Modulating long‐range energetics via helix stabilization: A case study using T4 lysozyme

Abstract

Cooperative protein folding requires distant regions of a protein to interact and provide mutual stabilization. The mechanism of this long‐distance coupling remains poorly understood. Here, we use T4 lysozyme (T4L*) as a model to investigate long‐range communications across two subdomains of a globular protein. T4L* is composed of two structurally distinct subdomains, although it behaves in a two‐state manner at equilibrium. The subdomains of T4L* are connected via two topological connections: the N‐terminal helix that is structurally part of the C‐terminal subdomain (the A‐helix) and a long helix that spans both subdomains (the C‐helix). To understand the role that the C‐helix plays in cooperative folding, we analyzed a circularly permuted version of T4L* (CP13*), whose subdomains are connected only by the C‐helix. We demonstrate that when isolated as individual fragments, both subdomains of CP13* can fold autonomously into marginally stable conformations. The energetics of the N‐terminal subdomain depend on the formation of a salt bridge known to be important for stability in the full‐length protein. We show that the energetic contribution of the salt bridge to the stability of the N‐terminal fragment increases when the C‐helix is stabilized, such as occurs upon folding of the C‐terminal subdomain. These results suggest a model where long‐range energetic coupling is mediated by helix stabilization and not specific tertiary interactions.

Introduction

Cooperativity is a hallmark of globular proteins. At equilibrium, many small (<200 residues) globular proteins fold in an apparent two‐state manner (U⇌N), populating either a completely folded or unfolded conformation.1, 2 The stability of the folded protein (ΔG _unf = G _U–G _N) is the sum of many interactions, both local and long range. While the chemical nature and mechanism of short‐range, local interactions in cooperativity can be easy to identify and investigate, it is difficult to probe how cooperativity occurs between structurally distinct regions. For repeat proteins, such as ankyrin domains, this coupling has been attributed to a large interfacial energy that, in some cases, overshadows the lack of intrinsic stability of the individual repeats.3, 4

Here, we explore coupling between distant regions of a protein using the protein T4 lysozyme* (T4L*, * denotes the cysteine‐free pseudo‐wild‐type variant5). T4L* is a well‐studied globular protein; the stabilities and structures of hundreds of single‐site mutants have provided a wealth of information about the role of side‐chain interactions and their effect on protein stability.6 The protein is composed of two distinct subdomains [Fig. 1(A)]: the N‐terminal subdomain (residues 13–74) and the C‐terminal subdomain (residues 75–164 and 1–12).7, 8 The full‐length protein unfolds in an apparent two‐state manner with no notable populated intermediates (ΔG _unf ~14 kcal/mol).8, 9 In isolation, however, the C‐terminal subdomain folds autonomously into a marginally stable structure (2.1 kcal/mol), while the isolated N‐terminal subdomain has been reported to be predominately unfolded.8 Together, these results indicate that in the context of the full‐length protein, folding of the domains is strongly coupled.

Figure 1.

CP13* structure, subdomain architecture, and constructs. (A) Ribbon diagram of CP13*. N‐terminal subdomain (dark blue) (PDB ID http://firstglance.jmol.org/fg.htm?mol=2O4W) and C‐terminal subdomain (gray) are connected by the long central C‐helix. Residues involved in salt bridge are colored His31 (orange) Asp70 (green). (B) Contact map of CP13*. Inter‐residue contacts in the N‐terminal subdomain in blue. Inter‐residue contacts in C‐terminal subdomain in gray. (C) Primary sequence representation of T4L*, CP13*, NTF*, LNTF*, and NTF*‐αH. N‐terminal subdomain in dark blue and C‐terminal subdomain in gray. Extended N‐terminal subdomain fragments LNTF (with residues 75–81 in light blue) and NTF‐αH with alanine helix residues in red

The nature of the strong coupling between the two subdomains is difficult to discern as almost all of the side‐chain contacts in T4L* are within the individual subdomains7, 8 [Fig. 1(B)]. Thus, communication between the subdomains must derive from the topology or backbone connections between the two. There are two such connections: one at the end of the A‐helix and one within the C‐helix (Fig. 1). The A‐helix (residues 1–12) is structurally part of the C‐terminal subdomain, making this subdomain discontinuous in sequence8 and connects the two subdomains between residues 12 and 13. The C‐helix is the central helix that spans both subdomains [residues 59–81, Fig. 1(A)].

