While characterizing the methyltransferome of S. cerevisiae, we identified and characterized TMT1. While we showed that this is the yeast homolog of the E. coli trans-aconitate methyltransferase, we also showed that this protein was responsible for methylating an alternate, more abundant, small molecule substrate. We decided to apply techniques of metabolomics to solve two problems in identifying this alternate substrate: The analysis of complex biological samples for a small, yet significant, difference and the identification of the molecule responsible for that difference.

We performed GC/MS profiling of crude cell extracts after they were acidified, ethyl acetate extracted and BSTFA derivatized. In order to compare the GC/MS profiles, we developed a software package (COMSPARI, http://www.biomechanic.org/comspari/) that allows for the rapid comparison of two datasets. Using this software we were readily able to see differences between the wild-type and knockout samples in both an in vitro labeled, anion exchange fractionated extract as well as in an in vivo unlabeled, unfractionated extract!

Using AMDIS, an EI spectra corresponding to the product of interest was abstracted from the dataset. There was no match to this spectrum in the NIST '02 EI fragmentation database. CI measurement with methane as the reagent gas yielded a peak pattern consistent with an M+H, M+C₂H₅, M+C₃H₅ pattern for a parent mass of 334. Comparison of spectra between BSTFA and MTBSTFA derivatized extracts suggested that there were two sites of derivatization. Finally, we performed an exact mass measurement of the compound. We combined this information with our chemical knowledge of the compound to predict a molecular formula of the underivatized product as C₈H₁₄O₅. This formula, combined with manual analysis of underivatized library spectra guided the organic synthesis of target compounds. We found that the BSTFA derivative of 3-isopropylmalate has the same EI spectrum and GC elution time as the BSTFA derivative of our unknown substrate. In order to determine which of the two possible sites is being methylated, we combined synthetic 3-isopropylmalate, crude lysate from a TMT1 overexpression strain and [3H]-methyl-S-adenosylmethionine. We then purified the product on a silica anion exchange column. HMBC NMR analysis of the purified compound shows that the methylation is on the one position.

The known source of 3-isopropylmalate in the cell is the leucine biosynthetic pathway. We have tried to show the interaction of TMT1 with the leucine pathway directly by assaying S. cerevisiae strains in which the LEU4, LEU9, and LEU1 enzymes have been knocked out. These are the known biosynthetic source of alpha ketoisovalerate and 2-isopropylmalate, consecutive precursors which are converted to 3-isopropylmalate BY leu1. However, we we are still able to show endogenous 3-isopropylmalate in these cells suggesting an alternate source of this compound in vivo!

Also, the leucine biosynthetic pathway shares similar intermediates and reaction chemistry with the portion of the citric acid cycle from oxaloacetate to alpha-ketoglutarate via cis-aconitate. While trans-aconitate is an inhibitor of the citric acid cycle (the methylation of which attenuates this effect), the analogous substrate in leucine biosynthesis is isopropylfumarate. TMT1 has poor activity towards isopropylfumarate yet the activity towards trans-aconitate and 3-isopropylmalate is comparable. This suggests that TMT1 is possibly in the class of "moonlighting" enzymes, performing detoxification in the citric acid cycle and possibly providing a branch point off of leucine biosynthesis into a novel biochemical pathway!