Journal of Muscle Research and Cell Motility 11, 378-391 (1990)

Expression of human r-myosin heavy chain fragments in Escherichia coli; localization of actin interfaces on cardiac myosin PATRICK ELDIN ~, M A R T I N E Le CUNFF I, KLAUS W. DIEDERICH 2, T H O M A S JAENICKE 2, B E R N A R D C O R N I L L O N I, D O M I N I Q U E M O R N E T I, HANS-PETER V O S B E R G 2 a n d JEAN J. LINGER~* Institut National de la Sant~ et de la Recherche M~dicale, INSERM U300, Facult~ de Pharmacie, Av. Ch.FlahauL 34060 Montpellier cedex 1, France. 2 Max Planck Institute for Medical Research, Department of Cell Physiology, ]ahnstrasse 29, D--6900 Heidelberg, FRG

Received 2 January I990; revised 23 April I990; accepted 23 April I990

i

Summary ,'~ A cDNA clone coding for an internal fragment of slow-cardiac//-myosin heavy chain was isolated from a 2gtl0 human skeletal muscle library. Six overlapping cDNA subclones, which span myosin heavy chain subregions and presumably interact with actin, were derived from this clone, fused to a//-galactosidase vector and expressed in Escherichia coil Three of the subclones were obtained by PCR (polymerase chain reaction) which enables gene or cDNA fragments to be amplified independently of preexisting restriction sites. Initially, various experiments were carried out using a long MHC (myosin heavy chain) fusion protein containing the 50 kDa-20 kDa connecting region, the whole 20 kDa region and the short subfragment 2 region. This MHC fusion protein was chemically or proteolytically cleaved in the same conditions as the native myosin molecule. Whole and truncated forms of the MHC fusion protein were separated on polyacrylamide gels, electroblotted on nitrocellulose sheets and renatured. They were then assayed in overlay experiments with F-actin and/or myosin light chains in solution. Specific antibodies were used to detect interactions between heavy chain fragments and F-actin or light chains. We thus observed that one long heavy chain fragment synthesized by E. coli behaved like proteolytic or chemical MHC preparations made from native myosin molecules. Two chymotryptic fragments of the MHC fusion protein, which are soluble at low ionic strength, cosedimented with F-actin in solution. Our results demonstrate that, in actin overlay experiments with whole fusion proteins, interactions seem to be due to the heavy chain fragment, not to the bacterial component. All interactions were non ATPsensitive. We further investigated the possible participation of the six recombinant MHC fragments in contributing to the actomyosin interfaces on the 50 kDa-20 kDa regions of the human cardiac fl-MHC. The present procedure, which enables the synthesis of any MHC fragment independent of any protease site, is a powerful new tool for studying structure-function relationships within the myosin molecule family.

Introduction Mammalian slow and fast skeletal muscles, ventricular and atrial myocardia all contain different specific myosins composed of many diverse but specific light and heavy chain hexameric arrangements. Nucleotide sequences of corresponding genes are currently being determined (for the most recently established sequences, see YamauchiTakihara et al., 1989; Feghali and Leinwand, 1989; KarschMizrachi et al., 1989; McNally et al., 1989; Kraft et al., 1989; Jaenicke et al., submitted). For over forty years, numerous proteases and chemical reagents have been applied to myosin molecules in order to identify their functional residues or domains. Thus, the N-terminal part of all heavY chains (the head) has long been known to *To whom correspondence should be addressed. 0142-4319/90 $03.00 + .12 9 1990 Chapman and Hall Ltd.

contain the ATP cleavage site, the actin binding site(s) and most of the light chain binding sites (Lowey et al., 1969). The C-terminal part of the myosin heavy chains (the rod) is involved in the formation of thick filaments. More recently, some defined amino acid residues and distinct sequences within myosin heavy chains of striated muscles were identified as being essential to important myosin functions, such as ATP and actin binding (for recent reviews, see Morner et al., 1989; Botts et al., 1989). Many different complementary experimental techniques have been used simultaneously to determine the spatial vicinity of certain residues, these include: chemical modifications of amino acids, blocking of proteolytic cleavage sites by actin, preparation of short fragments of myosin heavy chains which bind either actin or light chains,

Synthetic myosin heavy chain fragments recognize actin and light chains crosslinking and fluorescence resonance transfer experiments (reviewed in Audemard et al., 1988). Biophysical techniques such as NMR (Trayer et al., 1987), or immunological methods, such as immunoelectronmicroscopy (Miyasnishi et al., 1988), have also contributed to the identification of essential residues and subsequences within myosin heavy chains. Genetic engineering techniques were recently introduced as a new approach to the study of structure-function relationships of myosin molecules. Reinach and Fischman (1985) were the first to use a recombinant DNA protocol to define the primary structure of epitopes recognized by monoclonal antibodies directed against myosin light chain 2 from chicken skeletal muscle. In a further report, they examined the calcium binding caPacities of normal and mutated synthetic light chains (Shimizu et al., 1985; McNally et al. 1988) recently demonstrated the coexpression and assembly of intronless myosin heavy chains from Dictyostelium discoideum and myosin light chains from Acanthamoeba in Escherichia coil Maeda and co-workers (1989i have described paracrystals of rabbit skeletal muscle light meromyosin synthesized in Escherichia coiL, Rimm and co-workers (1989) recently used fusion proteins made with myosin heavy chain fragments from Acanthamoeba, to study the location of the head-rod junction. In most cases, the myosin fragments originated from cloned cDNA. Most of the functionally important myosin heavy chain areas are currently far from being completely understood. Concerning actin interface(s) on myosin heavy chains, for example, two very recent reports modify the common assumption that actin interfaces are mainly located in the 20 kDa fragment or the 10 kDa C-terminal end of the 50 kDa of myosin subfragment 1 (Morner et al., 1989; Botts et al., 1989). Furthermore, F-actin apparently interacts with the isolated 23 kDa fragment located at the Nterminal end of the myosin subfragment 1 (Miihlrad et al., I989) and non-polymerizable actin monomers can be crosslinked to the 50 kDa segment of the myosin subfragment 1 (Bettache et al., 1989). These experiments were carried out using myosin fragments isolated from native muscles. Hence, fragment isolation depends on the presence of specific proteolytic or chemical cleavage sites within the myosin chains under study. In addition to these limitations, it has only been possible to use myosins prepared from rabbit or chicken fast skeletal muscles in most experiments. Other striated muscle myosins, such as slow or cardiac myosins, have rarely been used. When myosin amino acid sequences from mammalian striated muscles are determined progressively, generally most of the essential residues or sequences seem to be relatively conserved. Hence, new experimental approaches are required in order to locate more precisely functional or essential areas on the myosin molecule, to understand why molecules with similar structures have such different activities and to obtain evidence on how subtle myosin primary structure changes may explain the observed

3 79

functional differences. In the course of our investigation on human cardiac myosin isoforms (Bouvagnet et al., 1984; Lichter et al., 1986; Dechesne et al., 1985, 1987; Diederich et al., 1989), we isolated a relatively long cDNA clone coding for an internal fragment of the slow-cardiac fl-myosin heavy chain. Six cDNA subclones were derived from this clone, they spanned overlapping myosin heavy chain domains in the 50 kDa-20kDa regions of the human cardiac/J-MHC and included known MHC binding sites in different recombinations. These subclones were integrated into pEX cloning vector and expressed as ]~-galactosidase fusion proteins in Escherichia coil The actin and myosin light chain binding properties of these fusion proteins, or of their subfragments, were studied in vitro.

