Identification of plant actin-binding proteins by F-actin affinity chromatography

Authors


*For correspondence (fax +1 336 758 6008; e-mail muday@wfu.edu).

Summary

Proteins that interact with the actin cytoskeleton often modulate the dynamics or organization of the cytoskeleton or use the cytoskeleton to control their localization. In plants, very few actin-binding proteins have been identified and most are thought to modulate cytoskeleton function. To identify actin-binding proteins that are unique to plants, the development of new biochemical procedures will be critical. Affinity columns using actin monomers (globular actin, G-actin) or actin filaments (filamentous actin, F-actin) have been used to identify actin-binding proteins from a wide variety of organisms. Monomeric actin from zucchini (Cucurbita pepo L.) hypocotyl tissue was purified to electrophoretic homogeneity and shown to be native and competent for polymerization to actin filaments. G-actin, F-actin and bovine serum albumin affinity columns were prepared and used to separate samples enriched in either soluble or membrane-associated actin-binding proteins. Extracts of soluble actin-binding proteins yield distinct patterns when eluted from the G-actin and F-actin columns, respectively, leading to the identification of a putative F-actin-binding protein of approximately 40 kDa. When plasma membrane-associated proteins were applied to these columns, two abundant polypeptides eluted selectively from the F-actin column and cross-reacted with antiserum against pea annexins. Additionally, a protein that binds auxin transport inhibitors, the naphthylphthalamic acid binding protein, which has been previously suggested to associate with the actin cytoskeleton, was eluted in a single peak from the F-actin column. These experiments provide a new approach that may help to identify novel actin-binding proteins from plants.

Introduction

The actin cytoskeleton is a dynamic filamentous network comprised of polymeric actin filaments and a myriad of associated proteins. The organization of filamentous actin (F-actin) and its assembly from monomeric globular actin (G-actin) can be dramatically altered to allow changes during growth and development and in response to a variety of stimuli ( Nick, 1999; Staiger, 2000). In eukaryotic cells, the actin cytoskeleton is required for localization of membrane proteins, cell motility, cell division, surface remodelling, control of cell shape, contraction, cell–substrate interactions, development of polarity, movement of vesicles, secretion and endocytosis ( Ayscough and Drubin, 1996; Meagher and Williamson, 1994). To understand how these processes are mediated by the actin cytoskeleton, it is necessary to identify the specific actin-binding proteins that modulate cytoskeletal organization and function ( Pollard et al., 1994 ). Numerous actin-binding proteins have been identified through genetic and biochemical approaches. Actin affinity chromatography has proven to be a powerful approach for isolating actin-binding proteins ( Miller and Alberts, 1989; Miller et al., 1991 ). This strategy has identified numerous novel actin-binding proteins from non-plant cells and confirmed the interaction of proteins suggested to bind to actin by other experimental approaches ( Adams et al., 1989 ; Aroian et al., 1997 ; Bearer and Abraham, 1999; Drubin et al., 1988 ; Field and Alberts, 1995; Miller et al., 1989 ; Terasaki et al., 1997 ).

The method that has been most successful in identification of plant actin-binding proteins has been to identify genes that have sequence similarity to actin-binding proteins from other organisms ( de Ruijter and Emons, 1999). Such an approach has lead to the identification of a number of proteins that have actin remodelling capability, including profilin ( Staiger et al., 1993 ; Valenta et al., 1993 ), actin-depolymerizing factor (ADF) ( Kim et al., 1993 ; Lopez et al., 1996 ), fimbrin ( Cruz-Ortega et al., 1997 ; McCurdy and Kim, 1998) and villin ( Klahre et al. 2000 ; Vidali et al., 1999 ). Another protein, with a potentially important role in connecting cell signalling to the actin cytoskeleton, is a Rho-like GTPase that was also identified by this approach ( Yang and Watson, 1993). Although the growing availability of plant DNA sequence information will facilitate the identification of additional plant actin-binding proteins, this approach may not be useful for identification of actin-binding proteins that are unique to plants.

Biochemical fractionation has lead to the identification of several plant actin-binding proteins. Several proteins have been suggested to be associated with the actin cytoskeleton based on their partitioning with actin during detergent extraction. These include phosphatidylinositol kinases ( Tan and Boss, 1992; Xu et al., 1992 ) and a protein that binds inhibitors of auxin transport, including napthylphthalamic acid (NPA) ( Butler et al., 1998 ; Cox and Muday, 1994). Both of these proteins have also been shown to bind actin filaments through cycles of depolymerization and repolymerization ( Cox and Muday, 1994; Tan and Boss, 1992); however, microtubules may cycle in these procedures as well ( Cox and Muday, 1994). A related technique to identify actin-binding proteins is co-sedimentation with purified animal actin. This approach has lead to the identification of plant myosin ( Yokota and Shimmen, 1994) and actin-bundling proteins ( Nakayasu et al., 1998 ; Yokota and Shimmen, 1995; Yokota et al., 1998 ), and has been used to demonstrate the actin interaction of sucrose synthase ( Winter et al., 1998 ). Although this technique provides initial evidence for actin-binding protein activity, additional chromatography steps and characterization are required to confirm this function ( Nakayasu et al., 1998 ; Yokota et al., 1998 ).

Actin affinity chromatography combines the strengths of actin co-sedimentation with the potential for selective protein interactions conveyed by chromatography. For example, the ability to vary salt or buffer conditions to selectively elute proteins from an affinity column offers the potential to identify proteins with variable affinities for G- or F-actin. Therefore, this approach can both identify potential actin-binding proteins and purify these proteins. There are no reports in the literature on the use of actin affinity chromatography to identify plant actin-binding proteins, although this approach has been used to show that plant annexins will bind to animal F-actin columns ( Calvert et al., 1996 ).

