Maintenance of F-actin turnover is essential for plant cell morphogenesis. Actin-binding protein mutants reveal that plants place emphasis on particular aspects of actin biochemistry distinct from animals and fungi. Here we show that mutants in CAP1, an A. thaliana member of the cyclase-associated protein family, display a phenotype that establishes CAP1 as a fundamental facilitator of actin dynamics over a wide range of plant tissues. Plants homozygous for cap1 alleles show a reduction in stature and morphogenetic disruption of multiple cell types. Pollen grains exhibit reduced germination efficiency, and cap1 pollen tubes and root hairs grow at a decreased rate and to a reduced length. Live cell imaging of growing root hairs reveals actin filament disruption and cytoplasmic disorganisation in the tip growth zone. Mutant cap1 alleles also show synthetic phenotypes when combined with mutants of the Arp2/3 complex pathway, which further suggests a contribution of CAP1 to in planta actin dynamics. In yeast, CAP interacts with adenylate cyclase in a Ras signalling cascade; but plants do not have Ras. Surprisingly, cap1 plants show disruption in plant signalling pathways required for co-ordinated organ expansion suggesting that plant CAP has evolved to attain plant-specific signalling functions.
Cyclase-associated protein (CAP) was identified in S. cerevisiae as an interactor of adenylate cyclase (AC) (Field et al., 1990). Mutations in CAP/SRV2 not only affect the regulation of AC by Ras (Fedor-Chaiken et al., 1990; Shima et al., 2000) but also cause actin organisational phenotypes (Vojtek et al., 1991). Investigations into the biochemical activity of CAP in the context of the actin cytoskeleton has defined CAP as an actin-binding protein (ABP) capable of associating with monomeric actin and facilitating actin treadmilling (Balcer et al., 2003; Mattila et al., 2004). The C-terminus of S. cerevisiae CAP is required in vivo and in vitro for the majority of cytoskeletal functions (Gerst et al., 1991; Mattila et al., 2004), while the N-terminus regulates AC activation in vivo. The functional division between signalling and actin organisation has led to CAP being considered a bifunctional protein.
CAP is conserved over a wide range of organisms. Cross-species complementation experiments have shown that heterologous CAP can consistently complement S. cerevisiae CAP-dependent cytoskeletal functions but not AC activation. The N-terminus of S. cerevisiae CAP is required to expose AC binding sites to Ras (Shima et al., 2000). S. pombe also requires CAP for AC activity (Kawamukai et al., 1992), but S. pombe AC is not activated by the Ras pathway. CAP in S. pombe must facilitate AC activation in a novel fashion and, consequently, the N-terminus of S. pombe CAP cannot complement S. cerevisiae cap mutants or vice-versa (Kawamukai et al., 1992). CAP isoforms from other species are also unable to complement S. cerevisiae AC activation (Matviw et al., 1992; Vojtek and Cooper, 1993; Yu et al., 1994; Zelicof et al., 1993) and have been argued to operate in their own species-specific signalling pathways (Hubberstey and Mottillo, 2002). The cross-species association of apparently independent signalling and cytoskeletal activities might reflect an as yet unidentified functional integration of the two roles (Vojtek and Cooper, 1993).
In addition to S. cerevisiae and S. pombe, CAP mutants have been identified and characterised in Drosophila (Baum et al., 2000; Benlali et al., 2000), in Dictyostelium (Noegel et al., 1999) and in mammals, where RNAi suppression of CAP function has been performed (Bertling et al., 2004). Phenotypes shared by these mutants are reductions in polarised cell morphology and cell motility coinciding with disorganisation of actin-rich structures. At the level of tissue organisation the cap phenotypes reveal a requirement for CAP in multicellular developmental signalling pathways. In Dictyostelium, CAP is required to perpetuate the cAMP relay signal to organise fruitbody formation (Noegel et al., 2004), and in Drosophila CAP is essential for Hedgehog-mediated eye development (Benlali et al., 2000).
Homologues of CAP have been identified in plants (Barrero et al., 2002; Kawai et al., 1998). The single Arabidopsis isoform has been shown to have the ability to bind actin and to complement the cytoskeletal defects of CAP-deficient yeast (Barrero et al., 2002), which suggests that plant CAP proteins have the potential to regulate the actin cytoskeleton, but the endogenous role of CAP in plant cells has remained uncharacterised.