Materials and methods

Materials Oligonucleotides synthezised by Operon Technologies (Alameda, CA94501, USA), were purified by electroelution from polyacrylamide gels containing urea (Moore et aI., 1987). Myofibrils or myosin were prepared from bovine left ventricular myocardium (Klotz et al., 1981). Myofibrils were partially proteolyzed with chymotrypsin to produce a mixture of rods and S-1, subfragment-1 (Weeds and Taylor, 1975). Myosin light chains were obtained from myosin treated with 6M guanidine chloride and separated on DEAE cellulose in the presence of urea (Franck and Weeds, 1974). F-actin was prepared from rabbit skeletal muscle as described by Momet and Ue (1984). Factin binding of MHC fragments was analyzed by ultracentrifugation as described by Katoh et al., (1985). Isolation of cDNAs /J-myosin heavy chain cDNA clones were isolated from a human skeletal muscle specific 2gt 10 cDNA library with approximately 3 x 105 independent recombinants (kindly provided by Dr A. Bach, ZMBH, Heidelberg). For clone identification by the plaque hybridization procedure (Benton and Davis, I977), different subfragments of the fl-myosin heavy chain gene were used (Diederich et al., 1989). Radioactive labeling of probes was done using the random primer extension procedure (Feinberg and Vogelstein, I983). Two overlapping cDNAs were identified (cEO2 with 3.26 kb and cLR with 2,15 kb) from 30 myosin heavy chain clones. These coded together for about 80% of the/~-myosin heavy chain, starting at the 5' end with amino acid 362 within the 50 kDa domain of the S-1 head, and extending to the poly(A) tail of the mRNA at the 3' end. The inserts were subcloned into the vector pBS M13 +. The six MHC cDNA's used in these experiments were prepared from cLR. Construction of expression plasmids and production of fusion proteins The myosin cDNA fragments were prepared either by amplifying different fragments of cLR DNA by PCR with primers having specificrestriction sites at their 5' ends (Saiki et al., 1988) or, by cutting cLR DNA or amplified products at naturally preexisting restriction sites. For PCR amplification (Taq DNA

E L D I N et al.

380 polymerase from AGS, Heidelberg/FRG), a newly developed computer-controlled robot (a small portable crane) for the cyclic transport of samples into preheated glycerol baths was used (B/ihler GmbH, T6bingen/FRG). Characteristics of the six MHC constructs and the sequences of the PCR primers are summarized in Table 1. All six inserts were finally purified by electroelution from polyacrylamide gels. Ligation of the inserts, transformation of the E. coli strain pop 2136 and production of fusion proteins were done as described in Moore and co-workers (1987). The pEX (Genofit) expression plasmids used are available in three versions with three different reading frames for cDNA doning (Stanley and Luzio, 1984). Tests o~ lengths and specificities of the inserts in the transformed plasmids were carried out by PCR amplification of the corresponding inserted DNAs with specific oligonudeotides corresponding to different parts of their sequences. PCR amplified fragments were sequenced after cloning to control sequence fidelity. Clones with sequence errors leading to amino acid exchanges were excluded from the experiments. Fusion proteins were isolated after successive treatments of bacterial walls with lysozyme and DNase (Nagai et al., 1985). Each fusion protein represented about 30% of the total E. coli protein (about 20 mg per 100 ml culture medium).

Chemical and proteolytic cleavage conditions Specific cleavage at the Asp-Pro bond was obtained with a 50% formic acid and 4.5 M urea treatment of the whole MHC, or of the fusion proteins, for about 2 or 3 days according to Sutoh

(1983). The conditions for chymotryptic treatment were as follows: myosin or fusion proteins were placed in 0.1 M KC1, 10 mM EDTA, 50 mM phosphate buffer, pH 7.5 at 10--20 mg ml - i, and cut with 0.1 mg m l - ~ chymotrypsin for I0 min. at 25 ~ (Weeds and Pope, 1977).

Renaturation procedures MHC fragments, isolated either chemically or through partial proteolysis, were renatured in solution according to the method of M/ihlrad and Morales (1984). Proteins or fragments, separated by PAGE and electroblotted, were renatured on nitrocellulose sheets. After electroblotting, nitrocellulose sheets were successively soaked for several hours in 20 ml of buffer A [0.2 M sucrose, 5 mM 2-mercaptoethanol, 50 mM Tris acetate, pH 7.8, (Katoh et al., 1985)1 containing first 6 M, then 3 M and finally no urea, respectively. The nitrocellulose sheets were subsequently blocked with buffer B (0.1 M Tris, 0.15 M NaC1, pH 7.4, containing 3% bovine serum albumin and 0.05% Tween 20). They were then used immediately in overlay experiments. Overlay F-actin at about I mg ml - 1 was solubilized in 0.2 mM dithiothreitol, 2 mM CaC12, 0.1 mM NaN 3, 0.1% bovine serum albumin, 2 mMMgC12, 2 mM Tris, pHS.0. Myosin light chains at 0.4 mg m l - ~ were solubilized in buffer B. F-actin or myosin light chains were incubated with renaturated myosin fragments on nitrocellulose sheets for one hour at room temperature. The

Table I. Summary of subclone construction. (A) Subclones obtained by direct cloning of restriction fragments, derived from

currently characterized clones, are indicated according to their amino acid positions. (B) For PCR produced subclones, sequences, lengths (in brackets) and restriction sites (underlined) of the required oligonucleotide primers are noted along with their corresponding amino acid positions. * and ** were ligated to vectors after BamHI digestion at the two natural sites of fl-MHC cDNA (amino acid positions 598 and 1038, respectively). All the PCR products were ligated to pEX3, except fl-HC2 and fl-HC3 which were cloned into pEX2.

Table I.A Name

aa position

8HC1 8HC3 8HC4

599-1 037 51 8 - 5 9 8 599- 1 037

Insertion site 5' 3' BamHI EcoRI BamHI

BamHI BamHI Sail

Origin cLR PCR-8HC2 PCR-13HC2

Table I.B

Name

aa position

8HC2

518-721

PCR oligonucleotides 5, primer 3' primer GGC/GAATTC/CAGGCCTGCAI-FGACCTCATCGAG (33) CG/GTCGAC/GCCTCTGCCGGAAGTCCCCGTAGAG (33)

8HC9

363-598

CTG/GGATCC/GGAG]-I'CAAGCTGAAGCAGCGG (31) CG/GTCGAC/GCCTCTGCCGGAAGTCCCCGTAGAG (33)

BamHI BamHl*

8HCll

670-1037

CTG/GGATCC/GGTACGTTGTATCATCCCTAAT (31) CTG/GGATCCGI-FA/G'I-I'CATCTCGATCTGCACG GA (34)

BamHI BamHl**

Insertion site EcoRI Sail

Synthetic myosin heavy chain fragments recognize actin and light chains nitrocellulose sheets were then incubated with blocking buffer B with or without 5 mM ATP for 30 min at 37 ~ (Kremer et al., 1988). Treatment with anti-actin or anti-myosin light chain antibodies was carried out as described by Matus and coworkers (1980). Alkaline phosphatase coupled anti-mouse or anti-rabbit antibodies were used to detect antigen-antibody complexes. Staining intensities of the bands were determined by densitometric measurements on negative prints of nitrocellulose sheets.

