doi:10.1016/j.jmb.2010.06.011 J. Mol. Biol. (2010) 401, 843–853

Available online at www.sciencedirect.com

The Structure of the FnIII Tandem A77-A78 Pointsto a Periodically Conserved Architecture in theMyosin-Binding Region of Titin

Rainer M. Bucher1, Dmitri I. Svergun2⁎, Claudia Muhle-Goll3

and Olga Mayans1,4⁎

1Division of Structural Biology,Biozentrum, University of Basel,Klingelbergstr. 70, CH-4056Basel, Switzerland2European Molecular BiologyLaboratory, HamburgOutstation, c/o DeutschesElektronen Synchrotron,Notkestrasse 85, D-22603Hamburg, Germany3European Molecular BiologyLaboratory, Meyerhofstrasse 1,69117 Heidelberg, Germany4School of Biological Sciences,University of Liverpool, CrownStreet, Liverpool L69 7ZB, UK

Received 8 February 2010;received in revised form1 June 2010;accepted 5 June 2010Available online11 June 2010

*Corresponding authors.O.MayansStreet, Liverpool L69 7ZB, UK. E-maPresent address: C. Muhle-Goll, In

Box 3640, 76021 Karlsruhe, GermanAbbreviations used: FnIII, fibrone

MM, molecular mass; RB, rigid bod

0022-2836/$ - see front matter. Crown C

Titin is a large intrasarcomeric protein that, among its many roles in muscle,is thought to modulate the in vivo assembly of the myosin motor filament.This is achieved through the molecular template properties of its A-bandregion, which is composed of fibronectin type III (FnIII) and immunoglob-ulin (Ig) domains organized into characteristic 7-domain (D-zone) and 11-domain (C-zone) superrepeats. Currently, there is little knowledge on thestructural details of this region of titin. Here we report the conformationalcharacterization of three FnIII tandems, A77-A78, A80-A82, and A84-A86,which are components of the representative fourth C-zone superrepeat. Thestructure of A77-A78 has been elucidated by X-ray crystallography to 1.65 Åresolution, while low-resolution models of A80-A82 and A84-A86 havebeen calculated using small-angle X-ray scattering. A77-A78 adopts anextended “up–down” domain arrangement, where domains are connectedby a hydrophilic three-residue linker sequence. The linker is embedded in arich network of polar contacts at the domain interface that results in a stiffmolecular conformation. The models of A80-A82 and A84-A86, whichcontain hydrophobic six-residue-long interdomain linkers, equally showedelongated molecular shapes, but with slightly coiled or zigzagged confor-mations. Small-angle X-ray scattering data further suggested that the longlinkers do not result in a noticeable increase in molecular flexibility but leadto semibent domain arrangements. Our findings indicate that the structuralcharacteristics of FnIII tandems from A-band titin contrast markedly withthose of poly-Ig tandems from the elastic I-band, which exhibit domaininterfaces depleted of interactions and compliant conformations. Further-more, the analysis of sequence conservation in FnIII domains from A-bandtitin points to the existence of conformationally defined interfaces at specificsuperrepeat positions, possibly leading to a periodic and locally orderedarchitecture supporting the molecular scaffold properties of this regionof titin.

Crown Copyright © 2010 Published by Elsevier Ltd. All rights reserved.

Keywords: A-band titin; FnIII tandem; X-ray protein crystallography; small-angle X-ray scattering (SAXS); sequence motif conservation

Edited by G. Schulz

is to be contacted at the School of Biological Sciences, University of Liverpool, Crownil addresses: Svergun@EMBL-Hamburg.de; Olga.Mayans@liv.ac.uk.stitut für Biologische Grenzflächen (IGB-2), Karlsruhe Institute of Technology, POy.ctin type III; MyBP-C, myosin-binding protein C; SAXS, small-angle X-ray scattering;y; MPD, 2-methyl-2,4-pentanediol; NCS, noncrystallographic symmetry.

opyright © 2010 Published by Elsevier Ltd. All rights reserved.

844 Structure of A77-A78 from Titin

Introduction

Titin is a sarcomeric component of vertebratestriated muscle that plays pivotal roles in theorganization of the actomyosin contractile appara-tus, as well as in the development, remodeling, andelasticity of myofibrils (recently reviewed byKontrogianni-Konstantopoulos et al.1). Titin extendsfrom the Z-disc to the M-line, spanning over 2 μmin situ. It is composed of N33,000 amino acids thatfold into a total of 195 immunoglobulin (Ig) and 132fibronectin type III (FnIII) domains, togetheramounting to over 90% of its mass.2 The Ig andFnIII domains are linked in tandem followingpatterns that correlate with the diverse functionalregions of the sarcomere. Tandems composedexclusively of Ig domains are found in the elastic I-band region, while the myosin-binding A-bandsegment comprises conserved superrepeats of Igand FnIII modules.3 Two superrepeat types are en-countered (Fig. 1a): (i) a short grouping of 7 domainsfollowing the pattern [Ig-(FnIII)2-Ig-(FnIII)3] occur-ring in the D-zone, at the edge of the A-band andcolocalizing with the tip of the myosin filament;and (ii) a long 11-domain repeat composed of [Ig(FnIII)2-Ig-(FnIII)3-Ig-(FnIII)3] that forms the C-zoneof the A-band extending up to the edge of the barezone. The D-zone contains six copies of the shortsuperrepeat, while the C-zone has 11 long super-

Fig. 1. Domain organization of A-band titin. (a) Organizatitin. FnIII modules are shown in orange, and Ig domains are sDomains with previously known atomic structure are highligdividing lines. Domain pairs in superrepeat positions equivalsuperrepeat under study in this work is underlined. (b) Compothe occurrence of short and long interdomain linkers is indicatthose subjected to recent characterization4 are shown in gray.

