Charge Interactions in the Rod Domain Drive Formation of Tetramers ...

VOl. 269, No. 28, Issue of July 15, pp. 18679-18685,1994 Printed in U.S.A.

T H E JOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Charge Interactions in the Rod Domain Drive Formationof Tetramers during Intermediate Filament Assembly* (Received for publication, February 25, 1994, and in revised form, April 13, 1994)

Jin-jun Meng, Sohaib Khan, and Wallace Ips From the Department of Anatomy and Cell Biology, University of Cincinnati College of Medicine, Cincinnati. Ohio 45267-0521

Protein subunits introduced into cells, either by DNA transfection or by protein microinjection, integrate into existing IF arrays rapidly andspecifically (reviewed by Albers and Fuchs (1992)). Moreover, exchange of subunits between a polymerized pool and a subunit pool has been demonstrated for the neuronal IF protein, NF-L (Angelides et al., 19891, and critical concentrations in the same order of magnitude as that of actin have been reported for NF-L (Angelides et al., 1989) and for desmin (Ip and Fellows, 1989; Chou et al., 1990a). Current evidence indicates that the assembly of IFs from their subunit proteins isa multistep process involving a number of intermediate oligomeric structures (Aebi, et al. 1983; moSteinert, 1991a; Steinert et al., 1993a, 1993b, 1993~). The nomeric IF polypeptide has a characteristic tripartite structure that consists of an a-helical rod domain flanked by nonhelical head and tail domains (Geisler and Weber, 1982; Hanukoglu and Fuchs, 1982). The sequence of the rod domain, which is believed t o constitute the backbone of an IF, exhibits a heptad a stretch of repeat inwhich the 1st and the 4th residues within 7 are hydrophobic residues. This motif is indicative of coiled coil-forming ability (Crick, 1953). Further analyses at the sequence and protein chemical levels have confirmed that the two-chain molecule (the dimer) isa coiled coil of polypeptides held together in a parallel and in-register arrangementby interactions between those hydrophobic residues (Parry et al., 1977, 1985; Quinlan et al., 1986). A second assembly intermeIntermediate filaments (IFs)’ are a major cytoskeletal com- diate that has been studied extensively is the four-chain mol(1992) and by ponent of eukaryotic cells. As they are relativelyinsoluble un- ecule (tetramer) (reviewed by Parry and Steinert der physiological ionic conditions, IFs are often misconstrued Stewart (1993)).While many issuesconcerning the structureof as static elementsof the cytoskeleton. However, this notion is the tetramer are still a matter of some debate, it isgenerally incompatible with the dramatic reorganizationof IFs that oc- acknowledged that it is bipolar (Geisler et al., 1985) and that curs during manybiological processes. For example, desmin is there may be polymorphism at this level of organization, pertranslocated from a filamentous cytoskeletal array to the Z- haps resulting ina range of molecular lengths (Potschka,1986; lines of myofibrils during early skeletal myogenesis (Granger Aebi et al., 1988; Steinert, 1991a).Many structural models and Lazarides, 1982; Gard and Lazarides, 1982; Tokuyasu et have been proposed as to how four-chain molecules may pack al., 1984). Rearrangement of cytoplasmic IFs into perinuclear into 10-12-nm diameter IFs (Crewther et al., 1983; Geisler et whorls (Blose and Chacko,1976), cyclical depolymerization, al., 1985; Ip et al., 1985a; Aebi et al., 1988; Hisanaga and and repolymerization (Chouet al., 1989) alsooccur in manycell Hirokawa, 1990; Coulombe and Fuchs, 1990; Steinert, 1991b; types during mitosis, probably under regulation by the cell Geisler et al., 1992; Steinert et al., 1993a, 1993b, 1993c; Heins et al., 1990b). Recent experimental et al., 1993).However, the lack of detailed structural informacycle kinase, ~ 3 4 “ ~(Chou “’ evidence also arguesconvincingly for a dynamic nature of IFs. tion at the tetrameric and higheroligomeric levels has made it difficult t o evaluate critically the validity of any one model. a long standing hypothThe purpose of this study was to test * This work was supported in part by Grant AR35973 from the National Institutes of Health and by the Muscular Dystrophy Association. esis regarding theforces that drive the formation of oligomeric The costs of publication of this article were defrayed in part by the structures beyond the dimer stage. has It been known forsome payment of page charges. Thisarticle must therefore be hereby marked time thaton the outsideof the IFcoiled coils the charged amino “advertisement” in accordance with 18 U.S.C.Section 1734 solely to acidsaredistributed periodically, suchthat positively and indicate this fact. negatively charged residues are essentially arranged in alter$ Recipient of an established investigator award from the American Heart Association National Center. To whom correspondence should be nating zones (Parry et al., 1977; McLachlan andStewart, adressed: Dept. of Anatomy and Cell Biology, University of Cincinnati 1982). This has given rise to the proposal that, if neighboring College of Medicine, ML0521, 231 Bethesda Ave., Cincinnati, OH coiled coils were staggered axially by an odd multiple of a 45267-0521. %I.: 513-558-3614;Fax: 513-558-4454. charged zone, electrostatic interactions between them could ‘The abbreviations used are: IF, intermediate filament; PCR,polymerase chain reaction; LMM, light meromyosin. provide the driving force for the assembly of higher order oli-

