Cytoskeletal proteins of insect muscle: location of zeelins in Lethocerus flight and leg muscle

The structure of the myofibrils in asynchronous insect flight muscle is adapted to oscillatory contraction. Tension generated in the fibres is increased, after a delay, by stretching the activated muscle a small amount, and this property allows the muscle to contract at high frequency in response to periodic applied length changes (Pringle, 1978; Tregear, 1983). Stretch activation is a property of the myofibrils, due either to the geometry of the filament lattice (Wray, 1979b) or to myofibrillar components specific to flight muscle. The fibres have a high resting stiffness and shorten very little during activity. Thick and thin filaments overlap nearly completely in the sarcomere and the Z-disc is more dense and wider than in synchronous insect muscles or vertebrate striated muscle. The hexagonal lattice of thick and thin filaments is extremely regular and the configuration of rigor crossbridges is known in some detail (Taylor et al., 1993). The regularity of the lattice is likely to be important for stretch activation and the lattice may be stabilised by subsidiary proteins associated with the filaments.

Z-discs have been isolated intact from bee, Lethocerusand Drosophilaflight muscle (Saide and Ulrick, 1974; Bullard and Sainsbury, 1977; Sainsbury and Hulmes, 1977; Saide et al., 1989). These preparations contain proteins that are associated with the Z-disc but may not be part of the Z-disc lattice, such as projectin, which links thick filaments to the Z-disc. Amino acid analyses showed that the Z-disc preparations from both bee and Lethocerushave a high proportion of proline residues. Two proline-rich proteins, zeelin 1 (35 kDa) and zeelin 2 (23 kDa), were originally isolated from preparations of LethocerusZ-discs (Sainsbury and Bullard, 1980). These proteins were insoluble in non-denaturing solvents and therefore likely to be cytoskeletal proteins; it was thought they were an integral part of the Z-disc. We show here that the zeelins are not in the Z-disc but are cytoskeletal proteins that are associated with preparations of isolated Z-discs.

Recently, Vigoreaux and colleagues (1993)have sequenced a Drosophilaflight muscle protein, flightin, of apparent molecular mass 27 to 30 kDa, which is associated with the A-band. The protein is found only in asynchronous flight muscle and it was suggested that it has an essential function in stretch activation of these muscles. There are some similarities between flightin and zeelin 2 and it is possible they are related proteins.

In this paper we describe proteins in the cytoskeletal fraction of Lethocerusflight muscle, four of which have been isolated. The assembly of zeelins into filaments and micelles has been studied and the position of the proteins in the myofibrillar lattice has been determined. A preliminary account of this work has been given previously (Ferguson et al., 1992).

Lethocerus indicuswere obtained from Thailand. Myofibrils were prepared from the indirect flight muscle, thoracic leg muscle and muscle from the proximal section of the leg. The thoracic leg muscle preparation also contained some non-fibrillar flight muscle. Myofibrils were washed with rigor solution (0.1 M KCl, 20 mM potassium phosphate buffer, pH 6.8, 5 mM MgCl₂, 5 mM EGTA, 5 mM NaN₃) containing 0.5% Triton X-100 (Bullard et al., 1985).

A fraction containing Z-discs and cytoskeletal proteins was obtained by extracting indirect flight muscle myofibrils from two waterbugs with 5 volumes of 1.4 M KCl, 0.1 M NaHCO₃(pH 8.9) for 1 hour on ice. Partially extracted myofibrils were removed by centrifuging at 3,000 gfor 10 minutes and the supernatant was centrifuged at 30,000 gfor 30 minutes to give a pellet containing Z-discs and associated zeelins (Sainsbury and Bullard, 1980). A high yield of zeelins was obtained by extracting myofibrils twice in the high ionic strength buffer, centrifuging at 30,000 gfor 20 minutes. The pellet was taken up in 5 to 10 ml of 8 M urea, 20 mM sodium acetate (pH 5.2), 1 mM β-mercaptoethanol, homogenised in a glass homogeniser and dialysed against the buffer. After clarifying by centrifuging at 100,000 gfor 1 hour, the sample was passed over a column of S-Sepharose (6 cm × 1 cm) equilibrated with the same buffer. The column was washed with the 8 M urea buffer and fractions of about 1.5 ml collected. Unbound protein was eluted in about 12 ml. The column was then washed with 40 ml of buffer containing 0.1 M NaCl, which eluted zeelin 2, followed by 40 ml of buffer with 0.2 M NaCl, which eluted zeelin 1. Fractions containing zeelins 1 or 2 were pooled and concentrated about eight times with a Centricon-10 (Amicon). The proteins were purified further on a Superose-12 FPLC column equilibrated with the 8 M urea buffer containing 0.1 M NaCl for zeelin 2 or 0.2 M NaCl for zeelin 1. Isoforms of zeelin 2 were separated on a Mono S FPLC column. Fractions containing zeelin 2 from the S-Sepharose column were desalted on a PD-10 Sephadex column and passed over a Mono S column equilibrated with the 8 M urea buffer. Protein was eluted with a gradient from 34 ml of column buffer to 34 ml of buffer with 0.2 M NaCl. The protein p17 was eluted at 0.12 M NaCl, zeelin 2 isoform 1 at 0.13 M and isoform 2 at 0.16 M NaCl. In some experiments, the initial extract of urea-soluble proteins was separated on a Mono S column eluted with a gradient of 0 to 0.3 M NaCl. p18 was eluted before the gradient and p17 at 0.13 M NaCl. Zeelin 2 was eluted at 0.15 M and 0.16 M NaCl (two isoforms) and zeelin 1 at 0.3 M NaCl.

