Abstract

Introduction

Hysteresis is a generally observed phenomenon across many systems in which the output is a function of not only the current input, but also of the function of the history of inputs to the system. The consequence of hysteresis can present itself in a number of ways across a broad range of models that can be presented physically and mathematically. In biology, for example, it is observed in activated T-cells with increased plasticity of their response as the threshold for their activation is lowered (1). With respect to proteins, hysteresis can be a signature of biphasic behavior or an apparent non-equilibrium behavior in the native state where the energetic barrier is large enough to distinguish two native conformations (2-6). This behavior can be attributed to the varying populations of folded states depending on where you are on the dual basin hysteretic energy landscape of Green Fluorescent Protein (GFP) (7, 8). Hysteresis in the folding of GFP (7-9) is mediated by the coupling of a proline-locking mechanism that alters the packing of the central functional chromophore. The dynamic lag manifests itself as a non-coincidence in the denaturing and refolding “thermodynamic” titrations, after the protein has been denatured for an extended period of time (7). While knots in proteins are an active area of research, the question of hysteresis and knotting is largely unexplored.

Hysteresis in the knotted protein, Minimal Tied Trefoil from Thermotoga maritima (MTT_Tm, PDB code 1O6D), was predicted and confirmed for this highly stable knotted protein because of the complex topology and high stability of the protein (10, 11). Originally knots were studied in biomolecules, such as nucleic acids and DNA, which are known to be flexible and can diffuse along the polymer chain (12-14). However, due to increased complexity inherent in a polypeptide chain, the same theory of flexibility and diffusion is not directly translatable to proteins (12). Due to the conformational restriction in a knotted protein, the geometric constraints can introduce a variety of other complex structural consequences. Simulations indicate an uncoupling of folding from untying; that is, the protein unties on timescales significantly longer than required for unfolding (10). In the current study, we use a combination of optical and solution Nuclear Magnetic Resonance (NMR) spectroscopies to elucidate differences in the refolded hysteretic protein from the native, under conditions strongly favoring the native state. Furthermore, we highlight molecular interactions we attribute to hysteresis, using ¹H-¹H-¹⁵N-NOESY-HSQCs to further visualize the side chain packing in relation to the knot topology. Our results indicate a noticeable switch in the knotting loop where a pronounced difference in the environment for initiating residue F65 is observed. In line with theoretical data (10), we find that the packing of the of the β-like turn with the phenylalanine-histidine pair in the knot element precludes true equilibrium, contributing to hysteresis in MTT_Tm. Our results indicate that the interactions that hinder reversible folding are side chain mediated for knot proteins. Consistent with simulations, early formation of this β-like turn requires backtracking, a localized unfolding event needed to allow folding to proceed, as the turn may be in the wrong orientation for proper formation of the knot. Here, we provide the initial experimental evidence supporting theoretical suggestions (11) of non-native interactions of the β-like turn.

Results and Discussion

The structure of MTT_Tm

MTT_Tm is identified as a member of the SPOUT family of methyltransferases, containing a five stranded β-sheet sandwiched between two α-helices on one side and three α-helices on the other (15-17) (Figure 1). The most distinct feature of the SPOUT family is the topologically unique deep C-terminal knot, classified as a 3₁ knot (trefoil knot). This represents the simplest possible knot found within protein structures, defined by a single loop threaded once by the backbone, where the chain crosses itself no more than three times (18). The knot loop in MTT_Tm is formed by β-strand 3 and α-helix 3 and is subsequently threaded by β-strand 5 (Figure 1). The knotting loop region is anchored by residue F65 through D92 (Figure 1, blue), while the threading β-strand (Figure 1, orange) encompasses residue H114 through L119.

Open in a separate window

Figure 1

Crystal structure of the knot protein, MTT_Tm (PDB code 1O6D). (A)Viewed from the side, the protein is represented as a cartoon emphasizing the knot with the threading β-strand (β 5, orange) threading through the knot loop (blue). (B) Same as A, viewed from the top of the protein. (C) Schematic of the protein, viewed in the same orientation as (B) following the same color scheme.