The role of this discontinuous subdomain architecture (the placement of helix A) was explored using a circular permutant, CP13*8, 9, 10, 11 [Fig. 1(A)]. CP13* begins at residue 13 with the A‐helix (residues 1–12) appended to the C‐terminus via a short flexible linker.8 Thus, in CP13*, the C‐terminal subdomain is contiguous in sequence. In spite of this rearrangement, CP13* and T4L* have similar structures (Root‐mean‐square‐deviation (RMSD)<1.0 Å)10 and CP13* retains two‐state behavior at equilibrium.8, 9, 10 Thus, the discontinuous subdomain architecture of T4L* is not fully responsible for the coupling between the subdomains. It should be noted, however, that this permutation does have a small effect on the coupling between the two subdomains as determined by native‐state hydrogen exchange (NSHX)10 and single molecule force spectroscopy.11 NSHX studies of the full‐length protein identified a rarely populated, high‐energy, partially unfolded conformation under native conditions that consists of an extended folded C‐terminal subdomain and an unfolded N‐terminal subdomain,12 and experiments with CP13* revealed an increase in the population of this intermediate.10 Thus, the discontinuous subdomain architecture of T4L* aids in coupling the two regions, but it is not the whole story. Interestingly, a longer version of the C‐terminal subdomain that includes the entire C‐helix greatly enhances the stability of the C‐terminal subdomain, implicating the C‐helix itself as a potential stabilizer.10

One of the most important side‐chain interactions within T4L* is a buried salt bridge between His31 and Asp70—a partially buried interaction within the N‐terminal subdomain of the protein. This interaction stabilizes the native state of T4L* and CP13* by 3–5 kcal/mol.8, 13 In the present work, we identify an additional role for this salt bridge: coupling the two subdomains of T4L*. We revisit the isolated N‐terminal subdomain and find that it is folding competent in conditions that promote the formation of the H31‐D70 salt bridge and its stability is dependent on this interaction. We also find that, similar to what we observed for the C‐terminal subdomain, extending the N‐terminal subdomain fragment to include the last seven residues of the C‐helix, which are part of the C‐terminal subdomain, stabilizes the interaction between these two residues and increases the stability of the N‐terminal subdomain. This stabilizing effect can be recapitulated by adding an autonomously folding alanine‐based helical sequence.

One of the partners in this salt bridge, residue 70, resides in the C‐helix and is part of the N‐terminal subdomain. We propose that helix stabilization provided by interactions at the other end of the C‐helix (in the C‐terminal subdomain) correctly orients D70 with respect to H31, thereby increasing its effective concentration and the salt‐bridge stability. Thus, the C‐helix‐mediated coupling between the two subdomains appears to be transmitted via the helical backbone. This mechanism—coupling via stabilizing a helix and the cooperative nature of the helix–coil transition—may be a general feature of long‐range communication within proteins.

Results

The isolated N‐terminal subdomain folds in a pH‐dependent manner

A fragment encoding the N‐terminal subdomain (residues 13–74 of T4L*, NTF*) was cloned, expressed and purified for biophysical studies (see Materials and Methods section). Previous studies on a similar construct characterized the fragment as “predominantly unfolded,”8 but we find that under slightly different experimental conditions (pH 5 versus 4.4), the fragment is folded as monitored by far‐ultraviolet (UV) circular dichroism (CD) [Fig. 2(A)]. The urea‐induced equilibrium denaturation (monitored by the change in CD signal at 222 nm, pH 5, 4 °C) shows a sigmoidal transition, indicating cooperative unfolding [Fig. 2(B)]. Although the transition is very broad, this denaturation curve can be fit to a two‐state model,14 yielding an extrapolated free energy (ΔG _unf(H2O) of 1.60 ± 0.13 kcal/mol and an m‐value =1.13 ± 0.01 kcal/mol*M [Fig. 2(B), Table 1]. The m‐value agrees with expectations based on the number of residues in the fragment (m‐values have been found to correlate with buried surface area and size of a given polypeptide15), consistent with folding of the entire fragment.

Figure 2.