Micelles of zeelin 2 were formed by dialysing the protein (0.1 mg/ml) in 6 or 8 M urea, 20 mM sodium acetate (pH 5.2), 1 mM β-mercaptoethanol, 0.1 M NaCl against 1% deoxycholate (DOC), 10 mM Tris-HCl (pH 8.0), 1 mM β-mercaptoethanol at 4°C. Filaments were formed by dialysing from the same buffer into 3 M urea, 10 mM potassium phosphate (pH 6.5), 1 mM β-mercaptoethanol for 3 to 6 hours and then into 10 mM Tris-HCl (pH 7.5), 1 mM β-mercaptoethanol, 1 mM NaN₃for at least 16 hours at 4°C. At pH 6.5 no filaments were formed and at pH 8.5 filaments were small and clumped. Longer filaments were formed by dialysing against the 10 mM Tris-HCl buffer with 0.5 M NH₄SCN for up to 2 days. Zeelin 1 and p17 and p18 formed micelles under conditions used to form zeelin 2 filaments. An attempt was made to form mixed filaments by mixing zeelin 1 and 2 in the 6 M urea buffer and then dialysing against the solutions used to form zeelin 2 filaments.

Zeelin 2 was polymerised in the presence of myosin. Lethocerusmyosin (1 μM), (prepared by the method of Hammond and Goll, 1975, and free of paramyosin) was mixed with zeelin 2 (3.5 μM); both proteins were in 3 M urea, 10 mM Tris-HCl (pH 7.5), 1 mM dithiothreitol, 1 mM NaN₃. The mixture was dialysed against the same buffer without urea for 16 hours at 4°C.

Binding between zeelin 2 filaments and myosin filaments was tested by mixing 10 μl myosin filaments (60 pmol in 0.15 M KCl, 20 mM potassium phosphate, pH 7.0, 1 mM NaN₃, 1 mM β-mercaptoethanol) and 10 μl zeelin 2 filaments (43 pmol in the 10 mM Tris-HCl buffer). After 1 hour at 25°C, filaments were centrifuged at 2,000 gfor 5 minutes and the pellets and supernatants were examined in the electron microscope and by SDS-polyacrylamide gel electrophoresis.

Washed myofibrils of Lethocerusflight muscle that had been stored in rigor solution with 50% glycerol at −20°C were digested with calpain to remove Z-discs before separating thick and thin filaments (Reedy et al., 1981). After digesting with calpain, thick and thin filaments can be separated without extensive homogenisation and shearing (Bullard et al., 1988), which might tear the filaments away from possible association with zeelins. If myofibrils are washed with rigor solution containing Triton X-100, zeelins are not digested by calpain (Bullard et al., 1990). Myofibrils (14 mg/10 ml) in calpain-activating solution (18 mM Tris-HCl, pH 7.5, 5 mM β-mercaptoethanol, 0.12 mM EDTA, 1 mM CaCl₂) were digested with 0.1 mg of turkey gizzard m-calpain (a gift from Dr D. E. Goll) at 0°C for 1 hour, or until examination in a phase-contrast microscope showed that most of the Z-discs were gone. Myofibrils were washed in rigor solution to remove calpain, adjusted to pH 6.0, 5 mM MgCl₂, 10 mM ATP, and the suspension was sheared through a syringe with a 26 gauge needle (Reedy et al., 1981). The suspension was centrifuged at 3,000 gfor 10 minutes to remove remaining myofibrils. The resulting supernatant was centrifuged at 20,000 gfor 30 minutes; the pellet was examined in the electron microscope and contained thick filaments and some thin filaments. The supernatant containing thin filaments was centrifuged at 100,000 gfor 2 hours and the pellet of thin filaments was homogenised in rigor solution. In order to remove thin filaments from the fraction containing thick filaments, the fraction was treated with gelsolin. Human plasma gelsolin was a gift from Dr A. G. Weeds; the protein was expressed in Escherichia coliand contained the entire sequence (Way et al., 1989). Gelsolin (2 mg/ml) was added to 100 μl of the thick filament fraction suspended in 0.15 M KCl, 5 mM MgCl₂, 2 mM CaCl₂, 5 mM ATP. The suspension was incubated for 16 hours at 4°C and then centrifuged at 20,000 gfor 20 minutes. The pellet was examined in the electron microscope and contained thick filaments but no thin filaments.

