Richard Treisman: Projects

Changes in cell identity and behaviour are in large part brought about through alterations in gene transcription patterns induced by mitogenic and growth factor stimuli, acting through cell surface receptors, and interactions with extracellular matrix, mediated by adhesion receptors. We study the  Serum Response Factor (SRF) network, a group of transcriptional regulators central to the response to extracellular signals.

The focal point of the network is SRF, which acts in partnership with two families of regulatory cofactors, which compete for a common site on the SRF DNA-binding domain. The myocardin-related transcription factors (MRTFs) are controlled by growth factor-induced changes in cellular G-actin concentration, while the ternary complex factors (TCFs) respond to the classical Ras-ERK signalling pathway. Actin binds to the MRTFs through a conserved domain containing three RPEL motifs, each of which constitutes a G-actin binding site. The study of other RPEL proteins such as the Phactr family of PP1 cofactors, particularly their role in the response to changes in cytoskeletal dynamics, has become a major interest in the group.

 

The SRF network and Rho-actin signal targets.

The SRF network and Rho-actin signal targets. Conserved domains or motifs are shown. The TCF B box and MRTF B1 boxes compete for a common surface on the SRF DNA-binding domain. RPEL1 and RPEL2 exhibit greatly reduced affinity for actin in Myocardin, which is constitutively nuclear. The conserved PP1 binding region in the Phactr proteins is shown in green. Only Phactr1 shuttles, although all domain are conserved throughout the family. Black boxes, NLS basic elements. (Click to view larger image)

Global analysis of the SRF network

Although SRF was first identified in studies of the fibroblast serum response over two decades ago, the extent to which it is responsible for the serum-induced immediate transcriptional response, and the roles played by its cofactors, have remained uncharacterised.

We have used ChIPseq and RNAseq to define the extent of the SRF network in fibroblasts, in conjunction with pathway-specific inhibitors to assess the role of signalling. A substantial majority of over 3100 SRF sites in fibroblasts recruit the MRTFs, with only a minority of sites binding TCFs, and the small number of sites that bind both cofactor families in general exhibit a clear preference for one family or the other. SRF appears to be the primary targeting agent for the MRTFs in fibroblasts, and so far we have found no evidence for MRTF recruitment by other sequence-specific DNA-binding proteins.

SRF binding sites are preferentially associated with transcriptionally active genes, the majority being located within gene features. SRF is overwhelmingly associated with serum-induced transcriptional activation, with over 14 times as many genes activated as were repressed. Intriguingly, over two-thirds of the genes associated with SRF sites are apparently constitutively transcribed, suggesting that they are somehow refractory to activation through SRF cofactors. Consistent with this, parallel studies show that only 8% of constitutively active SRF-linked genes are SRF-dependent, compared with up to 60 per cent of serum-inducible ones.

 

MRTF-SRF signalling adds a nuclear dimension to inside-out signalling

MRTF-SRF signalling adds a nuclear dimension to inside-out signalling. Classical cytoplasmic inside-out signalling is indicated in blue, with adhesion leading to Rho GTPase activation and consequent assembly of new focal contacts and adhesions. Concomitant activation of the MRTF-SRF pathway potentiates expression of genes involved in all aspects of adhesion signalling, including ECM components and modifiers (examples in fibroblasts are shown at right). Green dashes indicate the physical linkage between the nucleus, cytoskeleton and adhesions. (Click to view larger image)

We have used this data to define an MRTF-SRF target gene set of 921 genes, which bind MRTF ChIP and/or respond to MRTF-linked signals. This gene signature significantly overlaps with gene signatures characteristic of cancer cell invasion and metastasis, and the response to mechanical stress.

Integrin engagement with extracellular matrix (ECM) leads to focal adhesion (FA) assembly, requiring Rho-dependent actomyosin contractility, and this process, known as inside-out signalling, is strongly implicated in cancer cell invasiveness. The MRTF-SRF pathway thus forms an additional longterm nuclear arm of inside-out signalling, controlling expression of major FA force sensors such as Talin and Vinculin, alongside ECM components, actomyosin components and cytoskeletal regulators.

G-actin as an MRTF regulator

The MRTF proteins are novel G-actin binding proteins whose activity responds to fluctuations in G-actin concentration. A major point of regulation is at the level of their nuclear accumulation: G-actin binding prevents MRTF nuclear import, and is required for nuclear export.

Previous studies from the group showed that growth factor-induced Rho GTPase activation results in G-actin depletion and blocks nuclear export, leading to MRTF nuclear accumulation. In contrast, when G-actin concentration is elevated, G-actin binding inhibits MRTF nuclear import, because it prevents binding of importin α-β heterodimers to a bipartite NLS within the RPEL domain, and these properties are conserved in the RPEL protein Phactr1.