repeats. D-zone and C-zone repeats appear to haveresulted from gene duplication events, as suggestedby the fact that domains in equivalent positions indifferent repeats are more similar to each other thanto other domains within the same repeat.5

A-band titin is an integral component of the thickmyosin-based filament, part of the sliding motorsthat drive muscle contraction. Each thick filamentcontains a core of myosin molecules assembledthrough their α-helical coiled-coil tails, so that thelatter form the filament backbone while the headdomains remain accessible on the surface. Thefilamentous assembly of myosin follows a helicalarrangement with a pitch of ∼43 nm.6,7 This valueroughly coincides with the predicted contour lengthof the 11-domain superrepeat of C-zone titin (theapproximate Ig and FnIII domain length is 4.0–4.2 nm). The myosin pitch also agrees with theregular spacing at which myosin-binding protein C(MyBP-C) is found to be associated with the thickfilament.8 Such colocalization points to the in vivointeraction ofmyosin,MyBP-C, and titin. Effectively,evidence of complexation has been obtained throughdiverse binding studies.8–12 With recombinant sam-ples, MyBP-C has been shown to bind the first Igdomain in the long superrepeats of titin.12 The inter-action of titin and myosin has been proven usingboth native titin samples10,11 and short recombinantFnIII tandems,13 but identification of the interacting

tion of superrepeats in the D-zone and C-zone of A-bandhown in white. The titin kinase domain is shown in black.hted in yellow. Superrepeat boundaries are indicated byent to those of A77-A78 are indicated by blue circles. Thesition of the C-zone 11-domain superrepeat of titin, whereed. Constructs under study here are shown in black, while

Table 1. X-ray data statistics and model refinementparameters

X-ray source Elliott GX20Detector Mar345-IPWavelength (Å) 1.5405Space group C2221Unit cell parameters (Å) 115.9, 163.2, 65.2Resolution 20–1.65 (1.68–1.65)Unique reflections 74,213 (3775)Rsym (I) (%) 7.1 (36.8)I/σ(I) 11.9 (3.6)Multiplicity 3.90 (3.69)Completeness (%) 99.6 (99.5)R-factor/Rfree (%)a 17.9/20.8Number of protein residues 401Number of small ligands 3× MPDNumber of solvent atoms 695rmsd bond/angle (°) 0.006/1.019Ramachandran plot

Core/disallowed (%) 90.2/0a The Rfree set comprises 978 reflections corresponding to 1.32%

of the total data.

845Structure of A77-A78 from Titin

region in myosin remains controversial. Earlyreports that used extracted titin indicated that theinteraction is mediated by the tail region of myosin(light meromyosin).10,11 However, a later study13

that used recombinant titin fragments revealedbinding to myosin subfragment 1 containing thehead domain. Interaction and colocalization data,together with a presence in early myogenesis beforemuscle fibers acquire their striated nature,14 have ledto the hypothesis that titin as a molecular templateinvolved in the assembly of the myosin-basedfilament.15

Despite its central role in muscle, the A-bandregion of titin remains poorly characterized. Atomicstructures are restricted to the FnIII domain A71determined in isolation,16 while the first conforma-tional arrangements of its domain tandems havebeen calculated from small-angle X-ray scattering(SAXS) data only recently.4 This contrasts withnotable structural advances on the Z-disc, I-band,and M-line regions of titin. Multidomain structuresnow available for these regions include the Ig–Igduet Z1Z2 at the N-terminus of the filament,17,18poly-Ig I65-I70 from the elastic I-band,19 and the Ig–Ig–FnIII array A168-A170 directly preceding thekinase domain in the M-line.20 In the current work,we investigate further the structural features ofrepresentative FnIII tandems from A-band titin. Forthis, we have selected components of the fourth longsuperrepeat that have been shown to interact withmyosin and share a high sequence similarity withtheir counterparts in C-zone superrepeats,13 thusbeing good representatives of the myosin-bindingregion of titin. We have elucidated the crystal struc-ture of the FnIII pair A77-A78, spanning the secondand third positions of the superrepeat (Fig. 1a), andhave validated it against SAXS data in solution.SAXS is also used here to characterize the molecularconformation of the two remaining FnIII tandemcomponents of that same superrepeat A80-A82 andA84-A86. The FnIII tandems studied here arecomplementary to the A-band fragments recentlydescribed,4 which primarily explored Ig–FnIIIarrangements in the superrepeat (Fig. 1b). Structuraldata here obtained are interpreted in the light ofsequence conservation patterns across A-band mo-dules, and conclusions on titin A-band architectureare drawn.

Results

Crystal structure of A77-A78

The crystal structure of domain pair A77-A78 hasbeen elucidated to 1.65 Å resolution (statistics fordiffraction data and structural model are given inTable 1). A77-A78 consists of two FnIII modulesthat share a 38% sequence identity and a highstructural similarity (rmsd of 0.69 Å; calculatedusing PyMOL21). Characteristics typical of the cano-nical FnIII fold22,23 are present in both modules,

namely (i) an N-terminal PxPP motif that defines theconformation of the residue preceding the firstproline in the interdomain linker; (ii) a buriedtryptophan residue involved in N-terminal hydro-phobic core packing (W26 andW123 inA77 andA78,respectively); and (iii) a tyrosine residue in β-strandF (Y75 and Y171 in A77 and A78, respectively) thatforms the classical “tyrosine corner”24 and whosehydroxyl group binds a water molecule as observedin other FnIII domains (e.g., the cytoplasmic tail ofintegrin25). Domains A77 and A78 share a conservedhydrophobic core with A71 (NMR ensemble)16 andA170 (rmsd of 0.9 Å),20 the only other FnIII domainsof titin with known structure.The crystal structure of A77-A78 exhibits a semi-

