Structure and macromolecular assembly of two isoforms of the major sperm protein (MSP) from the amoeboid sperm of the nematode, Ascaris suum

Nematode sperm are unique amoeboid cells that display the crawling movements typically associated with actomyosin-rich cells although they contain little actin or myosin (reviewed by Roberts, 1987; Ward, 1986). Sperm crawl by extending a motile pseudopod that is packed with fine filaments composed of major sperm protein (MSP), a family of small, basic, spermspecific polypeptides (reviewed by Roberts et al. 1989). In sperm from the large parasitic nematode, Ascaris suum, these filaments are organized into 15–20 intermeshed, branched fiber complexes that extend from protrusions, called villipodia, along the leading edge of the pseudopod 20–25 μm rearward to the organelle-packed cell body. Each fiber complex contains multiple filaments that splay out radially from a central core to interdigitate with similar filaments from neighboring complexes, so that the MSP cytoskeleton is arranged as an interconnected unit (Sepsenwol et al. 1989).

As sperm crawl, the fiber complexes appear to flow, or treadmill, rearward due to continuous addition of filament mass at the distal end coupled to disassembly of the complexes at the opposite end (Sepsenwol et al. 1989; Sepsenwol and Taft, 1990). This pattern is generally similar to the centripetal flow of actin filament arrays in other amoeboid cells (Bray and White, 1988; Fisher et al. 1988; Heath and Holifield, 1991; Wang, 1985) and neuronal growth cones (Forscher and Smith, 1988; Smith, 1988). Like the flow of actin filaments in the lamellipodium of locomoting goldfish kératinocytes (Theriot and Mitchison, 1991), the MSP filament system in sperm moves centripetally at the same rate as the cell crawls forward (Roberts and King, 1991). This close association between cytoskeletal flow and cell locomotion argues that MSP is a central component in the molecular interactions that underly the amoeboid movement of the cell.

The novel and apparently simple motile machinery of Ascaris sperm, based primarily on MSP, makes these cells a valuable complement to actomyosin-based cells for understanding the fundamental principles of amoeboid movement. But to exploit this system, we need the same detailed knowledge about MSP as is available for actin. Therefore we investigated the nature of the macromolecular assemblies MSP forms in vivo and in vitro, isolated the protein, and characterized its molecular structure. Here we demonstrate that either chemical fiaxation or polyethylene glycol (PEG) stabilizes MSP filaments from lysed Ascaris sperm. We show that sperm contain two isoforms of MSP, designated α and β, which have similar but not identical protein sequences. We have devised conditions under which MSP can be induced to form filaments in vitro that resemble closely those in lysed cells. Moreover, we describe conditions whereby crystals of MSP can be produced in which the protein appears to be arranged into filaments that may parallel those found in vivo and in vitro.

Ascaris males were collected from the intestine of infected hogs at Premium Pork Processing (Moultrie, GA) and transported to the laboratory in phosphate-buffered saline at 38–39°C. Spermatids were isolated by draining the contents of the seminal vesicle into tubes containing HKB buffer (50 mM HEPES, 65 mM KC1, 10 mM NaHCO₃, pH 7.6), washed twice, and treated with vas deferens extract, prepared as described by Sepsenwol et al. (1986), to activate differentiation into motile spermatozoa. After activation, the cells were either used immediately or pelleted by brief centrifugation at 10,000 g, frozen rapidly in liquid N₂, and stored at −70°C.

Spermatozoa were pipetted onto carbon-coated, 200-mesh nickel grids rendered hyprophillic by glow discharge. Grids were checked by light microscopy for active, crawling spermatozoa before lysing by treatment with HKB containing 0.5% Triton X-100. MSP filaments in the demembranated cells were prevented from disassembling either by fixation in 1% glutaraldehyde within 2–3 s after lysis, or by addition of 20% polyethylene glycol (PEG; average M_r 18,500; Polysciences, Inc.) to the lysis solution followed by fixation in 1% glutaraldehyde. These specimens were washed with several drops of HKB, then either negatively stained with 1% aqueous uranyl acetate and air dried or stained with 1% uranyl acetate for 1 h, dehydrated to absolute ethanol, and critical point dried by the method of Ris (1985). Both types of preparation were examined in a JEOL 1200CM electron microscope operated at 100 kV. Some PEG-treated cells were prepared for scanning EM by the method described, except that uranyl acetate staining was omitted and the dried grids were coated with gold/palladium. These cells were examined in a JEOL 840 SEM operated at 10 kV.

