INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Profilins are small proteins that interact with at least three types of ligands: actin monomers (and the homologous Arp2 subunit of Arp2/3 complex); proteins with sequences adopting a type II poly-L-proline helix; and polyphosphoinositides (Machesky and Pollard, 1993; Schluter et al., 1997). Interaction with actin monomers suppresses growth at the pointed end of actin filaments but not the barbed end (Pollard and Cooper, 1984). This allows profilin-actin complexes to support the growth of uncapped barbed ends in cells. In addition, most profilins (amoeba [Mockrin and Korn, 1980; Nishida, 1985], vertebrate [Goldschmidt-Clermont et al., 1991; Perelroizen et al., 1996], yeast [Eads et al., 1998]) but not all profilins (plants [Perelroizen et al., 1996; Eads et al., 1998; Kovar et al., 2000]) stimulate the exchange of the adenine nucleotide bound to actin monomers. This nucleotide exchange activity is postulated to prevent ADP-actin monomers dissociated from filaments from being trapped in high-affinity complexes with ADF/cofilins (Blanchoin and Pollard, 1998; Didry et al., 1998). Potential protein ligands with proline-rich sequences include VASP (Reinhard et al., 1995), Enabled (Ahern-Djamali et al., 1999), Mena (Gertler et al., 1996; Lanier et al., 1999), N-WASP (Suetsugu et al., 1998), WAVE/Scar (Miki et al., 1998), verprolin/WIP (Ramesh et al., 1997), diaphanous/p140mDia (Watanabe et al., 1997), cappuccino (Manseau et al., 1996), Bni1p and Bnr1p (Evangelista et al., 1997; Imamura et al., 1997), cdc12 (Chang et al., 1997), drebrin and gephyrin (Mammoto et al., 1998), SMN (Giesemann et al., 1999), and aczonin (Wang et al., 1999). In the case of Mena, diaphanous, cappuccino, Bni1p, Bnr1p, and cdc12, genetic interactions support the biological significance of the interaction.

Null or conditional mutations established that loss of profilin function is lethal in mice (Lanier et al., 1999), flies (Cooley et al., 1992; Manseau et al., 1996), and fission yeast (Balasubramanian et al., 1994) and severely disabling in Dictyostelium (Haugwitz et al., 1994) and budding yeast (Haarer et al., 1990). The existence of more than one profilin gene complicates the situation in mice and Dictyostelium. The two yeast have single profilin genes.

Although profilin is just a small protein, its biological functions are sufficiently complex that it has not yet been possible to show definitively why it is essential for viability. To date 18 articles have described 66 designed mutations or combinations of mutations in human, bovine, budding yeast, Anacthamoeba, Dictyostelium, and plant profilins (supplementary material, Table 1). These results are consistent with the hypothesis that interactions of profilin with both poly-L-proline and with actin are important for biological function. However, with some exceptions, many of these experiments are incomplete in one way or another, especially in establishing the stability of mutant proteins and measuring quantitatively interactions with actin or poly-L-proline.

We report here the first comprehensive analysis of the interactions of profilin with actin and poly-L-proline with a large collection of mutants of fission yeast profilin. S. pombe is favorable for this analysis. An essentially complete genome sequence contains only one profilin gene, cdc3⁺. A null mutation of cdc3⁺ is lethal. The mutation E42K in the temperature-sensitive strain cdc3-124 disables profilin at 36°C but not at 25°C (Balasubramanian et al., 1994). At 25°C the cdc3-124 strain grows normally with morphology indistinguishable from wild-type. At 36°C cdc3-124 cells have defects in cytokinesis, resulting in elongated, dumbbell-shaped cells with up to eight nuclei and mislocalized actin filament patches. We made ~90 point mutations, each designed from an atomic model of the protein to interfere with interactions with actin or poly-L-proline. We tested the ability of each mutant profilin to complement a profilin null mutation and a profilin temperature-sensitive mutation. We purified a selection of >30 of complementing and noncomplementing mutant proteins and characterized their stability and interactions with actin and poly-L-proline. This large database of quantitative information revealed that both actin binding and poly-L-proline binding are required independently for biological function. In addition, the data show that viability depends on both the affinities of profilin for each of these ligands and other functional attributes of the mutant protein. In particular, novel mutations reveal that ability of profilin to catalyze nucleotide exchange by actin is essential for viability of fission yeast and that at least one poly-L-proline ligand has different binding requirements than poly-L-proline.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strains, Media, and Chemicals

S. pombe strains used in this study were diploid profilin null strain VSP38 (cdc3::ura4⁺/cdc3⁺, ade6-216/ade6-210, leu1-32/leu1-32, ura4-D18/ura4-D18, h⁺/h), haploid temperature-sensitive strain KYG491 (cdc3-124, ade6-216, ura4-D18, leu1-32, h⁺), and the matching wild-type haploid strain KGY247 (cdc3+, ade6-216, ura4-D18, leu1-32, h⁺). Dr. Kathy Gould of Vanderbilt University kindly provided these strains. EMM and ME media were purchased from Bio-101. ATP was from Molecular Probes (Eugene, OR). Poly-L-proline (MW 5000), phloxin-B, thiamine, amino acids, and nucleotides were from Sigma (St. Louis, MO). The poly-L-proline affinity column was prepared according to Kaiser et al. (1989).

Mutagenesis and Purification of S. pombe Profilin

Wild-type S. pombe profilin cDNA in expression vector pMW-172 was provided by Dr. Steven C. Almo of Albert Einstein College of Medicine. Selected residues of S. pombe profilin were mutated by reverse polymerase chain reaction (PCR) mutagenesis according to Ho et al. (1989), except that pfu polymerase (Stratagene, La Jolla, CA) was used to increase the fidelity and reduce adenine addition at the end of PCR products. Nucleotide sequences of all PCR products were confirmed by automated dideoxynucleotide sequencing. Plasmids of wild-type and mutant profilins were freshly transformed into competent Escherichia coli BL21(DE3) strain and cultured overnight without isopropyl--D-thiogalactopyranoside induction in LB media.

