In our study of hemoglobin structure in the MWC model, we developed the idea that there were two forms of hemoglobin in solution, the taut and relaxed form, which are pre-exisiting and interconvertible even in the absence of dioxygen. Oxygen was presumed to bind preferentially to the relaxed form. In the KNF model we saw that ligand binding can induce conformational changes in adjacent subunits, promoting cooperative binding of ligand. In general these two models distill down to combinations of two simpler models. The first might be called the conformational selection in which ligand binds tightly to a preexisting conformations in a "lock and key manner" without inducing subsequent macromolecular conformational change. Alternatively, the ligand might bind loosely and then alter the macromolecular conformation to produce tighter binding, an example of an induced fit model. For the binding of dioxygen to hemoglobin, thermodynamic cycles could be drawn showing either binding of ligand and subsequent conformational changes in protein structure or conformational changes in protein structure proceeding binding. Is there additional evidence to support the conformational selection model of binding of ligand to a protein that can, in the absence of ligand, exist in two conformations? The answer is yes.
Antibodies are immune system protein molecules than can bind "foreign" molecules and target them for biological neutralization. Many crystal structures have been determined of antibodies in the presence or absence of a "foreign" ligand molecule. In these cases, the conformation of the bound antibody is different from that of the free. Either an induced fit model for ligand binding or a lock and key model of binding of ligand to one of two different pre-existing conformations of the antibody could account for this observation. These different mechanisms could be differentiated experimentally by stop-flow kinetic technique since both display slow and fast phases that are affected differently by ligand concentration. Theoretically, in the induced fit model, only one ligand type could bind to the antibody which would undergo a conformational rearrangement to produce tighter binding. However, in the two preexisting conformational models, a different structural ligand might bind to each of the two main antibody conformations. James et al. have recently shown through stop flow kinetics techniques (to investigate binding) and x-ray crystallography (to investigate final structures) that one antibody molecule can, through existing in two different preexisting conformations, bind two different ligands (antigens). One antibody conformations binds small aromatic molecules with low affinity (including the small molecule 2,4-dinitrophenol, the immunizing molecule or hapten) and then rearranges to produce a high affinity binding complex in which the DNP is bound in a narrow cavity (reducing the effective off rate of the bound ligand. A second antibody conformation binds a protein ligand over a broad, flat binding site of the antibody molecule.
Lange (2008) et al, using a NMR technique, residual dipolar coupling, that allows sampling of structures in the microsecond time scale, have shown that the solution structure of ubiquitin (which we modeled in our first lab), in the absence of ligand, exists in an ensemble of conformational states. More importantly, these different conformational states are identical to those found in the 46 crystal structure of ligand complexed to various protein ligands, strongly supporting the concept of conformational selection. In all likelihood, a combination of both induced fit and conformational selection probably occurs within a 3D energy landscape in which an initial binding encounter by either a lock and key fit to the "optimal fit" conformer or to a higher energy conformer in which the bound state relaxes to a lower energy through the induction of shape changes in the binding protein.
Figure: Conformational Selection vs Induced Fit Binding (after Boehr and Wright, Science 320, 1429 (2008)
An interesting experimental model to distinguish conformational selection versus induced ligand binding was offered by Rea et al. They studied rabbit ileal bile acid binding protein (I-BABP). The wild type protein has a helix-turn-helix motif at its N terminus. They produced a mutant (Δa-I-BABP) that replaced this motif with a Gly-Gly-Ser-Gly linker, causing the protein to unfold. Next they conducted binding and folding studies on addition of taurochenodeoxycholate (TCDC) using stopped-flow fluorescence to measure the binding behavior. They wished to distinguish between two distinct mechanisms – folding before binding (or conformational selection) and binding before folding (or induced-fit model). The data support a two phase model. One phase did not depend on ligand and one did, suggesting binding followed by a conformational change).
P* in the conformational selection models represents a high affinity, pre-existing conformation of the protein. In this model, high ligand shifts the equilibrium to the right.
