Redox modification of nuclear actin by MICAL-2 regulates SRF signaling

Abstract

The serum response factor (SRF) binds to coactivators, such as myocardin-related transcription factor-A (MRTF-A), and mediates gene transcription elicited by diverse signaling pathways. SRF/MRTF-A-dependent gene transcription is activated when nuclear MRTF-A levels increase, enabling the formation of transcriptionally active SRF/MRTF-A complexes. The level of nuclear MRTF-A is regulated by nuclear G-actin, which binds to MRTF-A and promotes its nuclear export. However, pathways that regulate nuclear actin levels are poorly understood. Here, we show that MICAL-2, an atypical actin-regulatory protein, mediates SRF/MRTF-A-dependent gene transcription elicited by nerve growth factor and serum. MICAL-2 induces redox-dependent depolymerization of nuclear actin, which decreases nuclear G-actin and increases MRTF-A in the nucleus. Furthermore, we show that MICAL-2 is a target of CCG-1423, a small molecule inhibitor of SRF/MRTF-A-dependent transcription that exhibits efficacy in various preclinical disease models. These data identify redox modification of nuclear actin as a regulatory switch that mediates SRF/MRTF-A-dependent gene transcription.

Figure 1. MICAL-2 is a nuclear protein that depolymerizes actin. See also Figure S1

(A) Cherry-MICAL-2 and -3 expressed in HEK293T cells are localized to the nucleus (blue), while Cherry-MICAL-1 and NLS-mutant Cherry-MICALs (M2NLSMut and M3NLSMut) exhibited cytosolic localization. Scale bar, 10 μm. (B) Subcellular fractionation of MICAL isoforms. Endogenous MICAL-2 and -3 are localized to the nucleus while MICAL-1 was detected in the cytosol. MEK1/2, VDAC, Calnexin, and PCNA immunoblotting was used to confirm the purity of the cytosolic (Cyto), mitochondrial (Mito), microsomal (Micro), and nuclear (Nuc) fractions, respectively. (C) MICAL-2 and -3 each contain a putative bipartite NLS. The domain structure of MICAL-2 is indicated above. The enzymatic domain contains a GXGXXG, aspartate-glycine (DG), and a glycine-aspartate (GD) motif, which is the characteristic FAD-binding motif. The F-actin-binding CH domain and LIM domain are indicated. A potential bipartite NLS (red box) in MICAL-2 and MICAL-3 is not found in MICAL-1. Shown in blue letters are the amino acids that constitute the NLS in MICAL-2 and -3. Shown in red are the amino acids that are mutated in the M2NLSMut and M3NLSMut. Uppercase letters are amino acids conserved in MICAL-1, -2 and -3. (D) MICAL-2 and -3 depolymerize F-actin similarly to MICAL-1. MICAL-1^redoxCH, MICAL-2^redoxCH and MICAL-3^redoxCH were incubated with pyrene-labeled actin and NADPH as indicated. These data indicate that MICAL-2 and MICAL-3, like MICAL-1, induce NADPH-dependent F-actin depolymerization. Statistical significance was determined by one-way analysis of variance (ANOVA) (***p < 0.0007) with Dunnett multiple comparison post-test. ***p < 0.0005, n ≥ 12. Mean ± SEM. (E) Dominant-negative inhibition of MICAL-2 lead to increased nuclear F-actin. Expression of dominant-negative MICAL2 (MICAL-2CT or MICAL-2GV, green) resulted in the appearance of filaments (yellow arrows) throughout the nucleus (blue), as seen by phalloidin labeling (red staining, blue arrows). The dominant-negative constructs colocalized with the F-actin (yellow). Scale bar, 5 μm.

Figure 2. MICAL-2 induces the nuclear localization of MRTF-A. See also Figure S2

(A) MICAL-2 expression (green) induces nuclear localization of MRTF-A (red) under serum-starvation. Nuclei are outlined in white, based on DAPI staining (blue). Scale bar, 5 μm. (B) Quantification of (A). The average intensity of MRTF-A immunofluorescence was quantified in the nucleus and the cytoplasm. The nucleus:cytoplasmic ratio of MRTF-A increased 10.76-fold in GFP-MICAL-2-expressing cells compared to GFP-expressing cells. ***p < 0.0005, Student’s t-test, n ≥ 30. (C) MICAL-2 is required for NGF-induced nuclear localization of MRTF-A in PC12 cells. NGF treatment induces nuclear localization of MRTF-A in LacZ shRNA-expressing PC12 cells. PC12 cells expressing MICAL-2-specific shRNA (yellow arrows) showed reduced NGF-induced nuclear localization of MRTF-A compared to uninfected cells (blue arrows). Nuclear border indicated by dotted white lines. Scale bar, 10 μm. (D) Quantification of (C). ***p < 0.0005, ANOVA (***p < 0.0001) with Dunnett post-test n ≥ 30. All data in this figure is mean ± SEM. (E) The nuclear localization of MRTF-A in dissociated E14 DRG neurons cultured in the presence of NGF is dependent upon MICAL-2. Infection of DRG neurons with MICAL-2-specific shRNA results in a substantial nuclear depletion of MRTF-A (red). Scale bar, 5 μm. (F) Quantification of (E). ANOVA (***p < 0.0005) with Dunnett post-test. n ≥ 25.

