Small heat shock proteins: Simplicity meets complexity

Small heat shock proteins (sHsps) are a ubiquitous and ancient family of ATP-independent molecular chaperones. A key characteristic of sHsps is that they exist in ensembles of iso-energetic oligomeric species differing in size. This property arises from a unique mode of assembly involving several parts of the subunits in a flexible manner. Current evidence suggests that smaller oligomers are more active chaperones. Thus, a shift in the equilibrium of the sHsp ensemble allows regulating the chaperone activity. Different mechanisms have been identified that reversibly change the oligomer equilibrium. The promiscuous interaction with non-native proteins generates complexes that can form aggregate-like structures from which native proteins are restored by ATP-dependent chaperones such as Hsp70 family members. In recent years, this basic paradigm has been expanded, and new roles and new cofactors, as well as variations in structure and regulation of sHsps, have emerged.

The molecular stress response consists of a network of proteins that efficiently stabilize protein homeostasis under conditions that negatively affect protein conformation (e.g. heat, oxidative stress, etc.) (1). Molecular chaperones (2,3), which include the family of small heat shock proteins (sHsps), 4 represent the key players within the stress response. sHsps are the most ubiquitous family of chaperones, being present in all kingdoms of life. Compared with other sophisticated chaperone systems such as GroE, or the Hsp70 and the Hsp90 machineries, sHsps seem to be more restricted in their mode of action: they bind efficiently to a large variety of non-native proteins ranging from peptides to large proteins (4) and have a large binding capacity. Thus, they are efficient in preventing irreversible aggregation processes. However, unlike ATP-dependent chaperones, they are not able to further process the bound proteins. Conformational transitions after binding that may result in structural changes of the bound substrate proteins and their release have not been detected for sHsps. Their general role appears to be to trap non-native proteins in a state from which they can refold to the native state with the assistance of ATPdependent chaperones. This does not imply that sHsps generally prevent the formation of large aggregates; in a number of cases, it has been observed that they are rather part of these aggregates such as inclusion bodies in Escherichia coli (5)(6)(7), stressed eukaryotic cells (8 -12), and also in vitro (13). What is important in this context is that the non-native proteins can be readily recovered from these complexes and refolded to the native state, in contrast to the situation observed for aggregates formed in the absence of sHsps. Thus, sHsps represent the central cellular component inhibiting or modulating protein aggregation and improving disaggregation processes after stress (14,15).

Structure
The plasticity of their quaternary structure, resulting for most sHsps in an ensemble of oligomeric species, is a hallmark of the sHsp protein family and the basis for their biological activity. sHsps are characterized by their low molecular mass  and tripartite domain architecture (15)(16)(17)(18). Their primary structure can be dissected into a conserved, structured ␣-crystallin domain (ACD), which is flanked by a flexible N-terminal region (NTR) of variable length and sequence and a short C-terminal region (CTR) (Fig. 1A) (19,20).
The ACD (or Hsp20 domain, PF00011) is 90 -100 amino acids long and represents the signature motif of sHsps. It was already present in the last common ancestor of prokaryotes and eukaryotes (20 -22). Structurally, the ACD is a compact ␤-sandwich composed of two anti-parallel sheets of three and four ␤ strands, respectively (Fig. 1B), similar to the immunoglobulin fold (23,24). The ACDs of sHsps in mammals and nonplant higher eukaryotes contain an extended ␤ strand (referred to as ␤6 ϩ 7 strand) instead of a distinct ␤6 strand (Fig. 1B).
The CTR of sHsps is polar, solvent-exposed, not longer than 20 amino acids (20), and shows flexibility that increases markedly toward the very C terminus where it becomes comparable with that of isolated peptides of the same length (25)(26)(27). The CTR contains the characteristic three-residue IX(I/V) motif, which plays a prominent role in oligomerization (28) as described below. Interestingly, an IX(I/V) motif is also present in the N-terminal region of some sHsps, e.g. in tapeworm Tsp36 (29) and human HSPB6 (30).
