Folding and conformational changes in proteins

The global spatial arrangement of atoms in a protein is called protein conformation. Catalysis is not only determined by chemicals steps, but also by conformational changes in protein structure. Proteins must adopt a specific conformation, or in other words, a specific three-dimensional structure (native protein) to be catalytically active (Anfinsen et al., Reference Anfinsen, Redfield, Choate, Page and Carroll1954; Carrion-Vazquez et al., Reference Carrion-Vazquez, Oberhauser, Fowler, Marszalek, Broedel, Clarke and Fernandez1999; Klinman and Hammes-Schiffer, Reference Klinman and Hammes-Schiffer2013). In fact, proteins can be found mostly folded in physiological conditions. The stability of native proteins in folded state is only marginally higher compared to the unfolded state (Taverna and Goldstein, Reference Taverna and Goldstein2002; Magliery, Reference Magliery2015), as the difference in Gibbs free energy (ΔG) separating folded and unfolded states is only about 5 to 15 kcal/mol (Creighton, Reference Creighton1990; Voet and Voet, Reference Voet and Voet2010). Protein stability studies have usually been carried out by reversibly unfolding using denaturants (chaotropic agents such as guanidine chloride and urea), increasing or decreasing the temperature, varying the pH, applying high pressures, or cleaving disulfide bonds. These studies allow for the determination of the free energies necessary to maintain the folded state, the intermediate states if they exist, the folding energy landscape, and the population of the unfolded protein (Dill, Reference Dill1985; Shea and Brooks, Reference Shea and Brooks2001).

Many biochemical reactions proceed via large conformational changes within or between interacting molecules. In the past, protein conformational changes were viewed as rigid body movements coupled to ligand binding with just one lowest free energy conformation in each condition (with and without ligand). One of the best examples is hemoglobin, where the conformational changes are coupled to the binding of an oxygen molecule in a remote site of the active site (allosterism) (Edelstein, Reference Edelstein1975). However, the current concept of native proteins considers them as ensembles of related, interconverting, transient microstates that describe the canonical high-resolution structures observed by crystallography or NMR spectroscopy (Frauenfelder et al., Reference Frauenfelder, Parak and Young1988; Hilser et al., Reference Hilser, García-Moreno, Oas, Kapp and Whitten2006; Fenimore et al., Reference Fenimore, Frauenfelder, Magazù, McMahon, Mezei, Migliardo, Young and Stroe2013). Thus, protein movements are consequence of a collection of microstates that occur at different time scales, as seen in Figure 1.

Figure 1.

The many different conformational states produce a rocky landscape.

The conformations existing under a given set of conditions are usually the ones that are thermodynamically the most stable, having similar ΔG (Creighton, Reference Creighton1990; Voet and Voet, Reference Voet and Voet2010). The existence of multiple stable conformations reflects the changes that occur in proteins as they bind to other molecules or when they catalyze reactions, validating the idea that many of the physical and functional properties of proteins (stability, solubility, binding of ligands) are influenced by the same structural fluctuations that give rise to the ensemble without a ligand (Hilser et al., Reference Hilser, García-Moreno, Oas, Kapp and Whitten2006; Klinman and Hammes-Schiffer, Reference Klinman and Hammes-Schiffer2013).

Ligand-binding and conformational changes in catalysis

Protein conformational changes play a crucial role in catalysis, enabling proteins to bind to ligands, form oligomers, and perform mechanical work (Koshland, Reference Koshland1958; Whitford et al., Reference Whitford, Miyashita, Levy and Onuchic2007; Olsson and Wolf-Watz, Reference Olsson and Wolf-Watz2010). These conformational changes can occur in time scales from milliseconds to nanoseconds (Henzler-Wildman et al., Reference Henzler-Wildman, Thai, Lei, Ott, Wolf-Watz, Fenn, Pozharski, Wilson, Petsko, Karplus, Hübner and Kern2007) and involve domain movements. According to the ‘induced fit’ theory (Koshland, Reference Koshland1958), substrate binding triggers a large conformational change in the enzyme, leading to the correct positioning of the residues involved in catalysis and the release of the product. In this view of enzyme catalysis, binding interactions between the substrate and the enzyme provoke motions in enzyme’s structure that bring the substrate into the transition state, facilitating and pushing forward the catalytic process along its reaction coordinate (Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004; Voet and Voet, Reference Voet and Voet2010). Kinases are generally viewed as good examples of the induced-fit mechanism (Koshland, Reference Koshland1958; Schulz et al., Reference Schulz, Müller and Diederichs1990; Choi and Zocchi, Reference Choi and Zocchi2007).

