Résumés des 1ères JS > Résumés des conférences Flash

 Retrouvez ici le mini livre des résumés des 1ières JS Numériques du RFMF

Julia SOULLIER, Aurélie BATUT, Anthony TOURNADRE, Jessica DALLOUX-CHIOCCIOLI, Pauline LE FAOUDER, Justine BERTRAND-MICHEL

MetaboHUB-MetaToul Lipidomic Core Facility, I2MC, Inserm, Toulouse, France

Lipids are ubiquitous biomolecules essential to life, found in every cellular type, ranging from the human body and vegetal organisms, down to bacteria. They have many different functions in cell structuration, energy storage and signalling so they are natural biomarkers for different diseases like cancer, cardiovascular, neurodegenerative or lung disease. Their identification and quantification are crucial especially on large clinical cohort. The bottleneck of lipidomic study is the sample preparation which can be long and tedious. Different methods of lipids extraction are described in the literature: such as liquid-liquid extraction (LLE) as Bligh and dyer (B&D) or Methyl-tert-butyl ether (MTBE), protein precipitation or solid phase extraction. They may be the source of many errors, considering the experimenter-dependant possible repeated artefacts. To circumvent these points, and increase the analytical service delivery, MetaboHUB-MetaToul facility decided to automate the sample preparation on a fluent 780 TECAN robot. Since several years, robots are used in pharmaceutical industries, anti-doping, etc. LLE as MTBE is a current robotic extraction, but B&D isn’t automatized yet. This development represents different challenges due to the protocol complexity, the presence of organic solvents, and the scale of volume dispensed.

The presentation will show the optimizations done on the robot on three major points: water contamination, accuracy and repeatability of dispensed volume and recovery of extracted lipids. Two systems (with needle or removal tips) have been tried and tested on the robot in order to determine which one was the more appropriate for our uses. We will show the results obtained on mice liver and plasma samples for fatty acid, neutral lipid and phospholipid profiling by liquid or gaz chromatography coupled with mass spectrometry or flame ionization detector. The complete automated sample preparation will be compared to the manual one.

Hikmat Ghosson^1,2, Yannick Brunato^1,2, Delphine Raviglione^1,2, Marie-Virginie Salvia^1,2,3, Cédric Bertrand^1,2,3,4

1: PSL Université Paris: EPHE-UPVD-CNRS, USR 3278 CRIOBE, Université de Perpignan, 52 Avenue Paul Alduy, 66860 Perpignan Cedex, France

2: UFR Sciences Exactes et Expérimentales, Université de Perpignan Via Domitia, 52 Avenue Paul Alduy, 66860 Perpignan Cedex, France

3: Laboratoire d’Excellence « CORAIL », Université de Perpignan, 52 Avenue Paul Alduy, 66860 Perpignan Cedex, France

4: S.A.S. AkiNaO, Université de Perpignan, 52 Avenue Paul Alduy, 66860 Perpignan Cedex, France 

Untargeted metabolomics is an analytical chemistry approach dedicated for the analysis of small biomolecules called “metabolites”. One of its strengths is the ability of covering wide ranges of metabolic information in biological systems. However, exhaustive coverage of information is challenging from a chemical-analytical point of view. In fact, the widely different physical-chemical properties of molecules (e.g. polarity, acidity/basicity) make the study of various types of metabolites complicated. It demands critical methodological optimizations. Nevertheless, for untargeted approaches, “optimal conditions” are also hard to be defined and judged. 

The current work is focusing on the crucial step of meta-metabolome extraction. It seeks to develop an “exhaustive” extraction protocol able to extract different types of metabolites by covering a wide range of polarity. This is in the framework of developing the “Environmental Metabolic Footprinting” untargeted metabolomics-based approach^[1,2], aiming to assess the environmental fate and impact of complex (bio)pesticides application on soil. Indeed, optimizing a method able to analyze diverse kind of metabolites can assure a wide range of information needed to assess the fate of the pesticide (xenometabolites) and its impact on soil’s biodiversity (endometabolites). 