The plant actin network is required for a variety of processes including the regulation of transpiration, pathogen defence responses, and (most visibly) growth and development (reviewed by Hussey et al., 2006). Disruption of actin polymerisation by drugs (Baluska et al., 2001), and by some loss-of-function, gain-of-function and misexpression actin mutants (Gilliland et al., 2002; Kandasamy et al., 2002; Nishimura et al., 2003) results in dwarf plants with restricted and uncoordinated cell expansion phenotypes. Sequenced plant genomes contain homologues of many ABPs, some of which have been shown to modulate actin behaviour in planta. With the exception of AIP1 (Ketelaar et al., 2004a), most plant ABP mutants and suppression constructs affect the morphogenesis of only a variable subset of cell types. The tissue-specific nature of formin phenotypes (Deeks et al., 2005; Ingouff et al., 2005; Yi et al., 2005) and profilin (McKinney et al., 2001) can be considered to be a symptom of large gene families with the potential for genetic redundancy, but the relatively mild phenotypes of components of the Arp2/3 complex (Mathur et al., 2003) together with the unexpectedly severe Arabidopsis AIP1 phenotype suggests that plants place functional emphasis upon individual classes of ABPs in a pattern that differs from animals and fungi. Here, we show that the Arabidopsis homologue of CAP (CAP1) is essential for the development of multiple cell types and that null mutant phenotypes of these tissues correlate with actin organisational defects. Moreover, deactivation of CAP1 alters the growth behaviour of multiple organs in a novel fashion resulting in curled inflorescences and meandering roots consistent with CAP1 contributing to the function of plant-specific signalling pathways.
Disruption of CAP1 affects plant development
The biological role of the actin-binding protein CAP1 was investigated through the characterisation of T-DNA insertion alleles SALK_112802 (designated cap1-1) and GABI-KAT 453G08 (cap1-2; Fig. 1). Plants homozygous for the insertion alleles were identified among segregating populations using genomic PCR. RT-PCR designed to amplify full-length CAP1 demonstrated an absence of CAP1 transcript in cDNA generated from cap1-1 homozygote and cap1-2 homozygote RNA templates (Fig. 2A). No truncated CAP1 mRNA was detected in mutant plants.
Plants homozygous for either cap1-1 or cap1-2 showed a consistent co-segregating pleiotropic phenotype (n=70 and 65, respectively). The absence of the phenotype in heterozygotes and F1 plants derived from backcrosses defines the cap1-1 and cap1-2 alleles as recessive. F1 plants from crosses between cap1-1 and cap1-2 homozygotes confirmed allelism. All identified aspects of the phenotype are present in both cap1-1 and cap1-2 homozygote plants. The influence of maternal genotype on seedling and plant growth behaviour was negligible, as plants descended from outcrosses with cap1 homozygote plants as the maternal parent appeared normal. Moreover, cap1 homozygote plants descended from heterozygote parents exhibited all phenotypic aspects.
Mutant plants homozygous for cap1 alleles have severely reduced stature (Fig. 2B). Rosette diameters of mutant plants measured at 22 days after germination (DAG) are reduced compared with wild-type controls (20.7, 14.3 and 15.3 mm for WT, cap1-1 and cap1-2, respectively; n>30 for all lines) although the mean number of rosette organs is equal. Root growth is also impaired in cap1 seedlings, with a 44% reduction in primary root length compared with wild-type plants after a 5-day growth period on vertical plates. Wild-type and cap1 plants grown in parallel initiated inflorescences simultaneously but differed in rates of inflorescence growth (Fig. 2B). At 35 DAG wild-type, cap1-1 and cap1-2 inflorescences measured a mean height of 139.9, 89.9 and 90 mm, respectively. Inflorescences of cap1 plants produce floral buds at a slower mean rate than wild-type inflorescences, contributing to height differences. Epidermal peels taken from synchronous stem internodes of cap1 and wild-type inflorescences show a reduction in cell elongation (Fig. 2C,D).
Mutant cap1 pollen grains show reduced fertility
Comparison of microarray expression analysis experiments highlights maturing pollen grains as a major site of CAP1 expression. The viability of pollen with mutant cap1 alleles was assessed in vitro. Pollen grains and the tubes they produce provide a convenient model to study highly polar growth processes. Pollen derived from mutant plants showed a reduction in the rate of germination after 24 hours of incubation in growth medium when compared to wild-type pollen (Fig. 3A-C). The growth rates of tubes successfully produced by mutant pollen grains were compared with wild-type tubes 5 hours after the initiation of germination (Fig. 3E) and were found to grow at a mean speed of approximately 1.0 μm per minute, almost one-third of the rate of wild-type growth (2.8 μm per minute). After 24 hours of growth in vitro mutant pollen tubes do not reach the same terminal lengths as wild-type tubes (Fig. 3D).
