Α-Actinin/Titin Interaction: a Dynamic and Mechanically Stable Cluster of Bonds in the Muscle Z-Disk

α-Actinin/titin interaction: A dynamic and mechanically stable cluster of bonds in the muscle Z-disk

Marco Grisona, Ulrich Merkela, Julius Kostanb, Kristina Djinovic-Carugob,c, and Matthias Riefa,d,1

aPhysik Department E22, Technische Universität München, 85748 Garching, Germany; bDepartment of Structural and Computational Biology, Max F. Perutz Laboratories, University of Vienna, A-1030 Vienna, Austria; cDepartment of Biochemistry, Faculty of Chemistry and Chemical Technology, University of Ljubljana, SI-1000 Ljubljana, Slovenia; and dMunich Center for Integrated Protein Science, 81377 Munich, Germany

Edited by James A. Spudich, Stanford University School of Medicine, Stanford, CA, and approved December 16, 2016 (received for review August 2, 2016) Stable anchoring of titin within the muscle Z-disk is essential for In humans, the isoforms of titin exhibit four to seven Z-repeats preserving muscle integrity during passive stretching. One of the (15, 16, 20). The structure of the EF3-4 hands complex with titin main candidates for anchoring titin in the Z-disk is the actin cross- Z-repeat 7 shows the bound Z-repeat in an α-helical confor- linker α-actinin. The calmodulin-like domain of α-actinin binds to mation (21). In solution assays, binding affinities of various the Z-repeats of titin. However, the mechanical and kinetic prop- Z-repeats to EF3-4 were determined to lie in the micromolar erties of this important interaction are still unknown. Here, we use range (22). Micromolar affinity points only to a moderately a dual-beam optical tweezers assay to study the mechanics of this stable interaction, the kinetics of which are unknown. This raises interaction at the single-molecule level. A single interaction of the question of how such a relatively weak interaction can α -actinin and titin turns out to be surprisingly weak if force is achieve the task of firmly anchoring titin within the Z-disk even applied. Depending on the direction of force application, the un- under applied mechanical loads. To answer this question, we set binding forces can more than triple. Our results suggest a model out to investigate the mechanical as well as kinetic stability of α where multiple -actinin/Z-repeat interactions cooperate to en- this interaction directly in a single-molecule mechanical experi- sure long-term stable titin anchoring while allowing the individual ment. We provide evidence that the concerted action of several components to exchange dynamically. α-actinin/Z-repeat bonds can establish long-term mechanically stable anchoring, whereas the individual bonds can break and α-actinin | titin Z-repeats | Z-disk mechanics | optical tweezers reform on the second timescale. uscle is the tissue that is constantly subjected to high me- Results Mchanical loads. Whereas thick and thin filaments are re- Interaction Between α-Actinin and Titin Z-Repeat 7 (PullA-T7 Geometry). sponsible for active force production, the passive elasticity of To probe the mechanical strength of the interaction between muscle is dominated by titin/connectin filaments (1). Hence, α-actinin and titin Z-repeat 7 (T7), we prepared a construct where under passive stretching conditions the integrity of muscle relies 23 residues of T7 (21) were fused to the C terminus of EF3-4 of the ’ on titin s being firmly anchored within the sarcomere, preventing CaMD of α-actinin 2 via a 4 × (GGS) linker. To apply load to this A the interdigitated muscle filaments from falling apart (Fig. 1 ). fusion construct we tethered dsDNA linkers to cysteine residues at Whereas titin is firmly attached to thick filaments in the A-band the termini (Fig. 1C) that could be attached to 1-μm-sized beads in – and the M-line (2 6), it is much less clear how stable anchoring is the optical tweezers. For details on the assay see Materials and achieved in the Z-disk, where adjacent sarcomeres overlap. The Methods and SI Materials and Methods. Note that in this construct superstable titin/telethonin interaction within the Z-disk was α – the force propagates through both the -actinin EF3-4 domain and considered important for titin anchoring (7 9), but knockout the T7 peptide, mimicking how force propagates in the Z-disk (Fig. mutants later showed that it is not essential for muscle integrity 1B). We therefore called this construct PullA-T7. (10–12). Apart from a direct interaction between actin filaments BIOPHYSICS AND

