The successful application of high-throughput
screening (HTS) as a
tool for drug discovery research has
driven the development of both high-throughput organic synthesis (HTOS)
for compound collection expansion
and the need for high-throughput LC-MS analysis for compound library
quality assessment. Providing rapid
evaluation of both the purity and
quantity of material present in large
numbers of samples in the absence of
individual, fully characterized reference
standards (e.g., as required for
quantitation using UV detection) represents
an ongoing challenge for pharmaceutical
industry research.
A preliminary evaluation of charged aerosol detection (CAD) as an addition
to the LC-MS quantitation
toolbox was performed at Berlex
Biosciences
(Richmond, CA). In the
preliminary phase of this study, the
detector response of CAD was compared
to evaporative light scattering detection (ELSD) for a series of common
LC-MS calibration standards.
Based on the initial observations, a
set of diverse standards with appropriate
chromatographic characteristics
were selected to generate a calibration
curve for quantitation. Finally,
20 building blocks from the HTOS
reagent collection were analyzed to
test the general applicability of the
calibration approach.
Background
The ultimate goal of HTS of large
compound collections in drug discovery
research is the identification
of biologically active compounds of
pharmacological interest as starting points for drug discovery programs.
Improvements in both biological
screening methodology and HTOS
have made the overall hit-to-lead
process substantially more effective.
Clearly, the usefulness of the biological
data obtained is enhanced by
improving the quality of the samples
screened. For this purpose, quality can
be defined as both purity and quantity
of active ingredient in the solution
used for screening.
The purity of HTOS compounds is
often measured using HPLC separation
with UV detection coupled to mass spectrometry for identification
of the desired compound. However,
significant differences in UV molar
absorptivity for typical HTS collection
compounds mean that the method
cannot be used for quantitation without
well-characterized standards for
each sample.
Alternative detection techniques,
including ELSD and chemiluminescent
nitrogen detection (CLND),
have been investigated for quantitation
of HTOS compounds.1,2 Both of
these techniques have been successfully
applied under a narrow range of
conditions. However, ELSD has a limited
sensitivity range, leading to overestimation
of purity through underestimation
of volatile impurities. In
addition, ELSD response signals show
significant dependence on the organic modifier content of the mobile phase,
leading to nonuniform response factors
during gradient HPLC analyses.
By the nature of the technique,
CLND is only applicable for nitrogen
containing compounds. Thus, the
method does not detect impurities
that do not contain nitrogen, again
leading to overestimation of sample
purity. In addition, CLND cannot be
used with typical HPLC mobile phases
with nitrogen containing solvents or
mobile phase additives.
The recent introduction of a commercially
available CAD instrument
(ESA, Chelmsford, MA) has led to
considerable interest in evaluating
this method as a “universal” detector
for both purity detection and quantitation
in high-throughput analytical
laboratory applications.3,4 In common
with ELSD, the CAD method vaporizes
the HPLC mobile phase eluent in
a nebulizer chamber. The vapor phase
containing the analyte then flows into
a high-voltage chamber in which a
secondary carrier gas stream of positively
charged nitrogen is introduced.
A corona discharge produces positively
charged analyte particles that are carried
in the gas phase through an electrometer,
where the signal response is measured. Thus, the technique offers
the potential for higher sensitivity to
a wider range of molecular weight species
when compared to ELSD, which
relies on particle size dependent light
scattering detection of analytes.
Experimental
All studies at Berlex were performed
using a single-quadrupole ZMD 2000
mass spectrometer (Waters Corp.,
Milford, MA) with electrospray ionization
(ESI) interface, an LC-20AD HPLC pump (Shimadzu Scientific Instruments, Columbia, MD),
an HTC-PAL Compact autosampler
(LEAP Technologies, Carrboro,
NC), a model 2487 UV detector
(Waters Corp.), a Corona CAD
(ESA, Chelmsford, MA), and a Sedex
75 ELSD (SEDERE, Lawrenceville,
NJ). The CAD and ELSD data were
collected in separate experiments, in
which each detector was connected in
series to the outlet of the UV detector
in the LC-MS system. Standards
and test compounds were prepared as
dimethylsulfoxide (DMSO) solutions
and diluted to target concentrations
with a 1:1 methanol/water mixture.
Flow injection experiments used acetonitrile/water mixtures (with 0.01% trifluoroacetic acid [TFA]) as the carrier
phase with a 0.55 mL/min flow
rate. For LC-MS experiments, a Synergi
hydro-RP column, 2 × 50 mm, 4
μm (Phenomenex, Torrance, CA),
was used with a 1–99% acetonitrile/water (±0.01% TFA) gradient over 5
min with a 0.55 mL/min flow rate.
Results and discussion
Comparison of CAD and ELSD
Figure 1 - Flow injection MS-ELSD signal response.
A series of commonly used LC-MS
standards was analyzed to compare the
detector response of CAD with ELSD.
Flow injection analysis (FIA) was used
at constant concentration (125 µM)
across a range of solvent compositions
to confirm the previously reported
dependence of both ELSD and CAD
signals on the organic modifier content
of the mobile phase in LC-MS
studies.5,6 The results of the comparison
of the CAD and ELSD signal
responses for propranolol, theophylline,
and thymidine are summarized
in Figure 1 for ELSD and in Figure 2
for CAD.
Figure 2 - Flow injection MS-CAD signal response.
The ELSD and CAD signals both show
a molecular weight dependence on
response factor. However, the ELSD
signal exhibits a nonlinear response
with respect to the organic modifier
content of the carrier phase (Figure 1).
In comparison, the CAD signal has an
essentially linear response to carrier
phase organic modifier content (Figure
2). This observation suggested the
potential for developing a generalized
calibration curve for quantitation of
diverse compounds with similar elution
times in an LC-MS-CAD analytical
system. If the retention times of the
analytes are sufficiently close to the
standards chosen, then a reasonable
estimate of the quantity of material
present can potentially be obtained.
Calibration standard
selection
Figure 3 - LC-MS-CAD calibration samples.
Figure 4 - LC-MS-CAD calibration curves.
The series of calibration standards
(Figure 3) was tested to evaluate
the potential for a generalized CAD
calibration approach. The results for
LC-MS-CAD analysis for these standards
as a function of concentration are shown in Figure 4. Three of the
samples elute with similar retention
times (RT = 2.24–2.88 min, Line A).
Therefore, the mobile phase composition
at the point of detection
for L-tryptophan-2-naphthylamide,
propranolol, and diphenhydramine
should be comparable across the concentration
range. The exception is
adenosine (RT = 0.89 min, Line B);
therefore, this compound was not used
for the calibration curve when the
sample test set was analyzed.