J-domain protein chaperone circuits in proteostasis and disease

Abstract

The J-domain proteins (JDP) form the largest protein family within cellular chaperones and, in cooperation with the Hsp70 chaperone system, orchestrate a plethora of distinct functions, including those that help maintain cellular proteostasis and development. JDPs evolved largely through the fusion of a J-domain with other protein subdomains. The highly conserved J-domain facilitates the binding and activation of Hsp70s. How JDPs (re)wire JDP-Hsp70 chaperone circuits and promote functional diversity remains insufficiently explained. Here, we discuss recent developments that show the presence of a high degree of regulation built around J-domains to ensure correct pairing and assembly of JDP-Hsp70 machineries that operate on different clientele under various cellular growth conditions.

J-domain proteins form a front-line protein quality control layer in cellular proteostasis

Proteins by nature are dynamic, and newly synthesized polypeptides and non-native intermediates are at increased risk of misfolding and forming potentially cytotoxic non-native oligomers and aggregates. Through evolution, cells have invested in developing a highly conserved multitiered protein surveillance system, the proteostasis network (PN), (see Glossary) which integrates signals and regulates fluxes through different aspects of protein biogenesis (e.g. protein synthesis, folding, transport, and degradation) [1]. An essential wing of the PN consists of molecular chaperones (see Glossary), a set of proteins that guide and facilitate other proteins in acquiring native folds that support proper functionality. The human chaperome, namely the ensemble of chaperones and their co-chaperones, consists of more than 300 members [2], including the J-domain proteins (JDP, J-protein, and Hsp40), whose role is to recruit the 70 kDa Heat shock protein (Hsp70) into selected client proteins.

JDPs form one of the largest and most diverse chaperone families in metazoa, including humans. The evolutionary expansion of the JDP family paved the way for the Hsp70 chaperone system to gain new wide-ranging functions, including the formation of a sophisticated protein quality control (PQC) layer within the PN, a key aspect that may have contributed to multicellularity. The highly versatile JDP-Hsp70 chaperone machines support a host of housekeeping and stress-related activities that protect cells against proteotoxic stresses [3, 4]. Some key housekeeping activities guided by JDPs include de novo protein folding, transport of proteins across cellular membranes, assembly/disassembly of protein complexes, formation of biosensors, and regulation of protein activity (Table S1). In stressed cells, JDPs help to switch the Hsp70 machinery to avoid premature apoptosis and modulate cellular stress responses and protect aberrant proteins from aggregation, refold misfolded proteins, solubilize aggregated proteins, and cooperate with the ubiquitin proteasome system and selective autophagy to degrade kinetically trapped and damaged proteins (Figure 1A, Table S1). In addition to general housekeeping and stress-related functions, specialized JDPs guide Hsp70 chaperones to perform an assortment of non-PQC activities such as gene silencing [5] and mRNA processing [6].

Figure 1. Selection of client proteins by J-domain proteins for different Hsp70 chaperone functions.

(A) Client selection by JDPs to perform housekeeping and stress-related activities in cells. Shown examples include promiscuous and selective client bindings, as well as JDP mechanisms that do not involve direct client binding. The J-domain is represented as a blue filled circle and dash. The client protein is shown in magenta. (B) Schematic view of the canonical Hsp70 functional cycle. Client proteins are selected and handed over to Hsp70 bound to ATP (nucleotide binding domain (NBD) in dark grey; substrate binding domain (SBD) in light grey). The physical interaction between JDP and Hsp70 is mediated by the J-domain (JD, filled blue circle and dash). JD and client binding synergistically trigger Hsp70’s ATP hydrolysis, resulting in conformational changes in the SBD, which has a high affinity (ADP state) for the client. Subsequently, nucleotide exchange factors (NEF) induce ADP dissociation and re-binding of ATP, converts Hsp70 to low-affinity ATP state. This facilitates the release of the client protein. The released client can fold to its native state or, alternatively, re-enter the Hsp70 cycle.

Recent technological advances have considerably increased our mechanistic, structural, and functional understanding of the JDP family. Herein, we describe genetic and structural regulatory elements built around JDs that help modulate the selective assembly and activation of various JDP-Hsp70 machineries with distinct functions. Dysregulation of this central Hsp70 chaperone (re)wiring system can critically affect various PQC functions in the PN under normal and stressed cellular conditions leading to disease. We highlight some of the challenges and outstanding questions surrounding JDPs as they emerge as an important class of proteins associated with a broad range of human diseases.

The J-domain protein-Hsp70 chaperone cycle

JDPs emerged in terrestrial bacteria [7] and co-evolved to perform the important tasks of selecting client proteins for Hsp70 and allosterically stimulating ATP hydrolysis in Hsp70, which is crucial for client capture [8–10] (Figure 1B). The canonical chaperoning cycle is initiated with an Hsp70 (in the ATP-bound state) and a JDP loaded with a client protein. The J-domain (JD) of JDPs, an evolutionarily conserved namesake domain (consensus Pfam domain PF00226), mediates the primary physical interaction with Hsp70 [11–13] to form a transient ternary protein complex [14]. The formation of the Hsp70-JDP-client complex facilitates the transfer of client proteins from JDPs to Hsp70 (Box I). In parallel, this assembly induces ATP hydrolysis in the nucleotide binding domain (NBD), which in turn is necessary to change Hsp70 from an open substrate binding domain (SBD) (client receiving) conformation, with a high client binding rate, to a closed SBD conformation, where the trapped client has a very low dissociation rate [15, 16] (Figure 1B). This ATPase-driven transition results in accelerated ATP hydrolysis and energy consumption with a consequent non-equilibrium enhancement of the Hsp70 client binding affinity (ultra-affinity), which is higher than the affinity of the ATP- and ADP-bound states [15]. The energy released from ATP hydrolysis drives the iterative client binding cycles. These binding events allow the unfolding and refolding of some client proteins [17–19] thus pointing to different chaperoning actions of Hsp70 (e.g. holding versus folding). The clients are then released from Hsp70 via nucleotide exchange factor (NEF)-accelerated ADP release/ATP rebinding or spontaneously [20]. During eukaryotic evolution, the modulation of Hsp70 ATPase activity through JDP and client binding has undergone further fine-tuning [21]. How these changes translate into functional output remains elusive due to the poor correlation between these activities [22].