and titin at the Z-disk edge (13), the most prominent candidate COMPUTATIONAL BIOLOGY Significance for the anchoring of titin within the Z-disk is its interaction with α-actinin (Fig. 1B) (6, 12, 14). Four isoforms of human α-actinin have been identified: the Muscle is the tissue in our body experiencing most extreme calcium-insensitive muscle isoforms 2 and 3, which cross-link mechanical forces. The mechanism of active force generation actin filaments in sarcomere-delimiting Z-disk complexes, and has been investigated for more than 50 y and is fairly well calcium-sensitive nonmuscle isoforms 1 and 4. α-Actinin is an understood. However, despite its physiological significance, it antiparallel homodimer whose most prominent task is cross- is still unknown what mechanical linkages hold together the linking actin filaments of neighboring sarcomeres in the Z-disk muscle machinery under passive stretching forces. In this pa- B per, we show with direct mechanical single-molecule mea- (Fig. 1 ; reviewed in ref. 14). In each subunit, a flexible region α called the neck separates the actin binding domain (ABD) from surements that an array of titin/ -actinin bonds composes a four spectrin-like repeats (SR) forming the rod region (Fig. 1B dynamic network that can provide stable anchoring, main- and Fig. S1). The rod regions of the two subunits interact and taining the integrity of the muscle Z-disk even under load. This dynamic network explains how components of the Z-disk are provide a rigid spacer between the actin filaments. At the other able to rapidly rearrange and, at the same time, form a long- end of each subunit a calmodulin-like domain (CaMD) formed term stable mechanical structure. by two pairs of EF-hands (EF1-2 and EF3-4) is able to bind a – Z-disk region of titin formed by the so-called Z-repeats (15 17). Author contributions: M.G., K.D.-C., and M.R. designed research; M.G. performed re- The current model for α-actinin 2 dynamic regulation suggests search; U.M. and J.K. contributed new reagents/analytic tools; M.G. and M.R. analyzed that EF3-4 hands of one subunit bind to the neck region of the data; and M.G., K.D.-C., and M.R. wrote the paper. juxtaposed subunit, thus not being available for the interaction The authors declare no conflict of interest. with titin Z-repeats (Fig. S1) (18, 19). Upon activation of α-actinin This article is a PNAS Direct Submission. in the Z-disk by phosphatidylinositol 4,5-biphosphate (PIP2), EF3-4 1To whom correspondence should be addressed. Email: mrief@ph.tum.de. is released from the neck and binding of titin can be achieved in this This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. open conformation of α-actinin. 1073/pnas.1612681114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1612681114 PNAS | January 31, 2017 | vol. 114 | no. 5 | 1015–1020 Downloaded by guest on September 26, 2021 A D

B E

F G H

Fig. 1. Interaction between α-actinin EF3-4 and titin Z-repeat 7 in PullA-T7 geometry. (A) Schematic of the sarcomere organization under stretching conditions, showing titin elongation. (B) Arrangement of actin, titin, telethonin, and α-actinin within the Z-disk. (C) Schematic of the dual-beam optical tweezers experimental setup with a fusion construct [Protein Data Bank (PDB) ID code 1h8h]. Yellow dots mark cysteine residues, here and elsewhere. (D) Force- extension trace obtained by moving the beads apart at a constant velocity, here for the PullA-T7 fusion construct. Gray dots are full bandwidth data, and black dots are smoothed data. Dotted lines are WLC fits (SI Materials and Methods). (E) Passive-mode time trace of 5 s at an average force of 3.7 pN. Gray dots are full-bandwidth data, on which HMM analysis was performed assuming the kinetic network on the right. For clarity, smoothed data are colored, based on HMM-assigned states. (Top Right) Zoom of 1 s, where full-bandwidth data are colored and smoothed data are in black. (F) Force-dependent transition rates from UU to FU (red) and from FU to UU (green). Solid lines are fits that allow extrapolation to zero-force rates (SI Materials and Methods). (G) Same for transition rates from FB to FU (purple) and from FU to FB (green). (H) Population probability of the three states as a function of force, and fits as described in SI Materials and Methods.(I) Competition assay performed by addition of T7 in solution to evaluate the dissociation constant. A passive-mode sample trace of 5 s shows that the solution binding event (cyan) has the same force (and contour length) as the FU state but can be identified by the different lifetime.