Purification of wild-type and mutant profilins was based on Kaiser et al. (1989). Briefly, a pellet of bacteria from 300 ml of LB media was suspended in 35 ml of Buffer L (25 mM Tris-Cl pH 7.5, 50 mM sucrose, 10 mM EDTA, 5 mM dithiotreitol [DTT], 1 mM phenylmethylsulfonyl fluoride) with 2 M urea, lysed by sonication, and centrifuged at 95,000 × g for 20 min. If the recombinant profilin was soluble the extract was applied to a 2.5 cm (diameter) × 4-cm (height) column of Whatman DEAE-cellulose, DE52, and washed with Buffer L. The flow through containing profilin was dialyzed against Buffer D (20 mM Tris-Cl pH 8.0, 20 mM KCl, 1 mM EDTA, 1 mM DTT). Recombinant profilins were further purified by affinity chromatography on poly-L-proline Sepharose. After adsorbing the sample, the column was washed with 3 M urea in Buffer D. Bound profilin was eluted with 8 M urea in Buffer D and refolded by dialysis against 1 mM EDTA, 1 mM DTT, 20 mM Tris-Cl, pH 8.0. Profilin mutants that did not bind poly-L-proline were purified by gel filtration on Superdex 2000. If the recombinant profilin was insoluble after extracting the bacteria, the pellet was dissolved in 8 M urea in Buffer D at 4°C. After removal of insoluble material by centrifugation at 95,000 × g for 20 min, the supernatant was adjusted to 4 M urea by addition of Buffer D and run through a DE52 anion exchange column in the same buffer. Profilin in the flow through was dialyzed against Buffer D and purified further as described above.

The concentration of S. pombe profilin was determined by absorbance at 280 nm with an extinction coefficient of 1.63 OD/mg/ml for wild-type and most mutant profilins. The extinction coefficient of P107W profilin was 2.22 OD/mg/ml. These extinction coefficients were measured in a Beckman XLI analytical ultracentrifuge by using absorbance and Raleigh interference optics with an assumed fringe displacement of 3.32 per milligram per milliliter.

Rabbit muscle Ca-ATP actin was prepared by the method of Spudich and Watt (1971) and purified by gel filtration on Sephacryl S-300 in Buffer A (0.2 mM ATP, 0.5 mM DTT, 0.1 mM CaCl₂, 1 mM NaN₃, 2 mM Tris-Cl pH 8.0).

Urea Denaturation

Profilins were added at a final concentration of 1.5 µM to 1 mM EDTA, 1 mM DTT, 20 mM Tris-Cl, pH 8.0, containing 0 to 8 M urea at room temperature. Intrinsic fluorescence was measured with a PTI spectrofluorimeter (Photon Technology Instruments, Santa Clara, CA) with excitation at 295 nm. The emission peak of profilin shifts from 330 nm (without urea) to 352 nm and increases in intensity (with 8 M urea) (Kaiser and Pollard, 1996). The difference between the emission intensities at 352 and 330 nm was plotted against concentration of urea. The urea concentration at the midpoint of the transition in intensity was used to measure stability.

Interaction of Profilin with Poly-L-Proline

Profilins were added at a final concentration of 1.5 µM to 1 mM EDTA, 1 mM DTT, 20 mM Tris-Cl, pH 8.0, containing 0 to 5 mM poly-L-proline at room temperature. Intrinsic fluorescence was measured with excitation at 295 nm and emission at 323 nm. The dissociation equilibrium constant (K_d) was calculated using Kaleidagraph software and Eq. 1 from the dependence of fluorescence change on the concentration of poly-L-proline.

(1)

F is the relative fluorescence; [L] and [P] are the total concentrations of poly-L-proline and profilin.

Interaction of Profilin with Actin Monomers

Monomeric rabbit skeletal muscle actin at a concentration of 12 µM in 0.2 mM ATP, 0.5 mM DTT, 0.1 mM CaCl₂, 1 mM NaN₃, 2 mM Tris-Cl pH 8.0 was treated at 4°C with Dowex-1 (Bio-Rad AG1-x4 resin washed with water extensively and buffered with 1 M Tris-Cl, pH 7.5, and then buffered with 25 mM Tris-Cl pH 7.5 and 0.01% NaN₃) to remove free ATP. ATP Mg-actin was prepared by adding 200 µM ATP, 200 µM EGTA, and 50 µM MgCl₂ for 50 min at 4°C. Free ATP was removed with Dowex-1 and Mg-ATP actin was used within 3 h.

The time course of ATP dissociation from 1.2 µM Mg-ATP actin was measured by fluorescence with excitation at 360 nm and emission at 410 nm (Waechter and Engel, 1975; Nishida, 1985) after adding 500 µM ATP with 0-100 µM profilin. The rate constant (k₁) for the dissociation of ATP from actin was determined by fitting the time course with a single-exponential equation by using Kaleidagraph software. Profilin promotes dissociation of nucleotide from actin. The dependence of k₁ on profilin concentration gave the dissociation equilibrium constant (K_d) for actin and profilin according to Eq. 1, where F is k₁ and [L] and [P] represent the concentrations of profilin and actin.

Profilin mutant Y79R fails to promote nucleotide exchange, but inhibits nucleotide exchange by wild-type profilin. We used a competition assay to measure the affinity of Y79R profilin for actin in the presence of wild-type profilin. We measured the time course of dissociation of ATP from 1.2 µM ATP actin in the presence of 500 µM ATP, 1.5 µM wild-type profilin, and 0-100 µM Y79R. The apparent dissociation rate constant (k_obs) from each time course was determined by fitting a single-exponential equation.

The total actin A_T is either free A_f, bound to wild-type profilin (A_wt), or bound to mutant profilin (A_mu).

(2)

Three reactions contributed to the observed dissociation rate constant, k_obs: dissociation from free actin (A_f), actin bound to wild-type profilin (A_wt), and actin bound to Y79R profilin (A_mu).

(3)

where k_f, k_wt, and k_mu are rate constants for ATP dissociation from free actin, actin bound to wild-type profilin and actin bound to Y79R. All three of these rate constants were measured experimentally. Because k_f = k_mu, Eq. 3 is shortened to the following:

(4)

The dissociation equilibrium constant for actin and wild-type profilin is as follows:

(5)

where wt_T is total wild-type profilin. Under the conditions of this experiment essentially all of Y79R profilin is free, so the dissociation equilibrium constant for actin and Y79R profilin is as follows:

(6)

where mu_T is the total Y79R profilin.