Another view of enzyme catalysis, according to the Haldane–Pauling hypothesis (Pauling, Reference Pauling1946; Haldane, Reference Haldane1965), considers the idea that part of the enzymatic catalysis is ‘strain-induced’ (also called geometric destabilization theory (Jencks, Reference Jencks1975). It suggests that enzymes cannot be efficient catalysts if they are fully complementary to the substrates and they must exert strain on the substrate upon binding. Furthermore, it suggests that in order to catalyze reactions, an enzyme must be complementary to the transition state species, and not to the substrate molecule itself in its normal configuration. This hypothesis considers that the energy derived from the enzyme-substrate interaction (the binding energy) is used to accelerate the rate of the reaction. Experimental evidence of strain-induced catalysis has been indirect (Amyes and Richard, Reference Amyes and Richard2013). One example is using catalytic antibodies. They recognize and bind non-covalently to their complementary antigen, then catalyze bond-breaking or other chemical reactions involving the antigen (Hanson et al., Reference Hanson, Nishiyama and Paul2005). In those works, they used molecules that were analogs to the transition state which bind more tightly to the antibody than its actual antigen or substrate (Pollack et al., Reference Pollack, Jacobs and Schultz1986; Tramontano et al., Reference Tramontano, Janda and Lerner1986; Savinov et al., Reference Savinov, Hirschmann, Benkovic and Smith2003). Another example is the study of the strain in the carbonyl bond of the substrate of beta-lactamase (Hokenson et al., Reference Hokenson, Cope, Lewis, Oberg and Fink2000). Class A beta-lactamases hydrolyze penicillins and other beta-lactams via an acyl-enzyme catalytic mechanism. Comparison of the beta-lactam carbonyl stretch frequency in the free and enzyme-bound substrate revealed an average decrease of 13 cm⁻¹ in its frequency, indicating substantial strain/distortion of the lactam carbonyl when bound in the enzyme-substrate (ES) complex. The substrate binding to the active site induces substantial strain and distortion that contributes significantly to the overall rate enhancement in beta-lactamase catalysis (Hokenson et al., Reference Hokenson, Cope, Lewis, Oberg and Fink2000). Although these data help us to understand strain-induced catalysis, they are not enough to quantify the contribution of strain in catalysis. It is also not clear how this event is linked with the conformational changes coupled to ligand binding.

Brownian ratchet versus power stroke model

There are two general models to explain how molecular motors operate: the Brownian ratchet and the Power stroke model. In the Brownian ratchet model, the motor uses nucleotide binding and/or hydrolysis to direct and rectify its Brownian motion, with a net movement in one direction (Astumian, Reference Astumian1997). In the power stroke model, the motor utilizes the energy of ATP binding and/or hydrolysis or product release to drive the motion (Smith et al., Reference Smith, Tans, Smith, Grimes, Anderson and Bustamante2001). A typical Brownian motion ratchet mechanism molecular motor is the E. coli RNA polymerase with a work (stall force times step size) value close to 2 k_BT (k_B is the Boltzmann constant and T is the temperature) at stall force (force at which the velocity of the motor drops to zero); (Neuman et al., Reference Neuman, Abbondanzieri, Landick, Gelles and Block2003). In the case of the bacteriophage Phi29 (a power stroke motor), the work value is about 10 k_BT (Moffitt et al., Reference Moffitt, Chemla, Aathavan, Grimes, Jardine, Anderson and Bustamante2009). It is considered a power stroke model when the work is higher than 5 k_BT at the stall force (Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004). How do motors convert the chemical energy into mechanical work or movement? What is the mechano-chemistry of the motor? The force can be considered as a product of the nucleotide binding/hydrolysis/product release cycle, and the velocity of a motor often depends on the external force and is related to the mechanism of the motor (Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004). The rate of the motor will be force-dependent if, in the conditions of the experiments, the movement is the rate-limiting step. By systematically varying the concentration of ATP, or its hydrolysis products (ADP, Pi) at different forces, it is possible to determine the force-generating step during the chemical cycle of ATP hydrolysis and to study the mechano-chemistry of molecular motors (Wang et al., Reference Wang, Schnitzer, Yin, Landick, Gelles and Block1998; Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004). These studies have been done in many systems. For example, it has been shown that the rate constants of certain sub-steps in the mechanochemical cycle of molecular motors are force-dependent (or torque-dependent in rotary motors) as seen for Myosin and F1-ATPase (Oguchi et al., Reference Oguchi, Mikhailenko, Ohki, Olivares, De La Cruz and Ishiwata2008; Sellers and Veigel, Reference Sellers and Veigel2010; Adachi et al., Reference Adachi, Oiwa, Yoshida, Nishizaka and Kinosita2012; Watanabe et al., Reference Watanabe, Okuno, Sakakihara, Shimabukuro, Lino, Yoshida and Noji2012; Cossio et al., Reference Cossio, Hummer and Szabo2015; Volkán-Kacsó and Marcus, Reference Volkán-Kacsó and Marcus2015; Houdusse and Sweeney, Reference Houdusse and Sweeney2016). Molecular motors are enzymes. Thus to understand the mechanisms that govern their function we should focus on how enzymes work.