Female LOU/C rats were immunised with proteins as follows: MAC 348 and 349 were derived from a rat injected intramuscularly (i.m.) with 150 μg (day 0) of partially purified zeelin 2 obtained from a Mono S FPLC column. The protein was dissolved in 1% DOC, 10 mM Tris-HCl (pH 7.5) and emulsified in complete Freund’s adjuvant. A booster injection of 150 μg in the 1% DOC buffer with incomplete Freund’s adjuvant was given subcutaneously (s.c.) on day 34 and an intravenous (i.v.) boost of 200 μg in 1% DOC buffer on day 159. MAC 350 and MAC 358 were produced by the same procedure. Partially purified zeelin 1 (200 μg), also from a Mono S column, was dissolved in 0.1% SDS, 10 mM Tris-HCl (pH 7.5). The s.c. boost on day 34 and the i.v. boost on day 159 were both 200 μg. Immune spleens were removed 3 days after i.v. boosts and fused with cells of the myeloma line Y3Ag1.2.3. (Galfrè et al., 1979). Cultures producing antibodies were identified by the immunodot method using a Z-disc preparation as antigen and characterised further by immunoblotting. Four of the cell lines derived from the rat immunised with pure zeelin 1 produced antibodies to zeelin 1 and five produced antibodies specific to α-actinin. Eight cell lines producing antibodies to zeelin 2 were obtained from the rat immunised with zeelin 2. MAC 348 is IgM subclass, MAC 349 and MAC 358 are IgG2a, and MAC 350 is IgG1. Ascitic fluid containing MAC 349 or MAC 350 antibody was obtained as described by Newman et al. (1992). Anti-TnH was MAC 143 (Bullard et al., 1988). Anti-paramyosin was a polyclonal antibody to dung beetle paramyosin purified on a paramyosin affinity column (Bullard et al., 1977).

Proteins were electrophoresed in mini-slab SDS-polyacrylamide gels (12% acrylamide) with the Laemmli (1970)buffer system. Two-dimensional gel electrophoresis was also carried out in mini-gels with a pH gradient of 3 to 10 in the first dimension (O’Farrell, 1975). Myofibrils or Z-discs were homogenised in O’Farrell lysis buffer containing 0.5% SDS and centrifuged at 12,000 gfor 10 minutes to remove insoluble material before electrophoresis. Z-disc proteins were also separated by two-dimensional, non-equilibrium pH gradient electrophoresis (NEPHGE), which separates basic proteins (O’Farrell et al., 1977); the pH gradient was 3 to 10 in the first dimension. Samples were electrophoresed at 600 V for 3 hours, then 800 V for 1 hour. The ratio of actin to zeelins in myofibrils was determined by densitometry of SDS-polyacrylamide gels stained with Fast Green (Bullard et al., 1985).

Proteins were transferred from SDS-polyacrylamide gels to nitro-cellulose by electrophoresis in 10 mM Tris-glycine, pH 7.5, 20% methanol for 2 hours at 250 mA. Blots were incubated with hybridoma cell supernatant and developed as before (Lakey et al., 1990). Immunofluorescence was by the method described by Lakey et al. (1990). Washed myofibrils were glycerinated for 2 to 4 days and incubated with hybridoma supernatant (diluted 1:100) for up to 1 hour at 4°C. Myosin was extracted from some myofibrils by incubating with 1 M KCl, 20 mM potassium phosphate (pH 7.0), 10 mM sodium pyrophosphate, 1 mM MgCl₂for 30 minutes at 0°C (Bullard and Reedy, 1973), centrifuging at 5,000 gfor 10 minutes, washing in the same solution and resuspending in rigor solution.

Bundles of five or six fibres were dissected from the dorsal longitudinal muscle of freshly killed L. indicus. Fibres were skinned in rigor solution containing 1% Triton for 15 minutes, washed in three changes of rigor solution for 15 minutes each and fixed for 30 minutes in rigor solution with 3% paraformaldehyde, 0.1% glutaraldehyde. These procedures were in 200 μl multiwell plates (Nunc) on ice. The fibres were infused with 2.1 M sucrose in rigor solution for 15 minutes at 25°C and then frozen in liquid N₂. Cryosections 50 to 100 nm thick were labelled with antibody and Protein A-gold (8 to 10 nm) and stained with 0.3% uranyl acetate (Lakey et al., 1990). Lowicryl-embedded fibres were processed as described below.

Thoraces of L. indicuswere dissected to expose the dorsal longitudinal muscles and glycerinated by cycling between relaxing solution (rigor solution with 5 mM ATP) containing 1% Triton X-100 or 50% (v/v) glycerol, then soaked for 12 hours in relaxing solution with 75% (v/v) glycerol and 5 mM dithiothreitol; all these procedures were at 4 to 10°C. Thoraces were stored at −80°C for periods from 2 days to 18 months (Reedy and Reedy, 1985). Glycerinated thoraces were equilibrated at 4°C and strips were dissected from the dorsal longitudinal muscle and divided into bundles of about 5 fibres in a pool of relaxing glycerol on ice. Fibres were either processed directly for cryosectioning or mounted on a U-shaped pin. The time the fibres were at room temperature during mounting was minimised. Some mounted fibres were stretched in relaxing glycerol on ice by 20 to 100% rest length (Lakey et al., 1990; Reedy et al., 1994). Fibres were washed in three changes of rigor solution, fixed in rigor solution with 3% paraformaldehyde, 0.1% glutaraldehyde or 6% paraformaldehyde, all on ice, and infused with 2.1 M sucrose in rigor solution at 25°C, as described for fresh fibres. This procedure was varied to investigate the effect of glycerination on the migration of zeelins to the Z-disc. The time course of the glycerination effect was followed. Strips of the dorsal longitudinal muscle were glycerinated and fibres from a sample strip were dissected at 25°C, washed and fixed on ice and infused with sucrose at 25°C; fibres were cryosectioned and labelled with antizeelin 2. Remaining strips were stored at −80°C. Subsequently, sample strips were removed from storage every few days, cryosectioned and labelled; the procedure was repeated until label was observed on the Z-disc. Remaining fibres were stored for 5 months.