We canceled A_f and resolved [A_wt] from Eqs. 5 and 6:

(7)

Eq. 7 was substituted into Eq. 4 and fit with Kaleidagraph software to give K_mu

For reasons that we do not understand 100 µM concentrations of wild-type or mutant profilin inhibit nucleotide dissociation from actin ~10% compared with lower concentrations, so we corrected for this in the analysis of competition data.

Spontaneous actin polymerization was initiated by adding 20 µl of 10× KMEI (0.1 M imidazole, 0.5 M KCl, 10 mM MgCl₂, 10 mM EGTA, pH 7.0) to 180 µl of 5 µM Mg-ATP muscle actin monomers, 0.1 µM pyrene-labeled actin monomers, and 0-50 µM profilin. The polymer concentration was monitored by pyrene fluorescence with excitation at 365 nm and emission at 407 nm (Kouyama and Mihashi, 1981).

Genetic Complementation of Profilin-Null Strain and Temperature-sensitive Strain

Wild-type and mutant profilins cDNAs in pREP 81 plasmid under the control of the weakest thiamine-repressible nmt1 promoter were transformed into profilin null or ts strains by electroporation. Transformed cells were grown on plates with selective media at 32°C for the null strain and 25°C for the ts strain. Complementation of heterozygous null strains was assessed by random sporulation assays at 32°C.

Wild-type and ts profilin strains with complementing plasmids were grown in liquid media at 25°C. While still in exponential phase cells were diluted to OD₅₉₅ of 0.25 and then grown for 12 h at 36°C while maintaining the exponential phase by diluting the culture. Cells were mounted on glass slides and DIC micrographs were taken with a digital camera. The length and width of 100-200 cells were measured using Metamorph (Universal Imaging, West Chester, PA) software.

Fluorescence Microscopy

Wild type, ts strain, and ts strains with plasmids carrying wild-type or mutant profilins were grown at 25°C and then shifted to 36°C for 8 h. Cells were collected at both temperatures and fixed in 70% ethanol. Cells were rehydrated in H₂O and stained with 10 µg/ml 4,6-diamidino-2-phenylindole (DAPI) and 50 µg/ml Calcofluor in H₂O at room temperature for 10 min. Images of cells stained with Calcofluor and DAPI were taken sequentially (Balasubramanian et al., 1994) with a digital camera.

Immunoblot Analyses

Rabbit was immunized with purified recombinant S. pombe profilin and antibodies purified from serum by adsorption to and elution from an immunoblot with immobilized profilin (Pollard, 1984)

Cells growing exponentially in 20 ml of liquid medium at 25°C or for 4 h at 36°C were spun down and resuspended in 200 µl of lysis buffer (25% sucrose, 20 mM imidazole pH 7.0, 5 mM EDTA, 5 mM DTT, 1 mM phenylmethylsulfonyl fluoride and 1 tablet of protease inhibitor cocktail [Roche Molecular Biochemicals, Indianapolis, IN] per 25 ml). Cells were lysed at 4°C with glass beads in a Fastprep agitator (Bio-101) for 4 × 18 s at a speed of 5.5. We added 75 µl more lysis buffer and lysates were collected by poking a hole at the bottom of tube followed by centrifugation at 3000 rpm for 1 min. Lysates were centrifuged for 20 min at 4°C in an Eppendorf centrifuge. Supernatants were assayed for protein by Bradford assay (Bio-Rad, Richmond, CA) and prepared for gel electrophoresis by boiling 5 min in SDS sample buffer. Samples containing roughly 25 µg of protein were run on 16% polyacrylamide gels in SDS. Standards consisted of 0.3-50 ng of purified S. pombe profilin mixed with 25 µg of lysate of the cdc3 ts strain, which contains no detectable profilin under our lysis conditions. Proteins were transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA) overnight at 23 V in buffer consisting of 10 mM 3-(cyclohexylamino)propanesulfonic acid pH 11.0 and 10% methanol. The membrane was blocked with Tris-buffered saline/Tween 20 (TBST) (10 mM Tris-Cl pH 7.5, 150 mM NaCl, 0.1% Tween 20) containing 5% dry milk for 4 h and incubated with affinity-purified anti-profilin antibody (1:100 dilution) in TBST with 1% dry milk for 4 h. Membranes were washed three times in TBST over 20 min. Membranes were treated with a 1:10,000 dilution of goat anti-rabbit antibody conjugated with horseradish peroxidase (Zymed Laboratories, South San Francisco, CA) in TBST with 1% dry milk for 1 h and bound antibodies were visualized with enhanced chemiluminescence reagents (Amersham Pharmacia Biotech, Piscataway, NJ). Films were scanned and the profilin content of each sample calculated by comparison with the internal standard curve.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Design of Point Mutations to Affect Specific Ligand Interactions

Our aim was to find point mutations of profilin that are stably folded but eliminate interactions with one and only one class of ligand, actin (Arp2/3 complex), poly-L-proline, or phosphatidylinositol biphosphate (PIP₂), so that we could test which interactions are essential for viability. Our strategy was to replace amino acids known to be involved with each interaction. Structural studies of profilins from different species have defined the binding sites for actin (Schutt et al., 1993) and poly-L-proline (Archer et al., 1993; Mahoney et al., 1997, 1999). The PIP₂ binding site is less well defined (Sohn et al., 1995; Chaudhary et al., 1998). We used atomic structures to align the sequences of bovine (Schutt et al., 1993), Saccharomyces cerevisiae (Eads et al., 1998), and Acanthamoeba profilins (Vinson et al., 1993; Fedorov et al., 1994). The sequence of S. pombe profilin was aligned with these sequences without ambiguity. This alignment was used to identify residues of S. pombe profilin involved with each interaction (Figure 1). We used different strategies to eliminate the binding of each ligand.

View larger version (5K): [in this window] [in a new window]   Figure 1.   Mutations designed to cover all residues involved with binding various ligands, e.g., the binding with actin (a), poly-L-proline (p), and PIP2 (+), marked on top of the wild-type S. pombe profilin sequence. Some residues may participate in both actin and PIP2 binding. Each point mutant is represented by the mutated amino acid that aligns with the wild-type sequence. E42K (*) is the mutation of profilin in cdc3-124 ts strain. C89S (*) is a control mutant for structure stability. Mutants that failed to complement both null (32°C) and ts (36°C) strains are bold and underlined; mutants that complemented the ts at 36°C but not the null strain at 32°C are bold only. Four of these noncomplementing mutants complement null strain at 25°C (i).

