Title: Screening of leaf extraction and storage conditions for eco-metabolomics studies

We developed an on-site metabolite extraction method for leaf tissue samples from field 9 studies in challenging logistical circumstances. We highlight extract stability and reproducibility 10 compared to frozen or dried tissue.


Introduction
In agriculture, high-throughput phenotyping approaches have become essential to assess traits related to increased yield, as well as those that confer tolerance to environmental stresses in crops (Araus and Cairns, 2014).Metabolomics is a powerful analytical approach that can provide information on the patterns and nature of plant responses to the environment, by providing information on the chemical features, identity, and quantity of metabolites produced by plants in different conditions (Sardans et al., 2021).In this way, metabolomics can add the chemical dimension to the high-throughput crop phenotyping toolbox, as thousands of metabolic markers often representing hundreds of metabolites can be recovered from a single leaf sample (Brunetti et al., 2013).Investigations of plant stress responses commonly focus on specialized metabolites, which are not essential for cell growth and development and are instead synthesized or modified by plants in response to specific environmental triggers (Macel et al., 2010;Walker et al., 2022).
Nevertheless, high-throughput phenotyping platforms have been developed under refined conditions (i.e., greenhouse and growth chamber facilities proximate to laboratories) and only reliably work with specialized equipment, which limits their application when dealing with realistic (field) conditions (Araus and Cairns, 2014).Such limitations extend to the use of a metabolomics approach in agriculture, where sample preparation and storage is a crucial step towards obtaining high quality data.For instance, most protocols in plant metabolomics require liquid nitrogen to shock-freeze the tissue immediately upon collection and keep the material frozen during the sample handling procedure.While this approach offers the closest representation of the metabolites in the living plant, it requires uninterrupted cooling (usually at -80 °C) and rapid sample handling to avoid thawing and degradation (Ossipov et al., 2008;Sedio et al., 2018;Bakhtiari et al., 2021).
A common alternative, when cooling conditions are not meet, is to dry the plant tissue onsite and then transport and store the dried material (Fernandez-Conradi et al., 2022).This allows for reproducible sampling with minimal material requirement; however, little data is available on how the drying process changes the obtained metabolite profile due to differential stability of different metabolites.As a result, there is a need for a sample preparation method that ensures sample stability until the samples can be processed in the laboratory.This is particularly relevant when the sampling fields are located far from the laboratory facilities, and field campaigns are not easy or possible to repeat.
Here, we address limitations for the use of metabolomics in realistic agroecological conditions by describing and comparing sample handling methods.These methods were conceived in the context of a larger project aiming at understanding the metabolomic profile of maize grown under different conditions in tropical Africa, where weather and logistics conditions can make a metabolomics approach challenging.We first evaluated the suitability of two leaf preservation and six extraction methods, based on changes in metabolite profile across a 75-day storage period, to determine the method that resulted in the best apparent sample stability as judged by similarity to the metabolite profile obtained by the laboratory standard procedure: solid-phase extraction, or liquid-liquid extraction of flash-frozen and finely powdered leaf tissue within a day after harvest.We then conducted a follow-up study focussing on an on-site liquid-liquid extraction procedure in comparison to in-field air-drying followed by laboratory extraction, and the laboratory standard procedure.Our results demonstrate that this liquid-liquid extraction procedure generates reproducible metabolomic profiles while being feasible for field studies in terms of effort and stability of extracts.The methodology presented in this paper has the potential to be a viable alternative to the more established methods for plant metabolomics research in field studies and contribute to a better understanding of plant metabolism under realistic conditions.

Sample handling for broad method screening
Although we aim to develop a method practical for field research in tropical maize agroecosystems (i.e., central Africa), we required an experimental setting which allowed for comparison to extracts generated with an unbroken cooling chain.For this reason, maize plant tissue was collected from field-grown maize at the Strickhof Competence Centre of Agricultural Sciences (Eschikon, Switzerland, 47.4524090, 8.6806795) and used in eight different sample extraction and storage approaches.An overview of the employed methods is shown in Figure 1A and a detailed description of all procedures can be found in SI1.The samples were then stored at three different temperatures (30 °C, 4 °C, and -20 °C) for 1 day, 1 week, 1 month, and 75 days, respectively.At each of those timepoints four replicates of each method and temperature were analysed.