Glycerinated fibres mounted on U-pins, fixed and infiltrated with sucrose were freeze substituted and embedded in Lowicryl HM20 by the method of van Genderen et al. (1991). Fibres on U-pins were frozen in liquid N₂and processed in a Reichert CS Auto cryosubstitution apparatus. U-pins were removed when the cured blocks were trimmed. Sections 40 to 80 nm thick were cut on a Reichert OMU3 microtome with a diamond knife and mounted on bare 400 mesh grids or Formvar-coated grids. Sections were labelled with antibody and Protein A-gold and stained with 4% uranyl acetate (Lakey et al., 1990). Sections on bare grids were coated with carbon (100 Å thick) before examination in the Philips 400 electron microscope. Sections are exposed to antibody on one side and both cryo- and Lowicryl sections are labelled on this side only.

Immunolabelling of sections was carried out with hybridoma cell supernatant diluted 1:50, v/v (MAC 358, and 348), or ascitic fluid diluted 1:400, v/v (MAC 350), and 1:800 (MAC 349) or IgG of 15 μg/ml (MAC 143) and 20 μg/ml (anti-paramyosin). Quantification of gold particles was carried out by computer after digitising selected micrographs. The digitised images were displayed on a graphics terminal and gold particles detected and marked automatically. Particles were then counted in different regions of the sarcomere (Z-disc and A-band). Filaments and micelles of zeelins 1 and 2 were negatively stained with 1% aqueous uranyl acetate. STEM mass measurements of unstained zeelin 2 filaments were carried out as described previously using TMV as an internal standard (Reedy et al., 1981).

Preparations of isolated Z-discs contain cytoskeletal proteins in addition to the proteins that are components of the Z-disc. The major proteins in this fraction are actin, α-actinin, zeelins and a protein of 110 kDa; there are lesser amounts of projectin, kettin and proteins of 230 kDa, 120 kDa, 18 kDa and 17 kDa (Lakey et al., 1990, 1993; and see Fig. 2A). The proteins in the cytoskeletal fraction (a Z-disc preparation) are compared with those in the whole myofibril by two-dimensional electrophoresis in Fig. 1A,B; proteins were identified by incubating immunoblots in specific antibodies. The major proteins in the cytoskeletal fraction have multiple isoforms that are more basic than actin. There are two isoforms of α-actinin, as found in extracts of Drosophilathorax (Saide et al., 1989), and two isoforms of the 110 kDa protein. Zeelin 2 was separated into two major and one minor isoform with the same molecular mass but different isoelectric points. The isoelectric points of the zeelin 2 isoforms were in the same range as those of α-actinin isoforms, which are calculated to be around 5.74 from the amino acid sequence (Fyrberg et al., 1990). Zeelin 1 did not run into two-dimensional gels of the usual polarity but when the polarity was reversed, two isoforms of zeelin 1 were seen at the basic side of the gel; zeelin 2 was at the acidic side of this gel (Fig. 1D). Therefore, zeelin 1 is considerably more basic than zeelin 2 and both have differently charged isoforms. The isoforms of zeelin 2 are unlikely to be due to a change in charge produced by phosphorylation: when myofibrils were phosphorylated with the catalytic subunit of cAMP-dependent protein kinase, the three isoforms of zeelin 2 shifted to a lower isoelectric point but remained as separate spots on a 2-D gel (not shown).

Zeelins were isolated from preparations of Z-discs or from the residue left after extracting myofibrils extensively at high ionic strength. The fraction containing Z-discs and cytoskeletal proteins was dissolved in urea and zeelins were separated by ion exchange chromatography followed by gel filtration (Fig. 2A,B). Zeelin 2 was separated into two isoforms on a Mono S column (Fig. 2C). The proteins p17 and p18 were also isolated on a Mono S column (Fig. 2C). After passing through the Superose column some samples of zeelin 1 had an additional peptide of 23 to 25 kDa, which was probably a degradation product (Fig. 2B, lane 3); the peptide did not react with antibody to zeelin 2. Some preparations of zeelin 2 had altered mobility on SDS-polyacrylamide gels when stored in urea at 4°C and after about a month there were two peptides, one with lower and one with higher mobility than the original protein (Fig. 2C, lane 4).

Since both zeelins are hydrophobic, an attempt was made to renature the proteins after isolation in urea, by dialysing against a number of non-ionic detergents. Zeelins were insoluble in non-ionic detergents, but zeelin 2 formed a clear suspension in 1% deoxycholate, pH 8.0, which contained micelles.

The insolubility of zeelins resembles that of intermediate filament proteins and we tested the possibility that they might form filaments or other ordered assemblies at low ionic strength. The proteins in 6 M urea, pH 5.2, were equilibrated first in 3 M urea, pH 6.5, and then in 10 mM Tris-HCl, pH 7.5. Zeelin 2 formed short filaments about 10 nm wide. Longer filaments of 300 to 600 nm were formed in the presence of ammonium thiocyanate, which slows down assembly. The filaments appeared to be flexible and sometimes formed an annulus or figure of eight (Fig. 3A). In some cases short lengths of two filaments were aligned. In the presence of 0.1 M NaCl the filaments aggregated into loose tangles, which did not dissociate when NaCl was removed (Fig. 3C). Filaments formed by the isoforms of zeelin 2 were somewhat different. The less basic isoform 1 formed shorter filaments that tended to aggregate into ragged bundles (Fig. 3B) while isoform 2 formed filaments like those made from mixed isoforms.