Box I. Client handling by J-domain proteins.

Client binding to JDPs

The client-binding subdomains in JDPs come in various flavours [38], enabling these cochaperones to interact with a broad range of client proteins. Class A, class B, and a few class C JDPs appear to have some degree of promiscuous client binding, while other members display high substrate specificity. Class A/B JDPs can interact with a wide array of client proteins in non-native (e.g. partially-fully unfolded, misfolded or aggregated) states [14, 27, 62, 80, 102–105]. Generally, non-native client proteins expose peptide regions enriched in hydrophobic amino acids. During protein (re)folding, such regions get buried in the protein core. Class A and B JDPs recognize these exposed sites as discriminatory signatures between native and non-native clients and interact through the grooves located in the β-sandwich-type substrate/client binding domains (e.g. CTDI and CTDII, Figure 3B and C) [14, 48, 80, 106, 107]. The selectivity of clients comes partially from the preference of these grooves for peptide regions with distinct amino acid compositions [108, 109]. The two β-sandwich subdomains within the same JDP could have distinct peptide binding preferences (Box I, Figure IA). In bacterial DnaJ, both CTDI and CTDII bind to hydrophobic regions, but CTDI prefers peptide regions depleted of aromatic residues, while CTDII selects acidic and aromatic residue-enriched regions [14].

The multimeric state of JDPs could further influence the binding efficiencies towards client proteins. The class A/B JDPs that function as stable homodimers (Figure S1A and B) present four β-sandwich-type client-binding sites [110]. Higher-order oligomeric states among and between class A and B JDP dimers were recently reported, and these scaffolds help recognize certain clients such as aggregated proteins [27, 62]. Other JDP multimers, for example DNAJB6, which appears to target amyloidogenic proteins, have also been detected [111]. The avidity for aggregated [27] or unstructured [14] proteins may change due to the presence of increased client binding sites from JDP dimerization/oligomerization and different JDP binding modes [14, 112]. The requirement of various JDPs or JDP configurations to selectively target different aggregate types [27, 62, 104, 105, 113] may result from a culmination of these client-binding features. Additional client interaction sites could also exist. For example, intrinsically disordered G/F-rich regions in some JDPs could provide conformational flexibility to bind to differently shaped clients.

Highly specialized client protein interactions can also be observed in JDPs. For example, Jac1/DNAJC20 binding to Isu1 [114] (Box I,Figure IB) and Auxilin/DNAJC6 binding to clathrin triskelia [25] are prime examples of such interactions. The fine-tuning of client selection by these cochaperones has evolved to extreme levels in plants that contain a more complex network of JDPs and Hsp70s [104, 115]. All plant subcellular compartments (cytosol/nucleus, endoplasmic reticulum, mitochondria, and chloroplast) host an increased number of JDPs as a result of multiple rounds of gene duplication. It appears that some of the plant JDPs created through this expansion have evolved into acquisitions of new activities. However, the others remain largely functionally redundant, but express themselves differentially under specific growth conditions in distinct plant tissues [116].

JDP binding sites in clients appear to be somewhat mutually exclusive from the Hsp70 interaction sites [25, 27, 88, 102, 117]. In some cases, JDP binding has been shown to trigger conformational changes in the client [25, 27, 118], which could expose new Hsp70 binding sites [102, 103]. Access to JDP binding sites changes with the degree of (un)folding of a protein, and unique combinations of JDP and Hsp70 bound to those exposed sites can lead to different PQC outcomes [108, 117].

Delivery of client proteins

Numerous mechanisms support client transfer from JDPs to Hsp70s. Structural elements, such as the M2 motif in DNAJA1, play an important role in releasing clients to Hsp70 [50]. For particular clients, transfer factors such as NudC and Cdc37 could short-circuit the flow of clients to downstream chaperone systems (Box I Figure IC). For example, NudC can remove Hsp70 from JDP-client-Hsp70 complexes and transfer clients directly to Hsp90 [119, 120], thus bypassing the classical transfer of clients from Hsp70 to Hsp90 assisted by TPR binding proteins such as HOP (or Cdc37). Furthermore, sequence elements, such as the presence of an Ile versus Met immediately upstream of the EEVD motif in Hsp70 and Hsp90, help JDPs such as DNAJB4 discriminate between two main chaperone systems in cells [82] (Box I Figure IC). PTMs have also been implicated in driving conformational changes in JDPs that specify downstream partner proteins [121]. In summary, the above findings provide important features that tweak the client handover between JDP and major downstream chaperones/partner proteins that could lead to different functional outcomes.

Box I Figure I. Client binding and transfer from J-domain proteins to downstream chaperone systems.