The NTR of sHsps, which is highly variable in sequence and length, plays a driving role in oligomer formation and in the interaction with substrate proteins. It ranges from 24 residues in Hsp12.2 from Caenorhabditis elegans (31) to 247 residues in Hsp42 from Saccharomyces cerevisiae (32). The NTRs are enriched in hydrophobic residues, with phenylalanines and tryptophans being over-represented (20,33). In mammalian sHsps, the NTR harbors serines that are phosphorylated by specific kinases in vivo (34 -37). In almost all available crystal structures of sHsps, the NTR is either completely missing or only partially resolved. This is indicative of a pronounced flexibility, a view also supported by its susceptibility to proteolysis (38 -40) and high hydrogen/deuterium exchange rates (41,42). Accordingly, the NTR, in particular the stretch at its very beginning, is assumed to be in general intrinsically disordered. However, within the NTR, specific segments adopting secondary structure exist, as shown for Sulfolobus tokodaii Hsp14 (43,44), wheat HSP16.9 (24), Schizosaccharomyces pombe HSP16.0 (45), and human ␣B-crystallin (46), for example. These helical regions are highly dynamic and exist as structural ensembles of heterogeneous conformations (24,45). For the large yeast Hsp42, a prion-like domain and an intrinsically disordered subdomain were identified as parts of its NTR (47,48).
The ability to populate ensembles of large and inter-converting oligomers of varying subunit stoichiometries (typically ranging from 12-to 32-mers) is a key structural feature of sHsps. However, a few exceptions exist as sHsps have been described that are only tetramers (49) or dimers (50 -55). The small oligomer size of these sHsps correlates with distinct properties of their NTRs and CTRs: the members of the C. elegans Hsp12 protein family have the shortest NTRs and CTRs among sHsps (49,50); Hsp17.7 from Deinococcus radiodurans possesses also a short CTR (54); Tsp36 from the tapeworm Taenia saginata, which has two ACDs per monomer, completely lacks a C-terminal extension (53). The conserved IX(I/V) motif is absent in the short CTR of human HSPB6 (Hsp20) (52) as described below, and Arabidopsis thaliana Hsp18.5 lacks the ␤6 strand (55).
The dynamics and polydispersity of sHsps along with the structural plasticity of their NTRs and CTRs render a detailed analysis of the structure and mechanism of sHsps technically challenging. However, persistent efforts have culminated in atomic structures for several sHsps, some of them N-and C-terminally truncated and only a few full-length proteins. The key finding from these studies is that sHsp oligomers are assembled hierarchically via a series of weak and dynamic inter-subunit contacts involving all three sHsp domains (56).
The basic building block of the higher order sHsp structure is a homodimer formed by the ACDs of two protomers (Fig. 1C). Note that ACDs of mammalian and higher eukaryotic sHsps (with the exception of plants) lack a distinct ␤6 strand but contain an extended ␤7 strand (referred to as ␤6 ϩ 7 strand) (left). In ACDs of bacterial sHsps, for example, there is a distinct ␤6 strand located in the loop connecting ␤5 and ␤7 strands (right). C, structures of ␤7-interface ACD dimers (left) and ␤6-swapped ACD dimers (right) are shown. The ACDs of individual protomers are colored blue and gray. In human ␣B-crystallin (HSPB5) (left) (PDB code 2KLR) (62), the dimer interface is formed by the interaction of the extended ␤6 ϩ 7 strands from neighboring protomers. The ␤7-interface is marked by a dashed line. In Hsp17.7 from D. radiodurans (right) (PDB code 4FEI) (54), the ␤6 strand exchanges between partner chains at the dimer interface. D, groove along the ␤7-interface (ACD groove) in bovine ␣A-crystallin (PDB code 3L1F) (58) with the bound 2-methyl-3,4-pentanediol molecules (yellow).