Frozen sperm were thawed and homogenized by 10 strokes in a tight-fitting tissue grinder. The homogenate was centrifuged at 10,000 g; the pellet was resuspended in 1 ml of HKB buffer and respun. The pooled supernatants were spun at 100,000 g and the resulting supernatant (S100) concentrated 2- to 3-fold by vacuum dialysis and stored at −70°C.

SDS-PAGE gels were run according to Laemmli (1970) on 15% polyacrylamide slabs. Two dimensional non-equilibrium pH gradient electrophoresis (NEpHGE) of S100 was carried out using a BioRad minigel system according to the procedures outlined by O’Farrell et al. (1977). Samples of S100 dissolved in urea sample buffer were loaded onto ureaacrylamide tube gels containing 5% pH 3.5-10 Ampholines. Samples were run at 200 V for 10 min, then at 400 V for 90 min, extruded and equilibrated in SDS sample buffer, and loaded onto SDS-polyacrylamide slab gels for separation in the second dimension. Gels were stained in Coomassie brilliant blue or blotted onto nitrocellulose as described by Towbin et al. (1979). Blots were probed with 25 μg of antibody AZ10, a monoclonal anti-MSP antibody (Sepsenwol et al. 1989), followed by 25 ug/ml horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Jackson Immunoresearch Laboratories, West Grove, PA). Positions of labelled spots were detected with 4-chloro-l-naphthol.

MSP was partially purified from S100 by size-exclusion chromatography on a Sephadex G-75 superfine column (2.5 cm × 90 cm) equilibrated with HKN buffer (HKB with NaCl in place of NaHCO₃). MSP-enriched fractions were pooled and concentrated to 25–30 mg/ml total protein by vacuum dialysis. The two isoforms of MSP were separated by HPLC cation exchange chromatography on a Waters 1 cm × 7.5 cm SP-5PW column equilibrated with 5 mM NaH₂PO₄, pH 6. Typically 5–6 mg total protein were loaded and the column washed with equilibration buffer for 5 min at a flow rate of 1 ml/min. MSP was eluted in a 1%/min linear gradient of 100 mM Na₂HPOyNaH2PO₄, pH 7. Under these conditions one form of MSP eluted from the column at about 18 min followed by the other at 25 min. Fractions were collected by hand, concentrated by vacuum dialysis, and assessed for purity by SDS-PAGE and/or reversed-phase HPLC. The two forms also separated readily on reversed phase, eluting between 30 and 40% acetonitrile in a linear gradient of 0.25%/min with 0.1% aqueous trifluoroacetic acid (TFA) as mobile phase A and 0.08% TFA in acetonitrile as mobile phase B. Cation exchange was selected as the separation method of choice due to higher loading capacity and non-denaturing chromatographic conditions.

Purified MSP was lyophilized, resuspended in 0.1% TFA, filtered into a preweighed tube and lyophilized again. The dried protein was weighed, resuspended in 1 ml of 0.1% TFA and scanned by UV spectrophotometry. Alternatively, triplicate samples were filtered and scanned, then transferred to preweighed tubes, lyophilized and weighed. Additional samples were handled in 20 mM ammonium acetate, pH 6.8, in place of TFA. The mean extinction coefficient (0.1%, 280 nm, 1 cm) calculated for 12 samples was 0.9±0.1.

