Since the talin is anchored to integrins via the head domain, while actin binding sites exist downstream at the C-terminus of the H1–H12 bundle (see B), the force vector acting between the membrane-bound integrins and the cytoskeleton are likely transmitted through talin along the long axis of the H1–H12 rod. Unclear is, however, whether the force in full length talin acts directly on the two terminal atoms of the H1–H12 bundle, or given the tilt of the helices with respect to the long rod axis of talin [27
], whether it is distributed over force-bearing interfaces formed with the adjacent domains. The H1 helix might form a force-bearing interface with the talin head (the structure of the linker region connecting TH
is not known), while the H12 helix might be tightly packed against the rest of the structurally unresolved C-terminal talin rod (residues 890-2541). We thus conducted two sets of constant force simulations where the force is either applied locally to the two terminal Cα-atoms of the H1–H12 bundle, or is distributed along the length of the terminal helices, H1 and H12, to mimic the existence of such putative force-bearing interfaces. For both cases, we analyzed how stretching the talin rod with constant force would alter its structure. The little resistance to strainnig the bundle is due to the unfolding of secondary structure and leads to a rapid molecular extension with time. If a major energy barrier has to be overcome to allow the further unfolding of the rod, the protein domain will pause to extend as indicated by a plateau region in the extension-time plots ( and ). We thus asked how the structure of the H1–H12 bundle is changed with time as we pull on the rod with constant force, and whether one major event leads to the complete disintegration of the rod or whether intermediate states (I)
exist separated by energy barriers. Please note that the lifetimes of intermediate states increase as the forces are lowered to physiologically relevant forces (for further discussion see [31
]). Due to limited computational resources, however, the simulations shown here can only be run for several nanoseconds and higher forces are required to trigger the events described. For other molecular systems, however, it has been shown that these constant force approaches can correctly predict the positions of major energy barriers along physiologically significant unfolding pathways [31
Sequential Unfolding Trajectories and Associated Structures of the Talin Rod H1–H12 Extended under Constant Force Applied to Terminal Cα-atoms
Fragmentation of the Talin Rod into α-Helix Subbundles Leads to the Sequential Exposure of the Vinculin Binding Helices (VB Helices)
The Terminal Helices Tend To Unfold Easily if the Force Is Applied Directly to the Terminal Atoms of the H1–H12 Bundle
When applying various constant forces locally
to the terminal atoms of the H1–H12 bundle (100 pN, 200 pN, 300 pN, and 400 pN), the intial end-to-end distance of the talin rod H1–H12 of 3.2 nm increases rapidly with time due to progressive unfolding (loss of secondary structure) of the terminal helices (B). The ease by which they unravel even at low forces indicates that relatively little energy is required to sequentially break the backbone hydrogen bonds, turn-by-turn, that stabilize their helical secondary structure [39
]. As shown in A, a first short-lived plateau can be seen that has a lifetime of less than 1 ns if we pull at 100 pN (Intermediate State I1
). Just little activation is then needed to begin the unfolding of both of the terminal helices, H1 and H12, which leads to an extension with respect to the resting length of altogether 15 nm, until a second plateau is reached (Intermediate State I2
Notice though that both helices do not initially unfold completely. Only the N-terminal half of H1 (residues 499–503) is unraveled in state I2,
due to a bent in H1 at Ser502. The bending site seems to be defined by the side chain of residue Ser502, which competes for the backbone hydrogen bond between the Ile499 and Met503 (see B for presentation of such an event). Similarly for the H12, which is a vinculin binding helix, its C-terminal residues 874–879 are unraveled while its N-terminal part of the helix remains buried in the helix bundle.
Further extension results from the sequential turn-by-turn unraveling of the N-terminal end of H1 (D2; in all of 8 local force simulations; ), and the H12 by ~3 additional turns (Intermediate State I3) which leads to a total extension of 28 nm while pausing in the third plateau in A. This intermediate I3, which is seen to last for ~15 ns at 200 pN, is characterized by a completely unraveled H1, whereas residues 868–879 of H12 are unraveled while residues 849–867 of H12 are still in a helical conformation and in tight contact with the rest of the H9–H12 bundle. At higher forces, we see a rapid transitioning into I4 (multiple unraveled helices) where both H2 (~8 turns unraveled) and H12 (~6 turns unraveled) are mostly unraveled. I4 is characterized by an extension of 35–36 nm. Once the major energy barrier to rupture I4 is passed (as described below), the remaining bundle H2–H11 rapidly breaks down into the bundles H2–H5, H6–H8 and H9–H12 (T4 in ). Each of these three helix bundles unfolds independently from each other at later times as discussed below. Note that the two vinculin binding helices, H11 and H12, are already completely unraveled when T4 is reached, i.e., they convert early on into an extended polypeptide chain held under tension.