In many sHsps (especially of mammals and other nonplant higher eukaryotes), the dimer interface is formed by the interaction of the extended ␤6 ϩ 7 strands from neighboring protomers in an anti-parallel orientation ("␤7-interface dimer" (15)) ( Fig. 1C) (57)(58)(59)(60)(61)(62). This interface is relatively weak with a dissociation constant in the order of a few micromolars (63). The interaction of the ␤ strands results in different strand registers as observed in crystal structures of mammalian sHsps (57,58,64). The occurrence of mammalian sHsp oligomers with an odd number of subunits (65)(66)(67)(68)(69) is in support of the lability of the ␤7-interface as here at least one monomer is present in addition to the usual dimers. Crystal structures of truncated mammalian sHsp constructs (Fig. 1D) disclose a positively charged deep groove along the ␤7-interface of the respective dimers (ACD groove), which is able to bind polypeptides and small molecules (57,58,64).
Isolated ACDs alone are commonly not sufficient for oligomer formation (73). The formation of higher-order sHsp oligomers requires the involvement of the CTRs and NTRs in a next level of hierarchy (56). The CTR usually acts as a flexible "crosslinker" between dimers: the IX(I/V) motif of a CTR of one dimer binds into the hydrophobic groove formed by the ␤4 and ␤8 strands of a subunit in the neighboring dimer ( Fig. 2A). In some cases, bidirectional binding with respect to the ␤8 strand, e.g. in ␣Aand ␣B-crystallin, was observed (23,24,45,58,62,72,74). This inter-molecular IXI-␤4/␤8 interaction was long considered to be indispensable for sHsp oligomer assembly as sHsps lacking the IX(I/V) motif (75) as well as those with mutations in the same motif (73-76) typically populate small oligomers (77,78). However, there are also data demonstrating that the disruption of the IX(I/V) motif does not necessarily abolish oligomer formation (79,80).
In some sHsps, the IX(I/V) motif can bind into the ␤4/␤8 groove of the same protomer (81) inserting into the ACD groove ( Fig. 2B) (57,82). As these interactions are weak and dynamic, the IX(I/V) motif can even fluctuate between bound and entirely unbound states (83), and small modifications in protein sequence, solution conditions, and temperature can have strong effects on its dynamics regulating access to the ␤4/␤8 pocket (74,84). Thus, the initial view of the CTR as a static "cross-linker" has to give way to a more comprehensive representation of the variable and dynamic roles of the CTR in sHsp oligomerization.
The contribution of the NTR to the assembly of sHsps is also multifaceted. Because of its intrinsically disordered nature, the NTR is commonly fully unresolved or only partially resolved in crystal structures of full-length proteins. In the cases where structures for segments of the NTR could be resolved, the NTRs of the protomers of one dimer often display different conformations (24, 30,45). Multiple conformations of the NTR are also observed in solution by NMR, e.g. human ␣B-crystallin (85). The NTR employs either the ACD or the NTR of the neighboring monomer as a binding partner to mediate intermolecular contacts, but examples have been reported in which the NTR is also involved in intramolecular contacts. If the NTR harbors an IX(I/V) motif, e.g. in tapeworm Tsp36 (29) and human HspB6 (30), this N-terminal IX(I/V) motif can bind into the ␤4/␤8 pockets of the partner subunits (Fig. 2C). Parts of the NTR can also insert into the ACD groove ( Fig. 2D) (60). Moreover, intermolecular N-terminal interactions promote dimer association and the assembly of higher-order oligomers as suggested for human ␣B-crystallin (46,56,86) and as seen in the crystal structure of Sulfolobus solfataricus Hsp14.1 (Fig. 2E) (71). The tremendous variability and plasticity of the interac- JBC REVIEWS: Structure and function of sHsps tions of the terminal regions have recently been shown in the structure of human HSPB2/B3 (both full-length) heterotetramer (3:1 ratio), in which the NTR and the CTR bind to the ␤4/␤8 as well as to the ACD grooves (82).
