- Email: [email protected]
The Unfolding Story of Three-Dimensional Domain Swapping
Structure, Vol. 11, 243–251, March, 2003, 2003 Elsevier Science Ltd. All rights reserved.
PII S0969-2126(03)00029-7
The Unfolding Story of Three-Dimensional Domain Swapping Frederic Rousseau,1 Joost W.H. Schymkowitz,1,* and Laura S. Itzhaki2,3,* 1 European Molecular Biology Laboratory Meyerhofstrasse 1 69117 Heidelberg Germany 2 Cambridge University Chemical Laboratory MRC Centre for Protein Engineering Lensfield Road Cambridge, CB2 1EW United Kingdom
Summary Three-dimensional domain swapping is the event by which a monomer exchanges part of its structure with identical monomers to form an oligomer where each subunit has a similar structure to the monomer. The accumulating number of observations of this phenomenon in crystal structures has prompted speculation as to its biological relevance. Domain swapping was originally proposed to be a mechanism for the emergence of oligomeric proteins and as a means for functional regulation, but also to be a potentially harmful process leading to misfolding and aggregation. We highlight experimental studies carried out within the last few years that have led to a much greater understanding of the mechanism of domain swapping and of the residue- and structure-specific features that facilitate the process. We discuss the potential biological implications of domain swapping in light of these findings. Background Three-dimensional domain swapping is a process by which one protein molecule exchanges a domain with an identical partner (Figures 1A and 1B). Although it had been proposed as the mechanism for the formation of RNase A dimers as far back as 1962, domain swapping was first observed in crystal structures during the eighties and nineties, and the terminology was only formally introduced by Eisenberg and coworkers in 1994 to describe the crystal structure of diphtheria toxin [1, 2]. This structure led to a series of elegant theoretical papers by Eisenberg and coworkers that proposed how and why domain swapping might occur and the potential biological implications. The latter included domain swapping as a mechanism for regulating function, as an evolutionary strategy to create protein complexes, and as a mechanism for misfolding and aggregation (Figure 1C) [1, 2]. More than 30 crystal structures of domain-swapped molecules have been elucidated since [3, 4], but the *Correspondence: [email protected] (L.S.I.), [email protected] (J.W.H.S.) 3 Present address: Hutchison/MRC Centre, Hills Road, Cambridge, CB2 2XZ, United Kingdom.
Review
physiological significance of these highly intertwined oligomers remains unclear. Also, experimental data on the thermodynamics and the mechanism of domain swapping were, until very recently, almost only qualitative. In part, this was due to problems inherent in the experimental study of domain swapping. First, domainswapped oligomeric forms are often metastable; once formed, they take a long time to convert back to the more stable monomeric form and therefore any quantitative analysis, for example the measurement of equilibrium constants, becomes unfeasible. Second, tractable model systems are required, ones in which both monomer and domain-swapped forms can be isolated and studied easily in solution. In the last couple of years, these problems have been overcome and much has subsequently been revealed to explain domain swapping [5], in particular by the combined use of protein engineering and the biophysical tools that have been applied so successfully to investigate protein folding. As researchers become increasingly interested in understanding and quantifying protein aggregation, mechanisms of self-association such as domain swapping have become the focus of attention and in the future will continue to provide new insights into the process. Domain Swapping: Definition and Terminology Domain swapping is defined as the process by which two or more protein molecules exchange part of their structure to form intertwined oligomers. These oligomers are composed of subunits having the same structure as the original monomer, with the exception of the hinge loop that connects the exchanging part with the rest of the structure and which is usually folded back on itself in the monomer and extended in the domainswapped oligomer [1, 2]. The term “domain” is not used in a strict sense, as proteins have been reported to swap entire tertiary globular domains [2, 6] but also often only a single element of secondary structure, such as an ␣ helix [7, 8] or a single strand of  sheet [9]. Although initially only domain-swapped dimers were observed, barnase was later shown to be able to fold into a domainswapped trimer [10]. The exchanging part of the structure makes the same interactions in the oligomer as it makes in the monomer, but its interactions are formed inter- rather than intramolecularly. These interactions that are identical in the monomer and oligomer form the “primary” interface [11]. Because subunits are often close to each other in a domain-swapped oligomer, a new interaction interface that is absent in the monomer may be formed, and this is termed the “secondary” interface (Figure 1A) [11]. Biological Relevance of Domain Swapping: Hypotheses and Experimental Evidence Almost 10 years after the identification of domain swapping, popular model proteins such as chymotrypsin inhibitor 2 (CI2) [12], barnase [10], SH3 [13], and staphylococcal nuclease [8] have now crystallized in domain-
Structure 244
Figure 1. Basic Principles in Domain Swapping (A) Simplified representation of a monomer and a domain-swapped dimer, with the secondary surface highlighted in red (see text). (B) Schematic representation of structures of a monomer [70] and domain-swapped dimer [9, 54] of suc1 obtained using the program MOLSCRIPT [71]. The domain-swapping C-terminal  strand and the hinge loop are shown in red. (C) Schematic representation of open-ended and closed-ended domain-swapped oligomers.
swapped forms, as have many others. There is no apparent structural or functional connection between these different proteins and examples are appearing across organisms and structural and functional classifications. A more complete overview can be found elsewhere [3, 4], while below a few examples are presented to illustrate the diversity of domain swapping (see also Figure 2). Both SH2 and SH3 domains, probably the most studied protein-protein interaction domains in signal transduction, have been reported to crystallize as domainswapped dimers [13, 14]. The dimeric form of bovine seminal RNase was the first RNase found to domain swap [15], but later the bacterial RNase barnase was reported to form domain-swapped trimers [10], and bovine RNase A was recently shown to swap two different structural elements [7, 16, 17]. The first of these is formed by swapping an N-terminal ␣ helix and the second by swapping a C-terminal  strand. Other examples include a coat protein from Simian virus 40 [18], the lens eye protein B2-crystallin [19], the cell cycle regulatory proteins from the cks family [9, 20, 21], bovine odorant
binding protein OBP [22], the DNA recombination and repair protein from E. coli RecA [23], citrate synthase from chicken heart [24], and many more proteins carrying out diverse functions. However, clear evidence for a functional role of domain swapping is lacking and, as all these structures have been observed in crystal forms, they could easily be suspected to be an artifact of the nonphysiological conditions and high protein concentrations used for crystallization. Several hypotheses together with coincident lines of evidence nevertheless suggest a significance for domain swapping in vivo. The first potential biological role for domain swapping is the functional regulation of proteins. Here several interesting observations have been made. Glyoxalase I from Pseudomonas was shown to exist both in an active domain-swapped dimeric state and as a metastable and less active monomer. In vitro conversion from dimeric to metastable monomeric glyoxalase I can be triggered upon addition of glutathione [25, 26]. After removal of glutathione, glyoxalase I slowly reverts to a more stable and active domain-swapped form. Although no evidence exists for such a mechanism in vivo, this demon-
Review 245
Figure 2. Schematic Representations of Domain-Swapped Proteins (A) Bovine seminal ribonuclease (PDB ID code 11bg) [27]. (B) Suc1 (PDB ID code 1sce) [9, 54]. (C) Bovine odorant binding protein (PDB ID code 1g85) [72]. (D) Recombination endonuclease VII from [73] (PDB ID code 1e7d). (E) Rice yellow mottle virus coat protein (PDB ID code 1f2n) [29]. This figure was generated using PyMOL (http://www.pymol.org).
strates that functional regulation by domain swapping can, in principle, be achieved. Another interesting finding is that formation of domain-swapped oligomers can introduce allosteric regulation of the protein activity already present in the monomer, as shown in bovine seminal RNase [27]. Also, several receptor proteins, including G-coupled receptors, are thought to dimerize by domain swapping [28]. Finally, the viral capsid of an icosahedral virus that was recently found to be composed of domain-swapped dimers has higher thermostability than other viruses from the same family lacking the domainswapped architecture [28, 29]. A second proposed biological implication of domain swapping is a detrimental one. Domain swapping has the potential to be a mechanism for protein misfolding, aggregation, and amyloid formation (Figure 1C) [11, 30, 31] and thus it may be related to misfolding leading to amyloid and prion diseases [11, 17, 30, 32–34]. Here again, direct evidence for domain swapping as a mechanism for aggregation and amyloid formation is lacking, but several observations strongly suggest that it might indeed be the case. First, transient aggregation observed during refolding of several proteins has been suggested to occur by domain swapping [35–38]. Second, a correlation between domain swapping propensity and the rate of heat aggregation of suc1 was observed [5]. Third, a role for domain swapping in the processes of prion and amyloid formation is suggested by the crystallization of both the human prion protein [39] and the
amyloidogenic human cystatin C [40, 41] in domainswapped forms. Fourth, Eisenberg and coworkers were able to design both domain-swapped dimers and highorder oligomers from the same three-helix bundle structural motif but with different topologies [7, 16, 17]. An updown-down favored a reciprocal swap to form a closed domain-swapped dimer. An up-down-up topology was predicted to favor an extended swap, and fibrils were formed that were assumed to be domain swapped. Finally, domain swapping has been proposed to contribute to structural diversification and the emergence of oligomers during evolution [11], but here again, although the idea is elegant and simple, it is not clear what the relevance of this is in vivo. Interestingly, it was recently shown that core mutations could introduce oligomerization by domain swapping, with the formation of a new hydrophobic interface and structural rearrangement of other parts of the protein [42]. Energetic Contributions to Domain Swapping A quantitative understanding is needed, first of the factors that can contribute to the equilibrium between a monomer and its oligomeric alter ego, and second of how interconversion between the two forms occurs. Oligomers are always disfavored by their greater translational and rotational entropy [11]. Further, domainswapped oligomers are made up of subunits that are identical in structure to the monomer, except for (1) the hinge loop and (2) the secondary interface [11]. These
Structure 246
Table 1. Dissociation Constants for Suc1 Wild-Type and Mutants in the Hinge Loop Mutation
Kd (mM)
Wild-type P90A P92A E91P P90AP92A ⌬8789
1.9 890 0.2 0.0001 11.2 1 ⫻ 10⫺6
Values were taken from [5, 53].