The growth phenotype of cap1 pollen tubes was confirmed in vivo using pollen grains germinated on intact flowers. Emasculated wild-type stigmas were fertilised with either wild-type or cap1 pollen and after a minimum of 4 hours were dissected, fixed and stained with aniline blue to assess pollen viability. After 4 hours, wild-type pollen tubes had traversed most of the style, whereas mutant pollen tubes had yet to penetrate the stigma (Fig. 4). Mutant tube penetration was beginning to occur by 5.5 hours, by which time pioneering wild-type tubes were in contact with ovules (Fig. 4). Flowers observed 24 hours after pollination showed that mutant pollen tubes were capable of eventually reaching ovules after a sufficient growth period.
If the poor viability of mutant pollen is due to the debilitating effect of the male gametophyte inheriting a cap1 mutant allele, then the transmission frequency of cap1 alleles within segregating populations is likely to be reduced. The frequency of F2 cap1-1 and cap1-2 homozygotes from F1 heterozygote parents is low (3% compared with an expected value of 25%). Progeny of heterozygote plants were not observed to have an increase in mortality to explain the absence of homozygotes, which suggests that a fertility problem was causing the low frequency of cap1 mutants. Genotyping of F1 plants generated by crosses between wild-type plants and cap1 heterozygotes shows that the fertility of cap1 pollen is reduced: wild-type plants fertilised with pollen from cap1 heterozygote plants show a cap1 transmission frequency of 1.2% (n=85). By contrast, the reciprocal cross shows a transmission frequency of 44% (n=41). This indicates that pollen grains with a cap1 genotype display a fertilisation handicap when competing against wild-type pollen generated by the same parent, as an ovule is nearly 50 times more likely to be fertilised by a pollen grain inheriting a wild-type allele.
F-actin is disrupted in cap1 mutants
Mutant cap1 lines were crossed with plants carrying GFP:FABD2, a construct consisting of the second actin-binding domain of fimbrin fused to GFP under the control of the CaMV 35S promoter, to identify possible actin cytoskeletal disruption associated with the developmental abnormalities of cap1 mutants. Observing the pollen tube cytoskeleton with GFP:FABD2 was found to be prohibited as the CaMV 35S promoter does not stimulate expression within the gametophyte. Instead, we compared the actin arrays of root hairs, a sporophytic tissue used as a model for tip growth. Root hairs of cap1 homozygotes are severely shortened, bulbous, waved and occasionally branched when compared with wild-type root hairs grown in equivalent conditions (Fig. 5A,B). Mean growth speed is reduced to 0.36 μm per minute for mutant hairs compared with 0.79 μm per minute for wild-type hairs.
Elongating wild-type root hairs contain a population of longitudinally oriented actin cables within the shank of the hair that disperse approximately 10 μm from the growing tip (Fig. 5C). Dynamic F-actin populations at the tip are essential for directing and facilitating Golgi-derived vesicle fusion to the plasma membrane of the growth zone. Induced stabilisation of F-actin and subsequent invasion of the apical clear zone by actin bundles has been found to correlate with cessation of growth (Ketelaar et al., 2004a; Ketelaar et al., 2004b). Growing hairs lacking CAP1 contain short F-actin bundles that congregate at the cell cortex (Fig. 5D). These F-actin bodies often appear as aggregates rather than defined bundles (Fig. 5C,D). Frequently, the cortical F-actin aggregates extend to the very tip of growing cap1 root hairs (Fig. 5E,F). The absence of organised long actin cables coincides with an apparent reduction in long range transport. Labeling of mitochondria using mitotracker red showed that the wild-type pattern of reverse fountain streaming of organelles (supplementary material Movie 1) is not apparent in immature cap1-1 and cap1-2 root hairs (supplementary material Movie 2). Growing mutant root hairs often contain large zones of cytoplasm between the growing tip and central vacuole that are depleted in F-actin relative to the accumulations at the cortex (Fig. 5D).