When force is applied to PullA-T7, unbinding/unfolding We used a hidden Markov model (HMM) algorithm (23) to transitions occur already at surprisingly low forces, as shown in identify the transitions between this three-state system on the Fig. 1D. At the low pulling speed of 10 nm/s, only transitions unfiltered traces. From the dwell-time distributions we calculated between the fully folded/peptide-bound state and the unfolded/ transition rates assuming the transition network shown in Fig. 1E peptide-unbound state can be observed. This conclusion is cor- with the FU state as an obligatory on-pathway intermediate. Fig. roborated by worm-like chain (WLC) fits revealing a contour 1F shows force-dependent folding (red) and unfolding (green) length increase of 34 ± 1 (SEM) nm, which corresponds to both rates of EF3-4, and Fig. 1G shows binding (green) and unbinding detachment of the peptide (assuming that it is unstructured after (purple) rates of T7. A detailed discussion about the extraction of unbinding) as well as unfolding of the EF3-4. the unbinding rates of T7 in Fig. 1G is presented in Figs. S3 and To reveal more details of the force-induced unbinding/unfold- S4. The force dependence of the unbinding rate is weak, as can be ing kinetics, we performed a series of time traces, measuring force seen by an increase of only one order of magnitude within about fluctuations while keeping the distance between the two trap 10 pN of force. This weak force coupling is a consequence of the centers fixed (passive-mode assay with preload around 3–5pN).A transition state’s being close (2 nm) to the bound state. We find that time trace excerpt of 5 s is shown in Fig. 1E,togetherwitha the unbinding rate of T7 at zero force is 0.6 ± 0.1/s (Table 1). schematic of the expected unbinding/unfolding pathway. Close Force-dependent population probabilities (Fig. 1H) allow the inspection of the time traces revealed short dwells (green spikes; extraction of free-energy differences between the three states see also the zoom in Fig. 1E) in addition to the folded/bound using a model (fitting curve) (24) explained in SI Materials and (purple, hereafter named FB) and unfolded/unbound state (red, Methods. We find that the UU state has a free-energy difference UU). This third state is consistent with a conformation where of 0.5 kBT with respect to the FU state, and 4.7 kBT to the FB α-actinin is folded and T7 unbound (green, FU), because it also state (Table 1). shows up in a mutant without ligand, where we only pull on EF3-4 The measurements shown so far, using the PullA-T7 construct, (PullA construct; details in Fig. S2 and SI Materials and Methods). do not allow the measurement of solution binding rates or the

1016 | www.pnas.org/cgi/doi/10.1073/pnas.1612681114 Grison et al. Downloaded by guest on September 26, 2021 Table 1. Energetic and kinetic binding parameters of different target peptides Δ 0 μ F=0 = F=0 = Peptide Sequence fused to EF3-4 G , kBTKD, M kon ,1 s koff ,1 s T7 654_GKKAEAVATVVAAVDQARVREPR_676 4.7* 4.3 ± 2.1† 302 ± 17 0.64 ± 0.06 ‡ T1 422_ADKSAAVATVVAAVDMARVREPV_444 5.9* 1.3 ± 1.1 509 ± 97 0.46 ± 0.05 ‡ T3 518_GTEKAFVPKVVISAAKAKEQET_539 3.5* 14 ± 11 158 ± 13 11 ± 10 T2 472_EAEKIAVSKVVVAADKAKEQELK_494 <0.5§ > 287‡ —— T5 562_ETRKTVVPKVIVATPKVKEQDLV_584 <0.5§ > 287‡ —— ‡ Neck 259_AEQAETAANRIVKVLAVNQENERLME_284 0.7* 235 ± 160 137 ± 640± 13

Bold amino acids in the peptide sequences mark key hydrophobic binding positions of the α-helix, aligned following refs. 19, 21, and 22. Complete sequences of the constructs are given in SI Materials and Methods. The most relevant error in ΔG0 comes from the force calibration, and it is estimated to be about 10%. The zero-force extrapolated rate constants are obtained from the fits in Fig. 1G (T7), Fig. S6 (T1 and T3), and Fig. S8 (neck). The zero-force rates errors, here and elsewhere, are SD. *From the population probability vs. force plot obtained from passive-mode traces. † From the competition assay. ‡From Eq. 2. § The folding free-energy of the EF3-4 (0.5 kBT) sets a thereshold for the detection of the binding events.

dissociation constant, because the T7 peptide is tethered to the rebinding of the tethered peptide and stabilizing the folded EF3-4 domain. To this end, we used a competition assay with conformation of EF3-4. Because no force is applied to the free free T7 peptide in solution as described in ref. 25. Adding the T7 peptide, it will only dissociate at rates similar to the force-free case. peptide (650–698) to the solution now establishes a competition Indeed, from the dwell times of the cyan states, we find a zero force between the tethered and the solution peptides (Fig. 1I). Apart off-rate koff = 1.3 ± 0.5/s, within a factor of two from the extrapo- from the three states already discussed above, a new long-lived lated zero force value of the purple branch in Fig. 1G.Weesti- state (cyan) arises at a contour length identical to the green state mated the pseudo first-order on-rate of the peptide to kon = 17 ± 6/s (see the contour length vs. dwell-time scatter plot in Fig. S5B). In by summing up the total time the molecule spends in the green state this state, a peptide from solution has bound, thus blocking the between two subsequent cyan events (see also SI Materials and

AB C BIOPHYSICS AND COMPUTATIONAL BIOLOGY

Fig. 2. Interaction of α-actinin EF3-4 with different titin Z-repeats, with the same pulling geometry and the same color coding as in Fig. 1. (A–C) Interaction with T1 in A,T3inB, and T5 in C. Upper row: constant-velocity traces with fits (dashed lines) as in Fig. 1. The contour length increases for T1 and T3 are in agreement with the 34 ± 1 nm found for the PullA-T7 construct and expected from length measurements on the PDB structure. For T5, the dashed line is not a fit (forces are below resolution) but marks the expected contour length of the FB state to guide the eye. Middle row: 5-s passive-mode time traces. Similar traces in competition assay with T7 in solution are shown in Fig. S7, to confirm that the FU state can also bind T7 as in Fig. 1I. Bottom row: population probability plots of UU and FB states as a function of force. Solid lines are a global fit to the data and dotted gray lines are fits to data from the pullA-T7 construct, for comparison (Fig. 1H).