conformational differences can either favor or disfavor domain swapping and will greatly determine the equilibrium between monomer and domain-swapped forms. A convenient way to quantify the monomer-dimer equilibrium is by using the dissociation constant Kd, defined as Kd ⫽
[M]2 [D]
(1)
where [M] is the monomer concentration and [D] is the dimer concentration in monomeric units. The Kd can be seen as the protein concentration at which half of the protein molecules are dimeric. The free energy difference associated with this equilibrium can be obtained from ⌬G ⫽ ⫺RT ln Kd
(2)
The Kd for dimerization has been measured for a handful of domain-swapped dimers, and Table 1 shows the values for the protein suc1 and a subset of mutants. The Kd’s for the natural proteins measured to date are in the micromolar and millimolar range, which is far above the nanomolar range that would be expected for a biologically relevant process. However, the energy required to tip the balance is not large, considering that the entropic cost of fixing two molecules into a dimer is probably on the order of a few kilocalories [43]. From Equation 2, a free energy change of approximately 4 kcal mol⫺1 results in a change in the dissociation constant of three orders of magnitude. Thus, a few mutations in the hinge loop or in the secondary interface, designed to favor the dimer, could easily bring the Kd into the nanomolar range. This is illustrated in Table 1 for suc1, and the effects of the different mutations are explained below. The Hinge Loop and the Domain Swapping Propensity What are the specific sequence and structural requirements that turn a protein segment (generally a loop) into a hinge loop that facilitates domain swapping? Some early experiments addressed this question by focusing on the length of the hinge loop. Shortening the hinge loop in staphylococcal nuclease [8], suc1 [5], ckshs1 [44], CD2 [45], and single-chain Fv [46] destabilizes these monomeric proteins by making the hinge loop too tight to close the monomeric conformation. This is sufficient to induce formation of domain-swapped dimers in these proteins. Here, the entropic cost of association is more than compensated by the energetically
more favorable extended conformation of the shortened hinge loop in the domain-swapped form. Loop lengthening also has an effect on the oligomerization state of proteins. Monomeric CI2 forms domain-swapped dimers, as well as trimers and even higher order oligomers, when its active site loop is lengthened by 5–10 residues [12]. Moreover, the presence of high-order oligomeric forms increases with loop insertions consisting of repeats of Gln→Ala→Gly. The same observation was made in suc1: wild-type suc1 only forms monomers and domain-swapped dimers at protein concentrations in the low millimolar range. However, when both prolines of its hinge loop are mutated to alanine, trimers and tetramers are observed, even more so when both prolines are substituted by glycines (F.R., J.W.H.S., and L.S.I., unpublished results). Thus, making the hinge loop more flexible, either by lengthening it or by mutation, also induces swapping. There is a kinetic, as well as a thermodynamic component of the effect of loop lengthening, discussed later in the review. How else could the hinge loop sequence affect the domain swapping propensity? Bergdoll and coworkers observed that prolines are frequently found in the hinge loops of proteins that domain swap by arm exchange [47], and most recently the TorD chaperone was found as a domain-swapped dimer with a double-proline motif in its hinge [48]. Experiments on suc1 suggest a thermodynamic explanation. Suc1, which can fold as a monomer or an intertwined dimer with a Kd of 1.8 mM, has two prolines in its hinge loop [49]. Mutation of either proline to alanine dramatically shifts the monomer-dimer equilibrium, while mutation to alanine of any other residue in the hinge loop has very little effect [5] (see Table 2). Pro90 is unfavorable in the monomer hinge conformation; the mutation Pro90Ala shifts the equilibrium completely toward the monomer. Pro92, on the other hand, is unfavorable in the dimer; the mutation Pro92Ala shifts the equilibrium toward the dimer and has a Kd of 100 M (Table 1). Thus, the hinge loop in both monomer and dimer forms of the protein is in a “strained” conformation, more so in the monomer. The strain causes the hinge loop to act like a loaded molecular spring that relieves the tension by adopting an alternative conformation without altering the rest of the structure. In the case of suc1, the destabilization of the hinge loop by the prolines does not appear to arise from unfavorable interactions with other residues but rather from backbone strain that the prolines impose on each other as a double mutant cycle (Pro90AlaPro92Ala) demonstrated [5, 49]. We were able to engineer mutants with domain swapping propensities spanning nine orders of magnitude, by manipulating the strain in the hinge loop conformation of suc1 (Figure 3; Table 1) [5]. The principle that unfavorable interactions turn hinge loops into loaded springs also appears to be generally applicable. In protein L [50, 51], for instance, the monomeric hinge loop was destabilized by forcing residues into forbidden regions of the Ramachandran plot, thereby inducing an alternate, domain-swapped form. In conclusion, domain swapping ability can be introduced into a protein by manipulating loops in one of two ways. The first is by destabilizing the conformation of a loop in the monomer, either by shortening it or
Review 247
Table 2. Design Approaches for Domain Swapping Protein
Description
Hinge Shortening Staphylococcal nuclease [8] Suc1 [5] CD2 [45] Single-chain Fv antibody [74]
Deletion of 6 residues from hinge loop Deletion of 3 residues from hinge loop Deletion of 2 residues from hinge loop Length of linker between variable domains controls oligomerization state. Short linkers (2–10 residues) lead to oligomer formation; longer linkers (12–25 residues) lead to stable monomeric forms. Introduction of Strain in Hinge by Amino Acid Substitution Suc1 [5] Mutation of glutamate 91 in the hinge loop to proline Ckshs1 [44] Mutation of glutamate 63 in the hinge loop to proline Protein L [50] Mutations in the hinge loop A52V, N53P, and G55A Cyanovirin-N [75] Mutation of serine 52 to proline in hinge loop Hinge Lengthening CI2 [12] Insertion of glutamine repeat in hinge loop Introduction of Flexibility in Hinge Suc1 [5] Mutation of prolines 90 and 92 to glycine in hinge loop Other Mutations CD2 [45] Mutation of residues in the intermolecular interface GB1 [42] Mutation of residues in the hydrophobic core These mutational strategies were used to successfully convert monomeric proteins into obligate dimers. For terminology, see text.
by introducing unfavorable interactions. This forces the loop into a more favorable conformation in a domainswapped variant of the original monomeric structure. Because “spring loading” applies to rather short and inflexible loops, the propensity for domain swapping achieved in this way is high and is biased toward a specific oligomeric state. The second way to facilitate domain swapping is by inserting a long and/or flexible loop into the monomeric structure. Here the propensity to domain swap is generally lower than that achieved using the first approach, and also less specific because the flexibility of the loop allows it to adopt many conformations with similar energy. There is a kinetic component to these effects, as discussed later. Contribution of Residues Outside the Hinge to Domain Swapping Propensity How do these changes in domain swapping propensity compare to those that occur upon mutation elsewhere in the protein? Mutagenesis studies of CD2 were focused on the secondary interface—the new intermolecular interface that is created in the domain-swapped
form. These interactions contribute to the domain swapping propensity by stabilizing the dimer [46, 52]. Suc1 has no secondary interface, and therefore the equilibrium between monomer and dimer is solely determined by the hinge loop. In spite of this, mutations to alanine elsewhere in the protein were not neutral in their effect on the equilibrium between monomer and dimer. Although the effects were not as large as for mutation in the hinge loop, they were significant (changes in Kd of one to two orders of magnitude) [53]. Specifically, mutations to alanine in the  sheet shifted the equilibrium toward the monomer, showing that those interactions are more favorable in the dimer. Further, a phosphoprotein binding site of suc1 in the  sheet on the opposite side of the molecule to the hinge loop and at a distance of more than 20 A˚, has a different affinity for phosphosubstrates in the monomer compared with the dimer [53]. The crystal structures of the monomer and domainswapped dimer are highly superimposable outside of the hinge loop region, so that there is no obvious structural reason for these effects. The different response to mutation of the two forms of the protein appears to Figure 3. Scheme Showing the Consequences of Backbone Strain Created by Proline Residues The hinge residues are represented as colorcoded bricks. Prolines are shown in dark blue and their preceding peptide bond is in bold to indicate the restrictions that are imposed on it by the ring of the proline. Positions of backbone strain are shown by yellow zigzag lines. There is an equilibrium in the wild-type protein between the monomer conformation in which P90 is strained and the dimer conformation in which P92 is strained [5].
Structure 248
originate in the hinge loop. The conformation of the hinge loop is more strained in the monomer than in the dimer [5], and the strain is sensed throughout the  sheet. As a result, the  sheet is somewhat less satisfied when accompanied by the  turn hinge in the monomer conformation than with the extended hinge conformation in the dimer [53]. Indeed, removing the strain in the hinge loop by mutating both prolines to alanine reduces the difference in energetic behavior of the two forms. The biological implications of these observations are discussed later. In conclusion, however, the presence of an unfavorable loop conformation in a protein does not only promote domain swapping, but also subtly affects the energetics of the entire structure, probably by altering its dynamic behavior.
An Unfolding Mechanism for Domain Swapping Domain swapping is essentially a protein-folding phenomenon, as it gives a single polypeptide sequence access to two or more well-defined native states. Two questions immediately arise. First, what are the folding pathways to monomer and domain-swapped oligomer and how are they connected to each other? Second, can one control which state is produced upon refolding of a domain-swapping protein? Generally, interconversion between monomer and dimer is very slow (from days to months) [5, 11], and thus both states are separated by a large kinetic barrier. For interconversion to occur, many native interactions, often in the hydrophobic core, must be disrupted, to be replaced by identical interactions with another protein chain. Eisenberg and coworkers proposed that the transition state for interconversion is an “open” form of the structure, disrupting many native interactions while maintaining the overall native fold of the exchanging domains, which is both enthalpically and entropically unfavorable [11]. Such a scenario is most likely for cases that involve the swapping of true domains that can fold independently of each other and then associate. However, several domain-swapped structures may require a more intricate interplay between folding and association. In the case of barnase [10], for example, the parts of the structure that swap can only fold partially in an independent manner. The authors of the domainswapped barnase structure therefore proposed that association must occur at the latest at an intermediate stage of folding. The situation is even more extreme in the case of suc1 and its homologs, where a  strand that is a central part of the hydrophobic core is exchanged to form the domain-swapped dimer [9, 20, 54]. A protein engineering analysis of the folding mechanism of both monomeric [55] and dimeric [56] suc1, complemented by molecular dynamics simulations of unfolding of the monomer [57], shows that the interactions that connect this  strand to the rest of the protein are formed very early in the folding reaction and thus that their disruption requires substantial unfolding. Further, the time required for monomer-dimer equilibration in suc1 can be reduced from a few months to a few minutes simply by increasing the unfolding rate significantly by changing the temperature, denaturant concentration, or pH [5]. Likewise, the hu-
man homolog of suc1, ckshs1, unfolds much faster than suc1, and the monomer-dimer interconversion is also much faster compared with suc1 [45]. These data led to a model in which a suc1 molecule can fold either as a monomer or a domain-swapped dimer from the denatured state [5]. Interconversion has to proceed via the denatured state and the kinetic barrier between monomer and dimer under native conditions is a direct consequence of the coupling of two folding reactions in the denatured state. In a kinetic study of an engineered domain-swapped dimer of the N-terminal domain of CD2, in which an antiparallel  sheet is formed intermolecularly, the authors reach a similar conclusion that association occurs in the denatured state and folding of the intertwined dimer is a subsequent step [52]. To conclude, the mechanism of converting a monomeric protein into a domain-swapped oligomer is highly dependent on the folding mechanism of the particular protein. The more independent the structure of the swapping elements and the more framework-like the folding mechanism, the later during folding they can associate. In many cases, however, the swapped elements are highly intertwined and not independently stable, and the folding is a very concerted process; consequently, association will occur before too much of the structure is present. Domain Swapping, Refolding, and Aggregation A protein engineering analysis of the folding pathways of both the monomer and the dimer forms of suc1 revealed that the two folding processes are controlled by the same key interactions (or folding nucleus) [55, 56]. As a consequence, there is ambiguity in the refolding of suc1: either the interactions between the  strands are made intramolecularly and the monomer is formed, or two chains associate with the interactions between the  strands occurring intermolecularly and the domain-swapped dimer is formed. The competition between the collapse of a chain on itself to form a monomer and the probability of interacting with another chain before the key interactions are formed explains the effect of loop length on domain swapping. Loop lengthening slows down the rate of folding of the monomer because of the higher entropic cost of fixing a longer loop rather than a shorter one [58, 59], and this indirectly favors folding to the dimer because it increases the likelihood of association with another chain [58]. Loop lengthening thus promotes domain swapping by making it kinetically more accessible. Another consequence of increasing loop length or flexibility is the increased tendency to aggregate, as observed in suc1 and other proteins. And it is only one step further to suggest that protein aggregation could likewise occur using the same “native” nucleus but by swapping in an open-ended manner to form an extended chain. Indeed, several studies have illustrated the close relationship between aggregation and the monomeric folding reaction of proteins. Transient aggregation was observed during the refolding of suc1 as well as CI2, U1A [31, 35], and maltose binding protein [36], a dimeric folding intermediate has been characterized in the folding reaction of monomeric CheY [60], and a dimeric aciddenatured state has been observed for barnase [61].