Actin Binding. About 20 amino acids from profilin contribute to a large interface of ~2250 ² with actin in the complex of bovine profilin with -actin (Schutt et al., 1993). To create steric interference between the proteins, we first replaced many residues, one at a time with amino acids with bulky side chains such as Trp, Tyr, or Phe (Figure 1). Second, four conserved profilin residues with positive charges (K67, R72, K81, K84) pair with Glu residues on actin. This suggested that ionic interactions between these residues might stabilize the binding. We substituted up to six different residues for these four basic residues. We also introduced single Glu residues at seven places on the interface of profilin with actin to alter its electrostatic potential. In bovine profilin, two longer loops flank both ends of strand 6. Some residues in these loops interact with actin but have no counterpart in S. pombe profilin.

Poly-L-Proline Binding. Residues involved with poly-L-proline binding are the most conserved across all species. The crystal structure of poly-L-proline bound to profilin (Mahoney et al., 97) shows two types of interactions: hydrophobic interactions involving the pyrrolidine ring of poly-L-proline with aromatic side chains of conserved amino acids on profilin; and hydrogen bonds between backbone carbonyl oxygens of poly-L-proline and side chains on profilin. We substituted a variety of amino acids for conserved residues in the poly-L-proline binding site in an attempt to disrupt both types of interactions.

PIP₂ Binding. The binding site for PIP₂ is not well defined, but is likely to involve positively charged amino acids (Fedorov et al., 1994). S. pombe profilin has six Lys, four Arg, and one His. We mutated each to Glu, Ala, Leu, and Trp (or Tyr). Four of these residues are also part of actin binding interface (Figure 1).

Complementation of S. pombe Strains with Temperature-sensitive or Null Mutations of Profilin

We transformed a haploid strain carrying a profilin-ts allele (cdc3-124) and a diploid strain containing one profilin-null allele with plasmids carrying each of 85 profilin mutants under the regulation of the weakest thiamine-suppressible nmt1 promoter in plasmid Rep81. This test included 48 mutants designed to interfere with actin binding, nine mutants targeted to poly-L-proline binding, and 46 mutants targeted to PIP₂ binding (18 of these overlapped with mutants for actin binding). We tested two additional point mutants, E42K (the mutation found in the cdc3-124 ts allele) and C89S (a mildly unstable core mutation). Neither of these residues participates directly in any known binding interaction. We tested for complementation of the ts strain at 36°C (Figure 2, A-D, and Table 1) and did random sporulation assays at 32 and 25°C to examine whether mutants complemented the null allele (Table 1). A few mutants were also tested for complementation of the ts strain by using the strongest nmt1 promoter in plasmid Rep3 (Table 1).

View larger version (122K): [in this window] [in a new window]   Figure 2.   (A-D) Complementation of cdc3-124 ts strain at 36°C by expression of mutant profilins from the weakest nmt-1 promoter in pRep81 plasmid in the absence of thiamine. (A) Actin binding mutants. (B) Actin binding mutants, wild-type profilin (WT), mutant profilin (E42K) for cdc3-124 strain, and structure stability control mutant C89S. (C) Poly-L-proline binding mutants. (D) Mutants with profound loss of poly-L-proline binding Y5D, actin binding K81E, and nucleotide exchange activity Y79R, mutant with increased nucleotide exchange activity K81F, ts mutant (E42K), wild-type (WT) profilin, vector control as well as cdc3-124 strain. (E) Immunoblots to measure the expression levels of mutant profilins in cdc3-124 ts strain shifted to 36°C for 4 h. Approximately 25 µg of protein was loaded in each lane. Table 1 lists the quantitation of these results. No profilin was detected in the cdc3-124 strain without transformation or when transformed with mutant E42K.

Although profilin is essential for viability of S. pombe (Balasubramanian et al., 1994) and all of our point mutations were designed to interfere with ligand binding, a remarkable 64 of 87 mutants complemented both the null and ts mutations in the absence of thiamine. Eight mutants failed both complementation tests: Y5D, K67W, I71E, I71R, I71W, Y79R, K81E, and L121R. Eleven more mutants failed to complement the null allele, but managed to complement the ts allele (K67A, K67E, K67L, I69W, R72E, P107Y, P107W, A111E, Y120A, Y120D, Y126D). Another two mutants complemented the null allele at 25°C but not at 32°C (K84W and G86W). E42K and S77W complemented the ts strain and null strain at 25°C but not 32°C. The substitutions in all 23 noncomplementing mutants are located on a surface where profilin interacts with either actin or poly-L-proline except E42K, which is on the "back" side of the protein. Expression of ts profilin E42K from the Rep81 promoter in the absence of thiamine rescued the E42K ts mutant strain (cdc3-124) at the restrictive temperature of 36°C, suggesting that the concentration of active profilin in the cdc3-124 ts strain is too low for growth at the restrictive temperature. Mutants S77W, K84W, and G86W rescued the null at 25° but not at 32°C, indicating these profilin alleles are also temperature sensitive.

At the restrictive temperature used for complementation tests, 36°C, most mutant profilins were expressed from plasmids in the cdc3-124 background at approximately the same level as genomic wild-type profilin, judging from immunoblot assays (Figure 2E and Table 1). No profilin was detected in the cdc3-124 ts strain under our lysis conditions unless transformed with a profilin plasmid. Wild-type strain KGY247 has 0.68 µg of profilin per 10 mg of total protein. cdc3-124 transformed with wild-type profilin has 0.67 µg of profilin per 10 mg of total protein. All mutant profilins were expressed at a similar level (0.28-1.2 µg/10 mg protein), except for the cdc3-124 mutant E42K, which was undetectable.

