Report

Nuclear Actin Regulates Dynamic Subcellular Localization and Activity of the SRF Cofactor MAL

See allHide authors and affiliations

Science  22 Jun 2007:
Vol. 316, Issue 5832, pp. 1749-1752
DOI: 10.1126/science.1141084

Abstract

Actin, which is best known as a cytoskeletal component, also participates in the control of gene expression. We report a function of nuclear actin in the regulation of MAL, a coactivator of the transcription factor serum response factor (SRF). MAL, which binds monomeric actin, is cytoplasmic in many cells but accumulates in the nucleus upon serum-induced actin polymerization. MAL rapidly shuttles between cytoplasm and nucleus in unstimulated cells. Serum stimulation effectively blocks MAL nuclear export, which requires MAL-actin interaction. Nuclear MAL binds SRF target genes but remains inactive unless actin binding is disrupted. Fluorescence resonance energy transfer analysis demonstrates that the MAL-actin interaction responds to extracellular signals. Serum-induced signaling is thus communicated to nuclear actin to control a transcriptional regulator.

Small guanosine 5′-triphosphate (GTP)–binding proteins of the Rho family control the assembly of the actin cytoskeleton in response to extracellular signals. Activation of Rho leads to the accumulation of filamentous actin (F-actin) through both filament stabilization and de novo polymerization with concomitant depletion of cellular levels of monomeric actin (G-actin). In fibroblasts, Rho signaling regulates the subcellular localization and/or activity of MAL, a G-actin–binding SRF coactivator (13). Experiments with actin-binding drugs or actin overexpression have suggested that MAL activity responds to G-actin concentrations (46). Actin-binding drugs have distinct effects on MAL. Serum-induced nuclear accumulation of MAL and SRF activity is inhibited by latrunculin B (LatB), whereas drugs such as cytochalasin D (CD), swinholide A (SwA), and jasplakinolide induce MAL nuclear accumulation and SRF activation in the absence of signals (1, 4). CD and SwA also disrupt MAL-actin interaction in immunoprecipitation and protein-affinity precipitation assays (1, 6), but the role of actin binding in MAL regulation has remained unclear.

We first tested whether interaction with actin retains MAL in the cytoplasm or controls its continuous nucleocytoplasmic shuttling. In unstimulated cells, inactivation of the exportin Crm1 by its specific inhibitory drug leptomycin B (LMB) induced nuclear accumulation of MAL or MAL–green fluorescent protein (GFP), and this required the B2 region of MAL, a putative nuclear import signal (Fig. 1A and fig. S1). This shows that MAL continuously transits through the nucleus and allows investigation of the signaling requirements for its nuclear import in the absence of export. LMB-induced nuclear accumulation of MAL, but not control proteins, was inhibited by the G-actin–sequestering drug LatB and coexpression of C3 transferase, which irreversibly inactivates Rho, wild-type actin, and the nonpolymerizable actin Arg62→Asp62 (R62D) mutant (5, 7) (Fig. 1B and fig. S2). Thus, Rho and actin signaling control MAL nuclear import.

Fig. 1.

MAL subcellular localization is a regulated dynamic system. (A) (Top) LMB treatment induces relocalization of endogenous MAL (red) and MAL-GFP (green) in a stable NIH3T3 cell line. (Bottom) MAL-GFP, with functional elements (1, 3) and RPEL motif (PFAM 02755) mutations (7) indicated. FCS, fetal calf serum. (B) Rho-actin signaling is required for MAL nuclear import. Cyt, cytoplasmic; C/N, pancellular; Nuc, nuclear. (100 cells per point; n = 3 independent experiments; error bars indicate SEM). wt, wild type. (C) MAL rapidly accumulates in the nucleus (at least 12 cells per condition; error bars, SD).

LMB-induced MAL-GFP nuclear accumulation was rapid, being effectively complete within 5 min (Fig. 1C and fig. S3) and indicating that basal MAL nuclear export rates must be very high to maintain its cytoplasmic localization (see below). Even the maximum rate of serum-induced MAL-GFP nuclear accumulation was less than this basal import rate, suggesting that increased nuclear import is not the major mechanism of MAL relocalization (Fig. 1C). CD and jasplakinolide, which activate SRF (4), induced MAL-GFP nuclear accumulation at a rate comparable to that of LMB (Fig. 1C and fig. S3).

To analyze export directly, we fused MAL to photoactivatable GFP (PAGFP) (8, 9). Fluorescence was activated in the nucleus by focal-plane-restricted multiphoton excitation (10), and its subsequent decay measured (Fig. 2A and fig. S4). In resting cells, export of MAL-PAGFP was extremely rapid, with an apparent initial rate of 2.90% s–1 (probably an underestimate because the 10-s excitation period is comparable to the decay of nuclear fluorescence), and LMB-sensitive. Export was dramatically reduced after serum stimulation (0.48% s–1) and almost completely inhibited by drugs that induce MAL nuclear accumulation and SRF activation (1, 4), including CD, SwA, and jasplakinolide. MAL-GFP remained nuclear for several hours after serum stimulation; this reflects continued signaling, because MAL reaccumulated in the cytoplasm upon LatB treatment or after serum removal, with an initial rate comparable to that in serum-stimulated cells (Fig. 2B and fig. S5). Thus, nuclear export rather than import represents the major regulatory step in serum-induced nuclear accumulation of MAL.