Results

cLR clone encodes for a large M H C fragment The eDNA clone cLR (2166 bp) codes for a fragment of the ]/-myosin heavy chain. This clone contains coding information for about two thirds of the 50 kDa and for the entire 20 kDa subdomain of the S-1 head (447 amino acids) and for most of the S-2 region (276 amino acids). The fact that cLR codes for a ]/-specific heavy chain was determined by first comparing cLR sequences with exon sequences of the ]/-myosin HC gene and then protein sequences of different types of myosin HCs. About 60% (or I300 bp) of the cLR insert was sequenced. The ]/myosin HC gene has been cloned and sequenced along its entire length (Diederi~ch et al., 1989; Jaenicke et al., submitted). Sequenced parts of cLR were essentially identical to DNA sequences of exons 12-20 on the ]/-myosin HC gene (exon numbers correspond to Strehler et al., 1986). Only three base exchanges between the eDNA and the genomic DNA were found. Two are conservative and probably reflect DNA sequence polymorphisms. The

Antibodies The monoclonal antibody specific for the /J-MHC (F36.2C9) was prepared with human ventricular myosin as antigen. Its epitope was located at the N-terminal portion of the myosin rod through immunoelectronmicroscopy. The antibody did not react with myosin S-1 (see Table I in Dechesne et al., 1985). The monoclonal antibody specific for actin was provided by Biogenex Laboratories. Polyclonal antibodies against MLC (myosin light chain) were raised in a rabbit immunized with purified bovine cardiac MLC. Rabbit anti-mouse or goat anti-rabbit antibodies coupled with alkaline phosphatase were obtained commercially. Polyacrylamide gel electrophoresis and Western blotting were carried out according to Deschenne et al., (1985).

A

0

10001

eDNA'

2 0J0 0

381

30001

40001

5q00

60p8 bp

I

cEO2

2760 1170

R

l

I

cLR

B 3335

B

II

: BamHI

|

1920

B

I eDNA Genomic DNA

5'------TCTTCC CTC AAG T T G CT C AGC ACC C T G T T T GCC AAC T A T G CT G G G 5'------TCTTCC CTC AAG T T G CTC AGC ACC C T G T T T GCC AAC T A T G CT G G G

Exon16

|

Exon17

2006

I

G C T G A T G C G CC'I' A'I'T G A G A A G G G C A A A G G C A A G G C C AAGAAA-----3' G C T G A T G C G CC'I' A ' r T G A G A A G G G C AAA G G C A A G G C C AAGAAA-----3'

C SS

LKLLSt

Rat embryonic MHC

- -

n r - - A h - y - t

Rabbit Skeletal F a s t MHC

- k

m t

Huma n /I-MHC

LFANYRGRDAP

I EKGKGKA

KKGSS

Rabbit B-MHC f - t

t

- - d 66

- K - v a K

. . . . .

* - A f - - s g a q a g 9 9 g 6 G g K - - g K

. . . . .

"

Fig. 1. Summary of overlapping human/J-MHC subclones derived from cLR and the sequence of the hypervariable 50-20 kDa junction in the head domain of myosin HC. (A) cLR and cEO2 are two overlapping cDNA clones coding for 80% of human ventricular/J-MHC. (B) DNA sequences, the region between nucleotide positions 1920 and 2006 of the eDNA (or mRNA) is shown. Human/j-myosin HC sequences were derived from eDNA (clone cLR, upper line) and from exons 16 and 17 of the human/J-myosin HC gene (lower line). The position of the intron between exons 16 and 17 is indicated with an arrow. (C) DNA-derived protein sequences; the hypervariable region in human/J-myosin HC and in the embryonic skeletal myosin HC of rat (bottom). For clarity, the hypervariable 'core' (located at the 50 kDa-20 kDa junction) is separated from the conserved N- and C-terminal neighborhood. The corresponding rabbit fast skeletal muscle (Elzinga, 1971), Rabbit/J (Kavinsky et al., 1984) and rat embryonic (Strehler et al., 1986) myosins are given for comparison with human/J-myosin./J-myosin identical residues are indicated with dashes. Residues that are identical in at least two myosin sequences are capitalized. Asterisks represent gaps inserted to maximize homology.

ELDIN et al.

382 third one, very close to the 5' end of cLR, could cause a lys to glu exchange in a region where all known myosin protein sequences are conservative. Therefore, we believe that this base exchange is a cloning artefact. The relevant DNA sequences and DNA-derived amino acid sequence of the junction are shown in Fig. 1. This junction is hypervariable in different myosin HC isoforms over a distance of 25 amino acids. The cLR encoding human 50-20 kDa junction closely resembles the fl-myosin HC sequence of rabbit [2 out of 25 amino acids are exchanged, with one exchange being conservative (Kavinsky et al., 1984)]. All other known vertebrate sarcomeric myosin HC isoforms, such as rat embryonic myosin HC, have 16 or more exchanges in this subregion (for review see Warrick and Spudich, 1986). Thus, the very close similarity of cDNA with genomic fl-myosin HC sequences, and the striking difference between the protein sequence of the human fl-myosin heavy chain junction and that of corresponding junctions in other myosin HC isoforms, is firm evidence that cLR encodes a/J-myosin HC. This conclusion was corroborated by further results which show that most of the/J-MHC fusion proteins used in these experiments reacted with F36.2C9, a /J-specific monoclonal antibody. Immunoelectron microscopy demonstrated that its epitope is located on the N-terminal end of the myosin rod, at the junction between S-1 and the hinge (Dechesne et al., 1987).

Synthesis of a large M H C fragment (flHC1) The first MHC fragment constructed in this study originated from a cDNA subclone (1316bp), obtained by BamH1 cleavage of clone cLR (see Fig. 1A). This subclone coded for an MHC fragment which extended from amino acid positions 599 to 1037. This region was selected because it includes an MHC subsequence know to be

involved in interactions with actin and myosin light chains (Fig. 2) (nomenclature for actin-myosin contacts is from Audemard et al., 1988). The corresponding fusion protein (flHC1) was chosen to be the E. coil-synthesized MHC fragment prototype. It was used to prepare shorter flHC1 subfragments, tentatively free of any fl-galactosidase, and to determine the experimental conditions on the interaction of any MHC fragments with either F-actin or myosin light chains. Fusion protein flHC1 is composed of a 110 kDa flgalactosidase moiety located at the N-terminus of the fusion protein and of the 599-1037 residue MHC fragment. flHC1 and other fusion proteins were insoluble in salt buffers which are regularly used in biochemical studies of myosin molecules, flHC1 could only be solubilized in 6M urea or buffers containing SDS. The fl-galactosidase (flgal) without fused MHC fragment Was also expressed in E. coil According to PAGE, the molecular weights of flgal and/JHC1 were 110 and 160 kDa, respectively (Fig. 3A, lanes a and b). Immunoblotting experiments using the anti-flMHC monoclonal antibody F36.SB9 confirmed that flHC1 contained the corresponding epitope (Fig. 3B, lane b). No immune reaction with this anti-flMHC antibody was observed with the flgal fragment (Fig. 3B, lane a). flHC1 reacted positively with a polyclonal serum specific for fl-galactosidase (not shown).

