Decoding chaperone complexes: Insights from NMR spectroscopy Free

Proteins play a crucial role in virtually all biological processes, often requiring well-defined three-dimensional structures to function properly. However, inside the cell, protein folding is complicated by several factors that enhance the risk of misfolding and subsequent aggregation. Molecular chaperones are essential in assisting many proteins to attain their native structures, preventing irreversible aggregation of folding and misfolding intermediates.^1–6 Chaperones are also vital for maintaining protein homeostasis, as they help repair or refold proteins damaged by environmental stress, including, for example, the effects of reactive chemicals, oxidative stress, and heat. Moreover, disfunction within the chaperone machinery is linked to various diseases, including cancers and a variety of neurodegenerative conditions.

Although crucial for cell survival, chaperones do not exhibit highly specific substrate binding characteristics of many macromolecular machines. Rather, chaperones possess the ability to interact with a diverse array of substrates. This wide-ranging specificity supports a network of folding assistants that operate either with or without ATP, ensuring substrates are safeguarded from engaging in off-pathways along the protein folding trajectory.^1–3,7 Chaperones often interact transiently with their substrates due to the inherently disordered nature of unfolded and partially folded proteins. These transient interactions often result in an ensemble of chaperone-protein substrate complexes that have historically posed challenges for determining the structure of chaperone-substrate complexes by x-ray crystallography and cryo-electron microscopy. However, the dynamic nature of these complexes makes them suitable targets for nuclear magnetic resonance (NMR) spectroscopy.

NMR spectroscopy is an especially powerful method for studying transient chaperone–substrate complexes due to its ability to capture the dynamic and flexible nature of these interactions. Recent technological advances have extended the reach of NMR to larger protein complexes, offering unique insights into chaperone functions. NMR, as a tool, is unmatched when studying transient interactions and intrinsically disordered proteins, continuing to enrich our understanding of protein chemistry and chaperone mechanisms.

A key feature of NMR in studying transient interactions involving sparsely populated states entails the transfer of a magnetic resonance property (e.g., chemical shifts, various relaxation parameters, residual dipolar couplings, etc.) from an invisible “dark state” to an easily NMR observable major species through chemical exchange. Comprehensive reviews of the various NMR approaches are provided in Refs. 8–16. Broadly speaking, these experiments can be divided into three classes: those that rely on: (i) differences in chemical shifts between the major and minor species (e.g., relaxation dispersion and chemical exchange saturation transfer); (ii) the presence of shorter distances between a paramagnetic label and protons in the minor species than the major one (e.g., paramagnetic relaxation enhancement); and (iii) a much higher rotational correlation time (molecular weight) in the minor species than the major one (e.g., lifetime line-broadening and dark state exchange saturation transfer).

In Sec. II, we highlight key examples where NMR has been instrumental in advancing our understanding of chaperone function. In our laboratory, we have investigated the role of several chaperones toward different aggregating client substrates. For this purpose, we have utilized a wide range of NMR techniques to specifically focus on challenging aggregating proteins and the complexes they form with large molecular weight chaperones.

For cellular processes to function correctly, there must be a delicate balance between the synthesis, folding, and degradation of proteins. Molecular chaperones are crucial in maintaining the integrity of the proteome. Chaperones are primarily responsible for the proper folding of newly synthesized polypeptide chains. However, several cellular factors can influence the activity of chaperones. Rugged folding energy landscapes, polypeptides trapped in non-native intermediate states, or off-pathway conformations can lead to protein misfolding. Some of these aggregates are cytotoxic and are associated with a range of misfolding diseases, including Huntington's, Parkinson's, and Alzheimer's diseases, as well as type II diabetes. To mitigate the harmful effects of protein misfolding, certain chaperones are even dedicated to managing aggregation phenomena. The different functions of chaperones are summarized in Fig. 1. Overall, chaperones are essential for maintaining proteostasis, which is crucial for cell viability and function. In our laboratory, we have studied the mechanisms and activities of several chaperones on different client proteins, such as amyloid β42 and the huntingtin protein, which are involved in the etiology of Alzheimer's and Huntington's diseases, respectively.