Review 249
Biological Implications of Domain Swapping: A Balance between Optimizing Function and Minimizing Misfolding Aside from the regulatory switch that domain swapping may provide in a small number of proteins, a more general, and negative, picture for the significance of domain swapping in vivo is beginning to emerge. Loops play an important role in the function of proteins, and are often the location of binding sites or catalytic sites. The functional requirement for specific residues in specific orientations in these sites often results in the introduction of unfavorable interactions. The strain imposed, in suc1 for example, can even provide the means to facilitate signal transduction between remote parts of the structure. On the down side, however, the introduction of strain in loops can create a propensity to domain swap and thereby cause misfolding and aggregation. With this in mind, apparently neutral mutations such as those listed in the single nucleotide polymorphism (SNP) database should be evaluated further, as a side chain truncation that perturbs the stability even very little might transform a monomeric protein into an obligate dimer or higher oligomer with impaired function. Proteins may have evolved ways to protect against the vulnerability that can arise from optimizing function. One way is by compensatory mutations: for example, the first proline in the hinge loop of suc1 is conserved for the protein’s function, but it creates strain that makes the protein prone to domain swap; thus, as a balance, the second proline appears to be required to prevent misfolding and aggregation, thus acting as a gatekeeper for folding [35]. The second way is illustrated by the observation that, while mutations can be made that increase suc1’s equilibrium domain swapping propensity by many orders of magnitude, the product of refolding of all of these mutant proteins is still overwhelming the monomeric form (F. Rousseau et al., submitted). Thus, minimizing kinetic accessibility can limit the extent of domain swapping. Finally, the functional native state and a protein’s ability to fold to that state might also be protected in vivo with chaperones or other accessory proteins [62]. Related to this are the striking studies of Lindquist and coworkers, showing that the Hsp90 chaperones can protect proteins during evolution [63, 64]. They found that, by protecting the native state, Hsp90 acts a buffer or “capacitor” to allow a protein to accumulate multiple mutations that individually may be deleterious but that in combination can produce a new, advantageous phenotype. In several of the examples cited here, domainswapped oligomerization is transient in nature. An interesting observation in this regard is the reversible nature of bacterial inclusion bodies; upon chemical inhibition of protein synthesis, Carrio et al. observed resolubilization of inclusion bodies in E. coli [65]. Domain swapping might also have a role in aggregation processes that lead to disease. Studies on cystatin C, a mutant of which causes an amyloid disease, showed that the protein forms domain-swapped dimers that could be the precursors to amyloid formation [40, 41]. Also, the human prion protein is domain swapped in crystal form [39]. Although it is difficult to see how domain swapping could give rise to the final assembly of cross- structure that char-
acterizes amyloid fibrils, domain swapping could play a role in the early stages of fibril formation. Open-ended domain-swapped oligomers are alternative forms of the protein that are folded but unstable because they have unsatisfied surfaces at the termini. These species could therefore act as precursors or intermediates in fibril formation. They could provide a reservoir of partly folded or easily unfoldable molecules (see for example [66]), thereby giving rise to many nuclei for the formation of the ultimate fibril state. Of particular interest also is the yeast prion Ure2p, which retains its native ␣-helical conformation in fibrils, in contrast to a cross- structure, suggesting a mechanism for assembly similar to domain swapping [67]. Domain swapping may also facilitate other polymerization processes such as the formation of SNARE complexes in membrane fusion [68] and the assembly of pilus fibers [69]. Conclusions Protein engineering studies have greatly increased our understanding of the mechanism and sequence determinants of domain swapping, in particular by providing quantitative information. Modulation of the strain in the closed monomeric and more extended swapped hinge loop conformations appears to be a design strategy that is generally applicable. Mechanistic studies reveal a picture of domain swapping that is intimately linked to the folding reaction. By making use of the same folding nucleus, a protein can fold as a monomer or as a domainswapped oligomer. Higher order species could also assemble in a similar way and the open-ended domainswapped state provides a generic, structural model for the precursors in the formation of fibrillar diseased states. Finally, the model of a strained loop as the main driving force in domain swapping suggests that the phenomenon could be a rather common side effect, as many proteins are optimized for function at the cost of stability. References 1. Bennett, M.J., Choe, S., and Eisenberg, D. (1994). Domain swapping: entangling alliances between proteins. Proc. Natl. Acad. Sci. USA 91, 3127–3131. 2. Bennett, M.J., Schlunegger, M.P., and Eisenberg, D. (1995). 3D domain swapping: a mechanism for oligomer assembly. Protein Sci. 4, 2455–2468. 3. Liu, Y., and Eisenberg, D. (2002). 3D domain swapping: as domains continue to swap. Protein Sci. 11, 1285–1299. 4. Newcomer, M.E. (2002). Protein folding and three-dimensional domain swapping: a strained relationship? Curr. Opin. Struct. Biol. 12, 48–53. 5. Rousseau, F., Schymkowitz, J.W.H., Wilkinson, H.R., and Itzhaki, L.S. (2001). Three-dimensional domain swapping in p13suc1 occurs in the unfolded state and is controlled by conserved proline residues. Proc. Natl. Acad. Sci. USA 98, 5596– 5601. 6. Kortt, A.A., Malby, R.L., Caldwell, J.B., Gruen, L.C., Ivancic, N., Lawrence, M.C., Howlett, G.J., Webster, R.G., Hudson, P.J., and Colman, P.M. (1994). Recombinant anti-sialidase single-chain variable fragment antibody. Characterization, formation of dimer and higher-molecular-mass multimers and the solution of the crystal structure of the single-chain variable fragment/sialidase complex. Eur. J. Biochem. 221, 151–157. 7. Liu, Y., Hart, P.J., Schlunegger, M.P., and Eisenberg, D. (1998). The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1-A˚ resolution. Proc. Natl. Acad. Sci. USA 95, 3437–3442.
Structure 250
8. Green, S.M., Gittis, A.G., Meeker, A.K., and Lattman, E.E. (1995). One-step evolution of a dimer from a monomeric protein. Nat. Struct. Biol. 2, 746–751. 9. Khazanovich, N., Bateman, K., Chernaia, M., Michalak, M., and James, M. (1996). Crystal structure of the yeast cell-cycle control protein, p13suc1, in a strand-exchanged dimer. Structure 4, 299–309. 10. Zegers, I., Deswarte, J., and Wyns, L. (1999). Trimeric domainswapped barnase. Proc. Natl. Acad. Sci. USA 96, 818–822. 11. Schlunegger, M.P., Bennett, M.J., and Eisenberg, D. (1997). Oligomer formation by 3D domain swapping: a model for protein assembly and misassembly. Adv. Protein Chem. 50, 61–122. 12. Chen, Y.W., Stott, K., and Perutz, M.F. (1999). Crystal structure of a dimeric chymotrypsin inhibitor 2 mutant containing an inserted glutamine repeat. Proc. Natl. Acad. Sci. USA 96, 1257– 1261. 13. Kishan, K.V., Scita, G., Wong, W.T., Di Fiore, P.P., and Newcomer, M.E. (1997). The SH3 domain of Eps8 exists as a novel intertwined dimer. Nat. Struct. Biol. 4, 739–743. 14. Schiering, N., Casale, E., Caccia, P., Giordano, P., and Battistini, C. (2000). Dimer formation through domain swapping in the crystal structure of the Grb2-SH2-Ac-pYVNV complex. Biochemistry 39, 13376–13382. 15. Mazzarella, L., Capasso, S., Demasi, D., Dilorenzo, G., Mattia, C.A., and Zagari, A. (1993). Bovine seminal ribonuclease— structure at 1.9-Angstrom resolution. Acta Crystallogr. D Biol. Crystallogr. 49, 389–402. 16. Liu, Y., Gotte, G., Libonati, M., and Eisenberg, D. (2002). Structures of the two 3D domain-swapped RNase A trimers. Protein Sci. 11, 371–380. 17. Liu, Y.S., Gotte, G., Libonati, M., and Eisenberg, D. (2001). A domain-swapped RNase A dimer with implications for amyloid formation. Nat. Struct. Biol. 8, 211–214. 18. Stehle, T., and Harrison, S.C. (1996). Crystal structures of murine polyomavirus in complex with straight-chain and branchedchain sialyloligosaccharide receptor fragments. Structure 4, 183–194. 19. Bax, B., Lapatto, R., Nalini, V., Driessen, H., Lindley, P.F., Mahadevan, D., Blundell, T.L., and Slingsby, C. (1990). X-ray analysis of  B2-crystallin and evolution of oligomeric lens proteins. Nature 347, 776–780. 20. Bourne, Y., Watson, M.H., Arvai, A.S., Bernstein, S.L., Reed, S.I., and Tainer, J.A. (2000). Crystal structure and mutational analysis of the Saccharomyces cerevisiae cell cycle regulatory protein Cks1: implications for domain swapping, anion binding and protein interactions. Structure 8, 841–850. 21. Arvai, A.S., Bourne, Y., Hickey, M.J., and Tainer, J.A. (1995). Crystal structure of the human cell cycle protein CksHs1: single domain fold with similarity to kinase N-lobe domain. J. Mol. Biol. 249, 835–842. 22. Tegoni, M., Ramoni, R., Bignetti, E., Spinelli, S., and Cambillau, C. (1996). Domain swapping creates a third putative combining site in bovine odorant binding protein dimer. Nat. Struct. Biol. 3, 863–867. 23. Story, R.M., Weber, I.T., and Steitz, T.A. (1992). The structure of the E. coli recA protein monomer and polymer. Nature 355, 318–325. 24. Remington, S., Wiegand, G., and Huber, R. (1982). Crystallographic refinement and atomic models of two different forms of citrate synthase at 2.7 and 1.7 A˚ resolution. J. Mol. Biol. 158, 111–152. 25. SaintJean, A.P., Phillips, K.R., Creighton, D.J., and Stone, M.J. (1998). Active monomeric and dimeric forms of Pseudomonas putida glyoxalase I: evidence for 3D domain swapping. Biochemistry 37, 10345–10353. 26. SaintJean, A.P., and Creighton, D.J. (1999). Regulated 3D domain swapping of glyoxalase I. FASEB J. 13, A1560. 27. Vitagliano, L., Adinolfi, S., Sica, F., Merlino, A., Zagari, A., and Mazzarella, L. (1999). A potential allosteric subsite generated by domain swapping in bovine seminal ribonuclease. J. Mol. Biol. 293, 569–577. 28. Gouldson, P.R., Higgs, C., Smith, R.E., Dean, M.K., Gkoutos, G.V., and Reynolds, C.A. (2000). Dimerization and domain swap-
29.