Chemical cleavage of flHC I: production of a 48 kDa M H C subfragment The cloning linker of the pEX3 plasmid used for producing /JHC1 led to the generation of an asp-pro peptide bond at the junction between the fl-galactosidase and the inserted MHC sequence. It reconstituted the asp-pro bond present in native human fl-MHC and in rabbit skeletal muscle MHC. The peptide bond is cleaved by

Actin _ ~ A

25"1~'D~-'""I

B

I

50 kDa

2O4

MLC C

ao kDa

640

9

DMM

....

1171

808

[

1325 ....

SF2

f}HCl

fJlIC2 BIIC3

B]IC4

I}IIC11 fJllC9

363

520

599 670

721

1037

Fig. 2. Schematic drawing of human fl-MHC subclones expressed in E. coil Spanned regions are indicated according to their respective amino acid positions relative to the complete human fl-MHC amino acid sequence. In the upper part, corresponding tryptic MHC fragments are represented with the currently assumed actin contacts (A, B and C) and MLC (myosin light chain) binding regions (Audemard et al., I988).

Synthetic myosin heavy chain fragments recognize actin and light chains

383

Coomassie Blue Staining

F36.2C9 Immunoblot Fig. 3. SDS-PAGE and western blot analysis of recombinant protein/J-HC1 and its cleavage products. (A) Lane a, fl-galactosidase

(/Jgal.) non-recombinant protein; lane b,/J-HC1 recombinant protein; lanes c and d, urea-formic acid products of ]~gal and/J-HC1 respectively; lanes e and f, chymotrypsin cleavage products of fl-HC1 and bovine ventricular myofibrilsrespectively, lane e', soluble chymotrypsin cleavage products of fl-HC1 obtained after renaturation. (B) Corresponding lanes were electroblotted on nitrocellulose sheets and probed with F3&2C9 monoclonal antibody.

treatment with 50% formic acid and 4.5 M urea for about 2 days (Sutoh, 1983). The same chemical treatment applied to the ]~HC1 fusion protein resulted in its fragmentation into two main subfragments of about 48 and 52 kDa (Fig. 3A, lane d). The 48 kDa fragment, not found in the urea-formic acid treated ]~gal (Fig. 3A, lance c), presumably corresponded to the MHC fragment extending from proline 600 to the C-terminal end (residue I037). This was confirmed by the positive immune reaction of the 48 kDa fragment and the null reaction of the 52 kDa fragment with the anti-flMHC antibody (Fig. 3B, lane d and c respectively). The 52 kDa flHC1 subfragment(s) were der-

ived from the/J-galactosidase moiety, which contains at least two cleavable asp-pro sites (Gilbert and Maxam, 1973), and reacted positively with the polyclonal serum specific for /J-galactosidase (not shown). The 48 kDa MHC subfragments were as insoluble at the entire flHC1 fusion protein.

Chyrnotryptic cleavage of flHC I: production of two soluble MHC subfragments Partial chymotryptic fragmentation of ]~-HCI was assayed, as previously described for the myosin molecule,

384

ELDIN et al.

to determine whether this MHC fragment, which is synthesized by E. CoIL could give rise to soluble MHC subfragments in a low ionic strength solution. Solubilization was a prerequisite for the cosedimentation experiments with F-actin (see Materials and methods). A 10 min. chymotryptic treatment of the entire insoluble fi-HC1, in a low ionic strength buffer (Weeds and Pope, 1977), gave rise to numerous fi-HC1 subfragments reacting with the fl-MHC antibody (Fig. 3 A and B, lanes e). This peptide mixture, which was soluble in 6 M urea only, was then dialyzed against buffers at decreasing concentrations of urea to a final low ionic strength solution containing no urea. The mixture was ultracentrifuged at 100 000 g for one hour to remove protein precipitates which reappeared after dialysis (Katoh, 1985). Two subfragments of 28 and 22 kDa were found in the supernatant. They were barely detectable on PAGE with Coomassie Blue, but intensely reacted with the anti-flMHC antibody (Fig. 3A, B). None of the soluble subfragments reacted with the polyclonal serum specific for/J-galactosidase (not shown), indicating that they were derived from the MHC moiety of the ]~HC1 fusion protein. Beginning with one mg insoluble /JHC1 protein, about 10 #g of each soluble peptide was finally obtained. These soluble peptides were later used in cosedimentation experiments with actin. Chymotryptic treated ventricular myofibrils, which mainly contained authentic S1-HC, LC1, rods and actin (Weeds and Pope,

1977), were used as controls in the experiments using entire or subfragmented flHC1 (Fig. 3A and B, lanes f). Overlay experiments of the entire flHC1 or its subfragments, with actin or myosin light chains Overlay experiments were conducted to investigate potential interactions of the flHC1 fusion protein or its subfragments with rabbit skeletal muscle F-actin or myosin light chains prepared from bovine ventricular myosin (Franck and Weeds, 1974). Such interactions were detected either with a specific anti-actin mouse monoclonal antibody or a polyclonal serum raised in rabbifs immunized with bovine cardiac light chains. Alkaline phosphatase coupled anti-mouse or anti-rabbit antibodies detected the antigen-antibody complexes formed. We found that a renaturation procedure developed from that used for MHC cyanogen bromide peptides in solution (Katoh et al., 1985), or from overlay experiments on PAGE gels (Mitchell et al., 1986), was required to observe F-actin or myosin light chain binding. In the actin overlay experiments with ]~HC1 fusion protein and its 48 kDa urea-formic acid subfragment, Factin binding occurred with the complete flHC1 protein and with its 48 kDa subfragment (Fig. 4, lanes b and d, respectively). In contrast, using the same experimental conditions, F-actin did not bind with any of the other soluble chymotryptic flHC1 subfragments (not shown).

F-Actin Overlay Fig. 4. F-actin blot overlay on entire fl-HC1 protein and its formic acid cleavage products. After SDS-PAGE separation, and electroblotting on nitrocellulose, sheets were incubated with F-actin and then probed as outlined in 'ExperimentalProcedures'. Lane a, ]~gal;lane b, fl-HC1; lanes c and d, urea-formic acid products of/Jgal and//-HCI respectively; lane e, bovine ventricular myofibrils treated by chymotrypsin.

Synthetic myosin heavy chain fragments recognize actin and light chains The S-1 heavy chain controls, obtained by chymotryptic treatment of bovine cardiac myofibrils, reacted even more intensely with the anti-actin antibodies (Fig. 4, lane 3). The weak immune reaction of cardiac actin, present in the myofibril extract, resulted from the monoclonal anti-actin antibody being directed against skeletal actin. In control experiments, the complete flgal fragment, its 52 kDa flgal subfragment and the myosin rod did not react with actin (Fig. 4, lanes a, c and e). We conclude that F-actin recognizes the entire flHC1 protein and its 48kDa subfragment, but not smaller MHC subfragments, in a way qualitatively similar to its recognition of the authentic S-1 fragment. When 5 mM ATP was added to the blocking buffer B after F-actin incubation (see experimental procedures), F-actin still bound to flHC1 and to its 48 kDa subfragment and to the S-1 heavy chains. This contrasts with results obtained by Mitchell et al., (1986) using a gel overlay technique, which showed that S-1-Factin complexes were dissociated by adding ATP. The entire flHC1 protein and its subfragments were exposed to different solutions of cardiac bovine myosin light chains in overlay experiments. The analysis was

