Abstract

   Elucidation of complex molecular networks requires integrative analysis
   of molecular features and changes at different levels of information
   flow and regulation. Accordingly, high throughput functional genomics
   tools such as transcriptomics, proteomics, metabolomics, and lipidomics
   have emerged to provide system-wide investigations. Unfortunately,
   analysis of different types of biomolecules requires specific sample
   extraction procedures in combination with specific analytical
   instrumentation. The most efficient extraction protocols often only
   cover a restricted type of biomolecules due to their different
   physicochemical properties. Therefore, several sets/aliquots of samples
   are needed for extracting different molecules. Here we adapted a
   biphasic fractionation method to extract proteins, metabolites, and
   lipids from the same sample (3-in-1) for liquid chromatography-tandem
   mass spectrometry (LC-MS/MS) multi-omics. To demonstrate utility of the
   improved method, we used bacteria-primed Arabidopsis leaves to generate
   multi-omics datasets from the same sample. In total, we were able to
   analyze 1849 proteins, 1967 metabolites, and 424 lipid species in
   single samples. The molecules cover a wide range of biological and
   molecular processes, and allow quantitative analyses of different
   molecules and pathways. Our results have shown the clear advantages of
   the multi-omics method, including sample conservation, high
   reproducibility, and tight correlation between different types of
   biomolecules.

   Keywords: multi-omics, 3-in-1 method, proteomics, metabolomics,
   lipidomics, Arabidopsis, disease

Introduction

   Systems biology, the comprehensive study of biological components and
   their interactions within a cell or a tissue, is indispensable toward
   understanding complex cellular functions and processes. Multi-omic
   measurements and integration of the resulting information can transform
   our understanding of complex biological systems ([31]Dai and Chen,
   2012; [32]He et al., 2012; [33]Mostafa et al., 2016; [34]Meng et al.,
   2019). Multiple layers of information (DNA, RNA, protein, metabolite,
   and lipid) can provide important insights into cellular molecular
   networks. In recent years, rapid progress has been made in genomics and
   transcriptomics. Nevertheless, proteomics, metabolomics, and lipidomics
   have emerged as cornerstones in the field of systems biology because
   the essential information at protein, metabolite, and lipid levels
   cannot be predicted or deduced from genomics and transcriptomics
   ([35]He et al., 2012; [36]Geng et al., 2016, [37]2017; [38]Mostafa et
   al., 2016; [39]Meng et al., 2019).

   To conduct multi-omics, aliquots of the same sample are required for
   different extraction procedures optimized for different biomolecules.
   In addition to increased effort inherent to different parallel sample
   handling, the required sample amounts for multiple extractions are
   often not available. Meanwhile, the multi-components extracted from
   parallel sets of replicates can decrease consistency and comparability
   when performing multi-omics integration. Consequently, a versatile
   extraction method providing robust and reliable recovery of the
   molecular components from a single sample is desirable. Such a method
   decreases sample handling time and thus increases throughput.
   Importantly, it conserves critical samples and improves data accuracy
   and comparability because different molecules are all derived from the
   same sample. Common methods employed for fractionated extractions are
   based on a two-phase lipid extraction method developed in 1957
   ([40]Folch et al., 1957). It uses chloroform/methanol/water
   partitioning of polar and hydrophobic metabolites and was designed to
   increase the purity of lipids. Here we modified and optimized this
   method to obtain high quality proteins, metabolites, and lipids from a
   single sample, as a 3-in-1 method ([41]Figures 1, [42]2). This method
   can be easily applied to many types of materials. It should be noted
   that when applying to other sample types, the amount of samples may
   vary based on the types of samples and their water content, etc.
   Regardless of the source material, proteins, lipids, and metabolites
   have the same physicochemical properties, therefore, this method has
   broad application potential.

FIGURE 1.

   [43]FIGURE 1
   [44]Open in a new tab

   Diagram of 3-in-1 sample preparation method for profiling proteins,
   metabolites, and lipids from control and primed Arabidopsis leaves. The
   biphasic fractionation separates three types of biomolecules
   simultaneously, which are analyzed on the same mass spectrometry
   platform. The data are also analyzed using the same vendor’s software.
   A more detailed workflow of the extraction is shown in [45]Figure 2.

FIGURE 2.

   [46]FIGURE 2
   [47]Open in a new tab

   Detailed workflow of 3-in-1 sample extraction of proteins, metabolites,
   and lipids from control and primed Arabidopsis leaves. A
   chloroform/methanol/water extraction is used to separate the three
   fractions and each layer is carefully isolated using supplies of glass
   materials to avoid plastic contaminants in samples. The order of
   fractionated is important and labeled. Butylated hydroxytoluene (BHT)
   is added at the start of the extraction to avoid oxidation of lipids
   during the procedure. MeOH, methanol; PE, phosphatidylethanolamine; DG,
   diacylglycerol; BSA, bovine serum albumin; CHCl[3], chloroform; FA,
   formic acid; LC-MS/MS, liquid chromatography tandem mass spectrometry;
   IPA, isopropanol; ABC, ammonium bicarbonate; DTT, dithiothreitol; IAM,
   iodoacetamide.

