A Strategy for an Unknown Screening
Approach on Environmental Samples Using
HRAM Mass Spectrometry
Olaf Scheibner,1 Patrizia van Baar, 2 Florian Wode, 2 Uwe Dünnbier, 2 Kristi Akervik, 3 Jamie Humphrie, 3 Maciej Bromirski1
1
Thermo Fisher Scientific, Bremen, Germany; 2Berliner Wasserbetriebe, Berlin, Germany;
3
Thermo Fisher Scientific, Austin, TX, USA
Overview
FIGURE 2. Analytical gradien
Purpose: Run a general unknown screening approach in an automated fashion.
Methods: Surface water samples form the city of Berlin, Germany, were analyzed in
full scan/AIF mode with a Thermo Scientific™ Exactive™ Plus mass spectrometer and
analyzed in a widely automated workflow using Thermo Scientific™ TraceFinder™ 3.1
and Thermo Scientific™ SIEVE™ 2.1 software.
Results: Differences in the load of contaminants could be easily determined in the
different samples. Easy detection and identification of a significant number of
contaminants was achieved.
Introduction
The analysis of food and environmental samples for contaminants by LC-MS has
become a quick and cost-effective routine application when run in a targeted fashion,
but this method disregards events or circumstances not taken into account
beforehand. Run in a non-targeted fashion, this method is known to be laborious and
time-consuming, making it everything but a routine application. New-generation
software now links in quantitative and (unknown) screening approaches to one
smoothly integrated workflow, tying together component detection capabilities of
unknown screening workflows with the identification capabilities of targeted screening
and quantification software. Here we show how one data set can serve for routine high
throughput quantitative analysis and for versatile non-targeted investigations in a
highly automated manner.
Mass Spectrometry
Methods
Sample Preparation
Four samples of surface water from different sources were taken and analyzed without
any further treatment. In addition, one neat standard as a control sample and one tapwater sample as a reference sample were run in the same sequence.
Liquid Chromatography (or more generically, separations)
For online sample concentration and chromatographic separation, a Thermo
Scientific™ EQuan™ LC-MS system was used (Figure 1). A sample volume of
1000 µL was injected onto a Thermo Scientific™ Hypersil GOLD™ 20×2.1 mm
trapping column with subsequent elution onto a Thermo Scientific™ Accucore™
RP-MS 18 analytic column. A 6.7-minute solvent gradient was applied as shown in
Figure 2. This resulted in a total cycle time of 15 minutes for sample injection, online
concentration, and chromatographic separation.
FIGURE 1. Schematics of the EQuan online solid phase extraction and
separation system
Autosampler
Large
Volume
Sample Loop
For mass spectrometric detecti
run in full scan/all ion fragmenta
permanently alternated with AIF
(FWHM @ m/z 200) was used
(resp. m/z 70 to 870 and resolu
prepared for all possible contam
with the standard calibration mi
the instrument (compound tunin
FIGURE 3. Exactive Plus meth
Accela 600
Pump
6-Port
Valve
6-Port
Valve
3 µm Accucore RP-MS 18
Analytical Column
Conventional
Volume
Sample Loop
Accela 1250
Pump
6-Port
Valve
12 µm Hypersil GOLD
Preconcentration
Column
Data Analysis
For data analysis, SIEVE 2.1 an
acquired with TraceFinder softw
and then retransferred into Trac
reporting, and archiving took pla
2 A Strategy for an Unknown Screening Approach on Environmental Samples Using HRAM Mass Spectrometry
Results
FIGURE 2. Analytical gradient of the LC method
Suspect Screening
h in an automated fashion.