STEM mass measurements were carried out on unstained filaments of zeelin 2. Using TMV as an internal standard, the mass per unit length for 327 measurements was 30±7 kDa/nm. Taking a protein density of 800 Da per nm³and the maximum measured width of 10 nm, this would correspond to a filament thickness of about 3±0.9 nm. Therefore, the filaments are ribbon-shaped rather than cylindrical. Images of negatively stained filaments often show narrowing that is consistent with the appearance of a twisted ribbon.

Zeelin 1 formed irregular micelles about 20 nm in diameter when renatured under the same conditions as zeelin 2 (Fig. 3D), and p17 and p18 formed small micelles about 8 nm in diameter (not shown). A mixture of zeelin 1 and zeelin 2 formed clumped, elongated micelles, suggesting that zeelin 1 interacts with zeelin 2 to prevent filament formation. Fractions containing zeelin 2 and p17 also did not form filaments. Myosin had no effect on formation of zeelin 2 filaments, and electron micrographs showed that myosin molecules and zeelin filaments were not associated; nor did zeelin filaments associate with myosin filaments.

The relative amounts of zeelins and actin in flight muscle myofibrils were estimated from scans of SDS-polyacrylamide gels. The ratio of actin to zeelin was 6.7±0.67 moles for zeelin 1 and 8.2±0.95 moles for zeelin 2 (s.e.m. for 5 batches of myofibrils). The relative dye binding by zeelins and actin was not determined, so these figures give an approximate value for the proportions. The molar ratio of myosin to zeelins can be calculated from the lattice geometry (Bullard and Reedy, 1973). Thick and thin filaments overlap nearly completely in flight muscle and there are 3 thin filaments for every thick filament. There is one monomer every 2.75 nm of the thin filament, and assuming 6 out of 7 monomers are actin and 1 is arthrin (Bullard et al., 1985), there are on average 3× 36 kDa of actin per 2.75 nm or 0.93 moles/nm. A thick filament with 4 myosin molecules per 14.5 nm (Reedy et al., 1981; Morris et al., 1991) has 4× 470 kDa of myosin per 14.5 nm or 0.276 moles/nm. The molar ratio of myosin to actin would therefore be about 0.3. The molar ratio of myosin to zeelin is then 2.0 for zeelin 1 and 2.5 for zeelin 2.

Monoclonal antibodies were raised to zeelins 1 and 2, and the specificity of the antibodies was determined on immunoblots of myofibrils. Monoclonal antibodies to zeelin 1 reacted strongly with the 35 kDa zeelin in flight myofibrils and more weakly with two proteins of slightly lower molecular mass (Fig. 4). The antibody also labelled the two lower molecular mass proteins in preparations of pure zeelin 1. The proportions of the two smaller proteins were constant and it is likely they are minor isoforms of zeelin 1 rather than degradation products. They have a basic isoelectric point and are therefore not myosin light chains (Fig. 1D). Antibodies to zeelin 1 labelled proteins of 35 and 30 kDa in thoracic leg myofibrils equally strongly; the 35 kDa protein is the same size as the major flight muscle isoform and the 30 kDa protein is the same size as the larger of the two minor isoforms (Fig. 4). The amount of zeelin 1 in thoracic leg muscle may be rather small as the protein was barely visible in Coomassie Blue-stained gels, or the protein may only be in some fibres of this general type. The antibody did not label any protein in myofibrils from the muscle in the proximal section of the leg.

Monoclonal antibodies to zeelin 2 reacted specifically with the 23 kDa zeelin in flight myofibrils (Fig. 4). On two-dimensional immunoblots of myofibrils, the antibodies labelled the two main isoforms seen on Coomassie Blue-stained gels and a third more acidic isoform (Fig. 1C). The antibody did not label any protein on immunoblots of thoracic leg myofibrils (Fig. 4) or of myofibrils from the muscle in the proximal section of the leg. Thus, zeelin 1 is in both synchronous and asynchronous muscles, though with different proportions of isoforms, while zeelin 2 may only be in asynchronous muscle.

The position of zeelins in flight and leg myofibrils was determined by immunofluorescence. Anti-zeelin 2 labelled the A-band in two broad bands either side of the H-zone (Fig. 5A). Anti-zeelin 1 predominately labelled the A-band in the region of the H-zone and the Z-disc (Fig. 5B). In myofibrils from which myosin had been extracted at high ionic strength, antizeelin 1 labelling was in two broad bands similar to antizeelin 2 labelling (Fig. 5C).This suggests that zeelins have the same distribution in the A-band and that zeelin 1 is less accessible to antibody than zeelin 2 in the intact myofibril. If myofibrils were incubated with antibody for more than about an hour, antibodies to both proteins partially extracted the A-band, perhaps by disrupting thick filaments.

Labelling in leg myofibrils differed from that in flight myofibrils. Anti-zeelin 1 labelled across the A-band, including the H-zone (Fig. 5D); the antigen was more accessible than in flight myofibrils. Anti-zeelin 2 did not label leg myofibrils, in agreement with the immunoblotting results. Therefore the synchronous leg myofibrils contain zeelin 1 but not zeelin 2 in the A-band.

The position of zeelins in flight muscle myofibrils was determined by labelling sections with anti-zeelin and examining the position of gold markers by electron microscopy. Fibres fixed immediately after removal from a freshly killed Lethoceruswere labelled on the A-band by antibodies to both zeelins. Anti-zeelin 2 labelled sectioned myofibrils over the whole A-band except for a region in the middle of the sarcomere and another at the end of the sarcomere extending up to the Z-disc (Fig. 6A). The region between the end of the A-band labelling and the Z-disc was 227±8 nm wide and the unlabelled region in the middle of the sarcomere was 225±11 nm wide (s.e.m. for myofibrils from 9 different fibres). Anti-zeelin 1 labelled much more lightly but the distribution of label appeared to be the same as for zeelin 2 (Fig. 6B). Therefore, the zeelins are in a limited region of the A-band.