Figure 3. MICAL-2 induces SRF/MRTF-A-mediated luciferase expression. See also Figure S3

(A) MICAL-2 induces the SRF/MRTF-A transcriptional reporter in serum-starved (0.3% FBS)-treated cells. Catalytically inactive MICAL-2 mutants (MICAL-2GV and MICAL-2CT), as well as a MICAL-2 mutant that does not localize to the nucleus (M2NLSMUT) does not induce the reporter. ANOVA (**p < 0.005) with Dunnett post-test. n = 18. All data in this figure is mean ± SEM. (B) MICAL-2 does not activate the MRTF-A-independent SRF/TCF luciferase promoter. ANOVA (***p < 0.0001) with Dunnett post-test. n = 18. (C) MICAL-2, but not MICAL-1 or MICAL-3, activates the SRF/MRTF-A luciferase reporter. ANOVA (***p < 0.0005) with Dunnett post-test. n = 18. (D) MICAL-2-dependent induction of the SRF/MRTF-A reporter requires MRTF-A. Coexpression of dominant-negative MRTF-AΔTAD blocked MICAL-2-dependent induction of luciferase. ANOVA (***p < 0.0001) with Dunnett post-test. n = 18.

Figure 4. MICAL-2 is required for NGF-dependent neurite outgrowth in PC12 cells and DRG neurons. See also Figure S4 and Table S1

(A) MICAL-2 is required for NGF-induced neurite outgrowth in PC12 cells. NGF treatment (50 ng/mL) for 48 h resulted in prominent neurite outgrowth in LacZ shRNA (green) control as measured by Alex568-phalloidin staining (red). Knockdown of MICAL-2 with either of two shRNA significantly reduces neurite outgrowth. Scale bar, 10 μm. (B) Quantification of neurite length in (A). The average length of all neurite extensions were quantified in LacZ shRNA- and MICAL-2 shRNA-expressing PC12 cells. ANOVA (***p < 0.0001) with Dunnett post-test. *p < 0.05, n ≥ 40. All data in this figure is mean ± SEM. (C) Quantification of neurite number in (A). Knockdown of MICAL-2 significantly reduced the average number of neurites induced by NGF treatment. ANOVA (***p < 0.0001) with Dunnett post-test. ***p < 0.0005, **p < 0.005, n ≥ 40. (D) MICAL-2 knockdown decreases axon growth rates in DRG neurons. DRG neurons expressing either LacZ-specific or MICAL-2-specific shRNA were imaged at 0 (blue arrow) and 60 min (yellow arrow). Scale bar, 10 μm. (E) Quantification of rates in D. ANOVA (***p < 0.0004) with Dunnett post-test. ***p < 0.0005, **p < 0.005, n ≥ 20.

Figure 5. MICAL-2 regulates SRF/MRTF-A-dependent gene transcription during zebrafish development. See also Figure S5 and Table S2

(A,B) Representative control embryo derived from transgenic myl7:egfp reporter fish at 24 hpf, showing the size and position of a normal looping heart tube. By 48 hpf the control heart is fully looped. The myl7:egfp transgene directs eGFP expression to myocardial cells and facilitates detection of alterations in heart morphology. (C,D) mical2b regulates cardiogenesis during zebrafish development. The mical2b splice-blocking morphant embryo at 24 hpf shows a smaller heart tube that fails to loop normally and thus is positionally displaced and leads to a significant pericardial edema (PE). By 48 hpf the morphant heart tube is linear and dysmorphic. Similar results were obtained with the translation blocking morpholino. For A–D n>100. (E) Higher magnification view of the normal control heart at 52 hfp highlights the evenly spaced cardiomyocytes (white arrows) focused on the atrial chamber. (F–G) In contrast, cells in the smaller morphant atrium are irregularly spaced (white arrows). The representative embryonic hearts shown in F and G are for embryos injected with 2 or 4 ng of the splice-blocking morpholino, respectively (n>50). (H) mical2b knockdown impairs SRF/MRTF-A-dependent but not SRF/MRTF-A-independent gene transcription. Levels are compared to control wild-type (WT) embryos for the mical2b morphants (2B), and in embryos rescued by co-injected RNA encoding the full-length wild-type murine MICAL-2 or the MICAL-2GV mutant protein. Experimental gene levels were normalized by comparison to housekeeping gene EF1α. ANOVA (***p < 0.0005) with Dunnett post-test. ***p < 0.0005, **p < 0.005, **p < 0.05, n ≥ 6. Mean ± SEM.