Grison et al. PNAS | January 31, 2017 | vol. 114 | no. 5 | 1017 Downloaded by guest on September 26, 2021 Methods and Fig. S5A). The dissociation constant KD, with 52 μM seen from Fig. 1E, where the bound state is populated only about of T7 was estimated to 4 ± 2 μM, using the relation half of the time. To avoid forced unfolding of EF3-4 and getting more realistic koff rebinding kinetics, we designed a new construct, PullT7, where KD = ½Psolution. [1] kon the points of force application are changed, such that force is only applied to the tethered peptide (Fig. 3A, Inset). Constant- velocity curves (Fig. 3A) already show a notable increase in the Interaction of EF3-4 with Other Target Peptides. In a second set of forces required to stretch and unbind the tethered peptide from experiments, we measured the mechanical strength of other titin the EF3-4 domain. Detectable unbinding events start at 8 pN Z-repeats with the α-actinin 2 C-terminal EF3-4 domain, fusing and reach as high as 18 pN. Also, in passive-mode experiments together the proteins in analogy with the PullA-T7 construct A (Fig. 3B) binding/unbinding forces are higher. Force-dependent (Table 1). For comparison with T7, we investigated T1 (Fig. 2 ), D T2, T3 (Fig. 2B), and T5 (Fig. 2C). We could measure binding population probabilities (Fig. 3 ) show the midpoint for the for repeats T1 and T3, but T2 and T5 showed no detectable unbinding force at 13 pN, whereas the associated free energies binding. Constant-velocity traces (top row) depict unbinding/ are similar to the ones measured in the PullA-T7 construct. unfolding forces for the T1 fusion construct similar to T7, whereas T3 forces are lower. This result is confirmed by passive- mode traces (middle row), where the lifetimes of the peptide- A bound state are much longer for T1 than for T3 (T5 data re- semble those of the PullA construct in Fig. S2, where no peptide is tethered). In analogy with Fig. 1 F–H, we analyzed both the probability (bottom row) as well as the dwell time distributions of passive-mode traces. Force-dependent kinetics were estimated kF=0 kF=0 (Fig. S6) and zero-force rates on (FU to FB) and off (FB to FU) extracted (Table 1). The general behavior of the T1 and T3 constructs under force is similar to that observed for T7, as shown in the lower row of Fig. 2. The force-probability plot is slightly shifted toward higher forces for T1, and to lower forces for T3. This is related to a 0 0 0 higher free energy ΔG = GUU − GFB for T1 and a lower for T3. B Table 1 summarizes both energetic as well as kinetic parameters of the different tethered titin Z-repeats. Because we directly measured the solution affinity of T7 in the competition assay described above, KD values for the other peptides can now directly be obtained from the free-energy values measured for the tethered peptides. We designed the constructs to have the same effective concentration of the tethered peptides, by using the same linker length of PullA-T7 (for CD details see SI Materials and Methods). For each repeat Ti, we can 0 therefore relate the measured binding/folding free energy ΔGTi Ti directly to its dissociation constant KD using ! KT7 ΔG0 − ΔG0 = D [2] Ti T7 ln Ti . KD

Finally, we investigated the interaction of EF3-4 with the neck E region of α-actinin 2 (G258–M283), which is crucial for the regulation of α-actinin during sarcomere assembly and regulation (18, 19). Not only is the neck region unstructured in absence of a substrate (18, 19), but also no interaction of the peptide lacking the neighboring domains was ever observed. Our data (Fig. S8 and Table 1) show that in the fusion construct the neck peptide binds EF3-4, albeit with very low affinity. A detailed discussion of these experiments is presented in SI Materials and Methods.