Importance of forces in mechanochemical processes

Different processes inside the cell, such as chromosomal segregation, DNA transcription, DNA replication, the formation and liberation of vesicles, the packing of DNA during viral replication, and the translocation of protein through channels, are mechanical processes (Schekman, Reference Schekman1994; Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004; Cecconi et al., Reference Cecconi, Shank, Bustamante and Marqusee2005; Zhang et al., Reference Zhang, Halvorsen, Zhang, Wong and Springer2009). As mentioned before, protein folding and unfolding, conformational changes, and catalysis also represent good examples of mechanical processes studied by the application of a mechanical force. One of the advantages of studying the reactions with a focus on force is that the reaction coordinate is an easily quantifiable physical parameter. The kinetics of a reaction can be studied by the application of a mechanical force (Bell, Reference Bell1978). An applied mechanical force affects the free energy, equilibrium, and rate of a reaction occurring along a mechanical reaction coordinate, so thermodynamic and kinetic parameters could be determined at zero force extrapolating the information obtained at different forces. Recently, new transition state theories stemming from the application of Kramers’s theory (Kramers, Reference Kramers1940) have emerged (Dudko et al., Reference Dudko, Hummer and Szabo2008; Cossio et al., Reference Cossio, Hummer and Szabo2016; Bullerjahn et al., Reference Bullerjahn, Sturm and Kroy2020). Another interesting and useful application of force in the study of conformational changes and catalysis, called ‘the allosteric spring probe’ (ASP), was developed by the group of G. Zocchi (Choi and Zocchi, Reference Choi and Zocchi2007), using a guanylate kinase (GK) as a model. The novelty of this technique is in the attachment of a single-stranded DNA spring to a protein, followed by using complementary strands of different sizes to apply a mechanical force, generating useful information about the mechanism of the reaction and allowing to assess quantitatively the relevance of the induced-fit theory in this enzyme.

Force is also very important in many biological signaling processes. For example, in order to start the coagulation process the Von Willebrand factor unfolds via shear forces, and exposes the binding site to start the process (Zhang et al., Reference Zhang, Halvorsen, Zhang, Wong and Springer2009). It has been shown that the folding force of SNARE proteins is enough to achieve membrane fusion in the vesicle release at the synapse (Gao et al., Reference Gao, Zorman, Gundersen, Xi, Sirinakis, Rothman and Zhang2012; Zhang et al., Reference Zhang, Rebane, Ma, Ma, Li, Jiao, Qu, Pincet, Rothman and Zhang2016). Another study has shown that the refolding of the Top7 protein is enough to release a stall sequence from the ribosome (Goldman et al., Reference Goldman, Kaiser, Milin, Righini, Tinoco and Bustamante2015). All these studies show that refolding force is crucial for biochemical processes in all cells.

Dynamic of local unfolding and the generation of work

But how fast will the protein refold? For folding of a complete protein, the experimental and theoretical approaches predict a speed limit of approximately N/100 μs for a generic N-residue single-domain protein (Kubelka et al., Reference Kubelka, Hofrichter and Eaton2004; Dill et al., Reference Dill, Ozkan, Shell and Weikl2008). Then, a small portion of the protein could fold very fast because of its size which is compatible with the catalysis timing. For example, each catalytic cycle in AK takes 25–40 ms and the size of the local unfolded region is around 20 amino acids, predicting a refolding time of 0.2 μs, one order of magnitude faster than the turnover number (Pelz et al., Reference Pelz, Žoldák, Zeller, Zacharias and Rief2016).