extended conformation with a distinct “up–down”domain arrangement defined by a torsional angle of∼176°, an opening angle of 157°, and a distance of∼4.1 nm between the modular centers of mass(calculated as in Marino et al.17) (Fig. 2a). Thisdomain arrangement is displayed by both molecularcopies in the asymmetric unit of this crystal form(rmsd of 0.62 Å; calculated using PyMOL21), sug-gesting a stable conformation. Domains A77 andA78 are connected by a three-residue hydrophiliclinker S100-E101-R102 that is embedded in a richnetwork of polar contacts at the modular interface(Fig. 2b, Table 2). Accordingly, linker residues havelow solvent accessibility (S100 b1%, E101 ∼30%,and R102 ∼41%; calculated using the PISA server26)and are restricted in conformation, being hinderedfrom acting as a free mechanical hinge. Directinteractions between domains A77 and A78 arealso present (Table 2), involving the β-turn EF inA77 and the BC loop in A78 as well as β-strand G″ inA77 and β-hairpin FG in A78. These regionscontribute specific contacts (e.g., the salt bridgesE71→H128 and K98→E180), as well as severalglycine residues (G72 in A77; G130 and G131 inA78), to the interface (Fig. 2b). Glycine residues atthese positions are highly conserved in sequence

Fig. 2. Crystal structure of A77-A78. (a) Overall molecular conformation. Linker residues are displayed, and β-strandsare labeled. (b) Stereo view of the domain interface. Specific interactions are displayed as red broken lines. Connectingresidues are highlighted in green; these include residue A99, which is the last integral group of β-strand G and thereforedefines the end of domain A77, as well as the linker sequence S100-E101-R102.

846 Structure of A77-A78 from Titin

(Fig. S1), probably helping to prevent steric clashesbetween modules that would result in an alteredoverall conformation in such tight interface. Notice-ably, hydrophobic contacts are not present in theinterface. In summary, the multiple specific interac-tions in the interdomain region of A77-A78 point to arigid and well-defined conformation of this tandem.

Conservation of FnIII–FnIII interfaces in titin

To evaluate the possible conservation of FnIIItandem architecture in A-band titin, we investigated

the variation in interface groups (as observed inA77-A78) across all A-band domains. For this, thesequences of D-zone and C-zone FnIII moduleswere aligned according to their position in the super-repeat (Fig. S1). General conclusions on the conser-vation of domain characteristics (core and surface)were the same as those drawn from previousanalyses.13,27 The current study focuses on inter-domain features, namely linker length and compo-sition, and on residues involved in specific interfacialcontacts (i.e., side-chain–side-chain interactions).The analysis showed that none of the groups shaping

Table 2. Modular interactions in A77-A78

Domain A77 Linker Domain A78 Distance (Å)

K19 [N] E101 [OE1] 3.03E71 [OE1] H128 [NE2] 3.40E71 [OE2] H128 [NE2] 2.78K98 [NZ] E180 [OE2] 2.74K98 [O] K181 [NZ] 2.91T18 [O] S100 [N] 3.40

S100 [OG] S132 [OG] 2.74S100 [OG] K181 [NZ] 2.81S100 [O] K181 [NZ] 3.39E101 [N] G130 [O] 2.83E101 [O] N179 [ND2] 2.96

R102 [NH1] D185 [OD1] 3.15R102 [NH1] I183 [O] 2.95R102 [NH2] D185 [OD1] 3.19

Specific interactions between side chains are indicated inboldface.

847Structure of A77-A78 from Titin

the modular conformation of A77-A78 is strictlyconserved across A-band domains and that, in fact,notable variations in interface elements occur.However, substantial conservation of such interdo-main features is found in domains at equivalentsuperrepeat positions (Fig. 3). Applying this posi-tional criteria, we observed that interface features arewell conserved in C-zone domains belonging to thecentral (third to seventh) superrepeats, while D-zonedomains and the 11th domain of each long C-zonesuperrepeat are the most variable.Furthermore, the analysis revealed that modular

interactions are mediated by residues in the variablepositions of three otherwise highly conservedmotifs:(i) xPx[P]P at the N-terminus of the FnIII fold;(ii) Pxx[D]GGx in the BC loop; and (iii) AxNxxG inβ-hairpin FG (Fig. S1). These elements form a struc-turally conserved loop cluster at the N-terminal endof the FnIII fold, resembling equivalent motifs in therelated Ig domains of titin.28 Specifically, the highlyconserved asparagine residue in the AxNxxG se-quence acts as a central anchor, joining β-hairpin FGto the BC loop and linker sequence. This preservesthe architecture of the interface loops, preventing aloose arrangement. Variable residues in these motifsare, however, conserved across domains in equiva-lent positions in the superrepeat (Fig. 3a). This isparticularly evident in β-turns FG and EF. Inparticular, A77-A78 share detectable conservationwith C-zone tandems in equivalent superrepeatpositions and thus can be expected to be a goodrepresentative of this group. Also noteworthy is theconservation pattern of linker length and composi-tion (Fig. 3b). All D-zone and C-zone linkers are sixresidues long, except for those between domains 2and 3 and domains 6 and 7 that are shorter(predictably three and two residues in length,respectively). Short linkers are largely hydrophilic,while the longer ones incorporate several hydropho-bic residues, mostly bulky aromatics and prolinegroups that can be expected to restrict modulardynamics. It is thus possible that the increase inlinker length not necessarily translates into anincrease in modular flexibility. Deviations in the

linker length pattern only occur in the first and last C-zone superrepeats. This might reflect the molecularruler role of this region of titin, where regularspacing of attachment sites is crucial for establishinga functional ultrastructure. In brief, the analysis ofsequence conservation patterns across the FnIIIdomains of titin points to a periodic architecture ofthe A-band region of this filament.