Actin complexes on the MRTF RPEL domain

Actin complexes on the MRTF RPEL domain. A stable trimeric complex is detected in solution with actin bound to RPEL1 and RPEL2 and the intervening spacer. An unstable pentameric complex, in which RPEL3 and the second spacer are also bound by actin, is also detectable at high actin concentrations. Integrity of all the actin binding sites are required to keep the MRTFs cytoplasmic in resting cells. The NLS is indicated in grey, RPEL motifs in red. (Click to view larger image)

While we know that G-actin inhibits import, we know little about its function in export, other than Crm1 is involved. Although it has been reported by others that phosphorylation in the C-terminal region is required, our own studies have shown that MRTF C-terminal sequences are not essential for export. Instead, it appears that multiple Crm1-dependent NES elements cooperate for MRTF export, but the relation between their function and G-actin binding remains to be elucidated.

G-actin also regulates transcriptional activation by the MRTFs. In previous studies we found that confinement of MRTFs to the nucleus by inactivation of the Crm1 exportin does not inhibit their activation by growth factor stimuli, which still induces actin-MRTF dissociation. Thus, regulated MRTF nuclear accumulation cannot be the only step at which MRTF activity is regulated. Chromatin immunoprecipitation studies show that both Crm1 inactivation and targeting of MRTFs to the nucleus by fusion to a nuclear import signal results in their recruitment to SRF target genes.

We are using biochemical and genomic approaches to investigate the basis of this G-actin mediated nuclear transcriptional inhibition. Preliminary data suggest that G-actin inhibits both MRTF recruitment to SRF, and the ability of MRTF to recruit active RNA polymerase II, and that both phenomena requires the RPEL domain. A priority will be to elucidate the mechanism of this inhibition.

Two state model for MRTF-A nucleocytoplasmic shuttling

Two state model for MRTF-A nucleocytoplasmic shuttling. Changes in actin concentration affect assembly of import- and export-competent complexes. Allactin binding sites are required to maintain MRTF-A in the cytoplasm in resting cells suggesting that the pentavalent actin•MRTF-A complex is the export substrate. (Click to view larger image)

G-actin as a regulator of the Phactr family and other proteins

The Phactr protein family comprises four RPEL-containing proteins which are cofactors for PP1. The Phactrs have been implicated in cell cycle regulation, cytoskeletal dynamics, and development of cancers such as melanoma. These four proteins contain one N-terminal RPEL motif and a C-terminal cluster of 3 motifs abutting a conserved element which forms a PP1c binding site. The Phactr1 protein, but not the other family members exhibits regulated nucleocytoplasmic shuttling in response to changes in actin dynamics, his requires the integrity of its C-terminal RPEL motifs. Like the MRTFs, elevated G-actin levels inhibit signal-induced Phactr nuclear accumulation, and this requires the N-terminal RPEL motif and the C-terminal motif cluster. Phactr1 appears required for the actomyosin contractily, and invasiveness and motility of melanoma cells in culture.

The Phactr1 C-terminal domain binds three rather than five actins, in contrast to the MRTFs, suggesting that the mechanism by which these proteins respond to changes in actin dynamics is different. The similar spatial relationship between the RPEL actins in the two complexes reflects secondary contacts between each actin-bound RPEL motif and the neighbour actin. G-actin competes for PP1 binding to the C-terminal region, and with importin a-b binding to NLS sequences embedded within the C-terminal RPEL domain and adjacent to the N-terminal RPEL motif.We are studying the structural basis of Phactr1-PP1 interaction, and the mechanism by which G-actin regulates PP1 binding. The Phactr1-PP1 complex does not contain other cofactors, and Phactr1 binding does not occlude the PP1 substrate binding site. We are using proteomic and affinity-labelling approaches to identify potential Phactr1-PP1 substrates.

Phactr1 regulates the actin cytoskeleton

Phactr1 regulates the actin cytoskeleton. Left, siRNA-mediated knockdown of Phactr1 in CHL1 melanoma cells results in loss of transverse actomyosin stress fibres defective motility in the scratch-wound assay, and loss of invasive capability. Right, schematic summary of the role of Phactr1 in cytoskeletal dynamics. (Click to view larger image)

A number of other proteins also contain RPEL motifs, and we are studying whether these proteins bind G-actin, and if so, its relevance to their function. In at least one case, G-actin bound to the RPEL motif interacts in cis with sequences in a distant domain, setting a precedent for new regulatory mechanisms in RPEL proteins.

The SRF network in haematopoietic cells

T cell development provides a system in which to study the biological significance of TCF-SRF signalling. Our gene knockout studies showed that the SAP-1 TCF is essential for thymocyte positive selection in the mouse, and that SAP-1 functions redundantly with Elk-1, but not Net, in this system. Consistent with this, SRF inactivation completely blocks T cell development.

Studies with SRF mutants show that SRF mutants that cannot recruit cofactors are completely inactive in thymocyte development, and that fusion of the TCF regulatory domain is sufficient to restore T cell development. SRF is required for the initial effector T cell proliferative response to immune challenge, simultaneously suppressing early memory precursor differentiation, and TCF-SRF signalling is also required for homeostatic T cell proliferation. We are using T cell system to study the role of the SRF network in a the context of bona fide proliferative stimuli.