If the Tensile Force Is Distributed along the Length of the Terminal Helices (Mimicking Force-Bearing Interfaces), the Talin Rod Is Mechanically More Stable
To assess whether the just described intermediates are characteristic only for an isolated talin fragment where force is applied to the terminal atoms, or also for the case where force is distributed over the cross-section of the talin rod, we ran constant force simulations in which we distributed the force (200 pN, 250 pN, 300 pN, 400 pN) along the Cα
-atoms of the terminal helices H1 and H12 as shown in . Additional constant velocity simulations were carried out under otherwise similar conditions, but by applying force distributed along the C-terminal H12 (see Methods) and harmonically constraining H1 (Figure S2
, only stimulation with harmonical constrained helix).
Distributing the force over a putative interface had a major stabilizing impact on the resulting unfolding trajectories and the forces needed to activate the VBSs ( and ): The most notable difference is that the talin rod H1–H12 could withstand considerable higher forces before breaking apart (C), stably exceeding our simulation window of 20 ns in a prestretched intermediate state I1 () if pulled with a constant force (200 pN). Earlier, this force had resulted in the unraveling of the terminal helices within the first 5 ns (A) if force was only applied to the terminal atoms. Distributing the force over the terminal helices thus stabilized the entire H1–H12 bundle against force-induced breakage and terminal helix unraveling: unfolding of H1 was not detected as a first major unfolding event even at higher forces (7 simulations, 250–400 pN). Plotting the sequence by which hydrogen bonds between the helix bundles broke () reveals that straining the talin rod leads to the appearance of some new hydrogen bonds (for example bond 8 between residues Asn559 and Glu738) while others break (for example bond 5 between residues Ser658 and Pro725).
Side Chain Hydrogen Bonds Formed between the α-Helix Subbundles, H1–H5, H6–H8, H9–H12 in Talin1 as Identified in the Simulations
When H1–H12 is stretched with a constant force of 250–400 pN distributed over the terminal helices, interhelical bundle contacts start to loosen up after a few ns leading to a slight opening of the interface between the bundles H9–H12 and H1–H8 (I2). This passage leads to the breakdown of some hydrogen bonds in the interface between H1–H8 and H9–H12, namely between residues Gln887-Ser752, Gln886-Gln755, His784-Glu733, His788-Gln733 (referred to as bonds 19, 20, 21, 25 in ). Also the bonding between bundles H1–H5 and H6–H8 is weakened significantly by breakdown of hydrogen bonds between residues Asn559-Gln735, Ser729-Asp548, Gln715-Asn559 (referred to as bonds 1, 2, 3 in ). Further pulling results in a gradual opening of the interface between the bundles and water starts to penetrate between the helices (E). The stress applied leads finally to a bending of H9 around residue 770 due to breakage of the backbone hydrogen bond formed between residues Gly768 and Thr772 (B). After this, H9 stays attached to H1–H8 on the N-terminal end during the short-lived intermediate state I3 seen in many but not in all the simulations (5 of 7 simulations 250–400 pN). After passing the I3 intermediate, a rapid disintegration of the talin rod is seen (T3−) which then allows for the sequential unfolding of the now separated bundles.
Talin Rod Fragmentation Eliminates Force-Bearing Interfaces Which Then Weakens the Mechanical Stabilities of Its Fragments
Common to all simulations is the break-up of the talin rod, H1–H12, into smaller helix bundles that have far smaller mechanical stabilities. We observed that the first split occurred between H1–H8 and H9–H12 (6 times in 7 simulations where constant force is applied over terminal helices), after which H1–H8 splits into H1–H5 and H6–H8 (, , and ). In the other case observed only once in seven simulations, H1–H5 separated first followed by the H9–H12 separation. The break-up into these well-defined bundles as a major event is most clearly seen if the force is distributed over the terminal helices (). Once the force-bearing interfaces are broken apart, the bundles have no stabilizing effects upon each other any longer. Consequently, the force is thus transmitted at these later times through the N- and C-terminal atoms of the resulting helix bundles. We thus simulated separately the unfolding of the helix bundles H1–H9 (crystallographically determined structure of talin rod fragment, PDB 1SJ8), H2–H8 (intermediate found in terminal atom pulls), and H9–H12 (intermediate seen only in force-bearing interface simulations). The starting end-to-end distances of the terminal atoms of these helix bundles prior to stretching were 3.6 nm (H1–H9), 7.9 nm (H2–H8) and 1.5 nm (H9–H12). Most noticeable when looking at all their unfolding pathways () is the lack of well pronounced plateaus and thus clearly distinguishable intermediate states, especially in the case of H9–H12. If pulled apart at 300 pN, the H1–H12 bundle breaks into bundles H1–H8 and H9–H12 after 7 ns (). This indicates that different structural changes are happening in parallel at different positions.
Unfolding Trajectories of the Individual Talin Rod Subbundles
Furthermore, when simulating the mechanical stabilities of alternate talin rod fragments that were not truncated along a ‘natural' bundle-bundle interface, for example simulating H1–H9 instead of H2–H8, it is interesting to note that H9 is the least stable helix in those cases () since it belongs structurally to the H9–H12 bundle. Unraveling of H9 is the first major unfolding event of H1–H9 even if the force is distributed over terminal helices, since H9 is easily detached from the H1–H8 bundle.
