The role of caspase 3 in cell death-induced tumor cell proliferation in vitro and in vivo.
(a) The effect of caspase 3 deficiency on dying cells stimulated tumor growth in vitro. In all three groups, the difference between wild type and casp3^−/− MEF cells are statistically significant (P<0.01, t-test). Error bars: SEM, n=3.
(b) The effect of caspase 3 knockdown(Casp3-kn) in 4T1 cells used a feeder cells. The difference between the control groups and the casp3kn groups are statistically significant (P<0.01 between each of the control and each of the caspase 3 knock down clones, n=3, t-test). Error bars: SEM, n=3. Lower panel shows western analysis of caspase 3 levels.
(c) The effect of a dominant negative caspase 3 in dying, unlabeled 4T1 cells on 4T1Fluc tumor cell growth. Inset, western blot showing dominant negative caspase 3 expression. (d) Western blot analyses of key proteins involved in apoptosis.
(e) The effect of lethally irradiated wild type and Casp3^−/− MEF cells on growth 4T1Fluc cells in mice. Error bars: SEM, n=4. The difference between two groups were highly significant statistically (P<0.001 from day 3 on, one-way ANOVA test). Lower panel shows representative bioluminescent images.
(f) The effect of caspase 3 knockdown in lethally irradiated 4T1 cells on growth of 4T1Fluc cells in vivo. The difference between wild type 4T1 and 4T1-siCasp3 groups is statistically significant from day 5 (P<0.05, n=5, one-way ANOVA test).

An important role for caspase 3-activated iPLA₂ in facilitating cell death stimulated tumor cell repopulation.
(a) The effect of iPla₂ levels in dying cells on the growth of 4T1Fluc cells in vitro. The differences between wild type feeder cells (MEF or 4T1) and those with iPla₂ knockdown are significant, so is the difference between Casp3^−/− MEF cells and those transduced with a truncated, constitutively active iPla₂ (P<0.01, n=4, t-test).
b). The effect of a constitutively active iPLA₂ in Casp3^−/− MEF cells in vivo. The differences between the luciferase signals were statistically significant (P<0.01 on days 7, n=5, t-test). Error bars: SEM. The inset shows expression of truncated iPla₂.
c). The in vivo effect of iPLA₂ knockdown in wild type MEF cells. The difference between the two groups was statistically significant from day 3 (P<0.02, from day 7, n=5, one-way ANOVA). The inset shows western blot analysis of shRNA knockdown of iPla₂.

Deficiency in caspase 3 correlated with tumor response to therapy in mice as well as in human patients.
(a) Result of radiation therapy in tumors established from wild type caspase 3-deficient MCF-7 breast cancer cells and MFC-7CASP3 with a exogenous copy of caspase 3 Error bars: SEM. The differences between the two groups are highly significant after radiotherapy (P<10⁻⁶, n=10, one – way ANOVA).
(b) Results of radiation therapy B16F10 murine melanoma tumors grown in wild type C57BL6 and Casp3^−/− mice. Error bars: SEM. The difference between the two groups are significant after radiotherapy from day 11 (P<0.04, n=5, one-way ANOVA).
(c) Kaplan-Meier analysis of cancer recurrence in a cohort of head and neck squamous cell carcinoma (HNSCC) patients treated at Princess Margaret Hospital in Toronto, Canada. Log-rank test (p-value= 0.0114, HR=3.44, 95%CI: 1.35-8.75) was used for analysis of statistical significance.
(d). Kaplan-Meier analysis of survival in a cohort of advanced breast cancer patients treated at Shanghai No. 1 People’s Hospital. Log-rank test (Log-rank p-value = 0.0006, HR=5.29, 95%CI: 1.70-16.46) was used to analyze statistical significance of the difference in survival between the two groups.
(e) A schematic representation of the “Phoenix Rising” pathway for cell death-mediated tumor cell repopulation.

In vitro and in vivo evidence for the generation of strong growth-stimulating signals in dying cells.
(a) Stimulation of 4T1Fluc cellular growth in vitro by irradiated 4T1 cells. Top panel: growth of 4T1Fluc cells as observed by luciferase activities. The difference between each of the higher dose irradiated groups (8, 10, and 12 Gys) and controls (0 Gy and no feeder) is statistically significant (P<0.001, t-test). Error bars: SEM, n=4. Middle panel: representative images from bioluminescence imaging. Lower panel: selected photographs of cellular growth after crystal violet staining.
(b) Stimulation of Fluc-labeled cellular growth in vitro by irradiated mouse embryonic fibroblast (MEF) cells. Top panel, relative growth of MEF-supported tumor cells vs tumor cells seeded alone. Error bars: SEM, n=3. Lower panel, representative bioluminescence images. In all three cases, the difference in cellular growth between those with and those without MEF feeders are statistically significant (P<0.001, t test).
(c) Effect of tumor cell death on 4T1Fluc tumor cellular growth in vivo. Top panel: growth of 4T1Fluc cells as followed by luminescence signals. The difference between the two groups were highly significant (P<0.001 from day 4, one way ANOVA test). Error bars: SEM, n=5. Lower panel, representative images of mice at early and late stages of tumor growth.
(d) Effect of dying MEF cells on 4T1Fluc tumor cellular growth in vivo. Top panel, growth of 4T1Fluc cells. The difference between the two groups were again highly significant (P<0.001 from day 1, one way ANOVA test). Error bars: SEM, n=5. Lower panel, representative images of mice at early and late stages of tumor growth.

Relationship between caspase activation and growth of externally injected tumor cells in the irradiated tumor microenvironment.
(a) Caspase 3 activation in 4T1 tumors as detected by a caspase 3 reporter. Left panels depict the structure of a proteasome-based caspase 3 reporter (top left) and its principle of action (lower left). Right panels showed caspase 3 activities in 4T1 tumors transduced with the control as well as caspase reporter genes. The difference between the control and caspase 3 reporter groups are significant at days 3, 5, and 7 (P<0.01, n=5, t-test). Error bars: SEM.
(b) Growth of the 4T1Fluc cells injected into irradiated and non-irradiated established tumors. The difference between the two groups were statistically significant (P<0.05 from day 7, t test). Error bars: SEM, n=4. Lower panel shows representative images of tumor-bearing mice at early and late stages of observation.
(c) Imnnofluorescence analysis of growth of intratumorally injected GFP-labeled cells and key protein expression surrounding the injected cells. SMA, smooth muscle actin, a marker for blood vessels. The size bars represent 100μm.

(a) The role of caspase 3 in radiation-induced arachidonic acid release. Error bars: SEM (n=3). *P<0.02 (t-test).
(b) Regulation of radiation-induced PGE₂ secretion by caspase 3. Error bars: SEM (n=3). *P<0.05 (t-test).
(c) PGE₂ stimulated tumor growth from 1000 4T1Fluc tumor cells injected subcutaneously into nude mice. Error bars: SEM. The difference between the two groups are statistical significant from day 17 (P<0.05, n=5, one-way ANOVA).