This paper outlines the procedures that were used to purify plant actin, to demonstrate that the pure protein is functional, and to construct actin affinity columns. The ability of the F-actin and G-actin affinity columns to differentiate between known F- and G-actin-binding proteins verified that these columns have the appropriate function. The selective binding of the NPA binding protein to an F-actin column provides additional evidence that this auxin transport protein interacts with the actin cytoskeleton, and suggests a mechanism by which the asymmetric distribution of this auxin transport protein may be maintained. Finally, these experiments have resulted in the identification of an unknown soluble actin-binding protein of approximately 40 kDa that will be the target of further investigations.

Results

Actin purification from zucchini hypocotyls

Zucchini (Cucurbita pepo L.) hypocotyl G-actin was purified on a DNAase I affinity resin with a modified version of a protocol that was used to partially purify pea root actin ( Andersland et al., 1992 ). Hypocotyls were homogenized and the clarified supernatant was incubated with DNAase I coupled to Sepharose CL-4B. Actin was eluted from the DNAase I resin using 50% formamide, and the eluted proteins were immediately gel-filtered to remove the formamide, which was necessary to prevent actin denaturation. Removal of formamide by gel filtration allowed the polymerization of purified zucchini actin, which was recovered in the pellet after ultracentrifugation at 100 000 g. In Figure 1(a), the purity of the eluent from the DNAase I affinity matrix is compared to the supernatant and pellet after ultracentrifugation. Although the predominant band in the DNAase I column eluent is actin, there are clearly other contaminating proteins ( Figure 1a). The identity of the 42 kDa protein as actin was confirmed by immunoblot with a monoclonal antibody against actin ( Figure 1b). After ultracentrifugation, the actin recovered in the pellet is electrophoretically pure, as judged by a silver stain. In contrast to other reports, phalloidin was not necessary for polymerization of DNAase I-purified actin ( Andersland et al., 1992 ), although there was a slight increase in the amount of pelletable actin in the presence of phalloidin (Hu and Muday, unpublished observations). The yield of actin in the final sample recovered after ultracentrifugation was approximately 11.7 µg per g fresh weight of zucchini hypocotyl tissue, resulting in a typical yield of 3.5 mg from 300 g of hypocotyls.

Figure 1.

Purification of actin by DNAase I affinity chromatography followed by ultracentrifugation.

(a) Samples were subjected to SDS–PAGE and proteins were identified by silver stain. (b) Immunoblot of the same samples with a monoclonal antibody against actin. Samples were molecular weight standards (S), DNAase I matrix eluent, and pellet and supernatant samples after a subsequent ultracentrifugation step. The three actin samples had the same total volume and equal volumes were loaded on the gel. Actin is labelled ‘A’.

Demonstration that the purified actin is functional

Several approaches were used to demonstrate that the actin purified by DNAase I affinity chromatography was native. First, the ability of G-actin to bind to maize profilin was analysed by measuring the quenching of intrinsic tryptophan fluorescence upon complex formation. The Kd values (dissociation constants) for zucchini actin binding to profilin were determined and compared to maize pollen actin, which has been shown previously to be polymerization-competent ( Gibbon et al., 1997 ; Gibbon et al., 1998 ; Ren et al., 1997 ). The average Kd values between these two actin preparations were similar, as shown in Table 1.

Table 1.  Affinity (Kd, µm) of maize profilins for MgATP–zucchini actin and MgATP–maize pollen actin
Profilin isoformZucchini actinMaize pollen actin
  1. Values are means ± SD; the number of replicates is given in parentheses. The Kd values of each profilin isoform binding to zucchini or maize pollen actin were measured by recording changes in intrinsic tryptophan fluorescence under low-ionic strength conditions. No statistically significant differences were found between maize and zucchini actin or between the two profilin isoforms using a two-tailed t test.

ZmPRO11.73 ± 0.3 (6)1.73 ± 1.0 (3)
ZmPRO52.0 ± 0.7 (2)2.17 ± 1.3 (3)

Another measure of the activity of monomeric actin is its ability to assemble into filaments. To verify that the purified zucchini actin could polymerize, several approaches were used. First, during the purification procedure, actin was recovered in the pellet after ultracentrifugation, which was suggestive of polymerization. Second, native gel electrophoresis was used to examine the size of the actin complexes. F-actin stabilized by phalloidin will not enter the gel and should remain in the bottom of the sample well. The purified actin recovered after ultracentrifugation was polymerized in buffer containing 50 m m KCl and 5 m m MgCl2 in the absence or presence of 20 or 40 µm phalloidin. The results in Figure 2 show that, in the absence of phalloidin, the actin migrated as a low-molecular-weight band. In the presence of the phalloidin, none of the actin entered the gel, which is consistent with the formation of actin filaments. Since the same actin preparation sediments during ultracentrifugation without added phalloidin, it is surprising that by native gel electrophoresis there is no evidence of F-actin in the absence of phalloidin. Our interpretation is that there is depolymerization of actin filaments during the native gel electrophoresis and that phalloidin prevents this depolymerization.

Figure 2.

Polymerization of purified actin.

