Homo-oligomeric protein complexes are functionally vital and highly abundant in living cells. In the present article, we review our current understanding of their geometry and evolution, including aspects of the symmetry of these complexes and their interaction interfaces. Also, we briefly discuss the pathway of their assembly in solution.
- complex symmetry
- interaction interface
- protein complex
- quaternary structure conservation
In a living cell, there exist constant interactions between proteins, both specific and non-specific. Specific protein interactions mediate a myriad of biological functions , and can be classified into permanent and transient interactions on the basis of their strength and persistence . Permanently interacting proteins form multisubunit ‘quaternary complexes’ that constitute the key protein machinery in effectively all cellular processes, including the most fundamental ones. They participate, for instance, in DNA, RNA and protein synthesis (e.g. DNA and RNA polymerases, ribosomes), protein transport (e.g. transport protein particles) and protein degradation (e.g. proteasomes).
Oligomeric quaternary complexes composed of identical protein subunits are termed ‘homo-oligomers’ (homomers). These functionally important complexes are found in many central cellular processes such as transcription (e.g. adenylyl transferase), glycolysis (e.g. phosphoglycerate mutase) and lipid metabolism (e.g. phospholipase A2). Aberrant interactions in the assembly of the homomer complex can result in diseases . In pharmacology, homomeric complexes are viewed as potential drug targets, and their therapeutic importance has recently been reviewed by Cardinale et al. . These factors warrant the need to gain an understanding of the evolutionary principles that determine the formation of homomers. In the present mini-review, we discuss our current understanding of the prevalence of homomers and the evolutionary aspects of their symmetry and interaction interfaces. We also briefly discuss their assembly pathway in solution.
Prevalence of homomers
Homomers are formed by self-interacting copies of the same protein. Accumulation of large numbers of crystal structures in the Protein Data Bank (PDB), partly due to structural genomics projects , has enabled us to study quaternary structures systematically on a large scale.
An important aspect of any study on homomers is the determination of the physiologically relevant quaternary structure from the crystal structure. In addition to experimental methods such as analytical ultracentrifugation and macromolecular MS, many computational approaches help in this process. For instance, the PDB itself uses the PISA (Protein Interfaces, Surfaces and Assemblies) method to determine the correct physiological unit in the absence of any experimental annotation . With PISA, the most likely interface is predicted on the basis of size and energetic considerations with an 80–90% success rate. A Wiki-like system that includes symmetry considerations and evolutionary conservation to help in the manual annotation of homomers quaternary structure was published by Levy . We have incorporated this semi-automatic curation in our 3DComplex database .
Why is so much effort expended on the correct annotation of homomeric quaternary structure? One of the answers is their prevalence. It has been estimated from Escherichia coli protein annotation  and from crystallographic data  that over 50% of proteins form homomers. Also, in a recent large-scale analysis of the Mycobacterium pneumoniae protein complexes, this fraction was estimated at 50% . One reason for this prevalence could be stability, conferred by the hydrophobic homomer interface, increasing to the total buried surface area in these proteins. Functional reasons might also be involved, e.g. a change in function can be coupled to a change in oligomeric state, or simply to a conformational change which allows the homomer to be co-operative, as documented in many cases (e.g. ).
It has been shown that homomers predominantly adopt either cyclic or dihedral symmetries , as illustrated in Figure 1. In a few cases, homomers adopt cubic symmetries when the number of subunits is a multiple of 12 . Theories that attempt to explain symmetry in homomers have been reviewed in detail in .
Smaller complexes are observed to be more prevalent than larger ones, and those with even numbers of subunits are also favoured over those with odd numbers [8,9]. The abundance of complexes with an even number of subunits is simply explained by their ability to adopt both dihedral and cyclic symmetries, whereas odd numbers of subunits can only be achieved with a cyclic symmetry. As a matter of fact, dihedral complexes are approx. 10-fold more abundant than cyclic complexes . This could be due to the existence of multiple evolutionary pathways for the assembly of dihedral complexes from cyclic complexes. Additionally, it could be due to the observation that self-complementary interactions, which are present in dihedral homomers, are statistically stronger and thus more likely to form by mutation than the interactions between dissimilar surfaces such as those observed in heteromers and cyclic homomers [13,14].
