Probing Small Molecule Binding to Unfolded Polyprotein Based on its Elasticity and Refolding

Abstract

Unfolded protein, a disordered structure found before folding of newly synthesized protein or after protein denaturation, is a substrate for binding by many cellular factors such as heat-stable proteins, chaperones, and many small molecules. However, it is challenging to directly probe such interactions in physiological solution conditions because proteins are largely in their folded state. In this work we probed small molecule binding to mechanically unfolded polyprotein using sodium dodecyl sulfate (SDS) as an example. The effect of binding is quantified based on changes in the elasticity and refolding of the unfolded polyprotein in the presence of SDS. We show that this single-molecule mechanical detection of binding to unfolded polyprotein can serve, to our knowledge, as a novel label-free assay with a great potential to study many factors that interact with unfolded protein domains, which underlie many important biological processes.

Introduction

Proteins are essential components of every living organism, performing virtually every cellular function. They have varying lifetimes within cells and are constantly degraded and recycled. A protein is biosynthesized by ribosome through sequential addition of amino acids to a growing peptide chain. The nascent polypeptide chain then folds into a unique three-dimensional structure, often assisted by molecular chaperones. Damaged or abnormal proteins, as well as proteins that have finished performing their cellular regulatory duties, are then degraded or broken down into peptide chains and/or amino acids and reused for protein synthesis.

Unfolded protein is an important substrate for binding by many cellular factors, such as chaperones during or shortly after protein synthesis for proper folding (1, 2) and ATPases associated with various cellular activities (AAA) that mediate protein degradation for protein turnover (3, 4, 5, 6). In vitro, peptides binding small molecules such as sodium dodecyl sulfate (SDS), urea, and guanidine hydrochloride (GuHCl) have been extensively used as denaturants for studies of protein stability. There are multiple biochemical methods to study binding of small molecules or chaperones to unfolded protein. 1) Proteins can be denatured using physicochemical means such as heat and chemical denaturants to generate unfolded protein (7, 8, 9), before being diluted in solutions containing binding partners; 2) one can use short peptides to study binding (10, 11); and 3) one can use protein substrate with sequence mutations that result in permanently unfolded protein (7, 12).

While these methods allow study of binding to unfolded protein, each of them has its own shortcomings. In the case of studies using denaturants to generate unfolded protein, the presence of small amount of denaturants in solution after dilution may potentially interfere with binding partners (e.g., chaperone) or with the unfolded protein structure. In the case of studies using short peptide or mutationally destabilized proteins, while they make it possible to study binding of molecules to unfolded protein in physiological solution conditions, they do not allow studies of protein refolding. In addition, these binding assays are largely based on measurements in bulk using gel electrophoresis and spectroscopy approaches such as circular dichroism and fluorescence emission spectra (7, 12, 13), which measure the average property rather than the events taking place on individual unfolded protein.

To address the limitations of the existing methods in studies involving unfolded protein binding, in this work we developed a method to detect unfolded protein binding by small molecules in solution without interference from denaturing agents or sequence mutations at a single-molecule level. Binding is directly probed at a single-molecule level based on binding-induced conformational changes and the refolding of the unfolded protein under tension using magnetic tweezers. A single polyprotein containing eight tandem repeats of titin I27 immunoglobulin domain, (I27)₈, is tethered to a coverslip surface on one end and to a paramagnetic bead on the other end, and is subject to external force, F, applied by a pair of magnets through the bead to mechanically unfold the polyprotein into unfolded polyprotein. The end-to-end distance along the force direction, referred to as “extension”, at a set of forces is then measured. The extension of the resulting unfolded polyprotein at various forces serves as a means to detect binding of partner molecules. Using this method, unfolded polyprotein is generated via mechanical unfolding and so binding can be probed in physiological solution conditions. In addition, decreasing the mechanical force allows unfolded polyprotein to refold, making it possible to probe the binding based on the effect of the binding partner on the refolding of the unfolded polyprotein.

Partner binding to the unfolded polyprotein may cause a local deformation of the unfolded polyprotein, which can be detected using single-molecule manipulation technologies. The in-house-developed magnetic tweezers used in this study can detect extension change at nanometer resolution over a long timescale (14, 15, 16, 17). Depending on the nature of the binding partner, various types of unfolded polyprotein deformation can be induced. If a binding partner wraps unfolded polyprotein or promotes attractive interaction between remote residues in the unfolded polyprotein, a shorter extension upon binding is expected. On the other hand, if binding leads to stiffening of the unfolded polyprotein backbone or steric interactions between the bound molecules, a longer extension is expected (see sketch in Fig. 1).