Sample handling for liquid-liquid extraction optimisation
As a follow up study during the following cropping season, we evaluated metabolite stability in two extraction solutions and compared those results to air-dried and shock-frozen leaf storage.A detailed description of all procedures can be found in SI1.The samples were again stored at the same three different temperatures (30 °C, 4 °C, and -20 °C) and four replicates per timepoint, method and temperature were measured at six timepoints after 1 day to 8 weeks of storage time as shown in the timeline in Figure 1B.  the Agilent low concentration tune mix (13 compounds in acetonitrile, part number G1969-85020) prior to analysis.For additional mass accuracy, a calibration segment was programmed from 0.05 to 0.15 min at every UHPLC run with the help of a 6-port-valve with a 20 µL loop which contained a solution of 10 mM sodium formate clusters.The peak tables were exported in .csvformat (see Data Availability) and PCA data was exported in .csvformat to plot graphs using our python workflow (see SI3).

Recommended sample extraction procedure
For the full methods detailing all tested extraction procedures, see the detailed extraction protocols in SI1.Here, we detail the recommended extraction procedure.
An extraction solution consisting of MeOH / water in a 2:1 ratio and camphorsulphonic acid as an internal standard (20 ng / mL) was prepared, of which 200 µL were added to a 1.5 mL Eppendorf tube for each sample.This solution is appropriate for extracting mid to high polarity metabolites which are commonly studied and contain many specialised secondary metabolites.
Twelve leaf disks were collected with a 6 mm diameter hole punch (Milian, Vernier, Switzerland) directly into the extraction solution and the immersion in MeOH directly upon collection may reduce enzymatic activity in the sample (Maier et al., 2010) .The tubes were thoroughly shaken and transported in a common household cooling box containing ice packs.
The leaf tissue was ground inside the Eppendorf tubes using plastic micropestles having a tip with approximately the same volume as the point of the 1.5 mL Eppendorf tubes and attached to a household electric drill.It is recommended to use micropestles with a rough surface to facilitate leaf grinding, which we did by roughening the surface using 240 grit sandpaper.After the leaf tissue was ground to a paste, another 500 µL of the extraction solution was added before shaking thoroughly.The liquid-liquid extraction was performed through addition of 500 µL of chloroform to separate pigments and lipids, followed by thoroughly shaking.After letting the tubes rest for approximately 10 minutes at room temperature, the phase separation was completed, and the upper MeOH / water phase was transferred to fresh microcentrifuge tubes.

Suitability of internal standards
For the broad screening, we selected stevioside as an internal standard, but identified characteristics of the mass spectra during data evaluation which led us to seek alternatives.In the mass spectrum of stevioside in Figure 2A, the detected signals for the proton and ammonium adducts (805 and 822 m/z) are highlighted alongside the main signal at 319 m/z which matches the loss of all three hexose substructures.Additionally, signals were marked which match the loss of one and two hexose substructures.We attributed this to a possible in-source fragmentation and combined with a slight reduction in peak area observed with longer storage periods, the decision was made to include two additional possible internal standardscamphorsulphonic (CSA) and glycyrrhizic acidin the narrow method screening experiment.For comparison, the mass spectrum of CSA can be found below the stevioside spectrum in Figure 2B, and shows a single signal without any fragmentation.Figure S3 shows the intensity of each of the three compounds across the storage experiment.CSA showed a stable signal across the storage period and additionally matched better with our target compound mass range, so we recommend using CSA over stevioside or glycyrrhizic acid.For targeted metabolomic analyses, isotopically labelled reference compounds would be preferable.