The more simple way of screening is
components possibly present in a sa
condition-free component detection,
which could serve for confirmation b
more. In this case, a built-in databas
containing name, elemental compos
matching spectral library containing
inside the application. As a result, iso
library search were used for result co
in, Germany, were analyzed in
e™ Plus mass spectrometer and
mo Scientific™ TraceFinder™ 3.1
d be easily determined in the
a significant number of
FIGURE 4. Suspect screening resu
ontaminants by LC-MS has
when run in a targeted fashion,
ot taken into account
od is known to be laborious and
plication. New-generation
ening approaches to one
ent detection capabilities of
apabilities of targeted screening
ata set can serve for routine high
targeted investigations in a
Mass Spectrometry
were taken and analyzed without
as a control sample and one tapame sequence.
arations)
c separation, a Thermo
e 1). A sample volume of
ersil GOLD™ 20×2.1 mm
mo Scientific™ Accucore™
dient was applied as shown in
tes for sample injection, online
For mass spectrometric detection, an Exactive Plus mass spectrometer was used and
run in full scan/all ion fragmentation (AIF) mode. In this mode, full scans are
permanently alternated with AIF fragmentation scans. A resolution setting of 70,000
(FWHM @ m/z 200) was used (Figure 3). A mass range of m/z 103 to 900 was applied
(resp. m/z 70 to 870 and resolution setting 35,000 FWHM for the AIF scans) to be
prepared for all possible contaminants. The mass axis of the system was calibrated
with the standard calibration mix once prior to measurement. Further optimization of
the instrument (compound tuning) was not required.
As to be expected, it was possible to
a match on all three confirming point
of screening did not cover all possibl
standard (as used normally for targe
same batch. A clear benefit could be
library spectra were present for addi
FIGURE 3. Exactive Plus method setup
In addition, a larger database with 2,
the question of contaminants not fou
phase extraction and
Port
alve
Autosampler
Conventional
Volume
Sample Loop
FIGURE 5. Three stages of confirm
match, fragment search, and libra
B: Fragment overlay; C: Library co
Accela 600
Pump
Accela 1250
Pump
Port
alve
12 µm Hypersil GOLD
Preconcentration
Column
Data Analysis
For data analysis, SIEVE 2.1 and TraceFinder 3.1 software were used. All data
acquired with TraceFinder software was transferred automatically to SIEVE software,
and then retransferred into TraceFinder after processing in SIEVE. Final processing,
reporting, and archiving took place in TraceFinder software.
Thermo Scientific Poster Note • PN ASMS13_T578_OSheibner_E 07/13S 3
Results
hod
Unknown Screening
Suspect Screening
The more simple way of screening is the suspect screen, using a large list of
components possibly present in a sample (Figure 4). It avoids the critical step of
condition-free component detection, but works already without analytical standards
which could serve for confirmation by providing valid retention time, ion ratios, and
more. In this case, a built-in database with about 1,000 components was used,
containing name, elemental composition, and fragment information. Additionally, a
matching spectral library containing roughly 4,000 HR/AM MS2 spectra is available
inside the application. As a result, isotopic pattern match, fragment search, and MS2
library search were used for result confirmation.
FIGURE 4. Suspect screening result view
As a consequence of the limita
workflow was run. For this, the
application, SIEVE, for uncond
settings and parameters were
software automatically, the co
immediately. As a result, 5,000
all components regardless of t
needed. As part of the process
sample, so a simple filter could
from the result list, leaving 1,82
component analysis to this res
related, while one water sampl
different in its content, so the f
FIGURE 6. PCA result after fi
n
Plus mass spectrometer was used and
In this mode, full scans are
cans. A resolution setting of 70,000
s range of m/z 103 to 900 was applied
00 FWHM for the AIF scans) to be
ss axis of the system was calibrated
easurement. Further optimization of
red.
surface water 2 surface water 4
surface water 3
As to be expected, it was possible to identify a good number of contaminants, yielding
a match on all three confirming points. On the other hand, it was clear that this method
of screening did not cover all possible compounds, as was visible from the neat
standard (as used normally for target screening on these samples) measured in the
same batch. A clear benefit could be seen in the fact that fragment information and
library spectra were present for additional confirmation (Figure 5).
In addition, a larger database with 2,900 components was applied, but still left open
the question of contaminants not found because they may not be members of this list.
FIGURE 7. Confirmation of th
propiconazole taken as an e
determined retention time gi
isotope pattern match shows
FIGURE 5. Three stages of confirmation in suspect screen: isotope pattern
match, fragment search, and library search; A: Isotopic pattern overlay;
B: Fragment overlay; C: Library comparison
A
C
.1 software were used. All data
red automatically to SIEVE software,
ocessing in SIEVE. Final processing,
er software.