30. 31.
32. 33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
ping in G-protein-coupled receptors: a computational study. Neuropsychopharmacology 23, S60–S77. Qu, C., Liljas, L., Opalka, N., Brugidou, C., Yeager, M., Beachy, R.N., Fauquet, C.M., Johnson, J.E., and Lin, T. (2000). 3D domain swapping modulates the stability of members of an icosahedral virus group. Structure 8, 1095–1103. Fink, A.L. (1998). Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold. Des. 3, R9–R23. Silow, M., and Oliveberg, M. (1997). Transient aggregates in protein folding are easily mistaken for folding intermediates. Proc. Natl. Acad. Sci. USA 94, 6084–6086. Cohen, F.E., and Prusiner, S.B. (1998). Pathologic conformations of prion proteins. Annu. Rev. Biochem. 67, 793–819. Klafki, H.W., Pick, A.I., Pardowitz, I., Cole, T., Awni, L.A., Barnikol, H.U., Mayer, F., Kratzin, H.D., and Hilschmann, N. (1993). Reduction of disulfide bonds in an amyloidogenic Bence Jones protein leads to formation of “amyloid-like” fibrils in vitro. Biol. Chem. Hoppe Seyler 374, 1117–1122. Sinha, N., Tsai, C.J., and Nussinov, R. (2001). A proposed structural model for amyloid fibril elongation: domain swapping forms an interdigitating -structure polymer. Protein Eng. 14, 93–103. Silow, M., Tan, Y.J., Fersht, A.R., and Oliveberg, M. (1999). Formation of short-lived protein aggregates directly from the coil in two-state folding. Biochemistry 38, 13006–13012. Ganesh, C., Zaidi, F.N., Udgaonkar, J.B., and Varadarajan, R. (2001). Reversible formation of on-pathway macroscopic aggregates during the folding of maltose binding protein. Protein Sci. 10, 1635–1644. London, J., Skrzynia, C., and Goldberg, M.E. (1974). Renaturation of Escherichia coli tryptophanase after exposure to 8 M urea. Evidence for the existence of nucleation centers. Eur. J. Biochem. 47, 409–415. Mannervik, B., Cameron, A.D., Fernandez, E., Gustafsson, A., Hansson, L.O., Jemth, P., Jiang, F.Y., Jones, T.A., Larsson, A.K., Nilsson, L.O., et al. (1998). An evolutionary approach to the design of glutathione-linked enzymes. Chem. Biol. Interact. 112, 15–21. Knaus, K.J., Morillas, M., Swietnicki, W., Malone, M., Surewicz, W.K., and Yee, V.C. (2001). Crystal structure of the human prion protein reveals a mechanism for oligomerization. Nat. Struct. Biol. 8, 770–774. Staniforth, R.A., Giannini, S., Higgins, L.D., Conroy, M.J., Hounslow, A.M., Jerala, R., Craven, C.J., and Waltho, J.P. (2001). Three-dimensional domain swapping in the folded and moltenglobule states of cystatins, an amyloid-forming structural superfamily. EMBO J. 20, 4774–4781. Janowski, R., Kozak, M., Jankowska, E., Groznka, Z., Grubb, A., Abrahamson, M., and Jaskolski, M. (2001). Human cystatin C, an amyloidogenic protein, dimerizes through three-dimensional domain swapping. Nat. Struct. Biol. 8, 316–320. Frank, K.M., Dyda, F., Dobrodumov, A., and Gronenborn, A.M. (2002). Core mutations switch monomeric protein GB1 into an intertwined tetramer. Nat. Struct. Biol. 9, 877–885. Luo, H., and Sharp, K. (2002). On the calculation of absolute macromolecular binding free energies. Proc. Natl. Acad. Sci. USA 99, 10399–10404. Seeliger, M.A., Schymkowitz, J.W., Rousseau, F., Wilkinson, H.R., and Itzhaki, L.S. (2002). Folding and association of the human cell cycle regulatory proteins ckshs1 and ckshs2. Biochemistry 41, 1202–1210. Parker, M.J., Dempsey, C.E., Hosszu, L.L., Waltho, J.P., and Clarke, A.R. (1998). Topology, sequence evolution and folding dynamics of an immunoglobulin domain. Nat. Struct. Biol. 5, 194–198. Arndt, K.M., Muller, K.M., and Pluckthun, A. (1998). Factors influencing the dimer to monomer transition of an antibody single-chain Fv fragment. Biochemistry 37, 12918–12926. Bergdoll, M., Remy, M.H., Cagnon, C., Masson, J.M., and Dumas, P. (1997). Proline-dependent oligomerization with arm exchange. Structure 5, 391–401. Tranier, S., Iobbi-Nivol, C., Birck, C., Ilbert, M., Mortier-Barrie`re, I., Me´jean, V., and Samama, J.-P. (2003). A novel protein fold and extreme domain swapping in the dimeric TorD chaperone from Shewanella massilia. Structure 11, 165–174.
Review 251
49. Schymkowitz, J.W., Rousseau, F., and Itzhaki, L.S. (2000). Sequence conservation provides the best prediction of the role of proline residues in p13suc1. J. Mol. Biol. 301, 199–204. 50. Kuhlman, B., O’Neill, J.W., Kim, D.E., Zhang, K.Y., and Baker, D. (2001). Conversion of monomeric protein L to an obligate dimer by computational protein design. Proc. Natl. Acad. Sci. USA 98, 10687–10691. 51. O’Neill, J.W., Kim, D.E., Johnsen, K., Baker, D., and Zhang, K.Y. (2001). Single-site mutations induce 3D domain swapping in the B1 domain of protein L from Peptostreptococcus magnus. Structure 9, 1017–1027. 52. Hayes, M.V., Sessions, R.B., Brady, R.L., and Clarke, A.R. (1999). Engineered assembly of intertwined oligomers of an immunoglobulin chain. J. Mol. Biol. 285, 1857–1867. 53. Schymkowitz, J.W., Rousseau, F., Wilkinson, H.R., Friedler, A., and Itzhaki, L.S. (2001). Observation of signal transduction in three-dimensional domain swapping. Nat. Struct. Biol. 8, 888–892. 54. Bourne, Y., Arvai, A.S., Bernstein, S.L., Watson, M.H., Reed, S.I., Endicott, J.A., Noble, M.E., Johnson, L.N., and Tainer, J.A. (1995). Crystal structure of the cell cycle-regulatory protein suc1 reveals a -hinge conformational switch. Proc. Natl. Acad. Sci. USA 92, 10232–10236. 55. Schymkowitz, J.W.H., Rousseau, F., Irvine, L.R., and Itzhaki, L.S. (2000). The folding pathway of the cell-cycle regulatory protein p13suc1: clues for the mechanism of domain swapping. Structure 8, 89–100. 56. Rousseau, F., Schymkowitz, J.W.H., Wilkinson, H.R., and Itzhaki, L.S. (2002). The structure of the transition state for folding of domain-swapped dimeric p13suc1. Structure 10, 649–657. 57. Alonso, D.O., Alm, E., and Daggett, V. (2000). Characterization of the unfolding pathway of the cell-cycle protein p13suc1 by molecular dynamics simulations: implications for domain swapping. Structure 8, 101–110. 58. Ladurner, A.G., and Fersht, A.R. (1999). Upper limit of the time scale for diffusion and chain collapse in chymotrypsin inhibitor 2. Nat. Struct. Biol. 6, 28–31. 59. Viguera, A.R., and Serrano, L. (1997). Loop length, intramolecular diffusion and protein folding. Nat. Struct. Biol. 4, 939–946. 60. Garcia, P., Serrano, L., Rico, M., and Bruix, M. (2002). An NMR view of the folding process of a CheY mutant at the residue level. Structure 10, 1173–1185. 61. Sanz, J.M., Johnson, C.M., and Fersht, A.R. (1994). The A-state of barnase. Biochemistry 33, 11189–11199. 62. Todd, M.J., Lorimer, G.H., and Thirumalai, D. (1996). Chaperonin-facilitated protein folding: optimization of rate and yield by an iterative annealing mechanism. Proc. Natl. Acad. Sci. USA 93, 4030–4035. 63. Rutherford, S.L., and Lindquist, S. (1998). Hsp90 as a capacitor for morphological evolution. Nature 396, 336–342. 64. Queitsch, C., Sangster, T.A., and Lindquist, S. (2002). Hsp90 as a capacitor of phenotypic variation. Nature 417, 618–624. 65. Carrio, M.M., and Villaverde, A. (2001). Protein aggregation as bacterial inclusion bodies is reversible. FEBS Lett. 489, 29–33. 66. Dobson, C.M. (2001). The structural basis of protein folding and its links with human disease. Philos. Trans. R. Soc. Lond. B Biol. Sci. 356, 133–145. 67. Bousset, L., Thomson, N.H., Radford, S.E., and Melki, R. (2002). The yeast prion Ure2p retains its native ␣-helical conformation upon assembly into protein fibrils in vitro. EMBO J. 21, 2903– 2911. 68. Kweon, D.H., Chen, Y., Zhang, F., Poirier, M., Kim, C.S., and Shin, Y.K. (2002). Probing domain swapping for the neuronal SNARE complex with electron paramagnetic resonance. Biochemistry 41, 5449–5452. 69. Sauer, F.G., Pinkner, J.S., Waksman, G., and Hultgren, S.J. (2002). Chaperone priming of pilus subunits facilitates a topological transition that drives fiber formation. Cell 111, 543–551. 70. Endicott, J.A., Noble, M.E., Garman, E.F., Brown, N., Ramussen, R., Nurse, P., and Johnson, L.N. (1995). The crystal structure of p13suc1, a p34cdc2-interacting cell cycle control protein. EMBO J. 14, 1004–1014. 71. Kraulis, P.J. (1991). MOLSCRIPT—a program to produce both
72.
73.
74.
75.
detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946–950. Ramoni, R., Vincent, F., Ashcroft, A.E., Accornero, P., Grolli, S., Valencia, C., Tegoni, M., and Cambillau, C. (2002). Control of domain swapping in bovine odorant binding protein. Biochem. J. 365, 739–748. Raaijmakers, H., Vix, O., Toro, I., Golz, S., Kemper, B., and Suck, D. (1999). X-ray structure of T4 endonuclease VII: a DNA junction resolvase with a novel fold and unusual domain-swapped dimer architecture. EMBO J. 18, 1447–1458. Kortt, A.A., Dolezal, O., Power, B.E., and Hudson, P.J. (2001). Dimeric and trimeric antibodies: high avidity scFvs for cancer targeting. Biomol. Eng. 18, 95–108. Barrientos, L.G., Louis, J.M., Botos, I., Mori, T., Han, Z., O’Keefe, B.R., Boyd, M.R., Wlodawer, A., and Gronenborn, A.M. (2002). The domain-swapped dimer of cyanovirin-N is in a metastable folded state: reconciliation of X-ray and NMR structures. Structure 10, 673–686.
