The cytoskeletal organisation of actin filaments, has a well recognised role in maintaining the structure of cells. However, it is beginning to emerge that cytoskeletal organisation is a very complex and dynamic process, with an influence on almost every function of the cell. Regulation of actin polymerisation is a key event in allowing actin structures to form and/or be reorganised, and thus, a number of proteins are involved it this regulation (for review and references see Pollard et al., 2000). The simplest model proposed involves dynamic cycling between two forms of actin in the cell, monomeric (G-actin) and filamentous (F-actin). Monomeric actin subunits (ATP- bound) can polymerise from one end of the filament (barbed end), allowing the filament to grow (see Figure 1.9). Conversely, ADP-bound subunits can dissociate from the other end (pointed end), allowing filament to disassemble. The barbed end is also called the fast growing end, and pointed end, the slow growing end (Chen et a l,
2000). Regulation of F-actin formation can be achieved in at least in two ways: regulation of the activity of severing proteins, such as cofilin and gelsolin, and binding of stabilising proteins, such as tropomyosin. Rho family protein-induced regulation of actin stability can be explained by phosphorylation of cofilin. It has been reported that ROCK and PAK kinases activates LIM kinase, which in turn inactivates cofilin by phosphorylation .(Edwards et a l, 1999; Maekawa et al., 1999). When posphorylated cofilin is displaced from actin monomers allowing them polymerise from the barbed end (Hamburg et a l, 1999; Pollard et a l, 2000). The severing of actin filaments by gelsolin can be regulated by extracellular stimuli, that modulate free calcium ion concentration, since gelsolin activity is calcium dependent. Regulation of gelsolin
Chapter 1: Introduction
activity by LPA has been reported, suggesting that serpentine receptors can modulate actin polymerisation (Meerschaert et al., 1998).
In addition to stabilising filamentous actin, Cdc42 and Rac family GTPases have been implicated in initiation of new filament formation through the Arp2/3 complex (Machesky, 1997). Biochemical and jelectronmicroscopic data suggest that Arp2/3 complex localises to regions of lamellipodial protrusions and nucleates new branches of actin filaments (Machesky et a l, 1997; Svitkina and Borisy, 1999). Consequently, it has been demonstrated that N-WASP and Scarl promote Arp2/3 nucléation activity (Machesky and Insall, 1998yfor review, see Machesky and Gould, 1999; Machesky and Insall, 1999). Moreover, the WASP family protein family Ena/VASP has been demonstrated to recruit actin monomers and promote filament elongation by bridging monomers and filaments. Ena/VASP proteins contain the conserved amino-terminal domain EV H l (Ena/VASP homology) and proline-rich sequences that bind to profilin and to SH3 domains (Machesky and Schliwa, 2000; Prehoda et al., 1999; Symons et al., 1996). Based on these and other findings, the following model for signal induced regulation of actin polymerisation was proposed. Growth factor activation of receptor tyrosine kinases creates phosphorylation sites for docking proteins like grb2 and NCK, which then recruit WASP (or WASP family members) to the plasma membrane, which in turn activates the Arp2/3 complex and promote actin polymerisation. In addition, the relative activities of capping proteins are also regulated by signalUng intermediates. The rate that knew filament elongation depends on Ena/VASP proteins and profilin. Established filaments are severed and depolymerised by cofilin. Conversely, positive signals leading to LIM kinase activation inactivate cofilin’s severing ability, hence promoting polymerisation (reviewed by Machesky, 2000; Machesky and Schliwa, 2000); (see Figure 1.9).