Figure 1.

Schematic of the magnetic tweezers setup used to stretch and probe binding to unfolded polyprotein. Molecules that bind to the unfolded polyprotein may cause deformation to the unfolded polyprotein depending on the nature of binding. If a binding partner wraps unfolded polyprotein or promotes attractive interaction between remote residues, shorter extension upon binding is expected (scenario 2) as compared to the naked unfolded polyprotein (scenario 1). On the other hand, if binding leads to increased repulsive electrostatic or steric interactions between the residues, a longer extension is expected (scenario 3).

In this work, unfolded polyprotein was generated by mechanical unfolding of a folded (I27)₈. It is to be noted that according to the updated nomenclature, I27 is now renumbered as I91 (18). The mechanical stability and unfolding kinetics of titin I27 domain have been studied extensively by atomic force microscopy (AFM) at a higher force regime and recently by magnetic tweezers at a lower force regime (17, 19, 20, 21). A recent study by magnetic tweezers revealed an equilibrium critical force of ∼5 pN, above which the I27 is thermodynamically unstable in physiological solutions, and I27 is predominantly in their unfolded state. At forces <5 pN, quick refolding of the unfolded I27 domains takes place (17).

In our study, the force-extension curve of the resulting unfolded polyprotein was characterized before introduction of a binding partner. The effects of partner binding to the unfolded polyprotein are detected based on the shift to the force-extension curve or extension difference compared to naked unfolded polyprotein, as well as protein refolding. In the subsequent sections, we describe the method to directly measure the unfolded polyprotein extension and the application of this method in probing the binding of SDS, urea, and GuHCl.

Materials and Methods

High-force magnetic tweezers were used for stretching of proteins, as described in the literature (14, 17, 22), and schematized in Fig. S1 in the Supporting Material. A pair of magnet is used to apply force on a 2.8-μm superparamagnetic bead. The height change of the bead in response to a force change can be obtained with a spatial resolution of ∼2 nm for a bead stuck on a surface. For a tethered bead undergoing thermal motion, the resolution depends on the data acquisition time and force. Larger force decreases the thermal motion and shorter acquisition time is needed to achieve a good resolution.

The height change of the tethered bead in response to force change comprises of two contributing factors—the extension change of the molecule and the bead rotation due to off-center attachment caused by anisotropic magnetization of the bead (23) (see the Supporting Material). For protein tethers with an extension much longer than the bead diameter, the contribution from the bead rotation is negligible. However, for tethers comparable to or shorter than the bead diameter, the bead rotation contributes significantly to the bead height change. Therefore, to infer the extension change of the molecule based on the force-height data of the tethered bead, the contribution from the bead rotation must be subtracted.

To circumvent this challenge in measuring the extension of unfolded polyprotein, we used the force-height data of the same bead tethered to the (I27)₈ in the folded state as a reference. I27 exhibits high mechanical stability (24, 25, 26) and it remains folded during short exposure to forces ≫5 pN. Thus it enables us to obtain the reference force-height data of the bead tethered to the folded (I27)₈ by stepwise force increase through a set of force values up to ∼60 pN (Fig. S2 A). To minimize the probability of domain unfolding during this process, we restricted the data acquisition time to 2 s at each force with a sampling rate of ∼200 Hz.

The (I27)₈ protein construct that we used in our experiments had a HaloTag on the N-terminus, followed by eight repeats of I27 domains, and a biotinylated Avi-Tag on the C-terminus. A HaloTag-(I27)₈-biotin protein is tethered in between a HaloTag-ligand-coated coverslip surface and a streptavidin-coated magnetic bead for single-molecule stretching experiments. During cycles of stretching and refolding, HaloTag domain can occasionally unfold with a maximum contour length of ∼66 nm, which can be easily distinguished from the characteristic (I27)₈ unfolding signal and hence it does not interfere with our measurements. The (I27)₈ protein tether in our experiments can exist in two distinct states: folded polyprotein and unfolded polyprotein (or unstructured polypeptide chain), depending on force applied.