   To test the utility of our 3-in-1 extraction method, we used leaves of
   Arabidopsis thaliana (WS ecotype) that had been primed by a pathogenic
   Pseudomonas syringae pv. tomato DC3000 (Pst DC3000). Systemic Acquired
   Resistance (SAR), a salicylic acid (SA)-dependent immune response,
   improves immunity of systemic tissues after prior localized exposure to
   a pathogen. Arabidopsis knockout mutants defective in SAR response
   differ in disease resistance when compared to wild type plants. Here we
   used Arabidopsis wild type and a knockout mutant of a lipid transfer
   protein DIR1 (defective in induced resistance 1) to examine SAR in
   whole leaves. We have successfully used the 3-in-1 method and annotated
   424 lipids, which cover most of the lipid classes. In addition, we have
   identified 1967 metabolites using a LC-MSn method, and obtained 1849
   protein identifications. These results demonstrate the superior 3-in-1
   method can greatly facilitate multi-omics studies in systems biology.

Materials and Methods

Plant Growth and Bacterial Injection

   Arabidopsis thaliana wild type (WS ecotype) and dir1 mutant (in WS
   background) were grown in 8 h light/16 h dark with a light intensity of
   140 μmol/m^2 s. One mature rosette leaf of the 5-week-old plants was
   injected with either Pst DC3000 in 10 mM MgCl[2] (OD600 = 0.02) for
   treated plants or 10 mM MgCl[2] for mock plants using a needleless
   syringe. Fully expanded distal rosette leaves that were not injected
   were collected at 48 h after infiltration and directly frozen in liquid
   nitrogen and stored in −80°C for 3-in-1 extraction. Three biological
   replicates of treated and three biological replicates of mock leaves
   were used.

Multi-Omics Sample Preparation

   Three hundred milligrams fresh weight leaves were quickly immersed in
   glass tubes with 3 ml pre-heated 75°C methanol (MeOH) and 0.01%
   butylated hydroxytoluene (BHT) and incubated for 15 min. Internal
   standards were added to each sample as follows: for proteins: 60 fmol
   bovine serum albumin (BSA) tryptic peptides per 1 μg sample protein;
   for metabolites: 10 μL 0.1 nmol/μL lidocaine and camphorsulfonic acid;
   and for lipids: 10 μL 0.2 μg/μL deuterium labeled 15:0-18:1(d7)
   phosphatidylethanolamine (PE) and 15:0-18:1(d7) diacylglycerol (DG).

   For extraction of proteins, metabolites and lipids, 6 ml of chloroform
   and 2 ml of water (3:1, vol/vol) were added to each tube and 500 μl of
   MeOH was added to replenish the methanol that evaporated from boiling
   ([48]Folch et al., 1957). Samples were vortexed at 4°C for 1 h. The
   liquid was transferred from the extracts to glass centrifuge tubes for
   further phase separation. To improve collection of all 3 components, 2
   ml of chloroform/methanol (2:1 vol/vol) with 0.01% BHT was added to the
   leaves in the glass tubes and agitated for another 30 min at 4°C. This
   liquid was combined with the previous into the glass centrifuge tubes.
   This last extraction procedure was repeated twice on all the samples
   until the leaves appeared white. After extraction, leaves were dried at
   105°C overnight and weighed for dry weights.

   For phase separation, extracts were centrifuged at 10,000 rpm for 10
   min at 4°C. First the upper (metabolites in MeOH) phase was collected
   and transferred to plastic 2 ml centrifuge tubes, then the bottom
   (lipids in chloroform) phase was removed and transferred to glass
   tubes, leaving the middle (protein) layer for protein collection. The
   lipid extract was evaporated by the Nitrogen gas and dried sample tube
   filled with nitrogen gas was placed at −80°C. The lipid extract was
   dissolved in 1 ml isopropanol (IPA) for LC-MS analysis. Metabolites
   were lyophilized to dryness, then the tubes were filled with argon and
   placed at −80°C. Metabolites were solubilized in 100 μl of 0.1% formic
   acid (FA) for LC MS/MS analysis.