Fig. 3. Interaction between α-actinin EF3-4 and titin Z-repeat 7 in PullT7 PullT7 Geometry. Although the tethered constructs (e.g., PullA- geometry. (A) Constant-velocity trace at 10 nm/s and schematic of the pull- T7) used in the above experiments allow precise characterization ing geometry. The contour length increase obtained from the fits is 4.3 ± 0.5 of the unbinding rates, they may not precisely mimic the way (SEM) nm, in agreement with the expected stretching of the peptide helix. those molecules refold/rebind in the Z-disk. In our experiments, (B) Passive-mode trace, with the same color coding as in Fig.1. (C) Transition owing to the direct tethering of T7 and EF3-4, the latter is rates from FU to FB (green, on-rates) and from FB to FU (purple, off-rates). subjected to load even after the bond has broken, leading to From fits (solid lines), we extracted an off-rate at zero-force of 0.6 ± 0.3/s and ± rapid unfolding of the domain. In muscle, after unbinding of T7, an on-rate of 216 106/s. (D) Population probability plot. From fits (solid lines), a free energy difference between FU and FB of 5.8 k T was extracted. EF3-4 will not be under load and, hence, will not experience B (E) Competition assay with 10 μM of T7 in solution. As described for Fig. 1I,we forced unfolding (Fig. 1B). The low midpoint force we measure extracted a KD of 8 ± 4 μM from this data. Because in the absence of force the for the PullA-T7 construct (3.5 pN, Fig. 1H) results from the EF3-4 domain can still unfold (FU to UU rates in Fig. 1F), a correction of the difficulties of unfolded EF3-4 to refold under force, as can be kinetic parameters is necessary (SI Materials and Methods).

1018 | www.pnas.org/cgi/doi/10.1073/pnas.1612681114 Grison et al. Downloaded by guest on September 26, 2021 Furthermore, binding/unbinding kinetics in this geometry (Fig. to α-actinin and found measurable affinities to α-actinin 2 for T1, 3C) extrapolate to zero-force values similar to those of the T3, and T7. Consistent with those earlier results, we observed no PullA-T7 construct (Fig. 1G). A competition assay with T7 so- binding of T2 and T5, and we measured similar affinities for T1 lution peptide similar to the one of Fig. 1I yields identical so- (KD = 1 μM) and T7 (KD = 4 μM) and lower for T3 (KD = 14 μM; lution binding and unbinding kinetics (Fig. 3E). Table 1). Although Joseph et al. (22) found T1 to be a slightly weaker binder, it is rather more stable in our experiments (Fig. Discussion 2A). The differences in affinities are also reflected in the binding/ Interaction Between α-Actinin and Titin Z-Repeat 7. Generally, unbinding rates (Fig. S6). Whereas the differences in on-rates are proteins responsible for muscle mechanics and stability have small among the constructs, we observe significant difference in been shown to be very stable either mechanically or thermody- off-rates between the strongly binding peptides T1 and T7 vs. the namically. Titin Ig-domains unfold at forces around 200 pN (26, weakly binding peptide T3. Overall, the kinetic data suggest that 27) and the titin/telethonin interaction has been reported as one even the strongest Z-repeat/α-actinin bonds can last no longer of the most stable noncovalent bonds (∼800 pN) (9). Hence, it than a few seconds in the absence of force. may come as a surprise that in the PullA-T7 geometry we find that the mechanical stability of the titin Z-repeat/α-actinin in- Sarcomeric Z-Disk Model. Our single-molecule data indicate that, teraction involves protein structures that tolerate only a few for a single α-actinin/Z-repeat bond, the unbinding rates in the piconewtons of force before they detach or unfold. EF3-4 with- absence of load are faster than 2 s. How can this be reconciled out peptide bound already unfolds at around 1 pN (Fig. 1F). with a long-term stable arrangement of the Z-disk? Apparently, Even when the peptide is bound, the complex unbinds with a a single bond cannot provide enough long-term stability for the midpoint force of 3.5 pN (Fig. 1H). The relatively low mechan- α-actinin/titin arrangement. The picture, however, changes if we ical stability we find for the complex is in agreement with the take into account that, in the Z-disk, there is more than one moderate dissociation constants reported in the literature (KD = α-actinin/Z-repeat interaction for each titin molecule. Luther 0.1–0.25 μM) (18, 22). Our competition assay (Fig. 1I)yieldsa and Squire (30) have estimated, from the 19-nm α-actinin higher KD value (4 μM), which may be explained by different spacing found in electron microscopy studies, that every other experimental conditions. It is important to note that the KD Z-repeat can bind α-actinin in the Z-disk. Hence, if seven repeats values we measure for Z-repeats 1 and 