The Vinculin Binding Sites (VBS) Are Sequentially Exposed once the Talin Rod Breaks Up
Activation of the VBS requires that the VB helices H4, H6, H9, H11, and H12, are at least partially exposed to water. As seen in D, these five VB helices are buried in the H1–H12 bundle under equilibrium, and become sequentially exposed only after the fragmentation of the talin rod into smaller helix bundles has occurred. The asterix defines the point where the strain-exposed solvent-buried surface area of a VB helix is equal to the solvent-buried area of the helix when complexed to the vinculin head. The asterix thus marks the unique points in the unfolding trajectory in which each of the VB helices gets activated (D). Breaking the talin rod apart thus defines the highest energy barrier that has to be overcome to initiate the exposure and activation of the VBSs, and the implications thereof will be discussed below.
Hydrogen Bonding Analysis of the Interfaces between Helix Bundles
To better understand the molecular mechanism behind the interactions that regulate helix bundle separation, we analyzed the hydrogen bonding pattern between the defined helix bundles as shown in . The blue fluctuatuions represent bonds formed between H1–H5 and H6–H8, while the green represents bonds formed between H6–H8 and H9–H12. While most hydrogen bonds fluctuate between formation and breakage even during the equilibration, only a few of the side-chain hydrogen bonds, like the bond 2, are longer-lived. This, together with the fact that there appears to be considerable statistical variability in bond breaking events between different simulations suggests that at least most of these side-chain hydrogen bonds are not force-bearing. While it cannot be excluded that one or the other of these bonds slightly contributes to the mechanical stabilization, hydrophobic contacts between the helix bundles seem to play the dominant role in upregulating the mechanical stability of the N-terminal part of the talin rod.
A separate SMD analysis of the talin rod fragments H1–H5 was done recently by applying force to the polar side chains T498, S501 and S502 close to the N-terminus of H1 and Q635, Q646, E650, and Q653 close to the C-terminus of H5, assuming that the force-bearing interactions were mediated by side chain hydrogen bond formation across the interfaces of adjacent α-helix bundles [41
]. When using an implicit water model in which the protein structure was solvated in a dielectric medium, they observed a rotation of VBS1 (H4) under applied force and suggested this to be a potential activating mechanism. They also observed that the H4 rotation was strongly reduced when repeating the simulation in the presence of explicit water molecules (these computationally more elaborate conditions were used in our simulations as well). We thus analyzed for how long the polar side chains of H5 are hydrogen-bonded across the interface formed between H1–H5 and H6–H8. Our simulations of H1–H12 reveal that water penetrates into the interface H1–H5 and H6–H8, thus breaking these side-chain hydrogen bonds, even before we can see a major force-induced structural change within the H1–H5 bundle (Figure S3
). Among the residues that were previously used to model the contact interface between bundles H1–H5 and H6–H8 [41
], only residues Gln635 (bonded to Ser714) and Gln646 (bonded to Lys721) are hydrogen bonded across the interface, and those side-chain bonds show quite low stability during our simulations (, bonds 6 and 14, and S3), suggesting that those polar residues are also not force-bearing during the activation process of the H4 helix.
Comparative Analysis of Talin1 and Talin2
To compare the mechanical properties of the talin1 and talin2 rods, H1–H12, we generated a homology model of talin2 based on the talin1 structure (Figure S4
). Homology modeling is a reasonable approach since their sequences are highly similar (74% identical) in this region. Furthermore, a sequence analysis of the hydrogen bonding partners when comparing talin1 and talin2 from human, mouse, and chicken () revealed that 29 of the 43 residues participating in hydrogen bonds between helix bundles are fully conserved, eleven of the residues (Ser658, Ser688, Thr693, Thr720, Ser754, Gln755, Gln762, Arg765, Glu780, His788, Lys869) are similar and only three residues are non-conserved (Arg692, Ser752, Gly766) (Figure S1
). The RMSD for backbone atoms after energy minimization and thermalization was 1.3 Å. Also for talin 2, the fragmentation of the rod H1–H12 constituted the major energy barrier as seen in constant force and constant velocity SMD simulations, in which the force was distributed over force-bearing interfaces. The split of the rod occurred in identical positions as described above for talin1, namely between the bundles H1–H5, H6–H8 and H9–H12. Also the sequence of early events was similar: First, the H9–H12 bundles separated, followed by the rupture of the H1–H8 fragment. Our simulations of early events further indicate that the major energy barrier of fragmenting the rod are not greatly different between the talin1 and talin2 rods, H1–H12, at least within the stochastic variability that is intrinsic to single molecule studies (Figure S5
Forces and Timescale of Computer Simulations
Finally, it is important to note that the SMD simulations are carried out on time scales that differ significantly from those at which biological molecules are stressed. Since the force needed to unfold a protein is logarithmically dependent on the pulling velocity, significantly smaller forces may be able to cause the here described structural rearrangements at physiological timescales. Unfolding forces measured by SMD in nanosecond timescale are thus significantly higher compared to those measured using AFM on millisecond timescales, yet, SMD has correctly predicted in the past the relative mechanical stabilities of some protein domains and the position of key energy barriers [32