Despite the various hypotheses surrounding enzyme catalysis, there is still uncertainty about where the catalytic power of the enzyme resides. Most of these ideas have remained as hypotheses, partly due to the difficulty of designing experiments to demonstrate them. While the induced-fit mechanism is widely accepted (Klinman and Hammes-Schiffer, Reference Klinman and Hammes-Schiffer2013), the ‘strain-induced’ idea remains largely hypothetical as bulk chemical experiments provide only indirect support for its validity (Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004). Moreover, it is challenging to correlate conventional bulk experiments with in silico dynamic simulations to demonstrate the existence of cracking. Therefore, it is crucial to find a molecular description of the mechanisms by which enzymes achieve high catalytic efficiency coupled to conformational changes, demonstrating the importance of strain in these processes. A modern approach to address this question is to consider protein folding and unfolding, conformational changes, and catalysis as mechanical processes, where basic physical concepts such as force, torque, work, energy conversion efficiency, and mechanical advantage can be determined to describe them (Bustamante et al., Reference Bustamante, Chemla, Forde and Izhaky2004).

Based on this knowledge, we could hypothesize that the energy released by binding/catalysis in an enzyme is coupled to local unfolding and the refolding is the process that allows the mechanical work.

Another indication of local unfolding is the existence of intermediates. Intermediates of the folding pathway are very common even for very small proteins (Rivera et al., Reference Rivera, Mjaavatten, Smith, Baez and Wilson2023b). Even for typical two-states proteins, if we change more parameters to unfold the protein we could see an intermediate (Rivera et al., Reference Rivera, Mjaavatten, Smith, Baez and Wilson2023b). These intermediates could be related to the local unfolding of proteins and may be a universal characteristic. In the past unfolding regions were seen as non-functional regions of the protein. Today the view is completely contrary. There are many proteins that have disordered regions that are essential for their functions. Such proteins are known as intrinsically disordered proteins (IDPs)(Oldfield and Dunker, Reference Oldfield and Dunker2014). It has been shown that these disordered regions can fold upon binding to another protein (Malagrinò et al., Reference Malagrinò, Visconti, Pagano, Toto, Troilo and Gianni2020).

In our view of molecular motors, an unfolded region will be very important to perform work once it is re-folded. As previously mentioned, the refolding force is enough to perform work in some systems such as SNAREs, SecA, and IDPs which can fold under different conditions such as substrate binding, protein–protein interaction, and so on (Gao et al., Reference Gao, Zorman, Gundersen, Xi, Sirinakis, Rothman and Zhang2012; Goldman et al., Reference Goldman, Kaiser, Milin, Righini, Tinoco and Bustamante2015).

Aquifex aeolicus adenylate kinase

A suitable enzyme to study as a minimal molecular motor should involve a big conformational change upon substrate binding, it should display a correlation between turnover number and the fluctuations in the domain closure, the turnover number should not be too fast (slower than the detection system), and the molecule should be preferably a monomer (Figure 2a). In this context AK from the thermophilic organism Aquifex aeolicus and GK from Mycobacterium tuberculosis are good candidates. AK (EC 2.7.4.3) is a small (20–26 kDa) monomeric ubiquitous signal transducing enzyme that is responsible for maintaining cellular steady-state concentration of adenylate nucleotides in the cell (Schrank et al., Reference Schrank, Wrabl and Hilser2013). This enzyme has been studied in detail for decades and much is known about its structure, function, kinetics, and taxonomy. This enzyme catalyzes the reversible phosphoryl transfer reaction, using Mg²⁺ or Mn²⁺ as a cofactor, according to: $ {\mathrm{Mg}}^{2+}+\mathrm{ATP}+\mathrm{AMP}\leftrightarrow {\mathrm{Mg}}^{2+}+2\mathrm{ADP} $ (Hamada et al., Reference Hamada, Sumida, Kurokawa, Sunayashiki-Kusuzaki, Okuda, Watanabe and Kuby1985; Schulz et al., Reference Schulz, Müller and Diederichs1990). From a structural point of view, this protein is modular and is formed by three subdomains: the core subdomain that is very important for the overall stability, the ATP and AMP binding subdomains that have the ATP and AMP binding sites, and the lid subdomain that increases catalytic efficiency by covering the binding sites. The X-ray crystal structures of AK have been solved in an open (substrate-free) and a closed (inhibitor-bound) conformation (Lienhard and Secemski, Reference Lienhard and Secemski1973; Müller and Schulz, Reference Müller and Schulz1992). Analysis of the three-dimensional structures of AK, in both the bound and unbound states, reveals that the lid subdomains and the ATP and AMP subdomains undergo a significant conformational change in response to substrate binding (Schulz et al., Reference Schulz, Müller and Diederichs1990). This movement has been described as domain closure, which in the case of AK is 11 Å (Schulz et al., Reference Schulz, Müller and Diederichs1990; Sinev et al., Reference Sinev, Sineva, Ittah and Haas1996), being in the closed conformation where the phosphoryl transfer reaction occurs (Müller and Schulz, Reference Müller and Schulz1992; Ådén and Wolf-Watz, Reference Ådén and Wolf-Watz2007) and as explained previously has been shown that it undergoes cracking. Since the resolution limit of the technique is at least 10 Å, AK is a good candidate for the study of its catalysis with the optical tweezers (Wilson et al., Reference Wilson, Leachman, Cervantes, Ierokomos, Marqusee and Bustamante2013; Pelz et al., Reference Pelz, Žoldák, Zeller, Zacharias and Rief2016). AK has a catalytic constant (k _cat) of 40 sec⁻¹ (40 catalysis events per second) which is well within the instrumental capabilities, since it is possible to determine cycles of catalysis of 40 events or less with our bandwidth of 1000 Hz. The k _close and k _open of this enzyme were determined to be 44 and 1600 Hz, respectively (Henzler-Wildman et al., Reference Henzler-Wildman, Thai, Lei, Ott, Wolf-Watz, Fenn, Pozharski, Wilson, Petsko, Karplus, Hübner and Kern2007). Some correlations between the enzyme flexibility and catalysis have already been described using bulk techniques (Antikainen and Martin, Reference Antikainen and Martin2005). For instance, the work of Wolf-Watz et al. (Reference Wolf-Watz, Thai, Henzler-Wildman, Hadjipavlou, Eisenmesser and Kern2004) and Henzler-Wildman et al. (Reference Henzler-Wildman, Thai, Lei, Ott, Wolf-Watz, Fenn, Pozharski, Wilson, Petsko, Karplus, Hübner and Kern2007) found strong correlations between the k _open and the turnover number by comparing the mesophilic and the thermophilic AK. A work from (Ben Ishay et al., Reference Ben Ishay, Rahamim, Orevi, Hazan, Amir and Haas2012) using time-resolved FRET had been shown that AK can refold locally a domain of the protein within the dead time of the stopped-flow device, showing that this local refolding is below 5 ms. This shows that the protein could unfold and refold faster than the catalytic cycle. If this intermediate corresponds to the cracked region with faster techniques and/or transition paths analysis (Cossio et al., Reference Cossio, Hummer and Szabo2018) we should be able to isolate this metastable state.