Structural Stability of Mutants

We used urea denaturation to test the stability of noncomplementing mutant proteins (Figure 3, A and B, and Table 1) to rule out folding defects as a trivial explanation for the failure of mutants to complement either ts or null alleles. For controls in this and the following assays we examined up to six complementing mutants of the same residue. Wild-type S. pombe profilin undergoes a shift in intrinsic fluorescence over a narrow range of urea concentrations as the protein makes a transition from folded to unfolded. Making the usual two state (folded/unfolded) assumption (Tanford, 1968; Santoro and Bolen, 1988), the concentration of urea at the midpoint of this transition reflects the equilibrium constant for folding. A difference from wild type represents a change in stability. The transition urea concentration for both recombinant and native wild-type S. pombe profilin is 4.5 M urea. This is higher than S. cerevisiae and Acanthamoeba profilins (3.4-3.5 M). Substituting tryptophan for another residue changed the intrinsic fluorescence of the folded protein, but did not interfere with the urea denaturation experiments.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Urea denaturation of purified profilins monitored by intrinsic fluorescence. (A) Mutants that are equal to or more stable than C89S () were considered as stable, including poly-L-proline binding mutants Y5A (), Y5D (), Y120D (), Y120A (), Y126D () and actin binding mutants Y79R (), K81E (), K81Y (), P107Y (+), A111E (), and wild-type profilin (). (B) Unstable mutants E42K() and L121R () denature at lower urea concentrations than C89S () and wild-type profilin ().

Mutant C89S served as a control for the influence of stability on physiological function. Although much less stable than wild-type profilin, with a transition concentration of 3.2 M urea, C89S rescued both the profilin ts and null alleles as well as wild-type profilin (Figure 2B and Table 1). Cys 89 is buried on strand 7, outside the binding surface for actin, far from the poly-L-proline binding surface and is unlikely to contribute to PIP₂ binding. C89S bound poly-L-proline with a K_d value of 50 µM (Figure 4A) and actin with a K_d value of 0.21 µM (Figure 5A) (not significantly different from wild-type profilin). This argues that an abnormal phenotype for any mutant with a transition concentration >3.2 M urea is not due simply to unstable structure. Folding instability seems to account for the temperature sensitivity of cdc3-124 where the substitution E42K is far from any known ligand binding site. Its transition concentration was 2.2 M urea (Figure 3B). Recombinant E42K protein was extremely difficult to purify from bacteria and aggregated at 4°C without urea.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Binding of profilins to poly-L-proline measured by intrinsic fluorescence. The dependence of the fluorescence changes on profilin concentration is fit with binding isotherms to measure the equilibrium constant. (A) Wild-type () and actin binding mutants Y79R (), K81E (), K81Y (), P107W (), P107Y (), A111E (), and minimally stable mutant C89S (). Inset: Curve fits at low concentrations of poly-L-proline. (B) Wild-type profilin () and poly-L-proline binding mutants W30Y (), L121A (), Y120A (), Y120D (), Y126D (), and Y5D (). Inset: Curve fit for wild-type and Y5D over a much larger range of poly-L-proline (150 mM).

View larger version (11K): [in this window] [in a new window]   Figure 5.   Nucleotide release assays using 1.2 µM ATP muscle actin and a range of profilin concentrations. The dependence of the rates of nucleotide release relative to actin alone on the concentration of profilin are fit with binding isotherms to measure the equilibrium constant. (A) Wild-type (), C89S () and poly-L-proline mutants Y5A (), Y5D (), Y120A (), Y120D (), and Y126D (). (B) Actin binding mutants I71R (), I71W (), Y79R (), and K81E () have no detectable effect on the rate; P107W (), P107Y (), and A111E () required higher concentrations than wild-type to promote exchange but maximum rates similar to wild-type (); K81F () requires higher concentrations than wild-type but increases nucleotide release more than wild-type. (C) Effect of a range of concentrations of Y79R on nucleotide release catalyzed by 1.5 µM wild-type profilin.

We limited our biochemical characterization to stable mutants with transition midpoints higher than the C89S control at 3.2 M urea. This eliminated E42K and L121R (Figure 3B). We also set aside K67L, I69W, K84W, and S77W, because they lacked a distinct urea denaturation transition and appeared to be partially unfolded without urea (our unpublished results). We eliminated K67W, designed as an actin binding mutant, because it failed to bind the poly-L-proline affinity column. We did not test G86W, but assumed that its temperature-sensitive phenotype is due to structural instability. We set aside K67A and K67E, because they aggregated in the actin polymerization buffer.

This screen yielded 13 structurally stable mutants that failed to complement the profilin ts and/or profilin null strains, consisting of four mutants designed to compromise poly-L-proline binding (Y5D, Y120A, Y120D, Y126D) and nine designed to compromise actin binding (I71E, I71R, I71W, R72E, Y79R, K81E, P107W, P107Y, A111E). For controls, we included structurally stable mutants that complemented both null and ts strains: three designed to compromise poly-L-proline binding (Y5A, W30Y, L121A) and four potential actin binding mutants (K81A, K81F, K81L, K81Y).

Poly-L-Proline Binding by Wild-Type and Mutant S. pombe Profilins

We tested each purified profilin for poly-L-proline binding by using the enhancement of intrinsic fluorescence as the assay (Figure 4, A and B, and Table 1). The dissociation equilibrium constant (K_d) of recombinant wild-type S. pombe profilin was 55 µM proline residues. Native wild-type profiin has a K_d value of 46 µM. For a decamer, the minimal length of poly-L-proline for high-affinity binding (Perelroizen et al., 1994; Petrella et al., 1996), this corresponds to a K_d value of 5.5 and 4.6 µM, respectively. This affinity of S. pombe profilin for poly-L-proline is slightly higher than Acanthamoeba profilin (twofold) and fivefold higher than recombinant S. cerevisiae profilin (Petrella et al., 1996; Eads et al., 1998). All stable mutants in the actin binding site bound poly-L-proline normally (Figure 4A and Table 1).

The stable mutants that targeted the poly-L-proline binding site had a range of lower affinities for poly-L-proline (Figure 4B and Table 1), but all bound actin normally (Figure 5A and Table 1). Y5D, the only poly-L-proline mutant that failed to complement both null and ts allele, had the greatest loss of affinity for poly-L-proline, 94-fold. Mutants Y120A, Y120D, and Y126D had moderately lower affinity for poly-L-proline (between 27- and 32-fold; Figure 4B and Table 1) and complemented the profilin ts strain but not the profilin null strain (Figure 2C). Mutations Y5A, W30Y, and L121A reduced the affinity for poly-L-proline <10-fold (Figure 4B and Table 1). S. pombe tolerated this small reduction in poly-L-proline binding, because all of these mutants complemented both the ts and null strains (Figure 2C and Table 1). Thus, moderate loss of affinity for poly-L-proline compromised biological function and a 100-fold loss of affinity was incompatible with life.