The steps taken to subtract the contribution of bead rotation are as follows. We first obtained the force-height data of a bead tethered to the folded (I27)₈ before unfolding, $H_0\left(F\right)$, which is contributed from the force-extension curve of the folded polyprotein, bead rotation, and force response of any other linkers such as HaloTag in the tether. After protein unfolding, we obtained the force-height data of the same bead tethered to the unfolded (I27)₈, $H\left(F\right)$, which is contributed from the force-extension curve of unfolded polyprotein together with other contributions from bead rotation and linkers. At the same force, the orientation of an off-center attached bead remains the same before and after protein unfolding. Therefore, by subtracting $H_0\left(F\right)$ from $H\left(F\right)$, the contribution of bead rotation and linkers cancels. As a result, the difference between the heights of bead after protein unfolding and that before protein unfolding is equivalent to the extension difference between the unfolded polyprotein and the folded polyprotein. Adding back theoretical force-extension curve of folded (I27)₈, we finally obtained the extension of unfolded polyprotein:

$$z\left(F\right)=\frac{H\left(F\right)-H_0\left(F\right)+z_{\mathrm{0,8}}\left(F\right)}{8},$$ (1)

where $z\left(F\right)$ is the force-extension curve of the unfolded polyprotein normalized to one domain, and $z_{\mathrm{0,8}}\left(F\right)$ is the theoretical force-extension curve of the (I27)₈ protein with all the eight domains in the folded state. The above formula applies for a force range >5 pN at which all the unfolded I27 domains remain unfolded during the measurement of $H\left(F\right)$.

At forces <5 pN, domain refolding may take place during the measurement of $H\left(F\right)$. The number of folded domains can be known by jumping to a higher force (e.g., ∼20 pN) and comparing the bead height at this force with the reference force-height data $H_0$ (20 pN) that was recorded from the same tether before unfolding of any protein domains. The difference between the two heights divided by the step size of one domain unfolding at ∼20 pN is the number of domains refolded when the tether was held at the lower force, and eight minus this number is the number of domains that remain unfolded.

With this knowledge, the force-extension curve of unfolded polyprotein normalized to one domain is revised as follows:

$$z\left(F\right)=\frac{H\left(F\right)-H_0\left(F\right)+z_{0,n}\left(F\right)}{n},$$ (2)

where n is the number of domains that remain unfolded at the lower force, and $z_{0,n}\left(F\right)$ is the theoretical force-extension curve of the (I27)₈ protein with n domains in the folded state. The theoretical force-extension curve of n folded I27 domains, $z_{0,n}\left(F\right)$, is in between two extreme cases: 1) the domains are linked by completely free joints (i.e., zero bending rigidity), and 2) they are linked by completely rigid joints (i.e., infinite bending rigidity). The two models give almost identical results at forces >10 pN, with a maximum relative extension difference of only ∼7% at ∼4 pN (Fig. S3). Therefore, $z_{0,n}\left(F\right)$ predicted by either of the models can be used. Hereafter, the FJC model is used to obtain the force-extension curves of unfolded polyprotein:

$$z_{0,n}\left(F\right)=n\left[l_0\text{coth}\left(\frac{F{l}_0}{k_{\text{B}}T}\right)-\frac{k_{\text{B}}T}{F}\right],$$ (3)

where l₀ is the distance between N- and C- termini of a single folded I27 domain, which is ∼4 nm (17); F is the external force applied; k_B is the Boltzmann constant; and T is the temperature.

Results

Determining the intrinsic errors of the approach

Although the magnetic tweezers have a spatial resolution of ∼2 nm for height measurement of immobilized bead, here we would like to remind the readers again that the actual resolution for a tethered bead that undergoes thermal motion is compromised depending on the force applied to the bead and the acquisition time window. In the approach described above, the force-height data $H_0\left(F\right)$ were obtained by measuring the height at each force for 2 s. Furthermore, the approach is based on an assumption that the bead has only one anisotropic magnetization axis, i.e., the off-center tethered bead can only have one orientation at each force. If the tethered bead has more than one anisotropic magnetization axis, the contribution from bead rotation to the height measurement cannot be completely subtracted. Together, the approach may have a certain level of intrinsic uncertainty in height measurement.