Each of the two purified isoforms of MSP was reduced in 6 M guanidine HO, 2 mM EDTA, 10 mM dithiothreitol (DTT), 100 mM Tris-HCl, pH 8.2, at a protein concentration of 1–2 mg/ml. Solutions were sealed under Nj, incubated for 3 h at 37°C, then alkylated by addition of 12 mM iodoacetate for 90 min at 25°C in the dark. Preparations were desalted on a 0.7 cm × 10 cm column of Sephadex G-25 equilibrated in 50 mM ammonium bicarbonate and vacuum dialyzed to a protein concentration of at least 2 mg/ml.

Reduced and alkylated MSP, typically 200 μg in 100 μl of ammonium bicarbonate, was digested with trypsin (Calbiochem) at an enzyme:substrate ratio of 1:50 (w/w) or endoproteinase Lys-C (Calbiochem) at 0.5 unit/200 μg MSP. After 5 h at 37°C, digests either stored at −70°C or injected directly into an Aquapore RP-300 reversed-phase HPLC column with a TFA/acetonitrile solvent system as described above. Elution was monitored at 214 nm. Fractions were lyophilized and repurified by a second reversed-phase HPLC run prior to sequence analysis.

For cleavage at methionine residues, MSP was incubated with cyanogen bromide (CnBr:MSP=2; w/w) in 0.12 M HC1 with the volume adjusted to give a protein concentration of 2 mg/ml. After 17–24 h in the dark at 25°C, the reaction was diluted 5- to 10-fold with distilled water and lyophilized to remove excess reagent. Peptides were separated and repurified on reversed-phase HPLC as described above. We used the sequence of an Ascaris MSP cDNA clone reported by Bennett and Ward (1986) to predict likely cleavage sites and order sequenced peptides in regions of non-overlap.

Sequential degradation of the peptides was performed with an Applied Biosystems, Inc. model 477A pulsed-liquid phase sequenator using Normal-1 cycles. Amino acid residues were identified by on-line phenylthiohydantoin analysis using an Applied Biosystems model 120A HPLC.

Each isoform of MSP yielded a blocked N-terminal peptide (see Results). To sequence these fragments, each was hydrolyzed in 25% TFA for 2 h at 55°C. This yielded a set of overlapping, reproducible subfragments that were separated and repurified by reversed-phase HPLC and sequenced as described above. In subsequent trials the reaction time was reduced to 40 min, then to 10 min, again yielding reproducible cleavage patterns and sequencable peptides.

Samples of the two isoforms of MSP and their N-terminal peptides, purified as described above, were lyophilized from TFA/acetonitrile and sent to Dr. Alexander Buko, Abbott Laboratories, Abbott Park, IL, who generously conducted mass spectroscopic analyses. Samples were placed in solution onto a nitrocellulose-covered mylar target, dried, and inserted into an Applied Biosystems BIOION-20 plasma desorption mass spectrometer and ionized by irradiation with

Effects of protein precipitants on MSP in solution were tested by mixing purified MSP with precipitant to yield the desired final concentration of both. All precipitants were tested against 10 mg/ml MSP at concentrations up to 40%, except 1-butanol, 2-butanol, 2-octanol and isoamyl alcohol, which were tested as saturated solutions in HKB. After incubation for 15 min, a 4–5 μl sample was pipetted onto a carbon-coated, glow-discharged EM grid. After 10–15 s, the grid was washed with 4–5 drops of 1% uranyl acetate, negatively stained, and examined as described above. Preliminary experiments revealed that incubation in selected alcohols caused MSP to assemble into filaments.

Thin, needle-shaped crystals of both isoforms of MSP were formed by incubating purified protein at 3–5 mg/ml in HKB containing 15% PEG 18,500. Crystals for EM analysis were obtained by floating a carbon-coated copper grid on the solution of MSP in PEG for 5–10 min. Crystals adhering to the carbon film were washed free of excess PEG with several drops of uranyl acetate. Excess stain was wicked off with filter paper and the grid was air dried. Negatively stained crystals were examined in either a JEOL 1200CM or a Philips EM 400 electron microscope at either 80 or 100 kV.