The purpose ofthis study was to test a long standing hypothesis regarding the forces that drive the assembly of intermediate filaments (IFs). The initial step ofIF assembly is the formation of dimeric, a-helical coiled coils. On the outside of the coiled coils, charged amino acids are distributed periodically such that positively and negatively charged residues are arranged in alternating zones, 9.512 residues wide (Parry et al., 1977; McLachlan and Stewart, 1982). This structural feature has given rise to the hypothesis that, if neighboring coiled coils were staggered axially by an odd multiple of a charged zone, electrostatic interactions between them could provide the driving force for the assemblyof higher order oligomers or filaments (Fraser et al., 1986; Parry and Steinert, 1992).Using the IF protein vimentin as a modelsystem, we carried out deletion mutagenesis experiments to test this hypothesis. We generated mutant vimentin proteins lacking 14,21, and 28 residues in Helix 1B ofthe rod domain,and analyzed their assembly properties by DNA transfection into IF null cells, in vitro assembly, and chemical cross-linking. Results from these experiments are consistent with, and support, the hypothesis that charge complementation plays a key role in the assembly and stabilization of intermediate filaments.

18679

18680

Charge Interactions Intermediate and Filament Assembly

gomers or filaments (Fraser et al., 1986; Parry and Steinert, 1992). Until recently, support for this hypothesis had been derived largely from computational and sequence analyses; however, the advent of site-directed mutagenesis and bacterial expression of recombinant proteins shouldallow its direct testing in vitro. Recently, two studies havebeen carried out to address similar problems regarding the periodic distribution of actinbinding sites along the tropomyosin molecule (Hitchcock-DiGregori and Varnell, 1990)and thecomplementation of charged zones along the myosin rod (Atkinson and Stewart, 1992). In both cases, deletion mutagenesis wasused to generateproteins lacking specific segments in the a-helical domains. When the bacterially expressed mutant proteins were analyzedbiochemically and biophysically, interesting information bearing on the respective hypotheses were obtained. Motivated by the structural parallelism between the IF proteins and other a-helical proteins, we used a similar strategyto ask the question of whether the formation of tetramers and higher order assembly intermediates during vimentin polymerization was mediated by charge complementation. We generated mutantvimentin proteins lacking 14,21, and28 residues in the rod domain, and analyzed their assembly properties by DNA transfection, in vitro assembly, and chemical cross-linking. Results from these experiments are consistent with the hypothesis that electrostatic interactions between charged residues on the surface of IF coiled coils provide a major driving force for one or more steps during filament assembly. They also suggest there may be significant differences between the assembly process in vivo and in vitro.

E%&. -94

I-68

m :.

,.

-45

-29

I

wt

A14 A21 A28

M

FIG.1. SDS-polyacrylamidegel electrophoresis assessment of purity of recombinant vimentincarrying deletions. wt, wild type recombinant vimentin; A14, A21, and G'8, vimentin with 14-, 21-, and 28-residuedeletionsin Helix 1B; M, sizemarkerswithmolecular weights xlO-:%marked on the right of the figure.