Protein Precipitation and Trypsin Digestion

   Proteins were precipitated by addition of 80% acetone in the glass
   centrifuge tubes in −20°C. After 16 h, the samples were centrifuged at
   10,000 rpm for 10 min at 4°C. After removing acetone, proteins were
   resuspended in 100 μl of 50 mM ABC, reduced using 10 mM dithiothreitol
   (DTT) for 1 h at 22°C, and then alkylated with 55 mM iodoacetamide
   (IAM) for 1 h in darkness. The samples were digested with trypsin
   (1:100 w/w) for 16 h. All the samples were acidified by addition of
   0.1% FA to stop the digestion and stored at −80°C.

Liquid Chromatography Mass Spectrometry (LC-MS) and Omics Data Analysis

   Untargeted metabolomic, lipidomic, and proteomic methods were run on an
   Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific,
   Bremen, Germany). A Vanquish^TM UHPLC was used for lipids and
   metabolites, and an Easy-nLC was used for peptides. An Accucore C18
   (100 mm × 2.1 mm, 2 μm) column and an Acclaim C30 (2.1 mm × 250 mm, 3
   μm) were used for metabolites and lipids, respectively. The column
   chamber temperature was 55°C, and the pump flow rate was 0.45 ml/min.
   For metabolomics, solvent A (0.1% FA) and solvent B (0.1% FA and 99.9%
   acetonitrile) were used. The LC gradient is set to 0 min: 1% of solvent
   B, 5 min: 1% of B, 6 min: 40% of B, 7.5 min: 98% of B, 8.5 min: 98% of
   B, 9 min: 0.1% of B, 10 min stop run. To enhance identification, an
   AcquireX MSn data acquisition strategy was used which employs replicate
   injections for exhaustive sample interrogation and increases the number
   of compounds in the sample with distinguishable fragmentation spectra
   for identification ([49]David et al., 2021). Pooled samples were
   created using equal volumes of all the samples for quality control, and
   were run after each sample set. Electrospray ionization spray voltage
   for positive ions was 3500 and for negative ions was 2500. Sheath gas
   was set to 50, auxiliary gas was set at 1 and sweep gas was set to 1.
   The ion transfer tube temperature was set at 325°C and the vaporizer
   temperature was set at 350°C. Full MS1 used the Orbitrap mass analyzer
   with a resolution of 120,000, scan range (m/z) of 55–550, maximum
   injection time (MIT) of 50, automatic gain control (AGC) target of 2e5,
   1 microscan, and RF lens set to 50. For lipidomics, solution A
   consisted of 0.1% FA, 10 mM ammonium formate, and 60% acetonitrile.
   Solution B consisted of 0.1% FA, 10 mM ammonium formate, and 90:10
   acetonitrile: isopropyl alcohol. The LC gradient is set to 0 min: 32%
   of solvent B (i.e., 68% of solvent A), 1.5 min: 45% of B, 5 min: 52% of
   B, 8 min: 58% of B, 11 min: 66% of B, 14 min: 70% of B, 18 min: 75% of
   B, 21 min: 97% of B, 26 min: 32% of B, 32 min stop run. Full MS1 used
   the Orbitrap ion trap mass analyzer with a resolution of 70,000, 1
   microscan, AGC target set to 1e6, and a scan range from 200 to 2000 m/z
   for positive and negative polarity. The dd-MS^2 scan used 1 microscan,
   resolution of 35,000, AGC target 5e5, MIT of 46 ms, and loop count of
   3.

   The column used for peptides was the Acclaim PepMap^TM 100 pre-column
   (75 μm × 2 cm, nanoViper C18, 3 μm, 100 A) combined with an Acclaim
   PepMap^TM RSLC (75 μm × 25 cm, nanoViper C18, 2 μm, 100 A) analytical
   column. The LC runs a linear gradient of solvent B (0.1% FA, 99.9%
   Acetonitrile) from 1 to 30% for 90 min at 250 nL/min. The solvent A was
   0.1% FA. The MS was operated in data-dependent acquisition mode with a
   cycle time of 3 s. Eluted peptides were detected in the Orbitrap MS at
   a 120,000 resolution with a scan range of 350–1800 m/z. Most abundant
   ions bearing 2–7 charges were selected for MS/MS analysis. AGC for the
   full MS scan was set as 2e5 with MIT as 50 ms, and AGC Target of 1e4
   and MIT of 35 ms were set for the MS/MS scan. The normalized collision
   energy was 35, and ions were detected with an Ion Trap detector. A
   dynamic exclusion time of 30 s was applied to prevent repeated
   sequencing of the most abundant peptides.

   Proteome Discoverer^TM 2.4, Compound Discover^TM 3.0, and Lipid Search
   4.1^TM software (Thermo Fisher Scientific, Bremen, Germany) were used
   for proteomics, metabolomics and lipidomics data analyses, respectively
   ([50]Figure 1). Software scoring parameters used for metabolite, lipid,
   and protein identifications are briefly described here with references