(a) Actin polymers do not move into a non-denaturing gel, but are retained at the bottom of the well, as indicated by the arrow. Purified actin was treated with the indicated concentrations of phalloidin, and its mobility is compared to that of a heat-denatured actin standard also subjected to non-denaturing electrophoresis (S). Actin is visualized by detection with an actin monoclonal antibody, and the mobility of the monomeric actin is indicated by ‘A’. (b) Purified actin polymerizes into 6.8 nm helical filaments. Purified zucchini hypocotyl F-actin was negative-stained with 2% uranyl acetate and visualized by transmission electron microscopy. Bar = 50 nm.

Finally, the ultrastructure of filaments of purified actin was examined by transmission electron microscopy of negative-stained, in vitro-polymerized F-actin. Actin isolated by DNAase I affinity and ultracentrifugation was used at 50 µg ml−1, as determined by DNAase I inhibition assay. The actin samples were negative-stained with 2% uranyl acetate. Filaments were clearly identifiable in the absence of phalloidin, as shown in Figure 2(b). The width of these filaments was found to be 6.8 nm, which corresponds to the known width of actin filaments ( Sheterline et al., 1998 ). Addition of phalloidin resulted in a higher density of filaments of similar size (data not shown). The purified F-actin also interacted with the chymotryptic S-1 fragment of myosin, resulting in decorated actin filaments that were much thicker than undecorated actin (data not shown).

Validation of actin affinity columns

After verification that the purified actin was functional, we constructed F-actin, G-actin and bovine serum albumin (BSA) affinity columns. The specificity of the F-actin column was assessed by retention of a known F-actin-binding protein, animal α-actinin. The protein was eluted with KCl salt steps, as shown in Figure 3(a). α-Actinin binds tightly to the F-actin column and the elution requires high salt concentrations. In contrast, when an equal amount of α-actinin was applied to a G-actin column, there was significantly less retention of α-actinin, as shown in Figure 3(b). The retention of some α-actinin by the G-actin column is not surprising, since there may be some F-actin bound to the column, and there may also be a weak interaction between α-actinin and G-actin. Clearly, the amount of α-actinin eluting from the F-actin column is much greater than the amount eluted from the G-actin column, even though equal amounts of actin are used to prepare these columns. This result is consistent with the F-actin column specifically retaining F-actin-binding proteins.

Figure 3.

Purified animal α-actinin binds to strongly to zucchini F-actin columns and weakly to G-actin columns.

Purified α-actinin (0.1 mg) was applied to (a) an F-actin column or (b) a G-actin column, and eluted with the indicated concentration of KCl. The samples are 0.8 µg of α-actinin, and 40 µl of each 0.5 ml fraction.

Actin affinity chromatography of soluble actin-binding proteins

The demonstration of selective binding of purified actin-binding proteins was the first test of the specificity of the F-actin and G-actin columns. Another and perhaps better test is to show that there is differential retention of proteins by the two columns in an extract that contains multiple actin-binding proteins. Zucchini hypocotyls were homogenized in a buffer containing Tris, which will depolymerize actin filaments and allow the isolation of soluble actin-binding proteins. The samples were subjected to chromatography on G-actin, F-actin and BSA columns and eluted with buffer containing steps of increasing salt, and the resulting samples are shown in Figure 4(a and b), respectively. Under these conditions, several proteins are eluted from the F-actin column, although more proteins are eluted from the G-actin column. The pattern of eluted proteins from the BSA column differs from that of the G-actin and F-actin columns (data not shown). Of the proteins eluted from the F-actin column, the most interesting is a comparatively abundant protein of approximately 40 kDa that binds selectively to the F-actin and not the G-actin column. The identity of this protein is not yet known, but there are several actin-binding proteins from other organisms in this size range ( Sheterline et al., 1998 ). Alternatively, this may represent a novel plant actin-binding protein. Attempts to identify this 40 kDa protein by amino acid sequence analysis have not yet proven successful, as the N-terminus of the protein is blocked (Brady and Muday, unpublished observations).

Figure 4.

G-actin and F-actin columns selectively retain soluble actin-binding proteins.

Extracts enriched in soluble actin-binding proteins were subjected to (a) F-actin or (b) G-actin columns and eluted with increasing KCl concentrations. Samples include a purified actin standard (A) or profilin standard and starting material (St. Mat.). The fraction numbers are indicated and correspond to samples eluted with 0.25 m KCl (fractions 1–4) and 0.5 m KCl (fractions 5–8). Equal volumes of each fraction were subjected to SDS–PAGE and identified by silver stain (a,b) or by immunoblot with a maize profilin 5 antibody (c). Fractions eluted from both columns were subjected to immunoblot with the profilin 5 antibody. The mobility of actin at 45 kDa, of an F-actin-binding protein of approximately 40 kDa and of profilin are indicated with arrows and labelled ‘A’, ‘40 kDa’ and ‘P’, respectively.

Finally, the retention of the known G-actin binding protein, profilin, was examined by immunoblot of the fractions from the F-actin and G-actin columns, as shown in Figure 4(c). There was no detectable profilin in the F-actin fractions, but profilin was eluted in the fractions from the G-actin column.

F-actin affinity chromatography of membrane-associated proteins

In order to perform F-actin affinity chromatography with an extract of membrane-associated proteins, zucchini plasma membrane vesicles were treated with 1.0 m Tris in the presence of 0.5% Triton X-100. The supernatant was diluted tenfold and loaded onto F-actin, G-actin and BSA columns. After a buffer wash, bound proteins were eluted with sequential application of salt steps. The results are shown in Figure 5.

Figure 5.

Two annexin-like proteins bind specifically to the F-actin column.