Figure 2.

Single-molecule force spectroscopy techniques (open and closed conformations of the proteins to be studied). (a) Left: Open apo Aquifex AK (PDBid: 2RH5). Middle: Local unfolded or crack intermediate. Right: Close Aquifex AK in complex with the substrate analogue Zn^2 + ● Ap5A (light blue) (PDBid: 2RGX) (Olsson and Wolf-Watz, Reference Olsson and Wolf-Watz2010). In yellow is the lid domain and in gray is the core of the protein. Reprinted by permission from Springer Nature. (b) Left: Open configuration GK (PDBid: 1ZNW). Right: Close configuration of GK (PDBid: 1LVG) upon substrate binding. Alpha helices are colored magenta and beta sheets are colored yellow. (Alavi, Reference Alavi2017). (c) Open configuration BiP in the presence of ADP (PDBids: 5EVZ and 5E85) and right. Close configuration of BiP in the presence of ATP (PDBid: 5E84). In blue is the NBD and orange is SBD (Pobre et al., Reference Pobre, Poet and Hendershot2019). Images were made with 3D Protein imaging software (Tomasello et al., Reference Tomasello, Armenia and Molla2020).

Another interesting feature of AK is that this enzyme has an important role in health. There exists a lot of experimental data validating the direct relationship between AK and human diseases. For example, knockout of the major human AK isoform (AK1) demonstrated a deficit in vascular and myocardial AK catalysis blunts energetic communication, AMP, and adenosine metabolic signal transduction. This compromises post-ischemic coronary reflow, providing evidence for the role of AK as a metabolic monitor supporting regulation of the reactive vascular response in the stressed heart (Dzeja et al., Reference Dzeja, Bast, Pucar, Wieringa and Terzic2007). AK1 deficiency has also been related to hemolytic anemia (Abrusci et al., Reference Abrusci, Chiarelli, Galizzi, Fermo, Bianchi, Zanella and Valentini2007). It has been observed that human AK isoform (AK4) is upregulated in lung adenocarcinoma compared with normal cells, suggesting that AK4 promotes malignant progression and recurrence by promoting metastasis in a transcription factor ATF3-dependent manner (Jan et al., Reference Jan, Tsai, Yang, Huang, Yang, Lai, Lee, Jen, Huang, Su, Chuang and Hsiao2012). Therefore, understanding the mechanisms by which AK exerts its efficient catalysis is of great importance as it could help to unravel the causes and to develop therapies involving the AK function. In this respect, we have determined the sequence identity between Aquifex aeolicus adenylate kinase (Aquifex AK) and human AK, establishing that all of them have substantial identity (over 90% between Aquifex AK and Human AK1, AK2, and AK3 isoenzymes).