Morphological abnormalities associated with F-actin disruption also occur in cap1 cell types where expansion is localised to `diffuse' zones of cell wall. The epidermal cells of cap1 inflorescences are shorter with respect to the longitudinal axis of the inflorescence than wild-type cells (Fig. 2C,D) and contain a relatively sparse population of poorly aligned F-actin bundles. Trichome cells of the leaf epidermis also grow in a diffuse manner (Schwab et al., 2003) and are sensitive to disruptions in actin turnover (Mathur et al., 1999; Szymanski et al., 1999). Arabidopsis leaf trichome cells exposed to actin depolymerising drugs or produced by plants homozygous for null alleles of components of the Arp2/3 complex and associated signalling pathway display a `distorted' phenotype consisting of bloating and twisting of trichome stalks and branches. Trichomes from cap1 plants display a weak distorted phenotype: cap1 trichome branches are mildly twisted, and stalk inter-branch zones are often excessively elongated (Fig. 6A,B). The angle between trichome branches is also affected in a large proportion of trichomes that otherwise would have a wild-type appearance (Fig. 6A-C). Comparison of developing trichomes expressing GFP:FABD2 identified frequent excessive accumulation of F-actin in the core of elongating branches (Fig. 6D,E) that does not resemble the cohesive network of longitudinally aligned cables observed in the wild-type. This unusual F-actin array correlates with branches with diameters broader than those of wild-type branches of a comparable age. The phenotype is prevalent in trichomes between developmental stages 4 and 5, where branches are undergoing rapid expansion, but excessive F-actin accumulation within central cytoplasmic regions can be identified in trichomes as young as developmental stage 2. The central bundles of cap1 trichomes show some resistance to the action of the actin-deploymerising drug latrunculin B (data not shown), which possibly explains the absence of enhanced sensitivity of cap1 trichome morphogenesis to latrunculin B treatment (see supplementary material Fig. S2). The redistribution of F-actin in cap1 mutant trichomes is the reverse of microfilament redistribution in mutant root hairs, which suggests fundamental differences in the cytoskeletal organisation of tip growing and diffusely growing plant cells.
Synthetic phenotypes are produced between CAP1 alleles and the Arp2/3 complex pathway
The cap1 mutant alleles were crossed with previously characterised null alleles of SCAR2 and ARP2 (Basu et al., 2005; Le et al., 2003) to search for novel synthetic phenotypes that would reveal new functional roles for CAP1 and the Arp2/3 pathway during plant development. Null alleles of the Arp2/3 activator SCAR2 display a weak distorted trichome phenotype (Basu et al., 2005; Zhang et al., 2005). The scar2-1/cap1-1 and scar2-1/cap1-2 double homozygotes show an enhanced trichome phenotype with increased trichome distortion greater than either single mutant (Fig. 7A), which suggests that the actin-binding proteins SCAR2 and CAP1 act in parallel to control trichome branch expansion. No further synthetic phenotypes were detected in other tissues. The distorted shapes of arp2-1/cap1-1 and arp2-1/cap1-2 trichomes are similar to those of arp2 single mutants making ARP2 epistatic with respect to CAP1 during trichome development. This is expected as mutants of Arp2/3 complex components resemble trichomes grown in the presence of high concentrations of actin depolymerising drugs, and therefore may represent an actin phenotypic zenith of morphological distortion. Double mutants of either arp2-1/cap1-1 or arp2-1/cap1-2 showed a greater reduction in plant stature and a more severe inhibition of root hair elongation (Fig. 7B). The root hairs of the double mutant are arrested at the stage of bulge expansion at the surface of the root epidermis and do not initiate tip growth. Drug studies have previously shown that bulge formation is not as dependent upon the actin cytoskeleton as the later stage of tip growth (Baluska et al., 2000) explaining the arrest of the ABP double at this stage. Under standard growth conditions arp2 root hair development has previously been described as normal (Mathur et al., 2003) and no morphological abnormalities were identified in arp2-1 single mutant root hairs grown as controls in parallel with double mutants. Despite the absence of severe synergistic phenotypes in the double mutants, the additive arp2/cap1 root hair phenotype shows that the Arp2/3 pathway has a fundamental role in root hair tip growth that is only revealed in a cap1 background.