One way to differentiate these models is to look at the dependency of the different kinetic phases on ligand. In the conformation selection model, the slow step is the formation of the high affinity form of the protein, P*. The first slow step has a nonlinear dependence in L while the fast second step has a linear dependence. The data did not fit this model well.
In the induced fit, the ligand binds to a low affinity and perhaps unfolded form of the protein, which subsequently collapses to the bound form in a slow step.
Both ligand dependent and independent phases are evident in the equation for the slow step for the induced fit mechanism. At high ligand concentration (when L >> k-1/k1) , the slow step in the induced fit would be independent of ligand (kslow = k-2 + k2). The authors state the data is consistent with a variant of induced fit called the "fly casting model". In this model the protein first encounters ligand and forms a hydrophobic collapse intermediate (PL) in a fast step characterized by a linear dependence on ligand concentration. Then the intermediate slowly interconverts into a wild type like complex through conformational re-arrangement. Wild-type protein binds the ligand 1000x as quickly, suggesting entropic barrier to binding of the ligand to the unfolded state and rearrangement of the protein thereafter.
Junker et al used atomic force microscopy (AFM) to observe the effects of ligand binding on the folding/unfolding fluctuations of a single molecule of calmodulin (CaM), a calcium-binding protein that binds amphiphilic helicals peptides which leads to a large conformation change in the protein. To do this, they sandwiched a single CaM molecule between filamins that serve as attachment points for the AFM tip and a surface. A slow pulling force was then applied to the molecule, and the length gain was measured as the protein unfolded. The rapid fluctuations between folded and unfolded states were quantified and used to derive a complete energy landscape for the folding of CaM. They conducted these experiments in the presence of two ligands, Ca2+ and mastoparan (Mas), a wasp venom peptide. They found that Mas does not affect the folding rate of CaM, although it does stabilize the already folded form. This suggests that Mas does not bind to the transition state or the unfolded protein, but rather selects a particular conformation from an ensemble of possible choices. Ca2+ however, increases the folding rate, which suggests that it stabilizes both the transition state and the folded state. AFM offers a considerable degree of precision in drawing energy landscapes of protein folding and unfolding, and it has several applications that are yet to be explored.
Binding to Intrinsically Disorder Protein and MORFs
As described above, binding of a protein to a ligand (including another protein) could occur by a lock and key mechanism, possibly through a conformational selection process, or through an induced fit when an initial binding event is followed by a conformation rearrangement to form a more tightly bound complex. But how does binding to completely intrinsically disorder protein (which has been documented) occur? These cases are quite removed from those envisioned in simple induced fit mechanisms. To accommodate binding to IDPs, a new idea has emerged, that of Molecular Recognition Features (MoRFs).
MoRFs are typically contiguous but disordered sections of a protein that first encounter a binding partner (a protein for example). Using protein complexes in the Protein Data Bank, Mohan et al conducted a structural study of MoRFs by selecting short regions (less than 70 amino acids) from mostly disordered proteins that were bound to proteins of greater than 100 amino acids. They chose a sequence size of 70 amino acids and smaller since they would be most likely to display conformational flexibility before binding to a target. 2512 proteins fit their criteria. For comparison, they created a similar database of ordered monomeric proteins. The analysis showed that after they encounter a binding surface on another protein, the MoRF would adopt or "morp" into several types of new conformations, including alpha-helices (a-MoRFs), beta-strands (b-MoRFs), irregular strands (i-MoRFs) and combined secondary structure (complex-MoRFs), as shown in the figure below.