NTF* folds into a marginally stable structure and extending its sequence alters its structure and energetics. (A) CD spectrum of NTF* at pH 5 (blue solid line) and pH 7 (blue dashed line) NTF* H31N (orange line) NTF* D70N (green line). (B) Representative equilibrium denaturation curves of NTF* and variants monitored by CD at 222 nm at pH 5, of NTF* (blue circle), NTF* H31N (orange circles), and NTF* D70N (green circles). Inset of NTF* urea denaturation melt at pH 7 (blue circles). (C) CD spectra of LNTF* at pH 5 (light blue solid line) and pH 7 (light blue dashed line), LNTF* H31N at pH 5 (orange line), LNTF* D70N at pH 5 (green line), and NTF*‐αH at pH 5 (red). (D) Representative equilibrium denaturation curves normalized to fraction folded. LNTF* (light blue circles) and NTF*‐αH (red circles), LNTF* H31N (orange circles), and LNTF* D70N (green circles) at pH 5. Inset of LNTF* urea‐denaturation curve at pH 7 (light blue circles). All data taken at 4 °C

Table 1.

Energetics of Fragment and Full‐Length Sequences Derived from Urea‐Denaturant Melts [Link]

	ΔG unf (kcal/mol)	m‐value (kcal/mol*M)	Cm (M)	ΔΔG unf (kcal/mol)
NTF*	1.60 ± 0.01	1.13 ± 0.01	1.42	—
NTF* H31N#	~0.3	—	0.32	−1.22
NTF* D70N#	~0	—	0.11	−1.51
LNTF*	2.16 ± 0.04	1.20 ± 0.02	1.80	—
LNTF* H31N#	~−0.2	—	0.69	−2.36
LNTF* D70N#	~−0.9	—	0.96	−3.02
NTF*‐αH	2.44 ± 0.05	1.09 ± 0.02	2.23	—
CP13* pH 5	12.94 ± 0.17	2.82 ± 0.02	4.58	—
CP13* pH 7	10.80 ± 0.25	2.55 ± 0.06	4.23	—
CP13* H31N	8.43 ± 0.21	2.10 ± 0.05	4.01	−4.51
CP13* D70N	8.36 ± 0.13	2.17 ± 0.03	3.85	−4.58

^aHere, * denotes the cysteine‐free sequence and ^# denotes the energetics calculated using two‐state fit with a fixed folded baseline from wild‐type sequence.

The structure and energetics of NTF* are pH dependent. CD spectra of NTF* at pH 5 and pH 7 show a difference in magnitude and shape [Fig. 2(A)]. At pH 7, the fragment is no longer folded, as judged by the loss of a cooperative unfolding transition with urea [Fig. 2(B), inset]. This pH dependence may be explained in part by the partially buried salt bridge (H31‐D70) within the N‐terminal subdomain7, 8, 13 since the pKa of His is likely to be shifted above its normal value of around 6.8 depending on the strength of the buried salt bridge. In full‐length T4L*, this salt bridge contributes 3–5 kcal/mol and the protein is most stable at pH 5.13 It is important to note that the predicted isoelectric point (pI) of NTF* is 6.69, indicating that the overall charge of the protein changes sign in this range and is also likely playing a role in the noted pH dependence of stability.

H31 and D70 are important for the folding and stability of NTF*

To probe the role of these residues in NTF*, we generated two single‐site variants, NTF* H31N and NTF* D70N. The CD spectra suggest that, unlike NTF*, both variants are predominately unfolded under native‐like conditions [Fig. 2(A)]. Urea‐induced denaturation of these variants (pH 5, 4 °C) monitored by CD at 222 nm shows no notable folded baseline, demonstrating that, even under conditions of no or low denaturant, the fragments are in the transition zone [Fig. 2(B), Fig. S1]. To estimate how these mutations affect NTF*‘s stability, we carried out a two‐state fit on these data, fixing the folded baseline to the parameters from the wild‐type NTF* fit. This analysis yields an estimate of ΔG _unf ~ 0.3 kcal/mol for NTF* H31N and ΔG _unf ~ 0.0 kcal/mol for NTF* D70N (Table 1). The significant destabilization of NTF* that occurs upon mutating H31 and D70 suggests that these residues play a critical role in the stability of the isolated N‐terminal subdomain.