Taken together, the available structural data indicate that the modular architecture of sHsps is based on multiple, labile interfaces with the ACD at its core and the flanking regions contributing to the assembly in a flexible manner. The polydispersity often associated with sHsps is very likely the result of the weak and dynamic inter-subunit interactions leading to e.g. variations in the strand register of the ACD, bidirectional binding of the palindromic C-terminal IX(I/V) motif in mammalian sHsps, and domain swapping of the CTRs and NTRs. Notably, the NTR plays a key role in oligomer plasticity: variations in its length and/or sequence modulate the total number of subunits in the oligomers as highlighted by studies on bacterial Hsp16.5 (87)(88)(89)(90). Additionally, the labile interfaces, which sHsps employ to assemble into large oligomers, also permit subunit dissociation/re-association processes. The ensemble of different oligomeric states at equilibrium is highly dynamic with a constant exchange of subunits between oligomers (65,74,(91)(92)(93)(94)(95)(96). On the functional level, the ensemble dynamics is correlated to the recognition of substrates and the regulation of the chaperone activity (97,98).

Cellular function
Mechanistically, sHsps bind misfolded or unfolding proteins and prevent them from irreversible, uncontrolled aggregation in an ATP-independent fashion (Fig. 3) (16, 99 -101). Usually, sHsps do not display refolding activities but stabilize early unfolding intermediates of aggregation-prone proteins (102,103) that arise during diverse proteotoxic stress conditions. In this respect, sHsps have been originally termed "holdases" as they capture the unfolded proteins as a first line of defense creating a reservoir of proteins competent for disaggregation and refolding after stress survival. Some sHsps are also expressed at physiological conditions in the cell to contribute to the chaperone network. The stoichiometry of sHsp versus substrate protein determines, among other factors like temperature and the nature of the substrate, the architecture and size of sHsp/substrate complexes. According to in vitro studies, polydisperse yet soluble ensembles of sHsp/substrate complexes are formed when the sHsp is in excess as revealed by electron microscopic analyses (102, 104 -106) and MS (94). When, however, substrate proteins exceed, sHsps become incorporated in large, polydisperse, dynamic, and seemingly amorphous aggregates of the substrate protein (13, 104, 106, 107). The recruitment of sHsps to the insoluble fraction is also observed in vivo in many organisms: in E. coli upon protein overexpression (5) and heat stress (108); yeast, Arabidopsis thaliana, humans, and mammalian cells during heat stress (8, 10, 109 -111); and in C. elegans during aging (11).
Release and refolding of non-native proteins trapped in sHsp/substrate complexes do not occur spontaneously. Both processes require the cooperation of ATP-dependent chaperones (Fig. 3). As shown in vitro, using various substrate proteins and sHsps from humans, plants, cyanobacteria, and bacteria, Under stress conditions, substrate proteins are destabilized and begin to unfold (II), and the sHsp ensemble becomes activated by remodeling of the ensemble composition in favor of smaller species (often dimers) with exposed substrate-binding sites (III). Activated sHsps bind early unfolding intermediates of substrate proteins in an energy-independent manner (IV) and stabilize them in sHsp/substrate complexes of different forms (V). Bound substrates may reactivate spontaneously or are subsequently refolded by the ATP-dependent Hsp70 chaperone system composed of Hsp70, Hsp40, and a nucleotide exchange factor (NEF)) (VI). The effective disassembling of insoluble substrate aggregates with incorporated sHsps and substrate refolding requires the concerted action of the Hsp70 -Hsp100/ClpB bi-chaperone system. JBC REVIEWS: Structure and function of sHsps the Hsp70/Hsp40 system in the presence of ATP is sufficient for regaining active substrate proteins from soluble sHsp/substrate complexes (6,54,106,112,113). The effective disassembly of insoluble protein aggregates with incorporated sHsps and their refolding require additional chaperone power. Here, the action of Hsp70 -Hsp100 bi-chaperone systems, as shown in vitro with components from E. coli (114,115), cyanobacteria (112), and yeast (12,13), is required. This is in line with a plethora of in vivo observations. For example, in E. coli, the DnaK system together with ClpB quantitatively dissolves a wide spectrum of protein aggregates that formed during heat stress and include IbpB and IbpA (6,114,116). In yeast, Hsp26 incorporated into cytosolic aggregates apparently alters the nature of the aggregates, which then become efficiently disassembled and refolded by the Ssa1-Hsp104 system (12,13). In Arabidopsis, the release of the cytosolic class I and class II sHsps from the insoluble cell fraction depends on the Hsp100/ClpB system (109).