385

d o n e using polyclonal anti-LC antibodies. Only the 48 kDa urea-formic acid flHC1 subfragment was clearly positive with the anti-LC after overlay with LC1 and/or LC2. The induced staining intensities were quite similar (Fig. 5, lanes e-g). However, neither the entire flHC1 protein (Fig. 5, lanes a and b) nor the soluble chymotryptic flHC1 subfragments (not shown) reacted with the anti-LC. In positive control experiments, S-1 heavy chains obtained by chymotryptic treatment of bovine cardiac myofibrils were also found to react positively with both LCs in overlay experiments (Fig. 5, lane h). Negative controls with no anti-LC antibody staining were obtained with no MLCs or with the 52 kDa flgal subfragment (Fig. 5, lanes c and d). No ATP sensitivity on LC binding was detected. We conclude from these experiments that the intact flHC1 fusion protein and its small soluble fragments do not bind LC, but its 48 kDa fragment does. As an additional methodological test, the same overlay experiments with actin or myosin light chains were also performed with another MHC fusion protein, spanning an MHC region of the subfragment 2 between the 1038th and the 1164th residue derived from the cEO2 clone by

Myosin Light Chain Overlay Fig. 5. Myosin light chain blot overlay on entire fl-HCI protein and its formic acid cleavage products. After SDS-PAGE separation, and electroblotting on nitrocellulose, sheets were incubated with MLC and then probed as outlined in Materials and methods. Lane a, flgal; lane b, fl-HC1; lane c, 52 kDa fl-HC1 formic acid cleavage products; lanes d-g, 48 kDa fl-HC1 formic acid cleavage products; lane h, chymotryptic digest of cardiac myofibrils. Overlays were performed either with VLC1 alone (lane e), VLC2 alone (lane f), both MLC (lanes a, b, d, g and h) or no MLC (lane d).

ELDIN et al.

386 adequate PCR amplification (position 3198th to 3578th). This fusion protein, which contained no known areas of interactions with actin or light chains, reacted with none of the myosin natural partners (not shown).

F-actin cosedimentation of chymotrypsin treated flHC1 soluble fragments Since the renaturation of chymotryptic fl-HC1 fragments enables isolation of two soluble MHC subfragments, 10-20 pg F-actin at 1 mg m l - i was added to about 10 #g of the 22 and 28 kDa flHC1 fragments solubilized in the low ionic strength solution currently used for cosedimentation experiments (Katoh et aL, 1985) (Fig. 3, lane e'). A second sample containing the same peptides but without F-actin was also prepared. After one hour incubation and subsequent ultracentrifugation (30 min., 100 000 g), pellets and supematants of the samples with and without F-actin were analyzed on PAGE and compared with anti-flMHC antibody. After incubation with Factin, the 28 and 22 kDa fragments were found in the pellet (Fig. 6, lane 3). These fragments remained in the supernatant in the control experiment without F-actin (Fig. 6, lanes I and 2). When 5 mM ATP was added during the incubation with F-actin, both fragments still cosedimented. These results indicate that F-actin is able to bind soluble peptides prepared from the recombinant flHC1 in an ATP-insensitive manner. Construction of other M H C fusion proteins Based on data obtained with rabbit skeletal muscle myosins, the main actin interfaces were postulated to be located on at least three different MHC contacts or discontinuous MHC regions in the 20 kDa and 50 kDa tryptic MHC fragments. The hypothetical actin-myosin interacting sites (designated A, B and C in Fig. 2) were deduced from Audemard et al., (1988). In order to assign actin binding sites to regions on the C-terminal part of the S-1 domain of the cardiac fl-MHC, five additional MHC fusion proteins, designated fl-HC2, fl-HC3, fl-HC4, flHC9 and fl-HC11, were constructed (Fig. 2). The precise endpoints of these MHC fragments, and other points concerning the plasmid constructions, are noted in Table 1. Three of these fragments spanned the entire region or parts of it, between residues 518 and 721, hence including the three MHC sites which are thought to bind actin. The longest of these three subfragments (fl-HC2) was synthetized from the cDNA cLR by polymerase chain reaction using different pairs of oligonucleotide primers equipped with specific restriction sites at their 5' ends (Table 1). The fusion proteins (fl-HC3 and fl-HC4) were derived from flHC2 by cleavage with BamH1. Since fl-HC3 did not interact with actin, the longest fusion protein (fl-HC9) was constructed to test the interaction with actin at MHC site A. The fl-HC11 fragment was constructed to test the interaction with actin at the MHC site C. Since fl-HC9 is derived from the cLR clone, it has a glu residue at its N-

terminal end, but not the lys residue which is found in the genomic sequence.

Investigation of some actin interfaces on cardiac fl heavy chains The five additional MHC fusion proteins contained the same fl-galactosidase fusion carrier and were as insoluble as fl-HC1. Their molecular weights were analyzed by SDS on PAGE (see Fig. 7). Actin overlay experiments were simultaneously carried out with the six MHC fusion proteins as described for fl-HC1 and its subfragments (Fig. 4). We observed, in two sets of experiments with independently derived fusion proteins, different phosphatase staining intensities after F-actin binding to single MHC fusion fragments. A densitometric analysis of photographic negatives showed that the strongest relative signal was obtained with fl-HC2 and fl-HC4. fl-HC3 was at the opposite end of the scale and showed no staining at all. Intermediate staining intensities, below fl-HC2 and fl-HC4 values were obtained in the following order: S-1-HC > fl-HC2 ,,~ fl-HC4 > fl-HC1 > fl-HC9 ~ fl-HC11 > fl-HC3 (,,~0) (Fig. 7). We noted that fl-HC2 was the only MHC fragment which contained all of the three postulated actin binding sites. All other fragments contained only one or two putative actin binding sites, fl-HC1 and fl-HC4 contained the electrostatic actin binding site near the connector region (amino acid 633 to 643; site B), and the hydrophobic actin contact near the very reactive thiols (SH1/cys707 and SH2/cys697; site C). fl-HCll contained only the actin binding site C. fl-HC3, which contained only the putative actin binding site near the thrombic cleavage site (lys 560-ser 561; site A), unexpectedly did not react with Factin. There was no attempt to study the possible ATPdependency of the actin-MHC fusion protein complexes. Moreover, none of the five fusion MHC proteins reacted with light chains. Discussion

We have described the expression of different fragments of human ventricular d-myosin heavy chains in E. coli. MHC coding sequences, expressed in a prokaryotic system, give rise to 'functional' protein fragments, i.e. MHC fragments reacting specifically with their main natural partners: actin and myosin light chains. Our study expands on recent reports involving genetic engineering of the myosin molecule and concludes that no eukaryotic mechanisms are necessary for synthesizing mammalian MHC fragments in E. coli. This approach offers a new potential for the construction of functional chimeric myosins or fragments from mammalian genes. The recently developed PCR technique, its successive application and the use of different linkers with double recognition sites, wherein one part recognizes the sequence to be amplified and the other contains the restriction site required for insertion of the amplified