4 A Strategy for an Unknown Screening Approach on Environmental Samples Using HRAM Mass Spectrometry
This time the filter was set to lo
water 1 and surface water 2 (F
which were sent to the ChemS
1,529 identifications. Closing th
result list back to the TraceFind
database.
B
Unknown Screening
t screen, using a large list of
e 4). It avoids the critical step of
ready without analytical standards
valid retention time, ion ratios, and
1,000 components was used,
gment information. Additionally, a
0 HR/AM MS2 spectra is available
n match, fragment search, and MS2
As a consequence of the limitations of a suspect screen, an unknown screening
workflow was run. For this, the measured sequence was transferred to the screening
application, SIEVE, for unconditioned component detection. Since all necessary
settings and parameters were transferred from TraceFinder software to SIEVE
software automatically, the component detection process could be started
immediately. As a result, 5,000 components were detected. Since such a list contains
all components regardless of their significance, refinement of this list was clearly
needed. As part of the process, all samples were referenced against the tap-water
sample, so a simple filter could be applied to remove matrix and background signals
from the result list, leaving 1,829 components in the list. Application of a principal
component analysis to this result revealed that three water samples were closely
related, while one water sample (surface water 1, see Figure 6) seemed to be rather
different in its content, so the filter for significant components could be further refined.
FIGURE 6. PCA result after filtering for significant differences
For confirmation and reporting
normal suspect screening. The
TraceFinder software was to be
screen – in one application and
templates.
It became visible that some com
but it was still possible to extrac
maintaining full mass accuracy
example of the component Lox
roughly the same intensity as th
Still, the analyte signals are cle
so the compound can easily be
analyte and matrix signals is th
this analysis.
FIGURE 8. Importance of suff
components: the monoisotop
surrounded by matrix signals
means of the high resolving p
neat standard
surface water 2 surface water 4
surface water 3
surface water 1
ood number of contaminants, yielding
her hand, it was clear that this method
ds, as was visible from the neat
on these samples) measured in the
fact that fragment information and
mation (Figure 5).
nents was applied, but still left open
they may not be members of this list.
tap water
This time the filter was set to look only for significant changes in the samples surface
water 1 and surface water 2 (Figure 7). This reduced the list of components to 1,671,
which were sent to the ChemSpider database for identification. This search returned
1,529 identifications. Closing the SIEVE application automatically transferred this
result list back to the TraceFinder software, where it was imported as a new compound
database.
FIGURE 7. Confirmation of the unknown screening results from SIEVE software,
propiconazole taken as an example: the extracted ion chromatogram at the
determined retention time gives a clear signal free from interferences, the
isotope pattern match shows close to perfect overlay
spect screen: isotope pattern
A: Isotopic pattern overlay;
B
Since all final processing was d
and unknown screening could e
reporting and archiving one sin
applications is fully automated,
not been part of the initial targe
process.
Table 1. Selection of addition
suspect screen
Compound Name
Bisoprolol
Candesartan
Carbofuran
Dibenzylamine
Irbesartan
Loxoprofen
Mexacarbate
Oxazepam
Propiconazole
Tramadol
Formula
m/
C18H31NO4
3
C24H20N6O3
4
C12H15NO3
2
C14H15N
1
C25H28N6O
4
C15H18O3
2
C12H18N2O2
2
C15H11ClN2O2
2
C15H17Cl2N3O2 3
C16H25NO2
2
Conclusion
In this example of environmenta
the capabilities of target and su
general unknown screening wit
application. The resolving powe
the driving force behind the sele
this serves for the separation o
signals.
All used trademarks are the property of T
This information is not intended to encou
intellectual property rights of others.
Thermo Scientific Poster Note • PN ASMS13_T578_OSheibner_E 07/13S 5
screen, an unknown screening
ce was transferred to the screening
detection. Since all necessary
aceFinder software to SIEVE
process could be started
detected. Since such a list contains
efinement of this list was clearly
referenced against the tap-water
ove matrix and background signals
he list. Application of a principal
ree water samples were closely
, see Figure 6) seemed to be rather
components could be further refined.
cant differences
surface water 1
For confirmation and reporting of the results, this compound database was used for a
normal suspect screening. The advantage of looping the results back to the
TraceFinder software was to be able to handle all data – target, suspect, and unknown
screen – in one application and to be able to use the same data review and report
templates.
