Biochemical Society Transactions

7th International Conference on Protein Stabilization 2007

Enhancement of oligomeric stability by covalent linkage and its application to the human p53tet domain: thermodynamics and biological implications

G.M.K. Poon


The formation of oligomeric proteins proceeds at a major cost of reducing the translational and rotational entropy for their subunits in order to form the stabilizing interactions found in the oligomeric state. Unlike site-directed mutations, covalent linkage of subunits represents a generically applicable strategy for enhancing oligomeric stability by reducing the entropic driving force for dissociation. Although this can be realized by introducing de novo disulfide cross-links between subunits, issues with irreversible aggregation limit the utility of this approach. In contrast, tandem linkage of subunits in a single polypeptide chain offers a universal method of pre-paying the entropic cost of oligomer formation. In the present paper, thermodynamic, structural and experimental aspects of designing and characterizing tandem-linked oligomers are discussed with reference to engineering a stabilized tetramer of the oligomerization domain of the human p53 tumour-suppressor protein by tandem dimerization.

  • covalent linkage
  • oligomeric stability
  • oligomerization domain
  • p53
  • thermodynamics
  • tumour suppressor


In protein folding, the loss of configurational entropy is the primary thermodynamic cost which must be recovered from favourable processes such as hydrogen-bonding, electrostatic interactions, dispersion interactions and the hydrophobic effect. For oligomeric proteins, the formation of quaternary structure represents an additional entropic penalty wherein multiple translational and rotational degrees of freedom are lost in immobilizing the subunits within a single entity [1]. The most common strategy for engineering conformationally stabilized proteins is to introduce point mutations that increase favourable protein–solvent and/or protein–protein interactions in the folded state. However, the highly context-specific nature of this strategy means that, even with the ever-increasing predictive power of computational approaches, extensive screening of many, often combinatorial, mutations are required to ‘hit the jackpot’.

For oligomeric proteins, an altogether distinct, and generically applicable, alternative to point mutations is to address the dissociated (and often unfolded) state. In other words, since the tendency towards association is fundamentally dictated only by the difference in free energy of the oligomer relative to its constituent subunits, oligomerization can be favoured by destabilizing the dissociated state (Figure 1). This can be accomplished by covalently linking some or all of the subunits such that the entropy of dilution resulting from dissociation is minimized. Two methods of tethering subunits are available. One is the use of cysteine mutants which strategically cross-link adjacent subunits with disulfide bridges, although irreversible aggregation due to non-specific cross-linking is a common side effect. More preferably, subunits can be linearly joined as tandem repeats in a single polypeptide chain. The resultant configurational constraints on the dissociated but linked subunits significantly discount the entropic advantage of dissociation and drive the equilibrium towards oligomerization. Equivalently, this concept can be understood in terms of a large increase in the ‘local’ concentration of subunits within the excluded volume of the tandem-linked polymer [2,3]. In aqueous solution at 1 M standard state, the entropic cost for formation of a bimolecular complex has been estimated at up to ∼35 cal·mol−1·K−1 (1 cal=4.184 J), or over 10 kcal/mol in free energy at 25°C [4]. By ‘pre-paying’ part or all of this cost, it is clear that covalent linkage represents a much more effective method of stabilizing oligomeric proteins than site-directed mutagenesis (Figure 1).

Figure 1 Comparison of stability enhancement by site-directed mutagenesis and covalent linkage

The arrows represent the free energy change ΔG associated with the monomer/oligomer (M/O) equilibrium; their lengths represent the magnitudes of ΔG. M* denotes an ‘activated’ monomer that has been destabilized (in terms of its molar translational and rotational entropies relative to M) by covalent linkage.

To date, several examples of designer tandem-linked proteins have been reported [510]; in the case of transcription factors, the increase in oligomeric stability is leveraged to increase DNA-binding affinity [5,8,10,11]. Oligomeric enzymes may also derive a catalytic benefit to the extent that active conformations of their binding sites are coupled with oligomerization. Interestingly, evolution also exploits this stabilization strategy by gene duplication. Ca2+-binding proteins, such as calmodulin, consist of two tandem-linked EF-hand domains, and each domain in turn consists of two tandem-linked EF hands. In isolation, these structural components exhibit progressively destabilized conformational stabilities in the apo form as well as impaired co-operative binding of Ca2+ compared with their tandem-linked counterparts [12,13]. In the present paper, I address theoretical and practical aspects of engineering and characterizing covalently tandem-linked oligomeric proteins and illustrate these principles with the tetramerization domain of the human p53 tumour-suppressor protein [14]. Since oligomerization is intimately related to p53's biological properties, stabilization of this domain represents an important avenue for further understanding of this protein in normal cell-cycle control, as well as engineering therapeutic rescue constructs for cells in which p53 tumour-suppressive functions have been compromised.