In this objective, 5 extraction protocols based on solvents and mixtures with different polarities were developed and applied on 2 different types of soils, with 3 different environmental conditions (control, spiked with a synthetic pesticide, spiked with a natural pesticide). So far, 150 samples were extracted and analyzed with a broadband LC-HRMS method that was set-up for the purpose. The collected data were then handled and analyzed with various types of multivariate statistical analyses that were optimized in order to assess the “optimal” protocol, based on 4 main criteria, respectively: 1) the ability to widen the band of the extracted metabolites (qualitative), 2) the higher extraction yield (quantitative), 3) the repeatability of the extraction, and 4) the reliability in discrimination between spiked and control groups (EMF objective). 

The study showed that widening the polarity range for the extraction protocol, using miscible solvents with different properties was applicable and resulted for a significant ability in extracting a wide range of metabolites originating from both pesticide residues and soil endometabolome. It also showed comparable quantitative results and better repeatability comparing to other classical protocols. In addition, the study suggests multivariate analyses as suitable tool not only for data processing for biological studies, but also for developing analytical methods and protocols, as it can give holistic explanations about the large datasets acquired. Hence, in other words, untargeted metabolomics were used in order to improve the analytical method dedicated for an untargeted meta-metabolomics approach. 

Téo Hebra, Véronique Eparvierand David Touboul

Université Paris-Saclay, CNRS, Institut de Chimie des Substances Naturelles, UPR 2301, 91198, Gif-sur-Yvette, France. 

In order to explore the specialized metabolism of a strain or species, the generation of molecular networks based on fragmentation homologies is increasingly employed. In order to construct them, reverse phase liquid chromatography coupled with electrospray ionization tandem mass spectrometry (RP-LC-ESI-MS2) is the most widely used analytical strategy. However, other ionization methods compatible with liquid chromatography, such as atmospheric pressure chemical ionization or atmospheric pressure photoionization (APPI), are available. Can these other ionization methods compete with ESI for the exploration and analysis of specialized metabolism? 

In this study, we compare the standard analytical workflow using ESI or APPI. For that purpose, we analyzed ethyl acetate extracts from fungus and plants containing azaphilone and flavonoid derivatives, respectively. Then we processed the data and built a molecular network to compare the data associated to each ionization source. Due to the lack of reported analysis of specialized metabolites using APPI, we first optimized several parameters to obtain the best signal-to-noise ratio for several ions in our raw fungus extract. We established our experimental conditions as follows: acetone is used as dopant with a flow rate of 10 μL/min, the temperature of the vaporizer is set at 300°C and the optimal capillary voltage is 2000 V. Finally, we acquired and processed data from raw extracts of fungi and plants, and compared ESI and APPI ionization on several aspects: data processing, number of ions, their intensity, signal-to-noise ratio, detection of less abundant ions, analysis and interpretation of the molecular network. 

To conclude, ESI and APPI show good complementarity. For example, ESI is a more sensitive whereas APPI shows better signal-to-noise ratio and allows faster data processing and analysis due to the generation of only [M+H]⁺  ion.

Arnab Dey^[a], Benoît Charrier^[a], Estelle Martineau^[a,b], Elodie Gandriau^[a], Catherine Deborde^[c,d], Annick Moing^[c,d], Daniel Jacob^[c,d], Dmitry Eshchenko^[e1], Marc Schnell^[e1], Roberto Melzi^[e2], James Kempf^[e3], Dennis Kurzbach^[f], Morgan Ceillier^[g], Quentin Chappuis^[g], Samuel Cousin^[g], Sami Jannin^[g], Jean-Nicolas Dumez^[a], Patrick Giraudeau^[a]

[a] Université de Nantes, CNRS, CEISAM UMR 6230, F-44000 Nantes, France [b] SpectroMaitrise, CAPACITES SAS, Nantes, France [c] UMR Biologie du Fruit et Pathologie, INRAE, Univ. Bordeaux, Centre INRAE de Nouvelle Aquitaine-Bordeaux, 