3 (discussed below and are present, up to four α-actinins will bind to the same titin shown in Table 1) are also consistently higher than those reported molecule simultaneously. Under load, titin will remain anchored in the literature; however, relative stabilities agree very well in the Z-disk as long as at least one Z-repeat interacts with with literature. Our single-molecule competition experiment α-actinin at any given time. Additionally, interaction of the Zq also provides direct force-free measurements of the kinetics of region, which flanks Z-repeats, with α-actinin rod further con- the α-actinin/T7 repeat bond, and we find that the dynamics of tributes to stabilization of the titin/α-actinin interaction (17). To this bond are surprisingly high with an average lifetime of 0.7 ± assess the overall anchoring stability we therefore have to take 0.3 s (Fig. 1I). into account the avidity effect originating from those multiple From the model shown in Fig. 1B it becomes obvious that, in interactions (Fig. 4). the Z-disk, forces resulting from sarcomere length increase will Avidity (31) is a concept originally proposed for antibodies stretch titin and propagates through α-actinin, like in the probed where, in an early phase of the immune response, low-affinity PullA-T7 geometry, which we consider therefore a good model antibody fragments are combined in pentamers (IgM), thus in- for the force-dependent unbinding kinetics in muscle (FB-to-FU creasing the number of interaction sites from 2 to 10. Individual transition; Fig. 1G). Nonetheless, the midpoint force we measure interaction free energies add up owing to the increased valency. for the PullA-T7 construct (3.5 pN) is within the physiological In a similar way, the increased number of α-actinin/Z-repeat range of forces acting on a single titin molecule, estimated to be bonds will increase the overall strength of titin anchoring. For an 0–5 pN in a number of studies (27–29). The low measured forces estimate of lifetimes, it is important to note that α-actinin in in the PullA-T7 construct can be ascribed to the fact that EF3-4 muscle is also bound to actin, keeping each α-actinin close to its BIOPHYSICS AND readily unfolds after unbinding of T7, whereas this will not happen Z-repeat, even if it has transiently detached. Hence, we believe if titin and α-actinin are not tethered, as is the case in muscle. the fast on-rates we measure for our fusion constructs are good COMPUTATIONAL BIOLOGY The PullT7 construct resolves this problem because force is models for the on-rates in the muscle. Each interaction of the applied only across the Z-repeat 7 (Fig. 3). Hence, this construct three we identified in this study (T1, T3, and T7) has an off-rate better mimics the rebinding kinetics (FU-to-FB transition), be- between 0.5/s and 10/s and a rebinding rate of 150–500/s (Table cause the EF3-4 domain is not kept under force after unbinding 1). For simplicity, we assume an equal off-rate of 1/s and an on- of T7. Note that in PullT7 the unbinding kinetics are identical rate of 100/s of each interaction for the following estimate. At within errors to the kinetics of the PullA-T7 construct (Fig. 1G zero force, the probability for each interaction’s being unbound vs. Fig. 3C). Hence, force-dependent binding and unbinding kinetics measured with PullT7 are likely similar to those occur- ring in muscle. The midpoint force in PullT7 (Fig. 3D) now shifts from 3.5 to 13 pN, showing that the α-actinin/T7 bond can, indeed, resist forces considered physiologically relevant (27–29). The higher forces for this pulling direction are a consequence of the shorter difference in extension between the bound and the unbound conformations (4 nm of contour length, vs. 34 nm for the PullA-T7 construct, including unfolding of EF3-4): At equilibrium, interaction free energies can be related to the force multiplied by the elongation and the much shorter elongation in Fig. 4. Mechanical and kinetic model of titin anchoring within the Z-disk. the PullT7 construct leads to a higher midpoint unbinding force. α-Actinin spacing allows the binding of one actin cross-linker every two titin Z-repeats. The measured titin Z-repeats T1, T3, and T7 have unbinding rates Interaction with Other Titin Z-Repeats. The number of Z-repeats at zero load in the range of 0.5–10/s, whereas the binding rates are in the found in titin depends on the isoform and ranges between two range of 100–500/s. Fast rebinding is ensured by the additional anchoring of and seven across all species (20), with repeats T1 and T7 being α-actinin to the actin filaments. This picture holds up to forces of several always present. Joseph et al. (22) assessed affinities of all repeats piconewtons (physiological range), ensuring long-term stability.