l.S.2.1. Actin binding drugs
Jasplakinolide
Jasplakinolide is a cyclic peptide from marine sponge Jaspis johnstoni (Bubb
et al., 1994). It has been demonstrated that Jasplakinolide stabilises filamentous actin
in vitro, however in vivo data suggests that Jasplakinolide disrupts actin filaments and induces actin polymerisation in the amorphous phase (Bubb et al., 2000). This apparent paradox could be explained by the fact that Jasplakinolide induces nucléation of actin filaments, thus, resulting in an induction of actin monomer (G-actin)
Chapter 1: Introduction
polymerisation. It has been suggested that the disrupting effects on stress fibres and depletion of G-actin could happen by a two step mechanism: first, inducing the release of actin from its sequestering protein (p-thymosin or others), second, by nucléation of filament assembly, leading to a state where the levels of G-actin is limiting to maintain normal stress fibre turnover. Other reports support the evidence that Jasplakinolide leads to depletion of G-actin in a time and concentration dependent manner. The amount of the drug needed to titrate G-actin depends on the cell type used or state of the cell. For instance, serum starved cells are more sensitive to actin binding drugs than growing ones. Indeed, it is recognised, that serum starved fibroblasts have reduced levels of stress fibres, therefore, possibly containing a larger pre-existing pool of polymerisation-competent actin. However, the assessment of polymerisation- com petent actin levels is difficult due to its distribution, nucleotide content, postranslational modifications and/or sequestration by actin binding proteins, such as p-thymosin and profilin (Cao et a l, 1993; Goldschmidt-Clermont et a l, 1992).
Latrunculins
Latrunculin was isolated from Red Sea sponge Necombata magnifica, and identified as an inhibitor of actin polymerisation (Spector et al., 1983). Latrunculin binds to monomeric actin with a ratio 1:1, with a dissociation constant 0.2 pM (Coue
et al., 1987). Analysis of the crystal structure has revealed that latrunculin binds to the nucleotide binding cleft of actin (Morton et al., 2000), and therefore affects the nucleotide exchange rate of actin (Ayscough et al., 1997). Latrunculin has been reported to lower p-thymosin binding to actin by 1 0 fold, without affecting the
complex of profilin and actin in vitro. In addition, latrunculin did not affect binding of DNAse which is otherwise used as a measure of G-actin in cell culture experiments (Yarmola et al., 2000). However, the in vivo effects of latrunculin binding to actin, with respect to other actin binding proteins, remains poorly understood. Both Latrunculin A and Latrunculin B have similar effects on the actin cytoskeleton, however, they act with slightly different binding affinities.
Cytochalasin D
Although Cytochalasin D have been used for a number of years as an agent capable of disrupting the actin cytoskeleton, the reports concerning Cytochalasin D binding to actin in vivo are very limiting (for review see Cooper, 1987). Cytochalas in D binds to the barbed end of actin filaments, inhibiting both association and dissociation of subunits at the end. Affinity of Cytochalasin D for the barbed ends
Chapter 1: Introduction
is high (Kd=2nM), which is by 3 orders of magnitude lower when usually used in the experimental systems (Kj=2pM). Capping barbed actin filaments is the only known function of Cytochalasin D in vivo. In vitro, however, Cytochalasin D has been reported to bind both actin monomers and dinuners as well as promoting ATP hydrolysis (Brenner and Korn, 1981). Some reports suggests that cytochalasins might bind to the a subunit in the interior of an actin filament and thus break the filament suggesting that Cytochalasin D posses severing activities (Hartwig and Stossel, 1979). Despite the complexity of Cytochalasin D action, no reports suggest that it binds other molecules then actin. It is clear that Cytochalasin D disrupts supramolecular organisation of actin filaments, but not necessarily depolymerise actin filaments like latrunculins.