Fig. 2.

Actin binding and nuclear export. (A) Serum stimulation decreases MAL nuclear export rate. Decay kinetics of nuclear fluorescence after MAL-PAGFP nuclear photoactivation (>10 cells per condition; error bars, SD). (B) Nuclear accumulation of MAL-GFP requires continuous signaling. MAL-GFP localization after serum stimulation with or without additional serum washout and LatB treatment. h, hours. n = 3; error bars, SEM. (C) Sensitivity of a stable MAL-actin complex to actin-binding drugs and RPEL mutations. RPEL domain was bound to G-actin, and apparent molecular masses analyzed by gel filtration. Note that SwA dimerizes actin. Further details are in figs. S6 and S7. (D) GST affinity precipitation analysis of MAL-actin interaction. WB, Western blot. GST baits are shown in fig. S7. (E) Nuclear export requires interaction with actin. Nuclear export rates of wild-type or mutant MAL-GFP proteins measured by FLIP assay, quantified as in (A) (Student's t test, *P < 0.05). Error bars, SEM.

We next studied the interaction between recombinant MAL and purified actin. Gel filtration resolved a complex with a relative molecular mass of 252,000 and an apparent stoichiometry of 1:3 (Fig. 2C and figs. S6 and S7). Complex formation was insensitive to LatB but blocked by CD, SwA, or jasplakinolide. Substitution of the highly conserved positions 1 or 2 of each RPEL (7) motif with alanine (123-1A, 123-2A mutations, Fig. 1A) greatly reduced complex formation. Both gel filtration (Fig. 2C; see fig. S6 for further information) and a less-stringent glutathione S-transferase (GST)–MAL pulldown assay (Fig. 2D and fig. S7) indicated that MAL-123-2A exhibited somewhat greater residual affinity for actin than MAL-123-1A did. We used fluorescence loss in photobleaching (FLIP) (8) to compare the effect of RPEL mutations on MAL export with that of actin-binding drugs. MAL-123-1A and MAL-123-2A, which are nuclear in unstimulated cells, exhibited low export rates essentially identical to that of the wild-type protein in the presence of drugs that disrupt actin binding (Figs. 2E and 3B and fig. S8). Actin overexpression did not alter the subcellular localization of MAL-123-1A-GFP but slightly increased its export rate in the FLIP assay, which was prevented by CD (Fig. 2E and fig. S8). Actin overexpression redistributed MAL-123-2A to the cytoplasm, consistent with its greater residual affinity for actin, precluding analysis by FLIP (fig. S8). These data show that interaction with actin is required for Crm1-dependent MAL nuclear export.

Fig. 3.

Nuclear actin represses MAL activity. (A) LMB-induced MAL nuclear accumulation does not activate transcription of the SRF reporter (left) or endogenous SRF gene targets (right), measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR). 0.3%, serum-starved cells; 15%, serum-starved cells. n = 3; error bars, SEM. (B) Disruption of actin binding allows SRF activation by nuclear MAL derivatives. (Left) Localization of MAL derivatives in serum-starved cells. (Right) Activation of MAL-NLS by CD or overexpressed nuclear RPEL domain (261NLS) after 2-day starvation. (C) MAL nuclear accumulation allows association with target genes, measured by qPCR analysis of chromatin immunoprecipitates. n = 3; error bars, SEM.

Although it induced MAL nuclear accumulation, LMB treatment activated neither an SRF reporter nor transcription of the MAL-dependent SRF target genes Vcl, Srf, Cyr61, and Acta2 in the absence of serum or CD stimulation (Fig. 3A and fig. S9), suggesting that disruption of actin-MAL interaction is required for nuclear MAL to activate SRF; the MAL-independent SRF target gene Egr1 was unaffected. Consistent with this, CD potentiated activation of an SRF reporter by overexpression of MAL-NLS, which contains a heterologous nuclear import signal and is substantially nuclear-localized, but not activation by the constitutively nuclear MAL-123-1A mutant, which cannot bind actin (Fig. 3B and figs. S8 and S10). MAL-NLS activity was also potentiated by nuclear co-expression of the wild-type MAL RPEL domain (MAL2-261-NLS) but not by its 123-1A derivative, suggesting that MAL-NLS is repressed by actin (Fig. 3B and fig. S10). Consistent with these data, we previously found that NLS-actin expression relocalizes MAL to the nucleus but represses SRF activity (1, 5).