CAP mutants are affected in co-ordination of organ growth
In addition to being retarded in length, cap1 inflorescences curl during bolting and exhibit alterations in the direction of expansion that create `kinks' in the stem (Fig. 8A compared with Fig. 8B). The changes in growth angle occur at nodes, creating corners at points of lateral organ development. Zigzagging of the inflorescence has been reported in gravitropic mutants, but cap1 inflorescences remain gravitropic. Also unlike gravitropic mutants, cap1 secondary inflorescences regularly achieve 360 degree loops relative to the vector of primary inflorescence expansion (Fig. 8B). The looping process begins with secondary inflorescence heads bending downwards the gravity vector. At any one moment in time 37% of cap1 inflorescence heads are at an angle lower than the gravitational horizon (n=307). In the same environmental conditions only 1% of wild-type inflorescence heads grow at an angle below the same threshold (n=279). 6.5% of cap1 inflorescence heads over a period of 7 days achieved a complete 360 degree rotation relative to the axis of their own stem. The inflorescence rotation rarely exceeds one complete loop and is a temporary phenomenon; affected inflorescences uncurl hours to days after loop completion (supplementary material Movie 4). The pedicles of floral organs are also susceptible to curling (Fig. 8B) but these distortions are permanent. Time-lapse recording of wild-type and cap1 plants revealed that cap1 inflorescences do not undergo rotational circumnutation movements (supplementary material Movies 3, 4) but instead cap1 inflorescence-heads oscillate at irregular intervals within the vertical plane. An analogous phenotype can be observed in roots; cap1 roots are unable to grow in a straight line across the horizontal surface of agar medium (Fig. 8D,F), yet remain gravitropic. Microscopic analysis did not reveal twisting of epidermal cell files within affected organs, and the chirality of inflorescence curls occurs randomly between individual plants and between inflorescences of the same plant. These aspects of the cap1 phenotype indicate a loss of coordination in organ expansion. Bending of organs is achieved in plants by simultaneous differential expansion of opposed cell layers. Loss of circumnutation movements and initiation of novel curling motion can result from either corruption of growth signals or the interpretation of these signals in target tissues indicating an involvement of CAP in as yet unknown plant signalling pathways.
Endogenous CAP1 is required for Arabidopsis actin organisation
Disruption of the CAP1 gene causes phenotypes in multiple tissues that correlate with disturbances in the actin cytoskeleton. Elongating epidermal cells, tip growing cells, and trichomes show severe morphological abnormalities and unusual aggregation of F-actin. We have shown that the Arabidopsis CAP homologue is an actin monomer binding protein (supplementary material Fig. S1), and recent work has shown that CAP1 directly accelerates the exchange of ADP for ATP by actin monomers (Staiger and Blanchoin, 2006; Chaudhry et al., 2007) filling a functional space in actin biochemistry left by the absence of plant profilin nucleotide exchange activity. Drug studies have long shown that cap1 plant phenotypes are compatible with a suppression of actin biochemistry – sequestering, capping, or stabilising actin (with latrunculin B, cytochalasin D, or jaspokinolide, respectively) produces a similar suite of defects. A previous study has shown that the overexpression of plant CAP disassembles F-actin arrays in vivo and causes severe growth defects (Barrero et al., 2002), which is surprising when considering that the overexpression of CAP in other organisms does not result in gross phenotypic abnormalities. We have established that this potential for CAP1 to influence actin biochemistry is affirmed by the function of endogenous plant CAP in actin-dependent growth processes.
Reconciling CAP1 biochemistry with the CAP1 phenotype
The visible disruption to the Arabidopsis actin cytoskeleton resulting from absence of CAP1 consists of an F-actin re-arrangement into short bundles or aggregates that retain an unusual position within their respective cells types. Short bundles congregate to form a dense actin `core' in expanding trichome branches while in growing root hairs actin bundles diminish and withdraw to the cell cortex. The formations of F-actin adopted in cells lacking functional CAP vary from organism to organism. In S. cerevisiae the appearance of the F-actin arrays of dividing cells undergoes only a subtle alteration, as both actin patches and cables remain intact. Actin patch distribution is perturbed (Field et al., 1990) and the ASH1 mRNA polar marker is not anchored correctly after being transported along actin cables (Baum et al., 2000) possibly indicating a subtle cable defect. Animal cells with knocked down CAP amass large arrays of stable F-actin and lose dynamic F-actin arrays associated with lamellipodia (Baum et al., 2000; Benlali et al., 2000; Bertling et al., 2004; Rogers et al., 2003). These re-arrangements can be interpreted as evidence supporting biochemical observations that some CAP isoforms can sequester actin monomers in vitro and thus prevent excessive polymerisation (Freeman et al., 1995; Gieselmann and Mann, 1992; Gottwald et al., 1996). Plant F-actin does not homogenously accumulate in the absence of CAP as cap1 root hairs appear devoid of the organised bundles that amass in the shank of wild-type root hairs behind the growing tip. These observations suggest a more complex role for CAP1 in the turnover of actin filaments, possibly relating to the observed in vitro activity of accelerating actin monomer nucleotide exchange either directly (Moriyama and Yahara, 2002; Chaudhry et al., 2007) or through interplay with ADF (Mattila et al., 2004; Chaudhry et al., 2007).
