Whether proteins fold along defined stepwise pathways built from cooperative ~20-residue foldon units, or through many heterogeneous routes on a funneled energy landscape, remains a central unresolved question of protein biophysics. The foldon-pathway model is grounded in hydrogen-exchange pulse-labeling mass spectrometry and native-state hydrogen-exchange experiments on very few proteins (cytochrome c, ribonuclease H, maltose binding protein), and its generality across the diversity of protein folds, its applicability to kinetically two-state proteins, its extension to chaperone-assisted folding, its mechanistic basis (what defines foldon boundaries), and its reproducibility in physics-based simulation are all open. This problem asks for a determination of whether the foldon-based stepwise pathway picture describes protein folding in general, by direct foldon-resolved pathway mapping across structurally diverse proteins, by establishing whether hidden foldon units exist in two-state folders, and by reproducing foldon behavior in silico with realistic transferable force fields.

Background

Protein folding asks how an unfolded polypeptide chain reaches its native three-dimensional structure. Levinthal's paradox is the observation that an undirected random search over the astronomically many chain conformations could not reach the native state on biological timescales, which demands an explanation of how the search is accelerated. Two broad pictures have competed since the 1980s. The 'new view', derived from energy-landscape theory, holds that folding proceeds by amino-acid-level thermally driven dynamics down a funneled energy landscape through very many heterogeneous routes and intermediates. The classical 'defined pathway' view, revitalized by experiment from the Englander laboratory, holds that proteins are multistate assemblies of separately cooperative folding units called foldons — segments of roughly 15–35 residues that unfold and refold as units — and that folding proceeds by the ordered, stepwise formation and docking of one native-like foldon after another, each earlier foldon guiding and stabilizing the next (sequential stabilization). Key experimental tools: hydrogen exchange (HX) measures the protection of backbone amide hydrogens, which becomes slow when the amide is in a stable hydrogen-bonded structure; in pulse-labeling HX MS, a brief deuterium-to-hydrogen labeling pulse is applied at chosen times during kinetic refolding and read out fragment-by-fragment by mass spectrometry, giving population snapshots of which segments are already folded; native-state HX reads out the same foldon units at equilibrium under native conditions. Partially unfolded forms (PUFs) that differ by one foldon are the rungs of the equilibrium ladder. Using these methods, cytochrome c, ribonuclease H (Hu et al. 2013), and maltose binding protein (Walters et al. 2013) were each shown to refold through a stepwise sequence of native-like foldon intermediates that essentially the entire molecule population follows (<5% off-pathway), with limited branching when prior structure supports two equivalent subsequent steps (Krishna et al. 2007). The 2014 Perspective by Englander and Mayne synthesized this evidence, proposed a reconciliation — disordered microscopic multitrack searching forms the first foldon, then macroscopic foldon–foldon interactions bias the ordered stepwise assembly — and framed the remaining questions. A related 'predetermined pathway-optional error' (PPOE) hypothesis (Krishna & Englander 2007) explains residual kinetic heterogeneity as chance misfolding errors superimposed on one pathway. Many small proteins, however, show single-exponential (kinetically two-state) folding with no observable intermediates; NMR relaxation dispersion (which detects sparsely populated, transiently formed states through their effect on NMR line shapes) has revealed hidden intermediates in several such proteins (Sekhar & Kay 2013). On the computational side, millisecond all-atom molecular dynamics with realistic transferable force fields (Lindorff-Larsen et al. 2011) and Markov state models — kinetic models that coarse-grain many short trajectories into transitions between conformational states — built from such trajectories generally describe small fast-folding proteins as folding through multiple routes, and the Englander group's own review of the field (Englander et al. 2016) conceded that simplified models had not found cooperative foldons while a new generation of realistic-force-field analyses 'tends to emulate' the foldon picture — reliable in-silico reproduction of foldon behavior remained unachieved. The dispute over pathway uniqueness is explicit and adversarial: Eaton and Wolynes (2017) argued that theory, simulation, and single-molecule experiments 'show that proteins fold by multiple pathways', while Englander and Mayne (2017, and their Reply) argued the microscopic multiplicity belongs only to the prenucleation stage and that structure formation proceeds by defined foldon steps. Whether proteins fold by defined foldon pathways is not merely semantic: it determines what mechanistic information folding rates and intermediates carry, how misfolding and aggregation arise, and how folding in vivo (including in chaperone compartments) should be modeled.