The sHsp-assisted, reversible sequestering of misfolded substrates in large inclusions in vivo is apparently a protective strategy to enhance cellular fitness during stress (10,103,117). The aggregates of non-native proteins formed in the presence of sHsps differ in size, composition, and architecture from those formed in their absence. This avoids the exposure of hydrophobic surfaces to other cellular components and protects the substrates from proteases. sHsps seem to prevent tight interactions between the unfolded proteins thus facilitating the disaggregation and folding executed by the Hsp70 and Hsp100 chaperone systems (8,13,103). In a first step, Hsp70 seems to displace and release surface-bound sHsps from sHsp/substrate assemblies. Subsequent recruitment of Hsp100 by Hsp70 directs the aggregated proteins toward the solubilization and refolding pathway (6,7). This mechanism in which Hsp70 plays a key role is likely to be conserved with variations in all kingdoms of life. Metazoa, which lack Hsp100, possess a specialized Hsp70 chaperone machinery exhibiting powerful disaggregase activity (118,119).
The competitive displacement of sHsps from the aggregates by Hsp70 does not necessarily require a direct interaction or physical association between both chaperones, which has indeed not yet been reported. A direct interaction would imply species-specificity, which is not the case as sHsps and Hsp70s from different species have been shown to efficiently cooperate in vitro (54,107,112). To date, only an indirect physical link has been reported in the human system where the Hsp70 co-chaperone BAG3 mediates the association of sHsps with Hsp70 as a scaffolding protein via its BAG domain and IPV motifs (120).
Taken together, the recent findings reveal that the classical "holdase" activity of sHsps, i.e. keeping aggregation-prone proteins in a native-like conformation for subsequent refolding by ATP-dependent chaperones, does not fully account for their role in proteostasis networks as they also seem to enhance or regulate aggregation processes.
In some cases, sHsps even seem to regulate the interaction spectra and functions of native proteins, such as 14-3-3 proteins that interact with ϳ700 different partner proteins to coordinate vital cellular functions (121)(122)(123). The phosphorylated form of human HSPB6 was suggested to act as a displacer of other proteins from 14-3-3, thus affecting the balance of the 14-3-3 interactome and downstream cellular processes (30,121,123). In line with such a model, the 14-3-3/HSPB6 interaction was shown to be affected in a number of pathological conditions.

Substrate recognition
Several studies using proteomic approaches to identify substrate protein spectra in different organisms indicate that sHsps are very promiscuous and bind at least several hundred different cytosolic proteins under stress conditions (32, 40, 54, 60, 124 -127). Nevertheless, it is still enigmatic how sHsps recognize their substrates and which regions of the sHsp form contacts with the substrate. Apparently, the interaction depends on a charge-and/or hydrophobicity-driven capturing of substrates (18,97). The current view, which is based on results from crosslinking experiments followed by mass spectrometric analysis as well as peptide arrays, suggests that the interaction between substrates and sHsps involves multiple binding sites throughout the sHsp sequence with the NTR playing a major role (27, 41, 74, 80, 128 -134). For human ␣B-crystallin, it was shown that the fibril-forming Alzheimer's disease A␤(1-40) peptide is preferentially bound by the ACD, whereas the amorphously aggregating lysozyme is captured by the partially disordered NTR (85). Such substrate-dependent variations in the interaction sites would allow promiscuity and complexity in the substrate recognition of sHsps.
The observation that regions outside the ACD are involved in substrate recognition also might explain variations in the substrate specificity of different sHsps. This is further supported by the independent evolution of the NTR and CTR (20) rendering variations in the profile of substrates recognized by sHsps from evolutionally distant species likely. An additional line of evidence is based on the analysis of the evolution of paralogs of oligomeric sHsps. Here, especially their hierarchical assembly enables and favors the evolution of distinct functions of sHsp paralogs (135).
