The mechanical strength of single protein molecules can be investigated by using the atomic force microscope. By applying this technique to a wide range of proteins, it appears that the type of secondary structure and its orientation relative to the extension points are important determinants of mechanical strength. Unlike chemical denaturants, force acts locally and the mechanical strength of a protein may thus appear to be mechanically weak or strong by simply varying the region of the landscape through which the protein is unfolded. Similarly, the effect of ligand binding on the mechanical resistance of a protein may also depend on the relative locations of the binding site and force application. Mechanical deformation may thus facilitate the degradation or remodelling of thermodynamically stable proteins and their complexes in vivo.
- atomic force microscopy (AFM)
- energy landscape
- mechanical stability
- mechanical unfolding
Traditionally, the stability of a protein is quantified by measuring the extent to which a protein's tertiary structure can withstand changes of temperature, pH or the addition of chemical denaturants and this information is vital both to basic biochemical research and to its industrial applications. However, in vivo, many proteins need to withstand or respond to a mechanical stimulus such as extension as part of their function. While the importance of the effects of mechanical perturbation on macroscopic assemblies of proteins such as collagen, keratin, muscle and cartilage has long been appreciated, the ubiquity of force on the micrometre to nanometre scale has only recently been recognized. For example, as well as the generation of movement by processive proteins, mechanical force is also utilized in signal transduction , degradation  and the remodelling of proteins and their complexes . This interest, together with the development of both dedicated force probe instruments and steered molecular dynamics techniques, has led to an explosion of research activity and the birth of a new field, dynamic force spectroscopy. The application of this technique to delineate the factors that define a protein's mechanical response is the subject of this review.
Measuring the unfolding of a single polypeptide chain
Several techniques can be used to measure the response of a single biomolecule to the application of a force or a deformation. These include magnetic and laser tweezers, the BFP (biomembrane force probe) and AFM (atomic force microscopy). While each technique uses a relatively soft spring of known stiffness to relate a displacement to a force, each is optimal on different length and force scales  and, consequently, each is suited to particular biochemical problems. This short review focuses on the application of AFM to single molecule mechanical unfolding and the reader is directed elsewhere for details on laser or magnetic traps  and the BFP .
In the standard use of AFM, a picture (in terms of height or sample softness) of a biomolecule immobilized on to a substrate is acquired by raster scanning a small microfabricated silicon nitride cantilever over the substrate. The vertical displacement of the cantilever is monitored by measuring the deflection of a laser focused on the topside of the tip and, in this way, a topographical image of the substrate is built line by line. In force mode AFM, the cantilever does not usually raster across a surface. Instead, it is repetitively pushed on to the substrate and retracted at a preset rate. The cantilever acts as a Hookean spring, so the force applied on to the biological sample can be related to the change in deflection of the cantilever via the known spring constant of the cantilever. A typical unfolding experiment is shown schematically in Figure 1.
Is mechanical strength innate to all proteins?
Single molecule mechanical unfolding has now been performed on a diverse range of proteins and these results show that, when extended by the N- and C-termini, proteins unfold between the thermal noise limit of the instrument (∼20 pN; [7,8]) and several hundred piconewtons . As many of these proteins have known three-dimensional structures and well-characterized biophysical parameters (such as thermodynamic and kinetic stabilities), it is natural to compare these parameters with mechanical strength to delineate which features of the underlying energy landscape modulate the force response of a protein. This knowledge would thus allow the mechanical properties of novel proteins to be predicted and proteins with tailored mechanical properties to be designed.
A comparison between the mechanical strength of a protein domain and its biophysical properties first shows that there is no correlation between the thermodynamic stability of a domain and the force at which it is likely to unfold [10–12]. This may initially seem surprising but the observation that the unfolding force of a protein depends on the rate at which it is extended indicates that unfolding is a kinetic and not a thermodynamic process. Consequently, the unfolding force would be expected to depend on the height of the barrier that must be traversed to undergo a transition from the folded to unfolded state (described by the unfolding rate constant). However, once more, there is little correlation of this parameter with unfolding force (for example, barnase is kinetically stable but unfolds at a force of ∼60 pN ), suggesting that mechanical unfolding proceeds over a transition barrier distinct from the intrinsic unfolding pathway.
What then are the features of proteins that result in such a diverse response to force? An important determinant is clearly demonstrated by mechanically unfolding a heteropolymer consisting of alternating domains of I27 and Im9 (see Figure 2). These proteins have similar thermodynamic stabilities and unfolding rate constants but I27 (an all β-sheet protein) typically unfolds at ∼185 pN, whereas Im9, an all α-helical protein, unfolds at a force that cannot be detected (below ∼20 pN) . The type of secondary structure may thus be a key determinant of protein mechanical strength. Indeed, data accumulated to date on a wide variety of proteins suggest a hierarchy of mechanical strength whereby all α-helical proteins such as spectrin are generally mechanically weak (unfolding at less than 100 pN at typical extension rates [14–16]) and most β-sheet proteins such as Ig-like domains are mechanically robust  (see also Table 1 in ). The mechanical strength of proteins with mixed α+β folds is more variable but they generally have intermediate strength. This seemingly simple correlation was first observed by simulating the mechanical unfolding of proteins with different folds . In addition, these authors also noted that β-sheet proteins with directly hydrogen-bonded parallel terminal strands (such as Ig-like proteins) show very high mechanical strength. While subsequent experimental results appeared to concur with this observation, its generality was difficult to assess because most β-sheet proteins studied by this technique have Ig-like folds (a consequence of the fact that many force-bearing proteins in vivo consist of tandem arrays of such proteins interspersed with other folds).