The bacterial chaperonin GroEL is one of the most abundant proteins in E. coli cells and one of the most extensively studied chaperone systems. The human homolog Hsp60, which possesses ∼50% sequence identity to GroEL, contributes to mitochondrial proteostasis and is implicated in the cellular stress response to carcinogenesis.¹⁷ The crystal structure of GroEL reveals a cylindrical shape, 145 Å in height and 135 Å in diameter, with a central channel approximately 45 Å in diameter.^18,19 GroEL is composed of 14 identical subunits, each 57 kDa in mass, assembled into two interacting heptameric barrel-shaped structures stacked on top of each other. Each GroEL subunit, consisting of 547 amino acid residues, is divided into three distinct domains: apical, intermediate, and equatorial.²⁰ The apical domain of GroEL forms the ends of the cylinder and is responsible for binding both substrates and the co-chaperone GroES. The intermediate domain functions as a hinge, connecting the apical and equatorial domains, and facilitates cooperative conformational changes in the quaternary structure of GroEL. The equatorial domain contains the ATP-binding site and is crucial for forming the heptamer, providing most of the lateral interactions that define the heptameric structure as well as the interactions between the two heptameric rings.^20–22 The co-chaperone of GroEL, known as GroES, is composed of seven identical subunits, each with a molecular weight of 10 kDa.²³ 

Initial studies suggested that GroEL functions as a “foldase,” which initiates the protein folding cycle by binding to non-native substrates within the central cavity.²⁴ This is followed by the co-chaperone GroES binding to the GroEL-substrate complex. GroES interacts with GroEL in an ATP-dependent manner, causing conformational changes in GroEL. These changes are dramatic enough to increase the volume of the GroEL cavity to allow movement of the substrate/peptides inside the central cavity. The inner lining of the cavity walls is hydrophilic with a net negative charge, thus providing an environment conducive for protein folding.²⁵ Finally, upon hydrolysis of ATP, GroEL releases GroES, which causes the substrate to exit out of the cavity into solution.²⁶ Substrates that remain unfolded, misfolded, or partially folded undergo the entire refolding mechanism again either by the same or a different GroEL molecule, a process known as iterative annealing.^27,28 However, a completely different function of GroEL as an “unfoldase” has also been suggested.²⁹ Here, after binding to non-native substrates, GroEL does not facilitate folding but instead provides a hydrophobic surface that binds intermediates, preventing their aggregation and promoting unfolding of misfolded species.

In our laboratory, we have explored the foldase/unfoldase activity of GroEL using a series of multinuclear relaxation-based NMR techniques.^30,31 For a model substrate, we made use of a triple mutant of Fyn-SH3 that exists in equilibrium between its native folded state and a partially folded intermediate.³² We demonstrated that apo GroEL accelerates the overall interconversion rate between the native and intermediate states by 20-fold (Fig. 2). This acceleration in the overall exchange rate is achieved largely through a ∼500-fold increase in the rate constant for the GroEL-bound folded to intermediate state transition relative to that of free SH3. This activity of apo GroEL is likely due to enhanced hydrophobic interactions of the intermediate state with the GroEL cavity, as evidenced by a kinetic deuterium isotope effect that reduces the interconversion rate between the GroEL-bound folded and intermediate states by approximately threefold. These findings illustrate that apo GroEL has both intrinsic unfoldase and foldase catalytic activities, even in the absence of nucleotides or co-chaperones.³⁰ Using a combination of ¹⁵N lifetime line-broadening, dark state exchange saturation transfer (DEST), and Carr-Purcell Meinboom-Gill (CPMG) relaxation dispersion experiments, we further investigated the interaction of the intermediate and unfolded states of Fyn SH3 with apo GroEL.³¹ Our analysis revealed that apoGroEL stabilizes the intermediate folded state relative to the unfolded one.³¹ The unfolded state of the SH3 domain lacks hydrophobic GroEL consensus binding sequences, making it less favored within the GroEL cavity. In contrast, the intermediate state contains a hydrophobic patch that is competent for GroEL binding. Using paramagnetic relaxation enhancement (PRE) measurements with paramagnetic nitroxide labels located at various sites within the GroEL cavity, we demonstrated that Fyn SH3 not only interacts with the hydrophobic inner rim at the mouth of the cavity but also penetrates deeply within the cavity, transiently contacting the disordered C-terminal tail of GroEL.³³ In the case of the folding intermediate state of SH3, the substrate also reaches the base of the cavity. These transient interactions with the C-terminal tail likely facilitate substrate capture and retention before encapsulation.

We have also studied the interaction of GroEL with various intrinsically disordered polypeptide substrates that aggregate and form fibrils. Upon addition of amyloid β (Aβ) 40 and 42 to GroEL, sparsely populated (∼2%) NMR-invisible dark states of Aβ-bound GroEL are formed that can readily be investigated using relaxation-based NMR experiments.^34,35 Our findings revealed that GroEL binds to Aβ40 and Aβ42 at two and three distinct binding sites, respectively.^34,35 These primary sites of interaction are all hydrophobic in nature. The first two regions are the same for both Aβ40 and Aβ42. The additional binding site of Aβ42 is located in the C-terminal region, with Ile41 and Ala42 known to increase the aggregation propensity and neurotoxicity of Aβ42 compared to Aβ40.^36,37 We also demonstrated that GroEL can slow down fibril formation and even dissolve preformed amyloid aggregates.³⁴ Interestingly, upon interaction with the Het-S prion protein, GroEL accelerates aggregation and protofibril formation by two orders of magnitude.³⁸ Eventually, when fibrils are formed, they are densely decorated at regular intervals with GroEL. We also concluded, on the basis of solid state NMR relaxation experiments, that GroEL interaction with Het-S occurs between the mobile regions of the fibril and the apical domain of GroEL. On the other hand, in the case of the huntingtin exon-1 protein (htt^ex1), the interaction occurs between the N-terminal region of htt^ex1 and the apical domain of GroEL.³⁹ This interaction is weak with an affinity in the low millimolar range.

