Quantitative profiling of carbonyl metabolites directly in crude biological extracts using chemoselective tagging and nanoESI-FTMS

Abstract

The extensive range of chemical structures, wide range of abundances, and chemical instability of metabolites present in the metabolome pose major analytical challenges that are difficult to address with existing technologies. To address these issues, one approach is to target a subset of metabolites that share a functional group, such as ketones and aldehydes, using chemoselective tagging. Here we report a greatly improved chemoselective method for the quantitative analysis of hydrophilic and hydrophobic carbonyl-containing metabolites directly in biological samples. This method is based on direct tissue or cells extraction with simultaneous derivatization of stable and labile carbonylated metabolites using N-[2-(aminooxy)ethyl]-N,N-dimethyl-1-dodecylammonium (QDA) and ¹³CD₃ labeled QDA. We combined innovations of direct quenching of biological sample with frozen derivatization conditions under the catalyst N,N-dimethyl-p-phenylenediamine, which facilitated the formation of oxime stable-isotope ion pairs differing by m/z 4.02188 while minimizing metabolite degradation. The resulting oximes were extracted by HyperSep C8 tips to remove interfering compounds, and the products were detected using nano-electrospray ionization interfaced with a Thermo Fusion mass spectrometer. The quaternary ammonium tagging greatly increased electrospray MS detection sensitivity and the signature ions pairs enabled simple identification of carbonyl compounds. The improved method showed the lower limits of quantification for carbonyl standards to be in the range of 0.20-2 nM, with linearity of R² > 0.99 over 4 orders of magnitude. We have applied the method to assign 66 carbonyls in mouse tumor tissues, many of which could not be assigned solely by accurate mass and tandem MS. Fourteen of the metabolites were quantified using authentic standards. We also demonstrated the suitability of this method for determining ¹³C labeled isotopologues of carbonyl metabolites in ¹³C₆-glucose-based stable isotope-resolved metabolomic (SIRM) studies.

Figure 1.

Chemical reaction scheme between QDA/*QDA and carbonyls. DMP: N,N-Dimethyl-p-phenylenediamine

Figure 2.

UHR FT-MS analysis results of the adducts of QDA with a mixture of 14 carbonyl standard compounds under different conditions. The ordinate is normalized peak intensity from the UHR FT-MS as defined Methods. The reactions conditions are described in the method section. Blue bar: frozen at −80 °C for 18 h without DMP. Red bar: reaction at 40 °C for 23 h without DMP. Black bar: frozen at −80 °C for 18 h with DMP.

Figure 3.

Time-dependent reaction efficiency of QDA with carbonyl standards. Mixed standard solutions were reacted with QDA/*QDA using d₆-acetone as the internal standard in the presence of 0.2 mM DMP at −80 °C for different time period, followed by UHR FT-MS analysis.

Figure 4.

The extraction recovery of carbonyl standard compounds using HyperSep C8 tips. Standards were derivatized with QDA and the amount recovered after extraction using the known input was determined as described in the Methods. The recovery was determined at both the lower limit of quantification (LLOQ) and at the upper limit of quantification (ULOQ) as defined in Table 1. Black bars: LLOQ sample; blue bars: ULOQ sample.

Figure 5.

Atom-resolved map of ¹³C₆-glucose-derived carbonyl metabolites, their abundances relative to total protein, and their ¹³C fractional enrichment in A549 cells; the latter is corrected for natural abundance contributions. A549 cells were grown in the presence of ¹³C₆-Glc for 24 h before polar metabolites were extracted as described in the Methods. The detected carbonyl metabolites are labeled with blue numerals (indicating the atom number). Solid arrows indicate single-step reactions; dashed arrows indicate multiple reaction steps. All reactions are of course bidirectional, but single arrows are use to illustrate the observed net flow of ¹³C. The * indicate metabolites with mass isomers that are not resolved under the current methods. The levels of major isotopologues were determined using the calibration curves constructed with unlabeled standards. Unless indicated otherwise, the levels are expressed as nmol/mg protein (n = 3). : ¹²C; , : ¹³C from pyruvate dehydrogenase and pyruvate carboxylase-initiated Krebs cycle reactions, respectively; N1’: 1’-ribose attached to the nicotinamide ring.

Figure 6.

Example application of the QDA method to untargeted analysis of carbonyl metabolites in patient-derived mouse tumor xenograft (PDTX). A mouse PDTX extract was reacted with QDA/*QDA at −80°C as described in the Methods. Full-scan UHR FT-MS analysis of the derivatized extract and blank sample (A) showed sample-specific ion pairs at m/z 447/451 with the QDA/*QDA pair signature of Δm/z = 4.02188, which signals the presence of a carbonyl group and yielded the molecular formula C₆H₈O₇ for the metabolite. Search of this formula against the METLIN database gave five carbonyl metabolites. MS/MS analysis of the precursor ion pairs (B) reduced the number of degenerate assignments to 2-dehydro-3-deoxy-D glucarate and/or 5-dehydro-4-deoxy-D glucarate (DDG).

Figure 7.

UHR FT-MS profile of ¹³C labeled QDA adduct ion of DDG (m/z 447). A: full scan MS spectrum with expanded region to show the resolution of M+4 QDA adduct from the M+0 *QDA adduct. B: fractional distribution of ¹³C labeled isotopologues of DDG after natural abundance correction.