In order to assess the importance of directly hydrogen-bonded parallel terminal strands in the mechanical resistance of proteins, we scanned the protein structural database for a model test protein that was small, had parallel hydrogen-bonded N- and C-termini, was not an Ig-like protein and had no mechanical function in vivo. The B1 domain of Protein L (protein L) met these criteria (see Figure 3). Mechanical unfolding of a pentameric construct of protein L revealed that this protein shows significant mechanical strength at all extension rates tested (40–4000 nm/s). Similar results on a homologue, Protein G, have also been reported . High mechanical resistance has also been reported for the computationally designed protein Top7 [20,21]. Interestingly, the terminal strands of this protein are separated by intercalation of a third strand, which forms hydrogen bonds with both terminal strands (see Figure 3). These results are in accord with the hypothesis that the mechanical resistance of proteins can be rationalized simply by the type of secondary structure and, as we shall see later, the topology of this secondary structure relative to the position of the extension points.
These experiments clearly demonstrate that topology, not function, is an important determinant of mechanical strength. Sequence, however, strongly modulates this response. For example, I27 and I32 from human titin have high sequence identity (42%; ) but vary significantly in their unfolding forces (204 and 298 pN respectively; ). In addition, FNIII (fibronectin III) domains with almost identical structures  or proteins from the same fold family such as protein L , Protein G  and ubiquitin [23,24] also vary significantly in their mechanical properties, suggesting that other factors such as hydrophobic interactions in the core of the protein can also modulate protein mechanical strength.
Exploring the energy landscape of proteins by force spectroscopy
The importance of topology and the lack of correlation between intrinsic unfolding rates measured by addition of chemical denaturants and those measured by mechanical unfolding experiments can be explained by considering that mechanical unfolding is a probe of the stability of the protein local to the points being extended, whereas chemical denaturants act globally by solubilizing all parts of a protein. In mechanical unfolding experiments, proteins are usually expressed as tandem arrays and, in this case, force is always applied on to the N- and C-termini, limiting the mechanism of unfolding to a narrow pathway through the energy landscape. Importantly, there is no reason why this pathway should be the same as that of the intrinsic unfolding pathway and the rate-limiting barriers for chemical and mechanical unfolding may be quite distinct. Dynamic force spectroscopy is thus able to explore areas of the energy landscape that are inaccessible to standard ensemble techniques.
If force does act as a local denaturant, then a protein may be expected to behave anisotropically when force is applied to different points. A protein may either be mechanically strong when regions with high local stability are sheared apart (e.g. when proteins with parallel hydrogen-bonded terminal β-strands are extended by their N- and C-termini) or may display mechanical weakness when force is applied across a dynamic region or in such a way that interactions are broken in a stepwise manner. Investigation of the anisotropy of protein mechanical resistance has been carried out in a number of ways. In the first examples, two groups utilized the specific post-translational modifications performed on E2lip3  and ubiquitin  to unfold these proteins in two different geometries. These studies revealed that proteins do indeed show very different mechanical strengths when force is applied at different locations. These initial studies have now been complemented by other studies that have used either disulfide linkages  or circular permutation [27–29] to investigate the difference in force response of a protein when pulled in not two but up to five geometries . The power of this technique is exemplified by the wealth of knowledge now accumulated [26–28] for green fluorescent protein and its analogues, leading the way to the development of a force sensor whose sensitivity can be tailored by simply selecting the points to which force is applied.
What is the effect of complexation on the mechanical strength?
As well as the innate mechanical strength of a particular conformation, many proteins that have force-resistant or force-sensitive functions form complexes with ligands that may modulate their force response and therefore their function in vivo. However, while the determinants of protein mechanical strength have been examined in detail, the effect of binding ligands on mechanical strength has been investigated in only a few cases using small ligands [8,30–32].
In addition to small ligands, many proteins with a force-resistant or force-responsive function bind one or more protein ligands, typically involving the burial of significant areas of protein surface. The effect of protein–protein complexation on protein mechanical stability has been investigated using the high-affinity interaction between the nuclease domain of the colicin E9 and its cognate immunity protein Im9 . As shown in Figure 2, when concatenated into a construct of alternating I27 domains, Im9 is mechanically labile, as expected for an all α-helical protein. Surprisingly, complexation of E9 has little effect on the mechanical resistance of Im9 (unfolding force ∼30 pN) despite the high avidity of this complex . The mechanical lability of a protein in such a stable protein–protein complex is remarkable but accords with the hypothesis that force acts as a local probe of stability. In Im9, the points of force application are distal to the Im9–E9 binding interface, possibly rationalizing why the binding of E9 only marginally stabilizes Im9 to mechanical deformation .
These results highlight the ability of force to dissociate even the strongest protein–protein interaction when applied in a suitable geometry relative to the protein topology. This may suggest a method by which long-lasting complexes are rapidly dissociated in vivo by the application of a moderate force .
In the decade since the first report of single molecule mechanical unfolding , much progress has been made in our understanding of why some protein folds are more mechanically robust than others. However, attempts to predict and tailor the mechanical response of proteins with similar folds have met with varied success [12,20,34], emphasizing that, at the molecular level, much is still to be learned.
Financial support by the University of Leeds, the Biotechnology and Biological Sciences Research Council and the Electronic and Physical Research Council is gratefully acknowledged.
7th International Conference on Protein Stabilization 2007: Independent Meeting held at the University of Exeter, Exeter, U.K., 11–14 April 2007. Organized and Edited by J. Littlechild (Exeter, U.K.).
Abbreviations: AFM, atomic force microscopy; BFP, biomembrane force probe
- © The Authors Journal compilation © 2007 Biochemical Society