Conformation of FnIII tandems in solutionby SAXS

In order to establish whether the crystal structureof A77-A78 is representative of the molecule insolution and to explore the effect of interfacesequence variation on conformation, we studiedthe three FnIII tandems of the fourth C-zonesuperrepeat—A77-A78, A80-A82, and A84-A86—using SAXS. Experimental SAXS patterns (Fig. 4a)and overall molecular parameters (Table 3) aregiven. Molecular mass (MM) estimates indicatedthat all samples are predominantly monomeric insolution, and the experimental radii of gyration Rgand the maximum sizes Dmax pointed to elongatedmolecular shapes for all constructs. This was furtherconfirmed by the pair distance distribution functionp(r) that defines the frequency of distances r joiningtwo volume elements in the molecule. The p(r)functions of each titin fragment displayed a skewedappearance characteristic of extended particles(Fig. 4b). SAXS data also indicated that themolecular flexibility of these constructs was low.This was estimated by computing Vp, the hydratedvolume of a particle in solution. For a globularprotein, Vp (in Å3) is about 1.5–2 times the MM(in Da), but this ratio will noticeably exceed 2 if theprotein adopts different conformations in solution.29

The Vp values for all FnIII tandems were 1.6–1.9times the corresponding MM value, indicating littleconformational variability in their structures andsupporting the notion that the hydrophobic compo-sition of the longer interdomain linkers preventssignificant modular flexibility.The interpretation of SAXS data in terms of low-

resolution molecular shapes used ab initio calcula-tions in DAMMIN.30 In agreement with the dataabove, the most typical ab initio models (Fig. 4c)displayed extended structures accounting for twodomains in A77-A78 and three domains in bothA80-A82 and A84-A86. In the case of A77-A78, thescattering pattern computed from the crystallo-graphic model yielded an excellent fit to the mea-sured data (Fig. 4a, Table 3), and the atomic modelneatly superposed the ab initio SAXS shape (Fig. 4c).This agreement indicates that the crystal structure ofA77-A78 is a close representative of the sampleconformation in solution. The overall molecularshapes of A80-A82 and A84-A86 suggested possiblyzigzagged or slightly coiled conformations. In A80-A82, two of the domains adopt an extended con-formation, while the remaining domain is kinkedinto a semibent arrangement. A84-A86 displays amore regular shape, where all domain pairs follow

Fig. 3. Conservation of interfacial groups in FnIII modules from titin. (a) Residues in the variable positions (x) ofβ-turns EF and FG and the Pxx[D]GGx motif of the BC loop are displayed, with each line corresponding to a successivesuperrepeat (numbered). This figure displays selected interface motifs and is an extract from the complete sequencealignments provided in Fig. S1 of Supplementary Material. The conservation of the interfacial groups of A77-A78 acrossdomains in equivalent superrepeat positions is shown in red. Conserved motifs according to domain position arehighlighted in boldface, with those groups showing a particularly high sequence identity boxed. Sequence conservation incentral superrepeats is particularly noticeable. (b) Alignment of linker sequences, where linker residues are highlighted inboldface. Linkers have been defined here by an analysis of sequence conservation (Fig. S1), but with application of insightderived from the crystal structures of A77 and A78 to define domain boundaries, where the beginning of a domain isconsidered as the first proline in the N-terminal motif PxPP and where the end of β-strand G in the previous domain isattributed to a conserved small residue (often an alanine). Conserved groups are shown in red, with sequence identityindicated by an asterisk and with close conservation indicated by a colon. (a and b) The superrepeat under study in thiswork is boxed in green.

848 Structure of A77-A78 from Titin

semibent conformations. The differences betweenboth samples are also evident when comparing theirp(r) functions, with A84-A86 displaying a more pro-nounced secondary maximum at ∼48 Å (Fig. 4b).

This value corresponds to the average center-to-center distance between subsequent domains andindicates that the domains in A84-A86 are moreregularly spaced than those in A80-A82. This

Fig. 4. Experimental SAXS data and calculated models. (a) Experimental scattering is displayed as dots with errorbars, while the curves computed from the crystallographic (A77-A78) and RB (A80-A82 and A84-A86) models are given ascontinuous lines. Curves are displaced along the ordinate axis to aid visualization. (b) Distance distribution functions. Themaximum of the p(r) function represents the diameter of the cross section of a particle. The value recorded here,approximately 20 Å, matches the dimension of the short axis of an individual FnIII domain. To ease comparison, wescaled curves to the same maximum. (c) SAXS models. Ab initio models are shown as transparent molecular envelopes.The molecular cartoon superposed on the envelope of A77-A78 is the crystallographic model, while the best three RBmodels are shown for A80-A82 and A84-A86.

849Structure of A77-A78 from Titin

Table 3. Overall molecular parameters of FnIII tandemsfrom SAXS data

Sample A77-A78 A80-A82 A84-A86

MM (kDa) 23±2 33±3 31±3MMcalc (kDa) 21.7 32.7 31.5Rg (Å) 26±1 38±2 40±2Rgmod (Å) 25.6 37.2 37.5

Dmax (Å) 90±6 130±10 130±10Vp (×103 Å3) 35±4 60±7 60±7χab initio 1.01 1.02 0.98χmod 1.03 1.25 1.18

MM, Rg, Dmax, and Vp denote molecular mass, radius of gyration,maximum size, and hydrated volume, respectively. The para-meters without superscripts are experimental values obtainedfrom scattering data. MMcalc is the theoretical MM of constructscomputed from the primary sequence. χ is the discrepancybetween the experimental curve and those computed frommodels. Superscript “ab initio” refers to molecular shapes cal-culated ab initio. Superscript “mod” denotes the crystallographicmodel for A77-A78 and the models obtained by RB refinementagainst SAXS data for A80-A82 and A84-A86.