33140 Villenave d’Ornon, France [d] Bordeaux Metabolome, INRAE, 2018, MetaboHUB, Centre INRAE de Nouvelle Aquitaine-Bordeaux, 33140 Villenave d’Ornon, France. doi: 10.15454/1.5572412770331912E12 [e] Bruker Biospin [1] Industriestrasse 26, 8117 Fällanden, Switzerland [2] Viale V. Lancetti 43, 20158 Milano, Italy, [3] 15 Fortune Dr., Billerica, mA 01821 USA [f] University of Vienna, Faculty of Chemistry, Institute of Biological Chemistry, Währinger Str. 38, 1090 Vienna, Austria [g] Université de Lyon, CNRS, Université Claude Bernard Lyon 1, ENS de Lyon, Centre de RMN à Très Hauts Champs (CRMN), FRE 2034, 69100 Villeurbanne, France. 

Untargeted metabolomics is of particular importance since it brings crucial information on potential biomarkers without any a priori knowledge. NMR profiling plays a pivotal role in this field thanks to its high reproducibility and repeatability. However, for sensitivity reasons this application relies mostly on 1D ¹H NMR spectroscopy, which is limited by the strong peak overlap characterizing complex biological mixtures. ¹³C NMR offers a promising alternative due to its wide spectral range and high resolution compared to the relative crowding of ¹H spectra. However, the poor sensitivity of ¹³C NMR, only 1.7×10-4 that of ¹H due to low natural abundance and weaker magnetic interactions of ¹³C, strongly limits its application in metabolomics. Hyperpolarization methods offer an appealing solution for sensitive ¹³C NMR metabolomics, especially Dissolution Dynamic Nuclear Polarization¹ (d-DNP), a method that makes it possible to improve the sensitivity of solution-state NMR by factors up to 105. Preliminary d-DNP studies² showed the unique ability to detect ¹³C signals on plant and cancer cell extracts in a single scan at natural abundance, results that are inaccessible by conventional state-of-the-art high field NMR. Later, similar d-DNP studies reported a repeatability better than 4% for ¹³C signals on such biological extracts³, as suitable for analytical metabolomics. Here, for the first time, we introduce d-DNP into a complete workflow for untargeted metabolomics. This consists of sequential steps starting from biological extract preparation, d-DNP hyperpolarization, solution-state NMR acquisition, spectra processing to statistical analysis for discriminating metabolic marker identification. We have demonstrated the approach on a model plant system, two groups of tomato fruit extracts of the same variety at two stages of fruit development (mature-green and red-ripe)⁴. A principal component analysis (PCA) applied to hyperpolarized ¹³C NMR spectral data resulted in very clear group separation and highlighted several biomarkers in full agreement with previously reported studies⁵. We also describe the optimization of experimental parameters of the underlying d- DNP method. This uses semi-automated systems, where most steps can be performed by a single operator within a time compatible with high-throughput studies. These results pave the way towards future ¹³C -based “omics” studies for a variety of samples and diverse applications.

Katy DINIS^1,2, Lucie TSAMBA¹, Freddy THOMAS¹, Eric JAMIN¹, Valérie CAMEL²

1 Eurofins Analytics France, 44323 Nantes, France - 2 Université Paris-Saclay, INRAE, AgroParisTech, UMR SayFood, 91300, Massy, France

Food fraud is a worldwide increasing problem and food authenticity analyses are getting more and more importance. Fruit juice is one of the top ten products at risk of food fraud ^[1]. NMR and mass spectrometry are widely used in food authenticity assessment, particularly in liquid chromatography coupled to high resolution mass spectrometry ^[2-4]. Metabolomics based approaches appear to be a method of choice for food authenticity ^[5-7].