To estimate this intrinsic uncertainty in bead height measurement, we obtained the force-height data for a bead tethered to a 572 bp double-stranded DNA in eight cycles of force increase from 2 to 40 pN through a set of force values with 2 s acquisition time at each force. As DNA does not undergo any structural transition in this force range, variations of the bead height at any of the forces must be intrinsic uncertainty associated with the approach. Fig. 2 shows the average force-height data obtained from eight force-increase cycles on the same DNA tether, with error bars indicating the standard deviations of the bead height. The result reveals significant variations at forces <4 pN. At forces >4 pN, the standard deviation is <4 nm. As such, the force range that we use to measure the force-extension curve of an unfolded polyprotein and to detect small molecule binding will be limited to forces >4 pN.

Figure 2.

Average force-height curves of a single dsDNA tether obtained from eight force-increase cycles. (Error bars) Standard deviation of bead height calculated from eight cycles of measurements. The acquisition time window is 2 s at a sampling rate of ∼200 Hz. (Inset) Intrinsic error normalized to a single amino acid as detailed in Measurement of Unfolded Polyprotein Elasticity (see main text). The normalized intrinsic error between two data points is approximated by linear interpolation.

Measurement of unfolded polyprotein elasticity

Using the procedure described in Materials and Methods, we measured the force-extension curve of single native unfolded (I27)₈ polyprotein using magnetic tweezers. As protein folding started during the 2 s holding time at forces <5 pN, the extension measured could be a mixture of unfolded polyprotein and folded polyprotein. To obtain the extension of the unfolded polyprotein, we need to determine the number of the folded domain, subtract their contribution, and normalize the extension of the unfolded region into each domain according to Eq. 2. The number of the domains refolded during the 2 s time interval when the tether was held at a low force was obtained by jumping to a higher force and comparing the difference of the extension to extension obtained from the tether before any domains are unfolded at the same force. The difference divided by the step size of one domain unfolding at this higher force, rounded to the nearest integer, gives the number of the folded protein domains (further details can be found in Fig. S2). At forces >5 pN, the protein domains had no chance to refold within the 2 s measurement time at each force. The number of refolded domains was <4 at ∼4 pN, which is the lowest force that we use for our measurements.

Having determined the force-extension curve of unfolded polyprotein normalized to a single domain, we further normalized it to a single amino acid so the result can be compared with experiments using other protein domains. Fig. 3 shows the force-extension data of the unfolded (I27)₈ per amino acid in a force range of 4–25 pN, obtained from six independent measurements on different protein tethers. The error bar on each data point indicates the interpolated intrinsic error measured in the inset of Fig. 2, normalized to a single amino acid. At forces >5 pN where I27 domains do not refold, it is calculated by dividing the intrinsic errors in Fig. 2 with 8 × 89 (each I27 domain has 89 residues). At forces <5 pN, it is calculated by dividing with 4 × 89, because a maximum of four domains was refolded in the experiments at ∼4 pN.

Figure 3.

Force-extension data (per amino acid) obtained from six independent protein tethers. (Error bars) Intrinsic errors in Fig. 2 normalized to a single amino acid. Individual data sets are fitted with the WLC model, with a persistence length value of 1.00 ± 0.14 nm (mean ± SD). To see this figure in color, go online.

Previous studies have shown that the force-response of unfolded polyprotein chain can be well described by the wormlike chain (WLC) polymer model (27). Therefore, we fitted the data in Fig. 2 to the WLC model using the Marko-Siggia formula (28), and we obtained a bending persistence length of 1.00 ± 0.14 nm (mean ± SD) and a single I27 domain contour length of 32.1 ± 1.1 nm (mean ± SD). In general, the bending persistence length agrees with previous measurement using lock-in force spectroscopy techniques that can measure force-extension curve at low-force regime and magnetic tweezers based on the unfolding transition step sizes (17, 29). Having determined the force-extension curve of naked unfolded (I27)₈, we next investigated the effect of small molecules binding to the unfolded polyprotein based on the changes in force-extension data in this force regime.

Force response of unfolded polyprotein bound with small molecules

In this section, we demonstrate the use of this single-molecule approach to directly probe small molecule binding by investigating the effects of SDS, urea, and GuHCl binding to unfolded polyprotein, methods widely used to denature protein. Such binding can potentially alter the elastic properties of the unfolded polyprotein, which can be detected by studying the force response of the unfolded polyprotein before and after introduction of these molecules.