The MSP filament system in the pseudopods of Ascaris sperm is extrordinarily labile and usually the cytoskeleton disassembles within a few seconds after lysis (our unpublished observations). To investigate the structure of the MSP filaments more fully, we tried to develop conditions under which this depolymerization of MSP filaments would be inhibited or at least slowed. We explored a broad range of combinations of detergents, buffers, inorganic ions, nucleoside triphosphates, and pH, but were not able to define a suitable protocol. However, we were able to obtain some structural information by using chemical fixation or by using a known protein precipitant, polyethylene glycol (PEG), to stabilize the MSP filaments. For example, we were able to fix native filaments suitable for negative staining by plunging cells on carbon-coated EM grids into 1% glutaraldehyde within 1–2 seconds after lysis in 0.5% Triton X-100 (Fig. 1). In favorable preparations, much of the filament array was removed, leaving behind a thin mat of fibers on the support film. In some cases remnants of the fiber complexes could still be discerned. The individual filaments in these fixed preparations were about 10 nm wide; some appeared to contain a longitudinal cleft down the center, suggesting that the filament was composed of subfibrils wound around one another (Fig. 1, inset).

Inclusion of 20% PEG 18,500 in solutions containing 0.5% Triton X-100 stabilized the MSP filament array without requiring chemical fixation. PEG stabilization yielded demembranated sperm in which the cytoskeleton was stable for up to 6 h as judged by electron microscopy. Because much of the dense array of filaments that packs the pseudopod (Sepsenwol et al. 1989) remained, these preparations were usually too thick for satisfactory negative staining. It was, however, possible to examine positively stained, critical point-dried whole mounts by TEM or to view gold/palladium-coated whole mounts by SEM (Fig. 2). Lysis in PEG often resulted in loss of part of the cytoskeleton at the base of the pseudopod, but the remaining filaments appeared to be intact and uniform in diameter. Lower concentrations of PEG 18,500 and lower molecular weight PEGs were not effective in stabilizing the filament array.

To examine the structure, composition and assembly of these unique filaments, we isolated and purified MSP. In so doing we found that Ascaris sperm contain two isoforms of the protein. As shown in Fig. 3 (lane a; cf. Nelson and Ward, 1981), MSP is the most abundant protein in these cells, accounting for about 15% of the cellular total. MSP was enriched to approximately 40% of the soluble protein (S100) obtained by ultracentrifugation of sperm homogenates. Gel filtration chromatography yielded further enrichment and gave a fraction that separated into two electrophoretically pure species by cation exchange HPLC (Fig. 3, lanes c-e). Western blot analysis of two-dimensional NEpHGE/SDS-PAGE gels of S100 showed that these two species, which exhibit a slight difference in M_r but a large difference in pl, were both recognized by monoclonal anti-MSP antibody AZ10 (Fig. 4). Analytical isoelectric focussing (not shown) indicated that these two forms had pl values of 8.9 and 8.3.

To determine if these species were different isoforms of MSP rather than post-transcriptional modifications, we separated them by either cation exchange or re versed-phase HPLC and determined their amino acid sequences. The more basic form eluted first from the reversed-phase column, indicating lower hydrophobicity than the less basic form. The two forms, as expected from their pl values, eluted in the opposite order by cation exchange. Both procedures yielded purified proteins suitable for sequencing (Fig. 3). We analyzed 24 peptides from each form to obtain the complete sequences shown in Fig. 5. These data showed that the two species were independent polypeptides, each 126 residues long, that were identical in sequence except at four positions (14, 15, 54 and 67). In keeping with the nomenclature adopted for isoforms of actin (Rubenstein, 1990) and tubulin (Joshi and Cleveland,1990), and to discriminate between these polypeptides and sequenced MSP genes, which are identified by numerical suffixes (Ward et al. 1988), we designated the more basic form αMSP and the other βMSP. Alpha-MSP corresponds to the Ascaris MSP sequence predicted from a cDNA clone by Bennett and Ward (1986).