(A) Structural view of the interaction between PhoA (unstructured client protein) and CTDs of T. thermophilus DnaJ (PDB ID 6PSI) [14]. The DnaJ residues (grey) interacting with PhoA (pink) are highlighted on the CTDI (left zoom) and CTDII (right zoom). The grey spheres represent C-alpha atoms, and the black spheres represent C-beta atoms. (B) Structural view of the Isu1 binding site at the C-terminus of Jac1 of S. cerevisiae. Zoomed out inset shows residues interacting with Isu1 (PDB ID 3UO2) [114]. Isu1 is a scaffolding protein for the de novo synthesis of iron-sulfur clusters within mitochondria. (C) JDP-mediated client handover pathways to various downstream chaperone systems. The presence of an isolucine vs. a methionine upstream of the EEVD motif allows selective client (magenta) transfer between specific JDPs (e.g. class B) and Hsp70s. This difference can further differentially regulate interactions with Hsp90 (see inset). On the contrary, binding of NudC to class A and B JDPs could dislodge Hsp70 and facilitate the direct transfer of some clients to Hsp90. This bypasses the downstream client transfer between the two chaperone systems via the tetratricopeptide repeat (TPR) domain containing proteins (e.g. HOP and Cdc37). The inset shows how HOP scaffolds Hsp70 and Hsp90 to facilitate client transfer. The interaction of Hsp70 with E3 ubiquitin ligases such as CHIP through the EEVD motif could result in the ubiquitylation and degradation of clients that cannot be efficiently chaperoned.

In the classical Hsp70 cycle, upon transfer of the client protein (Box I), the JDP is released from Hsp70 (Figure 1B). This is the case where JDP Hsc20 dissociates upon client transfer to Ssq1 (a yeast Hsp70), which constitutes an essential step in the biogenesis of iron-sulfur clusters [23]. Whereas this may be the more general paradigm, alternative scenarios exist. For example, during the disassembly of the clathrin coat, DNAJC6 (Auxilin), while stably bound to its client, could recruit several Hsp70s (through multiple rounds of engagement) to maximize conformational distortions in triskelia units [24]. When cage disassembly occurs, DNAJC6 binding sites are disrupted, triggering JDP dissociation from its client protein [25, 26]. Similarly, emerging findings indicate that JDPs involved in protein disaggregation (e.g. DNAJB1) could remain associated with aggregated proteins to support more than one round of recruitment of Hsp70 [27].

The architecture of the J-domain

The JD was first identified in the Escherichia coli DnaJ protein [28] and is approximately 70 amino acids long. The basic architecture of this subdomain includes a hairpin structure and four helices (termed I-IV) (Figure 2A). Helices II and III form an antiparallel coiled coil through interchain hydrophobic interactions, while helices I and IV located at the N and C terminus of the JD further stabilize the antiparallel hydrophobic core [29, 30] (Figure 2A). Functional JDs contain a signature His-Pro-Asp (HPD) motif between helices II and III, which is essential for stimulation of ATPase activity in Hsp70s [12, 31] (Figure 2B and C). Most JDs carry a positively charged electrostatic patch rich in basic amino acids in helix II, which appears to help position the JD onto the negatively charged crevice in the “underbelly” of the ATP bound Hsp70.

Figure 2. J-domain mediated activation of Hsp70.

(A) Structural view of the J-domain (PDB ID: 1XBL). Helices I-IV are highlighted, as well as the highly conserved His-Pro-Asp (HPD) motif, crucial for biding and stimulating ATPase activity in Hsp70s. (B) Structure of the complex formed between the J-domain (JD, blue) of bacterial JDP (DnaJ) and Hsp70 (DnaK) (PDB ID: 5NRO). Inset panels highlight key residues that form the interaction interface between Hsp70 and the JD (helix II, helix III and HPD loop region). (C) Regulation of Hsp70 ATPase activity by JD and client binding. JD, Hsp70, and client are indicated in blue, grey, and black, respectively. Scenario 1: Binding of ATP induces the two NBD lobes of Hsp70 to rotate into a conformation that keeps the catalytic residues in an ATP hydrolysis-incompetent conformation. This confirmation allows SBDβ to dock onto the lobe-rotated NBD, which is stabilized through the interaction of residues D481 with I168 and K414 with D326. The dissociation of SBDβ from NBD in Hsp70 is infrequent in the absence of JDP and client binding, resulting in a low basal ATP hydrolysis rate. Scenario 2: Client binding can stimulate the dissociation of the SBDβ from the NBD, leading to enhanced ATP hydrolysis in Hsp70, but the linker often slips out of the lower crevice, failing to arrest the lobes in the competent conformation for ATP hydrolysis. This results in a marginal increase in ATPase activity that could prematurely release the client. Scenario 3: The binding of JD prevents the slipping of the linker from the lower crevice and increases the efficiency of the client-stimulated dissociation of SBDβ from NBD. This arrests the back-rotating NBD lobes in the ATP hydrolysis competent conformation, leading to efficient trapping of the client protein. (D) JDP-Hsp70 interaction network. The interaction data are extracted from the BioGrid database, displaying only interactions detected at the physical level. The red connection lines represent interactions that are supported by direct in vitro experimental evidence (“Reconstituted Complex” in BioGrid). ‘Atypical’ denotes putative Hsp70 chaperones with a non-canonical domain architecture (e.g. only the NBD present). The node sizes are scaled by the number of interactions reported for each JDP and Hsp70. Chaperones that are predominantly found in the ER and mitochondria are highlighted by red and green circles, respectively. The rest of the chaperones are located in the cytosol/nucleus.