(A) α-MoRF, Proteinase Inhibitor IA3, bound to Proteinase A (PDB entry 1DP5). (B) A β-MoRF, viral protein pVIc, bound to Human Adenovirus 2 Proteinase (PDB entry 1AVP). (C) An ι-MoRF, Amphiphysin, bound to α-adaptin C (PDB entry 1KY7). (D) A complex-MoRF, β-amyloid precursor protein (βAPP), bound to the PTB domain of the neuron specific protein X11 (PDB entry 1X11). Partner interfaces (gray surface) are also indicated. Vacic, V. et al. Journal of Proteome Research 6, 2351 (2007). Permission from Copyright Clearance Center's Rightslink /American Chemical Society
Vacic et al have further characterized the binding surfaces between MoRFs and their binding partners using structural data from PDB files. Interfaces were studied by determining the differences in accessible surface area between MoRFs and their binding partners, and the protein in unbound states. These were compared to ordered protein complexes, including homodimers and antibody-protein antigen interactions that were not characterized by disordered interactions. Their findings are summarized below.
- MoRF interfaces have more hydrophobic groups and fewer polar groups compared to the surface of monomers. This is true even as the overall amino acid composition of intrinsically disordered protein shows them to be enriched in polar amino acids, which leads them to adopt a variety of unfixed solution conformations.
- a-MoRFs have few prolines, which is expected as prolines are helix breakers.
- Methionine is enriched in both MoRFs and in their binding partner interface. Methionine is unbranched, flexibile, and contains sulfur, which is large and polarizable, making it an ideal side chain to be involved in London forces in a hydrophobic environment.
- Even though MoRFs have few residues, their binding interfaces were of similar or larger size than other protein binding interfaces, a result which also applies to IDPs as a whole. MoRFs interfaces also have a larger solvent-exposed surface area, similar to IDPs. This is consistent with the notion MoRFs are disordered before binding and that a defined structure is not possible with little buried surface area.
- As MoRFs have significant nonpolar character within a IDP that is highly enriched in polar amino acids, MoRFs should be highly predictable by search algorithms.
Binding, Intracellular Granules and Droplets
We've studied different types of protein aggregation, including aggregation of the native state (to form dimers, trimers, multimers, filaments) or alternate conformations (such as in prion protein aggregation and formation of inclusion bodies of misfolded proteins). It turns out that messenger RNA (which ultimately get translated into proteins) can aggregate with RNA binding proteins (RBPs) to form intracellular granules. These kinds of aggregates are commonly found in cell and have recently been recognized as non-membrane bound organelles. We've seen analogous particles, lipid droplets, which contain TAGs and cholesterol esters surrounded by a phospholipid monolayer with adsorbed protein, also promoted to the state of an organelle (in contrast to the recent demotion of the planet Pluto to a dwarf planet or plutoid). The lipid droplet however are surrounded by a "membrane", in this case a monolayer.
How do these and other granules form. A quick review of the Cell Tutorial (scroll to bottom) shows granule formation can be caused by a classic "phase transition", not unlike gaseous water can self associate through hydrogen bonds to form liquid drops which can freeze with the formation of more hydrogen bonds to form solids. Soluble biomolecules in cells can reversibly aggregate through the summation of multiple weak IMFs to form storage granules. This balance might be perturbed if storage granules aggregate further in a potentially irreversible process with health consequences. We've seen examples of the latter when neurodegenerative diseases like Alzheimer's and Mad Cow Disease. Lets delve into new insights into the processes involved in droplet formation.
Imagine small amounts of a sparing soluble oil added to an aqueous solution. Initially it is in solution, but at a higher concentration, London dispersion forces and the “hydrophobic effect” would drive the oil out of solution into liquid drops. This phase separation could also be called liquid-liquid demixing as two liquids (solubilized oil in water and separated oil drops) separate. This process has been shown to produce many types of non-membrane bound droplets (not to be confused with membrane bound vesicles) in the cell.
This phenomenon has also been seen with intrinsically disordered proteins. These are characterized by amorphous structures with repeated, often positively charged amino acids. Under the right condition, these can aggregate and “precipitate” from the solution. What is the nature of the precipitate? It might have properties more like distinct liquid droplets so this process could be called liquid-liquid demixing.
Properties of demixed drops would include reduced rates of diffusion of material into an out of the drop, coupled movements of materials in the drop, and probable weak hydrophobic-dependent aggregation making drops sensitive to agents like detergents. Liquid-like diffusion inside the drop is observed as evident by rapid recovery of fluorescence from partially photobleached internal components of the drop.