Extending the C‐terminal helix enhances its stability

In full‐length T4L*, the C‐helix (residues 59–81) spans both subdomains: the majority of the helix is part of the N‐terminal subdomain, but the final seven residues are in the C‐terminal subdomain. A previous study noted that for the isolated C‐terminal subdomain, including all of this helix (an extra 15 N‐terminal residues) stabilizes the C‐terminal subdomain.10 To determine the effect of extending NTF* to include the complete C‐helix, we generated a longer fragment encoding the N‐terminal subdomain with the entire C‐helix sequence, LNTF* (Long‐NTF*, residues 13–81). At pH 5, LNTF* is folded as monitored by far‐UV CD and unfolds cooperatively as a function of urea [Fig. 2(C,D), Fig. S2]. When fit with a two‐state linear extrapolation model,14 this unfolding curve results in a ΔG _unf = 2.16 ± 0.04 kcal/mol with an m‐value = 1.20 ± 0.02 kcal/mol*M. Therefore, under these conditions, an extended N‐terminal subdomain fragment is stabilized (0.6 kcal/mol) compared to NTF*, which has the shorter C‐helix (Table 1). Similar to NTF*, the folding of LNTF* is pH dependent, with less structure and a loss of cooperative unfolding at pH 7 compared to pH 5 [Fig. 2(C,D), inset].

The same single‐site variants investigated in the context of NTF* were also evaluated in the background of LNTF* (LNTF* H31N and LNTF* D70N). CD spectra and denaturation curves of these variants suggest that, like in the shorter fragment NTF*, they are predominately unfolded under native‐like conditions. Using a similar approach as above to get a general estimate of the energetics of these variants, these curves were fit using a two‐state model assuming the same baseline as LNTF*. Based on this analysis, the variants are significantly destabilized compared to LNTF* (LNTF* H31N: ΔG _unf ~ −0.2 kcal/mol; LNTF* D70N ΔG _unf ~ −0.9 kcal/mol) [Table 1, Fig. 2(C,D), Fig. S2].

To determine whether the increased stability of LNTF* as compared to NTF* is due to sequence‐specific interactions or simply the addition of a helical segment, we appended a sequence known to form a stable helix,16 (A(EAAAK)₃A), to the C‐terminus of NTF*, generating NTF*‐αH. When monitored by CD, NTF*‐αH is folded and has a cooperative urea‐induced unfolding curve [Fig. 2(D), Table 1] (ΔG _unf = 2.44 ± 0.05 kcal/mol, m‐value = 1.09 ± 0.02 kcal/mol*M [pH 5.0, 4 °C]). The addition of this stable alpha helix increases NTF* stability by ~0.84 ± 0.14 kcal/mol at pH 5.

Full‐length CP13* populates an equilibrium intermediate in the absence of the H31‐D70 salt bridge

The above fragment studies suggest that the H31‐D70 salt bridge is critical to the stability of the isolated N‐terminal subdomain (~1–2 kcal/mol). This same salt bridge is known to contribute 3–5 kcal/mol in full‐length T4L*.13 To probe the contribution of this interaction in full‐length CP13*, we investigated CP13*'s stability as a function of pH and mutation. Equilibrium urea‐denaturation studies analyzed with a two‐state assumption showed a notable difference in the free energy and calculated m‐value at pH 5 and 7 (pH 5: ΔG _unf = 12.94 ± 0.17 kcal/mol, m‐value = 2.82 ± 0.02 kcal/mol*M; pH 7: ΔG _unf = 10.82 ± 0.25 kcal/mol, m‐value = 2.55 ± 0.06 kcal/mol*M, Fig. 3, Table 1). The lower m‐value at pH 7 suggests a breakdown in the two‐state assumption due to a measureable population of an equilibrium intermediate,17, 18, 19 which could indicate selective unfolding of the N‐terminal subdomain at the higher pH.

Figure 3.

CP13* energetics are dependent on H31‐D70 salt bridge. Representative denaturation melts of CP13* at pH 5 (black solid line), pH 7 (black dashed line), and CP13* H31N (orange line) and CP13* D70N (green line) at pH 5. All data taken at 25 °C

To evaluate whether CP13* populates a partially folded intermediate at equilibrium, we performed proteolysis experiments at both pH 5 and pH 7. Proteolysis of CP13* was monitored as a function of time using the nonspecific protease thermolysin.20, 21 Quenched reactions from different time points were run on an sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) to quantify the remaining full‐length protein [Fig. 4(A) and Materials and Methods]. There is an exponential decrease in the intensity of full‐length protein at both pH 5 and pH 7. Concomitant with this disappearance of the full‐length protein (19 kda), a smaller band appears at just under 15 kDa (I _c) [Fig. 4(A)]. The size of this fragment is consistent with the size of the previously characterized equilibrium intermediate of T4L* and CP13* that includes a folded C‐terminal subdomain [Fig. 1(A) in gray] and residues 60–74 of the C‐helix.10, 12 Analysis of the fragment by mass spectrometry confirms that the fragment (residues 65–164 and 1–12) does indeed map onto the folded region of the intermediate observed by NSHX (I _c is the same as I _eq).10, 12 Together, these results suggest that the cleavable state consists of an unfolded N‐terminal subdomain and a folded C‐terminal domain.