Plasma membrane vesicles were treated with 0.5% Triton X-100 and 1 m Tris, and the solubilized proteins, after ultracentrifugation, were applied to either F-actin (F), G-actin (G) or BSA (B) columns and proteins eluted with the indicated concentrations of KCl. Equal volumes of all samples were loaded onto SDS–polyacrylamide gels. (a) Silver stain of the gel. (b) Immunoblot with a polyclonal antibody monospecific for pea annexin p35. The mobility of actin is indicated by ‘A’.

There were three specific bands in the sample from the F-actin column eluted with 1.0 m KCl, as shown in Figure 5(a). The two lowermost bands had molecular weights of approximately 33 and 35 kDa, which corresponds to the size of plant annexins ( Delmer and Potikha, 1997). The 42 kDa band eluted from the F-actin column is actin, which is released from the column as actin filaments break or depolymerize during the elutions, when there is a low concentration of Tris in the elution buffer. Samples from the F-actin column were subjected to immunoblot analysis, and, as shown in Figure 5(b), the two 33 and 35 kDa proteins cross-react with a pea annexin antiserum.

Elution of NPA binding activity from the F-actin affinity column

Fractions from the F-actin column loaded with the plasma-membrane derived samples were examined for NPA binding activity. As no detectable NPA binding activity was initially found, the sample preparation was modified. Zucchini plasma membrane vesicles were treated with Triton X-100 and Tris in the presence of 10 µm NPA, then subjected to ultracentrifugation, as this treatment has been previously shown to allow the recovery of soluble NPA binding activity after actin depolymerization ( Butler et al., 1998 ). The supernatant was diluted tenfold in 5% glycerol in water with 0.5 m m ATP, and loaded onto either the F-actin or BSA affinity column. Additionally, another F-actin column was loaded with buffer rather than supernatant protein. All three columns were washed, and loaded with 5 n m[3H]NPA, followed by another buffer wash. The elution was then carried out in buffer containing 5 n m[3H]NPA with salt steps of 0.25, 0.5 and 1.0 m KCl. The presence of NPA binding activity in each fraction was determined by filtration to assess the amount of [3H]NPA that was bound to a protein, and was performed in duplicate. Background was determined by addition of 10 µm unlabelled NPA to the fractions prior to filtration, and was similar in all fractions tested. The amount of total [3H]NPA binding is plotted in Figure 6. A peak of NPA binding activity was found in the 7th and/or 8th fractions in five separate experiments. There was no peak of NPA binding activity eluted either from the F-actin column that was pre-loaded with only buffer, or from the BSA column loaded with the diluted supernatant. Samples from these peaks were subjected to SDS–PAGE and proteins were visualized by silver stain, but there was no single polypeptide that was localized to only the fractions with NPA binding activity (data not shown). These eluted fractions were quite dilute and very little protein was visible by silver stain.

Figure 6.

[3H]NPA binding activity elutes as a single peak from F-actin columns.

Zucchini plasma membrane vesicles were treated with Tris, Triton X-100 and NPA, and subjected to ultracentrifugation. The supernatant was diluted tenfold, and the sample loaded onto the F-actin column or the BSA column. The eluent fractions were filtered to measure the amount of [3H]NPA bound. The third set of samples comprised fractions from an F-actin column that was not loaded with extract. Each value represents the average and standard deviation of two replicates. Similar results were obtained from five separate F-actin columns.

Discussion

The goal of these experiments was to identify plant actin-binding proteins using G- and F-actin affinity chromatography. These experiments focused on the identification of novel proteins, as well as verification of the actin association of two plasma-membrane associated proteins, annexins and the NPA binding protein, which had been suggested by other experimentation to bind actin. The choice of plant material for this analysis was carefully considered. Although many investigators have used pollen as a starting material for biochemical analyses of actin and actin-binding proteins, there are likely to be differences in the types or abundance of actin-binding proteins in other tissues. Because distinct actin isoforms are expressed in different tissues and these isoforms may have functional differences ( Kandasamy et al., 1999 ), it may be important to use actin from the tissues that will be used for identification of actin-binding proteins. Hypocotyls are likely to be a good tissue for identification of actin-binding proteins that control growth dynamics, because they are undergoing rapid elongation that is modulated by light and hormones. Etiolated zucchini hypocotyls are available in large quantities, are free of chlorophyll, and procedures for isolation of plasma membranes have been defined ( Muday, 2000). Finally, this is the tissue in which biochemical evidence linked the NPA binding protein to the actin cytoskeleton ( Butler et al., 1998 ).

To identify plant actin-binding proteins, actin affinity columns were prepared using purified zucchini hypocotyl actin. Before columns could be prepared, it was necessary to develop a procedure to purify actin from zucchini and demonstrate that the actin was functional. There are numerous reports in the animal literature of the application of affinity columns prepared with the enzyme DNAase I coupled to a solid support ( Sheterline et al., 1998 ; Zechel, 1980). DNAase I binds G-actin at a 1:1 ratio with high affinity, is commercially available, and has previously been used to partially purify actin from pea roots ( Andersland et al., 1992 ).

Actin was purified to electrophoretic homogeneity from zucchini hypocotyls using DNAase I chromatography followed by ultracentrifugation. As actin was eluted from the DNAase I resin with formamide, which can denature proteins, it was particularly critical to demonstrate that this actin was native. Ren et al. (1997) purified maize pollen actin using the ability of profilin–actin complex to bind to a poly- l-proline affinity matrix. Profilin binds G-actin with a high affinity (Kd values of 0.2–2 µm; Kovar et al., 2000 ), and this can be used for purification of G-actin as well as for quantification of the activity of purified actin to show that it is functional. The profilin-binding ability of purified maize pollen actin was compared with that of purified zucchini hypocotyl actin, using two isoforms of maize profilin. The resulting average Kd values for these two actin pools were not statistically different under these conditions. This similarity is striking considering the large differences in the two isolation procedures and the tissues from which the actin was purified.