The N-terminal peptide for each isoform, which was obtained by either proteolytic digestion or cyanogen bromide cleavage and which was identified by its amino acid composition, failed to sequence by automated Edman degradation. This observation suggested the presence of a blocked N terminus. Sequences from sets of overlapping subfragments obtained by limited acid hydrolysis of the N-terminal peptides, indicated that the initiator methionine was missing from the mature proteins. Mass spectroscopy of the N-terminal peptides confirmed their sequences and yielded an observed mass consistent with an acetylated N-terminal alanine for both isoforms (Table 1). The calculated M_r of intact αMSP after N-terminal processing is 14,302, 142 greater than β-MSP. These values are within the range of error of measurement of the observed masses of the intact polypeptides obtained by mass spectroscopy (Table 2).

The capacity of PEG to stabilize MSP filaments prompted us to test the effects of several protein precipitants. We found that MSP assembled into filaments in the presence of a wide range of water-miscible alcohols. Ethanol, methanol, isopropanol, n- propanol and 2-methyl-2,4-pentanediol (MPD) promoted filament assembly, whereas allyl alcohol, 1-butanol, 2-butanol, 2-octanol, isoamyl alcohol, polyvinyl alcohol, ethylene glycol, propylene glycol, PEG (average M_r 600, 1400, 6000 and 18,500) and glycerol did not. Fig. 6 shows filaments assembled in ethanol from a mixture of α- and β-MSP. When negatively stained, these filaments appeared to be about 10 nm wide with a characteristic substructure repeating axially at about 9 nm. The appearance of these filaments suggested that they were possibly constructed from two subfibrils (Fig. 6, inset). At this level of resolution, the diameter and substructure of filaments assembled in ethanol and native filaments isolated from detergentlysed, glutaraldehyde-fixed sperm were indistinguishable (Fig. 7).

Purified α and β-MSP assembled into filaments in ethanol under slightly different conditions. At high protein concentration (6 mg/ml), α-MSP formed filaments with ethanol concentrations of 20% and above, whereas at the same protein concentration α-MSP failed to form filaments at ethanol concentrations below 25%. Assembly of both isoforms occurred over a pH range from 5.7 to 9.5 and at temperatures from 2 to 39°C. Assembly reactions were normally carried out in HKB buffer, but none of the components of this buffer system was essential: MSP in deionized water, for example, assembled into filaments when ethanol was added to 30%.

We used pelleting assays to determine the critical concentration for assembly of α- and β-MSP in 30% ethanol (Fig. 8). During continuous centrifugation at 120,000 g the concentration of both isoforms decreased from a starting value of about 5 mg/ml to 2.6 mg/ml in 15 min and remained constant thereafter. The same equilibrium concentration was obtained for samples with starting concentrations ranging from 3 to 10 mg/ml, and the supernatants lacked filaments detectable by electron microscopy, indicating that the critical concentration for assembly under these conditions was about 0.2 mM for both isoforms. The filaments in the pellets from these preparations disassembled completely in HKB buffer, so that we could account for all of the starting protein in either the supernatant or the redissolved pellet. The critical concentration of MSP in HKB buffer alone was about 10 mg/ml. However, because of the large amount of unpolymerized material in these preparations, we were not able to demonstrate unequivocally the presence of filaments in the pellets by negative staining. Therefore, it may be that the critical concentration measured in HKB buffer alone represented that for precipitation rather than for actual filament formation.

Another indication of the rate at which MSP filaments assemble and disassemble was obtained by pipetting 5 ml of a solution of 5 mg/ml βMSP onto an EM grid. After withdrawing all but a thin film of the solution from the surface, the grid was plunged into 30% ethanol for 5 s, washed with several drops of uranyl acetate, and negatively stained. The grid contained numerous filaments, whereas no filaments were found on control grids for which incubation in ethanol was omitted. MSP filaments that formed on grids plunged into ethanol also disassembled rapidly. If these grids were transferred from ethanol to HKB buffer for 5 s before treatment with uranyl acetate, no filaments were found.

