Monitoring the conformational changes of an intrinsically disordered peptide using a quartz crystal microbalance

Abstract

Intrinsically disordered peptides (IDPs) have recently garnered much interest because of their role in biological processes such as molecular recognition and their ability to undergo stimulus-responsive conformational changes. The block V repeat-in-toxin motif of the Bordetella pertussis adenylate cyclase is an example of an IDP that undergoes a transition from a disordered state to an ordered beta roll conformation in the presence of calcium ions. In solution, a C-terminal capping domain is necessary for this transition to occur. To further explore the conformational behavior and folding requirements of this IDP, we have cysteine modified three previously characterized constructs, allowing for attachment to the gold surface of a quartz crystal microbalance (QCM). We demonstrate that, while immobilized, the C-terminally capped peptide exhibits similar calcium-binding properties to what have been observed in solution. In addition, immobilization on the solid surface appears to enable calcium-responsiveness in the uncapped peptides, in contrast to the behavior observed in solution. This work demonstrates the power of QCM as a tool to study the conformational changes of IDPs immobilized on surfaces and has implications for a range of potential applications where IDPs may be engineered and used including protein purification, biosensors, and other bionanotechnology applications.

Introduction

Intrinsically disordered peptides (IDPs) are of interest as they play critical roles in biological processes without well-defined three-dimensional structures. This is typically accomplished by transitioning from a disordered to an ordered state on interaction with a binding partner or a change in environment.^1–3 Examples of IDP's include certain domains of p53, phosphorylated kinase-inducible domain (pKID), neurofilament H, and others. ^4–7 The ability to control the transition between different structural states presents an opportunity for their utilization in applications such as protein purification, biosensors, development of “smart” therapeutics, and other bionanotechnologies. ^8,9 For example, the elastin-like peptides undergo a change in structure on changes in temperature and have been employed in applications including protein purification and temperature-controlled binding. ^10,11 The ability to switch between a disordered and an ordered structure with an intrinsic binding site may be useful in biosensor development where the ability to modulate binding is desired. Specifically, if an IDP with stimulus-dependent binding were immobilized in a functional form on a surface, this surface could be used as a substrate for the controlled binding (and perhaps release) of a target analyte. To this end, we are exploring the use of IDPs as potential scaffolds for engineered biomolecular recognition.

The repeat-in-toxin (RTX) peptide motif is found in certain secreted bacterial proteins.^12,13 The domain is intrinsically disordered in the absence of calcium, but forms a beta roll structure in the presence of calcium ions. ¹⁴ Previous work has characterized the end-capping requirements for calcium-responsive folding of block V of the RTX domain from the adenylate cyclase of Bordetella pertussis and a C-terminal capping domain has been found to be essential for calcium responsiveness. ^15,16 In the presence of calcium, aspartate residues in each repeat bind calcium and a beta roll structure is formed. ^17,18 Folding appears to occur with the highest affinity and cooperativity when the native C-terminal sequence is present, but has also been shown to occur when other well-folded domains are fused to the C-terminus. Caps may shield the hydrophobic core of the beta roll from solution, and it has been hypothesized that the C-terminal cap also acts as a nucleator of a polarized folding process which proceeds from C-terminus to N-terminus in the presence of calcium. ¹⁵ This is similar to what has been observed in other repeat proteins, such as the leucine rich repeats of internalin B, which fold in an N-terminally polarized fashion. ¹⁹ As multiple, unrelated proteins enable folding of the beta roll, we previously proposed that folding is enabled via entropic stabilization of the C-terminus. ¹⁵ Further, we reasoned that alternative nonprotein caps, including a solid surface, may be capable of enabling calcium-responsive structure formation.

IDPs are a promising scaffold for engineering biomolecular recognition and will likely find utility in a variety of applications. However, study of surface immobilized IDPs is difficult due to the incompatibility of most spectroscopic characterization techniques with such an environment. The quartz crystal microbalance (QCM) is a sensitive and versatile tool that has been applied to detect biomolecular interactions and conformational changes in peptides and proteins.^20,21 For example, the calcium-induced conformational change of calmodulin has been studied by QCM, where it was shown that the dehydration process and increased rigidity of the protein on calcium binding was detectable as an increase in frequency. ²² Considering the significant conformational change of the RTX domain in response to calcium, we hypothesized that this change would be detectable by QCM and that its behavior would depend on the N-terminal or C-terminal orientation of the peptide due to previous experiments where such dependencies have been observed. ²³ Here, we use QCM to study conformational changes of capped and uncapped RTX peptides and explore the ability of a surface to act as an entropic cap, enabling calcium-responsive folding in peptides which are not capable of folding in solution.