The ability of zucchini actin to form filaments was also demonstrated by sedimentation of F-actin during ultracentrifugation, decreased mobility of F-actin on native gels, and examination of actin filaments with electron microscopy. Ultrastructural examination of in vitro-polymerized actin lead to the identification of helical filaments with a width of 6.8 nm, consistent with the conformation and size of other plant actin filaments assembled in vitro ( Igarashi et al., 1999 ; Ren et al., 1997 ). Together, these results demonstrate that this procedure yields purified zucchini hypocotyl actin that is competent for polymerization.

Purified zucchini hypocotyl actin was used to prepare both G- and F-actin columns. The patterns of proteins eluted from the F-actin and G-actin columns were examined using two different starting materials. Many soluble proteins bound and were eluted from the G-actin column, whereas fewer proteins eluted from the F-actin column. A single protein of approximately 40 kDa was highly enriched in the fractions eluted from the F-actin column, but was not eluted from the G-actin column. This protein was also absent in the fractions eluted from the BSA column. Therefore, these columns are selective in their protein retention profiles and have led to the identification of a potentially novel F-actin-binding protein of approximately 40 kDa. As this protein is blocked at the N-terminus, amino acid sequence information is not yet available to facilitate identification of the protein.

To identify plasma membrane-associated actin-binding proteins, isolated zucchini plasma membranes were also used as a starting sample for chromatography on the actin columns. One major doublet of proteins at 33 and 35 kDa was identified in the high-salt elution from the F-actin, but not the G-actin and BSA columns. These two proteins were shown to cross-react with antisera raised against pea annexins. Given the demonstrated specificity of these antibodies ( Clark et al., 1992 ; Clark et al., 1998 ), and the fact that annexin doublets near 33 and 35 kDa have been observed repeatedly ( Clark and Roux, 1995), these results confirm that annexins bind to F-actin, and, in the present case, to a physiologically relevant source of actin.

It is clear that, in animal cells, some classes of annexins bind to F-actin, but in plant cells the association with actin has been less clear ( Delmer and Potikha, 1997). Several groups have been unable to detect annexins in detergent-insoluble cytoskeletal pellets ( Blackbourn et al., 1992 ; Calvert et al., 1996 ), although using zucchini plasma membranes, these two annexin-like polypeptides do partition with actin during detergent extraction (data not shown). Calvert et al. (1996) showed that purified tomato annexins would bind to F-actin columns prepared with animal actin, but a direct interaction with plant actin has not yet been reported. In this study, annexins are preferentially enriched during F-actin chromatography. Although the significance of the annexin interaction with the actin cytoskeleton is not completely clear, the linkage of annexins to both membranes and the actin cytoskeleton could be important for the organization and stabilization of various membrane domains, as discussed previously ( Gerke and Moss, 1997). In plants, a polar distribution of annexin has been found by immunocytochemical localization in fern rhizoids ( Clark et al., 1995 ), Lilium longiflorum pollen tubes ( Blackbourn et al., 1992 ) and pea plumules ( Clark et al., 2000 ) .

Although the NPA binding protein has been suggested to interact with actin filaments ( Butler et al., 1998 ; Cox and Muday, 1994), the previous evidence supporting the interaction with plant actin was indirect. The NPA binding protein has been found to partition with the actin cytoskeleton during detergent extraction, and drugs that alter the polymerization state of actin altered the partitioning of both actin and NPA binding activity ( Butler et al., 1998 ; Cox and Muday, 1994). Although these previous results suggested an interaction with actin, they did not show that the NPA binding protein directly interacts with actin ( Muday, 2000).

To provide complementary evidence indicating that the NPA binding activity directly interacts with the actin cytoskeleton, samples enriched in this protein were applied to actin affinity columns. NPA binding activity is retained by the F-actin column and reproducibly elutes with high salt concentrations. NPA binding activity was localized to one or two fractions in each of five separate F-actin affinity columns and was significantly greater than the activity eluting from a BSA column or an F-actin column to which no solubilized proteins were applied. The elution profile of this protein is different from that of annexins. Demonstration of the specific elution of the NPA binding protein from the F-actin column provides stronger evidence in support of the interaction of this protein with the actin cytoskeleton. The function of the interaction between the NPA binding protein and the actin cytoskeleton may be to fix the auxin efflux carrier complex on one face of the cell and to thereby develop asymmetric auxin transport. Recent evidence has demonstrated that the PIN gene products, which are believed to encode part of the auxin efflux carrier protein complex, are localized to one end of elongated plant cells ( Gälweiler et al., 1998 ; Müller et al., 1998 ). The regulatory subunit of the auxin efflux carrier, or NPA binding protein, may function to localize the efflux carrier complex through interaction with the actin cytoskeleton ( Muday, 2000).

The goal of this study was to develop procedures for construction of plant F-actin affinity columns and the application of affinity chromatography to identify plant actin-binding proteins. The selectivity of the G-actin and F-actin affinity columns was verified using known actin-binding proteins. Soluble and membrane-associated actin-binding proteins were subjected to chromatography on these columns, resulting in the identification of an unknown F-actin-binding protein of approximately 40 kDa, and verification of the F-actin interaction of annexins and the NPA binding protein .