Figure 4.

Proteolysis of CP13* and variants. (A) Representative gel of CP13* proteolysis. CP13* was incubated at 25 °C with 0.20 mg/ml of thermolysin at pH 5 or 7 (see Materials and Methods). Samples taken at designated time points and run on a SDS‐PAGE gel to determine k _p. Cleavage product denoted by (I _c). (B) ΔG _prot of CP13* pH 5 and 7. ΔG _prot calculated from k _p/k _int (K _prot). k _p was calculated from band intensities for full‐length CP13* fit to a single exponential. k _int was calculated based on the concentration of thermolysin at a given pH. (C) Boltzmann diagram of CP13* at pH 5 and 7 determined from proteolysis experiments

To gain more information about the difference in the energetics of CP13* under these two conditions, we carried out a quantitative analysis of the observed proteolysis rate (k _p), utilizing formalisms developed for hydrogen exchange.21, 22, 23 This approach allows for the determination of either the kinetics or thermodynamics of the opening reaction that leads to the formation of the cleavable conformation, depending on the conditions of the experiment. To determine whether proteolysis in our experiments was in the kinetic (EX1 exchange) regime or thermodynamic (EX2 exchange) regime, we monitored k _p as a function of the intrinsic rate of proteolysis, k _int, by modulating the concentration of thermolysin. If k _p is constant as a function of k _int, k _p reports on the kinetics of the partial unfolding reaction. If k _p changes linearly as a function of k _int, k _p provides thermodynamic information about the reaction, and the slope of this line is equal to the equilibrium constant of the opening reaction (K _prot).23, 24 There is a linear relationship between k _p and k _int for CP13* at pH 5 and 7 (slope_pH5 = 0.0018, slope_pH7 = 0.039) suggesting that the proteolysis occurs in the EX2 regime and that k _p reflects the thermodynamics of the unfolding reaction. Using the average k _p/k _int (K _prot) at each pH we find that ΔG _prot, the free‐energy difference between the ground state (N) and the cleavable conformation (I _c), is 3.5 kcal/mol at pH 5 and 1.5 kcal/mol at pH 7 [Fig. 4(B,C)]. These data suggest that destabilizing the salt bridge, increases the population of the intermediate which explains the difference in m‐values under these two conditions.

The pH dependence of the intermediate population implicates a role for the H31‐D70 salt bridge in coupling between the two subdomains of CP13*. To probe this directly, we performed equilibrium denaturation experiments with the single‐site variants of CP13* H31N and D70N. Even at pH 5, two‐state fits of the urea‐induced denaturation curves for both CP13* H31N and CP13* D70N result in m‐values significantly lower than that for CP13* (CP13* H31N: 2.10 ± 0.05 kcal/mol*M [Δm‐value = 0.72 kcal/mol*M]; CP13*D70N: 2.17 ± 0.03 kcal/mol*M [Δm‐value = 0.65 kcal/mol*M]) (Table 1, Fig. 4). Interestingly, these m‐values are quite similar to each other and to the m‐value of the stable fragment of the C‐terminal subdomain, longer c‐terminal fragment (LCTF), 2.06 ± 0.02 kcal/mol*M.9, 10, 12 These lower m‐values, together with the decreased m‐value of CP13* at pH 7 compared to pH 5, suggest that the salt bridge (H31‐D70) plays an important role in the folding cooperativity of the two subdomains in CP13*, without which there is a significant population of an intermediate. Indeed, proteolysis studies on these two variants show complete proteolysis to I _c within 1 min (the first time point). This rapid proteolysis suggests that, in these variants, the N‐terminal subdomain is substantially unfolded under native conditions.

Discussion

In this work, we investigate how a helix spanning the two subdomains of the T4L* variant CP13* couples distant regions of the protein. CP13* is a small globular protein composed of two distinct subdomains with almost no side‐chain interactions between the two,7, 8 yet it exhibits typical two‐state equilibrium unfolding behavior.6, 8, 9, 10 We investigate the role of the single topological connection between the subdomains, the C‐helix in coupling these subdomains.