PII S0969-2126(03)00029-7
The Unfolding Story of Three-Dimensional Domain Swapping Frederic Rousseau,1 Joost W.H. Schymkowitz,1,* and Laura S. Itzhaki2,3,* 1 European Molecular Biology Laboratory Meyerhofstrasse 1 69117 Heidelberg Germany 2 Cambridge University Chemical Laboratory MRC Centre for Protein Engineering Lensfield Road Cambridge, CB2 1EW United Kingdom
Summary Three-dimensional domain swapping is the event by which a monomer exchanges part of its structure with identical monomers to form an oligomer where each subunit has a similar structure to the monomer. The accumulating number of observations of this phenomenon in crystal structures has prompted speculation as to its biological relevance. Domain swapping was originally proposed to be a mechanism for the emergence of oligomeric proteins and as a means for functional regulation, but also to be a potentially harmful process leading to misfolding and aggregation. We highlight experimental studies carried out within the last few years that have led to a much greater understanding of the mechanism of domain swapping and of the residue- and structure-specific features that facilitate the process. We discuss the potential biological implications of domain swapping in light of these findings. Background Three-dimensional domain swapping is a process by which one protein molecule exchanges a domain with an identical partner (Figures 1A and 1B). Although it had been proposed as the mechanism for the formation of RNase A dimers as far back as 1962, domain swapping was first observed in crystal structures during the eighties and nineties, and the terminology was only formally introduced by Eisenberg and coworkers in 1994 to describe the crystal structure of diphtheria toxin [1, 2]. This structure led to a series of elegant theoretical papers by Eisenberg and coworkers that proposed how and why domain swapping might occur and the potential biological implications. The latter included domain swapping as a mechanism for regulating function, as an evolutionary strategy to create protein complexes, and as a mechanism for misfolding and aggregation (Figure 1C) [1, 2]. More than 30 crystal structures of domain-swapped molecules have been elucidated since [3, 4], but the *Correspondence: [email protected] (L.S.I.), [email protected] (J.W.H.S.) 3 Present address: Hutchison/MRC Centre, Hills Road, Cambridge, CB2 2XZ, United Kingdom.
Review
physiological significance of these highly intertwined oligomers remains unclear. Also, experimental data on the thermodynamics and the mechanism of domain swapping were, until very recently, almost only qualitative. In part, this was due to problems inherent in the experimental study of domain swapping. First, domainswapped oligomeric forms are often metastable; once formed, they take a long time to convert back to the more stable monomeric form and therefore any quantitative analysis, for example the measurement of equilibrium constants, becomes unfeasible. Second, tractable model systems are required, ones in which both monomer and domain-swapped forms can be isolated and studied easily in solution. In the last couple of years, these problems have been overcome and much has subsequently been revealed to explain domain swapping [5], in particular by the combined use of protein engineering and the biophysical tools that have been applied so successfully to investigate protein folding. As researchers become increasingly interested in understanding and quantifying protein aggregation, mechanisms of self-association such as domain swapping have become the focus of attention and in the future will continue to provide new insights into the process. Domain Swapping: Definition and Terminology Domain swapping is defined as the process by which two or more protein molecules exchange part of their structure to form intertwined oligomers. These oligomers are composed of subunits having the same structure as the original monomer, with the exception of the hinge loop that connects the exchanging part with the rest of the structure and which is usually folded back on itself in the monomer and extended in the domainswapped oligomer [1, 2]. The term “domain” is not used in a strict sense, as proteins have been reported to swap entire tertiary globular domains [2, 6] but also often only a single element of secondary structure, such as an ␣ helix [7, 8] or a single strand of  sheet [9]. Although initially only domain-swapped dimers were observed, barnase was later shown to be able to fold into a domainswapped trimer [10]. The exchanging part of the structure makes the same interactions in the oligomer as it makes in the monomer, but its interactions are formed inter- rather than intramolecularly. These interactions that are identical in the monomer and oligomer form the “primary” interface [11]. Because subunits are often close to each other in a domain-swapped oligomer, a new interaction interface that is absent in the monomer may be formed, and this is termed the “secondary” interface (Figure 1A) [11]. Biological Relevance of Domain Swapping: Hypotheses and Experimental Evidence Almost 10 years after the identification of domain swapping, popular model proteins such as chymotrypsin inhibitor 2 (CI2) [12], barnase [10], SH3 [13], and staphylococcal nuclease [8] have now crystallized in domain-
Structure 244
Figure 1. Basic Principles in Domain Swapping (A) Simplified representation of a monomer and a domain-swapped dimer, with the secondary surface highlighted in red (see text). (B) Schematic representation of structures of a monomer [70] and domain-swapped dimer [9, 54] of suc1 obtained using the program MOLSCRIPT [71]. The domain-swapping C-terminal  strand and the hinge loop are shown in red. (C) Schematic representation of open-ended and closed-ended domain-swapped oligomers.
swapped forms, as have many others. There is no apparent structural or functional connection between these different proteins and examples are appearing across organisms and structural and functional classifications. A more complete overview can be found elsewhere [3, 4], while below a few examples are presented to illustrate the diversity of domain swapping (see also Figure 2). Both SH2 and SH3 domains, probably the most studied protein-protein interaction domains in signal transduction, have been reported to crystallize as domainswapped dimers [13, 14]. The dimeric form of bovine seminal RNase was the first RNase found to domain swap [15], but later the bacterial RNase barnase was reported to form domain-swapped trimers [10], and bovine RNase A was recently shown to swap two different structural elements [7, 16, 17]. The first of these is formed by swapping an N-terminal ␣ helix and the second by swapping a C-terminal  strand. Other examples include a coat protein from Simian virus 40 [18], the lens eye protein B2-crystallin [19], the cell cycle regulatory proteins from the cks family [9, 20, 21], bovine odorant
binding protein OBP [22], the DNA recombination and repair protein from E. coli RecA [23], citrate synthase from chicken heart [24], and many more proteins carrying out diverse functions. However, clear evidence for a functional role of domain swapping is lacking and, as all these structures have been observed in crystal forms, they could easily be suspected to be an artifact of the nonphysiological conditions and high protein concentrations used for crystallization. Several hypotheses together with coincident lines of evidence nevertheless suggest a significance for domain swapping in vivo. The first potential biological role for domain swapping is the functional regulation of proteins. Here several interesting observations have been made. Glyoxalase I from Pseudomonas was shown to exist both in an active domain-swapped dimeric state and as a metastable and less active monomer. In vitro conversion from dimeric to metastable monomeric glyoxalase I can be triggered upon addition of glutathione [25, 26]. After removal of glutathione, glyoxalase I slowly reverts to a more stable and active domain-swapped form. Although no evidence exists for such a mechanism in vivo, this demon-
Review 245
Figure 2. Schematic Representations of Domain-Swapped Proteins (A) Bovine seminal ribonuclease (PDB ID code 11bg) [27]. (B) Suc1 (PDB ID code 1sce) [9, 54]. (C) Bovine odorant binding protein (PDB ID code 1g85) [72]. (D) Recombination endonuclease VII from [73] (PDB ID code 1e7d). (E) Rice yellow mottle virus coat protein (PDB ID code 1f2n) [29]. This figure was generated using PyMOL (http://www.pymol.org).
strates that functional regulation by domain swapping can, in principle, be achieved. Another interesting finding is that formation of domain-swapped oligomers can introduce allosteric regulation of the protein activity already present in the monomer, as shown in bovine seminal RNase [27]. Also, several receptor proteins, including G-coupled receptors, are thought to dimerize by domain swapping [28]. Finally, the viral capsid of an icosahedral virus that was recently found to be composed of domain-swapped dimers has higher thermostability than other viruses from the same family lacking the domainswapped architecture [28, 29]. A second proposed biological implication of domain swapping is a detrimental one. Domain swapping has the potential to be a mechanism for protein misfolding, aggregation, and amyloid formation (Figure 1C) [11, 30, 31] and thus it may be related to misfolding leading to amyloid and prion diseases [11, 17, 30, 32–34]. Here again, direct evidence for domain swapping as a mechanism for aggregation and amyloid formation is lacking, but several observations strongly suggest that it might indeed be the case. First, transient aggregation observed during refolding of several proteins has been suggested to occur by domain swapping [35–38]. Second, a correlation between domain swapping propensity and the rate of heat aggregation of suc1 was observed [5]. Third, a role for domain swapping in the processes of prion and amyloid formation is suggested by the crystallization of both the human prion protein [39] and the
amyloidogenic human cystatin C [40, 41] in domainswapped forms. Fourth, Eisenberg and coworkers were able to design both domain-swapped dimers and highorder oligomers from the same three-helix bundle structural motif but with different topologies [7, 16, 17]. An updown-down favored a reciprocal swap to form a closed domain-swapped dimer. An up-down-up topology was predicted to favor an extended swap, and fibrils were formed that were assumed to be domain swapped. Finally, domain swapping has been proposed to contribute to structural diversification and the emergence of oligomers during evolution [11], but here again, although the idea is elegant and simple, it is not clear what the relevance of this is in vivo. Interestingly, it was recently shown that core mutations could introduce oligomerization by domain swapping, with the formation of a new hydrophobic interface and structural rearrangement of other parts of the protein [42]. Energetic Contributions to Domain Swapping A quantitative understanding is needed, first of the factors that can contribute to the equilibrium between a monomer and its oligomeric alter ego, and second of how interconversion between the two forms occurs. Oligomers are always disfavored by their greater translational and rotational entropy [11]. Further, domainswapped oligomers are made up of subunits that are identical in structure to the monomer, except for (1) the hinge loop and (2) the secondary interface [11]. These
Structure 246
Table 1. Dissociation Constants for Suc1 Wild-Type and Mutants in the Hinge Loop Mutation
Kd (mM)
Wild-type P90A P92A E91P P90AP92A ⌬8789
1.9 890 0.2 0.0001 11.2 1 ⫻ 10⫺6
Values were taken from [5, 53].