Structural elements mediating the J-domain-Hsp70 interaction

We now have a precise map of the JD-Hsp70 interface showing structural elements of the JD required to stimulate ATP hydrolysis and allostery in Hsp70 [10, 13, 32–34]. This evolutionarily conserved interaction surface is formed by the helix II-loop (containing the HDP motif)-helix III, and a structurally contiguous region of Hsp70 involving lobe IIA of the NBD, the interdomain linker, and the β-sandwich of the SBD. The JD docks on top of the interdomain linker of ATP bound Hsp70, which is embedded between the NBD and SBDβ via polar and electrostatic interactions. The docking is stabilized by a) polar interactions between a positively charged electrostatic patch in helix II (proximal to the HPD motif) and the negatively charged under-cleft of the NBD, and b) hydrophobic interactions between helix III and SBDβ [13, 32–34] (Figure 2B). These interactions position the HPD motif to come into contact and activate an intramolecular signaling pathway that converges to the catalytic centre of Hsp70. These events pair client loading with JD action, which synergistically stimulates ATPase activity in Hsp70 [13, 35–37]. Initially, the JD holds the interdomain linker in place to prevent the NBD lobes from rotating to an optimal conformation for ATP hydrolysis and alters the polar interactions within the NBD to facilitate the passage of client-binding signals to the catalytic center [35, 36] (Figure 2C).

The J-domain protein family

Multiple assemblies of JDPs and Hsp70s exist in all organisms, including humans (Figure 2D). JDPs appear to have been largely derived through progressive fusion and/or loss of different protein domains over time and, as a consequence, exhibit high variations in both protein length, architecture, and function [29, 38–40]. Based on the overall composition of JDPs, 1945 different domain architectures with more than 1500 distinct domains were recently identified in all kingdoms of life [39] (see the Outstanding Questions box). The subdomains beyond the JD serve to a) regulate interactions between JDP and Hsp70 (e.g. regulating ATP hydrolysis in Hsp70), b) select and hand over client proteins to Hsp70, c) scaffold JDP-Hsp70 machinery with other partner proteins/assemblies, d) target JDP-Hsp70 machinery to distinct cellular localizations. Some of these well-established features have already been reviewed by others [38, 40, 41] (Box II).

Outstanding questions.

Specific JDPs driving known Hsp70 functions in cells remain largely unknown.

What is the extent of functional redundancy among JDPs in cells?

Can selective JDP inhibitors be generated with reduced off-targets?

How does cellular signaling influence (re)wiring of JDP-Hsp70 chaperone circuits?

How do PTMs modulate the JDP-Hsp70 chaperone network during cell growth and organismal development?

Is the JDP-Hsp70 chaperone network influenced by epigenetics? How do aging and disease impact JDP driven chaperone circuits?

What new cellular activities have arisen with the evolutionary expansion of the JDP family, for example, in humans?

Box II. J-domain independent functions.

Although rare, JDPs could have functions that do not require JD-mediated recruitment of Hsp70. For example, JD-truncated JDP RSP16 could still promote the assembly and maturation of the radial spoke required for the movement of the flagellar limb in C. reinhardtii [122]. Similarly, yeast JDP Sec63 lacking JD is capable of partially supporting the translocation of ER-targeted polypeptides by facilitating the opening of the Sec61 channel without the involvement of Bip, the ER-resident Hsp70 [123]. Furthermore, it was shown that a mutant of Cwc23 lacking JD could still help recycle spliceosomes during pre-mRNA splicing by directly interacting with the splicing factor Ntr1. However, JD-intact Cwc23 is required when additional defects are introduced into the spliceosome disassembly machinery [6].

The naturally occurring JD-lacking isoforms of two human JDPs (DNAJB12 and DNAJB14) were recently characterized. These isoforms operate differentially compared to their full-length JDPs with intact JDs. Interestingly, the short JD-lacking isoform of DNAJB12 (translated via an alternative promoter) significantly decreases polyQ-type amyloid aggregation in cells compared to its full-length counterpart, showing the opposite effect [87]. In contrast, the short isoform of DNAJB14, derived from alternative splicing, shows complete loss of the aggregation prevention function of the full-length protein [87]. This is not suprising given that this isoform lacks most part of the protein. These examples reveal a complex interplay between naturally-occurring isoforms of JDPs based on their ability to recruit Hsp70. A set of proteins homologous to some JDPs, but lacking functional JDs (lack of a JD or contains a pseudo-JD-like domain without an HPD motif), was recently identified in plants [124, 125]. These proteins may shed important information about the evolution and expansion of this cochaperone family. Although these proteins are not considered JDPs, they may contribute to the functional diversification and regulation of various Hsp70 chaperone activities by possibly competing for client proteins under different cell growth conditions. However, in general, caution must be taken when interpreting JD-independent functions of JDPs.

The classification of proteins containing JD has been a challenge because of the enormous structural and functional diversity of the JDP family. Initially, members of the JDP family were classified into three subdivisions (class A, B, and C) based on their structural similarity to that of E. coli DnaJ, the first JDP identified [42] (Figure 3A). JDPs belonging to class A were defined as full homologs of bacterial DnaJ consisting of an N-terminal JD followed by a G/F-rich region, a C-terminus containing two β-sandwich subdomains (CTDI and CTDII) with a zinc-finger-like region (ZFLR) inserted into CTDI [29]. Class A JDPs form stable homodimers with the aid of a dimerization domain located at the C terminal (Figure 3B; Figure S1). The JD, as explained previously, is primarily used to interact with Hsp70. The G/F-rich region is typically considered to be largely unstructured. This region plays a regulatory role in the binding of JDs to Hsp70 [44–47]. The G/F rich was recently shown to promote the liquid-liquid phase separation of some JDPs in mammalain cells [43]. The observed effects are, however, marginal under the tested conditions. The β-sandwich CTDs serve as interaction sites for a wide variety of unfolded, misfolded, or aggregated client proteins with varying specificities [14, 48] (Box I). The role of the ZFLR remains largely unclear and could possibly act as an additional client interaction site for class A members [49] and/or play a role in client transfer [50].

Figure 3. Characteristics of J-domain protein classes.