Experimental evidence further shows that different NTRs lead to differences in the mechanism of substrate interaction. For the two sHsps from bakers' yeast, Hsp26 and Hsp42, the respective NTRs determine different functionalities and their different sequestration (32,48,103). Hsp26 but not Hsp42 is temperature-regulated by a specific sensor domain in the NTR (102,136), rendering Hsp42 the constantly active and more general sHsp in the yeast cytosol (32). Hsp42 promotes protein aggregation and sequestration into Q-bodies (10,103,117,137). In contrast, substrate release by Hsp70/Hsp100 is more effective from Hsp26 than from Hsp42 (103).
As the NTRs are also engaged in subunit contacts within the sHsp oligomer, the exposure of the substrate-binding sites needs to be triggered. Accordingly, the higher oligomers within the ensemble in which the NTRs are maximally engaged in inter-subunit interactions are very likely to represent inactive storage forms. Activation requires a shift in the equilibrium to an ensemble weighted toward smaller species. These smaller species are thought to represent the "active-state oligomers" with exposed NTRs (15, 56, 97, 98, 138 -140). In line with this notion, some sHsps that are apparently dimeric, e.g. AtHsp18.5 JBC REVIEWS: Structure and function of sHsps from A. thaliana (55) or Hsp17.7 from D. radiodurans (54), show chaperone activity.
Besides the overexpression of many sHsps upon stress, several specific triggers that lead to a shift of the oligomeric distribution within the ensemble have been described: the presence of an unfolded substrate, an increase in the environmental temperature, changes in pH, and post-translational modifications such as phosphorylation and hetero-oligomer formation (98). In all cases, smaller species with higher activity are observed. The dissociation of higher-order oligomers allows the regulated exposure of substrate-binding sites that are otherwise sequestered in the oligomers (40,139,(141)(142)(143)(144). Upon stress, the presence of unfolding substrate proteins that interact with these binding sites induces a shift in the equilibrium of the sHsp ensemble toward the more active species (87,92,139). Regulating sHsp activity by phosphorylation or more generally by posttranslational modifications is especially found in eukaryotes (34,35,(145)(146)(147)(148)(149)(150). The most extensively studied example in this respect is human Hsp27 (HSPB1), which possesses three phosphorylation sites (Ser-15, Ser-78, and Ser-82) within its NTR that are modified via a for mitogen-activated protein kinase cascade (147,151,152). Phosphorylation of human Hsp27 (HSPB1) leads to an enrichment of tetramers that further dissociate into dimers (113,(153)(154)(155)(156). Similarly, the oligomer ensemble of ␣B-crystallin mainly consists of 12-mers, hexamers and dimers upon phosphorylation in vitro, as revealed by studies using phosphorylation-mimicking variants (40,66,157). These phosphorylation events primarily affect the N-terminal contacts in the oligomer destabilizing subunit interfaces by the additional negative charges. The close proximity of the three phosphorylation sites within the NTR of ␣B-crystallin suggests that the activation is tunable with increasing phosphorylation (40,86).
The existence of sHsp paralogs in the same compartment together with the dynamic exchange of subunits between sHsp oligomers provides the possibility to form hetero-oligomers with modulated chaperone activities. A prominent example in this respect is the pair of E. coli sHsps, where one (IbpA) has low chaperone activity alone but enhances the chaperone activity of the second (IbpB) upon hetero-oligomerization (115). However, whether this concept is at work also in higher organisms remains to be seen.