Experimental procedures

Chemicals and radiochemicals

[2,3,4,5(N)-3H]NPA (58 Ci mmol−1) was obtained from American Radiolabeled Chemicals Inc. (Saint Louis, MI, USA). NPA was purchased from Chemical Services (West Chester, PA, USA). Phalloidin was obtained from Molecular Probes (Eugene, OR, USA). DNAase I was purchased from Worthington Biochemical Corporation (Freehold, NJ, USA) and α-actinin was purchased from Cytoskeleton, Inc. (Denver, CO, USA). Sepharose CL-4B was obtained from Pharmacia Biotech Company (Washington DC, USA). Mouse anti-actin IgG monoclonal antibody (clone 4) was obtained from ICN Biomedical (Costa Mesa, CA, USA). Alkaline phosphatase-labelled goat-anti-mouse IgG1 antibody, nitroblue tetrazolium (NBT), 5-bromo-4-chloro-3-indoyl phosphate p-toluidine salt (BCIP), Scintiverse scintillation fluid and Whatman GF-B filters were obtained from Fisher Scientific (Pittsburgh, PA, USA). The polyclonal antibodies against pea annexin p35 (E6) and Zea mays profilin 5 (ZmPRO5) were generated and characterized as described previously ( Clark et al., 1992 ; Kovar et al., 2000 ). All other chemicals were purchased from Sigma Chemical Company (Saint Louis, MI, USA).

Attachment of DNAase I to CNBr-activated resin

DNAase I was coupled to CNBr-activated resin to make DNAase I affinity resin for actin purification. The CNBr-activated resin (25 ml) was washed with 250 ml cold coupling buffer (0.1 m sodium bicarbonate pH 8.5). The resin was mixed with DNAase I (20 mg DNAase I in 25 ml coupling buffer), coupled overnight at 4°C, and successively washed with 200 ml each of coupling buffer, 0.1 m NaCl and water.

To block the CNBr-activated group, the resin was resuspended in 25 ml 1.0 m ethanolamine, pH 9.0 with HCl, and mixed gently on a shaker at room temperature for 1 h. The coupled resin was successively washed with 500 ml each of water, 1.0 m NaCl and water, all at 4°C. The resin was stored in MEB (microtubule extraction buffer: 50 m m PIPES, pH 7.0, 0.5 m m MgSO4, 1 m m EGTA, 0.5 m m NaATP) with 0.02% sodium azide at 4°C for up to 6 months.

Actin purification

Zucchini (Cucurbita pepo L. cv. Burpee Fordhook) seedlings were grown for 5 days in the dark in vermiculite at 30°C. Hypocotyls were homogenized with a ratio of 1 ml per gram of tissue using HB (homogenization buffer: 0.25 m m Tris, pH 7.2, 3.0 m m EDTA, 250 m m sucrose, 0.25 m m PMSF, 1.0 m m DTT) containing 0.5 m m NaATP. After filtration through nylon cloth, cellular debris was removed by centrifugation at 7000 g for 10 min. The supernatant was incubated with DNAase I–Sepharose CL-4B resin at a ratio of 15–20 ml of the supernatant for 1 ml resin on a shaker for 30 min at 4°C. The resin was collected by brief centrifugation at 7000 g and was washed successively with 50 ml each of 0.2 m NH4Cl, 5% formamide in APMEB (actin purification MEB; MEB + 5% glycerol, 0.15 mg ml−1 DTT), 1.0 m KCl, 5% formamide in APMEB, and APMEB alone. Moist resin (1 ml) was transferred to 2 ml microcentrifuge tubes, centrifuged briefly at 16 000 g, and the supernatant was carefully removed. APMEB containing 50% (v/v) formamide equal in volume to the volume of resin was mixed into each tube. After a 10 min incubation with gently shaking, the tube was subjected to centrifugation at 16 000 g, the supernatant was removed and the resin was extracted again for 10 min. After centrifugation at 16 000 g, the supernatants were immediately gel-filtered over a column containing Sephadex G-25 equilibrated with APMEB. The purified G-actin could be kept on ice for 5 days without losing its ability to repolymerize. The recovery of actin was determined by the bicinchoninic protein assay procedure ( Smith et al., 1985 ) using BSA as a protein standard.

SDS–PAGE, native electrophoresis and immunoblot analysis

For SDS–PAGE, protein samples were denatured in sample buffer at 95°C for 10 min, loaded onto 10% or 12% SDS polyacrylamide gels, and subjected to electrophoresis followed by silver staining ( Wray et al., 1981 ). Images were captured using the Alpha-Ease gel capture system (Alpha Innotech) or onto Polaroid film.

Samples were also subjected to electrophoresis on native gels according to the procedure of Andersland and Parthasarathy (1992). The non-denaturing separating gel contained 10% acrylamide, 375 m m Tris–HCl, pH 8.8, 6% glycerol. The running buffer and the gels contained 0.2 m m CaCl2. An indicator of sample mobility, bromophenol blue, was added to an empty lane of each gel.

Both native and SDS gels were blotted onto BA-S nitrocellulose membrane (Schleicher & Schuell, NH, USA) and blocked with 5.0% powdered milk in PBS (10 m m sodium phosphate, pH 7.5, 150 m m NaCl). Blots were probed with a mouse actin IgG monoclonal antibody (clone 4 from ICN), which has been shown to recognize plant actin ( Butler et al., 1998 ), or a rabbit Zea mays profilin 5 antibody, diluted 1000- or 500-fold, respectively. Alkaline phosphatase-conjugated goat anti-mouse IgG or goat anti-rabbit IgG was added at 1000-fold dilution in 5.0% milk in PBS, and, after washing in PBS, blots were developed by adding NBT and BCIP in alkaline phosphatase buffer (100 m m Tris, pH 9.5, 100 m m NaCl and 50 m m MgCl2).