Comparison of leaf homogenisation efficiency
Both during the broad method screening and LLE optimisation experiments, different approaches were tested for leaf tissue homogenisation using steel ball mills, ceramic mortars and micropestles.When freezing tissue in liquid nitrogen while grinding, a powder is generally obtained.However, when homogenizing air-dried leaf tissue with either ball mills or ceramic mortars, we were unable to obtain a powder, as some leaf veins remained intact.A direct comparison of the powders obtained when grinding fresh leaf tissue and air-dried leaf tissue in liquid nitrogen is shown in Figure 3.Both of those methods still led to a more homogeneous product than attempting to grind tissue without using liquid nitrogen.Doing so with a ceramic mortar left the leaf tissue structure mostly intact, whereas with a micropestle, a chunky and more homogeneous paste could be obtained (Figure S4).

Selectivity of sample preparation methods
During the broad method screening, fundamentally different sample purification approaches were tested, most notably solid phase extraction (SPE) and liquid-liquid extraction (LLE).The two approaches lead to significant differences in the resulting metabolite profile.In our experiments, the profile after sample workup with SPE was shifted towards molecules with a higher molar mass and a lower polarity compared to samples prepared by LLE.This trend can already be observed in a base peak chromatogram, as shown in Figure 4, and becomes clearer when comparing the data in the peak tables.Some signals could only be observed in LLE samples and others only in SPE samples.
. CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is  The significant shift of the metabolite profile causes a challenge when it comes to multivariate data comparison, where principal component analysis (PCA) is a common approach.Any PCA which contains LLE and SPE samples will group the extraction approaches tightly together as shown in Figure S5, which masks the shifts in the profile across a storage period.Thus, all PCA results were plotted separately for LLE and SPE sample groups to allow a sensible interpretation.

Extract stability over time
Changes in the overall metabolite profile were assessed by PCA, which showed that in almost all cases, a higher temperature led to a greater shift of the profile.During the broad screening, the metabolite profile continued to shift for all evaluated sampling methods without reaching a stable result (which could occur after completing all possible molecular transformations).Examples of the PCA can be found in SI4 with special attention towards  Notably, the 95% confidence ellipses of the shock-frozen leaf tissue samples and the on-site procedure overlap, while air-dried samples are fully removed from the other two groups.All three groups are significantly different from each other when performing a pairwise comparison based on the first five PCs, but when comparing the on-site extract storage samples to shockfrozen leaf storage, the p-value was found to be 0.0006.Comparing shock-frozen and air-dried leaf storage returned a p-value below 0.0001, and the same p-value was obtained when comparing air-dried leaf storage to our on-site extract storage (see: SI5, posthoc analysis).

A note on storage of extracts on SPE cartridges
During the broad screening we found indications that metabolite storage on solid-phase extraction (SPE) cartridges (procedure CE-OA in SI1) could be a viable alternative for on-site sample preparation and storage.Figure 6 shows the samples stored on the SPE cartridge in comparison to elution and drying the eluent under nitrogen flow prior to storage (procedure FE-SPE in SI1).The samples stored on the cartridge seemed more reproducible (tighter grouping of replicates) and less affected by storage time (less distance between the timepoint groups) than the samples following the "FE-SPE" procedure.Due to material shortages at the time, the "CE-OA" approach was only evaluated at three storage timepoints, and we would therefore recommend more in-depth testing before employing this approach on a larger scale.

Liquid-liquid extraction -extract and leaf storage
Storage of samples after an LLE without shock-freezing of the leaf tissue showed promising results during the broad screening.All samples from the 7-and 30-day timepoints that were stored at reduced temperatures were tightly grouped together on the PCA (Figure S10).The samples stored at 4 °C and -20 °C showed a comparable metabolite profile, which led us to study the LLE approaches in more detail.During the narrow screening we could verify the minimal impact of storage in a freezer compared to refrigerator and obtained a highly reproducible metabolite profiles for both conditions.Overall, our on-site extraction procedure results in samples which more closely represent the metabolite profile of shock-frozen leaf tissue compared to air-dried leaf storage.Even when including the samples stored at room temperature, the profile is closer to our goal than air-dried samples, but there is a notable change depending on storage duration.As such, the storage duration of each sample would become an important factor to control for, which may not be required when storing the extracts at reduced temperatures.(Faijes et al., 2007;Link et al., 2008).We thus hypothesize that the immediate contact with MeOH assists with quenching of enzymatic activity for leaf tissue, not unlike flash-freezing with liquid nitrogen.The stability of the MeOH-immersed leaf tissue then becomes relatively independent of temperature and handling.Drying leaf tissue for storage and transport does not have such a quenching step after collection and drying takes more time than flash-freezing or penetration of leaf discs by MeOH solution.Similar effects have been