It is of special interest that not all sHsps found in the same compartment form hetero-oligomers indicating an evolutionary pressure toward homo-oligomerization (135). For example, in flowering plants (angiosperms) at least six subfamilies of sHsps, termed class I to VI, exist in parallel in the cytosol (97,158) with none of them co-assembling into hetero-oligomers (158,159). Similarly, in mosses and other plants, class I and II sHsps are present, and no hetero-oligomers were yet observed (160). Nevertheless, all members of one class (arising as paralogs from gene duplication events) readily form hetero-oligomers (24, 55,94). In humans, ␣Aand ␣B-crystallins in the eye lens form the most prominent sHsp hetero-oligomer. In human muscle cells up to seven cytosolic sHsps can be present in moderate to high amounts (149,161,162). Similar to their plant homologs, not all of them are able to form hetero-oligomers. The human sHsps can be divided roughly into two classes where most of the members only form hetero-oligomers within the same class (161,(163)(164)(165). Interestingly, however, some members seem to connect the two classes by their ability to interact with members of both classes. Taken together, this suggests that the formation of hetero-oligomeric and co-existing homo-oligomeric ensembles allows a tremendous variability in activity and specificity. However, the extent to which hetero-oligomerization may regulate the function and specificity for certain substrates remains to be explored in detail.

Role of sHSPs in diseases
In the human eye lens, the two ␣-crystallins, ␣Aand ␣B-crystallin, are supposed to counteract aggregation processes that lead to cataracts (166 -168). In addition, sHsps have been linked to a number of diseases in which a de-regulation of the physiological expression level of the respective human sHsps (termed HspBs) is observed, such as cancers, neurological disorders, myopathies, and multiple sclerosis (163, 169 -172). The common trait in all these diseases seems to be that elevated levels of mammalian sHsps may protect against the proteotoxic damage (162,172). However, enhanced levels of sHsps are not always favorable for the organism, e.g. the growth of some tumor cells is supported by increased sHsp levels, at the same time reducing the susceptibility for cancer therapy (171)(172)(173)(174)(175). Also, reduced or insufficient levels of sHsps can lead to diseases (176,177). In general, sHsps are presumed to not directly cause disease or regulate progression. The observed effects are thought to be secondary and correlate to the general fate of the cell (149,162). Exceptions are diseases that are caused by mutations in an sHsp, such as neuropathies caused by the K141N mutation in human HspB8 (178), a form of cataract and desmin-related myopathy caused by the R120G mutation of ␣B-crystallin (179 -182), cardiomyopathy induced by the same mutation of ␣B-crystallin (183,184), and dominant congenital cataract caused by the R116C mutation of ␣A-crystallin (181,185,186).
sHsps have also been implicated in aging. It had been shown that a member of the Hsp16 family in C. elegans is an important player in regulating the life span (187). Further biochemical studies showed that during aging in C. elegans, protein aggregation increases, and several sHsps are associated with these aggregates (11).
Concerning therapeutic intervention using sHsps as targets, ␣-crystallins have been in the focus. In search for molecules that bind and stabilize them in soluble forms, an oxysterol and lanosterol have been identified that are capable of reversing the aggregation of ␣Aand ␣B-crystallins in vitro and recovering (at least partially) transparency in animal models of hereditary cataract (188,189). This, of course, prompts further studies toward the use of related compounds in humans for cataract treatment.

Conclusions
In summary, sHsps are archaic and ubiquitous molecular chaperones found in different branches of life. Although they all share a conserved structural domain, the ACD, their divergent NTRs and CTRs dictate differences in their overall structure and consequently in their substrate interactions as well as JBC REVIEWS: Structure and function of sHsps functions in the cell. With few exceptions, sHsps assemble into large oligomers that are polydisperse and exist in a dynamic equilibrium with a rapid exchange of subunits between oligomers. Increasing evidence points to the critical role of the oligomer dynamics in the ability of sHsps to bind denaturing substrates and to maintain substrates in a soluble and foldingcompetent state or sequester them in reversible aggregates. These findings have led to a consensus model of sHsp function in combating irreversible aggregation together with other chaperones. However, important issues remain to be elucidated in more detail. The substrate range is still not sufficiently defined despite progress in recent years. Also, the mechanism of substrate recognition by sHsps or the potential functional differences between diverse family members remain largely open issues. This includes the cooperation between different sHsps in the same cellular compartment. Despite established links of sHsps to many diseases and their general importance for proteostasis as well as stress management, more work is clearly needed to increase our mechanistic understanding of these still enigmatic molecular chaperones.