Although PEG stabilized MSP filaments in vivo, we found that crystals rather than filaments formed when either α or βMSP was mixed with PEG 18,500 in vitro. The crystals were generally needle-shaped and their formation and growth was easily observed by light microscopy (Fig. 9). Often crystals up to 0.5 mm long and 0.1 mm wide could be produced in this way. Crystals started to form within 5 min after mixing MSP at a minimum of 3 mg/ml with 15% PEG at 22°C. Lower M_r PEGs did not promote crystallization under these conditions.

Electron microscopy of thin areas of the crystals that had been negatively stained (Fig. 10) showed a pattern of distinct longitudinal striations spaced 3.6 nm apart. The striations themselves tended to follow a slightly sinusoidal path, most easily appreciated if the micrograph in Fig. 10 is viewed at an angle from below. Optical diffraction patterns (Fig. 10, inset) showed three strong spots on row lines corresponding to a transverse spacing of 1/3.6 nm⁻¹, but also often showed a weak row line at a spacing of 1/7.2 nm⁻¹ (arrow in Fig. 10, inset), indicating that the true unit cell was double the striation spacing. The optical diffraction patterns did not show any strong reflections along the meridian, but the strong spots along the 1/3.6 nm⁻¹ row lines corresponded to an axial spacing of 9 nm and correlated with the sinusoidal path observed along the striations. There were, however, weak but quite distinct reflections corresponding to an axial spacing of about 45 nm between the strong reflections along each row line, and so it seems likely that the true unit cell in projection was rectangular, with dimensions of 7.2 nm × 45 nm.

Because they exhibit the same pattern of locomotion as their actin-rich counterparts, including pseudopod extension, protrusion, ruffling and cytoskeletal flow, nematode sperm are a valuable alternative for investigating the forces that propel crawling movement in eukaryotic cells. The simplicity of the molecular apparatus involved in nematode sperm motility, being based primarily on MSP, should facilitate investigation of the molecular mechanism of their locomotion and so complement studies on actin-based systems. Moreover, comparison of MSP- and actin-based systems should produce a more comprehensive understanding of the fundamental principles of amoeboid motility. To lay the foundation for investigating the molecular role of MSP in amoeboid motility we have characterized in detail the protein and the filaments it forms. We have examined stabilized filaments from Ascaris sperm, purified and sequenced the two isoforms of MSP, and defined conditions for in vitro assembly of filaments indistinguishable in structure from native fibers. In addition, we have identified conditions for formation of crystals with helical substructure that should enable both the molecular structure and its assembly into helical aggregates to be studied at high resolution.

The differences in the sequences of αand β-MSP are consistent with their observed differences in charge and SDS-PAGE mobility. The lysine residue at position 67 renders αMSP more basic than β MSP, which has serine at this position. The M_r of α MSP is slightly greater than that of β MSP although we suspect that the difference in SDS-PAGE M_t between the two isoforms probably arose from retention of small elements of secondary structure or different numbers of charged residues on each (see, for example, Noegel et al. 1989). Mass spectroscopy of the N-terminal fragments, together with sequence analysis of subfragments generated by acid hydrolysis, showed that both isoforms were modified post-translationally to yield an acetylated N-terminal alanine. This pattern of N-terminal processing is common to many eukaryotic cytoplasmic proteins (Aitken, 1990; Arfin and Bradshaw, 1988), including actins (Rubenstein and Martin, 1983; Solomon and Rubenstein, 1985). The close agreement between the observed and calculated masses of the intact Ascaris MSP polypeptides indicated that these isoforms, like those of Caenorhabditis elegans (Burke and Ward, 1983), undergo no further post-translational modification.

Alpha- and β-MSP are the first members of this family of polypeptides to be sequenced directly from isolated protein, although predicted amino acid sequences have been reported from cloned MSP cDNAs from Ascaris (Bennett and Ward, 1986), C. elegans (Klass et al. 1984, 1988; Ward et al. 1988) and Onchocerca volvulus (Scott et al. 1989). Our observation that a- and β MSP differ at only four residues is consistent with accumulated sequence data that show that the MSPs are highly conserved both within and between species. In C. elegans, for example, over 30 MSP genes are expressed in roughly equal amounts, but produce only three isoelectric variants of the protein (Klass et al. 1988). Likewise, two MSP genes in O. volvulus encode proteins that differ by only five amino acids (Scott et al. 1989). Among the three species for which sequence data are available, even the most dissimilar MSPs are over 80% homologous and most of the amino acid substitutions are conservative (Scott et al. 1989).