(A) Classifications of the JDP repertoires by class for H. sapiens, S. cerevisiae, and E. coli. The JDPs found predominantly in the ER and mitochondria are indicated by red and green circles, respectively. The rest of the chaperones are located in the cytosol/nucleus. (*) Proposed reclassification of some class B human JDPs as class C members according to Malinverni et al. [39] (B) Canonical domain architectures and structural view of class A JDPs. (C) Canonical domain architectures and structural view of class B JDPs. (D) Structural views of the nine most abundant JDP architectures in class C. The J-domains (JD) are colored blue. The models shown are full-length predictions obtained by AlphaFold2 as deposited in the EBI database [83]. Below each structure, we show the corresponding simplified domain architecture based on the PFAM domain annotations (excluding the J-domain (PFAM ID: DnaJ) for clarity). Examples of JDPs with the corresponding domain organization in H. sapiens (H.s.), S. cerevisiae (S.c.), and E. coli (E.c.) (when available) are also reported below each example structure. For DUF3444, which is found only in plants, we report an example of a JDP from A. thaliana.

Class B JDPs were loosely defined as members containing an N-terminally localized JD, a G/F-rich linker, and various C-terminal subdomains. Based on large-scale JDP sequence analysis, class B was recently redefined to contain members that contain two β-sandwich subdomains and a dimerization domain similar to that of class A members, except for the absence of a ZFLR [39] (Figure 3C; Figure S1).

Class C, on the other hand, is a catchall group that contains the rest of the JDPs. This class includes the lion’s share of JDPs and contains highly diverse members with only JD in common with DnaJ. The most abundant JDP architectures within this class represent members with distinct Pfam domain types DnaJ-X, DnaJ-like_C11_C, TerB, Znf_C2H2_jaz, zf-CSL, DUF1977, HSCB-C, Sec63 and TPR [39] (Figure 3D; Figure S1). The process of combining a JD(s) with different protein subdomains appears to have given rise to many class C members and can be viewed as a naturally occurring protein engineering feat of evolution. Through this process, Hsp70 has gained the ability to support a multitude of vastly different biological processes, a clear indication that this feature has provided advantages in cellular fitness [40, 51]. On the contrary, some members within this class have re-emerged during evolution with just a JD [39]. One such JDP in bacteria (AtcJ) is required for growth at low temperatures [52] while another in humans (DNAJB3) was shown to be associated with obesity and insulin signaling [53], indicating that some of these ‘dwarfed’ class C members could play crucial roles in cellular stress and metabolism.

The diversity within the JDP classes could further increase due to naturally occurring functional isoforms generated, for example, via alternative splicing of transcripts. Translated splice variants are believed to function in closely related biochemical pathways. For example, both full-length and short isoforms of human DNAJA3 (Tid1) participate in protein folding, but target different clientele in different cellular compartments (mitochondria vs. cytosol) [54]. Similar alternative splice variants with localization and functional differences have been reported for other JDPs (e.g. DNAJB2a vs DNAJB2b and DNAJB6a vs DNAJB6b) [38]. Recent genome-wide scans reveal that many JDP members have multiple isoforms, suggesting that this is a widespread phenomenon within this cochaperone family. These variants may be differentially expressed based on cellular location, cell/tissue type, or/and growth condition to modulate chaperone functions, adding yet another layer of complexity to this chaperone network (see Outstanding Questions box).

(Re)Wiring the interactions in the JDP-Hsp70 network via the J-domain

Organisms ranging from viruses to prokaryotes to eukaryotes rely on multiple JDP-containing Hsp70 systems to facilitate a wide array of biological activities [38–40]. There are 50 JDPs and 13 Hsp70s (8 canonical; 5 atypical) in human cells. Given the broad range of fundamental biological processes supported by the JDP family of proteins that affect both cell viability and organismal development (Table S1), it is not surprising that these cochaperones are involved in many human pathologies ranging from neurodegenerative disorders, cancer, metabolic disorders to infectious diseases (see section on disorders associated with J-domain proteins; Table S2).

In most cases, several JDPs partner with a single Hsp70 [55] (Figure 2D). Furthermore, these cochaperones generally act as substoichiometric catalysts of Hsp70s both in vitro and ex vivo [56]. Up-/down-regulation of multiple JDPs and drastic reorganization of the JDP-Hsp70 chaperone network have been observed to support cell differentiation and viability under various growth conditions [57–60]. Transcriptional programmes (e.g. heat shock response and unfolded protein response) that respond to various stresses change the gene expression of particular JDPs and Hsp70s [59, 61]. Assembly of distinct chaperone machines can be achieved by adjusting the celluar concentrations of certain pairs of JDP and Hsp70, which provides a powerful method to help rewire the JDP-Hsp70 circuits. However, this paradigm alone is insufficient to achieve tight regulation of this chaperone system considering the mere size and interconnectivity of the JDP-Hsp70 network (Figure 2D) where some components are redundantly used to assemble chaperone machines with different and sometimes opposing functions [62, 63] (Tables S1 and S2) (see Outstanding Questions box). Consistent with this idea, we observe a high degree of regulatory elements built in the JDs of JDPs to specifically direct the function of Hsp70 machines.

