101
Today, such screening of unknowns is done using data-dependent analysis [27,
28]. This technique, routinely used in global proteomics as well [29], produces fragment ion spectra of as many chemicals as possible that elute off the column at a
given moment. The information thus obtained includes (i) retention time, related to
the substance’s partitioning into the solid phase of the separation column, (ii) accurate mass of the compound, (iii) the relative intensities of its isotopes and (iv) structural information through its fragments [30]. A differential analysis, e.g. using
Compound Discoverer (ThermoFischer Scientific Inc., San Jose, CA, USA), then
compares compounds that are present at significantly different concentrations in the
active relative to an inactive sample and assignes statistical significance to the differences observed. Filtering for fold-change and p-value produces a list of features
that could potentially be responsible for the endocrine activity detected in the sample. In the case of the Lake Thun algal extracts we found a compound that was present at very high concentration in the estrogenic sample with a mass of 180.1020 Da.
The first step in identifying this unknown feature is to assign a sum formula. The
Fiehn group has established rules that allow to narrow down the sum formula of a
given feature, based on its accurate mass and isotopes [31]. They mined public mass
spectral databases (Natural Products Data Base, Wiley) and determined the ratios of
elements to carbon (elements: H, F, Cl, Br, N, O, P, S, Si) for all entries and thus
likely ranges for unknown sum formulas (covering 99.7% of the entries). Of course
the nitrogen rule from classical EI spectral interpretation also holds, only that in
positive ESI, even ions carry an additional proton, thus they have an uneven number
of nitrogens, while odd ions have zero or even numbers of nitrogens. With a 3 ppm
mass accuracy, easily obtained on current state-of-the-art MS, 64 candidate sum
formulas match a mass of 500 Da [31]. If the isotopic abundance is taken into
account (5%), the number of candidates is reduced to 3 only. Using the rules mentioned above, we determined the sum formula to be C 10 H 14 O 2 N, with a mass difference of 0.526 ppm, well within the accuracy obtainable on an Orbitrap. All elemental
ratios were within the range defined by Kind and Fiehn. The double bond and ring
equivalents (DBE) were determined to be 4.5, i.e. non-integer, which indicated that
this ion contains an even number of electrons (M + H
+
) and that it could be a phenolic compound (1 ring plus 3 double bonds), similar to the most important estrogenic
compounds (estrone, estriol, estradiol, ethinylestradiol, nonylphenol and bisphenol
A). The next best candidate (C 6 H 17 O 2 N 2 P) had a DBE of only 0.0 with a large mass
difference of −1.198 ppm, and hence could be discarded.
The MS/MS spectrum, which had been acquired in a data-dependent way,
showed two main fragments at m/z 138 and 121. A likely candidate that, according
to the fragmentation software Mass Frontier (ThermoFischer Scientific Inc., San
Jose, CA, USA), would produce these fragments, was found to be N-acetyltyramine
(CAS 1202-66-0). This molecule was then synthesized and analyzed by MS. The
fragments could be experimentally verified, but a test for estrogenicity in the YES
returned a negative response, suggesting that N-acetyltyramine could not be the
compound responsible for the estrogenic response observed in the positive sample
(Fig. 6.6).
6 Mass Spectrometry in Ecotoxicology
Today, such screening of unknowns is done using data-dependent analysis [27,
28]. This technique, routinely used in global proteomics as well [29], produces fragment ion spectra of as many chemicals as possible that elute off the column at a
given moment. The information thus obtained includes (i) retention time, related to
the substance’s partitioning into the solid phase of the separation column, (ii) accurate mass of the compound, (iii) the relative intensities of its isotopes and (iv) structural information through its fragments [30]. A differential analysis, e.g. using
Compound Discoverer (ThermoFischer Scientific Inc., San Jose, CA, USA), then
compares compounds that are present at significantly different concentrations in the
active relative to an inactive sample and assignes statistical significance to the differences observed. Filtering for fold-change and p-value produces a list of features
that could potentially be responsible for the endocrine activity detected in the sample. In the case of the Lake Thun algal extracts we found a compound that was present at very high concentration in the estrogenic sample with a mass of 180.1020 Da.
The first step in identifying this unknown feature is to assign a sum formula. The
Fiehn group has established rules that allow to narrow down the sum formula of a
given feature, based on its accurate mass and isotopes [31]. They mined public mass
spectral databases (Natural Products Data Base, Wiley) and determined the ratios of
elements to carbon (elements: H, F, Cl, Br, N, O, P, S, Si) for all entries and thus
likely ranges for unknown sum formulas (covering 99.7% of the entries). Of course
the nitrogen rule from classical EI spectral interpretation also holds, only that in
positive ESI, even ions carry an additional proton, thus they have an uneven number
of nitrogens, while odd ions have zero or even numbers of nitrogens. With a 3 ppm
mass accuracy, easily obtained on current state-of-the-art MS, 64 candidate sum
formulas match a mass of 500 Da [31]. If the isotopic abundance is taken into
account (5%), the number of candidates is reduced to 3 only. Using the rules mentioned above, we determined the sum formula to be C 10 H 14 O 2 N, with a mass difference of 0.526 ppm, well within the accuracy obtainable on an Orbitrap. All elemental
ratios were within the range defined by Kind and Fiehn. The double bond and ring
equivalents (DBE) were determined to be 4.5, i.e. non-integer, which indicated that
this ion contains an even number of electrons (M + H
+
) and that it could be a phenolic compound (1 ring plus 3 double bonds), similar to the most important estrogenic
compounds (estrone, estriol, estradiol, ethinylestradiol, nonylphenol and bisphenol
A). The next best candidate (C 6 H 17 O 2 N 2 P) had a DBE of only 0.0 with a large mass
difference of −1.198 ppm, and hence could be discarded.
The MS/MS spectrum, which had been acquired in a data-dependent way,
showed two main fragments at m/z 138 and 121. A likely candidate that, according
to the fragmentation software Mass Frontier (ThermoFischer Scientific Inc., San
Jose, CA, USA), would produce these fragments, was found to be N-acetyltyramine
(CAS 1202-66-0). This molecule was then synthesized and analyzed by MS. The
fragments could be experimentally verified, but a test for estrogenicity in the YES
returned a negative response, suggesting that N-acetyltyramine could not be the
compound responsible for the estrogenic response observed in the positive sample
(Fig. 6.6).
6 Mass Spectrometry in Ecotoxicology