We were unable to detect any homology between the MSP amino acid sequence and those of other cytoskeletal proteins such as actin, tubulin, myosin and intermediate filament proteins. Thus, MSP lacks the putative nucleotide and cation binding sites as well as the post-translational modifications associated with regulation of assembly of other types of cytoskeletal elements (reviewed by Aebi et al. 1988; Kirschner and Mitchison, 1986; Pollard and Cooper, 1986; Stewart,1990). Moreover, systematic searches of sequence data bases (EMBL, Swiss Protein) failed to recognise any similarities to other proteins except for MSPs from other nematode species. Examination of the likely secondary structure of MSP using techniques such as that described by Garnier et al. (1978) did not show any striking patterns that might indicate some unusual structural feature (such as an ohelical, coiled-coil conformation). Therefore, at least at the level of primary sequence, MSP appears to be a unique protein family. However, it may well be that structural similarities to other proteins do exist at the level of three-dimensional structure as found, for example, between actin and the heat shock cognate protein HSC-70 (Flaherty et al. 1991).

In nematode sperm MSP appears have the central role in amoeboid motility that is usually associated with actin in other eukaryotic cells. Such an observation invites comparison between the two proteins, but most of the similarities that emerge appear to be rather superficial. Most cells, for example, contain multiple isoforms of actin (Rubenstein, 1990) and, although a number of MSP isoforms also appear to be present (two in the case of Ascaris but 30 for C. elegans’, Ward et al.1988) it is not clear that any functional significance can be attached to the presence of isoforms. Both actin and MSP form helical filaments that have a strong twofold helical component, but these helices do not appear to have any marked structural similarities. Both filament systems do appear to be dynamic and it seems very likely that there is a relationship between filament polymerization/depolymerization and motility. Actin has a comparatively low critical concentration (about 0.1 μM under physiological conditions; Pollard, 1986), which is roughly 2000-fold lower than that of MSP. Moreover, actin polymerization is modulated by accessory proteins and involves ATP, whereas neither nucleotide binding nor accessory protein binding has been demonstrated for MSP.

The similarity between actin and MSP may be related more to the mechanism by which they produce motility rather than to their molecular structure and assembly dynamics. In this context, there may be an interesting feature of the compounds we used to induce the formation of macromolecular assemblies (both fila ments and crystals) of MSP. These compounds are commonly used to crystallize proteins. Although the mechanisms by which they operate are not fully understood, PEG is thought to alter the structure of water (McPherson, 1985b), whereas alcohols and MPD bind to water and alter its dielectric constant (McPherson, 1985a). In all cases, one property of these materials is probably to reduce the concentration of free water. These organic solvents all reduce the solubility of MSP in a way that allows individual molecules to interact in an orientation that promotes the formation of filaments. Thus, the polymerization process itself may result in the liberation of free water. Should this also occur in vivo, then the observation that fiber complexes assemble continuously at the front of the advancing pseudopod would support Oster’s (1989) proposal that protrusion of amoeboid cells is driven by local osmotic forces.

Definition of conditions that promote assembly of MSP filaments in vitro emerged from our observation that high concentrations of high molecular weight PEG stabilized the cytoskeleton of detergent-lysed sperm. PEG has been used for cytoskeletal stabilization in other types of cell, but usually at much lower concentrations than required for Ascaris sperm (Cande and Wolniak, 1978; Clark and Rosenbaum, 1982; Rozdzial and Haimo, 1986). Although PEG induced purified MSP in solution to form crystals rather than assemble into filaments, a range of other precipitants were able to produce filaments that appeared to be indistinguishable from those observed in vivo. However, the alcohols and MPD that promoted filament assembly of both isoforms did not induce crystal formation or stabilize filaments in lysed sperm.