Material & Methods: Apple juices from different processes were collected (both direct juices and concentrated juices). Samples were centrifuged during 10 min at 4500 rpm and the supernatant was diluted before analysis. Three replicates per samples were prepared. Sample vials were randomized in the analytical sequence. QC samples (pool of samples) and diluted QC samples were also prepared and analyzed every 10 injections. Analyses were performed on a ThermoFisher Vanquish Flex UHPLC system connected to a QExactive Plus mass spectrometer. The UHPLC separation was achieved using a C18 Hypersil Gold column (150 x 2.1 mm, 1.9μm) at a 0.3 ml/min flow. The mobile phases were water (A) and methanol (B), both acidified with 0.1% formic acid. The MS system was operated in positive ion mode with a mass range set at m/z 120-1000 in full scan mode with a resolution of 70,000. Raw data files were analyzed using Workflow4Metabolomics platform. After peaks extraction from the data files, some data filtration steps were used. Then, chemometrics analyses were realized for sample discrimination using Principal Component Analysis (PCA) and for sample classification using Orthogonal Partial Least Squares – Discriminant Analysis (OPLS-DA) (22 samples were used in the calibration data set and 8 samples in the validation data set).

Results: The data normalization steps allow reducing the number of features detected by deleting features present in blank samples and features with a high variation coefficient. The use of diluted QC samples also helps to detect useless features in order to delete them. These steps are critical because it selects the features used for the chemometrics analysis. In fact, it is important to remove irrelevant peaks while avoiding or reducing discriminant peaks loss. A trend seems to appear for the discrimination between direct juices and concentrated juices using PCA. The OPLS-DA model allowed the discrimination between the juice processes. It also gives good prediction and classification results for the validation data set. During the chemometrics analysis, variables with statistically relevant differences between groups have been determined using calculated p-values and fold changes. Some of them were tentatively identified using online databases.

Conclusion: Untargeted UHPLC-HRMS coupled with chemometrics analyses is a promising method for food fraud detection. Tools will be developed to classify apple juices according to their geographical origin and to discriminate organic and conventional juice samples. Such promising results suggest the capability to detect anomalies (adulteration) with this analytic approach thanks to databases for authentic juices.

Carriot N.^*, Barry-Martinet R., Briand J.-F., Ortalo-Magné A., Culioli G.

^{* 1}Université de Toulon, MAPIEM (EA 4323), Toulon, France 

Le biofouling est un processus naturel de colonisation biologique qui impacte toutes les surfaces naturelles ou artificielles immergées en milieu marin. Le développement des biofilms marins constitue une étape cruciale de ce processus et génère des problèmes économiques majeurs ^[1]. Cette étude vise à déterminer l'effet de la disponibilité en nutriments (phosphates) sur la production métabolique de biofilms formés in vitro par la souche bactérienne marine Pseudoalteromonas lipolytica TC8. Des milieux de culture préparés avec différentes concentrations de phosphates, choisies pour être pertinentes d’un point de vue écologique (de 10^-4 à 1,32 mM), ont été utilisés. Les extraits bactériens ainsi obtenus ont été analysés par métabolomique (UPLC-ESI-QToF) puis les données résultantes ont été extraites (MZmine2), filtrées et analysées statistiquement (MétaboAnalyst 4.0). Il apparait que les différents groupes d’échantillons présentent des différences notables en termes de production métabolique. Une analyse approfondie utilisant l’approche par réseaux moléculaires (GNPS) ^[2] a permis d’identifier une partie des métabolites présents dans ces échantillons, et notamment d’annoter les variables les plus discriminantes. 

Grâce à cette approche analytique, 53 métabolites ont été identifiés (principalement des phosphatidyléthanolamines, des phosphatidylglycérols, des ornithine lipides, des peptides et des céramides) dans les biofilms de P. lipolytica TC8 grâce à leur schéma de fragmentation MS/MS. Les résultats ont montré qu'une carence en phosphates génère des modifications notables de la composition lipidique de la membrane bactérienne avec une diminution drastique de la biosynthèse des phospholipides (notamment des phosphatidyléthanolamines) en faveur d’aminolipides polaires (notamment des ornithine lipides). 