Problem Statement

Determine whether the foldon-based stepwise pathway model describes protein folding in general, and what defines its limits. The single scientific objective is to establish - positively or negatively, with foldon-resolved evidence - the domain of validity of the claim that proteins fold by the ordered, cooperative formation and assembly of native-like foldon units along a predominantly defined pathway. This objective jointly requires the following linked determinations: (a) extending residue-resolved kinetic pathway mapping (e.g., hydrogen-exchange pulse-labeling mass spectrometry, NMR relaxation dispersion, or equivalent structure-resolved kinetic probes) beyond cytochrome c, ribonuclease H, and maltose binding protein to a protein panel spanning diverse folds, sizes, and folding classes, deciding for each whether folding proceeds by stepwise foldon assembly or by heterogeneous parallel routes, using the fraction of the population that follows a structurally distinct parallel route as the operational criterion, where two routes are structurally distinct when they form different sets of native-like foldon segments or add those segments in a different order (or into structurally different native-like units), counting as stepwise rather than as a distinct route both branching between near-equivalent next steps that prior structure permits and the residual kinetic heterogeneity that the model itself attributes to chance misfolding excursions (the optional-error / PPOE mechanism), so that only flux through a structurally distinct parallel route counts toward the fraction: a protein counts as stepwise when fewer than 5% of the population follows a structurally distinct parallel route (the <5% off-pathway level documented for the characterized proteins), and counts as heterogeneous when 5% or more does, so the 5% cut and its complement partition every protein with no intermediate band; (b) establishing whether small, kinetically two-state proteins that never populate observable intermediates nonetheless possess cooperative foldon units and stepwise pathways, by indirect validation or new methods; (c) demonstrating reliable reproduction of experimentally measured foldon behavior in silico with realistic, transferable (non-per-protein-tuned) force fields, or demonstrating the systematic failure of current transferable force fields to reproduce such behavior; and (d) characterizing the prevalence and mechanistic significance of pathway branching beyond the special case documented in cytochrome c. The following two questions are related field-level questions that inform the objective without being part of its linked determinations: identifying the structural and energetic determinants that make a set of residues a cooperative foldon (foldon-boundary definition); and extending the framework to chaperone-assisted folding beyond the single case studied to date, maltose binding protein folding inside GroEL (a barrel-shaped chaperonin machine that encapsulates unfolded substrate proteins in an isolated cavity). Progress on either would deepen a resolution. Facet (c) is a required determination: because the foldon model is a physical account of how proteins fold, whether realistic, transferable-force-field simulation can reproduce experimentally measured foldon behavior is an operational test of the model's mechanistic sufficiency, and facet (c) is resolved in either direction - by demonstrated reproduction, or by a systematic failure of current transferable force fields to reproduce foldon behavior across a representative panel of proteins. A systematic force-field failure is evidence about the adequacy of current simulation methodology, not by itself about the model's truth: the same failure is equally attributable to force-field inaccuracy, so it bears on the generality claim only when corroborated by foldon-resolved experimental evidence, and cannot on its own refute the model. Because the objective is a general claim about folding, the determinations in facets (a) and (d) support an affirmative general conclusion only when the underlying evidence meets an adequacy bar: the panel must cover at least the four major structural classes (all-α, all-β, α/β, and α+β), each represented by at least two non-homologous proteins, spanning sizes from small single-domain to multi-domain, with the stepwise-versus-heterogeneous and branching verdicts consistent across those classes and sizes. An affirmative resolution - that the foldon model describes folding in general - requires all four determinations (a)-(d) to be resolved with mutually consistent evidence. A negative resolution, by contrast, rests on foldon-resolved experimental observation alone and does not require (b), (c), or (d) to be resolved: a single foldon-resolved demonstration of heterogeneous multipathway folding in a protein the model predicts to fold stepwise, or a systematic foldon-resolved experimental failure to observe stepwise foldon assembly across a panel meeting the adequacy bar, is by itself decisive evidence against generality. A third outcome arises when (a), (b), and (d) are resolved affirmatively while (c) resolves as a systematic force-field failure: the foldon pathway model is then experimentally supported but not currently reproducible in silico. The foldon pathway model's scope - which proteins, which conditions, which stages of folding - is settled by foldon-resolved experimental evidence together with the facet (c) simulation determination, or by decisive evidence that folding generality belongs to heterogeneous multipathway behavior instead.

