Formins filter modified actin subunits during processive elongation

Abstract

Fission yeast cells reject actin subunits tagged with a fluorescent protein from the cytokinetic contractile ring, so cytokinesis fails and the cells die when the native actin gene is replaced by GFP–actin. The lack of a fluorescent actin probe has prevented a detailed study of actin filament dynamics in contractile rings, and left open questions regarding the mechanism of cytokinesis. To incorporate fluorescent actin into the contractile ring to study its dynamics, we introduced the coding sequence for a tetracysteine motif (FLNCCPGCCMEP) at 10 locations in the fission yeast actin gene and expressed the mutant proteins from the native actin locus in diploid cells with wild-type actin on the other chromosome. We labeled these tagged actins inside live cells with the FlAsH reagent. Cells incorporated some of these labeled actins into actin patches at sites of endocytosis, where Arp2/3 complex nucleates all of the actin filaments. However, the cells did not incorporate any of the FlAsH-actins into the contractile ring. Therefore, formin Cdc12p rejects actin subunits with a tag of ∼2 kDa, illustrating the stringent structural requirements for this formin to promote the elongation of actin filament barbed ends as it moves processively along the end of a growing filament.

Introduction

Formins are a family of homodimeric proteins, each comprising two conserved domains in addition to more variable regulatory elements (Goode and Eck, 2007, Paul and Pollard, 2009b). Formin homology 2 (FH2) domains are composed of alpha-helices and connected head-to-tail to the partner FH2 domain by flexible linkers (Xu et al., 2004). These linkers can stretch enough for the FH2 dimer to wrap around an actin filament (Otomo et al., 2005), with a higher affinity for the barbed end than any other region of the filament. Formin homology 1 (FH1) domains are flexible polypeptides containing 2–12 polyproline sequences that bind profilin–actin complexes (Chang et al., 1997). Diffusion of the FH1 domains brings profilin–actin into contact with the barbed end (Kovar et al., 2006, Paul and Pollard, 2008), so that the complex can bind to the filament, followed by rapid dissociation of the profilin. The rate limiting step of formin mediated assembly of actin filaments is profilin–actin binding to the multiple FH1 sites, followed by the rapid transfer of actin onto the barbed end at a rate >1000 s⁻¹ (Paul and Pollard, 2009a). As each new actin subunit adds to the barbed end, the formin steps reliably onto the new subunit, so formins move processively along growing barbed ends as more than 10,000 subunits are incorporated (Kovar et al., 2006, Paul and Pollard, 2008).

We are particularly interested in formin Cdc12p, which assembles actin filaments for the actomyosin contractile ring during cytokinesis in fission yeast. Fission yeast Schizosacchromyces pombe, is an excellent model organism to study the actin cytoskeleton, because powerful genetics, biochemistry and quantitative microscopy are available (for review see (Pollard and Wu, 2010)). These yeast cells have three structures composed of actin filaments: actin patches at sites of clathrin-mediated endocytosis, contractile rings for cytokinesis and actin cables for intracellular transport (for review, see (Kovar et al., 2011)). Genetics experiments showed that distinct proteins nucleate the actin filaments for these three structures: Arp2/3 complex nucleates filaments in actin patches (Morrell et al., 1999); formin Cdc12p nucleates filaments for the contractile ring (Chang et al., 1997); and formin For3p nucleates filaments for interphase cables (Feierbach and Chang, 2001). Deletion or temperature sensitive mutations of any of these actin nucleators leads to defects specifically in one type of actin structure but not the others. Each of these three nucleators localizes to distinct sites appropriate for the filaments they form in cells.

Although we know a great deal about the molecular components and assembly mechanism of contractile rings (Vavylonis et al., 2008, Wu and Pollard, 2005, Wu et al., 2003), we still know very little about the dynamics of actin filaments during the assembly and disassembly of the rings. This information would greatly enhance our understanding of the mechanism of cytokinesis. However, such an effort had been hampered by the lack of a fluorescence probe of actin filaments. Formins (Bni1, mDia1, mDia2, Cdc12p) incorporate actin subunits labeled with Oregon green (molecular weight 463) on cys374 or Alexa green (molecular weight 643) on a lysine side chain into actin filaments in biochemical assays (Kovar and Pollard, 2004). However, actin with an N-terminal fluorescent protein tag does not incorporate into the actin filaments of the contractile ring in fission yeast (Wu and Pollard, 2005). Since these actin filaments are essential for cell division, replacement of the native actin gene with a gene encoding a fusion of a fluorescent protein and actin results in the failure of cytokinesis and the cells die, apparently because formin Cdc12p rejects all of the fusion protein. Although fission yeast cells cannot survive on GFP–actin alone, they tolerate low levels of expression of GFP–actin in the presence of wild type levels of native actin (Wu and Pollard, 2005). These trace levels of GFP–actin can be used to make quantitative measurements of actin filaments in actin patches (Sirotkin et al., 2010) but not contractile rings.

Although probes such as fluorescent protein-tagged calponin homology domains (CHD) (Wachtler et al., 2003) and Lifeact (Riedl et al., 2008) can be used to label actin filaments indirectly in vivo, they have shortcomings. These probes bind actin filaments with K_ds of about 2–50 μM (Gimona et al., 2002, Mateer et al., 2004, Riedl et al., 2008), thus they must be expressed at a relatively high concentration to detect actin filaments in cells, which could perturb the actin cytoskeleton. More importantly, none of these indirect fluorescent probes can be used for quantitative fluorescence microscopy to characterize the dynamics of actin filaments in live cells, which requires direct labeling of the protein being studied (Wu and Pollard, 2005). Here we inserted a tetracysteine peptide at 10 sites in fission yeast actin, and labeled these proteins with the FlAsH reagent in live cells (Adams et al., 2002, Griffin et al., 1998). The combined size of the peptide and dye is less than 2 kDa. Some tetracysteine tagged actins were incorporated into actin filaments nucleated by Arp2/3 complex in actin patches, but to our surprise formin Cdc12p excluded all of these labeled actin molecules from contractile rings. Here we explore the structural basis for this rejection of labeled actin by formins.