The critical state of actin dynamics depends upon the balance of actin-binding protein activity. Overexpression of both plant CAP1 and AIP1 (Barrero et al., 2002; Ketelaar et al., 2007) mimic the phenotypic effects of reduced expression of these respective proteins (this study) (Ketelaar et al., 2004a). A sub-nominal level of CAP1 protein is likely to reduce the in vivo concentration of ATP-actin monomers, while an increase in CAP1 concentration or activity could enhance CAP1 sequestration of actin monomers from the G-actin pool. Any imbalance in ABP activity impacts upon the efficacy of actin turnover and, consequently, on cytoskeletal-driven cell growth.
Arabidopsis CAP1 is required for signalling
The curling of cap1 inflorescences and roots indicates a role for CAP1 in coordinating the expansion of tissues. The curling of cap1 organs is distinct from the twisting observed in mutants such as spiral or lefty, which show defects in microtubule organisation as the cap1 curls have no consistent chirality, and the turns of roots are not associated with visible twisting of epidermal cell files. However, the inflorescence curling phenomenon has a common pattern of behaviour: cap1 inflorescences always initiate a curl by turning towards the gravity vector. Recovery to a vertical position closes the curl, and the process is then rapidly reversed over the course of a few hours to re-straighten the inflorescence. Such movement requires the simultaneous differential expansion of many cell files, suggesting the involvement of an intercellular signal that is either misinterpreted or misdirected in the absence of CAP1.
The interpretation of mutant phenotypes in understanding the signalling role of CAP is complicated by the multiple consequences of cytoskeletal disturbance. Clonal analysis of Drosophila eye discs shows a requirement for CAP in signalling processes to organise photoreceptor differentiation (Benlali et al., 2000). Rather than directly transducing a signal in the manner of SRV2, Drosophila CAP was hypothesised to perturb hedgehog signalling by causing morphological abnormalities across the surface of the eye disc leading to the physical disruption of morphogen distribution. Therefore dissection of the plant CAP signalling phenotype should always be considered in light of the effects of actin disruption on morphogenesis and other basal cell processes such as exocytosis and endocytosis. Attempting to separate signalling and cytoskeletal activities through domain analysis must also be approached with caution. In S. cerevisiae, where this analysis was first performed, the `cytoskeletal' phenotypes were never fully complemented using only the C-terminus (Gerst et al., 1991) and recently the transduction of signals from Ras to adenylate cyclase within a cell-death pathway was found to be more reliant upon the C-terminus of SRV2 than the N-terminus (Gourlay and Ayscough, 2006).
The two biological activities of CAP have long been considered independent functions, but evidence is accumulating that interaction with actin monomers could be coupled with participation in signalling pathways. In Dictyostelium, CAP is required to perpetuate the cAMP chemotactic signal (Noegel et al., 2004) and to respond to the same signal by stimulating cytoskeletal based motility (Noegel et al., 1999). Recently, SRV2 of yeast was shown to have a remarkable association with the cytoskeleton during Ras-mediated signalling to apoptotic-like pathways (Gourlay and Ayscough, 2006). Suppression of actin dynamics leads to Ras activation, which in turn activates adenylate cyclase. One of the consequences of this pathway is further actin rearrangements, possibly via PKA effectors downstream from cAMP signalling (Gourlay and Ayscough, 2006). CAP is needed to transduce the signal from Ras to adenylate cyclase, and the actin-binding domain of CAP is required for this process. In this instance CAP appears to offer an input to the pathway dependent upon the actin-binding domain, even feasibly acting as some form of sensor to the free G-actin pool. As the effectors of the signalling pathway are very likely to include other actin-binding proteins, the cytoskeletal phenotypic effects of CAP knockouts in yeast are dependent on CAP signalling activities, totally integrating the two roles and confusing any phenotypic distinction.
The concept of integration invites a model for CAP function based upon feedback from the actin cytoskeleton. Latrunculin B treatment is known to exaggerate the gravitropically stimulated bending of roots (Hou et al., 2004) through an uncharacterised mechanism. The roots of cap1 plants also respond in an exaggerated fashion to gravistimulation (P.J.H., unpublished). The meandering behaviour of cap1 roots and the curling of inflorescences might result from an overcompensation to internal cues aimed at maintaining a controlled angle of tissue expansion. CAP1 could feasibly be required to monitor and respond to the status of the G-actin pool in expanding cells. Rapid dynamics could provoke the perpetuation of an intercellular compensatory signal via CAP1 to expand other cell files in an antagonistic manner and stimulate direct intracellular suppression of actin turnover.
In conclusion, Arabidopsis CAP1 is essential for healthy growth, but unlike AIP1 its deactivation is not lethal to plant life. The cap1 mutants could reveal aspects of the in vivo behaviour of other ABPs, both through the observation of F-actin formation in affected tissues and through double-mutant analysis. The accumulation of F-actin at the cortex of affected cells suggests the presence of so far unidentified F-actin machinery at sites of intense exocytosis. Arabidopsis cap1 is also required for the coordination of tissue expansion in the manner of a component of an intercellular signalling pathway.
