Extraction and analysis of dry seed metabolites
The dry seed metabolites were extracted using the method as previously described by Roessner et al., (2000) with small changes. In short, 10 mg seeds of each tomato line was homogenized using a micro dismembrator (Sartorius) in a precooled 2 ml Eppendorf tube with 2 iron balls (2.5 mm). A solution of 700 µl methanol/chloroform (4:3) together with a standard (0.2 mg/ml ribitol) was added to each Eppendorf tube and mixed thoroughly. After 10 minutes of sonication 200 µl Milli-Q water was added to the samples followed by vortexing and centrifugation (5 min, 13,500 rpm). Then, the methanol phase was collected and transferred to a new 2 ml tube and the remaining organic phase was extracted again with 500 µl methanol/chloroform. The solution was kept on ice for 10 minutes and afterwards 200 µl Milli-Q water was added. Again after vortexing and centrifugation (5 min, 13,500 rpm), the methanol phase was collected and combined with the former collected phase and mixed well. A solution of 100 µl of this mix was transferred to a glass vial and dried overnight using a speedvac centrifuge at 35°C (Savant SPD1211).
The gas chromatography-time of flight-mass spectrometry (GC-TOF-MS) method which was previously described by Carreno-Quintero et al., (2012) was used for analysis of the dry seed metabolites. Detector voltage was set at 1600 V. Analysis of the raw data was performed using chromaTOF software 2.0 (Leco instruments). Furthermore, the Metalign software was used for further analysis such as aligning the mass signals (Lommen 2009). The peak threshold for noise was set to 2 and the output was loaded in Metalign Output Transformer (METOT; Plant Research International, Wageningen) and MSClust (Tikunov et al. 2012) was used to construct Centrotypes. The Centrotypes were identified by matching the mass spectra to an in-house-constructed library, to the GOLM metabolome database (http://gmd.mpimp-golm.mpg.de/) and to the NIST05 library (National Institute of Standards and Technology, Gaithersburg, MD, USA; http://www.nist.gov/srd/mslist.htm). The identification was based on spectral similarities and comparing the retention indices calculated by a third order polynomial function (Strehmel et al. 2008).