By comparing the stability of the N‐terminal subdomain in different contexts—in isolation, with additional residues to complete the C‐helix, with additional residues encoding a stable helix, and in the context of the full‐length protein—we find that the energetics of this subdomain are tunable and that this change relies on the addition of a stable helical sequence to the N‐terminal subdomain, thus suggesting a role for the C‐helix in stabilizing this region. By investigating the pH dependence of the subdomain's stability and the effect of site‐specific mutations, we find evidence that these changes in stability directly correlate with a well‐known salt bridge within the N‐terminal subdomain.13 The stability of this interaction increases as the stability of the helix is increased remotely (its C‐terminus). In the context of the full‐length protein, where interactions between the C‐terminal end of the C‐helix and other residues of the C‐terminal subdomain maximally stabilize the helix, we find that removing the salt bridge decouples the two subdomains such that a conformation with an unfolded N‐terminal subdomain is populated at equilibrium. Because D70 is in the C‐helix, we posit that stabilizing the helix orients D70 increasing the energetic contribution from the D70‐H31 salt bridge. Thus, the stability of the C‐helix provides the conduit for long‐range energetic coupling of the two subdomains.

Coupling of the subdomains via the C‐helix

Our current work, together with previous studies of the C‐terminal subdomain, demonstrates that both subdomains can fold autonomously, as suggested by analyses of side‐chain contacts7 [see Fig. 1(B) for a contact map of CP13*]. Independently these regions are marginally stable, but when fused in the context of CP13* the stability is much greater than the sum of the parts (3.75 kcal/mol and 1.60 kcal/mol vs. 12 kcal/mol [Table 1]).

In CP13*, the two subdomains are joined only by a shared 23 amino‐acid helix, the C‐helix. The first 16 residues are part of the N‐terminal subdomain while the last seven are part of the C‐terminal subdomain.8 While helical in the full‐length protein, these residues in isolation do not encode a stable helix: a peptide corresponding to the isolated C‐helix is unstructured in aqueous solutions when monitored by far‐UV CD,25 and Agadir, an algorithm parameterized to predict peptide helix propensity, suggests a very low helical propensity for the C‐helix sequence (4.3%).26

The C‐helix is folded in the context of the C‐terminal subdomain. Previously, we determined the high‐resolution crystal structure of a fragment encoding the C‐terminal subdomain with the entire C‐helix (an additional 15 residues: 60–74), LCTF.10 LCTF has a native‐like fold and residues 65–81 form a helix, indicating that all but the first turn of the C‐helix (residues 59–64) folds in this context. NSHX data show that C‐helix residues 75–81, which make contacts with other residues in the C‐terminal subdomain, have higher ΔG _HX's than the remainder of the C‐helix.10, 12 Thus, the C‐terminal subdomain and its contacts with the last seven residues of the C‐helix (residues 75–81) stabilize the entire helix.10 Such contact‐assisted helix formation has been observed in molecular dynamic simulations of barnase and Protein A.27

The presence of the complete C‐helix sequence also appears to stabilize the C‐terminal subdomain. Interactions between the C‐helix and the rest of the C‐terminal subdomain provide more stability to that subdomain, as LCTF is 5.5 kcal/mol more stable than CTF (residues 75–164 and 1–12).9, 10 In the present work, we find that the C‐helix plays a similar role, albeit with a smaller magnitude, in stabilizing the N‐terminal subdomain sequence. Taken together, it appears that the presence of the complete C‐helix sequence stabilizes both subdomains in isolation.

Long‐distance coupling and communication via helix stability

In proteins, relatively weak bimolecular interactions are strengthened when they occur between components of a single polypeptide chain due to an increase in effective concentration.28 For example, the interaction between imidazole and carboxylic acid is quite weak; however, in the case of the H31‐D70 salt bridge in T4L*, it is worth 3–5 kcal/mol.13 The stabilization of this interaction can be thought of as a result of the increased effective concentration of H31 and D70 in the context of the protein.

Here, we observe an increase in the stability of the H31‐D70N interaction when the NTF* sequence is extended either by completion of the C‐helix (LNTF*), by appending a stable engineered alpha helix (NTF‐αH*), or within the context of the full‐length protein (CP13*). These data suggest that increasing the stability of the C‐terminal end of the C‐helix modulates the stability of the interaction between H31 and D70, presumably by positioning them in an optimal orientation for interaction (i.e. increasing their effective concentration).