body, V9 (Boehringer-Mannheim), were used to monitor vimentin expression. Double-label immunofluorescence microscopy was performed, also a s described by Makarova et al. (1994). Briefly, after exposure to DNA-calcium phosphate and additional incubationfor 48 h, cells were washed in phosphate-buffered saline,fixed in methanol at -20 "C, and sequentially incubated with the appropriate primary and secondary antibodies for 1 h each. Additional washing in phosphate-buffered saline followed, and the labeledcells were observed and photographed in MATERIALSANDMETHODS a Nikon Microphot under epifluorescence illumination. Expression in E. coli and Purification ofWild Dpe and Mutant Construction of Plasmids-cDNAs encoding wild type and mutant vimentins were subcloned into PET l l d (Novagen, Madison, WI) for Vimentin-The procedure for expressing wild type and deletion mutant on the protocol supplied by Novagen. Bacexpression in Escherichia coli. For expression in cultured cells, the vimentin proteins was based same cDNAs were cloned into pRc/RSV (Invitrogen, San Diego, CA), terial clones of BL21(DE3), transformed withpETvim.wt or pETvimAX, were cultured inLB broth to anoptical density of 0.6-1.0. Isopropyl-land for sequencing, Bluescript KS(I1) was used as thevector. was then added to a final concentration of A 1.8-kilobase EcoRI fragment of murine wild type vimentin cDNA thio-/3-D-galactopyranoside 3-5 h of shaking, cells were havested (Sarria etal., 1990) (a gift of Dr. Robert Evans, University of Colorado 0.8 mM to induce expression. After Medical School) was used as the starting materialfor all plasmid con- by centrifugation, and inclusionbodies were isolated a s described (Nastructions. An NcoI site was generated first at the 5' terminus of the gai and Thorgerson,1987). Bacterial genomic DNA was fragmented by cDNA, just before the ATG start signal,by oligonucleotide-directed mu- shearing in anultrasonicator. After centrifugation and washing, puritagenesis. This changed the sequence at the beginning of the coding fied inclusion bodies were dissolved in 10 mM Tris-HCI, pH 7.5, 0.2 mM region from CCATGT to CCATGG, resulting in the conversion of the 2nd EGTA, 5 mM dithiothreitol, and 8 M urea. Wild type and mutantvimentin proteins were further purifiedby anion exchange chromatography amino acid of the vimentin from serine to alanine. To construct the bacterial expression plasmid, pETvim.wt, the fragment in pBluescript on DEAF,-Sepharose CL-GB in 10 mM Tris-HCI, pH 7.5,6 M urea with a was cut withNcoI and BamHI and inserted into the corresponding sitelinear salt gradient of 0-300 mM. Peak fractions were lyophilized and stored a t -20 "C. Typically, 20 mg of vimentin were obtained from 100 of PET lld. ml of culture cells. Protein content was determinedby the Bradford or The same fragment, with the added NcoI restriction site, was also used as templatefor production of deletion constructs,which was car- the BCA assay (PierceChemical Co.). Purity of the recombinant protein ried out using the recombinant PCR protocol of Turner and Moyer preparations was assessed by SDS-polyacrylamide gel electrophoresis (Fig. 1)and immunoblotting, using the antibodies described above. The (1992). Briefly, oligonucleotide primers were designed such that the first round of PCR produced two overlapping fragments that, when mutant vimentins with 14-, 21-, and 28-residue deletions were desigannealed, constituted a heteroduplex containing the desired deletion. nated A VA14, VA21 and VA28, respectively. second round of PCR, with it as template and T3 and T7 primers, then In Vitro Filament Assembly and Electron Microscopy-In vitro assembly of filaments was carried out by dialysis in 4-mm flat width amplified the mutant fragment. Following PCR, the productswere either subcloned into the NcoVBamHI site of PET l l d for expression in Spectropor tubing as described by Ip et al. (1985a), with the following E. coli or into the HindIIVXbaI site of pRC/RSV vector for transfection modifications. Recombinant vimentin in 8 M urea, 5 mM dithiothreitol, of eukaryotic cells. For sequencing, the PCR products were digested 10 mM Tris-HCI, pH 8.0, was dialyzed stepwise against the same soluwith Hind111 and BamHI and subcloned into BluescriptKS(I1).Deletion tion containing first 4 M urea, then 2M urea, andfinally a solution of 10 mutants were designated a s pETvimAX or pRSVvimAX, where X de- mM Tris-HCI, pH 7.0. The resulting solution contained tetramericprotofilaments. To assemble IFs, protofilaments at a concentration of 50notes the number of amino acids deleted. Thus, pETvimAl4 had a 14-residue deletion (amino acids 153-166), pETvimA2l carried a dele- 200 pg/ml were first cleared of aggregates by centrifugation in an Airtion from amino acid 153 to 173, and pETvimA28 was deleted from fuge (105,000 x g for 10 min) and then dialyzed overnight a t 4 "C amino acid153to 180. The corresponding vectors for transfection stud- against assembly buffer (170 mM NaCI, 1 mM MgCl,, 10 mM Tris-HCI, pH 7.0). After dialysis, samples were negatively stained with 1% uranyl ies were pRSVvimAl4, pRSVvimA21, and pRSVvimA28. DNA Zkansfection a n d Immunofluoresent Labeling-Monolayer cul- acetate and analyzed by electron microscopy. Among the criteria used to of 10-12-nm-wide tures of the human adrenal carcinoma cell line, SW13 clone 2 (a gift determine if assembly was normal were the presence from Dr. Robert Evans, Universityof Colorado Medical School), and the filaments, uniformity of filament caliber, whether the filament walls fibrosarcoma cell line, HTD114 (a gift from Dr. Peter Stambrook, Uni- were smooth or rough, whether any surface detailscould be discerned, of the filaments.Decisions were primarversity of Cincinnati College of Medicine), were transfected with wild and, finally, the average length type andmutantvimentins cloned into pRc/RSV a s described ily madeby visual comparison with IFs reconstitutedfrom convention(Makarova etal., 1994). Arodent-specific anti-vimentin antiserum (also ally purified and recombinant vimentin; no attempts were made to supplied by Dr. Evans) and an anti-human vimentinmonoclonal anti- quantify anyof the parameters mentionedabove.

Charge Interactions and Intermediate Filament Assembly a.

-

s .. . .~*
rnctions hf.t\r.wn apolar residues at thr first and fourth I ;I and d I positions o f the, heptad repeat. Specific ionicintctrnctionx b t ~ t \ v w nopposittbly charged amino acids at thefifth and scvvnth 1 1 , a n d gr p o . q i t i o n 5 of the heptad stahilize thr dimeric .structure. ;Ind spc-cif\. t h t b orientationandstagger of t h r t\vo polyp(~ptidt~chain.