Synthetic myosin heavy chain fragments recognize actin and light chains fragment into a cloning vector (Orlandi et al., 1989), eliminates many previous time-consuming procedures. Hereafter, it will be possible to obtain rapidly MHC fragments from any particular amino acid. Using myosin cDNA, we are no longer dependent on the chemical reactivity of a few peptide bonds located at scattered points along the MHC, or on the protease sensitivity of MHC regions. The only requirement is MHC DNA sequence information. The interactions observed here between F-actin and different MHC fragments synthesized in E. coli are due to the MHC fragments, not to the fl-galactosidase portions of the fusion proteins. This function of genetically engineered MHC fragments was derived from studies made with native fl-HCI and its chemical or proteolytic fragments. The 48 kDa urea-formic acid/J-HC1 subfragment spans a MHC region of 436 amino acids, from proline 599 to aspartic acid 1037. This fl-HC1 subfragment was found to be normal sized and to be a pure •-MHC fragment which reacts with actin and myosin light chains. Furthermore, both soluble 22 and 28 kDa chymotryptic fl-HC1 subfragments, which contain the epitope of one/J-MHC

Fig. 6. F-actin co-sedimentation of chymotrypsin treated /JHC1 soluble fragments. Crude fragments obtained after chymotrypsin treatment of fl-HC1 were renatured, ultracentrifuged, placed with or without F-actin and again ultracentrifuged as outlined in Results (Katoh et al., 1985). All samples were immunoanalyzed with a specific anti-]/-MHC monoclonal antibody (F36.2C9). Soluble fragments in the absence of actin: (I) supematant and (2) pellet after ultracentrifugation. Soluble fragments in the presence of actin: (3) pellet after ultracentrifugation. Nothing was observed in the corresponding supemarant.

387

specific monoclonal antibody, interact with F-actin in solution. This epitope has been found at the N-terminal end of the chymotryptic rod (Dechesne, 1985), consequently it is located near the postulated chymotryptic site of native myosin (near phenylalanine 840) (Morner et al., 1981). The 22 kDa fragment, and possibly the 28 kDa fragment, span yet undetermined areas of about 200 and 250 residues respectively, near phenylalanine 840. Consequently, most of these chymotryptic fragments arise from the 20 kDa fragment of the myosin molecule and a fragment of the myosin rod, and are probably free of any fl-galactosidase residue. The fact that some MHC fusion proteins did not interact with F-actin, suggests that these observed interactions are neither artefactual nor due to the known sticky nature of F-actin. The recombinant MHC fragments investigated in this study are not comparable with fragments obtained from natural myosin molecules: in all of their properties or behaviour in solution their conformations and ATPsensitivities differ. For example, from the different reactions with the fl-HC1 fusion protein and its chymotryptic soluble subfragments, we can assume that, in the same low ionic strength buffer, fusion protein /J-HC1 and native myosin have different conformations, at least in the MHC region of the phenylalanine 840. These conformdtional-dependent variations in MHC cleavage around this phenylalanine residue have also been observed with fragments prepared from native myosin molecules thus they are not unique to genetically engineered MHCs. For example, Burke et al., (1983) observed that S-1, free of light chains and treated with trypsin, gave rise to a new S1 derivative containing the 23-50-18 kDa domains. This 18 kDa severed heavy chain fragment, which is a 20 kDa fragment shortened at its C-terminal end, was unable to reassociate with any light chain, as native S-1 heavy chain did. Such environmental differences could explain why whole fl-HC1 fusion proteins barely reacted with light chains, whereas their 48 kDa fragments did. A few comments should be made concerning ATPinsensitivity observed here with MHC fragments synthesized by E. CoIL First, the limits of the nitrocellulose overlay technique and its apparent inadequacy for measuring ATP-dependency of MHC-binding with F-actin should be discussed. In contrast to the observations of Mitchell et al., (1986) using gel overlays, even S-1 heavy chain-F-actin complexes bound to nitrocellulose are not dissociated by ATP. The reason for the differences between these two solid-phase systems is unknown. We cannot eliminate the possibility that the tighter binding of proteins on nitrocellulose after their electroblotting preclude their refolding. This could also be the reason why both soluble chymotryptic MHC fragments embedded in nitrocellulose do not bind F-actin in the overlay experiments. The use of MHC fragments prepared by drastic chemical treatments of natural myosin molecules or isolated as insoluble proteins from the inclusion bodies of bacterial

388

ELDIN et al.

extracts, required subsequent renaturation. In both cases, the efficiency of these recovery processes is a posteriori evaluated by functional tests. In this context, it is worth considering the origin of the functional differences between MHC fragments issued from the same region of the myosin molecule and prepared from either/J-HC1 fusion protein (as in this study) or from native myosin molecules as described by Griffiths and Trayer (1989). These authors used formic acid digestion, guanidinium chloride and urea

Coomassie Blue Stained

F-Actin Overlay Fig. 7. F-actin blot overlay on six different human fl-MHC subfragments expressed in E. Coli. F-actin overlay was performed as outlined in Materials and methods. Chymotrypsin treated myofibrils were used as control (lane a). Lane b, fl-HC1; lane c, fl-HC2; lane d, fl-HC3; lane e, fl-HC4; lane f, fl-HC9; lane g, fl-HCll (see Fig. 2 for the precise clone locations).

treatment successively to chymotryptic S-1 prepared from native myosins. They obtained a functional 26 kDa fragment which binds F-actin in an ATP-sensitive manner. We prepared soluble chymotryptic fragments of fl-HC1 fusion protein from the same myosin region as the 26 kDa fragment. The fragments revealed similar actin binding properties but in an ATP-insensitive manner. These actin binding tests, which are made in solution, are technically similar but we must consider the possibility that different MHCs have different conformations. The procedure used for refolding MHC fragments of bacterial origin may be insufficient to convert an artificial, abnormal conformation which these fragments may assume during their synthesis in E. coil These observations underline the limitations of using insoluble MHC fusion proteins made in E. coli, as in the present study. Recently, alternative vectors for expressing non-fused and soluble proteins have been proposed. One has been used successfully to express functional myosin rod fragments (Maeda et al., 1989). The same vectors are currently being tested for their ability to express soluble and functional myosin S-1 fragments so that they may eventually be used in solution, in cosedimentation or competition experiments and not only in solid phase systems like in overlay experiments. In spite of these technical limitations, the present experiments on the localization of acto-myosin interface(s) extend previous results using essentially fast rabbit skeletal muscle MHC and fragments, to now include human cardiac fl-MHC. Yamamoto (1989), in determining the amino acid residues involved in the actin-subfragment I interface, recently reported that the whole length of the S-1 site B (amino acid 633 to 643, site B in Fig. 2) fully interacts with the N-terminal acidic sequence of actin, at least in rigor states of the rabbit skeletal muscle myosin. We observed that cardiac MHC subfragments at different length (]~-HC2 and ]~-HC4), which contain this connector region, also reacted with actin. The relatively high levels of amino acid substitutions detected in the region between the cardiac and skeletal MHC, do not prevent actin binding. We noted that both cardiac and skeletal MHC actually contain the same proportion of lysine residues which are believed to be crucial for strong electrostatic interactions with actin (Fig. 1) (Chaussepied et al., 1988). The positive interaction of fl-HCll with actin suggests that regions around the very reactive thiols (cysteine residues 697 and 707; site C in Fig. 2) which are conserved in cardiac and skeletal MHC are involved in the actin interface, independently of the strong actin binding site B (Katoh et al., 1984). Our observation that the short cardiac fl-HC3 (located near the 50 kDaZ20 kDa junction) does not interact with F-actin, whereas the long overlapping (mutated at its first N-terminal residue) fl-HC9 does interact, differs from previous conclusions drawn from crosslinking and peptide experiments carried out on the same area of rabbit skeletal MHC (actin binding site A). Different hypotheses may be proposed. The first possibility is that there are differences in actin interaction domains