References

  1. S. Walter Englander, Leland Mayne (2014) The nature of protein folding pathways. Proceedings of the National Academy of Sciences 111(45):15873-15880. DOI 10.1073/pnas.1411798111. https://doi.org/10.1073/pnas.1411798111
  2. Wenbing Hu, Benjamin T. Walters, Zhong-Yuan Kan, Leland Mayne, Laura E. Rosen, Susan Marqusee, S. Walter Englander (2013) Stepwise protein folding at near amino acid resolution by hydrogen exchange and mass spectrometry. Proceedings of the National Academy of Sciences 110(19):7684-7689. DOI 10.1073/pnas.1305887110. https://doi.org/10.1073/pnas.1305887110
  3. Benjamin T. Walters, Leland Mayne, James R. Hinshaw, Tobin R. Sosnick, S. Walter Englander (2013) Folding of a large protein at high structural resolution. Proceedings of the National Academy of Sciences 110(47):18898-18903. DOI 10.1073/pnas.1319482110. https://doi.org/10.1073/pnas.1319482110
  4. Mallela M.G. Krishna, Haripada Maity, Jon N. Rumbley, S. Walter Englander (2007) Branching in the sequential folding pathway of cytochrome c. Protein Science 16(9):1946-1956. DOI 10.1110/ps.072922307. https://doi.org/10.1110/ps.072922307
  5. Mallela M.G. Krishna, S. Walter Englander (2007) A unified mechanism for protein folding: Predetermined pathways with optional errors. Protein Science 16(3):449-464. DOI 10.1110/ps.062655907. https://doi.org/10.1110/ps.062655907
  6. Ashok Sekhar, Lewis E. Kay (2013) NMR paves the way for atomic level descriptions of sparsely populated, transiently formed biomolecular conformers. Proceedings of the National Academy of Sciences 110(32):12867-12874. DOI 10.1073/pnas.1305688110. https://doi.org/10.1073/pnas.1305688110
  7. Kresten Lindorff-Larsen, Stefano Piana, Ron O. Dror, David E. Shaw (2011) How fast-folding proteins fold. Science 334(6055):517-520. DOI 10.1126/science.1208351. https://doi.org/10.1126/science.1208351
  8. S. Walter Englander, Leland Mayne, Zhong-Yuan Kan, Wenbing Hu (2016) Protein Folding-How and Why: By Hydrogen Exchange, Fragment Separation, and Mass Spectrometry. Annual Review of Biophysics 45:135-152. DOI 10.1146/annurev-biophys-062215-011121. https://doi.org/10.1146/annurev-biophys-062215-011121
  9. Wenbing Hu, Zhong-Yuan Kan, Leland Mayne, S. Walter Englander (2016) Cytochrome c folds through foldon-dependent native-like intermediates in an ordered pathway. Proceedings of the National Academy of Sciences 113(14):3809-3814. DOI 10.1073/pnas.1522674113. https://doi.org/10.1073/pnas.1522674113
  10. S. Walter Englander, Leland Mayne (2017) The case for defined protein folding pathways. Proceedings of the National Academy of Sciences 114(31):8253-8258. DOI 10.1073/pnas.1706196114. https://doi.org/10.1073/pnas.1706196114
  11. William A. Eaton, Peter G. Wolynes (2017) Theory, simulations, and experiments show that proteins fold by multiple pathways. Proceedings of the National Academy of Sciences 114(46):E9759-E9760. DOI 10.1073/pnas.1716444114. https://doi.org/10.1073/pnas.1716444114
  12. S. Walter Englander, Leland Mayne (2017) Reply to Eaton and Wolynes: How do proteins fold? Proceedings of the National Academy of Sciences 114(46):E9761-E9762. DOI 10.1073/pnas.1716929114. https://doi.org/10.1073/pnas.1716929114
  13. Xiang Ye, Leland Mayne, Zhong-yuan Kan, S. Walter Englander (2018) Folding of maltose binding protein outside of and in GroEL. Proceedings of the National Academy of Sciences 115(3):519-524. DOI 10.1073/pnas.1716168115. https://doi.org/10.1073/pnas.1716168115
  14. Hao Yu, Derek R. Dee, Xia Liu, Angela M. Brigley, Iveta Sosova, Michael T. Woodside (2015) Protein misfolding occurs by slow diffusion across multiple barriers in a rough energy landscape. Proceedings of the National Academy of Sciences 112(27):8308-8313. DOI 10.1073/pnas.1419197112. https://doi.org/10.1073/pnas.1419197112
  15. Krishna Neupane, Daniel A. N. Foster, Derek R. Dee, Hao Yu, Feng Wang, Michael T. Woodside (2016) Direct observation of transition paths during the folding of proteins and nucleic acids. Science 352(6282):239-242. DOI 10.1126/science.aad0637. https://doi.org/10.1126/science.aad0637
  16. Sandhya Bhatia, Guruswamy Krishnamoorthy, Jayant B. Udgaonkar (2021) Mapping Distinct Sequences of Structure Formation Differentiating Multiple Folding Pathways of a Small Protein. Journal of the American Chemical Society 143(3):1447-1457. DOI 10.1021/jacs.0c11097. https://doi.org/10.1021/jacs.0c11097
  17. Sandhya Bhatia, Jayant B. Udgaonkar (2022) Heterogeneity in Protein Folding and Unfolding Reactions. Chemical Reviews 122(9):8911-8935. DOI 10.1021/acs.chemrev.1c00704. https://doi.org/10.1021/acs.chemrev.1c00704
  18. Carlos Outeiral, Daniel A. Nissley, Charlotte M. Deane (2022) Current structure predictors are not learning the physics of protein folding. Bioinformatics 38(7):1881-1887. DOI 10.1093/bioinformatics/btab881. https://doi.org/10.1093/bioinformatics/btab881
  19. Kailong Zhao, Pengxin Zhao, Suhui Wang, Yuhao Xia, Guijun Zhang (2024) FoldPAthreader: predicting protein folding pathway using a novel folding force field model derived from known protein universe. Genome Biology 25(1):152. DOI 10.1186/s13059-024-03291-x. https://doi.org/10.1186/s13059-024-03291-x
  20. S. Walter Englander (2023) HX and Me: Understanding Allostery, Folding, and Protein Machines. Annual Review of Biophysics 52:1-18. DOI 10.1146/annurev-biophys-062122-093517. https://doi.org/10.1146/annurev-biophys-062122-093517
  21. Jorge A. Vila (2026) Decoding How Proteins Fold. Biophysica 6(2):36. DOI 10.3390/biophysica6020036. https://doi.org/10.3390/biophysica6020036