The distribution of zeelins was affected by glycerinating the fibres. Variable amounts of label were observed on the Z-disc in sections of glycerinated fibres and the factors affecting this labelling were investigated. Lethocerusmuscle is frequently stored in glycerol at −80°C and the effect of this on migration of proteins is of interest. Both zeelins moved to the Z-disc in glycerinated fibres but the effect was followed for zeelin 2 because antibody labelling was stronger. Two factors influenced the migration of zeelin 2 to the Z-disc: the length of time the fibres were stored in glycerol and the temperature at which the fibres were processed before and during fixation. Fig. 7A,Bshows sections from fibres glycerinated for 18 months. If the fibres were processed very rapidly on ice there was no zeelin in the Z-disc (Fig. 7A); there was a relatively large amount of label outside the myofibril, especially on membrane fragments, suggesting that zeelin was dissociated from the myofibril during long glycerination. Fibres processed for longer at 25°C, using standard methods for cryomicroscopy, had zeelin 2 in the Z-disc (Fig. 7B) as well as on membranes surrounding the myofibril. Fibres that were warmed to 25°C during dissection but otherwise kept on ice, also had zeelin 2 in the Z-disc. Therefore, long glycerination loosens zeelin 2 from the sites to which it is bound in the A-band and the free protein migrates to the Z-disc if fibres are brought to room temperature. Transverse sections labelled with anti-zeelin 2 had label throughout the Z-disc, which confirms that zeelin 2 penetrated the Z-disc lattice. The migration was prevented by fixing fibres before glycerination:anti-zeelin 2 labelled these fibres on the A-band only. The labelling in fibres from freshly killed insects was confined to the A-band, even after leaving fibres at 25°C for 2 hours. Some strips of muscle were washed with Triton, as in the glycerination procedure, but not treated with glycerol. After storing for two weeks at 4°C antibody label was on the A-band only. Therefore, zeelin migration is the result of glycerination.

The time course of the glycerination effect was followed. There was no appreciable anti-zeelin 2 label on the Z-disc until the fibres had been in glycerol at −80°C for about 11 days (Fig. 7C); the amount of label on the Z-disc continued to increase and after two months the density of label exceeded that on the A-band. Throughout this time, the label on the A-band changed very little. These results show that zeelin 2 becomes increasingly less firmly bound to the A-band while fibres are stored at −80°C.

In order to determine if zeelins in the A-band are associated with the thick or the thin filaments, fibres were stretched so that the two types of filament no longer overlapped. Antibodies to both zeelins labelled the A-band and not the thin filament region (Fig. 8). Therefore, zeelins are not associated with thin filaments. The irregularity of the thick filaments in stretched myofibrils is probably due to different amounts of extension in the connecting filaments between thick filaments and Z-disc. Zeelins are also irregularly arranged in stretched fibres, which suggests that they are connected to the thick filaments.

A striking feature of the labelling seen in thin oblique sections is the banded appearance of the labelled regions (Fig. 9). This effect is caused both by the well known interaction of the sectioning plane with the regular lattice of the myofibril (Reedy, 1968) and by the labelling of the section on one side only. Fig. 9Ashows a typical labelling pattern for anti-paramyosin. In this case, since paramyosin is a core protein in the thick filament, we would expect labelling to be confined to the region where the section plane cuts through the centre of the thick filament (split myosin region) and in Fig. 9Athe labelling is in a thin line along this region. Labelling by anti-zeelin 1 is similarly confined to the narrow split myosin region (Fig. 9B). In both cases the labelling is not symmetrical about the darker region in which thick filaments pass through the section, consistent with label being bound to one side of the section only. The oblique section of the stretched fibre in Fig. 8Balso shows anti-zeelin 1 labelling restricted to the split myosin region on one side of the section. Labelling by anti-zeelin 2 is on one side of the region in which thick filaments span the section but the label extends further from the split myosin region, indicating a less restricted distribution of the epitope (Fig. 9C,D). The effect is seen most clearly in Fig. 9C, which is a very thin section where the section plane runs from top left to bottom right. Fig. 9Eis a diagram showing how the type of labelling seen in thin oblique sections can be explained if zeelin 2 is on the outside of the thick filament. The labelling pattern of antizeelin 1 suggests that zeelin 1 is more closely confined to the thick filament.

It has been suggested that TnH might extend from the thin filament and form a link with the thick filament, which could be important in stretch activation of the flight muscle (Bullard et al., 1988). The Drosophilaprotein, flightin, has been proposed as a anchoring protein for TnH on the thick filament (Vigoreaux et al., 1993). The possibility that TnH is linked to zeelin 2 was investigated by comparing the distribution of the proteins across the sarcomere. The labelling pattern of antizeelin 2 and anti-TnH differed (Fig. 10). Anti-TnH label extended from the edge of the H-zone almost up to the Z-disc while zeelin 2 label left a wider gap in the middle of the sarcomere and stopped before the end of the A-band. Therefore, the proteins are not adjacent to each other across the whole sarcomere.