Synthetic myosin heavy chain fragments recognize actin and light chains between cardiac and skeletal MHC in the fl-HC3 region, or that the 80 residue fl-HC3 is too short to resume a conformation suitable for actin interaction, or a MHC fragment of the 50 kDa region interacts with F-actin. According to Bonet et al., (1988), crosslinking experiments of the acto-S1 complex using phenyldiglyoxal were restricted to the 50 kDa fragment, involving rabbit MHC arginine residues and spanning residues 239-455. This subregion is present in fl-HC9, but not in fl-HC3. According to Bettache et al., (1989), a covalent linkage which is restricted to the rabbit skeletal muscle 50 k D a M H C fragment is possible with non-polymerisable G-actin. Perhaps the actin binding site within the 50 kDa region is closer to its N-terminal end than postulated by Audemard et al., (1988). Actin interfaces with the 50 kDa region will be better defined via future testing with new MHC constructs. In conclusion, with our results, the notion that actin interfaces on MHC are discontinuous can now be extended to include cardiac fl-MHC. Additional experiments, involving smaller and/or overlapping, native or mutated MHC areas, derived from whole MHC genes, are now planned in order to locate precisely MHC areas interacting with actin, and to determine the possible cooperation between different actin contacts. Using this same approach, actin interactions in the 23 kDa Nterminal part of MHC (M/ihlrad, 1989), and the competition between ATP and actin for the N and C terminal S-I segments (Mornet et aI., 1989; Maruta et al., 1989), must now be analyzed. The anticipated close comparison between heavy chain domains from different myosin isoforms is possible. Competition and crosslinking experiments between recombinant and native MHC fragments could be highly informative. Finally, by providing any chimeric or mutated MHC fragment independently of proteolytic sites, the present procedure suggests a new way of investigating molecular mechanisms of the myosin molecule in the energy transduction process during muscle contraction.

Acknowledgments This work was supported by l'Institut de la Sant4 et de la Recherche M6dicale, le Centre National de la Recherche Scientifique et l'Association fran~aise des Myopathes. P.E. is a research fellow from the Minist6re de la Recherche et de la Technologie; K. W. D. is a research fellow from the Deutsche Forschung Gemeinschaft.

389

between the ATPase and actin sites in myosin. ]. Muscle Res. Cell. Mot. 9, 197-218. BENTON,W. D. ~ DAVIS,R. W. (1977) Screening lambda gt recombinant clones by hybridization to single plaques in situ. Science 196, 180-2. BETTACHE, N., BERTRAND, R. & KASSAB, R. (1989) Coupling of nonpolymerizable monomeric actin to the F-actin binding region of the myosin head. Proc. Natl. Acad. Sci. U.S.A. 86, 6028-32. BONET, A., AUDEMARD, E. & MORNET, D. (1988) The actin-myosin subfragment-1 complex stabilized by phenyldiglyotal. ]. Biol. Chem. 263, 14115-21. BOTTS, J., THOMASON, J. F. & MORALES, M. F. (1989) On the origin and transmission of force in actomyosin subfragment 1. Proc. Natl. Acad. Sci. U.S.A. 86, 2204-8. BOUVAGNET, P., LINGER, J. O. C., PONS, F., DECHESNE, C. & L(~GER, J. J, (1984) Fiber types and myosin types in human atrial and ventricular myocardium. Circ. Res. $5, 764-804. BURKE, M., SIVARAMAKRISHNAN, M. & KAMALAKANNAN, V.

(1983) Properties of the alkali light chain association to the heavy chain of myosin subfragment I. Biochemistry 22, 3046-53. CHAUSSEPIED, P. & MORALES, M. F. (1988) Modifying preselected sites on proteins: The stretch of residues 633-642 of the myosin heavy chain is part of the actin binding site. Biochemistry 27, 1778-85. DECHESNE, C., LINGER, J. O. C., BOUVAGNET, P., CLAVIEZ; M. &

LINGER,J. J. (1985) Fractionation and characterization of two molecular variants of myosin from adult human atrium. ]. Mol. Cell. Cardiol. 17, 753--67. DECHESNE, C., BOUVAGNET, P., WALZTHONY, D. & LINGER, J. J.

(1987) Visualization of cardiac ventricular myosin heavy chain homodimers and heterodimers by monoclonal antibody epitope mapping. ]. Cell. Biol. 105, 3031-7. DIEDERICH, K. W., EISELE, I., RIED, T., JAENICKE, T., LICHTER, P. &

VOSBERG,H-P. (1989) Isolation and characterization of the complete human r-myosin heavy chain gene. Human Genetics 81, 214-20. ELZINGA,M. (1971) Amino-acid sequence around 3methylhistidine in rabbit skeletal muscle actin. Biochemistry 10, 224-9. FEGHALI, R. & LEINWAND, L. A. (1989) Molecular genetic characterization of a developmentally regulated human perinatal myosin heavy chain. ]. Cell. Biol. 108, 1791-7. FEINBERG, A. P. & VOGESTEIN, B. (1983) A method for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 132, 6-13. FRANCK, G. & WEEDS, A. G. (1974) The amino acid sequence of the alkali light chain of rabbit skeletal muscle. Eur. ]. Biochem. 44, 317-34. GILBERT, W. & MAXAM, A. (1973) The complete nucleotide sequence of the lac operator. Proc. Natl. Acad. Sci. U.S.A. 70, 358I-4. GRIFFITHS, A. L & TRAYER, I. P. (1989) Selective cleavage of skeletal myosin subfragment-1 to form a 26 kDa peptide which shows ATP-sensitive actin binding. FEBS Letters 242, 275-8. KARSCH-MI, ZRACHI I, TRAVIS, M., BLAU, H. & LEINWAND, L. A.

References AUDEMARD, E., BERTRAND, R., BONET, A., CHAUSSEPIED, P. &

MORNET,D. (1988) Pathway for the communication

(1989) Expression and DNA sequence analysis of a human embryonic skeletal muscle myosin heavy chain. Nucl. Acids Res. 17, 6167-79. KATOH,T. (1984) Binding of F-actin to a region between SH1