First, we probed binding of SDS, which is known to disrupt noncovalent bonds in protein structure, thus denaturing the protein and resulting in unfolded polyprotein. We probed binding of SDS at concentrations of 0.001% up to 1% in PBS buffer (Fig. 4 A). At 0.001% SDS, the force-extension curve is similar to that of the naked unfolded polyprotein, indicating negligible effect of SDS on the force response of unfolded polyprotein at this concentration (orange data). At 0.01% SDS, significant reduction in extension is seen at forces <10 pN (blue data). Further increase of SDS concentration to 0.1% results in further decrease of extension in the force range used (green data). At 1% SDS, the extension is slightly longer compared to that obtained in 0.1% SDS at force range of 4–8 pN (yellow data). The inset shows the effect of SDS on the extension at a force of 6 pN, by plotting the extension difference from the naked unfolded polyprotein at different SDS concentrations.

Figure 4.

Effects of SDS binding on unfolded polyprotein force-extension curve. (A) Representative data obtained from a single protein tether at various concentrations of SDS in PBS buffer. Error bars represent intrinsic errors normalized to one amino acid. (Inset) Extension difference between an SDS-(unfolded-polyprotein) complex and an unfolded polyprotein at 6 pN obtained from multiple independent experiments. (B) Schematic diagram of the possible conformations adopted by SDS-(unfolded-polyprotein) complex at low concentration of SDS (scenario 2, wrapped mode), and high concentration of SDS (scenarios 3a and 3b, less wrapped micelle conformation and nonmicellar reorganization, respectively). To see this figure in color, go online.

The above results indicate a complex dependence of the unfolded polyprotein elasticity on SDS binding, which reveal several pieces of important information related to the nature of the interaction. The extension reduction observed in the force range tested in the experiments indicates that SDS is capable of wrapping the unfolded polyprotein in a low force regime. In addition, the level of unfolded polyprotein wrapping by SDS decreases as SDS concentration increases from 0.1 to 1% at forces <8 pN, which can be explained by a switch to a less wrapped binding mode that allows binding of more SDS molecules, a mechanism analogous to that observed for binding of DNA bending/wrapping proteins (30, 31). Fig. 4 B schematizes the possible conformations of SDS-(unfolded-polyprotein) complex that can explain our data (see Discussions for details).

We also probed the effect of urea and GuHCl binding to unfolded polyprotein. Although the nature of interactions between these chemical denaturants and unfolded polyprotein remains elusive, it is believed that the urea/GuHCl penetrate protein hydrophobic core by dispersion interactions and disrupt the tertiary structure by hydrogen bonding (32, 33, 34, 35, 36). In our experiments, we varied the concentration of urea from 8 to 800 mM and GuHCl from 6 to 600 mM in PBS buffer (see Fig. S4). In these concentration ranges, small shifts in the force-extension curve after introducing urea and GuHCl are observed. Considering potential interference from partial HaloTag unfolding, we cannot conclude such small shifts indicate binding of urea and GuHCl. Overall, these results likely indicate that urea and GuHCl in these concentration ranges do not strongly bind to the unfolded (I27)₈ in the range of concentrations used, or that their binding does not significantly affect the elastic property of the unfolded polyprotein. We could not increase the urea and GuHCl concentrations to several molars that are typically used in protein denaturation experiments, because at such high concentrations the refractive index of the solution increases significantly and prevented us from detecting high-resolution extension change data (37, 38).

We also note that in the presence of denaturants, we did not see significant increase in tether contour length, which indicates that the HaloTag protein fused in our (I27)₈ construct for surface immobilization (see Materials and Methods) is either not unfolded during experiments or is already unfolded. In the former scenario, the presence of denaturants does not affect the force-extension curve. In the latter scenario, the changes in the force-extension curve in the presence of denaturant is a contribution of both (I27)₈ and the HaloTag protein. When fully unfolded, the HaloTag protein has a maximum contour length of ∼66 nm. The contribution of HaloTag is small, however, relative to (I27)₈, because unfolded (I27)₈ has a much longer total contour length of ∼240 nm.