Figure 1 :
Figure 1: Overview of the evaluated sample extraction and storage methods.The broad method screening included all methods and samples were evaluated after storage for 1, 6, 30, and 75 days respectively.
Liquid chromatography was performed on a Vanquish Horizon UHPLC System by Thermo Fisher (Waltham, MA, USA) build from a Vanquish binary pump H, a Vanquish split sampler HT and a temperature-controllable Vanquish column compartment.Chromatographic separation was achieved on an ACQUITY Premier CSH C18 Column (130 Å, 1.7 µm, 2.1 × 50 mm, Waters, Milford, MA, USA) at 30 °C to reduce column backpressure.Eluent A consisted of H2O + 0.1% HCOOH and B of MeCN + 0.1% HCOOH.The solvent flow was kept at 0.6 mL/min with the following gradient: (i) 5% B isocratic from 0.0 to 0.4 min; (ii) linear increase to 35% B until 2.8 min; (iii) linear increase to 75% until 3.2 min; (iv) linear increase to 100% B until 3.3 min, (v) holding 100% B until 4.4 min (vi) back to the starting conditions of 5% B until 4.5 min; (vii) equilibration for 1.1 min until the next run.The injection volume is dependent on the employed extraction method and is specified in the detailed extraction protocols in SI1.A timsTOF Pro hybrid quadrupole-time-of-flight (QTOF) mass spectrometer equipped with trapped ion mobility spectrometry (TIMS) produced by Bruker (Bremen, Germany) was connected to the Vanquish UHPLC system and was used to acquire ion mobility and MS/MS data.Ionisation was performed in positive and negative ESI mode and the scan range was set to 20 to 1350 m/z at a 12 Hz acquisition rate.Mass and CSS calibration was performed using Figure S1 and S2).All parameters for the peak picking and data evaluation are shown in SI2.

Figure 2 :
Figure 2: Comparison of the full scan MS spectrum of the internal standards stevioside (red, A) and camphorsulphonic acid (CSA, blue, B) with signal annotation of matching m/z ratios.

Figure 3 :
Figure 3: Comparison of ground flash-frozen (left) versus air-dried (right) leaf tissue following the same pulverization procedures.

Figure 4 :
Figure 4: Overlaid chromatograms of a subset of four samples prepared by solid-phase extraction (SPE, blue) and four samples prepared by liquid-liquid extraction (LLE, red) highlighting generally higher abundance of high-polarity (shorter retention time) compounds when LLE is used.

Figures
Figures S6and S7, which show the comparison of all evaluated LLE and SPE methods.For the narrow screening the shift of the metabolite profile over time was significantly reduced.As

Figure 5 :
Figure 5: Principal components analysis (PCA) conducted on metabolite profiles of samples extracted and stored under different conditions: (A) LLE and extract storage at different temperatures, (B) LLE of differently handled leaf tissue samples, (C) on-site LLE and cold storage, (D) frozen storage of shock-frozen leaf tissue.The three sample groups screened during the LLE can also be directly compared to each other, which is shown in Figure 5B.The samples from air-dried leaf tissue separate along PC1,

Figure 6 :
Figure 6: Principal component analysis of samples stored on an SPE cartridge (CE-OA, squares) and samples prepared by SPE and dried down for storage(FE-SPE, hexagons).Samples stored at 30 °C are not included.

Figure
Figure 5C and D highlight the extract stability over storage duration, and notably a lower (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.It is