Figure 6. MICAL-2 regulates nuclear actin independent of RhoA. See also Figure S6

(A) MICAL-2 induces the SRF/MRTF-A luciferase reporter in a ROCK- and RhoA-independent manner. Treatment with either inhibitor reduced luciferase expression induced by serum-stimulation but not by MICAL-2 expression. ANOVA (***p < 0.0001) with Dunnett post-test. n =8. All data in this figure is mean ± SEM. (B) MICAL-2 induces nuclear MRTF-A in a ROCK- and RhoA-independent manner. HEK293T cells were infected with either GFP or GFP-MICAL-2. Nuclear accumulation of MRTF-A was unaffected in MICAL-2 expressing, serum-starved HEK293T cells treated with either 2 μg/mL C3-transferase, a RhoA inhibitor, or 100 μM Y27632, a ROCK inhibitor. ANOVA (***p < 0.0001) with Dunnett post-test. (C) MICAL-2 expression decreases the nuclear to cytosolic ratio of G-actin in HEK293T cells. In starved GFP-expressing cells, G-actin is readily detectable in the nucleus (blue), as measured by DNAse I staining (red). Serum stimulation and GFP-MICAL-2 significantly reduced the amount of G-actin in the nucleus. Scale bar, 5 μm. (D) Quantification of the nuclear:cytosolic DNAse I staining seen in C. ANOVA (***p < 0.0001) with Dunnett post-test. ***p < 0.0005, n ≥ 30. (E) MICAL-2 expression reduces nuclear levels of GFP-Actin but not GFP-actin M44L. When overexpressed, GFP-actin M44L is enriched in the nucleus (blue) when compared to wild-type GFP-Actin. MICAL-2 (red) coexpression further decreases the levels of nuclear wild-type GFP-actin (green) while having no effect on the nuclear levels of GFP-actin M44L. Scale bar, 5 μm. (F) Quantification of the nuclear:cytosolic GFP-actin staining seen in G. ANOVA (***p < 0.0001) with Dunnett post-test. ***p < 0.0005, n ≥ 25.

Figure 7. MICAL-2 is targeted by CCG-1423. See also Figure S7

(A) CCG-1423 inhibits MICAL-2-induced activation of the SRF/MRTF-A transcriptional reporter by both serum and MICAL-2 expression. ANOVA (***p < 0.0001) with Dunnett post-test. ***p < 0.0005, n = 24. All data in this figure is mean ± SEM. (B) CCG-1423 exhibits concentration-dependent thermal destabilization of MICAL2-EN. Recombinant MICAL2-EN, which comprises the enzymatic domain of MICAL-2, was incubated with increasing concentrations of CCG-1423, or the control compound, CCG-100594, and the T_m was calculated in a thermal denaturation assay. Incubation of CCG-1423 exhibited thermal destabilization of MICAL-2 with an IC₅₀ of 3.8 μM and Hill coefficient -1.1. ANOVA, Bonferroni’s multiple comparisons test *p < 0.05 10 μM 1423 vs DMSO; **p < 0.01 25 μM 1423 vs DMSO; ###, p < 0.001 25 μM 1423 vs 25 μM 100594. 10 μM at n=12 and 25 μM at n=7. (C) CCG-1423 inhibits MICAL-2 enzymatic activity. We monitored MICAL-2 activity using an NADPH consumption assay. MICAL-2 was incubated with either 5 μM CCG-1423 or the inactive control compound CCG-100594 in the presence of 2 μM F-actin and 10 μM NADPH. ***p < 0.0005, Student’s t-test, n = 9. (D) Enzyme kinetics of MICAL-2. Velocity of recombinant MICAL-2 activity as a function of NADPH and CCG-1423. MICAL-2^redoxCH was incubated with DMSO, 0.5 μM or 5 μM CCG-1423. CCG-1423 inhibits MICAL-2 with a K_i of 1.57 μM.