conformational differences can either favor or disfavor domain swapping and will greatly determine the equilibrium between monomer and domain-swapped forms. A convenient way to quantify the monomer-dimer equilibrium is by using the dissociation constant Kd, defined as Kd ⫽
[M]2 [D]
(1)
where [M] is the monomer concentration and [D] is the dimer concentration in monomeric units. The Kd can be seen as the protein concentration at which half of the protein molecules are dimeric. The free energy difference associated with this equilibrium can be obtained from ⌬G ⫽ ⫺RT ln Kd
(2)
The Kd for dimerization has been measured for a handful of domain-swapped dimers, and Table 1 shows the values for the protein suc1 and a subset of mutants. The Kd’s for the natural proteins measured to date are in the micromolar and millimolar range, which is far above the nanomolar range that would be expected for a biologically relevant process. However, the energy required to tip the balance is not large, considering that the entropic cost of fixing two molecules into a dimer is probably on the order of a few kilocalories [43]. From Equation 2, a free energy change of approximately 4 kcal mol⫺1 results in a change in the dissociation constant of three orders of magnitude. Thus, a few mutations in the hinge loop or in the secondary interface, designed to favor the dimer, could easily bring the Kd into the nanomolar range. This is illustrated in Table 1 for suc1, and the effects of the different mutations are explained below. The Hinge Loop and the Domain Swapping Propensity What are the specific sequence and structural requirements that turn a protein segment (generally a loop) into a hinge loop that facilitates domain swapping? Some early experiments addressed this question by focusing on the length of the hinge loop. Shortening the hinge loop in staphylococcal nuclease [8], suc1 [5], ckshs1 [44], CD2 [45], and single-chain Fv [46] destabilizes these monomeric proteins by making the hinge loop too tight to close the monomeric conformation. This is sufficient to induce formation of domain-swapped dimers in these proteins. Here, the entropic cost of association is more than compensated by the energetically
more favorable extended conformation of the shortened hinge loop in the domain-swapped form. Loop lengthening also has an effect on the oligomerization state of proteins. Monomeric CI2 forms domain-swapped dimers, as well as trimers and even higher order oligomers, when its active site loop is lengthened by 5–10 residues [12]. Moreover, the presence of high-order oligomeric forms increases with loop insertions consisting of repeats of Gln→Ala→Gly. The same observation was made in suc1: wild-type suc1 only forms monomers and domain-swapped dimers at protein concentrations in the low millimolar range. However, when both prolines of its hinge loop are mutated to alanine, trimers and tetramers are observed, even more so when both prolines are substituted by glycines (F.R., J.W.H.S., and L.S.I., unpublished results). Thus, making the hinge loop more flexible, either by lengthening it or by mutation, also induces swapping. There is a kinetic, as well as a thermodynamic component of the effect of loop lengthening, discussed later in the review. How else could the hinge loop sequence affect the domain swapping propensity? Bergdoll and coworkers observed that prolines are frequently found in the hinge loops of proteins that domain swap by arm exchange [47], and most recently the TorD chaperone was found as a domain-swapped dimer with a double-proline motif in its hinge [48]. Experiments on suc1 suggest a thermodynamic explanation. Suc1, which can fold as a monomer or an intertwined dimer with a Kd of 1.8 mM, has two prolines in its hinge loop [49]. Mutation of either proline to alanine dramatically shifts the monomer-dimer equilibrium, while mutation to alanine of any other residue in the hinge loop has very little effect [5] (see Table 2). Pro90 is unfavorable in the monomer hinge conformation; the mutation Pro90Ala shifts the equilibrium completely toward the monomer. Pro92, on the other hand, is unfavorable in the dimer; the mutation Pro92Ala shifts the equilibrium toward the dimer and has a Kd of 100 M (Table 1). Thus, the hinge loop in both monomer and dimer forms of the protein is in a “strained” conformation, more so in the monomer. The strain causes the hinge loop to act like a loaded molecular spring that relieves the tension by adopting an alternative conformation without altering the rest of the structure. In the case of suc1, the destabilization of the hinge loop by the prolines does not appear to arise from unfavorable interactions with other residues but rather from backbone strain that the prolines impose on each other as a double mutant cycle (Pro90AlaPro92Ala) demonstrated [5, 49]. We were able to engineer mutants with domain swapping propensities spanning nine orders of magnitude, by manipulating the strain in the hinge loop conformation of suc1 (Figure 3; Table 1) [5]. The principle that unfavorable interactions turn hinge loops into loaded springs also appears to be generally applicable. In protein L [50, 51], for instance, the monomeric hinge loop was destabilized by forcing residues into forbidden regions of the Ramachandran plot, thereby inducing an alternate, domain-swapped form. In conclusion, domain swapping ability can be introduced into a protein by manipulating loops in one of two ways. The first is by destabilizing the conformation of a loop in the monomer, either by shortening it or
Review 247
Table 2. Design Approaches for Domain Swapping Protein
Description
Hinge Shortening Staphylococcal nuclease [8] Suc1 [5] CD2 [45] Single-chain Fv antibody [74]
Deletion of 6 residues from hinge loop Deletion of 3 residues from hinge loop Deletion of 2 residues from hinge loop Length of linker between variable domains controls oligomerization state. Short linkers (2–10 residues) lead to oligomer formation; longer linkers (12–25 residues) lead to stable monomeric forms. Introduction of Strain in Hinge by Amino Acid Substitution Suc1 [5] Mutation of glutamate 91 in the hinge loop to proline Ckshs1 [44] Mutation of glutamate 63 in the hinge loop to proline Protein L [50] Mutations in the hinge loop A52V, N53P, and G55A Cyanovirin-N [75] Mutation of serine 52 to proline in hinge loop Hinge Lengthening CI2 [12] Insertion of glutamine repeat in hinge loop Introduction of Flexibility in Hinge Suc1 [5] Mutation of prolines 90 and 92 to glycine in hinge loop Other Mutations CD2 [45] Mutation of residues in the intermolecular interface GB1 [42] Mutation of residues in the hydrophobic core These mutational strategies were used to successfully convert monomeric proteins into obligate dimers. For terminology, see text.
by introducing unfavorable interactions. This forces the loop into a more favorable conformation in a domainswapped variant of the original monomeric structure. Because “spring loading” applies to rather short and inflexible loops, the propensity for domain swapping achieved in this way is high and is biased toward a specific oligomeric state. The second way to facilitate domain swapping is by inserting a long and/or flexible loop into the monomeric structure. Here the propensity to domain swap is generally lower than that achieved using the first approach, and also less specific because the flexibility of the loop allows it to adopt many conformations with similar energy. There is a kinetic component to these effects, as discussed later. Contribution of Residues Outside the Hinge to Domain Swapping Propensity How do these changes in domain swapping propensity compare to those that occur upon mutation elsewhere in the protein? Mutagenesis studies of CD2 were focused on the secondary interface—the new intermolecular interface that is created in the domain-swapped
form. These interactions contribute to the domain swapping propensity by stabilizing the dimer [46, 52]. Suc1 has no secondary interface, and therefore the equilibrium between monomer and dimer is solely determined by the hinge loop. In spite of this, mutations to alanine elsewhere in the protein were not neutral in their effect on the equilibrium between monomer and dimer. Although the effects were not as large as for mutation in the hinge loop, they were significant (changes in Kd of one to two orders of magnitude) [53]. Specifically, mutations to alanine in the  sheet shifted the equilibrium toward the monomer, showing that those interactions are more favorable in the dimer. Further, a phosphoprotein binding site of suc1 in the  sheet on the opposite side of the molecule to the hinge loop and at a distance of more than 20 A˚, has a different affinity for phosphosubstrates in the monomer compared with the dimer [53]. The crystal structures of the monomer and domainswapped dimer are highly superimposable outside of the hinge loop region, so that there is no obvious structural reason for these effects. The different response to mutation of the two forms of the protein appears to Figure 3. Scheme Showing the Consequences of Backbone Strain Created by Proline Residues The hinge residues are represented as colorcoded bricks. Prolines are shown in dark blue and their preceding peptide bond is in bold to indicate the restrictions that are imposed on it by the ring of the proline. Positions of backbone strain are shown by yellow zigzag lines. There is an equilibrium in the wild-type protein between the monomer conformation in which P90 is strained and the dimer conformation in which P92 is strained [5].
Structure 248
originate in the hinge loop. The conformation of the hinge loop is more strained in the monomer than in the dimer [5], and the strain is sensed throughout the  sheet. As a result, the  sheet is somewhat less satisfied when accompanied by the  turn hinge in the monomer conformation than with the extended hinge conformation in the dimer [53]. Indeed, removing the strain in the hinge loop by mutating both prolines to alanine reduces the difference in energetic behavior of the two forms. The biological implications of these observations are discussed later. In conclusion, however, the presence of an unfavorable loop conformation in a protein does not only promote domain swapping, but also subtly affects the energetics of the entire structure, probably by altering its dynamic behavior.
An Unfolding Mechanism for Domain Swapping Domain swapping is essentially a protein-folding phenomenon, as it gives a single polypeptide sequence access to two or more well-defined native states. Two questions immediately arise. First, what are the folding pathways to monomer and domain-swapped oligomer and how are they connected to each other? Second, can one control which state is produced upon refolding of a domain-swapping protein? Generally, interconversion between monomer and dimer is very slow (from days to months) [5, 11], and thus both states are separated by a large kinetic barrier. For interconversion to occur, many native interactions, often in the hydrophobic core, must be disrupted, to be replaced by identical interactions with another protein chain. Eisenberg and coworkers proposed that the transition state for interconversion is an “open” form of the structure, disrupting many native interactions while maintaining the overall native fold of the exchanging domains, which is both enthalpically and entropically unfavorable [11]. Such a scenario is most likely for cases that involve the swapping of true domains that can fold independently of each other and then associate. However, several domain-swapped structures may require a more intricate interplay between folding and association. In the case of barnase [10], for example, the parts of the structure that swap can only fold partially in an independent manner. The authors of the domainswapped barnase structure therefore proposed that association must occur at the latest at an intermediate stage of folding. The situation is even more extreme in the case of suc1 and its homologs, where a  strand that is a central part of the hydrophobic core is exchanged to form the domain-swapped dimer [9, 20, 54]. A protein engineering analysis of the folding mechanism of both monomeric [55] and dimeric [56] suc1, complemented by molecular dynamics simulations of unfolding of the monomer [57], shows that the interactions that connect this  strand to the rest of the protein are formed very early in the folding reaction and thus that their disruption requires substantial unfolding. Further, the time required for monomer-dimer equilibration in suc1 can be reduced from a few months to a few minutes simply by increasing the unfolding rate significantly by changing the temperature, denaturant concentration, or pH [5]. Likewise, the hu-
man homolog of suc1, ckshs1, unfolds much faster than suc1, and the monomer-dimer interconversion is also much faster compared with suc1 [45]. These data led to a model in which a suc1 molecule can fold either as a monomer or a domain-swapped dimer from the denatured state [5]. Interconversion has to proceed via the denatured state and the kinetic barrier between monomer and dimer under native conditions is a direct consequence of the coupling of two folding reactions in the denatured state. In a kinetic study of an engineered domain-swapped dimer of the N-terminal domain of CD2, in which an antiparallel  sheet is formed intermolecularly, the authors reach a similar conclusion that association occurs in the denatured state and folding of the intertwined dimer is a subsequent step [52]. To conclude, the mechanism of converting a monomeric protein into a domain-swapped oligomer is highly dependent on the folding mechanism of the particular protein. The more independent the structure of the swapping elements and the more framework-like the folding mechanism, the later during folding they can associate. In many cases, however, the swapped elements are highly intertwined and not independently stable, and the folding is a very concerted process; consequently, association will occur before too much of the structure is present. Domain Swapping, Refolding, and Aggregation A protein engineering analysis of the folding pathways of both the monomer and the dimer forms of suc1 revealed that the two folding processes are controlled by the same key interactions (or folding nucleus) [55, 56]. As a consequence, there is ambiguity in the refolding of suc1: either the interactions between the  strands are made intramolecularly and the monomer is formed, or two chains associate with the interactions between the  strands occurring intermolecularly and the domain-swapped dimer is formed. The competition between the collapse of a chain on itself to form a monomer and the probability of interacting with another chain before the key interactions are formed explains the effect of loop length on domain swapping. Loop lengthening slows down the rate of folding of the monomer because of the higher entropic cost of fixing a longer loop rather than a shorter one [58, 59], and this indirectly favors folding to the dimer because it increases the likelihood of association with another chain [58]. Loop lengthening thus promotes domain swapping by making it kinetically more accessible. Another consequence of increasing loop length or flexibility is the increased tendency to aggregate, as observed in suc1 and other proteins. And it is only one step further to suggest that protein aggregation could likewise occur using the same “native” nucleus but by swapping in an open-ended manner to form an extended chain. Indeed, several studies have illustrated the close relationship between aggregation and the monomeric folding reaction of proteins. Transient aggregation was observed during the refolding of suc1 as well as CI2, U1A [31, 35], and maltose binding protein [36], a dimeric folding intermediate has been characterized in the folding reaction of monomeric CheY [60], and a dimeric aciddenatured state has been observed for barnase [61].