390 and SH2 group of myosin subfragment 1 which may determine the high affinity of acto-subfragment 1 complex at rigor. ]. Biochem. (Tokyo) 95, 447-54. KATOH, T., KATOH,H. & MORITA,F. (1985) Actin binding peptide obtained by cyanogen bromide cleavage of the 20 kDa fragment of myosin subfragment 1. ]. Biol. Chem. 260, 6723-7. KAVINSKY,C. J., UMEDA,P. K., LEVIN,J. E., SINHA,A.M., NIGRO, J. M., JAKOVCIC,S. & RABINOVITZ,M. (1984) Analysis of cloned mRNA sequences encoding subfragment 2 and part of subfragment 1 of alpha- and beta-myosin heavy chains of rabbit heart. ]. Biol. Chem. 259, 2775-81. KLOTZ, C., SWYNGHEDAUW,B., MENDES,H., MAROTTE,F. & LINGER,J. J. (1981) Evidence for new forms of cardiac myosin heavy chains in mechanical heart overloading and ageing. Eur. ]. Biochem. 115, 415-21. KRAFT,R., BRAZOVEHNDER,M., TAYLOR,D. A. & LEINWAND,L. A. (1989) Complete nucleotide sequence of full length cDNA for rat alpha-cardiac myosin heavy chain. Nucl. Acids Res. 17, 7529-30. KREMER,L., DOMINGUEZ,J. E. & AVILA,J. (1988) Detection of tubulin-binding proteins by an overlay assay. Anal. Biochem. 175, 91-5. LICHTER, P., UMEDA,P. K., LEVIN,J. E. & VOSBERG,H. P. (1986) Partial characterization of the human d-myosin heavy chain which is expressed in heart and skeletal muscle. Eur. J. Biochem. 160, 419-26. LOWLY, S., SLAYER,H. S., WEEDS,A. G. & BAKER,H. (1969) Substructure of the myosin molecule. Subfragments of myosin by enzymatic degradations. ]. Mol. Biol. 42, 1-29. MAEDA, K., SCZAKIEL,G., HOFMAN,W., MENETRET,J. F. & WITTINGHOFER,A. (1989) Expression of native rabbit light meromyosin in Escherichia coil ]. Mol. Biol. 205, 269-73. MARUTA,S., MIYANISHI,T. & MATSUDA,G. (1989) Localization of the ATP-binding site in the 23-kDa and 20-kDa regions of the heavy chain of the skeletal muscle myosin head. Eur, ]. Biochem. 184, 213-21. MATUS, A., PEHLING,G., ACKERMANN,M. & MAEDER,J. (1980) Brain postsynaptic densities: their relationship to glial and neuronal filaments. J. Cell. Biol. 87, 346-59. McNALLY, E. M., GOODWIN,E. B., SPUDICH,J. A. & LEINWAND,L. A. (1988) Coexpression and assembly of myosin heavy chain and light chain in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 85, 7270-3. McNALLY, E. M., GIANOLA,K. M. & LEINWAND,L. A. (1989) Complete nucleotide sequence of full length cDNA for rat beta-cardiac myosin heavy chain. Nucl. Acids Res. 17, 7527-8. MITCHELL,E. J., JAKES,R. & KENDRICK-JONES,J. (1986) Localization of light chains and actin binding sites on myosin. Eur. J. Biochem. 161, 25-35. MIYANISHI, T., TOYOSHIMA,C., WAKABAYASHI,T. & MATSUDA, G. (1988) Electron microscopic study on the location of 23 kDa and 50 kDa fragments in skeletal myosin head. ]. Biochem. 103, 458-62. MOORE, D. D., SEIDMAN,J. G., SMITH,J. A. & STRUHLK. (1987) in Current protocols in molecular biology, (AUSUBEL,F. M., BRENT, R., KINGSTON,R. E., editors) John Wiley and Sons, NY.

E L D I N et al. MORNET, D., BERTRAND,R., PANTEL,P., AUDEMARD,E. & KASSAB, R. (1981) Proteolytic approach to structure and function of actin recognition site in myosin heads. Biochemistry 20, 2110-20. MORNET, D. & UE, K. (1984) Proteolysis and structure of skeletal muscle actin. Proc. Natl. Acad. Sci. U.S.A. 81, 3680-4. MORNET, D., BONET,A., AUDEMARD,E. & BONICEL,J. (1989) Functional sequences of the myosin head. ]. Muscle Res. Cell. Mot. 10, 10-24. MOHLRAD, A. & MORALES,M. F. (1984) Isolation and partial renaturation of proteolytic fragments of the myosin head. Proc. Natl. Acad. Sci. U.S.A. 81, 1003-7. MOHLRAD,A. (1989) Isolation and characterization of the Nterminal 23-kilodalton fragment of myosin subfragment 1. Biochemistry 28, 4002-10. NAGAI, K., PERUTZ,M. F. & POYART,C. (1985) Oxygen binding properties of human mutant hemoglobins synthesized in Escherichia coll. Proc. Natl. Acad. Sci. U.S.A. 82, 7252-5. ORLANDI, R., GOSSOW,D. H., JONES, P. T., & WINTER,G. (1989) Cloning immunoglobulin variable domains for expression by the polymerase chain reaction. Proc. Natl. Acad. Sci. U.S.A. 86, 3833-7. OSTERMANN, J., HORWICH,A. L., NEUPERT,W. & HARTL,F.-H. (1989) Protein folding in mitochondria requires complex formation with hsp60 and ATP hydrolysis Nature (London) 341, 125-30. OSTERMANN, J. et aL, (1989) Protein folding in mitochondria requires complex formation with hsp60 and ATP hydrolysis Nature (London) 341, 125-30. REINACH, F. & FISCHMAN,D. A. (1985) Recombinant DNA approach for defining the primary structure of monoclonal antibody epitope. ]. MoL Biol. 181, 411-22. RIMM, D. L., SINARD,J. H. & POLLARD,T. D. (1989) Location of the head-tail junction in myosin. ]. Cell. Biol. 108, 1783-9. SAIKI, R. K., GELFAND,D. H., STOFFEL,S., SCHARF,S. J., HIGUCHI, R., HORN, G. T., MULLIS,K. B. & EHRLICH,H. A. (1988) Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239, 487-91. SHIMIZU, T., REINACH,F. C., MASAKI,T. & FISCHMAN,D. A. (1985) Analysis of the metal-induced conformational change in myosin with a monoclonal antibody to light chain two. J. Mol. Biol. 183, 271-82. STANLEY,J. K. & LUZIO, R. (1984) Construction of a new family of high efficiency bacterial expression vectors: identification of cDNA clones coding for human liver proteins. Embo J. 3, 1429-34. STREHLER,E. E., STREHLER-PAGE,M-A., PERRIARD,J-C., PERIASAMY,M. & NADAL-GINARD,B. (1986) Complete nucleotide and encoded amino acid sequence of a mammalian heavy chain. J. Mol. Biol. 190, 291-317. SUTOH, K. (1983) Mapping of actin binding sites on the actin sequence. Biochemistry 22, 1579-85. TRAYER, I. P., TRAYER,H. R. & LEVINE,B. A. (1987) Evidence that N-terminal region of the A1 light chain and myosin interacts directly with the C-terminal region of actin. Eur. J. Biochem. 164, 259-66. WARRICH,H. M. & SPUDICH,J. A. (1986) Myosin structure and function in cell motility. Ann. Rev. Cell. Biol. 3, 379-421.

Synthetic myosin heavy chain fragments recognize actin and light chains (1975) Separation of subfragment 1 isoenzymes from rabbit skeletal muscle myosin. Nature (London) 25 7, 54-6. WEEDS,A. G. & POPE, B. (1977) Studies on the chymotryptic digestion of myosin. Effects of divalent cations on proteolytic susceptibility. ]. Mol. Biol. 1 I I , 129-57.

WEEDS, A. G. & TAYLOR, R. S.

YAMAUCHI-TAKIHARA, K., SOLE, M. J., LIEW, J., ING, D. & LIEW, C.

391

c. (1989) Characterization of human cardiac myosin heavy chain genes. Proc. Natl. Acad. Sci. U.S.A. 86, 3504-8, YAMAMOTO, K. (1989) Binding manner of actin to the lysinerich sequence of myosin subffagment 1 in the presence and absence of ATP. Biochemistry 28, 5573-7.