It became visible that some components were coeluting with higher amounts of matrix,
but it was still possible to extract significant signals from the surrounding matrix,
maintaining full mass accuracy despite the low signal intensity. Figure 8 shows the
example of the component Loxoprofen, where the surrounding matrix signals have
roughly the same intensity as the first and second isotope signal of the compound.
Still, the analyte signals are clearly resolved from the background and matrix signals,
so the compound can easily be detected and confirmed. Key to this clear separation of
analyte and matrix signals is the high resolving power of R = 70000 @ m/z 200 used in
this analysis.
FIGURE 8. Importance of sufficient resolution for unambiguous identification of
components: the monoisotopic signal (A) and the first isotope signal (B) are
surrounded by matrix signals of similar intensity, which are only separated by
means of the high resolving power used
tap water
ant changes in the samples surface
ced the list of components to 1,671,
identification. This search returned
on automatically transferred this
e it was imported as a new compound
ening results from SIEVE software,
cted ion chromatogram at the
free from interferences, the
overlay
Since all final processing was done in one application, the results of target, suspect,
and unknown screening could easily be combined into one result, making result
reporting and archiving one single step. Since all data transfer between the two
applications is fully automated, Table 1 shows a short selection of compounds that had
not been part of the initial target screening, but were found in the unknown screening
process.
Table 1. Selection of additional contaminants not present in previous target and
suspect screen
Compound Name
Bisoprolol
Candesartan
Carbofuran
Dibenzylamine
Irbesartan
Loxoprofen
Mexacarbate
Oxazepam
Propiconazole
Tramadol
Formula
m/z (Apex) m/z (Delta (ppm)) RT (Measured) Isotopic Pattern Score (%)
C18H31NO4
326.2330
0.57
5.12
100
C24H20N6O3
441.1671
-0.50
6.56
100
C12H15NO3
222.1127
-0.19
5.18
98
C14H15N
198.1277
-0.66
7.31
98
C25H28N6O
429.2401
-0.03
6.45
100
C15H18O3
247.1332
0.45
5.52
85
C12H18N2O2
223.1443
-0.06
5.53
96
C15H11ClN2O2
287.0584
0.48
6.29
96
C15H17Cl2N3O2 342.0774
0.21
7.43
89
C16H25NO2
264.1961
0.10
4.35
100
Conclusion
In this example of environmental analysis. we could show that it is possible to enhance
the capabilities of target and suspect screening with its limitations by a streamlined
general unknown screening with a high degree of automation from within one
application. The resolving power of the Exactive Plus benchtop Orbitrap MS system is
the driving force behind the selectivity and reliability of the obtained results because
this serves for the separation of the analyte peaks from background and matrix
signals.
All used trademarks are the property of Thermo Fisher Scientific and its subsidiaries.
This information is not intended to encourage use of these products in any manners that might infringe the
intellectual property rights of others.
6 A Strategy for an Unknown Screening Approach on Environmental Samples Using HRAM Mass Spectrometry
www.thermofisher.com
©2016 Thermo Fisher Scientific Inc. All rights reserved. All trademarks are the property of Thermo Fisher Scientific, Inc. and its
subsidiaries. Specifications, terms and pricing are subject to change. Not all products are available in all countries. Please consult
your local sales representative for details.
Africa-Other +27 11 570 1840
Australia +61 3 9757 4300
Austria +43 1 333 50 34 0
Belgium +32 53 73 42 41
Canada +1 800 530 8447
China +86 10 8419 3588
Denmark +45 70 23 62 60
Europe-Other +43 1 333 50 34 0
Finland/Norway/Sweden
+46 8 556 468 00
France +33 1 60 92 48 00
Germany +49 6103 408 1014
India +91 22 6742 9434
Italy +39 02 950 591
Japan +81 45 453 9100
Latin America +1 561 688 8700
Middle East +43 1 333 50 34 0
Netherlands +31 76 579 55 55
New Zealand +64 9 980 6700
Russia/CIS +43 1 333 50 34 0
South Africa +27 11 570 1840
Spain +34 914 845 965
Switzerland +41 61 716 77 00
UK +44 1442 233555
USA +1 800 532 4752
ASMS13_T578_OScheibner_E 08/16S