For the pea annexin p35 antibodies, the blots were blocked with 1% powdered milk in PBS. The blot was probed with a 50-fold dilution of guinea pig anti-pea annexin p35 polyclonal antibody. Alkaline phosphatase-conjugated goat anti-guinea pig IgG was added at 1000-fold dilution in 1% milk in PBS. After washing in PBS with 1% milk, the blot was quickly rinsed in water and developed as given above. The images were scaled and processed to minimize background using Adobe Photoshop 5.0.

Quantification of actin binding to maize profilin

The affinities of recombinant maize profilins ZmPRO1 and ZmPRO5 for zucchini MgATP–G-actin were determined by quenching of intrinsic tryptophan fluorescence ( Gibbon et al., 1997 ; Perelroizen et al., 1994 ). The purification and characterization of recombinant maize profilins has been described elsewhere ( Gibbon et al., 1997 ; Gibbon et al., 1998 ; Karakesisoglou et al., 1996 ; Kovar et al. 2000 ; Staiger et al., 1993 ). Actin purified by DNAase I affinity chromatography followed by ultracentrifugation was subjected to overnight dialysis in buffer G (5 m m Tris, 0.2 m m CaCl2, 0.01% w/v NaN3, 0.5 m m DTT, 0.4 m m ATP, pH 8.0). Quartz cuvettes were loaded with 0, 0.3 or 0.45 µm zucchini actin in buffer G supplemented with 0.2 m m EGTA and 50 µm MgCl2 for 10 min before each experiment. Profilin was added sequentially from stock solutions of concentrations between 225 and 275 µm, in buffer G. The total increase in volume after profilin addition was always less than 1.5%. Fluorescence was recorded on a spectrofluorimeter (model 8000; SLM Instruments, Urbana, IL, USA) with excitation at 295 nm and emission at 330 nm. The observed fluorescence minus the fluorescence of actin alone was plotted against the concentration of profilin at each point. The Kd value at each actin concentration was determined by fitting the resulting plots with the equation Fobs − Fa = Fp + (Fpa − Fa − Fp) [PA], where Fobs is the measured fluorescence, Fa, Fp and Fpa are the intrinsic fluorescence coefficients for actin, profilin and the profilin–actin complex, respectively, and [PA] is the concentration of the profilin–actin complex ( Perelroizen et al., 1994 ). Measurements were performed using two independent batches of each profilin and zucchini actin. The reported Kd is the average of both curves from one experiment. For comparison, equivalent measurements were also performed for ZmPRO1 and ZmPRO5 binding to maize pollen G-actin ( Gibbon et al., 1997 ; Gibbon et al., 1998 ; Ren et al., 1997 ).

Electron microscopy of F-actin

Actin purified by DNAase I affinity resins and sedimentation by ultracentrifugation was resuspended in the same volume as the supernatant using APMEB, with 50 m m KCl, 5 m m MgCl2, and the concentration of actin was 50 µg ml−1 as approximated by DNAase I inhibition assay. The repolymerized actin was applied to carbon-coated grids and negatively stained with 2% uranyl acetate for 2 min. The specimens were visualized by transmission electron microscope (model EM400, Philips Electronic Instruments, Inc.) at an accelerating voltage of 80 kV.

Preparation of actin affinity column matrices

A 40 ml sample of G-actin was eluted from DNAase I affinity resin, gel-filtered, and further purified by pelleting during ultracentrifugation at 100 000 g for 30 min. The pellet was resuspended in 2 ml APMEB. Freshly prepared CNBr-activated Sepharose CL-4B (2 ml) was packed in a column at 4°C and washed with 10 ml of cold APMEB. A 1 ml aliquot of polymerized actin in APMEB (actin concentration was about 0.5 mg ml−1 as measured by DNAase I inhibition assay) was added to the column and mixed gently, phalloidin was added to reach 10 µm and the resin was mixed again. Coupling was allowed to proceed overnight (between 16 and 24 h) without mixing. The unbound actin was removed by filtering the resin and the reaction was then terminated by adding 50 m m ethanolamine (pH 8.0) at a flow rate of 5 ml h−1 for 2 h.

The G-actin columns were prepared in two different ways. For the first approach, the G-actin column was packed in a similar manner to the F-actin affinity columns with the exception of the addition of phalloidin. Following addition of the actin, the column was gently mixed and coupled overnight at 4°C with gentle shaking. The unbound actin was removed by running the column and the reaction was terminated by adding 50 m m ethanolamine (pH 8.0) at a flow rate of 5 ml h−1 for 2 h. The second method used purified F-actin that was dialysed overnight in G buffer ( Ren et al., 1997 ) (5 m m Tris, pH 8.0, 0.2 m m CaCl2, 0.01% NaN3, 0.5 m m DTT, 0.4 m m ATP). The dialysed G-actin was then added directly to 2 ml of activated resin and allowed to incubate for 30 min with shaking at 4°C.

After coupling, the column was washed with APMEB to remove unbound actin and ethanolamine, and to pack the column bed. The flow properties of the F-actin columns were then checked by adding a small aliquot of APMEB containing phenol red to the column bed. When the dye band is washed through the column with APMEB, it should move through the bed evenly. If the dye ran around the edge of the matrix, the column was gently mixed in a small volume of APMEB plus 10 µm phalloidin and allowed to stand undisturbed for several hours. The column bed was repacked by washing with APMEB.