Characterization of Actin Binding by S. pombe Wild-Type and Mutant Profilins

Nucleotide Release Assay. We used the ability of profilin to promote nucleotide release from monomeric actin to measure the affinity of profilin for actin (Figure 5, A-C). In the low-salt conditions of our assay, Mg-ATP-rabbit muscle G-actin exchanged its bound ATP with an excess of unlabeled ATP with first order kinetics. The k_obs of 0.0065 s¹ is the dissociation rate constant for ATP (k₁), the rate-limiting step. A saturating concentration of wild-type S. pombe profilin increased the rate of exchange by 2.3-fold, up to 0.015 s¹, similar to S. cerevisiae profilin (Eads et al., 1998). The dissociation equilibrium constant (K_d) for profilin binding actin was determined from the dependence of k₁ on profilin concentration (Figure 5, A and B). By this criterion wild-type S. pombe profilin bound rabbit muscle ATP-actin monomers with a K_d value of 0.21 µM (Table 1). S. pombe profilin may bind its own actin somewhat better, because a similar profilin from another lower eukaryote, Acanthamoeba, binds its own ATP-actin more strongly (0.1 µM) than muscle actin (0.48 µM) (Vinson et al., 1998).

Effect on Actin Polymerization. Profilin inhibits spontaneous actin polymerization by preventing actin dimer and trimer formation and pointed end elongation (Pollard and Cooper, 1984). Although this assay does not measure affinity directly owing to complicated effects of profilin on multiple steps in the reaction, we could compare the activity of mutant profilins with the concentration dependence of inhibition by wild-type profilin (Figure 6, A and B, and Table 1). Our stability benchmark C89S and the severe poly-L-proline binding mutant Y5D inhibited spontaneous actin polymerization within a factor of 2 of wild-type profilin. Mutants that failed to bind actin monomers in the nucleotide exchange assay (I71E, I71R, I71W, K81E) also failed to inhibit spontaneous actin polymerization (Figure 6B and Table 1). Y79R inhibited spontaneous actin polymerization even though it did not affect nucleotide exchange (Figure 6, A and B). K81F and P107W both inhibited spontaneous actin polymerization better than they bound to actin in the nucleotide exchange assay, but neither bound actin well by this criterion.

View larger version (17K): [in this window] [in a new window]   Figure 6.   Effect of profilins on spontaneous polymerization of 5 µM Mg-actin with 0.1 µM pyrene-labeled actin measured by pyrene fluorescence. (A) Time course of polymerization of actin alone () and with wild-type profilin (), poly-L-proline binding mutant Y5D (), and actin binding mutants Y79R (), K81F (), and P107W (). (B) Dependence of the time required for half maximal polymerization on the concentrations of various profilin mutants. Symbols are same as in A with the addition of K81E () that does not inhibit actin polymerization at even 100 µM.

Phenotype of cdc3-124 Cells Expressing Wild-Type and Mutant Profilins

We studied the morphology of log phase cdc3-124 cells depending upon wild-type and mutant profilins for growth at the restrictive temperature by DIC microscopy (to observe shape and measure size) and fluorescence microscopy after staining with DAPI (to count the number of nuclei) and Calcofluor (to assess septum formation) (Figure 7, A-K, and Table 1). At 25°C, cdc3-124 cells were indistinguishable from wild-type cells in terms of their shape, size, number of nuclei, and septum staining. Most cells had one nucleus (76%); the remaining cells had two nuclei. Compact septa were present in the middle of 18% of cells. Expression of wild-type and mutant profilins in cdc3-124 strain at 25°C did not alter morphology.

View larger version (64K): [in this window] [in a new window]   Figure 7.   Morphology of cdc3-124 ts strain expressing various profilins from pRep 81 plasmid under the control of the weakest nmt1 promoter after incubation for 8 h at 36°C in the absence of thiamine. Cells were fixed and stained with DAPI and Calcofluor. Calcofluor staining was recorded first (A, C, E, G). After bleaching of the Calcofluor, DAPI staining was recorded (B, D, F, H). (A and B) cdc3-124 without transformation. (C and D) Y5D. (E and F) K81E. (G and H) Y79R. Calcofluor and DAPI were recorded in the same time for wild-type profilin (I) and K81F (J). (K) Quantitation of nucleus number in comparison with biochemical data.

At 36°C, cdc3-124 cells were defective in cytokinesis (Balasubramanian et al., 1994). After 8 h at 36°C elongated or dumbbell-shaped cells accumulated two (36%), four (54%), or even more (6.8%) nuclei (Figure 7K). Calcofluor staining was stronger than normal and most cells had abnormal septa, usually concentrated between the nuclei but also dispersed elsewhere (Figure 7, A and B). Lengths ranged from 12 to 57 µm. Episomal expression of wild-type profilin completely suppressed the cdc3-124 phenotype at 36°C. Cell size and the number of nuclei were indistinguishable from cdc3-124 at 25°C and wild-type cells at 25°C or 36°C (Figure 7, I and K, and Table 1).

Poly-L-Proline Binding Mutants. Y5D profilin with only 1% of wild-type affinity for poly-L-proline failed to complement either null or ts cells. cdc3-124 cells expressing Y5D at 36°C had a terminal phenotype similar to cdc3-124 cells (Figure 7, C, D, and K).

Actin Binding Mutants. Profilin mutants with actin binding reduced <1% of wild-type rescued neither null nor profilin ts cells. For example, cdc3-124 cells expressing profilin mutant K81E with <1% wild-type affinity did not grow at 36°C and had the same morphological defects as cdc3-124 cells (Figure 7, E, F, and K).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

By creating and characterizing a large collection of point mutations, we now have a clearer picture of the biochemical requirements for profilin function in fission yeast. We paid particular attention to two factors that might complicate interpretation of complementation experiments: protein stability and expression levels. We concentrated on those noncomplementing mutant profilins that are stable and expressed at levels close to endogenous profilin in the wild-type strain.