As with the formation of a crystalline solid from a liquid solution, the process must be seeded. For intrinsically disordered proteins, this process can be “catalyzed” by poly-(ADP-ribose), a nucleic acid-like polyanion. The negative charges would counter the positive charges in the disordered protein domain, which without neutralization, would interfere with protein/protein contacts necessary for aggregation/droplet formation and demixing. Aggregation in these cases may arise from hydrophobic interactions (even though hydrophobic side chains are underrepresented in the disordered domains).
Solubility of proteins in cells is a fascinating topic in itself. It was recently discovered that the high concentration of ATP (5 mM) in the cell actually helps to solubilize proteins. ATP is considered a hydrotrope. It’s a small molecule with a vary distinct polar part (polyphosphate and ribose) and a more nonpolar part (the adenosine ring). Hence it acts sort of like a mini-detergent (an amphiphle) but it doesn’t form micelles. It does help stabilize more nonpolar parts of proteins in solution and has been shown to inhibit aggregate formation and also disaggregate some aggregates. The figure below shows a nonprotonated form of energy-minimized ATP with its dipole moment shown as an arrow from + to - end (obtained using Spartan). The dipole moment would only be larger if the ATP was deprotonated and had negative charges.
Biochemists also use the term gel (examples include polyacrylamide gel or fibrin blood clots which are chemically cross-linked) and a gel in the gel to liquid-crystalline phase transitions in lipid bilayers, held together by weak noncovalent interactions), when they wish to describe a structure that is neither clearly solid nor liquid. A structure like the cytoskeleton or the actin-myosin network would be examples of the latter.
Noncovalent gels would be characterized by regulatable dissociation of subunits and hence short half-lifes. A gel (either covalent or noncovalent) with a high-water content could be called a hydrogel which would contain hydrophilic components. An example would be RNA-protein containing particles
Many of the granules contain RNA and proteins and are called ribonucleoprotein bodies (RNPs) or RNA granules. Specific examples of these include cytoplasmic processing bodies, neuronal and germ granules, as well as nuclear Cajal bodies, nucleoli and nuclear dots/bodies). Some granules just contain proteins , including inclusion bodies with misfolded and aggregated proteins and those with active protein involved in biosynthesis, including the purinosome (for purine biosynthesis) and cellusomes (for cellulose degradation).
Another feature found in some neurodegenerative disease is a trinucleotide repeat. In Fragile X syndrome, there 230-4000 repeats of the CGG codon in the noncoding parts of the genome, compared to less than 50 in the normal gene. In Huntington’s disease, the repeat CAG is found in the protein coding part of the affected gene. The translated protein has a string of glutamines which probably causes protein aggregation. Specific proteins may also bind to the string of CAGs.
If the trinucleotide expansion is in intronic DNA, deleterious effects are not associated with translated proteins but with the transcribed RNA in the nucleus. The intronic repeats would be spliced out of the primary RNA transcript. A CTG DNA repeat would produce a poly CUG containing RNAs (found in myotonic dystrophy), which could aggregate through non-perfect base pairing. In vitro experiment show that small complexes are soluble, but as the size increases, a liquid-liquid demixing phase separation (or alternatively a liquid-gel transition) can occur, forming spherical drop RNA particles. This would explain the observation that pathologies occur above a certain repeat length. If misfolded proteins are also present, these particles might combine to form larger gels.
In the control experiment, when the repeats were scrambled, demixing and spherical particle formation was not observed. In an experiment similar to the addition of 1,6-hexanediol to intrinsically disorderd proteins, if small antisense trinucleotide repeats, such as (CTG)8, which could interfere with the weak H bonds between G and C in the aggregates, were added, the size of RNA drops (foci) were reduced. In vivo experiments showed characteristic drop-like structures but only if the repeats were of sufficient size.
