RASSF1A elicits apoptosis through an MST2 pathway directing proapoptotic transcription by the p73 tumor suppressor protein

Abstract

RASSF1A is a tumor suppressor gene that is epigenetically silenced in a wide variety of sporadic human malignancies. Expression of alternative RASSF1 isoforms cannot substitute for RASSF1A-promoted cell-cycle arrest and apoptosis. Apoptosis can be driven by either activating Bax or by activation of MST kinases. The Raf1 proto-oncogene binds to MST2, preventing its activation and proapoptotic signaling. Here we show that key steps in RASSF1A-induced apoptosis are the disruption of the inhibitory Raf1-MST2 complex by RASSF1A and the concomitant enhancement of MST2 interaction with its substrate, LATS1. Subsequently, RASSF1A-activated LATS1 phosphorylates and releases the transcriptional regulator YAP1, allowing YAP1 to translocate to the nucleus and associate with p73, resulting in transcription of the proapoptotic target gene puma. Our results describe an MST2-dependent effector pathway for RASSF1A proapoptotic signaling and indicate that silencing of RASSF1A in tumors removes a proapoptotic signal emanating from p73.

Figure 1. RASSF1A Disrupts MST2 Interaction with Raf1

(A) MCF7 cells were transfected with 1 μg HA-RASSF1A or empty vector. After 16 hr serum starvation (0.1% serum), cells were lysed. MST2 immunoprecipitates were subjected to an in-gel kinase assay and western blotted with the indicated antibodies. (B) Endogenous MST2 and RASSF1A immunoprecipitates from serum-starved and anti-Fas CD95-treated HeLa cells were analyzed by western blotting with the indicated antibodies. (C) MCF7 cells were transfected with increasing amounts of HA-RASSF1A or HA-RASSF1C. After serum starvation, MST2 immunoprecipitates were western blotted for coprecipitating Raf1. (D) MCF7 cells were transfected with 50 ng/ml Raf1 or a nontargeting siRNA pool (control) and 1 μg of HA-RASSF1A or HA-RASSF1C. MST2 immunoprecipitates were examined for associated Raf1 by western blotting.

Figure 2. MST2 Is Required for Fas- and RASSF1A-Induced Apoptosis

(A) Serum-starved MCF7 cells were transfected with increasing amounts of HA-RASSF1A or HA-RASSF1C plasmids. After 20 hr of serum starvation, apoptosis was measured by assaying DNA fragmentation by FACS. Error bars show standard deviation. Aliquots of cell extracts were examined for the expression of transfected genes by using tubulin as loading control. (B) MCF7 cells were transfected with 0.5 μg of HA-RASSF1A and 50 ng/ml of MST2 siRNAi (left panel) or 50 ng/ml of Raf1 siRNA (right panel) or control siRNA and analyzed as in Figure 1. (C and D) MCF7 cells transfected with 0.5 μg of HA-RASSF1A and 50 ng/ml siRNAs for MST2, Raf1, or control were starved for 16 hr and treated with anti-Fas CD95 (50 nM) plus cyclohexamide (5 μg/ml) for 4 hr and assayed as above. Error bars show standard deviation. (E) Raf-1 is required for protection from Fas-induced hepatitis in adult mice. PCR analysis of genomic DNA isolated from the liver of c-raf-1^flox/flox and c-raf-1^Δ/Δ;MxCre⁺ littermates after induction of MxCre in vivo, showing successful conversion of the f to the Δ allele (upper panel). This also resulted in the elimination of Raf1 protein as shown by immunoblot analysis of liver lysates. Fas expression was not affected by the Raf1 knockout, and an ERK2 immunoblot is shown as a loading control. (F) c-raf-1^Δ/Δ mice are sensitized to challenge by an anti-Fas Abs (Jo-2, 200 μg/kg). Survival was monitored for 48 hr. Animals that did not die within the first 8 hr recovered. (G) Caspase 3 activation in the liver of c-raf-1^Δ/Δ animals was determined by immunoblotting whole-cell lysates with antisera recognizing the cleaved, active forms of caspase-3. An ERK immunoblot is shown as a loading control. The livers were isolated either before or after (2 hr) i.v. injection with anti-Fas.

Figure 3. MST2-LATS1 Association Is Induced by RASSF1A and Required for Fas Proapoptotic Signaling

(A) MCF7 cells were transfected with 0.5 μg of HA-RASSF1A and HA-RASSF1C constructs. Endogenous LATS1 immunoprecipitates from serum-starved cells were blotted with LATS1 and MST2 antibodies. (B) MCF7 cells were transfected with 0.1, 0.5, and 1 μg of HA-RASSF1A. Endogenous LATS1 immunoprecipitates from serum-starved cells were blotted with LATS1, MST2, and HA antibodies. (C) MCF7 cells were transfected with 0.5 μg HA-RASSF1A and HeLa cells with 50 ng/ml RASSF1A siRNA. After 16 hr serum starvation, cells were incubated with 50 nM CD95 antibody to Fas and cyclohexamide (5 μg/ml) for 4 hr where indicated and endogenous LATS1 immunoprecipitates were western blotted with LATS1 and MST2 antibodies. (D) HeLa cells were transfected with 50 ng/ml RASSF1A, MST2, LATS1, or control siRNAs, serum starved, and treated with anti-Fas CD95 (50 nM) plus cyclohexamide (5 μg/ml) for 20 hr. Apoptosis was determined as in Figure 2. Error bars show standard deviation. Cell lysates were western blotted with the indicated antibodies to assess the efficiency of knockdown. Endogenous expression levels of tubulin were used as loading control.

