CE21 - Alimentation et systèmes alimentaires

A MULTIdisciplinary approach to decipher the first step of aroma MIXture perception. – MULTIMIX

MULTIMIX : A MULTIdisciplinary approach to decipher the first step of aroma MIXtures perception

Odors that we detect in our environment result from the perception of mixtures of odorants. Several of these mixtures can lead to a homogeneous percept: odor blending or masking by one component.<br />MULTIMIX aims to identify the biological, molecular and sensory characteristics of odorants that contribute to a homogeneous percept. It will provide keys for the improvement of the flavoring of foods formulated according to nutritional recommendations.

Odor blending and masking: an attractive challenge to understand the processes involved in the homogeneous perception of odor mixtures at the peripheral level of the olfactory system

The conceptual approach of this project intends to acquire fundamental knowledge on aroma mixture homogeneous percept. A percept is heterogeneous when the specific odor quality of each aroma compound taken separately can be identified. Conversely, a percept is homogeneous when a single odor is discerned from the mixture. There are two categories of homogeneous percept: odor blending if the mixture of molecules carrying different odor notes is perceived to have a specific new odor, distinct from odors of the its components; masking, if one component completely covers the quality of the other components in the mixture. It is now accepted that aroma mixture perception results from interactions occurring from the peripheral olfactory system to the brain. Nevertheless, the precise pathways involved in the homogeneous perception of odor mixtures remain poorly understood, especially at olfactory receptors (ORs) and olfactory sensory neurons (OSNs) levels. To address this question, the objective of MULTIMIX is to identify the characteristics of odorants and ORs that contribute to a homogeneous percept through the study of four mixtures according two approaches:<br />• Target-based approach: ORs point of view. How do OSNs process the chemical signal of an odorant mixture? Do odorants have a common pattern of ORs targets? Is it solely the ORs, or is the transduction pathway also critical to integrate these mixtures? <br />• Ligand-based approach: odorants point of view. Do ligands have common molecular features? Can specific networks be generated between ligands through their odors? <br />Ultimately, the fundamental knowledge offered by MULTIMIX aims to constitute a finalized knowledge that will be used to better understand food flavor perception.

The MULTIMIX project is based on the study of four mixtures: two blending mixtures and two binary masking mixtures:
• «pineapple « blending mixture«: ethyl isobutyrate (EI), ethyl-maltol (EM), allyl hexanoate (AH) ;
• «red cordial« blending mixture: isoamyl acetate (IA), vanillin (V), frambinone (F) ;
• binary masking mixture 1 : octanal (O) and citronellal (C) ;
• binary masking mixture 2 : isoamyl acetate (IA) and whiskey lactone (W).
In the case of the “Pineapple” mixture, the perception of the mixture of ethyl isobutyrate (Et iB, strawberry-like odor) and ethyl maltol (Et-M, caramel-like odor) was investigated in comparison with a reference (allyl hexanoate, Al-H, pineapple-like odor) chosen to evoke an odor close to the one expected from the mixture.
The target-based approach focuses on ORs and OSNs: the aim is to identify and characterize ORs and OSNs activated by the molecules involved in the blending process through experimental and computational explorations. First, using the RNA-seq methodology we will identify ORs activated by the molecules involved in binary model mixtures. Then, the identified OR-odorant pairs, will be pooled to create a new computational OR-OR network followingbioinformatic and chemogenomic approaches. Finally, the identification of ORs targets will be completed by the analysis of the physiological responses of OSNs through the characterization of odorant response’s properties and OSNs’ transduction capacities.
The ligand-based approach aims to study molecular properties of odorants. The aims are to identify (i) the common molecular features by a pharmacophore approach and (ii) the links between the odor notes of the odorants involved in homogeneous percept by exploring networks of odor notes and odorants.

