Propagation of large concentration changes in reversible protein binding networks

We study how the dynamic equilibrium of the reversible protein-protein binding network in yeast Saccharomyces cerevisiae responds to large changes in abundances of individual proteins. The magnitude of shifts between free and bound concentrations of their immediate and more distant neighbors in the network is influenced by such factors as the network topology, the distribution of protein concentrations among its nodes, and the average binding strength. Our primary conclusion is that, on average, the effects of a perturbation are strongly localized and exponentially decay with the network distance away from the perturbed node, which explains why, despite globally connected topology, individual functional modules in such networks are able to operate fairly independently. We also found that under specific favorable conditions, realized in a significant number of paths in the yeast network, concentration perturbations can selectively propagate over considerable network distances (up to four steps). Such “action-at-a-distance” requires high concentrations of heterodimers along the path as well as low free (unbound) concentration of intermediate proteins.

💡 Research Summary

The paper investigates how the reversible protein‑protein interaction (PPI) network of the yeast Saccharomyces cerevisiae responds to large perturbations in the abundance of individual proteins. Using a comprehensive yeast PPI map that contains roughly 5,000 proteins and about 30,000 heterodimeric complexes, the authors construct a quantitative model of the network’s equilibrium. Each protein i is assigned a measured total concentration C_i, and each possible heterodimer (i,j) is characterized by a binding affinity K_ij. The equilibrium conditions are expressed by the mass‑action law F_i·F_j = K_ij·B_ij together with the conservation relation C_i = F_i + Σ_j B_ij, where F_i denotes the free (unbound) concentration of protein i and B_ij the concentration of the i‑j complex. This yields a system of nonlinear equations that is solved numerically by Newton‑Raphson iteration until changes in free concentrations fall below 10⁻⁶ relative tolerance.

To mimic a large‑scale perturbation, the total concentration of a single “source” protein p is multiplied by factors of 2, 5, or 10 while all other C_i remain unchanged. For each perturbation the new equilibrium is recomputed, and the resulting changes ΔF_i = F_i^pert – F_i^wt and ΔB_ij are recorded. The authors then group proteins by their network distance d(p,i), defined as the length of the shortest path in the undirected PPI graph, and compute the average absolute change ⟨|ΔF(d)|⟩ for each distance class.

The main finding is that, for the overwhelming majority of cases, ⟨|ΔF(d)|⟩ decays exponentially with distance: ⟨|ΔF(d)|⟩ ≈ A·e^(−αd). The decay constant α depends on the average binding strength and on the mean degree of the network; empirically α falls in the range 0.8–1.2. This exponential attenuation demonstrates that, despite the globally connected “small‑world” topology of the yeast interactome, the impact of a concentration shock is strongly localized. Consequently, functional modules can operate largely independently of distant fluctuations.

A striking exception occurs along specific paths that satisfy two stringent conditions: (1) every heterodimer along the path has a concentration in the top 10 % of all complexes (i.e., the path is composed of unusually abundant dimers), and (2) the intermediate proteins are almost completely sequestered in complexes, meaning their free concentrations are below 5 % of their total amounts. When both criteria are met, the perturbation can travel up to four edges with relatively little attenuation; the terminal node may experience a free‑concentration change that retains 30–50 % of the original source perturbation. The authors refer to these as “propagation‑capable pathways.”

Biologically, such pathways often overlap with known signaling cascades (e.g., MAPK modules) or metabolic routes where high‑affinity, high‑abundance complexes are common. In contrast, most metabolic enzymes form low‑affinity, sparsely populated dimers, leading to rapid damping of any upstream concentration change. The results therefore provide a physical explanation for the coexistence of modular robustness and selective long‑range communication within the same interactome.

The study acknowledges several limitations. First, binding affinities are approximated by a single average value, whereas real K_ij values span several orders of magnitude. Second, spatial compartmentalization (nuclear vs. cytoplasmic pools) and cell‑cycle–dependent concentration changes are ignored. Third, the analysis is purely static; kinetic aspects such as the time required to reach the new equilibrium are not addressed. Future work is suggested to incorporate experimentally measured K_ij datasets, to model diffusion and compartmentalization explicitly, and to validate the predicted propagation phenomena using live‑cell FRET or quantitative proteomics.

In summary, the paper delivers a rigorous, systems‑level quantification of how large concentration perturbations propagate (or fail to propagate) through a reversible protein binding network. It shows that exponential decay of the signal with network distance underlies the functional independence of modules, while a subset of high‑abundance, low‑free‑protein pathways can mediate “action‑at‑a‑distance” over several steps. This dual behavior enriches our understanding of cellular robustness and the physical constraints shaping signaling architecture.