Review 249
Biological Implications of Domain Swapping: A Balance between Optimizing Function and Minimizing Misfolding Aside from the regulatory switch that domain swapping may provide in a small number of proteins, a more general, and negative, picture for the significance of domain swapping in vivo is beginning to emerge. Loops play an important role in the function of proteins, and are often the location of binding sites or catalytic sites. The functional requirement for specific residues in specific orientations in these sites often results in the introduction of unfavorable interactions. The strain imposed, in suc1 for example, can even provide the means to facilitate signal transduction between remote parts of the structure. On the down side, however, the introduction of strain in loops can create a propensity to domain swap and thereby cause misfolding and aggregation. With this in mind, apparently neutral mutations such as those listed in the single nucleotide polymorphism (SNP) database should be evaluated further, as a side chain truncation that perturbs the stability even very little might transform a monomeric protein into an obligate dimer or higher oligomer with impaired function. Proteins may have evolved ways to protect against the vulnerability that can arise from optimizing function. One way is by compensatory mutations: for example, the first proline in the hinge loop of suc1 is conserved for the protein’s function, but it creates strain that makes the protein prone to domain swap; thus, as a balance, the second proline appears to be required to prevent misfolding and aggregation, thus acting as a gatekeeper for folding [35]. The second way is illustrated by the observation that, while mutations can be made that increase suc1’s equilibrium domain swapping propensity by many orders of magnitude, the product of refolding of all of these mutant proteins is still overwhelming the monomeric form (F. Rousseau et al., submitted). Thus, minimizing kinetic accessibility can limit the extent of domain swapping. Finally, the functional native state and a protein’s ability to fold to that state might also be protected in vivo with chaperones or other accessory proteins [62]. Related to this are the striking studies of Lindquist and coworkers, showing that the Hsp90 chaperones can protect proteins during evolution [63, 64]. They found that, by protecting the native state, Hsp90 acts a buffer or “capacitor” to allow a protein to accumulate multiple mutations that individually may be deleterious but that in combination can produce a new, advantageous phenotype. In several of the examples cited here, domainswapped oligomerization is transient in nature. An interesting observation in this regard is the reversible nature of bacterial inclusion bodies; upon chemical inhibition of protein synthesis, Carrio et al. observed resolubilization of inclusion bodies in E. coli [65]. Domain swapping might also have a role in aggregation processes that lead to disease. Studies on cystatin C, a mutant of which causes an amyloid disease, showed that the protein forms domain-swapped dimers that could be the precursors to amyloid formation [40, 41]. Also, the human prion protein is domain swapped in crystal form [39]. Although it is difficult to see how domain swapping could give rise to the final assembly of cross- structure that char-
acterizes amyloid fibrils, domain swapping could play a role in the early stages of fibril formation. Open-ended domain-swapped oligomers are alternative forms of the protein that are folded but unstable because they have unsatisfied surfaces at the termini. These species could therefore act as precursors or intermediates in fibril formation. They could provide a reservoir of partly folded or easily unfoldable molecules (see for example [66]), thereby giving rise to many nuclei for the formation of the ultimate fibril state. Of particular interest also is the yeast prion Ure2p, which retains its native ␣-helical conformation in fibrils, in contrast to a cross- structure, suggesting a mechanism for assembly similar to domain swapping [67]. Domain swapping may also facilitate other polymerization processes such as the formation of SNARE complexes in membrane fusion [68] and the assembly of pilus fibers [69]. Conclusions Protein engineering studies have greatly increased our understanding of the mechanism and sequence determinants of domain swapping, in particular by providing quantitative information. Modulation of the strain in the closed monomeric and more extended swapped hinge loop conformations appears to be a design strategy that is generally applicable. Mechanistic studies reveal a picture of domain swapping that is intimately linked to the folding reaction. By making use of the same folding nucleus, a protein can fold as a monomer or as a domainswapped oligomer. Higher order species could also assemble in a similar way and the open-ended domainswapped state provides a generic, structural model for the precursors in the formation of fibrillar diseased states. Finally, the model of a strained loop as the main driving force in domain swapping suggests that the phenomenon could be a rather common side effect, as many proteins are optimized for function at the cost of stability. References 1. Bennett, M.J., Choe, S., and Eisenberg, D. (1994). Domain swapping: entangling alliances between proteins. Proc. Natl. Acad. Sci. USA 91, 3127–3131. 2. Bennett, M.J., Schlunegger, M.P., and Eisenberg, D. (1995). 3D domain swapping: a mechanism for oligomer assembly. Protein Sci. 4, 2455–2468. 3. Liu, Y., and Eisenberg, D. (2002). 3D domain swapping: as domains continue to swap. Protein Sci. 11, 1285–1299. 4. Newcomer, M.E. (2002). Protein folding and three-dimensional domain swapping: a strained relationship? Curr. Opin. Struct. Biol. 12, 48–53. 5. Rousseau, F., Schymkowitz, J.W.H., Wilkinson, H.R., and Itzhaki, L.S. (2001). Three-dimensional domain swapping in p13suc1 occurs in the unfolded state and is controlled by conserved proline residues. Proc. Natl. Acad. Sci. USA 98, 5596– 5601. 6. Kortt, A.A., Malby, R.L., Caldwell, J.B., Gruen, L.C., Ivancic, N., Lawrence, M.C., Howlett, G.J., Webster, R.G., Hudson, P.J., and Colman, P.M. (1994). Recombinant anti-sialidase single-chain variable fragment antibody. Characterization, formation of dimer and higher-molecular-mass multimers and the solution of the crystal structure of the single-chain variable fragment/sialidase complex. Eur. J. Biochem. 221, 151–157. 7. Liu, Y., Hart, P.J., Schlunegger, M.P., and Eisenberg, D. (1998). The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1-A˚ resolution. Proc. Natl. Acad. Sci. USA 95, 3437–3442.
Structure 250
8. Green, S.M., Gittis, A.G., Meeker, A.K., and Lattman, E.E. (1995). One-step evolution of a dimer from a monomeric protein. Nat. Struct. Biol. 2, 746–751. 9. Khazanovich, N., Bateman, K., Chernaia, M., Michalak, M., and James, M. (1996). Crystal structure of the yeast cell-cycle control protein, p13suc1, in a strand-exchanged dimer. Structure 4, 299–309. 10. Zegers, I., Deswarte, J., and Wyns, L. (1999). Trimeric domainswapped barnase. Proc. Natl. Acad. Sci. USA 96, 818–822. 11. Schlunegger, M.P., Bennett, M.J., and Eisenberg, D. (1997). Oligomer formation by 3D domain swapping: a model for protein assembly and misassembly. Adv. Protein Chem. 50, 61–122. 12. Chen, Y.W., Stott, K., and Perutz, M.F. (1999). Crystal structure of a dimeric chymotrypsin inhibitor 2 mutant containing an inserted glutamine repeat. Proc. Natl. Acad. Sci. USA 96, 1257– 1261. 13. Kishan, K.V., Scita, G., Wong, W.T., Di Fiore, P.P., and Newcomer, M.E. (1997). The SH3 domain of Eps8 exists as a novel intertwined dimer. Nat. Struct. Biol. 4, 739–743. 14. Schiering, N., Casale, E., Caccia, P., Giordano, P., and Battistini, C. (2000). Dimer formation through domain swapping in the crystal structure of the Grb2-SH2-Ac-pYVNV complex. Biochemistry 39, 13376–13382. 15. Mazzarella, L., Capasso, S., Demasi, D., Dilorenzo, G., Mattia, C.A., and Zagari, A. (1993). Bovine seminal ribonuclease— structure at 1.9-Angstrom resolution. Acta Crystallogr. D Biol. Crystallogr. 49, 389–402. 16. Liu, Y., Gotte, G., Libonati, M., and Eisenberg, D. (2002). Structures of the two 3D domain-swapped RNase A trimers. Protein Sci. 11, 371–380. 17. Liu, Y.S., Gotte, G., Libonati, M., and Eisenberg, D. (2001). A domain-swapped RNase A dimer with implications for amyloid formation. Nat. Struct. Biol. 8, 211–214. 18. Stehle, T., and Harrison, S.C. (1996). Crystal structures of murine polyomavirus in complex with straight-chain and branchedchain sialyloligosaccharide receptor fragments. Structure 4, 183–194. 19. Bax, B., Lapatto, R., Nalini, V., Driessen, H., Lindley, P.F., Mahadevan, D., Blundell, T.L., and Slingsby, C. (1990). X-ray analysis of  B2-crystallin and evolution of oligomeric lens proteins. Nature 347, 776–780. 20. Bourne, Y., Watson, M.H., Arvai, A.S., Bernstein, S.L., Reed, S.I., and Tainer, J.A. (2000). Crystal structure and mutational analysis of the Saccharomyces cerevisiae cell cycle regulatory protein Cks1: implications for domain swapping, anion binding and protein interactions. Structure 8, 841–850. 21. Arvai, A.S., Bourne, Y., Hickey, M.J., and Tainer, J.A. (1995). Crystal structure of the human cell cycle protein CksHs1: single domain fold with similarity to kinase N-lobe domain. J. Mol. Biol. 249, 835–842. 22. Tegoni, M., Ramoni, R., Bignetti, E., Spinelli, S., and Cambillau, C. (1996). Domain swapping creates a third putative combining site in bovine odorant binding protein dimer. Nat. Struct. Biol. 3, 863–867. 23. Story, R.M., Weber, I.T., and Steitz, T.A. (1992). The structure of the E. coli recA protein monomer and polymer. Nature 355, 318–325. 24. Remington, S., Wiegand, G., and Huber, R. (1982). Crystallographic refinement and atomic models of two different forms of citrate synthase at 2.7 and 1.7 A˚ resolution. J. Mol. Biol. 158, 111–152. 25. SaintJean, A.P., Phillips, K.R., Creighton, D.J., and Stone, M.J. (1998). Active monomeric and dimeric forms of Pseudomonas putida glyoxalase I: evidence for 3D domain swapping. Biochemistry 37, 10345–10353. 26. SaintJean, A.P., and Creighton, D.J. (1999). Regulated 3D domain swapping of glyoxalase I. FASEB J. 13, A1560. 27. Vitagliano, L., Adinolfi, S., Sica, F., Merlino, A., Zagari, A., and Mazzarella, L. (1999). A potential allosteric subsite generated by domain swapping in bovine seminal ribonuclease. J. Mol. Biol. 293, 569–577. 28. Gouldson, P.R., Higgs, C., Smith, R.E., Dean, M.K., Gkoutos, G.V., and Reynolds, C.A. (2000). Dimerization and domain swap-
29.