Materials and Methods
Recombinant protein purification
For actin-binding studies, the full-length transcript of CAP1 was cloned into a Gateway derivative of pGEX-4T-1 to form a translational fusion with GST at the N-terminus of CAP1. Recombinant protein was expressed in E. coli strain BL21 DE3 pLysS Rosetta 2. Cultures were grown to OD600 0.8 and induced using IPTG (final concentration 1 mM) at 37°C for 2.5 hours or overnight at 14°C. Cells expressing GST-tagged proteins were resuspended in PBS, pH 7.3 and lysed using a freeze-thaw cycle. Cell supernatant was incubated with glutathione sepharose 4B according to the manufacturer's instructions (Amersham, UK) and washed three times with PBS.
Rabbit skeletal muscle actin was purified as described previously (Spudich and Watt, 1971), and as later modified (Winder et al., 1995). Briefly, rabbit muscle acetone powder was mixed with buffer G (2 mM Tris pH 8.0, 0.2 mM ATP, 0.5 mM DTT, 0.2 mM CaCl2, 1 mM NaN3). After a 30 minute incubation and spin, the supernatant was filtered. MgCl2 was added to a final concentration of 2 mM and KCl was added to 0.8 M. After polymerisation the F-actin was pelleted for 2 hours at 50,000 g. Actin was resuspended in G-buffer and dialysed for 2 days, then centrifuged at 50,000 g. The top two-thirds of the supernatant was gel-filtered using sephacryl S300 to isolate actin monomers.
ADP-actin was generated from purified ATP-actin monomers on the day of use by incubation with yeast 20 U/ml hexokinase and 1 mM glucose in G-buffer for 3 hours (Pollard, 1986). 0.1 M ADP (Sigma) stock solution was also treated with hexokinase and glucose to remove ATP contaminants.
For actin-depletion assays, G-actin was added to 225 μl of depletion buffer (10 mM Tris pH 7.5, 0.2 mM CaCl2, 0.5 mM DTT, 0.2 mM ADP, 1 mM MgCl2, 100 mM KCl) to make a final solution of 3 μM. 25 μl (12.5 μl bead volume) of glutathione sepharose beads (Amersham) coated with either GST or GST-CAP1 were immediately added to the actin solution to make a 9 μM bait protein suspension. Beads were incubated with the actin for either 5 minutes or 30 seconds with gentle agitation. Following incubation, beads were briefly spun to the bottom of the tube and 100 μl of supernatant removed and mixed with 2× SDS loading buffer. Samples were run on an 8% polyacrylamide SDS gel and stained using Coomassie solution. Native gels were polymerised at a final concentration of 10% Protogel acrylamide/bisacrylamide mix (National Diagnostics). 0.2 mM ADP (pre-treated with 20 U/ml hexokinase and 1 mM glucose) was present in both gel and running buffer (25 mM Tris, 200 mM glycine, 0.5 mM DTT). Recombinant GST-CAP1 and GST were removed from glutathione beads using elution buffer (10 mM reduced glutathione, 50 mM Tris-HCl, pH 8.0) and dialysed for 5 hours with 4 changes of G-buffer. Combinations of ADP-actin (final concentration 1 μM), GST-CAP1 (5 μM) and GST (5 μM) were assembled in G-buffer with a total volume of 20 μl and incubated on ice for five minutes before loading onto the native gel.
Arabidopsis seed was sterilised using 5% bleach (BDH) for 25 minutes with gentle agitation followed by 4 washes with water. Seeds were plated on to half-strength Murashige and Skoog salts (Sigma) with 0.8% plant agar. After germination all plants were grown either on compost or half MS plates in 16 hours light (at 22°C) and 8 hours dark (at 18°C). Salk T-DNA lines (Alonso et al., 2003) were created by SIGNAL (the Salk Institute Genomic Analysis Laboratory) and supplied by NASC (Nottingham, UK). GABI KAT line 453G08 (Rosso et al., 2003) was created and supplied by the Max Planck Institute for Plant Breeding Research (Cologne, Germany).
