| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Editorial Focus
Aab Cardiovascular Research Institute, University of Rochester School of Medicine & Dentistry, Rochester, New York
ALL LIVING CELLS, INCLUDING bacteria, require an intricate actin cytoskeleton for normal homeostasis. Once thought to be a static scaffold functioning solely to maintain cell shape, the actin cytoskeleton is now known to be a dynamic menagerie of proteins controlling such processes as migration and contraction, endocytosis and exocytosis, apoptosis, intracellular trafficking, and signal transduction. Many of the component parts of the actin cytoskeleton are also found in the so-called nucleoskeleton, which appears to play an important role in gene transcription (14). For example, nuclear actin and myosin have recently been implicated in interchromosomal associations that bring co-regulated genes in close proximity following hormone stimulation (10). In this manner, coexpressed gene loci may be regionalized to nuclear centers where key transcription factors are concentrated. There is also evidence for nucleoskeletal proteins directly mediating gene expression (14). The actin cytoskeleton appears to effect changes in gene expression indirectly through physical associations with transcription factors held in abeyance within the cytosol. One of the better understood transcription factors under such control is myocardin-related transcription factor A (MRTF-A).
The myocardin family of coregulators binds serum response factor (SRF) over a 10-base pair element called the CArG box (11). Based on the binding rules of SRF, there are 1,216 permutations of the CArG box with the vast majority of functional elements found in close proximity to >200 target genes (7). SRF is only weakly active in stimulating CArG-dependent target genes. However, the CArG-SRF-MRTF-A ternary complex comprises a potent transcriptional switch for gene expression (11). Whereas the founding member, myocardin, is constitutively nuclear, MRTF-A undergoes signal-dependent changes in cellular localization (4). Thus, under normal (e.g., quiescent) conditions, MRTF-A (aka MAL) physically associates with globular (G) actin in the cytosol and is unable to bind nuclear SRF. However, following agonist stimulation (e.g., serum), G-actin is polymerized into filamentous (F) actin, thereby releasing MRTF-A, allowing for nuclear translocation where MRTF-A engages SRF bound to CArG boxes in and around key target genes (8).
The phenomenon of G-actin polymerization into F-actin, and the latter's depolymerization or severing back into G-actin, is known as actin dynamics or actin treadmilling (12). Many proteins are required to coordinate the waves of G-actin polymerization (e.g., profilin) and subsequent F-actin depolymerization (e.g., cofilin). Loss of function in any number of actin treadmilling proteins has severe consequences for normal cell biology and organismal life (7). Despite the wealth of data describing actin dynamics, there has been no genome-wide investigation of changes in gene expression linked to actin dynamics, nor have there been many studies validating in vivo actin dynamics. Now, in this issue of Physiological Genomics, Verdoni et al. (18) provide startling new in vivo data demonstrating a link between defective actin dynamics and global changes in gene expression that include an enrichment of cytoskeletal genes under direct control of SRF (18).
To study the role of actin dynamics on gene expression control in vivo, Verdoni et al. (18) exploited two mouse strains exhibiting variations in expression of a critical actin-binding protein. The corneal disease-1 (corn1) mouse carries a 35-kb chromosomal deletion of the entire coding region of destrin (Dstn) (5). Dstn (aka actin depolymerizing factor, ADF) is a 19-kDa protein that depolymerizes F-actin. Dstn is related in primary amino acid sequence and function to cofilin and is expressed in essentially all mammalian tissues (9). Surprisingly, mice lacking Dstn (Dstncorn1) are viable and exhibit little pathology in most organ systems, suggesting compensation from other ADFs (e.g., cofilin). The Dstncorn1 mouse does, however, display autosomal recessive traits that manifest initially as corneal epithelial hyperproliferation and neovascularization followed by frank cataract formation (15). In a second mouse model (Dstncorn1-2J), a milder corneal phenotype is associated with a putative hypomorphic Dstn allele stemming from a nonsynonymous mutation (Pro106Ser) (5). Verdoni et al. first showed elevated phalloidin and beta actin staining within the epithelial layer of the cornea in both mutant mouse strains, consistent with Dstn's major function as an F-actin depolymerizing factor; though both mutants display F-actin accumulation, the Dstncorn1 mutant showed stronger staining. They then performed well-controlled microarrays from 14-day-old Dstncorn1 or Dstncorn1-2J corneas, a time coinciding with the emergence of F-actin accumulation and epithelial cell hyperproliferation. Consistent with the differential phalloidin staining, hierarchical clustering of array data revealed greater changes (log2-fold compared with control, wild-type mouse) in corneal gene expression from Dstncorn1 mice with 599 genes significantly elevated (compared with 150 in Dstncorn1-2J) and 627 genes reduced (compared with 52 in Dstncorn1-2J). Thus, the extent of F-actin accumulation appears to correlate with the level of gene expression providing compelling support for the actin cytoskeleton coordinating gene expression control. Using the database for annotation, visualization, and integrative discovery (DAVID), Verdoni et al. found cell cycle-associated genes to be enriched in the corneas of Dstncorn1 mice, but not the Dstncorn1-2J mice. These findings are congruent with the hyperproliferation that is more evident in the Dstncorn1 mouse (15). On the other hand, both mutants showed significant enrichment for gene ontology-annotated genes related to the actin cytoskeleton or components of the cell (e.g., adherens junction) directly connected to the cytoskeleton.