Mycobacterium tuberculosis guanylate kinase

The other protein of interest is GK which catalyzes the ATP-dependent phosphorylation of GMP into GDP (ATP + GMP $ \rightleftharpoons $ ADP + GDP) (Stehle and Schulz, Reference Stehle and Schulz1992). In humans, GK is the only known enzyme responsible for GDP production and thus is essential for cellular viability and might be a potential target for cancer therapeutics (Khan et al., Reference Khan, Shah, Ban, Trigo-Mouriño, Carneiro, DeLeeuw, Dean, Trent, Beverly, Konrad, Lee and Sabo2019). This protein is around 24 kDa and 4 nm long, similar in size and conformation to AK. As seen in Figure 2b, GK’s structure consists of a clamp-like cavity with three distinct regions: a central CORE domain, a LID which closes onto the CORE domain during catalysis and a GMP binding domain (GMP-BD). Additionally, a canonical P-loop motif is involved in ATP binding (Choi and Zocchi, Reference Choi and Zocchi2007; Khan et al., Reference Khan, Shah, Ban, Trigo-Mouriño, Carneiro, DeLeeuw, Dean, Trent, Beverly, Konrad, Lee and Sabo2019).

Upon substrate binding the two lobes of the clamp close through an ∼1-nm conformational change, most of which is induced by GMP binding. This conformational change that is relatively large compared to the size of the enzyme involves several direct and indirect (via water molecules) interactions which bring the LID and the GMP-BD jointly toward the CORE region (Choi and Zocchi, Reference Choi and Zocchi2007). This substrate-induced conformational change turns the enzyme from the open state to the close state, but it is not known if this conformational change takes place via cracking. In this view, the catalytic cycle of the enzyme can be described by the kinetic cycle of conformational changes, an idea that was proposed more than 50 years ago (Bliumenfel’d, Reference Bliumenfel’d1971) and experimentally studied using NMR and Nanorheology (Eisenmesser et al., Reference Eisenmesser, Bosco, Akke and Kern2002; Qu and Zocchi, Reference Qu and Zocchi2013).

In 2007, Choi and Zocchi performed an elegant experiment on GK by inserting an externally controllable molecular spring on the protein (an allosteric spring probe, ASP). In this experiment, 60 mer DNA was used as the probe which in ss form exerts essentially no tension but significantly rigidifies upon hybridization with a complementary strand and therefore exerts a mechanical stress on the protein (Choi and Zocchi, Reference Choi and Zocchi2007).

They found that when the open structure is favored through mechanical stress the binding affinity for GMP is drastically reduced, whereas the binding affinity for ATP and the catalytic rate were unaffected. This result shows that GMP binding does not allosterically control ATP binding but must allosterically control catalysis or at least the ATP hydrolysis step, as ATP would not hydrolyze in the absence of bound GMP.

Another novel experiment in Zocchi’s lab was studying GK’s mechanical properties through a technique they call nanorheology (Figure 3). In this technique the protein tethers Gold nanoparticles, to a gold surface. The particles are driven by an AC electric field while their displacement is synchronously detected by evanescent wave scattering, yielding the mechanical response function of the protein under study in the frequency domain (Wang and Zocchi, Reference Wang and Zocchi2010). This technique can be thought of as using a rheometer (Barnes and Hutton, Reference Barnes and Hutton1989) to measure the deformation of the substance under study, in this case, the protein, in nanometer scale. Analyzing the frequency response of the protein one can determine its dynamics. For GK the results showed that the amplitude of deformation decreases as 1/ $ \omega $ (the inverse of frequency of the applied force) for low frequencies and it plateaus at higher frequencies. Therefore, it was concluded that the protein can be modeled as a viscoelastic material: its mechanics are best described by a viscous fluid at low frequencies of the applied force and by an elastic spring at high frequencies. The frequency beyond which the protein shows elastic behavior is the corner frequency. Another interpretation of this result is that the system is dissipative in low frequencies and non-dissipative in high frequencies. It has been suggested (Alavi et al., Reference Alavi, Ariyaratne and Zocchi2015) that this internal dissipation can be associated to the dynamics of the hydration layer of the enzyme. It was shown that perturbing the hydration layer of GK through small amounts of chaotropic and kosmotropic agents can have a significant effect on its internal dissipation(Casanova-Morales et al., Reference Casanova-Morales, Alavi, Wilson and Zocchi2018a). Furthermore, the closed state of the enzyme (GMP bound) was shown to behave dramatically different than the open state in the viscoelastic regime while behaving similarly in the elastic regime (Wang and Zocchi, Reference Wang and Zocchi2010; Qu et al., Reference Qu, Landy and Zocchi2012). The open state is therefore ‘softer’ as it can access the soft more dissipative modes easier through the hinge motion which is related to the viscoelastic transition. In this view, the hinge would flow which can be a result of localized melting and refolding as was previously suggested in studies using structure-based coarse-grained simulations (Miyashita et al., Reference Miyashita, Onuchic and Wolynes2003; Whitford et al., Reference Whitford, Miyashita, Levy and Onuchic2007).