850 Structure of A77-A78 from Titin

observation is consistent with sequence data (see thetext above) that indicate similar linker lengths inA84-A86 (housing two six-residue linkers), but notin A80-A82 (containing a six-residue linker and atwo-residue linker) (Fig. 3b). In addition, the markedsecondary maximum in the p(r) function of A84-A86further supports a low dynamics of long interdo-main linkers in this construct, since modularmotions would result in a decrease or smearing ofthis feature.A further insight into the conformation of these

tandems was sought through the rigid body (RB)fitting of the atomic homology models of individualdomains to SAXS data. In the following, it shouldbe borne in mind that the FnIII fold has the overallshape of a prolate ellipsoid of rotation, with anoticeable long axis but a rather isometric crosssection. As a result, rotations around the longmodular axis in this study yielded little change inoverall molecular shape and were poorly distin-guishable by SAXS. Thus, domain positions in theresulting RB models are well defined in thedirection of the main axis, but poorly resolved inazimuthal orientations. The models calculated forboth constructs yielded good fits to the experimen-tal data within the available resolution (approxi-mately 15 Å) (Fig. 4a, Table 3) and were also inagreement with ab initio molecular shapes (Fig. 4c).Inspection of the models indicated that the inter-domain hinge angles in A84-A86 are similarly bentin the two interfaces, while one interface wasconsistently more extended than the other in A80-A82. The exact domains contributing to eachinterface in A80-A82 could not be resolved hereby fitting to the SAXS data because of the highconservation shared by the FnIII components.However, we speculate that both short linkers indomain pairs A77-A78 and A81-A82 result insimilarly extended arrangements, while the longerlinkers might allow for the more bent conforma-tions of A80-A81, A84-A85, and A85-A86.

Discussion

Titin, despite its gigantic dimensions, has a simpleand repetitive architecture consisting of seriallylinked Ig and FnIII modules. Whereas domains actas rigid building blocks, modular connectionsdictate the conformational dynamics of the chainand, thereby, its mechanical and cellular scaffoldingproperties. The current study addresses FnIII–FnIIIinterfaces in the A-band of the filament. The findingssuggest that these interfaces are conformationallywell defined, exhibit limited dynamics, and arelargely conserved. Conservation patterns (Fig. 3)point to the existence of distinct interfaces at specificsuperrepeat positions. Considered in broad terms,two main types of domain arrangements can beinferred from the characteristics of linker sequencesand SAXS data: (i) those defined by short (two orthree residues in length) linkers of hydrophilicnature that result in extended domain arrange-ments, and (ii) those containing longer six-residuelinkers rich in aromatic and/or proline residues thatappear to adopt semibent conformations. It has beenpreviously proposed that the long-range geometryderived from the relative arrangement of domainswithin the chain is an important mechanism of pro-tein recognition and productive binding by titin.17

Here linker length (i.e., intermodular distance) anddomain orientation might work together in achiev-ing a suitable distribution of binding sites to supportthe interaction of titin with myosin and MyBP-C inthe thick filament. A periodic and locally orderedinterdomain architecture in A-band titin is wellsuited for supporting the molecular template func-tion of this region of the filament.The characterization of FnIII tandems in this work

complements a recent study of FnIII–Ig fragmentsfrom the long superrepeat of A-band titin4 (Fig. 1b).Contrary to that work, sample dimerization was notdetected here. MM values calculated by SAXS (atechnique that is highly sensitive to the presence ofoligomeric species) (Table 3) were in excellent agree-ment with those derived from monomer sequences,and SAXS data could be modeled in terms of mono-meric particles up to the maximal concentrationsmeasured (8–14 mg/ml). Furthermore, the crystalstructure of A77-A78 appears to be monomeric, andmolecular pairs of possible physiological relevancecould not be identified in the crystalline lattice(Fig. S2). Thus, further work is required to establishwhether dimerization is a general property of titin'sA-band components and whether in vitro dimericformations represent physiological states.Atomic structures of Ig–Ig17–20 and Ig–FnIII20

interfaces in titin were previously available. Acomparative analysis shows that the interdomainparameters (i.e., hinge opening and torsion) of theFnIII tandem A77-A78 are well in agreement withthose calculated for Ig–Ig pairs of titin withcomparable linker length.19 As a result, the samesecondary structure elements influence the inter-face in Ig–Ig and FnIII–FnIII tandems, where theNxxG motif of β-turn FG plays a particularly

851Structure of A77-A78 from Titin

central role. However, Ig–Ig interfaces are light, aredepleted of interactions, and exhibit certain flexi-bility and compliance,17,31–33 as expected from therole of Ig tandems in chain elasticity. Ig–FnIIIinterfaces show a small set of interactions andappear structurally stiffer.20 However, compared toIg–Ig and Ig–FnIII interfaces, the FnIII–FnIII pair-ing of A77-A78 shows the most richly networked—and presumably the stiffest—domain interface. Stiffconformation should be understood here as thatwhich does not exhibit interdomain motions oflarge amplitude (i.e., modular movements, ifpresent, are small compared to the size of thedomains). In A77-A78, experimental support forrestricted conformational dynamics is provided bythe calculation of its Vp value and by the fact thatthe scattering computed from the crystal structureneatly fits the experimental SAXS pattern. Insummary, modular junctions in titin present abroad spectrum of dynamic properties that supportthe different local sarcomeric roles of the filamentin mechanics and scaffolding. In consequence, thehistorical view of titin as a chain of homogeneous“beads on a string” cannot be further held.