Specifying partner Hsp70s through discriminatory genetic signatures in the J-domain

A series of recent findings have revealed multiple regulatory mechanisms involving JD that tune and tightly modulate distinct JDP-Hsp70 assemblies. Although the tertiary structures of different JDs are remarkably similar, the interaction between some JDs and Hsp70s has specific outcomes [64, 65] (Figure 2D). Indirect experimental evidence using JDP-stimulated ATPase activity measurements in Hsp70s as a readout shows that JDs are not fully interchangeable with respect to their collaboration with different Hsp70 paralogs [66, 67]. Furthermore, exchanging JDs between JDPs frequently generates a non-functional chimera [65, 68]. For example, switching the JD of yeast Pam18, which functions with mitochondrial Hsp70 (mtHsp70) in the protein import machinery, to that of yeast Mdj1, which cooperates with the same mtHsp70 to support protein folding in the mitochondrial matrix, does not facilitate preprotein translocation [68]. There are, however, cases where JD swapping could also result in fully functional JDP chimeras [65]. The exchangeability seems to arise, in part, when the JDPs have a common Hsp70 partner (e.g. DnaJ and DjlA interact with the same Hsp70, DnaK) as opposed to different partners (e.g. DjlC and DnaJ interact with HscC and DnaK, respectively) [69, 70].

Residues distributed along with the Hsp70 binding interface, particularly charged amino acids, appear to play a dominant role in tweaking the JD-Hsp70 interaction. The replacement of the JD of yeast Sec63 with the JD of Sis1 results in the formation of a defective chimeric JDP, in vivo [71]. However, Sis1 JD containing Sec63 can be reactivated by changing a few charged exposed amino acids along helices II and III [71]. Similarly, a specific set of charged residues in the JD of Hsc20 helps to correctly pair it with its co-evolving Hsp70 [72]. A recent large-scale phylogenetic analysis (comprising >150’000 JDP sequences from prokaryotes and eukaryotes) identified a series of discriminatory sequence positions in JDs (that help identify JDPs according to type, class, cellular localization and phyla) overlapping with the Hsp70 binding site [39]. These positions located proximal to the HPD motif are occupied largely by charged residues (or residues that could get phosphorylated) suggesting that they are part of a discriminatory electrostatic potential signature (DEPS) that tweaks JD-Hsp70 interactions [39]. Figure 4A shows a zoomed-out snapshot of probable DEPSs embedded close to the Hsp70 binding sites in the JDs of human JDPs. A closer inspection shows that the more outlying electrostatic distribution patterns are largely visible in the JDs of the JDPs reported to have the lowest amount of Hsp70 partners (e.g. ≤3, Figure 2D). This provides a plausible explanation as to why JDs known to interact with specialized Hsp70s exhibit the most striking sequence divergence, with concomitant alterations in their cooperating Hsp70 partner [37, 39, 72]. In further agreement, Hsp70 homologues that do not have a classical JDP partner, such as ribosome-associated HSPA14 (Ssz) or cytosolic Hsp110s and ER Hsp170, which are members of the Hsp70 superfamily but act as NEFs, show little or no conservation in their JD-interacting surface [73]. The discriminatory genetic signatures embedded in JDs can be further modified by post-translational modifications (PTM) that could change default interactions between specific pairs of JDPs and Hsp70s, on demand [74], and link these chaperone circuits to cell signaling (see Outstanding Questions box). In general, the presence of these discriminatory genetic signatures in JDs opens the possibility to fine-tune the JD-Hsp70 interaction, thus establishing a gradient of interaction preferences within the JDP-Hsp70 cellular network.

Figure 4. Regulation of the J-domain protein-Hsp70 network through the J-domain.

(A) Distribution of exposed electrostatic potentials (iso-surfaces shown at +/− 1kcal/mol, blue: positive potential, red: negative potential) in the J-domains of 50 human JDPs. These electrostatic potential patterns and post-translational modifications could fine-tune the selection of Hsp70. In the absence of structural models, the J-domains were modelled by AlphaFold2. (B) Inhibition of the J-domain of CbpA (blue) by binding of CpbM (orange) (PDB ID 3UCS). The inset displays the critical residues involved in the JD-CbpM interaction interface. (C) Regulation of the J-domain by interactions with the G/F-rich region. Left: Interaction of the yeast Sis1 J-domain with its G/F-rich region through the salt bridge E50-R73 (PDB ID 6D6X). Right: Interaction between helix V and the J-domain of human DNAJB6 (PDB ID 6U3S). (D) Auto-inhibition of human DNAJB8 through the interaction of its J-domain with residues in the C-terminal domain (CTD) (Model provided by L. Joachimiak, Ph.D.).

Regulating JDP-Hsp70 assembly by inhibitor proteins of the J-domain

CbpM is a bacterial JDP inhibitor that can directly bind and block JDs. CbpM clasps the JD of CbpA and covers the helix II-loop-helix III region vital for the Hsp70 interaction [75] (Figure 4B). CbpM is highly selective in its targeting and binds specifically to the JD of CbpA and not to that of DnaJ, the most abundant JDP in bacteria. The discriminatory region is largely assigned to the area that encompasses residues located within the helix I-loop-helix III region [75, 76]. The CbpM gene lies downstream of CbpA in the same operon, suggesting that the CbpA-DnaK pairing is highly regulated by this JD inhibitor [75, 77]. During certain growth phases (e.g. stationary phase), CbpA is released from CbpM to interact with DnaK and form functional chaperone machines that exert specific cellular activities [78]. This JD inhibitor shows a high degree of structural similarity to MerR-like transcription regulators [76], but orthologs with similar activity have not been reported in eukaryotes. Another example can be found in yeast, where a regulatory component of the mitochondrial protein import machinery named Pam16 uses a JD-like domain (the invariant tripeptide motif HPD is replaced by DKE/S) to interact directly and inhibit the activity of the JD of Pam18. This interaction appears to result in the formation of a pseudosymmetric structural arrangement that prevents mHsp70 from interacting with Pam18 JD [68, 79]. Again, it is possible that PTMs at JD serve such regulatory activities in eukaryotic JDPs, and therefore distinct inhibitor proteins are not necessary.