Swinholide A
Swinholide A, another actin binding drug, has also been discovered whilst searching for marine natural products with biological activity. Swinholide A (isolated from marine sponge Theonella swinhoei) is a 44-carbon ring dimeric dilactone macrolide. This drug is highly cytotoxic, although the mechanism of cytotoxicity is unknown (Bubb et al., 1995). Swinholide A binds co-operatively to two actin subunits with a Ka 9x10 M'^. In addition to sequestering non-polymerised actin subunits, Swinholide A increases the number of filament ends by severing F-actin (Bubb and Spector, 1998). Swinholide A is a very powerful disrupter of the actin cytoskeleton: a concentration of lOnM is sufficient to cause substantial changes to cytoskeletal arrangements within 30 minutes. In contrast to cytochalasins, the effects of Swinholide A on actin are very specific, and appears to only sequester actin as a dim Î lers and sever F-actin. Thus, this is a promising reagent for use in the study of actin dynamics and cytoskeletal rearrangements in the future (Bubb and Spector,
1998).
I.5.2.2. The actin cycle and SRF activation
It has been previously demonstrated that Rho A activity is absolutely necessary for serum- and LPA-induced SRF activation in transfection assays (Hill et at., 1995). Subsequently, it has been demonstrated that Rho signalling impinges upon the actin treadmill cycle via two downstream effectors, m Dial and ROCK. Like Rho, mDia is very a strong activator of SRF (Sotiropoulos et at., 1999; Tominaga et al., 2000). The
Chapter 1: Introduction
exact mechanism by which this occurs is not clear, however, it has been demonstrated that D ial can interact with the actin binding protein profilin (Watanabe et al., 1999). Another powerful activator of SRF, LIMK has been identified in a screen looking for potential activators (Sotiropoulos et at., 1999).
LIMK activation of SRF
Several lines of evidence suggest that LIMK activate SRF through the actin pathway. First, the actin binding protein cofilin is the only known physiological target of LIMK. Second, mutation of phosphorylation site Ser3 of cofilin inhibit SRF activation. Third, the actin binding drug latrunculin blocks SRF activation by LIMK (Sotiropoulos et at., 1999). It has been reported that LIMK activity can be modulated by phosphorylation of Thr508, in its activation loop, by ROCK (Ohashi et al., 2000) or by PAK (Edwards et al., 1999). This suggests that both RhoA and Rac could contribute to regulation of LIM kinase activity through their downstream effectors, and thereby activate SRF. The relative contribution of each GTPase in SRF activation would be dependent on different signalling pathways and/or different cellular context.
Activation of SRF by Dia
Interestingly, in NIH 3T3 cells, a constitiwely active form of mDial strongly activates the SRF reporter gene (Copeland and Treisman, 2001). In contrast, another effector of RhoA, ROCK, is only a very weak activator of SRF, even though the constitutively active R0CKA3 is C3 sensitive. In addition, the specific Rho inhibitor, Y27632, does not significantly inhibit the SRF reporter activation after serum stimulation (Sahai et a l, 1999). These data strongly suggest that Dia is the main downstream effector of RhoA leading to SRF activation in NIH 3T3 cells. This is consistent with the fact that dominant negative form of LIMK (which is regulated by ROCK) does not substantially inhibit SRF activation in NIH3T3c cells (Sotiropoulos
et al., 1999). Conversely, in PC12 cells LIM kinase is a major effector of Rho GTPases leading to SRF activation, since cofilin S3A is able to inhibit signals activating SRF (Geneste et a l, 2001).
Actin binding dmgs and SRF activation
It has been reported previously that latrunculin has an adverse effect on expression of certain cytoskeletal genes, such as vinculin and actin (Ben-Ze'ev et a l,
1990; Bershadsky et a l, 1995; Lyubimova et a l, 1999). It has also been shown that 46
Chapter 1: Introduction
phalloidin treatment changes the levels of actin (Serpinskaya et al., 1990). The promoters of these genes have subsequently been shown to contain consensus SRF binding sites (Mohun et a l, 1987; Moiseyeva et a l, 1993). Activation of an SRF reporter gene by the actin binding drug Cytochalasin D has been demonstrated in the Treisman laboratory. Interestingly, Cytochalasin D action was demonstrated to be independent (downstream) of RhoA (C. Hill and R. Treisman, personal communication ; Sotiropoulos et al., 1999 ).