Both the filaments in lysed sperm and those formed by reassembly in organic solvent/water mixtures were about 10 nm wide and appeared to be two-stranded helices. The small size of MSP (expected, on the basis of its M_R to approximate to a 2 nm sphere) together with a general slight curvature of the filaments made it difficult to obtain more detailed information about their substructure using optical diffraction or Fourier-based computer image processing. Moreover, because of the high critical concentration for assembly of MSP and the ease with which filaments disassemble, negatively stained specimens for examination by EM always had high background concentrations of unassembled MSP, which further obscured fine detail. We are currently seeking methods to eliminate this background and to produce specimens with straighter filaments to obtain micrographs more suitable for high-resolution structural analysis.

Because they contain structural information from different levels of the specimen superimposed, great caution is required in interpreting the images we obtained from negatively stained crystals. However, the pattern of striations observed indicated that it was very likely that the crystals were formed by the side-to-side aggregation of filaments. Such an interpretation is supported by the sheer size of the unit cell. If the unit cell dimension parallel to the direction of view was of the same order as the 7.2 nm transverse spacing, its volume would correspond roughly to its containing between 60 and 100 protein molecules of M_T 14,000. Such a large number of molecules in the unit cell could only be accommodated by their being assembled into a regular macromolecular complex such as a helix (the elongated nature of the unit cell making the other alternative, an icosahedral shell structure, unlikely). It was striking that the axial repeat of the sinusoidal path followed by the longitudinal striations in the crystal was close to the axial repeat seen in both reassembled filaments and those from lysed cells. Such a modulation would be consistent with a helical structure with a pitch of about 9 nm or some integral multiple of this value. The spacing of the longitudinal striations in the crystals was somewhat less than the diameter of filaments in fixed material and of reassembled filaments. Therefore it may be that the filaments in the crystals represent subfilaments rather than the complete filaments present in the cell and might, for example, represent an intermediate in filament assembly. However, the striations could arise from larger structures being superimposed and so the true size of the helical filaments in the crystals could be some integral multiple of the 3.6 nm transverse spacing seen by eye. In this context, the indication by optical diffraction that the true transverse lattice constant was about 7.2 nm indicated that the diameter of the repeating unit (which was probably a helical filament of MSP) in these crystals was probably much closer to the 10 nm diameter seen in isolated filaments. Intercalation between helical filaments, for example, could both reduce the centre-to-centre spacing between filaments in the crystals and produce a superposition pattern of strong longitudinal striations. Clearly these issues can only be resolved by solving the structure of these crystals in three dimensions by using tilt series reconstruction, which we are currently undertaking.

The capacity of a- and β MSP to form filaments and crystals independently in vitro means that both isoforms need not be present for assembly. We do not know if native filaments contain one or both isoforms. Alpha-MSP, the less abundant isoform in the cell, forms filaments in vitro at both lower protein and precipitant concentrations than β MSP. This property could, for example, allow α MSP to nucleate filament assembly in the pseudopod or provide a way of lengthening existing filaments rapidly. Such features would be consistent with the rapid (up to 70 μm/min) and continuous assembly that occurs at the plasmalemmal end of fiber complexes in crawling sperm.

The capacity to stabilize native filaments, assemble filaments in vitro and form crystals, together with sequence data, will enable the molecular structure of MSP and the arrangement of subunits in MSP filaments to be investigated in considerable detail using electron microscopy and image processing (see Stewart, 1988a,b) and possibly also X-ray diffraction. Such structural data are likely, in turn, to provide insight into the mechanism and control of filament assembly and the regulation of cytoskeletal architecture. This information is fundamental to understanding the role of MSP in sperm locomotion and, by comparison with actomyosin-based systems, to understanding the general basis of amoeboid motility.

We are most grateful to our colleagues in Tallahassee and Cambridge, and to Tom Pollard and Richard Henderson for their helpful comments, criticisms and suggestions. We also thank Juliane Essig and Gregory Roberts for their technical assistance. This work was supported by NIH grant GM29994 and NATO CRG 910264.