Discussion

We first sought to confirm that three cysteine-modified RTX constructs [Fig. 1(a)] behaved, in solution, the same as previously characterized noncysteine-modified constructs.¹⁵ We wanted to be sure that the presence of salt did not alter the behavior of the peptides, because a higher salt environment was required for QCM experiments, and that the addition of the cysteine amino acids could lead to dimerization which could alter the beta roll behavior. The uncapped cys-RTX and RTX-cys constructs showed no change in secondary structure by circular dichroism (CD) in up to 100 mM CaCl₂, while the C-terminally capped cys-RTX-C construct exhibited an increase in secondary structure between 0 and 10 mM CaCl₂, similar to what has previously been observed. ¹⁵ Therefore, calcium titrations were performed with this C-terminally capped construct while following the CD signal at 220 nm to monitor β-sheet formation [Fig. 2(a)]. Data were fit to the Hill equation, yielding a dissociation constant (K_D) of 1340 μM and a Hill coefficient of 3.26 (Table 1), similar to previous observations. ¹⁵

Figure 1.

(a) Sequence of RTX and suspected calcium-bound conformation. cys-RTX-C consists of a cys residue at the N-terminus and includes the gray capping sequence at the C terminus. cys-RTX and RTX-cys possess cys residues at the N-terminus and C-terminus, respectively. (b) Frequency and resistance changes of cys-RTX-C on the QCM. Changes in the buffer composition are noted by arrows.

Figure 2.

(a) Titrations of cys-RTX-C with varying concentrations of calcium performed by measuring changes in tryptophan fluorescence, CD signal at 220 nm, and QCM. (b) Calcium titrations of cys-RTX, RTX-cys, and cys-RTX-C using QCM.

Table I.

Dissociation Constants (K_D) and Hill Coefficients (n) for the Three Constructs as Determined by QCM, CD, and Tryptophan Fluorescence

Construct	Technique	KD, μM	Hill coefficient, n
cys-RTX	QCM	5140 ± 160	1.45 ± 0.06
RTX-cys	QCM	4190 ± 390	0.94 ± 0.05
cys-RTX-C	QCM	728 ± 9	2.29 ± 0.06
cys-RTX-C	CD	1340 ± 60	3.26 ± 0.38
cys-RTX-C	Trp fluorescence	147 ± 5	1.43 ± 0.06

cys-RTX-C possesses a pair of tryptophan residues in the C-terminal cap, which can be used to track conformational changes in this region of the peptide. Therefore, an additional calcium titration was performed while tracking tryptophan fluorescence [Fig. 2(a)]. A K_D of 147 μM and a Hill coefficient of 1.43 were obtained (Table 1). In this case, although the Hill coefficient is low, cooperativity is a less meaningful parameter because the tryptophan signal is likely only reporting the conformational behavior of the cap. From these results, it seems that the C-terminal region of the peptide undergoes a conformational change at a concentration which is far lower than for the overall peptide. This suggests that the C-terminal region is the first to respond to calcium. The K_D obtained by tryptophan fluorescence is nearly identical to those that we have reported previously, despite the higher salt concentration.¹⁵ Using a Forster resonance energy transfer (FRET) construct, we previously reported a salt dependency on the binding affinity of the RTX-C construct. ²⁴ The lack of a salt dependency when tracking conformational changes in the cap via tryptophan fluorescence suggests that the RTX region is primarily responsible for the salt dependency of the binding affinity.

Next, we sought to determine if calcium-dependent conformational changes of the RTX domain could be detected by QCM. Therefore, we examined the cys-RTX-C construct, which was shown to respond to calcium in solution. This peptide was expected to undergo a significant and easily detectable conformation change and was tested in calcium-free and calcium-rich solution. The frequency change on immobilization to the gold modified QCM crystal corresponds to a mass loading as comparable to what has been observed in previous work, as computed by the Sauerbrey equation.²¹ On addition of 10 mM CaCl₂, a change in both frequency (12 Hz) and resistance (−1 ohm) was observed [Fig. 1(b)]. According to our previous work, using a manufacturer provided relationship, a 1-ohm resistance change contributes about 2 Hz to the total frequency change. ²¹ Therefore, most of the frequency change (∼10 Hz) is because of loss of mass at the surface, suggesting that a dehydration process likely takes place on beta roll formation. After switching to calcium-free buffer, a return to baseline was observed. This demonstrates that calcium binding is readily reversible, in agreement with previous results. The calcium-induced folding and unfolding process can be repeated on the same crystal, and very similar frequency responses were observed the second time (not shown). This is similar to the repeated folding and unfolding of the beta roll reported in solution. ²⁴ To determine whether the change is specific to calcium or an ionic effect, the experiment was repeated using 10 mM MgCl₂. No change in resistance and only a small (∼2 Hz) change in frequency was observed [Fig. 1(b)], possibly because of cation replacement. This supports the conclusion that the signal change is due to beta roll formation and is consistent with the observation of calcium-specific binding of the beta roll in solution. ²⁴

Next, we tested the uncapped cys-RTX and RTX-cys constructs. We hypothesized that cys-RTX would not be calcium responsive, as immobilization occurs on the N-terminus, while RTX-cys could be responsive, due to an entropic stabilization effect by the mass of the surface at the C-terminus. Again, the frequency change observed on incubation of these constructs with the QCM surface indicates robust attachment (not shown). Interestingly, on exposure to calcium, a frequency increase was observed for both peptides (5 Hz for cys-RTX and 7 Hz for RTX-cys). This effect was found to be specific to calcium and did not occur with magnesium. However, no significant change in resistance was observed for either construct. The calcium specificity of the frequency change suggests beta roll structure formation, while the lack of a resistance change may be due to the fact that these peptides are smaller in size and, therefore, do not undergo as significant a conformational change. Also, the frequency change cannot be attributed to nonspecific interactions with the QCM surface because this would result in a drop in frequency (due to mass gain), as was observed in testing of the system without an immobilized peptide (not shown).