Considerations in constructing covalently linked tandems

At a generic level, when designing a covalently linked tandem, three important questions should be considered.

What should be the valency of the tandem?

In other words, how many copies of the subunit should be fused into a single polypeptide chain? Although reported examples of tandem covalent linkage involve exclusively dimeric proteins, any number of the subunit's gene can be concatenated in a cassette format by using standing molecular biology techniques. Increasing the valency of a tandem leads to a corresponding reduction in the molecularity of the tethered oligomer, with the thermodynamic consequence that oligomerization becomes less sensitive to concentration. In the extreme case, if all subunits in an oligomer are tethered in a single tandem, then oligomerization becomes a unimolecular folding transition and independent of concentration. To illustrate the importance of molecularity, consider a simple two-state dissociation equilibrium in which the native oligomer Nn dissociates into its n constituent denatured subunits D in a coupled fashion: Embedded Image(1)

It should be noted that eqn (1) does not preclude the existence of lower-order kinetic intermediates which are not observable at equilibrium, and has been shown to describe the coupled equilibrium dissociation and unfolding of the heptameric co-chaperonin protein 10 [15]. Expressed in terms of the fraction dissociated/unfolded f, the dissociation/unfolding constant K {K can also be expressed as KD=K1/(n−1) (n>1), the concentration at which [Nn]=[D] [17,18]} depends on total subunit concentration pt=n[Nn]+[D] [16]: Embedded Image(2)

It is instructive to consider eqn (2) in some detail, specifically in terms of the dependence of f on K for different stoichiometries (Figure 2). For a monomeric protein, n=1, and we have the concentration-independent solution f=K/(K+1), which recapitulates the familiar situation of f=½ when K=1. Oligomers, in contrast, are progressively destabilized with increasing molecularity (Figure 2A) and the extent of destabilization is amplified as concentration decreases in the submolar range (Figure 2B). These effects are attributable to dilution entropy and the choice of the molar standard state in defining K (and ΔG°) [1,4]. Importantly, this behaviour implies that a reduction in the overall molecularity in oligomeric state can confer substantial stabilization in terms of f even under conditions in which the (standard) free energy of dissociation/unfolding for the tethered oligomer is lower than that of the untethered one.

Figure 2 Fractional dissociation f as a function of oligomeric stability in terms of K and ΔG for oligomers undergoing coupled dissociation/unfolding according to eqn (1)

Curves are generated by solving eqn (2) numerically for f by the Newton's method. Numbers indicate oligomeric molecularities. K (for dissociation/unfolding) is given as a dimensionless quantity that would be produced from ln K=−ΔG°/RT or when composed with activities. (A) Dispersion of f for dimeric to hexameric oligomers, each at 10 μM total subunit concentration, pt. This dispersive behaviour may give rise to an apparent stabilization afforded by a reduction in molecularity even if ΔG° per monomer is decreased by covalent linkage. Dispersion is minimized, but not abolished, at pt=1 M (the molar standard state); see eqn (2). (B) Comparison of f for a dimer and tetramer at pt=1 mM (dashed line) with 10 μM (solid line), to illustrate the increase in dispersion of f with decreasing total concentration.

In practice, a tandem of valency higher than, or not divisible by, the molecularity of the oligomer will result in cross-linked oligomers, and is generally not desirable. Thus, for a tetrameric oligomer, a tandem dimer or tetramer is appropriate, whereas a trimeric or pentameric oligomer requires a tandem of the same respective valency. Where available, structural, thermodynamic and kinetic data are extremely useful, since knowledge of structural symmetries and the mechanism of oligomerization will help guide the selection of an appropriate valency for a given oligomeric protein.

What should be the length of the linker joining each tandem repeat?

This is not a straightforward decision. A shorter linker will reduce the sampling space available to each subunit in the tandem and therefore further destabilize the dissociated state in favour of oligomeric formation. A lower limit on this length exists, however: if the linker is close to or shorter in length than the distance between the C-terminus of one subunit and the N-terminus of the next in the untethered oligomer, this will disrupt the formation of intra- and inter-subunit interactions that normally are present in the oligomer, or at worst, prevent the subunits from docking with each other in the required orientation. Clearly, structural data of the oligomer is essential for choosing the optimal length of the linker.