Researchers found that in vitro, RNA drop formation was inhibited by monovalent cations. In the presence of 0.1 M ammonium acetate, which permeates cells without affecting pH, 47× CAG RNA droplets in vitro disappeared.
Aggregation of mRNA might be one way to regulate its translation and hence indirectly regulate gene activity. There are advantages to regulating the translation of a protein from mRNA, especially if the "activity" of the mRNA could be dynamically regulated. This would be useful if new protein synthesis was immediately required. Hence one way to regulate mRNA activity (other than degradation) is through reversible aggregation.
Protein drops and granules
The cytoskeletal proteins actin and tubulin (heterodimer of alpha and beta chains) can exist in soluble (by analogy to water gaseous) states or in condensed filamentous state (actin filaments and microtubules respectively). GTP hydrolysis is required for tubulin formation. Actin binds ATP which is necessary for filament formation but ATP cleavage is required for depolymerization. Hence nucleotide binding/hydrolysis regulates the filament equilibrium which differentiates from simple phases changes such as in water.
Since only certain proteins form granules, they must have similar structural features that facilitate reversible binding interactions. There appear to be multiple sites with these protein that individually form weak binding interactions, but collectively through multivalent (multiple) binding interactions allow robust but not irreversible granule formation. Here are some characteristics of proteins found in granules:
- the protein NCK has 3 repeated domains (SH3) the bind to proline-rich motifs (PRMs) in the protein NWASP. These proteins are involved in actin polymerizaiton. In high concentration they precipitate from solution and coalesce to form larger droplets;
- repeating interaction domains are widely found especially among RNA binding proteins;
- some proteins contain Phe-Gly (FG) repeats separated by hydrophilic amino acids in portions of the protein that are intrinsically disordered.
- a biotinylated derivative of 5-aryl-isoxazole-3-carboxyamide (structure below) precipitates proteins which are enriched in those that bind RNA (RBPs). In general the precipitate proteins were intrinsically disordered characterized by low complexity sequences (LCS). One such example contained 27 repeats of the tripeptide sequence (G/S)Y(G/S). The proteins could also form hydrogels (made of hydrophilic polymers and crosslinks) and transition between soluble and gel phases with extensive hydrogen bond networks. The hydrogel gel phase gave x-ray diffraction patterns similar to beta structure-enriched amyloid proteins. Short ranged weak interactions between LCS might then drive reversible condensation to gel like granule states characterized by extensive hydrogen bonding (again similar to hydrogen bonding on ice formation). If this process goes awry, more continued and irreversible formation of a solid fibril (as seen in neurodegenerative diseases) might occur from the hydrogel state;
- RNAs appear in granules as protein bind them through RNA binding domains of proteins which interact through low complexity sequences leading to phase separation and hydrogel-like formation of granules. Around 500 RNA binding proteins have been found in the human RNA interactome. They are enriched in LCSs and have more tryosines than average proteins in the whole proteome in which the Tyr are often found in an (G/S)Y(G/S) motif. Phosphorylation of tyrosines (Y) in LCS may decrease association and hydrogel stability.
Given the many neurodegenerative diseases are associated with unfolded/misfolded protein aggregates, the high protein concentrations in protein-containing liquid drops might pose problems to cells. If high enough, the equilibrium might progress from the liquid drop to a solid precipitate, which would have severe cellular consequences. The progression to the solid state may irreversibly affect the cell.
Macromolecule Oligomer Formation and Symmetry
Many proteins are found in aggregated states and have quaternary structure. Hemoglobin consists of two alpha and two beta monomers (or protomers) which assemble to produce the biologically relevant heterotetrameric protein. As we discussed inChapter 5A - Reversible Binding I, a given monomer can self aggregate to form homooligomers (dimers, trimers, tetramers, or Mn). The polymers display symmetry with respect to the geometric arrangement of the subunits. Symmetry, as we have just seen, is an important component of the MWC model. Most oligomeric proteins contain protomers that are symmetrically arranged. What mechanism determines whether a monomeric protein forms a homooligomer? Why do they stop at a certain n value? Can proteins be engineered to do so? If mutation can induce oligomer formation, then fewer mutations would be required to produce a symmetric oligomer from subunits since fewer mutations would be required as a single mutation in a single monomer would be represented n times in a single oligomer of n monomers. This fact probably underlies the reason that oligomers display exquisite symmetry. Hence a basic knowledge of symmetry of protein oligomers is necessary.