We have tested the role of the MRTF cofactors in the haematopoietic system through gene inactivation studies. During haematopoietic development the transition of haematopoiesis moves from the liver to the fetal bone marrow just before birth. Inactivation of SRF early in hematopoietic development using vav-CRE results in a phenotype similar to that seen upon inactivation of the SDF-1 chemokine, with specific inability of HSC/P cells to colonise the fetal bone marrow, but not other tissues. SRF-null haematopoietic progenitors fail to migrate to CXCL12 and are defective in their migration on and through endothelial cell monolayers, consistent with SRF being required for the response to this chemokine. Transcriptome analysis demonstrates that as in the fibroblast system, SRF-dependent transcripts are highly enriched in those encoding cytoskeletal structural genes and regulators. Many of these genes are specific to the stem cells, showing that cell context is an important determinant of SRF target gene specificity.

The bone marrow colonisation and SDF-1 migratory phenotypes characteristic of SRF-null animals are also seen animals lacking both MRTF-A and MRTF-B in hematopoietic cells: inactivation of both MRTFs is required to block bone marrow colonisation, although MRTF-A null animals are partially defective. A single copy of either Mrtfa or Mrtfb in the hematopoietic system suffices to rescue viability, and thus in HSC/Ps the two MRTFs appear to exhibit effective functional redundancy, as previously observed in megakarocyte differentiation and other developmental systems.

 

MRTF-SRF signalling adds a nuclear dimension to inside-out signalling

MRTF-SRF signalling adds a nuclear dimension to inside-out signalling. Classical cytoplasmic inside-out signalling is indicated in blue, with adhesion leading to Rho GTPase activation and consequent assembly of new focal contacts and adhesions. Concomitant activation of the MRTF-SRF pathway potentiates expression of genes involved in all aspects of adhesion signalling, including ECM components and modifiers (examples in fibroblasts are shown at right). Green dashes indicate the physical linkage between the nucleus, cytoskeleton and adhesions. (Click to view larger image)

The SRF network in invasion and cancer

Numerous studies of both human cancers and mouse cancer models have implicated Rho signalling, and RhoC in particular, in cancer cell invasion and metastatic tumour spread, and this has been generally thought to reflect direct effects on the cytoskeleton. Since the control of MRTF activity by Rho provides a mechanism by which cytoskeletal gene expression can be coordinated with cytoskeletal regulation, and it is increasingly clear that the MRTF-SRF pathway influences cancer cell invasion and metastasis.

MRTF-SRF signalling is required for experimental metastasis

MRTF-SRF signalling is required for experimental metastasis. Left panels, B16F2 derived lung nodules following tail vein injection of B16F2 control or derivatives expressing MRTF or SRF shRNAs. Right panel, quantification of fluorescently labelled control or MRTF-depleted human MDA-BA-231 or rat MTLN3 breast cancer cells in lung sections at the indicated times after tail vein injection. (Click to view larger image)

Using knockdown approaches we have shown that MRTF-SRF signalling is required for invasion and experimental metastasis two highly metastatic tumour cell lines, human MDA-MB-231 breast carcinoma and mouse B16F2 melanoma in which Rho signalling plays an important role in invasion and migration. Depletion of MRTF or SRF activity in both lines impaired cell motility in the scratch-wound assay, reducing speed and directional persistence, reduced cell spreading and adhesion to fibronectin and collagen matrix, and decreased the ability of MDA-MB-231 cells to migrate through collagen-coated membrane toward serum-containing medium in a modified Boyden chamber assay. MRTF-SRF activity was also required for their invasiveness in an organotypic model, in which tumour cells move through collagen gel along tracks generated by stromal carcinoma-associated fibroblasts.

MRTF-SRF depleted cancer cells exhibit reduced cell motility in tumour xenografts, suggesting a defect in metastatic potential, and consistent with this, MRTF- and SRF-depleted tumour cells are unable to colonise the lung from the bloodstream in an experimental metastasis model, even though they initially arrive at the organ in similar numbers to non-depleted cells (see Figure). Actin-based cell behaviour and experimental metastasis thus requires Rho-dependent nuclear signalling through the MRTF-SRF network. Microarray analysis identified Myh9 and Myl9 as among a small number of shared MRTF-SRF targets required for these processes. Current studies in the laboratory aim to test the role of the SRF network directly in genetically-engineered mouse models of cancer.

Richard Treisman

richard.treisman@crick.ac.uk
+44 (0)20 379 61104

  • Qualifications and history
  • 1981 PhD, Imperial Cancer Research Fund/University College London, UK (Advisor: Bob Kamen)
  • 1981 Postdoctoral Fellow, Harvard University, USA (Advisor: Tom Maniatis)
  • 1984 Scientific Staff, MRC Laboratory of Molecular Biology, UK
  • 1988 Established lab at the Imperial Cancer Research Fund (ICRF; became Cancer Research UK in 2002)
  • 2000 Director of Laboratory Research, ICRF
  • 2002 Director, London Research Institute, Cancer Research UK
  • 2009 Research Director, the Francis Crick Institute, London, UK