Structural details of MTT_Tm indicate that there are unique features in the loop connecting β-strand 3 and α-helix 2. This tight loop resembles a β-turn, as it is a short loop in a restricted conformation. Prototypically, β-turns consist of four critically packed residues (labeled i, i+1, i+2, and i+3) and are defined by either having an inter-backbone hydrogen bond between the CO of residue i and the NH of residue i+3, or a distance of less than 7 Å between the Cα atoms of residue i and i+3 (19). In the case of MTT_Tm, the packing of the β-like turn is mediated by a hydrogen bond between the CO of L61 and the OH of S64, contributing to the restricted conformation of the turn (Figure 5). The trans-proline and glycine (P62 and G63) in between residues L61 and S64 orient them into a position that allows them to hydrogen bond forming a β-like turn (19). This creates a critical network of contacts that stabilizes the native state.

Open in a separate window

Figure 5

Schematic of the critical packing interactions around the knot region of MTT_Tm. (Top) The knotting loop and threading β-strand 5 are shown in blue and orange, respectively. The two conformations observed in F65 are represented as dashed lines, where the dominant state can interact with H114. The β-like turn is highlighted in dark green. The native trans-proline is represented by a dark green proline ring and side chain. The cis-conformation of the proline is shown in light green. (Bottom) Crystal structure of the β-like turn, following the same coloring scheme as above, highlighting the hydrogen bond (dashed green line) interaction critical for the loop formation between residue L61 and S64.

Characterization of hysteresis in MTT_Tm

MTT_Tm has been observed to undergo the unique phenomenon of hysteresis, which is dependent on the time scales allowed for denaturing. The hysteretic behavior of MTT_Tm is most clearly evident by evaluating thermodynamic titrations in denaturant (10). To further investigate the molecular details of hysteresis, we expanded upon previous work by evaluating the influence of incubation time in denaturant with spectroscopic methods, including residue specific details by NMR. The subsequent thermodynamic effects of the varied incubation times were evaluated under NMR conditions (Figure 2). MTT_Tm was incubated for 24 hours before refolding, a significantly longer time than ten times the t_1/2-value obtained from kinetics (Supplemental Figure S1). From these denaturing conditions, we obtain two superimposable titration curves for unfolding and refolding (Figure 2, Top). This suggests that the knot is still intact in MTT_Tm, even after incubating for 24 hours under denaturing conditions. This is supported by previously published in vitro and in silico studies indicating that the knot persists in the denatured state much longer than initially thought (20, 21).

Open in a separate window

Figure 2

Thermodynamic titrations of MTT_Tm. (Top) Overlay of a denaturing titration curve (black) with a refolding titration curve, after incubation in denaturant for 24 hours (blue). The two titration-curves are superimposable. (Bottom) Overlay of a denaturing titration curve (black) with a refolding titration curve (red), collected after incubation in denaturant for 4 weeks. The titration curves are no longer superimposable.

Hysteresis was initially identified in MTT_Tm after a six-month incubation in strong denaturant (10), we evaluate the time frame in which hysteresis is observed under our new NMR conditions (see methods below). After denaturing MTT_Tm for four weeks, hysteresis was clearly evident (Figure 2, Bottom). In addition to shifting the midpoint to lower denaturant, a new biphasic behavior emerges, indicating that there is a third state in the hysteretic protein. Due to the emergence of a third state and a shifting midpoint we wanted to evaluate whether the native protein and the hysteretic protein have the same native structure.

Previous NMR studies investigating the effects of hysteresis in the highly stable GFP revealed chemical shift heterogeneity in the native-like trapped hysteric protein (7). Using a similar approach, we acquired both ¹H-¹⁵N HSQCs of native and hysteretic protein to obtain residue specific information. To probe the dynamic solution state, MTT_Tm was assigned using a suite of NMR experiments. The chemical shifts are well dispersed in both native and hysteretic protein (Figure 3). The overlaid spectra are indistinguishable under these conditions, indicating that hysteresis is not attributable to perturbations in the backbone interactions. This is indicative that the hysteretic protein retains its native backbone structure despite hysteresis. Given the constrained geometry of the knotted MTT_Tm, it is reasonable that backbone limitations will influence both the position and relative dynamics of side chains, particularly those that have steric restrictions. As assessment of backbone interactions indicate that they are not principally involved in hysteresis, examination of side chain interactions by NMR proved more revealing. Figure 4 presents ¹H-¹H-¹⁵N-NOESY-HSQC 2D strip plots of key residues, with the most pronounced heterogeneity observed in F65. The heterogeneity in F65 is clearly observed as two distinct peaks, indicating that F65 populates two slowly exchanging conformations. F65 is in close proximity to H114, allowing the aromatic rings to form stabilizing π-stacking interactions (22-24). This π-stacking interaction can account for the presence of two conformations, one of which is stabilized, identified as the more dominant peak in the ¹H-¹H-¹⁵N-NOESY-HSQC (Figure 4). H114 is in a critical position as it is the initiating residue of the threading β-strand 5 (Figure 5). Interestingly, the knotted region is adjacent to the constrained β-like turn facilitating crosstalk through the network of contacts and the knot through the π-stacking of F65 and H114.