Methods

Protein production

Expression clones for A77-A78, A80-A82, and A84-A86from human titin have been previously described.13 Thefragments correspond to residues 22,877–23,070, 23,166–23,466, and 23,562–23,866, respectively (European Molec-ular Biology Laboratory access code X90568). For simplic-ity, a residue numbering of 1–195 will be used throughoutthis text to refer to A77-A78.Protein expression used cultures of Escherichia coli BL21

(DE3) Rosetta grown at 37 °C up to an OD600 of 0.6 inLuria–Bertani medium supplemented with 25 μg/mlkanamycin and 34 μg/ml chloramphenicol. Expressionwas induced with 1 mM isopropyl-β-D-thiogalactopyr-anoside, and growth continued for 16 h at 30 °C. Bacterialpellets, harvested by centrifugation at 3400g and 4 °C,were resuspended in lysis buffer composed of 100 mMNaCl, 50 mM Tris–HCl (pH 7.2), and 2 mM DTT in thepresence of a protease inhibitor cocktail (Boehringer) andDNase. Lysis was performed by addition of lysozyme andsonication. The homogenate was clarified by centrifuga-tion and applied to a Ni2+-chelating HisTrap column (GEHealthcare) equilibrated in lysis buffer. Elution wasperformed with 200 mM imidazole. The eluate wasdialyzed against 150 mM NaCl, 50 mM Tris–HCl(pH 8.0), and 2 mM DTT in the presence of TEV proteaseat 4 °C overnight for tag removal. Subsequent purificationused reverse metal-affinity chromatography on a HisTrapcolumn, followed by gel filtration on a Superdex 75HiLoad 16/60 column (GE Healthcare) in lysis buffer.Resulting samples were stored at 4 °C until further use.

Crystal structure elucidation

Crystals of A77-A78 were grown on VDX crystallizationplates at 4 °C using the hanging-drop method. Dropsconsisted of 1.5 μl of protein solution (concentration,

∼52mg/ml) and 0.5 μl of polyethylene glycol 400 at a finalconcentration of 5% (vol/vol). Reservoir solutionscontained 5% (vol/vol) polyethylene glycol 400 and 40%(vol/vol) 2-methyl-2,4-pentanediol (MPD). While theaddition of MPD to the hanging drop resulted in clustersof thin needles, its sole presence in the reservoir wasoptimal to achieve large single crystals (dimensions inexcess of 400 μm×400 μm×400 μm).X-ray diffraction data were collected from crystals in

native mother liquor at 100 K on an Elliott GX20 rotatinganode X-ray generator equipped with an Osmic confocalsystem and a MAR345 Image Plate detector. Data wereprocessed using the XDS package34 (statistics are pre-sented in Table 1). Crystals contained two molecularcopies of A77-A78 in the asymmetric unit (VM=3.6;solvent content, 65%). The rotational component ofnoncrystallographic symmetry (NCS) was identifiedusing POLARRFN,35 which revealed a 2-fold (κ=180°)NCS axis perpendicular to c (ω =90°) and roughlycontained in the ab plane, bisecting it (φ=45°). This resultsin a quadratic lattice, where protein molecules arearranged along the sides of a hypothetical square anddefine a large solvent-filled space. In this lattice, solvent-filled square channels coalign across unit cells, perforatingthe crystal as a micropore. Phasing was performed bymolecular replacement in PHASER36 using the crystalstructure of A170 from titin20 as search model. A170shares a 38% and 34% sequence identity with A77 andA78, respectively, and accounts for one-fourth of theasymmetric unit content. Subsequent model building,refinement, and validation used ARP/wARP,37 Coot,38and PHENIX,39 with individual domains defined as TLSgroups. Automated solvent building used PHENIX. Forrefinement, diffraction data were divided into a workingset and a test set using FREERFLAG.35 The final modelconsists of twoNCS copies of A77-A78, each containing allprotein residues, and one copy containing two N-terminalresidues that are remnants of the TEV cleavage site. Modelstatistics are given in Table 1.

SAXS experiment and data analysis

Synchrotron SAXS data from A77-A78, A80-A82, andA84-A86 samples were collected in several experimentalsessions on beamline X33 at the European MolecularBiology Laboratory (Deutsches Elektronen Synchrotron,Hamburg) using a MAR345 Image Plate detector. Allsamples were measured at five solute concentrations each,at the least. The starting concentration for all samples was1.2 mg/ml, while the maximal concentrations were14 mg/ml for A77-A78, 10 mg/ml for A80-A82, and8 mg/ml for A84-A86. The scattering intensity I in therange of momentum transfer 0.01b sb0.45 Å− 1 wasrecorded (s=4πsinθ/λ, where λ=1.5 Å is the X-raywavelength, and 2θ is the scattering angle) at a sample–detector distance of 2.7 m. Radiation damage, monitored byrepetitive 2-min exposures, was negligible. Backgroundscattering was subtracted, and data were reduced, normal-ized, and extrapolated to infinite dilution using PRIMUS.40

The forward scattering I(0) and the radius of gyration Rgwere evaluated using Guinier approximation41 assumingthat, at very small angles (sb1.3/Rg), the intensity isrepresented by I(s )= I(0)exp(−(sRg)

2/3). These parameterswere also computed from the entire scattering patternsusing the indirect transform package GNOM,42 providingalso the pair distribution function of the particle p(r) and themaximum sizeDmax. The excluded volume of the hydratedparticle Vp was computed using the Porod invariant.43

852 Structure of A77-A78 from Titin

Low-resolution molecular shapes were generated abinitio using the program DAMMIN,30 which representsparticle shape by an assembly of densely packed beadsand employs simulated annealing to construct a compactinterconnected model fitting the experimental data tominimize the discrepancy:

v =

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

N − 1

Xj

I sj� �

−cIcalc sj� �

r sj� �

" #2vuut ð1Þ

where N is the number of experimental points, c is ascaling factor, and Icalc(s) and σ(sj) are the calculatedintensity and experimental error at the momentumtransfer sj, respectively. Multiple DAMMIN calculationswere performed to assess the stability of resultingsolutions. The programs DAMAVER44 and SUPCOMB45

were used to average multiple models and to select thebest representative.The scattering pattern of A77-A78 was calculated from

crystallographic coordinates usingCRYSOL.46 For A80-A82and A84-A86 modeling, homology models of individualFnIII domains were first created using MODELLER47 withA77 and A78 as templates. The scattering amplitudes of thehomology models were precomputed in CRYSOL andSASREF,48 and the models were employed to assemble theconformations of A80-A82 and A84-A86 from individualdomains using RB refinement. Starting from a randomdomain arrangement, SASREF uses simulated annealing toguide the translations and rotations of domains tominimizethe discrepancy χ (Eq. (1)) between experimental data andcalculated data while maintaining chain connectivitywithout steric clashes. Multiple SASREF runs were per-formed, yielding stable and consistent RB models for eachsample.