Repression of protein refolding by SDS

The binding of small molecules to unfolded polyprotein also suggests repression of protein refolding. Therefore, this information can supplement the observation of unfolded polyprotein binding that we probed using force-extension curves in the previous section. To determine the effects of SDS on refolding of unfolded (I27)₈ protein, we performed multiple cycles of protein unfolding-refolding experiments in the absence and presence of SDS. Each cycle includes a force-increase scan from 1 to 80 pN at a loading rate of 2 pN/s. The tether was then held at 80 pN or larger until all domains were unfolded. After this, the force was reduced to 1 pN by increasing the bead-magnet distance with a speed of 150 μm/s. In this force-decrease procedure, the force decreases exponentially with time. The time taken to decrease the force from 80 to 5 pN is ∼12 s, thus allowing SDS to bind to the unfolded polyprotein as protein refolding was mechanically inhibited in this force range. It took another ∼13 s to further decrease the force to 1 pN, during which refolding may occur. The tether was held at 1 pN for an additional 10 s to give more time for protein refolding.

Fig. 5 A shows representative domain unfolding in three force-increase time traces in PBS buffer obtained in successive force cycles. Each force-increase time trace shows eight unfolding steps indicating refolding of all domains at forces <5 pN during the preceding force-decrease cycle. When such experiments were repeated in the presence of 0.001% SDS, fewer unfolding steps were observed (Fig. 5 B, orange data). At 0.01% SDS, domain unfolding is no longer observed (Fig. 5 B, blue data). These data with fewer unfolding steps are correlated with longer extensions compared to that obtained in the absence of SDS. These results indicate that SDS binding strongly inhibits protein refolding at low forces. The suppression effect of SDS on protein refolding was quantified in Fig. 5 C. In the absence of SDS, ∼95% of (I27)₈ domains are refolded during the unfolding-refolding cycle described in the preceding paragraph. In the presence of 0.001% SDS, the refolding percentage dropped to ∼78%, indicating slight repression of domain refolding due to SDS binding. Increasing the concentration of SDS to 0.01 and 0.1% results in an almost complete loss of protein refolding. Here we note that the above SDS concentration-dependent percentage of refolded I27 domains should not be interpreted as the equilibrium probability of refolding, because the measurement is nonequilibrium in nature. While these data show that SDS in the concentration range tested significantly suppresses protein refolding, it does not provide information on the equilibrium melting concentration.

Figure 5.

Effects of SDS on (I27)₈ refolding. (A) (Upper panel) Three force-increase time traces of extension obtained in three consecutive force cycles, each showing unfolding of (I27)₈ domains indicated by characteristic stepwise extension increases. (Lower panel) Force values during the force-increase scans. (B) Representative unfolding time trace data obtained from a single protein tether. (Upper panel) Three force-increase time traces of the extension of an (I27)₈ protein tether in the absence (black data, bottom) and in the presence of 0.01% (orange data, middle) and 0.1% SDS (blue data, top). (Lower panel) Force values during the force-increase scans. (C) Percentage of (I27)₈ domains that were refolded in the absence and presence of SDS. The statistics was obtained from multiple force-cycle experiments of five independent protein tethers at each condition. The error bars represent standard deviations obtained from all the unfolding-refolding cycles. To see this figure in color, go online.

In summary, we have developed a method to directly stretch a single unfolded polyprotein, and probe the binding of small molecules to the unfolded polyprotein based on the change in force-extension curves of the unfolded polyprotein. We demonstrate the use of this method to probe binding of SDS, urea, and GuHCl to unfolded polyprotein, and detected binding by SDS. The binding to unfolded polyprotein is confirmed with the concurrent observation of protein refolding suppression in the presence of SDS. These methods thus offer, to our knowledge, a novel and generic label-free method for studying direct molecular binding to unfolded polyprotein in physiological solution condition.

Discussion

In this study we demonstrate, to our knowledge, a novel method that allows us to measure the force-extension curve of an unfolded polyprotein in a low force regime using magnetic tweezers, and its application in detecting small molecule binding. Our approach is different from most AFM stretching measurements in which a polyprotein is tethered in between the tip of the cantilever and a surface. In a typical stretching procedure using AFM, the tip of the cantilever is moved away from the surface, resulting in the end-to-end distance of the tether increasing in a controlled manner. In such experiments, the external control is a constraint on the extension of the tether, and the force change is a response to changes in extension. In this extension-constraint mode, the extension can be kept constant or varied with time. This is in contrast to force constraint, where the external constraint is the force and the extension change is the response. In force-constraint mode, the force can be kept constant or varied with time. Magnetic tweezers apply force constraint on tethered molecules.