J-domain associated auto-inhibitory mechanisms determine the action of JDP-Hsp70 assemblies

Specific JDPs have evolved inbuilt auto-inhibitory mechanisms to modulate the interaction between their JDs with partner Hsp70s. In Sis1, a conserved intramolecular bridge between the JD and the G/F-rich region (residues E50 to R/K73) was recently shown to considerably limit the free movement of the JD (Figure 4C, left). The ability of Sis1-type JDPs to effectively participate in protein (re)folding seems to be controlled by the formation of this bridge [44, 45]. Subsequently, another auto-inhibitory element that regulates JD was identified in the G/F-rich region of DNAJB6 [46]. This element, termed helix V, interacts with helices II and III of the JD and competes with the Hsp70 docking (Figure 4C, right). Helix V is conserved in the G/F-rich regions of other JDPs such as DNAJB8 (sequencewise) and DNAJB1 [80]. This particular intramolecular autoinhibitory activity is important for amyloid disaggregation by Hsp70 disaggregases, in vitro [80]. In addition, a secondary contact formed between the negatively charged Glu-Glu-Val-Asp (EEVD) motif-containing region (located at the extreme C-terminus of eukaryotic cytosolic Hsp70s) and the positively charged groove in the CTDI of DNAJB1 appears to release the JD from this autoinhibition. However, the manner in which the EEVD motif-containing region is able to trigger the displacement of the JD from helix V, possibly in trans, remains unknown. In addition to further tuning of the Hsp70 binding, the EEVD motif may also help determine the architecture and stability of particular chaperone machines [60]. Together, the above findings show that, though seemingly inconsequential, the G/F-rich region plays an important role in regulating the binding of Hsp70 to the JD of selected JDPs, with important functional consequences.

The C-terminal domain of DNAJB8 also acts as an auto-inhibitory element that controls the binding of its JD to partner Hsp70 (Figure 4D). On the basis of structural analysis, this element somewhat mimics the topology of the JD-binding interface on Hsp70s [47]. Sequence alignment and in silico structure prediction (AlphaFold2) analysis indicate that DNAJB1, DNAJB6, and DNAJB8 contain more than one of the auto-inhibitory elements (e.g. DNAJB1: elements 1 and 2; DNAJB6: elements 2 and 3; DNAJB8: elements 1, 2, and 3). How these multipronged regulatory elements are integrated and coupled to the rewiring of Hsp70 and client protein interactions in cells requires further investigation.

Similar regulatory mechanisms may also exist in other JDPs. Low resolution protein structures and protein docking simulations indicate that JDs could interface with CTDs in class A members [62, 81]. There is also evidence that the region containing the EEVD motif could interact with ZFLR in some members of class A, but the functional consequences of this interaction are unclear [44, 45, 80, 82]. According to AlphaFold predictions, helix V-like structures with varying confidence scores could also be observed in the G/F-rich regions of non-class B members (e.g. DNAJC7) [83], indicating that some of these elements are more universal than previously assumed.

The formation of the J-domain-mediated JDP scaffolds promotes different chaperone machines

A set of metazoan Hsp70-based protein disaggregases is formed through transient heterooligomerization between homodimers of class A and class B JDPs [62, 63, 84]. A negatively charged electrostatic patch opposite the Hsp70 binding site on the JDs is used to interact with the CTD (close to the hinge region between CTDI and CTDII) of the opposite class member. Mixed class JDP scaffolds generated via JD-to-CTD interactions help form potent protein disaggregases that target amorphous protein aggregates [62, 85]. The formation of JDP scaffolds is a distinctive requirement for the assembly of Hsp70-based disaggregases over foldases [62], thus providing yet another example of how JD-mediated interactions drive the assembly of distinct chaperone machines. However, not all JDs can support JDP-JDP scaffolding due to a naturally occurring charge reversion at the CTD-binding interface (e.g. DNAJB2 and DNAJB8 that do not participate in protein disaggregation) [63]. This prevents the formation of functionally irrelevant Hsp70-JDP chaperone assemblies in cells.

Collectively, genetic and structural regulatory elements, particularly targeting the JD, play a key role in building up the plasticity of this complex chaperone network and allow it to operate in a highly dynamic manner and respond rapidly to different cellular needs. At the same time, it is important to note that additional mechanism(s) for correctly pairing Hsp70s and JDPs could also exist beyond the JD. Specificity may also be regulated by differential expression of specific JDP members in different types of cells / tissues [86] and subcellular locations to direct precise biological functions.

Pathological conditions associated with human J-domain proteins

The relatively high level of regulation built into this chaperone system suggests that there is limited redundancy among members of the JDP family in vivo, and specific disease-relevant JDP-Hsp70 subnetwork links can be established (see Outstanding Questions box). Over the past decade, the use of next-generation sequencing methods to understand human genetic diseases has considerably accelerated the discovery of pathological mutations in JDPs. A relatively large number of JDP mutations leading to loss of function have been discovered in a broad spectrum of protein conformational diseases (see Glossary) including neuro/neuromuscular disorders (Table S2). For example, mutations in the DNAJB2 and DNAJB6 genes have been reported to cause various hereditary neuropathies ranging from amyotrophic lateral sclerosis (ALS) to Huntington’s disease. Overexpression of particular isoforms of these JDPs has been shown to exert potent anti-neuropathological protein aggregation activity, resulting in improved neuronal survival in cellular and animal models of disease (Table S2). The two JDPs employ distinct mechanisms to suppress protein aggregation. Using ubiquitin-interacting motifs in the client binding region, DNAJB2 targets aberrant proteins for proteasomal degradation while DNAJB6 binds and blocks the primary nucleation of amyloidogenic proteins during fibril formation [38] (Table S2).