In the study of small molecules, chemists describe symmetry through the use of mathematical symmetry operations and elements, which find great use in analysis of structure and in molecular spectroscopy. These concepts are usually first encountered in physical and inorganic chemistry classes. A symmetry operation is a movement of an object like a molecule that leads to an identical, superimposable molecule.. Each operation has a symmetry element (point, line, or plane) about which the motion occurs. Some examples are shown below:
Table: Symmetry Elements and Operations
|Element (with Jmol link)||Operation|
|inversion center (i)||projection through center (point) of symmetry of point x,y,z to point -x,-y,-z|
|proper rotation axis (Cn)||rotation around a Cn axis by 360o/n where C denotes Cyclic|
|horizontal (σh) and vertical (σv) symmetry plane||reflection across a horizontal or vertical plane|
|improper rotation axis (Sn)||rotation around a Sn axis by 360o/n followed by reflection in plane perpendicular to the axis.|
Luckily for students trying to apply these rules to protein oligomers, biomolecules made up of chiral monomers (such as the L-amino acids of proteins) can not be converted to identical structures using inversion or reflection since the chirality of monomer would change - for proteins this would entail and L to D amino acid change. That excludes all but proper rotation axes (Cn) from the list above.
A point group is a collection of symmetry operations that define the symmetry about a point. The 4 types of symmetries around a point are those described above: rotational symmetry, inversion symmetry, mirror symmetry, and improper rotation. The types of point groups around a point include:
- cyclic (Cn) - contain one single Cn rotation axis. A biological example is the tobacco mosaic virus double disk (34 monomers, C17). In this point group note that the n in Cn is equal to the number of monomers and the angle of rotation is 360o/n.
Figure: C2 Symmetry
- dihedral (Dn) - These have mutually perpendicular rotation axes. Specifically they contain at least 1 C2 axis perpendicular to a Cn axis (Canter and Schimmel. Biophysical Chemistry - Part 1). The minimal number of subunits is n. Most protein oligomers fall into this category. The packing (or asymmetric) unit does not have to be a single monomer but could be a heteodimer.
- A D2 point group has 1 C2 axis and 2 perpendicular C2 axes, and 4 monomers (like Hb). These proteins can dissociate into two dimers (such as two α/β dimers for Hb). Note that a different arrangment of 4 monomers could produce a oligomer with C4 symmetry instead of D2.
- A D4 point group has 1 C4 axis and 4 C2 axes, along with 2n=8 subunits. An example of a D4 point group is ribulose bisphosphate carboxylase/oxygenase (RuBisCO) which has 8 subunits (where a subunit, or more technically the assymetric subunit, is a dimer of a small and large molecular weight protein). This point group could arise from quaternary structure of two C4 tetramers or four C2 dimers.
Figure: D2 Symmetry
- cubic - contain four C3 axes connecting opposite corners of a cube (so the lines are effectively diagonals) arranged as the four body diagonals (lines connecting opposite corners) of a cube. The tetrahedron (4 sides), cube (6), octahedron (8), and icosohedron (20), perfect Platonic solids (in which all faces, edges and angles are congruent) all have related 3 C3 axes (diagonals connecting opposite corners for cubes, diagonals from a vertex to the opposite face for tetrahedrons, line connecting two opposite faces for octahedron, etc ) so they all can be considered to be part of the cubic point group.
Cubes have a total of 13 symmetry axes comprising 3 types (three C4 axes passing through the centers of opposite faces, four C3 axes passing through opposite vertices, and six C2 axes passing through the the centers of opposite edges). On octahedron can be aligned with a cube and be shown to have the same symmetry axes.