In order to confirm that zeelins are associated with thick and not with thin filaments, isolated filaments were prepared. Thin filament preparations were free of thick filaments, as judged by electron microscopy and did not contain zeelins (Fig. 11, lane 1). The fraction containing thick filaments also had some thin filaments; SDS-polyacrylamide gels showed that zeelins were present in this fraction as well as thick and thin filament proteins (Fig. 11, lane 2). The thick filament fraction was treated with gelsolin to remove thin filaments. Actin, tropomyosin and other thin filament proteins were removed but zeelins remained with the thick filaments (Fig. 11, lane 3). This supports the conclusion from immuno-labelling experiments that zeelins are associated with thick and not with thin filaments.

Insect flight muscle myofibrils contain several proteins that are insoluble under conditions that dissolve thick and thin filaments. Two higher molecular mass proteins in this cytoskeletal fraction, α-actinin and kettin, are known to be associated with the Z-disc (Lakey et al., 1990, 1993). We have isolated four lower molecular mass cytoskeletal proteins: zeelin 1, zeelin 2, p17 and p18. The zeelins were previously thought to be components of the Z-disc (Sainsbury and Bullard, 1980). We now show that these proteins are in the A-band, although they have a high affinity for the Z-disc. The lower molecular mass cytoskeletal proteins form soluble macromolecular assemblies. Zeelin 1, p17 and p18 form micelles, which suggests that the molecules are amphiphilic, with a hydrophobic end inside the micelle and a hydrophilic end exposed to solvent (Tanford, 1980). In contrast, zeelin 2 forms filaments and is therefore probably not amphiphilic. Zeelin 2 may have hydrophobic regions within the filament and hydrophilic regions that are on the outside. The area of contact between zeelin 2 molecules in the filament is probably small, making the filaments flexible enough to form rings. Actin filaments, for example, do not form rings because the multiple contacts between monomers limit filament flexibility. The ribbon-like shape of zeelin 2 filaments would also increase flexibility. The side-to-side association of filaments formed from isoform 1 is probably due to a different surface charge on the filaments. Similarly, the loose tangles of filaments formed by mixed isoforms in 0.1 M NaCl may be due to masking of the surface charge. Formation of micelles of zeelin 1 and filaments of zeelin 2 suggests that the tertiary structures of the two proteins are different. Zeelins do not form mixed filaments.

Both zeelins in flight muscle have multiple isoforms of the same molecular mass and differing isoelectric points. In the case of zeelin 1, there are also minor isoforms of lower molecular mass. The proportion of low and high molecular mass isoforms of zeelin 1 differs in flight and leg muscle as does the position of the protein in the A-band. This probably reflects a difference in the function of zeelin 1 in synchronous and asynchronous muscle. Zeelin 2 is only found in flight muscle and is likely to have a special function in asynchronous muscle.

Zeelins are in a limited region of the A-band in flight muscle myofibrils and the antibody labelling pattern of stretched fibres shows clearly that both are associated with thick rather than thin filaments. The pattern of antibody labelling in isolated myofibrils, and in obliquely sectioned fibres, suggests that zeelin 1 is closer to the core of the thick filament and less accessible than zeelin 2. When myosin is extracted from myofibrils, zeelin 1 is exposed to labelling and retains a regular organisation in two discrete regions of the A-band. Therefore, an intact thick filament lattice is not essential for the ordered arrangement of zeelin 1 in the sarcomere. The extreme insolubility of zeelin 1 in the extracting solution allows myosin to be removed, leaving a skeleton of zeelin 1. Disruption of the A-band by long incubation with anti-zeelins suggests that the cohesion of the thick filament lattice may depend on zeelins.

The attachment of zeelins to the A-band is loosened by glycerol, even in fibres stored at −80°C; the proteins become more mobile on warming to room temperature and move to the Z-disc. The particular affinity of zeelins for the Z-disc may be due to association of hydrophobic regions on the molecules with hydrophobic sites on the Z-disc. Binding of dissociated zeelins to membrane fragments around the myofibrils supports this. Mechanical measurements are often made on flight muscle fibres stored in glycerol and dissociation of zeelins from the A-band may contribute to the progressive deterioration in oscillatory performance in these fibres.

There is about one mole of each of the zeelins for two moles of myosin. This means that zeelins are rather sparsely distributed and are unlikely to be in the form of filaments like those formed by zeelin 2 in vitro. Thick filaments in insect flight muscle are made up of 12 subfilaments (Wray, 1979a; Beinbrech et al., 1988) and the packing of myosin molecules in the subfilaments may make the rod region of some myosin molecules inaccessible; the association of zeelins with myosin may be limited to those myosin molecules having a rod region on the surface of the thick filament. Alternatively, zeelin 1 may be confined to those myosin molecules in which the rod is in the interior of the thick filament and zeelin 2 to those in which the rod is on the outside. The molar ratio of zeelins to actin is approximately the same as the molar ratio of troponin to actin (Bullard et al., 1985), which would be consistent with an association between zeelins on the thick filament and troponin on the thin filament. If an extended TnH forms a link between thick and thin filaments by binding to zeelin 2 on the thick filament, the two proteins would be expected to have the same distribution across the sarcomere. However, there are regions at the end and in the middle of the sarcomere in which there is TnH on the thin filaments but no zeelin on the thick filaments. Therefore, TnH is likely to have regulatory functions in addition to any crosslinks formed with zeelins.