Our model for the role of the C‐helix in coupling the N‐ and C‐terminal subdomains is that energetic information is transmitted through helix stabilization. We suggest that side‐chain interactions between residues in the C‐terminal subdomain—the most stable region of the protein—act to stabilize the formation of the C‐helix via interactions at its C‐terminus (the helix–coil transition is known to be a cooperative process—nucleation or stabilization at one end of the helix, will propagate to the rest of the helix29, 30). Stabilizing the C‐terminus of the helix will stabilize the entire helix (including the parts interacting with the N‐terminal subdomain). In short, when the C‐helix is stabilized, it increases the effective concentration of H31 and D70 allowing them to form a stronger salt bridge and stabilize the N‐terminal subdomain.

The C‐helix's lack of intrinsic helicity plays an important role in our model. If the C‐helix were to form a stably folded helix—not reliant on tertiary interactions with the C‐terminal subdomain—there would be less cooperativity between the two subdomains. The N‐terminal subdomain could be folded in the absence of a folded C‐terminal subdomain making it possible for the sequence to populate two partially‐folded equilibrium intermediates. Previous NSHX studies of T4L*12 and CP13*10 as well as the native‐state proteolysis in this study show evidence for a single equilibrium intermediate that has an unfolded N‐terminal subdomain. Taken together, the C‐helix's lack of intrinsic helicity may allow it to serve as a conduit of structural and energetic information.

Several studies on the nature of coupling between regions of a cooperatively folded protein have focused on side‐chain interactions at the interfaces between each region. For example, elegant studies on variants of ankyrin‐repeat proteins have highlighted the importance of the interface interactions between neighboring repeat elements and their relationship to the intrinsic stability of each unit in creating a cooperatively folded protein.31 For titin, a multi‐subunit protein with many Ig‐like β‐sheet modules, the domains are coupled by mutually stabilizing interactions between domains.32

Communication via a unit of secondary structure, such as an alpha helix, provides an alternative and important mechanism for coupling systems without notable interfaces or side‐chain contacts between the modules. A helix‐dependent mechanism of coupling has previously been identified in the model protein spectrin. In nature, spectrin exists as a multidomain protein, with each domain composed of three α‐helices. In the multidomain context, a central helix spans two domains. Similar to the subdomains of T4L*, isolated spectrin domains unfold in a two‐state manner at equilibrium, but they are more stable in the multidomain context, suggesting a role for the central helix in coupling the energetics of these domains. Despite the similarities between spectrin's behavior and our results for T4L*, there is a notable distinction. A study of how cooperativity is conferred in pairs of spectrin domains found that the sequence of the central helix is important for proper coupling between spectrin domains.18 This helix sequence‐dependent communication between these domains differs from our model, as it relies on the specific sequence of the linking helix, whereas the necessary feature in our model is a lack of encoded helicity.

Considerations for protein dissection

A common practice in biochemical studies of large multidomain proteins is to isolate a given domain or domains using sequence homology and secondary structure predictions as guides. Often regions with no predicted secondary structure are thought to serve as flexible linkers that play an insignificant role in the folding of neighboring domains, and are most often used as excision sites. The role of the C‐helix in the cooperativity of T4L*, in spite of its lack of intrinsic helicity, suggests a need for caution when evaluating seemingly inconsequential regions of protein sequence when determining boundaries for protein dissection as they might be critical to the stability and folding of neighboring sequences.

Implications for co‐translational folding

Our helix stabilization model might also be relevant to co‐translational folding. The ribosomal exit tunnel can accommodate33, 34, 35, 36 and stabilize helices,37 but the potential role of these transient helices remains unclear. A possible effect of stabilizing such transient helices might be to transmit stability to the exposed nascent polypeptide chain, thereby stabilizing folding of the emerging N‐terminus of a nascent chain.

Materials and Methods

Plasmid construction

The NTF* construct was subcloned from the CP13* sequence into a pET27a vector that included a sequence that encodes an N‐terminal Tobacco Etch Virus protease (TEV)‐ cleavable hexahistidine tag. Site‐directed mutagenesis was used to convert the TEV‐encoding site to a HRV‐3C protease cleavable site. The LNTF* construct was constructed from a CP13* construct containing encoded N‐terminal hexahisitidine tag and a HRV‐3C site by site‐directed mutagenesis to include a stop codon after residue 81. NTF‐αH* was created using the NTF* construct using site directed mutagenesis to create the alanine‐based helix sequence (AEAAAKEAAAKEAAAKA). All variants were made using site‐directed mutagenesis.