We were fortunate that the cdc3-124 mutation has a dramatic phenotype at the restrictive temperature, yet episomal expression of wild-type profilin completely suppresses these abnormalities. Interpretation of the functional tests in vivo depended on quantitative analysis of the biochemical properties. Comparison of quantitative biochemical assays with cellular structure and function established how affinity and catalytic activity are related to viability (Figure 8). Exploring a spectrum of biochemical properties required a survey of many different mutants, because it was impossible to predict these properties from the structures. Previous studies of mutant profilins relied largely on qualitative biochemical characterization. Few articles reported binding constants for profilin mutants tested in vivo (supplemental materials, Table 1).

View larger version (20K): [in this window] [in a new window]   Figure 8.   Dependence of complementation of null and ts mutants on the affinity of profilin mutants for actin and poly-L-proline. Reduced affinities within 10-fold of wild-type profilin (>0.1 WT) are tolerated both in profilin null and ts background; loss of affinities between 10- and 50-fold (>0.02 WT) are able to rescue the ts allele but not the null; >100-fold loss of affinities (<0.01 WT) are incompatible with life. Altered nucleotide exchange activity of actin binding mutants also determined their complementing ability. "?" indicates no mutants to test in these categories in the group with affinity >0.02 WT.

For practical reasons, we used readily available ligands (poly-L-proline and rabbit muscle actin) to characterize our mutant profilins rather than S. pombe actin and the potential poly-L-proline ligands Cdc12p (Chang et al., 1997), verprolin, and Wsp1p. Although using these yeast ligands would have been desirable, none has been purified in quantities sufficient for testing numerous mutants. We feel that the use of rabbit muscle actin is justified by the fact that the profilin binding sites are almost identical on vertebrate and S. pombe actins. Among 22 residues that interact directly with profilin in the crystal structure (Schutt et al., 1993) only residue 371 differs; histidine in muscle actin and tyrosine in S. pombe actin. This must account for the small differences of affinity of, for example, Acanthamoeba profilin for amoeba and muscle actin (Vinson et al., 1998) and the threefold difference in the ability of S. cerevisiae profilin to promote nucleotide release from rabbit muscle and yeast actins (Eads et al., 1998). Nevertheless, the loss of affinities of mutant yeast profilins for poly-L-proline and muscle actin parallel closely the severity of their phenotypes. On the other hand, profilin residues that interact with actin are very variable. Thus, differences in the ability of profilins to promote nucleotide exchange by actin are attributable to profilin alone.

Structural Stability

Protein instability is a potential trivial explanation for the failure of any point mutation in a complementation test. The cdc3-124 mutation E42K illustrates how an apparently harmless amino acid substitution on the surface of the protein well away from any known ligand binding site disables the protein by making it severely unstable. Deletions of residues from the C terminus (Kaiser and Pollard, 1996) also compromise the stability of profilins.

We measured the stability of mutated profilins by urea denaturation, a test similar to thermal denaturation. At the midpoint of these titrations half of the protein is unfolded. Few previous studies of profilin mutants included this key information (supplementary materials, Table 1). Because mutant C89S denatured at 3.2 M urea but was indistinguishable from wild-type profilin in ligand binding and ability to complement the profilin null mutation, we conclude that any mutant denaturing at a urea concentration >3.2 M is stable and that failure to function in vivo is attributable to loss of one or more biochemical activities. Of the profilin mutants with amino acid substitutions on the surface that failed to function in complementation tests, one-third were unstable by this criterion. Wild-type S. pombe profilin is remarkably stable (4.5 M) compared with S. cerevisiae profilin (3.4 M) and Acanthamoeba profilins (3.5 M). S. pombe profilin mutant K67A is also much more stable (4.4 M) than the corresponding S. cerevisiae mutant K66A (2.7 M) (Wolven et al., 2000). It is not known whether other organisms tolerate unstable mutant profilins better than S. pombe.

Strategy for the Poly-L-Proline Site. The poly-L-proline binding site on profilin includes the phylogenetically most conserved amino acids, notably, W2, Y5, W30, Y120, and Y126. These aromatic residues contribute to binding poly-L-proline through hydrophobic interactions and donating hydrogen bonds to backbone carbonyls (Petrella et al., 1996; Mahoney et al., 1997). We avoided tryptophan-2, because in our experience with Acanthamoeba profilin, mutation of this tryptophan usually compromised stability (Vinson and Pollard, unpublished observations). Similarly, human profilin mutant W3F eluted from a poly-L-proline affinity column with 3 M urea, which might indicate a loss of stability (Ostrander et al., 1999). Human profilin W3N lost not only poly-L-proline binding but also actin binding; it too may have been poorly folded, although this was not tested (Bjorkegren-Sjogren et al., 1997). We explored other residues in the poly-L-proline binding site, namely, Y5, Y120, L121, and Y126, because substitutions were tolerated and all are crucial for poly-L-proline binding.

Strategy for the Actin Site. Profilin tolerated many mutations of the actin binding surface without a loss of structural stability. However, mutants with tryptophan substitutions on the surface were frequently unstable (K67W, I69W, S77W, K84W).

Conditions for Survival

Ability of profilin to provide fission yeast with biological function depends on its affinity for both actin and poly-L-proline and ability to catalyze nucleotide exchange on actin (Figure 8 and Table 1). Profilin mutations reducing affinity for poly-L-proline or actin >100-fold were unable to complement either profilin null or ts strains, so both poly-L-proline binding and actin binding are essential for the organism. Mutant profilins with affinities for either ligand reduced 20-50-fold complement the ts strain but not the null. Residual function of E42K at the restrictive temperature appears to allow these partially disabled mutant proteins to support viability. The ability of E42K expression from the nmt1 promoter to complement the ts strain verifies residual function of E42K profilin at the restrictive temperature in spite of no protein detected on immunoblots. Profilins with near wild-type affinity complemented both null and ts strains.

Because the phenotype of cdc3-124 cells expressing profilin mutants with drastic losses in affinity for either poly-L-proline (Y5D) or actin (K81E) or actin nucleotide exchange (Y79R) were indistinguishable, all three activities are likely to be required simultaneously for normal function. Similarly, mutants of S. cerevisiae profilin designed to interfere with poly-L-proline binding (W29A, W2A, Y119A) have the same phenotype as profilins with mutations in the actin binding site (Wolven et al., 2000). These budding yeast profilin mutants may benefit from more detailed biochemical analysis, because it is impossible to predict the properties of these substitutions.

