Substantial equivalence (metabolomics) experiments
Substantial equivalence (metabolomics) experiments
Modern "metabolomics" methods can be used to automatically compare the levels of a number of structurally different compounds in a large number of samples. The technique is well suited for screening plant populations, including the detection of unintended effects caused by transgenes. Several metabolomics methods are available for substantial equivalence analysis. We developed a metabolomics method for screening plants using [ 1H]-NMR fingerprint recognition. The source of this experiment is "A Guide to Transgenic Technology and Field Identification Experiments in Wheat Crops" [English] H.D. Jones P.R . H.D. Jones P.R. Hewlett, eds.
Operation method
Substantial equivalence (metabolomics)
Materials and Instruments
SIMCA - P Multivariate Statistical Software Move The methods described in this chapter are best suited for analyzing wheat semolina, but have been modified for whole flour, bran, or other tissues such as leaves and roots. Key steps to consider before conducting a substantially equivalent metabolomics experiment, as with any other field or greenhouse experiment, include field experiment design (see Note 1), tissue sampling (see Note 2 ), and sample labeling (see Note 3). For more product details, please visit Aladdin Scientific website.
[ 1H ] - NMR Extractor
Eppendorf Polypropylene Tubes NMR Tubes NMR Spectrometer
3.1 Metabolite extraction and NMR sample preparation
( 1 ) From each biological replicate of white flour, three 30 mg (± 0.03 mg) samples were weighed into labeled 1.5 ml Eppendorf tubes. Both biological and technical replicates are randomized throughout the experiment (see Note 4).
( 2 ) Add 1 ml of NMR extractant (see above) and cap the tube.
( 3 ) Vortex the contents of the tube until the flour is completely dispersed (usually about 30 s) (see Note 5).
( 4 ) Heat the centrifuge tube accurately for 10 min at (50±1)C. This can be accomplished using a polystyrene float and a preheated water bath. The tube should be placed so that the contents are under the surface of the water bath.
( 5 ) After removal from the water bath, quickly transfer the tube to a microcentrifuge and centrifuge for 5 min.
( 6 ) Take 800 μl of supernatant from each tube and add to a clean, labeled 1.5 ml polypropylene tube.
( 7 ) Heat-stimulate the solution at (90±2)°C for 2 min (see Note 6), using a preheated water bath as described previously.
( 8 ) After removing the float from the water bath, the centrifuge tube is quickly transferred to a tube rack and placed in cold storage (4°C) for 45 min at this temperature.
( 9 ) While the samples were kept cold, the microcentrifuge was centrifuged at full speed for 5 min.
( 10 ) Take 0.70 ml of the supernatant and add it to a clean labeled 5 mm thin-walled NMR tube, capped and ready for analysis (see Note 7).
3.2 NMR Data Collection
( 1 ) Place the NMR tubes in the NMR autosampler rack and record their positions in the instrument workbook.
( 2 ) Confirm that the Variable Temperature Unit in the NMR Spectrometer is set to 300 K.
( 3 ) Enter the sample details in the sample list of the automated program, making sure to accurately enter the appropriate sample labels. For each sample entered, D2O is selected as the sample solvent, the number of scans required is 1024, and the parameter is set to WATERSUP (see Notes 8~10).
( 4 ) Start the automatic program. The NMR software then automatically loads each sample into the NMR magnet, searches for the D2O signal and locks on to it, optimizes the signal strength (through an automatic homogenization procedure), sets the receiver gain, and collects the NMR data. At the end of data collection, the NMR automated program routinely and automatically processes the data and saves the file before moving on to the next sample (see Note 11).
3.3 NMR data: visual inspection and quality assurance
( 1 ) At the end of the entire experiment, open the AMIX ( Analysis of Mixtures, Bmker Biospin, Germany) software and select File > Open from the X Win NMR command. enter the appropriate file location for storing the experimental data file in the pop-up window and click "OK ".
( 2 ) Select all sample records for the entire experiment in the pop-up box and click "OK".
( 3 ) All NMR spectra should be displayed in the main window at this point. Check the files to make sure you have satisfactory data (see Note 12).
( 4 ) Re-identify any samples with unsatisfactory data from step 3 (see Note 13).
( 5 ) When the data have been collected and quality assessed (see Note 14), remove the samples from the NMR tubes and transfer them to screw-cap glass bottles. Store these bottles in the refrigerator in case they are needed for future analysis.
3.4 Data processing, database and spectral storage
Before analyzing the data in the data package, further processing of the data was required. In this step, the preparation of the data in the Bmker NMR spectral database (SBase) (part of AMIX software) was included. The reason for this step is to ensure high comparability of the datasets and to reduce the complexity of the data by converting the 32k data points into a matrix containing about 1k data points. The process of "bucketing" eliminates the numerical calibration problems caused by displacement differences, as small differences in sample pH can lead to weak chemical shift differences in some signals.
( 1 ) Open the AMIX (Analysis of Mixtures, Bruker Biospin, Germany) software and select AMIX Tools > Prepare Data.
( 2 ) In the Prepare Data window, open the file to be added to SBase by selecting File > Open from the X Win NMR command. Enter the appropriate location for the experimental data file in the pop-up window and click "OK".
( 3 ) Select all records of the entire experiment in the pop-up box and click "OK".
( 4 ) All NMR spectra should be displayed in the main window at this time.
( 5 ) Zoom in on the d4 -TSP peak at 80.00. This is the internal standard peak. Set this peak to the maximum height (see Note 15).
( 6 ) Reset the zoom so that the entire width of the spectrum is rendered.
( 7 ) Select the batch function and choose the following options in the pop-up window:
( i ) Delete negative peaks;
( ii ) Use the user-defined range 810.0~10.2 for noise reduction.
( 8 ) Enter the sample name of the first sample (see Note 16).
( 9 ) Save this spectrum to the spectral database (SBase), close it, and enter the sample name of the next spectrum in the list.
( 10 ) After saving all data to SBase, close the Sample Preparation window and open the Buckets window ( AMIX Tools > Buckets).
( 11 ) Select "Create new bucket table" and choose "data from SBase" in the popup window.
( 12 ) Another window will pop up. Locate the bucket data to be bucketed.
( 13 ) Select in the Bucketing options window (see Note 17).
( i ) Range of data to be bucketed (89.5 to 0.5 ppm)
( ii ) Bucket width (0.01 ppm)
( iii ) Inclusion peaks (all positive peaks)
( iv ) Proportional Scaling Method (set scale by reference area)
( v ) Reference area (80.05~- 0.05)
( vi ) Exclusion (none)
( 14 ) Open the bucket table with a spreadsheet package such as Microsoft Excel.
( 15 ) Add additional row labels to aid in future data analysis (e.g., strain, treatment, time point).
( 16 ) Remove rows corresponding to residual water (84.775 to 4.865) and methanol (S3.285 to 3.335).
( 17 ) Save the file as an Excel worksheet.
3.5 Multivariate analysis (see Note 18)
( 1 ) Open the SIMCA-P software (other similar software packages are also available. The specific operation may be different, but the principle is the same).
( 2 ) Create a new project (nie > New) and select the analysis file (created in the previous step) (see Note 19).
( 3 ) Transpose the dataset (Commands > Transpose dataset ) so that chemical shifts are in columns and samples are in rows.
( 4 ) Highlight the first row of chemical shifts and select that row as the primary variable ID.
( 5 ) Highlight the column containing the sample name and select that column as the primary observation ID.
( 6 ) Highlight any descriptive columns and assign those columns as qualitative X data.
( 7) Continue to upload data. Once started, do not exclude any values (see Note 20).
( 8 ) Double-click on the data model and select the "workset" button.
( 9 ) Select the scaling option, highlight all the data and select the scaling method " ctr". This centers the data at zero by subtracting the mean.
( 10 ) Exclude any qualitative values from the PCA (Principal Component Analysis) model by selecting the Variable option and choosing a descriptive value. Exclude these entries.
( 11 ) Automatically fit and check the PCA model (see Note 21).
( 12 ) At this point the PCA score plots can be analyzed (see Note 22). Different component plots should be analyzed with different clustering patterns; PC1 must be tested against PC2 because these components often explain the most variation in the data set.
( 13 ) Use color-coded variables based on your descriptive information (strain, treatment, etc.) to make it easier to see trends in the data set.
( 14 ) For each score plot, two corresponding loadings plots should be generated (see Note 23) to characterize the metabolites that are responsible for the variation in the clusters.
( 15 ) Both positive and negative peaks in the load plot can be assigned values according to metabolite variation between clusters. This can be accomplished by comparing pure compounds that were run authentic under the same conditions (solvent, temperature, pulse program) using the same spectrometer and the NMR spectral libraries that were bucketed and collected under the same conditions using AMIX (see Note 24).
( 16 ) If clear clusters are not available, discriminant analysis can be performed. This involves setting categories to the dataset prior to modeling and using this information to "impose" differences in the dataset. The corresponding scores and loading plots can be observed as described earlier (see note 25).
( 17 ) In the process of generating the loadings diagrams, information is collected on the value points that are causing the differences. This only provides clues. It is also necessary to examine the raw data to confirm metabolite assignments and changes. All areas of the load map should be examined because in some cases, intensity changes in very small peaks can be more meaningful than small changes on very large peaks in the spectrum.
3.6 Determining Substantial Equivalence
PCA score plots enable rapid assessment of the degree of similarity and difference between samples. Load plots are used to identify the metabolites responsible for the differences. Currently, most studies [ 2, 3 ] consider differences between cultivars and differences in the environment to be more influential than transgenes. However, for imported metabolic enzymes or their regulators, the direct effect of the imported genes on specific metabolites should also be considered. There is a need for different PCA models to detect environmental effects on transgenic/non-transgenic crops.
If, after removing environmental and varietal differences, transgenic vs. non-transgenic score plots do not form independent sets, transgenics are considered substantially equivalent to non-transgenics within the limits imposed by the analytical technique. More advanced supervised models such as PLS or neural networks [ 8 ] can be used to further explore whether the data sets differ.