In summary, we showed that GroEL recognizes its client substrates in several ways. Even though the GroEL-bound substrates are sparsely populated, advanced NMR techniques can readily detect these species and provide detailed kinetic information about chaperone activity at atomic resolution (Fig. 3).

The Hsp40 family of proteins, all of which contain a J-domain (and hence often referred to in the literature as J-domain proteins), constitutes the largest family of chaperones in most organisms, playing a crucial role in protein homeostasis. They trigger the hydrolysis of ATP bound to Hsp70 (DnaK), stabilize interactions with substrates, and efficiently facilitate the transfer of misfolded or partially folded proteins to Hsp70.^40–42 J-domain proteins can also function independently to protect against cell death. DNAJB6b (Fig. 4), a prominent J-domain protein, has been implicated in various human diseases and functions as an anti-aggregation chaperone, significantly slowing down amyloid fibril formation. DNAJB6b has been shown to suppress amyloid formation in several proteins that cause Huntington's,^43–46 Parkinson's,^47,48 and Alzheimer's^49–51 diseases.

DNAJB6 contains two globular domains, the J-domain (JD) and a C-terminal domain (CTD) [Fig. 4(a)]. The S/T linker between the two domains is largely intrinsically disordered (except for a ∼10 residue stretch at the N-terminus of the S/T region that forms a helix packed against two helices of the JD domain resulting in an auto-inhibited resting state;⁵² and sites ST16 and ST17 at the C-terminus of the S/T region that form the first β-strand of the CTD domain⁵³) and contains numerous conserved serine (S), threonine (T), glycine, (G), and phenylalanine (F) residues. DNAJB6b has a strong propensity to form highly polydisperse oligomers, ranging from 27 kDa to 1 MDa in size.⁴³ The CTD and the S/T-rich regions on the linker are responsible for this tendency to self-assemble into oligomers. As such, structural studies for DNAJB6b become a challenge. Without structural information, the mechanism by which DNAJB6b impedes aggregation cannot be known in detail. To overcome this challenge, we initially studied a variant of DNAJB6b that can remain primarily in monomeric form.⁵² This allowed the structure of the JD and CTD domains to be solved along with the linker domain using NMR. The structural information of monomeric DNAJB6b revealed that the chaperone is highly dynamic, and several transient interdomain interactions occur, which are crucial for modulating Hsp70 function. DNAJB6b-mediated Hsp70 function and substrate transfer are shown in Fig. 4B. Our results also shed light on DNAJB6b oligomerization, showing that the CTD is responsible for forming self-assembled oligomers via β-strand/β-strand interactions.⁵² Furthermore, we identified residues in the CTD of DNAJB6b that are critical for oligomerization. The S/T-rich segment (ST16/ST17) in the N-terminal β-strand of the CTD induces a backbone twist in the β-strand that stabilizes the monomeric form.⁵³ The transition of the N-terminal β-strand of the CTD from the twisted to a straight configuration facilitates the reversible interconversion between monomers and oligomers of DNAJB6b via an excited state dimer. Since the ST16/ST17 sites in the CTD are important for oligomerization, any mutation in this region is predicted to hinder self-assembly. For example, the T193A mutation in the CTD has been shown to reduce the ability to oligomerize as well as dissociate client aggregate proteins,^53,54 consistent with studies that have linked the T193A mutation to Parkinson's disease.⁵¹ Using various relaxation-based NMR techniques, we demonstrated that the T193A mutation enhances backbone crank-shaft motions,^55,56 resulting in the formation of a partially folded state. This state is an off-pathway state of DNAJB6b, which prevents the formation of functional oligomers.⁵⁷ Overall, these NMR studies^52,57 have provided significant insights into the mechanisms underlying DNAJB6b oligomerization, which is a fundamental aspect of its chaperone activity.⁵⁸ 