Accession codes

Model coordinates and diffraction data have beendeposited with the Protein Data Bank under accessioncode 3LPW.

O.M. acknowledges the support of the SwissNational Science Foundation (3100A0-100852), D.I.S.acknowledges the support of the Human FrontierScience Program (RGP0055/2006-C), and C.M.-G.acknowledges the support of Deutsche Forschungs-gemeinschaft (Mu1606/2).

Acknowledgements

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2010.06.011

References

1. Kontrogianni-Konstantopoulos,A., Ackermann,M.A.,Bowman, A. L., Yap, S. V. & Bloch, R. J. (2009). Muscle

giants: molecular scaffolds in sarcomerogenesis.Physiol. Rev. 89, 1217–1267.

2. Bang, M. L., Centner, T., Fornoff, F., Geach, A. J.,Gotthardt, M., McNabb, M. et al. (2001). The completegene sequence of titin, expression of an unusualapproximately 700-kDa titin isoform, and its interac-tion with obscurin identify a novel Z-line to I-bandlinking system. Circ. Res. 89, 1065–1072.

3. Labeit, S. & Kolmerer, B. (1995). Titins: giant proteinsin charge of muscle ultrastructure and elasticity.Science, 270, 293–296.

4. Tskhovrebova, L., Walker, M. L., Grossmann, J. G.,Khan, G. N., Baron, A. & Trinick, J. (2010). Shape andflexibility in the titin 11-domain super-repeat. J. Mol.Biol. 397, 1092–1105.

5. Higgins, D. G., Labeit, S., Gautel, M. & Gibson, T. J.(1994). The evolution of titin and related giant muscleproteins. J. Mol. Evol. 38, 395–404.

6. Huxley, H. E. (1963). Electron microscope studies onthe structure of natural and synthetic protein fila-ments from striated muscle. J. Mol. Biol. 7, 281–308.

7. Huxley,H. E. & Brown,W. (1967). The low-angle X-raydiagram of vertebrate striated muscle and its beha-viour during contraction and rigor. J. Mol. Biol. 30,383–434.

8. Labeit, S., Gautel, M., Lakey, A. & Trinick, J. (1992).Towards a molecular understanding of titin. EMBO J.11, 1711–1716.

9. Fürst, D. O., Vinkemeier, U. & Weber, K. (1992).Mammalian skeletal muscle C-protein: purificationfrom bovine muscle, binding to titin and the charac-terization of a full-length human cDNA. J. Cell Sci. 102,769–778.

10. Soteriou, A., Gamage, M. & Trinick, J. (1993). A surveyof interactions made by the giant protein titin. J. CellSci. 104, 119–123.

11. Houmeida, A., Holt, J., Tskhovrebova, L. & Trinick, J.(1995). Studies of the interaction between titin andmyosin. J. Cell Biol. 131, 1471–1481.

12. Freiburg, A. & Gautel, M. (1996). A molecular map ofthe interactions between titin and myosin-bindingprotein C. Implications for sarcomeric assemblyin familial hypertrophic cardiomyopathy. Eur. J.Biochem. 235, 317–323.

13. Muhle-Goll, C., Habeck, M., Cazorla, O., Nilges, M.,Labeit, S. & Granzier, H. (2001). Structural andfunctional studies of titin's fn3 modules revealconserved surface patterns and binding to myosinS1—a possible role in the Frank–Starling mechanismof the heart. J. Mol. Biol. 313, 431–447.

14. Schultheiss, T., Lin, Z. X., Lu, M. H., Murray, J.,Fischman, D. A., Weber, K. et al. (1990). Differentialdistribution of subsets of myofibrillar proteins incardiac nonstriated and striated myofibrils. J. Cell Biol.110, 1159–1172.

15. Whiting, A., Wardale, J. & Trinick, J. (1989). Does titinregulate the length of muscle thick filaments? J. Mol.Biol. 205, 263–268.

16. Muhle-Goll, C., Pastore, A. & Nilges, M. (1998). Thethree-dimensional structure of a type I module fromtitin: a prototype of intracellular fibronectin type IIIdomains. Structure, 6, 1291–1302.

17. Marino, M., Zou, P., Svergun, D., Garcia, P., Edlich,C., Simon, B. et al. (2006). The Ig doublet Z1Z2: amodel system for the hybrid analysis of conforma-tional dynamics in Ig tandems from titin. Structure,14, 1437–1447.

18. Zou, P., Pinotsis, N., Lange, S., Song, Y. H., Popov, A.,Mavridis, I. et al. (2006). Palindromic assembly of the

853Structure of A77-A78 from Titin

giant muscle protein titin in the sarcomeric Z-disk.Nature, 439, 229–233.

19. von Castelmur, E., Marino, M., Svergun, D. I.,Kreplak, L., Ucurum-Fotiadis, Z., Konarev, P. V.et al. (2008). A regular pattern of Ig super-motifsdefines segmental flexibility as the elastic mechanismof the titin chain. Proc. Natl Acad. Sci. USA, 105,1186–1191.