30. 31.
32. 33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
ping in G-protein-coupled receptors: a computational study. Neuropsychopharmacology 23, S60–S77. Qu, C., Liljas, L., Opalka, N., Brugidou, C., Yeager, M., Beachy, R.N., Fauquet, C.M., Johnson, J.E., and Lin, T. (2000). 3D domain swapping modulates the stability of members of an icosahedral virus group. Structure 8, 1095–1103. Fink, A.L. (1998). Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold. Des. 3, R9–R23. Silow, M., and Oliveberg, M. (1997). Transient aggregates in protein folding are easily mistaken for folding intermediates. Proc. Natl. Acad. Sci. USA 94, 6084–6086. Cohen, F.E., and Prusiner, S.B. (1998). Pathologic conformations of prion proteins. Annu. Rev. Biochem. 67, 793–819. Klafki, H.W., Pick, A.I., Pardowitz, I., Cole, T., Awni, L.A., Barnikol, H.U., Mayer, F., Kratzin, H.D., and Hilschmann, N. (1993). Reduction of disulfide bonds in an amyloidogenic Bence Jones protein leads to formation of “amyloid-like” fibrils in vitro. Biol. Chem. Hoppe Seyler 374, 1117–1122. Sinha, N., Tsai, C.J., and Nussinov, R. (2001). A proposed structural model for amyloid fibril elongation: domain swapping forms an interdigitating -structure polymer. Protein Eng. 14, 93–103. Silow, M., Tan, Y.J., Fersht, A.R., and Oliveberg, M. (1999). Formation of short-lived protein aggregates directly from the coil in two-state folding. Biochemistry 38, 13006–13012. Ganesh, C., Zaidi, F.N., Udgaonkar, J.B., and Varadarajan, R. (2001). Reversible formation of on-pathway macroscopic aggregates during the folding of maltose binding protein. Protein Sci. 10, 1635–1644. London, J., Skrzynia, C., and Goldberg, M.E. (1974). Renaturation of Escherichia coli tryptophanase after exposure to 8 M urea. Evidence for the existence of nucleation centers. Eur. J. Biochem. 47, 409–415. Mannervik, B., Cameron, A.D., Fernandez, E., Gustafsson, A., Hansson, L.O., Jemth, P., Jiang, F.Y., Jones, T.A., Larsson, A.K., Nilsson, L.O., et al. (1998). An evolutionary approach to the design of glutathione-linked enzymes. Chem. Biol. Interact. 112, 15–21. Knaus, K.J., Morillas, M., Swietnicki, W., Malone, M., Surewicz, W.K., and Yee, V.C. (2001). Crystal structure of the human prion protein reveals a mechanism for oligomerization. Nat. Struct. Biol. 8, 770–774. Staniforth, R.A., Giannini, S., Higgins, L.D., Conroy, M.J., Hounslow, A.M., Jerala, R., Craven, C.J., and Waltho, J.P. (2001). Three-dimensional domain swapping in the folded and moltenglobule states of cystatins, an amyloid-forming structural superfamily. EMBO J. 20, 4774–4781. Janowski, R., Kozak, M., Jankowska, E., Groznka, Z., Grubb, A., Abrahamson, M., and Jaskolski, M. (2001). Human cystatin C, an amyloidogenic protein, dimerizes through three-dimensional domain swapping. Nat. Struct. Biol. 8, 316–320. Frank, K.M., Dyda, F., Dobrodumov, A., and Gronenborn, A.M. (2002). Core mutations switch monomeric protein GB1 into an intertwined tetramer. Nat. Struct. Biol. 9, 877–885. Luo, H., and Sharp, K. (2002). On the calculation of absolute macromolecular binding free energies. Proc. Natl. Acad. Sci. USA 99, 10399–10404. Seeliger, M.A., Schymkowitz, J.W., Rousseau, F., Wilkinson, H.R., and Itzhaki, L.S. (2002). Folding and association of the human cell cycle regulatory proteins ckshs1 and ckshs2. Biochemistry 41, 1202–1210. Parker, M.J., Dempsey, C.E., Hosszu, L.L., Waltho, J.P., and Clarke, A.R. (1998). Topology, sequence evolution and folding dynamics of an immunoglobulin domain. Nat. Struct. Biol. 5, 194–198. Arndt, K.M., Muller, K.M., and Pluckthun, A. (1998). Factors influencing the dimer to monomer transition of an antibody single-chain Fv fragment. Biochemistry 37, 12918–12926. Bergdoll, M., Remy, M.H., Cagnon, C., Masson, J.M., and Dumas, P. (1997). Proline-dependent oligomerization with arm exchange. Structure 5, 391–401. Tranier, S., Iobbi-Nivol, C., Birck, C., Ilbert, M., Mortier-Barrie`re, I., Me´jean, V., and Samama, J.-P. (2003). A novel protein fold and extreme domain swapping in the dimeric TorD chaperone from Shewanella massilia. Structure 11, 165–174.
Review 251
49. Schymkowitz, J.W., Rousseau, F., and Itzhaki, L.S. (2000). Sequence conservation provides the best prediction of the role of proline residues in p13suc1. J. Mol. Biol. 301, 199–204. 50. Kuhlman, B., O’Neill, J.W., Kim, D.E., Zhang, K.Y., and Baker, D. (2001). Conversion of monomeric protein L to an obligate dimer by computational protein design. Proc. Natl. Acad. Sci. USA 98, 10687–10691. 51. O’Neill, J.W., Kim, D.E., Johnsen, K., Baker, D., and Zhang, K.Y. (2001). Single-site mutations induce 3D domain swapping in the B1 domain of protein L from Peptostreptococcus magnus. Structure 9, 1017–1027. 52. Hayes, M.V., Sessions, R.B., Brady, R.L., and Clarke, A.R. (1999). Engineered assembly of intertwined oligomers of an immunoglobulin chain. J. Mol. Biol. 285, 1857–1867. 53. Schymkowitz, J.W., Rousseau, F., Wilkinson, H.R., Friedler, A., and Itzhaki, L.S. (2001). Observation of signal transduction in three-dimensional domain swapping. Nat. Struct. Biol. 8, 888–892. 54. Bourne, Y., Arvai, A.S., Bernstein, S.L., Watson, M.H., Reed, S.I., Endicott, J.A., Noble, M.E., Johnson, L.N., and Tainer, J.A. (1995). Crystal structure of the cell cycle-regulatory protein suc1 reveals a -hinge conformational switch. Proc. Natl. Acad. Sci. USA 92, 10232–10236. 55. Schymkowitz, J.W.H., Rousseau, F., Irvine, L.R., and Itzhaki, L.S. (2000). The folding pathway of the cell-cycle regulatory protein p13suc1: clues for the mechanism of domain swapping. Structure 8, 89–100. 56. Rousseau, F., Schymkowitz, J.W.H., Wilkinson, H.R., and Itzhaki, L.S. (2002). The structure of the transition state for folding of domain-swapped dimeric p13suc1. Structure 10, 649–657. 57. Alonso, D.O., Alm, E., and Daggett, V. (2000). Characterization of the unfolding pathway of the cell-cycle protein p13suc1 by molecular dynamics simulations: implications for domain swapping. Structure 8, 101–110. 58. Ladurner, A.G., and Fersht, A.R. (1999). Upper limit of the time scale for diffusion and chain collapse in chymotrypsin inhibitor 2. Nat. Struct. Biol. 6, 28–31. 59. Viguera, A.R., and Serrano, L. (1997). Loop length, intramolecular diffusion and protein folding. Nat. Struct. Biol. 4, 939–946. 60. Garcia, P., Serrano, L., Rico, M., and Bruix, M. (2002). An NMR view of the folding process of a CheY mutant at the residue level. Structure 10, 1173–1185. 61. Sanz, J.M., Johnson, C.M., and Fersht, A.R. (1994). The A-state of barnase. Biochemistry 33, 11189–11199. 62. Todd, M.J., Lorimer, G.H., and Thirumalai, D. (1996). Chaperonin-facilitated protein folding: optimization of rate and yield by an iterative annealing mechanism. Proc. Natl. Acad. Sci. USA 93, 4030–4035. 63. Rutherford, S.L., and Lindquist, S. (1998). Hsp90 as a capacitor for morphological evolution. Nature 396, 336–342. 64. Queitsch, C., Sangster, T.A., and Lindquist, S. (2002). Hsp90 as a capacitor of phenotypic variation. Nature 417, 618–624. 65. Carrio, M.M., and Villaverde, A. (2001). Protein aggregation as bacterial inclusion bodies is reversible. FEBS Lett. 489, 29–33. 66. Dobson, C.M. (2001). The structural basis of protein folding and its links with human disease. Philos. Trans. R. Soc. Lond. B Biol. Sci. 356, 133–145. 67. Bousset, L., Thomson, N.H., Radford, S.E., and Melki, R. (2002). The yeast prion Ure2p retains its native ␣-helical conformation upon assembly into protein fibrils in vitro. EMBO J. 21, 2903– 2911. 68. Kweon, D.H., Chen, Y., Zhang, F., Poirier, M., Kim, C.S., and Shin, Y.K. (2002). Probing domain swapping for the neuronal SNARE complex with electron paramagnetic resonance. Biochemistry 41, 5449–5452. 69. Sauer, F.G., Pinkner, J.S., Waksman, G., and Hultgren, S.J. (2002). Chaperone priming of pilus subunits facilitates a topological transition that drives fiber formation. Cell 111, 543–551. 70. Endicott, J.A., Noble, M.E., Garman, E.F., Brown, N., Ramussen, R., Nurse, P., and Johnson, L.N. (1995). The crystal structure of p13suc1, a p34cdc2-interacting cell cycle control protein. EMBO J. 14, 1004–1014. 71. Kraulis, P.J. (1991). MOLSCRIPT—a program to produce both
72.
73.
74.
75.
detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946–950. Ramoni, R., Vincent, F., Ashcroft, A.E., Accornero, P., Grolli, S., Valencia, C., Tegoni, M., and Cambillau, C. (2002). Control of domain swapping in bovine odorant binding protein. Biochem. J. 365, 739–748. Raaijmakers, H., Vix, O., Toro, I., Golz, S., Kemper, B., and Suck, D. (1999). X-ray structure of T4 endonuclease VII: a DNA junction resolvase with a novel fold and unusual domain-swapped dimer architecture. EMBO J. 18, 1447–1458. Kortt, A.A., Dolezal, O., Power, B.E., and Hudson, P.J. (2001). Dimeric and trimeric antibodies: high avidity scFvs for cancer targeting. Biomol. Eng. 18, 95–108. Barrientos, L.G., Louis, J.M., Botos, I., Mori, T., Han, Z., O’Keefe, B.R., Boyd, M.R., Wlodawer, A., and Gronenborn, A.M. (2002). The domain-swapped dimer of cyanovirin-N is in a metastable folded state: reconciliation of X-ray and NMR structures. Structure 10, 673–686.