There are several ways in which the limited distribution of zeelins in the flight muscle A-band might arise. Myosin molecules are arranged in a parallel manner in the tip of the thick filament with the heads towards the Z-disc and the end of the thick filament almost touching the Z-disc. The region at the end of the A-band that does not contain zeelins is 227 nm wide, which is somewhat greater than the full length of a myosin molecule (170 nm; Elliott and Offer, 1978). A gap of this length is thus consistent with zeelins being associated with the end of the myosin rod. The gap in the middle of the A-band that does not contain zeelins is 225 nm. The bare zone of the thick filament, in the middle of which the rod regions of myosin molecules overlap in an anti-parallel manner, is 174 to 250 nm in Lethocerus, as measured from isolated filaments (Reedy et al., 1981; Morris et al., 1991). If zeelins are associated with the end of the myosin rod, this would explain their exclusion from the region of anti-parallel overlap. The maximum possible overlap would be the full length of the myosin rod, which is about 150 nm (Elliott and Offer, 1978). A more likely overlap, based on the width of the bare zone, is about 100 nm. Thus the overlap of myosin cannot fully explain the exclusion of zeelins at the middle of the A-band. An alternative explanation may be the presence of the large M-line protein (400 kDa) (Lakey et al., 1990), which may extend across the bare zone in each half-sarcomere and prevent zeelin binding to myosin.

A second possibility is that the position of zeelins in the A-band is determined by a titin-like protein. Titin in vertebrate skeletal muscle extends from the Z-disc to the M-line (Fürst et al., 1988) and is largely made up of repeating domains of two types (Labeit et al., 1990). A-band proteins, such as C-protein, may bind to discrete regions of the titin molecule forming stripes across part of the A-band (Fürst et al., 1989; Labeit et al., 1992; Soteriou et al., 1993). There is a protein in insect muscles that is similar in size to titin (Lakey et al., 1993) and it is possible that this acts as a template for assembly of zeelins on the thick filament. The lack of any association between zeelin 2 and myosin in vitro, either as monomers or filaments, supports the idea that a titin-like protein is needed for the association of zeelins with the A-band. Thirdly, projectin in flight muscle extends about 300 nm along the end of the A-band (Bullard et al., 1977; Saide, 1981; Nave and Weber, 1990) and this protein might exclude zeelins from the end of the sarcomere while the M-line protein excludes zeelins from the middle of the sarcomere.

A periodicity of 14.5 nm or a multiple of this would be expected in the antibody labelling pattern if zeelins are associated regularly with myosin. And if the binding of zeelins is determined by a titin-like protein, some other periodicity depending on the multi-domain repeat would be possible. The double-labelling method we have used here on sections does not reveal any obvious periodicity. However, a higher-resolution labelling technique, such as that used for TnH and TnT (Reedy et al., 1994), may enable a regular zeelin repeat to be seen.

Zeelin 2 and Drosophilaflightin appear similar in many ways, but if the proteins are related they are not highly homologous in sequence. None of the antibodies to zeelins reacted with Drosophilaflightin.The apparent molecular mass of flightin (27 to 30 kDa) is slightly higher than that of zeelin 2. The amino acid compositions of zeelin 2 (Sainsbury and Bullard, 1980) and flightin (Vigoreaux et al., 1993) are generally similar, except that flightin has less of the hydrophobic residues Pro, Leu and Phe, less His and more Ala and Val. The sequences of short peptides of zeelin 2 have no homology with the sequence of flightin (Kellner and Bullard, unpublished result). Both proteins have differently charged isoforms: the isoelectric points of the main isoforms of flightin are 5.3 to 6.0 and those of zeelin 2 are around 5.7. There are additional isoforms of flightin with a lower apparent molecular mass that are phosphorylated but these have not been detected for zeelin 2. The greatest similarity between flightin and zeelin 2 is that both are associated with the A-band in asynchronous flight muscle and do not occur in synchronous muscles. Zeelin 1 differs from both zeelin 2 and flightin: the protein is basic not acidic and occurs in both synchronous and asynchronous muscles.

The function of zeelins may be to maintain the ordered structure of the thick filament and hence the regularity of the myofibrillar lattice. This is likely to be essential for generating oscillatory contraction in asynchronous flight muscle. Zeelin 1 is closely integrated with the thick filament backbone and we cannot exclude the possibility that zeelin 1 is a core protein of the thick filament like paramyosin. However, the migration of zeelin 1 to the Z-disc in glycerinated fibres argues against this. In intact thick filaments, paramyosin in the core is only accessible to antibody at the bare zone in the middle of the filament (Bullard et al., 1977; Bullard, 1983). Antibody labelling across the sarcomere in sectioned fibres confirms that paramyosin is in the core along the whole length of the thick filament. The position and likely function of p17 and p18 remain to be determined.

Zeelins are substrates for the Ca²⁺-activated protease, calpain (Bullard et al., 1990). The enzyme is present in muscle fibres and is probably active in the early stages of muscle autolysis. Zeelin digestion is promoted by lipids, which may provide additional control of the enzyme. The disassembly of the sarcomere probably begins with cleavage of the Z-disc protein, kettin by calpain (Lakey et al.,1993), followed by cleavage of zeelins and destabilisation of the filament lattice.

We are grateful to Dr A. G. Weeds for a gift of gelsolin, to Dr D. E. Goll for calpain and to Dr R. Sanit for supplying Lethocerus. STEM mass measurements were made by Dr W. Tichelaar. This work was partly funded by a grant from the Muscular Dystrophy Group of Great Britain.