In addition to mutations, changes in cellular JDP levels could disrupt JDP-Hsp70 circuits and modify pathological protein aggregation [87] (Table S2). The latest findings from human tissue analysis report that JDP expression levels (e.g. DNAJA2) in the brain inversely correlate with the progression of diseases such as taupathies. Moreover, certain tau mutants appear to escape recognition by JDPs [88], which could lead to increased protein aggregation that may contribute to rapid disease progression. Furthermore, age-related changes in cellular JDP levels [89] that arise in part from the gradual shutdown of stress-responsive transcriptional programs [84], can result in disruption of this central chaperone network. Such conditions could also contribute to the progression of these age-associated diseases. JDP-based therapies against neurodegeneration may not be too far away. A proof-of-principle experiment conducted on a mouse model of Huntington’s disease recently revealed that delivering DNAJB6 protein using extracellular vesicles could suppress polyQ aggregation in affected brain tissue [90].

In cancer, dysregulation of some nodes of the JDP-Hsp70 network appears to promote cancer cell survival, proliferation, metastasis, and resistance to chemotherapy [58]. Based on emerging findings, it is possible to speculate that by abnormally rewiring the JDP-Hsp70 circuits to boost some chaperone machines, malignant cells can activate anti-apoptotic pathways and gain increased buffering capacity against toxicities associated with protein aggregation stresses that can arise from rapid tumor cell proliferation and genome instabilities. Multiple JDPs have been identified to promote or negate tumorigenesis by chaperoning tumor suppressors or oncoproteins (Table S2). For example, the DNAJB1-Hsp70 machinery possesses the ability to bind and inactivate p53, a central tumor suppressor protein [91]. Furthermore, aberrant fusion of DNAJB1 JD with the catalytic subunit of cAMP-dependent protein kinase was shown to drive fibrolamellar hepatocellular carcinoma (Table S2).

The levels of some JDPs could even be used as diagnosis/prognostic markers to predict the clinical outcome of certain types of cancer [92]. A recently conducted systematic knockout study of ER-localized JDPs (ERdj1–8) indicates that there is sufficient specificity among JDPs to possibly develop selective anticancer therapies [93]. In addition to the role of JDPs in major human disorders, functional impairments in these cochaperones are also identified in a number of rare diseases including ciliary dyskinesia 43, mild hyperphenylalaninemia, retinitis pigmentosa and hypogammaglobulinemia, and bone marrow failure syndrome (Table S2).

JDPs also play an important role in microbial infections. In particular, viruses encode their own JDPs (e.g. T-antigen) to reprogram host Hsp70 chaperone systems to support virion biogenesis [94, 95]. In addition, multiple JDPs in the host cell can be hijacked by viruses to drive Hsp70 functions that allow them to replicate [96]. Recent reports show that some host JDPs can also exhibit antiviral activity when induced (Table S2), suggesting that this chaperone system plays a role in host defence mechanisms against microbial infections. Similarly to viruses, parasites, such as malaria, secrete specific JDPs to modify the functions of Hsp70 in the host cell to facilitate its life cycle [97, 98]. We are still in the infancy of understanding the role of this chaperone system in bacterial and fungal virulence. Overall, these findings indicate that the JDP-Hsp70 chaperone network undergoes a series of complex rewiring events during host-pathogen interactions. These examples also demonstrate the capacity and power with which these cochaperones operate to rapidly control and rewire Hsp70 functions that broadly affect cell growth and metabolism.

Given the high degree of non-overlapping activities of JDPs in disease, these cochaperones could provide valuable new drug targets against a broad range of infections that have no or limited treatment. This offers an exciting future direction for developing therapeutically valuable JDP modulators, of which some are already emerging [99–101] (see Outstanding Questions box).

Concluding Remarks

Regulatory elements built around JDs ensure the correct pairing and assembly of different Hsp70 chaperone machines. Beyond modulating Hsp70 activity, some JDs have evolved to mediate specific interactions with other JDPs, adding another level of complexity as to how the JDs guide the various assembly architectures of Hsp70 chaperone machines. JDs have undergone sufficient diversification to possess phylogenetic and functional signatures at multiple resolutions that permit a new functional subdivision of the JDP family. How these signatures and their posttranslational modifications influence JDP activity remains an open question (see Outstanding Questions box). The presence of disease-causing mutations and the apparent role of some JDP-Hsp70 circuits as disease drivers/modifiers (Table S2) highlight the clinical importance of understanding the functional regulation of this chaperone system. Future genome-wide association studies, bottom-up proteomics approaches, and gene editing technologies will undoubtedly help uncover processes that direct the (re)wiring of JDP-Hsp70 circuits to allow acute responses to changing cellular needs. Furthering our understanding of the mechanisms that regulate the functions of JDP-Hsp70 is critical to uncovering how dysregulation of this chaperone system causes human disease (see Outstanding Questions box).

Highlights.

J-domains play a critical role in (re)wiring the cellular JDP-Hsp70 network vital to maintaining proteostasis and organismal development.

Multiple genetic and structural regulatory elements are built around the J-domain to provide specificity for the assembly of Hsp70 chaperone machineries.

J-domains carry discriminatory phylogenetic signatures that could specify the type, class, cellular location, and phyla of JDPs.

Loss of function and/or aberrant wiring of JDPs with partnering Hsp70s could act as risk factors in a wide range of human diseases.

Glossary

Proteostasis (protein homeostasis)

Balance of a functional cellular proteome

The proteostasis network

Competing and integrated series of biochemical pathways that ensures a healthy cellular proteome.

Molecular chaperones

An ubiquitous family of cellular proteins that interact with and assist in the conformational (un)folding and (dis)assembly of other proteins.

Protein conformational diseases

Disorders characterized by a protein(s) adopting an abnormal conformational state(s) leading to cellular dysfunction.