Within the context of the target-based approach, the RNA-sequencing methodology (RNA-seq) led to identify potential targets of the odorants involved in the mixtures studied. The first results indicate that in addition to several ORs, potential biological targets are associated with xenobiotic metabolism, membrane transport, signaling and molecular transport. Measurements of functional activation by odorants of ORs in the cellular system is in progress. The results will provide the quantification of the activity of odorants regarding their ORs targets.
Electro-olfactography (EOG) experiments are currently carried out in mice to measure the changes in electrical potentials of the olfactory epithelium (OE), which reflect the response of the olfactory neurons to odorants and mixtures. The first results show different response intensities depending on the areas of the OE and the odorants. In the case of the «Pineapple« blending mixture, whatever the zone, the signals’ amplitude is the strongest both for the EI and the EI-EM mixture, while it is the weakest for EM, and the responses to the AH are intermediate.
With regard to the ligand-based approach, the molecules carrying the odor notes «strawberry« (STR), «caramellic« («CAR) and «pineapple« (PNA) were extracted from a database of almost 4000 odorant molecules to provide a set of 298 molecular structures and their related odors. The network formed by these odor descriptors has shown a main association between the STR and CAR notes and, to a lesser extent, between the STR and PNA notes. The statistical study of molecular descriptors revealed strong structural differences between «CAR« molecules and «PNA« molecules, while some similarities between these molecules are related to the distribution of their chemical features in the 3D space.

The discrimination and identification of thousands of odors is based on a combinatorial coding that occurs at the level of ORs, which implies several possibilities of interactions between odorants and ORs. Moreover, most of the mammalian ORs are still orphan, and due to the huge number of odorants, the analysis of all pairs OR-odorant is experimentally unfeasible. The experimental identification of target ORs and OSNs responding to several molecules is relevant to understand complex aroma perception. This will strongly contribute to the understanding of peripheral coding and olfactory signal processing.
Beyond olfactory perception stricto sensu, odorants and ORs can play a broader biological role. Food intake, mood and odors appear to be narrowly linked. The role of odor stimuli in satiety has been pointed, but without considering the biological role of odorants. For instance, the endocannabinoid system is involved not only in psychomodulatory effects, but also in food intake and metabolic regulation. This later effect is thought to take place through olfactory processes. More broadly, results of RNA-seq studies will experimentally reveal the other potential targets and biological pathways affected by the tested odorants. Such results will contribute to both understanding of the olfactory processes and the associated biological effects beyond olfactory perception.
Lastly, there is a growing interest towards ectopic expression of ORs, namely the expression of ORs in numerous non-chemosensory organs, such as gut. Because of the strong links between gut physiology, mood, food intake and olfaction, improving knowledge of the ligands of these ORs would be a worthwhile asset in the fight against obesity and detrimental food intake behaviors.

Achebouche R, Tromelin A, et al. (2022). Application of artificial intelligence to decode the relationships between smell, olfactory receptors and small molecules. Scientific Reports, 12 (1), 18817, dx.doi.org/10.1038/s41598-022-23176-y, hal.inrae.fr/hal-03845677, OA

Rugard M, Jaylet T, et al. (2021). Smell compounds classification using UMAP to increase knowledge of odors and molecular structures linkages. PLoS ONE, 16 (5), e0252486, dx.doi.org/10.1371/journal.pone.0252486, hal.inrae.fr/hal-03266298, OA

Tromelin A, Koensgen F, et al. (2020). Exploring the Characteristics of an Aroma-Blending Mixture by Investigating the Network of Shared Odors and the Molecular Features of Their Related Odorants. Molecules, 25 (13), dx.doi.org/10.3390/molecules25133032, hal.inrae.fr/hal-02904635, OA


Rugard M, Tromelin A, Audouze K (2022). [P27] Chemo-Biological Analysis Applied to the Olfaction Field. Presented at : 8. Strasbourg Summer School in Chemoinformatics, Strasbourg, France (2022-06-27 - 2022-07-01), hal.inrae.fr/hal-03810731

Tromelin A, Koensgen F, et al. (2021). Exploring the odorant and molecular characteristics of molecules sharing the odour notes of an aroma blending mixture. Presented at : 16. Weurman Flavour Research Symposium, Dijon (webinar), France (2021-05-04 - 2021-05-06), dx.doi.org/10.5281/zenodo.5948755, hal.inrae.fr/hal-03582946

Rugard M, Jaylet T, et al. (2021). Increasing knowledge of odors and molecular structures linkages of smell compounds by comparing UMAP method to other classification approaches. Presented at : SFCi2021, Lille (France), Virtual meeting, France (2021-09-21 - 2021-10-01), hal.inrae.fr/hal-03442859