Grison et al. PNAS | January 31, 2017 | vol. 114 | no. 5 | 1019 Downloaded by guest on September 26, 2021 is the ratio between on- and off-rate (i.e., 10−2). Starting from Materials and Methods two Z-repeats only, once per second the first Z-repeat de- All protein constructs were prepared using standard recombinant techniques. taches. Before the rebinding takes place, the probability that The amino acid sequences, as well as the experimental procedures, are −2 the second also detaches is 10 , meaning that on average it takes presented in SI Materials and Methods. 2 10 attempts (100 s) until both Z-repeats are contemporary de- The experiments were carried out using custom-built dual-beam optical tached and titin is released from the anchoring. With three re- tweezers, with a design similar to the one described in ref. 34, but using two peats, the average lifetime of the cluster would accordingly be steerable traps, one by means of acousto-optic deflectors (AODs) (AA Opto 4 1s× 10 = 10,000 s. This simple estimate shows that many dy- Electronic) and the other by a two-axis piezoelectric mirror tip/tilt actuator namic bonds, when acting together, can lead to a long-term stable (Mad City Labs). Constant-velocity traces were obtained by tilting the piezo anchor. Even under a mechanical load of 5 pN, the probability of mirror, and the AODs were used for the jump assays (SI Materials and finding a single Z-repeat bound (Fig. 3D), as well as the unbinding Methods), because of the higher steering speed. rate (Fig. 3C), would not change significantly, and even under The stiffness and sensitivity of both traps were calibrated using a standard those conditions the Z-disk would remain stable. The concerted method (35) with back focal plane detection. Data acquisition was carried effort of many bonds reconciles the apparently contradictory find- out at a sampling rate of 150 kHz, averaged to 30 kHz before recording. ings of a long-term stable Z-disk (half-life of titin ≈14 h) (32) and a After cross-talk correction for cross-polarization between the two traps (36), fast exchange of its individual components (half-lives of α-actinin the signal of the two traps was averaged to improve the signal-to-noise ratio. and other Z-disk proteins <1 min) (19, 33), as has been measured in various fluorescence recovery after photobleaching studies. ACKNOWLEDGMENTS. We thank Matthias Gautel, Andrea Ghisleni, and It is interesting to note that also the interaction of filamin to Euripedes de Almeida Ribeiro for helpful discussions; Maik Veelders for the von Willebrand receptor GP1b involves such dynamic bonds technical assistance; and Katarzyna Tych and Markus Jahn for comments which, in combination with a cluster of interaction sites, result in on the manuscript. This work was supported by Marie Curie Initial long-term stability (25). Dynamic bonds in combination with Training Network MUZIC Grant 238423; Austrian Science Fund Projects I525, I1593, P22276, and P19060; Federal Ministry of Economy, Family and avidity providing long-term stability may be a common motif for Youth through the initiative, “Laura Bassi Centres of Expertise” (K.D.C.); all situations where a mechanically stable connection has to be Center of Optimized Structural Studies Grant 253275 (to J.K.); the Univer- maintained while still providing reorganization, flexibility, and sity of Vienna; and Deutsche Forschungsgemeinschaft Grant FOR 1352 P8 dynamics of the components involved. (to M.R.).