Figure 4. RASSF1A Regulates Apoptosis via LATS1 and YAP1

(A) Endogenous LATS1 immunoprecipitates from MCF7 and HeLa cells were examined for associated endogenous YAP1 by western blotting. (B) MCF7 cells were transfected with empty vector, 0.5 or 1 μg of HA-RASSF1A plasmid (top panels), or 1 μg of HA-RASSF1C (bottom panels). After 16 hr serum starvation, cell extracts were split in two aliquots, and endogenous YAP1 or LATS1 was immunoprecipitated and examined for association by immunoblotting with the indicated antibodies. (C) MCF7 cells were transfected with 1 μg of HA-RASSF1A and 50 ng of MST2 or control siRNAs and examined for association of YAP1 and LATS1 as above. (D) MCF7 cells stably expressing shRNA against YAP1 or control (Basu et al., 2003) were transfected with 0.25, 0.5, 1, and 1.5 μg of HA-RASSF1A. Apoptosis was determined by assessing caspase activation by zVAD-fmk-FITC after serum starvation for 16 hr and anti-Fas CD95 (50 nM) plus cyclohexamide (5 μg/ml) treatment. Error bars show standard deviation. Tubulin and exogenous HA-RASSF1A levels were determined by western blotting.

Figure 5. RASSF1A Induces Subcellular Relocalization of YAP1 and p73, Phosphorylation of YAP1 by LATS1, and Formation of a YAP1-p73 Complex

(A) MCF7 cells were transfected with control, 1 μg of HA-RASSF1A, or 1 μg of HA-RASSF1C plasmids. Nuclear and cytoplasmic fractions were examined for the presence of p73 and YAP1. Lamin A and HSP70 were monitored as markers for the purity of the nuclear and cytoplasmic fractions, respectively. (B) MCF7 cells were transfected with control, 1 μg of HA-RASSF1A, or 1 μg of HA-RASSF1C plasmids. After 16 hr serum starvation, p73 immunoprecipitates were assessed for the copurification of YAP1 by western blotting. (C) Endogenous LATS1 immunoprecipitates prepared from RASSF1A-transfected or control MCF7 cells were incubated with purified bacterial GST-YAP1 in the presence of ³²P-γ-ATP, showing that RASSF1A stimulates YAP1 phosphorylation by LATS1. (D) A kinase-dead myc-tagged LATS-1 mutant (myc-LATS1-KD) did not phosphorylate YAP1, showing that the phosphorylation is due to LATS1. (E) Recombinant GST-YAP1 was phosphorylated by LATS1 in vitro. Association of phosphorylated (pYAP) and nonphosphorylated YAP with in vitro-translated HA-p73α was determined by coimmunoprecipitation and western blotting for GST in HA immunoprecipitates.

Figure 6. A YAP1-p73 Complex Induces PUMA Expression and Mediates RASSF1A-Induced Apoptosis

(A) MCF7 cells were transfected with control, 1 μg of HA-RASSF1A, or HA-RASSF1C plasmids and serum starved for 16 hr. Extracts were immunoblotted for endogenous PUMA levels, and tubulin was used as loading control. (B) MCF7 cells were transfected with 1 μg of HA-RASSF1A and 70 ng/ml siRNA against YAP1 or control in the indicated combinations. Cell extracts were immunoblotted for endogenous protein levels of PUMA, Bax, YAP1, and tubulin (as loading control). Transfected HA-RASSF1A was detected with anti-HA antibody. (C) MCF7 cells were transfected as in (B) and examined for apoptosis by using the zVAD-fmk-FITC method. Endogenous YAP1, tubulin, and exogenous HA-RASSF1A levels were determined by western blotting. (D) MCF7 cells were transfected with 1 μg of HA-RASSF1A and 50 ng/ml p73 siRNA in the combinations indicated. Apoptosis was determined as above. Endogenous p73, tubulin, and exogenous HA-RASSF1A levels were determined by western blotting. (E) Saos2 cells with doxycycline-inducible HA-p73α were transfected with Flag-RASSF1A or empty vector. Apoptosis was determined by FACS analysis of zVAD-fmk-FITC in the presence or absence of HA-p73α induction (doxycycline for 24 hr). In (C)–(E), error bars indicate standard deviation.

Figure 7. Model for RASSF1A Activation of puma Expression

Fas activation releases MST2 from the inhibitory complex with Raf-1 and induces binding to RASSF1A, leading to MST2 activation. RASSF1A further promotes the interaction of MST2 with its substrate, LATS1. This in turn induces the phosphorylation of YAP1 by LATS1 and dissociation. YAP1 phosphorylation increases the affinity for p73, permitting YAP1 to form a nuclear complex with p73, which induces the transcription of the proapoptotic puma gene.