Conservation of homomers and large interfaces
Conventional principles of evolutionary conservation of proteins have been predominantly based on domains and tertiary structures (e.g. Pfam , SCOP , CATH ). In the last 5 or 10 years, studies and databases that consider the interactions between individual pairs of domains have been published (as implemented in, e.g., iPfam , 3did ). In order to study the function and conservation of quaternary structure, there is a need for a hierarchical classification system of entire structural states in which they perform their function. To this end, 3DComplex , a database of hierarchical classification of quaternary structures, has been developed. 3DComplex allows easy browsing and analysis of both homomeric and heteromeric protein complexes. One of the uses of the 3DComplex database is that it allows rapid analysis of evolutionarily related protein complexes. For instance, it can reveal cases where homomers are structurally similar to heteromers. Such ‘homologous heteromers’ are believed to have evolved by gene-duplication events where protein interactions remained conserved between the resulting homologues .
Varying degrees of conservation of quaternary structure are observed for homomers across species. It has been shown that proteins with sequence identity greater than 90% share close to 100% quaternary structure conservation (Figure 2), whereas among those with sequence identity of 30–40%, symmetry conservation is approx. 70% . These numbers apply to homomeric complexes. In terms of larger and more diverse protein complexes, it is worth noting that there have been in-depth studies of complexes. There have been studies on the evolution of respiratory complex I  and the bacterial flagellum .
The interacting surfaces need to be considered separately for weakly and strongly interacting proteins, as the former stays in equilibrium with its monomeric state. In a recent study, Dey et al.  investigated weak homodimers, their interaction interface conservation and their evolution. They observed that, in 93% of cases, the residues in the interacting interfaces are more conserved than those on the rest of the protein surface.
In dihedral homomers, there are always at least two types of interfaces, and one is usually larger than the other. It has been shown that the larger interface is generally conserved in evolution, whereas the smaller is more easily made or broken [11,24]. We discuss below how this may relate to the assembly of homomers in solution.
The individual subunits of homomers need to fold and assemble after their synthesis by the ribosome. Large complexes such as the ribosome have a predefined assembly pathway that involves many accessory factors, and chaperones assist in the assembly of many other complexes. For small homomeric complexes, it is likely that the subunits fold and assemble spontaneously in the cell as in solution.
By analogy to the folding problem, assembly in solution would be slower if all possible assembly pathways were equally likely for a given homomer. Instead, we have shown that the main assembly pathway of dihedral complexes takes place via an intermediate that corresponds to the large interface in the final complex, as shown in Figure 3 .
The current state of our knowledge about quaternary complexes has been mainly from three-dimensional structural data, which limits our vision to static glimpses of functional complexes. To visualize a dynamic picture, we need a framework that considers the energetics of interactions. For this, Villar et al.  have developed the model to simulate the assembly dynamics of homomeric complexes. This model was able to reproduce the hierarchical assembly observed experimentally for dihedral complexes .
Conclusion and discussion
In the last decade, we have made significant progress in our understanding of homomeric complexes. We now have a good understanding of the structural geometry and evolutionary aspects of homomeric complexes. However, several other questions still remain unanswered. The dynamic properties of homomers upon ligand binding and their interactions with other proteins are poorly understood.
Whereas most of our present knowledge comes from globular proteins, little is known about the homomerization of membrane proteins such as G-protein-coupled receptors and ion channels. This could be due to under-representation of membrane proteins in the complexes that have been studied. Development of new experimental methods addressing the above-mentioned limitations will accelerate our understanding and shed light on this area.
E.D.L. is supported by a Human Frontier Science Program Postdoctoral Fellowship. A.J.V. is supported by a LMB-Cambridge Scholarship and St John's College, Cambridge.
Experimental Approaches to Protein–Protein Interactions: A Biochemical Society Focused Meeting held at University of Sheffield, Sheffield, U.K., 11–12 January 2010. Organized and Edited by Michael Sutcliffe (Manchester, U.K.) and Mike Williamson (Sheffield, U.K.).
Abbreviations: PISA, Protein Interfaces, Surfaces and Assemblies
- © The Authors Journal compilation © 2010 Biochemical Society