What should be the sequence of the linker?

The ideal linker would be ‘inert’ yet rigid so that it constrains the tandem repeats at the position and orientation found in the oligomeric state [19]. While this is often attempted in synthetic multivalent receptor ligands [20], rigid (and hence highly structured) designer peptide linkers that would not interact and therefore not perturb the core structure of the oligomer are extremely challenging, if not impossible, to construct. It is more practical, therefore, to choose sequences that are suitable as flexible surface loops. In this respect, they should consist of hydrophilic residues (such as a glycine or serine residue) that are not likely to interact with structural surface charges on the conjoined subunits. They should also not contain any sequence that may be proteolytic cleavage sites. If the oligomer being considered is a domain, unstructured segments flanking the domain may also be appropriate, if they satisfy the above properties.

Evaluation of covalent tandem-linked oligomers

Once a covalently linked tandem has been cloned, overexpressed and purified, a number of checks should be made to ensure that the tethered oligomer maintains the same essential properties as the untethered one. For oligomeric enzymes, functional assays should be made; kinetic parameters may differ from the untethered oligomer depending on the nature and extent to which oligomerization is coupled with catalytic activity [4]. In addition, the molecularity of the tethered oligomer must be checked (by size-exclusion chromatography, analytical ultracentrifugation or cross-linking experiments) to ensure that unintended cross-linking by tethered subunits between oligomers has not occurred. At a higher resolution, the secondary and tertiary structures of the tethered and untethered oligomer can be compared by spectroscopic techniques such as CD or NMR. Minor differences can usually be expected, since some interaction by the linker and perturbation of the core oligomer structure is generally unavoidable.

The increased stability of the tethered oligomer can be quantified, using spectroscopic (absorption, CD, fluorescence and NMR) or calorimetric techniques, by its enhanced resistance to heat, chemical denaturation or hydrostatic pressure. Among these, DSC (differential scanning calorimetry) has the tremendous advantage of affording the complete thermodynamics [the transition temperature, T°; enthalpy change at T°, ΔH(T°); and change in heat capacity, ΔCp] of the dissociation/unfolding transition, particularly with respect to ΔCp, which cannot be reliably obtained from spectroscopic data. DSC and analytical ultracentrifugation are also helpful in diagnosing potential equilibrium intermediates that result from stabilization of previously kinetic intermediates. Because of issues associated with changes in molecularity (as seen for example in eqn 2 and Figure 2), as well as statistical complications concerning the unfolding of modular tandem repeats [21], comparing the dissociation/unfolding thermodynamics of an oligomer and its tandem-linked counterpart on the basis of ΔG or K may be obscured, and f (as a function of temperature at a given total concentration, for example) is therefore a more practical and useful comparator in this regard. This highlights the need for fitting the experimental data to a mechanistically appropriate model (taking into account emergent equilibrium intermediates, for instance) [22], as unphysical thermodynamic parameters will yield incorrect calculations of the extent of oligomerization (i.e. 1−f) at concentrations below experimental detection. Since dissociation/unfolding transitions of oligomers are, unlike unimolecular transitions, dependent on concentration, consistency of a chosen model should be checked by experiments at different protein concentrations. Accordingly, illustrative experiments comparing the stabilities of tethered and untethered oligomers should be performed at equal repeat concentrations.

Tandem dimerization of the tetramerization domain of the human p53 tumour-suppressor protein

We have applied the forgoing principles of designing and characterizing tandem-linked oligomers to the tetramerization domain (p53tet residues 325–355) of the human tumour suppressor p53, a critical transcription factor implicated in cell-cycle control, programmed cell death and cellular differentiation [23]. It is a modular protein wherein the compact p53tet domain (residues 325–355), near the C-terminus, mediates the oligomeric state of the protein. Since the regulatory functions of p53 are directly linked to its quaternary structure [24], the oligomeric stability of the p53tet domain dictates the functional status of this protein. The wild-type p53tet tetramer undergoes coupled dissociation and unfolding as described by eqn (2) with K≈10−18 M−3 (or KD≈10−6 M) [18,25,26] under physiological conditions. This indicates that, at low copy number, the intracellular pool of p53 is predominantly monomeric and inactive. Moreover, p53 mutations in cancer cells typically occur in regions located in the DNA-binding domain and result in p53 mutants which self-associate or form heterotropic complexes with wild-type p53 pools [27,28]. These p53 heterotetramers exhibit various transdominant effects, such as longer cellular half-lives and altered p53 functions relative to their wild-type counterpart [29], and most significantly, fail to suppress tumour growth in vivo [30].

One strategy aimed at alleviating transdominant inhibition is to replace non-functional p53 mutants with exogenous, functional p53 constructs. However, simple supplementation of wild-type p53 (without correcting the existent mutations) cannot significantly restore tumour-suppressor activity, since they too are subject to heterotetramer formation. Thus an effective rescue p53 oligomer must be more stable than wild-type p53 and resist exchange with mutant p53 monomers. Since kinetic studies have suggested that p53 tetramerization occurs via a structured dimeric intermediate (a ‘primary dimer’) [31], covalent linkage of two p53tet domains in a tandem dimer (Figure 3) may effectively stabilize this intermediate and, by extension, the full oligomer. Such a construct, p53tetTD, was designed and produced by recombinant methods (see [14] for a complete report): it consists of two tandems of p53-(310–360), giving rise to an inert, flexible linker 22 residues in length (structural data have shown that wild-type residues adjacent to residues 325–355 are disordered [32]). Two equivalents of p53tetTD associate to form a four-domain dimer that is topologically equivalent to the native p53tet tetramer, as judged by size-exclusion chromatography, analytical ultracentrifugation, chemical cross-linking, and NMR spectroscopy. DSC and far-UV CD spectroscopy of p53tetTD at different concentrations demonstrate that tandem dimerization indeed stabilizes the primary dimer (i.e. a folded p53tetTD monomer) at equilibrium, which bears all of the secondary structure content found in the final oligomer. The identification of a (tethered) stabilized primary dimer at equilibrium leads to a three-state model for the analysis of the DSC data and the calculation of f: Embedded Image(3) where N2, I and U represent the four-domain p53tetTD dimer, the folded two-domain p53tetTD monomer and the unfolded p53tetTD monomer respectively. Based on thermodynamic parameters obtained by DSC analysis under physiological conditions, the p53tetTD dimer remains 50% oligomerized at a total peptide concentration of 0.1 nM, compared with approx. 0.1 μM for the wild-type p53tet tetramer. Moreover, the p53tetTD dimer is significantly more resistant to subunit exchange with monomeric p53tet than the wild-type p53 tetramer, suggesting that a full-length construct harbouring p53tetTD as its oligomerization domain would exhibit similar resistance to transdominant inhibition. Even with the already significant enhancement in oligomeric stability found in p53tetTD, it is expected that further refinement, particularly in terms of linker length, will result in additional gains in stability. These highly stabilized p53tet-based oligomers may serve as useful scaffolds for building full-length rescue p53 oligomers [33,34] as well as multivalent presentation of functional peptides [35,36].

Figure 3 The tetramerization domain of the human p53 tumour (p53tet) suppressor and its covalently tandem-linked dimer, p53tetTD

(A) The wild-type p53tet (residues 325–355) tetramer exhibits a ‘dimer of dimers’ architecture [32,3740]; the subunits in the two ‘primary dimers’ are coloured red and green. (B) p53tetTD is a linear double tandem of p53-(310–360) separated by a GT dipeptide, corresponding to two copies of the p53tet domain (red and green) connected by a 22-residue flexible linker (blue). Two p53tetTD peptides associate to give a four-domain structure equivalent to the wild-type p53tet tetramer. The predicted structure shown here is generated by homology modelling based on the wild-type p53tet tetramer (1OLG [37,38] using SWISS-MODEL) and has been verified biochemically and spectroscopically including by NMR [14].


As a member of his laboratory, I am indebted to Dr Jean Gariépy of the Ontario Cancer Institute for his support as principal investigator. This work was supported financially by the Canadian Breast Cancer Research Alliance in association with the Canadian Cancer Society.


  • 7th International Conference on Protein Stabilization 2007: Independent Meeting held at the University of Exeter, Exeter, U.K., 11–14 April 2007. Organized and Edited by J. Littlechild (Exeter, U.K.).

Abbreviations: DSC, differential scanning calorimetry; p53tet, human p53 tetramerization domain (residues 325–355); T°, transition temperature; ΔCp, transition heat capacity; ΔH(T°), transition enthalpy


View Abstract