Fig 1.
Reversing the allosteric communication leads to the identification of allosteric sites.
In allosteric communication, binding of effectors at allosteric sites causes a change of configurational work exerted in the distant functional sites. In the hypothesis of reversibility of allosteric communication, simulated ligand binding at the functional sites enables the identification of allosteric sites.
Table 1.
Free energy changes at the functional and allosteric sites upon reverse perturbation of the allosteric communication for the proteins of the classical set.
Fig 2.
Free energy profiles obtained as a result of the reverse perturbation for AnthS, ATCase, CAP and DAHPS.
(A) AnthS, a heterotetramer involved in the biosynthesis of anthranilate. (B) ATCase, a heterododecamer that catalyzes the first step in pyrimidine biosynthesis. (C) CAP, a homodimeric transcriptional activator. (D) DAHPS, a homotetramer involved in the biosynthesis of aromatic amino acids. The free energy changes are the average from corresponding residues in different subunits. The symbols indicate residues in corresponding ligand binding sites. Grey error bands indicate the standard deviation of Δgi across the subunits. In the middle, the structure of the protein in the Cα coarse-grained model is colored according to the free energy change (increased–blue, decreased–red). The tube radius is proportional to the value of the free energy change. Colored spheres indicate ligand binding sites in a simplified quaternary structure depicted on the right: red or orange colors are used for functional sites, blue–for allosteric sites.
Fig 3.
Free energy profiles obtained as a result of the reverse perturbation for DAK, NADME, PFK and PGDH.
(A) DAK, a dimeric arginine kinase. (B) NADME, a homotetramer for oxidative decarboxylation of malate to pyruvate. (C) PFK, the homotetramer is a key enzyme in glycolytic pathway. (D) PGDH, a homotetramer involved in the biosynthesis of serine.
Fig 4.
Free energy profiles obtained as a result of the reverse perturbation for PTP1B, SSUPRT and ThrS.
(A) PTP1B, a monomeric enzyme that regulates the insulin signalling pathway. (B) SSUPRT, a homotetramer that catalyzes the synthesis of uridine monophosphate. (C) ThrS, a homodimer for the biosynthesis of threonine.
Fig 5.
Results of the reverse perturbation of allosteric communication in the large BGDH and G6PD homohexamers.
Reverse perturbation of the allosteric communication suggested the presence of latent allosteric sites in the antenna region of the BGDH (A) and various locations of the G6PD (B). The antenna region of BGDH and the Cys219 in the core of G6PD are highlighted by circles in the free energy profiles. The large standard deviation of the per-residue free energy change is caused by the structural asymmetry of homologous subunits.
Fig 6.
The operational definition of allosteric sites.
The homologous FBPase1 from (A) E. coli and (B) Sus scrofa are used for the illustration of the communication versus physical interactions between residues of functional and allosteric sites. Residues in ligand binding sites are indicated by spheres on the left. Pairs of residues in different sites, with Cα pairwise distance not greater than 11 Å are shown by black lines. On the right, ligands from the ligand-bound crystal structures are superimposed to the analysed apo structures to show the location of the binding sites. Free energy changes of every residue upon reverse perturbation are calculated and displayed on the structures (colors and tube thickness) and in the free energy profiles.
Fig 7.
Receiver operating characteristic (ROC) curves.
(A) High predictive power can be achieved for most allosteric proteins with distant functional and allosteric sites. (B) Allosteric sites for which low predictive powers are obtained. (C) A histogram of the area under the ROC curves (AUCs) for the detection of allosteric sites in proteins from the additional benchmark set of 41 proteins.
Fig 8.
Allosteric signalling from overlapping activator/inhibitor binding sites in PFK.
Perturbation of the larger binding site ADP (activator, red) causes larger increase in free energy (0.70 kcal/mol) to F6P functional sites, compared to that (0.30 kcal/mol) obtained from the smaller site PEP (inhibitor, blue). Residues in the sites are displayed as spheres along with the residue numbers. ADP and PEP from ligand-bound crystal structures (PDB code: 4pfk and 6pfk, respectively) are superimposed on the structure for illustration. The common residues that bind both ligands are colored grey, residues that only bind ADP are colored red. The distance cutoff for interacting residues of PFK and corresponding ligands is 3.5 Å.
Fig 9.
Inducing the allosteric response in the NADME.
Left: reverse perturbation of allosteric communication in the NADME revealed presence of the repertoire of potential latent allosteric sites. Right: sites 1–5 are perturbed and the resulted free energy changes at functional NAD site are tabulated. Site 5 serves as a negative control.
Fig 10.
Fine-tuning of the allosteric response.
Fine-tuning of allosteric response at the NAD functional site of the NAD-dependent malic enzyme can be achieved by varying composition of the perturbed site 1. The free energy changes (kcal/mol) at the NAD functional site are shown. The black lines indicate possible structures of ligand frameworks that can interact with corresponding allosteric sites.