Tromelin A, Koensgen K, et al. (2021). Exploring the odorant and molecular characteristics of molecules sharing the odour notes of an aroma blending mixture. Presented at : 16. Weurman Flavour Research Symposium, Dijon, France (2021-05-04 - 2021-05-06), hal.inrae.fr/hal-03253812, OA

Rugard M, Zaarour D, et al. (2019). Chemo-biological analysis applied to the olfaction field. Presented at : 9. journées de la Société Française de Chémoinformatique (SFCI), Paris, France (2019-10-21 - 2019-10-22), hal.inrae.fr/hal-03800876

Soubeyre V, Taboureau O, et al. (2019). In vivo high-throughput identification of olfactory receptor repertoire coding for a blending mixture. Presented at : 12. Wartburg symposium on flavour chemistry and biology, Eisenach, Allemagne (2019-05-21 - 2019-05-24), hal.inrae.fr/hal-02936619

Tromelin A, Koensgen F et al. (2019). Exploring the network of odours shared by an aroma blending mixture. Presented at : 12. Wartburg symposium on flavour chemistry and biology, Eisenach, Allemagne (2019-05-21 - 2019-05-24), hal.inrae.fr/hal-02785784

According to the nutritional and health recommendations of the French National Program for Nutrition and Health (Programme National Nutrition Santé, PNNS), the development of food products combining nutritional and organoleptic qualities for the benefit of the consumers' health and well-being is of paramount importance. The reduction of salt, sugar and fat represents one of the means for the development of food with good nutritional quality. This reduction can however be associated with a loss of organoleptic quality of reformulated food. Meanwhile, hedonic properties of food represent a key factor for its palatability. Since consumers’ preferences lean towards a high liking for sweet and fat food, new means to avoid rejection of healthier food are required. Food can be regarded as a complex multimodal perceptual stimuli, involving several dimensions. The holistic perception of taste, odor and aroma constitutes the flavor, which gives its identity and typicality to a food product. It is difficult to distinguish between the taste and the aroma: what is often believed to be the "taste" of food is in reality its odor. Interestingly, due to the association of aroma and taste into a single object named flavor, when a taste–odor association exists, the odor alone may be sufficient to produce the flavor perception. For example, the taste of a sucrose solution is perceived sweeter after addition of a caramel aroma. Likewise, an odor associated to a salty taste can enhance the salty perception. In this way, flavorings can be used to counterbalance salt, sugar or fat reduction in foods.
Aromas are often perceived in mixtures, which can lead to a homogeneous percept when a single odor is perceived from the mixture. The homogeneous percept is the subject of important applications in food flavoring, either to give or restore flavor typicality through the use of blending aroma mixtures, or to mask off-flavors. Nevertheless, the processes involved in the homogeneous perception of odor mixtures are still poorly understood. In this context, a fundamental study of homogeneous percept can bring essential knowledge to manage the aroma formulation, which, in turn, will result in food that meets both consumer expectations and criteria of healthy food.
The objective of the MULTIMIX project is to identify the biological and molecular characteristics of odorants that induce a homogeneous percept. In this perspective, MULTIMIX proposes a strong combination of in silico, in vitro, ex-vivo and in vivo approaches. Together, these approaches will be used to improve the understanding of homogeneous perception of odor mixtures and to provide, in fine, an assemblage of models for decision-support in formulation of aroma blending. Peripheral mechanisms occurring in the nose have been proposed to play a decisive role in the processing of mixture information. The understanding of mechanisms involved in the perception of odor mixtures at the peripheral level of olfactory process is a challenging subject.
In the MULTIMIX project, we will capitalize on previous studies done by several partners of the MULTIMIX consortium to explore the properties of several mixtures of odorants eliciting blending and masking. For instance, we will investigate the mixture of two odorants having strawberry and caramel odors respectively, which gives rise to the configural odor of pineapple.
MULTIMIX is an ambitious project that will have a significant impact in several domains of food chemistry and olfactory perception (food science, psychophysics, computational chemistry, neurophysiology, cognitive neurosciences).

Project coordination


The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.


Inserm ERL U1133 Equipe CMPLI-BFA-Université Paris 7 Inserm_Equipe Computational Modeling of Protein Ligand interactions-U1133
MTi Molécules Thérapeutique in silico

Help of the ANR 598,966 euros
Beginning and duration of the scientific project: October 2018 - 48 Months

Useful links

Explorez notre base de projets financés



ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter