Build the chemical standard collection
Timing varies depending on the sources of the standard collection
1. Curate a library of metabolite standards that are commercially available as a collection (e.g., IROA Mass Spectrometry Metabolite Library of Standards, Metasci Organic Acids Library) and/or are individually purchased in manufacturer’s stock bottles (e.g., Sigma, Acros).
2. For commercial metabolite standard collections (e.g., IROA, Metasci), follow the manufacturer’s instructions on drying, reconstituting, and pooling. Whenever possible, reconstitute metabolite standards in 50% methanol (LC-MS grade) to reach a stock concentration of 10 mM.
3. For each individually curated metabolite in manufacturer’s stock tubes, weigh at least 3 mg, transfer to a 2 mL Eppendorf tube. The metabolite should be reconstituted in 50% methanol to reach a stock concentration of 10 mM.
4. Generate individual pools of 20-30 metabolite standards by combining stocks and diluting with 50% methanol. Each metabolite in the pool should reach a working concentration of 200 uM.
CRITICAL STEP The metabolite standards in each pool should not share the same molecular mass. Two metabolite standards that share the same molecular mass will confound the RT assignment to each metabolite standard in the pool (see Step 6 below).
5. Analyze individual pools by one or more of the three analytical methods (Steps 7-11 below).
(Optional) If users wish to construct a linear range for each metabolite to enable quantification, individual pools of metabolite standards can be serially diluted in 50% methanol or a biological matrix where the quantification will be done (e.g., serum). If the matrix already contains the metabolite, in some cases there are commercially available versions of the matrix readily depleted of small molecules. The dilution series can be analyzed using each of the three analytical methods.
Construct the m/z-RT reference library
Timing varies depending on the size of the standard collection
6. Construct the following library values in an excel file spreadsheet: Calculate the m/z of each metabolite standard by 1) retrieving its monoisotopic mass from the PubChem database and 2) adding or subtracting the mass of a proton (1.007276 Da) depending on the default adduct ion type. For example, the adduct ion [M+H]+ is the default adduct ion for electrospray ionization in the positive mode (ESI+) and [M-H]- is the default adduct ion for electrospray ionization in the negative mode (ESI-).
7. Run individual pools of metabolite standards for each analytical method and save all raw file output from the LC-MS instrument.
8. Open in the Agilent Qual (Agilent MassHunter Qualitative Data Analysis software, v.B.07.00) a raw file containing a known pool of metabolite standards (e.g., .d file produced by an Agilent Q-TOF). The steps below are carried out in Agilent Qual.
9. Right click on the “TIC scan” option in the left panel and select “extract chromatogram” in the dropdown menu to open the Extract Chromatogram window.
10. Select the settings as described (Fig. 6a-c). The m/z input here should be the comma-separated list of predicted m/z for all metabolite standards contained in this pool (see Step 6 above).
11. Inspect each EIC-scan chromatogram corresponding to each m/z value in the left panel. Record RT in the reference library for a compound only when there is a single, strong peak (e.g., greater than 104 in peak height) for the corresponding m/z value (Fig. 6d). Alternatively, there can be two or more strong peaks or no peak detected for a given m/z value (Fig. 6e-f). Save the library file as .txt after all compounds have either retention times associated with m/z values, or have been removed from the original list of compounds. This will be your library file that you specify when analyzing sample data below.
? TROUBLESHOOT (see Supplementary Table 3)
(Optional) For metabolites run in dilution series, RTs of the same metabolite at several concentrations can be used to produce an averaged RT in the reference library. This averaged RT value (1) increases the accuracy by averaging sample-to-sample variations; and (2) distinguishes the true signal from background noise by validating the peaks for which the ion counts proportionally increase with the concentration.
Prepare samples for LC-MS analysis
Timing 0.5 d
We have developed a metabolite extraction method with specific sections tailored to different sample types used in microbiome studies. Once metabolites are extracted from different sample types, they can either be analyzed immediately or stored at -80 °C for future analysis (Fig. 1).
Homogenize and precipitate samples (steps vary depending on each sample type)
Sample type 1: Bacterial supernatant
12. Remove frozen bacterial supernatant from -80 °C and thaw on ice prior to extraction.
13. Pipette 200 μL of each sample into a 2 mL 96-well microplate well.
14. Add 1 mL of extraction buffer (see ‘Buffer and solvent preparation’) to each well to precipitate proteins.
15. Seal the microplate with a silicone mat and vortex the sealed plate at the highest speed for 5 seconds.
16. Incubate samples on the benchtop at room temperature for 5 minutes.
17. Centrifuge samples for 10 minutes at 5,000 x g.
Sample type 2: Serum
12. Remove frozen serum samples from -80 °C and thaw on ice prior to extraction.
13. Pipette 200 μL of sample into a 2 mL 96-well microplate well.
14. Add 1 mL of extraction buffer (see ‘Buffer and solvent preparation) to each well to precipitate proteins.
15. Seal the microplate with a silicone mat and vortex the sealed plate at the highest speed for 5 seconds.
16. Incubate samples on the benchtop at room temperature for 5 minutes.
17. Centrifuge samples for 10 minutes at 5,000 x g.
Sample type 3: Urine
12. Remove frozen urine samples from -80 °C and thaw on ice prior to extraction.
13. Dilute samples 1:20 in water to reach a total volume of 200 μL (10 μL urine sample in 190 μL of water) in a 2 mL 96-well microplate well.
14. Add 1 mL of extraction buffer (see ‘Buffer and solvent preparation’) to each well to precipitate proteins.
15. Seal the microplate with a silicone mat and vortex the sealed plate at maximum speed for 5 seconds.
16. Incubate samples on the benchtop at room temperature for 5 minutes.
17. Centrifuge samples for 10 minutes at 5,000 x g.
Sample type 4: Feces and/or intestinal contents
12. Remove frozen samples from -80 °C and thaw on ice prior to extraction.
13. Add ~30 mg of sample to a 2 mL screw cap vial containing previously added ~30 mg of acid-washed glass beads.
14. Add 600 μL of water and 600 μL of extraction buffer (see ‘Buffer and solvent preparation’) to each vial to precipitate proteins.
15. Homogenize vials using a mini bead beater at 4 °C for 5 minutes at 3500 oscillations per minute.
16. Incubate samples on the benchtop at room temperature for 5 minutes.
17. Centrifuge samples for 10 minutes at 5,000 x g.
Extract metabolites from all sample types above
18. Aliquot 440 μL of samples into wells of a new 2 mL 96-well microplate. Every sample should be aliquoted into duplicate plates. One plate will be used for LC-MS analysis, and the other will be archived at -80 °C as a backup. A total of 880 μL of supernatant should be collected from each sample between the duplicate 2 mL 96-well microplates.
19. Dry both plates of samples under air at 65 psi at 37 °C in a Turbovap evaporator. Continue this step until the solvent in all wells is completely evaporated. Seal one plate with a silicone mat fitted for the 96-well plate and store at -80 °C for archives as backup in case of re-analysis (e.g., instrument error, sample loss during the first analysis).
20. Reconstitute the remaining samples in 200 μL of reconstitution buffer (see ‘Buffer and solvent preparation’) and seal the plate with a silicone mat.
21. Vortex the sample plate at the maximum speed for 5 seconds.
22. Centrifuge the sample plate for 1 minute at 2,000 x g to spin down the residual sample stuck on the side of the well and/or on the silicone mat.
23. Transfer all 200 μL of samples into a 0.22 μm 96-well filter plate that is placed on top of a new 1 mL 96-well microplate, and centrifuge this plate stack for 10 min at 2,000 x g. For fecal or intestinal samples, centrifuge at5,000 x g for 10 minutes to fully pass the samples through the filter into the bottom plate.
24. Seal the sample plate with a silicone mat and store at -80 °C until LC-MS analysis.
PAUSE POINT Samples can be stored at -80 °C for up to 3 months prior to analysis.
25. Prior to LC-MS analysis, bring plate with reconstituted samples from -80 °C to room temperature.
26. Mix 4 μL of each sample contained in the same experiment into a user-designated well location on the same plate reserved for quality control (QC). This QC mix consists of a pool of all samples in the same experiment to provide a representation of all metabolomic features.
LC-MS analysis on the Agilent 6545 LC/Q-TOF instrument
Timing varies depending on the number of samples to run
The user operation instructions vary between instruments (e.g., Agilent Q-TOF vs. Thermo Orbitrap). The instructions below are provided for the use of Agilent Q-TOF 6545. The user should be trained by the professional staff at the core facility before operating the instrument independently.
27. Open the Agilent MassHunter Workstation Data Acquisition software (v.10.1), which is the user platform for implementing the subsequent steps.
28. Assemble a set of column and guard column of the same composition (e.g., both are C18 or Amide) to each analytical method, following manufacturer’s instructions.
29. Load a column securely attached to a guard column on the column compartment.
CRITICAL STEP Ensure the column is connected properly and check for leaks from both ends of the column once connected. Leaky column reduces the column pressure and often terminates the run as leaked solvent accumulates in the column compartment.
30. Connect the mobile phase solvent bottles to the corresponding “A” and “B” solvent positions of the LC unit.
31. Select “Tune” in “Context” at the top left corner to calibrate the instrument at either the positive or the negative mode, depending on the method to run next (Fig. 7a-b for example tune settings). The Q-TOF instrument uses the Agilent tuning mix, which contains a collection of reference ions for tuning both positive and negative modes.
32. Select “Acquisition” in “Context” to return to the home page to set up the analysis run.
33. Create a worklist including the following required information: Sample Name (e.g., “mouse_serum_1”), Sample Position (e.g., “P1-A01”), Method (file path to the method file), Data File (file path to the folder storing the new data), Sample Type (e.g., “Sample” by default), and Injection Volume (e.g., “As Method”).
34. Introduce five blanks at the beginning of each worklist. The blank sample containing 50% methanol can be either stored in a 2 mL Agilent glass vial or an empty well on the same plate as the biological samples. The first three blanks are used for removing the residual contaminants in the system while equilibrating the column with fresh solvents. While the first three blanks often contain contaminants from the previously injections, the fourth and fifth blanks can be used to assess background noise intrinsic to the solvent mix.
35. Add one or more QC samples at the beginning, middle, and the end of the full worklist. This is to assess any intra-experimental shift in retention time.
36. Load each 96-well 1 mL sample plate onto the autosampler. The autosampler should be kept at 4oC throughout the run to ensure sample integrity and to minimize sample evaporation.
37. Select an analytical method (see “Equipment setup”) and click “run” to initiate the worklist.
38. As the instrument progresses through the worklist, each newly generated .d raw file populates the folder previously designated in the “Data File” column of the worklist.
Data analysis using the MS-DIAL software
Timing 0.5 d
39. We follow the workflow as described in the MS-DIAL documentation (https://mtbinfo-team.github.io/mtbinfo.github.io/) with the following specifications. We use the MS-DIAL version 3.83 that is downloadable via the archive folder.
40. Convert raw data files to Analysis Base File (.abf) format using an open source Abf Converter (https://www.reifycs.com/AbfConverter/). Depending on the mass spectrometry instrument, the raw data file types may vary. For example, Agilent Q-TOF generates .d files, and Thermo Orbitrap generates .raw files. Regardless of instruments, the final converted file(s) should have the .abf file extension before subsequent analysis in MS-DIAL.
41. Create a new project in MS-DIAL in “New Project Window” by specifying the file path containing all ABF files, and selecting the following options (Fig. 8a):
a. Ionization type: “Soft ionization”
b. Separation type: “Chromatography”
c. MS method type: “Conventional LC/MS or data dependent MS/MS”
d. Data type (MS1): “Centroid data” or “Profile” (depending on how data are collected on the mass spectrometry instrument)
e. Ion mode: “Positive ion mode” or “Negative ion mode” (depending on the analytical method used for data collection)
f. Target omics: Metabolomics
42. Designate each file in the “Type” column in “New Project Window” as “Sample”, “QC”, or “Blank”. “QC” or quality control samples contain a pool of all samples used in an experiment (Fig. 8b), as described above. Select the specific set of files to analyze, and modify the column “Type” to define each file as “sample”, “blank”, or “QC”. The “QC” files will be used to generate a master list of molecular features for peak alignment of all samples. The “blank” files will be used to provide a list of molecular features present in the solvent background. Do not load the first three blanks at the beginning of an experiment, as they are used for cleaning purposes.
43. Load a parameter settings file (.med file extension) containing recommended parameters (source data: c18positive_settings.med, c18negative_settings.med, or hilicpositive_settings.med) by clicking on the “Load” button (Fig. 9a). This will enable MS-DIAL to retrieve and auto-populate our recommended parameters specific to each method (Fig. 9a-g), using the C18 positive parameters as an example). The parameter settings provided in this protocol have been used in analyzing data generated by an Agilent Q-TOF instrument and are broadly applicable to data analysis generated via other instruments.
44. In the “Identification” tab, specify a reference library used for compound annotation based on the m/z and RT information. Upload a library file (e.g., the one created at the end of step 11). in the “Identification” tab, under “Advanced: Text file and post identification (retention time and accurate mass based) setting”.
CRITICAL STEP The reference library file needs to be reloaded for each new analysis and is not included in the recommended setting file. The choice for the library file depends on the analytical method used to collect the data.
45. Prior to analyzing the full dataset using the full library, carry out a preliminary analysis using a small number of samples from the full experiment (e.g., several QCs collected throughout an experiment). This preliminary analysis gauges the retention time shifts of the internal standards detected in the current experiment compared to those reported in the reference library. For the reference library, use one of the internal standard libraries specific to each analytical method. These libraries will instruct the software to only annotate peaks for the matching internal standards. If all internal standards of the experiment fall within +0.2 minute from the library’s retention time, proceed to analyze the full data using the full library.
? TROUBLESHOOT (see Supplementary Table 3)
46. Export data by selecting “Alignment result export” (Fig. 9h). Users can select different output files based on their preferences. The most common output is “Raw data matrix (Area),” where the area under the curve of each peak is used to quantify ion abundance of each metabolite. Export format as a .txt file, which can be opened as an excel spreadsheet. The MS-DIAL analysis generates a list of m/z, RT, and ion counts (area under the curve) for 1) annotated metabolites (matched to the reference library based on based on m/z and RT) and 2) detected but unknown molecular features (those not matched to the reference library).
47. Manually filter each set of aligned peaks corresponding to each annotated metabolite in MS-DIAL (Fig. 10a). Select the following settings in the graphic user interface: 1) the analysis file name in “Alignment navigator,” 2) “Identified” option in the “Peak spot navigator”, 3) “EIC of aligned spot” in the top middle window, and 4) “Alignment spot viewer” in the bottom middle window.
48. Visually inspect each aligned peak shape and the area under the curve. Remove the entire metabolite data row from the alignment result excel spreadsheet (see Step 45) for the problematic peaks: 1) odd peak curvature resulting in only a subset of the peak being counted for the area under the curve (Fig. 10b), or 2) the peak is only detected in the blank controls but not in the samples.
Data analysis using the bioinformatics workflow
One code repository (https://github.com/the-han-lab/Han_et_al_Metabolomics_Protocol_2022) can be downloaded to provide an example workflow (Fig. 2) for integrating MS-DIAL data output with sample metadata within and across experiments. Detailed documentation for input files and code logic are explained in detail in data_analysis.ipynb.