Revealing what gets buried first in protein folding

After decades of research, the nature of the rate-limiting step in the folding of globular proteins still elicits a range of opinions (1). The properties of the associated transition state (TS) provide critical insights into possible folding mechanisms. However, the characterization of the TS is challenging because atomic level methods cannot be applied to this minimally populated state, and lower resolution methods have produced divergent views. Even the existence of a generalized TS remains actively debated. In PNAS, Guinn et al. (2) dissect the burial properties of the TS for 13 proteins by analyzing the denaturant and temperature dependence of folding rates to distinguish the burial of hydrophobic surface from that of amide groups. With this capability, they propose that the TSs generally are very advanced and often contain the native 2° structure, a level higher than most prior methods have indicated.

The analysis by Guinn et al. (2) indicates that amide surface is preferentially buried in the TS. On average, 77% of the total amide surface is buried in the TS, whereas a smaller fraction, 60%, of the total hydrophobic surface is buried at this point in the reaction (although ∼twofold more hydrophobic surface is buried in the TS in absolute terms). Notably, the amount of amide burial often can be accounted for through the formation of the entire native 2° structure. Accordingly, Guinn et al. propose that TSs often contain the native 2° structure with most of the tertiary contacts forming post-TS. Structural modeling indicates that this level of 2° structure organizes the chain into the native fold before the TS and the rate-limiting step is the formation of a key set of tertiary contacts. Because the amide and hydrocarbon burial are preferentially buried pre- and post-TS, respectively, the authors suggest that these two quantities along with total burial are natural reaction coordinates.

The results by Guinn et al. emphasizing 2° structure over hydrophobicity (2) evoke a comparison with the long-standing debate on the determinants of protein stability. Pauling and others stressed the importance of hydrogen bonds (3), but Kauzmann later argued that the hydrophobic effect is the dominant factor (4). However, more recent work has revisited this conclusion (5). Analogously, early kinetics models focused on the role of 2° structure (6), whereas later work emphasized hydrophobic collapse and the possibility that 2°-structure formation is driven by the collapse process (7).

The analysis by Guinn et al. indicates that amide surface is preferentially buried in the TS.

The native level of 2° structure in the TS proposed by Guinn et al. (2) is higher than indicated by most other experimental methods (Table 1). The predominant method for studying the TS has been mutational ϕ analysis. This method has supported a variety of folding models, but it generally indicates that multiple 2° structural elements can be absent in the TS (8–10). The newer Ψ analysis method, which uses bi-histidine metal ion-binding sites to explicitly identify side chain–side chain contacts (11), produces a relatively uniform TS picture, which also is one closer to that envisioned by Guinn et al.; Ψ analysis indicates that TSs are native-like and large, with extensive amounts of 2° structure and ∼70% of the native topology (8, 10, 12). Nevertheless, Ψ values of zero can be found on helices and between strands, indicating that some 2°-structure elements are absent in the TS.

Table 1.

Methods for characterizing the folding TS and levels of 2°-structure formation

Method	Observation
Guinn et al. analysis (2): Amide and hydrophobic surface burial	On average, 77% and 60% of the total amide and hydrophobic surface are buried in the TS, respectively; the level of amide burial is consistent with native 2°-structure formation
ϕ analysis: Energetic effects of altering side chains	TSs generally lack some 2° structures, particularly in polarized TSs
Ψ analysis: Side chain-side chain contacts	Large with extensive amounts of 2°-structure and native-like topology, but with some unstructured regions
H/D amide isotope effects: backbone hydrogen bond formation	Fraction of (helical) hydrogen bonds correlates to surface burial in the TS, typically with 50–90% of the hydrogen bonds formed
HX, RD: structure of post-TS hidden intermediates	Hidden intermediates often lack some 2° structures and place an upper bound on structure in the TS

H/D amide kinetic isotope effects can directly probe backbone hydrogen-bond formation in the TS (13, 14) and, hence, are a useful complement to the method by Guinn et al. of quantifying amide surface burial (2). Although the H/D isotope effect primarily reports on helical hydrogen bonds, we observed that, for 10 proteins, about 50–84% of the (helical) hydrogen bonds are formed in the TS and the percentage for each protein matches the fraction of the total surface buried in its TS [estimated from the m_f/m_o ratio of the denaturant dependence of folding rates (chevron plot) and global stability]. This result is remarkably consistent with the finding by Guinn et al. that 50–90% of the total burial occurs in the TS, and it supports the view that hydrogen bonding and surface burial are largely matched in the TS.