Open in a separate window

Figure 3

Comparison of ¹H-¹⁵N HSQC spectra of native and hysteretic MTT_Tm. The spectral overlays of native (black) and hysteretic (red) protein are indistinguishable. The assignments shown in the spectra are highlighted as spheres on the inlayed molecular representation of MTT_Tm (L61 and S64 are shown in green, F65 shown in blue, and H114 shown in orange). Additionally, the knot is colored in blue (knotting loop) and orange (the threading β-strand 5), and the β-like turn in green.

Open in a separate window

Figure 4

¹H-¹H-¹⁵N-NOESY-HSQC strip plots highlighting side chain interactions. Shown are 2D strips of residues comparing the native (black) and hysteretic (red) proteins. The cyan box in F65 highlights both the heterogeneity of the side chain in MTT_Tm (black) and the differences in the side chain signature between native and hysteretic protein.

Comparison of native and hysteretic protein ¹H-¹H-¹⁵N-NOESY-HSQC spectra indicate no significant deviations between the two spectra, with the exception of F65 (Figure 4). A unique difference in the hysteretic protein is observed as a loss in heterogeneity in the side chain conformation compared to native. This change in conformation upon refolding is similar to other observations of “aromatic ring switches” where a subtle change in ring packing, swap in hydrogen bonding (25) or stacking interactions lead to heterogeneity. More importantly, while overall backbone interactions appear intact, subtle disruption of side chain interactions are evident, as noted by the differences presented here (Figure 4). Consistent with the coupling of long range geometric constraints observed in GFP (7), the observed conformational differences in the F65 side chain are attributable to the interplay between the specific network of contacts bridging the complex topology of the knot to the tight β-like turn.

Conclusions

While theoretical and experimental work has shown that over time knots will eventually untie, this event takes significantly longer than initially anticipated (10, 20, 21). As untying the knot of MTT_Tm is distinctly decoupled from the unfolding mechanism (10), the results presented here suggest a specific network of side chain contacts as the cause of hysteresis. We hypothesize that hysteresis in MTT_Tm is attributable to the specific packing of side chain contacts bridging the complex topology of the knot to the β-like turn, leading into the knot, mirroring some of the complex behavior of the proline-linked chromophore hula-twist packing challenges leading to hysteresis in GFP (7). The π-stacking interaction between H114 (Figure 5, orange) and F65 (Figure 5, blue) allows for direct cross talk through the adjacent β-like turn (Figure 5, green), which could “lock” P62 in a trans-conformation in the native state. The “locking” represents the introduction of a geometrical constraint in the denatured state by the knot, which inhibits the isomerization of P62 from a trans- to cis-conformation. Hence, there is no hysteresis after denaturing and refolding in a short period of time because the proline is held in the trans-conformation by the knot. However, if MTT_Tm denatures for an extended time the packing of β-like turn together with the knot can destabilize, so that P62 can eventually isomerize to a thermodynamic distribution of interconverting the cis-trans conformations in the denatured state. This then allows for refolding into a “native-like” protein, with the proline in a cis-conformation rather than trans. Thus, the time dependent isomerization state of the proline may contribute to the observed hysteretic shift in the thermodynamic titration. Interestingly previous simulation studies indicate that early formation of the P62 β-like turn “precludes the correct route towards folding” (11). It has yet to be seen if experimental manipulation of the β-like turn would allow for the recovery of the native packing and relieve hysteresis, which is an area of active study. Importantly, simple cis-trans proline isomerization will not lead to hysteresis on its own. When proline isomerization is packed in close proximity to complex topology in proteins, hysteretic bistability can occur. In the case of MTT_Tm, the β-like turn is packed against the knotting loop, which stabilizes this structure in the denatured state and precludes true equilibrium and causes the observed hysteretic behavior.

Methods

Molecular Representations

Molecular representations were generated using the crystal structure of MTT_Tm, PDB code 1O6D, and PyMOL (28).

Acknowledgements

References