Wild-type and cap1 pollen was germinated in vitro as previously described (Krishnakumar and Oppenheimer, 1999). 100 μl aliquots of agarose pollen germination medium [1 mM CaCl, 1 mM Ca(NO3)2, 1 mM MgCl2, 0.01% boric acid, 18% sucrose, 0.5% agarose, pH 6.0] were solidified on microscope slides to produce a smooth coating. A 5 μl drop of liquid germination medium (without agarose) was applied to the centre of the slide, and mature wild-type or cap1 pollen was released into the liquid from open pollen sacs. Two mature pollinated stigmas were placed on the slide within 5 mm of the samples. Samples were left either for 5 hours or overnight in a closed humid environment within standard growth room conditions. After application of coverslips, pollen tubes were observed using a Zeiss Axioskop microscope with 40× objective, and images were captured using a video camera (Roper Scientific) controlled by Openlab 3 software (Improvision, UK).
For in vivo growth assays, wild-type and cap1 pollen was used to fertilise WT stigmas. After a specific period of germination within standard growth room conditions, stigmas were prepared as described (Jiang et al., 2005) with minor modifications. Fertilised carpels were dissected longitudinally to bisect the septum. The dissected tissue was fixed in a 3:1 ethanol:acetic acid solution for 2 hours. The samples were then washed with water and incubated with 10 M NaOH for 2 hours. Tissue was subsequently washed 3 times with water and 3 times with 100 mM K2HPO4-KOH, pH 11. Samples were incubated in the dark for 2 hours in 0.1% aniline blue in 100 mM K2HPO4-KOH, pH 11 before being mounted in glycerol and observed with a Zeiss LSM510 confocal microscope under 405 nm blue diode laser excitation.
Inflorescence epidermal peels were imaged using an Eclipse TE300 inverted microscope (Nikon, Japan) with Orca camera (Hamamatsu, Japan). All imaging of GFP:FABD2 fluorescence was performed using a Zeiss LSM510 confocal microscope with 40× objective. For mitochondrial imaging, growing root hairs were labelled with 250 nM mitotracker red CMXros (Invitrogen). For microscopic analysis of growing root hairs, seedlings were grown in `biofoil sandwiches', as described previously (Ketelaar et al., 2004a). Images of various organs were taken using a SZH10 stereomicroscope (Olympus) and video camera (Roper Scientific).
DNA from plants was prepared as described by (Edwards et al., 1991). All PCR experiments used RedTaq (Bioline) polymerase in accordance with manufacturer's instructions. The CAP1 wild-type allele was amplified using primers:
CAP1-1 For: 5′-CTCCTGACTTCGCCATC-3′
CAP1-1 Rev: 5′-GCAACCTGAACAAGTAACTATA-3′
CAP1-2 For: 5′-CTCGTACCAGTAAACCGGCCTT-3′
CAP1-2 Rev: 5′-CCGTGAAAACAACACCCACTTT-3′.
The amplification of the T-DNA insertion alleles was achieved with primers:
CAP1-1 Rev and LBb1 (5′-GCGTGGACCGCTTGCTGCAACT-3′)
CAP1-2 Rev and cap1-2TDNA (5′-CCCATTTGGACGTGAATGTAGA-3′).
Total RNA was isolated from homozygous mutant and azygous plants using the RNeasy Plant Mini Kit (Qiagen). The total RNA was DNase treated with RNase-free DNase (Promega) following the manufacturer's instructions. The cDNA first-strand was synthesised using 5 pmol of oligo(dT) 12-18mer and 200 units of Superscript™II RNaseH-Reverse Transcriptase (Invitrogen), according to the manufacturer's instructions. The RNA was removed by addition of 2 units of Ribonuclease H (Promega) and incubation at 37°C for 20 minutes. 1 μl of the reaction mixture was used as a template for the PCR reaction with BioRed Taq DNA polymerase (Bioline).
Control primers to Arabidopsis glyceraldehyde-3-phosphate dehydrogenase C were:
Environmental scanning electron microscope (ESEM) preparation
The plant material (leaves from homozygous mutant and azygous plants, from each T-DNA line) was prepared for ESEM by fixation (PBS, 1% glutaldehyde, 0.1% (v/v) Tween) and dehydration by an ethanol series, followed by critical point drying with carbon dioxide. Samples were imaged using a Philips XL30 ESEM in low vacuum mode (0.4 Torr).
This work was supported by the Biotechnology and Biological Sciences Research Council (M.J.D., S.D., T.K., P.J.H.) and the FCT Portugal, SFRH/BD/8760/2002 (C.R., R.M.). We thank Steve Winder for purified actin, and Christopher Staiger and Laurent Blanchoin for personal communications concerning the biochemistry of Arabidopsis CAP1.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/120/15/2609/DC1
↵* These authors contributed equally to this work
- Accepted May 15, 2007.
- © The Company of Biologists Limited 2007