Protein expression and purification

Full‐length proteins were expressed as described previously.9, 10 All fragments were overexpressed in BL21 Codon plus cells. Cells were grown in Luria Broth containing kanamycin (50 μg/ml) at 37 °C for NTF*, LNTF*, and NTF*‐αH protein expression was induced with 1 mM IPTG at OD₆₀₀~0.6. Cells were harvested after 3 h post‐induction.

Pellets of cells that overexpressed NTF*, LNTF*, NTF*‐αH, and their variants, were resuspended in 20 mM Tris pH 8.0 500 mM NaCl 20 mM imidazole 6 M GdmCl. The resuspended pellets were sonicated and whole cell lysates were spun and filtered. Filtered lysate was run on a column packed with Ni‐NTA agarose beads. The protein was eluted from resin using 20 mM Tris pH 8.0 500 mM NaCl 500 mM imidazole and 6 M GdmCl. The eluent of NTF‐αH was dialyzed into 20 mM KOAc 50 mM KCl. The eluents of NTF, LNTF, and their variants were concentrated and buffer exchanged into 20 mM KOAc 50 mM KCl pH 5.0 and 4 M urea and concentrated further. Aliquots were dialyzed into denaturant‐free buffer to ~1 mg/ml prior to experiments.

Cell pellets containing overexpressed CP13* were resuspended into 20 mM Tris pH 8.0 10 mM NaCl. After sonication and centrifugation, the lysate was run on an S column with a gradient against 20 mM Tris pH 8.0 300 mM NaCl. The peak fractions were then diluted into 20 mM NaOAc 10 mM NaCl pH 4.5 and run on an S column with a gradient against 20 mM NaOAc 1 M sodium chloride pH 4.5. The peak fractions were collected, and dialyzed into 20 mM KOAc 50 mM KCl pH 5.0.

CP13* D70N and H31N were purified from inclusion bodies. The cells were lysed and centrifuged. Pellets were washed by sonication in 20 mM Tris pH 8.0 and 1% Triton X‐100 and centrifuged. The wash step was repeated without Triton and centrifuged. The pellet was solubilized by sonication in 20 mM Tris pH 8.0 6 M GdmCl and centrifuged. The solubilized protein was added dropwise into 20 mM Tris 10 mM NaCl pH 8.0 and purified over an S column as described above.

CD Spectra

Far‐UV experiments were performed on an Aviv 410 spectrophotometer. All data was taken in either 20 mM KOAc pH 5.0 50 mM KCl or 20 mM KPO4 pH 7.0 50 mM KCl. CD spectra were taken in an AVIV 410 CD spectrometer at 4 °C in a 0.1 cm quartz cuvette at protein concentrations ~500 μg/ml. Data were collected between 260 and 200 nm at 1 nm intervals with each data point an average of 5 s of data. Data with dynode above 480V were not included.

Equilibrium denaturant melts

All equilibrium experiments were carried out in aforementioned buffer conditions. Urea melts of NTF*, LNTF*, and NTF‐αH* and variants were carried out at 4 °C, those of CP13* and variants were carried out at room temperature. CD signal was monitored at 222 nm and carried out in a 1 cm quartz cuvette. Chemical denaturant melts were carried out using an automated titrator with 5‐min equilibration times at each denaturant concentration. Chemical denaturation was fully reversible for all variants studied. Protein concentrations ranged from 20 to 50 μg/ml.

Proteolysis

Proteolysis rates were measured by incubating protein (400 μg/ml) with various concentrations (0.01, 0.1, 0.2, and 0.4 mg/ml) of thermolysin at 25 °C at pH 5 or 7 using the buffers described above. Proteolysis reactions were quenched at various time points with 50 mM EDTA. The quenched reactions were run on SDS‐PAGE gels. Gels were stained with SyproRed (Lonza Rockland) and scanned using a Typhoon scanner (GE Healthcare). ImageJ (NIH) was used to quantify the band corresponding to the full‐length protein. The band intensities were plotted as a function of time in KaleidaGraph (Synergy Software) and fit to a first‐order rate equation to calculate k _p. The k _p at various thermolysin concentrations were plotted against k _int which was calculated from the k _cat/K _m values calculated previously.38 The data were fit to a line and the slope was used to confirm the exchange regime of the proteolysis reaction. After confirming the EX2 exchange regime, ΔG _prot was calculated (ΔG _prot (=−RTlnK_prot)) using the average k _p/k _int (which is K _prot).

Supporting information

supporting Information Figures S1 and S2. The denaturant melts of NTF*, LNTF*, and their respective salt bridge variants presented as CD signal at 222 nm instead of fraction folded.