1. Granzier HL, Labeit S (2004) The giant protein titin: A major player in myocardial 20. Luther PK (2009) The vertebrate muscle Z-disc: Sarcomere anchor for structure and mechanics, signaling, and disease. Circ Res 94(3):284–295. signalling. J Muscle Res Cell Motil 30(5-6):171–185. 2. Houmeida A, Holt J, Tskhovrebova L, Trinick J (1995) Studies of the interaction be- 21. Atkinson RA, et al. (2001) Ca2+-independent binding of an EF-hand domain to a tween titin and myosin. J Cell Biol 131(6 Pt 1):1471–1481. novel motif in the alpha-actinin-titin complex. Nat Struct Biol 8(10):853–857. 3. Freiburg A, Gautel M (1996) A molecular map of the interactions between titin and 22. Joseph C, et al. (2001) A structural characterization of the interactions between titin myosin-binding protein C. Implications for sarcomeric assembly in familial hypertro- Z-repeats and the alpha-actinin C-terminal domain. Biochemistry 40(16):4957–4965. phic cardiomyopathy. Eur J Biochem 235(1-2):317–323. 23. Stigler J, Rief M (2012) Hidden Markov analysis of trajectories in single-molecule 4. Muhle-Goll C, et al. (2001) Structural and functional studies of titin’s fn3 modules experiments and the effects of missed events. ChemPhysChem 13(4):1079–1086. reveal conserved surface patterns and binding to myosin S1–A possible role in the 24. Schlierf M, Berkemeier F, Rief M (2007) Direct observation of active protein folding Frank-Starling mechanism of the heart. J Mol Biol 313(2):431–447. using lock-in force spectroscopy. Biophys J 93(11):3989–3998. 5. Pernigo S, et al. (2010) Structural insight into M-band assembly and mechanics from 25. Rognoni L, Stigler J, Pelz B, Ylänne J, Rief M (2012) Dynamic force sensing of the titin-obscurin-like-1 complex. Proc Natl Acad Sci USA 107(7):2908–2913. filamin revealed in single-molecule experiments. Proc Natl Acad Sci USA 109(48): 6. Linke WA (2008) Sense and stretchability: The role of titin and titin-associated pro- 19679–19684. teins in myocardial stress-sensing and mechanical dysfunction. Cardiovasc Res 77(4): 26. Rief M, Gautel M, Oesterhelt F, Fernandez JM, Gaub HE (1997) Reversible unfolding of 637–648. individual titin immunoglobulin domains by AFM. Science 276(5315):1109–1112. 7. Gregorio CC, et al. (1998) The NH2 terminus of titin spans the Z-disc: Its interaction 27. Li H, et al. (2002) Reverse engineering of the giant muscle protein titin. Nature with a novel 19-kD ligand (T-cap) is required for sarcomeric integrity. J Cell Biol 143(4): 418(6901):998–1002. 1013–1027. 28. Watanabe K, et al. (2002) Molecular mechanics of cardiac titin’s PEVK and N2B spring 8. Pinotsis N, et al. (2006) Evidence for a dimeric assembly of two titin/telethonin com- elements. J Biol Chem 277(13):11549–11558. plexes induced by the telethonin C-terminus. J Struct Biol 155(2):239–250. 29. Anderson BR, Granzier HL (2012) Titin-based tension in the cardiac sarcomere: Mo- 9. Bertz M, Wilmanns M, Rief M (2009) The titin-telethonin complex is a directed, su- lecular origin and physiological adaptations. Prog Biophys Mol Biol 110(2-3):204–217. perstable molecular bond in the muscle Z-disk. Proc Natl Acad Sci USA 106(32): 30. Luther PK, Squire JM (2002) Muscle Z-band ultrastructure: Titin Z-repeats and Z-band 13307–13310. periodicities do not match. J Mol Biol 319(5):1157–1164. 10. Zhang R, Yang J, Zhu J, Xu X (2009) Depletion of zebrafish Tcap leads to muscular 31. Krishnamurthy VM, Estroff LA, Whitesides GM (2006) Multivalency in ligand design. dystrophy via disrupting sarcomere-membrane interaction, not sarcomere assembly. Fragment-Based Approaches in Drug Discovery, eds Jahnke W, Erlanson DA (Wiley- Hum Mol Genet 18(21):4130–4140. VCH, Weinheim, Germany), pp 11–53. 11. Ibrahim M, et al. (2013) A critical role for Telethonin in regulating t-tubule structure 32. da Silva Lopes K, Pietas A, Radke MH, Gotthardt M (2011) Titin visualization in real and function in the mammalian heart. Hum Mol Genet 22(2):372–383. time reveals an unexpected level of mobility within and between sarcomeres. J Cell 12. Gautel M (2011) The sarcomeric cytoskeleton: Who picks up the strain? Curr Opin Cell Biol 193(4):785–798. Biol 23(1):39–46. 33. Wang J, et al. (2005) Dynamics of Z-band based proteins in developing skeletal muscle 13. Linke WA, et al. (1997) Actin-titin interaction in cardiac myofibrils: Probing a physi- cells. Cell Motil Cytoskeleton 61(1):34–48. ological role. Biophys J 73(2):905–919. 34. von Hansen Y, Mehlich A, Pelz B, Rief M, Netz RR (2012) Auto- and cross-power 14. Gautel M, Djinovic-Carugo K (2016) The sarcomeric cytoskeleton: From molecules to spectral analysis of dual trap optical tweezer experiments using Bayesian inference. motion. J Exp Biol 219(Pt 2):135–145. Rev Sci Instrum 83(9):095116. 15. Gautel M, Goulding D, Bullard B, Weber K, Fürst DO (1996) The central Z-disk region 35. Tolic-Norrelykke̸ SF, et al. (2006) Calibration of optical tweezers with positional de- of titin is assembled from a novel repeat in variable copy numbers. J Cell Sci 109(Pt tection in the back focal plane. Rev Sci Instrum 77(10):103101. 11):2747–2754. 36. Atakhorrami M, Addas KM, Schmidt CF (2008) Twin optical traps for two-particle 16. Sorimachi H, et al. (1997) Tissue-specific expression and alpha-actinin binding prop- cross-correlation measurements: Eliminating cross-talk. Rev Sci Instrum 79(4):043103. erties of the Z-disc titin: Implications for the nature of vertebrate Z-discs. J Mol Biol 37. Schlierf M, Li H, Fernandez JM (2004) The unfolding kinetics of ubiquitin captured 270(5):688–695. with single-molecule force-clamp techniques. Proc Natl Acad Sci USA 101(19): 17. Young P, Ferguson C, Bañuelos S, Gautel M (1998) Molecular structure of the sarco- 7299–7304. meric Z-disk: Two types of titin interactions lead to an asymmetrical sorting of alpha- 38. Stigler J, Ziegler F, Gieseke A, Gebhardt JC, Rief M (2011) The complex folding net- actinin. EMBO J 17(6):1614–1624. work of single calmodulin molecules. Science 334(6055):512–516. 18. Young P, Gautel M (2000) The interaction of titin and alpha-actinin is controlled by a 39. Jahn M, et al. (2014) The charged linker of the molecular chaperone Hsp90 modulates phospholipid-regulated intramolecular pseudoligand mechanism. EMBO J 19(23): domain contacts and biological function. Proc Natl Acad Sci USA 111(50):17881–17886. 6331–6340. 40. Wang MD, Yin H, Landick R, Gelles J, Block SM (1997) Stretching DNA with optical 19. Ribeiro EdeA, Jr, et al. (2014) The structure and regulation of human muscle α-actinin. tweezers. Biophys J 72(3):1335–1346. Cell 159(6):1447–1460. 41. Efron B, Stein C (1981) The jackknife estimate of variance. Ann Stat 9:586–596.

1020 | www.pnas.org/cgi/doi/10.1073/pnas.1612681114 Grison et al. Downloaded by guest on September 26, 2021