Together, these results suggest that the surface of the QCM crystal provides the entropic stabilization required for beta roll formation independent of which terminus is immobilized. Protein domains are very different end-capping groups when compared with a solid surface and it is likely that the surface provides substantially more entropic stabilization. The case of a polymer approaching a solid surface has been addressed in detail by Chan et al., demonstrating that substantial conformational changes can be induced by the entropic constraint and we are likely observing a similar phenomenon here.²⁵ Also, the peptides are closely packed on the surface in comparison with the solution phase, providing additional confinement on the entropy of the peptides. This is analogous to the scenario where the peptide is in a high concentration polyethylene glycol (PEG) solution. It has previously been reported that in 50% PEG and 100 mM CaCl₂, the uncapped RTX peptide undergoes a CD-detectable conformational change. ¹⁵ However, although the results suggest that a conformational change is taking place in both tethering situations, one cannot determine whether the final folded structure is the same as that achieved with the native capping group.

To further understand the folding process, we performed calcium titrations on all three immobilized constructs [Fig. 2(b)]. The calculated Hill coefficients (n) and K_D for binding are summarized in Table 1. The uncapped constructs, although capable of undergoing a conformational change when immobilized, possess lower binding affinities as compared to cys-RTX-C. Further, the Hill coefficients are both lower than that of the capped construct, suggesting an alternative folding mechanism.

These results suggest that the native C-terminal cap of the RTX domain performs two important functions: (1) it provides entropic stabilization to allow for folding in solution and (2) it acts as a nucleator of folding, allowing for cooperative calcium binding. The fact that the uncapped constructs respond to calcium, but with lower affinity and cooperativity, suggests that the solid substrate, and perhaps the alternative protein caps reported previously,¹⁵ achieve the first function but not the second function as well as the native cap. It is also possible that beta roll formation proceeds through different folding mechanisms. Specifically, when the native cap is present, folding proceeds in a polarized fashion beginning from the C-terminus. However, when the native cap is replaced by a globular protein or solid surface, calcium ions bind, but the binding may not occur in a specific cooperative order. Further investigation will be necessary to explore how the native capping group interacts with the adjacent beta roll repeats and why this leads to a significant improvement in cooperativity and calcium affinity.

The increase in calcium affinity of the immobilized construct, as compared to solution, is likely due to additional surface entropic stabilization of the peptide, whereas the decrease in cooperativity is may be due to the loss of flexibility in the N-terminus of the peptide. It is possible that more cooperative folding, although requiring entropic stabilization, also requires some additional flexibility in the RTX region.

Here, we have demonstrated that a solid substrate can be used to entropically stabilize an RTX peptide and allow for calcium-responsive folding that is not observed in solution. Further, we have demonstrated the utility of QCM in studying conformational changes in immobilized peptides and have obtained further insight into the folding mechanisms of RTX domains. We demonstrate that the native capping group likely plays two roles, entropic stabilization and folding nucleation. Non-native caps can provide entropic stabilization, but not necessarily folding nucleation, resulting in reduced cooperativity. These results should provide more insight for future efforts to immobilize IDPs on solid surfaces for a variety of applications.

Materials and Methods

The RTX and RTX-C constructs reported previously were used as templates for preparation of the cysteine-modified constructs.¹⁵ Cysteine residues were introduced on the N-terminus of the RTX-C construct using the Quikchange site-directed mutagenesis kit (Stratagene, Santa Clara, CA) to create cys-RTX-C. The same protocol was used to create the cys-RTX and RTX-cys constructs. Expression and purification of the three constructs followed a previously described protocol. ¹⁵

QCM experiments were performed as described previously.²¹ All experiments were carried out in Tris buffered saline (TBS) (50 mM Tris-HCl, 150 mM NaCl, pH 7.4). Calcium chloride and magnesium chloride were used as the calcium and magnesium ion sources, respectively, and added as indicated. Cysteine backfilling was performed after immobilization of the peptide on the QCM surface to prevent nonspecific surface interactions.

CD experiments were performed using a J-815 CD spectrometer (Jasco, Easton, MD) equipped with a Peltier junction temperature control. Protein concentrations of 75–100 μM were analyzed in a 0.01-cm path length quartz cuvette at 20°C. Fluorescence measurements were performed in the same machine equipped with a FMO-427S fluorescence monochromator. Protein concentrations of 1 μM in a 1-cm path length quartz cuvette were excited at 280 nm and the fluorescence signal was followed at 340 nm.

All titration data were fitted to the Hill equation using Origin 8.0 (OriginLab, Northampton, MA). Fraction folded was determined by scaling all data to the endpoints of the titrations. Errors are reported as standard deviation.