Heat shock proteins or Hsps get their name from their ability to respond to cellular stress such as temperature. Hsp104 is one such molecular chaperone that is found in yeast with homologues present in bacteria, protozoa, chromista, and plants but curiously, absent in metazoans. In yeast, a primary role of Hsp104 is to sever/disaggregate prion amyloid fibrils that can then be propagated and transmitted to newly formed yeast buds.⁵⁹ Despite being present only in yeast, Hsp104 has been widely regarded as a powerful tool for studying the effect of chaperones on neurodegenerative diseases in humans. Hsp104 can act synergistically with mammalian chaperones such as Hsp70 (DnaK) and Hsp40 (DnaJ) to form a competent proteostasis molecular machine.⁶⁰ Transgenic mice expressing Hsp104 have shown no abnormalities^61,62 and, in fact, increase stress tolerance of mammalian cells in culture.^62,63 More importantly, Hsp104 can effectively function as a disaggregase in mammals, rescuing animal models of Parkinson's⁶⁴ and Huntington's^61,65,66 diseases. However, the unique power of Hsp104 lies in its ability to remodel extremely stable amyloid fibrils.^64,67–70 This activity is absent in the bacterial homologue of ClpB.^67,71,72 Even though metazoan cells do not possess Hsp104-encoding genes, there is a close relative, known as Hsp110, that can assist in protein disaggregation.^73,74 However, the Hsp104-equivalent mammalian disaggregase machinery, comprising Hsp110, Hsp70, and Hsp40, is unable to remodel mature amyloid substrates.^60,74

The Hsp104 monomer is 908 residues long, 102 kDa in molecular weight, and composed of five domains: the N-terminal domain (NTD), two nucleotide-binding domains (NBD1 and NBD2), a coiled-coil middle domain (MD), and a unique C-terminal domain (CTD). The NTD is essential for mediating initial substrate binding⁷⁵ and even facilitates binding to co-chaperones. The NTD is expected to have high operational plasticity to differentiate and bind different types of substrates, including soluble oligomers, small dimers, large and insoluble aggregates, and prions.^76–78 The two NBDs of Hsp104 are structurally highly conserved. Both NBDs bind to and hydrolyze ATP but display different catalytic properties.⁷⁹ The coiled-coil middle domain of Hsp104 is a key player in substrate disaggregation. Finally, the CTD allows binding to co-chaperones for the release of folded polypeptides^80,81 and also contributes toward the hexameric assembly of Hsp104.^80,82

Hsp104 has been shown to hinder the fibrillization of a wide array of substrates at substoichiometric ratios, yet the mechanism behind this functionality was a mystery. We investigated how Hsp104 prevents fibrillization of Aβ42 under intrinsic conditions, without ATP and co-chaperones, by NMR.⁸³ By monitoring the disappearance of monomeric Aβ42 by NMR, we showed that the kinetics of Aβ42 aggregation under the experimental conditions employed (50 μM Aβ42 and 20 °C) was best described by a branching scheme (Fig. 5): an irreversible on-pathway process leading to fibril formation (entailing primary nucleation, saturating elongation, and secondary nucleation) and a reversible off-pathway process resulting in oligomers that do not mature into fibrils. Using lifetime line-broadening experiments at two different fields of 600 and 800 MHz, we showed that Hsp104 interacts with Aβ42 monomers very weakly (K_Diss ∼2.4 mM) on the submillisecond timescale, and therefore, substoichiometric inhibition of Aβ42 fibrillization cannot involve monomeric Aβ42. Rather, Hsp104 interacts tightly and reversibly (⁠ $Kdissapp$ ∼14 nM) with elongation-competent Aβ42 nuclei that are only produced in nanomolar quantities during the time course of aggregation/fibril formation (Fig. 5), thereby explaining why Hsp104 inhibits fibril formation in an apparent substoichiometric manner. The Aβ42 nuclei bound to Hsp104 can no longer undergo elongation, and, hence, fibril formation is completely inhibited (Fig. 5). This mechanism of substoichiometric inhibition is likely to be general and probably applies to other chaperones that act in a substoichiometric manner, including DNAJB6b.

Molecular chaperones are crucial for maintaining protein homeostasis in living cells, and their interactions with client proteins, whether folded or unfolded, are highly dynamic. Solution NMR spectroscopy is a powerful method for studying chaperone–substrate interactions at atomic resolution, especially when transient, sparsely populated states are involved. While both crystallography and cryo-electron microscopy provide a wealth of detailed structural information on static structures, NMR is ideally suited for studying dynamic systems involving weak biomolecular interactions sparsely populated, transient, excited states that drive many essential biological processes.

This work was supported by the Intramural Program of the National Institute of Diabetes and Digestive and Kidney Diseases at the National Institutes of Health (to G.M.C., Grant DK-029023).

The authors have no conflicts to disclose.

Shreya Ghosh: Writing – original draft (equal); Writing – review & editing (equal). G. Marius Clore: Writing – original draft (equal); Writing – review & editing (equal).

Data sharing is not applicable to this article as no new data were created or analyzed in this review.