Figure 3.

Nanorheology setup, showing the flow chamber with enzyme-tethered GNPs (enzyme shown in green and GNPs shown as golden spheres), the parallel plates capacitor geometry used for mechanical excitation, and the evanescent wave scattering optics used for readout.

Therefore, the hinge motion can be regarded as a dissipative transition between the extended and compact conformations of the protein. When the enzyme is in its unbound state, the structure between the two conformations can be driven by applying an external oscillatory force. Then in each conformational cycle, the structure hops from one state to another on the energy landscape, this hopping is what is regarded as ‘viscoelastic transition’ in nanorheology experiments and ‘cracking’ in molecular dynamic simulations. In this view, the dissipative viscous regime is extended for the unbound state.

However, when the enzyme is in its bound state, the structure is more confined with minimal oscillations around the bound state, that is, smaller hops locally closer to the bound state on the energy landscape. One can think of this ‘viscoelastic transition’ or ‘cracking’ as local regions of the protein unfolding during conformational transitions in order to reduce the significant strain in the hinge region.

This viscoelastic model however only describes the behavior of the enzyme versus frequency and not versus force. A non-equilibrium phase diagram was obtained in the frequency force plane separating liner elasticity dynamics from softer viscoelastic dynamics (Qu et al., Reference Qu, Landy and Zocchi2012).

Therefore, enzyme’s mechanics is non-linear and displays a softening transition as a function of force as seen in experiments where the frequency of the applied force is kept constant while the amplitude of it varies. The critical force at which the transition happens depends on its frequency. Similarly, the corner frequency depends on the amplitude of the applied force. This nonlinearity can also be explained by the concept of barrier crossing in an energy landscape, similar to the problem of escape over a barrier for a particle in thermal bath which suggests local unfolding, namely cracking, can be involved in these conformational motions(Zocchi, Reference Zocchi2018). In this view, the corner frequency can be regarded as the rate of breaking the specific bond structure of the ground state confirmation of the molecule, namely the escape rate from a barrier in the energy landscape. This rate increases with the force, that is, ω_c increases as the amplitude of the external force increases. In the same view, linear elasticity regime would extend at higher frequencies: at higher frequencies (on shorter timescales) a bigger force is needed to drive the structure from one regime to another. This nonlinearity is consistent with the idea of local unfolding as a relaxing mechanism in regions with high strain. The rate of this cracking would depend on the amplitude of the applied force and a bigger force is required to have local unfolding in shorter timescales.

