Adaptation of olfactory receptor abundances for efficient coding

Tiberiu Teşileanu, Simona Cocco, Rémi Monasson, Vijay Balasubramanian

Abstract

Olfactory receptor usage is highly heterogeneous, with some receptor types being orders of magnitude more abundant than others. We propose an explanation for this striking fact: the receptor distribution is tuned to maximally represent information about the olfactory environment in a regime of efficient coding that is sensitive to the global context of correlated sensor responses. This model predicts that in mammals, where olfactory sensory neurons are replaced regularly, receptor abundances should continuously adapt to odor statistics. Experimentally, increased exposure to odorants leads variously, but reproducibly, to increased, decreased, or unchanged abundances of different activated receptors. We demonstrate that this diversity of effects is required for efficient coding when sensors are broadly correlated, and provide an algorithm for predicting which olfactory receptors should increase or decrease in abundance following specific environmental changes. Finally, we give simple dynamical rules for neural birth and death processes that might underlie this adaptation.

Introduction

In the mammalian olfactory system, hundreds of types of olfactory sensory neurons (OSNs) exist, each expressing a single type of receptor. Two biological observations have long remained unexplained. First, receptor heterogeneity: some receptor types are orders of magnitude more abundant than others. Second, sporadic adaptation: unlike the visual system, where increased stimulus typically leads to increased activity, exposing an animal to a specific scent can cause corresponding receptors to increase, decrease, or remain unchanged in number.

This paper addresses these puzzles by applying the efficient coding hypothesis. While previous theories often assume sensors are independent, olfactory receptors are "broadly tuned"—one odorant activates many receptors, and one receptor responds to many odorants. This creates significant correlations. The authors propose that these biological observations are the signature of a system mathematically optimized to extract maximum information from a noisy, correlated environment.

The Efficient Coding Model

The authors developed a mathematical framework based on information maximization. The olfactory response is modeled as a transformation of environmental odorant concentrations into neural signals, mediated by receptor abundances and hampered by noise.

Figure 1: Schematic of the olfactory pathway and the corresponding mathematical response model.

The core of the model is the Overlap Matrix ($Q$), which captures correlations between receptor responses based on chemical affinities and environmental statistics. Using Karush-Kuhn-Tucker (KKT) conditions, the authors find the optimal set of receptor abundances $\{K_a\}$ that maximizes mutual information subject to a fixed total neuron count. They also propose a dynamical birth-death process that allows the population to drift toward this mathematical optimum, reflecting the natural turnover of OSNs in mammals.

Key Insights and Results

The model reveals that the optimal abundance of a specific receptor is highly context-dependent.

Context-Dependent Adaptation

A striking finding is that the "correct" number of receptors for a given smell depends on the abundance of receptors for other, related smells. If two receptors provide redundant information, the system may down-regulate one to save resources, even if its target odor is prevalent. This "Inverse Overlap Rule" suggests that in high-signal environments, optimal abundance is inversely related to a receptor's predictability from its neighbors.

Diversity and Population Size

The model predicts a "regime change" based on total neuron count. Small populations should focus on a few high-sensitivity receptors, while larger populations hit "step transitions" where it becomes optimal to express a wider variety of receptors more uniformly.

Figure 2: Model predictions for receptor diversity scaling and validation against mammalian allometric data.

Empirical Validation

The model's predictions align with several datasets:

  • Diversity Scaling: Cross-species analysis shows a positive correlation between functional OR gene counts and a proxy for total OSN count (epithelium area adjusted for body mass), supporting the prediction that diversity grows with population size.
  • Experimental Match: The model successfully replicates the "reproducible but diverse" effects seen in mouse RNAseq experiments, where environmental changes lead to both increases and decreases in specific receptor abundances.

Discussion and Future Directions

While the model provides a powerful framework, it relies on a linear approximation of receptor responses. Future work could explore full non-linear treatments, although these are currently computationally intensive. Additionally, the model assumes all information is equally valuable; incorporating a "utility" function (e.g., prioritizing predator scents) would add another layer of biological realism. Finally, the exact biological feedback mechanisms—how the olfactory bulb might signal the epithelium to modulate neuron birth or death—remain to be fully characterized experimentally.

Summary updated

Abstract

Olfactory receptor usage is highly heterogeneous, with some receptor types being orders of magnitude more abundant than others. We propose an explanation for this striking fact: the receptor distribution is tuned to maximally represent information about the olfactory environment in a regime of efficient coding that is sensitive to the global context of correlated sensor responses. This model predicts that in mammals, where olfactory sensory neurons are replaced regularly, receptor abundances should continuously adapt to odor statistics. Experimentally, increased exposure to odorants leads variously, but reproducibly, to increased, decreased, or unchanged abundances of different activated receptors. We demonstrate that this diversity of effects is required for efficient coding when sensors are broadly correlated, and provide an algorithm for predicting which olfactory receptors should increase or decrease in abundance following specific environmental changes. Finally, we give simple dynamical rules for neural birth and death processes that might underlie this adaptation.
BibTeX
@inproceedings{tesileanu2019adaptation,
  author = {Tiberiu Teşileanu and Simona Cocco and Rémi Monasson and Vijay Balasubramanian},
  title = {{Adaptation of olfactory receptor abundances for efficient coding}},
  booktitle = {eLife},
  year = {2019},
  doi = {10.7554/elife.39279},
}
Figure