SLAS Discovery - Special Issue on Advances in MALDI Mass Spectrometry for Drug Discovery - Volume 22 Issue 10 - December 2017 - 1214

1214
certain class of molecules with a more specific MALDI
preparation.
To assess the impact of different matrices on the ability of
MALDI MS to detect small molecules, a panel of 13 matrices
was evaluated for low background and high signal to noise
(S/N) ratio for the low mass range (Suppl. Fig. S1A). Eight of
the 13 matrices tested yielded S/N > 100 and mean AUC > 5 ×
105 for the tested compounds at concentrations as little as 625
nM (Suppl. Fig. S1B). Co-crystallizing with 2,5-DHB showed
promising data with low background in the low mass range;
however, the signal intensity for each compound also suffered.
Although some matrices provided higher S/N and overall
intensities, CHCA had fewer impurities in the spectrum that
potentially overlap with, and complicate the analysis of, the
small molecules of interest. Overall, the CHCA matrix
achieved a good balance between S/N and total compound
intensity and was therefore selected for small-molecule detection using SAMDI for the remainder of this study.

Solution- versus Solid-Phase Capture for SAMDI
Small-Molecule Binding
In the SAMDI small-molecule binding assay workflow, we
first tested individual compounds known to bind and inhibit
BACE either in a solution-phase or solid-phase capture
to determine the best format for retaining specific smallmolecule binding to a protein target. In the solution-phase
format, compounds and BACE were incubated to form specific complexes in an assay plate and then transferred to a
neutravidin-coated SAMDI plate followed by a washing
step. In contrast, for the solid-phase capture format, the biotinylated BACE first bound directly onto the neutravidincoated SAMDI plate followed by a washing step before
compound addition for binding to occur.
An n = 4 for each of four compounds of varying molecular
weights and affinities were compared individually between
solution-phase and solid-phase capture (Fig. 2A). The
solution-phase capture exhibited approximately twofold
higher signal when compared with the solid-phase capture.
Although the solid-phase capture would result in a higher local
concentration of protein and may yield better detection of
weak binders, the solution-phase capture will allow more efficient compound-protein complex formation prior to SAMDI
capture, as seen by the overall increase in signal and reduced
variability within each sample. Based on these results, the
solution-phase capture was considered the ideal format for this
work.

Compound Dose Response Detected by SAMDI
Small-Molecule Assay
Using the solution-phase capture approach, four compounds
from a chemical series with varying affinities were titrated
individually and incubated with biotinylated BACE. The

SLAS Discovery 22(10)
binding activity of these compounds was previously characterized using BACE time-resolved fluorescence resonance
energy transfer (TR-FRET, HTRF CisBio) biochemical
assay.24 The four compounds showed that the SAMDI binding EC50 values (mean values plotted of n = 4 for each compound) ranked in a similar order with the BACE HTRF
activity data (Fig. 2B). In addition, the maximum percentage bound at 10 µM compound concentration in the SAMDI
binding assay correlated with the biochemical activity data,
where the weakest compound, B4 at 1.8 µM in the biochemical HTRF assay, showed a maximum percentage bound of
25% and an EC50 of 1.6 µM in the SAMDI binding format.
The most potent compound tested, B1 with an HTRF IC50
of 6 nM, showed 100% bound at 10 µM and the highest
binding affinity EC50 value of 80 nM in the SAMDI binding
format. The lower maximal bound of low-affinity compounds may be the result of faster off-rate compounds dissociating from the protein during the wash step. Optimization
of the wash conditions for each protein target will likely
improve the retention of weak binders.

Plate Uniformity
For HTS of compound libraries, data uniformity determines
the reproducibility of the assay format. Because of the spread
in kinetic energies after ionization, MALDI can have signal
variation for the same sample. We assessed the signal variation by alternating columns of samples by incubating 1 µM
compound, B1 (MW = 374 Da, HTRF IC50 = 6 nM), with and
without 1 µM of biotinylated-BACE protein, across a 384well microplate (Fig. 2C), which allowed assessment of consistency of signal across the plate. The mean AUC for
compound B1 for samples without BACE was <1000, consistent with the absence of small-molecule detection. In contrast, samples with BACE added resulted in a compound
mean AUC of 3 × 105 for compound B1. This represents a
300-fold compound signal difference in samples with BACE
compared with samples without BACE. Despite the large
AUC associated with target-specific compound signal, the
variability of the AUC values was significant with a coefficient of variation (CV) = 48%. To address this issue, we
incorporated a normalization step that took into account the
variability introduced at any given sample-processing step
and from MS detection. Because EG3 thiol comprises the
major species of the SAMDI monolayer with a MW of 335
Da, similar to the MW range of a typical small molecule, it
can serve as an internal control for each well. Normalizing to
the EG3 thiol signal minimized signal variability introduced
by the MALDI ionization process and decreased the standard
deviation for the compound signal. After EG3 thiol normalization (Fig. 2C), the %CV of the samples with BACE
decreased from 48% to 22%, and the S/N increased from
300-fold to a window of 400-fold. Future work to further
reduce variation will involve analyzing more sample spots to

Table of Contents for the Digital Edition of SLAS Discovery - Special Issue on Advances in MALDI Mass Spectrometry for Drug Discovery - Volume 22 Issue 10 - December 2017