Force constraint offers an advantage of applying equal tension throughout the protein construct (due to force balance) regardless of its structural state (folded or unfolded) or whether it is bound with a ligand. This is in contrast to the extension constraint in AFM measurements, where a local structural change or ligand binding/unbinding causes a global change of the tension in the construct, resulting in coupling of the states of all domains in the construct. This advantage of force constraint is particularly useful in studies of complex interactions between a tethered protein and ligands, which was demonstrated in our previous protein unfolding experiments (22, 39, 40) and is demonstrated in this work through studies of unfolded polyprotein interaction with SDS.

The use of force to unfold protein and study binding to unfolded polyprotein also differs from that of using denaturant. The apparent advantage of mechanical unfolding is that protein can be unfolded in physiological solution conditions. A notable difference between mechanically and chemically unfolded protein is that the former is extended with no significant contacts between different residues, while the latter is typically in a partially extended state with many nonnative contacts (41). As such, care should be taken during comparison between studies involving unfolded polyprotein binding, binding kinetics, and protein folding obtained using the two methods (42).

We show that it is possible to directly detect SDS binding to unfolded polyprotein based on its effect on the force-extension curve of the unfolded polyprotein. It is known that upon SDS treatment, the protein conformation is denatured and the backbone is negatively charged. It has been suggested that SDS binding to unfolded polyprotein is predominantly due to hydrophobic interaction between the dodecyl chains of the detergent with hydrophobic region of the protein rather than by the interaction between the polar headgroup to polar amino acid residues (43). The conformation of SDS-(unfolded-polyprotein) complex is less clear, however, and multiple models have been proposed, including rodlike particle (44), necklace (45, 46, 47, 48), and flexible helix models (49). It was concluded based on small-angle neutron scattering data and NMR experiments that a necklace-and-bead structure is the most consistent model to explain these observations (47, 48), in which globular micelles are connected with short polypeptide segments.

In this work we probed binding of SDS to unfolded polyprotein at a single-molecule level. In the buffer condition that we used in our experiment (PBS buffer), the critical micelle concentration is ∼0.027–0.042% (50). The extension reduction at >0.01% SDS in the force range from 4 to 20 pN is consistent with wrapping of unfolded polyprotein around the micelles (Fig. 4 B, scenario 2). We also found that increasing SDS concentration from 0.1% to 1% leads to a slightly decreased level of extension reduction in the force range of 4–8 pN, which can be explained by accommodation of more SDS micelles in the unfolded polyprotein through a switch to a less wrapped conformation for each bound SDS micelle (Fig. 4 B, scenario 3a). This behavior of switching from a wrapped mode to a less wrapped mode is also seen previously with DNA bending/wrapping proteins IHF and SSB (30, 31). Alternatively, there may be reorganization of SDS molecules into a nonmicellar conformation along the unfolded polyprotein to maximize binding occupation (Fig. 4 B, scenario 3b). The binding of SDS detected on the force-extension curve measurement is also reflected by its strong suppression on the refolding of the unfolded polyprotein at low forces when SDS concentration was increased to 0.01%, which is near its critical micelle concentration.

In our studies of urea and GuHCl in the submolar concentration range, we did not see a significant effect on binding to the unfolded polyprotein. This may indicate the binding is too weak at these concentrations, which is consistent with previous works showing that urea and GuHCl decrease the mechanical stability of the folded protein domains only at concentrations in the molar range (51, 52, 53, 54, 55).

Although the application of this method is demonstrated in this study for small molecule binding, we expect this approach may be even more sensitive in probing binding of proteins such as chaperones to unfolded polyprotein. Compared to small molecules, they have much larger molecular weight. Thus, they may elicit a much greater impact on the elastic property of unfolded polyprotein due to the significant distortion to the peptide at the binding site and steric interactions between adjacent chaperone proteins bound to the unfolded polyprotein. As chaperone proteins play critical biological functions in mediating proper protein folding, this label-free single-molecule binding assay may find important applications in investigation of the molecular mechanisms of chaperones.

Author Contributions

R.S.W. and Q.T. performed the experiments; J.C. measured the intrinsic error of the approach; J.Y. conceived the research; R.S.W. and J.Y. designed the experiments and interpreted the data; M.Y. contributed to experiments at the initial stage; R.S.W. and J.Y. wrote the article; and all authors read and commented on the article.

Editor: Jason Kahn.