20. Mrosek, M., Labeit, D., Witt, S., Heerklotz, H., vonCastelmur, E., Labeit, S. &Mayans,O. (2007).Moleculardeterminants for the recruitment of the ubiquitin-ligaseMuRF-1 onto M-line titin. FASEB J. 21, 1383–1392.

21. DeLano, W. L. (2008). The PyMOL Molecular GraphicsSystem. DeLano Scientific LLC, Palo Alto, CA; http://www.pymol.org.

22. Main, A. L., Harvey, T. S., Baron, M., Boyd, J. &Campbell, I. D. (1992). The three-dimensional struc-ture of the tenth type III module of fibronectin: aninsight into RGD-mediated interactions. Cell, 71,671–678.

23. Campbell, I. D. & Spitzfaden, C. (1994). Buildingproteins with fibronectin type III modules. Structure,2, 333–337.

24. Hamill, S. J., Cota, E., Chothia, C. & Clarke, J. (2000).Conservation of folding and stability within a proteinfamily: the tyrosine corner as an evolutionary cul-de-sac. J. Mol. Biol. 295, 641–649.

25. de Pereda, J. M., Wiche, G. & Liddington, R. C. (1999).Crystal structure of a tandem pair of fibronectin typeIII domains from the cytoplasmic tail of integrinalpha6beta4. EMBO J. 18, 4087–4095.

26. Krissinel, E. & Henrick, K. (2007). Inference ofmacromolecular assemblies from crystalline state. J.Mol. Biol. 372, 774–797.

27. Amodeo, P., Fraternali, F., Lesk, A. M. & Pastore, A.(2001). Modularity and homology: modelling of thetitin type I modules and their interfaces. J. Mol. Biol.311, 283–296.

28. Marino, M., Svergun, D. I., Kreplak, L., Konarev, P. V.,Maco, B., Labeit, D. & Mayans, O. (2005). Poly-Igtandems from I-band titin share extended domainarrangements irrespective of the distinct features oftheir modular constituents. J. Muscle Res. Cell Motil. 26,355–365.

29. Heller, W. T. (2005). Influence of multiple well definedconformations on small-angle scattering of proteins insolution. Acta Crystallogr. Sect. D, 61, 33–44.

30. Svergun, D. I. (1999). Restoring low resolutionstructure of biological macromolecules from solutionscattering using simulated annealing. Biophys. J. 76,2879–2886.

31. Improta, S., Krueger, J. K., Gautel, M., Atkinson,R. A., Lefèvre, J. F., Moulton, S. et al. (1998). Theassembly of immunoglobulin-like modules in titin:implications for muscle elasticity. J. Mol. Biol. 284,761–777.

32. Lee, E. H., Hsin, J., Mayans, O. & Schulten, K. (2007).Secondary and tertiary structure elasticity of titin

Z1Z2 and a titin chain model. Biophys. J. 93,1719–1735.

33. Lee, E. H., Hsin, J., von Castelmur, E., Mayans, O. &Schulten, K. (2010). Tertiary and secondary structureelasticity of a six-Ig titin chain. Biophys J. 98,1085–1095.

34. Kabsch, W. (1993). Automatic processing of rotationdiffraction data from crystals of initially unknownsymmetry and cell constants. J. Appl. Crystallogr. 26,795–800.

35. CCP4. (1994). The CCP4 suite: programs for proteincrystallography. Acta Crystallogr. Sect. D, 50, 760–763.

36. McCoy, A. J., Grosse-Kunstleve, R.W., Storoni, L. C. &Read, R. J. (2005). Likelihood-enhanced fast transla-tion functions. Acta Crystallogr. Sect. D, 61, 458–464.

37. Perrakis, A., Morris, R. & Lamzin, V. S. (1999).Automated protein model building combined withiterative structure refinement. Nat. Struct. Biol. 6,458–463.

38. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallogr. Sect. D,60, 2126–2132.

39. Adams, P. D., Grosse-Kunstleve, R. W., Hung, L. W.,Ioerger, T. R., McCoy, A. J., Moriarty, N. W. et al.(2002). PHENIX: building new software for auto-mated crystallographic structure determination. ActaCrystallogr. Sect. D, 58, 1948–1954.

40. Konarev, P. V., Volkov, V. V., Sokolova, A. V., Koch,M. H. J. & Svergun, D. I. (2003). PRIMUS—aWindows-PC based system for small-angle scatteringdata analysis. J. Appl. Crystallogr. 36, 1277–1282.

41. Guinier, A. (1939). La diffraction des rayons X aux trespetits angles; application a l'etude de phenomenesultramicroscopiques. Ann. Phys. (Paris), 12, 161–237.

42. Svergun, D. I. (1992). Determination of the regula-rization parameter in indirect transform methodsusing perceptual criteria. J. Appl. Crystallogr. 25,495–503.

43. Porod, G. (1982). General theory. In Small-Angle X-rayScattering (Glatter, O. & Kratky, O., eds), pp. 17–51,Academic Press, London, UK.

44. Volkov, V. V. & Svergun, D. I. (2003). Uniqueness ofab initio shape determination in small angle scat-tering. J. Appl. Crystallogr. 36, 860–864.

45. Kozin, M. B. & Svergun, D. I. (2001). Automatedmatching of high- and low-resolution structuralmodels. J. Appl. Crystallogr. 34, 33–41.

46. Svergun, D. I., Barberato, C. & Koch, M. H. J. (1995).CRYSOL—a program to evaluate X-ray solutionscattering of biological macromolecules from atomiccoordinates. J. Appl. Crystallogr. 28, 768–773.

47. Sali, A. & Blundell, T. L. (1993). Comparative proteinmodelling by satisfaction of spatial restraints. J. Mol.Biol. 234, 779–815.

48. Petoukhov, M. V. & Svergun, D. I. (2005). Globalrigid body modelling of macromolecular complexesagainst small-angle scattering data. Biophys. J. 89,1237–1250.