Hydrogen exchange (HX) (15) and NMR relaxation dispersion (RD) methods (16) have characterized “hidden” species (17) on the native side of the TS. These post-TS species serve as the upper bound for the amount of structure possibly present in the TS. Most of these species are well advanced, but they typically lack some elements. For example, Bai and coworkers found that one helix and a portion of another helix is unfolded in a late intermediate of reduced-apocytochrome b₅₆₂ (18), whereas a post-TS intermediate of T4 lysozyme lacks the smaller β domain (19). Cytochrome c has three post-TS intermediates with the least structured one lacking a helix and three large loops (15, 20). The post-TS intermediate in ubiquitin is quite advanced, but lacks a β-strand and a 3₁₀ helix (11, 21).

A direct comparison between the different methods can be made for five proteins investigated by Guinn et al. (2). The TSs for protein L (8, 9) and acylphosphatase (ACP) (10) have been characterized by ϕ, Ψ, isotope effects (14). One helix is unfolded in the TS of both proteins and one to two of the five β-strands are absent in the TS of ACP. Some 2° structures also are unfolded in the TS of Fyn-SH3 (16) and CI2 (22). As mentioned, HX studies indicate that the TS of T4 lysozyme lacks at least a small β domain; however, this protein is one of two where Guinn et al. find that the amount of amide burial buried in the TS is less than that associated with the native 2° structure (2). The previous methods indicate that TSs have the majority but not the entire native 2° structure. The discrepancy between these and the current study by Guinn et al. may be reconciled if the authors relax their assumption that amide burial in the TS is first assigned to 2° structure, rather than any portion of the chain (e.g., loops).

Simulations and theory also have produced a variety of results on the amount of 2° structure in the TS. All-atom simulations have found that for some helical proteins, the TS has near-native 2°-structure content (23), whereas, in other proteins, the TS has a lower amount of structure (23–25). Landscape theory has estimated that the TS has ∼60% of the native contacts, but 88% of the backbone angles are in the native configuration (26), values that are highly consistent with the burial fractions of Guinn et al. (2).

The relative degree of 2°-structure formation and surface burial in the TS speaks to possible folding mechanisms. One view is that local context dominates and folding proceeds by the assembly of preformed hydrogen-bonded elements (6). At the other extreme is a hydrophobic collapse with minimal 2° structure. In between these views is one where hydrogen bonding and hydrophobic surface become buried incrementally and in concert in a process of sequential stabilization (27). Here, nascent native-like substructures serve as templates for the formation of additional structure through the stepwise addition of cooperative folding subunits or “foldons” (11, 15, 20, 21). In this view, the early folding process involves an initial uphill search through a multitude of mostly unproductive unstable conformations. Some conformations serve as better templates for the addition of other foldons and so guide the folding process. Individual steps involve the buildup of elements of local structure supported by long-range hydrophobic contacts with nearly commensurate levels of hydrogen bonds and surface burial (11, 13, 14). This process allows intrinsic local structural propensities to be enhanced or overridden by environmental context. Computational support for this mechanism can be found through its application to predict folding pathways and structure without using homology-based information (28, 29).

The method by Guinn et al. of separating amide from hydrophobic burial (2) provides an important piece to the folding puzzle. It still leaves, however, unresolved questions that require further study if all of the pieces are to fit well together. Beyond the application to protein folding, their analysis should be a very useful (and easily applied) method to characterize a variety of biological processes, including those involving conformational change and binding as well as the folding of nucleic acids.