: after Asp9, Asp28 and Asp175 (ref. 2). The processing of procaspase-3 at these sites can be performed by caspase-3 (ref. 3), and it is known that wild-type procaspase-3 will also autoactivate by proteolysis at these sites^4,5. Thus, analysis of procaspase-3 activation by SDS-PAGE, or western blotting with a general caspase-3 antibody, will reveal a 'triplet' of bands clustered around ∼32 kDa, where the top band is procaspase-3, the middle band is procaspase-3 minus the first 9 residues, and the bottom band is procaspase-3 minus the first 28 residues. This triplet can be seen in several papers in the literature, most obviously in ref. 3. The western blot data in Figure 2c of our article¹ were obtained with an antibody to the His6 tag, which is attached to the N-terminus of the protein. Thus, we see the procaspase-3 band disappear and a smaller fragment appear. When we stated in the paper that we saw 50% processing, this was 50% processing of the top (full-length procaspase-3) band. Procaspase-3 is not yet fully activated in these experiments, as the two other bands of the triplet are present but not visible, having lost the N-terminal His6 tag.

To make this point unambiguously, we have now treated procaspase-3 with a His6 tag on its C-terminus with PAC-1. In these experiments, we once again saw clear activation of procaspase-3 by PAC-1. The western blot shows the banding pattern observed after procaspase-3 is treated with 50 μM PAC-1, directly compared to vehicle-treated procaspase-3 (Fig. 1b). The expected triplet banding pattern is evident in these experiments, and there is a considerable difference between the two samples.

Our own recent experiments have indicated that under conditions of low cell density and high compound concentration, PAC-1 will indeed induce death in MCF-7 cells. However, further examination has shown that this death is very different from that induced in cancer cell lines that contain procaspase-3, and appears to be more consistent with necrotic, not apoptotic, cell death (unpublished data).

Determining the precise mode of action of a small molecule is always a difficult task; however, all of the data we have obtained thus far are consistent with our proposed mechanism. We do not discount the notion that PAC-1 could have an additional or alternative cellular target that accounts for its powerful proapoptotic effect, and we are in the midst of several experiments designed to either bolster or disprove our hypothesis.

Finally, we note that we have sent frozen procaspase-3 to other researchers, and they have also observed activation of procaspase-3 with PAC-1. We are happy to provide this or any of our reagents to interested parties.