Estelle Deschamps^1,2, Annick Schaumann¹, Isabelle Schmitz-Afonso², Emmanuelle Dé¹, Carlos Afonso², Corinne Loutelier-Bourhis², Stéphane Alexandre¹ 

1. Laboratoire PBS, UMR6270 ; Université de Rouen ; CNRS, CURIB, 25 Rue Lucien Tesnière, 76130 Mont-Saint-Aignan (France) 

2. Laboratoire COBRA, UMR6014 ; Université de Rouen, INSA de Rouen ; CNRS, IRCOF, 1 rue Tesnière, 76821 Mont-Saint-Aignan Cedex (France) 

When studying bacterial metabolome, bacterial culture in laboratory is often required in order to have enough material for sensitive analyses. However, since bacterial growth in laboratory can impact the metabolic pathway, growth conditions should simulate as close as possible the bacterial natural environment. Therefore, particular caution must be taken when choosing the growth medium since nutrients have direct impact on the produced metabolites. 

In this study, we compared the influence of growth media on the lipidome of Pseudomonas aeruginosa —a bacterium classified as a critical priority for the R&D of new antibiotics by the World Health Organization. Due to the high pathogenicity of P. aeruginosa in cases of cystic fibrosis (CF), this study aimed to find a medium which simulates the CF lung sputum and is compatible with the lipidome analysis. 

In order to choose such synthetic medium, P. aeruginosa PAK was grown into two CF-like media and one commercial medium —used as reference. Prior to lipid extraction, several lipid standards —one for each of the lipid classes— were spiked. Lipid samples were then analysed by reversed phase liquid chromatography – mass spectrometry (LC-MS). Extracted ion chromatograms were automatically generated for each identified phospholipids. Semi-quantification was then performed using relative areas of phospholipids (area of the sample lipid divided by the area of the standard lipid) in order to determine phospholipid quantity variations depending on the growth medium. 

Quantity variations were observed on the proportion of the different classes of phospholipids of P. aeruginosa (phosphatidylethanolamines (PE), phosphatidylcholines (PC), phosphatidylglycerols (PG) and cardiolipins (CL)). Especially, PC were far less produced in the medium which was lacking choline nutrients. The influence of the medium on the proportion of the several phospholipid fatty acyl lengths was also observed —e.g. PE 32:1 and PE 36:1. Therefore, tandem mass spectrometry was performed in order to study the fatty acid composition of phospholipids. Using fatty acid intensity ratios, sn-1 and sn-2 fatty acid positions were determined. We showed that for several phospholipids, the major isomer combination differed depending on the growth medium.

Robin Gauff^1c, Stéphane Greff², Dominique Davoult¹, Christophe Lejeusne² 

1 Sorbonne université, Station biologique de Roscoff. 2 IMBE Marseille, Station d’Endoume c présenter and corresponding author: gauff.robin@yahoo.de 

L’Urbanisation, en combinaison avec le Changement Climatique, est l’un des facteurs de perturbations anthropiques les plus importants. A l’interface des milieux urbains et marins, les ports cristallisent des enjeux et problèmes littoraux. Tout en détruisant les habitats naturels, les infrastructures portuaires offrent de nouveaux substrats durs à coloniser par une grande variété d'organismes (biofouling). Cependant, elles ne sont pas de simples substituts aux écosystèmes naturels car : (i) les filtres environnementaux et les processus écologiques clés peuvent y différer, (ii) leur composition biotique diffère des habitats naturels, et (iii) elles sont caractérisées par une forte diversité et abondance, voire dominance des espèces introduites. Ces milieux encore mal compris sont profondément impactés par des perturbations humaines qui souvent sont intenses mais localisées, résultant dans des environnements caractérisées par des forts gradients ou des conditions parcellées. En effet, les ports sont caractérisés par une importante pollution chimique des eaux (rejets urbains et activités industrielles) constituant un filtre de sélection drastique. Notre étude a pour but de comprendre comment ces filtres environnementaux impactent les organismes vivants dans les ports sur une très petite échelle spatiale (dizaine de mètres). Pour cela des études sont menées entre autres sur le Port du Château de Brest. Nous cherchons à comprendre quels facteurs abiotiques (pollutions etc.) et biotiques (prédation) varient sur cette petite échelle et comment ces filtres impactent la communauté dans son ensemble ainsi que le métabolisme de certaines espèces clefs. Les observations indiquent une variabilité communautaire forte, impactée par les filtres, mais également des modifications du métabolome chez certaines espèces.