Once a satisfactory column had been prepared, it was washed with 1.0 m KCl in APMEB and subsequently with APMEB. The column was either used the same day or stored in APMEB and 0.02% NaN3, with the F-actin columns containing 10 µm of phalloidin, at 4°C for less than 5 days. Just before use, the column was equilibrated with 2–3 column volumes of the buffer in which the extract to be applied to the column was prepared.

Preparation of zucchini plasma membranes

Plasma membranes were prepared from zucchini (Cucurbita pepo L. cv. Burpee Fordhook) seedlings grown for 5 days in the dark in vermiculite at 30°C as described previously ( Dixon et al., 1996 ; Muday et al., 1993 ). The plasma membrane vesicles were resuspended either in NBB (NPA binding buffer, 20 m m Na citrate, pH 5.3, 1 m m MgCl2, 250 m m sucrose) or in MEB at a ratio of 1.0 ml for every 50 g starting material, and were frozen at −70°C.

Actin affinity chromatography

Purified animal actinin was diluted to 0.02 mg ml−1 and 5 ml was loaded onto either an F-actin or a G-actin column. Elutions were performed sequentially for each column with 2 ml of APMEB with 0.25, 0.5 or 1.0 m KCl. During the elution, fractions of 0.5 ml each were collected. Equal volumes of fractions were loaded onto SDS–PAGE. The gel was subjected to silver staining.

Extracts enriched in soluble actin-binding proteins were prepared from zucchini (Cucurbita pepo L. cv. Burpee Fordhook) seedlings grown for 5 days in the dark in vermiculite at 30°C. The upper third of the etiolated hypocotyls (approximately 4 cm) was harvested without cotyledons and homogenized with a ratio of 1 ml per gram of tissue using THB (Tris homogenization buffer: 10 m m Tris, pH 7.5, 0.5 m m CaCl2, 0.5 m m PMSF, 0.5 m m DTT, 0.4 m m NaATP). After filtration through nylon cloth, cellular debris was removed by ultracentrifugation at 21 000 g for 20 min. Following centrifugation, the supernatant was dialysed overnight in a 100-fold volume of APMEB with gentle stirring at 4°C, to remove excess Tris. This sample was applied directly to the G-actin or F-actin columns. After washing the columns with 5 ml of APMEB, elutions were performed sequentially for each column with 3 ml each of APMEB containing sequentially 0.25, 0.5 or 1.0 m KCl. During the elution, fractions of 1.0 ml each were collected. Equal volumes of each fraction were loaded onto SDS–PAGE and proteins were visualized by silver staining.

Plasma membranes were treated under conditions that depolymerize the actin cytoskeleton, to release actin-binding proteins into the soluble phase. Plasma membrane vesicles at 2 mg ml−1 in NBB with 5% glycerol, 0.5 m m PMSF, 0.4 m m NaATP, were treated with 1.0 m Tris at pH 7.0 in the presence of 0.5% Triton X-100. Samples were incubated at 4°C with gentle shaking for 30 min and subjected to ultracentrifugation at 100 000 g for 30 min. The supernatant was diluted tenfold with 5% glycerol and 5 ml was applied to the column. Columns were washed with 5 ml APMEB and 0.1 m Tris, pH 7. Samples were eluted and fractions subjected to SDS–PAGE, as described above.

For columns designed for the measurement of NPA binding activity, plasma membranes were treated as above, with several modifications that were designed to maximize the recovery of NPA binding activity. Zucchini plasma membrane vesicles, at 2 mg ml−1, were treated with 1.0 m Tris, pH 6.0, in NBB in the presence of 0.5% Triton X-100, 10 µm unlabelled NPA and 0.2 m m PMSF for 30 min at 4°C with gentle shaking, then subjected to ultracentrifugation at 100 000 g for 30 min. The supernatant was diluted tenfold with 5% glycerol in water with 0.5 m m ATP and 0.2 m m PMSF. The diluted supernatant was added to an F-actin affinity column in a 5 ml volume. After the sample ran through the column, the column was washed with 5 ml modified NBB (NBB with 5% glycerol, 50 m m KCl, 5 m m MgCl2, 0.5 m m ATP, 0.2 m m PMSF) with 5 n m[3H]NPA. Proteins were eluted sequentially with 2 ml volumes of modified NBB with 5 n m[3H]NPA containing 0.25, 0.5 or 1.0 m KCl. During the elution, fractions of 0.5 ml were collected. To measure the amount of [3H]NPA bound in each fraction, 150 µl of each fraction were filtered over 24 mm GF/B discs pre-treated with 0.3% polyethylenimine and washed with 5 ml of NBB. Scintillation fluid (2.5 ml) was added to each filter, followed by liquid scintillation counting for 2 min. Each value represents the average of two replicates. The background was measured by addition of unlabelled NPA to reach a final concentration of 10 µm, into 150 µl of each fraction.

Acknowledgements

We thank Brian Tague and Jim Curran for helpful suggestions during the course of this work. We appreciate the assistance of Mark Lively with the protein sequence analysis. We thank Aaron Rashotte, Sandy Ante, Hanya Chrispeels and Jennifer Waters Shuler for helpful comments during the preparation of this manuscript, and Jana Naron for her bibliographic assistance. This work was supported by grants to G.K.M. from the National Science Foundation (grant no. IBN-9318250) and National Aeronautics and Space Administration (grant no. NAG2-1203) and to C.J.S. from United States Department of Agriculture-NRICGP (grant nos. 97-37304-4876 and 99-35304-8640).

Ancillary