Historically, while the cell biology of actin cytoskeletal proteins has been studied in great detail, there is a paucity of data as to transcriptional control of the corresponding genes, particularly those involved directly with actin dynamics. Among the cytoskeletal genes Verdoni et al. (18) showed to be overrepresented in the cornea of both Dstn mutants, a significant number are known targets of SRF. This finding is based on previous genome-wide interrogations that have begun to define the mammalian CArGome, that is, all functional SRF binding sites controlling gene expression (2, 17, 19). Verdoni et al. used real-time RT-PCR to validate 17 known or hypothetical SRF target cytoskeletal genes in corneal epithelium from Dstncorn1 mice. Interestingly, Dstn, a known SRF target gene (17), was increased in the cornea of Dstncorn1-2J mice; no such induction was seen in Dstncorn1 mice since the endogenous Dstn locus is missing. In total, >50 SRF target genes were shown to be increased in the corneas of Dstn mutants. At least two mechanisms may be advanced to explain the bias in elevated SRF-dependent gene expression with F-actin accumulation in Dstn null mice. First, SRF itself could be a target of activation, a finding Verdoni et al. demonstrated using both real-time RT-PCR and immunohistochemistry. Second, the greater accumulation of F-actin in the corneas of Dstncorn1 mice could elevate the nuclear localization of MRTF-A, thereby enhancing SRF-dependent gene expression. Verdoni et al. showed that MRTF-A was increased; however, the increase occurred mainly in the cytosol with only marginal increases in the nucleus. It is unclear as to why MRTF-A does not show more robust nuclear translocation with heightened F-actin. Nevertheless, the increase in MRTF-A (and possibly other myocardin family members) could associate with nuclear SRF to effect elevations in target gene expression.
What might account for augmented SRF and MRTF-A expression in Dstn null mice? The SRF promoter harbors two functional CArG elements that serve to autoregulate its expression (1, 16). Dominant negative MRTF-A can suppress SRF mRNA expression (13), and myocardin factors can activate the SRF promoter (Ref. 3 and unpublished data J. M. Miano). Thus, the increase in SRF reported by Verdoni et al. (18) likely stems from elevated MRTF-A directly activating the SRF promoter, presumably through SRF's two conserved CArG elements. Enhanced expression of MRTF-A was limited to the protein level, and so it is presently unknown whether corresponding changes occur at the transcript level. Moreover, there is nothing known about the regulatory cis elements controlling MRTF-A gene transcription. It is intriguing to consider the possibility that SRF directly or indirectly activates the MRTF-A promoter.
The study by Verdoni et al. (18) is the first published paper linking defective actin dynamics with global changes in gene expression. The degree to which the corneal epithelial cell transcriptome increases or decreases appears to be a direct consequence of the level of F-actin, SRF, and MRTF-A abundance. The provocative findings reported by Verdoni et al. raise a number of research questions for future investigation. First, it will be instructive to determine whether similar changes in F-actin accumulation and gene expression occur in other cell types devoid of Dstn, particularly those with unusually low levels of SRF or MRTF-A expression (e.g., liver and skeletal muscle, respectively). Second, many of the cytoskeletal genes shown to be induced with increased F-actin are hypothetical SRF targets and will require further validation to assign them membership to the CArGome. Third, the molecular biology underlying increases in SRF and MRTF-A (especially) with loss in Dstn should be explored. For example, does knockdown of MRTF-A reduce SRF expression and the cytoskeletal transcriptome in corneas of Dstncorn1 mice? Why is MRTF-A not fully localized to the nucleus under conditions of F-actin accumulation, and might there be concurrent activation of other myocardin paralogs? What co-regulators of SRF (beyond MRTF-A) are recruited to activate CArG-dependent cytoskeletal genes? Much of the story reported by Verdoni et al. relates to the genes induced with F-actin accumulation in the Dstncorn1 mouse. However, an even larger number of genes were reported to be significantly repressed, and the gene ontology of these genes is comparatively less illuminating. Further study into the consequences of repressed genes may thus be of some value in understanding the nature of the Dstncorn1 phenotype. Moreover, myocardin family members can repress gene expression in certain contexts, independent of SRF (6). Thus, MRTF-A could conceivably be a repressor of a subset of down-regulated genes in the Dstncorn1 mouse. Finally, inspection of the UCSC Genome Browser reveals two closely juxtaposed nonsynonymous SNPs within the human DSTN gene (http://genome.ucsc.edu/). Whether these coding SNPs (or those residing in noncoding sequences) are functionally associated with altered actin dynamics and the predisposition to cataracts in humans is unknown.
GRANTS
The author is supported by National Institutes of Health Grants HL-62572, HL-070077, and AG-026950.
FOOTNOTES
Address for reprint requests and other correspondence: J. M. Miano, Aab Cardiovascular Research Inst., Univ. of Rochester School of Medicine & Dentistry, 211 Bailey Rd., W. Henrietta, NY 14586 (e-mail: j.m.miano{at}rochester.edu).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