Poly-L-Proline Ligand Binding

Poly-L-proline binding is essential for profilin function in fission yeast. With one interesting exception, the defects in poly-L-proline binding paralleled the loss of activity in complementation assays. Residue Y5 is noteworthy, because S. pombe mutant Y5A with only a 3-fold loss of affinity for poly-L-proline barely complemented the ts strain. We speculate that Y5 is important for interaction with a type II polyproline helix containing residues other than proline, perhaps for binding cdc12 during cytokinesis (Chang et al., 1997). This residue was mutated twice previously. Human profilin with the Y6F substitution eluted from a poly-L-proline affinity column with 3.5 M urea (7.5 M urea for wild type), so it either lost stability or affinity for poly-L-proline (Sohn et al., 1995). No other properties were reported for this mutant. Maize profilin with the same Y6F substitution bound actin normally and bound poly-L-proline with affinity twofold higher than wild type (Gibbon et al., 1998). This mutant profilin was more active than wild-type profilin in disrupting the actin cytoskeleton of stamen hair cells.

Actin Binding and Nucleotide Exchange Promoted by Profilin

Although viability generally correlated with affinity for actin, simple loss of affinity for actin cannot explain the phenotypes of all of the actin site mutants. Mutant profilins Y79R, A111E, P107W, and K81F all had actin affinities in the range of 2-5% of wild type but varied in biological function. Their ability to complement profilin null and ts strains paralleled their nucleotide exchange activity. K81F (also K81Y, Table 1) had better nucleotide exchange activity than wild-type profilin and complemented both null and ts strains. A111E and P107W had nucleotide exchange activity similar to wild-type profilin and complemented the ts but not the null strain. Y79R lacking nucleotide exchange could not even complement the ts strain, like mutants without detectable actin binding. These data strongly indicate that nucleotide release is an essential function of profilin in S. pombe.

The biological relevance of profilin promoting nucleotide exchange by actin monomers has been debated since its discovery (Mockrin and Korn, 1980). Profilins vary widely in ability to catalyze nucleotide exchange: human profilin increases nucleotide exchange 40- to 1000-fold relative to actin alone (Goldschmidt-Clermont et al., 1992; Perelroizen et al., 1996; Selden et al., 1999); Acanthamoeba profilin increases nucleotide exchange 8- to 17-fold (Mockrin and Korn, 1980; Vinson et al., 1998); and S. cerevisiae profilin increases nucleotide exchange only threefold (Eads et al., 1998). In each case, the rate of exchange depends not only on the profilin but also on the type of actin, bound nucleotide (ADP or ATP), bound divalent cation (Mg- or Ca-), and buffer conditions.

Plant profilins raise serious questions about the importance of nucleotide exchange activity. Profilins from Arabidopsis, birch, and maize lack nucleotide exchange activity in assays with muscle and plant actins (Perelroizen et al., 1996; Eads et al., 1998; Kovar et al., 2000), although they bind actin (Giehl et al., 1994; Kovar et al., 2000) and promote assembly of actin sequestered by thymosin-4 (Perelroizen et al., 1996; Ballweber et al., 1998). The relevance of the latter activity is not clear given the absence of thymosin from the Arabidopsis genome. Arabidopsis profilin complemented both the profilin null strain of S. cerevisiae and the cdc3-124 ts strain of S. pombe (Christensen et al., 1996). Expression of maize profilin in Dictyostelium lacking both endogenous profilins suppressed the aberrant cell shape, and increased actin filament staining and developmental defects (Karakesisoglou et al., 1996).

On the other hand, studies of actin filament dynamics have revived the interest in a role for the nucleotide exchange activity of profilin. In vitro profilin overcomes the inhibition of ADP exchange by ADF/cofilins and recycles ATP-actin monomers back to actin filaments (Blanchoin and Pollard, 1998, 1999; Didry et al., 1998; Wolven et al., 2000). When actin turnover is driven by continuous sonication, profilin increases the polymer concentration by accelerating the rate-limiting exchange of ATP for ADP bound to actin monomers (Selden et al., 1999). In S. cerevisiae an actin mutation with a high rate of intrinsic nucleotide exchange overcomes a profilin mutation that reduces its affinity for actin both in in vitro filament turnover assays and in vivo (Wolven et al., 2000). This supports the importance of actin nucleotide exchange under physiological conditions. Our analysis both in vivo and in vitro of fission yeast profilin mutants that only differ in their nucleotide exchange activity provides the strongest evidence so far of the crucial requirement of nucleotide exchange activity for viability, at least in fission yeast.

Additional work is required to reconcile these differences among species. We suggest that the following be considered. 1) S. pombe may rely on the nucleotide exchange activity of profilin more than S. cerevisiae and Dictyostelium, because S. pombe absolutely requires profilin, whereas S. cerevisiae and Dictyostelium are viable (although very sick) without profilin. 2) Other molecules may supplement the nucleotide exchange activity of profilin in S. cerevisiae and Dictyostelium. This is unlikely in S. pombe because null strains are not viable unless provided with a profilin having nucleotide exchange activity. 3) The crucial variable among these species may be the intrinsic rate of nucleotide exchange by actin alone. Maize pollen actin has a higher intrinsic exchange rate than other actins (Kovar et al., 2000) and S. cerevisiae actin with a high exchange rate suppresses defects in profilin (Wolven et al., 2000). Nothing is yet known about the nucleotide exchange rate of S. pombe actin and all of these actins still need to be compared rigorously under identical conditions. 4) The concentrations and activities of other actin binding proteins, particularly ADF/cofilins, need to be considered.

The residue corresponding to fission yeast profilin Y79 is potentially important for nucleotide exchange. Mutant Y79R lacks nucleotide exchange activity and all plant profilins known to lack nucleotide exchange activity (Arabidopsis, birch, and maize) have R at this position (Perelroizen et al., 1996; Eads et al., 1998; Kovar et al., 2000). In addition to the 10 profilins from these three species, we checked 26 other plant species with 39 profilin sequences. Out of a