BiP: The master regulator of the ER

One of the most important chaperones in the ER is the molecular motor BiP protein (immunoglobulin heavy-chain Binding Protein, around 72 kDa and 8 nm long, Figure 2c). BiP (also known as HSPA5 and Grp78), a monomeric ATPase conserved across species, has been referred to as the master regulator of the ER because of the broad and crucial roles that it plays in ER processes and functions (Hendershot, Reference Hendershot2004; Alfaro-Valdés et al., Reference Alfaro-Valdés, Burgos-Bravo, Casanova-Morales, Quiroga-Roger and Wilson2018), such as protein synthesis, folding, assembly, circadian cycle and as molecular motor in translocation across the ER (Zimmermann et al., Reference Zimmermann, Eyrisch, Ahmad and Helms2011; Behnke et al., Reference Behnke, Feige and Hendershot2015; Pickard et al., Reference Pickard, Chang, Alachkar, Calverley, Garva, Arvan, Meng and Kadler2019). Although at the molecular level, the study of BiP is still in its early stages, some research groups have published findings of great value. These findings suggest that this protein could be a key player in various fields, such as in the detection and treatment of serious diseases (neurodegenerative diseases, cancer, heart diseases, among others; (Shields et al., Reference Shields, Panayi and Corrigall2012; Kosakowska-Cholody et al., Reference Kosakowska-Cholody, Lin, Srideshikan, Scheffer, Tarasova and Acharya2014; Park et al., Reference Park, Eun Kim, Morales, Moda, Moreno-Gonzalez, Concha-Marambio, Lee, Hetz and Soto2017; Ichhaporia and Hendershot, Reference Ichhaporia and Hendershot2021). Also, it has been implicated recently in the binding of the coronavirus SARS-CoV-2 spike protein to its membrane receptor (Dores-Silva et al., Reference Dores-Silva, Cauvi, Coto, Kiraly, Borges and De Maio2020; Ibrahim et al., Reference Ibrahim, Abdelmalek, Elshahat and Elfiky2020; Shin et al., Reference Shin, Toyoda, Nishitani, Fukuhara, Kita, Otsuki and Shimomura2021). Structurally, BiP is formed by two domains: a nucleotide-binding domain (NBD) with ATPase activity, connected by a flexible hydrophobic linker to the substrate-binding domain (SBD). The SBD can be further divided into a compact β-sandwich domain harboring a cleft for substrate binding and an α-helical domain at its C-terminal end, the so-called ‘lid’ (Zhu et al., Reference Zhu, Zhao, Burkholder, Gragerov, Ogata, Gottesman and Hendrickson1996). Many conformational changes, such as the opening and closing movement of the lid and the variation in the distance between the SBD and NBD, have been associated with the ATPase cycle of BiP in the ER. Once BiP binds K⁺ and ATP, its NBD and SBD domains come into close proximity to each other and the lid of the SBD opens, adopting a form that binds substrate proteins with low affinity. Also, several BiP cofactors have been discovered that assist in controlling the substrate-binding cycle and its localization within the ER (Otero et al., Reference Otero, Lizák and Hendershot2010; Braakman and Hebert, Reference Braakman and Hebert2013; Pobre et al., Reference Pobre, Poet and Hendershot2019).

Nanorheology versus optical tweezers: Complimentary techniques to study local unfolding

Nanorheology and optical tweezers are complementary techniques because they operate in different force and frequency regimes. Usually in optical tweezers (for a detailed description of this instrument please see(Smith et al., Reference Smith, Cui and Bustamante2003; Bustamante et al., Reference Bustamante, Kaiser, Maillard, Goldman and Wilson2014; Sánchez et al., Reference Sánchez, Robeson, Carrasco, Figueroa, Burgos-Bravo, Wilson and Casanova-Morales2022) the force versus extension plot starts around a few picoNewtons to 67 piconewtons and in nanorheology the force is estimated to be in the low piconewton regime (Qu and Zocchi, Reference Qu and Zocchi2013). A typical tweezers pulling is around 100 nm/sec and the stiffness is 0.1 pN/nm, then the complete pulling (if you pull up to 30 pN for example) will take around 6 sec, meaning 0.1 Hz. The force versus extension plot obtained by optical tweezer can be fit to the worm-like chain interpolation formula (Bustamante et al., Reference Bustamante, Marko, Siggia and Smith1994) which does not change much when the speed of pulling is changed. Interestingly the WLC fitting works well for dsDNA in the whole range of forces, but in the case of proteins the force versus extension plot deviates from the WLC fit at low forces (Bechtluft et al., Reference Bechtluft, Van Leeuwen, Tyreman, Tomkiewicz, Nouwen, Tepper, Driessen and Tans2007; Kaiser et al., Reference Kaiser, Goldman, Chodera, Tinoco and Bustamante2011; Wilson, Reference Wilson2011; Bianco et al., Reference Bianco, Mártonfalvi, Naftz, Koszegi and Kellermayer2015; He et al., Reference He, Li, Gao, Xiao, Hu, Hu, Hu and Li2019; Li et al., Reference Li, Chen, Guo, Wang and Li2021), probably due to localized unfolding/folding. Polypeptides may exhibit non-WLC behavior at lower forces if localized structures form during relaxation, such as beads on a string caused by hydrophobic collapse, or off-pathway folding intermediates which delay the final state.

Complementary, in Nanorheology where the frequency of the applied force is 10–200 Hz, a viscoelastic behavior is seen at lower forces which may correspond to local unfolding and help explain the deviation in the force versus extension plot obtained in optical tweezers. Currently, there is no comprehensive model for this viscoelastic transition (Zocchi, Reference Zocchi2018). Developing such a model would be important for a complete description of protein’s behavior under force.