Together these results show that actin, or an actin-dependent cofactor, can repress MAL activity in the nucleus. This appears to occur at the level of gene activation rather than at DNA binding, because in chromatin immunoprecipitation experiments LMB treatment induced a substantial specific increase of MAL recruitment to its target genes Vcl, Cyr61, and Srf, comparable to that induced by CD or serum (Fig. 3C and fig. S11). Nuclear actin might recruit repressors to actin-MAL-SRF complexes or prevent recruitment of transcriptional co-activators. Previous studies have implicated actin in transcriptional control through regulation of RNA polymerases and chromatin-modification and -remodelling complexes (11, 12).

To gain direct insight into actin-MAL interactions in cells, we exploited fluorescence resonance energy transfer (FRET), detected by fluorescence lifetime imaging (13). MAL-GFP was used as donor, and Cy3-labeled anti-myc, recognizing co-expressed Myc-actin, as acceptor. Under the assay conditions, the SRF reporter gene remained regulated (fig. S12). In unstimulated cells, FRET was readily detectable between MAL and actin, indicating that they physically interact (Fig. 4A and fig. S13). Treatment with CD reduced this interaction to background level, whereas LatB treatment increased it, consistent with biochemical and functional data (1, 6). In contrast, no FRET was detected between MAL-123-1A and actin. Serum stimulation transiently reduced but did not abolish FRET between wild-type MAL and actin, which returned to its prestimulation level by 30 min (Fig. 4B). Similar results were observed with a MAL-GFP derivative lacking the B2 region, which remained cytoplasmic (Fig. 4B and fig. S1). In contrast, although LMB treatment induced MAL nuclear accumulation in unstimulated cells, it did not affect actin-MAL FRET, which could nevertheless be reduced to background level by CD treatment (Fig. 4A and fig. S12). In LMB-pretreated cells, serum stimulation also transiently reduced but did not abolish nuclear actin-MAL FRET, although recovery was slower than in untreated cells (Fig. 4B; see below). Thus, even when MAL is artificially confined to the nucleus, actin-MAL interaction can respond to serum-induced signals.

Fig. 4.

MAL interacts with actin in both cytoplasm and nucleus. (A) Detection of MAL-actin interaction in vivo by FRET and FLIM. FRET efficiencies shown as box-and-whisker plots with median (>20 cells per condition; Mann-Whitney test, ***P < 0.0005; **P < 0.005). (B) Serum stimulation transiently decreases median FRET efficiency (***P < 0.0005 relative to previous point). (C) Serum stimulation transiently reduces DNAseI-stainable G-actin (n = 3; error bars, SD). (D) Multiple roles for actin in MAL regulation. In unstimulated cells, high export rates ensure MAL is mainly cytoplasmic, whereas nuclear actin prevents SRF activation. Upon stimulation, decreased export induces nuclear MAL accumulation, and diminished interaction with actin allows SRF activation. Proteins are shown as monomers for simplicity.

In serum-stimulated cells, the reduced but significant actin-MAL FRET, detectable when MAL is entirely nuclear, must reflect generation of a subpopulation of actin-free MAL or reduced actin-binding stoichiometry. It is likely that this reduced interaction at least initially reflects a rapid drop in the availability of G-actin, because total cellular deoxyribonuclease I (DNaseI)–stainable actin exhibited a similarly rapid decrease, followed by a slower recovery to prestimulation level (Fig. 4C). Actin thus interacts with MAL in both nucleus and cytoplasm, and serum-induced signals and actin-binding drugs change this interaction in a way consistent with the functional data. Together, these data provide direct support for our proposal that MAL activation reflects reduced MAL-actin interaction arising from depletion of the cellular G-actin pool (1).

We have demonstrated that actin regulates MAL activity at three levels: nuclear import, nuclear export, and activation of target gene transcription (Fig. 4D). Nuclear actin plays an especially important role in regulation of MAL localization and activity. MAL shuttles continuously between cytoplasm and nucleus, even in unstimulated cells. Serum-induced signals primarily control MAL subcellular localization by reducing the rate of its nuclear export, which requires actin binding and Crm1. However, our preliminary data indicate that recombinant RPEL domain–Crm1 interaction in vitro does not require actin. Perhaps actin instead facilitates interaction of Crm1-MAL complexes with nucleoporins (14). Defective nuclear export might explain the ability of some actin mutants to induce MAL nuclear accumulation even though they bind MAL effectively (6). Actin overexpression can also inhibit MAL nuclear import, perhaps by masking the B2 nuclear import signal, although this does not appear to play a major point of regulation in the cells studied here. It remains to be seen whether actin enters or exits the nucleus while bound to MAL.

Our data suggest that there must be communication between the cytoplasmic and nuclear actin pools. Other nuclear functions of actin may therefore be sensitive to extracellular signals. Actin-profilin complexes are subject to apparently constitutive export to the cytoplasm via interaction with exportin 6 (15), but the mechanisms of actin nuclear import remain obscure, and MAL itself may contribute. It will be interesting to assess how the dynamics of cytoplasmic-nuclear shuttling of actin impinge on MAL subcellular localization and activity in different cell types.

Supporting Online Material

www.sciencemag.org/cgi/content/full/316/5832/1749/DC1

Materials and Methods

Figs. S1 to S13

References

References and Notes

View Abstract

Navigate This Article