- Flash movie: Cube (red) and Octrahedron (blue) with 13 symmetry axes.
Tetrahedrons have a total of 7 symmetry axes comprising two types (four C3 axes of the cube and three C2 axes which are the same as the C4 axes of the cube. First note the relationship between a cube and an inscribed tetrahedron.
- Flash movie: Tetrahedron with 7 symmetry axes.
A dodecahedron with 12 regular pentagon faces (green) and an icosohedron with 20 equilateral triangle sides (red) can be aligned with each other (as can cubes and octahedrons) and have 31 symmetry axes, as shown below. Note also the relations between a cube inside a dodecahedron and a octahedron inside of a dodecahedron that makes sharing of symmetry axes between these pairs more obvious.
- Flash movie: dodecahedron and icosohedron with 31 symmetry axes.
vrml files for movies from http://www.georgehart.com/virtual-polyhedra/symmetry_axes.html
Examples of protein complexes with these point groups are:
- aspartate-ß-decarboxylase, tetrahedral, 12 asymmetric units
- dihydrolipoyl transsuccinylase, octahedral, 24 asymmetric subunits
- many spherical viruses, icosahedral, 60 asymmetric units
Proteins, especially those involved in cytoskeletal filaments, can form fibers which contain helical symmetry which differs from those described above since the monomers at the ends of helical fibers, although they have the same tertiary structures as those in the middle of the helical fibers, do not contact the same number of monomers as monomers internal in the oligomer and hence have different microenvironments.
A recent article by Grueninger et al. addresses the question of whether the process of oligomerization can be programmed into the genome. Can simple amino acid substitutions lead to oligomerization? Remember that oligomerization can be beneficial (formation of cytoskeleton filaments) or detrimental (formation of fibers in sickle cell anemia and mad cow disease). Oligomers with long half-lives (for example cytoskeletal filament such as actin and tubulin) and short half-lives (for example proteins causes transient activities are regulated by oligomer formation) are both necessary.
It has long been noted that if a protein chain forms oligomers, then a single amino acid change in the chain would be found n times in an oligomer of n chains. Mutations could either promote chain contact and oligomer formation or dissociation into monomeric or other asymmetric subunit composition if the mutation were in a region involved in subunit association (a contact region). Experimental work in this field of study is hampered by the fact that mutants made by site-specific mutagenesis to prefer the monomeric state often fail to fold (due to hydrophobic exposure and aggregation. Studies have shown that most contact areas between monomers or other asymmetric units are hydrophobic in nature and the contact regions must be complementary in shape. Obviously mutations that replace hydrophobic side chains involved in subunit contact with polar, polar charged, or bulkier hydrophobic side chains would inhibit oligomer formation.
Grueninger et al were able to successfully engineer dimer formation and oligomer formation as well. First consider the simplest case of a mutation in a monomer that can produce a dimer with C2 symmetry. This is illustrated below which also shows how a mutation that produces a weak interaction in a monomer could also produce a long helical aggregate (which can't be crystallized) without symmetry (as described above). A mutation at 2 could promote either oligomer helix formation or dimerization.
(adapted from Grueninger et al. Science, 319, 206-209 (2008)
It should be noted that mutation could lead to dimer or oligomer formation by producing a more global conformational change in the monomer (not indicated in the example above) which leads to aggregate formation, as we have seen previously in the formation of dimers and aggregates of proteins associated with neurodegenerative diseases (like mad cow disease).
Grueninger produces mutants of two different proteins that showed dimer formation as analyzed by gel filtration chromatography (but did not crystallize so no 3D structures were determined). In addition the group modified urocanase, a C2 dimer, at 3 side chains to form a tetramer with D2 symmetry. Also, they modified L-rhamnulose-1-phosphae, a C4 tetramer, at a single position to form an octamer with D4 symmetry. The latter two were analyzed through x-ray crystallography. Their work suggests ways that complex symmetric protein structures arose in nature from simple mutation and evolutionary selection.